2013 ADA Posters 386-1338.indd - American Diabetes Association
Transcription
2013 ADA Posters 386-1338.indd - American Diabetes Association
CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE CATEGORY MONITORING AND SENSING training. A linear patient-specific state-space model, combined with generic compartment models describing the insulin pharmacokinetics and the glucose rate of appearance from the gut, was identified from the CGM and meal and insulin data, using the subspace method N4SID, to describe the post-prandial impact of these inputs. The CIR of each patient was calculated by taking the ratio between the summarized amounts of carbohydrates and bolus insulin over the entire study period, and compared to the CIR estimates using the 500rule and the model, with a correlation of 0.5 and 0.8, respectively. The results suggest that the suggested method could be used as a means to improve the estimates of CIR from a limited set of data. 858-P The Impact of a Video Phone Reminder System on Glycemic Control in Older Adults With Type 2 Diabetes Mellitus (T2DM) in a Retirement Home LUCIA NOVAK, SUSAN WALKER, STEPHANIE FONDA, VIRGINIA SCHMIDT, ROBERT VIGERSKY, Bethesda, MD Interactive behavior change technologies (IBCT) - web portals connecting patients to their providers and personal health information, web-based care management programs, and mobile phone applications - have been shown to be helpful for improving outcomes in diabetes. We have shown that daily, diabetes-related video-based tips and reminders, sent to patients’ mobile phones, were associated with a clinically significant improvement in A1C (Bell AM et al. J Diab Sci Tech 6:310-319, 2012). We have developed a similar intervention for older adults residing in the Armed Forces Retirement Home in which the tips were tailored to the specific needs of this population and produced by providers familiar to the subjects. We conducted a prospective, randomized pilot study in subjects ages 70-89 with T2DM for 4-30 years. They were randomized to usual care (UC) or to UC plus receiving daily 1530 second video message (VM) on a bedside phone for 6 months. A1C was measured at baseline and quarterly for up to 12 months. We found that 4 of 5 VM subjects and only 1 of 4 UC subjects improved their A1C at 3 months. The trend continued at 6 and 9 months but dropouts prevented analysis at 12 months (Figure). These data suggest that video tips and reminders about self care management behavior may improve glycemic control in people with DM in retirement home settings. Supported by: The European Union & 860-P Effect of Different Plasma Insulin Concentrations (PI) on Interstitial Glucose (IG): Implications for Continuous Glucose Monitoring (CGM) Supported by: United States Army Medical Research and Materiel Command CLINICAL THERAPEUTICS/NEW TECHNOLOGY— GLUCOSE MONITORING AND SENSING Guided Audio Tour: Utility of Continuous Glucose Monitoring (Posters: 859-P to 866-P), see page 15. & 859-P Model-Based Estimates of the Post-Prandial Response to Carbohydrate and Insulin and of the Carbohydrate-to-Insulin Ratio FREDRIK STÅHL, ROLF JOHANSSON, ERIC RENARD, Lund, Sweden, Montpellier, France The Carbohydrate-to-Insulin Ratio (CIR) is a measure of how many units of rapid acting insulin to administer to match the digested amount of carbohydrates, and this metric is commonly used in optimizing both the Multi-Dose Injection (MDI) and insulin pump regime of insulin dependent diabetes patients (IDDM). To estimate the value for a specific patient, a heuristic rule of dividing 500 by the Total Daily Insulin (TDI) has been suggested. This so-called 500-rule often produces poor estimates, and to improve these estimates a model-based approach is suggested. In total, 90 patient IDDM data sets were collected in a three day hospitalized trial. The patients were equipped with Continuous Glucose Measurement systems (Abbott Freestyle), received standardized meals for breakfast, lunch and dinner (45, 70 and 70 gram), and the amount of bolus and basal doses were noted. 47 (27 pump, 20 MDI, age 42 [22-68], BMI 24.5 [16.8-35.9], HbA1c 7.8 [5.6-9.7], TDI 42 [6-82]) of the 90 data sets fulfilled quality inclusion criteria (no missing meals/insulin doses in the record, no gaps in the CGM data). For each patient data set, a segment of 24 hours was selected for model ADA-Funded Research & Supported by: European Union (252085); Ministry of Science (DPI2011-28112C04-01) For author disclosure information, see page 829. Guided Audio Tour poster A217 POSTERS Current CGM accuracy during hypoglycemia is still poor. This might be explained by insulin-induced changes of the plasma-to-interstitium glucose dynamics, not considered by CGM calibration algorithms. The aim of the present study was to assess the role, if any, of different PI on IG. Fourteen subjects with type 1 diabetes (36.5±10.5 years, HbA1c 7.9±0.4%) were evaluated under conditions of hyperinsulinemic eu- and hypoglycemia. Following an insulin-feedback period to standardize initial plasma glucose (PG), each subject underwent two randomized crossover clamps with either a primed 0.3mU/kg/min (low insulin, LIS) or 1mU/kg/min (high insulin, HIS) insulin infusion. A hypoglycemic plateau of 45 minutes was achieved after a euglycemic phase, recovering later euglycemia (Figure). Each subject wore two CGM monitors per study (Paradigm VEO, Medtronic, CA). Due to different sensor sensitivities, the raw CGM signal (nA) was normalized (CGMn) with the median pre-clamp values before statistical analysis (ANOVA). Despite much greater mean PI concentrations in HIS vs LIS (58.8±1.9 vs 24.7±1.4µU/ml), neither PG (79.5±4.8 vs 80.2±4.3 mg/dl, p=0.13) nor IG expressed as CGMn (0.83±0.14 vs 0.84±0.13, p=0.95) were significantly different (Figure). In summary, differences of PI in the physiological range do not affect IG concentrations. Other factors responsible for poor CGM accuracy under hypoglycemia should be investigated. Clinical Diabetes/ Therapeutics PAOLO ROSSETTI, FRANCISCO JAVIER AMPUDIA-BLASCO, FÀTIMA BARCELÓRICO, JUAN F. ASCASO, JORGE BONDIA, Gandia, Spain, Valencia, Spain CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING & MAGE and CONGA were calculated from 12-14 days of continuous glucose monitoring. Cell adhesion molecules ICAM1 and VCAM were measured in plasma by ELISA. GV was significantly higher in patients with MiVC (n=17) while their HbA1C, ICAM1 and VCAM were comparable to patients without complications. No correlation of GV and ICAM1 or VCAM was observed in patients with MiVC, however, in the patients without MiVC, ICAM1 correlated with SD (r=0.66, p=0.007), MAGE (r=0.64, p=0.01) and CONGA (r=0.78, p<0.001) while VCAM correlated with MAGE (r=0.56, p=0.031). Although GV was higher in patients with MiVC, it correlated with ICAM1 and VCAM independently on the presence of MiVC. In our study, GV was associated with endothelial dysfunction prior to the development of MiVC. This may indicate a potential role of GV for the future development of MiVC in Type 1 diabetes and requires further research. 861-P Glucose Differences in Patients With Type 1 Diabetes Who Continue versus Discontinue CGM Use POSTERS Clinical Diabetes/ Therapeutics SAMUEL L. ELLIS, ROBERT B. MCQUEEN, DAVID MAAHS, KAVITA V. NAIR, HEATHER D. ANDERSON, ANNE M. LIBBY, JON D. CAMPBELL, Aurora, CO There is limited data on the clinical effectiveness supporting the use of continuous glucose monitors (CGM). We recognize there is a high discontinuation rate of CGM use in adult patients with type 1 diabetes within the first 6 months, but limited data evaluates the reasons why patients selfdiscontinue CGM and the effects of self-discontinuation on glucose control. We retrospectively identified adult T1D patients at the Barbara Davis Center who first initiated CGM from 2006 to 2011. We conducted surveys on these patients to determine who self-discontinued CGM use during the first year of use and the most commons reasons for this. Adherence to CGM was estimated from survey recall and defined as average number of days/ month of CGM use during follow-up. Change in A1c was calculated as the difference between the baseline value and lowest follow-up value while patients used CGM. A total of 66 patients completed the CGM survey. Baseline mean (SD) age for CGM users was 39 (12) years; duration of diabetes was 23 (12) years; and A1c was 7.47%. The most common CGM product was DexCom (42%). A total of 23 (35%) CGM users self-discontinued use within the first year. The most common reasons patients self-discontinued CGM were accuracy, numerous alarms, skin reactions/sensor discomfort and cost. CGM users who selfdiscontinued within the first year were most likely to do so within 6 months (>70%). While using CGM those individuals who self-discontinued had minimal A1c reduction compared to those who continued (0.63% vs 0.32%; p=0.07). The within group changes were significant for both the continuers and self-discontinuers p< 0.0001 and p=0.017, respectively. Further research needs to be done to improve adherence to CGM so patients can achieve maximal clinical benefits. & 862-P Supported by: Charles University Specificity and Precision Assessment of a Long Term, Fully Implantable Continuous Glucose Monitoring (CGM) System & CARRIE LORENZ, WENDOLYN QUINTANILLA, COLLEEN MDINGI, SRINIVASAN RAJARAMAN, MARK MORTELLARO, Germantown, MD CGM use has been shown to be associated with improvements in A1C. Currently available CGM systems, which use enzymes for glucose signaling, are approved for short durations of use (≤ 7 days). Senseonics, Inc. has developed a CGM system intended for up to 180 days of use that utilizes an abiotic glucose transducer rather than an enzyme. Specificity of the abiotic recognition system for glucose over potentially interfering substances has been assessed in vitro (PBS). No mono-saccharide was observed to interfere over physiologically relevant concentrations and no signal response was observed for di-saccharides (e.g., sucrose, maltose) over any concentration tested. Glucose measurement precision of the CGM system has been assessed in vitro and in vivo. Precision was characterized in pH 7.4 PBS buffer at 32°C and 37°C across 3 glucose levels using 53 CGM sensors from 3 different production lots. Percent coefficient of variation (% CV) for glucose concentrations > 75 mg/dL were below 3% and standard deviations (SD) for concentrations < 75 mg/dL were below 3 mg/dL. In vivo glucose precision of the CGM system was characterized using data obtained from a 28 day human clinical trial on a subset of 4 human subjects each inserted with 2 sensors (1 in each arm) read continuously and simultaneously for 3 days. All sensors were recalibrated twice per day and display blinded. An overall in vivo percent absolute relative difference (PARD) of 6.0% and % CV of 4.5% was obtained. & 864-P The Impact of Continuous Glucose Monitoring (CGM) on Low Interstitial Glucose Values and Low Blood Glucose Values Assessed by Point of Care Glucose Meters (POC-GM) NORBERT HERMANNS, BEATRIX SCHUMANN, BERNHARD KULZER, THOMAS HAAK, Bad Mergentheim, Germany The impact of CGM on the duration of periods with low glucose measured in interstitial fluid is well known. But studies showing an impact of CGM on low blood glucose values, measured by POC-GM are rare. This crossover study examines the impact of CGM on duration of periods spent in low interstitial glucose range (70 mg/dl) as well as on the proportion of low blood glucose (BG) values measured by POC-GM (Glukometer 3000, Bio Sensor Technology, Germany)and time until detection of low BG by use of POC-GM. 41 type 1 diabetic patients (age 42.0 ±11.3 yrs, diabetes duration 15.3 ±10.1; A1c 8.2 ±1.4%) used the DEXCOM 7 Plus CGM system twice; once participants were blinded against the results and in the other study phase (Open GCM) patients received real-time glucose values and current glucose trends, used the CGM data to determine their insulin dose, and were alerted if hypoglycaemic or hyperglycaemic glucose ranges were approached. In addition, BG was routinely measured 6 times a day by POC-GM. The order of study phases was randomized. The time spent in a hypoglycaemic glucose range (< 70 mg/dl) was reduced by open CGM from 180.6 ±125 to 125 ± 89.2 minutes per day (p=.005). Also time spent in the euglycemic glucose range was increased from 946 ±176 to 1023 ±168 minutes per day (p=.003). Open CGM reduced the proportion of low BG measurements by POC-GM from 10.3 ±7.6% to 7.4 ±5.8% (p=.039), whereas the euglycemic POC readings was increased from 68.3 ±12.1% to 73.7 ±12.1% (p=.007). The time until a hypoglycaemic BG value was detected by POC-GM was significantly (p.028) shortened by 33.1 (95% CI 3.8 -62.3) minutes. These results demonstrate that CGM used for insulin dosing and hypoglycaemia alerts not only diminishes low interstitial glucose values but also reduces the proportion of low BG measured by POC-GM. In addition the time for the detection of a hypoglycaemic episode by POC-GM was significantly shortened. 863-P Glycemic Variability Is Associated With Markers of Endothelial Dysfunction: Cell Adhesion Molecules ICAM-1 and VCAM in Type 1 Diabetes Without Complications MARTIN PRAZNY, JAN SOUPAL, MARTIN FAJMON, JAN SKRHA JR., JAN KVASNICKA, JAN SKRHA, Prague, Czech Republic Glycemic variability (GV) is probably associated with vascular complications in patients with diabetes. However, it is not clear whether increased GV directly contributes to their development. In this study we compared GV with markers of endothelial dysfunction in Type 1 diabetic patients with and without microvascular complications (MiVC). Thirty-two patients were included (age 44 ± 13 yrs, HbA1C 70 ± 9 mmol/M, no macrovascular complications). Glucose standard deviation (SD), coefficient of variation (CV), Supported by: DexCom, Inc. & For author disclosure information, see page 829. A218 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING Guided Audio Tour: Blood Glucose Monitoring (Posters: 867-P to 874-P), see page 15. & BRADLEY C. LIANG, TALY ENGEL, CYRUS A. ROUSHAN, STACIE HALLER-WICH, XIAOLONG LI, MEGAN E. LITTLE, KEITH NOGUEIRA, ASHLEY SULLIVAN, JOHN B. WELSH, JEROME S. FISCHER, RAJIV SHAH, FRANCINE R. KAUFMAN, SCOTT W. LEE, Northridge, CA, San Antonio, TX LAURENCE B. KATZ, LUPE R. MILLER, RANDEE A. RANDOLL, RIAD G. DIRANI, GANG LI, West Chester, PA, Milpitas, CA This study is a retrospective evaluation of the 7-day performance of a novel subcutaneous amperometric glucose oxidase-based glucose sensor. Its design utilizes high-density circular array electrodes on a flexible polyimide substrate over which an enzymatic layer and protective membranes are deposited. Electrodes and membranes were optimized for responsiveness, linearity, and signal consistency. The sensor is 9 mm in length (implant volume 0.17 mm3) with mechanical properties that minimize wear-induced tissue inflammation. Data were analyzed with a prototype real-time continuous glucose monitoring (CGM) algorithm. Twenty subjects with type 1 or type 2 diabetes were enrolled into a singlesite feasibility study. Each subject wore either 1 or 2 of the glucose sensors on the abdomen connected to iPro2 recording devices (Medtronic) over two 7-day wear periods. Blood glucose (BG) concentration was measured using Contour NEXT Link meters (Bayer); collected points were used for both calibration and accuracy determination. Subjects performed 3 frequent sampling tests (on days 1, 3-4, and 6-7), each consisting of 12 or more BG samples over a period of 3 hours. Sensor data were analyzed with the real-time calibration algorithm configured to select calibration points twice daily. Sensor data were evaluated against all collected BG points at times recorded by the meter (2333 evaluation points over 266 sensor days). The mean absolute relative difference (MARD) between sensor and reference values was 11.80% (14.37%, 10.02%, and 11.43% on days 1, 3-4, and 6-7) with just 1 of the 43 sensors (2.33%) having a calculated MARD >20%. The mean functional life was 6.17 d, 84.14% of readings met ISO15197:2003 accuracy criteria, and 98.29% were in Clarke Error Grid zones A+B. The system provided consistent and accurate readings when compared to a BG meter reference. Such improvements in the accuracy and reliability of Medtronic’s next-generation CGM system are part of the technology evolution required to support development of the Artificial Pancreas. & Improving glycemic control in patients with diabetes is challenging for patients and their healthcare providers (HCPs). Tools that allow rapid evaluation of blood glucose (BG) patterns by HCPs during short office visits and provide insight to patients between visits to enable positive changes may support these efforts. This study evaluated HCP experience with a BG meter under development (OneTouch®Verio™) which uses pattern recognition and other automatic messages to provide patients with insight into their BG results. The accuracy and time for 64 HCPs to identify high and low BG patterns according to meter algorithms in 5 simulated logbooks containing 30 days of BG data was compared to using meters containing identical data. In addition, HCPs were given a demonstration of possible automatic meter messages patients may see when testing on their own, and asked if these messages may help their patients manage their BG. HCPs took more time to find patterns using a logbook (4.0 min; 3.7-4.3 min, 95% Confidence Interval) compared to using the meter (1.0 min; 0.91.1 min) (p<0.001). The total error rate for finding patterns using a logbook compared to patterns in the meter was 26.0% (18-34%, p<0.0001) for all low and high patterns. Most (70%) of the HCPs felt using the meter was easier than reviewing a logbook. Between 77-97% of the HCPs found high value in the features of the meter and felt that insights provided by the automatic messages may help their patients improve management of BG between office visits. Ninety-two percent (92%) of the HCPs had a favorable level of satisfaction to the meter and 89% would recommend it to all their patients; 91% to their patients on insulin. In summary, using the new OneTouch® Verio™ meter was 4 times faster and more accurate than using logbooks to help HCPs identify BG patterns. Based on responses to a survey, HCPs liked the meter and would recommend it to their patients. 866-P & Relationship between Glycemic Variability and Blood Pressure Variability in Diabetic Patients With Hypertension: Jikei Variability of ABPM and CGM Study HUGH D. TILDESLEY, ADAM S. WHITE, MARY-ELLEN CONWAY, STUART A. ROSS, AUGUSTINE M. LEE, HAMISH G. TILDESLEY, JEREMY H.M. CHAN, ADEL B. MAZANDERANI, Vancouver, BC, Canada, Hamilton, ON, Canada, Calgary, AB, Canada, Hanover, NH, Nashville, TN, St. George’s, Grenada Although recent studies have reported that both glycemic variability (GV) and blood pressure (BP) variability contribute to the development of cardiovascular disease, little is known about the relationship between GV and BP variability in diabetic patients with hypertension. We evaluated GV, BP variability and progression of arteriosclerosis using continuous glucose monitoring (CGM), ambulatory blood pressure monitoring (ABPM) and Cardio Ankle Vascular Index (CAVI). ABPM and CGM underwent at same time for straight 2 days (48 hours). Hypertension was defined as diastolic blood pressure systolic blood pressure [SBP] ə130 mmHg or, [DBP] ə80 mmHg or were taking antihypertensive medicine. Variability was estimated by Coefficient of variation (CV: Standard Deviation/average ambulatory BP or glucose). This was a prospective study. 64 adult diabetic patients with hypertension were enrolled in this study (mean: Age 53±12 years, BMI 26.7±3.8kg/m2, HbA1c 8.2±1.6%, diabetes duration 9.2±3.6 years, Patients takingα-Glucosidase inhibitor or glinide or using rapid acting insulin were excluded). The mean 24-h blood glucose level (mg/dl) and CV (%) were 144.2±36.8 and 21.2±8.8. The mean 24-h BP level (mmHg) and CV (%) were as follows, Systolic:126.6±14.6 and 10.8±2.6, Diastolic: 78.0±9.9 and 12.9±3.6. CAVI was 8.1±1.3 (normal range: <8.0). There were significant correlations with GV and 24-h systolic pressure (Pearson’s r=0.30, P=0.02), daytime systolic BP variability (Pearson’s r=0.31, P=0.04) and pulse pressure (Pearson’s r=0.41, P=0.002). There was no relationship between GV and diastolic BP. Interestingly, we observed a strong correlation between pulse pressure and CAVI by correlation analysis (Pearson’s r=0.478, P<0.001). We conclude that increased systolic BP variability and pulse pressure together related to GV and that both might finally lead to development of cardiovascular disease. & 868-P A Review of the Effect of Internet Therapeutic Intervention in Patients With Type 1 and Type 2 Diabetes MASAYA SAKAMOTO, HIROFUMI SUZUKI, HIROYUKI IUCHI, KENNOSUKE OHHASHI, TAKESHI HAYASHI, RIMEI NISHIMURA, KATSUYOSHI TOJO, KAZUNORI UTSUNOMIYA, Tokyo, Japan ADA-Funded Research 867-P A Novel Blood Glucose Meter that Provides Real-Time Automatic Messaging: Benefi ts to HCPs and Patients It is now our standard of care to offer an Internet Based Glucose Monitoring System. We reviewed the first 409 patients registered, of whom 388 patients had baseline and follow-up A1C. Patients were instructed to send their results every two weeks to their endocrinologist for review and feedback. A1C and laboratory results were requested at 3 month intervals. We compared baseline values to follow-up, results summarized in table 1. Regardless of type or treatment, all groups improved A1C. Frequent reporters showed greater drops in A1C shown in table 2. The observed difference in drops of frequent vs. infrequent reporters were found to be significant across all groups (p < 0.05). Type 1 patients with A1C at goal reported sporadically. When we excluded patients with A1C at goal, the difference in frequent reporters vs. infrequent reporters was significant. Table - 1 PAIRED SAMPLES T-TEST BASELINE AND 6 MONTHS TYPE N BASELINE (SD) 6 MONTHS (SD) ALL USERS 388 8.20 (1.27) 7.87 (1.15) TYPE 1 * excluding A1C < 6.9 115 8.12 (1.38) 7.93 (1.17) ALL TYPE 2 256 8.36 (1.35) 7.91 (0.98) TYPE 2 - OHA’S 116 8.15 (0.98) 7.67 (1.29) TYPE 2 - INSULIN +/- OHA’S 140 8.53 (0.82) 8.12 (0.91) P VALUE 2.86 E -9 0.0064 1.04E-9 1.26 E -5 1.13 E -5 For author disclosure information, see page 829. Guided Audio Tour poster A219 POSTERS Clinical Evaluation of a Prototype 7-Day CGM System 865-P Clinical Diabetes/ Therapeutics & CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING these limits). Three POCT systems for professional use and the three lots of the system for self-monitoring had more than 95% of the measurement results within these more restrictive limits. This study demonstrated that BG systems for self-monitoring which have to fulfil strict requirements prior to market introduction can achieve similar system accuracy as POCT systems for professional use. Table 2 - PAIRED T-TEST BY FREQUENCY (Frequent = >Once a month, Infrequent = <Once a month) TYPE N REPORTING BASELINE (SD)6 MONTHS (SD) P VALUE TYPE 1 *excluding A1C < 7.4 44 FREQUENT 8.07 (0.84) 7.85 (1.02) 0.035 TYPE 1 *excluding A1C < 7.4 71 INFREQUENT 8.14 (1.15) 7.98 (1.00) 0.043 TYPE 2 - OHA’S 35 FREQUENT 8.03 (1.25) 7.48 (0.85) 0.0018 TYPE 2 - OHA’S 81 INFREQUENT 8.21 (1.42) 7.75 (1.27) 0.00074 TYPE 2 - INSULIN +/- OHA’S 41 FREQUENT 8.42 (1.40) 7.86 (1.39) 0.00007 TYPE 2 - INSULIN +/- OHA’S 99 INFREQUENT 8.57 (1.42) 8.22 (1.08) 0.0013 & Supported by: Sanofi & GREGORY G. FOTIEO, CHRISTOPHER S. WENDEL, JAYENDRA H. SHAH, ELENA V. PLUMMER, ALLISON E. MURATA, HEATHER C. BRISLEN, GERALD A. CHARLTON, GLEN H. MURATA, Albuquerque, NM, Tucson, AZ, Phoenix, AZ 869-P The purpose of this study was to determine if HbA1c is better correlated with mean glucose or the area under the curve (AUC) derived from a 7-point glucose profile. Sixty-nine patients from the VA Southwest Health Care Network participated in this study. Subjects provided a blood sample for HbA1c and then measured their blood glucoses before and 2 hours after 3 daily meals and at bedtime for 3 consecutive days. A 7-point profile was constructed for each subject from the average reading for each time of day and average sampling intervals. AUC was defined as the total area under the glucose versus time plot. Mean glucose was defined as the average of the 7 profile values. The fractional contribution of each reading to AUC was derived from the expression: reading * (mean of adjacent sampling periods)/ AUC. The mean (SD) age of the subjects was 64.5 (7.8) years, all were men, and 40% belonged to an ethnic minority. Simple linear regression showed a positive correlation between HbA1c and mean glucose (r=0.338; P=0.007). However, more of the variance in HbA1c was explained by regressing HbA1c against AUC (r=0.375; P=0.002). After adjusting for multiple comparisons, no differences were found between the pre-breakfast reading (PRB) and all other values or the bedtime reading (BED) and all other values. Moreover, no differences were found between the fractional contributions of PRB and BED to AUC [0.233 (0.044) vs 0.244 (0.046), respectively; P=NS]. However, their contributions to AUC were each more than twice that of any other reading (P<0.001 in all cases). Among diabetic veterans, AUC is somewhat better correlated with HbA1c than mean glucose derived from glucose profiles. More importantly, AUC provides information on the relative importance of readings by giving greater weight to those that bracket longer time intervals. Our study supports the concept that protein glycation is a function of glucose concentration and exposure time and suggests that choosing treatment targets should be based upon sampling times as well as glucose values. JAE HYOUNG CHO, HUN SUNG KIM, SEUNG HYU YOO, CHANG HEE JUNG, WOO JE LEE, CHUL YOUNG PARK, JOONG YEOL PARK, SUNG WOO PARK, HO YOUNG SON, KUN HO YOON, Seoul, Republic of Korea POSTERS Clinical Diabetes/ Therapeutics Large Scale Study of the Internet-Based Diabetes Management System With an Automatic Data Uploading Device for Type 2 Diabetes Patients in Korea Objective: To improve the quality of diabetes control, we developed an Internet based interactive communication system with an automatic data uploading device and investigated its effectiveness on controlling the changes in HbA1c levels and blood pressure. Research Design and Methods: We conducted a large-scale, multi-center, randomized clinical trial involving 484 patients who visited the outpatient clinic at the Seoul St. Mary’s Hospital, Seoul Asan Hospital and Samsung Kangbook Hospital for 6 months. Intervention group (n=244) were treated with the management system for 24 weeks, and the control group (n=240) received the usual outpatient management over the same period. HbA1c , blood pressure and other laboratory tests were performed at the beginning of the study, at 3 month and the end of the study. Results: There was no significant differences between the two groups with respect to age, sex, diabetes duration, BMI, blood pressure, HbA1c and other laboratory data. At the end of the study . On follow-up examination 24 weeks later, HbA1c level were significantly decreased from 7.86± 0.69 to 7.55 ± 0.86% within intervention group (p<0.001) compared to 7.81 ± 0.66 to 7.70 ± 0.88% within control group. A subgroup with baseline HbA1c more than 8% and good compliance achieved a big reduction of HbA1c by 0.8 ± 1.05%, compared to a small reduction of 0.37 ± 0.82% in a subgroup with bad compliance. Conclusions: This new internet-based diabetes management system with an automatic data uploading device resulted in a significant reduction of HbA1c during the study period and patients with good compliance showed a more reduction of HbA1c. We propose that such the system can be a new method for improving diabetes control. & 871-P Correlating HbA1c With Glucose Profiles: A Comparison of Two Methods Supported by: Roche Diagnostics Corporation & 872-P The Impact of Age on Interventions to Improve Glycemic Control in Diabetes Patients on Multiple Daily Insulin Injection (MDI) Therapy: Results from the Automated Bolus Advisor Control and Usability Study (ABACUS) 870-P Evaluating System Accuracy of Blood Glucose Monitoring Systems for Point of Care Testing IAIN CRANSTON, DAVID A. CAVAN, RALPH ZIEGLER, KATHARINE BARNARD, CHRISTOPHER G. PARKIN, WALTER KOEHLER, BETTINA PETERSEN, MATTHIAS A. SCHWEITZER, ROBIN S. WAGNER, Portsmouth, United Kingdom, Bournemouth, United Kingdom, Muenster, Germany, Southampton, United Kingdom, Boulder City, NV, Mannheim, Germany, Indianapolis, IN GUIDO FRECKMANN, ANNETTE BAUMSTARK, MANUELA LINK, STEFAN PLEUS, CORNELIA HAUG, JOCHEN SIEBER, Ulm, Germany, Frankfurt, Germany Point of care testing (POCT) is performed with blood glucose (BG) monitoring systems for professional use only and with systems designated for self-monitoring of BG. These systems are often used interchangeably, thus raising the question for comparability of measurement quality. In this study, we evaluated the accuracy of a BG system for self-monitoring (BGStar, Sanofi) and six different, commonly used systems for professional use (HemoCue Glucose 201+, HEMOCUE AB; StatStrip, nova biomedical; ACCU-CHEK Inform II, Roche Diagnostics; BIOSEN C_line, EKF-diagnostic; HITADO SUPER GL compact, Dr. Müller Gerätebau; GLUKOMETER PRO, BST Bio Sensor Technology). Investigational procedures were performed following the international standard ISO 15197. Measurement results from capillary blood samples of 100 subjects were analyzed by calculating the percentage of results within ±10%/±15% or within ±10 mg/dL/±15 mg/dL of the reference measurement results for BG concentrations above or below 100 mg/dL, respectively. Glucose oxidase was used as comparison method (YSI 2300 STAT Plus, YSI Life Sciences, USA). From each BG monitoring system two devices and one lot of reagents were used, except BGStar, for which three reagent lots were evaluated separately. For all evaluated systems 99% to 100% of the measurement results were within ±15 mg/dL or ±15% of the reference method results. When applying more restrictive limits of ±10 mg/dL or ±10% larger differences between the evaluated systems were observed (81.5% to 97.5% of the results within Managing younger patients with diabetes is often problematic. We assessed the impact of bolus advisor use in patients who participated in the ABACUS trial, a 26-week, prospective, randomized, controlled, multinational study that enrolled 218, poorly controlled MDI-treated diabetes patients (202 type 1, 16 type 2) with mean (SD) baseline A1C 8.9 (1.2)%, age 42.4 (14.0) years and BMI 26.5 (4.2) kg/m2. Baseline age of patients was stratified to 18-30 years (n=53) or > 30 years (n=165), with younger adults having higher A1C values than older patients (9.3 [1.6]% vs. 8.7 [1.0]%, p < 0.05) and shorter duration of diabetes (11.8 [6.3] years vs. 19.6 [11.6] years, p < 0.01). Patients randomized to the control group (CNL) used a standard blood glucose meter and calculated bolus insulin dosages manually; patients randomized to intervention group (EXP) used the Accu-Chek Expert meter with an automated bolus advisor to calculate insulin dosages. Among the 100 EXP patients who completed the study, bolus advisor use was associated with a significantly higher percentage of patients achieving the primary goal of > 0.5% A1C reduction than CNL patients (n=93): 56.0% vs. 34.4%, p < 0.01. Although older patients (> 30 years) in both groups were more likely to achieve the primary outcome than younger patients (49% vs. 33%), more younger EXP than CNL patients achieved > 0.5% A1C reductions: 53% vs. 15%, p < 0.05. The mean age of those who achieved the goal within the & For author disclosure information, see page 829. A220 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING with the instructions for use and with the ISO accuracy testing protocol in a clinical setting. EXP group was lower compared with those who did not: 42.4 (12.8) years vs. 44.3 (14.2) years. However, within the CNL group, the mean age was higher among those who achieved the A1C goal compared with those who did not: 49.4 (13.7) years vs. 41.4 (13.9) years. This suggests that while bolus advisor use is beneficial in both age groups, it may be especially useful in younger adults, a population in which glycemic control is often difficult to achieve. Supported by: Sanofi 875-P Use of an Automated Bolus Advisor Reduces Glycemic Variability in Poorly Controlled Diabetes Patients Treated With Multiple Daily Insulin Injection (MDI) Therapy: Results from the Automated Bolus Advisor Control and Usability Study (ABACUS) Supported by: Roche Diagnostics GmbH 873-P LESLIE J. KLAFF, RONALD L. BRAZG, KRISTEN HUGHES, ANN M. TIDEMAN, HOLLY C. SCHACHNER, PATRICIA STENGER, SCOTT PARDO, NANCY DUNNE, JOAN LEE PARKES, Renton, WA, Tarrytown, NY Calculating insulin dosages is challenging for most diabetic patients on MDI therapy. We examined the effect of bolus advisor use on glycemic variability (GV) as measured by mean amplitude of glucose excursions (MAGE) in a subset of patients enrolled in the ABACUS trial, a 26-week, prospective, randomized, controlled, multi-national study of poorly controlled, MDI-treated patients, with mean (SD) baseline A1C 8.9 (1.2)%. Patients randomized to the control group (CNL) used a standard blood glucose meter and calculated bolus insulin dosages manually; intervention patients (EXP) used the Accu-Chek Expert meter with an automated bolus advisor. Ninety-three patients (90 type 1 diabetes) wore a continuous glucose monitoring (CGM) device for 3 days at 3 time periods. Only EXP patients showed significant reductions in mean MAGE: -20.2 (41.1) mg/dL / -1.1 (2.3) mmol/L, p < 0.01, vs. -2.9 (32.1) mg/dL / -0.2 (1.8) mmol/L, p = NS. In patients with baseline A1C of 7.5-9.0%, improvement was even greater in EXP than CNL patients: -23.2 (35.3) mg/dL / -1.3 (2.0) mmol/L, p < 0.01, vs. -5.3 (33.2) mg/dL / -0.3 (1.8) mmol/L, p = NS. In patients who achieved ≤ 0.5% A1C improvement, EXP by not CNL patients showed significant MAGE reductions: -15.3 (29.3) mg/dL / -0.9 (1.6) mmol/L, p < 0.05, vs. 1.1 (35.8) mg/ dL / 0.1 (2.0) mmol/L, p = NS; a similar trend was seen in those who achieved > 0.5% A1C reductions. Changes in MAGE correlated with SD of glucose values (r = 0.87) and were independent of education level, diabetes duration, duration of MDI or gender. The percentage of blood glucose values < 50 mg/dL / 2.7 mmol/L was < 2% in both groups throughout the study. Use of the Accu-Chek Expert meter may reduce GV by facilitating more accurate dosage calculations without increasing the risk of hypoglycemia. This study compared the accuracy of the CONTOUR® NEXT BGMS with 5 other BGMSs (Accu-Chek® Aviva Nano, Freestyle Lite®, OneTouch® Ultra® 2, OneTouch® Verio™ Pro, and Truetrack®). Subjects (N = 146) were ≥18 years and had diabetes. Subjects’ glucose levels were safely lowered or raised to provide a wide range of glucose values. Study staff tested fingerstick samples on the 6 BGMSs. Extreme glucose values were achieved by blood sample glucose modification. The primary endpoint was comparison of the mean absolute difference (MAD) from the reference value (YSI) in the low glucose range (<70 mg/dL). In this sub-analysis, a comparison was done using mean absolute relative difference (MARD) between the BGMSs and YSI values in the low range and across the entire range for both modified and unmodified samples. For all samples, in the low glucose range and across the entire range (21-496 mg/dL), CONTOUR® NEXT had a statistically significantly lower MARD than all the other BGMSs (Table 1). In unmodified samples, in the low glucose range, CONTOUR® NEXT had a lower MARD than all BGMSs and was significantly lower (P <0.05) except for Freestyle Lite®. In unmodified samples over the entire range, CONTOUR® NEXT had a statistically significantly lower MARD than all the BGMSs. In conclusion, when compared with other BGMSs, CONTOUR® NEXT demonstrated the lowest mean deviation from the reference value in the low and overall glucose range (MARD) with modified and unmodified samples. Supported by: Roche Diagnostics GmbH 876-P Low Glucose Suspend Systems: CGM Signal Attenuation Detection to Reduce Undesirable Pump Suspensions NIHAT BAYSAL, FRASER CAMERON, BRUCE A. BUCKINGHAM, DARRELL M. WILSON, H. PETER CHASE, DAVID MAAHS, B. WAYNE BEQUETTE, Troy, NY, Stanford, CA, Aurora, CO & Low glucose suspend (pump shut-off) algorithms use continuous glucose monitor (CGM) signals to predict when glucose values are likely to violate a hypoglycemic threshold, and therefore shut-off the continuous insulin infusion pump. In in-patient and out-patient studies we have shown that our Kalman-filter based pump shut-off algorithm reduces the risk of overnight hypoglycemia by 50%, with an 11 mg/dL increase in mean blood glucose. Unfortunately there are a number of false hypoglycemic predictions that cause the pump to shut-off unnecessarily; many of these false detections are due to pressure-induced sensor attenuations (PISA) that can occur, e.g., when someone rolls over their sensor while in bed. We compare 3 different methods for mitigating the effect of PISAs to avoid undesirable pump shut-offs. Method 1 prevents suspensions when the CGM readings drop faster than 8 mg/dL/min, and prevents suspensions above 230 mg/dL. Method 2 adds to these rules the removal of CGM readings that are part of a sharp, exponential decay from a rising or gently falling glucose trajectory and the following exponential restoration. Method 3 removes readings between sharp declines and two consecutive second derivatives. We test the effectiveness of the mitigation against the base Kalman filter algorithm applied to the original dataset and to a “fixed” version where the PISAs (total of 282 CGM readings) were manually removed. The dataset consists of 60 nights (~6000 readings at 5 min intervals) of outpatient CGM data from pump shut-off studies. On the original and “fixed” datasets the base algorithm suspended for 1020 and 920 readings, respectively, so there were 100 readings or 500 min of undesired pump suspensions from PISAs. When the three methods are run on the original dataset 3/15/25% of these undesired pump suspensions were removed. These results show that the 874-P Current Blood Glucose Meters Perform in Compliance With New Draft ISO Guidelines MARCUS BORCHERT, CARINA HENGESBACH, FILIZ DEMIRCIK, CHRISTINA SCHIPPER, THOMAS FORST, ANDREAS PFÜTZNER, PETRA B. MUSHOLT, Mainz, Germany ISO15197 accuracy performance criteria have to be met by devices for blood glucose measurement: 95 % of readings have to be within ±15 mg/dL for values < 75 mg/dL, and within ±20 % for values ≥100 mg/dL. The currently reviewed new draft ISO guideline suggests stricter accuracy acceptance criteria (95 % of readings have to be within ±15 mg/dL for values < 100 mg/ dL, and within ±15 % for values ≥100 mg/dL). We studied the compliance six blood glucose meters with the new guideline by performing a post-hoc analysis of data obtained from an ISO-conform clinical accuracy performance study. This study had been performed with 106 patients, clinically presenting with blood glucose levels distributed over the entire measurement range and in line with the glucose distribution requirements as demanded by the ISO guideline. The YSI 2300 STAT Plus™ analyzer (glucose-oxidase) served as reference method. All tested meters were in compliance with the current ISO criteria. The following meters met the new criteria: Accu-Chek Aviva (95.6 %/99.1 %), BG*Star (99.3 %/97.5 %), iBG*Star (98.5 %/97.5 %), FreeStyle Freedom Lite (95.6 %/96.5 %), and OneTouch Ultra2 (95.6 %/96.3 %). One meter failed the low blood glucose range criteria (Contour: 90.4 %/97.5 %). Almost all branded blood glucose meters, including BG*Star and iBG*Star met the new draft ISO15197 acceptance criteria, when tested in accordance ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A221 POSTERS RALPH ZIEGLER, DAVID A. CAVAN, IAIN CRANSTON, KATHARINE BARNARD, CHRISTOPHER G. PARKIN, WALTER KOEHLER, BETTINA PETERSEN, MATTHIAS A. SCHWEITZER, ROBIN S. WAGNER, Muenster, Germany, Bournemouth, United Kingdom, Portsmouth, United Kingdom, Southampton, United Kingdom, Boulder City, NV, Mannheim, Germany, Indianapolis, IN Clinical Diabetes/ Therapeutics & Comparative Accuracy Evaluation of Six Blood Glucose Monitoring Systems (BGMSs) CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING application of a PISA detection algorithm would prevent unnecessary pump suspensions when CGM signals are attenuated. BG, number of BG >180 mg/dl, and with higher total insulin requirement (U/ day), all p<0.001 (Table). The number of hypoglycemic events (<70mg/dL) were inversely related to the admission HbA1c level (p= 0.046). There were no association between admission HbA1c and hospital complications. In summary, this analysis indicates that the admission HbA1c value is a useful marker in predicting inpatient glycemic control and hypoglycemic events in non-ICU patients with T2D. Future randomized controlled studies should take the admission HbA1c into consideration for selecting the starting insulin dose in patients with T2D. Supported by: JDRF (22-2011-647) 877-P Use of an Automated Bolus Advisor Improves Glycemic Control With Increased Treatment Satisfaction in Poorly Controlled Diabetes: Results from the Automated Bolus Advisor Control and Usability Study (ABACUS) POSTERS Clinical Diabetes/ Therapeutics DAVID A. CAVAN, RALPH ZIEGLER, IAIN CRANSTON, KATHARINE BARNARD, JACQUELINE RYDER, CLAUDIA VOGEL, CHRISTOPHER G. PARKIN, WALTER KOEHLER, BETTINA PETERSEN, MATTHIAS A. SCHWEITZER, ROBIN S. WAGNER, Bournemouth, United Kingdom, Muenster, Germany, Portsmouth, United Kingdom, Southampton, United Kingdom, Langen, Germany, Boulder City, NV, Mannheim, Germany, Indianapolis, IN A1C <7% A1C 7-9% A1C >9% p-value Admission BG, mg/dl 156± 53 192± 63 280± 99 p<0.001 Mean daily BG, 148±35 161±29 183±36 p<0.001 mg/dl BG >180 mg/dl during treatment, % 53% 70% 91% p<0.001 # units of insulin/day, U/day 26±13 33±15 38±16 p<0.001 # units of insulin/day, U/kg/day 0.3±0.1 0.3±0.2 0.4±0.2 p<0.001 Patients BG<70 mg/dl, n (%) 18 (18%) 16 (17%) 8 (8%) p=0.046 We conducted a study to assess the impact of using an insulin bolus advisor on glycemic control in patients treated with multiple daily insulin injection (MDI) therapy. The ABACUS trial, a large, 26-week, prospective, randomized, controlled, multi-national study, enrolled 218 poorly controlled MDI-treated patients with diabetes (202 type 1, 16 type 2), with mean (SD) baseline A1C 8.9 (1.2)%, age 42.4 (14.0) years, BMI 26.5 (4.2) kg/m2 and 17.7 (11.1) years duration of diabetes. Patients randomized to the control group (CNL) used a standard blood glucose meter and manual bolus calculation; patients randomized to the intervention group (EXP) used the Accu-Chek Expert meter with an automated bolus advisor to calculate insulin dosages. Glucose data were downloaded and used for therapy parameter adjustments in both groups. A total of 193 patients (CNL, n=93; EXP, n=100) completed the study. Significantly more EXP than CNL patients achieved > 0.5% A1C reduction: 56.0% vs. 34.4% (p < 0.01). (Figure) Improvement in treatment satisfaction (DTSQ scale) was significantly greater in EXP patients: 11.4 (6.0) vs. 9.0 (6.3); p<0.01. Percentage of blood glucose values <50 mg/dL / 2.7 mmol/L was <2% in both groups during the study. Use of the AccuChek Expert meter resulted in improved glycemic control and treatment satisfaction without increasing severe hypoglycemia. 879-P The Association between Self-Monitoring Blood Glucose and Glycemic Control in Insulin Treated Patients With Type 2 Diabetes MIYUKI FURUYA, YUKO HATAJI, HIROHITO KUWATA, YUSUKE ARAI, SATOSHI MATSUNAGA, SHINTARO OKAMURA, TSUYOSHI MASHITANI, MASAKO KITATANI, SHIGEKI IKUSHIMA, YASUAKI HAYASHINO, SATORU TSUJII, HITOSHI ISHII, TENRI COHORT STUDY GROUP, Tenri, Japan The aim of this study is to investigate whether frequency and timing of SMBG are related to glycemic control in insulin-treated patients with type 2 diabetes. We obtained data of 1496 patients [average age, 66.1 years; 42.0% of females] with insulin-treated type2 diabetes from a registry that registered all outpatients with diabetes visiting the Diabetes Center at the Tenri Hospital, Japan.The patients answered the questionnaires regarding the frequency, number of measurements per day and the timing of SMBG. We evaluated the association between the frequency and HbA1c values using linear regression analyses. Furthermore, we proceeded to subgroup analysis by the number of insulin injections (1, 2, 3, >or 4 daily injections or CSII). Next, patients were categorized into two groups based on the timing of SMBG, group A (n=1231): patients who perform only pre-prandial testing, group B (n=263); patients who perform not only pre-prandial but also postprandial testing at least once a week. We compared HbA1c value in both groups. On average, patients performed 1.8 measurements of SMBG per day. Mean HbA1c were 8.1, 8.2, 8.0, 7.7, 7.6 and 7.5% for groups of the frequency of SMBG: none, rarely, once/day, twice/day, 3 times/day and 4 times or more/ day, respectively. Higher frequency of measurements was associated with lower HbA1c levels (p<0.0001), and there were no changes in these findings no matter what times patients shot insulin in the day. Mean HbA1c was 7.8% in the group A and 7.8% in the group B (p=NS). There were no differences between the groups in glycemic control by the timing of SMBG. The crosssectional survey showed higher frequency of SMBG testing was associated with better glycemic control regardless of the number of insulin injections in patients with type 2 diabetes. Meanwhile we did not observe significant association between the timing of SMBG and glycemic control. 880-P Supported by: Roche Diagnostics GmbH Laboratory Study for Evaluation of the Influence of Temperature and Humidity on Blood Glucose Measurements With BGStar and iBGStar 878-P Value of Admission Hemoglobin A1c (HbA1c) in Predicting Inpatient Glycemic Control and Clinical Outcome in Non-ICU Patients With Type 2 Diabetes PETRA B. MUSHOLT, CARINA HENGESBACH, FILIZ DEMIRCIK, CHRISTINA SCHIPPER, THOMAS FORST, ANDREAS PFÜTZNER, Mainz, Germany Dynamic electrochemistry performs multiple measurements under different electrochemical conditions for one reading. It applies a mathematical algorithm for correction of confounding factors, based on the different kinetics involved in the signal generation of confounding factors. This laboratory study investigated the impact of temperature and humidity on blood glucose meters employing dynamic electrochemistry (BGStar and iBGStar, Sanofi) in comparison to 3 competitive blood glucose meters (Contour, Bayer; One Touch Ultra 2 and OneTouch Verio Pro, Lifescan). Ten meters and 3 strip lots per type were conditioned to the selected temperature (12°, 25°, 38°) and humidity (<25 %, 50-60 %, 80-90 %) in an approved Climate Cabinet for 1 h prior to the performance of the measurements. Heparinized blood samples were manipulated to contain 3 different levels of blood glucose (60-90 mg/dL, 150-200 mg/dL, 250-350 mg/dL). Normal oxygen pressure was confirmed and glucose was tested with the reference method (YSI Stat 2300). Samples were FRANCISCO PASQUEL, SAIRA ADEEL, FARNOOSH FARROKHI, ISABEL ANZOLA, LIMIN PENG, DAWN SMILEY, GUILLERMO E. UMPIERREZ, Atlanta, GA Clinical guidelines recommend measuring HbA1c on admission to assist tailoring insulin therapy in the hospital and after discharge in non-ICU patients with T2D; however, the predictive value of admission HbA1c on inpatient glycemic control and outcomes has not been determined. Accordingly, we analyzed medical records of 296 general medicine and surgery patients (age: 58±12 yrs, admission BG: 212±92 mg/dl, HbA1c: 8.6±2.5%) treated with basal bolus regimen starting at 0.4 U/kg/day if BG was between 140-200 mg/dl and 0.5 U/kg/day if BG was between 200-400 mg/dl, given half as glargine once daily and half as glulisine before meals. Patients were divided into three groups according to HbA1c value: <7% (n=99), 7-9% (n=92), and >9% (n=105). Higher HbA1c directly correlated with admission BG, hospital daily & For author disclosure information, see page 829. A222 Guided Audio Tour poster ADA-Funded Research mean interstitial glucose (MG) and glucose variability (GV) in the risk of hypoglycemia in patients with type 1 diabetes (T1D). Continuous glucose monitoring using CGMS® system, was performed in 130 T1D patients, with diabetes duration of 17,1±8,6 years, in intensive insulin therapy (49.8±17.9 UI per day). MG (in mg/dL), GV (SD of MG, in mg/dL), time per day spent in hypoglycemia (HT), interstitial glucose ≤70mg/dl, in hours, and episodes of nocturnal hypoglycemia (NH), hypoglycemia between midnight and 8 a.m., in %, were assessed. Patients were divided in group I (n = 84) with HbA1C≤7,5% and group II (n = 46) with HbA1C>7,5%. Statistical analysis was performed using SPSS, version 21.0. Group I presented a significantly lower MG (139.2±25.9 vs173.1±33.2mg/dL, p<0,05) and GV (58.4±18.8 vs. 70.3±18.6mg/dL, p<0,05) and more HT (1,85±1.68 vs. 1,30±1.04 hours p<0,05). NH episodes weren’t significantly different between groups (8.7 vs. 5.8%). HT was positively correlated with GV (r=0.23, p=0.01) and negatively with HbA1C and MG (r=-0.23 and r=-0.36; p=0.01). NH was correlated with MG (r= - 0.24, p<0.05). In multivariate analysis, GV and MG were associated with TH (β= 0.22 and β= -0.15, p < 0.01, respectively), independent of HbA1c, diabetes duration and insulin dose; NH was only associated with MG (OR=0.9, p< 0.05). NH was associated with 16-fold increased risk of asymptomatic hypoglycemia (OR:16.9; p<0.01). In conclusion, patients with high HbA1C still remain at risk of hypoglycemia. Glucose variability independently predicts daily time duration spent in hypoglycemia. At night, hypoglycemia only correlates with MG, suggesting daily fluctuations are probably due to inadequate insulin coverage of the meals. Higher MG during night can eventually reduce risk of asymptomatic hypoglycemia. Supported by: Sanofi 881-P Prediction of Plasma Glucose via Exhaled Breath Analysis in Obesity and Type 2 Diabetes STACY R. OLIVER, HEATHER SOPHER, MATTHEW K. CARLSON, SIMONE MEINARDI, DONALD R. BLAKE, PIETRO R. GALASSETTI, Irvine, CA The life-threatening cardiovascular complications of obesity (OW) and type 2 diabetes (T2D) are linked to glycemic control, currently based on invasive, blood-based testing of plasma glucose. Breath analysis is increasingly considered an alternative, non-invasive, painless methodology for glycemic testing, able to improve patients’ compliance/ metabolic control. Its practical therapeutic applications, though, remain limited. We hypothesized that accurate breath-based glucose predictions can be obtained with gas clusters specific for each patient group. We studied 17 T2D (10M) and 12 OW (6M) subjects over 4 h in basal, hyperglycemic (220 mg/dL) and euglycemic-hyperinsulinemic conditions. At 12 time points ~100 volatile organic compounds (VOC) were quantified by gas chromatography on room and breath samples, and matched with plasma glucose. Gas-based glycemic predictions were generated by least squares regression on several VOC clusters. The strongest correlations between measured and predicted glucose values were seen using acetone, butanone, EtONO2, 2- and 3-PeONO2, and 2-BuONO2 in OW (r = 0.88, range 0.66-0.97), and acetone, MeONO2, methanol, propane, 2-PeONO2, and ethanol, in T2D (r = 0.902, range 0.73-0.99). Our data show how accuracy of breath-based metabolic tests increases by tailoring gas profiles to specific dysmetabolic conditions, and confirm the growing potential of breath analysis for non-invasive diagnosis/metabolic monitoring in diabetes. 883-P Safety and Accuracy Evaluation of San MediTech’s Real-Time Continuous Glucose Monitoring (RT-CGM) Device HUASHI ZHANG, DAN WANG, WENBEI CHEN, JING WU, IRVINE, CA To evaluate the safety and accuracy of the San MediTech’s Real-Time CGM device with the comparison of the capillary and the venous blood glucose reference values, respectively.Twenty Subjects (4 Type I and 16 Type II diabetes mellitus) were enrolled with Beijing Jian Heng Diabetes Hospital from June, 2012 to October, 2012.Each patient wore two SMT’s RT-CGM devices. Each day 7 capillary glucose values were measured with the glucose meter (Bayer Breeze2, Germany) On Day 3 and Day 5, eight venous blood glucose references were measured for each subjects once every 30 min during a 4hr period. The total numbers of the capillary and the venous blood glucose reference values used for the accuracy study are 1118 pairs (<4.4mmol/L:4.2%;4.4-13.3mmol/L:65.3%; > 13.3mmol/L:30.5%) and 495 pairs (< 4.4mmol/L:4.8%;4.4-13.3mmol/L:56.6%;> 13.3mmol/L:38.6%), respectively.All the subjects finished and no safety adverse event was observed. The MARD (mean absolute relative difference) values of SMT’s RT-CGM device compared with the capillary and venous references values are 16.8% and 15.9%, respectively. The Clarke error grid analysis plot of the SMT’s RT-CGM values VS the capillary blood glucose references is shown in Figure 1.The combined A zone and B zone percentage is 96.8%. More importantly, for the hypo glycemic range (1.7-4.4 mmol/L, n=47 pairs), the SMT’s RT-CGM device has the MARD as 26.4% and the combined A+B percentage as 76.6%. Supported by: NIH (1UL1RR031985), (K24DK085223) 882-P HbA1C, Glucose Variability and Hypoglycemia Risk in Patients With Type 1 Diabetes Mellitus DANIELA F. GUELHO, LUÍSA BARROS, CARLA BAPTISTA, ISABEL PAIVA, SOFIA GOUVEIA, JOANA SARAIVA, CAROLINA MORENO, LUÍSA RUAS, MANUELA CARVALHEIRO, FRANCISCO CARRILHO, Coimbra, Portugal Supported by: America-Asia Diabetes Research Foundation Many diabetic patients experience high variability in glucose concentrations which is associated with a risk of hypoglycemia. The aim of this study was to compare the predictive value of hemoglobin A1C (HbA1C), ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A223 POSTERS briefly equilibrated to the cabinet conditions and measured within the cabinet (via small door inlets). Mean values and standard deviations for each meter type/sample combination were calculated, and the mean absolute relative deviation (MARD) and precision were determined. MARD was found to be 5.3±2.7 % for BGStar and 6.5 ± 4.0 % for iBGStar (Contour: 5.3±0.8 %, Ultra2: 8.6±0.3 %, Verio Pro: 4.0±0.1 %). Deviations were slightly higher under dry condition and at cold temperature. Cold temperature also led to more frequent devices errors. Mean precision was 3.4±1.9 % for BGStar and 3.7±1.6 % for iBGStar (Contour: 3.1±0.1 %, Ultra2: 3.3±1.2 %, Verio Pro: 2.9±0.2 %). With a general imprecision of 3-4 CV% and an inaccuracy well below 10 DEV%, iBGStar and BGStar performed in line with current standards of patient self blood glucose monitoring. Generally, measurements should not be performed at cold ambient temperature conditions. Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING POSTERS Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING 884-P 886-P Glycemic Variability Is Associated With Subclinical Atherosclerosis in Chinese Type 2 Diabetic Patients Differences between Continuous Glucose Monitoring (CGM) Systems may Influence Frequency of CGM Usage JIAN ZHOU, YIFEI MO, MEI LI, YUQIAN BAO, XIAOJING MA, DI LI, WEI LU, CHENG HU, WEIPING JIA, Shanghai, China JAMES CHAMBERLAIN, DANA DOPITA, EMILY GILGEN, Salt Lake City, UT Realization of the clinical benefits of continuous glucose monitoring (CGM) in diabetes management is dependent upon the frequency at which patients use their CGM devices. To determine whether frequency of sensor use is related to CGM technology, in general, or differences between available CGM systems, we surveyed 81 type 1 adult diabetes patients with at least one year of CGM experience. The 8-item survey was offered to patients on an “as seen” basis; all patients completed the survey; 43 patients (all on insulin pumps) used the Medtronic MiniLink REAL-Time CGM (ML) and 38 patients (25 on insulin pumps) used the Dexcom SEVEN PLUS (SP) device. Results showed that 19% of ML users wore their CGM device “almost daily” compared with 76% of SP patients; 65% if ML users wore their device ≤ 1 week per month compared with 3% of SP users. (Table) Most “almost daily” ML users reported “improved bG” as their main reason for frequent use. The most common reason for frequent SP users was “liked knowing bG values at all times”. Thirty-one percent of ML users reported ‘CGM did not seem accurate” as the main reason for less frequent use. Reasons for infrequent use among SP patients showed no pattern. More than 92% (n=35) of SP patients reported they would purchase the same CGM system again compared with 44% (n=19) of ML patients. Differences in perceptions of performance and usability may strongly influence frequency of CGM use, which can impact clinical outcomes. The contribution of glycemic variability to macrovascular complications remains unclear. We therefore investigated the association between glycemic variability and cervical and/or intracranial atherosclerosis in Chinese type 2 diabetic patients. We conducted a cross-sectional study in 216 type 2 diabetic patients with a hemoglobin A1c of 8.3 ± 1.7% and a median diabetes duration of 9.0 years. The standard deviation of blood glucose values (SDBG) and the mean amplitude of glycemic excursion (MAGE) were calculated from continuous glucose monitoring system data for assessing glycemic variability while 24h mean blood glucose (MBG) was calculated for measuring overall blood glucose level. Magnetic resonance angiography (MRA) was used to detect cervical and/or intracranial plaque, and ultrasonography was used to quantify carotid intima-media thickness (IMT) as an index of subclinical atherosclerosis. One hundred and fifty-three patients (70.8%) presented with cervical and/or intracranial lesions on MRA among 216 patients in the study. Elder age, increased systolic blood pressure, increased MBG and elevated low density lipoprotein cholesterol were independent contributors to plaque formation. In patients without stenosis (n = 63), SDBG (r = 0.412, P = 0.001) and MAGE (r = 0.365, P = 0.005) were both correlated with carotid IMT and these relationships remained significant in multiple linear regression analysis (multiple R2 = 0.314 for the model including SDBG and multiple R2 = 0.268 for the model including MAGE). However, SDBG and MAGE were not significantly different among patients with different stenosis degrees. Taken together, our study suggests that glycemic variability is associated with subclinicalatherosclerosis in Chinese type 2 diabetic patients. 885-P Use of an Automated Bolus Advisor Is Associated With Improved Competence in Carbohydrate Counting in Patients Treated With Multiple Daily Insulin Injections (MDI): Results from the Automated Bolus Advisor Control and Usability Study (ABACUS) DAVID A. CAVAN, IAIN CRANSTON, RALPH ZIEGLER, KATHARINE BARNARD, JACQUELINE RYDER, CLAUDIA VOGEL, CHRISTOPHER G. PARKIN, WALTER KOEHLER, BETTINA PETERSEN, MATTHIAS A. SCHWEITZER, ROBIN S. WAGNER, Bournemouth, United Kingdom, Portsmouth, United Kingdom, Muenster, Germany, Southampton, United Kingdom, Langen, Germany, Boulder City, NV, Mannheim, Germany, Indianapolis, IN 887-P 24-Hour Glucose Spectra Evaluation of Continuous Ambulatory Peritoneal Dialysis Patients QIN TAN, JIANBIN LIU, JUAN LIU, YANBING LI, Guangzhou, China We assessed the carbohydrate (CHO) counting competency of 218 patients, treated with multiple daily insulin injection (MDI) therapy, who participated in the Automated Bolus Advisor Control and Usability Study (ABACUS), a large, 26-week, multi-national, prospective, randomized, controlled trial that evaluated the use of an automated bolus advisor in poorly controlled MDI-treated type 1 diabetes and type 2 diabetes patients with baseline A1C 8.9 (1.2)%. At entry, patients were asked to assess the carbohydrate content (grams) of 10 standardized meals (using life-size photographs), from which the mean meal error (MME), an indicator of accuracy, and mean meal absolute error (MMAE), an indicator of variability were calculated. All patients then received refresher training and carbohydrate reference materials prior to randomization. Patients randomized to the control group (CNL) used a standard blood glucose meter and calculated bolus insulin dosages manually; patients randomized to the intervention group (EXP) used the Accu-Chek Expert meter with an automated bolus advisor to calculate insulin dosages. Patients were followed for six months, with regular reviews by their healthcare team. The carbohydrate counting assessment was repeated at the end of the study and showed significant improvement in MMAE (15.2 [9.0] grams to 12.4 [7.3] grams, p < 0.01) and a trend towards improvement in MME (1.0 [10.1] grams to 0.3 [7.1] grams, p = NS) in the EXP group but no change in CNL patients. This suggests that use of a bolus advisor may improve competency in carbohydrate counting by providing frequent feedback on patient accuracy in estimating the carbohydrate intake. Thus, automated bolus advisors may serve as educational tools that can improve the effectiveness of diabetes self-management in patients treated with MDI. Glucose-based peritoneal dialysis(PD) solution is commonly used in patients with diabetes undergoing continuous ambulatory peritoneal dialysis (CAPD), but how it affects glucose fluctuation during PD remains unknown. The aim of our study was to investigate 24-hour blood glucose profile and fluctuation following peritoneal exchange (PE) in patients undergoing CAPD with glucose-based PD solution evaluated by CGMS. Forty-two patients receiving CAPD were enrolled, of whom 22 subjects with diabetic nephropathy (DN group) and 20 were free of diabetes (NDN group). Data of 30 healthy volunteers (control group) was also analyzed. Compared to control group, patients in DN group presented higher blood glucose value and blood glucose fluctuations value (all p<0.05). Patients in DN group also presented similar changes as well as the increment following PEcompared with NDN group (all p<0.05). However, there were no significant difference of these CGMS parameters between NDN group and control group (P>0.05). (Fig.1) For the NDN group, the increment following PE was higher with the use of 2.5%peritoneal solution than 1.5% peritoneal solution (2.13±2.02 vs 2.76±1.51, P=0.005).But there was no significant difference of the impact on blood glucose level with the use of 1.5% or 2.5% peritoneal solution in the DN group(all p>0.05). CAPD with glucose-based PD solution may impose disturbance to glucose homeostasis in patients with ERD, with a more significant impact in those with DN. Supported by: Roche Diagnostics GmbH & For author disclosure information, see page 829. A224 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING of 2.8 times per day; however, daily frequency of advice sought decreased significantly (-0.2/day, p < 0.01) during the course of the study. Older patients tended to use the bolus advisor more often than younger patients: 2.9 vs. 2.3, p < 0.05. There was no difference in bolus advice use between patients who achieved > 0.5% A1C reductions (n= 56) vs. those who achieved ≤ 0.5% A1C reduction (n=44). However, patients with baseline A1C 7.5-9.0% used the bolus advisor more often throughout the study than patients with higher glycemic levels (A1C > 9.0%): 3.1/day vs. 2.2/day, (p < 0.01) The number of times patients accepted the bolus advice sought was similar between age groups, with 2.5/day for all patients. When modifications to bolus advice were made, patients were nearly twice as likely to reduce insulin (0.2/day) vs. increasing insulin (0.1/day), with no difference between the age groups. These results demonstrate that patients frequently sought bolus advice. Further, improvements in A1C were seen despite the small decrease in use over time, suggesting that patients placed a high level of trust in their bolus advisor, learned from experience about their insulin-to-meal response and, thus, required slightly less frequent bolus advice over time. 888-P Pilot Study to Assess the Efficacy of a Continuous Glucose MonitorBased Semi-Automated Basal Assessment Management Protocol ROGER MAZZE, ELLIE STROCK, MATTHEW MURPHY, RICHARD M. BERGENSTAL, Minneapolis, MN ANDREJ ORSZAG, LEIF E. LOVBLOM, HOLLY TSCHIRHART, PETER PICTON, JOSEPH A. CAFAZZO, BRUCE A. PERKINS, Toronto, ON, Canada Few studies use a systematic, evidence-based approach to decisionmaking when comparing SMBG or CGM guided decisions with routine care. We undertook a randomized controlled 36-week study in which patients with type 2 diabetes were assigned to routine care (Controls) or care guided by Staged Diabetes Management (SDM) clinical pathways employing scheduled episodic testing (SMBG) or continuous monitoring (CGM). All subjects wore blinded CGM to produce AGPs which include glucose exposure, variability, stability, and percent hypoglycemia. Sixty patients were enrolled. There was a 1:2 F:M ratio, with average age 60+/-8.6 years and duration 15+/-7 years. There was a significant (p<0.0001) improvement in HbA1c from baseline to end of study for the SMBG (N=18) (8.2 to 7.1%) and CGM (N=24) (8.4 to 7.4%) groups and no significant change for the controls (N=18) (8.0 to 7.7%). Forty-seven patients had sufficient blinded CGM data to produce glucose profiles. Shown in Figure 1 is the change in glucose exposure (SMBG or CGM p<0.004, Controls NS) for the three groups. The CGM group experienced a significant reduction in hypoglycemia (p<0.05) while the other groups did not. Change in glucose variability was not significant, while change in stability was (p<0.02) for SMBG and CGM. There is an urgent need for clinical protocols to guide adjustment of overnight basal rates in current clinical practice. We aimed to evaluate a continuous glucose monitoring (CGM)-based semi-automated basal assessment algorithm on overnight glycemic stability, overnight hypoglycemia incidence and glycemic control. We developed and piloted an algorithm that interpreted CGM data from overnight fasting protocols and recommended basal rate changes based on glucose excursions. Subjects uploaded data online using sensor-augmented insulin pumps (Medtronic 754 pumps and Carelink Personal, Medtronic Inc.) for evaluation by the algorithm after each of 5 overnight fasts conducted over 4-10 weeks. The algorithm was designed to be conservative and iterative: it made hourly basal rate changes that did not exceed 10% at each iteration. The Standard Deviation of CGM values from midnight to 7 am (SD12-7am) over 3 baseline and follow-up nights, hypoglycemia incidence (defined by CGM values <4.0 mmol/L) and A1c were compared by paired sign-rank tests. Twenty subjects (male/female 7/13) with mean age 38±13y and A1c 7.6±0.8 percent underwent the 5 iterations of basal assessments over 7±3 weeks. SD12-7am did not change from baseline to follow-up (1.57±0.8 and 1.63±0.8 mmol/l, p=0.35). Qualitatively, however, mean glucose values were lower between 2 to 3 am at baseline compared to follow-up and 3-night hypoglycemia incidence declined from 1.6±1.8 to 0.5±0.7 episodes (p=0.01). Despite attenuation of overnight declines in mean blood glucose and a decrease in hypoglycemia incidence, the A1c improved from 7.6±0.8 to 7.4±0.9 (p=0.03). Despite not demonstrating improvements in the overnight measures of glycemic variability, we observed significant decline in hypoglycemia incidence and short-term improvement in A1c. Future research must focus on clinical applicability of CGM-based basal adjustment algorithms. 889-P Usage of a Bolus Advisor: Results from the Automated Bolus Advisor Control and Usability Study (ABACUS) RALPH ZIEGLER, DAVID A. CAVAN, IAIN CRANSTON, KATHARINE BARNARD, CHRISTOPHER G. PARKIN, IRIS VESPER, MATTHIAS A. SCHWEITZER, ROBIN S. WAGNER, Muenster, Germany, Bournemouth, United Kingdom, Portsmouth, United Kingdom, Southampton, United Kingdom, Boulder City, NV, Mannheim, Germany, Indianapolis, IN The use of SDM to guide clinical decision-making based on SMBG or CGM produced significant improvement in glycemic control assessed by AGP when compared to routine care. Adjusting insulin doses according to blood glucose and carbohydrate intake is challenging for most diabetic patients on multiple daily insulin injection (MDI) therapy. We examined usage of the Accu-Chek Expert meter with an automated bolus advisor in patients enrolled in the Automated Bolus Advisor Control and Usability Study (ABACUS), a 26-week, prospective, randomized, multi-national study. Data from 105 poorly controlled diabetic patients (98 type 1, 7 type 2), with mean (SD) baseline A1C 8.9 (1.1)%, age 42.7 (13.5) years were analyzed. Patients used their bolus advisor for 73.5% of all possible bolus opportunities. No differences in age, gender or type of diabetes were seen. At the beginning of the study, patients sought bolus advice an average ADA-Funded Research & 891-P Performance Evaluation of a Continuous Glucose Monitoring System Under Conditions Similar to Daily Life STEFAN PLEUS, CORNELIA HAUG, MANUELA LINK, EVA ZSCHORNACK, GUIDO FRECKMANN, Ulm, Germany This study aimed at evaluating the new generation of a continuous glucose monitoring (CGM) system under daily life-like conditions. The DexCom™ G4 CGM system was used with two sensors in parallel by 10 type 1 diabetes subjects for the whole sensor lifetime (approx. 7 days). For author disclosure information, see page 829. Guided Audio Tour poster A225 POSTERS 890-P Randomized, Controlled 36 Week Study of Clinical Decision-Making Employing SMBG and CGM With Analysis by Ambulatory Glucose Profile (AGP) Clinical Diabetes/ Therapeutics Supported by: Roche Diagnostics GmbH CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING diabetes types and ages (currently over 18 years). At the beginning of the trial, each subject was calibrated. The measurements taken throughout the trial were based on this calibration. Subjects participated in the trials for up to 6 months, in order to verify the calibration validity period (6 months). Moreover, feedback regarding device usability and user satisfaction was analyzed. Clarke Error Grid pulled data analysis of 89 subjects (5050 points) over 6 months of operation shows 96% of the points in the clinically accepted A+B zones and ARDmean of 32%. No degradation in performance was noticed as a function of time elapsed from calibration. Feedback shows that 96% of the subjects found the device comfortable; 75% found the device easy to operate, 85% expressed willingness to use the device regularly and 80% declared they will use the device more frequently than the invasive one. GlucoTrack is technically available to use. Its key benefits include 6 months of operation without re-calibration and ability to perform numerous spot measurements with acceptable accuracy. Furthermore, it provides pain-free and inexpensive use, without the need to be continuously worn. These advantages can lead to better blood glucose monitoring adherence and tighter glucose control. Supported by: Roche Diagnostics GmbH POSTERS Clinical Diabetes/ Therapeutics The sensors were applied during study day 1 and removed during study day 8. Performance was assessed by calculation of mean absolute relative deviation (MARD) between CGM readings and paired capillary blood glucose (BG) readings (approx. 15 duplicate measurements per day and per subject) and precision absolute relative deviation (PARD), i.e. differences between the paired CGM readings of the first and the second sensor worn by each subject. Additionally, MARD was separately calculated for BG readings below 70 mg/dL, between 70 and 180 mg/dL and above 180 mg/dL. Total MARD of the sensors used in this study was 10.4%. The MARD for BG below 70 mg/dL was 13.6% (8.3% of BG readings), when BG was between 70 and 180 mg/dL 10.8% (69.5% of BG readings) and for BG above 180 mg/dL 8.0% (22.2% of BG readings). PARD was 7.3% for the complete experiments (n=10), improving from 12.9% and 9.2% on days 1 and 2, respectively, to 5.3% and 5.9% on days 7 and 8, respectively. In this study, the evaluated CGM system showed good overall MARD compared to results reported for other commercially available CGM systems. In the hypoglycemic range, where CGM performance is often reported to be low, the CGM system achieved comparably good MARD. Overall PARD was 7.3% with a marked improvement during the course of the study from mean PARD 12.9% on day 1 to mean PARD 5.9% on day 8. 894-P Poorly Controlled Patients on Multiple Daily Insulin Injections (MDI) Often Improperly Use Insulin Therapy Parameters for Manual Bolus Calculations: Results from the Automated Bolus Advisor Control and Usability Study (ABACUS) 892-P WITHDRAWN ROBIN S. WAGNER, RALPH ZIEGLER, DAVID A. CAVAN, IAIN CRANSTON, KATHARINE BARNARD, CHRISTOPHER PARKIN, WALTER KOEHLER, IRIS VESPER, MATTHIAS A. SCHWEITZER, Indianapolis, IN, Muenster, Germany, Bournemouth, United Kingdom, Portsmouth, United Kingdom, Southampton, United Kingdom, Boulder City, NV, Mannheim, Germany Diabetes patients on MDI therapy utilize insulin to carbohydrate ratios (I:CHO), insulin sensitivity factors (ISF), target range and blood glucose (bG) values to determine appropriate insulin doses. Clinicians use resultant insulin doses and bG values to adjust I:CHO and ISF. We examined use of I:CHO and ISF by control (CNL) patients enrolled in the ABACUS trial, a 26week, prospective, randomized, controlled, multi-national trial. Patients randomized to the CNL group used a standard blood glucose meter and calculated bolus insulin dosages manually; patients randomized to intervention group (EXP) used the Accu-Chek Expert meter with a bolus advisor to calculate insulin dosages. Diary data from 96 CNL patients (87 type 1, 9 type 2) with baseline A1C 8.8 (1.2)% were analyzed. Patientrecorded bolus calculations were checked to determine if the stated I:CHO and ISF rules were used in a mathematically correct manner for each bolus calculation. Patients calculated an average of 4.0 (1.0) boluses per day with no change in frequency during the study. Patients correctly used their I:CHO and ISF rules 1.4 (1.1) and 0.8 (0.7) times per day, respectively. Patients incorrectly used I:CHO and ISF rules 1.7 (0.9) and 2.3 (1.0) times per day, respectively. There was no change in correct or incorrect usage of either parameter during the study; only 34.4% (n=32) achieved the study goal of >0.5% reduction in A1C. The results demonstrate that these MDI patients did not use their parameters correctly for most bolus dose calculations. This may be due to mathematical errors and/ or a lack of trust in their parameters; in which case, patients may have found unknown “compensatory mechanisms”. Although this may explain improvements in those who achieved the A1C goal, this behavior could have a negative impact on therapy adjustment if clinicians are unaware of these discrepancies. Supported by: Roche Diagnostics GmbH 895-P 893-P Presenting a Truly Non-Invasive Glucose Monitor for Home Use Sample Handling Is of Critical Importance for Successful Performance of Laboratory Investigations With Blood Glucose Meters Employing Dynamic Electrochemistry AVNER GAL, ILANA HARMAN-BOEHM, EUGENE NAIDIS, YULIA MAYZEL, NETA GOLDSTEIN, KEREN HORMAN, Ashkelon, Israel, Beer Sheva, Israel GlucoTrack® is a Non-Invasive device for self-monitoring of blood glucose at home, or home-alike environment, combining 3 technologies: Ultrasonic, Electromagnetic and Thermal. The device comprises user friendly Main Unit (MU) and sensors per each technology, located at a Personal Ear Clip (PEC). The PEC is clipped externally to the earlobe for less than a minute, to conduct a real-time spot measurement, after individual calibration. Glucose readings are heard and displayed on the smartphone sized MU with a color touch screen, designed for diabetics needs. The PEC must be replaced every 6 months, to be followed by a calibration. GlucoTrack performances were evaluated in various clinical trials simulating home-alike environment, for subjects of diverse genders, BMI, SANJA RAMLJAK, DAVID C. KLONOFF, PETRA B. MUSHOLT, ELLEN AMBERS, THOMAS FORST, FILIZ DEMIRCIK, ANDREAS PFÜTZNER, Mainz, Germany, San Mateo, CA, San Jose, CA Laboratory proficiency testing is a frequent procedure in many hospitals prior to introduction of new blood glucose meters for in-house use and recommendation to patients. Glucose meters optimized for patient use may fail when tested with laboratory procedures previously used with older meter technologies. Meters employing dynamic electrochemistry (DE) appear to be sensitive to laboratory errors. This study tested the impact of sample handling by experienced and inexperienced personnel on the results of an accuracy and hematocrit (HCT) interference test performed & For author disclosure information, see page 829. A226 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING with iBGStar vs. YSI Stat 2300. Venous heparinized whole blood was immediately aliquoted after draw, and manipulated to contain 3 different blood glucose concentrations (50-80 mg/dL, 120-180 mg/dL, and 250-330 mg/dL) and 3 different HCT levels (30 %, 45 %, and 60 %). Each sample was tested 4 times with 4 devices. Mean absolute relative deviation (MARD) was calculated for all samples. Sample preparation was done independently under observation by three groups: a team very experienced with DE laboratory testing (A), a group experienced with other meters but not DE (B), and a team entirely inexperienced with meter testing (C). Team A ensured physiological pO2 and gentle sample preparation. Team B did not consider pO2 and used laboratory equipment for effective mixing. Team C did everything manually, but also ignored pO2. MARD was 4.3 % with team A, 9.2 % with team B and 5.2 % with team C. Team B had much higher readings in particular at low glucose levels and team C produced 100% of “sample composition” errors with high HCT levels. Laboratory proficiency testing of DE devices should only be performed by trained and experienced experts taking into consideration that the sample handling in the laboratory had a major impact on the accuracy results. Normal hospitals should rather evaluate modern devices like iBGStar with patients and in accordance with the instructions for use. Methods: Pooled, retrospective, descriptive analysis of 160 eligible CGM profiles from T2DM patients with drug treatment over ≥6 months recorded in RCTs in our center (GLA=78; M=82). CGM (Medtronic) was performed for 72h to measure interstitial glucose (iG). AUC-24h, standard deviation of mean iG (SD-iG), mean average glucose excursions (MAGE), and day-to-day variation of glucose lowering effect between day 2 and 3 of CGM were calculated as difference for each couple (day 2 and 3) every 5 minutes of glucose profile. Threshold for HE was <3.0 mmol/L iG. Results: Characteristics of patients for GLA and M treatment were: age 64.0/64.0yrs; BMI: 29.7/30.0 kg/m²; diabetes duration: 4.9/4.7 yrs; A1C: 6.1/6.4%. CGM results: mean iG at day 2: 6.8/7.0 mmol/L (ns); fasting iG: 6.0/6.8mmol/L (p=0.002) AUC-24h: 1936/2005 mmol/L/24h (ns); SD-iG: 1.66/1.40 mmol/L (ns); MAGE: 4.3/3.7 mmol/L (ns). No difference in dayto-day variability was found between groups (mean difference iG: 0.03/0.03 mmol/L (ns) with coefficient of variation: 73.5/75.0 %, ns). Overall HE duration was similar with GLA and M (18.5/10.1 min/24h, ns). Conclusion: Pooled CGM data from early stage T2DM showed no significant difference in glycaemic variability and hypoglycaemia risk with insulin glargine compared to metformin therapy even at A1C control <6.5%. Comparison of Inpatient Glycemic Control by Continuous Glucose Monitoring (CGM) and Capillary Point-Of-Care (POC) Testing in General Medicine Patients With Type 2 Diabetes Treated With Basal Bolus Insulin Regimen Interest of Continuous Glucose Monitoring to Improve Glycemic Control in Patients With Type 2 Diabetes and End-Stage Renal Disease Treated by Basal-Bolus Insulin Regimen LORI KÉPÉNÉKIAN, AGNIESZKA SMAGALA, LAURENT MEYER, OLIVIER IMHOFF, AICHA SISSOKO, LIVIU SERB, THIERRY KRUMMEL, FRANÇOIS DOREY, DOMINIQUE FLEURY, JEAN-PIERRE LE FLOCH, FRANÇOIS CHANTREL, LAURENCE KESSLER, Strasbourg, France, Colmar, France, Mulhouse, France, Valenciennes, France, Villecresnes, France ANA M. GOMEZ, GUILLERMO E. UMPIERREZ, PABLO ASCHNER, FELIPE HERRERA, CLAUDIA P. RUBIO, OSCAR MUÑOZ, Bogotá, Colombia, Atlanta, GA Point-of care testing with capillary glucose finger-stick tests is the universally accepted method of glucose monitoring in the hospital. No previous studies have compared the efficacy of CGM in the management of hyperglycemia in general medicine patients with T2D. Accordingly, this prospective randomized study compared glycemic control by CGM (Sofsensor iPro 2, Medtronic) to bedside POC testing in non-ICU patients treated with basal bolus regimen for ≥ 3 days. Both patients and hospital staff were blinded to the CGM data. POC testing measurements were performed before meals, 2-h after meals, at bedtime and 3 AM. The primary outcomes were differences in daily BG, and number of hypoglycemia (<70 mg/dl) and hyperglycemia (>180 mg/dl) events between groups. A total of 40 insulin-naïve patients (age: 65.8±8 yr, DM duration 14.7±9 yr, admission BG: 251±9 mg/dl, A1C: 9.7±2.4%, ±SD) were treated with glargine and glulisine at a starting total dose of 0.4 U/kg/day if BG was between 140-200 mg/dl and 0.5 U/kg/day if BG was between 200-400 mg/dl, given half as glargine once daily and half as glulisine before meals. We observed no difference in daily BG after the 1st day of treatment by CGM and POC testing (176.2±33.9 vs. 176.6±33.7 mg/dl, p=0.828). There were 10 patients with BG<70 mg/dl recognized by both methods, but CGM detected higher of number of events (55 vs. 12, p=0.0001) than POC testing, with 40% occurring between breakfast and dinner and 60% between dinner and 6 A.M. A total of 26.3% of hypoglycemia were asymptomatic and most (86.7%) were only identified by CGM. The proportion of BG >180 mg/dl was 36.8% by CGM and 42.1% by POC, p=0.828. In summary, the use of CGM recognized increased number of hypoglycemic events compared to POC testing. Our study indicates potential benefit of using real-time CGM in the hospital to detect hypoglycemia in a more timely fashion compared to POC testing. Patients with type 2 diabetes and end-stage renal disease (ESRD) are at high cardiovascular risk. Optimal glucose level control is determinant to reduce morbi-mortality but is difficult to obtain because of haemodialysis session. The continuous glucose monitoring (CGM) could be an useful tool to optimize glycemic control in this population. We conducted a pilot prospective multicenter study to evaluate the contribution of CGM on glucose level control after 3 months of treatment with basal long-acting insulin once daily and rapid-acting insulin 3 times daily (detemir-aspart) in haemodialyzed patients with type 2 diabetes. CGM (Navigator ®, Abbott) was performed during 54 hours including 2 haemodialysis consecutive sessions at 0, 1 and 3 months with determination of the mean CGM glucose value, maximal glucose excursions and MAGE. Insulin was adapted by the CGM values. Comparisons were performed using ANOVA for repeated measures. The analysis concerned 28 patients: 19 males and 9 females, age 65.7±9.4 yr, BMI 33.3±6.6 kg/m², diabetes duration: 22.8±9.8 yr, haemodialysis duration: 3.6±2.5 yr, coronary disease: 39%, amputation: 14%. After 3 months, the mean CGM glucose value significantly decreased from 176±35 to 160±37 mg/dL (p=0.05). The frequency of hyperglycaemia > 180 mg/dL significantly decreased from 41.3±21.9 to 30.1±22.4 % (p<0.05) without significant increase of hypoglycaemia. There was no modification of MAGE. Insulin requirements increased significantly from 70±51 to 82±77 IU/d (p<0.001) without significant change in body weight. CGM coupled to basal-bolus insulin regimen detemir-aspart improves glycemic control, without increasing hypoglycemia, in type 2 diabetes patients with ESRD at high cardiovascular risk. Supported by: Abbott Laboratories; Novo Nordisk, Inc. 899-P Accuracy of Blood Glucose Meters Employing Dynamic Electrochemistry in Comparison to Other Devices 897-P Hypoglycaemic Episodes and Glycaemic Variability in Ealy Stage Type 2 Diabetes Patients Treated With Monotherapy Insulin Glargine and Metformin, Using Continuous Glucose Monitoring FILIZ DEMIRCIK, PETRA B. MUSHOLT, JOCHEN SIEBER, FRANK FLACKE, THOMAS FORST, ANDREAS PFÜTZNER, Mainz, Germany, Frankfurt, Germany Dynamic electrochemistry (DE) provides mathematical algorithms to correct for a variety of interfering conditions, including hematocrit (HCT). HCT values can vary widely in community patient populations (25-60 %) and may be further influenced by daily activities, e.g. exercise etc. This laboratory study investigated the accuracy of four DE devices (BGStar and iBGStar, Sanofi; Wavesense Jazz, AgaMatrix; Wellion Linus, Wellion) in comparison to three other devices (GlucoDock, Medisana; OneTouch Verio Pro, LifeScan; FreeStyle InsuLinx, Abbott-Medisense) in blood samples with varying HCT concentrations. Venous heparinized whole blood was immediately aliquoted after draw, and manipulated to contain 3 different blood glucose concentrations (60-90 mg/dL, 155 mg/dL, and 310 mg/dL) and 5 different HCT levels (25 %, 35 %, 45 %, 55 %, and 60 %). The samples CARSTA KOEHLER, FRANK SCHAPER, FRANK PISTROSCH, ELENA HENKEL, CAREN HOFFMANN, JULIANE STEINER, RONNY STAUDTE, WOLFGANG LANDGRAF, SUSANNE BILZ, CONSTANZE SCHONER, MARKOLF HANEFELD, Dresden, Germany, Frankfurt, Germany Aim: Among antidiabetic treatments which can lower HbA1c to target insulin is discussed to show greatest hypoglycaemia risk (HR) and glycaemic variability (GV). In routine daily SMBG profiles asymptomatic hypoglycaemic episodes (HE) are often unrecognized and underreported. Continuous glucose monitoring (CGM) was used to assess GV and duration of HE in early type 2 diabetes patients (T2DM) well controlled for A1C under daily life conditions with monotherapy of insulin glargine (GLA) or metformin (M). ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A227 POSTERS 898-P 896-P Clinical Diabetes/ Therapeutics Supported by: sanofi-aventis Supported by: Sanofi CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING were confirmed to have normal pO2 and were measured in parallel 6 times with three devices and three strip lots of each meter and twice with YSI Stat 2300 (reference method). Precision and mean absolute relative deviation (MARD) from reference was calculated. Stability to HCT influence was assumed, when less than 10 % difference occurred between the highest and lowest mean glucose deviations in relation to HCT concentrations (HIF: HCT Interference Factor). Five of the investigated meters showed HCT stability: BGStar (Precision/MARD/HIF: 3.1%/8.3%/4.6%), iBGStar (3.6%/8.8%/6.6%), Wavesense Jazz (4.5%/8.9%/4.1%), Wellion Linus (3.6%/5.0%/8.5 %), and OneTouch Verio Pro (2.7%/6.4%/6.2%). The two other meters were influenced by HCT (FreeStyle InsuLinx: 4.0%/7.5%/17.8%; GlucoDock: 2.8%/7.4%/46.5%). In this study, all meters employing dynamic electrochemistry were shown to have a high accuracy and precision and to reliably correct for potential hematocrit influence on the meter results. 901-P Online Prediction of Abnormal Glycemia Events for Type 1 Diabetes Using Empirical Models and Confidence Limits CHUNHUI ZHAO, WENQING LI, LUPING ZHAO, Hangzhou, China, Hong Kong, China Accurate glucose predictions can be used to provide early warnings of impending abnormal glycemia events. Before taking any action to correct the future violations, a measure of the reliability of the predictions is important and necessary to assess the uncertainty of the predicted values. To address this problem, data-driven empirical models are developed based on continuous glucose measurements for glucose prediction in this paper. An error bound in the form of prediction intervals is calculated to evaluate the prediction errors based on history predictions. Also, it is updated real time based on the most recent predictions. The hypoglycemia and hyperglycemia are thus judged using the combination of predictions and the confidence limits. The above method is investigated by a combination of confidence limits and empirical models based on two groups of ambulatory subjects and two groups of in silico subjects using the FDA-accepted UVa/Padova metabolic simulator for online short-term (0-60min) glucose prediction. The different online alarming types of abnormal glycemia events are illustrated in Figure 1 where four classes of alarming degrees are defined. The results indicate that the use of confidence limits for reality check of glucose prediction provides a useful quantitative measure of the forecasted alarming events and can make the early warnings of impending abnormal glycemia events more reliable. Supported by: Sanofi 900-P POSTERS Clinical Diabetes/ Therapeutics Effectiveness of Continuous Glucose Monitoring for Type 1 Diabetes in Clinical Practice ROBERT B. MCQUEEN, SAM L. ELLIS, DAVID M. MAAHS, HEATHER D. ANDERSON, ANNE M. LIBBY, KAVITA V. NAIR, SATISH GARG, JONATHAN D. CAMPBELL, Aurora, CO There is little evidence on the use of continuous glucose monitoring (CGM) in clinical practice and its effect on change in A1c for adults with type 1 diabetes (T1D). We estimated the adherence and change in A1c with CGM use when compared to self-monitoring of blood glucose (SMBG) alone. We retrospectively identified 66 adult T1D patients at the Barbara Davis Center who first initiated CGM from 2006 to 2011 and 67 controls using SMBG. 12 months prior to index date was the baseline period with a maximum follow-up of 9-months post index. Adherence to CGM was estimated from survey recall and defined as average number of days/ month of CGM use during follow-up. Change in A1c was calculated as the difference between the baseline value and lowest follow-up value. Baseline mean (SD) age in years for CGM vs. SMBG was 39 (12) vs. 32 (12) (p<0.05); duration of diabetes in years was 23 (12) vs. 15 (10) (p<0.05); and A1c was 7.47% vs. 7.67% (p=0.30). N=32 (48%) used CGM < 21 days/ month. Reasons for low adherence included cost, high frequency of alarms, and discomfort. The between group mean difference in change in A1c, adjusted for demographics, was -0.12% (p = 0.36), whereas the subgroup with a baseline A1c ≥ 7.5% and users of CGM ≥ 21 days/month was -0.49% (p = 0.10) [Fig 1]. These clinical practice CGM data suggest a trend towards decreasing A1c for adults with T1D, especially in patients with higher baseline A1c and higher CGM adherence. More data are needed on how to efficiently target CGM therapy in clinical practice. Supported by: NSFC (61273166) 902-P Opportunities to Enhance Recruitment of Youth With Type 1 Diabetes (T1D) into Continuous Glucose Monitoring (CGM) Trials KAITLIN C. GAFFNEY, LISA K. VOLKENING, MARIAH M. BARSTOW, CHRISTINA L. SULLIVAN, MICHELLE L. KATZ, LORI M. LAFFEL, Boston, MA Consistent CGM use improves A1c and reduces severe hypoglycemia but few T1D youth sustain CGM use and derive glycemic benefit. Research provides opportunities to understand CGM use but requires enrollment of unselected samples to enhance generalizability. To understand youth recruitment into CGM trials, we assessed characteristics and enrollment outcomes of eligible T1D youth approached to participate in a 2-year RCT comparing CGM use with and without a behavioral intervention. Eligibility criteria included ages 8-17y, T1D for ≥1y, A1c 6.5-10%, daily insulin ≥0.5 U/ kg, BG monitoring ≥4x/d, and no current CGM use. At routine visits, study staff approached patients deemed eligible by prior EHR review to explain the study and obtain written informed consent/assent. Following the recruitment visit, we categorized patients as agreed, thinking, or declined. Over 13 months, we approached 408 eligible patients (50% male, mean A1c 8.0±0.8%): 105 (26%) agreed, 47 (12%) were thinking, and 256 (63%) declined. Decliners were older (13.5y) and had longer T1D duration (7.2y) & For author disclosure information, see page 829. A228 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSE MONITORING AND SENSING than both agreers (12.7y, 5.9y; each p=.01) and thinkers (12.7y, 6.8y; each p=.01). Pump use was more common in agreers (86%) than thinkers (77%) and decliners (74%) (each p=.05). Sex and A1c did not differ by recruitment status. Most (55%) decliners cited wearing CGM as reason for refusal; youth <15y declined more often due to CGM than youth ≥15y (61% Vs 41%, p=.004). Almost 60% (241) agreed or declined at first encounter; of the remaining 167, 21% later agreed, 51% later declined, and 28% (47) remained thinkers. Thinkers were 3.5 times more likely to agree if they made a decision within 1 month of initial visit (p=.003). Recruitment for CGM trials in T1D youth requires substantial resources and should focus on providing up-to-date information on CGM devices, attention to realities of sensor placement for younger patients, and aggressive follow-up within 1 month of initial invitation to maximize enrollment. Table 1. MARD Comparisons to the CONTOUR PLUS BGMS Overall glucose range Low glucose range High glucose range n=314 n=93 n=113 YSI 27-460 mg/dL YSI < 80 mg/dL YSI > 180 mg/dL Meter system MARD MARD MARD mg/dL mg/dL mg/dL 3.35 3.43 3.26 CONTOUR®PLUS Accu-Chek® Performa 4.95* 4.55 6.0* 5.84* 6.42* 5.58* Accu-Chek® Active OneTouch® SelectSimple 10.32* 12.87* 9.43* 14.69* 17.11* 13.49* FreeStyle Freedom™ * CONTOUR®PLUS MARD YSI-YSI glucose statistically significantly analyzer, YSI Inc. lower in this range. Yellow Springs, OH Supported by: 5R01DK089349 903-P A Comparison of Internet Monitoring With Continuous Glucose Monitoring in Insulin-Requiring Type 2 DM SHAILA V. KOTHIWALE, VEERAPPA A. KOTHIWALE, PURVI V. BHARGAVA, Belgaum, India We compared the effects of Real-Time Continuous Glucose Monitoring (RT-CGM) with an Internet Blood Glucose Monitoring System (IBGMS) on Hemoglobin A1C levels (A1C) in patients with type 2 diabetes mellitus (T2DM) treated with insulin in a randomized study. 57 patients with T2DM were randomly assigned to one of two groups. Group one had the results of their self-monitoring of blood glucose (SMBG) monitored every two weeks using an IBGMS. Group two utilized RTCGM and were monitored every two weeks. Both groups used a secure website to upload data and to receive feedback from an endocrinologist. A1C and laboratory test results were collected at 0 and 6 months. Baseline parameters were not significantly different. After a 6 month follow up, both IBGMS and RT-CGM demonstrated significant within-group improvements in A1C, and are summarized in Table 1. IBGMS and RT-CGM did not show significantly different A1C levels at baseline and 6 months (p > 0.05). Daily insulin dosages for both within and between group data after 6 months were not found to be significantly different from baseline. The use of both IBGMS and RT-CGM significantly improved A1C levels in patients with T2DM treated with insulin in a randomized trial over a six month period. Periodontal diseases are chronic infectious diseases characterized by microbial plaque which provoke a host response and may induce an elevated chronic systemic inflammatory state, contributing to increased insulin resistance and worsening of glycaemic control over time in Type 2 diabetes mellitus patients. Hence the present randomized control trial was conducted in 50 type 2 diabetic patients for a period of three months to investigate the effect of conventional non-invasive periodontal therapy on glycaemic control. The demographic status, Body Mass Index(BMI), lifestyle, oral hygiene practices, HbA1c and Community Periodontal Index(CPI) were recorded at baseline and three months. The patients were randomly assigned into study group and control group. The conventional non-invasive periodontal therapy included scaling and root planning to eliminate microbial plaque in the study group whereas no periodontal therapy was rendered in the control group. In results , paired t test for HbA1c showed a decrease in mean of 0.63±0.47 % with p-value <0.000 in the study group and control group showed an increase in mean of 0.03±0.08% with p=0.2 at three months. Karl Pearson’s correlation coefficient showed a positive correlation for CPI scores and HbA1c with r=0.4006,p=0.0472 from baseline to three months in the study group. The findings of the present study concluded that non-invasive periodontal treatment resulted in the reduction of clinical parameters of periodontal infection with the improvement in glycaemic levels. Knowledge regarding this inter-relationship should be imparted to the patients for their oral hygiene maintenance and metabolic control.Thus, periodontal treatment could be an integral component of a holistic diabetic patient care programme. Table 1. Measurements in A1C over study period. Group Baseline 6 month P = Baseline versus (SD) Follow-up (SD) 6-month follow-up. RT-CGM 8.80 (1.37) 7.49 (0.70) 0.0001 IBGMS 8.79 (1.25) 7.96 (1.30) 0.0170 P = RT-CGM versus IBGMS 0.496 0.081 — 904-P 906-P Comparative Accuracy Evaluation of 5 Blood Glucose Monitorng Systems (BGMSs) Impact of 3-Day Glycemic Profile Use in Poorly Controlled Diabetes Patients on Multiple Daily Insulin Injection (MDI) Therapy: Results from the Automated Bolus Advisor Control and Usability Study (ABACUS) NANCY DUNNE, MARIA T. VIGGIANI, SCOTT PARDO, CYNTHIA ROBINSON, MARY E. WARCHAL-WINDHAM, HOLLY SCHACHNER, JOAN LEE PARKES, Tarrytown, NY KATHARINE BARNARD, DAVID A. CAVAN, RALPH ZIEGLER, IAIN CRANSTON, CHRISTOPHER PARKIN, WALTER KOEHLER, IRIS VESPER, MATTHIAS A. SCHWEITZER, ROBIN S. WAGNER, Southampton, United Kingdom, Bournemouth, United Kingdom, Muenster, Germany, Portsmouth, United Kingdom, Boulder City, NV, Mannheim, Germany, Indianapolis, IN This study compared the accuracy of the CONTOUR®PLUS BGMS, as measured by mean absolute relative difference (MARD), with 4 other BGMSs (Accu-Chek® Performa, Accu-Chek® Active, FreeStyle Freedom™, OneTouch® SelectSimple™). Study staff tested fingerstick samples on 5 BGMSs from 106 subjects ≥ 18 years of age. Extreme glucose values were achieved by blood sample glucose modification. The primary endpoint was comparison of the MARD from the reference value (YSI) in the overall tested glucose range. Other endpoints were MARD in the low glucose (< 80mg/dl) and high glucose (> 180 mg/dl) ranges. CONTOUR®PLUS had a statistically significant lower MARD across the entire tested range (27-460 mg/dL) as measured by YSI, as well as in the high range compared to the other BGMSs (Table 1). In the low glucose range CONTOUR®PLUS had a lower MARD than all other BGMSs which was statistically significant except for AccuChek® Performa (Table 1). In conclusion, the primary endpoint of this study, assessing accuracy in the overall glucose range using MARD, was met; CONTOUR®PLUS had a statistically significant lower MARD, one indicator of accuracy. ADA-Funded Research & We examined the effect of using 3-day, 7-point profiles in patients treated with multiple daily insulin injection (MDI) therapy. Data were drawn from 194 patients enrolled in the ABACUS trial, a large, 26-week, prospective, randomized multi-national study of poorly controlled, MDI-treated people with T1DM and T2DM, with mean (SD) baseline A1C 8.9 (1.2)%. All patients were asked to perform 7-point blood glucose (bG) testing (pre-/postprandial, at bedtime) for 3 consecutive days prior to randomization and prior to 3 additional clinic visits. We compared bG levels obtained on profile days with values from non-profile days. All patients had significantly (p < 0.01) lower mean bG values (Table) and a higher percentage of bG values within target range (70-180 mg/dL / 3.9-10 mmol/L) on more profile days than nonprofile days. Glycemic variability (MAGE) was significantly (p < 0.01) lower with significantly fewer bG values in the hypoglycemic range (< 50 mg/dL / 2.8 mmol/L) and hyperglycemic range (> 300 mg/dL / 16.7 mmol/L) on most For author disclosure information, see page 829. Guided Audio Tour poster A229 POSTERS Effect of Non-Invasive Periodontal Therapy on Glycaemic Control In Type 2 Diabetes Mellitus Patients—A Randomized Control Trial Clinical Diabetes/ Therapeutics 905-P HUGH D. TILDESLEY, ADAM S. WHITE, ANTHONY M. WRIGHT, JEREMY H.S. CHAN, ADEL B. MAZANDERANI, TRICIA TANG, STUART A. ROSS, AUGUSTINE M. LEE, HAMISH G. TILDESLEY, Vancouver, BC, Canada, Victoria, BC, Canada, Nashville, TN, St. George’s, Grenada, Calgary, AB, Canada, Hanover, NH CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS profile days compared with non-profile days. No significant between-group differences were seen in these parameters. Performing periodic 3-day, 7-point bG profiles may be an effective stand-alone intervention to optimize glycemic control. +0.02±0.23g/L if initial ascending arrow (n=36); p=0.008 by ANOVA. Similar differences in capillary glucose levels outcomes were observed for the 222 patients with initially correct interstitial glucose levels between 0,80 and 1.60g/L, p=0.003 by ANOVA. From these observations, 10/32 had a final capillary glucose level lower than 0,80g/L after a descending arrow; 16/162 after a stable arrow and 0/28 after an ascending arrow (p<0.0001 by χ2). For a patient with a “normal” interstitial glucose level, the trending arrow is a good indication of hypoglycemia risk 15 minutes later: never if ascending; 10 % if stable and 30 % if descending. This argues for the interest of CGMS in the prevention of hypoglycemia in real life situations. Table. Comparison of bG values on profile days vs. non-profile days (Intentto-Treat cohort, n=194) Profile Days Baseline Visit 5 Visit 7 Visit 9 Mean bG on Profile Days (mg/dL / mmol/L) 173.7 (39.6) / 9.7 (2.2) 175.1 (39.1) / 9.7 (2.2) 173.9 (38.7) / 9.7 (2.2) 176.0 (37.1) / 9.8 (2.1) Mean bG on Non-Profile Days (mg/dL / mmol/L) 181.9 (42.4) / 10.1 (2.4) 183.5 (40.6) / 10.2 (2.3) 184.8 (40.7) / 10.3 (2.3) 179.8 (39.6) / 10.0 (2.2) P Value < 0.01 < 0.01 < 0.01 NS CLINICAL THERAPEUTICS/NEW TECHNOLOGY— INSULINS Guided Audio Tour: New Insulin Therapeutics (Posters: 909-P to 914-P), see page 19. 907-P & JOHN S. MELISH, Honolulu, HI POSTERS Clinical Diabetes/ Therapeutics Impact of Rest and Activity on Glucose Pharmacodynamic Using CGM Modeling Continuous Glucose Monitoring (CGM) data from mealrelated glucose tolerance testing can provide a pharmacokinetic picture of glucose metabolism glucose uptake and disposal. This study applies a pharmacokinetic model (Melish, 2011 and 2012) to a published CGM study (Manohar, C., et al. Diabetes Care 35:2459, 2012) of interstitial fluid dynamic responses to a standardized meal containing 50 g glucose in age and weight match controls (n=12) and Type 1 diabetics (n=12) while at rest and after low-intensity exercise. One meal/day was followed by rest and two meals were followed by activity on three separate days. Areas under the curve above baseline were compared and in both groups were reduced by measured activity. Boluses of insulin were used in all meals in the diabetic patients. The analysis applied to this data assumes a linear uptake/time of glucose from mixed gastric contents and applies a pharmacokinetic analysis to the glucose concentration curves generated following either mild activity or rest. The grouped results include volume of distribution (Vd, (dL)), fraction/min glucose disappearance (Kd (min-1)), glucose remaining after glucose extraction, dietary glucose (PHGR (mg)), and dietary glucose clearance (Cldiet (dL/min)). Control healthy patients showed increase Kd, Vd, Cl compared with Type 1 patients. Activity was associated with a 22% increase in Vd in controls; no change was seen with Type 1 patients. Pharmacokinetic Data Subjects Control (rest) Control (active) Type 1 (rest) Type 1 (active) Kd 0.00711 0.00702 0.00570 0.00516 Vd 107 130 76 76 PHGR 37502 36826 39256 36584 909-P Demonstration of the Additive Effect of IDeg and IAsp in Adipocytes ANDERS R. SØRENSEN, ERICA NISHIMURA, AAGE VØLUND, Måløv, Denmark, Charlottenlund, Denmark In the IDegAsp formulation, insulin degludec (IDeg) and insulin aspart (IAsp) are combined while still retaining the distinct ultra-long basal acting and fast acting properties of each analogue. Thus, it is important to establish the nature of interactions between IDeg and IAsp as well as with endogenous human insulin (HI), which is being produced in many people with diabetes requiring insulin therapy. Insulin dependent de novo lipogenesis (DNL), determined as insulin stimulated incorporation of tritium labelled glucose into lipid in primary rat adipocytes, was used to assess the functional interaction of IDeg, IAsp and HI. All three produced full dose-response curves in this system, but with different EC50 values. Possible interactions could be antagonism (less lipid is produced), or synergism (more lipid is produced), or the effects could be additive. The dose response surfaces obtained by incubating combinations of two ligands were examined (see example in figure below). This was done by using the four parameter logistic model with a reformatted expression for the ligand concentrations in order to accommodate for potency difference and possible interaction between ligands. The results revealed additivity of all possible combinations. From these studies it can be concluded that HI, IAsp and IDeg act independently through the insulin receptor to produce the lipogenic response in adipocytes, indicating that in IDegAsp, the effects of IDeg and IAsp would also be additive. Cldietl 0.76 0.91 0.43 0.39 908-P Can Trending Arrows in Continuous Monitoring Systems Predict Glycemia Below 0,80g/l ? CONCEPCION GONZALEZ, ELISA MAURY, ISABELLE BARCOS, HENRI GIN, VINCENT RIGALLEAU, Pessac, France Real-time continuous glucose monitoring system (CGMS) has been shown to help improve HbA1c levels in patients with Type 1 diabetes, but its preventive effect on hypoglycemia is not established. We assessed whether the CGMS trending arrows predict subsequent blood glucose level lower than 0,80g/L. Thirty three patients with type 1 diabetes were admitted for education in the use of CGMS Medtronic (n=23) or Navigator (n=10). They were requested to note at time 0: the interstitial glucose level, the direction of the arrow (ascending, stable, descending), the capillary glucose level and the same parameters 15 minutes later to check whether the arrow had predicted the glucose level course (Time 15 min-Time 0). The patients (age: 42±12 years; diabetes duration: 22±12 years; HbA1c: 8.4±1.3%) collected 386 observations (one observation = one arrow plus 2 successive interstitial and capillary glucose levels). Interstitial Glucose levels at time 0 and 15 minutes later correlated with capillary glucose levels (R1=0.81 and R2=0.78 p<0.0001) but were lower (p<0.0001). For the 386 observations, initial descending arrows (n=47) were associated with decreasing capillary glucose levels (-0.15±0.28g/L) whereas capillary glucose course were -0.01±0.23g/L if initial stable arrow (n=303) and Supported by: Novo Nordisk, Inc. & For author disclosure information, see page 829. A230 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS which decreased within two weeks. No differences were seen between arms in adverse events or standard safety parameters. In conclusion, addition of Lira to IDeg once daily + Met improved longterm glycemic control, with weight loss and less hypoglycemia compared with addition of a single daily dose of IAsp with the largest meal. 910-P Telematic Support for Therapeutic Decision and Remote Monitoring by PDA Phone or Voice Server Improves HbA1c in Patients With Type 2 Diabetes Starting a Basal Insulin Therapy: The TELEDIAB2 Study Supported by: Novo Nordisk, Inc. MICHAËL JOUBERT, SYLVIA FRANC, AHMED DAOUDI, CEDRIC FAGOUR, BEATRIX BOUCHERIE, ERIC BENAMO, MICHEL RODIER, LISE DUFAITRE, BRUNO GUERCI, HÉLÈNE AFFRES, LUCY CHAILLOUS, ANNIE CLERGEOT, DURED DARDARI, OLIVIER DUPUY, HELEN MOSNIER-PUDAR, DIDIER CHARITANSKI, LUC MILLOT, CATHERINE GILET, BÉATRICE ROCHE, DOMINIQUE HOULBERT, YVES REZNIK, PIERRE-YVES BENHAMOU, GUILLAUME CHARPENTIER, Caen, France, Evry, France, Fort de France, France, Avignon, France, Nimes, France, Marseille, France, Nancy, France, Kremlin Bicêtre, France, Nantes, France, Besançon, France, Saint Mandé, France, Paris, France, Dreux, France, Saint-Etienne, France, Montargis, France, Clermont Ferrand, France, Alencon, France, Grenoble, France & We previously reported that LY2605541 (LY), a novel basal insulin analog with a prolonged duration of action, has a preferential hepatic effect compared with regular human insulin (HI), when both were infused via peripheral vein (0.5 and 0.1 nmol .kg-1.h-1 for LY and HI, respectively) such that similar glucose infusion rates (GIRs) were required to maintain euglycemia. The current study examined two additional LY delivery rates to determine if the hepato-preferential action is dose-specific. Chronicallycatheterized overnight-fasted conscious dogs underwent clamp studies with portal infusion of glucagon at a basal rate and peripheral venous infusions of somatostatin, LY at 0.25 (n=4) or 0.375 (n=5) nmol.kg-1.h-1 (with 0.25 and 0.375 nmol/kg priming doses, respectively), and glucose to maintain euglycemia. Net hepatic glucose output was reduced from basal in the HI group during the clamp, but there was no net hepatic glucose uptake (NHGU). In contrast, the LY groups switched to NHGU, reaching rates of 0.3±0.3 and 1.8±0.3 mg.kg-1.min-1 within 5 h with LY 0.25 and 0.375 nmol, respectively (P<0.001 for 0.375 nmol vs HI). The ratio of increase in nonliver glucose uptake relative to change in net hepatic glucose balance from basal, calculated at a time when GIRs were ≈3.5 mg.kg-1.min-1 in all groups, was high in HI (1.17±0.38), reflecting a predominant effect on the non-hepatic tissues with less impact on the liver. In contrast, the LY groups’ ratios were 0.39±0.33 (0.25 nmol; P=0.07 vs HI), -0.01±0.13 (0.375 nmol; P<0.01 vs HI), and -0.01±0.08 (in the previously-reported 0.5 nmol group; P<0.01 vs HI), indicating that stimulation of glucose disposal was greater in the liver relative to non-hepatic tissues such as muscle. Similarly, the ratio of change from basal in tracer-determined glucose uptake vs endogenous glucose production was 1.4 (HI) vs 0.72, 0.45, and 0.55 (LY: 0.25, 0.375, and 0.5 nmol). In conclusion, at current and previous doses tested, LY has a clear hepato-preferential action. Supported by: Eli Lilly and Company Supported by: Novo Nordisk, Inc. & Comparison of Addition of Liraglutide to Insulin Degludec Plus Metformin vs. Addition of a Single Dose of Rapid-Acting Insulin Analog to Largest Meal in Type 2 Diabetes BRIAN G. TOPP, JEANNE S. GEISER, DANNY K.W. SOON, TIM HEISE, M. DODSON MICHAEL, SCOTT J. JACOBER, JOHN M. BEALS, VIKRAM P. SINHA, Indianapolis, IN, Neuss, Germany CHANTAL MATHIEU, HELENA W. RODBARD, BERTRAND CARIOU, YEHUDA HANDELSMAN, ATHENA PHILIS-TSIMIKAS, ANN MARIE OCAMPO FRANCISCO, AZHAR RANA, BERNARD ZINMAN, Leuven, Belgium, Rockville, MD, Nantes, France, Tarzana, CA, La Jolla, CA, Søborg, Denmark, Toronto, ON, Canada LY2605541 (LY) is a novel basal insulin analog with a large hydrodynamic size that slows absorption and reduces clearance. The effects of LY’s large size on tissue distribution and mechanism of action remain unclear. We developed a tissue distribution model for LY and Insulin glargine (G) and performed simulations (sims) of recently completed Phase 1 and Phase 2 clinical trials using a mathematical model of metabolic regulation. Due to secretion into the portal vein and first pass clearance at the liver, endogenous insulin is 2-4 fold higher at the liver than muscle. G is predicted to diffuse relatively rapidly from plasma to liver and muscle resulting in a 1:1 ratio. LY is predicted to diffuse rapidly from plasma to the liver via fenestrated vessels but to diffuse slowly across the tight capillary junctions in the muscle resulting in a 4:1 ratio. As a result the primary tissue of action is predicted to be muscle for G and liver for LY. Physiologic based pharmacodynamic simulations incorporating this tissue distribution model were consistent with Phase 1 and Phase 2 data. Following a single injection in healthy volunteers, G (0.8 U/kg) is predicted to display 228 mg/min of glucose infusion during a euglycemic clamp (data: 263 ± 105 mg/min). A single dose of LY (2.2 U/kg) is predicted to display a glucose infusion rate of 144 mg/min (data: 161 ± 49 mg/min). Following 14 days of dosing (1.0 U/ kg) in virtual patients with T2D, LY is predicted to display glucose infusion rates of 210 mg/min during a euglycemic clamp (data: 250 ± 36 mg/min). At the end of 12 weeks of titrate-to-target dosing in virtual patients with type 2 diabetes, predicted changes in A1C (sims: -0.6 ± 0.5, -0.5 ± 0.5% for G, LY) (data: -0.7 ± 0.8, -0.7 ± 0.7% for G, LY), and hypoglycemic events (sims: 0.40, Previous studies have evaluated the efficacy and safety of adding a glucagon-like peptide-1 (GLP-1) analog to basal insulin or vice versa in type 2 diabetes. In this treat-to-target trial, subjects who completed 104 weeks on insulin degludec (IDeg) once daily plus metformin (Met) with HbA1c ≥7% were randomized to either liraglutide (Lira) once daily (N=88) or insulin aspart (IAsp) with largest meal once daily (N=89) for intensification during this subsequent 26-week trial. Mean baseline HbA1c was 7.7% in both arms. IDeg + Lira reduced HbA1c (−0.74 %-points) and was superior to IDeg + IAsp (−0.39 %-points) at Week 26 (estimated treatment difference [ETD] {IDeg + Lira}−{IDeg + IAsp}: −0.32 %-points [−0.53; −0.12]95%CI; p=0.0024). After 26 weeks, mean HbA1c was 7.0% with IDeg + Lira and 7.3% with IDeg + IAsp. Significantly more subjects achieved the target HbA1c <7% without confirmed hypoglycemia (plasma glucose < 56 mg/dL or severe hypoglycemia) and without weight gain with IDeg + Lira (49.4%) vs IDeg + IAsp (7.2%); estimated odds ratio {IDeg + Lira}/ {IDeg + IAsp}: 13.8 [5.2; 36.3]; p<0.0001. With lower HbA1c at end of trial, IDeg + Lira subjects had 87% less confirmed hypoglycemia vs IDeg + IAsp (1.00 and 8.15 events/patient-yr, respectively; p<0.0001), 86% less nocturnal confirmed hypoglycemia (0.17 and 1.11 events/patient-yr, respectively; p=0.0002), and significant weight loss (−2.8 kg) vs IDeg + IAsp (+0.9 kg); ETD {IDeg + Lira}−{IDeg + IAsp}: −3.8 kg [−4.7; −2.8]; p<0.0001. IDeg + Lira subjects initially had more nausea, ADA-Funded Research & 913-P Effects of a Novel Basal Insulin, LY2605541, on Hepatic Glucose Output and Muscle Glucose Uptake: A Physiologic Based Simulation Analysis 911-P For author disclosure information, see page 829. Guided Audio Tour poster A231 POSTERS MARY C. MOORE, MARTA S. SMITH, VIKRAM P. SINHA, JOHN M. BEALS, M. DODSON MICHAEL, SCOTT J. JACOBER, ALAN D. CHERRINGTON, Nashville, TN, Indianapolis, IN Basal insulin is recommended in type 2 diabetes (T2D) when HbA1c > 7-7.5% despite a combination of antidiabetic agents, and its efficacy relies on the active titration of insulin dosage. Telemedecine devices may help optimizing insulin titration and metabolic control in T2D. We evaluated two telematic support systems, an interactive voice server (IVS) and a smartphone with a decision support system (DSS) for T2D on the model of the DIABEO system (DSS for basal insulin titration and educational coaching functions for postprandial glycemic control). This 4-month, openlabel, parallel-group, multicenter study enrolled 190 adult patients with T2D (>1 yr), BMI 29.5±4.9 Kg/m2, age 58.7±9.6 yrs, 64% men, diabetes duration 13.1±7.6 yrs, HbA1c 8.9±1.1% despite oral agents (>6 months). Patients were randomized to i) a standard follow-up (1 consultation at 4 months) (G1), ii) the daily use of an interactive voice server (IVS) with regular short phone consultations (G2), iii) the daily use of the smartphone with regular short phone consultations (G3) for basal insulin Detemir (levemir) dose titration. The primary endpoint was HbA1c value at 4 months. Insulin was initiated at a mean dose of 16.5±4.7 U/d with a 4-month titration of 35±23,(G1), 44±35 (G2) and 50±35 U/d (G3). The 4-month HbA1c reduction was greater in G2 (-1.44%) and G3 (-1.48%) vs. G1 (-0.92%)(p=0.0017, G2 vs G1 and G3 vs G1), but did not differ between the intervention groups. The percentage of patients with a normal 08.00h fasting blood glucose level was higher in G2 and G3 compared to G1 (82.5% and 82.7% vs. 41.8%). No severe hypoglycemia occurred and the frequency of mild hypoglycemia was similar in all groups (0.3±0.7per week). In conclusion, both PDA and IVS improved metabolic control in patients with T2D with a poor glycemic control thanks to a more active titration of insulin dose, with no increase of hypoglycemia frequency. & 912-P Basal Insulin LY2605541 Has Hepato-Preferential Action Across a Range of Delivery Rates Clinical Diabetes/ Therapeutics & CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS 0.23 per wk for G, LY) (data: 0.35, 0.33 per wk for G, LY) were consistent with data. Overall LY’s large size is predicted to result in hepatic preferential tissue distribution. In conclusion, LY treatment compared with GL in patients with T1D demonstrated lower daily mean blood glucose, less nocturnal hypoglycemia but more total hypoglycemia, resulting in a need for lower mealtime insulin doses, which may reflect a more prolonged duration of basal insulin action or greater suppression of hepatic glucose production. Supported by: Eli Lilly and Company & 914-P Lower Within-Subject Variability in Mean Blood Glucose Concentration With Insulin Degludec vs. Insulin Glargine: A Meta-Analysis of Patients With T2D POSTERS Clinical Diabetes/ Therapeutics LUIGI F. MENEGHINI, STEWART B. HARRIS, MARC EVANS, SØREN RASMUSSEN, THUE JOHANSEN, THOMAS R. PIEBER, Miami, FL, London, ON, Canada, Cardiff, United Kingdom, Søborg, Denmark, Graz, Austria Insulin degludec (IDeg) is a new ultra-long-acting basal insulin shown to have low day-to-day variability in a euglycemic clamp study. In this post-hoc meta-analysis of patients with T2D, we compared day-to-day variability in mean self-measured blood glucose (derived from 9-point profiles; 9P-SMPG) between IDeg and insulin glargine (IGlar). This patient-level meta-analysis included all fi ve phase 3a, randomized, open-label, treat-to-target trials (26 or 52 week) in which once-daily IDeg and IGlar have been compared. 9P-SMPG profiles comprised measurements made before and 90 min after the start of breakfast, lunch and main evening meal, before bedtime, at 4 AM, and before start of breakfast the next day. Within-subject variability (CV%) in the overall mean plasma glucose (PG) concentration of the 9P-SMPG profile (area under the profile) was estimated from profiles recorded at weeks 12, 16, and 26 (26-week trials) and weeks 12, 16, 26, 40, and 52 (52-week trials), using a linear mixed model. Estimated within-subject variability in mean 9P-SMPG was significantly lower by 7-10% for IDeg vs. IGlar for patients on basal insulin plus OAD therapy, as well as the subset of previously insulin-naïve patients (Table). In conclusion, IDeg is associated with significantly lower within-subject day-to-day variability in mean blood glucose concentration than IGlar in patients with T2D receiving basal insulin plus OAD therapy. Supported by: Eli Lilly and Company & MARKOLF HANEFELD, RACHELE BERRIA, JAY LIN, RONNIE ARONSON, PATRICE DARMON, MARC EVANS, LUC VAN GAAL, Dresden, Germany, Bridgewater, NJ, Flemington, NJ, Toronto, ON, Canada, Marseille, France, Cardiff, United Kingdom, Edegem, Belgium Population Insulin degludec Insulin glargine IDeg/IGlar (IDeg) (IGlar) (ratio [95% CI]) n CV% n CV% T2D (BB) 692 16.0 236 16.3 0.98 [0.93; 1.04] T2D (BOT) 1370 13.8 789 14.9 0.93 [0.89; 0.96]* T2D (BOT-IN) 1162 13.8 581 15.2 0.90 [0.86; 0.94]* n = number of subjects; CV% = coefficient of variation; T2D = Trials 3579, 3672, 3582, 3586 and 3668 (excluding IDeg flexible dosing group); BB = basalbolus therapy (Trial 3582); BOT = basal insulin plus OAD therapy (Trials 3579, 3672, 3586, 3668 (excluding IDeg flexible dosing group); BOT-IN = previously insulin-naïve patients on BOT therapy (Trials 3579, 3762 and 3586); *p<0.05% The number of elderly people with T2DM is increasing; therefore, efficacy and safety of antidiabetic drugs for vulnerable patients is a key question for individualized treatment. This meta-analysis of 5 Phase 3 lixisenatide (LIXI) randomized trials (GetGoal-M, -P, -S, -M-Asia, -F1) in 501 elderly patients (≥65 yrs) with T2DM compared LIXI to placebo (PBO) on top of OADs. Mean patient characteristics at baseline: age 69.3 yrs, BMI 29.4 kg/m2, diabetes duration 10.6 yrs, HbA1C LIXI 7.99%, PBO 7.92%; 90.0% of patients on MET, 44.7% on SU, 16.6% on TZD. LIXI significantly reduced HbA1C vs PBO (-0.54% [CI: -0.73, -0.36], p<0.00001) and was better than placebo in each of the following composite endpoints: proportion of patients achieving HbA1C <7% (49.3 vs 22.8%, p<0.0001); HbA1C <7% and no weight gain (41.8 vs 18.8%, p<0.0001); HbA1C <7% and no documented symptomatic hypoglycemia (44.4 vs 22.3%, p<0.0001); HbA1C <7%, no documented symptomatic hypoglycemia and no weight gain (37.2 vs 18.8%, p<0.0001). A higher trend for more frequent documented symptomatic hypoglycemia was seen with LIXI vs PBO (OR: 2.09 [0.88, 4.97] p=0.09) mainly driven by SU as background medication, and no incidence of severe hypoglycemia in either group. This post-hoc analysis suggests in elderly patients with T2DM inadequately controlled on OADs, LIXI significantly improved glycemic control, particularly in composite assessments that indicate limited risk of weight gain/hypoglycemia. Supported by: Novo Nordisk, Inc. Guided Audio Tour: Novel Agents for Diabetes Treatment (Posters: 915-P to 921-P), see page 19. & 916-P Efficacy and Safety of Lixisenatide in Elderly T2DM Patients: Subanalysis from the GetGoal Program 915-P Improved Glycemic Control Despite Reductions in Bolus Insulin Doses With Basal Insulin LY2605541 Compared With Basal Insulin Glargine in Patients With Type 1 Diabetes JULIO ROSENSTOCK, RICHARD M. BERGENSTAL, THOMAS BLEVINS, LINDA MORROW, YONGMING QU, SCOTT J. JACOBER, Dallas, TX, Minneapolis, MN, Austin, TX, Chula Vista, CA, Indianapolis, IN LY2605541 (LY) is a novel basal insulin analog with a prolonged duration of action. The objective of this analysis was to further explore the initial observation of a reduction in prandial insulin requirements with LY compared with insulin glargine (GL) for patients with type 1 diabetes (T1D) who completed a Phase 2 randomized, open-label, 2x2 crossover study (n=108). At 8 weeks, patients (n=108) treated with LY required significantly less bolus insulin compared with GL (least-squares mean differences±standard error [SE] for LY − GL): breakfast (-0.9±0.4 IU; p=0.021), lunch (-1.4±0.4 IU; p<0.001), dinner (-2.0±0.4 IU; p<0.001), and total daily bolus dose (-4.3±1.5 IU; p=0.005). Figure 1 presents total daily basal and bolus insulin doses over time by treatment. LY had lower daily mean blood glucose compared with GL (143.1 mg/dL vs. 151.7 mg/dL; p<0.001) (N=108). The rate of total hypoglycemia events was higher for LY vs. GL (9.2 vs. 8.1 events/30d; p=0.074), but the rate of nocturnal hypoglycemia events was lower (0.9 vs. 1.2 events/30d; p=0.007) (n=108 for both). & For author disclosure information, see page 829. A232 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS BIOD-530 vs. U-500R (Study 1) or BIOD-530 vs. IL (Study 2) in miniature diabetic swine using a crossover study design. Fasted swine were given 0.25 U/kg insulin s.c. and fed. Blood was sampled from -30 to 480 min post dose. Plasma insulin was measured by insulin ELISA and glucose by YSI. Tmax (min) and T50%max (min) values were 42±12*, 11±2* for BIOD-530 and 94±17, 27±5 for U-500R in Study 1 and 67±18, 9.7±1.6* for BIOD-530 and 55±8, 21±4 for IL in Study 2, respectively (*p<0.05 vs. U500R or IL). Concentration vs. time profiles for insulin and glucose are shown in Figure below. The rate of absorption and onset of action of BIOD-530 are similar to or faster than IL and more rapid than U-500R. The times for insulin to return to baseline were similar for BIOD-530, IL and U-500R, while the time for glucose to return to baseline was faster for IL vs. BIOD-530 and U-500R. If similar profiles are demonstrated in humans, the rapid absorption of concentrated insulin BIOD-530 may provide superior meal time glucose control relative to U-500R for patients with severe insulin resistance. Supported by: Sanofi & 917-P Clinical Diabetes/ Therapeutics LINDA A. MORROW, MARCUS HOMPESCH, SCOTT J. JACOBER, SIAK LENG CHOI, YONGMING QU, VIKRAM P. SINHA, Chula Vista, CA, Indianapolis, IN, Singapore, Singapore In an open-label, randomized Phase 2 crossover T1D substudy, subjects (n=23) underwent 24-hr Biostator-controlled euglycemic clamps after 8 wks treatment with GL or LY, a novel basal insulin analog with prolonged duration of action. Clinically titrated basal insulin doses (LY: 16 to 64 U, GL: 19 to 60 U) were administered the morning of the clamp. At baseline, mean BMI was 26.8 kg/m2 ± 4.2 [SD]; mean A1C was 7.7% ± 1.0. Endpoint dose (U/ kg) was LY: 0.43 ± 0.13, GL: 0.42 ± 0.10. Daily mean BG (mg/dL) was LY: 138 ± 22, GL: 142 ± 22, p=0.64. Mean total and nocturnal hypoglycemia rates/30 d were LY: 2.7 ± 2.3 and 0.5 ± 0.8, GL: 3.0 ± 2.4 and 0.7 ± 1.1, p=0.11 and 0.43. Mean glucose infusion rate (GIR) normalized to unit of insulin was less for LY than GL and persistent over 24 hrs. The GIR profile for LY is consistent with a peak to trough fluctuation ratio <1.5. Supported by: NIDDK (R43DK096604) & 919-P A Pilot Prospective Case-Control Study to Compare the Efficacy of Laparoscopic Placement of Gastric Contraction Modulator (TANTALUS II®) versus Supplementary Insulin Treatment in Obese Type 2 Diabetic Patients SIMON K. WONG, ALICE P.S. KONG, RISA OZAKI, VANESSA NG, HAROLD E. LEBOVITZ, ENDERS K.W. NG, JULIANA C. CHAN, Hong Kong, China, New York, NY Aims: We evaluated the efficacy of the TANTALUS system, a mealinitiated implantable gastric contractility modulator (GCM) in suboptimally controlled obese Chinese type 2 diabetic (T2D) patients, who failed oral antidiabetic drugs (OAD) compared to supplementary insulin (INS) therapy. Methods: Moderately obese (BMI 25-35kg/m2) T2D patients with HbA1c >7.5% despite receiving maximal doses of 2 or more OADs were included. Patients either received laparoscopic implantation of GCM or insulin therapy. We compared the body weight, waist circumference (WC) and glycemic control (HbA1C) before and 1 year after intervention. Of the 17 patients recruited, 8 received GCM therapy and 9 received insulin injection. GCM and INS groups had similar mean (SD) body weight [80.4(10.9)kg vs 86.3(8.5)kg, p=0.17], BMI [29.4(2.1) kg/m2 vs 30.1(2.3)kg/m2, p=0.42] and HbA1c [9.1(1.0)% vs 9.0(0.7)%, p=0.55] at baseline. The respective mean (SD) HbA1c reduction at 6 and 12 months in the GCM group was 1.6(1.1)% and 0.9(1.6)% (p<0.05 within group). The corresponding changes in the insulin group were 0.3(0.6)% and 0.3(0.7)% (p=0.27 within group) with a trend of between-group differences (p=0.06). In the GCM group, there was modest but significant reduction in bodyweight [-3.2(5.2)kg, p=0.04 with group) and WC [-3.8(4.5)cm, p=0.02] at 12 months. In the insulin group, the respective changes were 2.4(2.5)kg (p=0.03) and 1.5(2.5)kg (p=0.09) with significant between-group differences (p<0.05). At 1 year, 2 patients in the GCM group required rescue insulin therapy but the daily dose was less than that in the insulin group [0.08(0.04) units/kg vs 0.37(0.20) units/kg, p=0.07). Conclusion: In difficult-to-treat T2D patients with obesity who fail OADs, GCM implantation may be a safe and alternative treatment option to insulin therapy. Inter-subject GIR variability (SD) was LY: 0.67, GL: 0.53. Eight LY and 1 GL subjects had minimal GIRs over 24 hrs (defined as <800 mg/kg), indicating optimal glucose control. Mean total glucose infused (GTOT(0-24);g/kg) was LY: 1.22 ± 0.82, GL: 1.90 ± 1.01, p<0.01. Mean GTOT(0-6) (g/kg) was LY: 0.21 ± 0.22, GL: 0.41 ± 0.22, p<0.01. Mean GTOT(18-24) (g/kg) was LY: 0.28 ± 0.18, GL: 0.35 ± 0.23, p=0.20. LY has a flatter profile than GL, with potentially more stable and predictable metabolic control. LY may have a novel mechanism of action as LY GTOT(0-24) is less than GL with similar glycemic control. Supported by: Eli Lilly and Company & 918-P BIOD-530 U-400 Concentrated Recombinant Human Insulin: Pharmacokinetic (PK) and Pharmacodynamic (PD) Profiles in Diabetic Swine Compared to U-500 Regular Human Insulin (U-500R) and Insulin Lispro U-100 (IL) RODERIKE POHL, ROBERT HAUSER, MING LI, BRYAN R. WILSON, MARY GUINNESS, MARILYN JACKSON, ALAN KRASNER, ERROL DE SOUZA, Danbury, CT BIOD-530 is a U-400 formulation of recombinant human insulin, disodium EDTA and citrate under evaluation as an alternative to U-500 regular human insulin (Humulin R®) for the treatment of diabetic patients with severe insulin resistance. In the present studies, we evaluated the PK and PD profiles of ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A233 POSTERS LY2605541 (LY) Exhibits a Flatter Glucodynamic Profile than Insulin Glargine (GL) at Steady State in Subjects With Type 1 Diabetes (T1D) CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS & 922-P 920-P Inhibition of Glucagon Secretion in Rat Pancreatic Islets via Na+ Channels: Mechanism of Anti-Diabetic Effect of Ranolazine Euglycemic Clamp Profile of New Insulin Glargine U300 Formulation in Patients With Type 1 Diabetes (T1DM) Is Different from Glargine U100 MING YANG, RUTH CHU, LUIZ BELARDINELLI, ARVINDER DHALLA, Fremont, CA Ranolazine (RAN), a Na+ channel blocker, has been shown to reduce HbA1c and glucose levels in pre-clinical and clinical studies, but the mechanism of this effect is unclear. Pancreatic α-cells express voltage-gated Na+ channels (NaCh), which play a significant role in glucagon secretion. In this study we determined the effects of RAN and the NaCh blocker tetrodotoxin (TTX) on glucagon secretion from rat pancreatic islets. RAN and TTX significantly reduced glucagon secretion in a concentrationdependent manner, with IC50 of 4.0±2 µM and 2.8±2 nM, respectively. Veratridine, a NaCh activator, concentration-dependently increased glucagon secretion, causing an 8-fold increase at 100 µM (Fig. A). Veratridine (30 µM)-induced glucagon secretion was reduced by RAN and TTX, with IC50 of 2.8 ±1 µM and 3.0±1 nM, respectively (Fig. B and C). Physiologic stimuli epinephrine and arginine also concentration-dependently increased glucagon secretion, with EC50 of 2.9±2 µM and 29±8 mM, respectively. RAN (10 µM) reduced epinephrine (5 µM)- and arginine (20 mM)-induced glucagon secretion by 61±6 and 44±7%, respectively. qPCR analysis showed that Nav1.3 is the predominant isoform expressed in rat islets and α-cells. In summary, RAN inhibits glucagon secretion from rat islets mediated via blockade of Nav1.3. These data suggest that the anti-diabetic effect of RAN in vivo may be due to inhibition of glucagon secretion from pancreatic α-cells. POSTERS Clinical Diabetes/ Therapeutics JOACHIM TILLNER, KARIN BERGMANN, LENORE TEICHERT, RAPHAEL DAHMEN, TIM HEISE, REINHARD H.A. BECKER, Frankfurt, Germany, Neuss, Germany Insulin glargine U100 (Lantus®, GlarU100) has become a standard of care as it provides a 24-h basal insulin supply after single-dose subcutaneous injection. A novel formulation containing 300 U/ml insulin glargine (GlarU300) is being developed to improve the activity profile. The aim of this study was to assess the pharmacodynamic and pharmacokinetic properties of this novel formulation in patients with T1DM. Single SC doses of 0.4, 0.6, and 0.9 U/kg GlarU300 and 0.4 U/kg GlarU100 administered in 24 patients with T1DM were compared in this double-blind, randomized, 4-sequence, crossover study using the automated (Biostator) euglycemic clamp technique over 36 hours. The mean profiles of serum insulin concentrations (INS), blood glucose (BG), and smoothed (LOESS factor = 0.06) body-weight-standardized glucose infusion rate (GIR) were different for the 2 formulations. GlarU100 GIR increased until 11 h and declined thereafter towards 36 h, consistent with the observation of BG values staying at a euglycemic level (mean smoothed BG ≤ 105 mg/dl) beyond 24 h during the clamp and then slowly increasing until the end of the clamp. GIR profiles for 0.4, 0.6, and 0.9 U/kg GlarU300 increased dose dependently with a similar shape over the 36 clamp hours showing an increase from 2 h until around 12 h, followed by a slight decline thereafter with ongoing activity up to 36 h. As compared to GlarU100, GlarU300 showed flatter profiles with less fluctuation in individual GIRs and provided still full BG control at 36 h, end of the clamp. All treatments were well tolerated with no differences in safety-related parameters between treatments. In line with GIR profiles, INS profiles of GlarU300 increased dose dependently and were flatter as with GlarU100. The new insulin glargine U300 formulation provides even longer and flatter activity and PK profiles as compared to GlarU100. Supported by: Sanofi & 923-P 921-P WITHDRAWN Insulin Mediated Glucose Disposal Rate Rather than Improvement of Acute Insulin Response after Intensive Insulin Treatment Is the Major Contributor to Induce Long-Term Remission of Newly Diagnosed Type 2 Diabetes HUI WANG, JIAN KUANG, LI YAN, QIFU LI, SHIPING LIU, ZHENGNAN GAO, WEIGANG ZHAO, FAN ZHANG, YERONG YU, GUOCHUN LUO, GANGYI YANG, LING LI, GUANGWEI LI, Beijing, China, Guangzhou, China, Chongqing, China, Changsha, China, Dalian, China, Shenzhen, China, Chengdu, China, Shenyang, China Underline mechanisms of transient continuous subcutaneous insulin infusion (CSII) induced long-term remission in severe newly diagnosed type 2 diabetes (T2DM) remains unclear. Aim of the present study is to explore major predictors of the remission. Method: 134 newly diagnosed T2DM with HbA1c 11.1% accepted intensive CSII therapy for 2 weeks, then followed for at least 2 years. Glucose clamp tests and IVGTT were performed to assess glucose infusion rate (GIR), acute insulin response (AIR) and acute glucagon response (AGR) before and after CSII. Multivariate analysis was used to identify predictors of the remission, which was defined as the HbA1c less than 7.0% and PG2h less than10 mmol/l without any hypoglycaemic agents. Results: (1) GIR(6.2 vs. 3.8 mg.kg-1.min-1 p<0.0001), AIR(74.2 vs. 51.6mU.L-1.min,p<0.01) and AGR (1201.6 vs. 1111.3 pg.ml-1.min, p=0.002) were significantly improved after the CSII compared with it before. (2) The remission rate was 58.2%, 47.6%, 41.1% and 30.6% at 6, 12, 18 and 24 month after the CSII respectively. Cox model analysis showed that GIR (HR 0.5, p=0.027), AGR (HR 0.99,p=0.04) (but not AIR, HR=0.81,p=0.16) after the 2-week insulin therapy and age at diagnosis of diabetes (HR 1.027, p=0.02) independently associated with remission. However if the HOMA_IR and HOMA_β were used instead of GIR and AIR, it was showed that HOMA_IR before (HR1.58, p=0.018) and after CSII (HR 2.42, p=0.018), HOMA_β (HR 0.365, p<0.0001) after CSII independently associated with the remission. Of note, among which the HOMA_IR had the biggest contribution to the remission (R2=0.15, contributed to 50% of the total changes of remission, whereas the R2 of HOMA_β was 0.05). Conclusion: Insulin sensitivity before and after the intensive insulin therapy rather than the improvement of β-cell function is the major contributor of the long-term remission of newly diagnosed diabetes. & For author disclosure information, see page 829. A234 Guided Audio Tour poster ADA-Funded Research 926-P IDegAsp Shows Distinct Prandial and Basal Glucose-Lowering Effects at Steady State in Subjects With Type 1 Diabetes YUN NING, JENNY JIANG, MICHAEL VAN PETTEN, LUIZ BELARDINELLI, ARVINDER K. DHALLA, Fremont, CA TIM HEISE, LESZEK NOSEK, HANNE HASTRUP, SURESH CHENJI, OLIVER KLEIN, HANNE HAAHR, Neuss, Germany, Søborg, Denmark, Bangalore, India Hyperglucagonemia leads to increased hepatic glucose production and plays a central role in development of hyperglycemia. Ranolazine (RAN) is an anti-anginal drug and also has anti-diabetic properties. It inhibits glucagon release from pancreatic islets, suggesting that it may exert the anti-hyperglycemic effects via a novel glucagon-lowering mechanism. This study determined the effect of RAN on glucagon secretion in streptozotozin (STZ)-induced diabetic rat model. Two weeks after STZ injection (65 mg/kg, i.p.) rats had hyperglycemia (from 83 ± 3 to 367 ± 10 mg/dL) and hyperglucagonemia (from 126 ± 7 to 219 ± 16 pg/mL). During an oral glucose tolerance test (oGTT), administration of RAN (30 mg/kg, p.o.) to diabetic rats 15 min prior to glucose load decreased glucagon levels from 194 ± 21 to 157 ± 19 pg/mL. Glucagon levels were significantly higher in vehicletreated diabetic rats compared with non-diabetic animals (258 ± 24 pg/mL vs. 130 ± 10 pg/mL) at 15 min post-glucose load. RAN treatment significantly lowered glucagon levels in diabetic rats (AUC RAN: 408 ± 229 pg/mL*min vs. vehicle: 1232 ± 206 pg/mL*min, P<0.05). Consistent with the reduction in glucagon levels, glucose AUC was also significantly lower in RAN-treated rats (AUC RAN: 9662 ± 546 mg/dL*min vs. vehicle: 13200 ± 1233 mg/dL*min, P<0.05). To rule out the potential for RAN to interfere with hypoglycemiainduced glucagon release, the effect of RAN was also assessed during insulin-induced hypoglycemia in normal rats. RAN administration during the hypoglycemic clamp did not affect the normal glucagon response; whereas glibenclamide, an insulin secretagogue, significantly prevented hypoglycemia-induced glucagon release. In summary, data show that RAN normalized the exaggerated postprandial glucagon response in diabetic rats without affecting the hypoglycemia-induced glucagon secretion. This glucagon lowering effect of RAN may be the mechanism of the anti-diabetic effects of RAN in non-clinical and clinical studies. Insulin degludec (IDeg)/insulin aspart (IAsp) is a soluble combination of 70% IDeg and 30% IAsp providing both ultra-long-acting basal coverage and a prandial insulin bolus in a single injection. The pharmacodynamic properties of IDegAsp at steady state in subjects with T1DM were examined in a single-centre, multiple-dose trial. To achieve steady state of the basal component, subjects received once-daily IDeg (0.42 U/kg) for 5 consecutive days (with separate bolus IAsp as needed for safety and glycaemic control). On Day 6 a 30-h euglycaemic clamp procedure was performed (Biostator, glucose target: 100 mg/dL) after a single dose of IDegAsp (0.6 U/kg comprising 0.42 U IDeg and 0.18 U IAsp). Twenty-two subjects (mean age, 40 y; HbA1c, 7.9%; duration of diabetes, 23 y) were exposed to treatment. The mean glucose infusion rate (GIR) profile showed a rapid onset of action and a distinct peak followed by a flat basal action (Fig 1). Median time to maximum glucose-lowering effect of IDegAsp (tGIRmax) was 2.5 h. Duration of action (from dosing until blood glucose was consistently >150 mg/dL) extended beyond 30 h in all subjects. In conclusion, at steady state the glucose-lowering effect of IDegAsp in subjects with T1DM showed a distinct peak action due to IAsp, and a separate and stable basal action from IDeg sustained for >30 hours. This profile may constitute a clinical advance, allowing for the combination of specific meal coverage with full 24-h basal coverage. 925-P Chronic Dosing of a Liver-Specific Glucokinase Activator (LGKA) Decreases Fasting Hyperglycemia (HG) by Decreasing Hepatic Glucose Production (HGP) in Zucker Diabetic Fatty Rats (ZDF) ERIN C. HEALEY, TRACY P. O’BRIEN, GREGORY A. MCCOY, TIFFANY D. FARMER, DEREK M. ERION, JEFFREY A. PFEFFERKORN, MASAKAZU SHIOTA, Nashville, TN, Cambridge, MA Glucokinase plays an important role in regulating hepatic glucose metabolism and liver selective activation of the enzyme is emerging as a therapeutic strategy for T2DM. (S)-6-(3-cyclohexyl-2-(4-(trifluoromethyl)1H-imidazol-1-yl) propanamido) nicotinic acid is a potent (EC50: 79±3 nM for human and 170±7 nM for rat) LGKA that exhibits enhanced distribution to the liver (liver:plasma = 9) and impaired distribution into the pancreas (pancreas:plasma = 0.05). ZDF were dosed LGKA at 100 mg/kg (ZDF-G) or vehicle (ZDF-V) once daily for 6 wks. Blood was collected prior to treatment (W0) and at 2 (W2), 4 (W4) and 6 wks (W6). A HG clamp study was performed at W6 on 6-h fasted conscious animals. The ZDF-G group had a 30% decrease in HG within 2 wks of treatment compared to ZDF-V without alteration of plasma insulin (INS) and glucagon (GGN) concentrations. At W6, under the basal condition, ZDF-G decreased HG (19.0±1.8 vs 21.6±0.7) and also was associated with slightly lower glucose utilization rates (Rd, µmol/kg/min: 56±5 vs 65±5) and HGP (µmol/kg/ min: 60±5 vs 69±4), compared to ZDF-V. During HG clamp studies in which plasma glucose (PG) were similar between groups, the ZDF-G had higher Rd (73±5 vs 56±5) and further decreased HGP (50±8 vs 66±8) that was associated with reduced gluconeogenesis (29±5 vs 48±6) and higher PG incorporation into hepatic glycogen (µmol/g: 20.6±3.2 vs 8.0±2.1) compared to that in ZDF-V. In ZDF-G, plasma triglyceride (TG) increased by 40% within W2 (mg/ml: 2.8±0.2 vs 2.6±0.3 at W0 to 4.1±0.5 vs 2.3±0.2 at W2, 4.3±0.2 vs 2.8±0.4 at W4 and 4.3±0.3 vs 2.6±0.4 at W6) without changes in plasma FFAs. The TG content in liver at W6 (mg/g: 13.1±1.1 vs 14.1±2.6) and muscle (9.0±2.0 vs 12.6±2.4) were similar between groups. Chronic dosing of LGKA improved hyperglycemia by decreasing gluconeogenesis and increasing glycogen synthesis. LGKA treatment was associated with increased plasma but not liver or muscle TG. ADA-Funded Research & Supported by: Novo Nordisk, Inc. 927-P Impact of Short-Acting Insulin Analogs on Biomarkers of Oxidative Stress and Chronic Systemic Inflammation in Patients With Type 2 Diabetes: Results from a Pilot Study ANKE PFÜTZNER, THOMAS FORST, MICHAEL MITRI, ANDREAS LÖFFLER, JULIA HEISE, CLAUDIA FORKEL, ANDREAS PFÜTZNER, Mainz, Germany This pilot study investigated hypothesis generating data regarding the impact of short-acting insulin analogs (insulin aspart, IA, and insulin glulisine, IG) in comparison to regular human insulin (RHI) on biomarkers of inflammation and oxidative stress during an oral glucose challenge experiment (OGTT) in patients with type 2 diabetes. Twelve patients were enrolled into this trial (11 male, 1 female, age: 64±9 yrs., HbA1c: 7.1±0.6 %, BMI: 31.5±4.9 kg/m²). They were randomized (4/treatment arm) to intensive insulin therapy with insulin glargine and either IA, IG, or RHI as short-acting insulin component. An OGGT was performed at baseline and after six months with assessment of nitrotyrosine, hsCRP and mRNA macrophage activation markers (IL6, TNFalpha, eNOS, MAPK1) after 0, 1 and 2h. Reduction of HbA1c to near normal levels, and similar and unchanged glucose excursions during the OGTT were seen in all three groups. A reduction in nitrotyrosine levels was seen with both analogs (-33 %), while RHI led to an increase (+31 %, p<0.05 vs. the combined analogs). IL-6 expression decreased slightly with IG (-5 %), while it increased with IA (+142%) and RHI (+64 %). TNFalpha expression decreased in all three groups (IG: -35%, IA: -71%, RHI: -30%). Vasoprotective eNOS expression increased with IG (+25 %), decreased with IA (-28 %), and was stable with RHI (+1 %). Growth hormone effect indicating MAPK1-expression was reduced in all three groups (-16%, -23%, -33%). In parallel, hsCRP was significantly reduced by -78 % with IG only (IA: -2%, RHI: - 11%, p<0.05 vs. IG in both groups). In this pilot study, an improvement of the inflammatory vascular situation could generally be observed with For author disclosure information, see page 829. Guided Audio Tour poster A235 POSTERS 924-P Ranolazine Suppresses Exaggerated Postprandial Glucagon Response in STZ-Induced Diabetic Rats Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS all insulin regimens, while a particular improvement in cardiovascular risk as indicated by a consistent trend of improvement of all reported markers was seen with IG only. These results need be confirmed in appropriately designed larger clinical trials. proteins suggest that the improvement of endothelial function contributes to CBF augmentation. Supported by: Novo Nordisk, Inc. Supported by: Sanofi 930-P Optimizing Insulin PK to Improve Glucose Control With a Bionic Pancreas 928-P Increased 1,5-Anhydroglucitol Predicts Long-Term Glycemic Remission in Patients With Newly Diagnosed Type 2 Diabetes Treated With Short-Term Intensive Insulin Therapy MANASI SINHA, FIRAS H. EL-KHATIB, EDWARD R. DAMIANO, STEVEN J. RUSSELL, Boston, MA The objective is to correlate insulin lispro time-to-peak plasma levels after injection (tmax) with anti-insulin antibody (ab) titers in adults and children with T1DM. Plasma insulin lispro levels in adult (21-72 yo) and pediatric subjects (1221 yo) with T1DM participating in bionic pancreas trials were measured every 30 min. Models for pharmacokinetics of insulin lispro were derived by fitting a summation of the exponential accumulation and decay functions for each bolus to the measured insulin lispro levels using a least-squares minimization protocol. The tmax values were derived from the fitted model for each subject. Antibody titers against regular human insulin were measured by a competitive binding assay for the 1st adult study and by a non-competitive binding ELISA assay in the 2nd adult study and pediatric study. Values were normalized for both assays such that lowest value measured was normalized to 0 and highest to 1 (Fig). Insulin lispro tmax varied from 31-191 min (mean 83±43) in adults and 57-138 min (mean 79±24) in pediatric subjects. High tmax values correlated with high insulin antibody titers measured by both assays. There was a high degree of variability in the speed of insulin lispro absorption in both groups. Slow absorption may be mechanistically related to autoimmunity against insulin. A potential mechanism is binding of insulin lispro by antibodies in the interstitial fluid after injection, thereby slowing diffusion out of the injection depot and absorption into the blood. POSTERS Clinical Diabetes/ Therapeutics LIEHUA LIU, XUESI WAN, ZHIMIN HUANG, DONGHONG FANG, HAI LI, AILING CHEN CHEN, WANPING DENG, YANBING LI, Guangzhou, China Our previous study showed that in patients with newly diagnosed type 2 diabetes mellitus (T2DM) treated with short-term continuous subcutaneous insulin infusion (CSII), serous 1,5-anhydroglucitol (1,5 AG) 1 month after the therapy could predict a 3-month glycemic remission. However, the longterm predictive effect remains to be validated. We enrolled 61 patients with newly diagnosed T2DM. After anthropometric measurements, fasting plasma glucose (FPG), fasting insulin, 2h post-prandial glucose(PPG), glycated hemoglobin A1c (A1C) and 1,5 AG were determined. CSII was used in all patients to achieve and maintain euglycemia (fasting capillary blood glucose 4.4-6.1mmol/L, postprandial capillary blood glucose 4.4-7.8mmol/L) for 2 weeks. After the cessation of CSII, baseline assessments were repeated. The patients were then followed up monthly for 3 months and every 3 months afterwards with FPG and 2hPG monitored. 1,5AG were measured at 1-month follow-up. Maintenance of optimum glycemic control (FPG<7.0mmol/L,2hPG<10mmol/L) for 1 year was defined as remission. After the therapy, FPG (11.8±3.0 vs. 6.6±1.7mmol/L,P<0.001), PPG (17.6±5.7 vs. 8.6±3.0mmol/L, P<0.001) and A1C(11.1±1.7 % vs. 9.5±1.3%, P<0.001) decreased. 1,5AG elevated significantly over time (baseline, 2.8±3.0mg/L, after CSII, 7.9±3.8 mg/L, 1 month, 9.9±4.4 mg/L, P<0.001). Logistic regression analysis showed that 1,5 AG at 1-month follow-up predicted 1-year remission independent of age, BMI, HOMA IR, FPG and PPG after CSII (OR 1.37,95%CI [1.12,1.69], P=0.03). The area under the curve of 1,5AG in the receiver operating characteristic curve analysis was 0.84 (95% CI 0.74–0.94, P < 0.001), and the optimal cutoff point was 9.28 mg/L (sensitivity 81.8%,specificity 75.0%). We concluded that 1,5 AG 1 month after CSII therapy can be used as a predictor of long-term remission after short-term intensive insulin therapy in newly diagnosed T2DM. 929-P Increase in Microcirculatory Cutaneous Perfusion With Improved Endothelial Function after Insulin Treatment in Poorly Controlled Type 2 Diabetic Patients—The INSUVASC Study MARINOS FYSEKIDIS, KARIM TAKBOU, YAYA JABER, EMMANUEL COSSON, PAUL VALENSI, Bondy, France The aim of this study was to examine the changes in cutaneous blood flow (CBF) at fasting and post-prandially, in poorly controlled type 2 diabetic (T2D) patients after insulin treatment. The INSUVASC study was a unicenter pilot, open labeled study. We included 42 poorly controlled T2D patients (mean age 53.8±9.6 years, HbA1c 9.0±1.4%). CBF was measured with a laser doppler device for a 6-minute period at rest and for 3 minutes after acetylcholine iontophoretic administration, at fasting (H0) and one to two hours (H1, H2) after a standardised breakfast. Vascular sympathetic activity at rest was evaluated by the 0.1Hz CBF spectral peak. Patients randomly received one of the three insulin regimens (Aspart (A), Detemir (D) or Aspart combined with Detemir(AD)) during four weeks. A total of 34 patients had good quality CBF measurements (A: n=8, D: n= 14 or AD: n=12) and completed the study. Age, BMI, HbAlc did not differ at baseline across the groups. CBF changes (expressed as % of the first baseline value) increased after breakfast, before (H2: 57±14%, p< 0.001) and after (H2: 83±25%, p=0.005) insulin treatment. After insulin treatment, at fasting, CBF increased (H0:30±10%, p=0.009), without significant correlation between CBF and glycaemia, CBF response to acetylcholine improved (area under the curve, maximum perfusion and vasodilation duration, p<0.05), and serum VCAM (p=0.021) and E-selectin (p=0.002) decreased. Spectral analysis showed no increased sympathetic activation and no significant differences across insulin regimens. CBF increased at fasting after insulin treatment independently of blood glucose changes. The lack of sympathetic activity changes, the increase in acetylcholine stimulated response and the decrease in adhesion glyco- Supported by: Leona M. and Harry B. Helmsley Charitable Trust; NIH/NIDDK 931-P A Clinical Utility Index for Selecting an Optimal Insulin Dosing Algorithm for LY2605541 DAVID H. MANNER, JUNXIANG LUO, YONGMING QU, SCOTT BERRY, BRENDA GAYDOS, SCOTT J. JACOBER, Indianapolis, IN, Austin, TX Selecting an optimal insulin dosing algorithm is difficult due to multifaceted dependent outcomes. In a utility-based approach to this issue, a Clinical Utility Index (CUI) was created by a priori weighing the relative benefit/risk (via expert input) of 4 components (Figure 1). Individual component value functions were created and the values multiplied to compute a single utility value for each dosing algorithm for a novel long-acting basal insulin analog, LY2605541 (LY). The CUI was then slightly modified and retrospectively applied to data obtained from a Phase 2 type 2 diabetes trial in which patients were randomized to 1 of 2 LY dosing algorithms (LY1 or LY2) or insulin glargine (GL) (Table 1). The overall CUIs for LY1 and LY2 relative to GL showed that LY2 seemed to perform better than LY1. The overall CUI results are generally consistent with the interpretation of the components. Based on this CUI, LY2 is more likely to be a better dosing algorithm than LY1. & For author disclosure information, see page 829. A236 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS These data suggest that non-critically ill patients with hyperglycemia receiving steroids require a higher percentage of their insulin TDD as nutritional insulin to achieve normoglycemia. Table 1. Clinical Utility Index (CUI) Components and Computed Values for 2 LY Dosing Algorithms Relative to GL Dosing Components LY1a LY2a GLa LY1/GL CUI LY2/GL CUI (n=98) (n=97) (n=93) 1. LSmean Change from Baseline in HbA1c (%) (SE) -0.64 (0.07) -0.83 (0.07) -0.64 (0.08) 1.00 1.23 2. Percent of Patients Achieving HbA1c <7% 55.3% 51.7% 48.4% 1.17 1.08 3. LSmean Total Hypoglycemia Rate (Events/ 1.74 (0.33) 1.59 (0.31) 1.68 (0.36) 0.91 1.13 Patient/30 Days) (SE) 4. LSmean Time When 50% of Patients Reached 5.00 4.00 3.14 0.77 0.89 Steady-State Dose (Weeks) CUI LY1/GL: 0.82, LY2/GL: 1.35 (4 Components) CUI probabilitiesb for LY2>LY1, LY2=LY1, and LY2<LY1 are 48%, 24%, and 28%, respectively. CUI LY1/GL: 1.07, LY2/GL: 1.51 (Components 1-3) CUI probabilitiesb for LY2>LY1, LY2=LY1, and LY2<LY1 are 46%, 24%, and 30%, respectively. aBergenstal et al, Diabetes Care 2012;35(11):2140-7. bBased on 3000 bootstrap samples. Supported by: NIH 933-P The Distinct Prandial and Basal Pharmacodynamics of IDegAsp Observed in Younger Adults are Preserved in Elderly Subjects With Type 1 Diabetes Supported by: Eli Lilly and Company 932-P Insulin Requirements in Non-Critically Ill Hospitalized Patients With Steroid-Induced Hyperglycemia ELIAS SPANAKIS, NINA SHAH, KEYA MALHOTRA, TERRI KEMMERER, HSINCHIEH YEH, SHERITA H. GOLDEN, Baltimore, MD Supported by: Novo Nordisk, Inc. Steroid administration to hospitalized patients frequently results in hyperglycemia. Guidelines for glucose management in this setting are lacking. We conducted a retrospective chart review of all non-critically ill inpatients with diabetes or hyperglycemia receiving steroids from 1/2009 to 12/2010. Forty-one patients were identified from 514 consults. Normoglycemia was defined as a patient-day weighted mean blood glucose [PDWMBG] of 70-180 mg/dL and hyperglycemia as a PDWMBG>180 mg/ dL. Multivariable linear regression was used to compare the insulin total daily dose (TDD), percentage of TDD given as basal and nutritional insulin, and TDD/kilogram (kg) body weight between normoglycemic versus hyperglycemic patients, adjusting for age, sex, race, and diabetes type and duration. Twenty-one patients achieved normoglycemia during admission (PDWMBG=153±16 mg/dL) and 20 patients remained hyperglycemic (PDWMBG=233±33 mg/dL)(p<0.001). There was no difference in age, sex distribution, hypoglycemia, or type or duration of diabetes between the groups; however, those with hyperglycemia were more likely to be Black (65%) compared to those who achieved normoglycemia (28.5%; p=0.04). Following multivariable adjustment, compared to hyperglycemic patients, normoglycemic patients had significantly lower percentage of insulin TDD (basal + nutritional) given as basal (40%±4% vs 60%±4%; p<0.001) and a significantly higher percentage given as nutritional (45%±8% vs 19%±8%, p=0.009). Although not statistically significant, TDD/kilogram (kg) bodyweight was higher in normoglycemic versus hyperglycemic patients (0.92±0.14 units/ kg versus 0.71±0.14 units/kg; p=0.21). Results were similar in subsidiary analyses excluding individuals with hypoglycemia during admission. ADA-Funded Research & 934-P Physical Health Status and Nocturnal Hypoglycemia With Insulin Degludec vs. Insulin Glargine: A 2-Year Trial in Insulin-Naïve Patients With Type 2 Diabetes HELENA W. RODBARD, BERTRAND CARIOU, BERNARD ZINMAN, YEHUDA HANDELSMAN, MICHAEL L. WOLDEN, AZHAR RANA, CHANTAL MATHIEU, Rockville, MD, Nantes, France, Toronto, ON, Canada, Tarzana, CA, Søborg, Denmark, Leuven, Belgium Insulin degludec (IDeg) is a new basal insulin with an ultra-long and stable glucose-lowering effect. We compared once-daily IDeg and insulin glargine (IGlar) (randomized 3:1) both in combination with metformin ± DPP4 inhibitors in an open-label, treat-to-target trial in patients with type 2 diabetes (T2D). Health status was assessed at baseline and 105 weeks using the Short Form 36 (SF-36 v2) questionnaire. SF-36 scores were analyzed (ITT population) using ANOVA, with adjustments for relevant covariates. After an initial 1-yr study period IDeg improved overall SF-36 physical health and functioning compared to IGlar. This was followed by a 1-yr extension in which subjects maintained their randomized treatment assignment. Of 1,030 subjects, 725 entered the extension and 659 (IDeg: 505; IGlar: 154) completed 2 yrs of treatment. At 105 weeks, IDeg was similar to IGlar with respect to change in A1C from baseline, but was associated with a significantly greater reduction in FPG (-6.8 mg/dL [-12.6; -1.1] 95% CI, p=0.02). Rates of overall confirmed hypoglycemia (PG <56 mg/dL, or severe) were similar between groups, whereas the rate of nocturnal confirmed For author disclosure information, see page 829. Guided Audio Tour poster A237 POSTERS Insulin degludec (IDeg)/insulin aspart (IAsp) is a soluble combination of 70% IDeg and 30% IAsp, providing both ultra-long basal coverage and a prandial insulin bolus in 1 injection. The pharmacodynamic profiles of IDegAsp and biphasic IAsp 30 (BIAsp 30) were investigated in a singlecentre, double-blind, two-period, single-dose, crossover trial in elderly (≥65 y) and younger (18-35 y) adults with type 1 diabetes randomly assigned 2 0.5 U/kg single-dose administrations of IDegAsp or BIAsp 30. Following each dose a 26-h euglycaemic clamp was performed (glucose target: 100 mg/dL). Fifteen elderly (mean age 68 y, duration of diabetes 34 y, BMI 25.2 kg/m2, HbA1c 7.5%) and 13 younger (mean age 25 y, duration of diabetes 13 y, BMI 23.2 kg/m2, HbA1c 7.4%) adults were randomised and completed the trial. Mean 24-hour area under the glucose infusion rate curve (AUCGIR,0-24h,SD) for IDegAsp was similar for elderly (geometric mean 1794 mg/kg [CV 62%]) and younger (1786 mg/kg [CV 28%]) subjects (mean ratio elderly/younger adults 1.01, 95% CI: 0.69; 1.47). Prandial coverage (AUCGIR,0-6h,SD) was comparable between elderly (geometric mean 909 mg/kg [CV 45%]) and younger adults (1001 mg/kg [CV 25%]). In conclusion, the glucose-lowering effect of IDegAsp is preserved in elderly subjects with type 1 diabetes (Fig 1), consisting of a distinct peak action due to the prandial IAsp, in addition to the stable, sustained ultra-long action of IDeg. Clinical Diabetes/ Therapeutics THOMAS PIEBER, STEFAN KORSATKO, SIGRID DELLER, HARALD KOJZAR, CARSTEN ROEPSTORFF, ANNE LOUISE SVENDSEN, HANNE HAAHR, Graz, Austria, Søborg, Denmark CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS hypoglycemia (12:01AM to 5:59AM) was significantly lower (by 43%, p<0.01) for IDeg. Consistent with previously reported findings for the initial 1-yr trial period, the overall physical component score was significantly better with IDeg vs IGlar after 2-yrs (treatment contrast (TC): 1.1 [0.1; 2.1], p<0.05). This was largely due to significantly better physical functioning (TC: 1.1 [0.0; 2.3], p<0.05) and bodily pain sub-domain scores (TC: 1.5 [0.2; 2.9], p<0.05). Consistent with 1-yr data, other SF-36 domain scores showed no significant differences between groups. In conclusion, a significantly better physical health status and lower risk of nocturnal hypoglycemia is maintained for IDeg vs IGlar following the second year of treatment of insulin-naïve patients with T2D. The Clinical Characteristics of the groups Categories SW group (n=20) BMI 20.3±2.5 TDD (U) 25.7±6.3 eGFR 56.8±41.6 %basal (%) 30.6±9.3 Diabetic retinopathy(Non/background/ 4/4/1/11 prePDR/PDR or more advanced) NW group Significance (n=28) 22.3±2.8 p<0.01 35.0±10.4 p<0.001 87.6±23.5 p<0.01 24.5±5.6 p<0.01 17/6/0/5 p<0.005 Supported by: Japan Diabetes Foundation Supported by: Novo Nordisk, Inc. 937-P 935-P Characterization of Insulin Glargine Protraction Mechanism and Metabolism In Vivo NORBERT TENNAGELS, HANS-PAUL JURETSCHKE, RONALD SCHMIDT, TANIA ROMACHO-ROMERO, JUERGEN ECKEL, ULRICH WERNER, Frankfurt, Germany, Düsseldorf, Germany JITEN VORA, HELENA W. RODBARD, HENRIETTE MERSEBACH, RAHUL KAPUR, STEWART B. HARRIS, Liverpool, United Kingdom, Rockville, MD, Søborg, Denmark, London, ON, Canada POSTERS Clinical Diabetes/ Therapeutics Lower Risk of Hypoglycemia With Insulin Degludec vs. Insulin Glargine in Patients Diagnosed With Type 2 Diabetes for >10 Years: Meta-Analysis of Five Randomized Trials Insulin glargine (GLA) is a long-acting insulin analog with proven safe and effective 24-hour glycemic control. GLA differs from human insulin by substitution of Asp by Gly in position 21 of the A chain and by carboxyterminal extension of the B-chain by 2 Arg residues causing a pI-shift from pH 5.4 to 6.7. Following s.c. administration as a clear solution of pH 4, GLA forms a microprecipitate at the injection site, from which it is slowly released but then rapidly metabolized to its major metabolite M1 that is responsible for the pharmacodynamic effects in humans and animals. In this study we characterize the mechanism of protraction and metabolism of GLA in more detail. GLA injection into the subcutis of a rat neck and subsequent inspection of the injection site indicated the formation of a local amorphous depot. Protease activity as measured by near infrared spectroscopy in these animals, and previous observation of efficient degradation to M1 at the injection site in humans, suggests the contribution of proteases to the dissolution process and metabolism in the s.c. depot. In addition, i.v. studies in different species demonstrating a rapid formation of M1 indicated that further metabolism can take place in the circulation. Proteomic analysis of supernatants of human subcutaneous adipose tissue revealed several proteases that are secreted into the medium and might be involved in the rapid and efficient metabolism of GLA in the local tissue. Among these, metallocarboxypeptidase E (CPE) could be validated as effectively and rapidly degrading GLA to M1 in vitro. CPE is an adipokine released by human adipocytes whose expression is upregulated during adipogenesis. In summary our data indicate that local depot formation by GLA is followed by degradation already at the s.c. injection site and continues in the circulation. Consequently, injection of GLA results in the rapid formation of M1 as the principal circulating insulin that has a metabolic and mitogenic activity comparable to that of human insulin. Insulin degludec (IDeg), an ultra-long-acting basal insulin with a stable and consistent glucose-lowering effect, is shown to be associated with significantly lower rates of overall (17%) and nocturnal hypoglycemia (32%) compared to insulin glargine (IGlar) in a pre-planned meta-analysis of patients with type 2 diabetes (T2D). T2D is a progressive disease where the incidence and rates of recurrent hypoglycemia increase with diabetes duration and consequent intensification of insulin therapy. In this post-hoc meta-analysis, hypoglycemia rates were compared between IDeg and IGlar in a subset of T2D patients with a diabetes duration >10 years. The meta-analysis included all five phase 3a, open-label, randomized, treat-to-target (FPG < 90 mg/dL) clinical trials of 26 or 52 weeks’ duration in which once-daily IDeg and IGlar were compared in T2D patients. Analysis of A1C and FPG was based on an ANCOVA model; analysis of hypoglycemic episodes was based on a negative binomial regression model. In all, 1,651 (IDeg: n=1,143; IGlar: n=508) of 3,372 randomized patients had a duration of T2D >10 years and were included in the meta-analysis. Treatment groups were similar with respect to mean A1C at end-of-trial (treatment difference (TD) IDeg−IGlar: 0.08% [-0.02; 0.17]95% CI; NS). IDeg was associated with a significantly greater reduction from baseline in mean FPG (TD: -10.1 mg/ dL [-14.2; 5.8]; p<0.01). The rate of overall confirmed hypoglycemia (plasma glucose < 56 mg/dL and severe episodes requiring assistance) was 21% lower with IDeg vs IGlar (p<0.01); the rate of nocturnal confirmed hypoglycemia (onset from 12.01AM to 5.59AM) was 29% lower with IDeg vs IGlar (p<0.01). In conclusion, this meta-analysis demonstrates that treatment with IDeg provides important clinical advantages in patients with a long-term T2D. This includes significantly lower rates of both overall and nocturnal hypoglycemia than IGlar at similar A1C levels. Supported by: Sanofi Supported by: Novo Nordisk, Inc. 936-P 938-P The Characteristics of Type 1 Diabetic Patients Who Use Insulin Pump With Square-Wave Bolus Insulin Insulin Degludec Multi-Hexamers Retain the Native Protein Fold of Human Insulin AKIO KURODA, TAKESHI KONDO, KEN-ICHI AIHARA, ITSURO ENDO, TETSUYUKI YASUDA, HIDEAKI KANETO, TAKA-AKI MATSUOKA, TOSHIO MATSUMOTO, MUNEHIDE MATSUHISA, Tokushima, Japan, Suita, Japan DORTE B. STEENSGAARD, MATHIAS NORRMAN, JADE BEARHAM, CHRISTIAN B. ANDERSEN, ANDERS V. FRIDERICHSEN, HOLGER M. STRAUSS, GERD SCHLUCKEBIER, IB JONASSEN, Måløv, Denmark Backgrounds: Square-wave bolus insulin of insulin pump might be a beneficial option for patients with gastroparesis or with quick insulin absorption. It has not been investigated the insulin pump setting in the patients who use normal bolus insulin and square-wave bolus insulin. Methods: Fifty-two type 1 diabetic patients (T1DM), who were using insulin pump, were investigated during 2-3 weeks of hospitalization. Each meal omission was done to confirm basal insulin rate. The amount and duration time of each bolus insulin was adjusted to set blood glucose to 100 and 150mg/dL before and 2 hours after meal, respectively. Insulin pump setting and clinical characteristics were investigated. Results: The square-wave (30-120 minutes) bolus insulin was necessary to achieve target glycaemia in 25 patients. Twenty patients used squarewave bolus insulin (SW group) and 28 patients used normal bolus insulin (NW group) in all meals. The clinical characteristics were shown in the table. BMI (t=-2.105, p<0.05) and %basal (t=2.027, p<0.05) were the independent determinants for SW group according to the multiple regression analysis. Conclusions: Low BMI and high %basal are the independent determinants for the use of square-wave bolus insulin. Insulin degludec (IDeg), an ultra-long-acting basal insulin, forms a soluble and stable depot of multi-hexamers after subcutaneous injection with a subsequent slow release of monomers into circulation, resulting in a duration of action of greater than 42 hours. In a formulation that includes both zinc and phenol, IDeg is organized as finite di-hexamers, but upon injection, dissociation of phenol allows the soluble depot of IDeg multi-hexamers to form. The depot is composed of long strands of multi-hexamers with a width of 6.3 ± 0.9 nm as observed by transmission electron microscopy. We have conducted various types of spectroscopy to provide insight into the structure of the IDeg molecule and its organization and self-assembly in the multi-hexamer. Circular dichroism, Infrared-, and Raman- spectroscopy showed that the fold of the protein backbone in the multi-hexamer is very similar to that of native insulin. The alpha-helical band is predominant and no signs of increased beta-sheet structure relative to that of human insulin were observed. Small angle x-ray scattering demonstrated the presence of strongly elongated structures with a repeated structural unit distanced by 34.8 Å, as indicated by a Bragg peak. The repeated distance originates from hexameric stacking and is close to the inter-hexameric distance also found & For author disclosure information, see page 829. A238 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS in classical crystals of human insulin. In contrast, amyloid fibril structures are characterized by a Bragg peak of 4.7 Å and a predominant beta-sheet structure. In conclusion, the IDeg multi-hexamers are composed of micrometer long linear arrays of hundreds of insulin hexamers. The ultra-long duration of action for IDeg relies on the formation of soluble multi-hexamers in subcutaneous tissue, and the protein fold adopted by insulin degludec in these multi-hexamers is very similar to that of human insulin. Supported by: Novo Nordisk, Inc. 939-P Long- Acting Insulin Analogues for Type 1 Diabetes Mellitus: A Systematic Review and Meta-Analysis Clinical Diabetes/ Therapeutics There is uncertainty regarding the optimal use of long-acting insulin analogues in type 1 diabetic patients (T1DM) because they are much more expensive than human neutral protamine Hagedorn insulin (NPH), with similar clinical outcomes. A systematic review and meta-analysis of clinical outcomes associated with long-acting insulin analogues in T1DM was done by searching all relevant electronic databases till April 2012.The main search concepts were T1DM and long-acting insulin analogues compared to NPH. The outcomes considered were the reduction of glycated hemoglobin (HbA1c) concentration and cumulative rate of hypoglycemia events (total, nocturnal and severe).Two reviewers independently extracted data from the articles and disagreements were solved by consensus. Sequence analysis of titles, abstracts and full texts resulted in the selection of 16 trials (8 detemir, 8 glargin). Most of the studies were of short to medium duration and of low quality. There was a high degree of heterogeneity that was not explained by patient characteristics of treatments. 15 studies had a parallel design and 3 were crossover. The studies included 4192 T1DM in total, with 2496 in the intervention groups and 1941 in the control group. Differences between insulin analogues and NPH in terms of HbA1c were marginal (SMD for insulin detemir: -0.15%, 95% CI -0.32% to 0.02%; for insulin glargin: -1.17%, 95% CI -2.44% to 0.10). Detemir had a protective effect for nocturnal hypoglycemias (SMD: 0.61, 95% CI 0.54 to 0.69), and marginal effect on total hypoglycemias (SMD:0.79, 95% CI 0.77 to 0.99). Glargin insulin did not demonstrated advantages for total, nocturnal or severe hypoglycemias (SMD:1.01, 95% CI 0.88 to 1.16), SMD 0.76, 95% CI 0.47 to 1.23, SMD 1.04, 95% CI 0.79 to 1.23, respectively).The results of this meta-analysis demonstrate that in type 1 diabetic patients, long-acting insulin analogues offer little benefit relative to conventional insulins in terms of glycemic control or hypoglycemia events. Supported by: Sanofi 940-P 941-P Asian Treat to Target Lantus Study (ATLAS): A 24-Week Randomized, Multinational Study Once-Daily Lixisenatide as Add-On to Basal Insulin ± OADs in Patients With Type 2 Diabetes Selectively Reduces Postprandial Hyperglycemic Daytime Exposure SATISH K. GARG, KARIM ADMANE, NICK FREEMANTLE, ATLAS STUDY GROUP, Aurora, CO, Paris, France, London, United Kingdom Self-adjustment of insulin dose and patient empowerment have been effective in achieving better glucose control. The ATLAS study evaluated titration of insulin glargine in uncontrolled insulin-naïve patients on 2 OADs in Asia. A total of 552 patients were randomized into 2 titration arms: 275 patient-led (PA), 277 physician-led (PH). Insulin dose was adjusted using the same algorithm to achieve a target FBG of 110 mg/dL in both arms. The primary objective was non-inferiority of change in A1c at 24 weeks from baseline. Baseline demographics and A1c were similar in both groups. A significant decrease in A1c was observed at Week 12 (-1.2%) and Week 24 (-1.3%). LS mean change from baseline (PA versus PH) was -0.15 (95% CI: -0.29 to -0.00; p=0.04), showing superiority. Proportions of patients with A1c < 7.0% without severe hypoglycemia (HYPO) (32.0% versus 26.0%, p=0.11) were similar in both groups. HYPO rate was low. Compared to PH, severe HYPO was similar, but nocturnal (p=0.002) and symptomatic HYPO (p=0.02) were higher in PA. Mean (±SD) insulin daily dose was 8.2 U (2.7) at baseline and 24.4 U (16.5) at Week 24 with a greater increase in PA (p<0.001). Few (2.6%) unrelated serious adverse events were reported. We conclude that insulin glargine titration, patient or physician-led, is safe and effective in achieving near target glucose control in Asian patients uncontrolled on OADs. Contrary to common belief, Asian patients titrated insulin dose up effectively when guided, similar to patients in the West. ADA-Funded Research & MATTHEW C. RIDDLE, YUTAKA SEINO, BERTRAND CARIOU, RICARDO GÓMEZ HUELGAS, CHRISTINE ROY-DUVAL, CAROLE HECQUET, ANDRES DIGENIO, JULIO ROSENSTOCK, Portland, OR, Osaka, Japan, Nantes, France, Málaga, Spain, Chilly-Mazarin, France, Bridgewater, NJ, Dallas, TX Basal insulin reduces basal hyperglycemia (BHG) but A1C may remain high due to persisting postprandial hyperglycemia (PPHG). Lixisenatide (LIXI), a GLP-1 receptor agonist in development for the treatment of T2DM, can reduce PPHG and A1C with no weight gain, or with weight loss, and thus has properties complementary to those of basal insulin. Patient-level data were pooled from 3 randomized Phase III studies of either once-daily LIXI + standard of care (SOC; basal insulin ± oral agents) vs placebo (PBO) + SOC to quantify the effects of LIXI on BHG and PPHG exposures. BHG (24-hr area above 5.6 mmol/L and under the fasting level) and incremental PPHG (area above fasting and under self-monitored 7-point plasma glucose profiles [AUC24h]) exposures were calculated by a previously reported method (Diabetes Care 2011;34:2508-14). The 753 eligible patients with evaluable profile data had mean age 57 years, BMI 29.8 kg/m2, diabetes duration 11.5 years, A1C 8.15% and fasting glucose 7.5 mmol/L. At baseline, mean daytime BHG and PPHG exposures were 49.2 and 55.1 mmol/L*h, respectively. Mean BHG and PPHG contributions to hyperglycemia were 42 and 58%, respectively. After 24 weeks of treatment, adjusted LS mean A1C change was -0.77% with LIXI + SOC and -0.29% with PBO + SOC (p<0.0001). The BHG exposure values for LIXI + SOC and PBO + SOC were similar (adjusted LS mean change for AUC24h -13 mmol/L*h vs -11 mmol/L*h [NS]), but PPHG exposure was reduced more with LIXI + SOC compared For author disclosure information, see page 829. Guided Audio Tour poster A239 POSTERS LUCIANA R. BAHIA, HELENA CRAMER, MARCIO LASSANCE, BRAULIO SANTOS, BERNARDO TURA, Rio de Janeiro, Brazil CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS with PBO + SOC (adjusted LS mean change -21 mmol/L*h vs -10 mmol/L*h, p<0.0001). The mean BHG and PPHG contributions to hyperglycemia were 46% vs 54% with LIXI + SOC and 39% vs 61% with PBO + SOC. In conclusion, this analysis suggests that once-daily LIXI complements the effects of basal insulin on glycemic control in type 2 diabetes, decreasing A1C mainly by reducing PPHG. 943-P Encapsulated Islet Sheet for Transplantation in Diabetic Yucatan Pigs LOURDES ROBLES, TIANYI ZHOU, MORGAN LAMB, MICHAEL ALEXANDER, EARL STEWARD, REMICK STAHL, RAHUL KRISHNAN, RICHARD STORRS, RANDY DORIAN, SCOTT KING, JONATHAN R. LAKEY, Orange, CA, San Francisco, CA Supported by: sanofi-aventis Encapsulating islets in a matrix, which would protect the cells from the body’s immune system, would have immense value in the treatment of diabetes. Our focus is the development of a biocompatible and retrievable alginate encapsulation method for the transplantation of isolated islets. The aim of this study was to use a large animal model of diabetes to test the in vivo effect of implanted islets encapsulated in alginate sheets. Yucatan pigs (20-50 kg) were rendered diabetic by an intravenous injection of Streptozotocin (STZ, 150 mg/kg). Diabetes was confirmed by two consecutive days of blood glucoses >350 mg/dL. Insulin was given twice a day post diabetes confirmation. Subsequently, pigs received four intra-abdominal sheets, roughly the size of a business card, made of purified alginate. Sheets contained an average of 360,000-390,000 islet equivalents per pig. No immunosuppression was given to the transplant recipients. Within 24 hours of STZ injection, pigs developed hyperglycemia with blood glucose levels >600 mg/dL. Prior to transplantation, insulin requirements were on average 0.8 U/kg/day to maintain average glucose levels of 336 mg/dl. After Islet Sheet transplantation the pigs recovered well with no signs of morbidity throughout the entire experimental period. The average insulin U/kg/day did not change post transplantation, however, sheets were removed after 48 days and remained both dithizone positive and viable (Newport green/PI) (Pre txp: 80.5 +/-3%, Post Txp: 78.2 +/- 4%). This study provides a reliable and reproducible method of inducing diabetes in a large animal model and provides a safe method of implanting encapsulated islets. Although the sheets were found to contain significant adhesions, viable islets were observed in the alginate sheets after an extended period. Future plans will focus on solidifying the sheet structure for prolonged in vivo survival. 942-P Hospital Management of Patients Prescribed U-500 Regular Insulin POSTERS Clinical Diabetes/ Therapeutics PURNIMA TRIPATHY, CECILIA LANSANG, Cleveland, OH Patients on U500 regular insulin (U500) as home regimen are often switched to different insulins when hospitalized. This study aims to determine the insulin regimen used and glucose levels of U500 users who are hospitalized. Retrospective chart review done on adult patients on U500 as their home regimen, admitted to the Cleveland Clinic from 2001-2011. Patients were divided into those switched to a different regimen (Group A) and those continued on U500 (Group B). A subset of patients on U500 for >50% of their stay was studied. Glucose levels were compared by Wilcoxon rank sum test. Rates of hypo- and hyperglycemic days were compared by Poisson regression models. Total 61 patients, 59% males, BMI 41.4+10.8 (mean + SD), age 59+12 yr, A1C 9.1+1.9%, length of stay 6.6+4.7 days,60.6% had Endocrinology consult. 32.7% of patients were switched to a combination of long-, intermediate-, short- and fast-acting insulin. Mean glucose levels were not significantly different between groups (Table 1). Group A was given much less insulin compared to home regimen. Group B had more frequent hypoglycemia (15.3+21.3 vs 2.8+6.4%). Group B had higher frequency of severe hyperglycemia (16.8+ 21.9 % vs 6.3+9.8%). One-third of patients were switched to a different insulin regimen. The similar glucose levels between groups, lower admission insulin doses in Group A, more frequent hypoglycemia and severe hyperglycemia in Group B, suggest U500 may not be suitable to continue in some patients when hospitalized. Supported by: Hanuman Medical Foundation; University of California, Irvine 944-P Table 1 Group A Group B (not continued (continued on U500) on U500) N=20 N=41 Blood glucose during 207.9 237.6 admission (mg/dL) (128.2-335.6) (115.5-392.1) Median (range) Insulin dose prior to 100 235 admission (units) (10.0-500.0) (10.0-2450.0) Median (range) Insulin dose during 34.8 200 admission (units) (5.0-99.5) (19.0-1156.9) Median (range) % hypoglycemia- days 2.8 (6.4) 15.3 (21.3) (BG < 70 mg/dL) Mean (SD) % severe hypoglycemia1.3 (5.6) 0.1 (0.8) days (BG < 40 mg/dL) Mean (SD) % hyperglycemia- days 80.6 (24.2) 78.9 (28.0) (BG >200 mg/dL) Mean (SD) % severe hyperglycemia6.3 (9.8) 16.8 (21.9) ays (BG >400 mg/dL) Mean (SD) *NA=too few patients to compare p Value 0.48 Metabolism of Insulin Glargine in T2DM on Long-Term Glargine Use: A Double Blind, Randomized, Cross-Over, Dose Response Study Group B Group B p Value patients patients on U500 for on U500 for >50% of <50% of hospital stay hospital stay N=34 N=7 229.2 290.3 0.14 (115.5-392.1) (159.3-346.0) 0.068 235 207.5 (10.0-2450.0) (10.0-480.0) 0.79 <0.001 223.2 129.7 (19.0-1156.9) (32.5-387.5) 0.16 <0.001 12.8 (19.1) 15.8 (21.9) 0.26 NA* 0.0 (0.0) 0.2 (0.9) NA* 0.55 81.9 (28.4) 78.2 (28.3) 0.90 <0.001 32.5 (30.2) 13.6 (18.7) 0.20 PAOLA LUCIDI, FRANCESCA PORCELLATI, PAOLA CANDELORO, PATRIZIA CIOLI, ANNA MARINELLI ANDREOLI, STEFANIA MARZOTTI, RONALD SCHMIDT, GEREMIA B. BOLLI, CARMINE G. FANELLI, Perugia, Italy, Frankfurt, Germany After subcutaneous (s.c.) injection in humans, insulin glargine (GLA) undergoes cleavage of the di-arginine in 30-B position and subsequent loss of threonine (30-B) with formation of metabolites M1 and M2, which do not differ from human insulin for mitogenesis. Aim of the present cross-over study was to establish metabolism of GLA after s.c. injection of therapeutic and high doses of GLA in subjects with T2DM on long-term use of GLA. Ten subjects were studied during 24 h of fasting after s.c. injection of 0.4 and 0.8 U/kg of GLA given in the morning, on two separate occasions during euglycemic clamps. GLA, M1 and M2 over 24 h period were extracted using immunoaffinity columns and quantified by a specific liquid chromatographytandem mass spectrometry assay. Plasma M1 (AUC0-24 h) was detected in all subjects with both doses of GLA and increased by increasing the dose from 0.4 U/Kg to 0.8 U/Kg by 54% (95% CI: 39 to 91, p=0.008). GLA was detectable in plasma in only six and nine out of ten subjects after dosing 0.4 and 0.8 U/Kg, respectively. M2 was not detected. At the high dose of 0.8 U/Kg, plasma GLA concentration (AUC0-24h) represented only 9.7% (4.6 and 15) of the total amount of insulin measured in the blood. After s.c. injection of GLA, even at high dose, this is nearly totally metabolized to the active metabolite M1 and GLA parent is generally scarcely detectable in blood. In vivo, GLA does not exert its metabolic effects directly, but via its main metabolite M1. Supported by: Sanofi & For author disclosure information, see page 829. A240 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS 945-P 947-P Insulin Degludec (IDeg) and Insulin Aspart (IAsp) Can Be Co-Formulated such that the Formation of IDeg Multi-Hexamers and IAsp Monomers Is Retained Upon S.C. Injection IDegAsp Produces Dose-Proportional Glucose-Lowering Effect in Subjects With Type 1 Diabetes Insulin degludec/insulin aspart (IDegAsp) is a 70%/30% combination of insulin degludec (IDeg) and insulin aspart (IAsp), providing both an ultralong-acting basal component and a rapid-acting bolus component in 1 injection. The pharmacodynamic (PD) properties of 3 doses of IDegAsp (0.4, 0.6, 0.8 U/kg body weight) were investigated in a double-blind, single-dose incomplete block crossover trial in subjects with type 1 diabetes (T1DM). Following each dose a 26-h euglycaemic clamp procedure was performed (Biostator, glucose target: 100 mg/dL). A total of 33 C-peptide-negative subjects with T1DM (mean age 39 y, BMI 25.1 kg/m2, HbA1c 8.0%) were included. With increasing dose, the 24-hour area under the glucose infusion rate (GIR) curve (area under the curve [AUC]GIR,0-24h,SD) and the maximum GIR (GIRmax,SD) increased significantly and proportionally. Estimated log doselog AUCGIR,0-24h,SD (1.19, 95% CI: 0.99; 1.40) and log dose-log GIRmax,SD (0.89 [0.66,1.13]) both supported dose proportionality as 1.0 was included in the 95% CI. The 24-hour single-dose GIR profile of IDegAsp showed a distinct peak action due to prandial IAsp, and separate and stable basal action from IDeg (Fig 1). In conclusion, IDegAsp produces a proportional dose-response across 3 clinically relevant dose levels and a PD profile with separate prandial and basal components in T1DM. Supported by: Novo Nordisk, Inc. 946-P Development and Validation of a Clinical Score to Predict Insulin Requirement for Optimum Control of Blood Glucose during Glucocorticoid Treatment in Patients With Autoimmune Diseases HIROYUKI MORITA, ICHIRO MORI, KEI FUJIOKA, HIDEYUKI OKADA, TARO USUI, MAYUMI TANIMOTO, KAZUO KAJITA, TATSUO ISHIZUKA, Gifu, Japan Long-term glucocorticoid (GC) treatment may be necessary for patients with various types of autoimmune diseases. It is important to predict insulin requirement in advance for optimum control of GC-induced hyperglycemia in autoimmune diseases. The aim of the present study is to develop and validate a score to predict insulin requirement for optimum control of blood glucose during GC treatment. We retrospectively evaluated 146 patients with autoimmune diseases admitted between 2004 and 2011 to develop the prediction score. Inclusion criteria indicated that the dosage of prednisolone treatment should be more than 5 mg/day for 4 weeks at least. Exclusion criteria included insulin therapy before GC administration or use of a GLP-1 analog. Seventeen percent of GC-administered patients required treatment with insulin. A logistic regression analysis revealed that risk factors of insulin requirement during GC administration should be raised for males, FPG, HbA1c, and maximum dose of prednisolone (PSL). A ROC analysis showed that cutoff values of FPG, HbA1c, and PSL were 99 mg/dl, 6.7%, and 0.75 mg/kg, respectively. We scored patients profiles as follows; male, 1 point; FPG (mg/dl), 90-98,1 point, 99-109, 2 points, more than 110 mg/dl, 3 points; HbA1c (%), 5.8-6.6, 1 point, 6.7-6.9, 2 points, more than 7.0, 3 points; PSL (mg/kg), 0.50-0.74, 1 point 0.75-0.99, 2 points, more than 1.00, 3 points. Sum of the scores was 6.5±1.4 for insulin-required patients vs. 2.9±1.5 for others. When 5 points was defined as the cut-off value of the sum of the scores, the sensitivity, specificity and accuracy were 96%, 90% and 91%, respectively. We validated the scoring to another 42 patients admitted in 2012. The sensitivity, specificity and accuracy were 83%, 87% and 86%, respectively. The clinical score is a valid and reliable tool to predict insulin requirement for optimum control of blood glucose during GC treatment. ADA-Funded Research & Supported by: Novo Nordisk, Inc. 948-P Glargine vs. Premixed Insulin for Management of Type 2 Diabetes Patients Failing Oral Antidiabetic Drugs: The GALAPAGOS Study PABLO ASCHNER, BIPIN SETHI, FERNANDO GOMEZ-PERALTA, WOLFGANG LANDGRAF, MARIE-PAULE DAIN, VALERIE PILORGET, ABDURRAHMAN COMLEKCI, Bogotá, Colombia, Hyderabad, India, Segovia, Spain, Frankfurt, Germany, Paris, France, Izmir, Turkey GALAPAGOS was a 24-wk, open-label, multinational, superiority trial of glargine (GLA) (± glulisine) vs premixed insulin (PRE) in insulin-naïve T2DM patients (pts) uncontrolled on OADs. Pts were randomized to GLA QD or PRE (QD or BID) while continuing OADs (at least metformin ± insulin secretagogue [IS]). A 2nd PRE injection could be added any time; glulisine could be added with main meal in GLA QD pts with A1C ≥7% and FPG <126 mg/dL at ≥ week 12. IS was stopped with 2nd injection. Insulin titration targeted FPG ≤100 mg/dL. Primary endpoint was % of pts with A1C <7% and no symptomatic confirmed (PG ≤56 mg/dL) hypoglycemia at 24 wks. mITT population comprised 923 pts (GLA: 462; PRE: 461). Baseline characteristics were similar in both groups (mean T2DM duration: 9 y; A1C: 8.7%; FPG: 161 mg/dL). Table shows key study endpoints. Similar percentages of pts reached the primary endpoint in both groups, demonstrating noninferiority (prespecified margin: 25% of PRE rate), but not superiority, of GLA. More GLA (57.5%) than PRE pts (36.5%) stayed on 1 injection/day over 24 wks. Symptomatic hypoglycemia was less with GLA than PRE. Efficacy predictors in both arms were geographic region, diabetes duration, baseline A1C and number of OADs at baseline. For author disclosure information, see page 829. Guided Audio Tour poster A241 POSTERS Insulin degludec is an ultra-long-acting insulin which is the first basal insulin that can be co-formulated with a fast-acting insulin analogue (insulin aspart). It has been previously demonstrated in vitro that the prolonged absorption of IDeg and its resulting ultra-long duration of action are due to the formation of soluble high molar mass complexes known as multi-hexamers; whereas, the fast action of IAsp is achieved by rapid dissociation of hexameric complexes after s.c. injection. The aim of this study was to determine the mechanism underlying the ability of IDeg and IAsp to be combined without altering their individual absorption profiles after injection. Size exclusion chromatography (SEC) was used as an in vitro model. The column was eluted with buffered saline including varying phenol concentrations to simulate the disappearance of phenolic excipients from the formulation upon s.c. injection. Two desB30 basal insulin analogues acylated at LysB29Nɛ by hexadecandioyl-ɣ-Glu (IDeg) or lithocholyl-ɣ-Glu and IAsp were compared with human insulin. The analogues were examined by SEC, formulated alone and in different combinations of long acting and fast acting insulin at a 70:30 ratio. Combinations of these long acting and fast acting analogues were found to form mixed hexamers and di-hexamers, but by adjusting the zinc concentration, it was possible to avoid mixed hexamers and mixed di-hexamers of IDeg and IAsp and their individual SEC elution profiles could be retained. It is concluded that the ability of IDeg to form multi-hexamers in the presence of zinc is critical to avoid interactions with IAsp such that the individual absorption profiles upon injection can be retained. Clinical Diabetes/ Therapeutics TIM HEISE, LESZEK NOSEK, OLIVER KLEIN, HANS-VEIT COESTER, CARSTEN ROEPSTORFF, ANNE LOUISE SVENDSEN, HANNE HAAHR, Neuss, Germany, Søborg, Denmark SVEND HAVELUND, ULLA RIBEL, FRANTA HUBÁLEK, THOMAS HOEG-JENSEN, IB JONASSEN, Måløv, Denmark CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS A GLA- or PRE-based regimen resulted in similar % of well-controlled pts without hypoglycemia; however, overall hypoglycemia was greater with PRE and it should be considered in evaluating the treatment benefit/risk. 950-P Pharmacological Benefi ts of Once Weekly Combination Treatment using LAPS-Insulin and LAPS-Exendin-4 in Animal Models Basal insulin and GLP-1 receptor agonist combination therapy offers the most promising approach for DM patients based on its synergistic effects on glycemic control and complementary effects on body weight. However, the currently available combination regimens require frequent dosing which give inconvenience to patients. We are developing both a once-aweek LAPS-Insulin (HM12460A) and a LAPS-Exendin-4 (INN: langlenatide, once-a-week or once-a-month regimen) using the proprietary LAPSCOVERY technology. The objective of this study was to investigate the beneficial effects of a once-a-week combination of LAPS-Exendin-4 and LAPS-insulin in animal models. From the oral glucose tolerance study in normal mice, we confirmed that the combination regimen efficiently reduced both fasting and post-prandial blood glucose. The same result was also observed in normal dogs, where the synergistic pharmacological effects were sustained for 4 days post-injection. In db/db mice, once-a-week administration showed several beneficial effects compared with the LAPS-Insulin monotherapy, such as improved glycemic control, neutralizing body weight gain, and reduction of insulin dose. Based on these observations, we concluded that the once-a-week combination regimen of LAPS-Insulin and LAPS-Exendin-4 could be a therapeutic approach, which confers maximal convenience to DM patients in addition to superior glycemic control. POSTERS Clinical Diabetes/ Therapeutics CHANG KI LIM, YOUNG JIN PARK, IN YOUNG CHOI, GYU HWANG LEE, JA HOON KANG, YOUNG HOON KIM, JEEWOONG SON, SE CHANG KWON, Hwaseong-si, Republic of Korea, Seoul, Republic of Korea Supported by: Sanofi 949-P Association between TNF-Alpha (TNF) Promoter Allele and Effi cacy of Insulin Therapy in Chinese Type 2 Diabetic Patients HAIXIA XU, WEN XU, FEN XU, HUA LIANG, JINHUA YAN, JIANPING WENG, Guangzhou, China Objectives: To investigate the genetic association between tumor necrosis factor-α (TNF-α) polymorphisms [-1031T>C (rs1799964), -863C>A (rs1800630), -857C>T (rs1799724)] and the clinical efficacy of insulin therapy in treating Chinese type 2 diabetic patients. Methods: Three hundred sixty-eight newly diagnosed type 2 diabetic patients to 168 ethnically matched healthy individuals as controls were involved. The three single nucleotide polymorphisms (SNPs) were analyzed by PCR-direct sequencing and haplotype analysis were performed to identify possible risk haplotypes for type 2 diabetes. Then patients carrying the risk allele together with its controls received mixed protamine zinc recombinant human insulin Lispro (25R) injection twice daily to attain glycemia control. Fasting blood glucose (FBG), postprandial blood glucose (PBG), glycohemoglobin A1C (GHAlc), fasting serum insulin (FSI) were measured before and every three months after insulin therapy for a total of one year. Results: The frequency of the TNF allele designated -1031C, -863C, and -857C was significantly increased in patients with type 2 diabetes (odds ratio =2.665; P=0.015). Except for a lower level of HDL-c, carriers of -1031C/863C/-857C (CCC group) exhibited similar FBG, PBG, GHAlc, FSI and HOMAIR to those none CCC carriers (none CCC group) at baseline. After one year of Lispro25R treatment, beta cell function was partially restored in most patients as evidenced by dramatic increased FSI level in both groups. Moreover, CCC group presented even higher FSI than none CCC ones (p=0.033) while Homa-IR remained statistically comparable. Conclusions: TNF allele of -1031C, -863C, and -857C might be both a risk haplotype for developing type 2 diabetes and a predictive genetic factor for clinical efficacy of insulin therapy. 951-P Pharmacokinetics and Pharmacodynamics of Insulin Glargine, Changes in Glucose and Lipid Metabolism after Either Evening or Morning Injection in T2DM FRANCESCA PORCELLATI, PAOLA LUCIDI, PAOLA CANDELORO, PATRIZIA CIOLI, ANNA MARINELLI ANDREOLI, GEREMIA B. BOLLI, CARMINE G. FANELLI, Perugia, Italy We have recently shown the differential pharmacokinetics and pharmacodynamics (PK/PD) of insulin glargine (GLA), when administered in T2DM at different time of the day (Diabetes, 61 (Suppl.1): A249, 2012). Here we report GLA effects on glucose (G) fluxes (deuterated G), lipid metabolism (plasma FFA, β-OH-butyrate) in 9 persons with T2DM (mean±SD: age 62±3.5 yrs; BMI 28±3.6 kg/m2; A1C 7.2±1.1%, known T2DM duration 19±9.9 yrs), during a 24h euglycemic G clamp (cross-over study), with GLA given s.c. (0.4 U/kg), either at 10 AM or at 10 PM. Total G metabolism (GIR_AUC0-24h) was similar (1012±731, 1058±606 mg/kg, p=0.654), but AUC0-12h was greater in AM vs PM (589±397 vs 358±259 mg/kg, p=0.010), whereas the opposite was observed for AUC12-24h (424±357 vs 700±420 mg/kg, AM and PM, p=0.005). PK of GLA and C-peptide did not differ in the two studies. The different GIR in AM vs PM was explained by changes in endogenous G production (EGP) with greater suppression in initial 12 h and lower in second 12 h, when GLA was given at 10 AM; the opposite occurred whit GLA given at 10 PM (AUC012h 510±284 vs 744±384 mg/Kg, AM and PM, p=0.066). G utilization was not stimulated neither in AM nor in PM studies. Lipolysis was more suppressed in PM vs AM (FFA AUC0-24h 7.5±1.7 vs 9.2±1.7 mmol.h/L, p<0.001; β-OH AUC024h 6.8±5.0 vs 18±12 mmol.h/L, p<0.003). In conclusion: PD of insulin GLA in T2DM is dependent on the time of daily injection. Evening administration results in greater GIR requirements in the second 12 h of day (afternoon), due to greater suppression of EGP and lipolysis, in contrast to morning administration which displays a sensible reduction in metabolic activity from 12 h to 24 h after GLA injection. Since PK and C-peptide were similar in the 2 studies, the differential PD might reflect the contribution of diurnal fluctuation in insulin sensitivity (higher in afternoon vs early morning). & For author disclosure information, see page 829. A242 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS achieved on insulin may predict those patients who can later succeed in controlling glucose levels with diet and physical exercise only. 952-P Continuous Glucose Monitoring to Study the Effectiveness of U-500 Regular Insulin as Basal Insulin in Severely Insulin Resistant Patients 954-P SHUBHANGI G. SHIDHAM, VALERIE CHRISTENSEN, BEVERLY PINKSTON, Columbus, OH Determinants of Reversibility of Beta-Cell Dysfunction in Response to Short-Term Intensive Insulin Therapy in Patients With Early Type 2 Diabetes Severe insulin resistance is defined as requiring more than 200 units of insulin per day or more than 2 units of insulin per kg per day. This subset of patients is more likely to have uncontrolled diabetes mellitus. Studies have shown effectiveness of U500 regular insulin in this subset of patients and is considered to have both prandial and basal activity. However, in clinical practice we have found U-500 regular insulin to work best as a basal insulin. To evaluate this observation we undertook following prospective study. Fifteen severely insulin resistant patients (aged 59.9 ± 6.50, BMI 42.16 ± 6.79 with baseline HgA1c of 9.33% ± 1.33) were administered preprandial BID or TID doses of U-500 regular insulin. No bedtime doses were administered. Continuous Glucose Monitoring (CGM), HgA1c, and weight/ BMI measurements were performed before and 6 months after starting U500 regular insulin. HgA1C decreased from 9.33% to 7.98% (p=0.0004), BMI increased from 42.16 kg/m² to 44.4 kg/m² (p=0.004). The average glucose between 12 AM to 6 AM decreased from 210.2±59.2mg/dl to 180±41.5mg/dl (p=0.020) showing effectiveness of U-500 as basal insulin. Average glucose in post prandial periods also decreased, but these results were not statistically significant. We therefore conclude that U-500 insulin works most effectively as basal insulin supporting our clinical observation. BMI kg/m² HgA1c Average glucose 12MN to 6am mg/dl Average glucose post breakfast mg/dl Average glucose post lunch mg/dl Average glucose post dinner mg/dl 6 months After U-500 (mean± SD) 44.45 ± 7.76 7.98% ± 0.82 180.2 ± 41.5 234.3 ± 55.1 210.5 ± 65.2 0.35 220.6 ± 35.1 196.3 ± 46.8 0.064 216.5 ± 41.1 200.5 ± 35.3 0.15 P value 0.004 0.0004 0.020 955-P Metformin (M), Sulfonylurea (SU), or Both as Background OAD Therapy for Insulin Glargine (IG): A Pooled Analysis of Treat-ToTarget Trials Supported by: U.S. Dept. of Veterans Affairs J. HANS DEVRIES, ANTHONY H. BARNETT, TIMOTHY REID, MARIE-PAULE DAIN, WOLFGANG LANDGRAF, ALEKSANDRA VLAJNIC, LOUISE TRAYLOR, RICHARD M. BERGENSTAL, Amsterdam, Netherlands, Birmingham, United Kingdom, Janesville, WI, Paris, France, Frankfurt, Germany, Bridgewater, NJ, Minneapolis, MN 953-P This pooled, patient-level data analysis evaluated efficacy and safety of adding IG to M, SU, or M+SU in uncontrolled T2DM. Eligible studies were randomized, treat-to-target trials with FPG target <100 mg/dL and ≥24 weeks duration. From 15 studies, 634 patients were treated with IG+M, 906 with IG+SU, and 1297 with IG+M+SU (Table). Hypoglycemia definitions included: overall (with plasma glucose [PG] <70, <56 mg/dL or requiring assistance); nocturnal (0:01-5:59am); severe (requiring assistance); and documented severe (requiring assistance and PG <36 mg/dL). Event rates and incidence were modeled considering baseline characteristics. All treatments reduced A1C; fewer patients reached A1C <7.0% with IG+SU vs IG+M or IG+M+SU. IG+SU and IG+M+SU treatment increased overall hypoglycemia rates up to 3-fold compared with IG+M. Severe hypoglycemia was rare in all groups. More IG+M patients achieved A1C <7.0% without nocturnal hypoglycemia (PG <70 mg/dL), than IG+SU and IG+M+SU patients. IG+M resulted in less weight gain than the other combinations, despite a higher insulin dose. In an overall model including all patients, baseline BMI, T2DM duration, age, and OAD group were significant factors affecting hypoglycemia. Thus, there may be a clinical advantage for IG+M therapy compared with IG+M+SU or IG+SU in terms of body weight and minimizing hypoglycemia risk while achieving glycemic goals, albeit with a higher insulin dose. Transient Continuous Subcutaneous Insulin Infusion Therapy in Newly Diagnosed Type 2 Diabetes With Obesity YA DONG SUN, Changchun, China The objective of this study was to investigate whether long-term optimal glycemic control and recovery of β-cell function can be achieved without medication by transient continuous subcutaneous insulin infusion (CSII) in newly diagnosed type 2diabetes with obesity. A total of 179 subjects (52 ± 2 years old [range 35-64], BMI 30.8 ± 1.9 kg/m2) with newly diagnosed type 2 diabetes were enrolled and had a 3 week treatment of CSII and then discontinued. Intravenous glucose tolerance tests (IVGTTs) were performed, and blood glucose, HbA1c, insulin, and C-peptide were measured before and after CSII. Patients were followed longitudinally on diet and physical exercise only after withdraw of insulin. All 179 subjects tolerated the intensive insulin treatment well. 29 patients who failed to achieve excellent glycemic control on CSII within 3 weeks were considered as early therapeutic failure and excluded from further analysis. Before CSII, the glycemic control was poor in the 150 diabetic patients, with average FPG of 15.27 ± 2.32 mmol/l, PPG 22.18 ± 3.43 mmol/l, and HbA1c 14.95 ± 2.42%. After treatment, FPG and PPG levels were significantly reduced (5.52 ± 1.30 mmol/l and 7.18 ± 1.21 mmol/l, respectively; P < 0.001). After 3 weeks CSII, there was a marked decrease in HbA1c (from 14.95 ± 2.42 to 8.76 ± 1.33%, P < 0.05). The remission rates (percentages maintaining near euglycemia) at the 3, 8, 16, and 24 weeks were 83.8, 76.9, 70.1, and 67.1%, respectively. These results demonstrate that in newly diagnosed type 2 diabetes patients with obesity and elevated fasting glucose levels, a 3 week treatment of CSII can successfully lay a foundation for prolonged good glycemic control. The improvement of β-cell function, especially the restoration of first-phase insulin secretion, could be responsible for the remission. The ease with which normoglycemia is ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A243 POSTERS Short-term intensive insulin therapy (IIT) can improve beta-cell function when administered early in the course of type 2 diabetes (T2DM). However, the degree of reversibility of beta-cell dysfunction in response to this therapy varies between patients. Thus, we sought to characterize the determinants of reversibility of beta-cell dysfunction in response to shortterm IIT in early T2DM. In this study, 63 patients with mean 3.0 ±2.1 years duration of T2DM and HbA1c 6.8 ±0.8% underwent 4 weeks of IIT, with oral glucose tolerance test (OGTT) administered at baseline and one day postIIT. Beta-cell function before and after IIT was assessed on OGTT using the Insulin Secretion-Sensitivity Index-2 (ISSI-2). Reversibility of beta-cell dysfunction was defined as percentage change in ISSI-2 ≥25%. In the entire cohort, there was an increase in ISSI-2 from baseline to post-IIT (median 183 vs 190, P=0.01), with 1/3 of participants achieving ≥25% improvement in ISSI-2. Compared to their peers, those with increase in ISSI-2 ≥25% had significantly greater decrements in fasting glucose (-1.6 ±1.0 vs 0.1 ±0.9 mmol/L, P<0.001), HbA1c (-0.8% ±0.5 vs -0.3% ±0.3, P=0.001), ALT [median -3 vs 1 IU/L, P=0.04), AST [ median -3 vs 1 IU/L, P=0.02], and HOMA-IR [median -0.7 vs -0.1, P<0.001). On logistic regression analysis, baseline HbA1c (OR 2.83, 95%CI 1.16-6.88, P=0.02) and change in HOMA-IR (OR 0.008, 95%CI 0.0004-0.16, P=0.001) emerged as independent predictors of reversibility of beta-cell dysfunction. Indeed, only those participants in whom IIT yielded an improvement in HOMA-IR achieved reversibility of beta-cell dysfunction. In conclusion, decline in HOMA-IR is a key determinant of improvement of beta-cell function in response to short-term IIT, underscoring a fundamental contribution of insulin resistance to the reversible component of beta-cell dysfunction in early T2DM. Clinical Diabetes/ Therapeutics Before U-500 (mean± SD) 42.16 ± 6.79 9.33% ± 1.39 210.0 ± 59.2 CAROLINE K. KRAMER, HAYSOOK CHOI, BERNARD ZINMAN, RAVI RETNAKARAN, Toronto, ON, Canada CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS POSTERS Clinical Diabetes/ Therapeutics IG+M (n=634) 54.7 (9.6) 47.9 31.5 (5.8) 88.0 (19.7) 7.3 (5.7) 8.7 (1.1) 188 (59) 7.0 (1.0) −1.7 (1.2) 56.8 0.8 (4.0) 0.51 (0.27) IG+SU (n=906) 59.1 (10.4) 46.1 29.1 (5.2) 82.2 (18.0) 9.5 (6.6) 9.0 (1.1) 205 (58) 7.6 (1.2) −1.4 (1.2) 31.5 2.8 (3.9) 0.43 (0.24) IG+M+SU (n=1297) 58.2 (9.3) 45.4 31.2 (5.0) 88.9 (17.4) 9.5 (6.3) 8.7 (1.0) 193 (52) 7.1 (0.9) −1.6 (1.0) 49.3 2.0 (3.4) 0.42 (0.25) 957-P Pooled (N=2837) 57.7 (9.8) 46.2 30.6 (5.3) 86.6 (18.4) 9.0 (6.4) 8.8 (1.1) 196 (56) 7.2 (1.0) −1.6 (1.2) 45.3 2.0 (3.8) 0.44 (0.26) Interchangeability of Exhaled Gases for Plasma Insulin Prediction in Obesity and Type 2 Diabetes Age, y Women, % Baseline BMI, kg/m2 Baseline weight, kg T2DM duration, y Baseline A1C, % Baseline FPG, mg/dL Week 24 A1C, % A1C change from baseline, % Patients with A1C <7.0% at week 24, % Weight change from baseline, kg Week 24 insulin dose, U/kg Overall hypoglycemia, event rate/PY (SE)a <70 mg/dL 2.05 (0.24) 5.03 (0.84) 5.78 (0.55) 3.90 (0.21) <56 mg/dL 0.64 (0.10) 1.60 (0.34) 2.16 (0.26) 1.30 (0.09) a Overall hypoglycemia, incidence, % (SE) <70 mg/dL 37.0 (3.0) 52.0 (4.4) 50.7 (2.6) 46.5 (1.4) <56 mg/dL 16.3 (2.3) 30.1 (4.5) 34.1 (2.7) 26.0 (1.3) Nocturnal hypoglycemia, event rate/PY (SE)b <70 mg/dL 0.37 (0.07) 0.75 (0.21) 0.79 (0.12) 0.60 (0.05) <56 mg/dL 0.18 (0.04) 0.25 (0.10) 0.36 (0.07) 0.25 (0.03) Nocturnal hypoglycemia, incidence, % (SE)b <70 mg/dL 11.5 (1.8) 15.0 (3.5) 17.5 (2.1) 14.5 (1.0) <56 mg/dL 4.8 (1.2) 5.4 (2.3) 11.3 (2.1) 6.7 (0.7) Severe hypoglycemia, event rate/PY (SE)c 0.07 (0.03) 0.10 (0.03) 0.10 (0.03) 0.09 (0.02) Severe hypoglycemia, incidence, % (SE)c 2.2 (0.6) 2.2 (0.5) 1.9 (0.4) 2.1 (0.3) Documented severe hypoglycemia, event rate/PY (SE)d 0.01 (0.01) 0.007 (0.004) 0.002 (0.002) 0.005 (0.002) Documented severe hypoglycemia, incidence, % (SE)c 0.4 (0.3) 0.3 (0.2) 0.1 (0.1) 0.2 (0.1) A1C <7.0% at week 24 with no nocturnal hypoglycemia 48.0 27.0 35.8 35.7 (<70 mg/dL), % Table. Baseline and week 24 data. Data represent mean (SD) unless otherwise specified. PY = patient year. Hypoglycemia incidence and event rates estimated and analyzed using logistic and negative binomial regression: aadjusting for BMI, duration of diabetes, age, group, study, interaction of duration and group; badjusting for BMI, duration of diabetes, age, group, study; cadjusting for BMI, duration of diabetes, age, group; dadjusting for duration of diabetes, age, group. STACY R. OLIVER, MATTHEW K. CARLSON, HEATHER SOPHER, SIMONE MEINARDI, DONALD R. BLAKE, PIETRO GALASSETTI, Irvine, CA Exhaled volatile organic compounds (VOC) are ideal non invasive markers of metabolism, whose practical application, though, remains challenging. We previously estimated plasma glucose, insulin (INS) and lipids via VOC analysis in healthy and type 1 diabetics, obtaining stronger predictions with condition-specific gas patterns. Ideal predictive models, though, should incorporate the same gas clusters for as many patient groups as possible. INS testing is important in obesity (OW)/type 2 diabetes (T2D), where INS changes far precede hyperglycemia. We hypothesize that interchangeable exhaled gas clusters can yield similarly accurate INS predictions in OW and T2D. During 4 h of glycemic/INS fluctuations, we collected in 17 T2D and 12 OW subjects, 12 plasma, room air and breath samples; ~100 VOC were quantified by gas chromatography, matched with plasma INS, and gas-based INS predictions obtained by linear regression analysis. The gas clusters yielding the strongest measured/predicted INS correlations were: acetone, 2-BuONO2, α-pinene in OW (r = 0.90); 2-methylpentane, CH3Cl, acetone, n-heptane, 2-methylhexane in T2D (r = 0.93). Applying the T2D model to OW, r = 0.91; applying the OW model to T2D, r = 0.85. More interchangeable clusters were used between groups with r values > 0.8. We show the feasibility of estimating plasma INS in OW and T2D with interchangeable VOC clusters, a feature very relevant at the later stage of practical development of actual testing devices. Supported by: NIH (1UL1RR031985), (K24DK085223) Supported by: Sanofi U.S., Inc. 956-P 958-P Intensive Glucose Lowering Therapy Maintains Beta-Cell Function for 6 Years in Newly Diagnosed Type 2 Diabetes An Emergency Room Triggered Package of Measures in Hyperglycemic Patients Reduces Mortality and Hospital Stay—Results of the GLUC-EMERGE Follow-Up Study LINDSAY B. HARRISON, PHILIP RASKIN, BEVERLEY ADAMS-HUET, XILONG LI, ILDIKO LINGVAY, Dallas, TX THORSTEN SIEGMUND, EDIN ZELIHIC, BORIS PONELEIT, PETRA-MARIA SCHUMM-DRAEGER, CHRISTOPH DODT, Munich, Germany Diabetes is a progressive disease marked by a steady decline in beta-cell function which leads to increased complications and treatment difficulty. We assessed the efficacy of 6 years of early intensive therapy with either insulin plus metformin (INS) or triple oral therapy (TOT) with metformin, glyburide, and pioglitazone on glycemic control and beta-cell function. We conducted a randomized trial of 58 patients with treatment-naïve newly diagnosed type 2 diabetes. An initial 3-month lead-in period with insulin and metformin was followed by randomization to INS or TOT. We evaluated glycemic control, safety (hypoglycemia, weight gain), treatment satisfaction, and beta-cell function (using mixed meal challenge testing). Analyses were intention-to-treat and performed with repeated measures models. Beta-cell function was preserved with no significant change from the time of randomization (after 3 months of INS treatment) or difference between the two groups as measured by area under the curve (AUC) of c-peptide (p=0.14) or ratio of c-peptide to glucose AUC (p=0.7). Excellent glycemic control was maintained in both groups (end-of-study HbA1c 7.3±1.9% INS versus 6.8±1.7% TOT) with 63.2% and 68.8% respectively maintaining HbA1c <7% by the end of the study. BMI increased in both groups but significantly more in the TOT group (35.6±6.6 to 35.4±8.4 kg/ m2 in INS and 36.5±8.0 to 38.8±10.1 kg/m2 in TOT) (p=0.04). Hypoglycemic events decreased significantly over time but did not differ between groups. There were 8 failures in INS and 6 in TOT (defined as HbA1c >8% on two consecutive visits). Predictors of failure included randomization systolic blood pressure, BMI, HbA1c, HOMA-IR, and c-peptide. In conclusion, proactive intensive treatment at the time of type 2 diabetes diagnosis - initial short-term insulin treatment followed by either an insulinbased or oral hypoglycemic-based therapy - can preserve beta-cell function for at least 6 years. Introduction: Hyperglycemia at hospital admission is associated with an increased risk of complications and mortality. Only few trials investigated hyperglycemic patients (P) during the whole hospital stay, starting from the emergency room (ER) to the normal wards until discharge. This study examines the effects of early identification followed by a structured treatment of hyperglycemia. Methods: We performed a prospective study of P (n=2832) admitted to the ER of an academic teaching hospital over a time period of 3 months. According to actual recommendations, P with a blood glucose (BG) of ≥ 140 mg/dl received active follow up (n=463). In P with severe BG derailment (blood glucose ≥ 180 mg / dl, n=369), insulin was administered by an individualized but prefabricated basal-bolus scheme followed by diabetologists consultations. The effects were tested using a multivariate regression analysis, correcting for BG-independent risk factors such as age, creatinine and other laboratory parameters. Moreover, we performed a comparison with the results of the GLUC-EMERGE (GES) study, an observational study, when hyperglycemia was not treated actively with the ER triggered package of medical measures. Results: The GLUC-EMERGE follow-up study cohort showed a significant decrease of BG-related hospital stay prolongation compared to the GES cohort. It was possible to reduce the increased hospital stay by 50% considering all blood-glucose-disturbed patients (-0.6 days) and by 60% in patients with severe BZ-derailment (-1.4 days). Additionally we were able to decrease the original 2-fold increased mortality risk for P with severe BG derailment compared to normoglycemic P (GES, p <0.01) to the level of normoglycemic P. Supported by: Novo Nordisk, Inc. & For author disclosure information, see page 829. A244 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINS Conclusions: An early and systematic BG intervention in patients with a BG dysregulation leads to a significant reduction of hospital stay and mortality and thus has significant implications for the patient and hospital. 959-P Glycemic Variability as Predictor of Clinical Outcome in InsulinTreated Patients With Type 2 Diabetes FARNOOSH FARROKHI, DAWN SMILEY, FRANCISCO PASQUEL, LIMIN PENG, CHRISTOPHER NEWTON, DARIN E. OLSON, SOL JACOBS, GUILLERMO E. UMPIERREZ, Atlanta, GA Glycemic variability (GV) has been shown to be an independent predictor of mortality in critically ill patients; however, its impact in non-ICU patients with T2D is not known. Accordingly, this post hoc analysis of a prospective, randomized trial of insulin-treated medical and surgical subjects with T2D determined the association of GV with hospital outcomes. A total of 279 patients were randomized to receive a Basal Bolus regimen (n=146, age: 59±11 yr, admission BG: 210±88 mg/dl, A1C: 8.7±2.5%, ±SD) starting at total daily dose (TDD) of 0.5 U/kg, given half as glargine once daily and half as glulisine before meals and to Basal Plus supplements (n=133, age: 59±13 yr, admission BG: 207±83 mg/dl, A1C: 8.3±2.3%) with glargine once daily at a TDD of 0.25 U/kg plus correction doses of glulisine before meals for BG>140 mg/dl. GV was calculated from all BG values using mean delta daily BG, mean standard deviation (SD) and mean amplitude of glycemic excursions (MAGE). Basal plus regimen resulted in similar mean daily and fasting BG, number of hypoglycemic events, hospital length of stay and complications compared to basal bolus regimen (all, p=NS). There were no differences in GV between treatment groups measured by delta change (72.5±36 vs. 69.3±34 mg/dl), SD (38.5±18 vs. 37.1±16 mg/dl) and MAGE (67.5±33 vs. 66.1±38 mg/dl), all p=NS. GV was higher in surgery patients treated with basal bolus (delta change and SD: p=0.02, MAGE: p=0.009), but there were no differences in GV in medical patients. Patients with high GV had higher number of hypoglycemic events compared to those with low GV (p <0.001); however, no association was found between GV and hospital complications (p=NS). In summary, the basal plus regimen resulted in similar mean daily blood glucose, glycemic variability and frequency of hypoglycemia compared to a basal bolus regimen. This randomized controlled trial indicates that basal plus is an alternative to basal bolus insulin regimen in general medicine and surgery patients with T2D. HANNAH SLABU, PETER A. SENIOR, Edmonton, AB, Canada Hypoglycemia is a major barrier to tight glycemic control which reduces microvascular and macrovascular complications in type 1 diabetes. Since long acting insulin analogues reduce the risk of nocturnal hypoglycemia, they may facilitate improved glycemic control. We sought to examine the differences between target and achieved fasting plasma glucose (FPG) levels, HbA1c and hypoglycemia rates for basal insulin analogs vs comparator in clinical trials conducted in type 1 diabetes. A systematic review of 14 trials comparing long acting insulin analogs with a comparator was performed. Only trials reporting achieved FPG and hypoglycemia rates were included. The trials of detemir (n=8) and glargine (n=3) had NPH as comparator, while the trials of degludec (n=2) had glargine as the comparator. One trial compared detemir to glargine. Hypoglycemia rates were normalized to events per patient per year. The target FPG in the trials ranged between 5-7.3 (mean 6.6) mmol/l but was not achieved in any trial.The achieved FPG for analogs was 9.0 (range 7.6-10.7) mmol/l versus 9.8 (8.1-11.4) mmol/l for comparators. The gap between target and achieved FPG was lower for analogs verses comparators (2.3±0.9 v 3.2±1.1mmol/l, p=0.042). Nocturnal hypoglycemia rates were lower for analogs (6.7±3.9 v 9.3±3.9 events/patient/year, p = 0.02) but overall hypoglycemia rates did not differ (44.6±22.0 v 52.5 ± 21.8 events/patient/year, p=0.08). In keeping with the treat to target design, HbA1c levels did not differ (7.8±0.3 v 7.9±0.5%, p=0.3). Long acting insulin analogs were associated with both a smaller gap between target and achieved FPG and lower nocturnal hypoglycemia rates. However, FPG targets were not achieved and few participants achieved HbA1c levels <7% in these clinical trials. Achieving tight glycemic control in type 1 diabetes while avoiding hypoglycemia in routine clinical practice remains challenging despite the advantages of long acting insulin analogs. 960-P Factors Associated With Achieving Improved Glycemic Control: Findings From the Observational A1chieve Study LEE-MING CHUANG, JIANWEN CHEN, LEÓN E. LITWAK, NABIL K. EL NAGGAR, SERDAR GÜLER, Taipei, Taiwan, Zurich, Switzerland, Buenos Aires, Argentina, Jeddah, Saudi Arabia, Ankara, Turkey A1chieve was a 24-week non-interventional study evaluating the safety and clinical effectiveness of insulin analogs in people with type 2 diabetes (n=66726) in routine practice in 28 countries across 4 continents. This post-hoc analysis investigated factors predicting HbA1c change. 44,872 participants were insulin-naïve prior to the study (baseline mean[SD] HbA1c 9.5[1.7] %, age 53.2[11.6] yrs, diabetes duration 6.6[5.4] yrs). 21854 participants were prior insulin users (baseline mean[SD] HbA1c 9.4[1.8] %, age 55.6[12.5] yrs, diabetes duration 10.8[6.8] yrs). Factors significant in the univariate analysis were diabetes duration; oral glucose-lowering drug number; age; gender; hypoglycemia; baseline & pre-study treatment; region; baseline HbA1c, FPG and PPG; dose titration. These factors were retained in the multivariate analysis. The results showed that longer diabetes duration, older age, female gender, prior insulin use, lower FPG, higher post breakfast PPG and greater dose titration were associated with lesser HbA1c reduction, while greater hypoglycemia frequency was associated with greater HbA1c reduction (Table). In summary, this analysis identified and confirmed metabolic factors that contribute to changes in glycemic control. A more appropriate dose titration while avoiding hypoglycemia is strongly recommended. ADA-Funded Research & 962-P Comparison of Algorithms for Initiation of Basal Insulin in Different Age Populations GEORGE E. DAILEY, JASVINDER K. GILL, JOHN STEWART, RONG ZHOU, La Jolla, CA, Bridgewater, NJ, Laval, QC, Canada, Cincinnati, OH A range of titration algorithms are available for initiation/intensification of basal insulin in patients with T2DM; however, safety and efficacy may vary depending on age. This study compared 3 algorithms for insulin initiation over a range of ages. Data were pooled from 7 randomized controlled trials adding insulin glargine to metformin and sulfonylurea at 10 IU starting dose. The algorithms were 1) 1 IU daily when FPG was above target; 2) 2 IU/3 days; 3) treat-totarget, generally increasing 2-8 IU weekly based on 2-day mean FPG. Patients were analyzed by age: < 55, ≥ 55 to < 65, and ≥ 65 yrs. Change from baseline in endpoint variables was analyzed using a mixed model with algorithm as a factor and corresponding baseline measurement as a covariate. See Table for baseline characteristics and clinical outcomes. Despite differing baseline values, all patients achieved similar endpoint A1C. In those aged < 55 or ≥ 55 to < 65 yrs, there were fewer hypoglycemic events in patients titrated by algorithm 2 than 1 or 3; and more patients using algorithm 2 also achieved A1C < 7.0% with no hypoglycemic event < 56 mg/ dL in categories ≥ 55 to < 65 or ≥ 65 yrs. For author disclosure information, see page 829. Guided Audio Tour poster A245 POSTERS 961-P Fasting Glucose Targets are Not Achieved in (Treat to Target) Trials of Basal Insulin Analogues: Systematic Review of Trials in Type 1 Diabetes Clinical Diabetes/ Therapeutics Supported by: Novo Nordisk A/S CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS POSTERS Clinical Diabetes/ Therapeutics This pooled analysis of data from 7 studies suggests that, regardless of age, algorithm 2 leads to similar glycemic control with less hypoglycemia than algorithm 3; the small number of patients ≥ 65 yrs make comparison difficult. The variation in numbers for each subgroup, and the differences at baseline, must be considered when interpreting these results. Table. Baseline and outcomes for algorithms by age category. P<0.05: a: 2 vs 3, b: 1 vs 2, c: 1 vs 3 < 55 yrs ≥ 55 to < 65 yrs ≥ 65 yrs Algorithm 1 Algorithm 2 Algorithm 3 Algorithm 1 Algorithm 2 Algorithm 3 Algorithm 1 Algorithm 2 Algorithm 3 (n=33) (n=31) (n=304) (n=35) (n=35) (n=280) (n=15) (n=16) (n=195) Mean age, yrs 48 50 48 60 59 60 70 72 70 10a 10 12 12 Mean duration 7 6 7 10 8a of diabetes, yrs Mean baseline 8.83 8.60 8.78 8.45b 9.24a, b 8.75a 8.13c 8.44 8.73c A1C, % A1C change –1.63 –1.44 –1.66 –1.29 –2.13 –1.75 –1.31 –1.90 –1.70 from baseline, % Mean endpoint 7.20 7.16 7.11 7.16 7.11 7.00 6.81 6.54 7.03 A1C, % Endpoint weight- 0.55 0.67a 0.48a 0.36b 0.60a, b 0.42a 0.29 0.29 0.33 adjusted insulin dose, IU/kg Confirmed hypo- 12/33 (36) 9/31 (29) 132/304 9/35 (26) 4/35 (11)a 115/280 4/15 (27) 5/16 (31) 71/195 (36) glycemia < 56 (43) (41)a mg/dL, n/N (%) Confirmed hypo- 19/33 (58)b 10/31 (32)a, 184/304 19/35 (54)b 7/35 (20)a, b 169/280 5/15 (33) 6/16 (38) 96/195 (49) b (61)a (60)a glycemia < 70 mg/dL, n/N (%) Confirmed noc- 6/33 (18) 2/31 (6)a 89/304 (29)a 3/35 (9) 1/35 (3)a 64/280 (23)a 2/15 (13) 3/16 (19) 30/195 (15) turnal hypoglycemia < 56 mg/ dL, n/N (%) Endpoint A1C < 10/33 (30) 8/31 (26) 78/304 (26) 13/35 (37) 15/35 (43) 79/280 (28) 5/15 (33) 9/16 (56)a 54/195 (28)a 7.0% and no hypoglycemic events with glucose < 56 mg/dL, n/N (%) Supported by: Novo Nordisk A/S CLINICAL THERAPEUTICS/NEW TECHNOLOGY— INSULIN DELIVERY SYSTEMS Guided Audio Tour: Novel Insulin Delivery Methods (Posters: 964-P to 971-P), see page 15. & 964-P Randomised Controlled Study of 24h Closed-Loop Insulin Delivery in Type 2 Diabetes KAVITA KUMARESWARAN, HOOD THABIT, LALANTHA LEELARATHNA, KAREN CALDWELL, DANIELA ELLERI, JANET M. ALLEN, MARIANNA NODALE, MALGORZATA E. WILINSKA, MARK L. EVANS, ROMAN HOVORKA, Cambridge, United Kingdom Conventional insulin treatment protocols for managing glucose on noncritical care wards often result in inadequate glycemic control. Our aim was to evaluate feasibility and safety of a closed-loop system, linking continuous glucose monitoring (CGM) and continuous subcutaneous insulin infusion (CSII) using a model predictive control algorithm, in subjects with type 2 diabetes, prior to larger inpatient studies over longer duration. Twelve subjects (7M, age 57.2y, BMI 30.5kg/m2) with insulin-naïve type 2 diabetes (HbA1c 8.4%, diabetes duration 7.6y) were admitted to a clinical research facility for two 24h visits (closed-loop and control) in a randomised crossover design. During closed-loop, subjects’ routine diabetes therapy was replaced with CSII. Basal infusion rates were adjusted manually every 15min based on CGM, using the algorithm. Meals were unannounced and no additional insulin was administered for carbohydrates consumed. During control visits, usual diabetes regimen was continued (Metformin 92%, Sulfonylureas 58%, DPP-4 inhibitors 33%). On both visits, subjects consumed matched 50-80g carbohydrate meals and optional 15g carbohydrate snacks at regular intervals and remained largely sedentary, mimicking an inpatient stay on a general ward. Plasma glucose measurements evaluated closed-loop performance. Compared with conventional therapy, 24h of closed-loop insulin delivery increased time spent in target plasma glucose (3.9-8.0mmol/l) from 24% to 40% (p=0.016), and reduced time above 8.0mmol/l from 76% to 60% (p=0.016). The benefit of closed-loop was even more striking overnight with 78% (vs. 35%; p=0.041) time in target glucose range, and only 22% (vs. 65%; p=0.041) time in hyperglycemia. There were no episodes of hypoglycemia and no time spent below 3.9mmol/l during either intervention. Closed-loop insulin delivery in type 2 diabetes provides safe and effective glucose control, and has the potential to provide a more convenient mode of managing hyperglycemia in hospital. Supported by: Sanofi U.S., Inc. 963-P Impact of Insulin Detemir on Weight Change in Relation to Baseline BMI: Observations from the A1chieve Study PHILIP HOME, RACHID MALEK, VINAY PRUSTY, ZAFAR A. LATIF, JIHAD HADDAD, Newcastle upon Tyne, United Kingdom, Sétif, Algeria, Zurich, Switzerland, Dhaka, Bangladesh, Amman, Jordan Weight gain associated with starting insulin therapy is a significant perceived barrier for both people with type 2 diabetes and health-care professionals. Such barriers result in extended exposure to poor glycemic control. A1chieve was a global non-interventional study in non-Western nations evaluating the safety and clinical effectiveness of insulin analogs in people with type 2 diabetes (T2DM) in routine clinical care. The aim of this subanalysis of 12,078 people starting insulin detemir is to examine aspects of weight change after starting the insulin in insulin-naïve people with T2DM, continuing or not continuing oral agents. Participants were 55% male, age 54.0 [11.3] (mean [SD]) yr, BMI 28.2 [5.3] kg/m2, and duration of diabetes 7.6 [5.5] yr, with baseline HbA1c of 9.5 (1.6) % (80 [18] mmol/ mol). From a baseline weight of 76.5 [16.3] kg, change at 24 weeks was -0.3 [4.0] kg. When stratified by quartiles of baseline BMI, greater weight reduction was seen with increasing baseline BMI in the overweight/ obese groups (Table), despite similar decreases in HbA1c and comparable final insulin doses across BMI groups. At both baseline and 24 weeks a higher proportion of people with a higher baseline BMI were on more than two oral glucose-lowering agents (Table). Thus, in the A1chieve study, weight reduction was greater at 24 weeks in people with type 2 diabetes with higher BMI when beginning insulin detemir. Supported by: NIHR; BRC & For author disclosure information, see page 829. A246 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS & 965-P To improve type 1 diabetes management, we have developed a model predictive control (MPC) algorithm for closed-loop (C-L) glucose control. The algorithm is based on a second order stochastic model specified by 3 patient specific parameters: insulin sensitivity factor, insulin action time and basal insulin infusion rate at C-L start. We conducted 2 randomized cross-over studies. Study I compared C-L with open-loop (O-L) control. Study II compared glucose control after C-L initiation in the euglycemic (C-L-Eu) and hyperglycemic (C-L-Hyper) ranges, respectively. Patients were studied from 22:00 to 07:00 on 2 separate nights. The C-L system consisted of a continuous glucose monitor connected via a USB cable to a laptop running the MPC algorithm. Insulin dosing was computed every 15 min but manually administered via an insulin pump. Rescue I.V. glucose was given if plasma glucose declined below 54 mg/dl. Participants (N = 11): Female sex, 45%; Age 41 ± 9 years; BMI 24.9 ± 3.5 kg/ m2; HbA1c 7.2 ± 0.4%; all were insulin pump users. Table 1 shows overnight glucose control. Of the 6 cases of I.V. glucose administration during C-L in study I, 5 were given in 1 study night; however the same patient needed 6 administrations during O-L indicating that his usual basal insulin infusion rate was too high. This first clinical feasibility test of a novel MPC algorithm shows intriguing results. However we learned that algorithm performance was sensitive to sub-optimal empirical estimation of patient specific parameters. The InsuPad device has been developed to enhance insulin absorption by standardized warming of the insulin injection site. Primary objective of this prospective study is the impact of InsuPad use on prandial insulin dose and glycemic control when studied under real world conditions. This interim analysis was performed with 78 patients (23 female, 55 male, 7 type 1 and 71 type 2 patients, age: 61.2±8.8 yrs., HbA1c: 7.2±0.47%, body weight: 104.0±17.8 kg). After a run-in treatment optimization and stabilization period of up to 4 weeks, patients were randomized to continue therapy for three months without (Control) or with InsuPad. During run-in, HbA1c decreased from 7.2±0.5 % to 6.8±0.5 % (p<0.001), and further improved until study end (Control group: 6.3±0.5 % InsuPad: 6.2±0.5 %; both p<0.001 vs. baseline). Number of hypoglycemic events (blood glucose <63 mg/dL) during the observation period was higher in the control group (6.7±8.3/patient) than in the InsuPad group (3.4±4.4/patient, p<0.05). The similar improvements in glycemic control in both groups required an increase in the daily prandial insulin dose from baseline by 12 % (from 59.8±25.9U to 65.9±35.8U, n.s) in the control group (basal insulin dose: 43.9±13.7U vs. 43.6±15.2U, n.s.), but were achieved with significantly lower prandial insulin doses (from 66.2±44.9U to 52.9±35.3U; -20.1%, p<0.001) and a slight increase in the basal insulin dose (from 52.5±35.1U to 55.6±40.2U. p<0.05) with InsuPad. Total daily insulin dose increased in the control group (+5.2%) and decreased with InsuPad (-8.3%, p<0.001). Use of InsuPad for three months resulted in a lower frequency of hypoglycemic events and a reduction in insulin requirements as compared to the Control group under real-world conditions. InsuPad may be useful to achieve a safer and more efficient basal bolus therapy in insulin-treated patients with diabetes. Table 1. Data are mean [min; max]. Glucose values are CGM measurements (mg/dl). Study I Study II Open-Loop Closed-Loop Closed-Loop-EU Closed-Loop-Hyper Glucose at 22:00 / 00:00 97 [45; 169] / 97 [59; 133] / 113 [86 ; 139] / 203 [140 ; 288] / 115 [47; 202] 103 [59; 137] 92 [56 ; 113] 167 [111 ; 254] Mean glucose 22:00-07:00 110 [52; 207] 112 [94; 128] 124 [112 ; 133] 153 [103 ; 198] % time spent within 43.7 [3.4; 86.7] 78.5 [48.5; 98.0] 64.0 [17.6 ; 94.5] 39.5 [0.0 ; 100.0] 70-144 mg/dl % time spent within 52.7 [3.4; 100.0] 90.7 [77.9; 100.0] 82.1 [33.3 ; 100.0] 61.1[5.4 ; 100.0] 70-180 mg/dl % time spent < 70 mg/dl 30.4 [0.0; 96.6] 9.1 [0.0; 22.1] 6.9[0.0 ; 32.7] 6.8[0 ; 12.9] I.V. glucose administrations 9 6 1 0 Supported by: Insuline Medical 966-P Absorption Kinetics of Insulin Following Subcutaneous Bolus Administration With Different Bolus Durations WERNER REGITTNIG, MARTINA URSCHITZ, BARBARA LEHKI, JULIA MADER, MARKUS EHRMANN, HARALD KOJZAR, MARTIN ELLMERER, THOMAS R. PIEBER, ON BEHALF OF THE AP@HOME CONSORTIUM, Graz, Austria To cover meal-related insulin requirements, insulin pumps allow insulin to be delivered at high rates over a short period of time (bolus delivery). The length of this period (bolus duration) usually depends on the chosen bolus size and on the used insulin pump model. This study assessed subcutaneous absorption kinetics of rapid-acting insulin administered with bolus durations commonly employed in commercially available insulin pumps (i.e., 2 and 40 seconds for delivering 1 U). Twenty type 1 diabetic subjects were studied on 2 occasions, separated by at least 7 days, using the euglycemic clamp procedure. Following an overnight fast, subjects were given, in random order, a subcutaneous insulin bolus (15 U of insulin lispro, Eli Lilly) either over 30 seconds using an Animas IR2020 pump (short bolus) or over 10 minutes using a Medtronic Minimed Paradigm 512 pump (long bolus). On each occasion, a new infusion set of the same type was used. Subcutaneous insulin administration with short bolus duration resulted in an earlier onset of insulin action as compared to insulin administration with long bolus duration (21.0 ± 2.5 vs. 34.3 ± 2.7 min; p<0.002). Furthermore, time to reach maximum insulin effect was found to be 27 min earlier when insulin was administered with short bolus duration as compared with administrations using long bolus duration (98 ± 11 min vs. 125 ± 16 min; p<0.005). In addition, the area under the plasma insulin curve from 0 to 60 min was higher for the bolus administration with short bolus duration than the one with long bolus duration (10307 ± 1291 vs. 8192 ± 865 min·pmol/L; p= 0.027). In conclusion, our data suggest that insulin bolus administration with short bolus duration may result in faster insulin absorption and, thus, may provide a better control of meal-related glucose excursions than that obtained with bolus administrations using long bolus durations. Moreover, our findings may have important implications for the future design of the bolus delivery unit of insulin pumps. & ANDREAS LIEBL, BERNHARD GEHR, CLAUS KIEHLING, COSIMA C. RIEGER, CAROLINE PATTE, THOMAS FREI, Bad Heilbrunn, Germany, Bad Toelz, Germany, Mannheim, Germany, Burgdorf, Switzerland Introduction and aim: The new Accu-Chek® DiaPort system is a percutaneous port system for continuous intraperitoneal insulin infusion (CIPII). Compared to previous models, the new DiaPort has a substantially improved design, material, and implantation tools. The aim of this study was to investigate the clinical performance and safety of the new device. Method and Design: 12 adult patients with type 1 diabetes unsuccessfully treated with continuous subcutaneous insulin infusion (CSII), defined as frequent or severe hypoglycemia and/or HbA1c above 8.5%, got the new Accu-Chek DiaPort system implanted and were followed for 12 months (open und uncontrolled). This is the interim analysis after 6 months (1018 patient days). Mean duration of diabetes was 30 years ( ± 12 SD), mean baseline HbA1c 9.0% (± 1.1% SD). Results: Surgical procedure, ingrowth, local or intraperitoneal tolerability, and function of the new Accu-Chek DiaPort system were without major problems. One event of a superficial infection around the DiaPort could be controlled with oral antibiotic and local therapy. One implantation was revised due to non-compliance of the patient. Frequent catheter obstructions occurred when using insulin lispro. After changing to buffered human insulin, and successful exchange of 8 blocked intraperitoneal catheters, no more catheter obstructions occurred. HbA1c improved significantly to 7.57% (± 0.63% SD), p<0.001. Total daily insulin dose dropped from 49 to 45 U Supported by: FP7-ICT-2009-4; 247138 ADA-Funded Research & 968-P Evaluation of the New ACCU-CHEK® DIAPORT, a Port System for Continuous Intraperitoneal Insulin Infusion, in Patients With Type 1 Diabetes: First 6-Month Results For author disclosure information, see page 829. Guided Audio Tour poster A247 POSTERS SIGNE SCHMIDT, DIMITRI BOIROUX, ANNE KATRINE DUUN-HENRIKSEN, JOHN BAGTERP JØRGENSEN, NIELS KJØLSTAD POULSEN, HENRIK MADSEN, OLE SKYGGEBJERG, STEN MADSBAD, KIRSTEN NØRGAARD, Hvidovre, Denmark, Lyngby, Denmark ANDREAS PFÜTZNER, THOMAS FORST, KLAUS FUNKE, NORBERT HERMANNS, MICHAEL DERWAHL, GABRIEL BITTON, RON NAGAR, THOMAS HAAK, Mainz, Germany, Potsdam, Germany, Bad Mergentheim, Germany, Berlin, Germany, Petach Tikva, Israel & 967-P Model-Based Closed-Loop Glucose Control in Type 1 Diabetes Clinical Diabetes/ Therapeutics & Use of the InsuPad Device Results in Improvement of Glycemic Control, Reduced Insulin Requirements and Reduced Hypoglycemia Frequency in a Real-World Setting—Interim Results from the BARMER Study CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS (ns). CGM showed significant improvements in mean glucose and glucose values in target range [%]. 2 severe hypoglycemias in the 12 patients during 6 months were observed (vs. 18 severe hypoglycemia in the previous 12 months). Conclusion: First results indicate good performance of the new Accu-Chek DiaPort system for CIPII. Significant improvements in metabolic control with lower HbA1c and reduced risk of severe hypoglycemia could be observed in patients failing on CSII. insertion device. The Minimed Duo device was tested in 20 adult (9 male) type 1 diabetes patients who were accustomed to using pumps and sensors. Each patient wore one MiniMed Duo and one Enlite sensor (control) for 3 days. The MiniMed Duo was used for glucose signal recording and insulin delivery during 2 inpatient periods of 12 hours (7AM-7PM) on day 1 and day 3 during which subjects took 3 meals with insulin boluses and frequent blood glucose (BG) tests. The remainder of the time was spent in the outpatient setting. At the end of the study, a survey evaluating the experience with the MiniMed Duo device was given. Two-hour postprandial BG values were measured. On day 1 and 3 of device wear (19 subjects, 114 meals), the mean 2-h postprandial BGs were 138±58.5 mg/dL and 137.8±62.1 mg/dL, respectively; the inter-day difference was not significant (P=0.9). Mean time needed for post-prandial glucose values to return to pre-prandial levels was 128±65 min. Clarke Error Grid analysis of all paired points over 3 days demonstrated comparable accuracy of MiniMed Duo (A+B, 96.38%) and Enlite sensors (A+B, 97.07%, P=0.41). Patients were generally satisfied with the overall experience (mean of 6.6±0.74 on a 7-point Likert scale). This feasibility study demonstrated that the new combination insulin delivery-glucose sensing device can be effectively used for insulin delivery and accurate measurement of glucose. Use of the MiniMed Duo may enhance patient compliance and reduce the burden associated with diabetes management. Supported by: Roche Diabetes Care & 969-P Improved Postprandial Glucose after Mixed Meals With Preadministration of Hyaluronidase at Each Infusion Site Change During CSII in T1DM POSTERS Clinical Diabetes/ Therapeutics DOUGLAS B. MUCHMORE, LINDA MORROW, MARCUS HOMPESCH, DANIEL E. VAUGHN, San Diego, CA, Chula Vista, CA Recombinant human hyaluronidase (rHuPH20) is FDA-approved to increase dispersion and absorption of other injected drugs. A 2-way crossover double blind study evaluated its impact on post-breakfast glucose in 22 subjects with T1DM using CSII. Hyaluronidase (1 mL of Hylenex® recombinant) or sham was given by bolus via infusion cannula at each set change. Set changes were every 3 days for 4 cycles for each treatment. After ~2hr of IV insulin to bring FBS to 110 mg/dL, an insulin bolus was given by CSII just prior to a standardized solid breakfast. Hyaluronidase was well tolerated and reduced postprandial glucose excursions up to 29 mg/dL vs. sham. Early hypoglycemia risk (AUC0-3h<70 mg/dL) was similar but late hypoglycemia risk (AUC3-5h<70 mg/dL) was reduced ~42%. Hyaluronidase preadministration at infusion set change reduces PPG and lowers late hypoglycemic risk in T1DM using CSII. & 971-P Exercise-Induced Hypoglycemia in the ASPIRE Study: Lessons Learned SUIYING HUANG, JOHN B. WELSH, SCOTT W. LEE, JOHN SHIN, FRANCINE R. KAUFMAN, ASPIRE STUDY GROUP, Northridge, CA Hyaluronidase Effects on Post Prandial Glucose (PPG) Sham Hylenex Change PPG-60 min (mg/dL) 179 150 -29 mg/dL (p=.003) PPG-120 min (mg/dL) 162 139 -23 mg/dL (p=.022) %Time0-4h 70-140 mg/dL 42% 53% +26% (p=.025) %Time0-4h 60-180 mg/dL 64% 76% +19% (p=.002) AUC0-4h >180 mg/dL (min*mg/dL)) 3529 1922 -46% (p=.042) Early Hypo Risk AUC0-3h<70 mg/dL (min*mg/dL) 110 122 +11% (p=.75) 64 -42% (p=.046) Late Hypo Risk AUC3-5h<70 mg/dL (min*mg/dL) 110 The ASPIRE study showed that exercise-induced hypoglycemia is mitigated by the Threshold Suspend (TS) feature, which automatically suspends insulin delivery when hypoglycemia is detected. We studied the association between the initial glucose response to exercise and the development of hypoglycemia. Fifty subjects with type 1 diabetes were studied on 133 occasions. They began exercise with plasma glucose (YSI) values in the 100-140 mg/dL range, exercised to YSI ≤85 mg/dL, and rested until hypoglycemia (50 ≤ YSI < 70 mg/dL) was reached. YSI rates of change (ROC) were used to stratify the experiments: rising (ROC ≥0.3 mg/dL/min), falling (ROC ≤-0.3 mg/dL/min), or stable. Groups were similar for age, A1C, and BMI. Subjects with initially falling YSI values had fewer total exercise sessions, required less exercise time to reach 85 mg/dL, and required less time to reach 70 mg/dL than did subjects with initially rising or stable YSI values. The ROC in the 30 min prior to exercise relates to the ROC during exercise, and those in the rising group had the highest glucose value prior to exercise. Groups were similar with respect to night-before SG values and the ROC after exercise and before reaching 70 mg/dL (Table). Heterogeneity in the initial glycemic response to exercise may be partially attributable to glucose concentration and glucose ROC immediately before exercise. Critical evaluation of glucose and its ROC prior to exercise might lead to better glycemic control during and after physical activity. Table & Rising (n=20) Falling (n=90) Stable (n=23) P value 970-P Feasibility Assessment of the MiniMed Duo Device: Combined Insulin Delivery and Glucose Sensing PRIOR TO EXERCISE DURING AND AFTER EXERCISE Total Mean ROC Starting YSI Total ROC after Time to number of night-before 30 min glucose exercise exercise and reach target exercise SG (mg/dL) prior to (mg/dL) time (min) before reaching (min) sessions exercise 70 mg/dL 4.8±1.5 152.3±40.0 0.28±0.76 138.6±34.0 103.9±31.7 -0.24±0.47 190.3±77.5 3.1±1.4 147.3±46.2 -0.39±0.60 119.5±18.6 56.5±25.7 -0.24±0.32 131.1±70.7 4.8±1.7 132.3±36.4 -0.23±0.62 114.6±18.3 97.6±39.7 -0.19±0.10 214.7±74.6 <0.0001 0.14 0.016 0.0009 KIRSTEN NØRGAARD, GAYANE VOSKANYAN, SUMONA ADHYA, HENRIK EGESBORG, JULIE THEANDER, PRATIK AGRAWAL, RAJIV SHAH, Hvidovre, Denmark, Northridge, CA <0.0001 0.84 <0.0001 972-P Insulin Requirement Profiles and Characteristics of Type 2 Diabetic Patients With Continuous Subcutaneous Insulin Infusion Therapy A rising number of insulin-requiring patients simultaneously use insulin pumps and continuous glucose sensors, which currently are separately inserted devices. Medtronic developed a new device (MiniMed Duo) combining a sensor and an infusion catheter on the same platform. The device improves on the Combo-set and features a steel needle for insulin infusion and an Enlite sensor platform. It is inserted with a specialized XIUZHEN ZHANG, JINHUA YAN, HAIXIA XU, WEN XU, BIN YAO, YANHUA ZHU, GUOCHAO ZHANG, JIANPING WENG, Guangzhou, China Optimal pump setting is indispensable for successful continuous subcutaneous insulin infusion (CSII) therapy. Currently used protocol, which & For author disclosure information, see page 829. A248 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS determines initial total daily dose (TDD) according to body weight and allocates half of the TDD to total basal dose (TBD), derived from experience in type 1 diabetes. Limited relevant studies were conducted in type 2 diabetes mellitus (T2D). This study aimed to investigate insulin requirement profiles and analyze the related factors of insulin dose in T2D with CSII. A total of 296 patients (male/femal:178/118; age: 52.9±12.7 years; diabetes duration: 4.5±3.2 years; BMI, 24.4±3.1 kg/m2; HbA1c: 10.6±2.3%) were enrolled. The patients were admitted to hospital for 1-2 weeks of CSII therapy and provided with diets of 25-30 kcal/kg ideal body weight per day. The insulin infusion was adjusted according to the fasting and postprandial of three meals capillary blood glucose ([FBG] and [PBG]). After achieving normoglycemia (defined as FBG≤7mmol/L and PPG≤10mmol/L), the insulin requirement profiles were analyzed. TDD was 51.8±17.0 units , TDD/kg was 0.80±0.27 units/kg, TBD was 20.6±7.8units, and the ratio of TBD to TDD(%TBD) was 40.2±9.2 % . When stratified by BMI(normal: <23.9;overweight: 24-27.9;obesity: ≥28), it was found that the obesity group needed more basal infusion than overweight group(%TBD:43±8% versus 39±9.8%, p<0.05). Further multiple regression showed that TDD was positively correlated with waist circumference (r=0.309, p<0.001), and HbA1c(r=0.173, p<0.05). And %TBD was positively associated with waist circumference (r=0.162, p<0.05), and negatively correlated with 2-h postprandial C-peptide (r=-0.135, p<0.05). Neither TDD nor %TBD was associated with weight, lipid profile or disease duration. Basal insulin requirement was~40% of the TDD in Chinese T2D treated with CSII. TDD was associated with parameters indicating glycemic control and central obesity other than with body weight. 974-P Effect of 2-Hr Suspensions of Basal Insulin on Elevating Nighttime Sensor Glucose Concentrations 973-P Pre-Clinical Safety and Efficacy Study of an Overnight Closed-Loop System ANIRBAN ROY, BENYAMIN GROSMAN, NEHA PARIKH, JINO HAN, MIKHAIL LOUTSEIKO, NATALIE KURTZ, GAYANE VOSKANYAN, FRANCINE R. KAUFMAN, JOHN MASTROTOTARO, BARRY KEENAN, Northridge, CA An Overnight Closed Loop (OCL) insulin delivery system (comprising pump, sensor, and control algorithm) should maintain normoglycemia under optimal conditions, while assuring patient safety during sub-optimal circumstances, such as erroneous sensor glucose readings. Medtronic’s Android-based OCL system contains the ePID glucose control algorithm along with several failsafe algorithms that provide enhanced patient safety, particularly during sensor failure modes. In this work, safety and efficacy of the mobile OCL system were examined in 7 diabetic dogs. Closed-loop (CL) mode was started between 10:00 and 11:00 PM and data were collected until 6:00 AM the next day. Four dogs received a correction bolus immediately before the start of CL-mode. Adjustments to insulin delivery were based on sensor glucose values collected at 5-min intervals, and blood glucose (BG) values collected at 30-min intervals were used for retrospective sensor accuracy (MARD) calculations. The average percentage time in-range (70-180 mg/dL) was 95.4%, mean BG ranged from 99 to 153 mg/dL, and the BG concentration never fell below 80 mg/dL (Table). Sensors in dogs fail more often than they do in humans. Despite poor accuracy (high MARD values) of the sensors used in dogs 5 and 7, the sensors were not replaced and the OCL system maintained near-euglycemia at all times. Overall performance of the system in terms of maintaining euglycemia was excellent. The next step is inpatient human testing of the system. 975-P High Sensitivity Occlusion Detection Using Fluid Pressure Monitoring During Basal Insulin Infusion STEVEN KEITH, ELAINE MCVEY, RONALD J. PETTIS, Research Triangle Park, NC We examined pressure-based occlusion detection during insulin lispro basal infusion and correlated these to PK outcomes in a clinical study comparing intradermal (ID) and subcutaneous (SC) infusion routes during two 6h infusions (1 and 2U/h) separated by a 4h washout. The study was an open label, randomized crossover in 20 T1 diabetic males [mean age 38.5, BMI 26.1, HbA1c 7.56]. SC infusion utilized a 28 gauge, 6 mm stainless steel infusion cannula and ID a developmental 34 gauge, 1.5 mm microneedle set. New freshly primed and seated sets were used for each 6h infusion period. A programmable syringe pump delivered microboluses every 3min mimicking a pulsatile insulin pump basal profile. Fluid pressure was continuously monitored using an in-line pressure transducer. Serum insulin lispro concentrations were quantified by radioimmunoassay at predetermined timepoints. Previously reported PK endpoints (onset rate & AUC) were similar between routes, with faster ID insulin offset. Using a proprietary algorithm to process infusion fluid pressures, apparent occlusions were detectable in 5-25% of all set/rate combinations. Occluded infusions exhibited physiologically implausible insulin time-concentration profiles characterized by low initial serum insulin levels at start-up or premature insulin declines before infusion cessation. Occlusions paralleled periods of increasing or elevated fluid pressures that were below typical insulin pump alarm thresholds (<10psi). PK outcomes were subsequently modeled using flow input from the OCL system performance based on YSI values (3 hours after start of CL-mode until end of experiment) Dog # Time In-Range, Mean Blood Glu- Blood Glucose Mean Absolute Relative 70–180 mg/dL (%) cose (mg/dL) Range (mg/dL) Difference (MARD, %) 1 83.3 148 115 – 184 17.0 2 100.0 136 112 – 179 12.1 3 100.0 149 135 – 176 19.1 4 100.0 99 85 – 114 5.3 5 85.1 119 83 – 206 49.4 6 100.0 153 131 – 170 17.0 7 100.0 127 110 – 170 28.5 ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A249 POSTERS Low glucose suspend feature of the Medtronic Veo system automatically interrupts basal insulin for 2-hrs if patients fail to respond to low glucose alarms; a feature aimed at preventing prolonged hypoglycemia, especially at night. We recently reported that such 2-hr suspensions of basal insulin are safe, resulting in minimal increases in β-hydroxybuytrate levels (i.e., from 0.08 to 0.11 mmol/L) even when basal insulin is suspended on random nights over a range of starting glucose values. To examine the efficacy of basal suspensions in increasing nocturnal glucose levels, the 17 T1D subjects (age 24 ± 9 yr, A1c 7.3 ± 0.5) in the study wore a blinded sensor on 71 nights when the basal insulin infusion was pre-programmed to include a 2-hr zero basal rate at random times after 11:30p.m. At the start of the suspend, mean sensor glucose (SG) was 134 ± 62mg/dL and ranged between 40-295 mg/ dL. As shown in the Figure, SG rose by 26 ± 47mg/dL by the end of the 2-hr suspend and by 61 ± 60 mg/dL two hours later. This differed from 31 nonsuspend nights in which baseline SG was 124 ± 70mg/dL and change in SG was not appreciated at 2 and 4hrs (p=ns). On the 11 nights with baseline SG<70 mg/dL (mean 52 ± 10mg/dL), SG rose by 14 ± 35 mg/dL at 2 hrs and SG was >60mg/dL on 91% of nights after 4 hrs. Automatic suspension of basal insulin for 2-hrs is an effective and safe means of limiting the duration of nocturnal hypoglycemia in patients with T1D. Clinical Diabetes/ Therapeutics JENNIFER SHERR, LORI R. CARRIA, KATE WEYMAN, MELINDA ZGORSKI, AMY T. STEFFEN, EILEEN M. TICHY, MILADYS M. PALAU COLLAZO, EDA CENGIZ, CAMILLE MICHAUD, WILLIAM V. TAMBORLANE, STUART WEINZIMER, New Haven, CT CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS pressure algorithm, and were well correlated to experimentally determined PK profiles. This is the first study to correlate basal infusion flow mechanics to insulin PK and demonstrates that transient occlusions may occur that are unlikely to be accompanied by an infusion pressure alarm. Infusion pressure provides highly sensitive occlusion detection and appears related to PK outcomes using infusion model simulations. This novel personalization scheme can maximize clinical benefits of closed-loop BG control by ensuring safe delivery of insulin through appropriate attenuation of control action based on readily available clinical parameters. Clinical trials of this algorithm are underway at the Sansum Diabetes Research Institute. Supported by: JDRF Supported by: JDRF 978-P Cognitive Effects of Intranasal Insulin in Older T2DM Adults 976-P SARA M. MONTI, ANDREW GALICA, PETER NOVAK, MEDHA MUNSHI, AMIR ABDULJALIL, PAULA K. ROBERSON, BRAD MANOR, WILLIAM MILBERG, VERA NOVAK, Boston, MA, Worcester, MA, Columbus, OH, Little Rock, AR Home Overnight Closed-Loop Insulin Delivery Over Three Weeks: WITHDRAWN Randomised Controlled Trial in Adolescents with Type 1 Diabetes Accelerated cognitive decline and altered cerebral vasoreactivity are characteristic of Type 2 diabetes mellitus (T2DM). In non-diabetic adults with mild cognitive impairment, intranasal insulin administration has shown some therapeutic promise. This pilot safety and efficacy study investigates the acute effects of intranasal insulin on regional perfusion and cognition in older T2DM adults through a randomized, double-blind, cross-over, placebocontrol design. 14 T2DM (6 M, 8F, age 62.6+7.8 yrs) and 12 controls (9 M, 3 F, 59.8+9.9 yrs) completed baseline cognitive testing, single-dose administration of 40 IU of intranasal insulin (Novolin, Novo Nordisk) or saline placebo, and postadministration continuous arterial spin labeling (CASL) 3T MRI perfusion studies and neuropsychological tests during a 2-night inpatient stay. The protocol was well tolerated, with no serious adverse events or excess expected mild adverse events. Plasma glucose and vital signs remained stable during 1 hour post-insulin. On insulin - relative to placebo, performance on visuospatial memory tasks improved in both groups (MANOVA, p=0.03). Higher visuospatial delayed recall scores correlated with perfusion increases during baseline CASL in the left insular cortex (p=0.05)(LS model adjusted for group, age, gender, and ΔCO2). Better visuospatial learning scores on insulin correlated with increased vasoreactivity in the insular cortex (p=0.04) (LS model adjusted for age, group, gender). Specific to the T2DM group, total visuospatial memory recall was associated with increased vasodilatation in the lepto MCA territory on insulin (p=0.035)(LS model adjusted for age, group, gender), an effect not shown on placebo or in controls. This study provides preliminary evidence that intranasal insulin may modify perfusion and cognitive function in older T2DM patients. POSTERS Clinical Diabetes/ Therapeutics ROMAN HOVORKA, DANIELA ELLERI, JANET M. ALLEN, MARIANNA NODALE, HOOD THABIT, KAVITA KUMARESWARAN, LALANTHA LEELARATHNA, KAREN CALDWELL, ESTHER O’SULLIVAN, HELEN R. MURPHY, ARTI GULATI, JASDIP MANGAT, CRAIG KOLLMAN, PETER CALHOUN, MALGORZATA E. WILINSKA, CARLO L. ACERINI, DAVID B. DUNGER, Cambridge, United Kingdom, Tampa, FL Free-living studies are needed to assess benefits of closed-loop (CL) insulin delivery to build on encouraging observations taken during controlled clinical centre conditions. We evaluated home use of overnight CL, initially in seven pump-treated adolescents with type 1 diabetes [M 4; age 15.5(2.1) yrs; A1C 7.9%(0.8); BMI 23.3(2.2)kg/m2, duration of diabetes 6.1(3.8)yrs; total daily dose 0.9(0.2)U/kg/day; mean(SD)] who, after training on study devices and passing competency assessment, underwent two 3-week periods of sensor augmented pump therapy (FreeStyle Navigator Transmitter and Combined Controller, Abbott Diabetes Care; Dana R Diabecare pump, Sooil) separated by 1 to 3 week washout. During one 3-week period, randomly assigned, overnight insulin delivery was automatically modulated by adaptive model predictive controller running on an ultraportable laptop. Apart from overnight CL, diabetes management was identical during the two study periods. Overall, CL was operational for at least 4 hrs on 112 (76%) nights; sensor data were available for at least 4 hrs on 107 (73%) non-CL nights (night defined between 11pm and 7am). CL increased the time glucose was in target range between 3.9 and 8.0mmol/L at night [58% (40, 77) vs. 42% (13, 69), CL vs. non-CL; P<0.001, median (IQR)]. Mean overnight glucose was reduced during CL [7.9 (2.1) vs. 8.7 (3.1)mmol/L; P = 0.006, mean(SD)] as was time > 8.0mmol/L [35.5% (17.7, 57.8) vs. 47.1% (17.7, 87.2), P = 0.006]. Time < 3.9mmmol/L was comparable [1.4% (0.3, 4.4) vs. 0.5% (0.0, 9.7), P = 0.24] but the percentage of nights with glucose < 3.5mmol/l for at least 20 min was halved (7% vs. 19%, P = 0.01). No serious adverse event or severe hypoglycaemia was observed. We conclude that overnight closed-loop insulin delivery is feasible in home setting providing benefits similar to those observed under controlled conditions in adolescents with type 1 diabetes. Supported by: NIH/NIDDK (5R21-DK084463-02); NIH/NIA (1R01AG028076-A2) 979-P Insulin Therapy With a 4mm x 32G Pen Needle vs. Larger Needles in Obese Subjects Supported by: JDRF; BRC; NIHR RICHARD M. BERGENSTAL, ELLIE STROCK, DIANA PEREMISLOV, VALENTIN PARVU, MICHAEL GIBNEY, LAURENCE HIRSCH, Minneapolis, MN, Franklin Lakes, NJ 977-P We performed a prospective, randomized, multicenter, crossover noninferiority trial to compare glycemic control in obese subjects with a 4mm x 32G pen needle (PN) vs two larger PNs. Subjects (N=293) with BMI ≥ 30 kg/m2, HbA1c 5.5-9.5% and who injected insulin ≥ 2 mos using only pens entered a 3-week wash-in period, using each PN one week. They were then randomized to either 4mm vs 8mm x 31G (N=127) or 4mm vs 12.7mm x 29G PN (N=147) study arms, with order of PN use controlled. The 1° outcome was HbA1c after each 12-week study period; 2° outcomes included: pain by Visual Analog Scale (VAS), preference, ease of use, ease of insertion, injection anxiety, insulin leakage from skin, and safety. Of 274 subjects randomized, 52% were male; 75% Caucasian, 18% Black; 92% T2DM. Mean age ± SD = 56.7 ± 11.0 yrs; BMI 37.0 ± 6.1, range 29.1-59.9 kg/m2; insulin use duration 6.9 ± 7.7 years; total daily dose 78.4 ± 52.9, range 6-350 units; HbA1c 7.5 ± 0.9%; 226 subjects are included in the 1° A1c analysis. HbA1c with 4mm PNs was 0.08% lower (95% CI -0.21, 0.06) vs 8mm PNs, and 0.10% lower (-0.19, 0.00) vs 12.7mm PNs, both within the ± 0.4% equivalence margin. Median insulin dose changes were -2 to +5 units within study periods. Differences in perceived relative pain by 150 mm VAS were 12.4mm (8.3%) and 30.1mm (20.1%) lower for the 4mm vs the 8mm and 12.7mm PNs, respectively (both p < 0.05). Subjects preferred the 4mm vs the 12.7mm PN (p < 0.05), but NS vs the 8mm PN. The 4mm PN was rated superior for ease of use, ease of injection, and anxiety vs both larger PNs (p<0.05). Of ~ 131,000 injections, subjects reported skin leakage in 4.2%, 4.1%, and 4.3% of injections, respectively, with the 4mm, 8mm, and 12.7mm PNs (NS). Hyperglycemic (BG > 400 mg/dL) and hypoglycemic (BG < 50 mg/ dL) event rates per 1000 patient days (1.9-3.0, and 8.0-9.9, respectively) did not differ between PNs. In conclusion, for obese subjects a 4mm x 32G PN is safe, provides equivalent glycemic control to larger PNs, is less painful, better tolerated, and does not increase insulin leakage. A Novel Model-Based Approach to Safe Personalized PID Controller Design for Artificial Pancreas JOON BOK LEE, EYAL DASSAU, DALE SEBORG, HOWARD C. ZISSER, FRANCIS J. DOYLE III, Santa Barbara, CA Prevention of hypoglycemia caused by excess insulin delivery is paramount in artificial pancreas (AP) design for blood glucose (BG) regulation. We present a personalization scheme based on readily available clinical parameters that is applied to a control-relevant model of insulin-glucose response. This model is incorporated through an internal model control based approach into a proportional-integral-derivative (PID) controller for an AP device. Individual basal levels based on daily insulin profiles are used to optimize closed-loop performance. The controller calculates relative deviation of the subject’s basal profile compared to recommended basal levels calculated from individual clinical parameters and adjusts insulin delivery accordingly. An insulin feedback mechanism ensures safety of the design by preventing excess insulin delivery when given previous deliveries. The controller was evaluated using a 31 hour protocol and challenged with three unannounced meal disturbances of 50g, 40g, and 50g CHO. Insilico simulations of this design on 100 subjects through the FDA accepted Univ. of Virginia/Padova metabolic simulator provided excellent results. 93% of overall simulation time was spent within acceptable clinical range of 70-180mg/dl and 65% in tight glycemic range of 80-140mg/dl. Nocturnal glucose control (10:30PM – 7:00AM) was also exceptionally good, with only one hypoglycemic event (BG <65mg/dl) over 100 subjects, and with 96% of simulation time spent within acceptable clinical range and 78% spent within tight glycemic range. Overall, 95% of all subjects were maintained within the A-B zone of the control variability grid analysis system. & For author disclosure information, see page 829. A250 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS Fig.1. Changes in individual C-peptide-genic index and Insulin sensitivity index. Red dot indicate at baseline and blue dot indicate at 2 year-CSII. 980-P Disposable Insulin Pump to Improve Diabetes Treatment Clinical Diabetes/ Therapeutics Intensive insulin therapy for type 2 diabetes requires the patient’s persistent attention to adhere with injections. Several factors can contribute to missed insulin injections including forgetfulness, a busy work schedule, needle phobia, poor psycho-mental adaptation to the diabetes, and the costs associated with the needles and syringes. We have employed a new insulin delivery system for its ability to overcome these problems, i.e. a relatively inexpensive, disposable insulin pump that administers only short acting insulin. V-GoTM (Valeritas, Inc, Bridgewater, NJ) is a novel insulin delivery system delivering up to 76 units/day. V-GoTM can deliver basal insulin continuously (3 different settings allow basal delivery of 0.83, 1.25 or, 1.66 units/h) for 24 hours. In addition, by clicking a button on the unit, it delivers bolus insulin in 2 unit increments up to 36 units/day. Instead of multiple daily injections, patients deliver insulin by affixing the device to the skin once a day. To date, we have initiated the disposable pump insulin treatment for 7 patients with uncontrolled diabetes, six with type 2, and one with type 1. Prior to V-GoTM -based treatment, despite total daily insulin doses of 137 ± 37 units (mean± SD) injected 3 ± 1 times daily, all had elevated glycohemoglobin (12.0 ±1.0 %, range 8.9% - 14.0%). We reasoned that the V-GoTM insulin delivery system would be more convenient and would promote insulin compliance. After less than 1 hour of education, all 7 patients successfully implemented V-GoTM insulin therapy. All patients have preferred the system’s convenience. For the three individuals for whom follow up data is available at least 35 days (range 35-51 days) after they initiated V-GoTM -based treatment, the average daily blood glucose significantly decreased from 305 ± 60 mg/dL to 177 ± 21 mg/dL (p<0.02) and total daily insulin use decreased from 139 ± 42 units to 59 ± 10 units (p=0.6). Our observations suggest that the V-GoTM insulin delivery system is well accepted by patients and can improve blood glucose control. 982-P A Phase 1, Open-Label, Randomized Dose Proportionality Study of Technosphere® Insulin Inhalation Powder (TI) Doses up to 80 U Administered With the Gen2 Inhaler in Healthy Subjects ROBERT A. BAUGHMAN, NIKHIL AMIN, ELAINE WATKINS, PING-CHANG CHUNG, JAMES P. NEZAMIS, ANDERS H. BOSS, Danbury, CT, Paramus, NJ, Chula Vista, CA 981-P Changes in Pancreatic Beta Cell Function and Insulin Sensitivity in Type 2 Diabetes by Continuous Subcutaneous Insulin Infusion Therapy for 2 Years This study incorporated a 4-way crossover of TI doses (10, 30, 60 and 80 U) with a fifth treatment of sc insulin in subjects maintained via hyperinsulinemic, euglycemic clamp. Clinical experience has that > 95% of patient doses are 60 U TI or less. Thirty-two of the 35 randomized subjects (57.1% male, 42.9% female; mean age 34.3 ± 9.7 years) completed all study treatments and visits. Primary PK parameters (area under the curve [AUC], maximum serum concentration [Cmax]) were C-peptide corrected. Dose proportionality was evaluated by comparing the proportionality slope and associated 90% CIs for each PK parameter against predefined 90% CIs. Relative bioavailability (Frel) was determined by dose-normalized AUC comparisons of the various TI doses to sc insulin. The PD effect was assessed by the GIR required to maintain euglycemia (90 ± 10 mg/dL) for 4 hours post-dosing. The insulin exposure over three hours post-dosing (AUC0-180) of C-peptide corrected data was the primary parameter, and was shown to be doseproportional (slope = 1.00 (90% CI: 0.939, 1.061) from 10 to 80 U. Evaluation of exposure extrapolated to infinity (AUC0-inf) provided a slope = 0.949 (90% CI: 0.880, 1.019). The median bioavailability of TI using the Gen2 inhaler was approximately 24% relative to sc injection of 15 IU insulin. The glucose infusion rate (GIR) during the clamp after TI dosing reached a maximum within 30 - 50 minutes, whereas after sc insulin the GIR peaked at 170 - 234 minutes. The most frequent adverse event was mild, asymptomatic hypoglycemia, reported for 85% of subjects after TI dosing and 12.5% following sc insulin, which was classified as related to study procedure (ie, inability of glucose infusion to respond at a rate to maintain euglycemia post-dosing). Insulin exposure following administration of TI doses from 10 to 80 U via the Gen2 inhaler was shown to be dose-proportional. SOO BONG CHOI, HYUN-JU AN, KYUNG-JIN KIM, EUN-KYOUNG JEON, YUNHEE NOH, Seoul, Republic of Korea Type 2 diabetes is characterized with impaired beta cell function and insulin sensitivity. To see if these abnormalities can be improved in type 2 diabetic patients through long-term continuous insulin infusion (CSII) therapy, we examined changes in serum C-peptide-genic Index and modified Matsuda Index during treatment for 2 years. We discontinued oral antidiabetic drugs and applied CSII therapy to subjects with type 2 diabetes who had failed to control hyperglycaemia (number, 309 with 61 % of male; age, 57.5 ± 11.8 years; duration, 10.1 ± 7.8 years; HbA1c 8.5 ± 2.0 %). After the 2 year-CSII treatment, the mean HbA1c significantly decreased from 8.5 ± 2.0 % to 6.3 ± 0.45 % (p < 0.001)and the mean serum C-peptide level increased from 4.7 ± 2.1 to 6.0 ± 2.24 ng/ml (p < 0.001) and the mean serum C-peptide-gentic Index significantly increased from 0.025 ± 0.001 to 0.035 ± 0.002 (p < 0.0001) at 120 minutes after mixed meal ingestion and modified Matsuda Index did not change after 2 years. The disposition index increased from 0.077± 0.077 to 0.103 ± 0.723 (p < 0.05). The resolution of glucotoxicity and maintenance of euglycaemia through long-term CSII therapy may contribute to the restoration of β-cell function but may have no effect on insulin sensitivity in type 2 diabetic patients. ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A251 POSTERS ABDULKADIR OMER, ROBERTA POWELL, JOHN PAUL LOCK, SAMIR MALKANI, DAVID M. HARLAN, Worcester, MA, Boston, MA CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS A prospective, observational, open-label, multicenter study is being conducted in patients (pts) with type 2 diabetes previously treated with any of the following: oral antidiabetic medications alone (OADs); OADs + incretins; QD or BID intermediate or long-acting insulin (LAI) ± OADs; premixed insulin ± OADs; at least 3 insulin injections a day (MDI) ± OADs. The pts are followed on previous therapy for 4-6 weeks before switching to V-Go. Pts are treated by conventional practice by their physicians. A planned interim analysis is presented. To date, 47 pts had continuous V-Go use for 3 months. Average patient age was 60 yrs with diabetes duration of 13 yrs. Most pts were previously on LAI (34%) or MDI (55%) ± OADs. During the pre-randomization phase A1C was reduced by 0.4% (9.0 to 8.6%) reflecting a clinical study effect. After V-Go initiation A1C further improved and was significantly lower (P<0.05) across the total population and MDI subgroup (table). Five pts reported AE’s probably related to V-Go (primarily skin irritation). Six pts reported hypoglycemia <70 mg/dl without severe events. 983-P How Shall We Predict Patient’s Autonomy for CSII Utilization in Type 2 Diabetes? POSTERS Clinical Diabetes/ Therapeutics YVES REZNIK, AMEL ZENIA, REMY MORELLO, MICHAEL JOUBERT, Caen, France Insulin pump therapy is a valuable tool for insulin intensification in type 2 diabetes patients uncontrolled by multiple daily injections (1,2). Unrecognized cognitive disability may compromise the use of insulin pump. In order to predict the patient’s ability to manage his pump, 39 type 2 on CSII were administered i) a 31 items CSII questionnaire for cognitive and operative capacities, ii) the Montreal Cognitive Assessment (MOCA) questionnaire to detect mild cognitive dysfunction. iii) the HADS and iv) the DTSQ questionnaire. At discharge from 5 d training, patients had complete (n=13) or partial autonomy (n=13) or were dependent from a nurse (n=13). The 3 groups were matched for age (mean 61 ± 8 yr) . Results: on pump, HbA1c dropped from 8.87 to 7.5% at1 yr. DTSQ score showed a high level of satisfaction. HADS showed anxious (33%) and depressive items (18%). At 1 yr, 30 patients had complete autonomy, 7 partial autonomy and 2 were still dependent. Autonomy score correlated with MOCA score (r=0.771, p<0.001). By ROC analysis, MOCA identified autonomous vs dependent patients at cut-off score 26 (AUC : 0.893, sensitivity 72%, specificity 80%). On follow-up, all patients with partial autonomy/dependency had a MOCA score <26, while 74% patients with complete autonomy had a MOCA score >26. Conclusion: type 2 diabetes patients with partial autonomy may progress to complete autonomy, and MOCA score may predict patient’s ability to deal with the pump device. 1- Reznik et al, Diabetes Technol Therap 2010;12:931 2- Berthe et al, Horm Metab Res 2007;39:224 Mean *P<0.05 N Month 0 A1C (%) Month 3 A1C (%) 3 Month A1C change (%) Change in Total Daily Insulin Dose Change in Weight, Month 0 – 3 (lbs) Total Population 47 8.6 (6.5 – 14.4) 7.8 (5.4 – 10.9) -0.9(1.08)* -11.7 U +0.7 LAI +/- OAD Subgroup 16 8.5 (7.1 – 10.9) 7.6 (5.4 – 10.9) -0.9(0.91) -1.6 U -2.7 MDI +/- OAD Subgroup 26 8.7 (6.5 – 14.4) 7.8 (6.1 – 10.9) -0.9(1.2)* -17.5 U +2.7 The interim results suggest improved glycemic control across the population and major subgroups as well as reductions in total daily insulin dose in pts switching to the V-Go. 984-P 986-P Predictors of Nocturnal Hypoglycemia During the Run-In Period of the ASPIRE-2 Study Insulin Pump Therapy Optimization Using Run-to-Run Control With Batch-Varying Gain BRUCE W. BODE, SCOTT W. LEE, FRANCINE R. KAUFMAN, ASPIRE-2 STUDY GROUP, Atlanta, GA, Northridge, CA HUI WANG, YOUQING WANG, Beijing, China Optimizing insulin pump therapy is a significant challenge for physicians, and run-to-run (R2R) control has been recognized a promising candidate. However, in the reported studies, the gain of R2R is always fixed and designed based on experience, so an intelligent method is needed to adjust the gain adaptively. First, using least square method, an ARX glucose-insulin model was identified, which can be used to predict the glucose level. Second, taking square sum of glucose prediction error as a fitness function, particle swarm optimization method was used to calculate the optimal gain for R2R in each day. At last, R2R used the optimal gain to update the insulin pump therapy from day to day. The proposed method was evaluated on the UVa/Padova simulator. In 15 days, all ten virtual adult subjects followed a protocol of 3 meals at 7am, noon, and 6pm of 40g, 70g, and 60g of CHO, respectively. In terms of blood glucose index (BGI), the proposed method was compared with the conventional R2R with fixed gain in the following figure. For all subjects, the proposed method can achieve less BGI values, which means better glycemic control performance. In addition, all subjects’ blood glucose (BG) values can be kept in the safe range (60-180mg/dL) without hyper- or hypo-glycemia from 10 th day. Through learning and optimization, the proposed method can adjust insulin pump therapy automatically. Therefore, it can work as a capable assistant for physicians and it will reduce physicians’ work load significantly. ASPIRE-2 is a randomized controlled trial of the Medtronic Veo insulin pump system with Threshold Suspend (TS) which stops insulin delivery when a specified glucose threshold value is reached. Evaluation of nocturnal hypoglycemia was performed during the run-in phase, prior to randomization. Run-in period was conducted in hypoglycemia-prone subjects with type 1 diabetes, age 16-70, while using the sensor-augmented insulin pump system without the TS feature. Nocturnal hypoglycemic events (defined as SG <65 mg/dL for >20 min between 10pm and 8am without subject intervention) during the last 2 wk of the run-in period were evaluated. As of 12/03/2012, 314 subjects completed the run-in phase: 248 were eligible for randomization, 66 were ineligible, and enrollment is now complete. Hypoglycemic events (N=1313) occurred at a rate of 0.3 per patient-night. Subjects with lower baseline A1C values had more hypoglycemic events than those with higher values (p<0.001). More hypoglycemia was noted among subjects with higher glycemic variability (GV) as measured by coefficient of variation (p< 0.001). Older subjects (>50 yr) and those with longer diabetes durations (>15 yr) had less hypoglycemia (p=0.002 for age, p=0.002 for duration). The effects of age and diabetes duration, when adjusted for GV, were not significant (p=0.96 for age, p=0.63 for duration). Those with higher basal/bolus ratios experienced more hypoglycemic events than those with lower basal/bolus ratios. When adjusted for GV, the basal/bolus ratio was found to be significant as either a continuous variable (p=0.009) or as a categorical variable (≤1 vs. >1, p=0.041). As expected, baseline A1C and GV influenced nocturnal hypoglycemia rates. The finding of higher rates of nocturnal hypoglycemia in younger subjects with shorter diabetes duration appears to result from greater GV in these subjects. Assessing GV may be useful to identify hypoglycemia-prone individuals in clinical practice as well as during recruitment for hypoglycemia studies. 985-P Effectiveness of V-Go® for Patients With Diabetes in a Real-World Setting (SIMPLE) GEORGE GRUNBERGER, BRUCE W. BODE, KENNETH HERSHON, CHERYL ROSENFELD, POUL STRANGE, Bloomfield Hills, MI, Atlanta, GA, New Hyde Park, NY, Denville, NJ, Princeton Junction, NJ The V-Go is a disposable 24-hour device that delivers a continuous basal rate of insulin and on-demand bolus dosing. The purpose of this study is to observe changes in glycemic control, insulin requirements and safety related to its use. Supported by: EFSD; CDS/Eli Lilly and Company; NSFC (61074081) & For author disclosure information, see page 829. A252 Guided Audio Tour poster ADA-Funded Research 989-P Effect of the InsuPad Device on Post Meal Glucose Levels With Delayed Insulin Bolus Injection XIU-HONG LIN, MINGTONG XU, FENG LI, XIAOYUN WANG, LIFANG MAI, YAN LI, LI YAN, Guangzhou, China GABRIEL BITTON, DMITRY FELDMAN, TAL ALON, RON NAGAR, ANDREAS PFÜTZNER, ITAMAR RAZ, Petach Tikvah, Israel, Holon, Israel, Mainz, Germany, Jerusalem, Israel Recent studies show that lipoprotein-associated phospholipase A2 (LP-PLA2) and secretory phospholipase A2 (sPLA2) are closely related to chronic inflammation and atherosclerosis. In this study, we evaluate plasma level of LP-PLA2 and sPLA2 in patients with newly diagnosed T2DM, and investigate the effect of short-term intensive insulin treatment on it. We enrolled 90 patients with newly diagnosed T2DM and 58 People with normal glucose, 30~60 years old, from Oct.2010 to Mar. 2012. Plasma LP-PLA2 and sPLA2 were determined in all subjects with an empty stomach. All patients received continuous subcutaneous insulin infusion with an insulin pump for 10~14 days. IVGTT and OGTT were carried out before and after treatment. As to gender, age, BMI, waist-hip ratio, blood pressure and lipid profile, there were no statistically differences between these two groups (P>0.05). Plasma levels of LP-PLA2 and sPLA2 were significantly higher in diabetics, especially in those with macroangiopathy (P<0.05). After treatment, level of sPLA2 was significantly reduced (P<0.05), while there was no statistically difference of LP-PLA2 (P>0.05). Regarding to β cell function, AUCIns of IVGTT and OGTT, AIR, ΔIns30/ΔG30, MBCI and HOMA-βwere all increased after treatment even when corrected by the influence of insulin resistance (P <0.001). With respect to insulin resistance, HOMA-IR was lower after treatment, and SIM, IAI and QUICKI were higher (P <0.05). Correlation analysis showed that LP-PLA2 and sPLA2 were positively correlated with HOMA-IR in all the subjects (P <0.05). In a multiple linear regression analysis, LP-PLA2 and sPLA2 were independent correlative factors of HOMA-IR (P<0.05). Thus we conclude that plasma levels of LP-PLA2 and sPLA2 are closely related to T2DM, insulin resistance and macroangiopathy of diabetics. And we inferred that intensive insulin therapy might be helpful for the improvement of insulin resistance as well as the protection of diabetic macroangiopathy. The InsuPad device is intended for use by diabetic subjects using rapid acting insulin analogs for meal time bolus injections. The device applies local controlled heat to the skin in the vicinity of the injection site which promotes local blood perfusion and enables faster absorption of the insulin from the injection site. This increased speed of absorption may enable flexibility with the timing of meal time insulin bolus injections. Previous studies have shown improved insulin absorption and better post prandial control when the InsuPad device is used. In this study we tested the effect of the InsuPad on post meal glucose levels when the meal time insulin bolus was injected 30 minutes post meal compared to bolus injection before meal when the InsuPad device was not used. Type 2 diabetic patients on basal bolus insulin therapy were admitted after overnight fast for a meal tolerance test. Subjects consumed standardized liquid meals. The study was conducted twice: with the InsuPad device (0.2U/kg insulin bolus injected 30 min post meal; Test) and without the InsuPad device (0.2U/kg injected before meal; Control). Blood samples for glucose measurements were taken from a venous line during the study. The aim of the study was to show non-inferiority of post meal glucose levels under Test compared to the Control. 15 type 2 diabetic subjects participated in the study (4 females), aged 58.4 ± 6.8 years, with HbA1c of 8.6 ± 1.1 and BMI of 28.2 ± 4.7 kg/m^2. Mean maximum glucose excursion was lower when the InsuPad device was used (129±15 mg/dl vs. 142±15 mg/dl p=0.013). Area under the curve of the glucose excursion during five hours post meal was lower, when the InsuPad device was used (54 ±9 mg/dl/Hr vs. 70±9 mg/dl/Hr p=0.017). Similar results were obtained for glucose excursions at 3 and 4 hours post meal. Using the InsuPad device enabled 30 minutes post meal injection with better post meal glucose control. This added flexibility when using the InsuPad device may have benefit in subjects with unpredictable eating patterns. 988-P Effects of Glargine Insulin Delivery Method (Pen-Device vs. VialSyringes) on Glycemic Control and Patient Preferences STACEY SEGGELKE, MATTHEW HAWKINS, JOANNA GIBBS, ELIZABETH COHLMIA, ELIZABETH COHLMIA, NEDA RASOULI, CECILIA WANG, BORIS DRAZNIN, Aurora, CO 990-P A Feasibility Assessment of PaQ®, a Simple 3-Day Basal/Bolus Insulin Delivery Device, in Patients With Type 2 Diabetes We evaluated the effects of 2 glargine insulin delivery methods (pen vs. vial-syringes) on glycemic control and patient preferences in a randomized, open-label, crossover study. Thirty-one patients discharged from the University of Colorado Hospital were recruited for this study. In the hospital, all patients were treated with basal/bolus regimen. Upon discharge, 21 patients received glargine by pen for 3 months and were then switched to vial-syringes for the next 3 months (Group 1). Group 2 consisted of 10 patients discharged on vial-syringes and converted to pens after 3 months. Hemoglobin A1c (HbA1c) was measured at enrollment, and at 3 and 6 months. A questionnaire assessing patient preference was administered at 3 and 6 months. The groups had similar baseline HbA1c (10.7±2.2% and 11.2±2.5%, Groups 1 and 2, respectively) and similar reduction in HbA1c at 3 months (7.8±1.7% and 7.3±1.4%, Groups 1 and 2, respectively, p< 0.001 vs. baseline). However after crossover, the changes in HbA1c from 3 to 6 months were significantly different between groups. The HbA1c increased to 8.5±2.0% at 6 months in Group 1 after switching to vial-syringes, but remained unchanged (7.1±1.6%) in Group 2 after switching to pens (p<0.01, Group 1 vs. 2). Patient questionnaires after each phase of the trial revealed that when asked “How satisfied would you be to continue with your present form of therapy?”, patients were more satisfied with the pen-device than with vial-syringe (5.8+0.4 vs. 2.7+0.7, pen vs. syringe, p<0.0001, on the scale from 0 to 6, with 6 being “very satisfied” and 0 being “very dissatisfied”). Furthermore, patients found the pen-device more convenient and were more likely to recommend this insulin delivery method to someone else. In conclusion, patients switching to glargine pen achieved lower HbA1c at 6 months of follow-up. Patients in both groups overwhelmingly preferred glargine pens over vial-syringes. JULIA K. MADER, LESLIE C. LILLY, FELIX ABERER, JÖRG PACHATZ, STEFAN KORSATKO, ELLIE STROCK, ROGER MAZZE, PETER DAMSBO, THOMAS R. PIEBER, Graz, Austria, Marlborough, MA, Minneapolis, MN PaQ (CeQur, SA) is a simple patch-on device that provides set basal rates and bolus insulin on demand. This 6-week open-label single-arm study assessed; feasibility of use, method of transition from multiple daily injections (MDI), ability to maintain glycemic control and safety in patients with type 2 diabetes (T2D) treated with PaQ. Twenty patients (age 59 ±5 y, T2D duration 15±7 y, A1C 7.7 ±0.7 %), on a stable MDI regimen were enrolled. The study was comprised of three 2-week periods; baseline (MDI), transition to PaQ, and PaQ therapy. The first PaQ basal rate (dose/ day) selected was ≤ the patient’s basal MDI dose. There was no attempt to treat to target. Eighteen patients completed the study. All were able to assemble and use PaQ. Mean total daily dose (TDD) for all patients at the end of PaQ therapy (57±15 U) was not different from baseline (60.4 ± 19 U). Changes in self-monitored blood glucose values (mean ± SD mg/dL) during PaQ therapy showed a trend toward better glycemic control compared to baseline; pre- and post-breakfast -10.7±28 (P=0.12), -13.0±34 (P =0.13), pre-and post-lunch 8.6±41 (P =0.39), -2.6±36 (P =0.76), pre- and post-dinner -2.7±27 (P =0.67), 10.6±61 (P =0.48) and bedtime -17.9±45 (P =0.12). Blinded continuous glucose monitoring (CGM) data during PaQ therapy revealed a trend towards improved glycemic control, with a mean change in average 24 hour glucose exposure of -190.3 mg/dL (P = 0.18) compared to baseline. The reduction in glucose exposure occurred overnight and during the day. CGM revealed no episodes of severe hypoglycemia. The improved glucose exposure was consistent with a mean change in A1C of -0.3±0.4%. No cannula site infections occurred and PaQ was well tolerated. MDI treated patients with T2D were easily and safely transitioned from MDI to PaQ. Despite similar TDD during MDI and PaQ study periods, there was a trend toward improved glycemic control with PaQ therapy. Future studies will assess longer-term PaQ efficacy and safety. Supported by: Sanofi Supported by: CeQUr Corporation ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A253 POSTERS 987-P The Effect of Intensive Insulin Treatment on Plasma Level of LPPLA2 and sPLA2 in Patients With Newly Diagnosed T2DM Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS 991-P Dynamic Bayesian Network (DBN) Model for Blood Glucose Prediction JUSTIN LEE, ROBERT KIRCHER, DON MATHESON, RICHARD MAUSETH, Redmond, WA The goal of this project is to develop an algorithm that produces a DBN statistical model that accurately predicts future BG values, given glucose and insulin on board (IOB) information up to and including the current time. The current goal is 2 hours. At each time T the algorithm selects the BG value B that maximizes the conditional probability: POSTERS Clinical Diabetes/ Therapeutics p(Bt>=T |B t<T, I t<T ) where B is blood glucose and I is IOB. The preceptron learning algorithm uses a set of patterns or curves that are designed to match medically important aspects in the respective BG and IOB histories. Each pattern is a sequence of integers representing 15-minute sample times. The algorithm also uses a features metric h(t) for each such pattern, which indicates how well, at time t the current BG and AI histories matched the patterns. The output from the training algorithm is a real-valued [1 x J+K] vector lambda L, where each Li (1<=i<=J+K) expresses how important the ith feature is at predicting BG. J and K are the number of BG and IOB feature patterns respectively. The above conditional probability at time t is calculated as edot(L,h(t)). The BG and IOB training datasets come from our 2011 JDRF-funded clinical study. A 5-minute sample time was used. The IOB sequence is calculated from the dosing values using a 3.5-hour insulin action curve. Currently there are 21,000 BG and IOB 1-hour integer-valued feature patterns. Initial results show that during fasting periods the DBN model predicts BG values 75 minutes in the future within 8.8 % of the actual clinical values. Using the same interval, post meal accuracy is within 6.6-15.8%. The training algorithm is independent of subject and will be validated on each of the 2011 clinical subjects. In addition to shutting off insulin when low blood sugars are predicted, the model will be used postprandially to adjust dosing based on the predicted BG values. Refinements and testing are ongoing for the planned incorporation into the version 3.0 controller planned for use in our 2013 NIHfunded clinical studies. Supported by: RG56692 993-P Efficacy of Outpatient Closed-Loop Control (CLC) BORIS P. KOVATCHEV, CLAUDIO COBELLI, ERIC RENARD, HOWARD ZISSER, Charlottesville, VA, Padova, Italy, Montpellier, France, Santa Barbara, CA Following early feasibility tests of a portable artificial pancreas - the Diabetes Assistant (DiAs) based on Android smart phone - we now report first randomized cross-over trials (RCT) of outpatient CLC. Twenty type 1 diabetes patients enrolled at 4 sites in the U.S. and Europe, N=5/site for two 38-hour outpatient sessions: CGM-augmented insulin pump therapy vs. CLC. Each session includes meals at local restaurants and 45-minute light exercise (walk in town). During both sessions the patient operates DiAs via graphical user interface; DiAs receives CGM data from a Dexcom Gen 4 sensor, runs the CLC algorithm, and controls a Tandem t:slim insulin pump. Study personnel monitor the trials remotely via 3G or WiFi and are available on site for assistance. Preliminary results: at this time, one 24-hour pilot and 7 (out of 40) sessions have been completed. Figure 1 illustrates the system action during the pilot: DiAs manipulates basal rate to maintain normoglycemia overnight, administers corrections during the day, suggests pre-meal insulin upon entry of estimated meal carbohydrates, and predicts/ mitigates hypoglycemia. On open- vs. closed-loop control, the average glucose was 120 vs. 130mg/ dl; the time within target (70-180mg/dl) was 85% vs. 84% (90% vs. 96% overnight); there were 3 vs. 2 hypoglycemic episodes. Complete results are expected in April, 2013. First RCT of outpatient CLC show control performance similar to previous inpatient studies and comparable to state-of-the-art sensor-augmented insulin treatment. Supported by: NIH; JDRF 992-P Usability of FlorenceD Closed-Loop System during Overnight Home Use in Adolescents With Type 1 Diabetes JANET M. ALLEN, DANIELA ELLERI, MARIANNA NODALE, ARTI GULATI, HOOD THABIT, LALANTHA LEELARATHNA, KAREN CALDWELL, MALGORZATA E. WILINSKA, ESTHER O’SULLIVAN, HELEN R. MURPHY, CARLO L. ACERINI, DAVID B. DUNGER, TIMOTHY T. WYSOCKI, ROMAN HOVORKA, Cambridge, United Kingdom, Jacksonville, FL Assessing user-friendliness of new technology is important to identify system strengths and limitations. We are undertaking overnight closedloop studies at home in adolescents with type 1 diabetes using a purpose built FlorenceD system consisting of an insulin pump, CGM device and ultra-portable computer running model predictive algorithm. We used an eight point questionnaire to evaluate the usability of the system. Data relate to 8 adolescents (age 15.2±2.1years, HbA1c 8.0±0.7%, duration of diabetes 6.5±3.7years) following 3 weeks of overnight closed-loop at home. Response options for each question ranged from 1=Terrible, 2=Poor, 3=Fair, 4=Good, to 5=Excellent. Results are summarised in Table. Items relating to accuracy and reliability of closed-loop performance scored highest, with fashion issues, battery life and flexibility of functions scoring lowest. Ease of start-up, instructions and technical support were rated as Good or Excellent by majority of subjects (range 62.5-71.4%). Over 87% of subjects rated screen information and alarms as Fair, Good or Excellent. Battery life and ease of replacement was rated poorly by 50% of subjects. Responses to items 2 and 3 show two groups of adolescents (62.5% Excellent/Good, 25% Poor). Further qualitative assessments are warranted to understand these differences. Next generation FlorenceD system will focus on miniaturisation of the system, improving battery life and increasing flexibility of functions. Supported by: JDRF (22-2011-649); NIH (RO1DK085623) & For author disclosure information, see page 829. A254 Guided Audio Tour poster ADA-Funded Research Table 1. Parameter Group A Group B Group C P value n=29 n=15 n=6 (ANOVA) Mean Glucose (mg/dl) 161.73 (147 - 176) 179.83 (146 - 214) 186.91 (135 - 239) P 0.34 (95% CI) (ON CSII) (ON CSII) (OFF CSII) 292.81 (37 - 548) 194.51 (166 - 223) 187.49 (156 - 219) P 0.78 (ON & OFF CSII) (ON & OFF CSII) (ON & OFF CSII) Mean Frequency (%) of Hyperglycemic 26.57 (17 - 36) 34.79 (23 - 47) 42.45 (12 - 73) P 0.27 (>200 mg/dl) events (ON CSII) (ON CSII) (OFF CSII) (95% CI) 28.75 (20 - 37) 40.77 (31 - 51) 41.08 (22 - 60) P 0.12 (ON & OFF CSII) (ON & OFF CSII) (ON & OFF CSII) Mean Frequency (%) of Hypoglycemic 6.87 (4 - 10) 9.71 (0 - 19) 7.7(0 - 15) P 0.73 (<70 mg/dl) events (ON CSII) (ON CSII) (OFF CSII) (95% CI) 6.99 (4 - 10) 5.13 (2 - 8) 7.12 (1 - 13) P 0.66 (ON & OFF CSII) (ON & OFF CSII) (ON & OFF CSII) “ON CSII” = glucose readings only while the patients were on the insulin pump “ON & OFF CSII” = glucose readings while on pump and off the pump (such as for surgical procedures, on insulin infusion) “OFF CSII” = glucose readings off insulin pump LUTZ HEINEMANN, DIETMAR WEBER, MATTHIAS KALTHEUNER, NIKOLAUS SCHEPER, GABRIELE FABER-HEINEMANN, DOROTHEA REICHERT, Düsseldorf, Germany Aim: Surprisingly little data are available about the success of insulin pump therapy under daily life conditions and which problems show up. Methods: 1142 patients treated in 40 practices specialized on diabetes therapy (DSP) answered a two-page patient questionnaire in the period from 01.04.2012 to 30.06.2012. Results: 59% (668) of the patients were female / 41% (471) men, their mean (SD) age was 42±15 years, duration of diabetes 21±13 years, HbA1c 7.2±1.1% and the duration of CSII therapy 7.9±6.1 years. 69% of the patients got the pump in a DSP, 21% in a diabetes clinic, 9% and 4% in a clinic by their GP. 55% used a pump from Medtronic, 41% of one of Roche Diagnostics. 97% of the patients are satisfied with their therapy. 76% of patients use a temporary basal rate, 63% a bolus calculator, and 49% one of the bolus options. More Teflon catheters are used then steel catheters (58% vs. 39%). The wear time of steel catheters was shorter in comparison to that of Teflon catheters (p<0.001). Most catheters are inserted vertically and not with an angle (82% vs. 12%). 91% of the catheters are placed in the abdomen. 39% of the patients use a needle length of 8 mm (6 mm 22%). 45% use a tube length of 60 cm (80 cm 21%). 52% change their catheter every 2 day (3 days 39%). 35% of patients reported differences in blood glucose curve during the wearing time of the catheter. More patients observed increases in glycemia with steel catheters (yes 210 (52%) vs. no 195 (48%)) in comparison to Teflon catheters (313 (59%) vs. 222 (41%); p<0.05). 39% of patients reported catheter problems: bending 19%, leakage 12%, air bubbles 12%, others 33%. Conclusion: This evaluation shows that the reality of pump therapy usage differ in some aspects from the usual recommendations. The type of catheter used has an impact on the metabolic control. Such evaluations also provide hints for improvement of insulin pump therapy. 996-P Poor Numeracy as a Limiting Factor for Successful Treatment of Type 1 Diabetes Mellitus (T1DM) Patients on Continuous Subcutaneous Insulin Infusion (CSII) EVGENYA M. PATRAKEEVA, NATALIA N. ROMANOVA, ALSU G. ZALEVSKAYA, EVGENYA V. SHLYAKHTO, Saint Petersburg, Russian Federation Background: poor numeracy is one of the main factors with great influence on quality of glycemic control in T1DM patients (D. Kerr, 2010 (1)). Aim: to evaluate possibilities to calculate simple mathematical task and it´s relation to quality of glycemic control in T1DM patients on CSII. Subjects and methods: 43 T1DM patients were switched to CSII therapy after traditional structural educational course. After 3 months of CSII patients were divided into 2 groups: group 1-28 patients with frequent (more than 1 square (S) and/or dual wave (DW) boluses per day) and group 2-15 patients with less frequent usage of different types of boluses. HbA1c level investigation and continuous glucose monitoring with standard deviation (SD) measurement and amount of hypoglycemia episodes were calculated. According to idea of importance of numeracy for all patients simple mathematical task was given (ranged as a 3d-class task for elementary school) to all patients. Due to results of task completion the decision about re-education of different boluses lessons was obtained. Results: HbA1c level after 3 months was comparable in both groups (7,4±1,1 vs 7,6±1,2), but difference in SD and amount of non-severe hypoglycemia episodes were statistically significant (p< 0,001). Right answer for mathematical task was given by 86% of group 1 patients and 33% of group 2 patients and there were no any differences in educational level, age or diabetes duration in both groups. In a month, after additional education of group 2 patients (for 4 hours) the amount of S/DW boluses was increased. Conclusion: numeracy assessment is an essential tool for successful education of patients on CSII. 1 - D.Kerr. Poor Numeracy: The Elephant in the Diabetes Technology Room. Journal of Diabetes Science and Technology. 2010; Volume 4, Issue 6:1284-1287 995-P Patient Characteristics and Glycemic Control in Patients on Insulin Pump SUBRAMANIAN KANNAN, ELLEN CALOGERAS, PATRICIA LOCK, MARIA CECILIA LANSANG, ANKITA SATRA, Cleveland, OH When hospitalized patients are allowed to use their insulin pump (CSII), multiple factors affect glycemic control. This study aimed to determine glycemic control and characteristics of CSII-using patients which can affect their glycemic control in the hospital. Starting Jan 2012, our certified diabetes educators standardized their assessment of hospitalized patients on pumps. Data on diabetes history, CSII knowledge assessment and glycemic control, were collected by chart review on adults hospitalized between January to July 2012. Patients were categorized into 3 groups: those who continued on CSII and did not need any education (Group A), those who continued on CSII but needed education (Group B) and those who were considered unsuitable for CSII (Group C) and were treated with insulin injections. There were 50 patients, 57% males, mean age 48 + 13 years, 86% T1DM. Mean DM duration was 26 + 14 years, mean duration of CSII was 8.7 + 6 years, mean A1c was 7.6 + 1.4 %. Mean duration of hospital stay was 5 + 4 days. Mean blood glucose levels (MBG), frequency of hyper/hypoglycemic events among the three groups is shown in Table 1. On univariate analysis, DM duration, knowledge of pump settings, and knowledge of hypoglycemia correction significantly correlated with glucose control, but only the latter was significant on multivariate analysis. (P < 0.001). MBG was not different among the three groups. Knowledge of hypoglycemia correction significantly correlated with MBG during hospitalization. 997-P Glycemic Control, Pain and Leakage With 4mm vs. Larger Pen Needles in Obese Patients Treated With Lantus® or High Insulin Doses: Pre-Specified Subgroup Analyses ELLIE STROCK, RICHARD M. BERGENSTAL, VALENTIN PARVU, LAURENCE HIRSCH, MICHAEL GIBNEY, Minneapolis, MN, Franklin Lakes, NJ Despite controlled studies that show 4mm x 32G pen needles (PNs) provide equivalent glycemic control with less injection pain, without increased skin leakage vs larger PNs (CMRO 2010;26:1531-41; DTT 2012;14:1084-90), there is clinician reluctance to recommend or to use 4mm PNs in patients taking high insulin doses and/or Lantus (glargine, IG). From a study of obese insulin-requiring subjects, we report on pre-defined subgroups taking high insulin doses (≥ one daily injection ≥ 40 units) or IG. This was a multi-center, prospective, open-label, randomized cross-over non-inferiority study with two arms - either 4 vs 8mm x 31G or 4 vs 12.7mm x 29G PNs, with order of PN use controlled. HbA1c was measured after each 12-week period; relative pain by 150 mm Visual Analog Scale and insulin backflow from skin was also evaluated. In 274 subjects randomized, baseline mean ±SD BMI = ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A255 POSTERS 994-P Reality of Insulin Pump Treatment in Germany: Results from a Survey With 1142 Patients Treated in 40 Specialized Practices Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULIN DELIVERY SYSTEMS CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES 37±6.1 kg/m2, HbA1c 7.5±0.9%, total daily dose (TDD) 78.4±52.9, range 6 to 350 units, 92% were T2DM.There were 118 subjects (43%) who took high dose injections (TDD 113.5±54.9, range 40 to 350 units), 174 (64%) used IG (TDD 77.5±53.3, range 7 to 350 units); 107 and 154 subjects, respectively, completed. HbA1c levels with the 4mm PN were -0.09% (95% CL -0.23, 0.05) compared to pooled 8mm and 12.7mm PNs in the high-dose subjects, and -0.08% (-0.16, 0.01) vs the longer PNs in the IG group, both within preset HbA1c equivalence criteria (±0.4%). The 4mm PN was rated significantly less painful than 8mm PNs in high-dose subjects, and vs 12.7mm PNs in the IG subjects, both p < 0.05. Pain ratings trended similarly in the other subgroups but were NS. Reports of leakage for the three PN lengths were infrequent and NS, ranging from 3.4 to 5.1% in high-dose subjects, and 3.7 to 4.7% in IG users. In conclusion, in obese subjects taking high insulin doses and/ or using insulin glargine, the 4mm x 32G PN provides equivalent glycemic control vs longer PNs and does not increase insulin leakage or pain. 999-P Artificial Pancreas With Diurnal Control Law for Safe Overnight Glucose Control at Home: Interim Clinical Results RAVI GONDHALEKAR, EYAL DASSAU, HOWARD C. ZISSER, REBECCA A. HARVEY, FRANCIS J. DOYLE III, Santa Barbara, CA Effective nighttime glucose control is crucial for people with type 1 diabetes mellitus (T1DM). An artificial pancreas that automatically delivers insulin based on continuous glucose monitoring must strive to prevent over-delivery of insulin, especially during sleep, and thus avoid nocturnal hypoglycemia, with minimal user interaction. A novel “diurnal” control strategy for insulin dosing was proposed for this purpose. The proposed strategy is based on periodic zone model predictive control, and was evaluated in clinic and subsequently compared to a time-invariant model predictive control strategy. The diurnal strategy is fully automatic at all times, modulates its action based on the time of day, and is designed to deliver insulin more conservatively at night. It thus strategically alleviates the risk of nocturnal hypoglycemia. A preliminary study of 4 subjects with T1DM for 7-43 years, ages 2856, underwent a 24h study with 2 unannounced meals and 30min exercise. Subjects served as their own control group, based on an identical protocol and the invariant strategy. Here we focus on the overnight (10pm-7am) results. The proposed strategy strategically performed up to 2.5h pump suspensions to circumnavigate impending hypoglycemia, possibly induced by an over-delivery in response to dinner. Specifically, the amount of insulin delivered was 5.2+-1.9 [U], compared with 6.76+-1.8 [U] in the invariant case. Furthermore, the proposed strategy resulted in a 1.26+-0.8 [U] underdelivery from basal, as opposed to a 0.31+-0.3 [U] over-delivery in the invariant case. However, the resulting average (diurnal 129+-19 [mg/dL], invariant 137+-10 [mg/dL]) and breakfast time (7am, diurnal 121+-22 [mg/ dL], invariant 117+-21 [mg/dL]) glucose values demonstrated no clinically significant differences. The proposed diurnal control strategy is a significant step towards safe and continuous operation of artificial pancreases in outpatient environments. Supported by: Becton Dickinson CHAOFENG YAN, YOUQING WANG, Beijing, China POSTERS Clinical Diabetes/ Therapeutics 998-P A Novel Physician-Friendly Closed-Loop Control Algorithm for Artificial Pancreas (AP) The existing AP control algorithms are generally too complicated for physician to understand and also those algorithms do not use clinical parameters sufficiently. By using standard clinical parameters, a model-free algorithm was proposed in this study, which is very easy to understand for physicians. Proportional controller was used to design the current insulin infusion rate; i.e., the infusion rate is proportional to the difference between the glucose prediction and setpoint, where the proportional gain is related with subject-specific correction factor (CF). Then, linear insulin-on-board (IOB) curve was employed to eliminate the influence of insulin cumulation. Because CF and IOB are involved, the proposed method is termed as Clinical Parameter based Proportional Controller (CPPC). CPPC was evaluated in silico using the UVa/Padova simulator. Ten virtual adult subjects followed a 24-h protocol of 3 meals at 7am, noon, and 6pm of 40g, 70g, and 60g of carbohydrates, respectively. As a reference, CPPC was compared with the standard model predictive control (MPC). As shown in the following table, CPPC can evidently increase the percentage of time when BG within the safe range and decrease the mean value and standard derivation of BG. CPPC is effective in achieving normoglycaemia and has excellent robustness to unannounced meals. The simplicity of the proposed method makes it physician-friendly, so it has great potential for extensive use. Supported by: NIH CLINICAL THERAPEUTICS/NEW TECHNOLOGY— NON-INSULIN INJECTABLES Guided Audio Tour: Clinical Use of GLP-1 Agonists (Posters: 1000-P to 1007-P), see page 17. & 1000-P Effects of the GLP-1 Receptor Agonist on Fatty Liver in Patients With Type 2 Diabetes JUN OGINO, CHIHIRO YONEDA, KANAKO TAJIMA, YASUHIKO ICHIMURA, SAYAKA FUKUSHIMA, TAKENORI HARUKI, YUKIE SAKUMA, RIE IWAI, TAKAYOSHI NISHINO, YOSHIFUMI SUZUKI, NAOTAKE HASHIMOTO, Yachiyo, Chiba, Japan, Asahi, Japan Glucagon-like peptide-1 (GLP-1) has extra-pancreatic effects on a variety of tissues in addition to the pancreatic actions. We evaluated the effects of the GLP-1 receptor agonist on fatty liver in patients with type 2 diabetes (T2D). Twenty two T2D patients (11 males, 11 females; mean age 56.9±9.4 years) with fatty liver who had no custom of alcohol abuse, viral and autoimmune hepatitis were examined in clinical characteristics, metabolic parameters, insulin secretion, biochemical markers for hepatic inflammation and fibrosis. Hepatic fat content was measured by 1H-magnetic resonance spectroscopy (1H-MRS) before and after GLP-1 receptor agonist liraglutide treatment. Baseline mean body mass index (BMI) was 31.5±4.8, duration of T2D; 9.9±6.4 years, HbA1c; 8.6±1.4%, postprandial plasma glucose; 197±77mg/dl, postprandial C-peptide; 3.2±2.5ng/ml. BMI, HbA1c, CPI (CPR index), AST and ALT levels were significantly improved (p<0.05) at 3 months after liraglutide treatment. Serum ferritin and cytokeratin-18 levels had a tendency to decrease. Hepatic fat concentration measured by 1H-MRS decreased from 34.1% to 18.2% (p<0.05) at 6 months after liraglutide treatment. Patients who improved intrahepatic fat accumulation had a tendency to have shorter T2D duration, more improvement of BMI, ALT, and YKL-40 than those who didn’t. Patients who improved HbA1c after liraglutide had a tendency to have shorter T2D duration, higher baseline CPR, lower baseline YKL-40, more improvement of BMI, ΔALT, and hepatic fat concentration. Supported by: EFSD; CDS/Eli Lilly and Company; NSFC (61074081) & For author disclosure information, see page 829. A256 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES In conclusion, metabolic parameters, liver function, biomarkers for hepatic inflammation and fibrosis, and intrahepatic fat accumulation improved after liraglutide treatment. Although it is unclear whether these results were caused by the direct effect on hepatocytes or indirect effect through weight reduction by liraglutide, GLP-1 receptor agonist may potentially ameliorate fatty liver. & 1001-P Effect of a Dual Agonist of Glucagon-Like Peptide-1/Glucose-Dependent Insulinotropic Peptide (MAR701) on Insulin Secretion Rate and Gastric Emptying in Healthy Volunteers & Clinical Diabetes/ Therapeutics MAR701 is a novel peptide analog with activity on both glucagon-like peptide-1 and glucose-dependent insulinotropic peptide receptors. The effects of MAR701 on insulin secretion rate (ISR) and gastric emptying (GE) were assessed in this phase I, two part, placebo- and active-control, sequential, single blind study in healthy volunteers. Part 1: After a >10 h overnight fast subjects (Cohort 1; n=6, all male, mean age 35.0 years, mean BMI 26.9 kg/m2) received 2 SC injections of placebo (-120 and 0 min) in the abdomen on Day 1 followed by 2 SC injections of Byetta® 5 µg (10 µg total at -120 and 0 min) in the abdomen on Day 2. Part 2: Subjects (n=18, 15 male, mean age 35.9 years, mean BMI 26.9 kg/ m2) received placebo on Day 1 followed by MAR701 8 mg on Day 2 (Cohort 2, n=6), placebo on Day 1 followed by MAR701 4 mg on Day 2 (Cohort 3, n=6), and placebo on Day 1 followed by MAR701 16 mg on Day 2 (Cohort 4, n=6). ISR was assessed using a 2.5 hr graded glucose infusion (GGI) at 2, 4, 6, 8, and 12 mg/kg/min (30 min per step of 20% dextrose, IV) with sampling for glucose, insulin and C-peptide. GE was assessed by measuring absorption of 1000 mg oral acetaminophen elixir over 240 min.. A dose-dependent increase in ISR was observed in response to Byetta at each GGI rate compared with placebo (p<0.001). MAR701 at all doses significantly increased ISR in a dose-dependent manner during each GGI step (p<0.0001). In contrast to Byetta, treatment with MAR701 did not significantly delay GE at any dose. Treatment with MAR701 was generally safe and well tolerated. The most common adverse events associated with treatment with MAR701 were hypoglycemia (occurring after discontinuation of the GGI) and mild nausea that was not dose-dependent. In conclusion, MAR701 significantly increased ISR at doses that had either no or minimal effects on GE. Supported by: Sanofi Effect of GLP1R/GCGR Dual Agonist in Monkeys Peptide-based analogs of the gut-derived incretin hormone glucagon-like peptide 1 (GLP-1) stimulate insulin secretion in a glucose-dependent manner. Such therapy improves the management of T2D and is often accompanied by modest weight loss. This has promoted the search for a safe and effective approach to enhance such efficacy. We have previously reported that balanced co-agonism at the GLP-1 receptor (GLP1R) and glucagon receptor (GCGR) can improve glucose metabolism and exhibit superior weight loss when compared to selective GLP1R agonism in diabetic-obese rodents. We now report the translation of these observations to obese, non-diabetic rhesus monkeys, which were treated with a protease-resistant GLP1R/ GCGR dual agonist. Daily administration of the dual agonist at a nearly sevenfold lower dose (3µg/kg s.c.) to commercially-sourced GLP1R selective agonist (Victoza®) (20µg/kg s.c.) caused superior weight loss. The impact on glycemic control and the potential diabetogenic risk of GCGR activation was assessed in diabetic rhesus monkeys. Victoza® (10 µg/kg s.c.) was again used as a comparative benchmark for selective GLP1R agonism. All monkeys after appropriate wash-out were also treated with the dual agonist at a tenfold lower dose (1 µg/kg s.c.). Sizable reductions in fasting glucose were achieved through one week of treatment. At end of study improved glucose tolerance was observed in the absence of changes in body weight and food intake. These results demonstrate successful translation of the superior pharmacology of GLP1R/GCGR dual agonists from rodents to non-human primates, deepening our belief that this approach constitutes a promising new mechanism in the treatment of T2DM. 1002-P THOMAS KISSNER, MARTINA DORAU, AUREL PERREN, ANNIKA BLANK, MARTIN HEINRICHS, GABRIELE DIETERT, Frankfurt, Germany, Bern, Switzerland In C cells isolated from human thyroid gland tissue, only marginal GLP-1R expression has been observed. This study assessed the expression of GLP1R in human thyroid tissue with proliferative C-cell findings compared to tissue without C-cell pathology. 49 tissue samples were examined from 3 studies; human thyroid tissue without C-cell pathology (n=19; Group A), C-cell carcinoma (sporadic; n=10; Group B), C-cell carcinoma (multiple endocrine neoplasia; n=10; Group C), & non-neoplastic & neoplastic C-cell hyperplasia (Groups D1 [n=5] & D2 [n=5]). C- & follicular cells were examined separately for GLP-1R expression, with calcitonin expression levels determined to ensure proper separation of cell fractions. Calcitonin expression levels in the follicular cells were low or undetected in Groups A (cycle time [CT]=32.1) & D (CT=28.4) & higher in Groups B (CT=20.1) & C (CT=23.9). In the C-cell fractions the levels were consistently higher with CTs of 21.5-26.6, 23.7, 25.0, 29.2 & 27.0 in Groups A, B, C, D1 & D2, respectively. In both C- & follicular cell samples GLP-1R expression levels were low or undetected in all groups. Calcitonin RNA was expressed in the C cell & to a much lower extent the follicular cell samples demonstrating that they had been successfully separated. In conclusion, no difference in GLP-1R expression was observed between thyroid tissue samples with or without proliferative C-cell findings, neither in follicular nor C-cell fractions. & 1003-P JULIE LAO, BARBARA C. HANSEN, RICHARD DIMARCHI, ALESSANDRO POCAI, Rahway, NJ, Tampa, FL, Bloomington, IN Comparison of GLP-1R Expression in Human C-Cell Thyroid Tumor Tissue and Human Thyroid Tissue Without C-Cell Pathology ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A257 POSTERS MARCUS HOMPESCH, LINDA MORROW, KHIN WIN, LOUIS VIGNATI, JONATHAN HAUPTMAN, Chula Vista, CA, Carmel, IN, Nutley, NJ CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES & Liraglutide 0.6 Estimated Liraglutide Estimated Liraglutide Estimated treatmg/ treatment dif- 1.2 mg/ treatment dif- 1.8 mg/ ment difference placebo ference (ETD) placebo ference (ETD) placebo (ETD) or Ratio or Ratio (R) or Ratio (R) (R) [95% CI] [95% CI] [95% CI] p-value p-value p-value R = 0.91 28.8/37.2 R = 0.77 28.4/37.5 R = 0.76 Glucagon (pg/mL) at 34.7/38.1 [0.66 ; 1.25] [0.56 ; 1.08] [0.55 ; 1.05] nadira (primary endpoint) P = 0.555 p = 0.126 p = 0.092 Change in mean glu- 3.9/4.1 ETD = -0.2 3.6/2.5 ETD = 1.1 4.3/6.0 ETD = -1.7 cagon (pg/mL) from [-4.0 ; 3.6] [-2.8 ; 5.1] [-5.7 ; 2.2] PG 5.5 mmol/L to p = 0.910 p = 0.569 p = 0.379 nadir (post-hoc) Adrenaline (pg/mL) 133.8/139.9 ETD = 0.96 100.2/61.4 ETD = 1.63 127.1/122.1 ETD = 1.04 [0.54 ; 1.70] [0.90 ; 2.96] [0.58 ; 1.87] at nadira p = 0.875 p = 0.104 p = 0.891 Noradrenaline (pg/ 151.3/141.9 ETD = 1.07 148.9/100.5 ETD = 1.48 133.3/136.7 ETD = 0.97 [0.67 ; 1.71] [0.91 ; 2.42] [0.60 ; 1.58] mL) at nadira p = 0.784 p = 0.112 p = 0.916 AUCGIR (mg/kg) at 42.6/58.5 R = 0.73 57.8/69.4 R = 0.83 44.0/68.3 R = 0.64 nadira [0.55 ; 0.97] [0.62 ; 1.12] [0.48 ; 0.86] p = 0.029 p = 0.213 p = 0.004 AUCGIR, (mg/kg) dur- 93.1/103.8 R = 0.90 108.3/153.8 R = 0.70 75.4/98.2 R = 0.77 ing recoveryb [0.67 ; 1.19] [0.52 ; 0.95] [0.57 ; 1.04] p = 0.442 p = 0.022 p = 0.087 Note: Data are presented as estimated means. a Nadir: PG = 2.5 mmol/L. b Recovery = from nadir to PG 4.0 mmol/L Abbreviations: AUC = area under the curve; GIR = glucose infusion rate; PG = plasma glucose 1004-P Safety and Efficacy of Dulaglutide vs. Sitagliptin after 104 Weeks in Type 2 Diabetes (AWARD-5) POSTERS Clinical Diabetes/ Therapeutics RUTH S. WEINSTOCK, GUILLERMO E. UMPIERREZ, BRUNO GUERCI, MICHAEL A. NAUCK, KAREN L. BOLEYN, ZACHARY SKRIVANEK, ZVONKO MILICEVIC, Syracuse, NY, Atlanta, GA, Nancy, France, Bad Lauterberg, Germany, Indianapolis, IN, Vienna, Austria This Phase 3, adaptive, double-blind, parallel arm trial compared dulaglutide (DU), a once-weekly, long-acting GLP-1 receptor agonist, with sitagliptin (sita) and placebo (PL) in metformin-treated type 2 diabetes. Primary endpoint was 52 weeks (wks); final endpoint was 104 wks. Patients (N=1098; mean baseline age 54; A1C 8.1%; weight 86.4 kg; diabetes duration 7 years) were randomized to DU 1.5 mg or 0.75 mg, sita 100 mg, or PL (to 26 wks). Similar proportions of DU and sita patients completed 104 wks. Adverse events (AE) were the most common reason for discontinuation. Both DU doses reported higher incidence of treatment-emergent AEs vs sita due to gastrointestinal (GI) events (Table 1). Incidence of GI AEs peaked during the first 12 wks, with no difference after 26 wks. Both DU doses were superior vs sita for change in A1C (Table 2). The AE profile and superior A1C vs sita after 104 wks indicate an acceptable benefit/risk profile of DU over a longer time. Table 1. Summary of Select Gastrointestinal TEAEs Over 104 Weeks TEAE DU 1.5 mg DU 0.75 mg Sita 100 mg (N=304) (N=302) (N=315) Nausea, n (%) 53 (17.4) 44 (14.6) 21 (6.7) Vomiting, n (%) 41 (13.5) 25 (8.3) 11 (3.5) Diarrhea, n (%) 49 (16.1) 36 (11.9) 18 (5.7) Constipation, n (%) 14 (4.6) 16 (5.3) 4 (1.3) Abdominal distension, n (%) 13 (4.3) 15 (5.0) 3 (1.0) Supported by: Novo Nordisk A/S & Table 2. Summary of Efficacy Measures Over 104 Weeks Final Endpoint DU 1.5 mg DU 0.75 mg Sita 100 mg (104 Weeks, ITT, LOCF) (N=304) (N=302) (N=315) A1C change (%), LS Mean (SE) -1.00±0.06†† -0.71±0.07†† -0.32±0.06 % of patients with A1C <7% 54.3# 44.8# 31.1 # Weight change (kg), LS Mean (SE) -2.88±0.25 -2.39±0.26## -1.75±0.25 ††multiplicity adjusted (based on tree gatekeeping) 1-sided p<0.001 for superiority vs sita for A1C change only. #2-sided p<0.001 vs sita. ##2-sided p=0.054 vs sita. PARESH DANDONA, HUSAM GHANIM, KELLY GREEN, SANAA ABUAYSHEH, SANDEEP DHINDSA, AJAY CHAUDHURI, ANTOINE MAKDISSI, MANAV BATRA, REEMA PATEL, Buffalo, NY We have recently demonstrated that exenatide exerts a rapid and a potent anti-inflammatory effect including the suppression of IL-1β expression in peripheral blood mononuclear cells (MNC). Suppression of IL-1β secretion or activity has significant implications for insulin sensitivity and secretion and on glycemic control since it modulates the synthesis of SOCS-3 which interferes with insulin signaling and it mediates the inflammatory damage of the β-cell. We have now hypothesized that exenatide induces an increase in plasma concentrations of IL-1RA which is known to improve glycemia and β-cell function in type 2 diabetics. Twenty four obese type 2 diabetics were prospectively randomized to be injected subcutaneously with either exenatide 10 µg twice daily (n=12, mean age: 56 ±3 years; mean HbA1c:8.6±0.4%) or placebo twice daily (n=12, mean age: 54±4 years; mean HbA1c:8.5±0.3%) for 12 weeks. HbA1c fell by 0.5% while there was no change in body weight. IL1β expression was suppressed by 22±10% in MNC, plasma concentration of IL-18 fell by 19% (from 546±58 to 436±46 pg/ml) while IL-1 RA concentrations increased significantly by 61±18% (from 318±53 to 456±88 pg/ml; p<0.05 for all). Exenatide also suppressed the mRNA expression in MNC of SOCS-3 and PTP-1B, two proteins that interfere with insulin signaling, by 31±10% and 18±5%, respectively. These modulators of inflammation and insulin signaling did not change in the placebo group. We conclude that exenatide induces an increase in IL-1RA concentrations in addition to suppressing IL-1β expression and plasma concentrations of IL-18. The combination of these effects would potentially be of benefit to both the integrity and function of the β-cell and insulin resistance. Supported by: Eli Lilly and Company & 1006-P Exenatide Increases Interleukin-1 Receptor Antagonist (IL1-RA) Concentration 1005-P Effects of Liraglutide as Adjunct to Insulin on Counter-Regulatory Hormone Responses to Hypoglycemia in Type 1 Diabetes: A Randomized, Double-Blind, Crossover Trial THOMAS PIEBER, SIGRID DELLER, MARTINA BRUNNER, LENE JENSEN, ERIK CHRISTIANSEN, FUMIAKI KIYOMI, SIMON HELLER, Graz, Austria, Søborg, Denmark, Tokyo, Japan, Sheffield, United Kingdom This trial investigated the counter-regulatory hormone (CRH) response to hypoglycemia, an important protective mechanism in type 1 diabetes (T1D), with liraglutide as adjunct to insulin. Adults with T1D (n=45) were allocated to 1 of 3 dose groups of liraglutide/ placebo for 4 weeks in a crossover design. At the end of each treatment period, a stepwise hyperinsulinemic, hypoglycemic clamp assessed glucagon, other CRHs, and vital signs at PG levels 5.5, 3.5, 2.5 mmol/L (nadir) and 4.0 mmol/L (recovery). Glucose infusion rate (GIR) was also measured. Baseline characteristics: age 34.5±11.2 yrs, BMI 23.9±2.4 kg/m2, A1C 7.6±0.8%, T1D duration 16.6±9.4 yrs [mean±SD]. At nadir, no significant differences in glucagon and no systematic differences in catecholamines (Table) or other CRHs were seen between treatments (not shown). However, a trend toward lower glucagon at increasing liraglutide dose was seen. No counter-regulatory differences were seen in change in glucagon or other CRH from PG 5.5 mmol/L to nadir. Pulse was higher for liraglutide at nadir, but with similar change from 5.5 mmol/L to nadir across groups (not shown). GIR for liraglutide indicated a tendency for less glucose needed at recovery and other PG levels. To conclude, no safety concerns related to CRH responses, including recovery from hypoglycemia, were raised with liraglutide as adjunct to insulin in T1D. & 1007-P Liraglutide as Adjunct to Insulin in Type 1 Diabetes: Effects on Glycemic Control and Safety in a Randomized, Double-Blind, Crossover Trial SIMON R. HELLER, STEFAN KORSATKO, JAMALA GURBAN, LENE JENSEN, ERIK CHRISTIANSEN, FUMIAKI KIYOMI, THOMAS R. PIEBER, Sheffield, United Kingdom, Graz, Austria, Søborg, Denmark, Tokyo, Japan Treatment with a GLP-1 analog as adjunct to insulin may improve glycemic control in type 1 diabetes (T1D). In a trial investigating effects of liraglutide as adjunct to insulin on counter-regulatory responses to hypoglycemia, we also assessed glycemic control and safety. & For author disclosure information, see page 829. A258 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES Adults with T1D (n=45) were allocated to 1 of 3 dose groups of liraglutide/ placebo for 4 weeks in a crossover design. Liraglutide was started at 0.6 mg/day with weekly 0.6 mg increases until target dose for each arm was reached. Baseline characteristics: age 34.5±11.2 years, BMI 23.9±2.4 kg/m2, A1C 7.6±0.8%, T1D duration 16.6±9.4 years [mean±SD]. A significant decrease in insulin dose was seen with 1.2 and 1.8 mg vs. placebo. No change in A1C was seen. Frequency of hypoglycemic events (none severe) was unaffected. As expected, gastrointestinal (GI) adverse events (AEs) were more frequent with liraglutide; none were severe. 1 of 3 AE withdrawals was treatmentrelated. Liraglutide induced up to 3.3 kg weight loss, not considered a safety issue. To conclude, the decrease in insulin dose indicates that liraglutide has a pharmacologic effect in T1D that may contribute to glycemic control. Besides the known GI side effects, no liraglutide-related safety or tolerability issues were identified in this short-term trial in T1D. The observed weight loss is of potential clinical benefit and worth investigating in larger and longer term trials. these functional data, relative GLP-1R mRNA expression levels were ~150fold higher in the rat C-cell line 6-23 than in human TT cells. In summary, signaling through the GLP-1R by all 4 specific agonists was strong only in the rat C-cell line and weak in the human C-cell line, while there was strong GIPR signaling by native GIP in both species. Activity at the GLP-1R correlated with mRNA expression levels in both species. In conclusion, there are significant species differences in GLP-1R expression and responsiveness between rats and humans. This suggests that GLP1R-mediated C-cell proliferation seen in rodents, after long-term exposure to high concentrations of GLP-1R agonists, might be a rodent-specific phenomenon and does not seem to be relevant to humans. Supported by: Sanofi & 1009-P Liraglutide Estimated treat- Liraglutide Estimated treat- Liraglutide Estimated treat0.6 mg/ ment difference 1.2 mg / ment difference 1.8 mg / ment difference placebo (ETD) placebo (ETD) placebo (ETD) [95% CI] [95% CI] [95% CI] p-value p-value p-value Changea in daily insulin -2.0 / 2.7 ETD = -4.7 -10.4 / 0.2 ETD = -10.6 -11.7 / -2.2 ETD = -9.5 dose (U) [-10.5 ; 1.0] [-16.5 ; -4.6] [-15.3 ; -3.6] p = 0.104 p < 0.001 p = 0.002 -0.10 / -0.01 ETD = -0.09 -0.12 / -0.12 ETD = 0.01 -0.04 / -0.02 ETD = -0.02 Changea in A1C (%) [-0.36 ; 0.18] [-0.27 ; 0.28] [-0.29 ; 0.25] p = 0.504 p = 0.971 p = 0.891 Hypoglycemicb events 14 (93.3) 289 / N/A 13 (92.9) 263 / N/A 14 (100) 316 / N/A N (%) E 14 (100) 313 13 (100) 326 15 (100) 295 Changea in pulse (beats/ 5.7 / 4.8 ETD = 0.9 9.1/ 4.7 ETD = 4.3 4.9 / -0.4 ETD = 5.3 min) [-5.9 ; 7.6] [-2.7 ; 11.3] [-1.6 ; 12.1] p = 0.794 p = 0.218 p = 0.128 GI AEs 10 (66.7) 12 / N/A 12 (85.7) 24 / N/A 13 (92.9) 31 / N/A N (%) E 2 (14.3) 2 1 (7.7) 2 3 (20.0) 5 Changea in BW (kg) -1.2 / 0.8 ETD = -2.0 -3.3 / 0.4 ETD = -3.7 -2.9 / 0.4 ETD = -3.3 [-3.1 ; -0.9] [-4.9 ; -2.6] [-4.4 ; -2.2] p < 0.001 p < 0.001 p < 0.001 Note: Unless otherwise stated, data are presented as estimated means. a = change from baseline to end of treatment. b = assessed according to ADA standard. Abbreviations: AE = adverse event; BW = body weight; CI = confidence interval; E = events; GI = gastrointestinal; N = number; N/A = not applicable; U = units. Critical illness-induced hyperglycemia is associated with poor outcomes and most patients do not have diabetes. Exogenous glucagon-like peptide-1 (GLP-1) attenuates, but does not normalize, the glycemic response to enteral nutrition in the critically ill. In health, co-administration of glucose-dependent insulinotropic polypeptide (GIP) and GLP-1 has an additive insulinotropic effect. We aimed to evaluate whether GIP, when administered with GLP-1, confers additional glucose-lowering effects in the critically ill. Twenty critically ill patients without diabetes participated in a prospective, randomised, double-blinded, crossover study. Patients were fasted and insulin withheld. Between T0 and T420 minutes GLP-1 (1.2 pmol/kg/min), together with GIP (2 pmol/kg/min) or saline control, were administered intravenously on consecutive days. From T60 liquid nutrient was infused into the small intestine. Blood was sampled frequently. Baseline blood glucose, insulin and glucagon concentrations were no different. The addition of exogenous GIP doubled plasma GIP concentrations, but was not associated with any reduction in peak blood glucose (P=0.43) or overall glycemic response to nutrient (Figure 1). This study indicates that the addition of exogenous GIP to GLP-1 does not have additional glucose-lowering or insulinotropic effects in critically ill patients, and that further evaluation of incretin-based therapies should focus on GLP-1 and its agonists. Supported by: Novo Nordisk A/S Guided Audio Tour: New Insights in GLP-1 Therapy (Posters: 1008-P to 1013-P), see page 17. & 1008-P Functional Activity of Lixisenatide and GLP-1 Receptor Expression in In-Vitro Thyroid C-Cells of Rat and Human Origin U. SCHWAHN, S. STENGELIN, ULRICH WERNER, Frankfurt, Germany GLP-1 receptor (GLP-1R) agonists are a new class of antidiabetic drugs. However, some GLP-1R agonists have been shown to induce proliferation of rodent thyroid C-cells in preclinical safety studies. Whether this is a class effect in rodents or a ligand-specific phenomenon is still under debate. Therefore, we studied the effect of the GLP-1R agonist lixisenatide (LIXI) in two thyroid C-cell lines derived from rats or humans. LIXI, exendin-4 (EXE), GLP-1(7-36) amide (GLP-1) and liraglutide (LIRA) were tested in a functional cAMP assay for effects on the rat (6-23) and human (TT) thyroid C-cell carcinoma cell lines. GIP, the natural ligand of the GIP receptor (GIPR), which also uses cAMP as 2nd messenger, was tested for comparison. There was a strong response to all 4 GLP-1R agonists in the rat cell line 6-23 (Emax% ~100%) compared to a weak response in human TT cells (Emax% ~20%). In rat-derived cells EC50 for LIXI, EXE, GLP-1 and LIRA was 8, 4, 7 and 715 pM, resp. while it was 56 and 38,600 pM for GLP-1 and LIRA, resp. in human TT cells. Due to marginal signals no numerical values could be determined for LIXI and EXE. In contrast, GIP induced strong responses in both cell lines (Emax% = 76% and 102%; EC50 = 1320 and 181 pM). In line with ADA-Funded Research & Supported by: NHMRC (Australia) & 1010-P Treatment Satisfaction With Albiglutide and Liraglutide in Patients With Type 2 Diabetes Uncontrolled With Oral Therapy ALAN A. MARTIN, SUSAN L. JOHNSON, JUNE YE, RICHARD E. PRATLEY, Uxbridge, United Kingdom, Research Triangle Park, NC, Orlando, FL We assessed treatment satisfaction with albiglutide (ALB) a QW GLP1 agonist and QD liraglutide (LIR) in patients with Type 2 diabetes (T2D) inadequately controlled by oral antidiabetics. In a 32-week, open-label trial (HARMONY 7), T2D patients treated with oral antidiabetics with A1C 7%-10% were randomized to ALB QW 30mg titrated to 50mg (n = 404) or LIR 0.6mg titrated to 1.8 mg QD (n = 408). Primary endpoint was change from baseline in A1C. Treatment satisfaction was measured In US patients (728 out of 812 subjects randomized) using a For author disclosure information, see page 829. Guided Audio Tour poster A259 POSTERS MICHAEL LEE, MAHESH UMAPATHYSIVAM, ADAM M. DEANE, MARIANNE CHAPMAN, KRISHNASWAMY SUNDARARAJAN, JONATHAN FRASER, CAROLINE COUSINS, CHRISTOPHER ANNINK, MICHAEL HOROWITZ, JURIS J. MEIER, CHRISTOPHER K. RAYNER, Adelaide, Australia, Bochum, Denmark Clinical Diabetes/ Therapeutics The Effect of Exogenous GIP in Combination With GLP-1 on Glycaemia in Critically Ill Patients CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES for 14 weeks. All subjects met with a dietician once a week and were advised to decrease calorie intake by 500 kcal/day. To quantify changes in insulin resistance, steady-state plasma glucose (SSPG) concentration was measured during the insulin suppression test. We also measured blood pressure, pulse, fasting glucose, and lipoprotein concentrations to assess changes in CVD risk factors. Eleven out of 35 individuals (31%) assigned to liraglutide dropped from the study compared with 6 out 33 (18%) on placebo (p=0.26). Subjects who remained on liraglutide (n=24) lost twice as much weight as those on placebo (n=27): 6.8 vs. 3.3 kg, p<0.001. In addition, liraglutide-treated subjects had a significant improvement in SSPG concentration (-58 vs. 3 mg/dL, p <0.001), as well as significantly greater lowering of systolic blood pressure (-8.1 vs. -2.6 mmHg, p=0.04), fasting glucose (-9.5 vs. 0.3 mg/dL, p<0.001) and triglyceride (-37 vs -13 mg/dL, p=0.04) concentration. However, pulse increased after liraglutide treatment (6.4 vs. -0.9 beats/min, p=0.001). In summary, individuals with prediabetes who were able to tolerate liraglutide treatment experienced twice as much weight loss as those on placebo. Weight loss associated with liraglutide treatment led to improvement in insulin resistance and several CVD risk factors but was accompanied by increase in pulse. Liraglutide can significantly augment weight loss and improve many CVD risk factors in this population at high risk to develop T2DM and CVD. POSTERS Clinical Diabetes/ Therapeutics validated instrument, the Diabetes Medication Satisfaction questionnaire (DMS), which measures satisfaction in three domains, treatment burden, efficacy and unwanted side effects (symptoms), and overall satisfaction. All scores range from 0 (worst) to 100 (best). At baseline, demographics were well matched and mean DMS scores were similar. At 32 weeks, change from baseline in A1C was −0.78% with ALB and −0.99% with LIR, which did not meet prespecified non-inferiority criteria (0.3) vs. LIR (95% CI 0.08-0.34%). Weight loss was greater and nausea and vomiting more frequent with LIR (abstract 945-P ADA 2012). DMS overall score improved in both groups, and the greatest improvements were in the efficacy domain (Table). No between group differences reached significance. Treatment satisfaction improved with both albiglutide and liraglutide mainly due to satisfaction with perceived efficacy. & 1013-P Direct Effect of Exenatide on Endothelial Function in Humans JURAJ KOSKA, KALYANI RARAVIKAR, DAWN C. SCHWENKE, DAVID A. D’ALESSIO, PETER D. REAVEN, Phoenix, AZ, Cincinnati, OH & Exenatide (Ex) improves postprandial endothelial function (EF), an effect that is partially, but not entirely explained by reduced postprandial glucose or lipid levels. Since endothelium express active GLP-1 receptors, we hypothesized that Ex may also improve EF under fasting conditions directly via GLP-1 receptor mediated mechanisms. Following an overnight fast, 32 participants (31 males, age 60 [mean] ± 6 [SD] years, BMI 33 ± 12 kg/m2) with impaired glucose tolerance (IGT) or diet controlled type 2 diabetes received intravenous a) Ex (50ng [12 pmol]/ min), b) Ex with GLP-1 receptor antagonist exendin-9 (Ex9, 600 pmol/kg·min), or c) placebo (Pl) in a randomized three-period crossover fashion. EF was measured as reactive-hyperemia index (RHI) by peripheral arterial tonometry before and during the last 15 minutes of the infusions (Figure). 1011-P Albiglutide Does Not Impact the Counter Regulatory Hormone Response or Recovery Time from Hypoglycemia in Patients With Type 2 Diabetes (T2DM): A Randomized, Double-Blind, Placebo Controlled, Stepped Automated Glucose Clamp Study MARCUS HOMPESCH, ANGELA LEONE-JONES, MOLLY C. CARR, JESSICA MATTHEWS, HUI ZHI, MALCOLM A. YOUNG, LINDA MORROW, RICKEY REINHARDT, Chula Vista, CA, King of Prussia, PA The impact of Albiglutide (Albi) a selective GLP-1 receptor agonist on counter regulatory hormones and recovery time from hypoglycemia was assessed in a glucose clamp study in T2DM patients. Participants received either a single SC dose of Albi (50mg, N=22) or P (N=22) 3 days before the clamp procedure (glycemic targets: 9.0, 5.0, 4.0, 3.3, and 2.8 mmol/L). Assessments included safety, hormonal responses (glucagon , c-peptide (CP), de-convoluted insulin secretion rate (ISR), cortisol, epinephrine, norepinephrine, growth hormone) at all glycemic targets (30 min duration each) and recovery time (RT) to 3.9 mmol/L after release of IV insulin infusion at the end of the 2.8 mmol/L period. Demographics were similar between groups (mean age 49 yrs, BMI 33.4 kg/m2, 46% male). In patients on oral anti-hyperglycemic drugs, therapy was discontinued 2 weeks before study drug dosing; except metformin which was discontinued only on the clamp day. Hormonal responses for all counter regulatory hormones at all glycemic targets were similar between treatment groups. ISR (de-convoluted from CP) was significantly higher at the end of the hyper- and euglycemic plateaus for Albi (0.32 vs 0.22 and 0.15 vs 0.08 mU/h, p< .05) with no differences between Albi and P at any of the hypoglycemic plateaus. RT to 3.9 mmol/L was similar between Albi and P, median (95% CI; 35 (CI 30-45) vs 30 (CI 25-40) min). There were no clinically relevant differences in safety data between Albi and P. Treatment emergent adverse events were similar between Albi (9) and P (10). Two mild TEAEs in the Albi group were related to study drug. In conclusion, a single dose of Albi in T2DM patients did not impair the counter regulatory hormone response to hypoglycemia and had no impact on the RT from hypoglycemia. It was confirmed that Albi stimulates ISR in a glucose dependent manner. & The increment in RHI was greater with Ex than Pl and this was completely abolished with Ex9 (Figure). The differences in RHI between Ex alone versus Pl or versus Ex+Ex9 remained significant after adjustment for changes in glucose (p=0.02), insulin (p=0.007), and both glucose and insulin (p=0.02) concentrations. There were no differences in responses to treatments between those with IGT and type 2 diabetes (p=0.95). These findings are consistent with improvement of EF by Ex via a direct action on endothelial GLP-1 receptors. Combined with our previous data, Ex appears to improve EF by both direct and indirect (improved postprandial glucose and lipid values) mechanisms. 1012-P Liraglutide for Overweight, Older Individuals With Prediabetes SUN KIM, FAHIM ABBASI, CINDY LAMENDOLA, ALICE LIU, DANIT ARIEL, PATRICIA SCHAAF, KAYLENE GROVE, GERALD REAVEN, Stanford, CA Older, overweight/obese individuals with prediabetes are at high risk to develop type 2 diabetes (T2DM) and cardiovascular disease (CVD). The aim of the present study was to evaluate the efficacy of liraglutide, a glucagonlike peptide 1 (GLP-1) analogue, to augment weight loss and improve CVD risk factors in this population. We randomized 68 older individuals (40-70 years of age) with overweight/obesity (BMI 25-40 kg/m2) and prediabetes to this double-blind study of liraglutide 1.8 mg versus matching placebo injections Supported by: Amylin Pharmaceuticals, LLC.; Eli Lilly and Company & For author disclosure information, see page 829. A260 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES 1014-P 1016-P Relationship between Postprandial Glucagon-Like Peptide 1, Glucagon Levels and Short and Long-Acting Insulin Requirement in CPeptide Negative Type 1 Diabetic Patients GLP-1 (32-36) Amide, a Novel Pentapeptide Cleavage Product of GLP-1, Modulates Whole Body Glucose Metabolism in Dogs We have previously demonstrated that under euglycemic clamp conditions in human subjects GLP-1(9-36) amide infusions inhibited endogenous glucose production without substantial insulinotropic effects. Earlier reports indicate that GLP-1(9-36)amide is cleaved in vivo to a nonapeptide, GLP-1(28-36)amide and a pentapeptide GLP-1(32-36)amide (LVKGRamide). Here we study the effects of the pentapeptide on whole body glucose disposal during hyperglycemic clamp studies. Five dogs underwent indwelling catheterizations. Following recovery, dogs underwent a 180 min hyperglycemic clamp (basal glucose+98 mg/dl) in cross-over studies. Saline or pentapeptide (30 pg·kg-1·min-1) was infused during the last 120 min after commencement of hyperglycemic clamp in a primed continuous manner. During the last 20 min of the pentapeptide infusion, glucose utilization (M) significantly increased to 21.3 ± 3.00 mg·kg1·min 1compared to M of 15.4 ± 1.6 mg/kg/min during the saline infusion (P=0.04). No significant differences in insulin (27 ± 3.4 vs. 24 ± 2.6 µU/ml, P=NS) or glucagon secretion (35 ± 2.7 vs. 33 ± 1.1 pg/ml, P=NS) were observed. These findings demonstrate that under hyperglycemic clamp studies, the GLP-1-derived pentapeptide, GLP-1(32-36) amide has biological actions in the modulation of glucose metabolism by stimulation of whole-body glucose disposal. Further, the findings suggest that the metabolic benefits previously observed during GLP-1(9-36) amide infusions in humans might be due, at least in part, to the metabolic effects of the pentapeptide that is cleaved from the pro-peptide, GLP-1(9-36) amide in the circulation. 1017-P Optimizing Glycemic Control in T2DM Patients Previously Treated With Intensive Insulin Therapy and Switched to Exenatide-Insulin Glargine Combination IAKOVOS AVRAMIDIS, CHRISTINA SPYRIDONIDOU, DIMITRIS STOIMENIS, ANTIGONI LALIA, Thessaloniki, Greece 1015-P Pharmacokinetics and Pharmacodynamics of a Chemically Stable Micro-Dosed Glucagon in a Diabetic Swine Model of Type 1 Diabetes The introduction of insulin in T2DM patients is frequently complicated with weight gain and high occurrence of hypoglycemia. Moreover, in many trials exenatide has been associated with improved glycemic control with less hypoglycemic events and weight reduction. The present study evaluated the glycemic control in patients who were previously treated with intensive insulin therapy and were switched to exenatide-insulin glargine. 56 patients (aged 68.50 years (8), BMI 35.42 kg/m2 (5.71), HbA1c 8.10% (1), diabetes duration 17.46 years (6.50), insulin therapy duration 6.25 years (4.11), hypoglycemic episodes/month 5 (5); means (SD)) were included in the study. All patients received basal insulin and 3 doses of prandial insulin (basal insulin dose 44 IU (21), total insulin dose 74 IU (34)). Subjects continued taking their metformin dose and insulin glargine. Pioglitazone was added in 17 patients. Subjects served as their own controls. The average follow-up was 12.5 months (7, range 3-27). Mean HbA1c reduction was 1% (1.22) and weight loss was 8 kg (6.33). The basal and total insulin dose reduction from baseline was 4 IU (15) and 34 IU (25) respectively (p=0.020 and p<0.001). Diabetes and insulin therapy duration were not associated with response to treatment (p=0.620 and p=0.712 respectively). Mean HbA1c reduction was 0.80% (1) during the first 6 months and the additional reduction was 0.40% (0.90) from 6 months to the end of followup. Weight reduction was 4.60 kg (4.20) and the further reduction was 3.50 kg (4.50). Based on the target of HbA1c of <7% the percentage of patients achieving goal glycemic status differed at follow-up (53.57%) compared with baseline (5.36%, p<0.001). Hypoglycemic episodes were reported only by one patient. Our study indicated that in a group of patients the exenatide-basal insulin combination achieved sustained glucose control with weight improvement and significant lower incidence of hypoglycemia. JOHN JIANG, KATHERINE M. MCKEON, FIRAS H. EL-KHATIB, STEVEN J. PRESTRELSKI, NANCY L. SCOTT, BRETT J. NEWSWANGER, PATRICK SLUSS, STEVEN J. RUSSELL, EDWARD R. DAMIANO, Boston, MA, Austin, TX In order to market a bihormonal bionic endocrine pancreas that provides both insulin and glucagon replacement therapy for people with type I diabetes, a pumpable glucagon formulation is needed that is bioactive on the liver and chemically stable for up to up to three days at near body temperatures. We analyzed the pharmacokinetics (PK) and pharmacodynamics (PD) of a stable, non-aqueous, concentrated (5 mg/ml), pumpable, liquid glucagon formulation (XeriSol™, Xeris Pharmaceuticals) in a porcine model of type 1 diabetes. Doses of XeriSol that were fresh or aged 7-8 days with agitation were administered subcutaneously using an insulin pump. Experiments began ~5 hours postprandially with blood glucose regulated to 60-140 mg/dl. Two consecutive glucagon doses (separated by > 60 minutes) were administered for each experiment. Venous blood glucose concentration was measured every 10 minutes. Plasma samples were drawn every 10 minutes for offline analysis of plasma glucagon concentration. Identical experiments were conducted using a freshly reconstituted commercially available glucagon formulation (Eli Lilly), which is chemically stable in solution for only a few hours. The second dose response for each experiment (n = 3 experiments) was considered when calculating the increase in blood glucose in response to glucagon dosing. Every dose was considered in the calculation of glucagon tmax. The mean tmax for fresh and aged XeriSol were 9 + 2 and 12 + 5 minutes, respectively and 10 + 5 minutes for fresh Lilly glucagon. The average increase in blood glucose response for fresh and aged XeriSol was 14 + 6 and 18 + 2 mg/dL/10µg, respectively (n = 3 experiments), compared with 19 + 4 mg/dl/10µg for fresh Lilly glucagon (n = 2 experiments). These results suggest that both aged and fresh XeriSol shows comparable efficacy to freshly reconstituted Lilly glucagon in STZ-treated diabetic pigs. This paves the way for an upcoming first-in-human study. 1018-P Add-On Treatment With Exenatide Once Weekly vs. Daily Basal Insulin in Patients With A1C ≥8.5% ERICH BLASE, STEVEN C. BRUNELL, YAN LI, MICHAEL GRIMM, San Diego, CA Typically, basal insulin (b-INS) is the add-on treatment of choice for patients with severe hyperglycemia, but it can be questioned whether b-INS is the best option. The current post hoc analyses compared the efficacy and tolerability of exenatide once weekly (EQW) with those of b-INS in patients Supported by: Frederick Banting Foundation ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A261 POSTERS It has been suggested that glucagon-like peptide 1 (GLP-1) in C-peptide negative type 1 diabetic (T1DM) patients decreases glucose concentration through reducing glucagon secretion. Although mechanisms by which GLP-1 affects glucagon secretion has not been well understood, in patients with preserved GLP-1 secretion decreased glucagon secretion and improved glucose control were documented. We assessed the difference in postprandial total GLP-1 level in relation to glucagon secretion and both, short and long-acting insulin requirement in 79 C-peptide negative T1DM patients (median age 46 years, T1DM duration of median 21 years, hemoglobin A1c median level 7.3%, short-acting insulin requirement mean 0.1 and long-acting insulin requirement median 0.3 unit/kg/day). Plasma total GLP-1 and glucagon levels were measured by ELISA assay (DRG Diagnostic, Germany). The group of patients with higher postprandial total GLP-1 concentration (≥2.6 pmol/L, n=40) required lower dose of long-acting insulin (0.28 vs 0.31, p=0.043), had higher fasting total GLP-1 concentration (1.25 vs 0.82, p=0.007), fasting glucagon (118.1 vs 95.3 pg/mL, p=0.048) and postprandial glucagon (126.6 vs 99.1, p=0.007) concentration. There was no difference in short-acting insulin requirement (p=0.053). Inappropriate elevation of glucagon could be explained by lack of inhibition of glucagon secretion due to low total GLP-1 concentration documented in our patients. Lower total GLP-1 concentration in our patients might be explained by insulin therapy causing an increase in serum dipeptidyl peptidase IV activity, which was recently reported in T1DM patients. Endogenously secreted GLP-1 plays an important role in glucoregulation in T1DM by modulating glucagon levels. The complex interplay between GLP-1, glucagon secretion and exogenously administered insulin should be investigated in future clinical trials of T1DM patients. Clinical Diabetes/ Therapeutics AMIN VAKILIPOUR, FRANCA S. ANGELI, OLGA D. CARLSON, LI CHEN, YOU-TANG SHEN, EVA TOMAS-FALCO, JOEL F. HABENER, RICHARD P. SHANNON, JOSEPHINE M. EGAN, DARIUSH ELAHI, Philadelphia, PA, Baltimore, MD, Boston, MA KARIN ZIBAR, TOMISLAV BULUM, KRISTINA BLASLOV, LEA DUVNJAK, Zagreb, Croatia POSTERS Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES with T2DM on metformin ± SFU with baseline A1C ≥8.5%. Data were pooled from two 26-week, randomized, controlled studies (EQW vs insulin glargine and EQW vs insulin detemir; N = 137 [EQW], 126 [b-INS]). Patients treated with EQW had a significantly greater decrease in A1C from baseline than those treated with b-INS (LS mean ± SE change: -2.0% ± 0.08 vs -1.6% ± 0.08; P=0.0008) and were significantly more likely to reach an A1C goal of <7.0% (39.4% vs 23.0%; P=0.002). Decreases in FPG were smaller in the EQW group (-52.2 ± 3.70 vs -61.9 ± 3.74 mg/dL; P=0.062]). Mean ± SE weight loss with EQW was -2.4 ± 0.23 kg, whereas weight gain with b-INS was 2.0 ± 0.24 kg (LS mean difference between groups, -4.4 kg; P<0.0001). Patients in the EQW group were significantly more likely to achieve a composite goal (A1C <7.0%, no weight gain, and no hypoglycemia [requiring assistance or self-treated with blood glucose <54 mg/dL]) than were patients in the b-INS group (33.6% vs 3.2%; P<0.0001). Hypoglycemia occurred at a rate of 0.08 exposure-adjusted events per patient year in the EQW group and 0.37 events in the b-INS group. The most common adverse events were nausea (28.5%), nasopharyngitis (16.8%), and hypoglycemia (14.6%) in the EQW group and hypoglycemia (38.9%) and nasopharyngitis (19.0%) in the b-INS group. Results of these analyses show that treatment with EQW was associated with significantly greater reductions in A1C and body weight, and a lower rate of hypoglycemia than treatment with b-INS. Results from this post hoc analysis suggest that EQW is a treatment alternative to b-INS in patients with A1C ≥8.5% who are receiving treatment with oral antihyperglycemic medications and are concerned about weight gain and the risk of hypoglycemia. weeks (5 ug bid×4 weeks followed by 10 ug bid×8 weeks). Eighteen healthy subjects (age 49.6±9.2 yrs, BMI 25.2±1.9 kg/m2, Male/Female 8/10) were recruited as normal control. The effects of exenatide on the expression of cytokines, toll-like receptor (TLR)-2, c-Jun N-terminal kinase (JNK)-1 and IκB kinase (IKK) β were investigated.The plasma levels of tumor necrosis factor (TNF)-α, interleukin (IL)-1β and IL-6 were significantly higher in T2D subjects than those of normal control (18.2±2.4 vs 12.5±3.6 pg/ml, 7.3±1.5 pg/ml vs 4.3±0.8 pg/ml, 49.4±30.4 vs 6.2±2.9 pg/ml, respectively (P < 0.001 for all). After treatment with exenatide for 12 weeks, fasting blood glucose fell from 9.6±1.8 to 7.1±1.2 mmol/L, postprandial blood glucose from 16.2±3.8 to 10.8±2.8 mmol/L, HbA1c from 8.5±0.9 to 6.5±0.9 % with BMI from 27.1±3.7 to 25.7±3.9 kg/m2 (P < 0.05). There was a significant reduction in plasma levels of TNF-α, IL-1β and IL-6 by 37.6±14.9, 35.8±21.0 and 61.7±16.1% , respectively, and the mRNA expression of TNF-α, IL-1β and IL-6 in PBMCs by 54.4±14.2, 75.6±18.4 and 51.2±12.5 %, respectively (P < 0.05 for all) after 12 weeks of exenatide. The expression of TLR-2, JNK-1 and IKKβ in PBMCs was suppressed by 72.1±19.4, 60.1±33.6 and 59.3±31.2%, respectively (P <0.05 for all). The study suggested that exenatide exerts an antiinflammation effect in PBMCs of newly diagnosed and drug-naïve T2D patients. 1019-P JULIO ROSENSTOCK, MAEVA GERME, EDWARD WANG, JAY LIN, RICCARDO C. BONADONNA, PEDRO DE PABLOS-VELASCO, RONAN ROUSSEL, DENIS RACCAH, Dallas, TX, Paris, France, Bridgewater, NJ, Verona, Italy, Las Palmas, Spain, Poitiers, France, Marseille, France Supported by: NSFC (81025005), (81100556) 1021-P Expanding the Basal-Plus Regimen: Basal Insulin + Lixisenatide Is More Likely to Achieve the Composite Outcome Of Hba1c <7%, No Documented Symptomatic Hypoglycemia and No Weight Gain Compared With Basal + Prandial Insulin Biological Profile of LY2405319, a Human FGF21 Variant and a Clinical Candidate ALEXEI KHARITONENKOV, JOHN M. BEALS, TAMER COSKUN, JAMES D. DUNBAR, BARBARA C. HANSEN, THOMAS F. BUMOL, DAVID E. MOLLER, Indianapolis, IN, Tampa, FL Lixisenatide (LIXI), a novel once-daily prandial GLP-1 receptor agonist, is in development for T2DM management as add-on to oral antidiabetic drugs ± basal insulin. An analysis of patient-level data from 5 randomized trials (GetGoal-Duo1, GetGoal-L, OPAL, ELEONOR, 1-2-3) was conducted to evaluate basal insulin + LIXI 20 µg QD vs a basal-plus regimen (basal + 1 dose of prandial insulin). Composite efficacy endpoints were evaluated using linear or logistic multivariate regression analyses in 1184 overall patients (519 basal + LIXI; 665 basal-plus). A slightly higher proportion of patients in the basal + LIXI group achieved HbA1C <7% vs basal-plus (OR: 1.62 [95% CI: 1.06-2.48] p=0.025), while basal-plus was more likely to result in weight gain (estimated difference: 1.38 kg [95% CI: 0.83−1.93] p<0.0001) and more documented symptomatic hypoglycemia (OR: 1.51 [1.01−2.25] p=0.045). Basal + LIXI was 2.5x more likely to result in HbA1C <7% and no weight gain, and nearly 3x more likely to result in HbA1C <7%, no documented symptomatic hypoglycemia and no weight gain (Table). In conclusion, both basal + LIXI and basal-plus regimens are effective in reducing HbA1C but basal + LIXI is significantly more likely to result in HbA1C <7% with no documented hypoglycemia and no weight gain vs basal-plus in T2DM patients not reaching HbA1C goals on basal insulin. Fibroblast growth factor 21 (FGF21) is a novel hormone-like metabolic regulator which has been implicated as a therapeutic target for type 2 diabetes, obesity, hepatic steatosis, and cardiovascular disease. However, development of human recombinant wild type FGF21 as a therapeutic agent is challenging due its poor biopharmaceutical properties. In order to improve protein stability and facilitate homogenous protein production, we designed a novel FGF21 variant, LY2405319 that contained four engineered changes: L118C, A134C, S167A substitutions and deletion of the four N-terminal amino acids, HPIP. LY2405319 was profiled for bioactivity in a panel of in vitro and in vivo assays which confirmed that biological properties of LY2405319 were essentially identical to FGF21. In mouse and human cells, LY2405319 and FGF21 have comparable potency while the critical features of FGF21 mechanism, KLB-dependence together with the non-mitogenic character of action, remained intact in LY2405319. Administration of LY2405319 in mice over 14 days resulted in a 25-50 % lowering of plasma glucose coupled with a 10-30 % reduction in body weight. The FGF21 variant also induced a variety of metabolic effects in diabetic rhesus monkeys. Treatment consisted of three subcutaneous LY2405319 dose levels, 1.5, 9 and 50mg/kg (daily), for seven weeks. All 6 monkeys at the end of LY2405319 administration showed 12-25% body weight loss, 90% mean reduction in serum triglycerides, lowering of total and LDL-cholesterol (50 % and 30%, respectively), and a ≈3 fold elevation in HDL-cholesterol; hyperglycemia was completely normalized in 5 out of 6 animals. In summary, LY2405319 exhibited biological properties that were indistinguishable from native FGF21. Thus, the profile of LY2405319 had features required for initiation of a clinical program designed to test the hypothesis that administration of exogenous FGF21-based molecule could result in efficacy with respect to important disease-related metabolic parameters in humans. Basal + LIXI vs basal-plus: composite endpoints Endpoints Basal + Basal- Odds 95% CI p value LIXI* plus† ratio n=519 n=665 % patients Regression analysis 28.3 29.8 1.89 1.20−2.97 p=0.0057 HbA1C <7% and no documented symptomatic hypoglycemia 27.9 22.5 2.56 1.63−4.01 p<0.0001 HbA1C <7% and no weight gain HbA1C <7%, no documented symptom19.6 15.8 2.76 1.64−4.64 p=0.0001 atic hypoglycemia and no weight gain *Includes GetGoal-Duo 1 (NCT00975286) and GetGoal-L (NCT00715624); †Includes OPAL (NCT00272012), ELEONOR (NCT00272064), 1-2-3 (NCT00135083). CI=confidence interval; HbA1C=glycated hemoglobin; Basal + LIXI=basal insulin + lixisenatide; Basal-plus=basal insulin + one dose of prandial insulin; documented symptomatic hypoglycemia=blood glucose <3.33 mmol/L. 1020-P Antiinflammation Effect of Exenatide in Chinese Newly Diagnosed and Drug-Naïve Type 2 Diabetic Patients XIAOHUAN XING, JINHUA YAN, WEN XU, HAIXIA XU, HUA LIANG, FEN XU, JIANPING WENG, Guangzhou, China Type 2 diabetes has been considered as a chronic inflammatory disease. Previous studies indicated that exenatide, a glucagon-like peptide-1 (GLP-1) receptor agonist, might have the effect of attenuating the inflammation of type 2 diabetic (T2D) patients. The study aims to investigate the antiinflammation effect of exenatide in peripheral blood mononuclear cells (PBMCs) of newly diagnosed and drug-naïve T2D patients. Nine newly diagnosed and drugnaïve T2D patients [age 45.1±6.1 yrs, body mass index (BMI) 27.1±3.7 kg/m2, Male/Female 3/6] were enrolled and received exenatide treatment for 12 Supported by: sanofi-aventis & For author disclosure information, see page 829. A262 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES 1022-P 1024-P Targeting Insulin Resistance in Type 2 Diabetes via Immune Modulation of Cord Blood-Derived Multipotent Stem Cells (CB-SCs) in Stem Cell Educator Therapy A Novel Long-Acting GLP-1R Agonist, SKL-18287, Has Unique Properties Compared to Liraglutide Supported by: Chinese Government Funding 1023-P 1025-P Reductions in A1C and Fasting Glucose With Exenatide Are Not Affected by Relative β-Cell Function in T2DM Patients Chronic Continuous Exenatide Infusion Does Not Cause Pancreatitis or Ductal Hyperplasia in Baboons: A 14 Weeks Longitudinal Controlled Study MICHAEL GRIMM, YAN LI, JENNY HAN, RICHARD PENCEK, San Diego, CA A1C reduction associated with the GLP-1 receptor agonist exenatide has been attributed in part to improvement in β-cell function and sensitivity, but it is unclear if it is dependent upon baseline β-cell function. This analysis of pooled data from 11 randomized, controlled, 24-30 week, clinical trials of exenatide twice daily (EBID, 7 trials) and once weekly (EQW, 4 trials) assessed glycemic control in relation to baseline β-cell function. Baseline HOMA-B, an estimate of steady-state β-cell function, was calculated from the homeostatic model assessment using fasting glucose (FG) and C-peptide. Patients treated with both EBID and EQW were divided into tertiles according to baseline HOMA-B with the lowest values in tertile 1 (table). Baseline A1C and FG were highest in the 1st HOMA-B tertiles and decreased in higher HOMA-B tertiles, implying a manifest association in baseline characteristics between glycemic control and β-cell function. Changes in A1C and FG were evaluated by ANCOVA models adjusted for baseline values of A1C or FG. All tertiles exhibited statistically significant reductions in A1C and FG (P<0.05). There were no statistically significant differences in changes in A1C or FG among tertiles within EBID or EQW groups. Adverse events were consistent with the known exenatide safety profile. In summary, both exenatide formulations resulted in meaningful improvements in A1C and FG regardless of underlying β-cell function. TERESA VANESSA FIORENTINO, MICHAEL OWSTON, GREGORY ABRAHAMIAN, STEFANO LAROSA, ALESSANDRO MARANDO, GIOVANNA FINZI, FAUSTO SESSA, SUBHASH KAMATH, FRANCESCA CASIRAGHI, GIUSEPPE DANIELE, FRANCESCO ANDREOZZI, ANTHONY COMUZZIE, MARK SHARP, PRISCILLA WILLIAMS, RALPH A. DEFRONZO, GLENN HALFF, EDWARD DICK, FRANCO FOLLI, San Antonio, TX, Varese, Italy The effects of GLP-1-analogues and DPP-IV inhibitors on exocrine pancreas are subject to intense investigations. Here, we evaluated the effects of chronic exenatide (EXE) treatment on exocrine pancreas structure and cell replication of non-human primates (baboons). 52 animals underwent 10-30% resection of pancreatic tail followed by continuous infusion of EXE (0.014 ug/kg-hour) or Saline (SAL) for 14 weeks. EXE plasma levels (753±244 pg/ mL) were 3-4 fold higher than those achieved in treated diabetic patients. Morphometry and immunocytochemistry for exocrine cell proliferation (Ki67) and apoptosis (M-30) were performed on baseline (tail) and after treatment samples (body/head). Pancreatitis and PanIN (pancreatic intraepithelial neoplasia) were never observed in EXE and SAL group. Ductal hyperplasia was increased by ~ 4 fold at the end of the study in both groups (EXE: 17 vs 4, p=0.003; SAL: 12 vs 3, p=0.006) suggesting a strong surgery effect. At the end of study, Ki67 positive acinar cells number did not significantly change, as compared to baseline, in SAL and EXE group. Ki67 positive ductal cells were significantly increased after EXE treatment (1.83±0.34 vs 0.71±0.22, p=0.04), however the delta of Ki67 positive ductal cells in EXE group was not significantly different from SAL (1.17±0.32 vs 0.17±0.55, p=0.13). The number of M-30 positive acinar and ductal cells did not change after SAL and EXE treatment in comparison to baseline. In conclusion, continuous infusion of EXE for 14 weeks did not produce any negative effects on exocrine pancreas, neither promoting pancreatic inflammation nor hyperplasia in non-human primates. Supported by: NIH (R01DK080148); Amylin/Lilly ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A263 POSTERS SKL-18287 is a novel long-acting GLP-1R agonist designed to exhibit DPP4 resistance and multimer formation for improving in vivo stability with several natural amino acid modifications of native GLP-1 that is currently being developed as QD and QW GLP-1R agonist product. To clarify further characteristic of SKL-18287, we investigated and compared it with liraglutide. SKL-18287 showed agonistic activity to mouse and human GLP-1R in vitro with ED50 values of 1.2 and 274 nmol/L, respectively. The competitive binding curve of SKL-18287 to human GLP-1R was not affected by increment of albumin concentration in the system while that of liraglutide was shifted toward the right side. It is noteworthy that radioactivity was distributed to the pancreas at nearly the same level as that to the plasma which was much more than to the other tissues after subcutaneous injection of 3H-[Tyr]SKL18287 to Sprague Dawley (SD) rats. As a consequent of this unique tissue distribution of SKL-18287 in vivo, 4 weeks repeated injection of SKL-18287 (6-12 nmol/kg) significantly and dose-dependently lowered HbA1c in GotoKakizaki T2DM rats. The reduction in HbA1c during 4 weeks of SKL-18287 administration (-0.26%) was more potent than that with liraglutide (-0.10%) when the same dosage was used. Interestingly, the inhibitory effect on gastric empty of SKL-18287 (12 nmol/kg) was nearly the same as that of liraglutide (8 nmol/kg) in SD rats. It is also known that GLP-1R agonists increase plasma calcitonin via GLP-1R in the thyroid C-cell in rodents. Because of its unique tissue distribution, the increment effect of SKL-18287 on plasma calcitonin in alloxan-induced diabetic mice was weaker than that of liraglutide and exenatide when SKL-18287 exhibited potent blood glucose lowering effect. These results suggest that SKL-18287 is a novel long-acting GLP-1R agonist which keeps a balance between efficacy and adverse effects. Considering these properties, we believe that SKL-18287 has potential as a new class of GLP-1R agonist. Mounting evidence points to the involvement of immune dysfunction in insulin resistance in T2D, suggesting that immune modulation may be a useful tool in treating the disease. We have developed a Stem Cell Educator therapy in which a patient’s blood is circulated through a closedloop system that separates mononuclear cells by aphaeresis, briefly cocultures them with adherent CB-SCs, and returns the educated cells to the patient’s circulation. In an open-label, phase1/phase 2 study, patients (N = 36) with long-standing T2D were divided into three groups (Group A, oral medications, n = 18; Group B, oral medications + insulin, n = 11; and Group C with impaired β-cell function, oral medications + insulin, n = 7). Median age was 51 years, and median diabetic duration was 9 years. All subjects received one Stem Cell Educator therapy. Clinical findings indicate that T2D patients achieve improved metabolic control and reduced inflammation markers after receiving Stem Cell Educator therapy. Median glycated hemoglobin (HbA1C) in Group A and B was significantly reduced from 8.61% ± 1.12 at baseline to 7.9% ± 1.22 at 4 weeks (p = 0.026), 7.25% ± 0.58 at 12 weeks (p = 2.62E-06), and 7.33% ± 1.02 at one year post treatment (p = 0.0002). Homeostasis model assessment (HOMA) of insulin resistance (HOMA-IR) demonstrated that insulin sensitivity was improved post treatment. Notably, the β-cell function in Group C subjects was markedly recovered, as indicated by the restoration of C-peptide levels (0.36 ± 0.19 ng/ml at baseline vs 1.12 ± 0.33 ng/ml at one year post treatment, p = 0.00045). Mechanistic studies revealed that Stem Cell Educator therapy reverses immune dysfunctions through immune modulation on monocytes and balancing Th1/Th2/Th3 cytokine production. Stem Cell Educator therapy is a safe and innovative approach that produces lasting improvement in metabolic control for individuals with moderate or severe T2D who receive a single treatment. Clinical Diabetes/ Therapeutics YU YAMAGUCHI, MIYUKI TAMURA, HIROSHI KINOSHITA, MASAYUKI OKAMOTO, RYUJI OKAMOTO, MITSUAKI TAKEUCHI, YUKIE MIZUNO, SHINJI FURUTA, KATSURA TSUKAMOTO, SANWA KAGAKU KENKYUSHO, Inabe, Japan YONG ZHAO, ZHAOSHUN JIANG, ZHAOHUI YIN, YE ZHANG, JIE SHEN, TINGBAO ZHAO, MINGLIANG YE, YALIN DIAO, YUNXIANG LI, HATIM THAKER, SUMMIT JAIN, HENG LI, Lisle, IL, Jinan, China, Chicago, IL CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES 1026-P 1028-P Early Liraglutide Treatment Improves β-Cell Function in Patients With Type 2 Diabetes Acute Metabolic Effects of GLP-1 Analogue in Type 1 Diabetes POSTERS Clinical Diabetes/ Therapeutics TARA GHAZI, LINDA RINK, JENNIFER L. SHERR, KEVAN C. HEROLD, New Haven, CT YOSHINOBU KONDO, SHINOBU SATOH, YUICHIRO INOUE, MASAYO KIMURA, MASAYUKI HIRAMA, URU NEZU, YASUO TERAUCHI, Kanagawa, Japan, Okinawa, Japan Some studies have suggested that GLP-1 receptor agonists may be beneficial to patients with Type 1 diabetes (T1D); however, the mechanisms of action are not clear. We studied the acute effects of exenatide (Byetta 5 µg) on insulin secretion, glucagon, GLP-1and GIP levels, and gastric emptying to oral mixed meal tolerance tests (MMTT) or an intravenous glucose tolerance test (IVGTT) in subjects with T1D, with (n=7) or without (n=5) residual insulin secretion (age: 30.4±9.5 years, HbA1c: 7.3±0.9 %, BMI: 27.9±5.2 kg/ m2, insulin usage: 0.62±0.2 U/kg/d, diabetes duration: 17.5±10.2 years). We calculated the area under the curve (AUC) for each metabolite. MMTT and IVGTT were performed while patients took basal but not bolus insulin, with and without pretreatment with exenatide. The glucose levels during the MMTT were reduced when exenatide was given (71957.7±26043.8 vs. 42921.9±13917.4 mg/dL, P=0.0005) but were unaffected during the IVGTT (3555.7±469.9 vs. 3612.1±489 mg/dL, P=0.687). The improved oral meal tolerance was accompanied by reduced glucagon levels (16106.9±3430.1 vs. 12835.9±5685.7 pg/mL, P=0.027) and delayed gastric emptying of acetaminophen (2034.6±517.1 vs. 606.4±365.6 µg/mL, P= 0.022) when exenatide was given. In subjects with residual insulin production there was a trend for reduced insulin secretion with exenatide (14321.9±17606.7 vs. 9272.5±8395.7 pmol/min, P=0.438). The fasting levels of GIP were lower when exenatide was given. In conclusion, our results show that exenatide suppresses glycemic excursion during MMTT regardless of β- cell function. The effect seemed to be mediated by decreased glucagon levels and delayed gastric emptying. Insulin secretion tended to be lower which may be the result of the latter. Preclinical studies with liraglutide have suggested improvements in β-cell function. Therefore, we investigated these effects using the glucagon stimulation test (GST). Sixty-six insulin-treated patients with type 2 diabetes were switched to liraglutide monotherapy and their β-cell function was measured with 1 mg intravenous GST at weeks 0 and 24. Liraglutide was initiated at 0.3 mg/day and titrated up to 0.9 mg/day. Its effect on β-cell function was assessed by the change in the area under the curve (AUC) of serum C-peptide immunoreactivity (CPR) during GST (AUC-CPR = (sum of CPR levels before and 6 min after GST) x 6 min/2). In full-set analysis, AUC-CPR increased after 24 weeks of liraglutide treatment (10.25 ± 5.04 to 11.90 ± 4.86 ng/ml·min, p = 0.005). In univariate regression model analysis, a negative correlation was seen between change in AUC-CPR (ΔAUC-CPR = (AUC-CPR at week 24) − (AUC-CPR at week 0)) and duration of diabetes (duration: β = −0.28, 95% confidence interval (CI), −0.43 to −0.14, p < 0.001, R2 = 0.19). In multivariate regression model analysis, the effect of treatment duration on ΔAUC-CPR was confirmed even after adjustments for age, gender, change in body weight, change in glycated hemoglobin, and insulin dose before liraglutide initiation (duration: β = −0.24, 95% CI, −0.42 to −0.06, p = 0.009, R2 = 0.28). In tertile analysis by duration of diabetes, early liraglutide treatment (duration ≤ 3 years) significantly improved AUC-CPR (10.00 ± 6.03 to 13.44 ± 5.78 ng/ml·min, p < 0.001), whereas late liraglutide treatment (duration ≥ 11 years) did not (10.68 ± 5.29 to 10.07 ± 4.37 ng/ml·min, p = 0.59). In receiver operating characteristic (ROC) analysis, the cut-off value for treatment duration for positive liraglutide effect on ΔAUC-CPR was 12 years (area under the ROC curve 0.73, sensitivity 86% and specificity 59%). These findings suggest that early liraglutide treatment potentially improves β-cell function in patients with type 2 diabetes. 1027-P Patient-Reported Outcomes With Dulaglutide vs. Metformin (AWARD-3) MATTHEW REANEY, BETH D. MITCHELL, PING WANG, VALERIA PECHTNER, BRAD CURTIS, KATE VAN BRUNT, Indianapolis, IN 1029-P This 52-week (wk) Phase 3, randomized, double-blind, parallel-arm, monotherapy study compared safety and efficacy of dulaglutide (DU), a longacting GLP-1 receptor agonist, to metformin (MET) in patients (pts) with early type 2 diabetes (A1C 6.5-9.5%). Pts received once-weekly DU 1.5 mg or 0.75 mg or MET 1000 mg twice daily. Both DU doses at 26 wks and DU 1.5 mg at 52 wks demonstrated superior glycemic control compared to MET (p<0.025). Weight reductions were observed in all groups. Patient-reported outcome measures of treatment satisfaction (DTSQ), weight-related self-perception (IW-SP), diabetes symptoms (DSC-R), and ability to perform activities of daily living (APPADL) were administered at baseline (BL), 26 and 52 wks. Pts in all groups achieved significant (p<0.05) LS mean changes from BL in total DTSQ and IW-SP scores at 26 and 52 wks. There were no significant LS mean changes from BL for any treatment group in total DSC-R and APPADL scores at 26 and 52 wks. LS mean improvements (decreases in score) from BL in frequency of perceived hyperglycemia (DTSQ) were observed in all groups at 26 and 52 wks, and improvement of hyperglycemia symptoms (DSC-R) was observed in both DU doses at 26 and 52 wks (Table). In conclusion, both once-weekly DU doses and MET demonstrated improvements from BL in DTSQ and IW-SP at 26 and 52 wks. Improvement of patient-perceived hyperglycemia was observed in both DU doses at 26 and 52 wks. Change from BL Score ITT, LOCF DTSQ, total DU 1.5 mg DU 0.75 mg MET DU 1.5 mg DU 0.75 mg (N=269) (N=270) (N=268) (N=269) (N=270) 26 wks 26 wks 26 wks 52 wks 52 wks 1.93 1.81 2.04 1.82 1.29 (0.39)‡ (0.38)‡ (0.39)‡ (0.44)‡ (0.43)‡ IW-SP, total 0.72 0.63 0.79 0.45 0.61 (0.19)‡ (0.18)‡ (0.18)‡ (0.19)‡ (0.19)‡ DSC-R, total 0.61 -0.40 1.02 1.23 1.05 (0.89) (0.88) (0.88) (0.99) (0.97) APPADL, total 0.09 0.19 0.02 0.39 -0.05 (0.33) (0.32) (0.32) (0.33) (0.33) -1.09 -0.86 -1.09 -1.10 DTSQ (Hyperglycemia) -1.25 (0.14)‡ (0.14)‡ (0.14)‡# (0.14)‡# (0.14)‡# DSC-R (Hyperglycemia) -0.20 -0.20 -0.02 -0.21 -0.16 (0.07)‡# (0.06)‡# (0.06) (0.07)‡# (0.07)‡ Data presented are LS mean (SE). ‡, # 2-sided p<0.05 vs. BL, and MET, respectively. Comparing Effectiveness of Prandial Insulin and Glucagon-Like Peptide-1 Treatment Regimens in Patients With Type 2 Diabetes on Background Basal Insulin in a Real-World Setting in the United States ANDRES DIGENIO, SUDEEP KARVE, SEAN CANDRILLI, MEHUL DALAL, Bridgewater, NJ, Research Triangle Park, NC This study documented demographic characteristics and outcomes among patients with type 2 diabetes mellitus (T2DM) initiating prandial insulin (INS) or glucagon-like peptide-1 (GLP-1) analog while on basal INS therapy. Electronic medical records (EMR) of patients with T2DM aged ≥18 years who had an EMR activity of ≥6 mo prior to, and ≥12 mo after their prandial INS or GLP-1 analog initiation date, were analyzed. Primary study endpoints included glycated hemoglobin (HbA1c), body weight, and health care resource utilization. The unmatched cohort comprised 1,962 GLP-1 analog users and 31,848 prandial INS users. Differences in patient characteristics were controlled by matching the groups on baseline variables including age, HbA1c, and body mass index. The resulting matched cohort comprised 1,143 GLP-1 analog users and 5,557 prandial insulin users, with the following mean baseline characteristics: age of 56.0 and 56.6 years; HbA1c of 8.5% and 8.6%; and weight of 111.1 and 108.9 kg, respectively. The outcomes of the matched analysis are presented in the Table. This study showed that, in a real-world setting, the addition of a GLP1 analog to basal INS in patients with T2DM was associated with similar glycemic control, but significantly higher reduction in body weight, lower incidence of hypoglycemia, and lower health care resource utilization compared with prandial INS. MET (N=268) 52 wks 1.94 (0.44)‡ 0.75 (0.19)‡ 1.48 (0.97) 0.28 (0.33) -0.66 (0.14)‡ -0.03 (0.07) & For author disclosure information, see page 829. A264 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES TRAcP5b (U/L) Baseline Week24 Change P value Lumbar BMD Baseline (g/m2) Week24 Change P value 2.879±1.243 3.181±1.322 0.302±0.149 0.461 0.980±0.040 1.003±0.039 0.024±0.016 0.679 3.415±1.86 3.539±1.076 -0.124±0.395 0.786 0.895±0.053 0.886±0.058 -0.006±0.017 0.892 2.529±1.216 3.115±1.170 0.586±0.183 0.126 0.926±0.029 0.929±0.029 -0.001±0.016 0.943 0.195 0.551 0.5580.303 0.129 0.382- 1031-P Patient-Reported Outcomes With Dulaglutide, Exenatide, or Placebo (AWARD-1) 1030-P Effect of Exenatide, Insulin and Pioglitazone on Bone Metabolism in Newly Diagnosed T2DM RENYUAN LI, HAIXIA XU, WEN XU, JINHUA YAN, BIN YAO, GUOCHAO ZHANG, JIANPING WENG, Guangzhou, China Preclinical studies suggested that insulin, incretin and thiazolidinediones had effect on regulation of bone metabolism. But clinical data is limited. We assessed the effects of these antihyperglycemic agents on bone metabolism in newly diagnosed T2DM. Sixty-two newly diagnosed and drug naïve T2DM were randomized to exenatide (20), mixed Protamine Zinc Recombinant Human Insulin Lispro Injection (25R) (21) or pioglitazone (21) group for 24-week treatment. HbA1c, bone turnover markers, including osteoclasin (OC), C-telopeptide of type I collagen (CTX) and tartrate-resistant alkaline phosphatase 5b (TRAcP5b), body mineral density (BMD) and fat mass index (FMI) were assessed at baseline and week 24. Baseline characteristics were similar among groups (Table 1). At week 24, HbA1c improved in all patients. Patients treated with exenatide lost body weight remarkably. Meanwhile, there was no significant change of body weight in insulin or pioglitazone group, maybe for lifestyle improvement. Greater reduction of FMI was found in exenatide group compared with that in insulin and pioglitazone group. But no significant improvement of OC, CTX, TRAcP5b or BMD was observed at week 24. 24-week treatment of exenatide, insulin and pioglitazone had improved glucose control in newly diagnosed T2DM. But with 24-week treatment, no significant change of OC, CTX, TRAcP5b or BMD was demonstrated. Among the three agents, exenatide showed beneficial effects on body weight and FMI. Table1 Change from BL Score DU 1.5 mg ITT, LOCF, LS Mean (SE) (N=279) 26 wks DTSQ, total 2.40 (0.34)‡*# IW-SP, total 0.56 (0.15)‡ APPADL, total 0.18 (0.27) EQ-5D Index 0.01 (0.01) EQ-5D VAS 4.50 (0.85)‡* DTSQ (Hyperglycemia) -1.93 (0.13)‡*# DTSQ (Hypoglycemia) 0.11 (0.11)# Comparison of treatment groups at baseline, after treatments and change from baseline Variable Exenatide Insulin Pioglitazone Pvalue Age (years) 45.7±2.3 53.0±2.4 51.3±1.8 0.147 Gender,M/F 13/7 10/11 9/12 0.230 HbA1c Baseline 8.4±0.2 8.7±0.2 8.6±0.2 0.759 (%) Week24 6.0±0.1 6.4±0.1 6.6±0.2 0.265 Change -2.4±0.3 -2.4±0.3 -2.0±0.2 0.200P value <0.001 <0.001 <0.001 Body Baseline 72.6±2.3 67.6±2.3 67.1±2.7 0.893 weight Week24 67.8±2.2 66.9±2.4 66.1±2.8 0.177 (Kg) Change -4.7±0.8 -0.7±0.8 -1.1±0.9 0.021 P value 0.013 0.774 0.682 FMI Baseline 8.0±0.6 7.6±0.5 8.1±0.4 0.684 (Kg/m2) Week24 7.1±0.7 7.5±0.5 8.3±0.5 0.227 Change -0.9±0.3 -0.1±0.2 0.2±0.2 0.009P value 0.025 0.930 0.778 OC Baseline 11.658±4.169 11.535±6.745 11.226±4.977 0.639 (ng/ml) Week24 11.040±5.465 12.172±6.058 11.077±5.273 0.825 Change -0.619±0.728 0.637±0.787 -0.150±0.691 0.289P value 0.690 0.731 0.926 CTX Baseline 0.634±0.332 0.825±0.415 0.673±0.346 0.227 (ng/ml) Week24 0.781±0.416 0.813±0.380 0.746±0.452 0.631 Change 0.147±0.046 -0.012±0.074 0.073±0.094 0.361P value 0.225 0.926 0.551 ADA-Funded Research & ‡, *, #2-sided DU 0.75 mg (N=280) 26 wks 2.56 (0.33)‡*# 0.47 (0.15)‡ 0.12 (0.27) 0.01 (0.01) 2.41 (0.85)‡ -1.77 (0.13)‡*# 0.16 (0.11)# EX PL (N=276) (N=141) 26 wks 26 wks 0.85 0.49 (0.33)‡ (0.45) 0.46 0.45 (0.15)‡ (0.20)‡ 0.47 0.03 (0.27) (0.36) 0.00 0.00 (0.01) (0.02) 3.94 0.71 (0.85)‡* (1.15) -1.10 -0.47 (0.13)‡* (0.18)‡ 0.48 0.27 (0.11)‡ (0.15) DU 1.5 mg (N=279) 52 wks 2.05 (0.36)‡# 0.50 (0.15)‡ 0.18 (0.29) 0.02 (0.01) 5.15 (0.89)‡ -1.89 (0.14)‡# -0.05 (0.11)# DU 0.75 mg (N=280) 52 wks 2.11 (0.36)‡# 0.47 (0.15)‡ -0.18 (0.29) 0.01 (0.01) 3.52 (0.89)‡ -1.63 (0.14)‡# 0.03 (0.11)# EX (N=276) 52 wks 0.69 (0.36) 0.64 (0.15)‡ 0.35 (0.29) 0.00 (0.01) 3.51 (0.89)‡ -1.13 (0.14)‡ 0.34 (0.11)‡ p<0.05 vs BL, PL and EX, respectively 1032-P The Effect of Prolonged Glucagon-Like Peptide-1 Receptor Stimulation on Gastric Emptying and Glycemia MAHESH M. UMAPATHYSIVAM, MICHAEL Y. LEE, KAREN L. JONES, CHRISTOPHER ANNINK, CAROLINE E. COUSINS, RAJ SARDANA, LAURENCE TRAHAIR, MARIANNE CHAPMAN, CHRISTOPHER K. RAYNER, MICHAEL A. NAUCK, MICHAEL HOROWITZ, ADAM M. DEANE, Adelaide, Australia, Bad Lauterberg, Germany Exogenous glucagon-like peptide-1 (GLP-1) and GLP-1 agonists, when given acutely slow gastric emptying (GE), which is the major mechanism for reducing postprandial glycemia. There is evidence that this slowing of GE is attenuated with prolonged stimulation of the GLP-1 receptor. We aimed to determine whether such attenuation occurs with prolonged GLP-1 infusion, and if it can be avoided with intermittent GLP-1 dosing. Subjects were randomised to receive 3 regimens of IV GLP-1 (0.8 pmol. kg-1.min-1); (i) ‘Prolonged’ - a continuous 24 hour infusion, (ii) ‘Intermittent’ two 4.5 hour infusions separated by 19.5 hours of 0.9% NaCl, (iii) ‘Acute’ For author disclosure information, see page 829. Guided Audio Tour poster A265 POSTERS This 52-week (wk), Phase 3, randomized study compared once weekly, long acting GLP-1 receptor agonist dulaglutide (DU) 1.5 mg and 0.75 mg to twice daily exenatide (EX) and placebo (PL, 26 wks only) in patients (pts) with type 2 diabetes. Both DU doses were superior to PL (26 wks) and EX (26 and 52 wks) in A1C change from baseline (BL; p<0.001). Weight reductions were observed in DU 1.5 mg and EX. Patient reported outcome measures for treatment satisfaction (DTSQ), weight related self-perception (IW-SP), ability to perform activities of daily living (APPADL), and health status (EQ-5D) were administered at BL, 26, and 52 wks. Significant (p<0.05) improvements from BL were observed in IW-SP (26 and 52 wks, all groups) and total DTSQ scores (26 wks, both DU doses and EX; 52 wks, both DU doses). A decrease in frequency of perceived hyperglycemia (all groups) was observed at 26 and 52 wks, but an increase in perceived hypoglycemia was observed with EX at 26 and 52 wks. EQ-5D visual analog scale (VAS) score showed significant improvements for both DU doses and EX at 26 and 52 wks. No significant changes were observed in APPADL and EQ-5D index scores for any group. Some significant between group differences were observed (Table). In conclusion, both once weekly DU doses and EX showed improvements from BL for IW-SP, total DTSQ, perceived hyperglycemia, and EQ-5D VAS scores. Improvement in total DTSQ and perceived hyperglycemia scores were greater in both DU doses vs EX. Clinical Diabetes/ Therapeutics MATTHEW REANEY, PING WANG, MARK LAKSHMANAN, KATE VAN BRUNT, BRAD CURTIS, BETH MITCHELL, Indianapolis, IN CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES POSTERS Clinical Diabetes/ Therapeutics a 4.5 hour infusion and Placebo (0.9% NaCl). GE of a mashed potato meal (643kcal), labeled with 20MBq 99mTc-calcium phytate, was measured by scintigraphy in 10 fasted healthy men (24±3yr). GLP-1 acutely slowed GE compared to placebo (P<0.01). Slowing of GE was attenuated with prolonged infusion (P=0.02), whereas slowing was maintained with intermittent dosing. The magnitude of the reduction in postprandial glycemia was diminished with prolonged infusion, but maintained with intermittent infusion, of GLP-1 (P=0.01). Sustained receptor activation attenuates, but does not abolish slowing of GE induced by GLP-1. This is consistent with the concept that short acting GLP-1 agonists are superior to long acting GLP-1 agonists in minimizing postprandial glycemic excursions. Supported by: Novo Nordisk A/S 1034-P Quantification of the Benefi t-Risk Relationship Between Glycemic Control and Hypoglycemia: A Comparison of Exenatide Once Weekly With Titrated Insulin Glargine JENNY HAN, MING ZHOU, JARET MALLOY, LARRY SHEN, San Diego, CA The DURATION-3 trial reported significantly greater A1C reduction and lower risk of hypoglycemia (hypo) with the GLP-1 receptor agonist exenatide once weekly (EQW) compared with insulin glargine (IG) over 3 years. As the risk of hypo typically increases when better glucose control is achieved, these analyses further quantify the benefit-risk relationship between A1C and hypo by baseline characteristics in 233 EQW and 223 IG patients (mean baseline A1C 8.3%). Exposure-adjusted event rates (EAER) of hypo in relation to the lowest or the weighted average A1C over 26 weeks, 1 year, and 3 years were evaluated by the Poisson regression model. Time to first hypo episode with EQW compared with IG, adjusted for endpoint A1C, was evaluated by the Cox proportional hazard model. Treatment (EQW or IG), concomitant antidiabetes therapy (with or without SFU), BMI, duration of diabetes, and A1C (lowest, endpoint, or weighted average A1C) were statistically significant (P<0.05) factors that impacted the risk of hypo. The EAER of hypo by week 26 with EQW, adjusted for lowest A1C, was 0.37 of that observed with IG (figure). The estimated survival function (probability of no hypo) at week 26, adjusted for endpoint A1C (figure) provided a hazard ratio of hypo between EQW and IG of 0.33. In conclusion, although risk of hypo increased with the benefit of decreased A1C, such risk was significantly reduced with EQW compared with IG. 1033-P The Effect of Liraglutide on A1c and Body Weight Is Largely Dependent on Diabetes Duration JOCHEN SEUFERT, TIMOTHY S. BAILEY, CLAUS B. SVENDSEN, MORTEN DONSMARK, MICHAEL A. NAUCK, Freiburg, Germany, Escondido, CA, Søborg, Denmark, Bad Lauterberg, Germany The LEAD phase 3 clinical program demonstrated liraglutide’s (lira) effectiveness across the continuum of type 2 diabetes (T2D). Using 26-28week data from 7 phase 3 trials, this pooled analysis evaluated the effect of baseline diabetes duration (DD) on changes in A1C and body weight (BW) from baseline with lira 1.8 mg (n=1581), lira 1.2 mg (n=1117) and placebo (n=524). DD ranged from <1 to >40 years (y), with a mean of ~8y for pooled groups. The slope for change in A1C or BW vs DD was determined by linear regression. There was a clinically nonrelevant trend for a greater A1C reduction with shorter DD with lira (Figure 1); statistical significance was achieved for pooled lira 1.2 mg (p<0.05) but the difference only equated to -0.2% A1C/10y shorter DD. The effect of lira on BW was independent of DD (Figure 2). In conclusion, the proven effectiveness of lira to reduce A1C and BW is largely independent of DD. Therefore, lira can be successfully used across the continuum of T2D. 1035-P Efficacy and Tolerability of Exenatide Twice Daily in Asian vs. White Patients With T2DM WAYNE H.H. SHEU, WENYING HUANG, STEVEN C. BRUNELL, ERICH BLASE, MARK HORNG, STEVE CHEN, Taichung, Taiwan, San Diego, CA, Taipei, Taiwan The 2012 ADA/EASD position statement on hyperglycemia management highlights the need for individualized treatment. Previous studies have shown that postprandial glucose concentrations in response to identical meals are greater in Asian than in White patients (pts). Exenatide twice daily & For author disclosure information, see page 829. A266 Guided Audio Tour poster ADA-Funded Research 1038-P Meta-Analysis of GLP-1 Agonist Lixisenatide Use in Patients Insufficiently Controlled With Oral Antidiabetic Drugs Suggests Improved Results in Efficacy, Hypoglycemia & Weight Gain 1036-P DENIS RACCAH, ANTONIO CERIELLO, PIERRE GOURDY, LUC SAGNARD, JAY LIN, JULIO ROSENSTOCK, Marseille, France, Barcelona, Spain, Toulouse, France, Paris, France, Flemington, NJ, Dallas, TX Effects of Pramlintide on A1C, Weight, and Hypoglycemia in Patients With Type 1 or Type 2 Diabetes: Subgroup Analysis by Duration of Diabetes The safety & efficacy of lixisenatide (LIXI) has been evaluated as addon to oral antidiabetic drugs (OADs) in patients with T2DM in 5 GetGoal Phase 3 studies (GetGoal-M, -P, -S, -M-Asia, and -F1). We report on a metaanalysis on composite efficacy and safety outcomes vs placebo. Baseline characteristics (no significant differences between LIXI & placebo (PBO) groups): mean age 55.5 yrs, mean BMI 31.5 kg/m2, duration of diabetes 7.4 yrs, duration of OAD use 4.5 yrs, metformin use 92.3%, SU use 35.6%, TZD use 16.5%. The total patients studied were 1828 LIXI & 932 PBO. A significantly greater proportion of LIXI patients achieved HbA1C <7% (odds ratio [OR] 2.66, [2.09, 3.38], p<0.0001). LIXI was also significantly better than PBO for the composite endpoints of HbA1C <7% & no weight gain (OR 2.64 [2.00, 3.48], p<0.0001), HbA1C <7% & no documented symptomatic hypoglycemia (OR 2.56 [2.04, 3.20], p<0.0001) & HbA1C <7%, no weight gain & no documented symptomatic hypoglycemia (OR 2.53 [1.94, 3.28], p<0.0001) (Table). Only 1 patient suffered documented severe hypoglycemia (0.05%) (in the LIXI group, add-on to SU). In summary, treatment with LIXI in patients with T2DM inadequately controlled on OADs significantly reduces HbA1C (-0.87% vs -0.33%; p<0.0001) & is 2.5 times more likely to result in HbA1C <7% with no documented symptomatic hypoglycemia and/or no weight gain than PBO. KATHRIN HERRMANN, STEVEN C. BRUNELL, KEVIN SHAN, STEVE CHEN, San Diego, CA Duration of disease affects various factors including insulin resistance and hypoglycemia awareness in patients with either type 1 (T1D) or type 2 (T2D) diabetes. This retrospective analysis evaluated the relationship between duration of diabetes and response to pramlintide treatment in T1D or T2D. Patients who received pre-meal pramlintide (30 or 60 mcg for T1D [n=714]; 120 mcg for T2D [n=292]); or placebo (n=537 vs T1D; 284 vs T2D) as an adjunct to insulin in 5 randomized trials were stratified by duration of diabetes tertile (Tert) within each diabetes type. Across all groups, baseline mean age ranged from 37 to 61 years, A1C from 8.6% to 9.5%, and duration of diabetes from 6 to 30 years. Across all Terts (in both T1D and T2D), pramlintide reduced A1C more than placebo. Pramlintide was associated with weight loss across all Terts and diabetes type, whereas placebo was associated with weight gain (except for Tert 3 in T2D). Generally, the incidence (but not the event rate) of severe hypoglycemia was higher among pramlintidetreated patients than among placebo-treated patients. The incidence and event rate of severe hypoglycemia rose with increasing duration of T1D, but not T2D, in both pramlintide- and placebo-treated patients. These results indicate that, compared with placebo treatment, pramlintide treatment was associated with improved A1C and decreased body weight regardless of baseline duration of diabetes or diabetes type. Meta-analysis* results: composite endpoints LIXI (% PBO Odds Ratio 95% p value (OR) CI of of Pts) (% of Pts) Lixi vs PBO OR HbA1C <7%, no weight gain† 35.83 18.88 2.64 2.00, <0.0001 3.48 HbA1C <7%, no documented symp- 40.86 22.96 2.56 2.04, <0.0001 tomatic hypoglycemia‡ 3.20 2.53 1.94, <0.0001 HbA1C <7%, no weight gain, no doc- 33.53 18.24 umented symptomatic hypoglyce3.28 mia†,‡ *Meta-analysis includes GetGoal-M (NCT00712673), GetGoal-P (NCT00763815), GetGoal-S (NCT00713830), GetGoal-M-Asia (NCT01169779) and GetGoal-F1 (NCT00763451); †Δ Weight: Endpoint -Baseline≤0; ‡symptomatic hypoglycemia with BG <60 mg/dL OR=odds ratio; CI=confidence interval; HbA1C=glycated hemoglobin; PBO=placebo; LIXI=lixisenatide. Lixisenatide is currently in development for the treatment of type 2 diabetes mellitus 1037-P Does Liraglutide Therapy Affect the Metabolic Response in Patients With Elevated Alanine Aminotransferase and Type 2 Diabetes Mellitus?: The ABCD Nationwide Liraglutide Audit PIYA SEN GUPTA, KEN Y. THONG, MATTHEW ARMSTRONG, PHILIP N. NEWSOME, PETER WINOCOUR, REJ RYDER, Birmingham, United Kingdom, Welwyn Garden City, United Kingdom The aims were to evaluate the effect of 1) liraglutide on metabolic response in patients with elevated ALT; 2) baseline ALT on metabolic response to liraglutide. Data was obtained from the UK ABCD audit of liraglutide in realclinical use (2009-2012, n=5650). Inclusion criteria: patients with baseline ADA-Funded Research & Supported by: Sanofi For author disclosure information, see page 829. Guided Audio Tour poster A267 POSTERS and follow-up ALT (>6 weeks). Exclusion criteria: liraglutide cessation. Descriptive statistics were performed (expressed as %frequency, mean±SD, median(IQR)). Patients were categorised into normal or abnormal baseline ALT groups according to gender-based levels (normal ALT males ≤30.0, females ALT≤19.0U/l). Changes in ALT, weight and HbA1c over time were calculated within and between ALT groups (ANCOVA, Wilcoxon tests). Spearman’s correlation was used to assess the relationship between baseline variables including ALT and metabolic response. Of 1309 patients (aged 55.5±11.3years, 54.8% male, 66.6% Caucasian, diabetes duration 9.0(6.0-13.0 years)), baseline ALT was 28(21-42U/l), weight 110.1±22.3kg, BMI 38.7±7.3kg/m2 and HbA1c 9.1±1.70%. Over 8.5 months(5.1-13.0), median ALT reduction was 1.0U/l(-8.0 to 4.0), BMI 0.8kg/ m2(-1.7-0.0) and HbA1c 0.7%(-1.7-0.1) (all P<0.0001). Comparing the normal baseline ALT group (n=524, 40.0%) to high ALT group (n=784, 59.9%), ALT changed from 19.0(16.0-24.0) to 20.0(15.0-26.0) (slight rise; P<0.001) and 38.0(29.0-50.0) to 33.0U/l(25.0-47.0) (P<0.0001), respectively. Baseline ALT did not correlate with weight or HbA1c response but correlated with ALT change (correlation coefficient -0.40, P<0.0001). Baseline HbA1c and weight did not correlate with ALT response. This analysis of serum ALT may have implications regarding non-alcoholic fatty liver disease, associated with T2DM. We conclude that liraglutide in real clinical use has a clinically significant impact on ALT reduction in T2DM patients with a high baseline ALT. (EBID; administered before the 2 main meals of the day) has been shown to potently reduce postprandial glucose in pts with T2DM. The current post hoc analysis explored the effects of EBID in Asian (N=787) and White (N=2223) pts using data pooled from 21 randomized, controlled studies (duration, 1230 weeks). At baseline, Asian pts had a lower body weight (71 vs 95 kg), higher A1C (8.4% vs 8.1%), lower FPG (166 vs 175 mg/dL), and more SFU use (75% vs 42%) than White pts. Self-monitored blood glucose data showed that, at baseline, Asian pts had higher postprandial glucose concentrations. At endpoint, EBID decreased preprandial glucose in both groups and had a greater effect on postprandial glucose in Asian pts. Changes in A1C differed at endpoint in Asian and White pts (-1.1% and -0.9%). Nausea was reported in 27% of Asian and 36% of White pts. In both Asian and White pts, hypoglycemia was more common in pts treated with an SFU (31% and 19%) than in those not treated with an SFU (3% and 10%). EBID was effective and tolerated in pts of both races; the reduction in postprandial glucose levels with EBID is of particular relevance to Asian patients considering their higher postprandial glucose excursions. Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES significant difference in weight loss, A1C, lipid profile or blood pressure when GLP-1 analogs were used in the management of patients with type 2 diabetes who participated in intensive weight management program. 1039-P Perioperative Liraglutide Therapy for Japanese Subjects With T2DM POSTERS Clinical Diabetes/ Therapeutics NAOKO KATAGIRI, SHIZUKA KANEKO, YUMIKO TAHARA, YUICHI SATO, KUMIKO HAMANO, Takatsuki, Japan, Kawasaki, Japan 1041-P Diabetic patients with limited exercise during perioperative period tend to gain weight, and their glycemic control frequently deteriorates. However, temporary insulin therapy has a risk of hypoglycemia. To investigate the efficacy and safety of perioperative liraglutide therapy, we performed a retrospective case analysis. 61 Japanese subjects with T2DM (male/female;39/22, Age;67.7±12.8, HbA1c;8.2±2.0, BMI;25.6±3.9) were initiated with liraglutide therapy before elective operations, e.g. orthopedic operations (23 cases), cardiac catheterizations for ischemic heart disease (21), cataract (12), prostate cancer (2), and breast cancer (1). In case of hyperglycemia induced by stress hormones during operation, regular insulin was added. As liraglutide decreases gastrointestinal motility, surgical cases with gastrointestinal tracts were excluded. BMI change, glucose profile and perioperative complications were analyzed. The BMI decreased (-1.7±0.7) before operation. Liraglutide therapy achieved good glycemic control throughout the perioperative period (Fig.). Additional regular insulin was not needed except 5 patients. HbA1c stayed in good range (6.8±0.8 after 6 months). Hypoglycemic episodes, wound healing retardations, or other complications were not observed. Liraglutide provides an effective and optional way to safely achieve good glycemic control in perioperative subjects with T2DM, especially those with limited exercise ability and at risks of hypoglycemia. The Novel Long Acting GLP-1/Glucagon Dual Agonist HM12525A Reduces Body Weight and Improves Glycermic Control in Rodent Models SUNG YEOB JUNG, JIN SUN KIM, IN YOUNG CHOI, KYU HANG LEE, YOUNGMI LEE, YOUNG HOON KIM, JAHOON KANG, SE CHANG KWON, Hwaseong-si, Republic of Korea, Seoul, Republic of Korea Oxyntomodulin is believed to exert its biological effects by activating both the GLP-1 and the glucagon receptor. The combined mechanism of food intake inhibition and increased energy expenditure is expected to enhance body weight loss. In addition, lipid metabolism is also expected to improve via the hypolipidemic effects of glucagon. HM12525A was developed by conjugating a novel GLP-1/Glucagon dual agonist with the constant region of the human IgG via a non peptidyl linker. HM12525A (LAPS-GLP-1/Glucagon dual agonist) is a potent dual agonist with well balanced affinities for the individual receptors. It has a pK profile compatible with once weekly or less frequent dosing regimens. The pharmacological effects of HM12525A were studied in HF DIO mice and in db/db mice. The HF DIO mice were treated once weekly sc with HM12525A (1, 3, and 5 nmol/kg) for 2 wks, and 100 nmol/kg of liraglutide was treated daily as an acitive comparator. HM12525A induced dose dependent body weight loss (-3, -10, -31%) compared to that of liraglutide (-17%). Also, the glucose lowering efficacy was evaluated in db/db mice by once-a-week administration of 6 nmol/kg HM12525A for 4 wks, and 60 and 100 nmol/kg of liraglutide was administered daily. The result showed that 6 nmol/kg of weekly HM12525A had comparable HbA1c reduction to the 100 nmol/kg of daily liraglutide. In conclusion, HM12525A showed potent body weight loss as well as glucose lowering efficacy in rodent models. 1040-P The Role of GLP-1 Analogs in Long-Term Weight Management Within a Multidisciplinary Intensive Lifestyle Intervention for Patients With Type 2 Diabetes OSAMA HAMDY, NUHA ELSAYED, AMR MORSI, ADHAM MOTTALIB, MARTIN J. ABRAHAMSON, Boston, MA 1042-P GLP-1 analogs use is associated with reductions in A1C and body weight. To investigate the impact of their use within a multidisciplinary weight management program, we evaluated body weight, A1C and other cardiometabolic outcomes in 104 obese patients with type 2 diabetes who enrolled in the Weight Achievement and Intensive Treatment (Why WAIT) program, a 12-week weight management program, focused on intensive lifestyle modifications. At entry, mean weight was 111.3±18.6 kg; BMI 38.7±5 kg/m², age 54.5±10.1 yrs and A1C 7.4±1.2%. Fifty five patients were started on GLP-1 analogs at the beginning of the program. Only 35 remained on GLP1 analogs for the 4 years of follow up (group A) and 49 patients never used GLP-1 analogs (group B). At baseline no difference in BMI (39±5.7 vs. 38.2±4.7 kg/m²), A1C (7.2±1.2 vs. 7.4±1.3%), diabetes duration, blood pressure or lipid parameters were noted between the 2 groups. At 12 weeks, weight loss was higher in group A but not significantly different from group B (-12.1±5.3 vs. 10.5±3.9 kg). A1C followed the same trend (6.2±0.7 vs. 6.5±0.9%). Weight loss in group A over 4 years of follow up tends to be higher (-10.1, -7.8, -8.0, -8.4 kg respectively) but was not significantly different from group B (-7.9, -6.2, -6.2, -6.7 kg respectively). Similar observation was seen in A1C. At 4 year, A1C in group A was 6.9±1.3 vs. 7.5±1.3% in group B (p=0.07). LDL was significantly lower in group A vs. group B at 3 and 4 yrs (86.4 vs.100 mg/dL, p<0.05) but other lipid parameters and blood pressure were not significantly different over the 4 years. In conclusion, except for LDL there was no WITHDRAWN & For author disclosure information, see page 829. A268 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES ΔQTcI and ΔQTcF. The resulting mean placebo-corrected ΔQTcI (ΔΔQTcI) was small with the largest ΔΔQTcI of 1.1 msec (upper bound of 90% CI: 3.8 msec) on Day 4 (Albi 30 mg) and -0.6 msec (upper bound of CI: 1.8 msec) on Day 39 (Albi 50 mg). Results from the analysis of QTcF were consistent with QTcI. Moxifloxacin caused a peak effect on ΔΔQTcI of 10.9 msec and the lower bound of the CI was above 5 msec at all preselected time points, which demonstrated assay sensitivity. The slope of the relationship between Albi plasma levels and ΔΔQTcI was negative, demonstrating that increasing exposures of Albi were not associated with QT prolongation. Albiglutide was well-tolerated and there were no clinically relevant differences in treatment safety data between Albi and P. In conclusion, Albi at doses up to 50mg in healthy subjects did not prolong the QTc interval in a clinically relevant way. Basal+LIXI vs basal-bolus: composite endpoints Endpoints Basal+LIXI BB Odds Wald 95% CI p value n=519* n=1165† ratio % patients Regression analysis (unadjusted) HbA1C <7% and no documented 28.32 18.63 3.71 2.52−5.48 p<0.0001 symptomatic hypoglycemia HbA1C <7% and no weight gain 27.94 14.55 3.63 2.47−5.32 p<0.0001 HbA1C <7%, no documented symptom19.65 5.34 8.95 5.26−15.25 p<0.0001 atic hypoglycemia and no weight gain *Includes GetGoal-Duo1 (NCT00975286) and GetGoal-L (NCT00715624); †Includes HMR1964A/3002, HMR1964A/3005, HMR1964A/3502 (NCT00135057) and HMR1964A/3511 (NCT00135083); BB=basal-bolus; CI=confidence interval; HbA1C=glycated hemoglobin; documented symptomatic hypoglycemia=blood glucose <3.33 mmol/L. 1045-P An adaptive, double-blind, placebo (PL)- and sitagliptin (sita)-controlled Phase 3 trial with dose-finding, efficacy and safety objectives to assess dulaglutide (DU), a long-acting GLP-1 receptor agonist, in type 2 diabetes mellitus (T2DM). We present the dose-finding results that selected doses for the DU clinical development plan. Patients with T2DM (A1c ≥7.0% to ≤9.5%), treated with diet and exercise only, oral antihyperglycemic medication (OAM) monotherapy, or metformin (MET) combined with OAM were randomized 1:1:3 to PL, sita 100 mg QD, or 1 of 7 once-weekly DU doses (0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 3.0 mg), after stabilization on MET monotherapy. Of 1202 randomized patients, the initial 230 participated in dose-finding. Allocation to DU occurred with adaptive randomization, updated bi-weekly with accumulating data for a prespecified efficacy and safety composite endpoint based on model predicted A1c, weight, pulse rate, diastolic blood pressure (Table). The algorithm continued adaptive randomization until 2 doses were selected that demonstrated clinically meaningful benefit based on prespecified selection criteria including ≥ two-fold concentration difference. A Data Monitoring Committee discontinued DU 3.0 mg before dose selection. The adaptive algorithm selected DU 0.75 and 1.5 mg; randomization continued to DU 0.75 and 1.5 mg and comparators to assess long-term efficacy and safety. 1043-P The Novel Anti-Glucagon Spiegelmer NOX-G16 Ameliorates Hyperglycemia in a Murine Type 1 Diabetes Model AXEL VATER, PRZEMYSLAW KACZMAREK, EWA PRUSZYNSKA-OSZMALEK, SIMONE SELL, CHRISTIAN MAASCH, KLAUS BUCHNER, PAWEL KOLODZIEJSKI, WERNER G. PURSCHKE, KRZYSZTOF W. NOWAK, MATHIAS Z. STROWSKI, SVEN KLUSSMANN, Berlin, Germany, Poznan, Poland In the absence of postprandial insulin increases in type 1 diabetes mellitus (T1DM), excessive secretion of glucagon leads to abnormally high glucose production in the liver thus contributing to postprandial hyperglycemia. We have identified NOX-G16, a PEGylated mirror-image (L-)oligonucleotide (a so-called Spiegelmer ®) that binds glucagon with an affinity of 1.7 nM and blocks glucagon-induced cAMP generation in vitro with IC50 of 2.4 nM. Due to its mirror-image configuration NOX-G16 is not cleaved by plasma DNases or RNases and exhibits a plasma half-life of approx. 6 h in mice. NOX-G16 was administered ip once daily to mice with streptozotocin (STZ)-induced insulinopenic diabetes as a model for T1DM. In intraperitoneal glucose tolerance tests (ipGTT) performed 2 h after dosing on days 1, 3 and 7 glucose excursions were dose-dependently diminished as seen by reduced peak glucose levels and a significantly lower area under the curve for glucose, compared to vehicle treatment. Endogenous gluconeogenesis was significantly diminished. Free fatty acids were slightly and ketone bodies were clearly reduced, while plasma triglycerides were comparable to vehicle-treated animals. Importantly, neither healthy controls nor T1DM mice experienced hypoglycemia. In summary, having accomplished anti-diabetic proof of concept in this model of T1DM, NOX-G16 may be a drug candidate for insulin sparing in T1DM patients. Additional pre-clinical work in T2DM is warranted. Model Adjusted Mean (95% CI) Changes from Baseline Treatment A1c Weight Pulse Rate (%) (kg) (beats/min) 52-week sitagliptin- 26-week placebo- 26-week placeboadjusted adjusted adjusted 0.25 mg -0.12 (-0.38; 0.20) -0.49 (-1.36; 0.44) -1.16 (-4.40; 1.90) dulaglutide n=13 n=16 n=16 0.50 mg -0.24 (-0.44; 0.02) -0.97 (-1.80; -0.12) -1.32 (-4.30; 1.80) dulaglutide n=16 n=22 n=22 0.75 mg -0.23 (-0.44; 0.08) -0.44 (-1.16; 0.44) -1.80 (-5.50; 2.10) dulaglutide n=20 n=20 n=20 1.00 mg -0.30 (-0.58; 0.08) -1.53 (-2.44; -0.60) -0.71 (-4.00; 2.50) dulaglutide n=7 n=9 n=9 1.50 mg -0.63 (-0.98; -0.20) -1.67 (-2.48; -0.92) -0.07 (-3.20; 2.70) dulaglutide n=18 n=22 n=22 2.00 mg -0.58 (-0.76; -0.32) -1.99 (-2.88; -1.20) 0.78 (-2.10; 3.80) dulaglutide n=23 n=29 n=29 3.00 mg -0.29 (-0.62; -0.02) -3.93 (-4.84; -2.88) 2.98 (-0.60; 8.70) dulaglutide n=9 n=13 n=13 1044-P Albiglutide Does Not Prolong the QTc Interval in Healthy Subjects: A Thorough QT/QTc Study BORJE DARPO, MEIJIAN ZHOU, JESSICA MATTHEWS, HUI ZHI, CAROLINE R. PERRY, RICKEY REINHARDT, Solna, Sweden, Rochester, NY, Research Triangle Park, NC, Upper Merion, PA Supported by: Eli Lilly and Company The effect of albiglutide (Albi), a selective once-weekly GLP-1 receptor agonist being developed for the treatment of T2DM, on cardiac repolarization (QTc interval), was assessed in a randomized, double blind, placebo (P) controlled, parallel group study in healthy subjects with a nested cross-over comparison for moxifloxacin (MX). Subjects were randomized to Albi (n= 78) or placebo (n= 79) and received injections of 30 mg Albi/P on Days 1 and 8 and injections of 50 mg Albi/P on Days 15, 22, 29, and 36. In the placebo group, moxifloxacin (MX) was administered on Day -1 in half of the subject and on Day 40 in the other half. Albi PK was drawn on Days 4 and 39 and serial ECG were extracted from continuous recordings on Day -2 (baseline), -1, Days 4, 39, and 40. Demographics were generally similar between Albi and P: mean age 2930 yrs; BMI 25 kg/m2. Mean change from Baseline (BL) in QTcI and QTcF on Albi 30mg (~ Cmax after 30mg SD, Day 4) and Albi 50mg (~ Cmax after 50mg RD, Day 39) was similar to the P response. MX caused a clear prolongation of ADA-Funded Research & Diastolic Blood Pressure (mm Hg) 26-week placeboadjusted 0.35 (-2.60; 3.60) n=16 0.11 (-2.40; 3.20) n=22 -0.27 (-3.00; 2.50) n=20 -0.55 (-3.90; 2.20) n=9 -0.62 (-3.40; 2.30) n=22 -0.41 (-2.60; 3.20) n=29 -0.19 (-2.70; 2.80) n=13 1046-P Injection Force (IF) of Rusable Insulin Pens in India: NovoPen® 4 (NP4), NovoPen® 3 (NP3), HumaPen® Ergo (HPE), Glaritus® Pen Royale (WR), INSUPen® (IP) and AllStar ® (AS) ARND FRIEDRICHS, VOLKER KORGER, STEFFEN ADLER, Brannenburg, Germany, Frankfurt, Germany The IF of an insulin pen is an important practical aspect of therapy for patients with diabetes mellitus, especially those with limited dexterity, and is a key element in the design and characteristics of insulin pens. We compared the IF of 6 reusable insulin pens available in India: NP4, NP3 (Novo Nordisk), HPE (Eli Lilly), WR (Wockhardt), IP (Biocon) and AS (Sanofi). There were 20 measurements per pen type per test series. Mean IF was measured by dispensing 60 U of insulin. IF was tested with dispense rates of 6 and For author disclosure information, see page 829. Guided Audio Tour poster A269 POSTERS ZACHARY SKRIVANEK, JENNY Y. CHIEN, BRENDA GAYDOS, MICHAEL HEATHMAN, MARY JANE GEIGER, ZVONKO MILICEVIC, Indianapolis, IN, Santa Clara, CA, Vienna, Austria Clinical Diabetes/ Therapeutics Dose-Finding Results in an Adaptive Trial of Dulaglutide Combined With Metformin in Type 2 Diabetes (AWARD-5) CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULIN INJECTABLES Meta-analysis results: composite endpoints* LIXI + basal PBO + basal Odds ratio (OR) 95% CI p insulin insulin LIXI + basal of OR value (n=665; % (n=533; % insulin vs PBO + of patients) of patients) basal insulin 26.32 17.26 2.65 1.30, 0.007 HbA1C <7%, no documented symptomatic hypoglycemia† 5.38 27.37 12.38 3.35 1.66, 0.0008 HbA1C <7%, no weight gain 6.77 HbA1C <7%, no weight gain, 18.50 9.76 2.64 1.48, 0.0009 no documented symptomatic 4.70 hypoglycemia *Includes GetGoal-Duo1 (NCT00975286), GetGoal-L (NCT00715624) and GetGoal-L Asia (NCT00866658); †symptomatic hypoglycemia with blood glucose <60 mg/dL; LIXI is currently in development for the treatment of type 2 diabetes mellitus LIXI=lixisenatide; PBO=placebo; CI=confidence interval; HbA1C=glycated hemoglobin Supported by: sanofi-aventis POSTERS Clinical Diabetes/ Therapeutics 10 U/s (constant volume flow rate). Pens were fitted with different needles as recommended by the manufacturers’ instructions (Becton Dickinson Micro-Fine® 31G x 5mm for HPE, WR, IP, AS and NovoFine® 31G 6mm for NP4 and NP3) and tested with different insulins: Lantus® for AS, Glaritus® for WR, Humulin® 30/70 for HPE, Basalog® for IP, NovoMix® 30 for NP3 and NovoRapid® for NP4. AS required a significantly lower mean IF vs NP3, HPE and IP (Table). Mean IF for NP4 and WR was similar to AS. Mean IF was 56, 49, 21 and 7% higher with NP3, HPE, IP, WR vs AS, respectively (p<0.05) at 6 U/s. Mean IF in all pens increased at 10 U/s, but IF remained lower for AS, NP4 and WR vs other pens (p<0.05). AS, NP4 and WR require a lower IF at different dispense rates vs other reusable insulin pens. This may enable patients, especially those with dexterity problems, to administer insulin more easily and improve management of their diabetes. 1048-P Efficacy and Safety of Lixisenatide in Japanese Patients With Type 2 Diabetes Mellitus: A Meta-Analysis of 2 Randomized Controlled Trials YUTAKA SEINO, NOBUYA INAGAKI, YUKIKO ONISHI, YUKIO IKEDA, KARIM ADMANE, JAY LIN, LUC VAN GAAL, Osaka, Japan, Kyoto, Japan, Tokyo, Japan, Paris, France, Flemington, NJ, Antwerp, Belgium The incidence of diabetes is increasing in Japan, with approximately 13.5% of the population now diagnosed with T2DM or impaired glucose tolerance. We conducted a meta-analysis of data from 2 randomized, double-blind trials to evaluate the efficacy & safety of adding lixisenatide (LIXI) to SU ± MET (GetGoal-S) or basal insulin therapy ± SU (GetGoal-L Asia) vs placebo (PBO) in Japanese patients with T2DM (LIXI, n=143; PBO, n=136). LIXI significantly reduced HbA1C (-1.08% vs PBO; p<0.0001), as well as FBG & PPG (-12.93 mg/dL [p=0.0072] & -149.82 mg/dL [p<0.0001] vs PBO, respectively). A significantly higher proportion of patients achieved HbA1c <7% with LIXI (31.47 vs 2.21% with PBO, p<0.0001). There was a trend towards weight reduction with LIXI (-0.3 kg vs PBO, p=0.22). The incidence of symptomatic hypoglycemia was 20.98% with LIXI vs 9.56% with PBO (all patients were treated with SU &/or basal insulin). There were no cases of severe hypoglycemia in either group. A significantly higher proportion of patients achieved a composite endpoint comprising HbA1C <7%, no weight gain & no documented symptomatic hypoglycemia in the LIXI group vs PBO group (Table). In Japanese patients with T2DM inadequately controlled on SU ± MET or basal insulin therapy, LIXI may improve glycemic control without weight gain and with no severe hypoglycaemia. LIXI may also enable a higher proportion of patients to achieve composite endpoints. Supported by: Sanofi 1047-P Meta-Analysis of Randomized Controlled Trials of Lixisenatide as Add-On to Basal Insulin in Patients With Type 2 Diabetes Mellitus BERNARD CHARBONNEL, EDWARD WANG, JAY LIN, MELANIE DAVIES, VIVIAN FONSECA, Nantes, France, Bridgewater, NJ, Leicester, United Kingdom, New Orleans, LA The safety and efficacy of lixisenatide (LIXI), a new once-daily glucagon-like peptide-1 receptor agonist, has been evaluated as add-on to basal insulin ± oral antidiabetic drugs (OADs) in 3 Phase III, randomized, placebo (PBO)controlled trials (GetGoal-L, -Duo1 and -L Asia) in T2DM. Endpoints included HbA1C, weight, insulin dose, fasting blood glucose, postprandial glucose and hypoglycemia. A meta-analysis was performed on the safety and efficacy outcomes in 1198 patients (mean age: 57.2 yrs; diabetes duration: 11.7 yrs; BMI: 30.3 kg/m2) enrolled to these trials, using a random effects model (RevMan). A significantly higher proportion of LIXI-treated patients achieved HbA1C <7% vs PBO (odds ratio [OR] 3.67 [1.64, 8.19] p=0.002). Furthermore, LIXI was more than 3× as likely to result in HbA1C <7% and no weight gain (OR 3.35 [1.66, 6.77] p=0.0008), and more than 2.5× as likely to result in either i) HbA1C <7% + no documented symptomatic hypoglycemia (OR 2.65 [1.30, 5.38] p=0.007), or ii) HbA1C <7% + no weight gain + no documented symptomatic hypoglycemia (OR 2.64 [1.48, 4.70] p=0.0009) (Table). In patients with T2DM, LIXI as add-on to basal insulin ± OADs is significantly more effective than PBO in achieving HbA1C <7%, and is more than 2.5 times as likely to result in HbA1C <7% with no documented hypoglycemia and no weight gain; it is, therefore, an attractive potential option as add-on to treatment with basal insulin. Analysis of composite endpoints* LIXI† PBO p value N % N % HbA1C 32 22.38 2 1.47 <0.0001 45 31.47 3 2.21 <0.0001 HbA1C HbA1C 36 25.17 3 2.21 <0.0001 25 17.48 2 1.47 <0.0001 HbA1C HbA1C 36 25.17 3 2.21 <0.0001 *Japanese subanalysis of GetGoal-S (NCT00713830) and GetGoal-L Asia (NCT00866658) trials; †Lixisenatide is currently in development for the treatment of type 2 diabetes mellitus; LIXI=lixisenatide; PBO=placebo; HbA1C=glycated hemoglobin; documented symptomatic hypoglycemia=blood glucose Supported by: Sanofi & For author disclosure information, see page 829. A270 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1049-P & Evaluation of Effect of Exenatide Treatment in Serum Calcitonin in Levels in Morbidly Obese Type 2 Diabetic Patient MUSTAFA DEMIRPENCE, ALIYE PELIN TÜTÜNCÜOGLU, DEVRIM DOLEK, FUSUN SALGUR, GONCA ORUK, MITAT BAHCECI, Izmir, Turkey DAVID J. KAZIERAD, JEFFREY A. PFEFFERKORN, ARTHUR BERGMAN, XIN WANG, TIMOTHY P. ROLPH, JAMES M. RUSNAK, Cambridge, MA, Groton, CT Background and aim: Exenatide provides a reduction in glycated hemoglobin with no weight gain. It was shown that exenatide might cause an increased incidence in thyroid C-cell tumors in rats. We aimed was to determine serum calcitonine levels during 6 monthly exenatide treatment period in type 2 diabetic patients. Subjects and methods: Obese (BMI≥35 kg/m2) newly diagnosed 33 type 2 diabetic patients aged with 48.4±12.9 yr were included to study. Exenatide therapy was started with 5µg b.i.d in addition to metformin (2 gr/day). At the 1st month of treatment an additional visit was scheduled for all subjects to verify the compliance to drug prescription and the occurrence of possible side effects. Exenatide dosage was increased to 10 µg b.i.d (20 µg/day) in appropriate patients at the 1st month of treatment. Serum calcitonin, fasting and postprandial glucose, amylase, A1c, lipid levels; and body weight and body mass index (BMI) were measured in three visits (baseline, 3rd and 6th month of treatment). Results: Serum calcitonin levels did not show any significant difference during with exenatide treatment (baseline: 2.44±4.17, 3rd month: 2.35 ±4.13, 6th month: 2.52±4.04 ). During the treatment period serum calcitonin levels remained in normal range. Whereas A1c levels (baseline: 8.17±1.83, 3rd month: 7.45 ±1.20, 6th month: 6.80±1.10), fasting (baseline: 186±59.7, 3rd month: 141 ±27.8, 6th month: 120±17.07) and postprandial glucose levels (baseline: 254±86.6, 3rd month: 183 ±43.8, 6th month: 143±20.4), body weight (kg) (baseline: 126.2±28.9, 3rd month: 120.7 ±27.7, 6th month: 116.2±28.2) and BMI (kg/m2) (baseline: 48.1±9.88, 3rd month: 45.9 ±9.58, 6th month: 44.1±9.48) showed significant improvements (p<0.05). Conclusion: Exenatide treatment could not give rise to increment in calcitonine levels at least 6 monthly period, whereas this treatment may ameliorate fasting and postprandial glucose, A1c levels and body weight in morbidly obese type-2 diabetic patients. PF-04991532 (PF) is a liver selective GKA in development for T2DM. Two 12-week, Phase 2, randomized, blinded studies investigated the efficacy and safety of PF in T2DM patients receiving stable background metformin therapy. In 1 study, patients were randomized to once daily oral administration of 1 of 3 doses of PF, placebo, or sitagliptin. The mean ranges for baseline demographics were similar across the dose groups: age 55-58 yrs; HbA1c 8.0-8.6%; duration of T2DM 7.1-8.4 yrs. The second study, of the same design and with similar baseline demographics, was conducted using a BID regimen at total daily doses of 50-600 mg. Results of both studies will be discussed, with the representative QD study data summarized below. Supported by: Pfizer, Inc. & 1052-P SLC47A1 Gene rs2289669 G>A Variant Influences the Hypoglycemic Effect of Metformin in T2DM Patients FANG LIU, DAN DAN ZHANG, TAI SHAN ZHENG, JUN LING TANG, HUI JUAN LU, WEIPING JIA, Shanghai, China 1050-P Background and aims: The SLC47A1 gene encodes for multidrug and toxin extrusion 1(MATE1) which plays a pivotal role in the metabolism and excretion of metformin. The present study is to investigate whether the SLC47A1 gene polymorphism influence the therapeutic effect of metformin in Chinese Hans who suffered from type 2 diabetes mellitus (T2DM). Methods: The SLC47A1 single nucleotide polymorphism (SNP) rs2289669 G/A was genotyped in 497 T2DM patients, including a metformin treated group (n = 224) and a non-metformin-treated group (n = 288) with PCR-restriction fragment length polymorphism method. HbA1c was followed up for 6 months after metformin treatment. Results: Three SLC47A1 genotypes, GA, GG and AA were found and the frequency of the SLC47A1 rs2289669 G/A and A allele was 46.99% in T2DM patients with the similar frequency in metformin group(47.1%) and non-metformin group(46.9%). The plasma lactate levels was significantly higher in AA carrier patients(1.18±0.47mmol/L ) than that in patients who carried GA(1.04±0.39mmol/L) and GG(1.04±0.34mmol/L, P=0.008), and the incidence of hyperlactatemia increased significantly in patients with SLC47A1 rs2289669 AA genotype(5.9%) compared to GA(3.4%) and GG genotype(1.5%, P<0.01). The 6-month follow-up of 250 patients revealed that there was greatest reduction of HbA1c in patients with SLC47A1 rs2289669 AA genotype than that of GA and GG genotype (ΔHbA1c -2.3% in AA genotype vs. -1.3% in GA, and -1.1% in GG genotype, P<0.01).Conclusions: The SLC47A1 gene rs2289669 G>A variant influences the hypoglycemic effect of metfotmin in Chinese T2DM patients, and the patients with AA genotype got better HbA1c reduction. SIVARAM PILLARISETTI, ISH KHANNA, La Chaux-de-Fonds, Switzerland Insulin resistance is the underlying defect in >90% of patients with type 2 diabetes mellitus and is the key pathologic mechanism for associated susceptibility to premature cardiovascular complications. Currently there is no safe insulin sensitizer in market that targets skeletal muscle insulin resistance. AMP-activated protein kinase (AMPK) is an energy-sensing enzyme that is expressed in many insulin-responsive tissues including liver and skeletal muscle. Activation of AMPK promotes glucose uptake in skeletal muscle and inhibits gluconeogenesis and lipogenesis in liver. We have developed a series of non-AMP mimetic, small molecule activators of AMPK. In cell based assays, a lead molecule KU-5039 activated AMPK with sub-micromolar potency compared to AICAR at 500 uM. KU-5039 exhibited excellent oral availability including distribution and AMPK modulation across targeted tissues (liver and skeletal muscle). In dbdb model of diabetes and insulin resistance, KU-5039 stabilized plasma glucose to near normal levels within 7 days of treatment. The plasma insulin levels were undetectable and insulin resistance, as measured by HOMA, was normalized by day 7. Unlike pioglitazone whose insulin sensitization was associated with weight gain, KU-5039 treatment of dbdb mice resulted in no weight gain. Similarly KU-5039 showed markedly improved insulin sensitization with no effect on plasma lactic acid levels versus metformin. The lead KU-5039 represents a promising approach for the treatment of insulin resistance in type 2 diabetes without the side effects associated with marketed insulin sensitizers. A detailed profile of the lead KU-5039 will be presented Supported by: NSFC (81070650) For author disclosure information, see page 829. Guided Audio Tour poster A271 POSTERS Sitagliptin 100 mg QD (N=54) After 12-weeks of dosing, 60.5% of patients receiving PF 750 mg QD doses achieved target HbA1c values <7%, compared to 16.3% and 39.6% in the placebo and sitagliptin groups, respectively. Overall, PF was safe and well tolerated across a range of QD and BID doses. There appeared to be doserelated trends for increases in LFTs and serum triglyceride values in the PF groups; and 3 subjects discontinued from the BID study due to increased TG values. Administration of PF-04991532, either QD or BID, resulted in clinically meaningful decreases in HbA1c and minimal episodes of hypoglycemia. Discovery of an Orally Efficacious, Novel AMPK Activator as Potent Insulin Sensitizer & PF-04991532 QD 150 mg 450 mg 750 mg (N=52) (N=54) (N=53) HbA1c (%)* 0.08 (-0.13, 0.29) -0.49 (-0.71, -0.28) -0.58 (-0.80, -0.36) -0.71 (-0.92, -0.50) Triglycerides (%)# 0.96 (0.83, 1.11) 1.12 (0.96, 1.30) 1.16 (1.00, 1.34) 0.92 (0.80, 1.07) Hypoglycemia** 1 (1.9) 1 (1.9) 2 (3.7) 0 (0.0) 2 (3.7) * Mean (80% CI) placebo-corrected change from baseline for the mixed model repeated measures analysis at 12 weeks # Geometric mean ratio to placebo (95% CI) change from baseline ** Episodes (% of total N) occurring over 12 weeks Guided Audio Tour: Innovative Oral Agents—Innovative Discoveries (Posters: 1050-P to 1057-P), see page 19. ADA-Funded Research Placebo (N=53) Clinical Diabetes/ Therapeutics Parameter CLINICAL THERAPEUTICS/NEW TECHNOLOGY— ORAL AGENTS & 1051-P The Hepatoselective Glucokinase Activator (GKA) PF-04991532, Lowers HbA1c after 12-Weeks of Dosing in Patients With Type 2 Diabetes (T2DM) CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS & 1053-P The Mechanism of the Anti-Diabetic Effect of Ranolazine: Inhibition of Glucagon Secretion in Rat Pancreatic Islets via Na+ Channel Blockade POSTERS Clinical Diabetes/ Therapeutics KRIS KAHLIG, MING YANG, CHENG XIE, SRIDHARAN RAJAMANI, ARVINDER K. DHALLA, LUIZ BELARDINELLI, Fremont, CA The anti-anginal drug Ranolazine (RAN) is a Na+ channel blocker that has been shown to reduce HbA1c levels in subjects with diabetes mellitus in clinical trials. However the mechanism of action of its anti-diabetic effects is unclear. Pancreatic α-cells express Na+ channels, which participate in generating the electrical activity necessary for glucagon secretion. We hypothesize that RAN inhibits Na+ current (INa) in pancreatic α-cells thus reducing glucagon secretion. The effect of RAN on INa and electrical activity in rat pancreatic α-cells was tested using perforated patch-clamp technique. RAN (10µM) reduced INa by 25±4% (-98.0±18pA to -74.1±15pA, n=7, p<0.01) and spontaneous electrical activity by 70±14% (99 ± 22 mV*s to 32 ± 6 mV*s, n=3, p<0.05). The reduced activity translated to a 36±6% reduction in glucagon secretion measured from intact islets in the presence of RAN. The role of INa was confirmed using selective (GS-458967) and broad spectrum (tetrodotoxin, TTX) Na+ channel inhibitors, which both significantly reduced α-cell INa and spontaneous electrical activity. Glucagon release from intact islets was inhibited by 3µM GS-458967 (53±8%) or 100nM TTX (47±6%). The INa activator veratridine (15µM) increased spontaneous electrical activity by 6-fold) and glucagon release from islets by 3.5-fold. Veratridine-evoked hyperexcitability was reduced by either 10µM RAN (290 ± 17 mV*s to 81 ± 35 mV*s, n=4, p<0.05) or 300nM GS-458967 (392 ± 42 mV*s to 55 ± 9 mV*s, n=3, p<0.05). Likewise, veratridine-evoked glucagon release was reduced by 10µM RAN (69±11%), 3µM GS-458967 (100%) or 100nM TTX (100%). In summary, Na+ channel blockers reduce electrical activity of α-cells and inhibit glucagon release. These data suggest the anti-diabetic effect of RAN is mediated by the inhibition of islet glucagon secretion secondary to the inhibition of α-cell INa. This mechanism of action reflects a novel target for the development of anti-diabetic therapeutics. & & 1056-P Combination Naltrexone/Bupropion Therapy Resulted in Clinically Meaningful Improvements in Weight and Quality of Life (QoL): Integrated Analysis of Four Phase 3 Trials RONETTE L. KOLOTKIN, COLLEEN BURNS, BRANDON WALSH, PRESTON KLASSEN, Durham, NC, La Jolla, CA Comprehensive treatment for obesity ideally will improve physical, psychological, and social aspects of health in addition to facilitating weight loss. This integrated analysis of the four Phase 3, 56-week trials of combination naltrexone sustained-release (SR) / bupropion SR (NB) vs placebo (PBO) investigated changes in body weight and weight-related QOL using the Impact of Weight on Quality of Life-Lite (IWQOL-Lite) scale at baseline and at 8, 16, 28, and 56 weeks of treatment. All patients (NB [n=2043], PBO [n=1319]) were overweight/obese (BMI: ≥27 and ≤45 kg/ m2). Treatment group differences (LS mean±SE) in the modified ITT-LOCF population (≥1 post-baseline weight while on study drug) were evaluated by ANCOVA with treatment, study, and baseline values as covariates. Baseline characteristics (mean±SD) were similar between treatment arms: 81% female, 79% Caucasian, 46±11 y, 36±4 kg/m2, type 2 diabetes: 12.6%, IWQOL-Lite total score: 71±18. Completion rate was 66% (NB) and 59% (PBO). Weight loss with NB was significantly greater than PBO at all time points (Week 56: -7.0±0.2% NB vs -2.3±0.2% PBO; p<0.001). At Week 56, NB resulted in significantly greater improvement in IWQOL-Lite total score (+11.9±0.3 vs +8.2±0.3; p<0.001) and all subscores (Physical function, Selfesteem, Sexual life, Public distress, Work; all p<0.05 vs PBO). At Week 56, significantly more patients reduced body weight by ≥5% with NB (53% vs 21%; p<0.001), and greater mean improvement in IWQOL-Lite total score was observed in these patients. The safety/tolerability profile of NB was consistent with its individual components; the most common adverse events were nausea (32%), constipation (19%), headache (18%), and vomiting (11%). Thus, NB improved both weight and psychosocial outcomes, with 36% of NB patients (vs 12% PBO; p<0.0001) achieving clinically meaningful improvement in both IWQOL-Lite and body weight (≥5% reduction) after 1 year. 1054-P Dose Response to Oral Insulin Capsules in Fasting, Healthy Subjects ROY ELDOR, EHUD ARBIT, DANIEL SCHURR, MIRIAM KIDRON, AVRAM HERSKO, Jerusalem, Israel, Haifa, Israel Oral alternatives to injectable insulins are expected to better mimic physiological portal vein insulin gradients, while inducing far fewer side effects than systemically delivered insulins. After establishing safety and efficacy of an insulin capsule (ORMD-0801), the presented research efforts set out to explore the ideal dose and active ingredient:adjuvant ratios by evaluating dose-responses of healthy, fasting individuals treated with orally administered insulin (ORMD-0801). One capsule containing 8 mg insulin, two capsules of 8 mg insulin each (8+8mg) or one capsule of 16 mg insulin were administered to subjects, followed by a 300-min monitoring period. Interim analyses demonstrated responses in 7/10 subjects, with maximal blood glucose responses for all treatments observed following a lag of ≥60 min from administration, as expected of an enteric-coated capsule. Dose response was manifested by significantly lower mean blood glucose Cmin following the 8+8mg (47.9 ± 11.3 mg/dL, p=0.006) and 16mg (57.3 ± 8.4 mg/ dL, p=0.001) treatments, when compared to that which followed 8mg dosing (64.6 ± 6.2 mg/dL). Significant reductions in glucose area under the curve (AUC) values were observed following both 8+8mg and 16mg treatments (13.2% and 8.1%, respectively), when compared to 8mg (p=0.003 and 0.008, respectively). No adverse events were reported or observed following any of the treatments. In conclusion, increasing doses of ORMD-0801 were well tolerated by healthy individuals and response intensity positively correlated with the administered dose. & 1057-P A Dose-Ranging Study of Diacerein in Patients With Type 2 Diabetes CARL O. BROWN, WEISHU LU, EMILY LIN, CALVIN CHEN, Taipei, Taiwan The cytokine interleukin (IL)-1Beta, a regulator of inflammatory response, is implicated in the progression of type 2 diabetes mellitus (T2DM) through both insulin resistance and beta-cell function mechanisms. We assessed the safety and efficacy of diacerein, an orally available small molecule that modulates IL-1Beta signaling, in patients with T2DM uncontrolled on a stable regimen of 1 to 3 oral antidiabetic medications (OADs). In Study AC-201-DM-001, a randomized, double-blind, placebo-controlled, dose-ranging, 24-week Phase 2 study, 259 patients were randomized 1:1:1:1 to receive twice daily (BID) 25 mg, 50 mg, or 75 mg diacerein, or placebo. At baseline, patients with mean (SD) glycosylated hemoglobin (A1C) of 8.4 (0.7) % and fasting plasma glucose of 172 (43) mg/dL were enrolled and treated at 21 sites in the US (N=150) and Taiwan (N=109). In the Full Analysis Set (N=249), diacerein showed placebo-corrected least squares mean (SE) A1C reductions of 0.20 (0.17) %, 0.29 (0.18) %, and 1055-P WITHDRAWN & For author disclosure information, see page 829. A272 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 0.35 (0.18) %* (*p<0.05) after 24 weeks of treatment in the 25 mg, 50 mg, and 75 mg groups, respectively. In the per-protocol (PP) population (N=184), the placebo-corrected A1C reduction was 0.37 (0.19) %, 0.42 (0.18) %*, and 0.49 (0.19) %*, for 25 mg, 50 mg, and 75 mg groups, respectively. Further, the US cohort had placebo-corrected A1C reductions in the PP population (N=104) of 0.61 (0.27) %*, 0.72 (0.25) %*, and 0.93 (0.26) %*, for 25 mg, 50 mg, and 75 mg groups, respectively. Differences in patient profiles such as background therapy (54% in Taiwan vs. 9% in US on 3 OADs) are possible reasons for the differences in efficacy results. The most frequent adverse event with diacerein was diarrhea (9%, 19%, 16%, and 32% in placebo, 25 mg, 50 mg, and 75 mg groups, respectively), which was mostly mild. Diacerein showed good dose response and was well tolerated up to 75 mg BID, demonstrating potential as a treatment for T2DM with a unique mode of action targeting the inflammation pathway associated with both impaired beta-cell function and insulin resistance. ΔI30/ ΔG30) improved 7 and 10 fold in the K and G gps. When adjusted for insulin sensitivity (EIR/HOMA-IR) the improvement was 2 fold greater in the K vs. G gp. Hypoglycemia (BG 50-69 mg/dL) occurred less in the K gp (7.4% vs. 21.2%). Combination of saxagliptin and metformin is an effective initial therapy for glycemic and beta cell function improvement in T2DM pts with severe hyperglycemia. 1058-P A Comparative Study of the Effects of a Dipeptidyl Peptidase-IV Inhibitor and Sulfonylurea on Glucose Variability in Type 2 Diabetic Patients With Inadequate Glycemic Control on Metformin HUN-SUNG KIM, HAE KYUNG YANG, JEONG AH SHIN, SEUNG-HWAN LEE, JAEHYOUNG CHO, HO-YOUNG SON, KUN-HO YOON, Seoul, Republic of Korea Objective: This study aimed to compare the effects of sitagliptin on glycemic change and 24-hour blood glucose variability to those of the sulfonylurea glimepiride. Methods: A 4-week randomized double blind-labeled prospective design was used. We recruited 33 patients who had been treated with metformin for at least 2 months (≥1000 mg). Each participant prescribed with metformin was randomly assigned to either the sitagliptin (100mg) or the glimepiride (2mg) group. A continuous glucose monitoring (CGM) was used to monitor glycemic changes for 3 successive days in both group at baseline and at the 4-week follow-up. Glycemic changes and glucose variability were obtained using CGM and these data were averaged over all subjects. Results: The comparison of HbA1c between baseline and the 4-week follow-up showed that HbA1c was significantly reduced in the sitagliptin group (7.0 ± 0.5 to 6.6 ± 0.4 %, p<0.001) and glimepiride group (7.3 ± 0.4 to 6.9 ± 0.4, p<0.001). The sitagliptin and glimepiride groups had similar HbA1c and fasting glucose levels after 4 weeks, and there were no significant differences between the 2 groups. The mean amplitude of glycemic excursions (MAGE) decreased significantly in the sitagliptin group (88.8 ± 18.5 to 67.3 ± 16.6 mg/ dL, p<0.001), but no significant difference was observed in the glimepiride group (103.0±27.2 to 89.3±26.1 mg/dL, p=0.175). The SD and oxidative stress markers did not differ significantly between the 2 groups. Conclusions: When sitagliptin was combined with metformin, the patients showed much more efficient blood glucose controlling effects not only the 3 indexes fasting blood glucose, postprandial blood glucose, and HbA1c but also MAGE. Supported by: Bristol-Myers Squibb & DANIEL M. FORD, MERAV BAZ-HECHT, NABIL A. SAID, BEN TAO, KENNETH E. TRUITT, HUBERT S. CHOU, Edison, NJ Previous studies evaluated the effects of COL combined with metformin-, sulfonylurea-, or insulin-based therapy, but it had not previously been extensively studied in combination with thiazolidinediones. A study was recently conducted to evaluate COL as an add-on to PIO-based therapy in adults with T2DM who had inadequate glycemic control (hemoglobin A1C ≥7.5% to ≤9.5%) on a stable dose of PIO (30 or 45 mg/day), with or without 1-2 other oral antidiabetes agents (metformin, a sulfonylurea, or a dipeptidyl peptidase [DPP]-4 inhibitor). Eligible subjects were randomized to receive COL 3.75 g/day (n=280) or matching placebo (PBO; n=282) in addition to existing PIO-based therapy for 24 weeks. Subjects’ mean age was 56.0 years, 56.9% were male, and the mean duration of T2DM diagnosis was 8.4 years. Concomitant antidiabetes medications also included metformin (83.3%), a sulfonylurea (32.6%), and a DPP-4 inhibitor (3.2%). At Week 24, with last observation carried forward, COL treatment showed reductions from baseline in A1C (least squares mean treatment difference -0.32%; P<0.001) and fasting plasma glucose (-14.7 mg/dL; P<0.0001) vs PBO. COL also decreased LDL cholesterol (-16.4%), total cholesterol (-6.5%), non-HDL cholesterol (-9.8%), and apolipoprotein (apo) B (-8.8%; all P<0.0001), and increased apoA1 (+3.4%) and triglycerides (median treatment difference +11.3%; both P<0.001) vs PBO. The most frequently reported adverse events in the COL group were hypoglycemia (COL, 4.0%; PBO, 6.1%), constipation (COL, 4.0%; PBO, 2.9%), increased blood creatinine phosphokinase (COL, 3.7%; PBO, 1.1%), urinary tract infection (COL, 3.3%; PBO, 4.3%), upper respiratory tract infection (COL, 2.9%; PBO, 3.2%), nausea (COL, 2.9%; PBO, 1.1%), and peripheral edema (2.9% for both). In conclusion, and similar to prior add-on studies, adding COL to PIO-based therapy significantly improved glycemic and most lipid parameters (triglycerides excepted) in patients with T2DM. Supported by: Merck & Co., Inc. & 1059-P Combination of Saxagliptin and Metformin Is Effective as Initial Therapy in Type 2 Diabetes Mellitus (T2DM) With Severe Hyperglycemia AMBIKA BABU, SAMIH BARCHAM, LOUIS DANIEL, LEON FOGELFELD, Chicago, IL Performance of two oral hypoglycemic regimens was tested in newly diagnosed T2DM patients (pts) with severe hyperglycemia presenting to different acute care sites. In a 12-week open label, RCT of 100 T2DM pts with blood glucose (BG) of 300-450 mg/dL were randomized to Glipizide XL 10 mg daily (G group [gp]) versus combination of Saxagliptin with Metformin extended release 5/2000 mg daily (Kombiglyze XR: K gp). The primary outcome was maintenance of a fasting and random BG of <300 and <400 mg/dL upto 6 weeks and <250 and <300 mg/dL thereafter, and no return ED visits. Interim data are presented for 60 pts (27 in K gp, 33 in G gp). Baseline characteristics were similar between both gps except for age (47, 41 in K, G). Two pts in K gp did not achieve the primary outcome. The enrollment BG of 340 (K gp) and 339 mg/dL (G gp) declined significantly by week 1 to 245 and 239 mg/dL and by week 12 to 131 and 126 mg/dL, respectively (figure). A1c improved significantly; 10.9 to 6.9% (K gp) and 11.1 to 7.2% (G gp). HOMA β-cell improved significantly 6 and 4 fold and early insulin response (EIR: ADA-Funded Research & 1060-P Efficacy and Safety of Colesevelam HCl (COL) as an Add-On to Pioglitazone (PIO) for Type 2 Diabetes Mellitus (T2DM) For author disclosure information, see page 829. Guided Audio Tour poster A273 POSTERS & Clinical Diabetes/ Therapeutics Guided Audio Tour: Clinical Pearls (Posters: 1058-P to 1065-P), see page 19. CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS & & 1061-P Durability of the Efficacy and Safety of Alogliptin Compared to Glipizide When Used in Combination With Metformin: 1 Year Interim Analysis OLIVIA J. PHUNG, DIANA M. SOBIERAJ, SAMUEL S. ENGEL, SWAPNIL N. RAJPATHAK, Pomona, CA, Storrs, CT, North Wales, PA, Whitehouse Station, NJ POSTERS STEFANO DEL PRATO, RICCARDO CAMISASCA, CRAIG WILSON, PENNY FLECK, Pisa, Italy, Deerfield, IL Clinical Diabetes/ Therapeutics 1063-P Early Dual Therapy Is Associated With Improved Glycemia in Type 2 Diabetes: A Meta-Analysis Treatment guidelines for type 2 diabetes recommend add-on agents when metformin alone fails to provide adequate glycemic control. However, early dual therapy may benefit patient outcomes. A systematic literature search in MEDLINE and CENTRAL through 07/2012 sought RCTs on the initiation of dual therapy with metformin vs metformin monotherapy. A random effects model yielded weighted mean differences (WMDs) and RRs. The metaanalysis included 15 RCTs (6693 pts; mean age range: 48.4-62.7 yrs; mean baseline A1c 7.2-9.9%, and mean diabetes duration: 1.6-4.1 yrs.) with median follow-up of 6 mos, with 10 reporting A1c change, 11 reporting A1c goal attainment of <7% and 10 reporting change in fasting plasma glucose (FPG). Add-on drugs were TZDs in 5, insulin secretagogues in 4, DPP-4 inhibitors in 4 and SGLT-2 inhibitors in 2 RCTs. Compared to metformin alone, dual therapy provided statistically significant reductions in A1c (WMD -0.43%, 95%CI -0.56,-0.30), increases in attainment of A1c goal of less than 7% (RR 1.40, 95%CI 1.33-1.48), and reductions in FPG (WMD -14.30 mg/dL, 95%CI -16.09,-12.51) (Figure). Heterogeneity was seen in all outcomes (I2 52-76%) except A1c goal (I2=0%). These results suggest a potential benefit of initial dual therapy on glycemic outcomes in diabetes compared to metformin monotherapy across a wide range of baseline A1c levels. Further research should explore if early dual treatment may also affect longer term health outcomes in diabetes. The durability of the efficacy and safety of alogliptin (ALO) compared to glipizide when used in combination with metformin (MET) in subjects with type 2 diabetes was evaluated using a 3-arm multicenter, randomized, double-blind, active-controlled study. The total duration of the study will be 104 weeks; however, these data are from the prospective 1 year interim analysis. The 3 treatment arms included ALO 12.5 mg QD + MET (ALO 12.5) (n=880), ALO 25 mg QD + MET (ALO 25) (n=885), and glipizide 5 mg titrated up to a max of 20 mg + MET (GLIP) (n=873), with the primary endpoint being change from A1c at 52 weeks. The majority of subjects were white (62%), women (50.3%), mean age of 55 years, body mass index of 31, diabetes duration of 5.5 years, and baseline A1c of 7.6%. Reductions in A1c at week 52 (primary endpoint) were -0.62%, -0.61%, and -0.52% for ALO 12, ALO 25, and GLIP, respectively. Significantly more subjects achieved an A1c <7% with ALO 25 (55.3%) vs GLIP (47.4%) (P<0.001). Reductions in FPG were -5.0 mgl/ dL for ALO 12.5, -7.2 mg/dL for ALO 25, and 0.9 mg/dL for GLIP; reductions for ALO were significantly greater for both doses compared to GLIP (P<0.001). Change in weight from baseline at 52 weeks was -0.64, -0.91, and 0.89 for ALO 12.5, ALO 25, and GLIP, respectively; both doses of ALO were significantly different from GLIP (P<0.001). A higher percentage of subjects in the GLIP group (23.8%) experienced ≥ 1 hypoglycemic event compared to ALO 12.5 (2.5%) and ALO 25 (1.4%). A similar number of subjects experienced ≥ 1 adverse event (AE) in all 3 groups; in addition, a comparable number of subjects experienced an AE leading to treatment discontinuation. A total of 8 deaths occurred in the study with 2 in the ALO 12.5 group, 3 in the ALO 25 group, and 3 in the GLIP group. In summary, a significantly higher number of subjects achieved the A1c goal of <7%. The safety profile was similar between all treatment arms; however, significantly less hypoglycemia was observed in the ALO dose groups. Supported by: Takeda Global Research & Development Center, Inc. & 1062-P Vitamin B12 Deficiency and Hyperhomocysteinemia in Patients With Type 2 Diabetes Receiving Metformin YUKA SATO, KEISHI YAMAUCHI, YOSHIKO FUNASE, TORU AIZAWA, Matsumoto, Japan Metformin (Mf) therapy is often associated with B12 deficiency which is a cause hyperhomocysteinemia, a risk factor for vascular complications. However, status of plasma homocysteine (Hcy) in relation to vascular complication in Mf -treated patients with T2DM has not been studied. We determined plasma concentration of B12 and Hcy and status of vascular complications in T2DM patients with and without Mf therapy. Those with vitamin supplementation, renal dysfunction (Scr >115 µmol/L) or a history of gastrectomy were excluded. In 62 consecutive Mf-treated patients (M/F 49/13, the mean age 62 yrs, HbA1C 7.5%, Mf dosing 1.1 g/day and duration of Mf therapy 2.9 yrs), B12 deficiency (<150 pmol/L) was found in 8 (13%) and borderline deficiency (150-220 pmol/L) in 18 (29%). The larger the Mf dosing, the lower the B12 (r = -0.30, P = 0.02). B12 was lower in smokers than in non-smokers (222±85 vs 300±83 pmol/L, P <0.01). However, Mf dosing was correlated with B12 even after adjustment for smoking. B12 in Mf-treated patients was lower than the age- and sex-matched non-Mf-treated T2DM patients (n = 44 each, 256 ± 99 vs 309 ± 120 µmol/L, P = 0.03). In the Mf-treated patients, there was no significant relation between B12 concentration and RBC, eGFR, and macrovascular complications and no anemia in those with B12 deficiency. Plasma Hcy was inversely correlated with B12 (Spearman’s ρ = -0.42, P <0.01) and Hcy concentration was >10 µmol/L in 20 (45%). Importantly, prevalence of diabetic retinopathy was 8, 14 and 27% in those belonging to homocyteine Tertile 1 (T1, < 9.0 µmol/L), T2 (9.0-11.0 µmol/L) and T3 (> 11.0 µmol/L), respectively (P = 0.13). Oral B12 administration (1.5 mg/day), without cessation of Mf, normalized B12 concentration. In conclusion, Mf therapy was dose-dependently associated with lowering of B12, which was in turn correlated with elevation of Hcy. Deleterious effect of hyperhomocysteinemia on diabetic retinopathy was suggested. Mf-induced B12 deficiency was responsive to B12 supplementation. Supported by: Merck & Co., Inc. & 1064-P Differential Regulation of Metabolic Parameters between SuperResponders and Non-Responders Treated With Sitagliptin EIJI KUTOH, YASUHIRO UKAI, Tokyo, Japan The aim of this work is to investigate the clinical characterization of responders vs. non-responders treated with sitagliptin. Treatment naïve subjects with T2DM received 50 mg/day sitagliptin monotherapy (n=41). At 3 months, levels of metabolic parameters were compared with those at baseline. Since it is known that the response to sitagliptin is proportional to the baseline HbA1c levels, the novel index called A1c index (ΔHbA1c/baseline HbA1c) was used to assess the glycemic efficacy of sitagliptin. The subjects were divided into three groups according to this index; super-responders (A1c index <-0.2, n=20, average ΔHbA1c: -3.84 %), responders (A1c index between -0.2~-0.1, n=10, average ΔHbA1c -0.96%) and non-responders (A1c index >-0.1, n=11, average ΔHbA1c -0.11%). At baseline, TG (140.35 vs. 438.71 mg/dl, p<0.005), non-HDL-C (156.45 vs. 188.14 mg/dl, p<0.01), atherogenic index (AI: log10 [TG/HDL-C], 0.381 vs. 0.782, p<0.02), HOMA-R (2.75 vs. 5.42, p<0.05) and BMI (23.66 vs. 27.26, p<0.05) were significantly lower in superresponders than non-responders. At 3 months, significant increases of HOMA-B (from 17.57 to 48.57, p<0.003) and BMI (23.66 to 24.10, p<0.05) were observed in super- responders, while these and other parameters did not change in non-responders or responders. In super-responders, Supported by: Chiiki Igaku Kenkyu Kikin Foundation & For author disclosure information, see page 829. A274 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS normal bladder cells and J82 cells (from 24h to 72h, 10M PGZ) were showed no changes at all. However, the expressions of CyclinD1 and P53 from J82 cells under 10M PGZ (for 8 to 10d) were decreased in spite of cells’ normal growth. Finally, PGZ addition may not elevate the risk of bladder cancer in normal urothelial cells nor promote J82 cell proliferation. But prolonged usage may weaken the carcinogenicity of bladder cancer. PGZ can cause normal bladder cell apoptosis in a tissue-specific way, which needs further investigations to clarify its significance. significant correlations were observed between ΔHbA1c and ΔHOMA-B (R=0.50951, p<0.03) or ΔBMI (R=-0.4589, p<0.05). These results suggest that 1) sitagliptin is more effective with those with lower levels of insulin resistance, TG, non-HDL-C, AI or BMI. 2) significant body weight gain was observed in super-responders and the improvement of glycemic control was associated with body weight gain. Potential involvement of GIP pathway in super-responders will be presented. Supported by: SHPJ1408400 & 1067-P TYLER DRAKE, DON HIRE, ELIZABETH SEAQUIST, Minneapolis, MN, WinstonSalem, NC Does Sulfonylureas Increase the Incidence of Cancers in Type 2 Diabetes? A Systematic Review and Meta-Analysis Achieving target glycemia in older patients with type 2 diabetes is a challenge in clinical practice. In clinical trials, patients are managed by protocols to achieve glycemic goals, but information from these protocols has not been readily transferred to the clinic. Here, we sought to indentify factors in the standard therapy arm of ACCORD (goal A1C 7.1-7.9%) that were associated with failure to achieve an A1C similar to guideline recommendations. ACCORD patients reached target glycemia by 12 mo and then maintained that level. Standard patients whose A1C was ≥8.0% at 12 mo were compared to those whose 12 mo A1C was 7.1-7.9%. Patients were analyzed based on baseline characteristics, medication classes used, weight gain, and hypoglycemic events. Of the 5123 patients randomized to the standard group, complete data were available for 4693 (1988 with A1C 7.1-7.9%, 1493 with A1C ≥8.0% at 12 mo). At baseline, patients with an A1C ≥8.0% at 12 mo were younger (61.3 vs 62.8 yrs, p<0.0001), had a higher BMI (32.5 vs 32.0 kg/ m2, p=0.0148), longer duration of diabetes (12.0 vs 10.8 yrs, p<0.0001), and had a higher baseline A1C (8.7 vs 8.2%, p<0.0001). They were more likely to be female (39.7 vs 36.4%, p=0.0458) and to be Hispanic or Black (8.8 vs 5.8%, p=0.0007 and 22.2 vs 15.6%, p<0.0001, respectively). There was no difference in the use of metformin, secretagogues, glucosidase inhibitors, or incretins at 12 mo, but patients above goal were more likely to be on thiazolidinediones (40.1 vs 36.4%, p=0.0294). Patients above goal had more hypoglycemic events (1 event: 2.8 vs 1.2%, p=0.0023; >1 event: 0.5 vs 0.4%, p=0.0023) and more weight gain (1.0 vs 0.1 kg, p<0.0001) at 12 mo. Patients in the standard therapy group of the ACCORD trial who failed to reach their A1C goal 12 mo after randomization had significant differences in both baseline characteristics and clinical features when compared to patients who reached goal A1C. These characteristics may prove useful in identifying patients who may or may not reach the desired A1C in clinical practice. HUI GAO, YUNFENG SHEN, XIAOYANG LAI, JIANPING WENG, HUA LIANG, Nanchang, China, Guangzhou, China Sulfonylurea is one of main drugs for treating type 2 diabetes. A wide variety of sulfonylureas are currently available in practice. But these agents were concerned to increase the potential risk of cancers. In this study, we assessed whether sulfonylureas increased the incidence of cancers with a meta-analysis of randomized controlled trials and cohort studies. Also, the risks of various cancers were evaluated. The following sources were used: the Cochrane library; EMBASE, MEDLINE, Web of Science, CBM, VIP, CNKI. Search terms including diabetes mellitus, type 2 diabetes mellitus, sulfonylurea, tolbutamide, chlorpropamide, glibenclamide, gliclazide, glipizide, gliquidone, glimepiride, malignancy, tumor, cancer, neoplasm, carcinoma, tumorigenesis in meta-analysis, randomized controlled trials and cohort studies. New cancer data were available for 308,972 patients in one randomized clinical trial and four cohort studies. Data on common types of solid organ tumors were available for 6,649 patients from these studies. Patients receiving sulfonylureas had no significant increased risks in new cancer occurrence compared with patients in controls, with an RR of 1.24 (95%CI 0.91~1.69) for metformin, 1.04 (95%CI 0.88~1.23) for insulin and 1.17 (95%CI 0.93~1.47) for other anti-diabetics. Among specific solid organ tumors examined, colorectal (RR 0.95, 95%CI 0.76~1.17), liver (RR 1.17, 95%CI 0.60~2.29) and pancreatic (RR 1.34, 95%CI 0.87~2.06) cancers were no significantly higher risks in patients receiving sulfonylureas. Interestingly, sulfonylureas treatment significantly reduced risk of bladder cancer (RR 0.64, 95%CI 0.46 to 0.90). In conclusion, our results suggested that sulfonylureas were not associated with increased risk of new cancer. Surprisingly, sulfonylureas reduced the bladder cancer risk. These findings warrant further investigation. & Guided Audio Tour: Off-Target Effects of Glucose Lowering Drugs (Posters: 1066-P to 1071-P), see page 15. & MATTIJS OUT, ADRIAAN KOOY, PHILIPPE LEHERT, CASPER G. SCHALKWIJK, COEN D.A. STEHOUWER, Hoogeveen, Netherlands, Louvain, Belgium, Maastricht, Netherlands 1066-P Effect of Pioglitazone on the Normal Bladder Cell and Bladder Cancer Cell Metformin is a key treatment option in type 2 diabetes (T2D), but is associated with lowering of vitamin B12 (B12) levels. Although it is commonly assumed that lower B12 levels reflect biological B12 deficiency, this is in fact not certain. Accumulation of methylmalonic acid (MMA) may reflect a biologically important B12 deficit. Therefore, we investigated whether treatment with metformin during 52 months affects serum levels of MMA. In this placebo-controlled trial, 390 patients with T2D were randomized to metformin or placebo added to insulin therapy. Baseline characteristics have been described previously (BMJ 2010). MMA levels were measured by ultra performance liquid chromatography tandem mass spectrometry (UPLCMSMS). Multivariate regression showed that metformin treatment significantly raised MMA serum concentrations. Compared to placebo, metformin increased MMA by 0.03 nmol/l (95% 0.003 to 0.063; p=0.03; after adjustment for MMA at baseline). Renal function and age did not affect this increase. In addition, this effect appeared to be dose-related (P<0.001). This long-term placebo controlled trial is the first study to show that in patients with T2D metformin not only reduces serum levels of B12, but also increases serum MMA, thus supporting the view that metformin-associated lowering of B12 levels reflects biological B12 deficiency. JIJIAO WANG, HONG LI, BENLI SU, XIAOLAN YUAN, JIAOJIAO ZHOU, CHUNYAN LI, BHAVNA TOHOOLOO, NAVINA PRIYA JHUMMON, HUI SHENG, SHEN QU, Shanghai, China, Dalian, China Recently, pioglitazone (PGZ) has become the focus for the possible relation with bladder cancer. Though many large-scale clinical studies have been terminated, more observational studies are needed for further interpretation.Our study aims to investigate the effect of PGZ on the normal bladder cells and bladder cancer cell line (J82). Normal urothelial transitional epithelium cells (from bladder of SD-rats) and J82 cells were treated with different concentration of PGZ (from 5 to 40M) respectively. We observed the surviving status of cells and detected the expressions of key mRNA and protein including P53 and CyclinD1 as well as apoptosis associated gene Bcl-2 and Bax at from 0h to 10d, after treatment with PGZ. We observed that morphology of normal bladder cells was shown significant decrease of cell viability after 48h in the 10M PGZ group. The inhibitory effects were time and dose dependent. But PGZ used, visibly did not inhibit the growth of J82 cells. The results of flow cytometry technology (FCT) further showed the apoptosis ratio of normal bladder cells (for 72h, 10M PGZ, vs. control group) was 11.3% vs. 49.7%. Liver cells, kidney cells and vascular endothelial cells of SD-rats treated with PGZ were shown no cell deaths in any of them. Accordingly, we hypothesized that PGZ exerts ‘toxicity’ which is highly tissue selective for bladder cells.Meanwhile, tested by western blot and Rt-PCR, the expressions of Bax and Bcl-2, as well as of CyclinD1 and P53 from both ADA-Funded Research & 1068-P Metformin Treatment Is Associated With Increased Serum Levels of Methylmalonic Acid in Patients With Type 2 Diabetes Treated With Insulin: A Placebo-Controlled 4-Year Trial For author disclosure information, see page 829. Guided Audio Tour poster A275 POSTERS 1065-P Clinical Diabetes/ Therapeutics & Achieving Goal Hemoglobin A1C in the Standard Therapy Group of the ACCORD Trial CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS & 1070-P Sulfonylureas Associated With Higher Emergency Room use in Elderly With Diabetes SWAPNIL RAJPATHAK, CHUNMAY FU, KIMBERLY BRODOVICZ, SAMUEL S. ENGEL, North Wales, PA Emergency room (ER) use due to drug adverse events pose significant economic burden on US healthcare system. Recent evidence among elderly suggests that drugs for type 2 diabetes account for 25% of ER hospitalizations, often linked with hypoglycemia. However, it is yet unknown whether sulfonylureas (SU) are associated with ER use. We conducted a retrospective study using MarketScan® (2009-10) to evaluate this association. The study included 112,607 patients aged ≥65 years on oral monotherapy for diabetes at baseline (34.3% on SU, 52.0% on metformin, 10.8% on TZDs and 2.9% on other oral agents; insulin users excluded). Mean emergency room (ER) visits was 0.67 among SU users compared to 0.47 among non-users in the one year period (p<0.0001). Further, SU use was reported in 31.6% with no ER visit, 37.0% with 1 and 44.0% with ≥2 ER visits. Hypoglycemia, which was more common among SU users, was associated with more frequent ER use (Table). In a multivariable logistic regression simultaneously accounting for propensity to use SU and for potential confounders, the adjusted ORs were 1.16 (95% CI: 1.12-1.19) for 1 ER visit and 1.27 (95% CI: 1.22-1.32) for ≥2 ER visits. For specific comparison of SU vs. metformin monotherapy, these ORs were 1.21 (95% CI: 1.16-1.23) for 1 and 1.36 (95% CI: 1.30-1.42) ≥2 ER visits. Results suggest that elderly SU users are more likely to use ER compared to non-users. Further research should confirm this and consider evaluation of factors associated with ER use among SU users. & POSTERS Clinical Diabetes/ Therapeutics Supported by: Altana AG; Lifescan, Inc.; E. Merck/Santé; Merck Sharpe & Dohme; Novo Nordisk, Inc. 1069-P Genital Mycotic Infections With Canagliflozin (CANA) in Subjects With Type 2 Diabetes Mellitus (T2DM) PAUL NYIRJESY, JACK SOBEL, ALBERT FUNG, CRISTIANA GASSMANN-MAYER, KIRK WAYS, KEITH USISKIN, Philadelphia, PA, Detroit, MI, Raritan, NJ, Titusville, NJ CANA, an SGLT2 inhibitor, lowers plasma glucose by increasing urinary glucose excretion. Genital mycotic infections (GMIs) with CANA were assessed in subjects with T2DM (N = 2,313; mean age, 56 y; A1C, 8%; BMI, 32 kg/m2) from 4 randomized, double-blind, 26-wk, placebo (PBO)-controlled studies. Rates of GMIs were 11% vs 3% (women) and 4% vs 1% (men) with CANA relative to PBO (Table). GMIs were generally considered mild or moderate in severity by investigators and few led to discontinuation; most events with CANA were treated with antifungal agents (women, 82%; men, 83%). CANA-treated women with GMIs were more likely to have a history of vulvovaginitis (29% vs 12%), be pre-menopausal (35% vs 27%), and reside in North America (60% vs 42%) than those without GMIs. CANA-treated men with GMIs were more likely to have a history of balanitis/balanoposthitis (25% vs 2%), have a slightly longer mean T2DM duration (9 vs 7 y), and to reside in Europe (44% vs 26%) than men without GMIs. With CANA treatment, 2% and 1% of females and males, respectively, had >1 GMI. In a larger dataset (N = 9,439) from 8 studies with longer mean exposure (CANA, 68 wks; control, 64 wks), a higher rate of male GMIs was seen (n = 5,493; 8% and 2% for CANA and control), more commonly in uncircumcised than circumcised men (11% vs 3%) and infrequently associated with phimosis or the need for circumcision. In summary, CANA was associated with modestly higher rates of GMIs that were generally manageable with standard treatments. No ER visits 1 ER visit ≥2 ER visits SU users (n=38,282) Total N (%) 25,245(65.9) 7,262(19.0) 5,775(15.1) Hypoglycemia, N (%) 235(0.9) 222(3.1) 398(6.9) Non-users (n=74,325) Total N(%) 54,632(73.5) 12,327(16.6) 7,366(9.9) Hypoglycemia, N (%) 251(0.5) 190(1.5) 220(3.0) OR (95% CI) Hip Fracture comparing SU users vs. non-users Unadjusted 1 1.43 (1.40-1.47) 1.70 (1.63-1.76) Adjusted* 1 1.16 (1.12-1.19) 1.27 (1.22-1.32) *Age, gender, health insurance, geographic region, history of CVD, stroke, hypertension and dyslipidemia, and use of benzodiazepines and anti-convulsant therapy, and propensity to use SU & 1071-P The Effect of DPP-4 Inhibitor, Alogliptin, on Bone Metabolism in Patients With Type 2 Diabetes Mellitus YUICHIRO YAMAUCHI, SHIN TUNEKAWA, YUSUKE SEINO, EITA UENISHI, TETUJI OKAWA, ATUSHI FUJIYA, KOUTA ISHIKAWA, HIDETADA OGATA, ATUSHI IIDA, RYOICHI BANNO, HIROSHI ARIMA, YOJI HAMADA, YUTAKA OISO, Nagoya, Japan Type 2 diabetes (T2D) has been associated with a high risk of bone fractures. Recent studies showed that treatment with DPP-4 inhibitors could be associated with a reduced risk of bone fractures in the meta-analysis and incretin hormones increased bone density in experimental models. However, the clinical effects of DPP-4 inhibitors on bone metabolism remain to be clarified. Therefore, we examined the effect of DPP-4 inhibitor, Alogliptin (Alo), on bone metabolism in Japanese patients with T2D. This prospective observation study was performed in 20 patients including 8 men and 12 postmenopausal women. The patients were treated with 25mg Alo once a day for 16 weeks, and GIP, calcitonin, biochemical markers of bone metabolism and bone mineral density were evaluated in a fasting state at baseline and post-treatment. Wilcoxon signed-rank test revealed that Alo significantly decreased HbA1c from 7.1±0.5% to 6.6±0.5% and serum bone alkaline phosphatase (BAP) from 18.3±11.9µg/L to 15.3±7.6µg/L, respectively. In contrast, Alo did not affect GIP and other bone metabolic markers such as osteocalcin(OC), serum NTX and OC/BAP ratio. Bone mineral density in lumbar vertebra and proximal femur did not change significantly during 16 weeks, but ΔBAP was associated with the elevation of bone density in proximal femur. It is noteworthy that Alo significantly elevated the level of calcitonin that was known to be increased by GLP-1 and to reduce fracture risk. (baseline: 22.7±5.7 pg/ml, post-treatment: 26.5±5.1 pg/ml). However, the elevation was not associated with the change of bone density. Multivariate regression analysis revealed that ΔBAP and Δcalcitonin were independent of ΔHbA1c, respectively. These data suggest that the long-term use of DPP-4 & For author disclosure information, see page 829. A276 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS Conclusion: Reduced β-cell function (IS/IR index) and impaired rise in plasma adiponectin characterize non-responsiveness to PIO. inhibitor may exert beneficial effects on bone quality by suppressing the unsuitable bone metabolism in T2D. The role of elevated calcitonin by the DPP-4 inhibitor on bone metabolism remains obscured at this study and further investigation is necessary. 1074-P The Sodium Glucose Co-Transporter-2 (SGLT2) Inhibitor Empagliflozin Improves Glycemic Control in Patients With Type 1 Diabetes: A Single-Arm Clinical Trial 1072-P Ranolazine Inhibits Glucagon Secretion from Human Pancreatic Islets via Blockade of Nav 1.3 Channels in α-Cells Adjunctive therapy to basal-bolus insulin regimens may provide an added strategy to improve glycemic control in patients with type 1 diabetes. In an 8-week single-arm open-label pilot study of the SGLT2 inhibitor empagliflozin 25 mg po QD in subjects receiving optimized standard of type 1 diabetes care, we evaluated the efficacy on glycemia and rates of hypoglycemia compared to a 2-week placebo run-in period (NCT01392560). Other outcome measures were urinary glucose excretion (UGE), insulin requirement, and anthropometrics. We recruited 42 normoalbuminuric patients (28 insulin pump, 14 multiple daily injection) with mean±SD age 24±5 years, A1c 8.0±0.9 %, and fasting plasma glucose (FPG) 10.0±4.8 mmol/L. 40 patients completed 8 weeks of treatment with empagliflozin and the mean A1c decreased by 0.4% to 7.6±0.9% (p<0.0001). FPG decreased non-significantly to 8.6±3.1 mmol/L (p=0.06). Symptomatic hypoglycemia (<3.0 mmol/L) declined from 0.12 to 0.04 events per day (p=0.005) and mean daily insulin dose declined from 55±20 to 46±19 U/day from the placebo run in period (p<0.0001). The mean UGE increased from 19±19 at baseline to 134±61 g/day at 8 weeks (p<0.0001). Weight decreased by 2.7 kg from 72.6±12.7 (p<0.0001) and waist circumference decreased by 3.8 cm from 82.9±8.7 (p<0.0001). Apart from hypoglycemia, polyuria (79%) and thirst (74%) were the most frequent adverse events. 2 subjects were withdrawn after the early occurrence of diabetic ketoacidosis (one in the setting of gastroenteritis, the other in the setting of insulin pump failure). In patients with type 1 diabetes, empagliflozin as adjunctive to insulin therapy was generally well-tolerated and associated with short-term improvement in glycemic control, a reduction in rates of hypoglycemia and reduced insulin requirement and body weight. A randomized clinical trial of the efficacy of adjunctive empagliflozin in type 1 diabetes is warranted. 1073-P Reduced Beta Cell Function and Inadequate Rise in Plasma Adiponectin Characterizes Non-Responsiveness to Pioglitazone in Individuals With Impaired Glucose Tolerance in ACT NOW Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1075-P DEVJIT TRIPATHY, DAWN C. SCHWENKE, MARY ANN BANERJI, GEORGE A. BRAY, THOMAS A. BUCHANAN, STEPHEN CLEMENT, ROBERT R. HENRY, ABBAS E. KITABCHI, SUNDER MUDALIAR, ROBERT E. RATNER, FRANKIE B. STENTZ, PETER D. REAVEN, NICOLAS MUSI, RALPH A. DEFRONZO, San Antonio, TX, Phoenix, AZ, Brooklyn, NY, Baton Rouge, LA, La Canada, CA, San Diego, CA, Memphis, TN, Alexandria, VA Efficacy and Safety of Canagliflozin (CANA) in Subjects With Type 2 Diabetes Mellitus (T2DM) and Chronic Kidney Disease (CKD) Over 52 Weeks JEAN-FRANCOIS YALE, GEORGE BAKRIS, BERTRAND CARIOU, JAVIER NIETO IGLESIAS, EWA WAJS, KATHERINE FIGUEROA, JOEL JIANG, KEITH USISKIN, GARY MEININGER, Montreal, QC, Canada, Chicago, IL, Nantes, France, Ciudad Real, Spain, Beerse, Belgium, Raritan, NJ Multiple studies have shown thiazolidinediones (TZDs) delay/prevent onset of type 2 diabetes mellitus (T2DM). However, a significant number of individuals with IGT fail to respond to TZDs. We examined factors that predict lack of response to pioglitazone in ACT NOW, a randomized doubleblind, placebo-controlled study to test whether pioglitazone (45 mg/day) can prevent/delay development of T2DM in IGT (n= 602). Insulin secretion and insulin sensitivity indices (derived from plasma glucose, insulin, C peptide during OGTT and frequently sampled intravenous glucose tolerance test) and plasma adiponectin were measured at baseline and study end (median 2.4 yrs). Diabetes developed in 50 PLAC-treated subjects vs. 15 PIO-treated subjects (HR=0.28: 95% CI= 0.16,-0.49; p<0.001). Of 213 PIO-treated subjects who completed the study, 101 (47%) reverted to NGT (responders) and 112 (53%) either remained IGT or developed T2DM (non-responders) At baseline there were no differences in FPG, 2-h PG, Matsuda Index of insulin sensitivity (MI) and plasma adiponectin between responders and non-responders. At study end all indices of beta cell function were lower in non-responders versus responders: insulin secretion/insulin resistance index (ΔI0-120/ ΔG0-120 x MI) (2.90 ± 0.2 vs. 3.57 ± 0.2, p=0.002); acute insulin response (312±27vs 415± 39, p=0.03); disposition index from FSIVGTT (684±72 vs 908± 69, p=0.02). Plasma adiponectin increased 2.6 fold (12± 0.7 to 32±2.0 µg/ml) in non-responders versus 4.2 fold (11± 0.5 to 46± 4.0 µg/ml) in responders (p<0.001). The increase in adiponectin correlated with the improvement in FPG (r=0.322, p<0.001), 2-h PG (r=0.281, p<0.001), Matsuda Index (r=0.526, p<0.005) and ΔI0-120/ΔG0-120 x MI (r=0.306, p<0.005). ADA-Funded Research & CANA is an SGLT2 inhibitor in development for the treatment of T2DM. This randomized, double-blind study enrolled subjects with T2DM and Stage 3 CKD (eGFR ≥30 and <50 mL/min/1.73 m2). Subjects (N = 269; mean age, 68.5 y; A1C, 8.0%; fasting plasma glucose [FPG], 164.0 mg/dL; BMI, 33.0 kg/m2; eGFR, 39 mL/min/1.73 m2; median albumin/creatinine ratio [ACR], 30 µg/mg) received CANA 100 or 300 mg or placebo [PBO] added to current therapy (74% on insulin) during a 26-week core period followed by a 26-week extension (n = 207). Over 52 weeks, CANA 100 and 300 mg reduced A1C, FPG, body weight, and systolic BP, with small increases in HDL-C and small decreases in LDL-C relative to PBO. Overall AE rates were 86%, 81%, and 87% and serious AE rates were 20%, 24%, and 27% with CANA 100 and 300 mg and PBO; AE-related discontinuation rates were low across groups. Proportion of subjects with hypoglycemia (64%, 59%, 46%) and rates of osmotic diuresis-related AEs (eg, pollakiuria; <5% per group) were higher with CANA 100 and 300 mg than PBO; rates of UTIs and AEs related to reduced intravascular volume (eg, postural dizziness) were higher with CANA 300 mg than PBO. Increases in BUN (12%, 16%, 5%) and decreases in eGFR (-4%, -8%, -3%) and median ACR (-1, -7, 2 µg/mg) were seen with CANA 100 and 300 mg relative to PBO. In summary, CANA improved glycemic control and was generally well tolerated in subjects with T2DM and Stage 3 CKD over 52 weeks, similar to findings at Week 26. For author disclosure information, see page 829. Guided Audio Tour poster A277 POSTERS Ranolazine (RAN) inhibits sodium current and electrical activity of rat pancreatic α-cells by blocking Na+ channels (NaCh), which play an important role in generating action potentials for glucagon secretion. We determined the effects of RAN and other NaCh blockers (GS-458967 and tetrodotoxin, TTX) on glucagon secretion in human pancreatic islets and identified the NaCh isoform responsible for these effects. RAN and GS-458967 significantly reduced glucagon secretion in human islets in a concentration-dependent manner, with 25±3 and 51±9% reduction for RAN (10 µM) and GS-458967 (3 µM), respectively. Veratridine, a NaCh activator, increased glucagon secretion in human islets in a concentrationdependent manner, with a 10-fold increase at 30 µM and an EC50 of 33±6 µM. RAN, GS-458967 and TTX, in a concentration-dependent manner, reduced the increase in glucagon secretion caused by 30 µM veratridine. Glucagon secretion was reduced by 36±11, 71±10 and 86±8% for RAN (10 µM), GS458967 (3 µM) and TTX (30 nM), respectively. Based on qPCR analysis of whole pancreatic islets, the major isoforms of NaCh expressed are Nav1.3 and 1.7, followed by 1.2 and 1.6. In dispersed human islet cells, transfection with siRNAs for the Nav1.3 or 1.7 isoform resulted in specific knockdown of their gene expression by 65±6 or 55±3%, respectively, compared to the control (n=6). Both veratridine- and L-arginine-induced glucagon secretion was significantly reduced (≥60%) by Nav1.3 but not by Nav1.7 knockdown (n=4), compared to the control. In summary, RAN and other NaCh blockers inhibit glucagon secretion from human islets by blocking Nav1.3 channels. Data suggest that NaCh play an important role in the regulation of glucagon secretion in human islets. We propose that the anti-diabetic effects of RAN in humans are due to a reduction of glucagon secretion by blockade of Nav1.3 channels in pancreatic α-cells. Clinical Diabetes/ Therapeutics BRUCE A. PERKINS, DAVID Z. CHERNEY, HELEN PARTRIDGE, NIMA SOLEYMANLOU, HOLLY TSCHIRHART, BERNARD ZINMAN, NORA FAGAN, STEFAN KASPERS, HANS-JUERGEN WOERLE, ULI C. BROEDL, ODD ERIK JOHANSEN, Toronto, ON, Canada, Mississauga, ON, Canada, Ridgefield, CT, Ingelheim, Germany, Rud, Norway ARVINDER DHALLA, MING YANG, RUTH CHU, LUIZ BELARDINELLI, Fremont, CA CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS CANA in Phase 3 studies. In summary, CANA reduced BP compared with PBO in subjects with a range of baseline BP; absolute reductions were greater in subjects with higher BP, but PBO-subtracted decreases were similar across groups. 1076-P SARA T. HJULER, MICHAEL FEIGH, KIM V. ANDREASSEN, MORTEN A. KARSDAL, KIM HENRIKSEN, Herlev, Denmark POSTERS Clinical Diabetes/ Therapeutics Oral sCT Intervention Improves Glucose Homeostasis and Insulin Sensitivity in an Animal Model of Type 2 Diabetes Oral salmon calcitonin (sCT) has previously been reported to have positive effects on glucose homeostasis and it has been suggested that this effect could mediated through increased insulin sensitivity. In this study the effect of chronic administration of oral sCT was evaluated in regards to fasting and non-fasting blood glucose, glycated hemoglobin (HbA1c), insulin sensitivity, and glucose tolerance in a rat-model of type 2 diabetes. Diet-induced obese (DIO) rats was given a low dose (25 mg/kg) of streptozotocin (STZ) resulting in diabetic hyperglycemia without hypoinsulinemia. The DIO-STZ rats were orally dosed with either 0.5 mg/kg salmon calcitonin (sCT) or its oral vehicle 5-CNAC once daily for 21 days. Postprandial blood glucose (PPG) was measured on day 1, 3, 5, 14, and 21. Oral sCT improved hyperglycemia during the intervention period resulting in 5 mM reduction of PPG at study end when compared to vehicle (p<0.01). During oral glucose tolerance test, oral sCT improved glucose intolerance by reducing the area under the curve with ~35 % when compared to vehicle (p<0.001). Furthermore, oral sCT reduced fasting blood glucose and HbA1c at study end by ~7 mM and 1 % respectively (p<0.05). Oral sCT did not exert an insulinotropic action. In contrast, compared to vehicle group, oral sCT caused a 2-fold increased in glucose disappearance rate (KITT) during intravenous insulin tolerance test (p<0.01) indicating increased insulin sensitivity. Both the vehicle and sCT groups lost weight after injections with STZ with no significant difference between groups. Finally oral sCT improved hyperleptinemia by 11 % compared to vehicle (p<0.05). In conclusion oral sCT improved the glucose homeostasis and insulin sensitivity independent of body weight in DIO-STZ rats. This study demonstrates oral sCT as a therapeutic agent against of type 2 diabetes in an intervention model of diabetes. 1078-P Inhibition of Phosphodiesterase as a Mechanism of Action for Silymarin-Induced Insulin Secretion in HIT-T15 Cells GINA MENG, JANA MAHADEVAN, SARA VALLERIE, ELIZABETH OSEID, JOSEPH BEAVO, R. PAUL ROBERTSON, Seattle, WA Silymarin (SIL) is an oral antioxidant that has been shown in several placebocontrolled human trials to significantly improve fasting glycemia and HbA1c levels in type 2 diabetic subjects. These beneficial effects were attributed to nullification of oxidative stress; however, no measurements were made of markers of oxidative stress or of insulin secretion. To define mechanisms of action for SIL, we examined insulin responses to 11.1 mM glucose (GSIS) using the glucose-responsive β-cell line HIT-T15. In a drug concentrationdependent fashion, SIL had a bimodal effect; initial enhancement of GSIS at drug concentrations of 25-100 µM (EC50 = 38 µM), but with concentrations >200 µM diminishment of GSIS was observed. The SIL concentrations that augmented GSIS were shown also to decrease phosphodiesterase activity in the presence of Ca2+/calmodulin (IC50=26 µM). The higher SIL concentrations that had diminished GSIS were shown by florescent microscopy to have undergone apoptosis. Only the higher concentrations of SIL demonstrated anti-oxidant effects on the accumulation of peroxides in cell culture media (100-500 µM). We conclude that the beneficial effects of SIL on GSIS may be primarily due to decreased phosphodiesterase activity and consequently greater endogenous c-AMP levels with resultant enhancement of glucoseinduced insulin secretion rather than via its antioxidant effects. 1079-P Supported by: The Danish Ministry of Science INdividualized Treatment Targets for EldeRly Patients Using VildaWITHDRAWN gliptin Add-On or Lone therapy (INTERVAL) Study W. DAVID STRAIN, VALENTINA LUKASHEVICH, WOLFGANG KOTHNY, MARIEJOSE HOELLINGER, PÄIVI M. PALDÁNIUS, Exeter, United Kingdom, East Hanover, NJ, Basel, Switzerland 1077-P Lower Blood Pressure (BP) With Canagliflozin (CANA) in Subjects With Type 2 Diabetes Mellitus (T2DM) Management of type 2 diabetes mellitus (T2DM) in elderly patients is challenging and complicated by multiple comorbidities, polypharmacy and increased risk of hypoglycemia. Current guidelines suggest individualizing treatment targets for these patients, despite no evidence-base for such an approach. We evaluated the feasibility of such an approach in parallel with conventional HbA1c reduction using vildagliptin vs. placebo. We enrolled 278 drug-naïve or inadequately controlled (HbA1c ≥7.0% to ≤10.0%) T2DM patients aged ≥70 years in a multinational, double-blind, 24-week study. Investigators set individualized treatment targets based on age, baseline HbA1c, comorbidities and frailty status before 1:1 randomization to vildagliptin (50 mg qd or bid as per label) or placebo. Vildagliptin had a higher likelihood of achieving the individualized treatment targets (OR=3.16), accompanied by clinically relevant reductions in overall HbA1c (Table). Vildagliptin was well tolerated in this elderly cohort, with no new safety signals emerging. INTERVAL is the first study to introduce investigator-defined individualized HbA1c targets as an endpoint in any T2DM population. It demonstrates the feasibility of evaluating individualized targets when exploring diabetes management in the growing elderly population. It also confirmed that individualized glycemic target levels are achievable using vildagliptin without any tolerability issues. MATTHEW R. WEIR, ANDRZEJ JANUSZEWICZ, RICHARD GILBERT, FERNANDO LAVALLE GONZÁLEZ, GARY MEININGER, Baltimore, MD, Warsaw, Poland, Toronto, ON, Canada, Monterrey, Mexico, Raritan, NJ Hypertension is a common comorbidity of T2DM. The BP-lowering response of CANA, an SGLT2 inhibitor, was evaluated using pooled data from 6 randomized, double-blind, placebo (PBO)-controlled Phase 3 studies in subjects with T2DM (N = 4,158, mean age, 59 y; A1C, 8%; BMI, 33 kg/ m2). Relative to PBO, CANA 100 and 300 mg provided modest reductions in systolic BP (SBP; -3.3 and -4.5 mmHg) and diastolic BP (DBP; -1.5 and -1.9 mmHg) in the overall population (Table). In subjects with SBP ≥140 mmHg (n = 1,267), reductions in SBP were seen with both CANA doses versus PBO (P <0.001); the PBO-subtracted decrease was slightly larger with CANA 300 mg than was seen in the overall population, and similar with CANA 100 mg. In subjects with DBP ≥90 mmHg (n = 355), a reduction in DBP was seen with CANA 300 mg versus PBO (P = 0.028); the PBO-subtracted decrease was modestly larger with CANA 300 mg than was seen in the overall population, and smaller with CANA 100 mg. PBO-subtracted BP decreases with both CANA doses were similar among subjects on antihypertensives and those not on these agents. Changes in heart rate were -0.6, -0.4, and -0.1 beats/min with CANA 100 and 300 mg and PBO, respectively. With CANA’s glucoretic effect, a low incidence of osmotic diuresis adverse events was seen with & For author disclosure information, see page 829. A278 Guided Audio Tour poster ADA-Funded Research Supported by: Novartis Pharma AG 1080-P Canagliflozin (CANA) Added on to Dipeptidyl Peptidase-4 Inhibitors (DPP-4i) or Glucagon-Like Peptide-1 (GLP-1) Agonists With or Without Other Antihyperglycemic Agents (AHAs) in Type 2 Diabetes Mellitus (T2DM) CAROL H. WYSHAM, VINCENT C. WOO, CHANTAL MATHIEU, MEHUL DESAI, MARIA ALBA, GEORGE CAPUANO, GARY MEININGER, Spokane, WA, Winnipeg, MB, Canada, Leuven, Belgium, Raritan, NJ 1082-P Empagliflozin as Add-On to Metformin Plus Sulfonylurea (SU) for 24 Weeks Improves Glycemic Control in Patients With Type 2 Diabetes (T2DM) CANA is an SGLT2 inhibitor in development for the treatment of T2DM. This post hoc analysis of data from the CANagliflozin cardioVascular Assessment Study (CANVAS; subjects with T2DM and a history/high risk of cardiovascular disease) evaluated CANA 100 and 300 mg compared with placebo (PBO) in subsets of subjects who were on DPP-4i (N = 316; mean age, 63 y; A1C, 8.1%; BMI, 32.3 kg/m2) or GLP-1 agonists (N = 95; mean age, 61 y; A1C, 8.1%; BMI, 37.4 kg/m2) as monotherapy or in combination with other AHAs. At Week 18, CANA 100 and 300 mg reduced A1C and body weight relative to PBO in both subsets. Overall adverse event (AE) rates were higher with CANA than with PBO in the DPP-4i subset, and comparable or lower with CANA relative to PBO in the GLP-1 agonist subset; serious AE rates were generally higher with CANA than with PBO in both subsets. Documented hypoglycemia rates were higher with CANA 100 and 300 mg than PBO in subjects on insulin, sulfonylurea, or meglitinide in the DPP-4i subset (17/70, 29/87, and 12/74) and the GLP-1 agonist subset (11/29, 11/22, and 4/26); only 2 and 1 CANA-treated subjects who were not on these concomitant agents reported hypoglycemia in the DPP-4i and GLP-1 agonist subsets, respectively. In summary, CANA added on to DPP-4i or GLP-1 agonists (+/- other AHAs) lowered A1C, reduced body weight, and was generally well tolerated in subjects with T2DM at 18 weeks. HANS-ULRICH HÄRING, LUDWIG MERKER, ELKE SEEWALDT-BECKER, MARC WEIMER, THOMAS MEINICKE, HANS J. WOERLE, ULI C. BROEDL, Tübingen, Germany, Dormagen, Germany, Biberach, Germany, Ingelheim, Germany A Phase III trial investigated the efficacy and safety of empagliflozin (EMPA) in patients with T2DM on metformin plus SU. Patients (mean age 57.1 [SD 9.2] years; mean BMI 28.2 [SD 5.3] kg/m2) were randomized doubleblind and treated with EMPA 10 mg (n=225), 25 mg qd (n=216) or placebo (PBO; n=225) for 24 weeks. The primary endpoint was change from baseline in HbA1c at week 24. Key secondary endpoints were change from baseline in weight and mean daily glucose (MDG) at week 24. EMPA significantly reduced HbA1c, weight and MDG vs PBO. Further analyses showed significant reductions in systolic blood pressure (SBP), FPG and 2h PPG vs PBO. Weight loss >5% was achieved by 27.6% of patients on EMPA 10 mg, 23.6% on EMPA 25 mg and 5.8% on PBO. Adverse events (AEs) were reported by 67.9%, 64.1% and 62.7% of patients on EMPA 10 mg, 25 mg and PBO, respectively. Hypoglycemia (plasma glucose ≤70 mg/dL and/or requiring assistance) was reported in 16.1% of patients on EMPA 10 mg, 11.5% on EMPA 25 mg and 8.4% on PBO; none required assistance. AEs consistent with urinary tract infection were reported in 10.3% of patients on EMPA 10 mg, 8.3% on EMPA 25 mg and 8.0% on PBO. AEs consistent with genital infection were reported in 2.7% of patients on EMPA 10 mg, 2.3% on EMPA 25 mg and 0.9% on PBO. To conclude, EMPA 10 mg and 25 mg for 24 weeks as add-on to metformin plus SU improved glycemic control, and reduced weight and SBP versus PBO, and were well tolerated in patients with T2DM. 1081-P A Novel Oral Peptide-Mimetic (UGP302) Enhance Insulin Action and Improves Glucose Homeostasis in Type 2 Diabetic Rats MICHAEL FEIGH, SARA T. HJULER, KIM V. ANDREASSEN, NOZER METHA, WILLIAM STERN, CLAUS CHRISTIANSEN, MORTEN A. KARSDAL, KIM HENRIKSEN, Herlev, Denmark, Boonton, NJ Controversy exists whether the peptide hormone salmon Calcitonin (sCT) exerts a diabetogenic action. We demonstrated that oral administration of sCT improved diabetic hyperglycemia in animal models of type 2 diabetes. UGP302 is a novel oral dual-action amylin and calcitonin receptor agonist with an enhanced pharmacological profile to native sCT. In here, we investiADA-Funded Research & Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. For author disclosure information, see page 829. Guided Audio Tour poster A279 POSTERS gate the anti-diabetic efficacy of oral UGP302 vs oral sCT in a proof of concept study using Zucker diabetic fatty (ZDF) rats. Male ZDF rats were treated with oral sCT or UGP302 (0.5, 1 or 2 mg/kg) or oral vehicle twice daily for 7 weeks. Fasting and non-fasted plasma glucose and HbA1c were determined to evaluate glucose homeostasis. Additionally, OGTT and ITT were performed to investigate treatment effects on glucose intolerance and insulin resistance. Oral UGP302 attenuated diabetic hyperglycemia during the intervention period. At study end, oral UGP302 at doses of 1 mg/kg and 2 mg/kg reduced fasting and non-fasted plasma glucose by approx 8 mmol/l and HbA1c by 1.6% when compared to vehicle (p < 0.001). Furthermore, oral UGP302 improved glucose intolerance during OGTT by reducing incremental area under the curve by approx 50% at doses of 1 mg/kg and 2 mg/kg (p < 0.001). In contrast, only at highest dosing regimen (2 mg/kg) did oral sCT induce a comparable glucoregulatory effect. Finally, oral UGP302, at doses of 1 mg/ kg and 2 mg/kg, exerts comparable insulin sensitizing effects and enhanced plasma glucose disposal during ITT by approx 23% compared to vehicle (p < 0.001). In contrast, a comparable alleviation of insulin resistance for oral sCT was only observed at 2 mg/kg dose. Oral UGP302 is a superior dual-action amylin and calcitonin receptor agonist to native salmon Calcitonin. Oral UGP302 attenuates diabetic hyperglycemia in ZDF rats by improving fasting and postprandial glycemic control and alleviating insulin resistance. These data indicate the clinical usefulness of oral UGP302 as anti-diabetic treatment. Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1083-P Favorable Gastrointestinal and Genitourinary Safety Profile of LX4211 Added-On to Metformin in a Phase 2b Study IKE OGBAA, GUI-LAN YE, PABLO LAPUERTA, BRIAN ZAMBROWICZ, PHILLIP BANKS, KENNY FRAZIER, ARTHUR SANDS, The Woodlands, TX LX4211 is a dual inhibitor of SGLT1 and SGLT2, the major glucose transporters in the gastrointestinal (GI) tract and kidney. It has shown a favorable GI and genitourinary (GU) safety profile in healthy subjects and small studies of patients with type 2 diabetes. In a 12-week dose ranging study, 299 patients with inadequately controlled T2DM on metformin were randomly assigned to LX4211 (75 mg qd, 200 mg qd, 200 mg bid, or 400 mg qd) or placebo (PBO). The GI and GU tolerability and overall safety of LX4211 were evaluated. POSTERS Clinical Diabetes/ Therapeutics 1085-P Empagliflozin Monotherapy Improves Glucose Control in DrugNaïve Patients With Type 2 Diabetes (T2DM) MICHAEL RODEN, JIANPING WENG, JENS EILBRACHT, BRUNO DELAFONT, GABRIEL KIM, HANS J. WOERLE, ULI C. BROEDL, Düsseldorf, Germany, Canton, China, Biberach, Germany, Reims, France, Ingelheim, Germany A Phase III trial investigated the efficacy and safety of empagliflozin (EMPA) in drug-naïve patients with T2DM. Patients (mean age 55.0 years; mean BMI 28.4 kg/m2) were randomized double-blind and treated with EMPA 10 or 25 mg qd (n=224 each), sitagliptin 100 mg qd (SITA; n=223), or placebo (PBO; n=228) for 24 weeks. Patients with HbA1c >10% (n=87) received openlabel EMPA 25 mg qd for 24 weeks. Primary endpoint was change from baseline in HbA1c at week 24. Key secondary endpoints were change from baseline in weight, systolic blood pressure (SBP) and diastolic BP (DBP) at week 24. EMPA significantly reduced HbA1c, weight, and SBP versus PBO. DBP changes were not significant versus PBO. Weight loss of >5% was achieved by 22.8% of patients on EMPA 10 mg, 29.0% on EMPA 25 mg, 6.3% on SITA and 4.4% on PBO. Proportion of randomized patients with adverse events (AEs) was similar with PBO (61.1%), EMPA (54.9-60.5%), and SITA (53.4%). Hypoglycemia (plasma glucose ≤70 mg/dL and/or requiring assistance) was reported by 1 (0.4%) patient per randomized group; none required assistance. AEs consistent with urinary tract infection were reported in 5.46.7% of randomized patients on EMPA, 5.2% on PBO and 4.9% on SITA. AEs consistent with genital infection were reported in 3.1-4.0% of randomized patients on EMPA, 0 on PBO and 0.9% on SITA. To conclude, EMPA 10 mg and 25 mg for 24 weeks reduced HbA1c, weight, and SBP versus PBO, and were well tolerated in drug-naïve patients with T2DM. In the safety population, the incidences of diarrhea and nausea were comparable between all LX4211-treated patients and PBO. Diarrhea was 7.2 vs 6.7% and nausea was 6.8 vs 5.0% in all LX4211-treated patients vs PBO, respectively. Diarrhea was not dose-dependent or related to initiation of dosing. Nausea and diarrhea were generally mild and resolved without treatment. Constipation and vomiting were less common with LX4211 vs PBO; constipation 3.4 vs 6.7%, and vomiting 0.4 vs 5.0%. Urinary tract infections occurred evenly among treatment groups. Vaginal yeast or bacterial infections were seen only with LX4211; they were mostly mild, their incidence was low, and none led to therapy discontinuation. There were no deaths, treatment related serious AEs, or episodes of hypoglycemia reported. LX4211 combined with metformin was safe and well-tolerated in this study. Studies of greater exposure duration are needed to characterize long term safety of LX4211. 1084-P Effects of Canagliflozin Added on to Basal Insulin +/- Other Antihyperglycemic Agents in Type 2 Diabetes JULIO ROSENSTOCK, MELANIE DAVIES, RICHARD DUMAS, MEHUL DESAI, MARIA ALBA, GEORGE CAPUANO, GARY MEININGER, Dallas, TX, Leicester, United Kingdom, Montreal, QC, Canada, Raritan, NJ Canagliflozin (CANA) is an SGLT2 inhibitor in development for the treatment of type 2 diabetes mellitus (T2DM). This exploratory analysis from the CANagliflozin cardioVascular Assessment Study (CANVAS; in T2DM with a history or high risk of cardiovascular disease) evaluated CANA 100 and 300 mg versus placebo (PBO) in a subset of subjects in the insulin substudy on stable, non-titrated basal (but not prandial) insulin (≥30 IU/d basal insulin at study entry) +/- other antihyperglycemic agents (N = 278; baseline mean age, 63 y; A1C, 8.3%; fasting plasma glucose [FPG], 159 mg/dL; BMI, 34.4 kg/ m2; insulin dose, 59 IU/d). At Week 18, CANA 100 and 300 mg reduced A1C relative to PBO (-0.86% and -0.89%); both CANA doses reduced FPG (by 2531 mg/dL) and body weight (by 1.8%-2.7%) versus PBO (Table). Post-baseline daily insulin dose (prior to glycemic rescue) was unchanged for 93% of PBOtreated subjects versus 85% and 86% of subjects with CANA 100 and 300 mg (mostly dose reductions with CANA). Overall AE, AE-related discontinuation, and serious AE rates were higher with CANA than PBO. The proportion of subjects with documented hypoglycemia (≤70 mg/dL or severe events) was higher with CANA 100 and 300 mg than PBO (42%, 43%, 25%), with severe hypoglycemia rates of 0%, 1%, and 2%, respectively. In summary, CANA added on to basal insulin improved glycemic control, reduced body weight, and was generally well tolerated, with some increase in hypoglycemia risk, in subjects with T2DM at 18 weeks. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1086-P Na Channel Blockers Exert Anti-Diabetic Effects via Lowering Glucagon Levels in ZDF Diabetic Rats YUN NING, MICHAEL VAN PETTEN, JENNY JIANG, LUIZ BELARDINELLI, ARVINDER K. DHALLA, Fremont, CA Ranolazine (RAN), an anti-anginal drug, is a Na channel blocker (NCB) that also has shown anti-diabetic effects in clinical and non-clinical studies. RAN appears to exert its anti-hyperglycemic effects via a novel glucagon-related mechanism. We tested the hypothesis that blockade of Na channels in & For author disclosure information, see page 829. A280 Guided Audio Tour poster ADA-Funded Research Glucago (pg/mL) Baseline Week 10 Vehicle (n=7) 130 ± 14 313 ± 15 GS-458967(n=8) 100 ± 4 254 ± 17 * RAN (n=8) 135 ± 10 243 ± 3 * Sitagliptin (n=8) 106 ± 10 221 ± 7 * FPG (mg/dL) Baseline Week 10 105 ± 5 315 ± 18 106 ± 3 166 ± 33 ** 103 ± 4 177 ± 25 ** 108 ± 4 223 ± 26 ** HbA1c (%) Baseline Week 10 1.7 ± 0.03 8.3 ± 0.2 1.7 ± 0.04 4.2 ± 0.5 ** 1.7 ± 0.04 4.3 ± 0.4 ** 1.7 ± 0.06 5.5 ± 0.8 ** 1089-P Glucagon-Like Peptide (GLP-1) Response to an Oral Glucose Tolerance Test in Impaired Glucose Tolerance: Does GLP-1 Play a Significant Role? 1087-P MEHMET SARGIN, HATICE DEGISMEZ, SAKIN TEKIN, ASUMAN ORCUN, BUKET TEKIN, NIHAL YUCEL, TULAY KARABAYRAKTAR, MEHMET ALIUSTAOGLU, Istanbul, Turkey Dissociation Between Metformin Plasma Exposure and its GlucoseLowering Effect: A Novel Gut-Mediated Mechanism of Action Glucagon-like peptide 1 (GLP-1) is an insulinotropic hormone that regulates glucose levels by potentiating glucose- induced insulin secretion and inhibiting glucagon release. There has been much speculation about the role of GLP-1 in the pathogenesis of type 2 diabetes, especially the progression from normal to impaired glucose tolerance. Our aim is to determine the effect of fasting and postchallange concentrations of total GLP-1 on postchallange glucose and insulin in impaired glucose tolerance (IGT). A total of 64 subjects (female/male: 47/17) participated in the cross-sectional study. Mean age was 46.8±7.2 years and body mass index (BMI) was 33.9±5.0 kg/ m2. Subjects were divided into two groups who had normal (n: 32) and IGT (n: 32). The two groups were matched according to gender, BMI and body fat percentage. We performed a 75 g oral glucose tolerance test (OGTT) with insulin, glucagon, c-peptide and GLP-1 measurements. Postchallange GLP-1 levels positively correlated with glucose (r: 0.59, p<0.001) and insulin (r: 0.39, p<0.005) concentrations, but this correlation was not significantly associated with glucose tolerance status. Area under curve (AUC) and time to reach maximum concentrations of total GLP-1 indicated no significant association with glucose tolerance status. Peak GLP-1 and insulin concentrations were achieved at 30. and 60. min in both groups, respectively. GLP-1 secretion does not seem to contribute significantly to the pathogenesis of IGT because of the lack of association of GLP-1 concentrations with glucose tolerance status. However, studies with large numbers of subjects are necessary in order to determine the role of GLP-1 in the pathogenesis of IGT. RALPH A. DEFRONZO, JOHN B. BUSE, TERRI KIM, SHARON SKARE, ALAIN BARON, MARK FINEMAN, San Antonio, TX, Chapel Hill, NC, San Diego, CA We designed a delayed release (DR) metformin (Met) tablet that targets the lower bowel to reduce Met bioavailability and test whether Met plasma exposure is required for its glucose-lowering effects. Met DR was compared to commercially available immediate release (IR) Met in a randomized, double-blind, crossover study. Subjects with type 2 diabetes (N=24, age 51 y, BMI 33 kg/m2) were withdrawn from oral diabetes therapy for 2 wks, then received each treatment orally twice daily (BID) for 5 days, separated by a 9—12 day washout: 1000mg IR, 1000mg DR, 500mg DR. Plasma Met, glucose, insulin, GLP-1 and PYY were measured from 7am—5pm (7am breakfast; 12pm lunch) at baseline (BL) and day 5. Compared to 1000mg IR, Met exposure was reduced by 45 and 57% with 1000mg and 500mg DR (N=19, both p<0.0001). The reduction in fasting plasma glucose from BL (197—200 mg/dl) was similar for all treatments (LSmean±SE: 22.5±6.8 mg/dl, 1000mg IR; -19.9±5.0 mg/dl, 1000mg DR; 16.4±3.8 mg/dl, 500mg DR; all p<0.01 vs. BL). The 10h glucose AUC was similarly reduced in all treatments by 8—14% (all p<0.0001 vs. BL). Insulin AUC was unchanged from BL. All treatments increased fasting and postprandial PYY and GLP-1 (all p<0.05 vs. BL). The GLP-1 AUC increased by 87, 62 and 69% and PYY AUC increased by 55, 38, and 46% for 1000mg IR, 1000mg DR and 500mg DR (all p<0.0001 vs. BL). Nausea (9%) and vomiting (9%) occurred with Met IR but not Met DR. Diarrhea (10—15%) occurred in all treatments. Despite substantial reductions in exposure, a Met DR formulation that targets the lower bowel was as effective as Met IR with better tolerability. Met IR and DR similarly increased plasma gut peptide levels. This evidence indicates that the glucose-lowering effects of Met are in a large part due to Met actions on gut enteroendocrine cells. By directly targeting the lower bowel, Met DR may provide maximum efficacy with improved tolerability at lower doses (≤1000mg total daily) than typically used with currently available formulations. 1090-P Linagliptin vs. Placebo Followed by Glimepiride in Type 2 Diabetes Patients With Moderate to Severe Renal Impairment MARKKU LAAKSO, JULIO ROSENSTOCK, PER-HENRIK GROOP, UWE HEHNKE, ILKKA TAMMINEN, SANJAY PATEL, MAXIMILIAN VON EYNATTEN, HANS-JUERGEN WOERLE, Kuopio, Finland, Dallas, TX, Helsinki, Finland, Ingelheim, Germany, Bracknell, United Kingdom 1088-P Renal impairment (RI) is a serious T2DM complication that restricts options for managing hyperglycemia and associated increased cardiovascular risk. This randomized double blind trial evaluated the efficacy and tolerability of the dipeptidyl peptidase 4 inhibitor linagliptin (LINA) in T2DM pts (HbA1c 7-10%) with moderate to severe RI (estimated glomerular filtration rate [eGFR] <60 mL/min/1.73m2; not on dialysis). Pts received LINA 5 mg qd (n=113) or placebo (PBO) (n=122) for 12 wks, then PBO pts were switched to glimepiride 1-4 mg qd (GLIM) and treatment continued to wk 52. Primary endpoint was HbA1c change from baseline at wk 12. At baseline, 63.4% were male, mean age 67±9 y, HbA1c 8.1±0.9% and eGFR 37±13 mL/min/1.73 m2. Most had T2DM for >10 y (76.4%) and were on insulin (85.8%). At wk 12, adjusted mean±SE HbA1c change with LINA was -0.50±0.06% (change with Features in Subjects With Increased Fasting Glucagon Concentration after Administration of a DPP-4 Inhibitor TOMOKO MORITA, CHIOKO MORISAWA, NATSUKO OSHITANI, YOSHIMI ABE, YOSHITAKA AKIYAMA, MASAKO YAZAWA, AKIFUMI KUSHIYAMA, SHOJI KAWAZU, MASAFUMI MATSUDA, Kawagoe, Japan, Tokyo, Japan, Kawagoe-shi, Japan Abnormal regulation of glucagon secretion is a feature of type 2 diabetes mellitus. Dipeptidyl peptidase 4 inhibitors (DPP4-Is) may ameliorate the glucagon regulation. After administration of a DPP4-I, however, an increased fasting glucagon conc. paradoxically has been observed in some patients. To identify the features in those subjects, we conducted oral glucose ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A281 POSTERS tolerance tests (OGTTs) in 10 such subjects (GlunH) and compared with 10 subjects without increased fasting glucagon conc (C). Total 20 patients (M/F:11/9, age=66±7 y.o., duration of diabetes=12±9 y., HbA1c=6.5±0.5%, BMI = 25±4kg/m2) received OGTTs. 16 patients received sitagliptin, and four received vildagliptin. Insulin secretion and insulin sensitivity indices obtained from OGTT were compared with the fasting glucose, glucagon, insulin, C-peptide, and proinsulin conc. Basal glucagon conc. before administration of a DPP4-I was slightly lower in the GlunH group (72±17 vs 82±21pg/ml, NS), while glucagon levels were significantly increased at least two weeks after DPP4-I administration (by +16±9 vs -7±12 pg/ml). FPG was decreased in both groups (104±20 to 102±17 vs 131±29 to 116±23 mg/dl in C). Fasting insulin conc. was significant lower in the GlunH group (3.1±2.2 vs 4.7±2.5 microU/ml in C, p<0.05) before DPP4-I and slightly increased by 1.8±2.0 vs -0.1±1.0 microU/ml in C, p=NS. Matsuda index was significantly higher in the GlunH group (15.6±6.8 vs 7.7±2.9 in C, p<0.01), while insulinogenic index and HOMA-IR were not statistically different between the two groups (0.11±0.13 vs 0.10±0.09, and 0.9±.0.8 vs 1.6±0.6, respectively). Increment of glucagon conc. 30 min after an oral glucose intake was similar (4.0±13.0 vs 4.4 ±25.0 pg/ml in C, p=NS). In conclusion, paradoxical increase of fasting plasma glucagon conc. after DPP4-I administration coincides with increased whole body insulin sensitivity, suggesting the counteracting mechanism to avoid hypoglycemia by GLP-1 activation even during fasting state. α-cells leads to glucagon suppression resulting in a glucose lowering effect. We determined the effects of another Na channel blocker, GS-458967 along with RAN and sitagliptin (positive control) in ZDF rats on glucagon, fasting plasma glucose (FPG), HbA1c levels and pancreatic morphology. Male ZDF rats (6 weeks old) were treated for 10 weeks with GS-458967 (0.6 mg/kg/d), RAN (150-180 mg/kg/d), or sitagliptin (30 mg/kg/d) added to chow. Vehicle-treated rats had hyperglucagonemia and hyperglycemia at week 10 (table). Chronic treatment with NCBs (GS-458967 and RAN) led to significant lowering in glucagon levels, which was associated with significant decreases in FPG and HbA1c. The anti-diabetic effect of NCBs was comparable to those of sitagliptin. Morphology data showed that pancreatic islets from vehicle-treated animals had less insulin-positive cells and more glucagon-positive cells. In contrast, all three compounds reversed the abnormal insulin:glucagon ratio by decreasing glucagon-positive and increasing insulin-positive cells. In conclusion, the anti-hyperglycemic effect of RAN is due to glucagon suppression mediated by blockade of Na channels in α-cells. Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1092-P Empagliflozin as Add-On to Metformin for 24 Weeks Improves Glycemic Control in Patients With Type 2 Diabetes (T2DM) HANS-ULRICH HÄRING, LUDWIG MERKER, ELKE SEEWALDT-BECKER, MARC WEIMER, THOMAS MEINICKE, ULI C. BROEDL, HANS J. WOERLE, Tübingen, Germany, Dormagen, Germany, Biberach, Germany, Ingelheim, Germany A Phase III trial investigated the efficacy and safety of empagliflozin (EMPA) in patients with T2DM on background metformin. Patients (mean age 55.7 [SD 9.9] years; mean BMI 29.2 [SD 5.5] kg/m2) were randomized double-blind and treated with EMPA 10 mg (n=217) or 25 mg qd (n=213) or placebo (PBO; n=207) for 24 weeks. The primary endpoint was change from baseline in HbA1c at week 24. Key secondary endpoints were change from baseline in weight and mean daily glucose (MDG) at week 24. EMPA significantly reduced HbA1c, weight and MDG vs PBO (table). Further analyses showed significant reductions in systolic blood pressure (SBP), FPG and 2h PPG vs PBO. Weight loss of >5% was achieved by 21.2% of patients on EMPA 10 mg, 23.0% on EMPA 25 mg and 4.8% on PBO. Adverse events (AEs) were reported by 57.1%, 49.5% and 58.7% of patients on EMPA 10 mg, 25 mg and PBO, respectively. Hypoglycemia (plasma glucose ≤70 mg/dL and/or requiring assistance) was reported in 1.8% of patients on EMPA 10 mg, 1.4% on EMPA 25 mg and 0.5% on PBO; none required assistance. AEs consistent with urinary tract infection were reported in 5.1% of patients on EMPA 10 mg, 5.6% on EMPA 25 mg and 4.9% on PBO. AEs consistent with genital infection were reported in 3.7% of patients on EMPA 10 mg, 4.7% on EMPA 25 mg and none on PBO. To conclude, EMPA 10 mg and 25 mg for 24 weeks as add-on to metformin improved glycemic control, and reduced weight and SBP versus PBO, and were well tolerated in patients with T2DM. POSTERS Clinical Diabetes/ Therapeutics PBO -0.08±0.07%: difference -0.42%, 95% CI -0.60 to -0.24; p<0.0001). In the 40 wk extension, HbA1c was lower with LINA than GLIM (Figure). Incidence of drug-related adverse events was similar in the first 12 wks (LINA 23.9%, PBO 24.6%), and lower with LINA in the extension (LINA 38.3%, GLIM 46.5%). Hypoglycemia was less frequent with LINA (LINA 57.9%, GLIM 69.3%). Mean adjusted weight increase after 52 wks was 0.06 kg (LINA) and 1.74 kg (PBO/GLIM). In conclusion, LINA was efficacious and well tolerated in T2DM pts with moderate to severe RI, with less hypoglycemia and relative weight loss vs GLIM. NCT01087502. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1091-P The Effect of Combination Linagliptin and Voglibose on Glucose Control and Body Weight KATIE HEADLAND, STEVEN VICKERS, SHARON CHEETHAM, ROBERT JONES, MICHAEL MARK, THOMAS KLEIN, Nottingham, United Kingdom, Biberach, Germany We investigated whether combination therapy with the dipeptidyl peptidase-4 inhibitor linagliptin and the α-glucosidase inhibitor (AGI) voglibose can improve glycemic and body weight control. Male ZDF-Leprfa/ Crl (diabetic fa/fa) rats fed standard chow were allocated (n=10/group) based on body weight and fasting plasma glucose (FPG). Rats were dosed (po) daily for 4 days in 2 studies - S1: vehicle, linagliptin (1 mg/kg), high-dose voglibose (10 mg/kg), or linagliptin + voglibose; S2: as S1 except low-dose voglibose (1 mg/kg). An oral sucrose tolerance test (4 g/kg po) was performed on Day 4 and body weight was recorded daily. Mean vehicle FPG levels on Day 4 were 7.39 and 8.18 mM in S1 and S2, respectively, and mean insulin levels were 2.01 and 3.76 ng/mL. Improved glucose control AUC (0-30 min) was observed with linagliptin (S1: -10%, S2: -17%; both p<0.05), voglibose (S1: -33%, p<0.001; S2: -18%, p<0.01), linagliptin + voglibose (S1: -33%, S2: -33%; both p<0.001) compared with vehicle. Improvements in glucose control were potentiated with linagliptin + low-dose voglibose compared with either drug alone (p<0.01). Plasma active GLP-1 was increased 5 min after the sucrose load with linagliptin (S1: 160%, p<0.01; S2: 144%, p<0.001) and linagliptin + voglibose (S1: 834%, S2: 639%; both p<0.001) compared with vehicle, and was larger than linagliptin or voglibose alone (p<0.001 for all). Compared with vehicle, linagliptin-induced improvements in glucose control were body-weight independent (S1: +0.7%, S2: -0.2%; both p=ns). In contrast, voglibose (S1: -3.0%; S2: -1.7%) and linagliptin + voglibose (S1: -3.4%; S2: -2.0%) reduced overall weight compared with vehicle (p<0.001 for all). Therapy with linagliptin + voglibose potentiates improvements in glucose control. This combination may minimize the side effects of AGIs because lower doses of voglibose may be required to maintain glycemic control. In addition, beneficial effects due to the supra-additive increase in active GLP1 levels may be evident. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. & For author disclosure information, see page 829. A282 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1093-P 1095-P Linagliptin Improved Glycemic Control Without Weight Gain or Hypoglycemia in Patients With Type 2 Diabetes Inadequately Controlled by a Combination of Metformin and Pioglitazone Metformin Use in Octogenarians With the increasing life expectancy more and more subjects with Type 2 diabetes are aged over 80. According to DECODE study, over 40% of octogenarians suffer from diabetes. There are no specific guidelines for the treatment of this group of elderly patients, and metformin use in this population is controversial. We have assessed cross-sectionally modes of type 2 diabetes therapy used in randomly selected subjects aged 8090 (n=158) and retrospectively analyzed 3-year safety of thereof. The comparative group comprised Type 2 diabetes subjects aged 60-70 years (n=112). The treatment modes and cardiovascular risk factors distribution is presented in the Table 1 (mean±SD or %). Symptomatic hypoglycemia was recorded in 21% of octogenarians and 45% of the controls (p<0.01) in the 3-year period preceding the cross-sectional analysis. No case of ketoacidosis or lactic acidosis was noted in the studied subjects. In conclusion, almost two thirds of octogenarians are treated safely with metformin. Moreover, glucose control and cardiovascular risk profile is more favorable in the octogenarians than in younger subjects, suggesting the effective use of available therapies. Therefore, very elderly subjects with Type 2 diabetes, despite differing at many instances from younger patients, can be effectively and safely treated with traditional therapies, including metformin. This multicenter, Phase IIIb, randomized, double-blind, placebo (PBO)controlled study examined the efficacy and safety of the DPP-4 inhibitor linagliptin (LINA) in type 2 diabetes patients inadequately controlled by a combination of metformin (MET) and pioglitazone (PIOG). All patients entered a 2-wk PBO run-in period before being randomized to receive either LINA 5 mg qd (n = 183) or PBO (n = 89) in addition to MET and PIOG for 24 wks. The primary endpoint was the change from baseline in HbA1c after 24 wks of treatment. Baseline demographics were similar between LINA and PBO groups. The majority of patients were Asian (68.8%), 51.5% were female, the mean ± SD age was 53.8 ± 9.3 years and mean BMI was 28.2 ± 5.3 kg/m2. The mean ± SD baseline HbA1c was 8.39 ± 0.84% in the LINA group and 8.47 ± 0.78% in the PBO group. After 24 wks of treatment with LINA, the PBOadjusted mean ± SE change from baseline in HbA1c was -0.58 ± 0.13% (p < 0.0001), which was largely driven by the Asian population (-0.90 ± 0.16%). In patients with baseline HbA1c ≥7.0%, 32.4% of patients in the LINA group and 13.8% in the PBO group achieved HbA1c <7.0% (odds ratio [OR] 2.94; p = 0.0033). Patients in the LINA group were more likely to achieve an HbA1c reduction of ≥0.5% (65.4% LINA, 49.4% PBO, OR 2.06; p = 0.0071). The PBOadjusted mean ± SE change from baseline in FPG at wk 24 was -10.4 ± 4.7 mg/dL (p = 0.0280). The % of patients who reported ≥1 adverse event (AE) was 62.3% with LINA and 48.3% with PBO. The % of serious AEs was 2.2% with LINA and 3.4% with PBO. Investigator-defined hypoglycemia occurred in 5.5% of the LINA group and 5.6% of the PBO group. No meaningful changes in mean body weight were noted for either group. In summary, LINA as addon therapy to MET and PIOG produced significant and clinically meaningful improvements in glycemic control, largely attributed to the results in the Asian population. There was no additional risk of hypoglycemia and no weight gain. NCT00996658 Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1094-P Pioglitazone Improves Indices of Skeletal Muscle Insulin Sensitivity Without Change in Energy Balance or Mitochondrial Metabolism in Patients With Type 2 Diabetes KEVIN E. CONLEY, SUDIP BAJPEYI, LAUREN M. SPARKS, STEVEN R. SMITH, Seattle, WA, Baton Rouge, LA, Orlando, FL Thiazolidinediones (TZDs) are oral anti·diabetic drugs that improve metabolic control in patients with type 2 diabetes (T2D) through the improvement of insulin sensitivity, energy balance and mitochondrial function. Here we evaluated the impact of pioglitazone in T2D patients on skeletal muscle indices of insulin sensitivity and metabolism after 8 weeks of either drug (n= 14) vs. placebo (n=8) treated T2D patients with low insulin sensitivity (GDR < 5). Paired non-invasive phosphorus magnetic resonance (31P MRS) and optical spectroscopic measurements were made on vastus lateralis muscle. Pioglitazone treatment resulted in significant improvements in cell indices of insulin sensitivity (lowering phosphodiesters, PDE, by 15 .7%, P<0.009) and glucose balance (lower cell pH, P<0.002). In contrast, stable cell and mitochondrial metabolism was evident from unchanged values in an index of cell energy balance (PCr/Pi), ATP and O2 fluxes, as well as mitochondrial ATP synthesis capacity. The finding of reduced PDE is consistent with the effect of TZDs on lipoprotein lipase, while lower cell pH indicates improved carbohydrate metabolism leading to greater cell lactic acid accumulation. Thus non-invasive tools applied in vivo indicate that pioglitazone treatment in T2D patients improves muscle insulin sensitivity and carbohydrate metabolism without impact on cell energy balance or mitochondrial metabolism. Supported by: Medical University of Lodz 1096-P A Promising Combination for Future Treatment of Type 2 Diabetes: Coadministration of Empagliflozin (SGLT-2 Inhibitor) With Linagliptin (DPP-4 Inhibitor) MATTHIAS KERN, NORA KLÖTING, ROLF GREMPLER, ERIC MAYOUX, MICHAEL MARK, THOMAS KLEIN, MATTHIAS BLÜHER, Leipzig, Germany, Biberach, Germany Combining different drug classes to improve glycemic control is one treatment strategy for type 2 diabetes. Here we investigated the effects of long-term treatment with empagliflozin (EMPA), a sodium glucose cotransporter-2 (SGLT-2) inhibitor in clinical development, alone and in combination with the DPP-4 inhibitor linagliptin (LINA; an approved antidiabetes drug) on whole body and organ-specific insulin sensitivity in db/db mice using euglycemic-hyperinsulinemic clamps. In 8-week old female db/db mice (n=15/group) treated for 8 weeks, the glucose disposal rate was improved in the 10 mg/kg/d EMPA group (5.9 mg/kg/min; p<0.001), the 3 mg/kg/d LINA group (3.4 mg/kg/min; p<0.01) and the EMPA+LINA group (combination 10 mg/kg/d EMPA plus 3 mg/kg/d LINA: 7.8 mg/kg/ min; p<0.001) compared with vehicle (1.9 mg/kg/min). Insulin-mediated suppression of hepatic glucose production (HGP) was significantly (p<0.05) higher in the EMPA (13.1 mg/kg/min) and EMPA+LINA groups (11.8 mg/kg/ min) compared with vehicle (26.3 mg/kg/min). LINA monotherapy decreased HGP (21.8 mg/kg/min), although statistical significance was not achieved. Tissue-specific labeled glucose uptake was higher in liver and kidney with EMPA (liver: 822 dpm/g/mL; kidney: 1580 dpm/g/mL; p<0.05 for both), LINA (liver: 935 dpm/g/mL; kidney: 1660 dpm/g/mL; p<0.05 for both) and Supported by: NIH ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A283 POSTERS MANDEEP BAJAJ, RONALD GILMAN, SANJAY PATEL, JOAN KEMPTHORNERAWSON, HANS-JUERGEN WOERLE, Houston, TX, East Providence, RI, Bracknell, United Kingdom, Ridgefield, NJ, Ingelheim, Germany Clinical Diabetes/ Therapeutics LESZEK CZUPRYNIAK, NATALIA SZYLLO, ELEKTRA SZYMANSKA-GARBACZ, MACIEJ PAWLOWSKI, MALGORZATA SARYUSZ-WOLSKA, JERZY LOBA, Lodz, Poland CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS EMPA+LINA (liver: 1040 dpm/g/mL; kidney: 2190 dpm/g/mL; p<0.01 for both) compared with vehicle (liver: 610 dpm/g/mL; kidney: 1030 dpm/g/mL). Glucose uptake into muscle and adipose tissue was not affected by any treatment. Improvements in liver triglyceride content were also shown with EMPA (9.1%; p<0.05), LINA (12.1%; p<0.01) and EMPA+LINA (5.7%; p<0.01) compared with vehicle (14.6%). In conclusion, EMPA+LINA is superior to the respective monotherapies in improving insulin sensitivity, in particular, during conditions of severe insulin resistance. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1097-P Single-/Multiple-Dose Pharmacokinetics (PK) and Pharmacodynamics (PD) of Omarigliptin, a Once-Weekly Dipeptidyl Peptidase-4 (DPP-4) Inhibitor, in Healthy Subjects POSTERS Clinical Diabetes/ Therapeutics RAJESH KRISHNA, CAROL ADDY, DANIEL TATOSIAN, XIAOLI S. HOU, ISAIAS N. GENDRANO, ASHLEY N. MARTUCCI, JOHN A. WAGNER, S. AUBREY STOCH, Whitehouse Station, NJ 1099-P Omarigliptin is an oral, once-weekly, DPP-4 inhibitor being developed for the treatment of type 2 diabetes mellitus. Two studies were conducted to evaluate safety, tolerability, PK and PD following single-/multipledose administration in healthy subjects. Two double-blind, randomized, placebo-controlled, multiple-panel, rising-dose studies were conducted. Male subjects received single doses (0.5-400 mg; n=6/panel), multiple doses (10, 25, 50, and 100 mg; n=6/panel) once weekly for three weeks, or matching placebo (n=2/panel) in the fasted state. Plasma was obtained to evaluate concentrations of omarigliptin, DPP-4 activity and active glucagon-like peptide (GLP)-1. Omarigliptin was generally well tolerated. There were no hypoglycemic events and no significant changes in safety parameters. Absorption was rapid with median Tmax of 1-3 hours and steady state plasma concentrations were achieved after ~2-3 weeks. Weekly AUC and Cmax displayed dose proportionality within the dose range studied at steady state. At steady state, ~71% of the 10-mg dose was excreted in urine, with an average renal clearance of ~2 L/hr. The point estimates of treatment difference (active vs. placebo) in DPP-4 inhibition ranged from 77%-89% at 168 hrs following the last of 3 once-weekly doses over the dose range studied. Omarigliptin administration resulted in ~2-fold increases in weighted average post-prandial active GLP-1. Administration of omarigliptin was generally well tolerated in healthy subjects, and the PK profile supports once-weekly dosing. Omarigliptin acts by stabilizing active GLP-1, which is consistent with its mechanism of action as a DPP-4 inhibitor. Efficacy and Safety of Lobeglitazone Monotherapy in Patients With Type 2 Diabetes Mellitus over 24 Weeks: Multicenter, Randomized, Double-Blind, Parallel-Group, Placebo Controlled Study SIN GON KIM, DOO MAN KIM, JEONG-TAEK WOO, HAK CHUL JANG, CHOON HEE CHUNG, KYUNG SOO KO, JEONG HYUN PARK, YONG SOO PARK, SANG JIN KIM, DONG SEOP CHOI, Seoul, Republic of Korea, Bundang, Republic of Korea, Wonju, Republic of Korea, Pusan, Republic of Korea, Guri, Republic of Korea, Cheonan, Republic of Korea Lobeglitazone is a novel PPAR- γ agonist with a thiazolidinedione structural motif. The aim of this study was to assess the glucose-lowering and lipidmodifying effects as well as the safety profile of lobeglitazone compared to placebo as a monotherapy in patients with type 2 diabetes. In this 24 week, multicenter, randomized, double-blind, parallel-group, placebo control, therapeutic confirmatory study, 173 patients were randomly assigned (a 2:1 ratio) to 0.5mg lobeglitazone (n= 115) or matching placebo (n= 58) orally once daily. The primary endpoint was the change in HbA1c from baseline to the end of treatment. The secondary endpoints included various glycemic parameters, lipid parameters and safety profile. This study is registered with ClinicalTrials.gov number NCT01001611. At 24 weeks, a significant reduction in HbA1c was observed with lobeglitazone versus placebo (-0.44% vs 0.16%, mean difference -0.6%, p<0.0001). The goal of HbA1c <7% was achieved significantly more in the lobeglitazone group compared to the placebo group (44% vs 12%, p<0.0001). Fasting plasma glucose (p <0.0001), homeostasis model assessment (HOMA) insulin resistance index (p=0.002) and quantitative insulin sensitivity check index (QUICKI, P<0.0001) were also improved in the lobeglitazone group. In addition, lobeglitazone treatment significantly improved triglycerides, HDL cholesterol, small dense LDL cholesterol, free fatty acid, Apo-B and Apo-CIII compared to placebo (p<0.01, respectively). More weight gain was observed in the lobeglitazone group than in the placebo group (0.89kg vs - 0.63kg, mean difference 1.52kg, p<0.0001). The safety profile was comparable between the two groups and lobeglitazone was also well tolerated. Lobeglitazone 0.5 mg showed a favorable balance in the efficacy and safety profile. The results support a potential role of lobeglitazone in treating type 2 diabetes. Supported by: Merck, Sharp & Dohme 1098-P Efficacy and Safety of Canagliflozin (CANA) in Subjects With Type 2 Diabetes Mellitus (T2DM) on Metformin (MET) and Pioglitazone (PIO) Over 52 Weeks THOMAS FORST, ROBERT GUTHRIE, RONALD GOLDENBERG, JACQUELINE YEE, UJJWALA VIJAPURKAR, GARY MEININGER, PETER STEIN, Mainz, Germany, Columbus, OH, Thornhill, ON, Canada, Raritan, NJ CANA is an SGLT2 inhibitor in development for the treatment of T2DM. This randomized, double-blind study enrolled subjects with T2DM on MET and PIO (N = 342) and included a 26-week core period (CANA 100 and 300 mg vs placebo [PBO]) and a 26-week extension period (n = 275; mean age, 56.9 y; A1C, 7.9%; fasting plasma glucose [FPG], 162.2 mg/dL; BMI, 32.5 kg/ m2; PBO group switched to sitagliptin 100 mg [PBO/SITA]). Efficacy data at 52 weeks are reported for CANA 100 and 300 mg (SITA used to maintain double-blind and as a control group, not as efficacy comparator); safety data are shown for both CANA doses and PBO/SITA. At Week 52, CANA 100 and 300 mg lowered A1C and FPG from baseline with a substantial proportion of subjects reaching A1C <7.0%; reductions in body weight and systolic BP, and increases in HDL-C, LDL-C, and triglycerides were also seen. Overall AE rates were lower with CANA 100 mg than CANA 300 mg and PBO/SITA (70%, 76%, 77%). Rates of serious AEs, AE-related discontinuations, UTIs, and hypoglycemia were similar across groups. Higher rates of AEs related to osmotic diuresis (ie, thirst, pollakiuria; <7% per AE), reduced intravascular volume (ie, postural dizziness, orthostatic hypotension; <2% per AE), and genital mycotic infections were seen in the pooled CANA group versus PBO/SITA. In summary, CANA improved glycemic control, reduced body weight, and was generally well tolerated in subjects with T2DM on MET and PIO over 52 weeks. Supported by: Chong Kun Dang Pharmaceutical Corporation 1100-P Small-Molecule Inducers of the Beta-Cell Master Gene Pax4 in Pancreatic Alpha Cells KAIHUI HU HE, JAMES SPOONAMORE, BRIDGET K. WAGNER, Cambridge, MA Type 1 diabetes is a severe metabolic disease caused by selective destruction of pancreatic insulin-producing beta cells. Small molecules that restore beta-cell mass or function may enable insulin independence, and improve the quality of life of type 1 diabetes patients. Recently, efforts to replenish beta-cell mass have focused on reprogramming other pancreatic cell types into beta cells. Previous in vivo studies in mice have shown that the misexpression of just one transcription factor, Pax4, in alpha cells converted them into beta cells, both during embryogenesis and adulthood. To answer the questions of whether small-molecule induction of Pax4 causes a similar alpha-to-beta conversion in mouse islets, and whether this result translates to human islets, we developed a qPCR-based assay for high-throughput screening to identify small molecules capable of increasing Pax4 expression in the mouse alpha-TC1 cell line. & For author disclosure information, see page 829. A284 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS Key secondary endpoints were changes from baseline in insulin dose and HbA1c at week 78. EMPA significantly reduced HbA1c at week 18 and 78, and insulin dose at week 78, vs PBO (Table). Further analyses showed reductions in FPG, body weight, and systolic BP (Table). Hypoglycemia (glucose ≤70 mg/dL and/or requiring assistance) was reported similarly in 36.1% of patients on EMPA 10 mg and 25 mg and 35.3% on PBO; 2 patients on EMPA 25 mg required assistance. AEs consistent with urinary tract infection were reported in 14.8%, 11.6% and 8.8% on EMPA 10 mg, EMPA 25 mg and PBO, respectively. AEs consistent with genital infection were reported in 7.7%, 5.2% and 1.8% on EMPA 10 mg, EMPA 25 mg and PBO, respectively. In conclusion, EMPA 10 mg and 25 mg for 78 weeks as add-on to basal insulin improved glycemic control, led to reductions in body weight without increase in hypoglycemia risk, and were well tolerated except for increased genitourinary infections. Hydrolysis probe-based detection of Pax4 mRNA resulted in a robust assay, with a Z’-factor of 0.7. We screened nearly 100,000 compounds derived from diversity-oriented synthesis (DOS) at the Broad Institute, treating cells for 3 days before cell lysis and qPCR detection. We reasoned that the structural diversity of DOS libraries, representing multiple chemical scaffolds, may increase the likelihood of targeting “undruggable” transcription factors, such as Pax4. Eighty compounds (hit rate 0.1%) showed significant increase in signal by qPCR, and retesting in cell lines and primary islets is underway. Small molecules identified in this screen may serve as tools to demonstrate alpha-to-beta transdifferentiation in human cells for the first time, and could provide a way to restore functional beta-cell mass in type 1 diabetes. Supported by: JDRF 1101-P Real-Life Effectiveness of Vildagliptin Compared to Sulfonylureas in Type 2 Diabetes Patients in Germany Clinical Diabetes/ Therapeutics Metformin is an established treatment for type 2 diabetes mellitus (T2DM) patients but therapy intensification is usually required over time. The EDGE-study compared effectiveness and safety of vildagliptin and other oral antidiabetic drugs in 45868 patients with T2DM not controlled by monotherapy under real life conditions. Add-on treatment was chosen by the physician based on patient’s need. Effectiveness was assessed by HbA1c drop and by a composite endpoint (HbA1c <7.0% without hypoglycemia and weight gain) after 12 months of treatment. Here, results for 8887 patients in Germany receiving vildagliptin or sulfonylurea (SU) add-on to metformin are presented. 7410 patients (age: 62.6±11.1 yr; baseline HbA1c: 7.8±1.6; BMI: 30.8±5.5 kg/m²) received study medication (metformin+ vildagliptin (4960) or metformin+SU (819)). HbA1c decreased significantly greater with vildagliptin compared to SU (vildagliptin:-0.69%; SU:-0.46%; Δ-0.23%; p<0.001) and a higher proportion of vildagliptin patients reached the composite endpoint (vildagliptin:25.5%; SU:17.7%; p< 0.001). The incidence of hypoglycemic events was 4-fold higher in the SU cohort when compared to the vildagliptin cohort. The data suggest, that in real life clinical practice vildagliptin is associated with a greater HbA1c-drop, less hypoglycemic events and a higher proportion of patients reaching target HbA1c without hypoglycemia and weight gain compared to SU. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1103-P Tofogliflozin Improves Insulin Resistance as well as Glucose Tolerance by Ameliorating Fatty Liver and Obesity ATSUSHI OBATA, NAOTO KUBOTA, TETSUYA KUBOTA, HIROYUKI SATO, YOSHITAKA SAKURAI, MASANORI FUKAZAWA, MASAYUKI SUZUKI, KIYOFUMI HONDA, YOSHIYUKI SUZUKI, SACHIYA IKEDA, KOHJIRO UEKI, TAKASHI KADOWAKI, Tokyo, Japan, Bunkyo-ku, Japan, Gotemba, Japan SGLT2 is mainly present in S1 segment of renal proximal tubule and accounts for approximately 90% of total urinary glucose reabsorption. To elucidate the anti-diabetic effects of SGLT2 inhibitors, we conducted long term administration of Tofogliflozin (Tofo), a novel SGLT2 inhibitor, to diet induced obese model mice. C57BL/6J mice were fed a high fat diet or a high fat diet which contains Tofo for 8 weeks. Although food intake was increased, body weight and fat mass were significantly decreased in Tofo treated mice. Blood glucose levels were significantly lower in Tofo treated mice with reduction of serum insulin levels. OGTT conducted after drug washout revealed improved glucose tolerance in Tofo treated mice with significantly lower insulin levels. HOMA-R was also significantly improved in Tofo treated mice. In addition, Tofo treated mice demonstrated elevated serum free fatty acid levels and diminished adipocyte size, which indicated accelerated lipolysis in Tofo treated mice. Reduced respiratory quotient and elevated serum ketone body suggested acceleration of β-oxidation in Tofo treated mice. In fact, CPT1α expression levels were significantly increased in the 1102-P Empagliflozin as Add-On to Basal Insulin for 78 Weeks Improves Glycemic Control With Weight Loss in Insulin-Treated Type 2 Diabetes (T2DM) JULIO ROSENSTOCK, ANTE JELASKA, FEI WANG, GABRIEL KIM, ULI C. BROEDL, HANS J. WOERLE, Dallas, TX, Ridgefield, CT, Ingelheim, Germany Empagliflozin (EMPA) is a selective SGLT2 inhibitor in development for the treatment of T2DM. We assessed the efficacy and safety of EMPA added-on to basal insulin in T2DM patients (mean age 58.8 yrs; HbA1c 8.2%; BMI 32.2 kg/m2) randomized to double-blind EMPA 10 mg (n=169) or 25 mg qd (n=155) or placebo (PBO; n=170) for 78 weeks. Basal insulin dose remained constant for the first 18 weeks, then adjustments were allowed at investigator discretion. Primary endpoint was change from baseline in HbA1c at week 18. ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A285 POSTERS MARKUS DWORAK, JEAN-BERNARD GRUENBERGER, GIOVANNI BADER, Nuernberg, Germany, Basel, Switzerland CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS liver of these mice. Decreased gene expressions of SCD-1, FAS, DGAT-1 and DGAT-2 were consistent with reduced liver triglyceride contents. As food intake of Tofo treated mice was increased, we also conducted a pair feeding experiment by adjusting the total amount of daily food intake to exclude the influence of overdosing. The pair feeding experiment revealed more significant results compared to the ad libitum experiment. Moreover, the expression levels of TNFα, MCP-1 and IL-6 in white adipose tissue were decreased in Tofo treated mice. In conclusion, Tofo may shift the energy consumption from glucose to lipid by increasing urinary glucose excretion and improve insulin resistance by reducing fat mass and liver triglyceride contents and ameliorating inflammation in white adipose tissue. 1104-P Empagliflozin in Patients With Type 2 Diabetes Mellitus (T2DM) and Renal Impairment (RI) POSTERS Clinical Diabetes/ Therapeutics ANTHONY H. BARNETT, AMBRISH MITHAL, JENNY MANASSIE, RUSSELL JONES, HENNING RATTUNDE, HANS J. WOERLE, ULI C. BROEDL, Birmingham, United Kingdom, Delhi, India, Berkshire, United Kingdom, Ingelheim, Germany A Phase III trial investigated the efficacy and safety of empagliflozin (EMPA) as add-on to existing therapy for 52 weeks in patients with T2DM and RI. Patients with mild RI (eGFR [MDRD equation] ≥60 to <90 mL/min/1.73 m2; n=290; mean age 62.6 years; mean BMI 31.5 kg/m2) received EMPA 10 or 25 mg qd or placebo (PBO). Patients with moderate RI (eGFR ≥30 to <60 mL/ min/1.73 m2; n=374; mean age 64.9 years; mean BMI 30.2 kg/m2) received EMPA 25 mg qd or PBO. The primary endpoint was change from baseline in HbA1c at week 24. Exploratory endpoints included changes from baseline in fasting plasma glucose (FPG), weight and blood pressure (BP) at week 24. EMPA significantly reduced HbA1c vs PBO at week 24. Further analyses showed significant reductions in FPG, weight and BP. At week 24, adverse events (AEs) were reported by 79.6%, 75.4% and 72.7% of all patients (including an exploratory group with severe RI [n=74] on EMPA 25 mg or PBO) on EMPA 10 mg, 25 mg and PBO, respectively. Hypoglycemia (plasma glucose ≤70 mg/dL and/ or requiring assistance) was reported in 23.5% of patients on EMPA 10 mg, 22.1% on EMPA 25 mg and 22.9% on PBO. AEs consistent with urinary tract infection were reported in 10.2% of patients on EMPA 10 mg, 9.0% on EMPA 25 mg and 8.2% on PBO. AEs consistent with genital infection were reported in 6.1% of patients on EMPA 10 mg, 2.5% on EMPA 25 mg and 1.3% on PBO. To conclude, in patients with T2DM and mild or moderate RI, EMPA reduced HbA1c, weight, and BP vs PBO, and was well tolerated. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1105-P Dapagliflozin as Monotherapy in Drug-Naïve Asian Patients With T2DM Inadequately Controlled on Diet and Exercise LINONG JI, JIANHUA J. MA, HONGMEI LI, TRACI A. MANSFIELD, CAROLINE L. T’JOEN, NAYYAR IQBAL, AGATA PTASZYNSKA, JAMES F. LIST, Beijing, China, Nanjing, China, Shanghai, China, Princeton, NJ, Braine-l’Alleud, Belgium Dapagliflozin (DAPA), a selective SGLT2 inhibitor, improves glycemic control in diverse patient populations, as monotherapy or in combination with other glucose-lowering drugs. We assess DAPA as monotherapy in drug-naïve Asian patients inadequately controlled on diet and exercise (HbA1c 7.0-10.5%) in a 24-week double-blind study (NCT01095653) that randomized patients to placebo (PBO; N=132) or DAPA (5 or 10 mg; N=128 and 133, respectively) groups. Baseline characteristics were balanced across groups. Most patients (89%) were Chinese. Median duration of T2DM was 0.2 years. At week 24 (LOCF), DAPA 5 and 10 mg provided dose-dependent, statistically significant reductions in HbA1c vs PBO (primary endpoint) and in FPG, PPG, body weight and proportion of patients with HbA1c <7% (secondary endpoints) (Table). Adverse events (AEs) were balanced across groups. Few patients had serious AEs or AEs leading to discontinuation. Hypoglycemia was uncommon (1.5, 0.8 and 0.8% in the PBO, DAPA 5 mg and 10 mg groups, respectively); none led to discontinuation. Genital infections were more common with DAPA; 0.8, 3.1 and 4.5% of PBO, DAPA 5 mg or 10 mg patients respectively. UTIs occurred in 3.0, 3.9 and 5.3% of patients, respectively. No AEs of renal failure or pyelonephritis and no deaths occurred. In summary, DAPA as monotherapy in drug-naïve Asian patients was well-tolerated, significantly improving glycemic control with the additional benefit of modest weight loss. & For author disclosure information, see page 829. A286 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS was generally well tolerated, and there were no hypoglycemic events. PK profiles in subjects with and without T2DM were similar to what has been observed in non-obese subjects (historical data) and are supportive of onceweekly dosing. Trough DPP-4 activity was inhibited by ~90%, post-prandial (PP) 4-hour weighted mean active GLP-1 concentrations were increased ~2fold, and post-meal glucose was significantly reduced relative to placebo in the treatment population. Omarigliptin was generally well tolerated in the treatment population and was associated with a significant reduction in PP glucose, which was achieved through inhibition of DPP-4 and increased active GLP-1. Omarigliptin may provide an important once-weekly treatment option for patients with T2DM. Dapagliflozin Efficacy and Safety at 24 weeks in Asian Patients Outcomes at 24 weeks Placebo Dapagliflozin−5mg Dapagliflozin−10mg (N=132) (N=128) (N=133) … HbA1c, % … 122 … n 127 8.15 (0.75) 127 BL mean (SD) 8.33 (0.93) −1.04 8.28 (0.93) Change−from−BL* −0.29 (−0.43,−0.16) (−1.18,−0.90) −1.11 (−1.24,−0.98) (95%CI) … −0.75 −0.82 (−1.01,−0.63)† Difference−vs−PBO (95%CI) … (−0.94,−0.56)† FPG, mg/dL … n 131 … … BL mean (SD) 166.6 (42.6) 127 128 Change−from−BL* 2.5 154.2 (31.8) 161.8 (43.0) (95%CI) (−1.9,6.9) −25.1 (−29.6,−20.6) −31.6 (−36.1,−27.2) Difference−vs−PBO … −27.7 (−34.0,−21.4)‡ −34.2 (−40.4,−27.9)‡ (95%CI) … 2-hr PPG§, mg/dL … … … n 105 100 105 BL mean (SD) 250.4 (63.6) 224.0 (58.4) 230.3 (67.4) Change−from−BL* 1.1 (−7.9,10.1) −46.8 (−56.0,−37.6) −54.9 (−63.8,−46.0) (95%CI) … −47.9 (−60.8,−35.0)‡ −56.0 (−68.7,−43.3)‡ Difference−vs−PBO … (95%CI) Body weight, kg n … … … BL mean (SD) 132 128 128 Change−from−BL* 72.18 (13.23) 68.89 (11.43) 70.76 (11.70) (95%CI) −0.27 (−0.72,0.18) −1.64 (−2.09,−1.18) −2.25 (−2.70,−1.80) Difference−vs−PBO … −1.37 (−2.01,−0.73)‡ −1.98 (−2.62,−1.34)‡ (95%CI) … Pts w HbA1c<7% … … … n 127 122 127 Proportion# 21.3% (14.8,27.8) 42.6% 49.8% (95%CI) … (34.2,51.1) (41.9,57.7) Difference−vs−PBO … 21.3% 28.5% (95%CI) (11.1,31.6)‡ (18.6,38.3)‡ … … … Safety data 84 (63.6) 79 (61.7) 81 (60.9) ≥1 AE, n (%) 1 (0.8) 3 (2.3) 3 (2.3) AE→disc, n (%) ≥1 SAE, n (%) 2 (1.5) 5 (3.9) 4 (3.0) SAE→disc,n(%) 0 1 (0.8) 2 (1.5) ≥1 hypo, n (%) 2 (1.5) 1 (0.8) 1 (0.8) *Data are adjusted mean change from baseline at 24 weeks derived from analysis of covariance with treatment group as an effect and baseline value as a covariate using last observation carried forward (LOCF) and excluding data after rescue therapy. †p<0.0001 vs placebo for primary endpoint tested at alpha=0.027 applying Dunnett’s adjustment. ‡p<0.0001 vs placebo for secondary endpoints tested following a sequential procedure at alpha=0.05. §As measured after ingestion of liquid meal. #Data are proportions adjusted for baseline HbA1c derived from logistic regression using LOCF and excluding data after rescue therapy; BL, baseline; disc, discontinuation; FPG, fasting plasma glucose; hr, hour; hypo, hypoglycemia; N, number of randomized patients who took at least one dose of double-blind study medication; n, number of randomized patients with non-missing baseline and week 24 (LOCF) values; PBO, placebo; PPG, post-prandial plasma glucose Supported by: Merck, Sharp & Dohme 1107-P Saxagliptin Is Well Tolerated and Improves Glycemic Control in Adult Diabetes Patients With GAD Antibodies 1108-P Pharmacokinetics (PK) and Pharmacodynamics (PD) of PF-04991532, a Hepatoselective Glucokinase Activator (GKA), Administered as Monotherapy in Japanese and Non-Japanese T2DM Patients DAVID KAZIERAD, ARTHUR BERGMAN, JEFFREY A. PFEFFERKORN, XIN WANG, TIMOTHY ROLPH, JAMES M. RUSNAK, Cambridge, MA, Groton, CT The purpose of this study was to investigate the PK, PD dose response and safety/tolerability of multiple oral doses of PF-04991532. Twice daily doses of placebo, 25, 75, 150 or 300 mg PF were administered to T2DM treatment naïve patients or patients washed off a single oral diabetic agent in a randomized fashion for 14 days (N=8 Japanese and N=8 non-Japanese subjects per treatment group). PF-04991532 AUC and C max increased proportionally with dose, and Tmax values ranged from 1-2 hours. The pharmaco kinetic profile was generally similar between Japanese and non-Japanese subjects. PF-04991532 was associated with reductions in mean daily glucose (MDG), fasting plasma glucose (FPG) and glucose excursion following a meal tolerance test (MTT) by Day 14 of dosing, with the largest MDG reductions occurring at the highest doses studied. Additionally, there was a trend for larger MDG reductions (~2-fold at the 300-mg BID dose) in Japanese subjects (N=38 naïve, N=2 washout) compared to non-Japanese subjects (N=12 naïve, N= 28 washout). PF-04991532 was found to be safe and well tolerated at all doses and there were no clinically meaningful changes in lab safety values, vital signs or ECG results. In conclusion, PF-04991532 was associated with robust effects on plasma glucose following 2 weeks of dosing to patients on no other anti-diabetic agents. Supported by: Bristol-Myers Squibb/AstraZeneca 1106-P Pharmacokinetic (PK) and Pharmacodynamic (PD) Effects of Multiple-Dose Administration of Omarigliptin, a Novel Once-Weekly Dipeptidyl Peptidase-4 (DPP-4) Inhibitor, in Obese Subjects With and Without Type 2 Diabetes Mellitus (T2DM) CAROL ADDY, DANIEL TATOSIAN, XIAOLI S. HOU, ISAIAS N. GENDRANO, ASHLEY MARTUCCI, MICHELLE GROFF, JOHN A. WAGNER, S. AUBREY STOCH, Whitehouse Station, NJ Omarigliptin is a potent, oral, long-acting DPP-4 inhibitor that is being developed as a once-weekly treatment for T2DM. This study was aimed to investigate the PK and PD effects of omarigliptin in obese subjects with and without T2DM. This was a randomized, double-blind, placebo-controlled, multiple-dose study of 50-mg omaragliptin given once weekly for 4 weeks. Subjects included 24 obese but otherwise healthy subjects (n=18 active/ n=6 placebo) and 8 obese subjects with T2DM (n=6 active/n=2 placebo; treatment-naïve, HbA1c ≥6.5% and ≤10.0%). Subjects were 45-65 years of age with 30 ≤ body mass index ≤ 40 kg/m2. Blood sampling occurred at select time points to evaluate the PK of omarigliptin, DPP-4 activity, active glucagon-like peptide (GLP)-1, and plasma glucose. Omarigliptin ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A287 POSTERS Latent autoimmune diabetes in adults (LADA) is recognized by the presence of glutamic acid decarboxylase (GAD) antibodies, and there is a need for effective therapies for these patients. Patients from 5 placebo (PBO)-controlled studies from the saxagliptin (SAXA) phase 3 program were pooled and classified by GAD status (GAD-positive: values ≥ lower limit of quantitation [LLOQ]; GAD-negative: values <LLOQ). To assess similarity in SAXA efficacy by GAD status, we analyzed treatment effects (glycated hemoglobin [A1C], fasting plasma glucose [FPG], 120-min postprandial glucose [PPG], adverse events [AEs], serious AEs, hypoglycemia) using analysis of covariance, with test for treatment by subgroup interaction. The GAD-positive group included 98 SAXA patients and 35 PBO patients; the GAD-negative group included 1849 SAXA and 727 PBO patients. Age, sex, and race were balanced across subgroups. For GAD-positive and GADnegative patients, baseline mean A1C was 8.2% in both subgroups, mean FPG was 177 and 170 mg/dL, and mean C-peptide was 3.3 and 3.4 nmol/L. SAXA showed greater decreases from baseline than PBO, with consistent treatment effects in GAD-positive and GAD-negative patients: mean change from baseline A1C was ‒0.64% and ‒0.62%, respectively (P=0.93) and PPG was ‒39.1 and ‒39.7 mg/dL, respectively (P=0.97). Differences in achieving A1C <7% were also consistent between GAD-positive and GAD-negative patients. There was a larger treatment effect for FPG for GAD-positive vs GAD-negative patients (‒33.9 vs. ‒13.8 mg/dL; P<0.01). General similarity was shown with SAXA vs PBO for both GAD categories for incidence of AEs (65% vs 80% for GAD-positive; 73% vs 70% for GAD-negative), serious AEs (2.0% vs 5.7% and 3.5% vs 3.4%, respectively), and hypoglycemic events (5.1% vs 5.7% and 7.4% vs 6.5%, respectively). SAXA was effective and generally well tolerated in patients with type 2 diabetes regardless of GAD status. These data suggest that SAXA is effective in patients with LADA. Clinical Diabetes/ Therapeutics PAOLO POZZILLI, RAFFAELLA BUZZETTI, ROBERT FREDERICH, NAYYAR IQBAL, BOAZ HIRSHBERG, Rome, Italy, Princeton, NJ CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS PF-04991532 Day 14 Pharmacodynamic Effects Dose MDG Change from FPG Change from Glucose AUC ratio baseline [90% CI] baseline [90% CI] following MTT (mg/dL) (mg/dL) (active/placebo)[90% CI] 25 mg -41.8 [-23.6, -60.0] -27.0 [-13.4, -40.6] 0.87 [0.81, 0.93] 75 mg -30.2 [-11.9, -48.4] -24.0 [-10.3, -37.6] 0.91 [0.85, 0.97] 150 mg -34.5 [-16.3, -52.7] -17.6 [-3.9, -31.2] 0.87 [0.81, 0.93] 300 mg -55.9 [-37.6, -74.2] -21.4 [-7.8, -35.1] 0.78 [0.73, 0.83] 1109-P No Effect of Exenatide or Sitagliptin Therapy on β-Cell Secretory Capacity in Early Type 2 Diabetes POSTERS Clinical Diabetes/ Therapeutics LALITHA GUDIPATY, CARISSA FULLER, NORA ROSENFELD, MICHEAL R. RICKELS, Philadelphia, PA The main pathophysiologic abnormalities in Type 2 Diabetes (T2D) are impaired tissue sensitivity to insulin action (i.e. insulin resistance) and impaired β cell insulin secretion. The impaired β-cell insulin secretion in T2D is currently best explained by a combination of reduced β-cell mass and a functional defect in insulin secretion from the available β-cells. Presently no anti-diabetogenic agent has been shown to prevent or reverse the progressive decline in β-cell insulin secretion, novel agents such as glucagons-like peptide-1 (GLP-1) hold some promise as GLP-1 enhances β cell insulin secretion and inhibits α-cell glucagon secretion in a glucosedependent fashion while exerting extra-pancreatic effects of GLP-1, most notably appetite suppression and delayed gastric emptying to limit glucose excursions following a meal. To evaluate GLP-1 effects on β-cell secretory capacity in vivo early in the progression of T2D, we performed glucose-potentiated arginine tests in 47 subjects with early T2D or pre-diabetes randomized receive GLP-1 analog exenatide, DPP4 inhibitor sitagliptin (allows for 2-fold increase in levels of active GLP-1) or a comparator insulin secretagogue glimepiride for 6 months. The acute insulin responses to 5 g IV arginine were measured at baseline and after 6 months of therapy during fasting (AIRarg) and 230 mg/dl (AIR pot) and 340 mg/dl (AIR max) glucose clamps. From before to after 6 months of therapy, there was no difference in the acute insulin responses to arginine with either exenatide or sitagliptin therapy. The mean difference after 6 months in AIRmax was -37 ± 180 uU/mL in the exenatide group vs. 8 ± 72 uU/mL in the sitagliptin group. This was not statistically different from the comparator insulin secretagogue, glimepiride where the mean difference in AIRmax was 57 ± 126 uU/mL. These data indicate that exenatide or sitagliptin therapy that increase GLP-1 levels do not have a significant effect on β cell secretory capacity in early Type 2 diabetic subjects. 1111-P Ultrasonography Modifications of Visceral and Subcutaneous Adipose Tissue After Pioglitazone or Glibenclamide Therapy Combined With Rosuvastatin in Type 2 Diabetic Patients Not Well Controlled by Metformin PAMELA MAFFIOLI, ANGELA D’ANGELO, TIZIANO PERRONE, ELENA FOGARI, GIUSEPPE DEROSA, Pavia, Italy The aim of this study was to compare pioglitazone or glibenclamide alone and in combination with rosuvastatin on hepatic steatosis in type 2 diabetic patients. After 3 months of metformin, patients were randomised to pioglitazone 15 mg twice a day or glibenclamide 5 mg twice a day for six months, then rosuvastatin was added for other six months. Patients underwent an ultrasound examination to evaluate steatosis degree, subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) diameter, an euglycemic hyperinsulinemic clamp to evaluate glucose infusion rate (GIR), and a blood sample collection to evaluate: glycemic control, fasting plasma insulin (FPI), lipid profile, adipocytokines at randomisation, and after 6, and 12 months. Both pioglitazone and glibenclamide improved glycemic control. Pioglitazone reduced FPI, while glibenclamide increased it. Pioglitazone increased GIR compared to glibenclamide therapy. Pioglitazone reduced total cholesterol, and triglycerides, and increased high density lipoprotein cholesterol. When rosuvastatin was added, there was a greater improvement with pioglitazone and rosuvastatin. Adiponectin was increased by pioglitazone, with a further increase when rosuvastatin was added. A significant decrease of leptin, and interleukin-6 was recorded with pioglitazone; a similar trend was maintained after rosuvastatin addition. The addition of rosuvastatin to pioglitazone decreased tumor necrosis factor-α. Pioglitazone + rosuvastatin were superior to glibenclamide + rosuvastatin in reducing high sensitivity C-reactive protein. Only pioglitazone decreased ultrasound parameters, and the addition of rosuvastatin further decreased them. We can conclude that pioglitazone was more effective than glibenclamide in improving inflammation and the hepatic steatosis indices. Supported by: Pennsylvania Dept. of Health 1110-P Canagliflozin (CANA) Lowers A1C and Blood Pressure (BP) Through Weight Loss-Independent (WL-I) and Weight Loss-Associated (WLA) Mechanisms LAWRENCE BLONDE, JOHN WILDING, JEAN-LOUIS CHIASSON, DAVID POLIDORI, GARY MEININGER, PETER STEIN, New Orleans, LA, Liverpool, United Kingdom, Montreal, QC, Canada, San Diego, CA, Raritan, NJ 1112-P CANA is an SGLT2 inhibitor that increases urinary glucose excretion, leading to decreased A1C and weight loss (WL) in subjects with type 2 diabetes. Systolic BP (SBP) also decreases, possibly due in part to diuresis. WL alone reduces A1C and SBP; WL-I and WL-A effects of CANA were estimated using a pooled analysis of 4 previously reported 26-week Phase 3 studies (N = 2250). Each study had 3 groups: placebo (PBO), CANA 100 mg, CANA 300 mg. Mean baseline values of A1C = 8.0%, body weight (BW) = 89 kg, and SBP = 128 mmHg were similar across groups. Within each group, mean changes in A1C and SBP were calculated for each decile of WL and ANCOVA analysis was done with ΔA1C or ΔSBP as response and ΔBW (%) as covariate. Greater reductions in A1C and SBP were seen with greater WL, with similar slopes in each group (Figure). WL-I effects were defined as the vertical distance between lines (ie, difference in response observed when WL is the same in each group). Total PBO-subtracted mean reduction in A1C was 0.72% with CANA 100 mg (0.60% [85%] of the effect was WL-I) and 0.90% with CANA 300 mg (0.75% [85%] WL-I). Total PBO-subtracted reduction in SBP was 3.5 mmHg with CANA 100 mg (2.0 mmHg [57%] WL-I) and 4.7 mmHg with CANA 300 mg (2.7 mmHg [59%] WL-I); relative contributions of fat and fluid loss to WL-A effects cannot be determined from these data. CANA lowers both A1C and BP through WL-I and WL-A mechanisms; a greater proportion of the reduction in A1C is WL-I compared to SBP. The Effect of Vildagliptin Relative to Sulphonylureas in Muslim Patients With Type 2 Diabetes Fasting During Ramadan: The VIRTUE Study MONIRA AL AROUJ, AHMED A.K. HASSOUN, RITA MEDLEJ, FARUQUE M. PATHAN, INASS SHALTOUT, MANOJ S. CHAWLA, PARNIA GERANSAR, SASHKA HRISTOSKOVA, SHELLEY DITOMMASO, MAHOMED Y. KADWA, Dasman, Kuwait, Dubai, United Arab Emirates, Beirut, Lebanon, Dhaka, Bangladesh, Cairo, Egypt, Mumbai, India, Basel, Switzerland VIRTUE was a prospective, observational study that assessed the risk of hypoglycemic events (HE) in patients with type 2 diabetes mellitus (T2DM) who fast during Ramadan. The study enrolled Muslim patients with T2DM in two cohorts: vildagliptin (n=684) or sulphonylurea (SU; n=631), both given either as monotherapy or dual therapy with metformin. Patients were recruited from 10 countries in the Middle East and Asia. Mean HBA1c was 7.34% and 7.44% in the vildagliptin and SU cohorts, respectively. The primary endpoint was the proportion of patients with ≥1 HE during Ramadan. Secondary endpoints included change in HbA1c, change in body weight, treatment adherence and overall safety. The study showed that significantly (p<0.001) fewer patients experienced 1 or more HE with vildagliptin (n=36/669, 5.4%) compared with SU (n=123/621, & For author disclosure information, see page 829. A288 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS index; (iii) having a trial duration of at least 12 weeks. Finally, a total of 16 articles reporting 22 comparisons met inclusion criteria, which included 6,567 study participants, 3,560 treated with sitagliptin and 3,007 treated with placebo or a comparator drug. In all studies, HOMA-IR and HOMA-β were analyzed as weighted mean differences (WMD) under a fixed effects model because no heterogeneity was found. The meta-analysis showed HOMAIR and HOMA-β was significantly improved (WMD = -0.3, 95% confidence interval (CI), -0.5 to -0.1; WMD = 11.7, 95%CI, 9.9 to 13.5, respectively). There was possible publication bias in the analysis of HOMA-β (Egger’s regression test: P=0.05), but not in HOMA-IR (P=0.92). In conclusion, it is suggested that sitagliptin may have a beneficial effect on insulin resistance as well as beta-cell function. 19.8%). No patients reported a grade 2 HE (requiring third-party assistance) with vildagliptin compared with 4 patients receiving SUs (p=0.053). The mean change in HbA1c, pre- to post-Ramadan, was -0.24% in the vildagliptin group compared with +0.02% in the SU group (-0.26% between treatment difference; p<0.001). Greater body weight reductions were observed with vildagliptin compared with SU (-0.76 vs. -0.13 kg; -0.63kg between treatment difference; p<0.001). Treatment adherence was high with a similar number of missed doses between cohorts. A higher proportion of SU-treated patients experienced adverse events compared with vildagliptin (22.8% vs. 10.2%), with the difference driven by hypoglycemia, which was the most common adverse event. In conclusion, vildagliptin therapy was associated with significantly fewer patients experiencing hypoglycemia compared with SU therapy in this large cohort of fasting patients with T2DM. This outcome is particularly meaningful when viewed in the context of good glycemic and weight control accompanied by favorable tolerability observed in vildagliptin-treated patients who fast. 1115-P Glycemic Control of Pioglitazone in a Diabetic Rat Model Is Improved when Combined With Empagliflozin Dipeptidyl Peptidase-4 Inhibitor Ameliorates PPARγ Agonist-Induced Body Weight Gain via Reduction in Body Fat Mass, but Not Fluid Volume TAKAHIRO MASUDA, JAN CZOGALLA, AKIKO EGUCHI, MICHAEL A. ROSE, MARIA GERASIMOVA, VOLKER VALLON, La Jolla, CA Peroxisome proliferator-activated receptor γ (PPARγ) agonists, like pioglitazone (PG) are effective anti-diabetic drugs, but induce fluid retention and body weight (BW) gain. Dipeptidyl peptidase-4 (DPP-4) inhibitors are new anti-diabetic drugs that enhance renal Na+ and fluid excretion. We therefore examined whether the DPP-4 inhibitor alogliptin (AL) ameliorates PG-induced BW gain. Male SV129 mice were treated with vehicle (VE; repelleted diet), PG (220 mg/kg diet), AL (300 mg/kg diet) or combination of PG and AL (COM)(n=8-10 per group) for 14 days. * P<0.05 vs VE; # P<0.05 vs PG. PG increased, and AL lowered BW vs VE (delta BW: 4.2±0.8*, -0.1±0.5*, and 1.7±0.3 %). COM (0.8±0.8 % #) prevented the increase in BW observed in PG alone. COM decreased hematocrit to a greater degree than PG alone (46.9±0.9 # vs 49.9±0.3 %; each *) while AL had no effect vs VE (50.4±0.3 vs 50.8±0.3 %). PG tended to increase and COM significantly increased fluid content of abdominal fat pads vs VE (9.1±0.9, 9.5±0.6*, 7.4±1.0 %). COM increased body fluid content determined by bioimpedance spectroscopy (BIS) to a similar degree as PG (in absolute terms and related to BW; each by 10-20% vs VE; each *). BIS revealed that AL had no effect on fat mass (FM) vs VE (FM/BW: 22.4±1.4 vs 20.5±1.9 %), but COM decreased FM to lower levels than PG (7.6±1.0 # vs 13.7±2.4 %; each *; similar results were obtained for absolute values). The latter was confirmed by MRI. AL, but not PG, modestly decreased food intake (by 6-8%*) and plasma free fatty acid concentrations (FFAs) (by 25-36%*). Vice versa, PG, but not AL, increased mRNA expression of thermogenesis mediator uncoupling protein 1 (UCP1) in epididymal white adipose tissue (by 808-1160%*). In summary, DPP-4 inhibition did not attenuate PPARγ agonist-induced fluid retention, but prevented BW gain via a further reduction in FM. The latter may involve distinct and complementary effects on determinants of FM like plasma FFAs and adipose tissue UCP1 expression. 1116-P Response to Dapagliflozin by Baseline HbA1c in Head-to-Head Comparisons ROBERT R. HENRY, ALEXANDER V. MURRAY, MICHAEL A. NAUCK, KATJA ROHWEDDER, EVA JOHNSSON, SHAMIK J. PARIKH, AGATA PTASZYNSKA, San Diego, CA, Greensboro, NC, Bad Lauterberg, Germany, Wedel, Germany, Wilmington, DE, Princeton, NJ Dapagliflozin (DAPA), a selective SGLT2 inhibitor, reduces hyperglycemia by removing excess blood glucose via the urine. The objective of this analysis was to directly compare the glycemic efficacy of DAPA with other oral antidiabetic drugs (OADs) across a wide range of baseline HbA1c. DAPA was compared with other OADs in 2 randomized, double-blind clinical trials: Study 1 (NCT00660907), a 52-week trial of DAPA (≤10 mg) vs glipizide (GLIP, ≤20 mg) as add-on to metformin (MET); Study 2 (NCT00859898), a 24-week trial of DAPA 10 mg vs MET extended release 2000 mg as monotherapies. Mean baseline HbA1c was 7.7% (Study 1) and 9.1% (Study 2). Non-inferiority of DAPA to comparators for changes in HbA1c (primary endpoint) was previously described. Here we present analyses by baseline HbA1c levels (Figure). Both DAPA and comparators provided greater reductions in HbA1c in the higher baseline HbA1c subgroups. Furthermore, within each study, reductions by DAPA and comparator were similar for each baseline HbA1c category. In both studies, DAPA had a low intrinsic propensity for hypoglycemia (Study 1: DAPA 3.5% vs GLIP 40.8%; Study 2: DAPA 0.9% vs MET 2.9%), consistent with the insulin-independent mechanism of action. In summary, as with the overall populations, DAPA demonstrated glycemic responses similar to other OADs in subgroups stratified by baseline glycemic control, producing larger reductions in HbA1c for individuals starting with higher baseline levels. Supported by: Takeda Pharmaceuticals U.S.A., Inc. 1114-P Effects of Sitagliptin on Insulin Sensitivity and Beta-Cell Function by the Homeostasis Model Assessment in Patients With Type 2 Diabetes: A Meta-Analysis TAKAFUMI TSUCHIYA, MISATO KAWANISHI, SHINTAROU SAKURAI, TATSUHIKO SUZUKI, KENNICHIROU HORI, TOMOKO TERASAWA, RIKA NARUSE, KENJI HARA, KOHZO TAKEBAYASHI, TOSHIHIKO INUKAI, Koshigaya, Japan The homeostatic model assessment is a method used to quantify insulin sensitivity and beta-cell function. The objective of this study is to evaluate the impacts of a dipeptidyl peptidase-4 (DPP-4) inhibitor, sitagliptin, on markers of insulin sensitivity and beta-cell function in patients with type 2 diabetes mellitus. For a meta-analysis, randomized controlled trials using sitagliptin as an add-on therapy or a monotherapy treating type 2 diabetes mellitus were identified through electronic searches up to and including December 2011. Studies were included if they met the following inclusion criteria: (i) randomized controlled trials that compared sitagliptin with placebo or a sitagliptin/metformin combination drug with metformin monotherapy in adult people with type 2 diabetes; (ii) reporting homeostatic model assessment of insulin resistance (HOMA-IR) or of beta-cell (HOMA-β) ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A289 POSTERS Empagliflozin is a novel, potent and selective sodium glucose cotransporter 2 (SGLT2) inhibitor in clinical development for the treatment of type 2 diabetes. We investigated whether empagliflozin could be used in combination with pioglitazone, an insulin sensitizer, to improve glycemic control in type 2 diabetic animals (ZDF rats). ZDF rats were treated once daily for 15 days with empagliflozin (1 mg/kg), pioglitazone (10 mg/kg) or both these drugs or vehicle. At day 15, blood glucose level increased by 18.6% in the control group, but decreased significantly by 36.9% (p<0.001) and 48.2% (p<0.05) vs control with empagliflozin and pioglitazone monotherapy, respectively, and by 75.7% (p<0.001) vs control with the combination. At the end of the study, fed blood glucose levels were 26.7mM, 16.9mM and 13.8mM in the control, empagliflozin and pioglitazone groups, respectively, and 6.5mM in the combination group. HbA1c increased from 3.19% to 5.79% between day 0 and day 15 in the control group. The increase in HbA1c was significantly less in the empagliflozin and in the pioglitazone group than in the control group, while the combination group had the lowest increase. Baseline-corrected changes in HbA1c were 2.6%, 1.94%, 1.03% and 0.5% with vehicle, empagliflozin, pioglitazone and the combination, respectively. This study shows that empagliflozin, with a mechanism of action that is independent of insulin pathways and beta-cell function, can be combined with an insulin sensitizer to improve glycemic control in animal model of type 2 diabetes. In conclusion these results suggest that, empagliflozin, could be combined with pioglitazone for a better control of glucose homeostasis in type 2 diabetic patients. 1113-P Clinical Diabetes/ Therapeutics LEO THOMAS, ROLF GREMPLER, MICHAEL MARK, ERIC MAYOUX, GERD LUIPPOLD, Biberach, Germany Supported by: Novartis Pharma AG CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS Blood Glucose (mmol/L*min) Insulin (pmol/L*min) Intact Proinsulin (pmol/L*min) GLP-1 (pmol/L*min) PAI-1 (ng/mL*min) cGMP (pmol/L*min) ADMA (µmol/L*min) Glimepiride -522±111 * 22595±560 * 1359±642 -7±185 863±455 92±98 -1.8±2.7 Linagliptin -309±68 * -8007±4204 $ -1771±426 *,$ 4149±658 *,$ -410±261 $ 111±128 -4.1±2.3 Change from baseline in the pp AUC (mean±SEM; *=p<0.05 vs. baseline; $ =p<0.05 between groups) Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1118-P Glucagon Receptor Antagonist LGD-6972 Is Efficacious in Streptozotocin-Induced Diabetic Mice Glucagon plays an important role in insulin dependent diabetes and glucagon receptor (GCGR) antagonism has been effective in clinical trials with T2DM patients. Glucagon is also critical in insulin deficient, streptozotocin (STZ) treated mice as observed in experiments with GCGR -/mice. We examined whether pharmacological antagonism of the GCGR would have similar effects in STZ treated mice. LGD-6972 is a potent and selective antagonist of the GCGR that has previously demonstrated activity in pre-clinical models of T2DM. Male BALB/c mice were injected with STZ and left untreated until hyperglycemia was confirmed. Hyperglycemic mice were orally dosed with LGD-6972 or vehicle for 28 days. Non-fasting blood glucose was significantly reduced in LGD-6972 treated mice (average = 80 mg/dL throughout the experiment). Larger effects were observed on fasting blood glucose (Vehicle: 382 mg/ dL, LGD-6972: 234 mg/dL, non-diabetic controls: 118 mg/dL). An oral glucose tolerance test revealed a large increase in glucose levels and a slow recovery to baseline levels in STZ mice relative to non-diabetic controls, consistent with an insulin deficient state. LGD-6972 treated mice had lower glucose levels than STZ mice throughout the glucose tolerance test (average decrease of 100 mg/dL) but displayed a similar delayed return to baseline. Terminal measurements of HbA1c, 3-hydroxybutyrate, and free fatty acids (FFA) were significantly elevated in STZ mice indicating prolonged hyperglycemia and ketoacidosis. LGD-6972 partially restored HbA1c and 3-hydroxybutyrate to non-diabetic levels but did not alter FFA levels.In addition to efficacy observed in T2DM patients, pharmacological inhibition of GCGR may be an effective treatment for T1DM and could potentially be used in an insulinsparing regimen. GCGR antagonism was not able, however, to completely normalize the hyperglycemic state, unlike the effects observed with GCGR -/mice. Further studies to investigate these differences may be useful. POSTERS Clinical Diabetes/ Therapeutics ERIC G. VAJDA, LIN ZHI, KEITH B. MARSCHKE, La Jolla, CA Supported by: Bristol-Myers Squibb/AstraZeneca 1119-P The Beneficial Effect of Metformin on β-Cell Function and its Relationship to Insulin Sensitivity in Lean Subjects With Newly Diagnosed Type 2 Diabetes 1117-P Linagliptin Improves Beta Cell Function and Vascular Biomarkers in Patients With Type 2 Diabetes YAN BI, DALONG ZHU, HUIJIE YANG, GUOYU TONG, Nanjing, China MICHAEL MITRY, ANDREAS PFÜTZNER, THOMAS FORST, Mainz, Germany Studies with metformin have suggested beneficial effects on β-cell function in obese patients, but it remains unclear whether a similar effect is observed in non-obese individuals. Here we investigated the effects of metformin or glipizide GITS on β-cell function in lean and overweight patients with newly diagnosed type 2 diabetes. 160 newly diagnosed patients with fasting plasma glucose 7.0-13.0 mmol/l and body mass index < 30 kg/m2 were randomized to metformin or glipizide GITS and treated for 24 weeks. β-Cell function (insulinogenic index as measured by the ratio of area under the curve (AUC) of insulin and C-peptide to glucose AUC) and insulin sensitivity (Matsuda index ISIM) were assessed during the standard meal tolerance test before and after therapy. Analysis was done by intention-to-treat. We found that HbA1c was significantly reduced to a similar extent with metformin and glipizide GITS treatment (-1.7±1.3% metfromin vs. -2.0±1.5% glipizide GITS, P=0.256). Plasma adiponectin and glucagon-like peptide-1 response were markedly increased from baseline in two groups. Metformin significantly increased ISIM from baseline (P=0.029), but subgroup analysis identified that the increase in ISIM in lean subgroup was not significant (P=0.134). When corrected for alterations in insulin sensitivity, parameters regarding insulin secretion were improved significantly (P<0.05) and comparable with metformin and glipizide GITS, which the trends persisted across lean and overweight subgroups with each treatment. Postprandial metabolism and the amelioration of endothelial function are major components in the development of vascular complications in T2DM. The goal of this study was to investigate the effects of Linagliptin compared to Glimepiride on beta cell function and several vascular biomarkers in the postprandial state. Thirtynine patients on metformin treatment (age: 64±7 years; duration of DM: 7.9±4.5years, 27 male; HbA1c: 57.2±6.9 mmol/mol; mean±SD) were randomized to add Linagliptin 5mg (n=19) or Glimepiride up to 4 mg (n=20) for a study duration of 12 weeks. Blood glucose, insulin, intact proinsulin, GLP-1, PAI-1, c-GMP, and ADMA levels were measured fasting and postprandial (pp) at 30 minutes intervals for a duration of 5 h. The areas under the curve (AUC0-300min) were calculated. HbA1c, fasting and pp glucose levels improved in both groups. Insulin, intact proinsulin, and PAI-1 levels increased during treatment with Glimepiride and decreased during treatment with Linagliptin. In both groups a slight increase in pp c-GMP and a slight decrease in pp ADMA could be observed (n.s.). A linear correlation could be observed between pp intact proinsulin and pp c- GMP (r=-0.14; p<0.0001), and between pp GLP-1 and pp PAI-1 (r=-0.08; p=0.003). In addition to metformin Linagliptin significantly improves pp beta cell function and reduces pp PAI-1 levels vs. Glimepiride. & For author disclosure information, see page 829. A290 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS Therefore, metformin’s beneficial effects on β-cell function were observed in both lean and overweight subjects with newly diagnosed type 2 diabetes, and which were independent of the change in insulin sensitivity, in part, in lean subjects. This study provides evidence-based data to support metformin use in non-obese patients with type 2 diabetes as an agent which can improve both insulin sensitivity and β-cell function. 1121-P The Hepatoselective Glucokinase (GK) Activator PF-04991532 Provides Additive Improvements in Glycemic Control With Exercise in Goto-Kakizaki Rats Drugs that exert additional glycemic control in combination with exercise training have the potential to improve patient outcomes. Recently, metformin was reported to show a lack of benefit in combination with exercise relative to either treatment alone in pre-diabetic subjects. Pharmacologic agents which improve glycemic control and complement the increased peripheral glucose disposal derived from exercise training have the potential to show additive benefit with physical activity. We hypothesized that PF-04991532, a hepatoselective glucokinase activator which improves hepatic glycemic control, would exert additive benefit in combination with exercise training. In the present study, the interplay between PF-04991532 and exercise training was examined. Male Goto-Kakizaki rats were exercised for 12 wks at ~50% peak power (1 hr/day, 5 days/wk). This exercise intervention was weight neutral but significantly improved markers of glycemic control [fasting plasma glucose (FPG), fasting insulin and HOMA-IR] relative to sedentary controls. 48-hrs following the last exercise session, a single oral dose of PF-04991532 (60 mg/kg) or vehicle was administered by oral gavage. An oral glucose tolerance test was performed 1-hr post compound dose. PF-04991532 produced improvements in fasting insulin levels, FPG, postprandial glucose and HOMA-IR in both exercise trained and sedentary rats. The effect of PF-04991532 was additive with exercise training in improving fasting insulin levels, FPG, post-prandial glucose and HOMA-IR. Improvements in post-prandial insulin glucose ratio (a marker for insulin sensitivity) were only observed in exercised but not non-exercised rats, and this effect was potentiated by PF-04991532.These data suggest that PF-04991532 may work additively with exercise to improve glucose homeostasis; subsequent studies will evaluate the effect of chronic GK activation in combination with exercise training. CHRISTOPHER KOVACS, VEERASWAMY SESHIAH, ROS SWALLOW, RUSSELL JONES, HENNING RATTUNDE, HANS J. WOERLE, ULI C. BROEDL, St. John’s, NL, Canada, Chennai, India, Berkshire, United Kingdom, Ingelheim, Germany A Phase III trial investigated the efficacy and safety of empagliflozin (EMPA) in patients with T2DM on a stable pioglitazone regimen with or without metformin. Patients (mean age 54.5 [SD 9.8] years; mean BMI 29.2 [SD 5.5] kg/m2) were randomized and treated double-blind with EMPA 10 mg (n=165), 25 mg qd (n=168) or placebo (PBO; n=165) for 24 weeks. Primary endpoint was change from baseline in HbA1c at week 24. Key secondary endpoints were change from baseline in FPG and weight at week 24. EMPA significantly reduced HbA1c, FPG, and weight versus PBO. Further analyses showed significant reductions in blood pressure (BP) vs PBO. Weight loss of >5% was achieved by 18.8% of patients on EMPA 10 mg, 13.7% on EMPA 25 mg and 5.5% on PBO. Adverse events (AEs) were reported by 67.3%, 71.4% and 72.7% of patients on EMPA 10 mg, 25 mg and PBO, respectively. Hypoglycemia (plasma glucose ≤70 mg/dL and/or requiring assistance) was reported in 1.2% of patients on EMPA 10 mg, 2.4% on EMPA 25 mg and 1.8% on PBO; none required assistance. AEs consistent with urinary tract infection were reported in 17.0% of patients on EMPA 10 mg, 11.9% on EMPA 25 mg and 16.4% on PBO. AEs consistent with genital infection were reported in 8.5% of patients on EMPA 10 mg, 3.6% on EMPA 25 mg and 2.4% on PBO. To conclude, EMPA 10 mg and 25 mg for 24 weeks as add-on to pioglitazone with or without metformin reduced HbA1c, FPG, weight and BP versus PBO, and was well tolerated in patients with T2DM. 1122-P Effect of Canagliflozin (CANA) 300 mg on C-peptide Clearance (CL Cpep) SUE SHA, DAVID POLIDORI, TIM HEISE, JAYA NATARAJAN, EUNICE ARTIS, SHEAN-SHENG WANG, NICOLE VACCARO, PAUL ROTHENBERG, ALIN STIRBAN, Raritan, NJ, San Diego, CA, Neuss, Germany CANA, a sodium glucose co-transporter 2 (SGLT2) inhibitor, lowers plasma glucose and A1C in type 2 diabetic subjects by increasing urinary glucose excretion. In clinical trials, CANA doses ≥100 mg improved indices of β-cell function (βCF) based on circulating C-peptide and glucose concentrations (eg, HOMA2-%B and meal tolerance test-based indices). However, as C-peptide is mostly cleared renally, these results could be confounded if CANA affects CLCpep. Therefore, this 2-period crossover study assessed whether a single 300 mg CANA dose alters CLCpep in 10 healthy subjects (mean age = 36 y, body weight = 79 kg). After an overnight fast, subjects received iv injection of 150 µg of synthetic human C-peptide 3 h after CANA or placebo (P) dosing and 1 h after initiating a constant iv somatostatin infusion to suppress endogenous C-peptide secretion. Serum C-peptide was measured frequently for 3 h; profiles were fitted to a 2-compartment model to obtain CLCpep. Urinary C-peptide was measured to calculate renal clearance (CLR). Serum C-peptide profiles were similar following CANA or P treatment (Fig). CLCpep was comparable with CANA and P (mean [SD] = 190 [37] vs. 197 [30] mL/min; ratio (CANA/P) [90% CI] = 96% [93%; 99%]). CLR and other kinetic parameters were similar between treatments. CANA was well tolerated. These results show that CANA 300 mg does not meaningfully alter CLCpep and confirm that C-peptide-based measurements of insulin secretion are appropriate for assessing βCF in CANA-treated subjects. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A291 POSTERS 1120-P Empagliflozin as Add-On to Pioglitazone With or Without Metformin Improves Glycemic Control in Patients With Type 2 Diabetes (T2DM) Clinical Diabetes/ Therapeutics TRENTON T. ROSS, DAVID A. BEEBE, DEREK M. ERION, PAUL A. AMOR, YIMIN ZHU, GREGORY J. TESZ, QINGYUN YAN, SANTOS CARVAJAL-GONZALEZ, TIMOTHY P. ROLPH, JEFFREY A. PFEFFERKORN, WILLIAM P. ESLER, Cambridge, MA CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1125-P 1123-P POSTERS Clinical Diabetes/ Therapeutics higher with CANA 300 mg (12.5%). There were no severe hypoglycemic episodes. In conclusion, CANA added to ongoing SU monotherapy produced significant improvements in a number of efficacy parameters important in T2DM management with an increased incidence of several specific AEs probably related to SGLT2 inhibition. Plasma Aminoacids Following a Mixed Meal in T2D Patients: Effect of Sitagliptin Treatment The Effect of High Dose Vitamin D Supplementation on Glycemic Control in Healthy Type 2 Diabetic Patients ELZA MUSCELLI, SILVIA FRASCERRA, ARTURO CASOLARO, SIMONA BALDI, WALT GALL, ELE FERRANNINI, Pisa, Italy OHK HYUN RYU, SUNGHWA LEE, MOON-GI CHOI, JAE MYUNG YU, DOO-MAN KIM, SUNG-HEE IHM, HYUNG JOON YOO, Chuncheon, Republic of Korea, Seoul, Republic of Korea, Pyungchon, Republic of Korea Aminoacid (AA) metabolism is altered in T2D, and fasting levels of α-hydroxybutyrate (α-HB), a biomarker for insulin resistance and incident diabetes, have been suggested to track AA metabolism. We investigated the changes in AA and α-HB induced by a mixed meal (MTT) and the effects of chronic sitagliptin treatment. Forty-seven T2D patients (age 56±7 yrs, BMI 29.9±4.2 kg/m2) were randomized to sitagliptin (100 mg/d) or placebo for 6 weeks. Seven age- and BMI-matched nondiabetic subjects served as control (CT). During a 5-hr MTT, plasma branched-chain AA (BCAA) peaked earlier in T2D than in CT (75[25] vs 62[3] mmol/L over 2 hrs, median[interquartile range], p=0.05), and rose higher (31[23] vs 19[24] mmol/L, ∂AUC over 5 hrs, p=0.05). Fasting α-HB was higher in T2D than CT (7.5[2.7] vs 5.9[1.3] µg/mL, p=0.04), increased at 60 min in T2D but not CT, and its ∂AUC over 2 hrs postMTT was higher in T2D (24[99] vs -41[86] µg/mL, p=0.005). Plasma FFA declined sharply during MTT, but their AUC was greater in patients (53±16 vs 35±10 mEq/L over 3 hrs, p=0.005). Sitagliptin treatment was associated with lower fasting and postmeal glycemia, and improved β-cell sensitivity (52.9[52.1] vs 31.1[35.9] pmol.min-1.m-2.mM-1, p=0.002) and insulin sensitivity (273[69] vs 239[69] mL.min-1.m-2, p=0.0002). As compared to placebo, both BCAA (∂AUC -6.4[21.1] vs 3.4[17.9] mmol.L-1.5h, p=0.01) and α-HB (∂AUC -114[250] vs 114[428] µg.mL-1.2h, p=0.002) decreased with sitagliptin, and meal-induced FFA suppression was greater. Changes in BCAA and α-HB AUCs were both reciprocally related to changes in insulin sensitivity (rho=-0.50 and -0.35, respectively, p≤0.03) T2D is associated with a hyperaminoacidemic response to a mixed meal, which circulating α-hydroxybutyrate levels track. Sitagliptin-induced glycemic improvement was associated with consensual reductions in BCAA and α-HB excursions (as well as FFA suppression) in parallel with improved insulin sensitivity, confirming that α-HB is a readout of metabolic overload. Background: Vitamin D deficiency is very common in Korea.Epidemiologic studies showed the striking inverse relationship between vitamin D level and insulin resistance/glucose intolerance. But there are few interventional studies that evaluate the glucose-lowering effect in diabetic patients. Aims: We investigated the glucose-lowering effect of high dose vitamin D in type 2 diabetic patients. Methods: We enrolled type 2 diabetic patients who took antidiabetic agents or managed diabetes by lifestyle modification. The participants was also satisfied the following criteria: age (<70 years) and HbA1C (<8.5%). We excluded patients who took vitamin D or calcium supplements. We also excluded chronic kidney disease patients (in men: Cr>1.5mg/dL, in women: Cr>1.4mg/dL), and physically disabled patients. We randomized 158 participants into 2 groups: intervention group (n=79) (Vitamin D 2000IU/ day+Calcium 200mg/day) or placebo group (Calcium 200mg/day) (n=79). Results: There were no difference in baseline characteristics, such as HbA1c (HbA1c 7.29±0.59% vs 7.30±0.61%),vitamin D level (10.1±3.9ng/ml vs 10.8±5.1ng/ml) diabetes duration, oral antidiabetic agents, and outdoor physical activity in placebo and intervention group, respectively. Although the achieved vitamin D level was different (15.6±7.1 vs 30.2±10.8ng/ml; P<0.001), there were no difference in HbA1c (7.27±0.87% vs 7.40±0.90%; P=0.415) between placebo and intervention group. Conclusions: In healthy type 2 diabetic patients, long-term supplementation of high dose vitamin D was no effect in glycemic control. 1124-P Canagliflozin (CANA) in Subjects With Type 2 Diabetes Mellitus (T2DM) Inadequately Controlled on Sulfonylurea (SU) Monotherapy: A CANVAS Substudy GREG FULCHER, DAVID MATTHEWS, VLADO PERKOVIC, KENNETH W. MAHAFFEY, ROBERT WEISS, JULIO ROSENSTOCK, GEORGE CAPUANO, MEHUL DESAI, WAYNE SHAW, FRANK VERCRUYSSE, GARY MEININGER, BRUCE NEAL, Sydney, Australia, Oxford, United Kingdom, Durham, NC, Auburn, ME, Dallas, TX, Raritan, NJ, Beerse, Belgium 1126-P Phase 1 Clinical Evaluation of the CCR2 Antagonist CCX872-B PIROW BEKKER, TREVOR CHARVAT, SHICHANG MIAO, LISA LOHR, TIMOTHY SULLIVAN, ZHENHUA MIAO, JAY POWERS, JUAN JAEN, THOMAS J. SCHALL, Mountain View, CA Within a pre-specified substudy of the Canagliflozin Cardiovascular Assessment Study (CANVAS), we evaluated the efficacy, safety and tolerability of CANA in subjects with T2DM inadequately controlled on SU monotherapy (n = 127; 45 placebo [PBO], 42 CANA 100 mg and 40 CANA 300 mg). The primary endpoint was change from baseline in HbA1c at week 18. Mean baseline age was 65 y, males (57%), HbA1c 8.4%, BMI 29.9 kg/m2, FPG 10.0 mmol/L, systolic blood pressure (SBP) 136.3 mmHg and eGFR 69.3 mL/ min/1.73 m2. At randomization, 35% were taking glimepiride, 29% glyburide/ glibenclamide and 27% gliclazide modified release (ND). Compared to PBO, both CANA doses significantly improved glycemia (Table). The overall incidence of adverse events (AEs) with CANA 100 mg (26.2%) was lower relative to PBO (64.4%) and CANA 300 mg (45%). Specific AEs of male and female genital infections, pollakiuria, and thirst were more common with CANA. AEs leading to discontinuation in the CANA groups were low (300 mg; 2.5%, 100 mg; 2.4%), with none reported in the PBO group. Documented hypoglycemia was similar with CANA 100 mg (4.1%) and PBO (5.8%) and CCX872-B is an orally administered, specific antagonist of the C-C chemokine receptor 2 (CCR2). This receptor and its most prominent chemokine ligand MCP-1 (also called CCL2) have been implicated in a variety of diseases, including various forms of renal inflammation/fibrosis and type 2 diabetes. We have shown that CCX872-B is effective in improving renal disease, as measured by albuminuria and several histologic measurements such as podocyte number, glomerular size, glomerular basement membrane thickness, and macrophage infiltration in several mouse models of diabetic nephropathy, including the db/db, BTBR ob/ob, and eNOS-/-db/db mouse models. CCX872-B also reduces hyperglycemia in mouse models of diabetes, such as the db/db mouse model. We describe the results of a randomized, double-blind, placebo-controlled, single- and multiple ascending dose Phase 1 clinical trial in 40 healthy volunteers. The primary objective of the trial is to evaluate the safety and tolerability of orally administered CCX872-B. Secondary objectives include evaluation of the pharmacokinetic (PK) and & For author disclosure information, see page 829. A292 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS pharmacodynamic profiles of CCX872-B. Oral doses of 3 mg up to 300 mg CCX872-B or placebo are given in sequential single and 7-day multiple dose periods in the trial. CCX872-B is given as hard gelatin capsules. A total of 40 subjects have been enrolled. CCX872-B was well tolerated at all dose levels tested in the study. No serious adverse events and no withdrawals from the study due to adverse events have been observed. The preliminary PK profile indicates that CCX872 plasma concentrations increased relatively dose-proportionately. CCX872-B appeared to be readily absorbed following oral dosing. The Tmax was 2 to 3 hours, Cmax ranged from 148 ng/mL at 3 mg to 12800 ng/mL at 300 mg CCX872-B, AUCinf ranged from 2830 hr.ng/mL to 270000 hr.ng/mL, and t1/2 was approximately 35 hours. In summary, orally administered CCX872-B was well tolerated in humans at doses ranging from 3 mg to up to 300 mg. CCX872-B has an excellent human PK profile. Lipid parameters did not consistently improve with either compound; mild increases in triglycerides and VLDL-cholesterol (vs. PBO) with RO838 were not significant. Both compounds were well tolerated. Slight changes in systolic blood pressure seen with 5 mg RO151 (decreases in day, night and 24h values) and 200 mg RO838 (increases in semi-supine values which were not repeated in the 24 hour ABPM) were not significant. Increased concentrations of ACTH and adrenal androgen precursors were found with RO151, but not with RO838. In conclusion, modest metabolic improvements were seen, particularly with the higher dose of RO151. Longer studies are needed to investigate potential benefits and risks of these compounds. 1127-P Administration of a GPR40 (FFAR1) Agonist plus Metformin Produces Glucose Normalization and Insulin Sensitization in a Preclinical Model of Insulin Resistance 1129-P CS-1050, a Novel PPARγ Ligand, Does Not Elicit the Cardiac Side Effects Associated With Thiazolidinediones LY2881835 is a potent, efficacious, and selective GPR40 agonist which lowers glucose levels and enhances insulin secretion when examined in preclinical models of T2D. In contrast, metformin (MET) lowers glucose levels via multiple mechanisms including insulin sensitization and reduced hepatic glucose production. LY2881835, MET, or LY2881835 plus MET were administered orally to Zucker fa/fa rats, a rodent model of insulin resistance, for 14 days. OGTTs were performed on days 1, 7 and 14 with fasting glucose levels measured 60 minutes following drug administration, i.e., just prior to administration of the oral glucose bolus. On days 1, 7, and 14 the AUCs for glucose were significantly lower in animals receiving LY2881835 or MET compared to the vehicle-treated fa/fa controls and equal to glucose AUCs for the learn controls. However, glucose AUCs for animals receiving the combination of LY2881835 plus MET were significantly lower compared to glucose AUCs for both the fa/fa and lean controls. Glucose levels, fasting and during the OGTTs, were never below 97 mg/dl for any animal receiving treatment during the study. These results demonstrate that the combination of LY2881835 plus MET normalizes glucose levels in a preclinical model of insulin resistance without producing hypoglycemia. Insulin AUCs during the OGTTs were elevated in animals receiving LY2881835 and decreased in animals receiving MET (trends, not significant) reflecting their differences in glucose lowering mechanisms. Animals receiving LY2881835 plus MET demonstrated insulin AUCs that were significantly lower than those in animals receiving only LY2881835 on days 7 and 14. The enhanced insulin sensitization with combination therapy may explain the significantly lower glucose in these animals. In summary, LY2881835 plus MET produces normalization of glucose levels during an OGTT in a preclinical model of insulin resistance without a risk of hypoglycemia. Type 2 diabetes (T2DM) is increasing in epidemic proportions and requires multiple therapeutic options, including insulin sensitizers. Thiazolidinediones (TZD), such as rosiglitazone (RSG), are highly useful to enhance insulin sensitization, but have detrimental effects of edema leading to heart failure and weight gain limiting their use. CS-1050, a derivative of cercosporamide, is a potent and selective non-TZD PPARγ partial agonist. Similar to RSG, CS-1050 has anti-diabetic effects but, unlike RSG, CS-1050 did not induce plasma volume increase or cardiac enlargement in rat models of T2DM. We studied cardiac side effects of CS-1050 and RSG in 3-month-old male B6.129S2-Serpine1tm1Mlg/J mice lacking plasminogen activator inhibitor-1, infused with angiotensin II (AngII; 2.5 ug/kg/min) via subcutaneous osmotic pumps, which caused cardiac hypertrophy and extensive fibrosis. Seven days after AngII infusion, mice were assigned to: 1) control (n=20); 2) 0.6g RSG/kg chow (n=30); or 3) 1.0g CS-1050/kg chow (n=19) for 27 days. Survival rates in CS-1050 (84%) and control (90%) mice were higher than RSG-treated mice (53%, p<0.05). Echocardiography at day 21 revealed reduced ejection fraction in RSG-treated mice (38%, p<0.05) compared to CS-1050 (49%) and control mice (51%). Heart weight and trichrome-stained fibrosis were similar in all groups. Expression of marker genes for heart failure (Anp), fibrosis (Tgfβ) and extracellular matrix genes were not significantly different among the groups. However, mRNA levels of arrhythmia gene K+ channel Kcna5 were decreased (p<0.05) in RSG vs. CS-1050 and control mice. In this model of cardiac fibrosis and hypertrophy, RSG increased mortality, depressed cardiac function and decreased expression of a key gene involved in maintenance of cardiac rhythm, while CS-1050 did not have these effects. The selective PPARγ partial agonist CS-1050 may have a different cardiac safety profile than TZDs for the treatment of T2DM. 1130-P An Integrated Analysis of Weight Loss With Combination Naltrexone/Bupropion Therapy by BMI (Obesity) Classification 1128-P Metabolic Effects and Safety of Two Selective 11β-HSD1 Inhibitors (RO5093151 (RO151) and RO5027838 (RO838) in Metformin-Treated Patients With Type 2 Diabetes (T2D) CAROLINE APOVIAN, COLLEEN BURNS, BRANDON WALSH, KRISTIN TAYLOR, Boston, MA, La Jolla, CA Naltrexone sustained-release (SR) / bupropion SR (NB) significantly reduced body weight vs placebo (PBO) in the four, 56-week, Phase 3 trials of overweight/obese patients (BMI ≥27 and ≤45 kg/m2). This integrated analysis investigated weight loss in these 4 trials when stratified by baseline BMI (obesity class): Class-I (30.0-34.9 kg/m2), Class-II (35.0-39.9 kg/m2), ClassIII (≥40 kg/m2). Treatment group differences (LS mean±SE) in the modified ITT-LOCF population (≥1 post-baseline weight on study drug) were evaluated by ANCOVA with treatment, study, and baseline values as covariates. Baseline characteristics (mean±SD) were similar between treatment arms (NB [n=2043] or PBO [n=1319): 81% female; 79% Caucasian; 46±11 y; 36±4 kg/m2; obesity Class-I: 37%, Class-II: 36%, Class-III: 25%. Completion rate was 66% for NB and 59% for PBO. At Week 56, NB resulted in significantly greater weight loss vs PBO (-7.0 ±0.2% NB vs -2.3±0.2% PBO; p<0.001) and proportion of patients who attained ≥5% weight loss (53% NB vs 21% PBO; p<0.001). NB patients experienced favorable shifts in BMI class from baseline relative to PBO (NB vs PBO: improved = 45% vs 20%, no change = 53% vs 74%, worsened = 2% vs 6%; p<0.001), and a larger proportion of NB vs PBO subjects shifted to a non-obese BMI from Class-I obesity (17% NB vs 6% PBO) and Class-II obesity (3% vs 1%). Of note, weight loss with NB was similar across the 3 obesity classes (NB range: -6.1% to -7.3%; PBOcorrected range: -4.0% to -5.0%, all p<0.001), as was achievement of ≥5% weight loss (NB range: 49% to 54%, odds ratio relative to PBO: 3.5 to 4.3; LINDA MORROW, TIM HEISE, MARCUS HOMPESCH, THOMAS R. PIEBER, HANSULRICH HAERING, CHRISTOPH KAPITZA, MARKUS ABT, MARKUS RAMSAUER, SABINE FUERST-RECKTENWALD, Chula Vista, CA, Neuss, Germany, Graz, Austria, Tübingen, Germany, Basel, Switzerland Two selective 11β-HSD1 inhibitors (RO151 and RO838) were assessed in a randomized, controlled study in patients with T2D who either received placebo (PBO) (21 pts), RO151 5mg BID (24 pts) or 200mg BID (20 pts), or RO838 50 mg QD (21 pts) or 200mg QD (24 pts) for 28 days plus a stable dose of metformin. Metabolic assessments (including oral glucose tolerance and/or standardized meal tests) and safety assessments (including ACTH stimulation test) were done at baseline and around 14 and 28 days of treatment. Key demographics at baseline were similar between groups (mean ranges: age 53-57 yrs, BMI 32-33 kg/m2, diabetes duration 6.5-9.4 yrs, FPG 170 -174 mg/dL ). Despite the short treatment duration, both RO151 and RO838 showed trends for improved HbA1c. Improvements in several insulin sensitivity parameters were seen with 200 mg RO151 but were not significant. While there was a slight decrease in body weight with PBO (-0.28 kg), greater reductions (-0.86 to -1.67 kg) occurred in all active treatment groups (reaching statistical significance vs.PBO in the RO151 200mg BID group only (LSmean -1.39 kg, 95% CI -2.54, -0.23 kg; p=0.019). ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A293 POSTERS YANYUN CHEN, MIN SONG, PRANAB MAITI, CHAFIQ HAMDOUCHI, ANNE REIFEL MILLER, Indianapolis, IN, Bangalore, India Clinical Diabetes/ Therapeutics MARICELA R. RAMIREZ, ALAN R. COLLINS, YUELAN REN, CHRISTOPHER J. LYON, MASANORI KUROHA, KAZUSHI ARAKI, JUN TANAKA, SATOKO WAKIMOTO, NAOKO NAKAI, TAKAHIRO NAGAYAMA, JUN OHSUMI, HUBERT CHOU, WILLA HSUEH, Houston, TX, Tokyo, Japan, Edison, NJ CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS all p<0.001). The safety/tolerability profile of NB32 was consistent with its individual components; the most common adverse events were nausea (32%), constipation (19%), headache (18%), and vomiting (11%). Thus, the clinically significant weight loss associated with NB was consistent across a wide BMI range, and resulted in an approximately 3-fold greater transition from Class-I/II obese to non-obese BMI. 1131-P LX4211 Efficacy Improves With Increasing Baseline Hemoglobin A1C POSTERS Clinical Diabetes/ Therapeutics IKE OGBAA, HALLIE ROZANSKY, PABLO LAPUERTA, BRIAN ZAMBROWICZ, PHILLIP BANKS, KENNY FRAZIER, ARTHUR SANDS, The Woodlands, TX Baseline hemoglobin A1C (A1C) influences the choice of antidiabetic therapy, as well as its efficacy, in patients with type 2 diabetes (T2DM). An unmet medical need remains for antidiabetic agents, with low intrinsic risk for causing hypoglycemia, which can provide clinically meaningful A1C reductions across a wide range of baseline A1Cs. LX4211 reduces A1C in a glucose dependent manner through dual inhibition of SGLT1 and SGLT2, the major sodium-glucose transporters in the gastrointestinal tract and kidney. In a 12-week dose-ranging study, 299 patients with inadequately controlled T2DM on metformin were randomly assigned to receive LX4211 (doses of 75 mg qd, 200 mg qd, 200 mg bid, or 400 mg qd) or placebo. The primary endpoint was change in A1C. Mean demographics were: age 55.9 years, A1C 8.1%, and body mass index 33.1 kg/m2. We report a subgroup analysis of A1C Supported by: Novartis Pharma AG 1133-P Saxagliptin Reduces A1C and Is Well Tolerated in Patients With Type 2 Diabetes Receiving Concomitant Statin Therapy BRIAN BRYZINSKI, ELSIE ALLEN, WILLIAM COOK, BOAZ HIRSHBERG, Wilmington, DE, Princeton, NJ The JUPITER study and other trials have reported an increased risk for development of diabetes with statin treatment. Given these findings, we examined whether glycemic control with saxagliptin (SAXA) 2.5 and 5 mg/d was affected by concomitant statin therapy. Pooled data from 9 placebocontrolled phase 3 studies with a primary 24-week treatment period were analyzed (4 monotherapy, 2 add-on to metformin, and 1 each of add-on to sulfonylurea, thiazolidinedione, or insulin). Safety was assessed in 11 studies and in an extended pool of 20 phase 2/3 studies. Greater mean reductions in glycated hemoglobin (A1C) with SAXA, compared with control, were observed for patients with any statin use and no statin use (Table). No treatment group-by-baseline statin use interaction was observed (interaction P=0.47). The proportions of patients with adverse events (AEs) in the treatment groups were comparable with or without statin use. Serious AEs, deaths, and symptomatic confirmed hypoglycemia (fingerstick glucose ≤50 mg/dL) were few and similar in patients with any statin use compared with no statin use. Similar safety findings were observed in the extended safety pool. SAXA improves glycemic control and is generally well tolerated in patients with type 2 diabetes and is not affected by concomitant statin therapy. reductions in patients stratified by baseline median A1C: >7.8% or ≤7.8%. LX4211 led to robust reductions in A1C with greater reductions achieved in patients with higher baseline A1C levels in all active treatment groups. 1132-P Efficacy and Safety of Vildagliptin in Triple Combination With Metformin Plus Sulfonylurea in Subgroup of Patients With T2DM and Baseline HbA1c ≤8% VALENTINA LUKASHEVICH, MAKO ARAGA, WOLFGANG KOTHNY, East Hanover, NJ The broadly used combination of metformin (Met) and sulfonylurea (SU) often fails to bring patients to glycemic goal and the addition of a third oral antidiabetic drug (OAD) or insulin is required. Our previously reported doubleblind placebo-controlled 24-week study in T2DM patients inadequately controlled with Met plus SU (HbA1c ≥7.5% and <11%) demonstrated that the addition of vildagliptin (Vilda) 50 mg bid reduced HbA1c by 1.0% from a baseline of 8.8%. We now assessed which patient population could actually achieve glycemic control with this triple combination. We present the efficacy and safety of Vilda in a subgroup of patients with baseline HbA1c ≤8% (N=84). In this subgroup with background demographic characteristics similar to the overall study population, Vilda demonstrated a reduction in HbA1c of 0.6% compared to an increase of 0.2% with placebo (Figure). Significantly more Vilda patients (38.6%) achieved an HbA1c target of <7.0% vs. placebo (13.9%) (p=0.014). Overall, Vilda demonstrated a good safety profile with low and comparable to placebo incidence of hypoglycemia. Vilda could be successfully used as a third OAD in patients with T2DM and HbA1c ≤8% failing on Met plus SU, as a significant part of this patient population could achieve HbA1c treatment target before initiating an insulin therapy. Supported by: Bristol-Myers Squibb/AstraZeneca & For author disclosure information, see page 829. A294 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS confirmed hypoglycemia (fingerstick glucose ≤50 mg/dL, ≤1%) were few and similar across treatment and CV risk groups. Two patients in the high CV risk PBO group died. SAXA is effective, with overall AEs similar to PBO and no increased risk of hypoglycemia in high CV risk patients. 1134-P Saxagliptin (SAXA) as Add-On to Metformin (MET) + Sulfonylurea (SU): Outcomes Stratified by Baseline A1C and Patient Characteristics ROBERT G. MOSES, SANJAY KALRA, JOHN MONYAK, HELEN YEH, Wollongong, Australia, Karnal, India, Wilmington, DE Supported by: Bristol-Myers Squibb/AstraZeneca 1136-P Association Between PPAR-γ Gene Polymorphism and Response to Pioglitazone Therapy in Bangladeshi T2DM Subjects MASUMA PARVIN, AK AZAD CHOWDHURY, TAPASH ROY, BEGUM ROKEYA, Dhaka, Bangladesh, London, United Kingdom Present study was undertaken to assess association between PPAR-γ gene variants and response to pioglitazone with T2DM subjects in Bangladesh. Evaluating case history 77 patients were randomized to receive pioglitazone (30 mg once daily) for three months. DNAs were extracted from whole blood of patients, amplified with exon B using primers forward: 5-ACT CTG GGA GAT TCT CCT ATT GGC and primers reverse 5-CTG GAA GAC AAA CTA CAA GAG introducing HaeIII recognition site for digestion and size polymorphism to distinguish alleles. 61 patients [HbA1c 7.2±0.7%, M ±SD] completed study. Outcome measures included BMI, HbA1c, fasting (FBG) and 2 hours blood glucose (2hBG), insulin levels, total cholesterol (TC), high (HDL-C) and low density lipoprotein cholesterol (LDL-C),triglyceroides, SGPT and creatinine for each month. Response to therapy was defined as either ≥10% decrease in Fasting Blood Glucose (FBG) or ≥1% decrease in HbA1c after 3 months treatment. Frequencies and response rate were 29.5% and 83.33% in Pro12Ala, 70.5% and 76.74% in Pro12Pro variants respectively. No Ala12Ala was found. In Pro12Pro HbA1c was reduced from 7.2 to 6.8 (p <0.005), Fasting Serum Insulin [µU/ml] from 15.9±8.0 to 11.0±5.6 (p <0.006). Enhancement in HOMA%S from 48.0 to 73.0 (p <0.000), QUICKI from 0.51 to 0.58 (p <0.001), decrease of HOMA IR from 5.3 to 2.7 (p <0.001) were observed. LDL-C (mg/ dl) reduction (p <0.046) and BMI [kg/m2] enhancement from 27.1 to 27.5 (p <0.002) were in Pro12Pro. In Pro12Ala reduction of HbA1c from 7.4 to 6.5 (p <0.036), increase of HOMA%B from 146.8 to 169.0 (p <0.515) and other nonsignificant positive changes were noticed. In conclusion, before treatment no difference in baseline bio-markers between two groups was observed but after treatment Pro12Ala differed significantly from Pro12Pro and was better in FBG (p <0.039), HOMA% B (p <0.006), BMI (p <0.094) and FSI (p <0.093) with lower risk. Pro12Ala achieved lower FBG and more active β cell function compared to Pro12Pro variants. Supported by: Bristol-Myers Squibb/AstraZeneca 1135-P Saxagliptin (SAXA) Reduces A1C and Is Well Tolerated in Patients With Type 2 Diabetes (T2DM) and High Framingham 10-Year Cardiovascular (CV) Risk ENZO BONORA, ELSIE ALLEN, BRIAN BRYZINSKI, BOAZ HIRSHBERG, WILLIAM COOK, Verona, Italy, Princeton, NJ, Wilmington, DE Recent trial results in patients with T2DM suggest that glycemic targets should be personalized and the burden of hypoglycemia and comorbid conditions, such as CV disease, carefully considered. A post hoc analysis was performed of data pooled from 5 phase 3, placebo (PBO)-controlled, 24-week studies of SAXA (2 of SAXA 5 mg/d monotherapy and 1 each of SAXA 5 mg/d add-on to metformin, glyburide, or thiazolidinedione), stratified by Framingham 10-year CV risk score (<20% vs ≥20%). Patients (52% women) had a mean age of 55 (range, 18-77) years and a mean body mass index of 30 (range, 17-45) kg/m2. Week 24 improvements in glycated hemoglobin (A1C), fasting plasma glucose, and 2-hour postprandial glucose were greater with SAXA vs PBO (Table). Differences between CV risk groups were investigated, with no important differences between groups based on Framingham risk. More patients achieved A1C <7% with SAXA vs PBO in both CV risk groups. In the high CV risk group, the proportion of patients with adverse events (AEs) was similar with SAXA (74%) and PBO (73%). Serious AEs (2%-5%) and reported hypoglycemia (7%-8%) and symptomatic ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A295 POSTERS Clinical Diabetes/ Therapeutics In a placebo (PBO)-controlled, 24-week phase 3b trial in patients with T2DM inadequately controlled with MET+SU (NCT01128153), adding SAXA 5 mg/d to MET+SU significantly reduced A1C vs PBO. A post hoc analysis was performed with data stratified by patient age (<65 y, ≥65 y), race (white vs Asian), baseline A1C (<8%, 8%-<9%, ≥9%), and BMI (<30 kg/m2, ≥30 kg/m2). Across categories of age (interaction P value=0.40), race (P=0.36), baseline A1C (P=0.12), and BMI (P=0.99), A1C was reduced more with SAXA vs PBO (Tables 1, 2). Adverse events were comparable across treatment groups and categories and were reported by 58%-85% of patients. Symptomatic confirmed hypoglycemia (fingerstick glucose ≤50 mg/dL) was reported by 2 Asian patients receiving SAXA, with baseline A1C <8% and BMI <30 kg/m2. When added to MET+SU, SAXA improves A1C across categories of age, race, baseline A1C, and BMI and is generally well tolerated. CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS -53.7 ± 8.6 mg/dL (p<0.0001). In summary, this pooled analysis from a large, global Phase 3 program shows that once-daily linagliptin improves measures of β-cell function relative to placebo. The effect of long-term treatment with linagliptin on β-cell function remains to be determined. 1137-P WITHDRAWN Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1139-P Urinary Tract Infection (UTI) With Canagliflozin (CANA) in Subjects With Type 2 Diabetes Mellitus (T2DM) CANA, an SGLT2 inhibitor, inhibits renal glucose reabsorption, resulting in increased glucosuria. The impact of this on UTI was assessed in subjects with T2DM in placebo (PBO)-controlled, Phase 3 studies of CANA using a UTI case report form to characterize the presentation and severity of events. Findings are reported for pooled analyses of subjects from four 26wk studies (DS1; N = 2,313) and subjects with moderate renal impairment (eGFR ≥30 and <60 mL/min/1.73 m2) from four 18- or 26-wk studies (DS2; N = 1,085), and for the CANagliflozin cardioVascular Assessment Study (CANVAS; 52-wk interim safety analysis) in subjects with a history or high risk of cardiovascular disease (N = 4,327); subjects with a history of UTI were not excluded from these studies. Across datasets analyzed, the proportion of subjects with UTI was slightly higher with CANA than PBO, with no increase in recurrent or serious events or in UTI-related discontinuations; upper UTI rates were low and similar across groups. UTI was more common in women than men, but the increased incidence with CANA was consistent between genders. UTIs generally occurred within 26 wks of starting CANA, with a subsequent attenuation in incidences. Median time to first symptomatic UTI tended to be earlier with CANA, but median duration and severity of symptoms of UTIs were similar with CANA and PBO. In summary, a slight increase in UTI rates, with no increase in serious or upper UTI, was seen with CANA in subjects with T2DM. POSTERS Clinical Diabetes/ Therapeutics LINDSAY E. NICOLLE, GEORGE CAPUANO, ALBERT FUNG, KEITH USISKIN, Winnipeg, MB, Canada, Raritan, NJ 1138-P The Dipeptidyl Peptidase (DPP)-4 Inhibitor Linagliptin Improves β-Cell Function and Postprandial Glucose in Type 2 Diabetes (T2D) SANJAY PATEL, TIM HEISE, MARTIN LARBIG, SONJA WEBER, THOMAS SECK, UWE HEHNKE, HANS J. WOERLE, KLAUS DUGI, Bracknell, United Kingdom, Neuss, Germany, Ingelheim, Germany In patients with T2D, progressive deterioration of β-cell function contributes to the development of hyperglycemia. Treatment with incretinbased therapies may potentially improve β-cell function and delay this progressive deterioration. Here we investigated the effect of the DPP4 inhibitor linagliptin on β-cell function, as assessed by changes in Homeostasis Assessment Model (HOMA)-%β index and incremental 2-hour postprandial glucose (PPG) levels after a meal tolerance test. Data from six randomized, 24-week, placebo-controlled, Phase 3 trials of linagliptin 5 mg qd were pooled for this analysis. In total, 2960 patients were available: linagliptin, n=2077; placebo, n=883. Mean ± SD baseline characteristics for this pooled dataset were as follows: age, 56.5 ± 10.2 years; BMI, 29.2 ± 5.1 kg/m2; HbA1c, 8.2 ± 0.9%; FPG, 167.1 ± 44.9 mg/dL. The % of patients with T2D >5 years was 53.8%. Mean drug exposure was 163 ± 33 days. Subgroup analyses were conducted in patients who had baseline and week-24 data for HOMA-%β (n=1770) and PPG (n=224) assessments. Mean ± SD baseline HOMA-%β and PPG for linagliptin vs placebo were 54.6 ± 101.6 vs 49.4 ± 47.7 (mU/L)/(mmol/L) and 267.3 ± 73.7 vs 253.2 ± 73.6 mg/dL, respectively. After 24 weeks’ treatment, the adjusted mean ± SE change from baseline in HOMA%β with linagliptin vs placebo was 16.7 ± 4.3 vs 0.2 ± 5.1 (mU/L)/(mmol/L); the placebo-adjusted mean change for linagliptin was 16.5 ± 4.6 (mU/L)/ (mmol/L) (p=0.0003). Similarly, after 24 weeks’ treatment, the adjusted mean ± SE change from baseline in PPG with linagliptin vs placebo was -44.6 ± 6.0 vs 9.1 ± 7.9 mg/dL; the placebo-adjusted mean change for linagliptin was 1140-P Potent Anti-Hyperglycemic Effects of Panax Notoginseng Extract Containing Dammarane-Type Triterpenes in Human Subjects MITSURU NOMURA, HIDEAKI IWASAKI, NORIYUKI SUZUKI, AYUMI MURATA, TAKU IWAMOTO, KUMIKO KITAMURA, YUSUKE TAKAMURA, MISAO KOIDE, MICHIAKI MURAKOSHI, HOYOKU NISHINO, Odawara, Japan, Kyoto, Japan The root of Panax notoginseng has been used as an herbal treatment in traditional Chinese medicine for centuries, and a group of saponins is considered as the major active ingredients. Since saponins are converted into dammarane-type triterpenes as aglycones during the digestion process, we produced a dammarane-type triterpene extract (DTE) from the root of Panax notoginseng in which all sugars were fully removed from saponins. Previously we reported that DTE has hyperglycemia inhibitory effect in mouse obese/diabetic models by enhancement of glucose uptake in skeletal muscle (0109-LB, ADA 71st Scientific Sessions, 2011). Panaxatriol, one of the dammarane-type triterpenes in DTE, has been found to improve whole & For author disclosure information, see page 829. A296 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1141-P MIAN WU, JUNXI LU, YAN HUANG, KAIFENG GUO, HAOYONG YU, LEI ZHANG, YUQIAN BAO, HAIBING CHEN, WEIPING JIA, Shanghai, China Aims: At present, conclusive opinions does not exist to support the tenet that vitamin D is associated with type 2 diabetes mellitus (DM) or its supplementation improve glucose metabolism of type 2 DM. In addition, reports about Chinese population were quite limited. So this study aimed to elucidate the relationship between serum 25-hydroxyvitamin D3 (25(OH)D3) level and β-cell function in Chinese subjects, and the effect of vitamin D3 supplementation on patients with type 2 DM. Methods: Two human studies were performed: (1) a cross-sectional study of 78 healthy, glucose-tolerant volunteers and 399 type 2 DM subjects, in whom data were used to assess the relationship between serum 25(OH) D3 level and β-cell function; and (2) a longitudinal study of 22 type 2 DM patients treated with 800 IU cholecalciferol per day for 6 months and a control group of 22 matched type 2 DM patients, baseline and follow-up records were collected to explore the effects of intervention. C-peptide and its derivatives were used to assess β-cell function. Results: In the cross-sectional study, serum 25(OH)D3 was independently correlated with β-cell function in men. However, the correlation between serum 25(OH)D3 and β-cell function in female group was quenched after adjustment for a number of confounding factors. In the longitudinal study, serum fasting C-peptide (FCP) was significantly increased (P = 0.030) in 22 type 2 DM patients after supplementation with cholecalciferol, whereas there was no statistical change in the control group (P = 0.811). Conclusions: There was a sex difference in the relationship between serum 25(OH)D3 and β-cell function in a Chinese population. Treatment with 800 IU cholecalciferol daily for 6 months increased insulin secretion. P value 1143-P DPP4 Inhibition Ameliorates Intramuscular Fat Content in Patients With IGT or Type 2 DM: Comparing Alogliptin and Voglibose MASANOBU TSUCHIYA, HIROSHI MAEGAWA, Yanai, Japan, Otsu, Japan We reported that the DPP4 inhibitor, Alogliptin treatment(ALO) ameliorates both liver fat content andLDL-size, but Voglibose treatment (VOG) does not. To clarify whether ALO maydecrease intramuscular fat content, the 37 patients comprised of 25 IGT and 12 DM were recruited. The 37 patients were randomly assigned to the treatment for 12weeks with either ALO (25mg daily) or VOG (0.6mg daily). A 75-g OGTT,laboratory measurements and computedtomography (CT) to determine intramuscular CT values (Hounsfield unit,HU) were performed at baseline and after the treatment. Earlyinsulin secretion and postprandial hyperglycemia were assessed by delta IRI / deltaPG for 30. Intramuscular fat content was evaluated by Psoas major muscle CT values and Paraspinal muscle CT values in terms of the mean values for two different sites respectively. 12 weeks after the treatment, ALO significantly decreased plasma glucose level at 0min (P<0.05), 30 min (P<0.05), 60 min (P<0.01) and 120min (p<0.001) after the glucose load and also showed higher IRI/ PG(0.65±0.37, 1.27±1.31, P<0.05). VOG significantly decreased plasma glucose level at 120m (P<0.001). ALO showed higher Psoas major muscle CT values (48.6± 5.7, 51.1 ± 4.8 HU, P<0.01), Paraspinal muscle CT values (53.2 ± 5.4, 54.5 ± 5.3 HU, P<0.01), HDL-c (P<0.05) and had lower triglyceride (P<0.01), as compared with those of the baseline but VOG does not. In this treatment, before and after treatment, there were no differences in BMI, HOMA-R, adiponectin, abdominal visceral fat area, subcutaneous fat area in both groups. In conclusion, DPP4 inhibition may decrease intramuscular fat content in patients with IGT or type 2 DM. 1142-P Comparative Effects of Aleglitazar- a PPAR-α/γ Agonist- versus Pioglitazone on Glycemia, Insulin Sensitivity and Lipids in Patients With Type 2 Diabetes and Stage 3 Chronic Kidney Disease MARKOLF HANEFELD, GIANCARLO VIBERTI, KLAS MALMBERG, NADEJDA MUDIE, TERESA URBANOWSKA, SYLVIE MEYER REIGNER, MATTHIAS HERZ, Dresden, Germany, London, United Kingdom, Stockholm, Sweden, Basel, Switzerland Aleglitazar (ALE) is a balanced PPAR-α/γ agonist in Phase III for cardiovascular risk reduction following acute coronary syndrome in patients with type 2 diabetes (T2D). As well as evaluating renal effects of ALE in stage 3 chronic kidney disease in patients with T2D, this Phase IIb study (AleNephro) assessed effects on glycemia, insulin sensitivity and lipids vs the established PPAR-γ agonist, pioglitazone (PIO). Patients were randomized to 52 weeks’ double-blind treatment with ALE 150 µg/d or PIO 45 mg/d in addition to pre-existing diabetes therapy (5%, drug-naïve; 44%, monotherapy; 51%, combination therapy), and 8 weeks’ follow-up. Endpoints included change from baseline in HbA1c, homeostasis model assessment of insulin resistance (HOMA-IR), C-peptide and lipids at 52 weeks. Similar HbA1c reductions were seen with ALE vs PIO (Table), with greater reductions (> 1%) in patients with HbA1c > 7.5% at baseline. In both treatment groups, reductions in HbA1c were similar in patients taking/not taking metformin at baseline but greater in those on sulfonylurea at baseline vs those without. Similar reductions in HOMA-IR and C-peptide and greater improvements in lipids were seen with ALE vs PIO (Table). In AleNephro, while ALE 150 µg/d had comparable effects to PIO 45 mg/d on diabetes control and insulin sensitivity, a significantly better effect was seen on atherogenic dyslipidemia. & PIO 45 mg (n = 152) HbA1c, % All patients Baseline 7.5 ± 0.9 7.6 ± 0.9 0.41 Absolute change at EOT (95% CI) -0.67 (-0.87, -0.48) -0.76 (-0.96, -0.56) Patients with baseline HbA1c > 7.5† Baseline 8.5 ± 0.8 8.4 ± 0.8 Absolute change at EOT (95% CI) -1.02 (-1.38, -0.66) -1.23 (-1.59, -0.87) 0.24 HOMA-IR (calculated), µU/mL*mg/dL‡ Baseline 3.15 (1.99-5.82) 3.37 (2.16-4.98) 0.26 % change at EOT (95% CI) -47.1 (-54.3, -38.7) -42.3 (-50.1, -33.2) C-peptide, ng/mL Baseline 3.33 ± 1.83 3.15 ± 1.67 0.22 Absolute change at EOT (95% CI) -0.85 (-1.13 -0.58) -0.68 (-0.95, -0.40) Triglycerides, mg/dL§ Baseline 204.1 ± 143.3 192.1 ± 105.2 < 0.001 % change at EOT (95% CI) -33.6 (-41.1, -26.1) -14.1 (-21.7, -6.5) HDL-C, mg/dL§ Baseline 46.4 ± 12.7 47.3 ± 12.5 < 0.001 % change at EOT (95% CI) 22.0 (17.4, 26.6) 11.6 (6.9, 16.3) LDL-C, mg/dL‡§ Baseline 104.0 (82-130) 112.0 (90-135) 0.04 % change at EOT (95% CI) -7.3 (-13.2, -1.0) -0.3 (-6.8, 6.6) Data from safety analysis set. ALE = aleglitazar, PIO = pioglitazone, EOT = end of treatment. * Mean ± SD values provided apart from median (interquartile range) for HOMA-IR and LDL-C at baseline. † At baseline: ALE, n = 58; PIO, n = 70. ‡ Analysis on log transformed scale. § Data from full analysis set: ALE, n = 149; PIO, n = 148. Oral Supplementation With Cholecalciferol 800 IU Improved β-Cell Function in Chinese Type 2 Diabetic Patients ADA-Funded Research ALE 150 µg (n = 149) 1144-P Empagliflozin Monotherapy for 12 Weeks Improves Glycemic Control in Japanese Patients With Type 2 Diabetes (T2DM) TAKASHI KADOWAKI, MASAKAZU HANEDA, NOBUYA INAGAKI, ATSUSHI TANIGUCHI, MASASHI SAKAMOTO, KAZUKI KOIWAI, HENNING RATTUNDE, HANS J. WOERLE, ULI C. BROEDL, Tokyo, Japan, Asahikawa, Japan, Kyoto, Japan, Ingelheim, Germany A Phase II trial investigated the efficacy and safety of empagliflozin (EMPA) monotherapy in Japanese patients with T2DM. Patients (mean age 57.5 [SD 9.9] years; mean BMI 25.5 [SD 3.9] kg/m2) were randomized doubleblind and treated with EMPA 5 mg (n=110), 10 mg (n=109), 25 mg (n=109) or 50 mg (n=110) qd or placebo (PBO; n=109) for 12 weeks. The primary endpoint was change from baseline in HbA1c at week 12. Secondary endpoints were the proportion of patients with baseline HbA1c ≥7% who reached HbA1c <7% and change from baseline in FPG at week 12. For author disclosure information, see page 829. Guided Audio Tour poster A297 POSTERS Parameter* Clinical Diabetes/ Therapeutics body insulin resistance by using the hyperinsulinemic-euglycemic clamp in mice and enhance phosphorylation of Akt in skeletal muscle (1933-P, ADA 72nd Scientific Sessions, 2012). The aim of this study was to evaluate the anti-hyperglycemic effects of DTE in human subjects. Using a double-blind, placebo-controlled design, Japanese men and women (n=53; aged 39-69 years) with slightly higher blood glucose levels (FBG: 79.0-135.0 mg/dL, HbA1c: 5.0-6.5 %, GA: 11.6-18.1 %), consumed 180 mg/d DTE or placebo tablets for 16 weeks. The time for ingestion of test tablets was not limited, but it was recommended that the subjects take test tablets after breakfast to maintain compliance. The blood-sugar monitoring revealed a significant reduction in FBG (-2.73±6.17 mg/dL) in the DTE group, as compared with that in the placebo controls (+1.07±6.66 mg/dL; P=0.036). Throughout this study, no adverse effects were observed. For confirmation of safety, 16 healthy subjects were given DTE tablets for 16 weeks. And none indicated changes in hematological and relevant biochemical parameters were apparent. From these results, DTE appears to be a promising agent for the control of blood-sugar level. CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1146-P Efficacy of Concomitant Therapy in Japanese Type 2 Diabetes Mellitus Patients With Insulin and Sitagliptin, a DPP-4 Inhibitor TOMOKO NAKAGAMI, RISA IDE, NAOKO IWASAKI, MAKIKO OGATA, JUNKO OYA, MARI OSAWA, NOBUE TANAKA, HIROKO TAKAIKE, ASAKO SATO, JUNNOSUKE MIURA, YASUKO UCHIGATA, Tokyo, Japan This open-label observational study has examined the efficacy of the additional administration of sitagliptin with respect to lowering HbA1c and body weight (BW) in type 2 diabetes (T2DM) patients insufficiently glycemic controlled by insulin. The study comprises 64 T2DM patients with mean age 59±15 years old, duration of DM 20±7 years, and BMI 25.8±4.4 kg/m2. Plasma glucose (PG), C-peptide (CPR) and glucagon on meal tolerance test (MTT), HbA1c and BW were compared between just before and 4 months after the administration by paired t-test. Multivariate stepwise logistic regression model with backward form was used to identify characteristics of the patients in whom HbA1c lowered without BW gain by this regimen. Four months after the administration, there were significant decreases in mean HbA1c (8.6±1.2 to 7.8±1.2%) and mean PG at 60 min (244.2±62.0 to 215.0±61.9 mg/dL) and 120 min (231.5±72.2 to 198.3±67.5 mg/dL) on MTT (all p<0.001); significant increases in mean CPR before (1.1±0.8 to 1.2±0.8 ng/ mL) and 60 min (1.7±1.10 to 1.8±1.1 ng/mL) on MTT (all p<0.05); and significant decreases in mean glucagon at 60 min (88.3±27.8 to 81.0±25.1 pg/mL) and 120 min (85.7±30.7 to 79.5±27.4 pg/mL) on MTT (all p<0.05). There was no change in the basal insulin dose after 4 months, but the additional insulin dose showed a significant decrease (0.26±0.20 to 0.25±0.19 units/kg, p=0.005). There was no significant change in mean BW and no severe hypoglycemia. Eventually, HbA1c decreased in 50 patients among 64, but 29 of 50 increased their BW. Logistic regression analysis showed that the patients in whom HbA1c lowered without BW gain independently related to CPR (1.0 ng/mL) at 60 min and 120 min on MTT at the pre-administration phase with odds ratios (95% CI) of 9.8 (1.2-82.3) and 0.1 (0.02-0.6), respectively and did not relate to glucagon level. Additional administration of sitagliptin may be effective in insulin treated T2DM patients who have non-delayed pattern of insulin secretion. POSTERS Clinical Diabetes/ Therapeutics EMPA significantly reduced HbA1c and FPG vs PBO; more patients reached HbA1c <7% with EMPA than with PBO. Further analyses showed significant reductions in weight vs PBO with all doses of EMPA, and significant reductions in systolic blood pressure (SBP) vs PBO with all doses except EMPA 5 mg. Adverse events (AEs) occurred in 32.7%, 37.6%, 36.7%, 38.2% and 42.2% of patients on EMPA 5 mg, 10 mg, 25 mg, 50 mg and PBO, respectively. Hypoglycemia (plasma glucose ≤70 mg/dL and/or requiring assistance) was rare; none required assistance. AEs consistent with urinary tract infection were reported in 1 patient (0.9%) each on EMPA 10, 25, 50 mg and PBO and 0 on EMPA 5 mg. AEs consistent with genital infection were reported in 1 patient (0.9%) each on EMPA 5, 10, and 50 mg and 0 on EMPA 25 mg or PBO. To conclude, EMPA as monotherapy for 12 weeks significantly reduced HbA1c, FPG, weight and SBP versus PBO, and was well tolerated in Japanese patients with T2DM. 1147-P Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. SNPs Related to the Glucose Lowering Effect of Metformin: Do they Affect Plasma Concentrations of Metformin (MPC) and their Effects on Insulin Requirements (DDI)? 1145-P MATTIJS OUT, ADRIAAN KOOY, MATTHIJS L. BECKER, RON H.N. VAN SCHAIK, PHILIPPE LEHERT, COEN D.A. STEHOUWER, Hoogeveen, Netherlands, Rotterdam, Netherlands, Louvain, Belgium, Maastricht, Netherlands WITHDRAWN Several genes involved in the pharmacokinetics and pharmacodynamics of metformin may influence the glucose-lowering effect of metformin. However, their clinical importance is yet unknown. In the population of the HOME trial (a placebo-controlled, randomised trial with patients with type 2 diabetes, treated with insulin) we genotyped 4 single nucleotide polymorphisms (SNPs) of such genes to detect their potential influence on MPC. Furthermore, we studied the potential of these SNPs and the MPC to predict the reduction of DDI, an important action of metformin in insulin therapy. We genotyped 164 metformin users of the HOME trial for polymorphisms in the genes coding for organic cation transporter 1 (OCT1, rs12208357 and rs622342), multidrug and toxin extrusion 1 transporter (MATE1, rs2289669) and Ataxia Telangiectasia Mutated (ATM, rs11212617). The daily dose of metformin (DDM) and DDI have been reported previously (Arch Int Med 2009). MPC was measured using HPLC-UV. Stepwise linear analysis showed that the SNP rs11212617 (ATM) had a significant effect on MPC (p < 0.001), but not on DDI. Other SNPs studied did not affect MPC or DDI (P values ranging between 0.30-0.92 and 0.40-0.72, respectively). Overall, we found a significant relationship between MPC and DDI (most pronounced after adjustment for renal clearance): an increase of MPC with 1mg/ml was associated with a decrease of DDI with 3.51 IU (95% CI: -6.70 to -0.32; p=0.01). DDM predicted the MPC (p<0.001) and was strongly correlated with a reduction of DDI: an increase of DDM with 1 gram/day was associated with a mean decrease of DDI with 10.57 IU (95% CI: -17.20 to -3.94; p<0.001). Of the SNPs studied, only the ATM SNP has an effect on MPC, but its clinical relevance is unclear, since this SNP has no effect on DDI. DDM is a strong predictor of the effect of metformin on DDI, independent of all SNPs studied. Supported by: Altana AG; Lifescan, Inc.; E. Merck/Santé; Merck Sharpe & Dohme; Novo Nordisk, Inc. & For author disclosure information, see page 829. A298 Guided Audio Tour poster ADA-Funded Research 1150-P Efficacy and Safety of Alogliptin in Subjects With Type 2 Diabetes: A Multicenter, Randomized, Double-Blind, Placebo-Controlled, Phase 3 Study in Mainland China, Taiwan and Hong Kong ELIZABETH ROUND, YUE SHENTU, GREGORY T. GOLM, EDWARD A. O’NEILL, IRA GANTZ, SAMUEL S. ENGEL, KEITH D. KAUFMAN, BARRY J. GOLDSTEIN, Whitehouse Station, NJ CHANGYU PAN, PING HAN, QIUHE JI, CHENGJIANG LI, JUMING LU, JINKUI YANG, WENHUI LI, JIAOE ZENG, JULIANNA CHAN, AN-TSZ HSIEH, Beijing, China, Shenyang, China, Xi’an, China, Hangzhou, China, Wuhan, China, Sha Tin, Hong Kong, Taipei, Taiwan Patients being treated with a sulfonylurea (SU) in combination with metformin (MET) for at least 10 weeks were randomized in a 1:1 ratio to sitagliptin (SITA) 100 mg/day or placebo (PBO) for 24 weeks in a multicenter, double-blind study. Of 427 patients randomized, 388 (90.9%) completed the study. At Week 24, the least squares mean (95% confidence interval [CI]) changes from baseline in HbA1c (%, SITA, -0.84 [-0.97, -0.71] vs. PBO, -0.16 [-0.28, -0.03]), 2-hour post-meal glucose (mg/dL, SITA, -36.8 [-45.2, -28.4] vs. PBO, -3.3 [-11.7, 5.0]) and fasting plasma glucose (mg/dL, SITA, -13.2 [-18.8, -7.7] vs. PBO, 5.3 [-0.2, 10.9]) were greater with SITA vs. PBO (p<0.001). The incidence of drug-related adverse events was numerically greater with SITA (n=13, 6.2%) vs. PBO (n=5, 2.4%). The between-group difference in drugrelated adverse events (95% CI) was 3.8% (-0.0, 8.2). This difference was driven mostly by a higher incidence of drug-related hypoglycemia (HYPO) with SITA (n=12, 5.7%) vs. PBO (n=4, 1.9%); these findings are consistent with previous observations, where a drug not known to cause HYPO increases the incidence of hypoglycemic events when added to SU. The overall incidence of symptomatic HYPO with SITA was n=27 (12.9%) and with PBO was n=7 (3.3%). The between-group difference in the overall incidence of symptomatic HYPO (95% CI) was 9.6% (4.6, 15.1), p<0.001. The two treatment groups were generally similar in other aspects of their safety profiles. Changes in body weight were small in both groups at Week 24; the between-group difference (95% CI) of -0.2 (-0.7, 0.3) kg was not significant (p=0.521). In this study, SITA 100 mg/day was generally well tolerated and provided good glycemic efficacy when added to the combination of SU and MET in patients with type 2 diabetes. This multicenter clinical trial in Mainland China, Taiwan and Hong Kong was done to determine the efficacy and safety of once-daily oral Alogliptin (AL) as monotherapy, add-on to ongoing metformin (M) therapy, and add-on to pioglitazone (P) therapy (with or without M) in patients with type 2 diabetes. A total of 506 subjects (185 in the monotherapy group, 197 in the add-on to M therapy group, and 124 in the add-on to P group) were randomized to receive AL 25mg QD or placebo (PBO) QD for 16 weeks. The primary endpoint was change from baseline (BL) in HbA1c at week 16. The secondary endpoints were change in fasting plasma glucose (FPG), incidence of marked hyperglycemia, clinical HbA1c response, and change in body weight from BL. For monotherapy, the decrease of HbA1C from BL to week 16 in the AL and PBO arms were 0.99% and 0.42%, respectively (P<0.001). For the add-on to M group, AL decreased HbA1c by 0.91% compared to PBO 0.22% (P<0.001). For add-on to P group, AL decreased HbA1c by 0.76% vs PBO 0.25% (P<0.001). The decreases in FPG were greater in the AL arms than in the PBO arms. The percentage of subjects able to achieve an HbA1c target of ≤ 6.5%, ≤7.0% and ≤7.5% were all significantly higher (P<0.001) in the AL arms than in the PBO arms. AL treatment showed lower incidence of hyperglycemia than PBO in all groups. No weight gain was observed in either the AL or PBO arms, with weight loss occurring in both arms in all groups. Pooled analysis showed that the percentage of subjects who experienced hypoglycemia was 1.6% in the AL arm vs 0.8% in PBO arm. All episodes of hypoglycemia were considered mild or moderate. No significant changes were observed between the AL and PBO arms in the laboratory parameters compared with BL. AL 25mg QD significantly reduced HbA1c and FPG, and enhanced clinical HbA1c response compared to PBO when used as monotherapy, add-on to M, and add-on to P. AL 25mg QD showed a safety profile similar to PBO. Supported by: Merck, Sharp & Dohme 1149-P Early Improvement in Control of Eating Is Associated With LongTerm Weight Loss—Integrated Analysis of Four Phase 3 Trials of Combination Naltrexone/Bupropion Treatment Supported by: Takeda Global Research & Development Center (Asia) Pte. Ltd. KEN FUJIOKA, BRANDON WALSH, COLLEEN BURNS, PRESTON KLASSEN, La Jolla, CA Early Onset of Increased Hypoglycemic Incidence With Glipizide (GLIP) vs. Saxagliptin (SAXA) in Type 2 Diabetes Patients on Metformin 1151-P Combination treatment with naltrexone sustained-release (SR) / bupropion SR (NB) results in substantial and sustained weight loss that is postulated to be partially mediated by effects on the mesolimbic dopaminergic reward system that modulate control of eating behavior. In the Phase 3 trials of NB in overweight/obese individuals, eating behavior was evaluated with the Control of Eating Questionnaire (CoEQ, twenty 100mm visual analog scales) at baseline and Weeks 8, 16, 28, and 56. This integrated analysis of all four Phase 3, 56-week clinical trials of NB in overweight/obese patients (BMI ≥27 and ≤45 kg/m2) evaluated the change in CoEQ question 19 (CoEQ 19: “Generally how difficult has it been to control your eating?”) and the relationship between changes in CoEQ 19 and body weight. Treatment group differences (LS mean±SE) in the modified ITT-LOCF population (≥1 postbaseline weight on study drug) were evaluated by ANCOVA with treatment, study, and baseline values as covariates. Baseline characteristics (mean±SD) were similar between treatment arms (NB [n=2043] or placebo [PBO; n=1319]): 81% female, 79% Caucasian, 46±11 y, 36.3±4.3 kg/m2. Completion rate was 66% for NB and 59% for PBO. At Week 56, weight loss was significantly greater with NB (-7.0±0.2%) vs PBO (-2.3±0.2%; p<0.001). NB-treated subjects reported significantly greater improvement in CoEQ 19 at all time points (p<0.001 vs PBO), with the greatest effect observed at Week 8 (-23.3 vs -13.5 mm; p<0.001). In both groups, reduction in CoEQ 19 at Week 8 was positively correlated with body weight reduction at Week 56 (r=0.20 (NB), r=0.22 (PBO); both p<0.001), such that the quartile of NB subjects with the greatest CoEQ 19 improvement at Week 8 (> 43 mm) exhibited -9.4% mean weight change at Week 56. These results suggest that NB is associated with rapid improvement in control of eating behavior that may, in part, contribute to the weight loss associated with treatment. ADA-Funded Research & ROBERT FREDERICH, SHAMIK J. PARIKH, ELSIE ALLEN, BRIAN BRYZINSKI, WILLIAM COOK, BOAZ HIRSHBERG, Princeton, NJ, Wilmington, DE A goal of diabetes therapy is to achieve glycemic control without hypoglycemia. In a double-blind, 52-week (with 52-week extension) trial (NCT00575588), patients with A1C 6.5%-10.0% on metformin (MET ≥1500 mg/d) were randomized 1:1 to SAXA 5 mg/d or GLIP 5-20 mg/d (titrated to optimal effect or highest tolerable dose in 18 weeks). Hypoglycemic episodes were recorded in patient diaries per protocol. 1. Despite titration, a much higher fraction of those randomized to GLIP experienced hypoglycemic events and had a higher number of total events vs those randomized to SAXA. 2. Severity of events judged by discontinuation and fingerstick glucose was greater in those randomized to GLIP. 3. For those randomized to GLIP, 38.4% had ≥1 hypoglycemic event. Of these, 65% had their first event within the first 10 weeks. Only 15 (3.5%) of those randomized to SAXA experienced ≥1 event, about half (n=7) with an event in the first 10 weeks. For author disclosure information, see page 829. Guided Audio Tour poster A299 POSTERS 1148-P Safety and Efficacy of Sitagliptin Added to the Combination of Sulfonylurea and Metformin in Patients With Type 2 Diabetes Mellitus and Inadequate Glycemic Control Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1153-P Withdrawing GLP-1 Analog, Liraglutide on Drug-Naive Type 2 Diabetes Patients—Does it Induce Glycemic Remission? HITOMI FUJII, MITSUTOSHI KATO, NORIKO KATO, KAZUO KANNO, HIROKO KONDO, HIROSHI TAKAMURA, DAISUKE MATSUDA, MASAYUKI SHIGETA, YUKO WATANABE, TAKAICHI MIYAKAWA, AKIO UEKI, TAKAFUMI KITAOKA, Tama, Japan, Katsushika-ku, Japan, Mitaka, Japan, Kodaira, Japan, Fussa, Japan, Higashiyamato, Japan Objectives: The GLP-1 analog Liraglutide has been prescribed for two years in Japan. To evaluate the efficacy and safety of Liraglutide, we conducted a multi-centered, observational study involving eight diabetes clinics in Tokyo and approved by the IRB of West Tokyo Clinical Diabetes Study Group. Methods: Type 2 diabetes patients on diet and exercise, not taking oral hypoglycemic agents, were recruited. They injected with 0.9mg per day (the maximum dose covered by Japanese national insurance) of Liraglutide with a target to reduce HbA1c levels to less than 7.0 %. Diabetologists decided to stop injection if their HbA1c level reduced to less than 6.2%. Results: 45 registered. 32 were male. Average age: 53.8±12.6; Duration of illness: 3.7±3.2 years; BMI: 25.8±5.81; HbA1c at start: 11.2±1.90 %. 22 (47.8%) and 20(43.5% of the cases) subjects achieved the target HbA1c levels at month 6 and 12 respectively. During 12 months, 18 were stopped injection because of achieving HbA1c less than 6% twice consecutive times (Group1: withdrawal). Nine stopped visiting (Group2: stopped coming) Eight cases were changed the treatment because of poor response to Liraglutide (Group 3: changed treatment). Residual 11 continued Liraglutide injection (Group 4: continuing). Among group1, after stopping Liraglutide, 3 out of 18 started Sulfonylureas, DPP-4 Inhibitors, and biguanides within 6 months. In group 3, 3 started insulin, 2 started exenatide, and others stated oral agents. Comparing basic characteristics of groups, group1 tended to have shorter duration of illness, larger waist circumference and higher insulin secretion. Conclusion: In conclusion, 0.9mg of Liraglutide per day was effective for patients who were drug-naïve and had relatively short duration of illness. More than 40% of the patients were able to achieve and sustained HbA1c level less than 7%. 1152-P POSTERS Clinical Diabetes/ Therapeutics Supported by: Bristol-Myers Squibb/AstraZeneca Real-World Improvements in A1C With Colesevelam in an Electronic Medical Records Dataset RICHARD A. HANSEN, JOEL F. FARLEY, MATTHEW L. MACIEJEWSKI, XIN YE, CHUNLIN QIAN, BENJAMIN J. POWERS, Auburn, AL, Chapel Hill, NC, Durham, NC, Parsippany, NJ, Boise, ID Colesevelam HCI, added to other anti-diabetic therapy, reduced A1C by ~0.5% compared to placebo in randomized clinical trials (RCTs). Evidence on real-world effectiveness is unknown. We used GE Centricity electronic medical records data from 2000-2011 to evaluate A1C changes in realworld clinical practice among patients with type 2 diabetes mellitus (T2DM) initiating colesevelam. A retrospective cohort of patients with T2DM was examined to evaluate changes in A1C associated with colesevelam initiation. Patients were included if they were 18 years or older, diagnosed with T2DM, and had database activity 395 days before and after colesevelam initiation. The sample was further restricted to patients with a baseline A1C > 7% within 90 days prior to starting colesevelam and at least one A1C result between 42 to 210 days after initiation. Three time intervals were created for A1C measurement, including 16-weeks (days 42-140), 26-weeks (days 42-210), and 52-weeks (days 42-395) following therapy initiation. The last observed A1C lab measurement during each interval was used to define change from baseline. Mean change in A1C was examined using paired t-tests and compared with RCT results. Of 1,709,393 patients in the GE database with T2DM, 1,769 met inclusion criteria. The cohort was 58% female, 38% age 65 and older, and 54% white. For the 16-week endpoint (N=1,405), A1C dropped from a mean of 8.22% to 7.75% (mean change -0.46%; P<0.001). For the 26and 52-week endpoints (N=1,769), A1C dropped from a mean of 8.24% to 7.81% (mean change -0.44%; P<0.001) and 8.24% to 7.78% (mean change -0.47%; P<0.001), respectively. The 0.44% to 0.47% A1C reduction observed in this study was similar to the reduction observed in RCTs, supporting the real-world effectiveness of colesevelam in reducing A1C. Planned subsequent analyses will compare colesevelam effectiveness with alternative treatment strategies and adjust for other covariates to fully consider aspects of treatment in real-world clinical practice. 1154-P Effect of Metformin on Weight Gain and Diastolic Blood Pressure in WITHDRAWN Type 1 Diabetes (T1D) ESTHER S. O’SULLIVAN, FIONA MACNAIR, RONAN J. CANAVAN, MALACHI J. MCKENNA, DONAL O’SHEA, Dublin, Ireland Metformin is commonly added to insulin therapy in T1D patients that are thought to have insulin resistance, but data to support this is limited. We examined medical records of 61 patients with T1D on Metformin, mean age 50.5 (25.8 to 87.7), 33 females; mean duration of T1D 21.0 years (0.8 to 45.8); total T1D person years 1155.1, total Metformin person years 275.7 (0.02 to 20.4). Total insulin, basal and bolus doses, HbA1c, weight, systolic and diastolic BP(sBP, dBP),total cholesterol (TC), triglycerides (TGs), HDL and LDL results were compared at baseline with the corresponding measurements just before the addition of Metformin, and with the most recent data available. The change in each parameter is expressed as rate of change prior to, and after the commencement of Metformin. The rate of weight gain after the addition of Metformin was -0.1kg per year, compared with 2.7kg per year before Metformin was added (p=0.02). The rate of change of dBP was also significantly better -2.0 vs 0.4mmHg/yr respectively, p=0.045. There were trends toward improvements in TC (p=0.08) and LDL (p=0.07) and reduced total insulin dose (-4.1 IU/yr vs 1.4 IU/yr p=0.2). Overall total and bolus insulin doses only changed significantly between baseline and starting Metformin, there was no overall change in HbA1c (figure 1). Our data although limited to retrospective analysis demonstrates the potential benefits to weight and dBP from the addition of Metformin in T1D. Supported by: Daiichi Sankyo, Inc. & For author disclosure information, see page 829. A300 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1155-P 1157-P Effects of Age, Gender and Obesity on the Single Dose Pharmacokinetics (PK) of Omarigliptin, a Novel Once-Weekly Dipeptidyl Peptidase-4 (DPP-4) Inhibitor WITHDRAWN Supported by: Merck, Sharp & Dohme 1158-P 1156-P A Novel Tripeptide Diapin Effectively Lowers Blood Glucose Levels in Type 2 Diabetes Mice by Increasing Blood Levels of Insulin and GLP-1 Efficacy and Safety of the Dipeptidyl Peptidase-4 Inhibitor, Sitagliptin Compared With Alpha-Glucosidase Inhibitor in Japanese Patients With Type 2 Diabetes Treated With Sulfonylurea Alone (SUCCESS2): A Multicenter, Randomized, Open-Label, Clinical Trial JIFENG ZHANG, ZHONGMIN A. MA, YUQING EUGENE CHEN, Ann Arbor, MI The prevalence of type 2 diabetes (T2D) is rapidly increasing worldwide and it is urgent to develop innovative therapeutic approaches for preventing such condition since several effective therapies, such as insulin and GLP1 require injections, which are costly, inconvenient, and less patient compliance. Many studies reported that co-injection of protein hydrolysates and amino acids can substantially increase plasma insulin responses and reduce the prevalence of hyperglycemia in T2D patients. Here, we report an identification of a novel tripeptide with significantly potential to treat T2D. The peptide, referred to as Diapin, is comprised of three natural L-amino acids. Glucose tolerance tests showed that orally administration of Diapin effectively inhibits blood glucose rise after oral glucose loading in both T2D mouse models, including KKay, db/db, and ob/ob mice and in normal C57BL/6J mice. In addition, Diapin reduces random blood glucose in KKay diabetic mice in a time- and dose-dependent manner. The amino acid composition of Diapin or the dipeptides with same amino acid composition, however, has no significant effect on blood glucose of 57BL/6J mice. Furthermore, we found that serum GLP-1 and insulin levels in diabetic models with Diapin treatment are significantly increased compared to that in the controls. A pharmacokinetic study using 13C-labeled Diapin has shown that Diapin exists in the blood stream at low concentrations. In summary, our finding has established a novel concept that a peptide with minimum of three amino acids has potential to improve glucose homeostasis and Diapin shows promise as a novel pharmaceutical anti-diabetic agent to treat patients with T2D, possible in the mechanism of its dual effects on both GLP-1 and insulin secretion. KAZUKI KOBAYASHI, HIDETAKA YOKOH, YASUNORI SATO, MINORU TAKEMOTO, DAIGAKU UCHIDA, AZUMA KANATSUKA, NOBUICHI KURIBAYASHI, KENICHI SAKURAI, TAKASHI TERANO, NAOTAKE HASHIMOTO, HIDEKI HANAOKA, KO ISHIKAWA, SHUNICHIRO ONISHI, KOUTARO YOKOTE, THE SUCCESS STUDY GROUP, Chiba, Japan, Kisarazu, Japan, Funabashi, Japan, Yachiyo, Japan Dipeptodyl peptidase-4 inhibitors (DPP-4i) have been used widely, but its most appropriate position in medication of type 2 diabetes has not been well established. We compared the efficacy and safety of sitagliptin (Sita), a DPP-4i and α-glucosidase inhibitor (αGI) in Japanese patients with type 2 diabetes (T2DM) in the setting of combination therapy with sulfonylurea (SU). A multicenter, prospective, randomized, open-label trial was conducted at 37 hospitals and clinics in Chiba prefecture in Japan from January 2011 to June 2012. The one hundred and twenty participants were 20-79 years old patients with T2DM who had inadequate glycemic control (glycated haemoglobin A1c [HbA1c] 6.9-8.8%) with 2 mg/day or less glimepiride (Gri) alone. They were randomly assigned to two groups, and either Sita (50 mg daily) or αGI (0.6 mg voglibose or 150 mg miglitol daily) was added onto Gri. The primary endpoint was the difference of change in HbA1c (ΔHbA1c) between the two groups at 12 weeks (w). Mean values of HbA1c at baseline was 7.6-7.7% in both group. At 12w, HbA1c was reduced by -0.72% in Sita and -0.28% in αGI group. The difference of ΔHbA1c Sita compared to αGI group was -0.49% (95% CI -0.66 to -0.32, p<0.0001) indicating the superiority of efficacy of Sita. At 24w, ΔHbA1c was similar between the two groups (-0.49%; 95% CI -0.34 to 0.16, p=0.47). Gastrointestinal symptoms were more common with αGI than with Sita (20 patients [37%] and 5 [8.6%], respectively, p=0.0003). Minor hypoglycemia occurred in about 4-5% of patients in each group. Sita was superior to αGI for reduction of HbA1c at 12w, and not inferior at 24w after administration on Japanese patients with T2DM treated with a small dose of Gri alone. It was also well tolerated with minimum risk of gastrointestinal symptoms and hypoglycemia. Supported by: Waksman Financial Group ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A301 POSTERS Omarigliptin is being developed as a potent, once-weekly, oral DPP4 inhibitor for the treatment of type 2 diabetes mellitus. The aim of this study was to evaluate the effects of age, gender and obesity on the PK of omarigliptin. This was a double-blind, randomized, placebo-controlled study. A single 10-mg oral dose of omarigliptin (n=6/panel) or placebo (n=2/ panel) was administered in the fasted state to healthy elderly men and women, healthy young obese [30 ≤ body mass index (BMI) ≤ 35 kg/m2] men and women, and to healthy young women of non-childbearing potential. Plasma was collected at selected post-dose time points for evaluation of omarigliptin concentrations. PK parameters were compared to historical data from a previously conducted single-dose study in young, non-obese male subjects. Omarigliptin was generally well tolerated and there were no serious adverse events or hypoglycemic events. There were no clinically significant differences in omarigliptin AUC0-∞ based on age, gender or BMI (young female vs. young male, pooled elderly vs. pooled young, young obese vs. young non-obese subjects, and pooled female vs. pooled male). Omarigliptin Cmax values were ~40% higher in females vs. males, regardless of age, but were similar in elderly vs. young subjects within each gender. Cmax was ~24% lower for young obese vs. young non-obese females; however, Cmax was similar in young obese vs. young non-obese males. Omarigliptin was generally well tolerated. Demographic factors and BMI do not have a clinically meaningful effect on omarigliptin PK. The data suggest that omarigliptin dose adjustment is not needed on the basis of age, gender or body weight. Clinical Diabetes/ Therapeutics CAROL ADDY, DANIEL TATOSIAN, XIAOLI S. HOU, ISAIAS N. GENDRANO, ASHLEY MARTUCCI, JOHN A. WAGNER, S. AUBREY STOCH, Whitehouse Station, NJ CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1159-P 1161-P Efficacy and Safety of Linagliptin Monotherapy in Asian Patients With Inadequately Controlled Type 2 Diabetes Mellitus: A 24-Week, Randomized, Phase III Clinical Trial Effect of Colesevelam HCl (COL) as an Add-On to Pioglitazone (PIO) on Lipid Particles in Type 2 Diabetes Mellitus (T2DM) MICHAEL R. JONES, DANIEL M. FORD, MERAV BAZ-HECHT, BEN TAO, KENNETH E. TRUITT, HUBERT S. CHOU, Parsippany, NJ, Edison, NJ POSTERS Clinical Diabetes/ Therapeutics YUHONG CHEN, GUANG NING, CHANGJIANG WANG, YAN GONG, HANSJUERGEN WOERLE, WEIQING WANG, Shanghai, China, Hefei, China, Ingelheim, Germany Previous studies evaluated the effects of COL in combination with metformin-, sulfonylurea-, or insulin-based therapy, but it had not previously been extensively studied in combination with thiazolidinediones. A study was recently conducted to evaluate COL as an add-on to PIO-based therapy in adults with T2DM who had inadequate glycemic control (hemoglobin A1C ≥7.5% to ≤9.5%) on a stable dose of PIO (30 or 45 mg/day), with or without 1-2 other oral antidiabetes agents (metformin, a sulfonylurea, or a dipeptidyl peptidase-4 inhibitor). Patients were randomized to receive COL 3.75 g/day (n=280) or placebo (PBO; n=282) in addition to their existing PIO-based therapy for 24 weeks. In addition to the typical glycemic and lipid parameters, changes in lipid particle concentrations and sizes were determined by nuclear magnetic resonance spectroscopy. At Week 24, with last observation carried forward, COL versus PBO produced a significant reduction in total LDL particle concentration (LDL-P; least squares [LS] mean treatment difference -126 nmol/L; P<0.0001); significant (P<0.05) reductions were also seen in subfractions of large (-50 nmol/L), medium small (-14 nmol/L), small (-75 nmol/L), and very small (-61 nmol/L) LDL-P. There were increases in VLDL particle size (LS mean treatment difference +2.0 nm) and associated increases in concentrations of large VLDL and chylomicron particles (+1 nmol/L) and medium VLDL particles (+7 nmol/L; all P<0.001), and a reduction in small VLDL particle concentration (-4 nmol/L; P<0.01). With COL versus PBO there was also increased HDL particle size (LS mean treatment difference +0.06 nm), associated with an increase in large HDL particle concentration (+0.6 µmol/L; both P<0.01). In conclusion, the addition of COL to PIO-based therapy resulted in generally favorable changes to the lipoprotein particle profile compared with PBO in patients with T2DM. This randomized, Phase III, placebo-controlled, double-blind 24-week study evaluated the efficacy and safety of the DPP-4 inhibitor linagliptin as monotherapy in patients with inadequately controlled type 2 diabetes mellitus (T2DM) in Asian countries (NCT01214239). Patients who were treatment naïve or who had been treated with 1 antidiabetes drug were randomized to either linagliptin 5 mg daily or placebo following 4-week washout of prior drugs. The primary endpoint was change from baseline in mean HbA1c after 24 weeks. A total of 300 Asian patients with T2DM (China n=261, Malaysia n=22 and the Philippines n=17) were randomized to linagliptin (n=201) or placebo (n=99) at a 2:1 ratio. Baseline characteristics were well matched between groups, with mean (SD) HbA1c of 7.95 ± 0.89% in the linagliptin group and 8.09 ± 0.91% in the placebo group. Following 24 weeks of treatment, adjusted mean (SE) HbA1c decreased by −0.68 ± 0.07% in the linagliptin group and −0.18 ± 0.10% in the placebo group (placebo-corrected difference, −0.50 ± 0.11; 95% CI: −0.71, −0.28; P<0.0001). In the pre-defined subgroup of patients with baseline HbA1c ≥8.5%, the placebo-corrected decrease in HbA1c was −0.91 ± 0.20% (P<0.0001). Placebo-corrected adjusted mean (SE) change in FPG with linagliptin was −9.6 ± 3.8 mg/dL (95% CI: −17.1, −2.2; P=0.01). Results for the pre-defined Chinese subgroup were similar to the overall Asian population. Linagliptin was well tolerated. Adverse events occurred in 28.0% of linagliptin patients and 28.3% of placebo patients, while drugrelated AEs occurred in 3.0% of linagliptin patients and 2.0% of placebo patients. Investigator-defined hypoglycemic events were reported in 0.5% and 0.0% of linagliptin and placebo patients, respectively. In conclusion, in Asian patients and a Chinese subgroup with inadequately controlled T2DM, linagliptin 5 mg as monotherapy was efficacious and well tolerated over 24 weeks. 1162-P WITHDRAWN Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1160-P Change in HbA1c after 24 Weeks as a Function of Baseline HbA1c: Comparison Between Vildagliptin and Sulfonylureas in Interventional Clinical Trials and Real-Life Conditions GIOVANNI BADER, ANJA SCHWEIZER, JAMES E. FOLEY, Basel, Switzerland, East Hanover, NJ T2DM patients have been shown to achieve lower HbA1c levels with vildagliptin (Vilda) than sulfonylureas (SUs) as add-on to metformin under real-life conditions (RLC) (EDGE observational study) despite similar efficacy in interventional controlled clinical trials (ICCT) and the known superior pharmacological action of SUs. The aim of this post hoc analysis was to study the predictors of HbA1c change after 24 weeks of treatment in ICCT and RLC and the interplay of treatments with predictors of HbA1c change after 24 weeks. We pooled data from 5 clinical trials carried out in 4480 patients [Vilda, N=2788 or SUs, N=1692 (glimepiride, n=1259; gliclazide, n=433)] and compared with a large sample of patients extracted from the EDGE study (Vilda, N=7002 or SUs, N=3702). The impact of baseline (BL) HbA1c, age, gender, and weight on change in HbA1c after 24 weeks were assessed by linear regression model. Overall, 30% of the change in HbA1c was correlated to BL HbA1c [slope -0.52 (95% CI: -0.53, -0.51; p<0.001)], while age, gender, and BL weight were poorly correlated. With Vilda the slope was -0.55 (95% CI: -0.56, -0.53; p<0.001) and no interaction was detected between BL HbA1c and study setting (ICCT or RLC). With SUs the slope was -0.54 (95% CI: -0.56, -0.52; p <0.001). However, there was a stronger association between HbA1c change and BL in the ICCT as compared with RLC as shown by the significant interaction effect [regression coefficient -0.32 (95% CI: -0.37, -0.27; p<0.001)]. This study confirms a strong association between change in HbA1c and BL HbA1c. Moreover, the data suggest that with Vilda the association under RLC is consistent with the ICCT, while with SUs this association appears to be blunted under RLC. Albeit speculative, these data suggest that pharmacologically SUs are very efficacious, but under RLC doses may be limited due to fear of hypoglycemia and weight gain associated with defensive eating to avoid hypoglycemia. Supported by: Novartis Pharma AG & For author disclosure information, see page 829. A302 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS KOHEI KAKU, ARIHIRO KIYOSUE, SATOSHI INOUE, NAOHIKO UEDA, TAKUTO TOKUDOME, JISIN YANG, ANNA MARIA LANGKILDE, Okayama, Japan, Tokyo, Japan, Osaka, Japan, Mölndal, Sweden PBO (N=87) HbA1c BL, % (SD) All patients Adj. mean change from BL, % (SE) (P value vs PBO) Patients with BL HbA1c < 7% Adj. mean change from BL, % (SE) Patients with BL HbA1c ≥ 7%–< 8% Adj. mean change from BL, % (SE) Patients with BL HbA1c ≥ 8%–< 9% Adj. mean change from BL, % (SE) FPG BL mean, mg/dL (SD) Adj. mean change from BL, mg/dL (SE) (P value vs PBO) BW BL mean, kg (SD) Adj. mean change from BL, kg (SE) (P value vs PBO) *Significant P value P<0.05. 5 mg DAPA (N=86) Clinical Diabetes/ Therapeutics This phase III study (NCT01294423) evaluated the efficacy and safety of the selective sodium-glucose cotransporter 2 inhibitor dapagliflozin (DAPA) in Japanese patients with type 2 diabetes mellitus (T2DM) inadequately controlled with diet or exercise. Patients (N=261, 34-81 y, with modestly elevated mean baseline [BL] HbA1c 7.5%) received placebo (PBO) or DAPA (5 or 10 mg) QD for 24 weeks. Estimated glomerular filtration rate at BL was ≥ 45–< 60 mL/min/1.73 m2 in 28% of patients. Adjusted mean reductions from BL in HbA1c, fasting plasma glucose (FPG) and body weight (BW) were significantly greater with DAPA (5 mg and 10 mg) vs PBO (Table). Similar to other DAPA trials, greater HbA1c reductions were observed in patients with higher BL HbA1c. Overall, 47.7% and 64.8% of patients with DAPA 5 and 10 mg and 51.7% with PBO experienced ≥ 1 adverse event (AE).The majority of AEs were mild or moderate in intensity and unrelated to study treatment. Hypoglycemia was reported by 2 patients with DAPA 10 mg. Four patients reported events suggestive of genital infection (DAPA 10 mg, 2; DAPA 5 mg, 1; PBO, 1) and 4 reported events suggestive of urinary tract infection (DAPA 10 mg, 2; PBO, 2). No events of pyelonephritis were reported. In summary, DAPA (5 and 10 mg) was well tolerated and effective in reducing HbA1c, FPG, and BW over 24 weeks in Japanese patients with T2DM inadequately controlled with diet and exercise. 10 mg DAPA (N=88) 7.50 (0.63) -0.06 (0.06) 7.50 (0.72) -0.41 (0.06) (< 0.0001)* 7.46 (0.61) -0.45 (0.06) (< 0.0001)* (n=21) -0.02 (0.12) (n=45) –0.12 (0.08) (n=19) –0.03 (0.13) (n=20) -0.16 (0.12) (n=50) –0.33 (0.08) (n=12) –0.73 (0.16) (n=19) -0.12 (0.12) (n=49) –0.38 (0.08) (n=19) –0.94 (0.13) 139.8 (21.7) 5.8 (2.2) 137.5 (24.4) –8.6 (2.2) (< 0.0001)* 138.7 (22.3) –13.7 (2.2) (< 0.0001)* 1165-P Pharmacokinetics (PK) & Pharmacodynamics (PD) of the GPR40 Agonist TAK-875 and Glimepiride Following Co-Administration in Subjects With Type II Diabetes (T2DM) MAX TSAI, RONALD D. LEE, MAJID VAKILYNEJAD, WENCAN ZHANG, JOHN MARCINAK, XUEJUN PENG, Deerfield, IL This phase 1, single-blind, placebo-controlled, sequential, single-site study evaluated the effects of multiple oral doses of TAK-875 on the PK of a single oral dose of glimepiride and effects of a single dose of glimepiride on steady-state PK and PD of TAK-875 in T2DM subjects. T2DM subjects (n=30) received a single dose of glimepiride 2 mg on Day 1 followed by TAK-875 50 mg QD on Days 3-19. On Day 18, glimepiride was also given with TAK-875. Serial plasma PK samples were collected on Days 1-3 & Days 18-20 for glimepiride and on Days 13-19 for TAK-875 & M-I. Serial PD samples for markers of fasting glucose, insulin, C-peptide, and glucagon were collected on Days 17-18. All PK & PD parameters were estimated using noncompartmental methods. PK interactions were assessed by point estimates and 90% confidence intervals (CIs) of the ratios of the central values for Cmax and AUC values when treatments were coadministered relative to each treatment alone using paired t-test. The effects of glimepiride on TAK875 PD response (AUEC) were similarly assessed for each marker. Glimepiride was absorbed (Tmax ~ 3 hr) and eliminated (T1/2 ~ 7.5 hr) following dosing of either glimepiride alone or with TAK-875. TAK-875 was absorbed (Tmax ~ 4 hr) following dosing of either TAK-875 alone or with glimepiride. The 90% CIs for Cmax and AUC of glimepiride, TAK-875, and M-I fell within the no-effect range of 0.80 to 1.25. The mean AUEC values were lower for fasting glucose and higher for other PD markers after dosing of TAK-875 with glimepiride relative to TAK-875 alone; differences were statistically significant (p<0.005) for all markers. Co-administration of TAK-875 with glimepiride was well-tolerated with no hypoglycemia and no effects on the PK of glimepiride, TAK-875, or M-I. Changes in PD markers with coadministration of glimepiride suggest a potential synergistic effect between TAK-875 and glimepiride on insulin secretion mediated via different pathways. 65.96 (12.91) 65.81 (14.37) 69.70 (13.82) –0.84 (0.26) –2.13 (0.27) –2.22 (0.26) (0.0003)* (0.0001)* Supported by: AstraZeneca/Bristol-Myers Squibb 1164-P Efficacy and Safety of Canagliflozin (CANA) Monotherapy in Subjects With Type 2 Diabetes Mellitus (T2DM) Over 52 Weeks KAJ STENLÖF, WILLIAM T. CEFALU, KYOUNG-AH KIM, ESTEBAN JODAR, MARIA ALBA, ROBERT EDWARDS, CINDY TONG, WILLIAM CANOVATCHEL, GARY MEININGER, Gothenburg, Sweden, Baton Rouge, LA, Goyang, Republic of Korea, Madrid, Spain, Raritan, NJ Supported by: Takeda Global Research & Development Center, Inc. CANA is an SGLT2 inhibitor in development for the treatment of T2DM. This randomized, double-blind study enrolled subjects with T2DM inadequately controlled with diet and exercise (N = 584), and included a placebo (PBO)controlled, 26-week core period (CANA 100 and 300 mg vs PBO) and a 26-week extension (n = 451; mean age, 55.5 y; A1C, 7.9%; fasting plasma glucose [FPG], 165.8 mg/dL; BMI, 31.8 kg/m2; blinded switch of PBO group to sitagliptin 100 mg [PBO/SITA]). Efficacy data at 52 weeks are reported for CANA 100 and 300 mg (SITA used only to maintain double-blind and control group); safety data are shown for CANA and PBO/SITA. At Week 52, CANA showed numerical, dose-related reductions from baseline in A1C, FPG, and ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A303 POSTERS body weight; decreases in systolic BP and triglycerides, and increases in HDL-C and LDL-C, were also seen. Overall AE rates were 67%, 66%, and 64% with CANA 100 and 300 mg and PBO/SITA. Rates of serious AEs, AE-related discontinuations, and hypoglycemia were low across groups. Genital mycotic infection rates were higher in the pooled CANA group than PBO/SITA (females, 10% vs 5%; males, 8% vs 0%). Higher rates of AEs related to osmotic diuresis (5%, 8%, 2%) and reduced intravascular volume (≤1% per AE) were seen with CANA 100 and 300 mg versus PBO/SITA; UTI rates were 8%, 7%, and 6%. In summary, CANA monotherapy provided sustained improvement in glycemic control and body weight reduction, and was generally well tolerated in subjects with T2DM over 52 weeks. 1163-P Efficacy and Safety of Dapagliflozin Monotherapy in Japanese Patients With Type 2 Diabetes Inadequately Controlled With Diet and Exercise CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1166-P Eating Behavior and Taste Threshold as Possible Predictors of the Efficacy of DPP-4 Inhibitors POSTERS Clinical Diabetes/ Therapeutics CHIEKO SAKAI, HISASHI SHIMOHIRO, YOSHIE OTA, HIRONORI SAKINADA, TATSUO TAKEUCHI, SHIN-ICHI TANIGUCHI, KEIICHI HANAKI, Yonago, Japan Extrapancreatic action of GLP-1, especially appetite regulation, has recently been emphasized. To clarify the association between hypoglycemic efficacy of DPP-4 inhibitors and indices for food intake such as eating behavior or taste threshold, we conducted a comparative study on T2D patients. Thirty-seven diabetic patients, either medicated (except incretin-based drugs) or not, were randomly recruited from an outpatient diabetes clinic (25 males, 12 females, age: 61.0±11.5 years, BMI: 25.4±3.7, A1C: 7.9±1.0%). They were divided into 2 groups depending on whether any DPP-4 inhibitor was started or not during the study period. The treatment group (n=20), on DPP-4 inhibitor, was further divided into a responder and a non-responder groups at 1 or 3 months after the start of the treatment. Eating behavior was evaluated using a 40-item questionnaire categorized as “diversion eating”, “impulse for eating”, “sensation of eating heartily”, “quick eating”, “high-fat/fast food preference”, “irregular meal times”, “hungry sensation”. Thresholds of recognition for 5 basic tastes (sweet, salty, sour, bitter, umami) were measured using a filter paper disk method. 1) Comparison between before and after the period: There was no significant weight gain or loss in the treatment group, although significant gain in the control group. 2) Between the treatment and control groups: Indices for eating behavior or taste thresholds were comparable. 3) Between the responder and non-responder groups: A high-fat diet/fast food preference score measured before the treatment was significantly higher in the responder than in the non-responder group (11.6±2.8 vs 7.3±2.2, p<0.05). As for taste, a bitter taste threshold before the treatment was significantly higher in the responder than in the non-responder group (3.0±1.4 vs 1.0±0.0, p<0.05). These results suggest that the efficacy of DPP-4 inhibitors might be predicted by evaluating the characteristics of eating behavior and taste threshold in patients with T2D. Supported by: Tehran University 1168-P Development and Characterization of Irreversible Inhibitors of Insulin-Degrading Enzyme SAMER O. ABDUL-HAY, THOMAS BANNISTER, MALCOLM A. LEISSRING, Jacksonville, FL, Jupiter, FL Insulin-degrading enzyme (IDE) is an atypical zinc-metallopeptidase that plays an essential role in terminating the insulin response and is also linked to the pathogenesis of type-2 diabetes mellitus. Despite widespread interest in IDE spanning more than a half century, pharmacological inhibitors of this important peptidase have remained elusive. Here we describe the discovery, optimization and characterization of the first small-molecule inhibitors of IDE. Using a fluorescence polarization-based IDE activity assay, we conducted an ultra-high-throughput (1536-well) cell-based compound screening campaign on a library of ~325,000 small molecules conforming to Lipinski’s Rule of Five. Potent (IC50 >10 µM) hits were further subjected to counterscreens to sensitive to cytotoxicity, fluorescence artifacts, and ability to inhibit IDE in vitro. Among the top hits we identified several isothiazolones that inhibit IDE with IC50 values in the low micromolar range. Using diverse approaches, we show that the mechanism of inhibition involves the irreversible modification of a specific cysteine residue within IDE’s active site (C819), which occurs via formation of a disulfide bond. Through subsequent structure-activity relation (SAR) studies, novel variants were synthesized with improved IC50 values as low as ~100 nM. The resulting compounds represent the first potent small-molecule inhibitors of IDE, which should have utility both as chemical probes for interrogating IDE’s role in insulin signaling and also as potential pharmacophores for future drug development. 1167-P Cichorium Intybus L. Extract (CI) as an Insulin Sensitizer AZIN NOWROUZI, YASIN POURFARJAM, AZITA AMIRI, Tehran, Islamic Republic of Iran, Huntsville, AL Cichorium intybus L. seed extract (CI) was shown previously to improve glucose tolerance test (GTT) profile in a group of rats that were made mildly diabetic using a combination of Streptozotocin (STZ) and Niacinamide (NIA), called NIA/STZ group. The same effect was not observed in severe diabetic rats, STZ group (made diabetic using STZ alone) presumably due to the absence of insulin (INS) in this group of rats. The present study re-examined the effect of crude chicory seed extract on GTT in the presence and absence of INS. GTT was performed four times in a fasting group of rats (n=8) with severe diabetes. The test was performed in separate days by intraperitoneal administration of Glucose, Glucose + CI, Glucose + INS, or Glucose + CI + INS. The concentrations of Glucose and CI, which were dissolved in normal saline, were 2 g and 125 mg, per kg body weight, respectively; 5 U of regular INS was used for all the rats. Blood sugar was measured in blood from the tail by a glucometer every 30 minutes. The results showed that blood sugar decreased under the effect of INS, but CI significantly improved INS action in lowering blood sugar during two hours of GTT. The results suggest that the glucose lowering property of chicory depends on the presence of insulin; CI acts as an insulin sensitizer leading to alleviation of insulin resistance in target organs in type 2 diabetes mellitus. 1169-P Improved Liver Histology in a Patient With Non-Alcoholic Steatohepatitis after Treatment With the Glucagon-Like Peptide-1 Receptor Agonist Liraglutide-A Case ANDERS E. JUNKER, LISE LOTTE GLUUD, FILIP K. KNOP, TINA VILSBØLL, Copenhagen, Denmark Non-alcoholic steatohepatitis (NASH) affects up to 30% of obese patients with type 2 diabetes and is expected to be the leading cause of liver transplantation by 2020. Recent studies have demonstrated beneficial effects of glucagon-like peptide-1 receptor (GLP-1R) agonists on liver function tests (LFTs). A 25-year old obese woman (body weight: 89 kg; body mass index: 32 kg/m2) with a 2-year history of type 2 diabetes (HbA1c: 6.3%) treated with metformin (1000 mg twice-daily) and insulin (30 IE once-daily) presented with elevated LFTs (alanine transaminases (ALT): 133 U/l; aspartatetransaminases (AST): 76 U/l) and NASH (liver biopsy: non-alcoholic fatty liver disease (NAFLD) score 5; hepatocytes with steatosis: 66%). Treatment with the GLP-1R agonist liraglutide was initiated at 0.6 mg once-daily and uptitrated to 1.8 mg once-daily after two weeks. Seven weeks after liraglutide initiation the patient had lost 7 kg and ALT had decreased to 66 U/l and AST to 42 U/l. After 46 weeks of liraglutide treatment LFTs were normalized (ALT: 29 U/l, AST: 25 U/l), histology was improved (NAFLD score 3; hepatocytes with steatosis: 40%) and glycemia was well controlled (HbA1c: 5.6%) without insulin treatment. The total weight loss was 16 kg. Liraglutide markedly improved LFTs and histology in this patient with NASH and type 2 diabetes. Randomized controlled trials are needed to evaluate whether the present findings hold promise for the treatment of NASH. & For author disclosure information, see page 829. A304 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1170-P 1172-P Single Doses of an iBAT (SLC10A2) Inhibitor Suppress 24h Serum Bile Acid Concentrations and Enhance Bile Acid Synthesis The Anti-Diabetes Efficacy of Bromocriptine-QR in Type 2 Diabetes Mellitus (T2DM) Subjects Is Enhanced among those With Elevated Blood Pressure and Plasma Triglyceride Levels ROBERT L. DOBBINS, JIANG LIN, ROBIN O’CONNOR-SEMMES, SUSAN L. MCMULLEN, MICHAEL J. BISHOP, LIHONG CHEN, Research Triangle Park, NC, Durham, NC Figure 1. 1171-P Bile Acid Sequestrants for Glycemic Control in Type 2 Diabetes: A Systematic Review With Meta-Analysis of Randomized Controlled Trials MORTEN HANSEN, KRISTIAN H. MIKKELSEN, DAVID P. SONNE, LISE L. GLUUD, TINA VILSBØLL, FILIP K. KNOP, Hellerup, Denmark These findings suggest that T2DM subjects with higher rather than lower BP and TG levels may on average respond better to the anti-diabetes effects of B-QR. We aimed to evaluate the glucose-lowering effect and safety of bile acid sequestrants (BASs) in patients with type 2 diabetes. We conducted a systematic review with meta-analyses of randomized controlled trials (RCTs). Manual and electronic searches (The Cochrane Library, MEDLINE and EMBASE) were combined. The interventions assessed were BASs (colesevelam, colestimide, colestipol or colestyramine) vs. placebo, oral antidiabetic drugs or insulin. The primary endpoint was change in HbA1c. Secondary endpoints were changes in fasting plasma glucose (FPG) and adverse events. Meta-analyses were performed using random effects models due to an expected clinical heterogeneity between trials. We included 17 RCTs (10 trials with colesevelam and 7 trials with colestimide) with a duration of 2 to 52 weeks, including 2,118 patients with type 2 diabetes. In random effects meta-analysis, BASs had a beneficial effect on end of treatment HbA1c (weighted mean difference (WMD) [95% confidence interval (CI)]: -0.59% [-0.70,-0.48]) and end of treatment FPG (-0.52% [-0.68,-0.37]). No evidence of publication bias or other small study effects was identified. BASs did not increase the risk of serious adverse events, but increased the number of non-serious adverse events (relative risk: 1.89 [1.32,2.67]), constipation being the most common. This review suggests that BASs improves glycemic control in patients with type 2 diabetes without increasing the risk of serious adverse events. ADA-Funded Research & 1173-P The Changes of Leukocyte Telomere Length and Telomerase Activity after Intervention of Sitagliptin in Initial Type 2 Diabetes DELIN MA, YANG YAN, Wuhan, China The aim of this study was to determine the effect of sitagliptin on leukocyte telomere length and telomerase activity in initial type 2 diabetic patients. We selected initial type 2 diabetic patients (n=38) and matched non-diabetic subjects (n=31) from the outpatient of Tongji Hospital, Tongji Medical College, Huazhong university of Science and Technology. Leukocyte telomere length ratio(T/S ratio) was measured using a quantitative PCR and analyzed. The activity of telomerase was measured by TRAP-ELISA method. Peripheral insulin resistance (homeostasis model assessment) was calculated from fasting plasma glucose and fasting plasma insulin. The results showed that telomere length(T/S ratio) of the type 2 diabetic patients (1.58±0.57) was significantly shorter than those of control subjects (3.98±0.90), and was significantly elongated after intervention by sitagliptin. There was no significant difference between the T2D and control group in telomerase activity, and the treatment of sitagliptin in T2D group showed no significant effect on the telomerase activity. Thus, we came to the conclusion that sitagliptin might protect β-cell in pancreases by elongating the telomere length in initial type 2 diabetes. For author disclosure information, see page 829. Guided Audio Tour poster A305 POSTERS Available evidence suggests that Bromocriptine-QR (B-QR) improves glycemic control in T2DM subjects via the central nervous system in part by reducing overactive sympathetic drive (OSD) to the periphery which is also a contributor to metabolic (insulin resistance) syndrome, particularly to hypertension and elevated plasma triglyceride (TG) levels thereof. Therefore, this study investigated whether the anti-diabetes effect of B-QR may be enhanced in T2DM subjects with BP and TG levels at or above the NCEP ATP III definition of metabolic syndrome (BP≥ 130/85 mmHg and TG≥ 150 mg/ dl) (surrogate for OSD) versus in those with normal blood pressure without anti-hypertensive therapy (<120/80 mmHg) and normal TG levels (< 150 mg/ dl). T2DM subjects on sulfonylurea therapy (HbA1c: 9.4) were randomized to B-QR (N=183) or placebo (N=211) and treated for 24 weeks while fasting and postprandial plasma glucose and HbA1c levels were monitored. The between group difference, B-QR versus Placebo, in change from such baseline values was determined by linear regression adjusting for Weight at baseline for the entire study population and for subsets after stratification by baseline BP and TG levels. Results are summarized in Figure 1. This clinical trial examined safety, tolerability, pharmacokinetics and pharmacodynamics of an ileal bile acid transport (iBAT) inhibitor, GSK2330672, previously shown to improve glucose and triglyceride levels in Zucker diabetic rats. It was a single blind, randomized, placebo-controlled, dose escalating, 4-way crossover trial. Each dosing period required a 4 night/3 day clinical admission. Subjects received identical meals each period and study drug was administered on Day 1 after an overnight fast. Healthy volunteers (n=17) were randomized to placebo plus 3 doses of GSK2330672. Groups of 6-7 individuals received 0.1, 0.3, 1, 3, 10, 30 and 60 mg before breakfast or 30 mg before breakfast and supper. Safety measures included clinical lab tests, ECGs, vital signs, cardiac telemetry and spirometry testing. Stool form and frequency of bowel movements were recorded. Fecal samples were collected for 48 h after start of dosing to assess bile acid excretion. Blood samples were obtained on Day 1 for assays of GSK2330672, serum bile acid and 7 alpha-hydroxy-4-cholesten-3-one (C4) concentrations. The most frequently reported adverse events included headache, dizziness, diarrhea, abdominal pain/discomfort and flatulence. Side effects were more likely to be reported at doses ≥30 mg. No clinically meaningful, treatmentrelated findings were observed for safety measures during the study. There was no measurable plasma GSK2330672 exposure at total daily doses up to 60 mg. Doses of 10mg QD, 60mg QD and 30mg BID significantly increased fecal bile acids vs. placebo and suppressed the initial peak in serum total bile acid concentrations that follows 1.5 h after breakfast. Doses of 60mg QD and 30mg BID consistently elevated concentrations of the bile acid precursor, C4, for 24.5 h. Single dose escalation of GSK2330672, a GI lumen-restricted iBAT inhibitor, demonstrated significant effects on bile acid excretion and synthesis at doses that were generally safe and well-tolerated. Clinical Diabetes/ Therapeutics ANTHONY H. CINCOTTA, MICHAEL EZROKHI, Tiverton, RI CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS 1175-P Efficacy and Safety of the Dipeptidyl Peptidase-4 Inhibitor, Sitagliptin Compared With α-Glucosidase Inhibitor in Japanese Patients With Type 2 Diabetes Inadequately Controlled on Metformin or Pioglitazone Alone (SUCCESS1): A Multicenter, Randomized, Openlabel, Clinical Trial HIDETAKA YOKOH, KAZUAKI KOBAYASHI, YASUNORI SATO, MINORU TAKEMOTO, DAIGAKU UCHIDA, AZUMA KANATSUKA, NOBUICHI KURIBAYASHI, KENICHI SAKURAI, TAKASHI TERANO, NAOTAKE HASHIMOTO, HIDEKI HANAOKA, KO ISHIKAWA, SHUNICHIRO ONISHI, KOUTARO YOKOTE, THE SUCCESS STUDY GROUP, Yachio Chiba, Japan, Funabashi, Japan Dipeptidyl peptidase-4 (DPP-4) inhibitors have been used widely lately, but its most appropriate position of DPP-4 inhibitors in combination therapy in type 2 diabetes has not been well established. Therefore, we examined the efficacy and safety of a DPP-4 inhibitor, sitagliptin in comparison with α-glcosidase inhibitor (αGI), a widely used class of hypoglycemic agent in Japan, in type 2 diabetes in the setting of combination therapy with metformin or pioglitazone. A multicenter, prospective, randomized, open-label trial was conducted at 22 hospitals in Japan from January 2011 to June 2012. Participants aged 2079 years with type 2 diabetes who had treated with metformin or pioglitazone alone but did not achieve target HbA1c (6.9-8.8%) were randomly assigned to sitagliptin (50 mg daily) or αGI (0.6 mg voglibose or 150 mg miglitol daily). The primary endpoint was change in HbA1c from baseline at 12 weeks. All data were analyzed according to the intention-to-treat principle. Sixty patients entered in the sitagliptin group and 59 in the αGI group. Mean values of HbA1c at baseline was 7.6% in the both groups. After 12 weeks, sitagliptin treatment led to significant reductions compared with αGI in difference of HbA1c (−0.49%). Significant difference was found between two groups at 24 weeks. Gastrointestinal disorders was more common with αGI than with sitagliptin (23 [40%] patients vs 6 [11%], p=0.0003). Minor hypoglycemia occurred in about 4% of patients in each group. Sitagliptin was superior to αGI for reduction of HbA1c, and was well tolerated with minimum risk of gastrointestinal disorders and hypoglycemia in type 2 diabetes inadequately controlled on metformin or pioglitazone alone. These findings support the use of sitagliptin in the setting of combination therapy with metformin or pioglitazone. 1174-P POSTERS Clinical Diabetes/ Therapeutics Efficacy and Tolerability of Saxagliptin (SAXA) in Patients With Type 2 Diabetes Mellitus (T2DM) and a History of Cardiovascular Disease (CVD) GIANMARIA MINERVINI, WILLIAM COOK, ELSIE ALLEN, Wilmington, DE, Princeton, NJ Subgroup analysis of data from patients with T2DM and a history of CVD (previous event or CVD diagnosis) included data from 3 phase 3 studies: SAXA+metformin (MET) vs MET as initial therapy (NCT00327015), SAXA+MET vs glipizide (GLIP)+MET (NCT00575588), and SAXA vs placebo (PBO) add-on to insulin (INS)±MET (NCT00757588). Glycemic efficacy and adverse events (AEs) for SAXA vs PBO or comparator were analyzed in patients with/without a history of CVD. No treatment-by-subgroup interactions (P<0.1) were noted for change from baseline in A1C (Table 1). AEs were comparable across CVD groups and reported by 47%-69% of patients. Symptomatic confirmed hypoglycemia (fingerstick glucose ≤50 mg/dL) was reported in 0%-9% of patients, with more episodes reported for patients receiving GLIP (Table 2). SAXA improves glycemic control and is generally well tolerated in patients with T2DM regardless of CVD history. Supported by: Waksman Financial Group 1176-P Dapagliflozin as Part of Triple Combination Therapy Helps Reduce HbA1c and Body Weight in Patients With Type 2 Diabetes SERGE JABBOUR, ELISE HARDY, TJERK W. DEBRUIN, INGRID GAUSE-NILSSON, KATJA ROHWEDDER, PAULA MARTIN, SHAMIK J. PARIKH, Philadelphia, PA, Wilmington, DE, Mölndal, Sweden, Wedel, Germany Dapagliflozin (DAPA) is an SGLT2 inhibitor that reduces hyperglycemia in patients with type 2 diabetes mellitus (T2DM) through the urinary excretion of glucose that is also associated with a reduction in body weight (BW). DAPA has demonstrated efficacy as monotherapy in treatment naïve patients and in patients inadequately controlled with single oral antidiabetic drugs (OADs) or insulin (+/- OADs). In the progression of T2DM, second and even third agents are often needed to intensify therapy. To investigate the efficacy of DAPA added to two stable medications, we performed subgroup analyses of four 24 week studies. Primary results from the full populations of these studies have been presented previously. Patients in the subgroups received placebo (PBO) or DAPA 10 mg in the presence of either sitagliptin (SITA) and metformin (MET) (prespecified stratification); or sulfonylurea (SU) + MET(HbA1c prespecified analyses no multiplicity testing, BW post hoc analyses); or insulin (INS) + MET (post hoc analyses). DAPA reduced HbA1c and BW compared with PBO at 24 weeks in each of the four subgroups (Table). These results are generally comparable to those observed across the DAPA program, confirming that, even when added to two different antidiabetic agents, DAPA is effective in reducing glycemia and BW in T2DM. This observation is consistent with an insulin-independent mechanism of action that works regardless of residual beta cell function. Supported by: Bristol-Myers Squibb/AstraZeneca & For author disclosure information, see page 829. A306 Guided Audio Tour poster ADA-Funded Research A total of 306 Asian patients (China n=264, Malaysia n=17 and the Philippines n=24) with T2DM were randomized to linagliptin (n=205) or placebo (n=101) at a 2:1 ratio. Baseline characteristics were well matched between groups, with mean (SD) baseline HbA1c of 7.99 ± 0.83% in the linagliptin group and 8.00 ± 0.80% in the placebo group. After 24 weeks, adjusted mean (SE) HbA1c decreased by −0.66 ± 0.05% in the linagliptin group and −0.14 ± 0.07% in the placebo group (placebo-corrected difference, −0.52 ± 0.09%; 95% CI: −0.70, −0.34; P<0.0001). In the pre-defined subgroup of patients with baseline HbA1c ≥8.5%, the placebo-corrected decrease in HbA1c was −0.89 ± 0.17% (P<0.0001). Adjusted mean (SE) FPG decreased by −10.7 ± 2.5 and −1.1 ± 3.5 mg/dL in the linagliptin and placebo groups, respectively (placebo-corrected difference, −9.6 ± 4.2 mg/dL; 95% CI: −17.8, −1.3; P=0.02). Results for the pre-defined Chinese subgroup were similar to the overall Asian population. Linagliptin was well tolerated, with adverse events occurring in a similar proportion of patients as in the placebo group (27.3% and 28.0%, respectively). Drug-related AEs were reported for 2.4% and 0.0% of linagliptin and placebo patients, respectively. Investigatordefined hypoglycemia occurred in 1.0% in both groups. In summary, linagliptin 5 mg was efficacious and well tolerated over 24 weeks in Asian patients and a Chinese subgroup with T2DM inadequately controlled on metformin. DAPA 10 mg HbA1c (%) SITA + MET (NCT00984867) n 113 113 BL HbA1c [% (SD)] 7.87 (0.75) 7.80 (0.81) a 24-wk Change from BL (95% CI) -0.02 (-0.15, 0.10) -0.43 (-0.55, -0.30) Difference from placebo --0.40 (-0.58, -0.23)b SU + MET (study 1) (NCT01031680) n 114 113 BL HbA1c [% (SD)] 8.10 (0.77) 8.06 (0.78) 24-wk Change from BL (95% CI)a -0.01 (-0.17, 0.15) -0.55 (-0.72, -0.39) Difference from placebo --0.54 (-0.73, -0.36)d SU + MET (Study 2) (NCT01042977) n 109 105 BL HbA1c [% (SD)] 7.96 (0.80) 8.03 (0.79) 24-wk Change from BL (95% CI)a -0.07 (-0.26, 0.12) -0.55 (-0.75, -0.36) Difference from placebo --0.48 (-0.69, -0.27)d INS + MET (NCT00673231) n 77 83 BL HbA1c [% (SD)] 8.43 (0.73) 8.52 (0.79) a 24-wk Change from BL (95% CI) -0.31 (-0.47, -0.16) -0.93 (-1.08, -0.78) Difference from placebo ---0.61 (-0.83, -0.40)d Body Weight (kg) SITA + MET (NCT00984867) n 113 113 BL BW [kg (SD)] 94.2 (20.9) 94.0 (19.8) a 24-wk Change from BL (95% CI) -0.47 (-1.00, 0.05) -2.35 (-2.87, -1.82) Difference from placebo --1.87 (-2.61, -1.13)b SU + MET (Study 1) (NCT01031680) n 114 112 BL BW [kg (SD)] 89.0 (15.5) 87.6 (19.0) 24-wk Change from BL (95% CI) a 0.0 (-0.6, 0.6) -2.2 (-2.8, -1.6) Difference from placebo ---2.2 (-2.9, -1.5)d SU + MET (Study 2) (NCT01042977) n 110 106 BL BW [kg (SD)] 90.1 (15.2) 93.4 (19.5) 24-wk Change from BL (95% CI) a -0.8 (-1.5, -0.2) -1.9 (-2.5, -1.2) Difference from placebo ---1.1 (-1.8, -0.3)c INS + MET (NCT00673231) n 77 83 BL BW [kg (SD)] 98.7 (17.5) 95.7 (16.2) a 24-wk Change from BL (95% CI) -0.06 (-0.61, 0.49) -1.77 (-2.30, -1.24) Difference from placebo --1.71 (-2.47, -0.95)d a Adjusted mean change from baseline (LOCF); b prespecified p-value <0.0001 for comparison of DAPA and PBO c Nominal p-value = 0.0047; d Nominal p-value <0.0001; BL - baseline Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1178-P MTBL0036, a New Promising Antidiabetic Drug Candidate: In Vivo Preclinical Studies MAHA EL HAGE, GÉRARD MOINET, BERNARD FERRIER, RÉMI NAZARET, GUY MARTIN, AGNÈS CONJARD-DUPLANY, GABRIEL BAVEREL, Lyon, France The objective of the present study was to establish the characteristics of the effect of MTBL0036, a new optimized and patented antidiabetic drug candidate, in 8-12 month-old db/db mice, a good animal model of type 2 diabetes. For this, MTBL0036 was administered per os as a suspension in 2% methylcellulose and the control animals received per os only the vehicle which itself did not influence glycemia. The glycemia was measured at various time points after small blood collections by small incisions of the tail, before and after the oral administration of a glucose load. The glycemia was measured using a glucometer (Lifescan One Touch Ultra, Lifescan, J&J Company, USA). The Areas Under the Curve (AUCs) were calculated using the GraphCalc software. The reduction (by 25-30%) of the blood glucose level in diabetic mice treated with 200 mg MTBL0036/kg body weight was observed one hour after the administration and persisted for at least 4 hours. No anti-hyperglycemic effect was observed after the administration of 200 mg metformin/kg body weight. When administered orally 120 min before the glucose load, two doses of MTBL0036 (25 and 50 mg/kg) reduced by 60% and 80%, respectively, the AUCs for glucose after the glucose load in 17h-fasted diabetic mice. Under the same conditions, metformin (50 mg/kg) did not reduce the AUC for glucose. In fed db/db mice treated for 5 days with a daily dose of 50 mg MTBL0036/kg, the level of neither insulin nor triglycerides was found to be elevated. MTBL0036 was found to be eliminated, at least in part, in urine in an unchanged form. It is concluded that, in db/db mice, MTBL0036 is more efficacious than metformin, the gold-standard treatment for type 2 diabetes. In addition, this promising antidiabetic drug candidate seems to be devoid of hypoglycemic and cardiovascular risks. Moreover, the risk for interacting with the metabolism of other drugs seems to be limited. Supported by: AstraZeneca/Bristol-Myers Squibb 1177-P Efficacy and Safety of Linagliptin in Asian Patients With Type 2 Diabetes Mellitus Inadequately Controlled on Metformin: A 24-Week, Randomized, Phase III Clinical Trial WEIQING WANG, JINKUI YANG, GANGYI YANG, YAN GONG, HANS-JUERGEN WOERLE, GUANG NING, Shanghai, China, Beijing, China, Chongqing, China, Ingelheim, Germany This randomized, Phase III, placebo-controlled, double-blind 24-week study evaluated the efficacy and safety of the DPP-4 inhibitor linagliptin added to metformin in patients with type 2 diabetes mellitus (T2DM) in Asian countries (NCT01215097). Antidiabetes drugs other than metformin were washed out for 4 weeks prior to randomization to either linagliptin 5 mg daily or placebo added to metformin. The primary endpoint of the study was change from baseline in mean HbA1c after 24 weeks. ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A307 POSTERS PBO Clinical Diabetes/ Therapeutics CLINICAL THERAPEUTICS/NEW TECHNOLOGY—ORAL AGENTS CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS increased after donepezil treatment in both SHR (0.321 ± 0.009 vs. 0.296 ± 0.004, p < 0.02) and WKY (0.339 ± 0.008 vs 0.313 ± 0.008, p < 0.03) when compared with placebo. There was no significant difference in body weight among the 4 groups. We conclude donepezil treatment ameliorates IR in SHR and may potentially be repurposed as adjunctive treatment for diabetes. CLINICAL THERAPEUTICS/NEW TECHNOLOGY— PHARMACOLOGIC TREATMENT OF COMPLICATIONS Guided Audio Tour: Management of Complications of Diabetes (Posters: 1179-P to 1186-P), see page 17. & & 1179-P MASAHIRO OKOUCHI, NAOTSUKA OKAYAMA, TAKASHI JOH, Inazawa, Japan, Nagoya, Japan POSTERS TALI CUKIERMAN-YAFFE, JACKIE BOSCH, LEANNE DYAL, PETER JOHNSTON, HERTZEL C. GERSTEIN, ON BEHALF OF THE ORIGIN INVESTIGATORS, Ramat Gan, Israel, Hamilton, ON, Canada, Bridgewater, NJ Clinical Diabetes/ Therapeutics 1181-P GLP-1 Analogs Represent a Novel Treatment Strategy for Alzheimer’s Disease in Patients With Type 2 Diabetes Effect of Basal Insulin Glargine and Omega 3 FA on Cognitive Decline and Dementia in People With Dysglycemia Diabetic patients commonly develop cognitive impairment of Alzheimer’s disease type, defined as diabetic encephalopathy, which is involved in oxidative stress-induced hippocampal neuronal apoptosis. However, the treatment strategy has not been elucidated. GLP-1, biguanides (BG) or thiazolidinediones (TZD) have been shown to exert neuroprotective actions in Alzheimer’s disease. GLP-1 receptors are widely expressed in neurons throughout the brain, and GLP-1 readily crosses the blood-brain barrier. Therefore, in vivo and in vitro, the differential neuroprotective effect of GLP-1, BG or TZD was investigated. First, we compared the difference of hippocampal atrophy on MRI using VSRAD among 300 type 2 diabetic patients treated with GLP-1 analogs, BG or TZD, and 200 type 2 diabetic patients without those medications, and 100 control subjects. Diabetic patients had more hippocampal atrophy than control subjects. Addionally, diabetic patients treated with GLP-1 analogs, BG or TZD had less hippocampal atrophy than diabetic patients without those medications. In vitro, GLP-1 protected against oxidative stress-induced neuronal apoptosis through activating GLP1 receptor mediated EGFR transactivation, followed by PI3K/Akt/mTOR/GCL signaling. We also found an increase in progenitor cells or young neurons in the hippocampal regions after GLP-1 analogs treatment. BG protected against oxidative stress-induced neuronal apoptosis through inhibiting PTP opening, cytochrome c release and caspase-9 and -3 activation. TZD protected against oxidative stress-induced apoptosis in association with the upregulation of Bcl-2 protein, consequently led to the inhibition of caspase-9 and -3 activation. In conclusion, GLP-1, BG or TZD afford neuroprotection through enhancing different anti-apoptotic signaling and increasing neural stem/progenitor cell proliferation. Thus, those treatments represent a promising treatment modality for diabetic encephalopathy. Recent evidence suggests that diabetes and nondiabetic dysglycemia are risk factors for an accelerated rate of cognitive decline. Whether normalizing glucose levels with insulin glargine and/or administering Omega 3 FA in this population may slow this is unknown. The ORIGIN trial which tested the cardiovascular effect of these two interventions in 12,537 participants with prediabetes or early diabetes followed for a median of 6.2 years, also assessed the effects of the interventions on the rate of cognitive decline and the development of probable dementia. Cognitive function was assessed utilizing the Mini Mental State Examination & Digit Symbol Substitution. The effect of the intervention on cognitive function over time; the annualized change in test scores; and the development of probable dementia defined as either reported dementia or a MMSE score < 24 were determined. There was no significant difference in the rate of change in cognitive test scores in the insulin-glargine comparison versus the standard care group or in the omega-3 versus the placebo group. Similarly, after adjustment for age, and the factorialized treatment arm, the incidence of probable dementia did not statistically differ between either the glargine group and standard care group or the omega-3 and placebo group (Table). In people with dysglycemia, insulin mediated normoglycemia and omega-3 had a neutral effect on the rate of cognitive decline and on incident probable dementia. & 1182-P Exenatide Once Weekly: Association between Weight Response, Glycemic Control, and Markers of Cardiovascular Risk Supported by: Sanofi (NCT00069784) & LAWRENCE BLONDE, LEIGH MACCONELL, WENYING HUANG, RICHARD PENCEK, New Orleans, LA, San Diego, CA Exenatide once weekly (EQW) has been shown to improve glycemic control and body weight (BW) in patients with T2DM. To better describe the relationship of glycemia, cardiovascular (CV) risk factors, and BW, a post hoc analysis based on ΔBW quartiles was conducted using pooled data (intentto-treat, N=1830; treated with EQW ± oral antidiabetes medications) from 8 randomized, controlled 24-30 wk trials. LOCF changes from baseline in A1C, fasting plasma glucose (FPG), BW, lipids and blood pressure were summarized by BW response quartile. Overall, clinically relevant decreases in A1C (-1.4%), FPG (-36 mg/dL), BW (-2.4 kg), and systolic blood pressure (SBP; -3 mmHg) were observed, as were improvements in other CV risk markers. Safety and tolerability with EQW were consistent with previous reports. Patients in 3 of 4 ΔBW quartiles (75%) showed weight reductions. Clinically relevant A1C reductions were observed across all ΔBW quartiles. EQW treatment was also associated with improvement in CV risk factors in at least 3 of 4 ΔBW quartiles including: total cholesterol, LDL-C, HDL-C, triglycerides, and SBP. The greatest improvement in cardiovascular risk factors was seen in the patient quartile with the greatest reduction in weight. In conclusion, clinically important A1C improvement was seen across all quartiles, including those with little or no weight loss. The majority of patients achieved reductions in BW associated with improvements in markers of CV risk. 1180-P Donepezil, an Acetylcholinesterase Inhibitor Used to Treat Alzheimer Dementia, Ameliorates Insulin Resistance in Spontaneously Hypertensive Rats (SHR) DONGMEI ZHANG, XIAO JIAN SUN, SOOHYUN P. KIM, STEPHEN L. WILLS, MICHAEL J. QUON, Baltimore, MD, Easton, MD, Bethesda, MD Donepezil, an acetylcholinesterase inhibitor used to treat Alzheimer dementia, may improve brain blood flow by raising acetylcholine levels that stimulate production of nitric oxide (NO). NO from endothelium increases blood flow in skeletal muscle to determine ~40% of insulin-mediated glucose uptake. Therefore, we hypothesized donepezil therapy may ameliorate insulin resistance (IR) in SHR, characterized by hypertension, endothelial dysfunction, and IR. SHR and WKY littermate control male rats (9 wk old) were treated with donepezil (10 mg/kg/d in 0.5% methylcellulose) or placebo (0.5% methylcellulose) by gavage for 12 wk (n = 6/group). After 4 wk, 2 h glucose tolerance tests (GTT, 2 g glucose/kg, i.p.) were conducted. Improvement in glucose tolerance was noted in SHR treated with donepezil vs placebo (AUCglucose = 7340 ± 770 vs 11100 ± 1360 (mg/dl)·min, p < 0.008). No significant differences in GTT were noted among WKY rats. After 12 wk, 2 h insulin tolerance tests (ITT, 0.75 IU insulin/kg, i.p.) were conducted. We observed a tendency for insulin sensitivity to improve in SHR treated with donepezil vs placebo (AOCglucose = 6110 ± 170 vs 5540 ± 240 (mg/dl)·min, p < 0.06). No significant differences in ITT were noted among WKY rats. After 12 wk, fasting insulin was reduced by donezepil treatment when compared with placebo in both SHR (14 ± 2 vs 21 ± 2 IU/ml, p < 0.03) and WKY (9 ± 1 vs. 15 ± 2, p < 0.014) rats. Fasting glucose was reduced in SHR (105 ± 5 vs. 117 ± 3 mg/dl, p < 0.04) and tended to be reduced in WKY (101 ± 5 vs 112 ± 3, p < 0.09) when compared with placebo. QUICKI, an index of insulin sensitivity, was & For author disclosure information, see page 829. A308 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS 1183-P Co-Administration of the DPP-4 Inhibitor Linagliptin and Native GLP-1 Induce Body Weight Loss and Appetite Suppression HENRIK HANSEN, GITTE HANSEN, SARAH PAULSEN, MICHAEL MARK, NIELS VRANG, THOMAS KLEIN, Hørsholm, Denmark, Biberach, Germany Linagliptin is a dipeptidyl peptidase (DPP)-4 inhibitor approved for the treatment of type 2 diabetes. DPP-4 inhibitors are weight-neutral, suggesting that elevation of endogenous incretin levels is not sufficient to promote weight loss per se. Here we evaluated the effect of subcutaneous co-administration of linagliptin and native GLP-1(7-36) in rats. In normalweight rats, acute linagliptin treatment (0.5 mg/kg, sc BID) had no effect on nocturnal food intake, whereas GLP-1 treatment (0.4 mg/kg, sc BID) evoked a modest and short-lived suppression of food intake. In contrast, linagliptin and GLP-1 co-administration induced a robust acute anorectic response. In dietinduced obese (DIO) rats, 14 days of linagliptin or GLP-1 monotherapy had no effect on body weight, whereas continuation with combined linagliptin (0.5 mg/kg, sc BID) and GLP-1 treatment (0.4 mg/kg, sc BID) for an additional 14 days induced a sustained decrease in food intake and body weight (-6.4 ± 0.8% compared with baseline body weight). Interestingly, the body weight-lowering effect of combined linagliptin and GLP-1 treatment was associated with a marked increase in chow preference at the expense of palatable high-fat-carbohydrate diet intake. In addition, combined linagliptin and GLP-1 treatment specifically increased preprodynorphin mRNA levels in the nucleus accumbens. These data demonstrate that combined treatment with linagliptin and GLP-1 synergistically reduces body weight in obese rats. This anti-obesity effect is caused by appetite suppression and change in diet preference, presumably associated with increased dynorphin activity in dopaminergic forebrain regions involved in reward anticipation and habit learning. In conclusion, linagliptin and GLP-1 co-administration may therefore hold promise as a novel therapeutic principle for combined weight and diabetes management in obese patients. & KENNETH CUSI, BEVERLY ORSAK, ROMINA LOMONACO, JOAN FINCH, CAROLINA ORTIZ-LOPEZ, FERNANDO BRIL, NORMA DIAZ, ROSE KAMINSKI-GRAHAM, CELIA DARLAND, NICOLAS MUSI, AMY WEBB, JEAN HARDIES, FERMIN TIO, Gainesville, FL, San Antonio, TX Nonalcoholic steatohepatitis (NASH), the more severe form of fatty liver associated with necroinflammation and risk of cirrhosis, is common in T2DM. The only effective therapy to date in this population is pioglitazone (PIO), but it has been studied only in a small 6-month randomized controlled trial (RCT) (Belfort et al, NEJM 2006). To examine its long-term safety and efficacy, we treated 101 patients (pts) with prediabetes or T2DM (51%) and NASH (age=51±1 yr, BMI: 34.4 ± 0.5 kg/m2, A1c: 6.3 ± 0.1%) for 18 months in a RCT of diet and PIO vs. diet and placebo. We assessed at 0 and 18 months: 1) Liver histology by biopsy (primary endpoint); 2) Liver fat by magnetic resonance and spectroscopy (MRS); 3) Total body fat (TBF) by DXA; 4) Liver/ muscle (Rd) insulin sensitivity (insulin clamp with 3-3H glucose); and 5) fasting hepatic insulin resistance (IR) (HIRi= EGP x fasting plasma insulin [FPI]) and adipose tissue IR (AdipoIRi= FFA x FPI). Placebo- and PIO-treated pts were well-matched at baseline for clinical, biochemical and metabolic characteristics (all NS). Diet plus placebo significantly lowered ALT (61±5 to 44±4 U/L; p<0.01) but this did not translate into any significant change in liver histology or metabolic parameters. In contrast, PIO compared to placebo significantly improved LFTs, FPG/A1c, FPI, HOMA and insulin sensitivity in adipose tissue (AdipoIRi: -37±3% vs. -10±1%; low-dose insulin-suppr of FFA: 66±4% vs. 44±4%), liver (HIRi: 10.5±1 vs. 28±2 mg/kg-1·min-1·µU/ml; lowdose insulin-suppr of EGP: -50±3% vs. -37±3%) and muscle (Rd: 9.3±0.6 vs. 5.3±0.6 mg·kgLBM-1·min-1; all p<0.01 vs. placebo). PIO decreased 60% liver fat (MRS) and markedly improved steatohepatitis (NAFLD activity score: 4.2±0.2 vs. 2.4±0.2, both p<0.001 vs. placebo), but fibrosis was unchanged. No pt on PIO discontinued the study due to an adverse event. Conclusion: PIO treatment for 18 months is safe and effective to reverse metabolic and liver histological abnormalities in pts with prediabetes/T2DM and NASH. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. & 1184-P Vitamin B12 Deficiency in the Diabetes Prevention Program Outcome Study (DPPOS) JILL P. CRANDALL, SHARON L. EDELSTEIN, VANITA ARODA, GEORGE A. BRAY, SARAH FOWLER, RONALD GOLDBERG, MARY BETH MURPHY, WILLIAM C. KNOWLER, SANTICA M. MARCOVINA, TREVOR J. ORCHARD, DAVID S. SCHADE, MARINELLA TEMPROSA, Bronx, NY, Washington, DC, Hyattsville, MD, Baton Rouge, LA, Rockville, MD, Coral Gables, FL, Memphis, TN, Phoenix, AZ, Seattle, WA, Pittsburgh, PA, Albuquerque, NM Vitamin B12 deficiency occurs with long-term metformin treatment, although it is infrequently recognized. Current guidelines do not recommend routine B12 testing. We measured vitamin B12 levels in metformin (MET) and placebo (PLA) participants in the DPPOS, the follow-up to a randomized clinical trial of diabetes prevention in high-risk adults. B12 levels were assessed by randomized treatment arm. Years of MET exposure were computed using protocol-MET exposure and physician-prescribed MET, reported by participants, for diabetes treatment outside of the protocol. ADA-Funded Research & 1185-P Safety and Efficacy of Long-Term Pioglitazone Treatment for Patients With Prediabetes or T2DM and NASH Supported by: Burroughs Wellcome Fund For author disclosure information, see page 829. Guided Audio Tour poster A309 POSTERS & B12 levels did not differ by age in either MET or PLA. The prevalence of anemia did not differ by B12 status in DPPOS year 1 (MET or PLA) or year 9 (measured in MET participants only). Total MET-years of exposure was the only significant predictor of B12 deficiency in a multivariate logistic model with age, sex, diabetes status, weight change and acid suppressant use. The OR (95% CI) associated with B12 deficiency per year of MET use was 1.25 (0.98-1.56) in year 1 and 1.11 (1.03-1.20) in year 9 in MET, and 1.56 (0.92-2.63) and 1.19 (1.07-1.32) in PLA. Over a mean 13 years follow-up, biochemical vitamin B12 deficiency occurred in over 7% of participants randomized to MET. Prevalence of deficiency increased with longer duration of metformin use. Anemia is not an adequate indicator of B12 deficiency. Given the potential neurologic consequences of untreated B12 deficiency, routine testing of MET treated patients should be considered. Clinical Diabetes/ Therapeutics B12 levels and metformin use by DPP treatment group and study visit DPPOS year 1 DPPOS year 9 (Mean 5 years follow-up) (Mean 13 years follow-up) MET PLA p MET PLA p (N=859) (N=856) (N=753) (N=736) Total metformin-years of exposure (mean +/- SD) * 2.9 ± 1.7 0.1 ± 0.6 -- 9.5 ± 4.4 1.7 ± 2.8 -Percent with Diabetes 33.2% 39.0% -- 52.3% 59.7% -B12 deficiency (≤203 pg/ml) ** 4.3% 2.4% 0.02 7.4% 5.3% 0.12 B12 deficiency or borderline-low B12 (≤298 pg/ml)** 19.1% 9.5% <0.01 20.3% 15.6% 0.02 B12 deficiency (≤203 pg/ml)** among participants 5.2% NA --- 9.2% NA --currently taking study-provided MET (*) includes protocol assigned and non-study metformin use; (**) B12 deficiency defined according to American Society of Hematology, 2003. CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS & of DAPA on lipids in patients with T2DM are reported here. Data from 3731 patients in 12 phase 2b/3 double-blind controlled trials receiving DAPA 5 or 10 mg or placebo (PBO) for up to 24 weeks were analyzed. Changes in the plasma lipid profile were studied as exploratory endpoints in these studies. Demographic and baseline characteristics were balanced between DAPA and PBO groups. Improved glycemic control with DAPA versus PBO has been previously reported in these studies. Changes in lipid parameters from individual trials showed variability across studies. In a 24-week pooled analysis, changes from baseline in HDL-C were +6.5% and +5.5% for DAPA 5 and 10 mg, respectively and +3.8% for PBO; LDL-C were +0.6% and +2.7% for DAPA 5 and 10 mg, respectively, and −1.9% for PBO; total cholesterol were +1.1% and +1.4% for DAPA 5 and 10 mg, respectively, and −0.4% for PBO. To conclude, DAPA is associated with small mean changes in fasting lipid parameters versus PBO at 24 weeks; these changes are not clinically meaningful. 1186-P Specific miRNA Inhibitors Accelerate Wound Healing in Diabetes POSTERS Clinical Diabetes/ Therapeutics BIAO FENG, LINBO ZHANG, SUBRATA CHAKRABARTI, London, ON, Canada miRNAs play an important role in the post-transcriptional regulation of gene expression and controlling various physiological and pathological processes. Our previous studies showed that miR-200b regulates VEGF mediated angiogenesis and miR-146a regulates extracellular matrix protein, fibronectin (FN) production. In this study, we examined the effects of these two miRNAs in wound healing (WH) in the diabetic and control animals. Microvascular endothelial cells (ECs) were exposed to low (5mM) and high (25mM) glucose levels and were tested for VEGF and FN production and examined for angiogenesis. They were also transfected with miRNA mimics and inhibitors of these miRNAs and examined similarly. Furthermore, skin wounds in the mice with or without STZ induced diabetes were treated with these inhibitors, alone or in combinations. Wound areas were measured and tissues from the wounds were examined histologically for scar tissue and for mRNA and miRNA expression by real time RT-PCR. In the ECs, glucose induced increased VEGF and FN production and angiogenesis. These mRNA expression and angiogenesis in the ECs exposed to low concentration glucose were also augmented by miR-200b or miR-146a inhibitor transfection. In the animals, compared to transfection reagent only, WH processes were accelerated by miR-200b or miR-146a inhibitor. However, such acceleration was most pronounced (50% reduction) when these two inhibitors were combined. In the diabetic animals, although overall WH process was delayed, combined inhibitor treatment significantly accelerated such process. Such wound healing was associated with increased VEGF and FN mRNA production and well developed scar tissues These results indicated that combination of miR-146a and miR-200b inhibitors are useful in the accelerating WH both in normal and in diabetic conditions. miRNA mediated WH may potentially constitute a novel therapeutic approach. Plasma lipid changes from baseline (%, 95% CI) in the pooled DAPA studies at 24 weeks PBO DAPA 5 mg DAPA 10 mg (N=1393) (N=1145) (N=1193) TC . . . n 989 888 834 BL mean, mg/dL 195.22 194.48 195.88 Mean change, % −0.4% +1.1% +1.4% (95% CI) (−1.4, +0.6) (0.0, +2.2) (+0.2, +2.6) HDL-C . . . n 990 889 834 BL mean, mg/dL 44.54 44.79 45.04 Mean change, % +3.8% +6.5% +5.5% (95% CI) (+2.8, +4.8) (+5.3, +7.8) (+4.3, +6.7) LDL-C . . . n 985 884 828 BL mean, mg/dL 114.72 113.24 114.09 Mean change, % −1.9% +0.6% +2.7% (95% CI) (−3.5, −0.3) (−1.2, +2.4) (+0.8, +4.7) TG . . . n 984 886 831 BL mean, mg/dL 187.46 190.40 194.21 Mean change, % −0.7% −3.2% −5.4% (95% CI) (−3.0, +1.7) (−5.8, −0.6) (−7.9, −2.8) FFA . . . n 838 732 694 BL mean, meq/L 0.56 0.58 0.56 Mean change, % −5.7% −0.5% +1.2% (95% CI) (−8.9, −2.5) (−3.7, +2.8) (−2.4, +5.0) Data include observations after rescue. N is the number of treated patients. n is the number of treated patients with non-missing baseline and week 24 values. BL=baseline, CI=confidence interval, DAPA=dapagliflozin, FFA=free fatty acids, HDL-C=high-density lipoprotein cholesterol, LDL-C=low-density lipoprotein cholesterol, PBO=placebo, TC=total cholesterol, TG=triglycerides. Supported by: CDA; CIHR Guided Audio Tour: Cardiovascular Implications of Diabetic Therapies (Posters: 1187-P to 1194-P), see page 17. & 1187-P Limiting Cardiac Ischemic Injury by Pharmacologic Augmentation of Translocation of Glucose Transporter JINGYING WANG, LIN LENG, LAWRENCE YOUNG, WILLIAM JORGENSEN, JI LI, RICHARD BUCALA, Buffalo, NY, New Haven, CT Background: Macrophage migration inhibitory factor (MIF) exerts a protective effect on ischemic myocardium by activating AMP-activated protein kinase (AMPK). Small molecules that increase the affinity of MIF for its receptor have been recently designed, and we hypothesized that such agonists may enhance AMPK activation and limit ischemic tissue injury. Methods and Results: Treatment of cardiomyocytes with the candidate MIF agonist, MIF20, augmented AMPK phosphorylation, increased by 50% the surface localization of glucose transporter, and enhanced by 25% cellular glucose uptake when compared to MIF alone. In mouse hearts perfused with MIF20 prior to no-flow ischemia and reperfusion, post-ischemic left ventricular function improved commensurately with an increase in cardiac MIF-AMPK activation and an augmentation in myocardial glucose uptake. By contrast, small molecule MIF agonism was not effective in cells or tissues genetically deficient in MIF or the MIF receptor, verifying the specificity of MIF20 for MIF-dependent AMPK signaling. The protective effect of MIF20 also was evident in an in vivo regional ischemia model. Mice pre-treated with MIF20 followed by left coronary artery occlusion and reperfusion showed a significant reduction in infarcted myocardium. Conclusions: These data support the pharmacologic utility of small molecule MIF agonists in enhancing AMPK activation in cardiac ischemic injury. Supported by: Bristol-Myers Squibb/AstraZeneca & ULRICH WERNER, K. HISS, DANIEL HEIN, THOMAS HÜBSCHLE, PAULUS WOHLFART, HARTMUT RUETTEN, Frankfurt, Germany Cardioprotection by the GLP-1 receptor agonist lixisenatide (LIXI) has recently been described in ex-vivo and in-vivo ischemia/reperfusion rat studies. Here we assessed the cardioprotective activity of LIXI in a more severe mouse model of myocardial infarction (MI), induced by permanent ligation of the left anterior descending coronary artery. The protective effects of once-daily treatment with LIXI for 4 weeks were compared with placebo and with sham-operated controls without MI. The ACE inhibitor ramipril served as reference. Treatment was started 3 days prior to induction of ischemia. Male C57BL/6 mice (6-7 weeks of age) were randomized into 5 groups: Sham, MI-control (placebo injection QD); MI-LIXI 10 (10 µg/kg s.c. QD), MI-LIXI 150 (150 µg/kg s.c. QD) and MI-ramipril (2.5 mg/ kg*day in drinking water). Post MI survival was monitored throughout the study and cardiac function was assessed by pressure-volume loop analysis at study end. Four weeks after infarct induction, MI-control animals showed dramatically reduced global cardiac parameters with significantly depressed systolic contractility and impaired diastolic function. This was reflected in significantly increased mortality (survival <60%). Chronic daily injection with LIXI (both 10 and 150 µg/kg) resulted in significant (40-47%) improvements in heart function Supported by: NIH & 1189-P Effects of Lixisenatide on Survival and Cardiac Function in a Mouse Model of Chronic Myocardial Infarction 1188-P Dapagliflozin Effects on the Lipid Profile of Patients With Type 2 Diabetes Mellitus ELISE HARDY, AGATA PTASZYNSKA, TJERK W.A. DE BRUIN, EVA JOHNSSON, SHAMIK J. PARIKH, JAMES F. LIST, Wilmington, DE, Princeton, NJ Dapagliflozin (DAPA), an SGLT2 inhibitor, increases urinary glucose excretion, and improves hyperglycemia in patients with type 2 diabetes mellitus (T2DM). DAPA also favorably affects cardiovascular risk factors by reducing weight, blood pressure and uric acid in these patients. The effects & For author disclosure information, see page 829. A310 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS Figure 1. parameters, such as ejection fraction, stroke volume and cardiac output, translating into a significantly increased survival rate of >70%. Ramipril also improved global cardiac parameters but the effect was not superior to that of LIXI. The survival rate with ramipril, however, was ~90%. In conclusion, the cardioprotection following ischemia/reperfusion in the rat which was demonstrated by LIXI in previous studies, could be confirmed and even extended towards a more aggravated MI induced heart failure mouse model with permanent coronary artery occlusion. Supported by: Sanofi & 1190-P Sustained Improvement of Postprandial Endothelial Function in Type 2 Diabetes With Exenatide Supported by: Novartis Pharmaceuticals Corporation; National Center for Research (UL1RR025758) & HIROYOSHI YOKOI, MASATO NAKAMURA, TOSHIYA MURAMATSU, HISAYUKI OKADA, MASAHIKO OCHIAI, SATORU SUWA, YUTAKA MATSUYAMA, SHINSUKE NANTO, ON BEHALF OF THE J-DESSERT INVESTIGATORS, Kitakyushu, Japan, Tokyo, Japan, Yokohama, Japan, Hamamatsu, Japan, Izunokuni, Japan, Suita, Japan Pioglitazone (Pio) is widely used for glycemic control in patients with type2 diabetes mellitus (DM), and is associated with a lower risk of cardiovascular events according to a meta-analysis of randomized trials. To evaluate the effect of Pio on ischemic cardiovascular events in Japanese patients with DM and coronary artery disease, after drug-eluting stent (DES) implantation, the data from the Japan-Drug Eluting Stents Evaluation; a Randomized Trial (J-DESsERT) was analyzed. In this prospective, multicenter trial, 3,533 patients were randomized 1:1 to undergo coronary stenting with Sirolimus-eluting stents or Paclitaxel-eluting stents. Lesion lengths were ≤46 mm, with vessel diameters from ≥2.5 mm to ≤3.75 mm. Randomization was stratified based on the presence or absence of DM. Definitions for allocation into the DM group at the time of this trial were: 1. Previous DM diagnosis; 2. Currently on diabetic medication (oral hypoglycemic drugs or insulin injections); 3. HbA1c ≥ 6.5% within 30 days before the procedure. Patients who met one or more of the above criteria were allocated to the DM group. A total of 1,692 patients (48%) with DM were analyzed from the J-DESsERT trial. Cardiac death, myocardial infarction, or stroke, occurred in 11 of 403 patients (2.7%) receiving Pio and 71 of 1,289 patients (5.5%) receiving the other therapy, without Pio, at 12 months (P=0.024). Pioglitazone is associated with a significantly lower risk of death, myocardial infarction, or stroke in Japanese patients with DM after DES implantation. Supported by: Amylin Pharmaceuticals, LLC.; Eli Lilly and Company & 1191-P Aliskiren Improves Vascular Smooth Muscle Function in the Skin Microcirculation of Type 2 Diabetic Patients With Normal Renal Function JODY DUSHAY, FRANCESCO TECILAZICH, ANTONIOS KAFANAS, MARY MAGARGEE, MICHAEL E. AUSTER, CHARALAMBOS GNARDELLIS, THANH DINH, ARISTIDIS VEVES, Boston, MA, Messolonghi, Greece & & 1193-P Bromocriptine-QR Attenuates the Progression to Both Cardiovascular (CV) Averse Events and Worsening Hyperglycemia in Type 2 Diabetes Mellitus (T2DM) Subjects in Good Glycemic Control Skin microcirculation endothelium dependent and independent vasodilation are impaired in T2DM and pre-diabetes. We studied the vascular effects of daily 150 mg aliskiren, a direct renin inhibitor, in a 12-week randomized, double-blind, placebo-controlled trial in T2DM and at-risk individuals (1st degree relative with DM or history of impaired glucose homeostasis or gestational DM). Forty seven T2DM subjects without vascular complications [age 58±12 years (mean ± sd), A1c 7.3%±1.2) and 41 at-risk (age 52±11, A1c 5.8±0.3) were enrolled. Twenty six (55%) T2DM and 4 (8%) at-risk subjects were also treated with ACE inhibitors or Angiotensin II Receptor Blockers. Aliskiren resulted in lower systolic (p <0.05) and diastolic (p <0.01) blood pressure and improved forearm skin endothelium-independent vasodilation in T2DM subjects (p<0.05, Figure 1). No additional effects of aliskiren were seen in serum or tissue (skin biopsy) markers of inflammation or in the macrocirculation in either study group. Importantly, there was no hypotension or change in potassium or creatinine levels with aliskiren treatment. We conclude that aliskiren is well tolerated and improves blood pressure and vascular smooth muscle function in skin microcirculation of T2DM patients with normal renal function. ADA-Funded Research 1192-P Impact of Pioglitazone on Cardiovascular Events in Patients With Type-2 Diabetes Mellitus after Drug-Eluting Stent Implantation ANTHONY H. CINCOTTA, RICHARD E. SCRANTON, MICHAEL EZROKHI, J. MICHAEL GAZIANO, Tiverton, RI, Boston, MA The one-year Cycloset Safety Trial (CST) demonstrated the CV safety and anti-diabetes efficacy of bromocriptine-QR (B-QR) versus placebo in a large (3070) randomized population of T2DM subjects whose hyperglycemia was treated with diet, or one or two anti-diabetes medications (including insulin) (Diabetes Care. 2010 Jul;33(7):1503-8, Endocr Pract. 2012 Nov 1;18(6):93143). The present study however tested whether B-QR may retard both CV and metabolic disease progression in T2DM subjects already in good metabolic control (i.e., that had a baseline HbA1c of ≤ 7.0) derived from the CST (BQR N= 1219, placebo N= 615, mean HbA1c: each 6.3). A Cox proportionalhazards regression analysis was used to determine the frequency of major CV events, pre-specified as a composite of first myocardial infarction, stroke, coronary revascularization, or hospitalization for angina or congestive heart failure that occurred after randomization. Both on-treatment (OT) (exposure time restricted to time on study drug) and intention to treat (ITT) analyses (exposure time equal to early termination, time to event, or one year) were conducted. The impact of B-QR on loss of good glycemic control was For author disclosure information, see page 829. Guided Audio Tour poster A311 POSTERS We previously demonstrated that exenatide (Ex) improved endothelial function (EF) and plasma triglycerides (TG) 3.5 hours after a single, fatenriched breakfast meal in subjects with impaired glucose tolerance or new onset diet controlled type 2 diabetes. In this study, we tested whether the vascular effect of Ex is sustained through multiple meals and whether it is effective in those with type 2 diabetes of longer duration. Participants with diagnosed type 2 diabetes of duration < 3 years (n=19) and > 5 years (n=21) were randomized to receive twice a day for 10 days in a cross-over fashion Ex (Byetta®; dose per injection: 5 µg for days 1-5, 10 µg for days 6-10) or placebo (Pl) . On day 11 after the morning injection, participants received fat-enriched breakfast and lunch meals (400 kcal/m2 per meal, 45% fat, 40% carbohydrate, 15% protein) separated by 4 hours. EF as reactive hyperemia index (RHI) by peripheral arterial tonometry, plasma glucose, insulin and lipid concentrations were measured before, and every 2 hours after the study drug, for a total 8 hours. 36 participants completed both study periods. Over 8 hours, exenatide increased RHI (mean percent change in area under the curve ± SD: 10 ± 18, p=0.008), decreased glucose and TG concentrations (by 16 ± 13 % and 23 ± 21 %, respectively, both p<0.0001), and had no effect on insulin or total, HDL and LDL cholesterol concentrations. Augmentation of RHI with Ex was similar in those with diabetes duration > 5 years and < 3 years (12 ± 19 % and 10 ± 17 %, respectively, p=0.8) and remained significant after adjustment for glucose (p=0.006) and TG (p=0.04) concentrations. In conclusion, exenatide administration was followed by sustained improvement of endothelial function over 8-hour period including fatenriched breakfast and lunch meals. The favorable effect of exenatide on endothelial function was observed regardless of diabetes duration and was only partially accounted for by sustained reductions in postprandial triglycerides. Clinical Diabetes/ Therapeutics JURAJ KOSKA, JAMES LIU, KALYANI RARAVIKAR, DAWN C. SCHWENKE, ERIC A. SCHWARTZ, PETER D. REAVEN, Phoenix, AZ CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS demonstrate any association between insulin sensitivity and these muscle transcripts, and protein abundance of SDHA (complex II) and UQCRC1 (complex III), and Thr172 phosphorylation of AMPK were unaffected by testosterone therapy. In conclusion, we show that in elderly men with subnormal testosterone levels, insulin sensitivity is not associated with expression of genes involved in oxidative metabolism, and the beneficial effect of testosterone treatment on lipid oxidation is not caused by a general increase in the abundance of genes and enzymes involved in OxPhos and lipid metabolism in skeletal muscle. analyzed by assessing among those study subjects not having a change in concomitant anti-diabetes medication (B-QR N=586, placebo N=328), the percent whose HbA1c level increased above 7.0 after 52 weeks of therapy (B-QR N=104, placebo N=84) via Fisher’s Exact Test. B-QR treatment reduced the time to first CV event by 50% (ITT) (HR: 0.51; CI 0.28 to 0.96,) and by 52% (OT) (HR: 0.48; CI 0.24 to 0.95). B-QR reduced the percent of subjects with loss of good glycemic control by 31% (from 26% to 18%, P=0.006) over this one-year study period. These findings suggest that B-QR therapy to T2DM subjects in good glycemic control may offer a possible means of retarding progression of both cardiovascular and metabolic disease among such T2DM subjects. 1194-P 1196-P Prior Treatment With Dipeptidyl-Peptidase IV Inhibitors Is Associated With Favorable Outcome in Patients With Acute Ischemic Stroke Sodium Glucose Cotransport-2 Inhibition With Empagliflozin Attenuates Renal Hyperfiltration in Patients With Type 1 Diabetes DAVID Z. CHERNEY, BRUCE A. PERKINS, NIMA SOLEYMANLOU, VESTA LAI, MARIA MAIONE, ALANA LEE, NORA FAGAN, HANS-JUERGEN WOERLE, ODD ERIK JOHANSEN, ULI C. BROEDL, MAXIMILIAN VON EYNATTEN, Toronto, ON, Canada, Burlington, ON, Canada, Ridgefield, CT, Ingelheim, Germany KONSTANTINOS TZIOMALOS, STELLA BOUZIANA, ATHINODOROS PAVLIDIS, MARIANNA SPANOU, VASILIOS GIAMPATZIS, MARIA PAPADOPOULOU, FOTINI GIANNAKIDOU, FOTIOS ILIADIS, TRIANTAFILLOS DIDANGELOS, CHRISTOS SAVOPOULOS, APOSTOLOS HATZITOLIOS, Thessaloniki, Greece POSTERS Clinical Diabetes/ Therapeutics & Supported by: Novo Nordisk, Inc. Renal hyperfiltration promotes the development of diabetic nephropathy and has been attributed to both hemodynamic and tubular mechanisms. Our primary objective was to determine the impact of 8 weeks of sodium glucose co-transport 2 inhibition with empagliflozin 25 mg QD on renal hyperfiltration in subjects with type 1 diabetes (T1D) (NCT01392560). Inulin (GFR) and paraaminohippurate (effective renal plasma flow [ERPF]) clearances were used in patients stratified based on having either hyperfiltration (T1D-H, GFR ≥ 135 ml/min/1.73m2, n=27) or normal GFR (T1D-N, n=13) at baseline during clamped euglycemia. Blood pressure, renal function, circulating levels of aldosterone, angiotensin II and nitric oxide (NO) were measured under clamped euglycemic (4-6 mmol/L) and hyperglycemic (9-11 mmol/L) conditions at baseline and again after empagliflozin. During clamped euglycemia, hyperfiltration was attenuated with empagliflozin in T1D-H (GFR 172±23 to 139±25 ml/min/1.73 m2, p<0.0001). This effect was accompanied by declines in plasma NO (p=0.0016), ERPF and renal blood flow and an increase in renal vascular resistance (p<0.0001). Results were similar under clamped hyperglycemic conditions. Systolic blood pressure declined with empagliflozin in T1D-H during clamped euglycemia (-3.1±8 mmHg, p=0.0495), despite a modest rise in circulating aldosterone (p=0.0189) and angiotensin II (p=0.0439). In T1D-N, renal function, blood pressure and NO remained unchanged. In summary, empagliflozin attenuated renal hyperfiltration in T1D-H but did not influence renal function in T1D-N patients. Our findings suggest for the first time that the effect of SGLT2 inhibition on tubuloglomerular feedback may play a favourable role against the pathogenesis of renal hyperfiltration in patients with T1D. We aimed to assess the effects of prior antidiabetic treatment on the severity and outcome of acute ischemic stroke (IS). We prospectively studied 378 consecutive patients (38.9% males, age 78.8±6.5 years) who were admitted for IS. Stroke severity was evaluated with the National Institute of Health stroke scale (NIHSS) at admission and the outcome was assessed with in-hospital mortality and with the modified Rankin scale (RSS) at discharge. Type 2 diabetes mellitus (T2DM) was present in 32.3% of the patients. Prior to stroke, antidiabetic treatment included metformin, sulphonylureas (SU), insulin and dipeptidyl-peptidase IV (DPP-IV) inhibitors, alone or in combination, in 60.0, 46.0, 24.0 and 27.0% of patients, respectively. Stroke severity and outcome did not differ between patients who were on monotherapy with metformin, SU or insulin. Patients who were treated prior to stroke with DPP-IV inhibitors, alone or in combination with other antidiabetic agents, showed a trend for less severe stroke compared with patients who were not receiving DPP-IV inhibitors (NIHSS score at admission 6.1±7.5 vs. 10.0±9.2, respectively; p=0.079) and had lower inhospital mortality (0.0% vs. 15.1%, respectively; p<0.05) and lower RSS at discharge (2.1±1.9 vs. 3.2±2.1, respectively; p<0.05), even though they had a higher prevalence of hypertension (100.0% vs. 84.9%, respectively; p<0.05). Other cardiovascular risk factors did not differ between the two groups. Stroke severity and outcome did not differ between patients who were treated prior to stroke with metformin, SU or insulin, alone or in combination with other antidiabetic agents, and patients who were not receiving the respective antidiabetic agent. In conclusion, in patients admitted for IS, prior treatment with DPP-IV inhibitors appears to be associated with more favorable in-hospital outcome and with less severe stroke compared with prior treatment with other antidiabetic agents. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1197-P Pharmacological Prevention of Diet Induced Metabolic Dysfunction by the SIRT1 Activator SRT3025 1195-P VIPIN SURI, MEGHAN L. DAVIS, QING NIE, ELDEN LAINEZ, SHI LIANG, YONG QI, JAMES L. ELLIS, GEORGE P. VLASUK, Cambridge, MA The Effect of Androgen Therapy on Genes and Proteins Involved in Oxidative Phosphorylation and Lipid Metabolism in Skeletal Muscle of Men With Subnormal Testosterone Levels The NAD+-dependent protein deacetylase SIRT1 has been suggested to be involved in the regulation of a number of metabolic pathways. Transgenic overexpression of SIRT1 confers resistance to the metabolic dysfunction induced by a high fat diet (HFD). We investigated the effect of pharmacological SIRT1 activation on prevention of HFD-induced metabolic dysfunction in mice. We have previously shown that the small molecule SIRT1 activator SRT3025 reverses high fat diet induced obesity, insulin resistance and dyslipidemia. To explore the preventative effect of SRT3025, we placed 6 week old C57/BL6 mice on HFD or HFD supplemented with SRT3025 (HFD3025) for 15 weeks. Despite similar food intake, the HFD-3025 group was completely protected from the rapid weight gain observed in the HFD group and was indistinguishable from the chow-fed animals (Figure 1). HFD-3025 also prevented development of hyperinsulinemia (-92%), hyperglycemia (-27%), glucose intolerance (-36% AUC, intraperitoneal glucose tolerance test) and adiposity (-62% fat mass) compared to the HFD group. Replacement of the HFD-3025 diet with HFD resulted in an increased rate of weight gain similar to the HFD group. Our data clearly demonstrates that in addition to reversing the metabolic dysfunction induced by HFD, SRT3025 can effectively prevent the development of metabolic dysfunction in mice fed a HFD and has the potential to be clinically beneficial in Obesity and Type 2 Diabetes. STINE J. PETERSSON, LOUISE FREDERIKSEN, JONAS M. KRISTENSEN, RIKKE K. HENRIKSEN, MARIANNE ANDERSEN, KURT HØJLUND, Odense, Denmark Recent studies have reported that low serum testosterone in men is associated with low insulin sensitivity and reduced expression of oxidative phosphorylation (OxPhos) genes, and that treatment with testosterone increases lipid oxidation. Here, we aimed to investigate the effect of testosterone therapy on genes and proteins involved in OxPhos and lipid metabolism in skeletal muscle in vivo in men with low serum testosterone. In 25 elderly men with subnormal serum levels of testosterone, skeletal muscle biopsies were taken before and after treatment with either testosterone gel (n=12) or placebo (n=13) for 6 months. Insulin sensitivity and substrate oxidation were assessed by euglycemic hyperinsulinemic clamps combined with indirect calorimetry. Muscle mRNA levels and protein abundance of a selected set of enzymes involved in OxPhos and lipid metabolism and phosphorylation of AMPK were examined by qRT-PCR and western blotting. Despite a placebo-controlled increase in lipid oxidation (p<0.05), testosterone therapy had no effect on insulin sensitivity or muscle transcript levels of genes involved in OxPhos (NDUFS1, ETFA, SDHA, UQCRC1, COX5B, PPARGC1A), lipid oxidation (ACADVL, CD36, CPT1B, HADH, PDK4) or adiponectin-AMPK signaling (ADIPOR1, ADIPOR2, PRKAA2, PRKAG3). Moreover, we could not & For author disclosure information, see page 829. A312 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS into an ApoE -/- background. Diabetes was produced by streptozotocin at 3 months of age. Vitamin E or placebo was administered in the drinking water beginning one month after the onset of DM. Plaque size and progression were assessed repetitively in vivo using a Vevo-2100 high frequency microultrasound system (Visualsonics) in the brachiocephalic artery at multiple time points. We observed markedly larger plaques in Hp 2-2 DM mice at all time points as compared to Hp 1-1 DM mice and mice without DM . Furthermore, vitamin E completely prevented progression of plaques in Hp 2-2 DM mice. Vitamin E can prevent plaque progression in Hp 2-2 DM mice recapitulating the ability of vitamin E to provide cardiovascular benefit to Hp 2-2 DM humans. Supported by: NIH (R0IDK085226) 1200-P OLEG TSUPRYKOV, MARKUS L. ALTER, KAROLINE VON WEBSKY, LYUBOV CHAYKOVSKA, CHRISTOPH REICHETZEDER, VIKTORIIA ANTONENKO, JAN RAHNENFÜHRER, THOMAS KLEIN, BERTHOLD HOCHER, Potsdam, Germany, Berlin, Germany, Zürich, Switzerland, Biberach, Germany IRENE G. MANDERS, MICHAELA DIAMANT, KATRIN STOECKLEIN, CAROLINE H.C. LUBACH, JUDITH BIJL-OELDRICH, PRABATH NANAYAKKARA, JAN A. RAUWERDA, MARK H.H. KRAMER, ELISABETH M.W. EEKHOFF, Amsterdam, Netherlands DPP-4 inhibitors may have kidney-protective properties independent of glucose control. We compared the effect of linagliptin with an angiotensin II receptor antagonist - the current clinical gold standard - on preventing renal disease progression in a nondiabetic rat model of chronic renal failure. Male Wistar rats were allocated to 4 groups: sham operated; 5/6 nephrectomy (5/6 Nx) plus placebo treatment (placebo); 5/6 Nx plus linagliptin (LIN; 0.083 mg/kg/d in chow); 5/6 Nx plus telmisartan (TEL; 5 mg/kg/d in drinking water). Study duration was 130 days; blood pressure and albumin excretion were assessed repeatedly. Kidneys and plasma biomarkers were analyzed at study end. Interstitial fibrosis increased by 69% in placebo rats vs SHAM rats (p<0.05), and decreased by 48% with LIN (p<0.05) and 24% with TEL (p=ns) vs placebo. Glomerular size increased by 28% in placebo vs SHAM rats (p<0.01), and decreased by 18% (p<0.001) with LIN but not significantly with TEL vs placebo rats. The glomerulosclerosis index was significantly increased in placebo rats vs SHAM rats. There was a trend towards decreased glomerulosclerosis with LIN and TEL. Analysis of collagen type I and III mRNA and protein levels confirmed histopathologic findings. The urinary albumin/creatinine ratio increased 14-fold in placebo rats vs SHAM rats (p<0.001), and decreased by 66% with LIN (p<0.05) and 92% with TEL (p<0.01) vs placebo rats. Blood pressure was lowered by TEL (31 mmHg; p<0.05) and unaffected by LIN. TIMP-1, calbindin, osteopontin and β2 microglobulin (B2M) were significantly increased in the placebo rats vs SHAM rats. LIN decreased plasma levels of TIMP-1, calbindin, osteopontin and B2M vs placebo rats (all p<0.05), whereas TEL significantly decreased osteopontin and TGF-β expression. In summary, LIN is as effective as TEL in preventing renal disease progression in rats with 5/6 nephrectomy. Furthermore, the underlying molecular mechanisms appear to be different. Optimal glycemic control in hospitalized patients with diabetes mellitus in non-Intensive Care Units (non-ICU) may improve outcome and shorten admission duration. However, achieving these targets is difficult and development and implementation of appropriate protocols remain major challenges. A Nurse-driven Diabetes In-hospital Treatment protocol (N-DIABIT) was developed, i.e. a sliding-scale protocol carried out by trained ward nurses, supervised by registered diabetes nurses. We investigated the feasibility, safety and efficacy of the N-DIABIT in patients with diabetes mellitus, admitted to non-ICU. We prospectively collected data of 210 patients with diabetes admitted in the five-month period after N-DIABIT introduction (intervention group) and compared these to a historical control group, consisting of 200 patients admitted in the five-month period before starting N-DIABIT. Additional analyses were repeated in patients in whom protocol adherence was ≥ 70% with respect to blood glucose measurements during admission (intervention and matched control subgroup; n = 240). Mean protocol adherence was >75%. Only the intervention- versus control subgroups showed significant lower mean blood glucose levels (P = 0.044), significantly less consecutive hyperglycemic episodes (≥10.0 mmol/L; P = 0.049) and a numerically shorter admission duration (mean 1.3 days shorter). Hypoglycemia incidence (<4.0 mmol/L) was similar before and after protocol implementation. Implementation of the N-DIABIT by trained ward nurses in non-ICU hospital diabetes care is feasible, effective and safe in improving glycemic control and in case of appropriate protocol adherence, may ultimately reduce admission duration. Supported by: Boehringer Ingelheim Pharmaceuticals, Inc. 1201-P Clinical Outcomes Following Discharge from a Pharmacist-Led Diabetes Intense Medical Management Clinic 1199-P Vitamin E Arrests Atherosclerotic Plaque Progression in Haptoglobin 2-2 Diabetic Mice CANDIS M. MORELLO, ANDREA BECHTOLD, JAN D. HIRSCH, La Jolla, CA A collaborative pharmacist-endocrinologist Diabetes Intense Medical Management (DIMM) Clinic (held ½ day per week) was developed at the Veterans Affairs San Diego Healthcare System to help primary care providers meet diabetes specific compliance measures and help patients achieve metabolic goals. Referred type 2 diabetes patients with A1C > 9% were treated by a pharmacist-CDE who combined clinical care with patient-specific diabetes education. Outcomes show mean A1C was reduced significantly from 10.2% at enrollment to 8.8% at 3 months and 8.0% at 6 months (p<0.001). Percent of patients achieving goals was significantly increased for Triglycerides (TG) (3 and 6 months) and Fasting Plasma Glucose (FPG) (6 months) compared to baseline (p<0.05). After achieving A1C goals, patients were discharged back to their primary care provider. The primary objective was to compare A1C at 6 and 12 months post discharge to A1C at discharge from the DIMM Clinic. Secondary outcomes were also compared [i.e. mean FPG, weight, BMI, cholesterol (LDL, HDL), TG, and blood pressure (BP)]. Thirty-one patients have been discharged from the DIMM Clinic after meeting metabolic goals. Time to discharge for these patients was ≤6 months (44%), ≤9 months (31%) and >9 months (25%). Follow up data was available for 22 patients at 6 and 12 months post-discharge. HILLA LEE VIENER, RACHEL LOTAN, ROSTIC GORBATOV, NINA S. LEVY, ANDREW P. LEVY, Haifa, Israel Nine independent longitudinal studies have demonstrated that individuals with the Hp 2-2 genotype and Diabetes Mellitus (DM), which constitutes approximately 40% of all individuals with DM, have a 2-3 fold increased incidence of atherosclerotic cardiovascular disease (CVD) and its sequelae as compared to DM individuals without the Hp 2-2 genotype. The mechanism responsible for this interaction between the Hp genotype and CVD in DM appears to be due to oxidative modification of HDL inHp 2-2 DM. Proof of concept for this hypothesis has been that antioxidant supplementation with vitamin E can restore HDL function in Hp 2-2 DM and vitamin E has been shown in two trials to significantly reduce MI and CVD death in Hp 2-2 DM individuals. We sought to determine if there was an interaction between atherosclerotic plaque progression and the Hp genotype that could be mitigated by vitamin E in mice. The Hp 2 allele, which is present only in man, is defined by the presence of a 1.7kb in-frame duplication of exons 3 and 4 of the Hp 1 allele. We engineered a murine Hp 2 allele and introduced it into the murine Hp locus by homologous recombination. We backcrossed Hp 2 and wild type (Hp 1) mice ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A313 POSTERS 1198-P Adherence to a Nurse-Driven Sliding Scale Protocol Improves Glycemic Control and Shortens Admission Duration in Hospitalized Non-ICU Patients With Diabetes Mellitus Clinical Diabetes/ Therapeutics Linagliptin Is as Efficacious as Telmisartan in Preventing Renal Disease Progression in Rats With 5/6 Nephrectomy CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS Mean age was 60±8 years. Mean duration of diabetes was 16±9 years and 100% of patients were male. A1C at 6 months post-discharge was 7.6±1.6% compared to 7.1±0.7% at discharge (p=0.20). At 12 months post-discharge A1C was 7.5±1.2% (p=0.34). Systolic BP was higher at 6 months (138±25 mmHg) and 12 months post-discharge (134±12 mmHg) versus 122±11 mmHg at discharge (p=0.01). Other secondary outcomes were not significantly different than at discharge. These results show significant A1C reductions that had been achieved by patients attending a pharmacist-led DIMM Clinic were maintained at 6 and 12 months after discharge back to their primary care provider. 1202-P Effects of Herbal Fomula AMP-1915 on Metabolic Syndrome POSTERS Clinical Diabetes/ Therapeutics SHENGJIE FAN, GAIGAI LIU, LU GUO, LIGANG ZHOU, CHENG HUANG, Shanghai, China Supported by: Research Ministry of Romania (52164/2008) Herbal formulae have been used in Traditional Chinese Medicine (TCM) widely for thousands years. Huangqisan (AMP-1915), consistent of Astragalus membranaceus (Fisch.) Bge, Cortex Mori radices and Pueraria lobata (Willd.) Ohwi, was recorded in 1117 AD firstly for the treatment of diabetes. However, there is no experimental evidence to support the therapeutic effects of the formula for diabetes. Ethanol extract of AMP-1915 was prepared from Astragalus membranaceus (Fisch.) Bge, Cortex Mori radices and Pueraria lobata (Willd.) Ohwi and HPLC study was carried out to analyze the major chemical components of the extract. High-fat diet-fed C57BL/6 obese (DIO) mice and db/db mice were treated with AMP-1915 and the fasting blood glucose, glucose tolerance test, serum lipid profile and the livers lipid contents were analyzed. Gene expression assay was performed with a real time RT-PCR and RNA-seq, and protein levels were determined by Western blot. Our results showed that eight monomer components were detected in AMP1915 by HPLC assay, including Astragaloside IV, Morin, Daidzein, etc.. AMP1915 (2% and 5%) significantly reduced fasting blood glucose level, improved glucose tolerance test and serum lipid profiles in the DIO mice. In db/db mice, AMP-1915 lowered fasting blood glucose and serum total choleterol content. Real time RT-PCR assay showed that AMP-1915 upregulated the expression of PPARγ and its target genes in the DIO mouse liver. Western blot assay revealed that AMP-1915 increased PPAR γ and phosphorylation of AMPK in HepG2 cells. The results indicate that AMP-1915 could lower blood glucose levels and improve serum lipid profiles both in DIO mice and db/db mice through the enhancement of PPARγ and AMPK signaling. Our data suggest that the herbal formulae may be used as therapeutic drug for the patients mixed with hyperglycemia and hyperlipidemia. 1204-P Comparative Study of Azelnidipine With Trichlormethiazide in Japanese Type 2 Diabetic Patients With Hypertension: The COAT Randomized Controlled Trial MASAHIRO TAKIHATA, AKINOBU NAKAMURA, HIROSHI KAMIYAMA, YUKO DOBASHI, TAKASHI MIYAZAKI, HARUKA TAMURA, RIKA SAKAMOTO, MINORI MATSUURA, YOSHINOBU KONDO, SATSUKI KAWASAKI, MARI KIMURA, YASUO TERAUCHI, Yokohama, Japan, Chigasaki, Kanagawa, Japan, Fujisawa, Japan Few randomized trials that compare calcium antagonists with diuretics in type 2 diabetic patients with inadequately controlled hypertension with angiotensin II receptor blocker have been published. The aim of this study is to compare the efficacy and safety of these two agents and the impact on surrogate markers related to diabetic and hypertensive complications. In a multicenter, open-label trial, 240 patients with adequately controlled diabetes (HbA1c ≤ 6.5%) and inadequately controlled hypertension (systolic blood pressure [sBP] ≥ 130 mmHg or diastolic blood pressure [dBP] ≥ 80 mmHg) with olmesartan were randomly assigned to a azelnidipine group (16 mg/day) or a trichlormethiazide group (1 mg/day) and were followed up for 48 weeks. Main outcome measure was the difference in the changes in HbA1c level from the baseline at 48 weeks between these two groups. Of the 240 subjects enrolled, 209 subjects (azelnidipine group: 103 patients, trichlormethiazide group: 106 patients) completed this trial. At 0 week, mean HbA1c level, sBP, and dBP were 6.26 ± 0.62% vs. 6.23 ± 0.60%, 142 ± 11 mmHg vs. 141 ± 11 mmHg, and 80 ± 6 mmHg vs. 79 ± 7 mmHg. At 48 weeks, the changes in HbA1c level, sBP, dBP, estimated glomerular filtration rate, and albumin-to-creatinine ratio from the baseline were 0.19 ± 0.52% vs. 0.19 ± 0.54% (n.s.), -10.7 ± 9.6 mmHg vs. -7.1 ± 7.7 mmHg (P < 0.001), -6.6 ± 6.6 mmHg vs. -3.3 ± 6.1 mmHg (P < 0.001), -2.6 ± 11.9 mL/min vs. -3.8 ± 9.2 mL/ min (n.s.), and -12 ± 105 mg/g Cr vs. -35 ± 94 mg/g Cr (P = 0.041). Dizziness (12 patients, 11.7% vs. 16 patients, 15.1%), and edema (16 patients, 15.5% vs. 7 patients, 6.6%, P = 0.047) were observed for 48 weeks. In conclusion, azelnidipine was more effective for BP control than trichlormethiazide, and trichlormethiazide was more effective for reduction of albuminuria than azelnidipine. Both of them, however, similarly exacerbated glycemic control in type 2 diabetic patients with hypertension. 1203-P The Influence of Thiamine Hydrochloride on Kidney Disease in a Model of Wistar Diabetic Rat CIPRIAN CONSTANTIN, AURELIAN RANETTI, GEORGIANA CONSTANTIN, CRISTIAN SERAFINCEANU, DAN CHETA, Bucharest, Romania Previous experiments on streptozotocined Wistar rats, including our own, showed kidney diabetic damage even if model of type 2 diabetes is with a good metabolic control. On the other hand, thiamine compounds are known for its kidney protective effects. In order to quantify the kidney changes and study the possible beneficial effect of a thiamine compound, 48 Wistar Rats are used to obtain a model of type 2 diabetes using streptozotocin. 3 Groups were matched for sex and metabolic control. For a period of 180 days 0g/l, 1g/l, 2g/l thiamine hydrochloride solution was administered daily to each rat in the intervention groups (A,B,C), using the water as a support. Weight, glucosuria and glycemia were determined periodically. At the end of the study period, rats were sacrificed. Liver, kidneys and heart were the organs investigated using optical microscopy. The most interesting findings were the kidney structural changes. 3rats (18.75%) in Agroup, 4rats (25%) in Bgroup and 7rats (43,75%) in Cgroup had a normal kidney structure (p<0.5). The rest had various degrees of structural alteration ranging from vacuolar dystrophy to polymorphic inflammatory infiltrate. The frequency of each finding is shown in the figure. Our data show that, even in pretty well controlled model of type 2 diabetes, kidney disease is very frequent in initial stages of disease. More data were collected to highlight the protective effects of this thiamine compound has on the kidney. & For author disclosure information, see page 829. A314 Guided Audio Tour poster ADA-Funded Research CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGIC TREATMENT OF COMPLICATIONS at baseline (8.9 vs 7.9%) and at time of insulin initiation or event in non-users (9.0 vs 7.4%) and experienced a longer time period without complications (40.8 vs 21.6 months). This analysis demonstrates that despite having a higher A1C at baseline, those initiating insulin during follow-up spent a significantly longer time period free of microvascular complications compared to those not initiating insulin. Approaches to facilitate early insulinization should be identified. 1205-P Impact of Ramadan Fasting on HbA1c in Relation to Medication Use MELANIE SIAW, NUR HIDAYAH SHAMSURI, DANIEL CHEW, RINKOO DALAN, SHAIKH ABDUL ABDUL SHAKOOR, NOORANI OTHMAN, THERESA CHOO, SITI NURHANNAH ABDUL KARIM, OLIVE LAI, JOYCE Y. LEE, Singapore, Singapore Muslims worldwide observe religious fasting of Ramadan from dawn to sunset. While the practice of fasting often affects the blood glucose of patients with diabetes as reported in the literature, little is known about the use of pharmacologic agents in relation to the degree of changes in HbA1c during this period. In this prospective study, we aimed to examine the change of HbA1c during Ramadan in relation to pharmacotherapy use by tracking HbA1c values and medication use before, during and after Ramadan for a total of 13 months. A total of 136 Muslim patients were eligible for this study. The mean age was 56.4 ± 8.9 years with 40.4% male and 59.6% female. Of these, 63.2% were on insulin containing therapies, 34.6% were on oral hypoglycemic agents (OHA) only and 2.2% were not on medication. During Ramadan, mean HbA1c improved by 0.66% and 0.27% for patients who were on OHA only and insulin containing regimen, respectively. After adjusting for demographics, health status and baseline HbA1c, patients who were on OHA only were 5.9 times more likely to observe improvement in HbA1c during Ramadan compared to those given insulin containing regimen (p<0.05). Furthermore, there were no significant hypoglycemic events related to fasting during Ramadan. Overall, the improvement in HbA1c during Ramadan appeared to be more prominent in Muslim patients who take oral antidiabetic medications without insulin. Supported by: sanofi-aventis 1207-P New Inhibitors to Preserve and Protect Islet Function in T1DM Supported by: CDMRP (PR093521) 1208-P Derivatives of 2-Aminoimidazole from Sea Sponges have Potent Anti-Glycating Activity RANDALL J. BASARABA, DAVID ACKART, BRENDAN PODELL, JESSICA HAUGEN, ROBERTA WORTHINGTON, CHRISTIAN MELANDER, Fort Collins, CO, Raleigh, NC Currently there are no approved drugs for humans that inhibit the formation and accumulation of advanced glycation end products (AGEs). AGEs accumulate as a consequence of uncontrolled diabetes, aging and other chronic inflammatory diseases. In diabetes the combination of high blood glucose levels combined with oxidative stress accelerates AGE formation. This is particularly true when diabetes is combined with infectious or noninfectious causes of myeloid cell-mediated inflammation. We have generated a library of over 700 small molecular weight compounds based on the 2-aminoimidazole (2-AI) subunit from the naturally occurring sea sponge-derived metabolites oroidin and bromoageliferin, some of which have potent anti-glycating activity in vitro. We used a high throughput in vitro fluorescence assay to screen the 2-AI library for compounds that both prevented the formation and interrupted covalent cross linking of serum albumin for seven days with the potent glycating agent and glycolytic intermediate glycolaldehyde. In the initial screen, seven compounds were found to have significant biological activity when compared to the known anti-glycating drugs aminoguanidine (AG) and ALT-711. One compound designated 2C4 was found to be active at inhibiting protein glycation by 65% at less than 5 micromolar whereas AG inhibited 35% only at the 400 micromolar dose. An additional 2-AI derivative (2C8) had biological activity in the 40 micromolar range. These data demonstrate that 2-AI derivatives have therapeutic potential as potent anti-glycating compounds. The synthesis of these biologically active compounds represents a focused and rational design strategy for the discovery of new anti-glycating compounds. 1206-P Timing of Insulin Initiation and Diabetes Complications in a Large Health System: Developing a Profile for Secondary Prevention JANICE C. ZGIBOR, LI CHUAN TU, SHIH CHEN KUO, WEI HSUAN LO-CIGANIC, FRANCIS SOLANO, KRISTINE RUPPERT, Pittsburgh, PA Diabetes and its complications contribute to significant morbidity and mortality, however, approximately one half of patients remain above the recommended A1C goal. Currently, there is a paucity of data from real world settings examining longitudinal outcomes in those using various treatment regimens. Identifying the ideal time for insulin intervention, from a population perspective, remains elusive and could provide insight into missed opportunities for prevention. This analysis used existing electronic health record data from the University of Pittsburgh Medical Center, to examine the association between timing of insulin initiation and incidence of microvascular complications. 1279 patients seen between 1/2/200306/08/2011 with type 2 diabetes were identified using previously validated algorithms. Average follow-up was 5.3 years. Retinopathy, nephropathy, and neuropathy were defined by ICD-9 and CPT codes. Prevalent cases were excluded. Follow-up started at the first available A1C (baseline) ≥ 7% and ended at the time of event or last contact. Time from baseline to development of complications was assessed. Four hundred sixty-five patients (36%) developed complications over the follow up period. Among those with complications, 98 (21%) started insulin, an average of 10.9 months post-baseline. Insulin users were significantly (p<0.001) younger (53 vs 61 yrs), more likely to be African American (47 vs 29%), had a higher A1C ADA-Funded Research & Supported by: NIH For author disclosure information, see page 829. Guided Audio Tour poster A315 POSTERS Curative strategies for type 1 diabetes are likely to combine beta cell regeneration with agents to uncouple destructive autoimmunity. Lisofylline (LSF) is a compound that disrupts autoimmune processes and protects beta cells from inflammatory injury. As a therapy, LSF is limited by poor bioavailability. Novel molecules from a LSF-activity screen have been identified. Compounds exhibit enhanced drug-like properties, having passed predictive in silico and ADME screens. Select compounds have been characterized using in vitro efficacy assays conducted in beta cell lines (INS1, βTC3), primary mouse islets and primary human donor islets. The efficacy of select compounds to preserve and protect beta cells from the harmful effects of pro-inflammatory cytokines (PICs) has been assessed. Following acute exposure to PICs, beta cells have elevated apoptosis, loss of glucosestimulated insulin secretion (GSIS) and enhanced gene expression of IL-12 ligand, MCP-1, and IFNγ. Assessment of five select compounds (DT09, DT021, DT042, DT047, DT079) at 50µM, showed significant protection to PIC-induced apoptosis measured by caspase-3 activation and YO-PRO1/PI fluorescent microscopy in beta cell lines and primary (mouse/human) islets (p<0.05). Compounds preserved GSIS in INS-1 cells exposed to PICs (p<0.05). IL-12ligand, IFNγ and MCP-1 gene expression induced by PIC-stimulation was inhibited by select compounds (p<0.05). Compounds were well tolerated in vivo and insulitis score in pre-diabetic NOD mice treated for six weeks was assessed. Collectively, the activity-profile of each compound varies, reflecting different molecular structures. Relative to LSF, DT042 and DT047 exhibited equal or greater activity to preserve and protect beta cell function following PIC exposure. These chemotypes are candidates for proof-ofprinciple combination therapy and provide new opportunities to protect beta cells exposed to an inflammatory environment. Clinical Diabetes/ Therapeutics DAVID A. TAYLOR-FISHWICK, JESSICA R. WEAVER, LINDSEY GRIER, WOJCIECH GRZESIK, Norfolk, VA HEALTH CARE DELIVERY—ECONOMICS 1209-P Comparative Assessment of the Effects of Strong Statins on Residual Risk in Diabetes With Accumulated Metabolic Risk FactorsTOHO-LIP Diabetes Sub-Analysis at 96 Weeks POSTERS Clinical Diabetes/ Therapeutics TERUO SHIBA, SUMIE OKAHATA, KENTARO SAKAMOTO, FUMIHIKO HARA, SHUUJI NANJO, SHINJI HISATAKE, ATSUSHI NAMIKI, SHOHEI YAMASHINA, KENJI WAGATSUMA, JUNICHI YAMAZAKI, GEN YOSHINO, TAKAHISA HIROSE, YASUO IWASAKI, SHIGEO YAMAMURA, HIDEHIKO HARA, KAORU SUGI, TOSHIKI FUJIOKA, MAO TAKAHASHI, HIROFUMI NOIKE, KOHJI SHIRAI, Tokyo, Japan, Hyogo, Japan, Chiba, Japan A recent analysis has revealed diabetes patients still bear residual risk after strong statin therapy. Most notably, HDL- C and non-HDL-C both seem to be of considerable importance in diabetics with multiple metabolic risk factors (MR). The TOHO-LIP (TOHO-Lipid Intervention Trial Using Pitavastatin) is a direct head-to-head comparison between pitavastatin (2mg/day) (P) and atrovastatin (10 mg/day) (A) conducted at three TOHO University hospitals in Japan on 652 enrolled patients with high-risk hypercholesterolemia. We conducted a diabetes sub-analysis on 414 diabetes patients to evaluate the statin effects in lipid profiles and safety at 96 weeks of treatment. Along with accumulation of MR, baseline HDL-C values were significantly reduced. Baseline non HDL-C values were increased only to a statistically significant level in the diabetics (p < 0.0001 , ANOVA). Both statin treatments brought about significant improvements of non HDL-C (-33%,p<0.0001 (P), -36%, p<0.0001 (A), respectively) and other atherogenic lipids at 96 weeks in the diabetics. In patients with densely accumulated MR,HDL-C at 96 weeks was increased significantly only in the pitavastatin-treated group (9.2% increase; p=0.0158). No adverse effects on glucose metabolism parameters (FPG or HbA1c) were observed in either group. The statin doses found to affect cardiac outcome are reported to range significantly from high to standard doses. There have few prospective studies comparing strong statins administered at standard doses. In our comparison pitavastatin, but not atrovastatin, improved the HDL-C level, one of the primary residual risks, in diabetics with accumulated MR. Pitavastatin thus appears to effectively prevent cardiovascular events. TOHO-LIP will determine the cardiovascular outcome brought about by pitavastatin’s favorable effects on residual risk and follow the patients for 5-7 years at maximum. & MAN WO TSANG, CHI SANG HUNG, SZE YUEN FUNG, YU CHO WOO, MAN-FUK LEUNG, PHILIP TSANG, Hong Kong, China The aims are to enhance diabetic management and reduce diabetes related hospital admission. Phase one is to develop a web-based glucose monitoring system based on One-Touch glucometer so that once the meter is connected to a net- book the glucometer readings will automatically upload to the server based at our institution. Phase two is a one year randomized, controlled prospective study of elderly with type 2 diabetes mellitus using the web-based computer assisted diabetes monitoring system, DMS. Patients were randomly assigned on block to control group or study group. Written consent was obtained.The aged home staff performed at least twice per week blood glucose haemoglucostix for all patients. Only data from study group were to be transmitted via a net-book to our centre. Two diabetologists took terms in reviewing the results and recommended feedback to the care providers at the aged home on management. The feed- back included dietary advice, drug adjustment or early follow up as indicated. The control group had the usual care and twice weekly glucose monitoring. 113 patients were recruited. 13 (4 from control group and 9 from study group) died and were excluded from this analysis. Mean age for study vs control group 83.30±5.1 vs 78.78±7.60 (P<0.001). Baseline Hba1c between study and control group was comparable, 7.42±1.36 vs 7.6±1.64 , (P=0.57).The decrease in HbA1c, at the end of one year study was -0.70±1.57 vs -0.46±1.37 in study and control group respectively (p=0.041). The causes of death were due to pneumonia, renal failure, stroke and ischaemic heart disease. The number of admission due hypoglycemia was too low for meaningful comparison. The higher death rate in the study group might partially accounted for by their older age. The study suggested that compared with usual care application of web-base diabetes monitoring system improved diabetes management among elderly with diabetes mellitus. HEALTH CARE DELIVERY—ECONOMICS Guided Audio Tour: Improving Diabetes Care (Posters: 1210-P to 1217-P), see page 15. & 1211-P A Web-Based Remote Tele-Monitoring System to Monitor Blood Glucose Levels in Aged Home Residents With Type-2 Diabetes Mellitus 1210-P Weight Loss Outcomes Among Older and Younger Adults in an Adapted Diabetes Prevention Program SARAH BROKAW, DIANE ARAVE, MARCENE BUTCHER, STEVEN D. HELGERSON, TODD S. HARWELL, Helena, MT The purpose of this study was to evaluate if there are differences in selfmonitoring (SM) behaviors, physical activity (PA) levels, and achievement of 7% weight loss goal among older and younger adults enrolled in an adapted diabetes prevention program (DPP). From 2008 through 2011, adults (N = 1,820) at high-risk for CVD and diabetes were enrolled in a group-based lifestyle intervention. Multiple logistic regression (LR) analyses were used to identify factors associated with SM of fat intake, achievement of the PA goal, and achievement of the 7% weight loss goal. Twenty-five percent of participants (n = 455) were 18 to 44 years of age, 60% (n = 1,097) were 45 to 64 years of age, 14% (n = 258) were 65 years of age and older. In bivariate analyses, participants aged >65 years were significantly more likely to achieve the weight loss goal (44%) compared to participants 45-64 years of age (40%), and participants 18-44 years of age (28%). Using multiple LR analyses adjusting for age, sex, baseline BMI, achievement of the PA goal, and number of weeks SM fat intake, older participants were significantly more likely to SM their fat intake, and to achieve the PA goal compared to younger participants (Table). However, age was not independently associated with achievement of the weight loss goal. Our findings indicate that older participants in an adapted real world DPP can achieve weight loss related behaviors and the weight loss goal as well as younger participants can. Supported by: Johnson & Johnson; OTDMS & For author disclosure information, see page 829. A316 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS & 1212-P We performed a multicenter study in insulin-injecting Spanish adults to assess the frequency of lipohypertrophy (LH) and determine its relationships to site rotation, needle reuse, glucose variability, hypoglycemia and consumption of insulin. The study involved 430 outpatients, 177 with DM1 and 253 with DM2. Multiple injection parameters were assessed by questionnaire; then each subject was examined by a study nurse for the presence of LH and for correct site rotation technique. Subjects were roughly half male, half female with ages ranging from 5 to 76 years (mean 49 ±23 [SD] years, 41 in DM1 and 55 in DM 2). Median number of injections/day was 4 for DM1 and 2 for DM2. 2/3 of patients (64.4%) had LH, more commonly in DM 1 (72.3%) than DM2 (53.4%). There was a strong relationship between the presence of LH and non-rotation of sites (p=0.01), with correct rotation technique having the strongest protective value (p=0.001). Of the patients without LH, 94% were observed to rotate injection sites correctly. Of patients with LH, 95% were observed to not rotate or to rotate incorrectly. Of those with LH 39.1% had unexplained hypoglycemia; for those without it was 5.9% (p=0.03). Of those with LH 49.1% had glycemic variability; for those without it was 6.5% (p=0.02). LH was also related to needle reuse, with risk rising significantly when use was more than 5 times (p=0.008). Longer duration of insulin therapy (p=0.02) and higher number of injections per day (p=0.01) were also risk factors for LH. Subjects with LH consumed on average 56 IU/day while those without consumed 41 IU/day (p<0.001). This 15IU difference multiplied over the number of daily injections into LH and assuming a cost of 0.0243 euros/IU equates to total annual cost to the Spanish health care system of over 122 million euros. In conclusion, correct injection site rotation is the critical factor in preventing LH. Avoidance of LH is associated with reduced glucose variability, hypoglycemia, insulin consumption and health care costs. Many diabetes patients fail to attain glycemic targets. We investigated factors influencing the setting and attainment of A1c targets in a sample of patients with type 1 diabetes (T1D). We recruited 1,893 adults (>18 years) from the T1D Exchange registry to complete a web-based survey on diabetes management, self-reported glycemic targets, and hypoglycemia. Responses were linked with registry data, including A1c laboratory results. We tested for differences between respondents with and without A1c targets, and across respondents with aggressive (<6.5%), normal (6.5 to 7%), and less aggressive (>7%) targets. We estimated logistic models to identify correlates of setting a less aggressive target than physician due to fear of hypoglycemia and correlates of target attainment. Standard errors were clustered to account for within clinic treatment patterns and propensity weighting was used to adjust for non-response bias. Respondents who reported having an A1c target (n=1771) were more likely to be married, have a graduate degree, and hold private insurance than those without targets (all p<0.05). Approximately half (54%) set targets below 6.5%. Respondents with depression (odds ratio=1.48), anxiety (1.76), and more frequent hypoglycemic events in the past month (1.03) or a major event in the past year (1.28) were more likely to have less ambitious targets than physician due to fear of hypoglycemia; those with a graduate degree (0.79) and married (0.74) were less likely (all p<0.05). Respondents with a bachelor (1.31) or graduate (1.44) degree, married (1.36), reporting more frequent testing of blood sugar (1.13), or using a continuous glucose monitor (1.40) came closer to attaining A1c targets; those with fear of fainting (0.54) and depression (0.68) were further from targets (all p<0.05). Our findings suggest that glycemic targeting may partially explain socioeconomic gradients in diabetes outcomes. Interventions identifying and alleviating symptoms of hypoglycemia early may improve A1c targeting and attainment. Supported by: Institute for Health Technology Studies 1213-P Now or Later? Real-World Study of Impact of Timing of Insulin Initiation on Outcomes Among Elderly Medicare Patients With Type 2 Diabetes Mellitus (T2DM) & USHA SAMBAMOORTHI, WENHUI WEI, STEVE ZHOU, RITUPARNA BHATTACHARYA, JOHN LING, MAYANK AJMERA, Morgantown, WV, Bridgewater, NJ PHILIP C. MCEWAN, MARK LAMOTTE, DAVID GRANT, JAMES PALMER, ADAM LLOYD, VOLKER FOOS, Cardiff, United Kingdom, Vilvoorde, Belgium, London, United Kingdom, Basel, Switzerland Initiation of insulin after failing oral antidiabetic drugs (OADs) has been suggested to be beneficial for patients with T2DM, but real-world data on the impact of timing of insulin initiation on clinical and economic outcomes among the elderly (≥65 y) is limited. Using the Humana Medicare claims database, this retrospective study analyzed elderly T2DM patients initiating basal insulin between 2007 and 2011. The number of baseline OADs was used as a proxy for the timing of insulin initiation. One-year follow-up outcome measures were treatment persistence, A1C, hypoglycemia rates, health care utilization, and costs. Of the 14,669 elderly patients included (mean age 74 y, female 49%, A1C 8.60% [where available]), 32% (n=4702) initiated insulin after 1 OAD, 48% (n=6980) after 2 OADs, and 20% (n=2987) after 3+ OADs. At baseline, compared with 2 or 3+ OAD patients, 1 OAD patients were similar in A1C (8.5%), but were sicker and had higher health care utilizations and costs. During 1-year follow-up, despite the 1 OAD group showing a lower insulin treatment persistence rate compared to the 2 and 3+ OAD groups (58.3% vs 64.6% and 70.2%; both P<0.0001), they had the highest reduction in A1C values (−0.88 vs −0.72 and −0.52; both P<0.05). Hypoglycemia rates increased, but decreased during the second half year. Significant reduction from baseline in annualized health care costs was observed in the 1 and 2 OAD groups (−$4913 and −$1599, respectively; both P<0.001) during 1-year follow-up, mainly due to decreased inpatient costs offsetting increasing drug costs. Health care costs stayed the same in the 3+ OAD group. Among elderly Medicare T2DM patients, significant improvements were observed after insulin initiation, not only in clinical but also economic outcomes, particularly among the 1 OAD group. A limitation of this study is that the duration of T2DM of each group is not known. Further study is needed to explore the heterogeneity of this population. Accurate estimation of baseline cardiovascular (CV) risk and relative risk reduction (RRR) is crucial to ensure that economic evaluations of new health technologies for the treatment of type 2 diabetes (T2DM) are robust. Many economic models (such as the CORE Diabetes Model) use risk equations (RE) derived from UKPDS and concerns persist regarding their validity; particularly as new equations are published. The objective of this study was to compare the consistency of predicted CV risk using RE derived from various T2DM populations. All CV equations identified from a recent systematic review, derived from populations with T2DM, were coded and validated. Equations from Australia (Fremantle), New Zealand (DCS), Sweden (Cederholm), China (Yang), Scotland (DARTS), USA (ARIC) and UK (UKPDS) were included. Predicted 5-year CV risk was obtained using baseline cohort characteristics taken from ACCORD. Relative risk reductions (RRR) were obtained by applying a 10% relative reduction in HbA1c, total cholesterol and SBP both individually and in combination. Mean 5-year predicted risk of CVD was 11.0% (SE 1.9%); minimum of 3.4% (ARIC) and maximum 20.7% (DARTS). A 10% reduction in HbA1c, TC and SBP resulted in a mean RRR of 6.4%(SE 0.7%), 6.8% (SE 1.5%)and 9.8% (SE 2.3%) respectively. The DCS equation (New Zealand) predicted the lowest RRR for HbA1c, TC and SBP reduction (4.3%, 1.0% and 3.5% respectively). The highest RRR for HbA1c change was Cederholm (8.3%) and the DARTS equation for TC and SBP, 10.3% and 18.9%, respectively. The difference in absolute risk across these equations does not appear dependent on geographical location or study recruitment period. Generally, the UKPDS equations produced consistent absolute CV risk and RRR estimates close to the group averages; this is of reassurance given their widespread use. Healthcare policy decisions that rely on CV risk estimation should perform sensitivity analysis across multiple equations where practicable. Supported by: Sanofi U.S., Inc. ADA-Funded Research & 1215-P Assessing the Consistency of Absolute Cardiovascular Risk Prediction and Relative Risk Reduction in Type 2 Diabetes Mellitus For author disclosure information, see page 829. Guided Audio Tour poster A317 POSTERS PETER HUCKFELDT, TARA K. KNIGHT, YAWEN JIANG, ROY BECK, KELLEE M. MILLER, ANNE L. PETERS, DANA GOLDMAN, Santa Monica, CA, Los Angeles, CA, Tampa, FL KENNETH STRAUSS, MARTA BLANCO, M. TERESA HERNÁNDEZ, MARISA AMAYA, Erembodegem, Belgium, Barcelona, Spain, Algeciras, Spain & 1214-P Patient Factors Associated With Setting and Attaining A1c Targets Clinical Diabetes/ Therapeutics & Prevalence, Risk Factors and Estimated Costs of Lipohypertrophy in Insulin-Injecting Patients With Diabetes HEALTH CARE DELIVERY—ECONOMICS & Guided Audio Tour: Moving Towards Better Diabetes Health Care Delivery (Posters: 1218-P to 1225-P), see page 15. Actionable Data to Improve Quality Diabetes Care 1216-P & POSTERS Clinical Diabetes/ Therapeutics SARAH G. IMERSHEIN, RICHARD A. JACKSON, Boston, MA Current clinical DM guidelines are important guideposts for clinicians to inform treatment decisions, but do not take into account individuals’ DM complexity or the results of recent clinical trials such as ACCORD and EDIC. To address this, we developed an algorithm that stratifies patient populations using medication regimen as a proxy for disease complexity. This algorithm places people with DM into 5 different groups of glycemic complexity, depending on their current medication regimen, assigning different A1C targets for each group. This algorithm was used with a large national network of affiliated diabetes centers. Stratified reports were delivered to each center. For 5 providers whose A1C score was < 70%, we provided individual reports and a brief discussion of their results by the individual groupings for glycemic complexity. To date 3874 charts have been assessed at 28 centers. Of these, 2191 (57%) are at 11 centers that have completed both a baseline and followup assessment. The network average for glycemic control in centers that have completed follow-up improved from 63.1% to 68.9%, p=0.002. More importantly, the 5 providers given reports followed by a brief discussion showed greater improvement than those who did not participate in brief discussions (Table 1). ROZALINA G. MCCOY, JIAQUAN FAN, STEVEN A. SMITH, JAMES R. DEMING, JEANETTE Y. ZIEGENFUSS, NILAY D. SHAH, Rochester, MN, Tomah, WI, Minneapolis, MN Quality metrics are used to measure the quality of diabetes-related care and guide quality improvement efforts. The D5 measures (HbA1c <8%, blood pressure <130/80 mmHg, LDL cholesterol <100 mg/dL, non-smoking, aspirin therapy), are approved by the National Quality Forum and are used by the majority of healthcare providers and payers, including the Centers for Medicare and Medicaid Services. Recent focus on patient-centered care led to the introduction of a patient-reported metric of chronic disease management: the Patient Assessment of Chronic Illness Care (PACIC). We sought to establish whether the PACIC correlates with traditional intermediate clinical outcome metrics, such as the D5. A postal survey was mailed to 4796 patients randomly selected from a diabetes registry maintained by 34 clinics in a Midwest practice-based research network in November 2009. The survey included the 20-item PACIC survey and questions on patient health, diabetes control, and demography. D5 metrics were obtained from electronic health records as of 2009. Logistic regression models were used to evaluate the association of outcomes with the PACIC score, adjusting for age, gender, diabetes duration, marital status, education level, and ethnicity. The survey response rate was 42.8% (2055 of 4796). Respondents were predominantly married (69.2%) and insured (61.9% public, 36.8% private); mean age was 65 years, 50.4% were male, and 93.4% had type 2 diabetes. Only 21.6% of patients (443 of 2055) met all five D5 quality metrics. PACIC scores were positively and significantly associated with confidence in diabetes self-management (p<0.001) and self-rating of health (p=0.02). However, there was no association between PACIC and the D5 (p=0.3). The PACIC may serve as an important tool to measure and improve the quality of patient-centered diabetes care that is distinct from and complementary to the currently available quality measures. Table 1. Percent of patients achieving A1C targets among providers with lower A1C scores, with and without additional discussion of results vs. the Network Average Lower Provider Network Average Scores (<70%) Discussed Did NOT Results discuss results (n=5) (n=6) At A1C Target Baseline 53.9% 57.6% 63.1% At A1C Target Follow-up 66.5% 62.2% 68.9% Difference +12.6%* +4.6%* +5.8% * Weighted average difference of difference, p<0.001 & Providing actionable data directly to providers, plus a brief discussion, may improve the number of patients achieving appropriate A1C targets. & 1218-P Patient Assessment of Diabetes Care Reflects Patient-Centered Outcomes But Not Traditional Outcome Quality Metrics 1219-P Comparing Apples With Apples: Diabetes Care by Endocrinologists vs. Internists LAWRENCE S. PHILLIPS, ARUN V. MOHAN, MARJAN KHOSRAVANIPOUR, ANNE TOMOLO, WENQIONG XUE, QI LONG, DIANA BARB, SANDRA L. JACKSON, DARIN E. OLSON, MARY K. RHEE, SONYA HAW, PHYLLIS WATSON-WILLIAMS, Atlanta, GA, Decatur, GA 1217-P Transitional Care Clinic for Patients With Diabetes May Prevent ReHospitalizations CECILIA C. LOW WANG, STACEY SEGGELKE, MATTHEW HAWKINS, ELIZABETH COHLMIA, JOANNA GIBBS, NEDA RASOULI, BORIS DRAZNIN, Aurora, CO Patients, insurers, and other stakeholders need to be able to assess diabetes care by healthcare institutions and providers, but we lack a good metric. As a common and costly disorder, diabetes is a model problem for evaluation of quality and value. However, simple performance measures may not take into account the time needed to improve glycemic control, and the complexities of care in patients who need insulin. To determine the need for such granularity, we used the Emory Healthcare data warehouse to compare the most recent A1c levels in patients seen by 8 endocrinologists (ENDOs) and 8 internists (GIMs). In all 5880 patients with use of the diabetes ICD-9 code 250.xx over 24 mo, the proportion with A1c >7% was higher for ENDOs vs. GIMs (51% vs. 38%, p<0.001). We then restricted analysis to the 3735 patients seen >3 times over 24 mo, and >1 time over 12 mo, and categorized as using only oral agents ± incretins (n=1880), also using basal insulin (n=324), and also using mealtime insulin (n=1531). Insulin-using patients were less well controlled: A1c was >7% in 66% of those using mealtime insulin and 55% of those using basal insulin, vs. 21% of those without insulin (p<0.0001 for trend). Moreover, ENDOs had more insulin-using patients than GIMs: 53% vs. 22% using mealtime insulin (p<0.0001), 10% vs. 7% using basal insulin (p=0.02), and 37% vs. 71% without insulin (p<0.0001), respectively. However, within each group, control was comparable or better for ENDO vs. GIM patients; A1c >7% was 18.8% vs. 23.4% without insulin (p=0.01), 53.7% vs. 57.1% with basal insulin (p=0.6), and 65.6% vs. 65.4% with mealtime insulin (p=0.9), respectively. Conclusion: Assessing performance in diabetes management on the basis of recent A1c levels in all patients may be misleading and inaccurate. Since patients seen by ENDOs are more complex than those seen by GIMs, assessments need to take into account the time needed to improve glycemic control and the difficulties of treatment with insulin. Our approach may be a useful model for evaluating care. Transitioning from the inpatient to the outpatient setting is often a problematic aspect of diabetes care. Different factors during hospitalization may adversely affect glycemic control. Patients are frequently discharged on regimens that differ markedly from pre-hospitalization outpatient regimens. Moreover, these regimens may have not been tested adequately during a relatively short length of stay, posing a significant threat to patient safety. We hypothesized that seeing patients with diabetes shortly after discharge would reduce readmissions. In our Transitional Care Clinic (TCC), patients with diabetes are seen within 2-5 days of discharge. We evaluated the effectiveness of TCC in preventing hospital readmissions within 90 days. One hundred indigent patients with diabetes discharged from the hospital were randomized to intervention group (INT, n=50) and seen in the TCC, vs. control group (CON, n=50). Since these patients lacked health insurance, timely follow-up for CON was often not available. The INT patients were seen by a diabetes specialist for a single 30 minute visit to adjust medications. All patients were contacted 90 days after discharge to collect information about re-admissions. Fourteen CON (28%) and 10 INT patients (20%) were re-admitted during follow-up (p=NS). Among patients originally admitted for diabetes-related issues (14 CON and 16 INT), 6 in CON (42.8%) and 2 (12.5%) in INT were re-admitted during the follow-up period (p< 0.05). In conclusion, TCC was effective for prevention of re-hospitalizations in indigent patients admitted for diabetes-related problems. However, the diabetes-specific TCC was not effective for patients originally admitted for other medical problems. Supported by: University of Colorado Hospital Supported by: U.S. Dept. of Veterans Affairs & For author disclosure information, see page 829. A318 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS & 1220-P Improving Health Outcomes in Emerging Adults With Diabetes Supported by: CDC & ERIN S. LEBLANC, ANA G. ROSALES, SUMESH KACHROO, JAYANTI MUKHERJEE, KRISTINE L. FUNK, GREGORY A. NICHOLS, Portland, OR, Wallingford, CT Diabetes patients often exceed desired A1C levels for months prior to medication adjustments. To better understand this clinical inertia, we examined patient and provider characteristics associated with optimal and poor glycemic control.149 primary care providers at Kaiser Permanente Northwest (KPNW) were identified as the primary provider for at least 10 diabetes patients. Using hierarchical linear modeling, we assessed both patient (n=14,430) and provider characteristics to determine predictors of optimal glycemic control (all A1C values <7%) as well as poor control (at least 1 A1C >9%). Of 14,430 patients, 5,823 (40.4%) were in optimal control and 2,446 (17.0%) were in poor control. Patient characteristics associated with optimal control included older age, lower baseline A1C, shorter diabetes duration, not using insulin, and fewer primary care visits (p<0.001 for all). The inverse of these variables also predicted poor control. Higher comorbidity burden and more visits to specialists, including endocrinologists, were not associated with optimal glycemic control. Males and patients who had more visits to specialists other than endocrinology were more likely to have poor control. No provider characteristics [age, gender, time with KPNW, mean comorbidity burden of patients, percentage of patients with diabetes, specialty (internal medicine vs family practice), degree (MD vs NP/PA) or clinic] were associated with optimal glycemic control. Only 1 provider characteristic, diabetes patients as a percent of total patients, was associated with poor control (p=0.03); each additional percentage point increased the odds of an A1C being ≥ 9% by 4%. More patient than provider characteristics predicted glycemic control. To better understand clinical inertia, future research should focus on understanding how provider decision making in response to poor glycemic control is influenced by individual patient characteristics, needs, beliefs and attitudes. Supported by: CDC & 1222-P Understanding Clinical Inertia: Are Patient or Provider Characteristics Predictive of Glycemic Control? 1221-P Does Group Size Impact Weight Loss Outcomes among Participants in an Adapted Diabetes Prevention Program? SARAH M. BROKAW, DIANE ARAVE, DEREK N. EMERSON, MARCENE K. BUTCHER, STEVEN D. HELGERSON, TODD S. HARWELL, Helena, MT The purpose of this study was to assess if group size is associated with weight loss goal among participants in an adapted diabetes prevention program (DPP). Adults at high-risk (N = 841) for CVD and diabetes were enrolled in the lifestyle intervention in 2011. The median group size was 16 (range 8-38). Multiple logistic regression (LR) analyses were used to determine if group size (smaller group <16 participants; larger group >16 participants) was independently associated with weight loss goal among participants. The mean age of participants was 53.4, and 82% were female. The mean number of core sessions attended was 13.0, and 36% of participants achieved the 7% weight loss goal. In the bivariate analyses, compared to participants in the larger groups those in smaller groups were more likely to have a higher BMI at baseline, attended fewer intervention sessions, were less likely to self-monitor their fat intake for >14 weeks, and lost less weight during the intervention. However, using multiple LR analyses adjusting for age, sex, baseline BMI, achievement of the physical activity goal, number of weeks self-monitoring fat intake, and group size, only two factors were independently associated with achievement of the weight loss goal: frequency of self-monitoring of fat intake and achievement of the physical activity goal (Table). Our findings indicate that group size is not associated with achievement of weight loss goal in an adapted DPP. Supported by: Bristol-Myers Squibb/AstraZeneca & 1223-P Visit-Based EMR Reminders With Exception Reporting Improve Diabetes Outcomes MICHAEL E. BOWEN, JASON FISH, DEEPA BHAT, BRETT MORAN, TEMPLE HOWELL-STAMPLEY, LYNNE KIRK, KIM BATCHELOR, ETHAN HALM, Dallas, TX Although best practice alerts (BPAs) may improve care via just-in-time reminders, the impact of provider engagement by exception reporting is poorly understood. This project examines the impact of BPAs with exception reporting on nationally recognized diabetes quality metrics. BPAs were developed and implemented in an internal medicine clinic. Established patients with diabetes were identified according to NCQA definitions. Providers voluntarily reported exceptions to BPA adherence. We describe the nature and frequency of BPA exceptions and assess the impact of BPAs on quality metrics using time series linear regression models. Patients with exceptions indicating the absence of diabetes, medical, or ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A319 POSTERS Studies of emerging adults (ages 18-24; EA) with diabetes often focus on risk factors for poor disease outcomes. Features of positive health outcomes [high health satisfaction (HS) and good glycemic control] are less defined. In particular, the role of buffering factors [anticipatory guidance of adult screening guidelines, diabetes related social support, frequency of physical activity and quality of life (QoL)] is unknown. We tested the hypothesis that higher HS survey ratings were associated with higher amounts of buffering factors and lower A1c. The study population consisted of participants in the SEARCH for Diabetes in Youth Study (prevalent diabetes in 2001) who completed the Quality of Care and Pediatric Quality of Life Generic Core Young Adult measures in 2008. Respondents (n= 342) had a mean age and diabetes duration of 21.9 + 2.4 and 11.7 + 3.7 yrs., respectively. The sample included: 71.9% type 1 diabetes, 79.8% minority status, 56.1% female, 77.6% continuously insured for the past 12 months. HS score was defined as a composite of overall health satisfaction and ratings of health care in the preceding 12 months, (range 2-10). Four buffering factors (described above), were also derived from the survey. Associations between HS and both the 4 buffering factors and self-reported A1c were assessed using Pearson correlations. HS scores were positively correlated with the levels of the 4 buffering factors: anticipatory guidance (r= 0.26, p=<0.0001), social supports (r= 0.16, p= 0.003), physical activity (r= 0.14, p=0.01), QoL total score (r=0.45, p=<0.0001); while HS was inversely related to A1c (r=-0.31, p=<.0001). There was no significant difference in the relationship between buffering factors and HS across diabetes type. These data demonstrate that higher HS scores are associated with higher levels of buffering factors and lower A1c in a diverse cohort of EA with longstanding diabetes. Further study of buffering factors in EA may identify interventions for improving health outcomes in this vulnerable population. Clinical Diabetes/ Therapeutics SARAH D. CORATHERS, JESSICA C. KICHLER, DANIEL P. BEAVERS, DANA DABELEA, JEAN M. LAWRENCE, ANGELA D. LIESE, JENNIFER RAYMOND, SHARON H. SAYDAH, MICHAEL SEID, JOYCE YI-FRAZIER, LAWRENCE M. DOLAN, SEARCH FOR DIABETES IN YOUTH STUDY GROUP, Cincinnati, OH, Winston-Salem, NC, Aurora, CO, Pasadena, CA, Columbia, SC, Atlanta, GA, Seattle, WA HEALTH CARE DELIVERY—ECONOMICS patient BPA exceptions were excluded from the denominator in outcome calculations. Between July 2011 and June 2012, 1761 established patients age 1875 with diabetes completed 5180 encounters with 37 providers after BPA implementation. At least one diabetes BPA was triggered in 2923 encounters for a total of 4794 BPAs. Of the 230 exceptions reported, 13% did not have diabetes, 37% had medical exceptions, 26% had patient exceptions, and 24% received services outside the health system. BPAs improved rates of glycemic control, LDL measurement, and LDL control after controlling for secular trends. Rates of A1C measurement were high at baseline and did not change post-intervention. (Table 1) BPAs engaging providers in exception reporting improved performance on A1C and LDL metrics. POSTERS Clinical Diabetes/ Therapeutics Supported by: Eli Lilly and Company & Supported by: University of Texas Health System 1225-P & Japan Diabetes Outcome Intervention Trial-1 (J-DOIT1): A Cluster Randomized Controlled Trial of Type 2 Diabetes (T2DM) Prevention by Telephone-Delivered Lifestyle Support for High-Risk Subjects NAOKI SAKANE, KENTARO OKAZAKI, KAZUHIKO KOTANI, KAORU TAKAHASHI, YOSHIKO SANO, KOKORO TSUZAKI, JUICHI SATO, SADAO SUZUKI, SATOSHI MORITA, YOSHITAKE OSHIMA, KAZUO IZUMI, MASAYUKI KATO, NAOKI ISHIZUKA, MITSUHIKO NODA, HIDESHI KUZUYA, Kyoto, Japan, Nagoya, Japan, Yokohama, Japan, Kobe, Japan, Tokyo, Japan 1224-P Is the Current Standard of Care Leading to Cost-Effective Outcomes for Patients With Type 2 Diabetes Requiring Insulin in the UK? To contain the growing T2DM population, it is critical to clarify effective and efficient modalities for intervention delivery with a wide population reach. The J-DOIT 1 was designed to test an effectiveness of goal-focused lifestyle support delivered by telephone for subjects with impaired fasting glucose (IFG) detected at health checkups. Forty-three groups (clusters), formed from 17 health checkup divisions in communities and worksites across the country, were randomly assigned to an intervention or a control arm. Study candidates, aged 20-65 years with fasting plasma glucose (FPG) of 5.6-6.9 mmol/l, were identified in each group using health checkup data in 2006. A total of 2840 subjects, 1336 from the intervention and 1504 from the control arm, were enrolled. The intervention was delivered by healthcare providers over a one year period, followed by 3 years’ follow-up period by annual health checkups. Primary outcome is the development of diabetes defined as FPG≥7.0 mmol/l. After one-year intervention, proportions of those who achieved the goals in terms of weight control, vegetable intake, and restriction on alcohol intake were modestly but significantly increased in the intervention arm. In spite of these favorable changes, overall, we could not find an improvement in the 5 years’ cumulative incidence of diabetes, with the hazard ratio (HR) in intervention group being 0.97 (95% confidence interval (CI), 0.75-1.25). In the sub-analyses with obese subjects (Body Mass Index≥25), no difference was found between the arms. However, the intervention significantly reduced the HR to 0.36 (95% CI, 0.15-0.83) among subjects with nonalcoholic fatty liver disease (NAFLD). Thus the effects of intervention by telephone appear limited in a real-world primary care setting. NAFLD subjects with IFG may be promising target populations for preventing T2DM. WILLIAM VALENTINE, BRADLEY H. CURTIS, RICHARD POLLOCK, KRISTINA S. BOYE, KATE VAN BRUNT, DAVID M. KENDALL, MICHAEL BRANDLE, Basel, Switzerland, Indianapolis, IN, Windlesham, Surrey, United Kingdom, St. Gallen, Switzerland To investigate whether insulin intensification, based on use of intensive insulin regimens recommended by the current standard of care in clinical practice would be cost-effective for patients with type 2 diabetes (T2DM) in the UK. Data were derived from a retrospective analysis of 3,185 patients with T2DM on basal insulin in The Health Improvement Network general practice database. Of those who changed therapy, 48% (614 patients) intensified insulin therapy, defined as adding bolus or premix insulin to a basal regimen. Projection of clinical outcome and cost (2011 GBP) over patient lifetime was made using a recently validated model of T2DM. Insulin intensification was associated with improvements in life expectancy, quality-adjusted life expectancy and time to onset of complications versus no intensification or delaying intensification by 2, 4, 6, or 8 years. Direct costs were higher with the insulin intensification strategy (due to the acquisition costs of insulin). Incremental cost-effectiveness ratios for insulin intensification were GBP 32,560, GBP 35,187, GBP 40,006, GBP 48,187 and GBP 55,431 per QALY gained versus delaying intensification 2, 4, 6 and 8 years, and no intensification, respectively. Although associated with improved clinical outcomes, insulin intensification as currently practiced in the UK has a relatively high cost per QALY and may not lead to cost-effective outcomes in the UK for patients with type 2 diabetes. Supported by: Health and Labour Sciences & For author disclosure information, see page 829. A320 Guided Audio Tour poster ADA-Funded Research 1228-P Improving Community Diabetes Care Through a Novel, Educational Outreach Visit Program BELINDA A. BROOKS, FRANZISKA H. LIMACHER, LYNDA MOLYNEAUX, MARIA CONSTANTINO, DENNIS K. YUE, Camperdown, Australia, Sydney, Australia JILL E. VOLLBRECHT, Traverse City, MI In recent years, increased emphasis has been placed on tracking outcomes of evidenced-based care of chronic disease. It is therefore essential that community leaders explore new methods of educating medical providers that result in sustained outcome improvements. In this program, an endocrinologist designed a curriculum based on ADA guidelines for management of diabetes, and reviewed it with each of nine primary care physicians (PCP). The endocrinologist and PCP then evaluated four diabetic patients together, and together formulated a treatment plan for each. The effect of this site visit was analyzed on previously-controlled diabetic patients (HbA1c <7%, n = 291; LDL <100 mg/dL, n=186; urine microalbumin:creatinine (ACR) <30 µg/mg, n=170) and uncontrolled patients (HbA1c >7%, n =216; LDL >100, n=148; ACR >30 µg/mg, n=42). Nine months following the education, statistically significant reductions in median LDL (123 to 106 mg/dL, p=0.008) and ACR (64 to 24.8 µg/mg, p=0.003) were demonstrated in previouslyuncontrolled patients. A trend toward improvement in median A1c was noted (7.9 to 7.7%, p=0.26). Eighteen months after the education, the effect was sustained (median LDL 80 mg/dL, p=0.001; median ACR 8 µg/mg, p=0.032; median HbA1c 7.8%, p=0.5). No improvements were noted in any outcome measures on analysis of data from previously-controlled patients. Total diabetic education referrals from the participants increased 23% following the site visit (142 to 175). The number of diabetic patients referred to endocrinology for diabetic management remained stable at 4% of the total diabetic population of the participants. PCP satisfaction was assessed using pre- and post-visit surveys; PCP’s reported a subjective increase in knowledge of and confidence in treating diabetes. The results indicate that a half-day educational session with an experienced diabetologist can lead to sustained, improved glycemic and lipid control in previously-uncontrolled diabetic patients. Ideally, clinical trial participants (CTP) should be representative of patients in clinical practice but rarely has this been tested in diabetes. We examined clinical parameters and individual case fatality data (matched with the Australian National Registry) of our CTP (n = 444). Overall, more males took part in clinical trials (68% vs 56%, p<0.0001). We matched CTP 1:5 with non clinical trial participants (NCTP) (n = 2220) for age, gender, type and duration of diabetes according to the year the trial was conducted. CTP were more likely to be anglo-celtic, less likely to smoke, have lower case fatality and die at an older age (Table 1). To overcome the possibility that CTP had less severe diabetes, we further matched a subgroup of CTP with comprehensive clinical data (n = 247) with NCTP (n = 1720) for glycaemic control, microvascular and macrovascular complications. Whilst case fatality rates were not significantly different (10.1 vs 14.3%; p = 0.07), CTP lived longer (73.4 vs 68.3 yrs; p = 0.01). CTP were also on less blood pressure and lipid lowering treatment despite having similar blood pressure and higher cholesterol than NCTP. Our findings indicate that the applicability of clinical trial results to the diabetic population at large could be altered if the efficacy of the treatment modality being tested is affected by gender, ethnicity or smoking status. These subtle differences in CTP may distort the calculation of the cost benefit ratio for the general diabetes community. Matched 1:5 Age (years) Males (%) Duration of Diabetes (years) Case Fatality (%) Age at Death (years) Duration of Diabetes at Death (years) Current Smoker (%) CTP n = 444 NCTP n = 2220 Statistic 54.0 ± 13.2 54.0 ± 13.3 p = 1.0 68 68 p = 1.0 8.3 [2.6 - 15.3] 8.3 [2.8 - 15.1] p = 0.5 11.7 16.5 p = 0.01 72.2 ± 11.1 68.5 ± 10.3 p = 0.02 19.2 [12.9 - 28.3] 14.8 [9.4 - 21.4] p = 0.001 9.5 24.1 Supported by: Michigan Dept. of Community Health 1229-P p = 0.001 WITHDRAWN 1227-P Association of ≥5% Weight Loss and Self-Reported Adherence With 6-Month Glycemic Control in Type 2 Diabetes Mellitus (T2DM): the DELTA Study CARRIE MCADAM-MARX, BRANDON K. BELLOWS, GAIL D. WYGANT, JAYANTI MUKHERJEE, SUDHIR UNNI, XIANGYANG YE, JOSHUA LIBERMAN, UCHENNA H. ILOEJE, DIANA BRIXNER, Salt Lake City, UT, Princeton, NJ, Wallingford, CT, Sacramento, CA The association between weight loss, adherence and glycemic control in patients with uncontrolled T2DM remains largely uncharacterized. T2DM patients ≥18 years were identified in a US integrated health system (1/1/2009 to 10/31/2011). Those prescribed a new class of anti-diabetic therapy (AD) on index date not previously used and with baseline HbA1c ≥7.0% were included. Outcomes of HbA1c goal attainment (<7.0%) and weight loss of ≥5% were defined at 6 months. Outcomes were evaluated by literature defined weight-effect properties of prescribed AD (weight loss - Metformin, GLP-1 agonists) or no weight loss (sulfonylureas, thiazolidinediones, insulin, DPP-4 inhibitors, and others) and by patient-reported medication compliance per the Medication Adherence Reporting Scale. Structural equation modeling described simultaneous associations between adherence, weight loss, and glycemic control. A total of 477 patients met inclusion criteria; mean (SD) age was 59.1 (11.6) years; 50.9% were women; and 30.4% were treatment naïve. Mean baseline HbA1c was 8.6% (1.6) and weight was 102.0 kg (23.0). A majority of patients (67.9%) reported being adherent to the index AD. At 6 months 15.9% had weight loss of ≥5% and 42.8% attained HbA1c goal. Mean weight change was -1.3 (5.1) kg, (p=0.39); mean HbA1c reduction was -1.2% (1.8) (p<0.001). Patients prescribed an AD with weight loss properties (OR 2.62; p=0.001) were more likely to experience weight loss ≥5% relative to those prescribed an AD not associated with weight loss. Weight loss of ≥5% (OR 6.40; p<.001) and being adherent with AD therapy (OR 1.85; p=0.005) were associated with HbA1c goal attainment. Though weight loss ≥5% and adherence were associated with glycemic control in T2DM, weight loss was a stronger predictor of HbA1c goal attainment than medication adherence in this study population. It is important to consider weight-effect properties, in addition to adherence counseling, when prescribing ADs. 1230-P Flexible Insulin Dosing Improves Health-Related Quality of Life (HRQoL) in a Basal Only Treatment Regimen: A Time Trade-Off Survey MARC EVANS, HENRIK H. JENSEN, METTE BØGELUND, JENS GUNDGAARD, BARRIE CHUBB, KAMLESH KHUNTI, Cardiff, United Kingdom, Holte, Denmark, Søborg, Denmark, Crawley, United Kingdom, Leicester, United Kingdom Rigidity of insulin regimens may negatively impact HRQoL and therapy adherence, with the requirement for basal insulin dosing at the same time every day being a significant contributing factor. We examined the HRQoL impact of both flexible dose timing and the number of basal insulin injections in a basal-only regimen using time trade-off (TTO) methods. HRQoL was quantified via an online TTO survey in the UK, Canada and Sweden with separate analyses of 1,121 respondents from the general population and 192 people with type 2 diabetes (T2D). HRQoL was measured on a utility scale: 0 (dead), 1 (perfect health). Respondents traded-off length of life for improving HRQoL in described health states. Respondents evaluated health states with diabetes and once-daily injections with flexible timing, once-daily fixed time injections, and twice-daily fixed time injections. In the general population, time flexible once-daily injection was associated with 0.016 (95% CI 0.011; 0.022) higher utility vs. a fixed time of injection as shown in table 1. The people with diabetes confirmed these results with a utility benefit of 0.015 (95% CI 0.004; 0.027). Once-daily injections had significantly higher utility compared to twice-daily. Flexible dosing and fewer injections have a positive HRQoL impact, which may enhance therapy adherence and potentially contribute to improved long-term outcomes in people treated with basal only insulin regimens. Supported by: Bristol-Myers Squibb ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A321 POSTERS 1226-P Diabetic Clinical Trial Participants Have Less Cardiovascular Risk Factors and Lower Case Fatality Clinical Diabetes/ Therapeutics HEALTH CARE DELIVERY—ECONOMICS HEALTH CARE DELIVERY—ECONOMICS Table 1. Utility differences from the time trade-off survey N Utility CI95 CI95 Lower Upper General population 1 Flex vs. 1 Fixed 1121 0.016 0.011 0.022 1 Fixed vs. 2 Fixed 1121 0.039 0.032 0.046 1 Flex vs. 2 Fixed 1121 0.055 0.048 0.063 Patients with T2D 1 Flex vs. 1 Fixed 192 0.015 0.004 0.027 1 Fixed vs. 2 Fixed 192 0.042 0.025 0.061 1 Flex vs. 2 Fixed 192 0.057 0.040 0.076 SE when HbA1c exceeded 8.0%, first by adding sulfonylurea (SU), then by discontinuing SU and adding basal insulin, and then by adding prandial insulin. Alternative scenarios were run using metformin as the first rescue therapy. Health utilities and costs associated with microvascular and macrovascular events were obtained from established sources. Sensitivity analyses were performed. Both algorithms with CANA 100 and 300 mg were estimated to reduce diabetes complications vs LMA: macrovascular events between 5% to 17% and microvascular events up to 36%. The CANA algorithms postponed the start of SU and insulin vs LMA. Within 1 year, 20% and 13% of the CANA 100 or 300 mg groups, respectively, vs 43% of the LMA group required SU. By year 10, only 27% and 19% of the CANA 100 and 300 mg groups, respectively, vs 66% of the LMA group were estimated to require insulin. In base case and sensitivity analyses, improvements in outcomes associated with the CANA algorithms led to lower costs and improved quality of life vs LMA over 30 years. p-value 0.003 <0.0001 0.003 <0.0001 0.004 <0.0001 0.006 0.0095 0.009 <0.0001 0.009 <0.0001 Supported by: Novo Nordisk, Inc. 1231-P POSTERS Clinical Diabetes/ Therapeutics Limited Effectiveness of Collaborative Primary Care Model for Diabetes and Depression: An Implementation Trial 1233-P Community One-Stop Diabetes Care Center in Cooperation With HMO Can Provide Effective Management for Diabetics in a Specific Ethnic Group JEFFREY A. JOHNSON, FATIMA AL SAYAH, LISA WOZNIAK, SANDRA REES, ALLISON SOPROVICH, WEIYU QUI, CONSTANCE CHIK, PIERRE CHUE, PETER FLORENCE, JENNIFER JACQUIER, PAULINE LYSAK, ANDREA OPGENORTH, WAYNE J. KATON, SUMIT R. MAJUMDAR, Edmonton, AB, Canada, Seattle, WA GEORGE LIU, VIVIEN HUI, TA-MIN CHANG, New York, NY Diabetes mellitus is the 7th leading cause of death in the US, affecting 8.3% of US population in 2011. Increasing cases of diabetes have occurred in Chinese Americans, affecting 15% of Chinese community population in New York City including 48% of patients at age of 65 years or older. Poor medication and diet compliance, lack of diabetic education, cost of medications, language/cultural differences, and depression are main obstacles for management of diabetes in Chinese American community. The Asian Diabetic Center (ADC) is established with a grant supported by United Health Group Inc. to promote the health of local Asian American community in New York City by providing accessible and high-quality medical care including consultations in Endocrinology, Nutrition, Podiatry, Nephrology, Vascular, Ophthalmology, and diabetes education. The purpose of this study is to determine the efficiency of one-stop health care management of 350 diabetic patients at ADC from March 17, 2011 to December 31, 2011. Significant first year improvement in various clinical parameters (mean ± SE, P<0.0001) were observed between first visit and after ADC consultation. Significant changes were: body mass index (from 26.21 ± 0.45 to 25.56 ± 0.41; 48.8% of patients had weight loss of 5.07±0.51 lb), HbA1C (from 7.87 ± 0.14 to 7.12 ± 0.11 %; the number of patients with ≤7% HbA1C increased from 38.1% to 59.2% and patients with ≥9% HbA1C decreased from 22.2% to 6.4%), LDL (from 83.65 ± 3.09 to 72.37 ± 2.66 mg/dl; the number of patients with ≤100 mg/dl increased from 70.8% to 82.9%), non-HDL (122.05 ± 3.46 to 107.62 ± 3.27 mg/dl; ≤130 mg/dl patients increased from 59.4% to 72.7%), and blood pressures (number of patients with ≤130/80 increased from 74.8% to 90.8%; ≤140/90, from 86.3% to 96.2%; >140/90, decreased from 13.7% to 3.8%). These observations indicate our one-stop care of diabetic patients is highly effective for the Asian American community. Depression is common in those with diabetes; when the two co-exist it worsens outcomes and increases health care costs. We evaluated the implementation of a nurse case-manager based collaborative team model to improve depressive symptoms and diabetes management in primary care in Alberta, Canada. We conducted a controlled implementation trial in 4 non-metro primary care networks. Eligible patients had type 2 diabetes and screened positive for depression, based on Patient Health Questionnaire (PHQ-9) scores >10. Patients were allocated using an “Intervention-On or -Off” monthly timeseries. Intervention consisted of case-managers working 1:1 with patients and their physicians, using motivational interviewing, evidence-based treatment algorithms, and specialist consultations as needed to deliver individualized care that prioritized depression, then cardio-metabolic treatment. The main outcome was improvement in PHQ-9 scores at 12-months; secondary endpoints included improvements in A1c, systolic pressure and LDL. We compared outcomes using random effects models, with intent-to-treat analyses. Of 1924 patients screened, 476 (25%) had PHQ >10. Of these, 95 were allocated to intervention and 62 to control. There were no baseline differences between groups: mean age 57.8 (SD 9.8) years, 55% women, 53% had BMI>35, mean PHQ-9 score 14.5 (SD 3.7), A1c 7.6% (SD 1.8), systolic pressure 125.2 mmHg (SD 16.0), and LDL was 2.2 mmol/L (SD 0.8). Preliminary analyses showed that intervention and control patients had similar 12-month improvements in PHQ-9 (6.4 [SD 5.7] vs 5.1 [SD 5.7] for controls; difference 1.1 [95%CI -2.8, 0.5]; p=0.18) and similar rates of depression remission (61% vs 58% of controls, p=0.71). No differences were observed in secondary outcomes. In patients with type 2 diabetes who screened positive for depression, a case-manager collaborative care model did not lead to improved outcomes compared to enhanced usual care in a Canadian primary care setting. Supported by: United Health Group 1234-P A Provider Satisfaction Inventory for Prediabetes Management Supported by: Alberta Health; CIHR (OTG-88588) JORDAN SILL, JEANINE ALBU, NANCY SOHLER, BRENDA MATTI, EDWIN YOUNG, GARY BURKE, New York, NY 1232-P Training of both providers and care teams is essential in the context of patient-centered medical home; assessing and measuring their satisfaction with management of type 2 diabetes (DM) has been reported but such methodology is not described for DM prevention. We administered a Provider Satisfaction Inventory (PSI) before and immediately after 1 hour training sessions we had designed for teaching the principles of systematic DM screening and prediabetes management. The PSI was a previously published scale (Montori et al Endocr Pract 2002) used to evaluate satisfaction with DM care on 4 categories: Chronic Disease Management, Collaborative Team Practice, Outcomes and Supportive Environment. The PSI was adapted to determine satisfaction with pre-diabetes care for this study. Forty providers (primary care physicians and extenders), 135 Internal Medicine residents (years 1-3), and 54 ancillary personnel (office coordinators, medical assistants, licensed practical nurses) from an inner city primary care network participated. Our baseline data shows the adapted instrument with adequate reliability, similar to the original version: correlations of each item to total PSI score with that item removed ranged from 0.49 to 0.74, and the Cronbach alpha for each scale ranged from 0.81 to 0.90 (0.94 for the PSI). Our training was successful: mean scores improved significantly A Health Economic Analysis of the Long-Term Benefi ts and Associated Cost Offsets of Canagliflozin Monotherapy in the U.S. CHERYL NESLUSAN, PIERRE JOHANSEN, MICHAEL WILLIS, SILAS MARTIN, Raritan, NJ, Lund, Sweden Canagliflozin (CANA) is a novel inhibitor of the sodium glucose cotransporter 2 in development for treating type 2 diabetes mellitus (T2DM). In a previously reported 26-week, randomized, double-blind, placebo-controlled Phase 3 study of 584 subjects inadequately controlled with diet and exercise, CANA 100 and 300 mg significantly decreased HbA1c vs placebo by 0.91% and 1.16%, respectively. We simulated how the observed changes in biomarkers (HbA1c, body weight, systolic blood pressure, cholesterol) may affect future outcomes and costs. We compared treatment algorithms beginning with lifestyle management and CANA 100 or 300 mg as monotherapy with one beginning with lifestyle management alone (LMA) using a validated micro-simulation model, the Economic and Health Outcomes Model (ECHO)-T2DM. Baseline patient demographics, baseline biomarker data, and treatment effects were sourced from the trial. In the base case, simulated treatment was intensified & For author disclosure information, see page 829. A322 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS insulin after 1 OAD, 48% (n=6980) after 2 OADs, and 20% (n=2987) after 3+ OADs. Multivariate analysis showed that those with older age (≥75 y), baseline hospitalization, diabetes complications, and discordant comorbidities were less likely to delay insulin initiation. Patients with more office visits, higher A1C levels, and polypharmacy at baseline were more likely to delay. These results suggest that current diabetes medical practice of insulin use is focusing on the sicker patients for earlier treatment, possibly because they may encounter urgent situations where they have to take insulin. However, the broader diabetes population needing insulin treatment by treatment guidelines is more likely to delay. Polypharmacy, which is common in the elderly, is another major obstacle. Our study highlights the importance of a patient-centered approach when evaluating timing of insulin initiation among the heterogeneous elderly population. after the training for each of the four scales and all provider groups (p<0.05), with the exception of two scales for which the ancillary personnel group had relatively high ratings at baseline. Though our training was successful, differences in satisfaction across the groups indicate a need for targeted training for specific personnel. For example ancillary personnel reported significantly lower satisfaction on Chronic Disease Management at both time periods than both providers and residents. Our results underscore the importance of satisfaction evaluations as part of the chronic disease prevention and management system for DM. Supported by: CDC (5U58DP002717) 1235-P Burden of Type 2 Diabetes and Out-of-Pocket Expenditures in Argentina Supported by: Sanofi U.S., Inc. JORGE ELGART, LORENA GONZÁLEZ, SANTIAGO ASTEAZARÁN, JOAQUÍN E. CAPORALE, J. DE LA FUENTE, CECILIA CAMILLUCCI, JUAN J. GAGLIARDINO, La Plata, Argentina, Córdoba, Argentina 1237-P DEBORAH J. WEXLER, STEPHANIE EISENSTAT, ELIZABETH GEAGAN, GIANNA WILKINS, LINDA DELAHANTY, MARCY BERGERON, ANNE THORNDIKE, MIRA KAUTZKY, BARBARA CHASE, MELANIE PEARSALL, ADRIAN ZAI, Boston, MA The patient centered medical home (PCMH) model offers novel opportunities for integrated, collaborative type 2 diabetes care, especially for complex care processes such as insulin initiation, which can be delayed between 2-7 years in primary care. Our objective was to reengineer the available clinical care team in the primary care practice to improve the quality and efficiency of insulin initiation. The team included a practice-based nurse diabetes care manager (not necessarily CDE-certified but called a Diabetes Champion) with a PCMH-based dietitian trained in diabetes medical nutrition therapy to coordinate care with the PCP. Three primary care practices were trained in protocols for insulin initiation, medical nutrition therapy, and behavioral assessment and motivational interviewing. The Diabetes Champions at each site used a patient registry to identify patients in the practice in need of insulin initiation using HbA1c and medication criteria. Over a 4 month intervention period, 42 patients were eligible and participated in the pilot. Mean age was 59 (SD 13) years (56% White, 31% Black, 10% Hispanic, 3% Other). The pre-intervention HbA1c for all participants was 8.8% (SD 2.2). Post-intervention, HbA1c was 7.7% (SD 1.6), weight decreased 0.04 (SD 5.6) kg, and there was a 5% rate of mild hypoglycemia. Fourteen patients had dietitian visits. Among the subgroup that started insulin, the pre-intervention HbA1c was 10.1% (SD 2.20). Post-intervention, HbA1c decreased to 7.3% (SD 1.2), a mean HbA1c reduction of 2.5% (SD 3.1); weight increased 0.1 (SD 1.05) kg; and there was an 8% rate of mild hypoglycemia. The usual rate of HbA1c decline in our population is 0.4% (95% CI 0.08-0.8) over 4 months. On average, the process required 1.2 nurse visits and 6.3 phone calls per patient. This model, teaming a Diabetes Champion nurse care manager with a dietitian based within the PCMH, improved the quality and efficiency of diabetes care. The next phase is to apply the model more broadly to other diabetes-related processes in the PCMH. Quarterly direct medical (MC) and Non Medical (NMC) costs (median [interquartile ratio] (number of cases) MC C Physician office visit 24 [12-48] (190) Laboratory test 114 [77-176] (86) Drugs 217 [61-473] (77) Practices 200 [90-481] (116) Hospitalizations 1,456 [674-4,388] (15) Subtotal 251 [72-666] (204) NMC 10 [0-34] (414) T2DM T2DM-C 48 [24-72] (201) 60 [24-96] (208) 159 [104-214] (154) 172 [120-247] (166) 676 [174-791] (155) 676 [296-1,414] (181) 159 [82-409] (125) 226 [130-583] (150) 453 [344-1,435] (11) 3,265 [1,473-6,569] (22) 573 [243-1,207] (215) 1138 [575-2,341] (213) 10 [0-40] (265) 20 [0-60] (277) P<0.05 compared to # ”C and *T2DM. Supported by: Novo Nordisk, Inc. 1238-P Trends in Insulin Initiation and Treatment Intensification Among Patients With Type 2 Diabetes 1236-P AMANDA R. PATRICK, MICHAEL A. FISCHER, NITEESH K. CHOUDHRY, WILLIAM H. SHRANK, JOHN D. SEEGER, JUN LIU, JERRY AVORN, JENNIFER M. POLINSKI, Boston, MA Real-World Study of Delayed Initiation of Insulin and Associated Factors Among Elderly Patients With Type 2 Diabetes Mellitus (T2DM) Many patients with type 2 diabetes eventually receive insulin, yet little is known about the patterns and quality of pharmacologic care following insulin initiation. Guidelines from the American Diabetes Association and the European Association for the Study of Diabetes recommend that insulin secretagogues such as sulfonylureas and meglitinides be discontinued at insulin initiation to reduce the risk of hypoglycemia. Our objective was to describe pharmacologic treatment patterns over time among adults initiating insulin. We identified a large commercially-insured population of adult patients without recorded type 1 diabetes who initiated insulin. We evaluated changes in non-insulin antidiabetic medication use during the 60 days immediately following insulin initiation, increases in insulin dose and/or dosing frequency during the 270 days following an insulin initiation period of 90 days, and rates of insulin discontinuation. 7,932 patients initiated insulin during 2003-2008, with the majority (61%) initiating basal insulin only. Metformin (55%), sulfonylureas (39%), glitazones (30%), and meglitinides (9%) were commonly used prior to insulin initiation. Basal or pre-mixed insulin dose and/or dosing frequency increased among RITUPARNA BHATTACHARYA, STEVE ZHOU, WENHUI WEI, MAYANK AJMERA, JOHN LING, USHA SAMBAMOORTHI, Morgantown, WV, Bridgewater, NJ Timely initiation of insulin in elderly patients with T2DM is a challenging and critical task in diabetes care, which requires evidence to help understand the patterns, limitations, and issues of current practice. Few real-world studies have investigated this topic, mainly due to the difficulties of defining “timely initiation of insulin” and limited access to treatment data on elderly patients with T2DM. Using the Humana Medicare claims database, this retrospective study analyzed elderly (≥65 y) patients with T2DM who initiated basal insulin between 2007 and 2011. The number of baseline oral antidiabetic drugs (OADs) was used as a proxy for the delay in insulin initiation. Multinomial logistic regression was used to assess the independent contributions of patients’ sociodemographic, clinical, and elderly-specific complexities to delayed insulin initiation. Included were 14,669 elderly patients (female 49%, white 71%, baseline mean age 74 y, A1C 8.60% [where available]), with 32% (n=4702) initiating ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A323 POSTERS We performed a case-control study to a) assess the impact of diabetes on different social and economic variables in Argentina, and b) compare them with those recorded in non diabetic people (C). We compared cases with Type 2 diabetes with (T2DM-C, n=387) or without chronic complications (T2DM, n=387) and C (n=774) paired by age, gender and social context. Data were obtained by structured telephone interviews (March-April 2011) and from clinical records. Statistical analysis was done using ANOVA/KruskalWallis and Chi-square (in proportions) tests. All subjects provided oral informed consent. In each group 54.8% were female. Diabetes duration in T2DM-C was significantly higher than in T2DM (8.1±8.4 vs. 10.8±9.0). T2DM-C had significantly lower salaries, higher monthly expenditures (with lower expenditure/income ratio), lower percentage of people with university degree, and higher hospitalization rate and resource consumption; they also lost more working days compared to T2DM or C. T2DM-C had significantly higher direct medical and indirect costs. The data show the heavy burden of diabetes and its complications for the patient, his/her family, health systems and society overall in Argentina. They also act as a warning to health authorities to strengthen the implementation of prevention strategies to reduce this burden. Clinical Diabetes/ Therapeutics Diabetes Home: A New Team-Care Model for Insulin Initiation in the PCMH HEALTH CARE DELIVERY—ECONOMICS with reductions in the no-show rate from 42.7% to 38.2% and the mean A1C from 10.4% to 9.8%. The top barriers addressed by navigators were related to transportation, finances, check-in processes, insurance, and housing. Although the study is ongoing, current data indicates PNPs in diabetes have promise for improving attendance and clinical outcomes in a inner-city patient population with healthcare disparities. 21% of patients, mealtime dose increased for 18%. Insulin was discontinued by 27% of patients. Among patients continuing insulin, metformin was continued by 62% of patients following insulin initiation, and nearly as often sulfonylureas (56%) and meglitinides (53%). We found evidence of substantial departures from guideline-recommended diabetes pharmacotherapy. Insulin secretagogues were frequently coprescribed with insulin. The majority of patients had no evidence of treatment intensification after insulin initiation, although this finding is difficult to interpret without HbA1c values. Addressing divergence from treatment guidelines following insulin initiation may improve patient outcomes. Supported by: William Randolph Hearst Foundation 1241-P WITHDRAWN Supported by: Eli Lilly and Company (F3Z-MC-B010) 1239-P Cost Predictors in Type 2 Diabetes Mellitus: A Retrospective Claims Database Analysis POSTERS Clinical Diabetes/ Therapeutics MORGAN BRON, ANNIE GUÉRIN, DOMINICK LATREMOUILLE-VIAU, RALUCA IONESCU-ITTU, JENNIFER SAMP, ERIC WU, Deerfield, IL, Boston, MA This study aimed to identify cost predictors in type 2 diabetes mellitus (T2DM) patients using oral antidiabetic therapies (OADs). Adult patients with T2DM using OADs were identified in the Truven Health Analytics MarketScan® databases (2004-2010). Patients with ≥4 HbA1c tests within 1 year were classified into cohorts based on quartiles (Q) of the distribution of total 12-month diabetes-related costs following a randomly selected OAD prescription date (i.e., index date). Diabetes-related costs were defined as the sum of costs for medical services with a diagnosis for diabetes and antidiabetic (AD) medications. Patient demographics, comorbidity profile, AD use, HbA1c level, and resource use were measured over the 12-month pre-index period and compared between cohorts with the highest costs (i.e., Q4) and lower costs (i.e., Q1-Q3). Multivariate logistic regression was used to identify predictors of the highest diabetes-related costs. Among the 3,921 selected patients with HbA1c tests (i.e., 5% of the eligible T2DM patients using OADs), the average 12-month diabetes-related cost was $12,623 for the highest cost cohort (N=981) and $1,871 for the lower cost cohort (N=2,940). Compared to the lower cost cohort, during the pre-index period, fewer patients in the highest cost cohort were female and achieved the glycemic goal (HbA1c ≤7%), more patients used insulin, and patients had a higher comorbidity burden, in particular microvascular diabetic complications, hypertension, and lipid disorders (all p<0.01). The main predictors (by greatest ORs) of the highest diabetes-related costs include insulin use (OR=3.08), coronary artery disease (OR=2.27), endocrinologist visit (OR=1.69), number of AD classes used (OR=1.56), diabetic retinopathy (OR=1.53), neuropathy (OR=1.44), and lipid disorder (OR=1.43) (all p<0.05). In a sample of T2DM patients receiving OADs, the main predictors for high diabetes-related costs included factors suggesting greater T2DM severity and diabetic complications. Supported by: Takeda Pharmaceuticals International, Inc. 1240-P 1242-P Navigation Program for Patients With Diabetes at an Inner-City Hospital Reduces A1C and Improves Clinic Attendance Evaluating the Patient Experience in the Asian Treat to Target Lantus Study (ATLAS): A 24-Week Randomized, Multinational Study WILJEANA J. GLOVER, VASSILIKI PRAVODELOV, ROBERTA CAPELSON, MELISSA NORGAISSE, DANIEL O’SHEA, VARSHA G. VIMALANANDA, JAMES L. ROSENZWEIG, Boston, MA NICK FREEMANTLE, KARIM ADMANE, SATISH K. GARG, ATLAS STUDY GROUP, London, United Kingdom, Paris, France, Aurora, CO Patients who keep their appointments manage their diabetes better than those who do not; patient navigation programs (PNPs) have been implemented to overcome barriers that may limit appointment attendance, as well as provide health education and psychosocial support. Navigators are used extensively in cancer care, but to a limited extent in diabetes care. We performed a pilot study to determine if the use of patient navigators would be associated with a decline in no-show rates and improved clinical outcomes. After nine months (1/30/2012-11/11/2012) of the PNP in Diabetes Services at the Boston Medical Center (BMC) using two patient navigators, 625 patients were accepted into the program based on the criteria of having an A1C > 8.5 and more than two no-shows or being recommended for the program by a provider [mean age 54 +/- 14; Black/Afr Am 52%, Hisp. 24%, White 12%, Other 12%; English 73%; Spanish 17%, Other 10%; Unemployed 39%, Full-time employed 12%, Part-time employed 4%, Other 45%]. Of these, 128 patients (20.5%) had participated in the program for >= 180 days, had >= 1 interaction with a navigator, and had been patients at BMC prior to 1/30/2012. We find a statistically significant difference in the no-show rate and A1c for these 128 participants. Specifically, PNP participation was associated The ATLAS study was designed to evaluate titration of insulin glargine in uncontrolled insulin-naïve patients on 2 OADs in Asia. A total of 552 patients were randomized: 275 patient-led and 277 physician-led. Insulin dose was adjusted using the same algorithm to achieve a target FBG of 110 mg/dL. ATLAS demonstrated noninferiority of patient-led compared to physician-led insulin titration. The overall scores for the Diabetes Treatment Satisfaction Questionnaire status (DTSQs) and change (DTSQc) were similar in the 2 groups. At week 24, DTSQs and DTSQc improved significantly for both groups (p<0.001); between-group difference was not significant. There were no differences in any DTSQc item. Health utility elicited with the EuroQol EQ-5D was high at baseline despite the need for insulin initiation. Baseline mean utility was 0.86 (95% CI 0.83 - 0.88) and 0.88 (95% CI 0.85 - 0.90) for each group respectively. Baseline EQ-5D was similar between countryrandomized groups but different between countries. Russian patients had the lowest health utility - mean 0.73 (95% CI 0.68 - 0.78) while Japanese patients had the highest - mean 0.94 (95% CI 0.91 - 0.96). No systematic differences between country randomized groups existed in the end of study utility. The ATLAS results confirm that Asian patients initiating basal insulin with a self-titration plan may do so without compromising health related quality of life or treatment satisfaction. & For author disclosure information, see page 829. A324 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS ($50,000–$100,000 per QALY or LYG), or not cost-effective (>$100,000 per QALY or LYG) and whether evidence was strong or supportive. Costs were in 2012 U.S. dollars. A total of 16 studies were included in the review (table).Sixteen studies met our inclusion criteria and were included in the review. Overwhelming evidence has shown that both lifestyle and metformin for preventing type 2 diabetes are efficient uses of heath care resources and should be implemented in a large scale in the US and the world (table). Table. Summary of the Cost-effectiveness Studies on Diabetes Primary Prevention 1243-P Type 2 Diabetes Mellitus Patients Who Are Trying to Lose Weight: Satisfaction With Oral Antihyperglycemic Agents (AHA) Lifestyle Standard High-risk counseling in care patients primary care setting CommunityNo Overweight supported campaign adults media campaign Intensive life- Standard Women with style modificare history of cation GDM,NGT or IGT ROBERT BAILEY, JANICE LOPEZ, KATHY ANNUNZIATA, LIEN VO, SILAS MARTIN, Raritan, NJ, Princeton, NJ Weight loss is recommended for overweight or obese individuals with type 2 diabetes mellitus (T2DM); however, excess weight and obesity continue to be a challenge to effective T2DM management. Some AHAs are associated with weight gain. An understanding of the relationship between weight gain and a patient’s satisfaction with medication may help inform treatment decisions under the recent ADA recommendation for the individualization of T2DM treatment. The study aim was to provide an assessment of T2DM patients’ satisfaction level with their current oral AHA. Results were taken from the 2011 US National Health and Wellness Survey, an online self-administered survey of a nationally representative sample of n = 75,000 adults aged ≥18 y. Medication satisfaction for all current AHAs was measured on a scale from 1 (“extremely dissatisfied”) to 7 (“extremely satisfied”); values of 6-7 = “satisfied”. Respondents were asked if they are currently taking steps to lose weight (TSLW) and were asked whether they gained weight, lost weight, or stayed the same over a 6-month period. The analysis focused on 7,828 T2DM patients (44% female; mean age = 58.5 y). About two-thirds of T2DM patients were TSLW. Their overall oral medication satisfaction was similar to those not TSLW (63% vs 62%). However, among those TSLW, those who lost weight were the most satisfied (67%) with their medication compared with those who gained weight (56%) or stayed the same (61%). Patients TSLW with HbA1c ≥7% had the most weight gain and the lowest satisfaction scores (21% gained and 55% satisfied) vs those TSLW with HbA1c <7% (18% gained and 70% satisfied). Weight loss is an important therapeutic objective along with AHA for T2DM. These data suggest that patients TSLW who failed to lose weight report lower levels of satisfaction with their AHA than those who lost or gained weight. Further study on the impact of specific AHAs in obese patients TSLW and satisfaction is warranted. 2 2 European countries 1 Australia CS 2 Australia, Israel, and India CS CS CS-$2700/LYG 1245-P Physician Specialty and Glycemic Control of Insulin Treated Patients in the U.S. MICHAEL GRABNER, SCOTT ABBOTT, MATT NGUYEN, YONG CHEN, RALPH QUIMBO, Wilmington, DE, Bridgewater, NJ Many patients [pts] on insulin do not reach the ADA-recommended level for A1C (<7%). Treatment decisions and A1C targets may vary by physician specialty, but there is little real-world evidence on this issue. This study compared the therapeutic management of insulin-treated pts between those with ≥1 prescription from an endocrinologist [Endo pts] and those without. Pts aged ≥18 years, with ≥1 insulin fill and health plan enrollment in 2011 were selected from the HealthCore Integrated Research DatabaseSM (representing a national health insurer with ~700,000 diabetes pts). Demographic and clinical characteristics were assessed. We identified 74,950 qualifying pts, of whom 28,154 (37.6%) were treated by 1,644 endocrinologists. A1C results were available for 16.4% of all pts. Compared to non-Endo pts, Endo pts were younger (54.1 vs. 58.3 yrs), more likely to be female (47.8% vs. 45.5%), have type 1 diabetes (15.3% vs. 7.1%), and have advanced disease (e.g. neuropathy, 22.7% vs. 16.7%). They were less likely to receive OADs (41.2% vs. 55.3%) but were more likely to receive basal-bolus insulin (48.5% vs. 37.9%). Average A1C was lower among Endo pts (8.36% vs. 8.62%) and they were less likely to have A1C ≥9% (Figure). We observed significant differences in patient profiles and treatment patterns across Endo and non-Endo pts. Regardless of specialty, nearly 80% of pts receiving insulin had an A1C ≥7%. These results highlight the unmet need for improved glycemic control among insulin treated pts. 1244-P RUI LI, PING ZHANG, Atlanta, GA Both lifestyle and metformin are effective in preventing type 2 diabetes and their cost-effectiveness (CE) has been evaluated in different delivery settings. We conducted a systematic review on all CE studies on diabetes primary prevention published between January 1985 and December 2012 using the procedure defined in the Cochrane Reviewers Handbook. Seven electronic bibliographic databases were searched to identify relevant studies. Quality of the study was assessed and studies that did not meet the criteria were excluded. Interventions were classified according to costeffectiveness ratios as cost saving, cost-effective (0–$50,000 per qualityadjusted life-year [QALY] or life year gained [LYG]), marginally cost-effective & Saving $5.7 billion nationwide or $1.8-$7.3 billion for Medicare in 10-25 years $750/QALY $250-$2700 CS: cost-saving; GDM: gestational diabetes; NGT: normal glucose tolerance; IGT: impaired glucose tolerance Cost-Effectiveness of Primary Prevention of Type 2 Diabetes: An Update of Systemetic Review ADA-Funded Research CS-$82,000/ QALY For author disclosure information, see page 829. Guided Audio Tour poster A325 POSTERS Supported by: Sanofi Range of the cost-effectiveness ratios CS-$91,500/ QALY Clinical Diabetes/ Therapeutics Intervention Comparison Intervention Number Country Median of the Population of stud- Settings cost-effectiveies ness ratios Intensive life- Standard IGT 11 US and 7 $1,400/QALY style modifi- lifestyle European cation recommencountries dation or no intervention Metformin Placebo IGT, Obese 8 US and 6 CS European countries A nationwide No this US popula2 US CS community- program tion at high based DPP risk for diatranslational betes program HEALTH CARE DELIVERY—ECONOMICS Supported by: Valeritas, Inc. POSTERS Clinical Diabetes/ Therapeutics quality indicators. The performance of foot examination was uniformly low. That of albuminuria test had a wide variation across doctors. HbA1c levels increased with the number of patients per one clinic. Although there were variations in their performances, they were within 3 SDs. In conclusion, we did not find an abnormal variation. However, unfavorable quality of the foot examination rate was reported. HbA1c levels increased with the number of patients per one clinic, suggesting a busy practice could lead to diminished quality of care. 1246-P Use of Insulin in the VA Is Improving DIANA BARB, WENQIONG XUE, CHRISTINE JASIEN, DARIN E. OLSON, JEEHEA S. HAW, MARY RHEE, ARUN V. MOHAN, ANNE TOMOLO, SANDRA L. JACKSON, QI LONG, LAWRENCE S. PHILLIPS, Decatur, GA, Atlanta, GA Most diabetes patients eventually come to need insulin, but primary care physicians (PCPs) - who manage most patients - often have difficulty using insulin, and a recent European study found that A1c was high at the time of insulin initiation and did not improve from 2005 to 2010 (EASD 2011). We asked if use of insulin is improving in the VA, where A1c levels are a “quality indicator” for PCP performance, and can impact PCP pay. We used a database of veterans from SC, GA, and AL, and identified nine type 2 diabetes cohorts with initiation of basal insulin 2001 through 2009. Metrics included A1c when insulin was initiated, A1c 12 months after initiation, nadir A1c within 3 years after initiation, and time to reach nadir A1c. 1583 veterans had mean age 67.7 years and BMI 32.6, and were 96.5% male and 38% black. From 2001 through 2005, there was no improvement in A1c at initiation, A1c at 12 months, or nadir A1c (all p=ns for linear trend). In contrast, from 2005-2009, there were highly significant trends for improvement: A1c at initiation fell 0.27 per year (p<.0001), A1c at 12 months fell 0.11 per year (p=.005), and nadir A1c was similar but the time to achieve nadir A1c was 7.7 weeks earlier per year (p<.0001). In 2005 vs. 2009, median A1c at initiation was 9.70 vs. 8.40 (p<.0001), A1c at 12 months was 8.10 vs. 7.70 (p=.09), and time to nadir was 111 vs. 61 weeks (p<.0001), respectively. Moreover, in 2005 vs. 2009, the proportion with A1c ≥8.0% at initiation decreased from 85% to 61%, and the proportion with A1c <8.0% at 12 months increased from 48% to 59% (both p<.05) - but those with A1c <7.0% at 12 months did not change (22% vs. 26%, p=ns). Conclusion: Although use of insulin in the VA changed little from 2001 through 2005, it improved 2005-2009, with significant decreases in A1c at initiation, A1c after 12 months, and time to reach nadir A1c. While further research is required to determine the impact of these improvements on patient outcomes and the VA, the VA’s use of A1c levels as a performance “quality indicator” which can affect pay may be a useful model for improving diabetes management in the US. 1248-P Assessing the Economic Impact of a Podiatrist: Analysis of Clinic Activity after Temporary Loss of 50% of the Podiatrists from a Tertiary Specialist Foot Clinic CATHERINE GOODAY, RACHEL MURCHISON, KETAN K. DHATARIYA, Norwich, United Kingdom Podiatrists form an integral part of the multidisciplinary foot team, however, their importance is often overlooked by purchasers of specialist services. A set of unforeseen circumstances within our specialist diabetes foot service meant that we lost 50% of our podiatry team within a few days for 6 months. Some of this time was filled by non-specialist community podiatrists. We wanted to assess the economic impact of this loss by examining data for the 5 years prior to this 6 month interruption, & for the 2 years after ‘normal service’ was resumed The results are shown in the table. Clinic No of Activity Admissions 1247-P Introduction of Statistical Process Control into Diabetes Care 2005 2006 2007 2008 2009 2010 2835 2921 3325 4197 4799 4058 30 43 39 50 58 72 % of Total Activity 1 1.5 1.1 1.2 1.2 1.8 2011 2012 4294 5270 41 45 0.95 0.89 Total Bed Days 515 775 570 919 867 1194 838 733 Mean Length of Hospital Stay (±SD) 17.2 (9.2) 17.2 (19.2) 14.6 (11.3) 18.4 (16.8) 14.7 (11.3) 16.5 (12.3) Year of interruption 20.4 (16.6) 16.2 (15.1) Our data shows that the loss of the podiatrists led to a significant rise in the numbers of admissions into hospital. At our institution a single bed day cost is $363. In the first month after the cut in service, we had 8 extra admissions, accounting for a total of 132 extra bed days. This equates to $47965. During the 12 months the extra costs increased by a total of $147333. Had the community help not been available, & admission rates stayed the same for the year, estimated costs would have risen to $547255 Our data does not look at the costs saved from avoiding surgery (currently set at $15270 for major amputations). Thus the actual costs incurred are likely to be higher. The annual salary of a senior podiatrist in the UK is set at $56394, thus they more than pay for themselves. MAKI KAWASAKI, YUSUKE KABEYA, MARI OKISUGI, MASUOMI TOMITA, TAKESHI KATSUKI, YOICHI OIKAWA, KIYOE KATO, YOSHIHITO ATSUMI, AKIRA SHIMADA, Tokyo, Japan Quality of diabetes care has been addressed recently. However, a universal method to control the quality has not been established. The aim of the study was to introduce a method derived from statistical process control (SPC), which is often used in manufacturing lines to control the quality of products, into diabetes care. Our hospital has a diabetes department, where 16 doctors provide diabetes care to approximately 8,000 patients. We sampled 1,105 regular visitors who attended the department from 23rd July to 4th August 2012. We calculated the overall and doctor-specific performances of quality indicators. Then, control charts were described. The measured values were plotted on the axis of each doctor’s average number of patients per one outpatient clinic. We tried to detect an abnormal variation in the performance from random variations. As a general rule of SPC,a value exceeding 3 SDs from the mean was defined as abnormal. Figure shows control charts of the & For author disclosure information, see page 829. A326 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS 1249-P Diabetes Care for Emerging Young Adults With Type 1 Diabetes: A National Survey of Adult Endocrinologists in the United States KATHARINE C. GARVEY, JONATHAN A. FINKELSTEIN, PETER FORBES, LORI M. LAFFEL, Boston, MA Previous work has addressed challenges reported by pediatric providers caring for older teens and young adults (YA) with type 1 diabetes (T1D), yet little is known about the barriers faced by adult endocrinologists (AEs) accepting T1D patients in this vulnerable developmental stage. To assess AE experiences, resources and barriers in caring for this population, we fielded a web-based survey to AEs in the AMA Masterfile. We received responses from 536 of 4214 AEs (13% response rate) representing 47 states. Respondents were 57% male and 79% Caucasian; 64% had practiced >10 years and 42% worked at an academic center. Most AEs (72%) felt that reviewing pediatric records of transitioning YA with T1D prior to the first visit was important/very important for patient care, but only 36% often/ always had the opportunity to do so. While most AEs reported access to diabetes educators (94%) and dietitians (95%), less than half (42%) reported easy access to mental health care (MH) and 54% desired more access to MH for YA patients. Access to MH for YA with T1D was more likely to be reported by AEs in academic settings (p = 0.0006). In a series of patient scenarios, controlling for academic setting and years in practice, AEs without easy access to MH were 2 to 6 times more likely to report barriers to T1D management for YA with depression, eating disorders, substance abuse, and developmental disability (all p<0.0001). In contrast, AEs reported minimal barriers to YA management for glucose sensor use, obesity, recurrent DKA, elite athletics, and fear of hypoglycemia. To improve care, AEs endorsed support groups for YA with T1D (82%), improved reimbursement (76%), and continuing medical education (CME) about YA behaviors (57%). Our findings underscore the importance of enhanced information transfer between pediatric and adult providers and increased MH access for YA with T1D, particularly in non-academic settings. Provider CME modules should focus on behavioral issues specific to YA with T1D. KALPITA MAJUMDAR, MARIA BARNARD, London, United Kingdom Secondary care diabetes specialists in the United Kingdom take many referrals from primary care, and provide expertise in complex cases. We studied referrals into our diabetes clinic to assess our care objectively, and to study specialist interventions offered to these patients. In a retrospective survey, we studied the source and reason for referrals to secondary care over 12 months. Each patient was followed over 18 months. Data was collected on documentation of 9 care processes and changes in three measurable targets (blood pressure, cholesterol, glycated haemoglobin) as set out in the UK National Diabetes Audit 2010-11, and referrals to allied diabetes professionals or other experts. A total of 355 patients were referred over the study period, excluding patients who can only be managed in specialist settings. 71% patients were referred by their primary care physician. Reasons for referrals were for a combination of poor glycemic control (201), type 1 diabetes (99), significant or worsening complications (105), and associated co-morbidities (91). We evaluated clinical care on 289 patients, excluding non-attenders. Data was collected for age, gender, type of diabetes. After analysing 100 cases, 79% patients had 9 care processes documented. A positive impact on at least one of three measurable targets was noted in 69% patients. Medications were started/altered in 86%, and 78% were seen by at least one other allied professional in the diabetes team. Referrals for other specialist opinion, structured education and/or psychological input were made for 45% patients. In times of economic austerity, with burgeoning numbers of diabetes patients, models of diabetes care are under greater scrutiny to assess their impact. Our study proves that patients are being referred for multiple and complex needs that cannot be comprehensively met in primary care. These patients require high quality diabetes expertise and multiple reviews, along with appropriate contributions from allied and expert specialists to optimise their metabolic milieu. Supported by: William Randolph Hearst Foundation 1250-P Effects of Insulin Adherence and Delivery Device on Real World Outcomes Among Patients With Type 2 Diabetes Mellitus (T2DM) RAJEEV AYYAGARI, WENHUI WEI, DAVID CHENG, CHUNSHEN PAN, ERIC Q. WU, Boston, MA, Bridgewater, NJ, Boca Raton, FL Treatment adherence affects real world insulin effectiveness and may itself be impacted by the delivery device (pen or vial/syringe). This study aims to understand the combined effects of insulin adherence and device on real world health outcomes. A national managed care administrative database was used to identify T2DM adults initiating insulin, who had continuous health plan insurance ≥6 months prior to initiation (baseline) and ≥1 y after. Outcomes included 1 y follow-up A1C reduction, hospitalization rate, total health care costs, and pharmacy costs. To account for the time-varying relationship between adherence (ie, all days covered in each calendar quarter) and outcomes, marginal structural generalized linear models were used to estimate joint causal effect of adherence and device. Mean outcomes were predicted for different combinations of adherence and device. Among the 13,428 patients (mean age 54 y, 46% female, baseline A1C 9.3%), adherent pen users had greater A1C and hospitalization rate reductions and higher pharmacy costs vs non-adherent vial users. Pen use and adherent vial use decreased hospitalization rate and increased pharmacy but not total costs (Figure). Our study suggests that adherence and pen use both have beneficial effects on patients’ real world outcomes, with the most favorable effects attributed to adherent pen use. Naturalistic studies of insulin therapy are needed to inform optimization of therapy for T2DM. ADA-Funded Research & Supported by: Whittington Hospital 1252-P Utilization of Electronic Health Records (EHRs) to Evaluate Prediabetes and Diabetes Screening Guidelines VANITA R. ARODA, SAMEER DESALE, CHRIS ST CLAIR, JOHN JANAS, DANIEL MERENSTEIN, PETER BASCH, ROBERT E. RATNER, NAWAR SHARA, Hyattsville, MD, Concord, NH, Washington DC, Columbia, MD, Alexandria, VA Many guidelines advocate early screening of individuals at risk of type 2 diabetes. Guidelines from the United States Preventive Services Task Force (USPSTF) and the ADA, however, differ in the recommended screening populations. The objective of this analysis was to evaluate the extent to which these two guidelines have been translated into practice, and the yield of this screening. We examined EHR data from 46,779 adult patients seen from 5/1/11-5/1/12 across 70 adult primary care practices within MedStar Health, a large healthcare system in the Maryland-DC region. Based on data within the EHR, an estimated 7.8% of this adult cohort aged 20 years or older carried a diagnosis consistent with prediabetes. Of the 46,779 patients, USPSTF guidelines identified 11768 (25%) with an indication for diabetes screening, of whom 4720 (40%) had evidence of screening by fasting glucose, HbA1c, or 2-h OGTT. When screened, the diagnostic yield, i.e. lab results consistent with prediabetes or diabetes, was For author disclosure information, see page 829. Guided Audio Tour poster A327 POSTERS 1251-P Secondary Care for Diabetes: Are We Worth It? Clinical Diabetes/ Therapeutics Supported by: Sanofi U.S., Inc. HEALTH CARE DELIVERY—ECONOMICS evident in 45% and 35%, respectively. 42% of those with labs consistent with prediabetes had prediabetes listed in the active problem list within the EHR. By contrast, 30,646 (66%) of our cohort met ADA guideline indications for screening, of which 11,530 (38% of these) had evidence of screening, with diagnostic yields of 47% for prediabetes and 33% for diabetes, identifying > 2.5 fold as many patients with prediabetes as captured by USPSTF guidelines (5339 versus 2073). Still, only 39% of those with labs consistent with prediabetes had prediabetes listed in the active problem list within the EHR. In sum, significant gaps in diabetes screening and prevention are in determining which individuals are at risk (USPSTF vs ADA), screening these individuals identified as at risk, and then consistently using the screening data to make an appropriate diagnosis of prediabetes. These findings can inform further efforts to improve earlier diagnosis and treatment of patients with prediabetes and diabetes. Medication Classes POSTERS Patients without DM** p-Value n (%) Mean MPR (Median) Oral antidiabetic MPR at 1 yr 11,452 (50.7) 0.74 (0.86) MPR at 3 yrs 4,873 (53.2) 0.65 (0.75) Insulin MPR at 1 yr 5,111 (22.6) 0.56 (0.59) MPR at 3 yrs 1,929 (21.1) 0.50 (0.53) Antihypertensives < 0.0001 MPR at 1 yr 16,059 (71.1) 0.82 (0.93) 31,260 (64.7) 0.78 (0.91) < 0.0001 MPR at 3 yrs 6,614 (72.2) 0.73 (0.86) 12,807 (64.8) 0.68 (0.80) Statins 0.0386 MPR at 1 yr 11,735 (52.0) 0.70 (0.81) 24,289 (50.2) 0.71 (0.83) 0.8759 MPR at 3 yrs 4,712 (51.4) 0.59 (0.66) 9,863 (50.0) 0.60 (0.68) Antiplatelets < 0.0001 MPR at 1 yr 9,263 (41.0) 0.70 (0.85) 19,332 (39.9) 0.73 (0.89) 0.2977 MPR at 3 yrs 3,891 (42.5) 0.55 (0.61) 8,039 (40.7) 0.56 (0.61) Anticoagulants 0.0096 MPR at 1 yr 2,259 (10.0) 0.55 (0.59) 3,704 (7.7) 0.53 (0.56) 0.1740 MPR at 3 yrs 888 (9.7) 0.42 (0.38) 1,507 (7.7) 0.40 (0.33) * Patients with T2DM are defined as having a) two or more claims of T2DM at least 30 days apart; and/or b) ≥ 1 claim of T2DM and ≥ 1 claim of oral or injectable antidiabetics; and/or c) ≥ 2 prescriptions of oral antidiabetics or GLP-1 agonist medications during the study period. ** Patients without DM are defined as not meeting criteria for T2DM and having no claim for type 1 DM during the study period. 1253-P Clinical Diabetes/ Therapeutics Patients with T2DM* n (%) Mean MPR (Median) Sub-Optimal Medication Initiation and Adherence Following Acute Coronary Syndrome in Real-World Patients With and Without Diabetes Mellitus: 2006–2011 VANESSA S. REDDY, RAKESH LUTHRA, YAPING XU, KENNETH WILHELM, THOMAS P. POWER, MAXINE D. FISHER, MARK J. CZIRAKY, San Francisco, CA, Wilmington, DE, Deerfield, IL Medication adherence is critical for optimal management of chronic disorders such as type 2 diabetes mellitus (T2DM) and coronary artery disease. This study assessed mean drug initiation and medication possession ratios (MPRs) for guideline-recommended acute coronary syndrome (ACS) and antidiabetic treatments 1 and 3 years following hospitalization for ACS, by presence or absence of T2DM. ACS patients hospitalized from Jan 2006 to Sept 2011 were identified from the HealthCore Integrated Research Database using pharmacy claims data. Of 120,398 ACS patients identified, 32% had T2DM. Patients with T2DM were on average older and had more baseline comorbidities than patients without DM (Table). Following the index ACS event, only 51% of patients with T2DM filled a prescription for an oral antidiabetic, 23% insulin, 52% statin, 71% antihypertensive (antiHTN) and 10% anticoagulant medications (Table). After 3 years of followup, patients with T2DM were most adherent to anti-HTN (MPR = 0.73) and least adherent to anticoagulants (MPR = 0.42) (Table). Overall, patients with and without T2DM had comparable MPRs across assessed drug classes; however, adherence continued to decline at 3 years in both cohorts. This analysis identifies critical areas for improving both initiation and long-term adherence to guideline-recommended treatments following an ACS event. 1254-P Vulvovaginal Candidiasis (VVC) and Urinary Tract Infection (UTI) in Women With and Without Type 2 Diabetes Mellitus (T2DM) ROBERT STELLHORN, WING CHOW, SILAS MARTIN, MANEESHA MEHRA, Raritan, NJ Women with T2DM have an increased risk of VVC and UTI. This analysis compared the occurrence of VVC and UTI and the associated treatment costs between women with T2DM and women without T2DM using the Pharmetrics commercial claims database. Data were examined for a 2-year period from 1/1/2010 to 12/31/2011 using a 10% sample. A cohort of women aged ≥25 years with a diagnosis of T2DM and/or use of antihyperglycemic agents before 1/1/2010 (N = 44,371) and a matched control cohort (N = 42,471) were identified. Diagnosis for VVC (ICD-9 codes: 112.1, 616.10, 616.11, 627.3) and UTI (ICD-9 codes: 590.10, 590.11, 590.80, 590.81, 595.0, 595.9, 599.0) and prescriptions for antifungal or antibiotic drugs in these cohorts were assessed over the 2-year period. In the T2DM and control cohorts, 9.7% and 9.1% of women, respectively, were diagnosed with VVC. Among these women, 36.4% (1,559/4,287) used prescription antifungals in the T2DM cohort compared with 36.3% (1,402/3,863) in the control cohort. For women diagnosed with VVC, median prescription antifungal treatment cost per person over the 2-year period was $10.00 for the T2DM cohort and $8.02 for the control cohort. Similar trends were observed between the T2DM and control cohorts when the occurrence of VVC and antifungal cost were assessed in women who had either a VVC diagnosis or an antifungal drug prescription. A diagnosis of UTI was made for 18.7% and 12.3% of women in the T2DM and control cohorts, respectively. Among women with a UTI diagnosis, antibiotics were used by 48.4% (4,009/8,276) of those in the T2DM cohort compared with 50.3% (2,629/5,224) in the control cohort. Median antibiotic treatment cost per person over 2 years was $8.00 for the T2DM cohort and $7.30 for the control cohort. In summary, similar occurrences of diagnosed VVC and higher occurrences of diagnosed UTI were observed in women with T2DM compared with women without T2DM, with minimal antifungal or antibiotic treatment costs that were comparable among the cohorts. Table: Medication Possession Ratios by Diabetes Status, 1 and 3 Years Patients with T2DM* Patients without DM** p-Value Baseline Mean or % Median Mean or % Median Characteristics Age, years 68.2 69.0 65.6 64.0 < 0.0001 Male 58.9% 58.7% 0.5337 Follow-Up Rate, years 1.8 1.4 1.8 1.4 0.9679 Hypertension 74.7% 52.5% < 0.0001 Dyslipidemia 57.5% 40.7% < 0.0001 Heart Failure 24.3% 11.5% < 0.0001 Renal Failure 16.5% 4.8% < 0.0001 & For author disclosure information, see page 829. A328 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS 1255-P 1257-P Medication Costs for Patients With Type 2 Diabetes Mellitus: How They Vary With Body Mass Index Status, Age, Gender and Co-Morbidity Analysis of a Large Payer Claims Database Suggests Elderly Patients With Diabetes are Under-Treated This retrospective cross-sectional study utilized medical, pharmacy and laboratory claims from a large Medicare Advantage with Prescription Drug Coverage (MAPD) payer. MAPD members (mean age 72 years) diagnosed with T2DM between 2009 and 2011 were eligible for inclusion. A 12-month baseline period before the first A1c value (the index date) was evaluated for demographic and clinical differences. Antidiabetic therapy was evaluated during a 6-month post-index period. The study population was stratified into three cohorts based on index A1c value: controlled (≤7%), uncontrolled (>7% and ≤9%), and severely uncontrolled (>9%). Despite elevated A1c values (>7%) 7-12% of patients were not on antidiabetic therapy during the post-index period. Metformin and sulfonylureas were the oral antidiabetics (OADs) most frequently used as monotherapy. In cohorts with A1c >7%, insulin use was higher while biguanide and sulfonylurea use was lower. Of those on combination therapy, the majority were on 2 or more OADs and higher injectable use was observed in the severely uncontrolled cohort. Metformin was included in 62% of the combination regimens with ‘Metformin + sulfonylurea’ being the most common. This study suggests suboptimal treatment of those not in control (A1c >7%). Many patients classified as uncontrolled based on A1c received only monotherapy postindex. Opportunities exist for treatment modification within this population to achieve tighter glycemic control. Proportion of Elderly Patients Treated with Various Levels of Antidiabetic Therapy Therapy A1c ≤7% A1c >7% A1c >9% P-value and ≤9% N=100,023 N=34,766 N=8,551 No antidiabetic prescription 48,072 48.1% 4,327 12.4% 631 7.4% <0.01 claims Monotherapy 36,602 36.6% 14,179 40.8% 3,031 35.4% <0.01 Biguanide 23,410 23.4% 7,200 20.7% 1,187 13.9% Sulfonylurea 10,279 10.3% 5,597 16.1% 1,051 12.3% Insulin 2,026 2.0% 2,981 8.6% 1,657 19.4% Combination Therapy 15,349 15.3% 16,260 46.8% 4,889 57.2% <0.01 Supported by: National University of Ireland 1256-P The Burden of Hospitalizations due to Hypoglycemia in Belgium MARK LAMOTTE, PIERRE CHEVALIER, TOM VANDEBROUCK, Vilvoorde, Belgium, Brussels, Belgium Strict control of glycemia in diabetic patients is impeded by the risk of hypoglycemia, especially when insulin is used. However, not all insulins have the same risk of hypoglycemia. The aim of this study was to study the cost of hypoglycemia requiring hospitalization (at least one overnight stay) in type 1 and 2 diabetes in Belgium. A retrospective database study was conducted using the longitudinal IMS Hospital Disease Database (HDD 2009), covering 34.3% of Belgian hospital beds. In this database hypoglycemic events can be identified based on ICD-9 codes (Diabetes hypoglycemic coma/insulin coma: 250.3; Diabetic hypoglycemia/hypoglycemic shock: 250.8; Anti-diabetic agent poisoning and adverse events: E932.3, 962.3). Number of stays for hypoglycemia, average length of stay (LOS) and on intensive care (ICU), in hospital mortality, rehospitalizations for hypoglycemia and costs were estimated using SAS software. The number of patients with diabetes in Belgium was estimated to be 600,000. In the HDD 3,674 patients were hospitalized for hypoglycemia. In 2009, each patient had on average 1.16 hospitalizations. Extrapolated to Belgium this means 10,806 patients hospitalized for hypoglycemia. Eightythree percent of those stays were in patients older than 55. The average LOS was 17.32 days (with 0.21 days on ICU) strongly driven by stays in the population >55 years (12 vs. 18 days; p<0.0001). Mortality during hospitalization was 5.83% driven by the age group older than 65 (2.89% vs. 7.60%; p<0.0001). The average cost of a hospitalization was €9,283 [€9,037-9,529]. Hospitalizations for hypoglycemia are expensive and result in an important in hospital mortality. Especially in the older age (>55), the costs for the health care payer are high due to extensive LOS. Interventions that can help reduce the risk of hypoglycemia, and as a consequence the burden on hospitals and society, without compromising glycemic control will help further improve diabetes management. Supported by: Novo Nordisk, Inc.; Humana, Inc. 1258-P Treatment Intensification after Early Failure of Metformin or Sulfonylurea among VA Patients With Type 2 Diabetes LI WANG, WENHUI WEI, MEHUL DALAL, ONUR BASER, Dallas, TX, Bridgewater, NJ, Ann Arbor, MI Randomized trials have documented the long-term cardiovascular (CV) benefits of treatment intensification in patients with type 2 diabetes mellitus (T2DM). However, there is little real world data on the factors affecting intensification choices among patients failing metformin (MET) or sulfonylurea (SU) soon after initiation. Using Veterans Health Administration (VHA) health care claims data from October 2005 to September 2011, a retrospective cohort study was conducted among T2DM patients with early treatment failure of MET or SU (defined as having A1C ≥ 9.0% or ≥ 2 A1C ≥ 7.0% dated > 90 days apart within 1 y of initiation). Factors associated with the time from treatment failure to intensification, and choices of intensification (OAD or insulin) were identified. A total of 27,556 patients were included (68.3% MET and 31.7% SU; 95.8% male; 55.4% white; mean age 61 y; A1C 9.0%; BMI 34.9 kg/m2; median follow-up 29 months). After treatment failure, 64.4% (n=17,742) intensified their treatment during follow-up (median time: 4 months; 79.2% with a 2nd OAD, 9.9% basal insulin, and 9.2% basal-bolus/premix insulin), with 36.9% (n=10,174) intensifying within 1 y. Independent factors associated with intensification included: white race; younger age; later index year; higher A1C; higher BMI; SU use; shorter time from initiation to failure; diabetes education; comorbid neuropathy, retinopathy, or mental illness; and absence of chronic kidney disease (CKD) (all P < 0.05). Patients intensifying with insulin within 1 y (n=1,865) were more likely to have initiated with SU, and were associated with a higher comorbidity burden, CV disease, CKD, and baseline hypoglycemic events (all P < 0.01). This study showed that among VA patients who failed MET or SU early, only 1/3 intensified their treatment within a year, and significant differences Supported by: Novo Nordisk A/S ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A329 POSTERS While type 2 diabetes is characterised by insulin resistance and relative beta cell failure, there is heterogeneity in the degree to which excess adiposity contributes to the individual diabetes phenotype, reflected in different treatment modalities for different patients. We sought to determine whether the degree of overweight or obesity, measured with body mass index (BMI), influences the cost of drug therapy for type 2 diabetes patients. We identified a subgroup of 185 adult type 2 diabetes patients attending our university hospital-based diabetes clinic, for whom detailed information about drug therapy and comorbidities (obesity, hypertension, dyslipidaemia) was available. Data relating to the cost of medications from the Medical Information Management System (MIMS) Ireland and the Health Service Executive Primary care Reimbursement Service (PCRS) were used to generate a Net Present Value (NPV) of the drug cost, discounting the annual cost over the average life expectancy of the patient group. There was a positive association between BMI and medication costs in linear regression analyses. Mean estimated annual costs were €615 higher in patients with a BMI greater than 35 kg m-2 compared to those with BMI less than 30 kg m-2 (p less than 0.01). Of note, the peak cost occurred in the BMI range 35-40 kg m-2 with an unanticipated decline in costs in those with BMI greater than 40 kg m-2. Overall, diabetic patients with obesity, hypertension and dyslipidaemia had an estimated annual medication cost that was €418 higher than those diabetic patients without any such comorbidity (p less than 0.05). These findings suggest that the substantial health economic costs associated with the rising prevalence of type 2 diabetes are differential with respect to the BMI status of affected individuals. In considering the risks and benefits of interventions such as bariatric surgery in this patient group, these additional medication costs need to be borne in mind. Clinical Diabetes/ Therapeutics S. LANE SLABAUGH, YIHUA XU, JANE STACY, NICK PATEL, YUNUS A. MEAH, JEAN C. BALTZ, JONATHAN BOUCHARD, Louisville, KY, Princeton, NJ MICHELLE QUEALLY, CIARAN O’NEILL, FRANCIS M. FINUCANE, Galway, Ireland HEALTH CARE DELIVERY—ECONOMICS existed between those who intensified with OAD vs insulin. This needs to be carefully evaluated when conducting outcomes studies in this population. 1261-P Successful Implementation of a T2D Prevention Program in a Community Health Center Setting Supported by: Sanofi U.S., Inc. ANNE CAMP, WILLIAM V. TAMBORLANE, GRACE KIM, ELIZABETH MAGENHEIMER, ABMARIDEL MONTOSA, New Haven, CT 1259-P The Diabetes Prevention Program (DPP) showed that an intensive lifestyle intervention (ILI) reduces the development of T2DM in adults with prediabetes. However these results have not been widely translated into practice, especially in Community Health Centers (CHCs) that care for poor minority populations at high risk for T2D. This study evaluated a DPP lifestyle intervention adapted for a CHC in an inner-city Hispanic neighborhood in which 68% of women are obese. We randomly assigned 109 women (92% Hispanic, age 42.2 ± 10.5 yr) with prediabetes determined by OGTT to a 14 week group-based ILI or to usual care control group at Fair Haven CHC. Primary outcome was the difference in changes in body weight after 1 year. Eighty-nine subjects (82%) completed 1 year follow-up. Age, weight, BMI, lipids, blood pressure and OGTT were matched in the 2 groups at the start of the study. After 1 year, the ILI group experienced significantly greater decreases in body weight compared with controls (-4.3 vs +0.2 kg; P<0.0001), % body weight (-5.1% vs +0.1% ; P<0.0001), BMI (-1.77 vs -0.02 kg/m2; P<0.0001) and triglycerides (-21 vs +35 mg/dL; P=0.036). Changes in secondary outcomes are shown in the Table. Despite limited resources, a DPP-style ILI can be effectively delivered in a CHC setting that serves poor minority patients at high risk for T2D. CHCs are a promising health care delivery system to disseminate diabetes prevention programs among disadvantaged groups. Correlation between Hypoglycemia and Adherence With Oral Antihyperglycemic Agents in Patients With Type 2 Diabetes POSTERS Clinical Diabetes/ Therapeutics JANICE LOPEZ, ROBERT BAILEY, KATHY ANNUNZIATA, SILAS MARTIN, Raritan, NJ, Princeton, NJ Patients with type 2 diabetes mellitus (T2DM) on oral antihyperglycemic agents (AHAs) continue to be impacted by hypoglycemia (HG), a known adverse event of medications used to control blood glucose. In this study, we explore the association between HG and medication adherence as adherence is important for maintaining glucose control, avoiding complications, and improving long-term outcomes. The 2011 US National Health and Wellness Survey recruited a nationally representative sample of 75,000 adults (aged ≥18 y) using a self-administered survey. Participants reported their experiences with HG (occurring ever and/ or in the past 3 months) and answered questions on the Morisky Medication Adherence Scale, which was used to determine adherence to diabetes medications. Some patients also reported recent HbA1c results. A total of 4,516 participants reported a diagnosis of T2DM and use of oral AHAs only (no insulin, no GLP-1 agonists). The mean age was 59.8 ± 13.2 y and 57% were male. Overall, 47.1% ever experienced HG and 23.5% reported recent HG (past 3 months). Of the T2DM patients who reported HG (ever or in the past 3 months), 58.8% were adherent versus patients with no HG who reported 69.1% adherence. 24.8% of T2DM patients with HbA1c <7% experienced HG in the past 3 months, compared to 29.3% of T2DM patients with HbA1c ≥7%. Of those with HbA1c <7%, those who had not experienced HG reported better medication adherence than those with HG (those with HbA1c ≥7% had similar adherence for both HG and no HG). Despite newer AHA treatment options, a large proportion of T2DM patients report experiencing HG. Those who reported a HG episode were more likely to report poorer medication adherence than those who did not experience HG. Further investigation is warranted on the types of AHA used and the relationship of HG and adherence. Outcome Weight (kg) % Weight Change BMI (kg/m2) Intervention Gp Control Gp Treatment (n=44) (n=45) Effect -4.3 (-5.4,-3.1) +0.2 (-0.7,1.0) -4.5 (-6.0,-2.8) -5.1(-6.3,-3.6) +0.1(-1.0, 1.2) -5.2(-6.8,-3.3) -1.77 (-2.25, -1.29) -0.02 (-0.36, 0.33) -1.75 (-2.34,-1.16) Fasting glucose (mg/dL) -4.4(-7.6,-1.2) -4.1(-7.8,0.5) 2 hr glucose (mg/dL) -19 (-30,-7) -11 (-25,3) Triglycerides (mg/dL) -21 (-37,-6) 35 (-16,85) Data are means (95% Confidence Intervals) 1260-P Resource Use of Type 2 Diabetes Mellitus Patients Following Initiation With Saxagliptin or Sitagliptin JASMINA IVANOVA, ANNA KALTENBOECK, KELLY BELL, NINA THOMAS, FRANCES SCHWIEP, YANA YUSHKINA, HOWARD BIRNBAUM, New York, NY, Boston, MA, Menlo Park, CA P-value <0.0001 <0.0001 <0.0001 -0.3 (-4.8,4.9) 0.88 -7 (-28,13) -56 (-105,-6) 0.40 0.036 Supported by: Donaghue Foundation 1262-P This analysis compared healthcare resource use by type 2 diabetes mellitus (T2DM) patients treated with saxagliptin or sitagliptin in the 6 months following treatment initiation. Patients with T2DM treated with saxagliptin (N=13,929) or sitagliptin (N=36,813) in 2010 or later were identified in the Truven MarketScan database of commercially insured beneficiaries. Patients were required to have no saxagliptin or sitagliptin claims in the 6 months pre-initiation, at least 6 months of continuous eligibility pre & post initiation and be 18 or older. Demographics and comorbidities were evaluated in the baseline period. All-cause & diabetes-related resource use was evaluated in the study period. Chi-squared tests assessed differences in resource use. Logistic regression assessed treatment effects on study period hospitalization, adjusting for baseline patient characteristics & hospitalization. Saxagliptin patients compared with sitagliptin patients were younger (58 vs. 60 yrs; p<0.001) with similar gender distribution (47.0% vs. 47.1% female; p=0.852). Saxagliptin patients had a lower mean CCI (Charlson Comorbidity Index) (0.48 vs. 0.65; p<0.001), but higher rates of dyslipidemia & hypertension (44.8% vs. 41.9%, & 51.8% vs. 49.8%). During the study period, they experienced lower rates of all-cause hospitalization, ED visits, & other medical visits (7.2% vs. 10.6%, 14.1% vs. 17.5%, and 67.0% vs. 70.4%; p<0.001). Saxagliptin patients were also less likely to have diabetes-related hospitalizations (4.0% vs. 6.6%; p<0.001). Odds of all-cause or diabetes related hospitalizations were lower for saxagliptin patients after adjusting for baseline characteristics. (OR 0.80 & 0.74, respectively, p<0.001). Saxagliptin patients had less evidence of acute comorbidities, higher rates of cardiometabolic risk factors & following treatment initiation were less likely to experience hospitalizations. What Is the Delay in Insulin Therapy Initiation in Patients With T2DM Not Responding to Oral Glucose Lowering Agents? RUTH MAST, DANIELLE JANSEN, AMBER AWA VAN DER HEIJDEN, JACQUELINE M. DEKKER, JACQUELINE G. HUGTENBURG, ROBERT J. HEINE, GIEL NIJPELS, Amsterdam, Netherlands, Indianapolis, IN What is the delay in insulin therapy initiation in patients with T2DM not responding to oral glucose lowering agents? Available data suggest that insulin therapy is initiated late in patients failing to respond adequately to oral glucose lowering agents (OGA). Determinants of this delay are unknown. The aim of this study was to assess the time to insulin initiation and HbA1c value in T2DM patients inadequately controlled with OGA. Also, the influence of BMI, diabetes duration, and general practitioner (GP) practice on timing of insulin initiation was investigated. All 3772 T2DM patients who entered a regional diabetes care system (DCS) in the Netherlands between 1998 and 2011, aged 40 years and over, with a follow-up of a minimum of 2 years, with at least once a HbA1c ≥ 58 mmol/mol (7,5%) whilst on OGA. The time to insulin initiation was calculated (delay time). GP practices were categorized in three types; having a practice nurse (PN) and initiating insulin; or with PN but initiating insulin by DCS; or without PN and initiating insulin by DCS. A Cox proportional hazard model was used to determine the influence of BMI, diabetes duration, and GP practice on insulin initiation. The influence of BMI, diabetes duration, and GP practice on the delay time were analyzed with regression analyses. All adjusted for age and gender. The median time to insulin initiation was 5.5 yr (range 1- 14.2 yr). The mean HbA1c level preceding initiation of insulin was 66 mmol/mol (8.2%). Only diabetes duration showed a significant association with insulin initiation. With a diabetes duration of 0-2 yr as reference category, the hazard ratios were respectively for 2-5 years 1.24 (95 % CI 1.05-1.5) and for 5-10 years 1.27 (95% CI 1.058-1.529) at insulin initiation. Supported by: Bristol-Myers Squibb & For author disclosure information, see page 829. A330 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS In conclusion, despite current guidelines that advice insulin therapy in patients not responding adequately to OGA the mean time to insulin initiation was 5.5 years. Only diabetes duration showed a weak association with the timing of insulin initiation. 1263-P Determination of Normal Inpatient Glucose Values in the National VA Population Supported by: Health Workforce (New Zealand) 1265-P Medical Costs Attributed to Arthritis among Persons With Diabetes RUI LI, MIRIAM CISTERNAS, SUNDAR SHRESTHA, JUSTIN TROGDON, LOUISE MURPHY, Atlanta, GA, San Diego, CA, Research Triangle Park, NC Nearly half of US population 18 years and older with diabetes also have arthritis. Unlike other common comorbidities of diabetes, such as cardiovascular disease, little is known about the economic impact of arthritis among people with diabetes. We estimated arthritis-attributable medical cost among US adults with diabetes using 2009 and 2010 Medical Expenditure Panel Survey data (MEPS); MEPS represents the US community (non-institutionalized civilian) population. Mean medical costs (total and by service [inpatient care, outpatient care, prescription drugs, and other services]) attributable to arthritis among persons with diabetes were estimated using regressions that controlled for demographic characteristics, socioeconomic status, insurance status, and eight common comorbidities of diabetes. All costs are in 2010 US dollars. In the MEPS population, costs attributable to treating arthritis accounted for 16.0% of total medical cost among persons with diabetes. Mean costs among people with both diabetes and arthritis were consistently greater than among those with diabetes only: excess total medical was $2,897 (22.8% more); inpatient care was $620 (17.1% more); outpatient care was $1025 (43.0% more), prescription drugs were $849 (32.1% more), and other services were $446 (38.7% more). Considering the high prevalence of arthritis among people with diabetes and substantial excess arthritis-attributable medical costs among persons with diabetes, diabetes control programs should consider addressing arthritis as an integral component of control strategies; meeting physical activity recommendations and participating in chronic disease self-management education courses are evidence-based strategies for managing both conditions. Supported by: U.S. Dept. of Veterans Affairs 1264-P Diabetes Nurse Specialist Prescribing: The New Zealand Experience HELEN J. SNELL, TIMOTHY F. CUNDY, PAUL L. DRURY, CLAIRE BUDGE, Palmerston North, New Zealand, Auckland, New Zealand, Nelson, New Zealand In New Zealand (NZ) prescribing has resided with medical or nurse practitioners. In 2011 a new regulation in the Medicines Act enabled Diabetes Nurse Specialists (DNS) to prescribe under supervision. The NZ Society for the Study of Diabetes (NZSSD) tested safety and effectiveness over 6 months in 4 demonstration sites. Twelve DNSs made 3396 prescribing decisions for 1274 patients from a range of diabetes and cardiovascular medicines. Safety and effectiveness was determined via a previously presented external evaluation. In 2012 NZSSD undertook a follow-up study to examine continuing safety and efficacy. Clinical data was analysed for a subgroup of patients from the 2011 project using SPSS v19. The data was added to that collected during the 2011 project to enable a comparison over 3 time points; 2011 project baseline and endpoint and 2012 follow up. Data was collected on 205 patients, 49% male, mean age 57 years, 58% NZ European, 31.2% Maori and Pacific people, 11% other. Most (74%) had Type 2 diabetes. HbA1c improved from 81.8 to 72.8 mmol/mol and systolic blood pressure improved from 132.1 to 123.5 mmHg. Table 1 provides a summary of the clinical data for all 205 patients at the three time points . This study demonstrated that DNS prescribing continues to be both safe and effective in the long term with clinical parameters remaining stable or improved. The incomplete nature of the data limited a direct longitudinal comparison but findings for this small set of patients were reassuring. ADA-Funded Research & 1266-P Cost-Effectiveness of “Team of 4” versus Standard of Care in Diabetes SWARMA VARMA, TONY AMOS, NICOLE T. ANSANI, Pittsburgh, PA ABC goal attainment (A1C <7%, blood pressure <130/80mm/Hg, and cholesterol (LDL) <100mg/dL) is difficult to achieve and complications are costly. We analyzed the economic impact of utilizing a “team of 4” approach in patients with diabetes. An outcomes evaluation analyzed the incidence of ABC goal attainment and cardiovascular outcomes (CVD) in a “team of 4” approach (defined as involvement & accountability of physician, staff, patient, and family member). These data were compared to standard of care (NHANES population statistics for patients with diabetes). 22% of “team of 4” patients achieved ABC goals vs 7% in standard of care (SOC); CVD was seen in 18% and 24%, respectively. Two decision tree analyses, using TreeAge software, were used to determine the cost-effectiveness relative to ABC goals or CVD from a managed care perspective over a 10 year period. Clinical inputs came from the outcomes evaluation for the “team of 4” and NHANES data for SOC. Costs were from the literature and discounted to 2012 U.S. dollars. Sensitivity analyses were performed for all inputs. Key assumptions are: outcome probabilities and costs from the literature represent practice; costs of administering care to “team of 4” is equal to SOC; and a 500,000 member health plan will have 50,000 patients with diabetes. For author disclosure information, see page 829. Guided Audio Tour poster A331 POSTERS Inpatient hyperglycemia is associated with poor outcomes, but the range of normal hospital glucose is not known. Since identifying a normal range could help determine targets for inpatient management, we evaluated hospital glucose levels in 23,019 veterans using the VA Informatics and Computing Infrastructure (VINCI) database - hospitalized between 2000 and 2009 for ≥3 days on medical/surgical services, with ≥1 primary care visit within 2 years before and annually for 3 years afterwards. 7,556 patients without use of the ICD-9 code 250.xx or a diabetes drug throughout this period were defined as normal (NL), and their glucose values were compared with those in 12,131 patients with diabetes throughout (DM PRIOR), 366 initially diagnosed during hospitalization (DM HOSP), and those diagnosed after discharge - 719 DM 1y, 568 DM 2y, 550 DM 3y, and 1,129 DM 4-5y. Patients had mean age 59.4 yr and BMI 35.8, and were 92% male, 74% white, 22% black, 4% other/unknown. Using three glucose values per patient (minimum, maximum, median) to offset variability in sample numbers, mean (sd) glucose values were: NL 114 mg/dl (40 mg/dl), DM PRIOR 169 (99), DM HOSP 265 (247), DM 1y 140 (59), DM 2y 125 (43), DM 3y 122 (41) and DM 4-5y 119 (44). NL group glucose values were significantly lower than those in the other groups (all p<0.05), but were higher when compared to random glucose levels (mean 93 mg/dl, sd 16, p<0.0001) in 979 outpatient adult volunteers with normal OGTTs. NL group 25th, 50th, 75th, 90th, and 95th percentile hospital glucoses were 94, 106, 125, 150, and 169 mg/dl, respectively. Conclusions: Patients not diagnosed with diabetes before, during, or within 3 years after hospitalization (“normal”) have inpatient glucose levels lower than those in patients who are diagnosed during that time period. Since the 95th percentile of “normal” hospital glucose is 169 mg/dl, screening for diabetes could be considered in patients with glucose values above this level, and keeping inpatient glucose below this level might be a reasonable goal. Clinical Diabetes/ Therapeutics MARY K. RHEE, SANDRA L. JACKSON, WENQIONG XUE, DARIN E. OLSON, DIANA BARB, J. SONYA HAW, ANNE TOMOLO, ARUN V. MOHAN, QI LONG, PHYLLIS WATSON-WILLIAMS, LAWRENCE S. PHILLIPS, Atlanta, GA, Decatur, GA HEALTH CARE DELIVERY—ECONOMICS SH was reported as primary or secondary diagnosis, respectively. The LoS was 5 days (interquartile range 2.5-10; Mean±SD: 7.79±8.70) in T1 diabetes and 7 days (interquartile range 4-11; Mean±SD: 9.13±8.94) in T2 diabetes. The costs increased with greater LoS (quartiles): from € 1,894 for LoS <5 days to € 4,393 for LoS >11 days. If the estimates referred to Region Puglia were applied to Italy (3,000,000 of patients with diabetes), we can estimate 15,000 hospitalizations/year for SH and a total expenditure of € 45,000,000/year. Results showed “team of 4” approach relative to ABC goals and CVD was ~ $180 and $480 less costly/patient vs SOC. Overall costs of “team of 4” vs SOC were $9213 and $9394, for ABC attainment and $7556 and $8037 for CVD, respectively. These data extrapolated (500,000 member plan) show a potential cost avoidance of the “team of 4” approach is $189,943 for ABC control and $24,044,500 for CVD costs over 10 years. Sensitivity analyses of outcome probabilities and costs showed minimal impact on each model. Adopting the “team of 4” approach can lead to significant cost savings while increasing ABC goal attainment and decreasing CVD. This data may serve as rationale for the development of education and adoption of systems to improve patient care processes. 1269-P Type I Diabetes in Medicare: Use of Services & Program Expenditures in 2010 1267-P CLAUDIA GRAHAM, DAN WALDO, San Diego, CA, Annandale, VA On-Demand On-Line Specialist Assistance from Hospital to General Practice: Telemedical Treatment of Diabetic Foot Ulcers Differences in utilization and program expenditures between controlled and uncontrolled Type 1 (T1) Medicare enrollees were analyzed. Analysis was done using the 2007-2010 Medicare Limited Data Set (LDS) Five-Percent Standard Analytic Files (5% SAFs). The 5% SAFs are comprised of fee-for-service claims for a sample of Medicare enrollees; sample inclusion is determined by the last 2 digits of the Health Insurance Claim Number, ensuring a random draw from the population. Count regressions and generalized linear model regressions were used to explore relationships between diabetes diagnosis and use of services or program expenditures. SAS and Stata software were used for analysis. Based on the LDS 5% SAF, in 2010 there were more than 500,000 persons with T1 in the Medicare population. Of those identified as T1, 229,300 (25%) had claims indicating that diabetes was uncontrolled. Diabetes identification and the state of control were determined from claims data, using ICD 9 code 250.XX. On average, enrollees with uncontrolled T1 have 5% higher expected medical costs per capita than those with controlled T1. This is based on a regression of the logarithm of enrollees’ hierarchical condition category (HCC) risk scores, controlling for age, sex, original reason for Medicare entitlement (age, disability, or end-stage renal disease), and dual eligibility for Medicaid. In terms of service use, enrollees with uncontrolled TI had a greater number of emergency department (ED) visits per capita than did enrollees with controlled TI , and almost twice the rate of ED visits for hypoglycemia (ICD9 code 250.8x). Overall, they had slightly more inpatient days of care and rates of inpatient admission, but relatively higher rates of admission corresponding with more hospital days per capita for hypoglycemia. Results suggest further characterization of hypoglycemia events in Medicare T1 is needed. POSTERS Clinical Diabetes/ Therapeutics BENJAMIN S. RASMUSSEN, JEANETT BRAENDSTRUP, MICHAEL VAEGGEMOSE, LARS D. MADSEN, KNUD YDERSTRAEDE, NIELS EJSKJAER, Odense, Denmark, Aarhus, Denmark, Holstebro, Denmark Objective: To assess the effects of implementation of on-demand on-line telemedical specialist assistance from hospitals to general practice in order to optimise the course of treatment of diabetic foot ulcers. Method: A priori conventional (historic) treatment courses for 30 individual patients were registered with regard to 1) time lines for diagnosis and treatment, 2) delay in first contact to foot ulcer specialist, 3) delay in first screening for diabetes late complications and contact to a diabetologist. After implementation of telemedicine all patients with diabetic foot ulcers in 6 general practices were consecutively recruited and clinical data and pictures documented in the internet-based communication system thereby time-stamping all events to be compared with the “historic” data. All parties (patients, relatives, staff in general practices and hospitals) are subjected to focus group interviews and questionnares describing satisfaction/dissatisfaction and advantages/disadvantage with the current treatment and treatment in this study. More than 40 patients will have been included from January 1st 2011 to 31st December 2012. Result: Preliminary results reveal 1) great patient/ relative satisfaction sparing the patient transport to hospital and allowing treatment in more comfortable and well known surroundings 2) satisfaction in general practice with on-demand assistance, but reluctance with regard to implementing the communication system 3) satisfaction among hospital staff as patients are stratified according to severity 4) overall diminished delays in diagnosis, treatment and first contact to specialists. Conclusion: On-demand on-line specialist assistance from hospitals to general practices is feasible. In our study treatment was optimized and patients and relatives were overall satisfied as was hospital staff. Barriers include reimbursement issues and apprehension towards yet another technology to be implemented. 1270-P An Intensive Discharge Intervention for Patients With Diabetes JEFFREY L. SCHNIPPER, JORGE CHIQUIE BORGES, NYRYAN V. NOLIDO, CHERLIE MAGNY-NORMILUS, STEPHANIE LABONVILLE, JUDY CHENG, BETH A. HIRNING, IWONA E. RYBAK, HEATHER DELL’ORFANO, MARGO HUDSON, MAUREEN MCQUEENEY, Boston, MA 1268-P Costs Related to Hospitalization for Hypoglycemic Episodes in Italy Patients with diabetes and comorbid cardiovascular disease may be at particularly high risk for post-discharge medication non-adherence and adverse outcomes. We randomly assigned inpatients on medicine and cardiology services with type 2 diabetes and comorbid cardiac disease, likely to be discharged home on insulin, to an intensive transitional intervention or to usual care. The intervention includes: 1) a nurse practitioner (NP) to coordinate care and patient education; 2) an inpatient pharmacist to perform intensive medication reconciliation and patient counseling; 3) standardized visiting nurse visits; 4) a post-discharge clinic visit staffed by the NP and a pharmacist; and 5) telemonitoring of point-of-care glucose readings. The primary end point is adherence to insulin 90 days after discharge based on pharmacy prescription refill data. To date, we have enrolled 125 of a planned 200 patients, including 37 with complete 90-day post-discharge pharmacy refill data. The mean medication possession ratio for all insulin types was 85.9% (SD 26.5) among intervention patients and 79.0% (SD 22.2) among usual care patients (difference 6.9%, 95% CI -10.4 to 24.2, p=0.23). Intervention patients had a 1.02 decrease in their 90-day A1c levels compared with a 0.83 decrease with usual care (p=0.61). No significant differences have been found to date in 30-day emergency department or readmission rates or rates of hypo- or severe hyperglycemia per monitored patient-day, although there was a trend towards increased self-reported hypoglycemia in the intervention arm (48.1% vs. 25% of patients with at least one episode, p=0.10). To date, the intervention is associated with a non-significant but promising trend towards increased adherence to insulin in the 90 days after discharge. MONICA FRANCIOSI, GIORGIA DE BERARDIS, FABIO PELLEGRINI, ANTONIO D’ETTORRE, LUCIA BISCEGLIA, FRANCESCO GIORGINO, VITO LEPORE, GIUSEPPE LUCISANO, FABIO ROBUSTO, ANTONIO NICOLUCCI, Santa Maria Imbaro, Italy, Bari, Italy The aim of the study was to evaluate the incidence and cost of hospitalization for severe hypoglycemia (SH) using the Administrative Databases related to 12 local Health Authorities of Region Puglia. We conducted a population-based cohort study using a record-linkage analysis of hospital discharge records, prescription databases, and civil registry, including data on 4.1 million citizens during the period 2002-2010. All subjects with pharmacologically treated diabetes were eligible regardless of gender, age and hypoglycemic treatment; 385,527 subjects (92% with T2 diabetes) were included. We identified 10,362 hospitalizations relative to 9,021 subjects. SH was reported as primary and secondary diagnosis in 40.46% and 59.54% of hospitalizations, respectively. Overall, in the period 2002-2010 the total cost related to hospitalization for SH was € 31,256,985. The average cost of hospitalization was of € 3,016. Despite a decrease in the number of events, the costs rose after 2006 and remained stable in the following years (up to € 3,374 in 2010). The average cost of hospitalization was € 2,326 if SH was reported as primary diagnosis and € 3,489 if SH was reported as secondary diagnosis. The greatest cost increase was observed in patients with T2 diabetes in Basal Oral Therapy (+56%). The median length of stay (LoS) for SH was 7 days (interquartile range 4-11 days; Mean±SD: 9.03±7.9): 6 days (interquartile range 3-10; Mean±SD: 7.53±7.35) and 8 days (interquartile range 4-13; Mean±SD: 10.13±9.99) when & For author disclosure information, see page 829. A332 Guided Audio Tour poster ADA-Funded Research HEALTH CARE DELIVERY—ECONOMICS The study demonstrates the feasibility of implementing a multi-disciplinary transitional intervention in medically complex patients with diabetes, and in the use of a promising method of determining insulin adherence using pharmacy refill data. Conclusions: Taking into account the costs of managing overt disease in Catalonia this primary care-led intensive lifestyle intervention was not only effective but also efficient in delaying progression to type2 diabetes 1271-P Effectiveness of an Inpatient Insulin Order Set Using Human Insulin in an Indian Health Service Hospital 1273-P Group Diabetes Visits to Support Self-Management Goals: A Model for Care in an Urban, Resource-Poor Community Supported by: Merck Company Foundation 1272-P Efficiency of a Primary Care-Led Lifestyle Intervention in Delaying Progression to Type 2 Diabetes Among High-Risk MediterraneanSpanish Individuals XAVIER COS, BERNARDO COSTA, JOAN J. CABRE, ORIOL SOLA-MORALES, RAMON SAGARRA, BERTA SUNYER, FRANCISCO BARRIO, JAAKKO TUOMILEHTO, THE DE-PLAN-CAT/PREDICE RESEARCH GROUP, Barcelona, Spain, Reus-Tarragona, Spain, Helsinki, Finland Objective: Transferring the findings of diabetes prevention research into primary healthcare was the objective of the Diabetes in Europe_ Prevention using Lifestyle, Physical Activity and Nutritional intervention project, developed in Catalonia (DE-PLAN-CAT). We report the 4-year costeffectiveness assessment of the intervention. Research Design and Methods: Incidence of diabetes, resource utilization, cost and quality of life (15D questionnaire) were collected alongside a reallife prospective cohort study conducted in 18 primary care centres. WhiteEuropean without diabetes aged 45-75 years (n=2,054) were screened using the FINDRISC and a 2-h OGTT. Where feasible, high-risk identified individuals who agreed to participate (n=552) were allocated sequentially to standard care (n=219), a group-based (n=230) or an individual level (n=103) intensive lifestyle DE-PLAN intervention. Analyses after 4.2-year median follow-up were performed based on the intention-to-treat principle. Results: The incidences of diabetes were 7.2 and 4.6 cases per 100 person-years in the standard care group and the intensive intervention group_4 and 3.6 cases in the group-based and the individual-based intensive intervention which represented 36.5% relative risk reduction (p<0.005) reaching to 49.3% in the individual intensive intervention group. The corresponding incremental cost-effectiveness ratios were 108€ (group) and 746€ (individual) per averted case of diabetes compared to standard care group. At study close quality of life was higher in the intensive lifestyle intervention group (0.93/0.91, p<0.01) being the incremental cost-utility ratio 3,243€ (4,250 USD) per quality-adjusted life-years gained. ADA-Funded Research & 1274-P A Randomized, Controlled, Open-Label Study to Evaluate the Effi cacy of Smart Phone Application Based Diabetes Self-Management Program in Patients With Type 2 Diabetes JONG DAI KIM, SEUNG HYUN YOO, YU MI HA, SUNG AH MIN, JIN SUN CHOI, YOO MI BAE, KYONG PIL MIN, SE EUN PARK, EUN JUNG RHEE, CHEOL YOUNG PARK, WON YOUNG LEE, KI WON OH, SUNG WOO PARK, Seoul, Republic of Korea, Sungnam, Republic of Korea The aim of this study is to evaluate the efficacy of smart phone application based diabetes self-management program (DSMP) for diabetic patients. A total 35 type 2 diabetic patients were enrolled (control group n=12 and intervention group n=23). And patients were followed during 6 months. The smart phone application based DSMP is interactive program based on analyzing data between self-monitoring of blood glucose, blood pressure, weight and life style pattern (meal, exercise, etc.). This program also provides alarm system and specific recommendation related to diet and exercise according to the results for improving self-management. For author disclosure information, see page 829. Guided Audio Tour poster A333 POSTERS Basal-bolus-correctional (BBC) insulin therapy is superior to “sliding scale” insulin alone in controlling hyperglycemia and improving outcomes in inpatients. An insulin order set for BBC using primarily analog insulins has been most studied for use outside the ICU. The Gallup Indian Medical Center recently made available a BBC order set, using human insulins due to resource constraints. This study assessed the effectiveness of the availability of this order set. Admissions to non-ICU services of adults with type 2 diabetes were reviewed for 4-month periods in 2011 (N=274) before and in 2012 (N=302) after availability of the order set and physician training. Comparing the cohorts, use of BBC therapy increased from 11% to 28% (p<.001) and use of any basal insulin increased from 50% to 63% (p<.001). The estimated mean blood glucose (by random regression adjusting for age, sex, initial glucose, A1C, service, and correlation of observations within individuals) for the first 9 days after admission showed that after day 2 the blood glucose levels in the 2012 cohort were lower than the 2011 cohort (Figure), with a difference for days 3-9 of -14.4 mg/dl (SE 6.2; p=.020). The incidence of hypoglycemia did not change. In conclusion, availability of a BBC human insulin order set in this rural community hospital was associated with improved physician ordering practices and improved control of hyperglycemia. A program of group diabetes visits (GDVs) in an urban, resource-poor community (Camden, NJ) was initiated to provide a team approach to ongoing patient-centered diabetes (DM) care and offer a model for sustaining behavioral change after diabetes self-management education (DSME). Patients who complete 6.5 hrs of DSME are offered a quarterly GDV with their primary care provider (PCP) and a certified diabetes educator (CDE). The CDE and a program assistant review charts prior to class to develop a DM Care Plan based on history, lab results and goals. The PCP and CDE discuss the patient’s clinical measure and behavior goals in advance of the GDV. Each two-hour program is divided into three components: DSME, clinical care, and an interactive Q&A. At the visit, the PCP completes the care plan for each patient. An endocrinologist attends the GDV for the first 6 months to model clinical decision making. After the program the CDE and PCP review the medical decisions, behavior goals and plan follow up care. Patient’s demographics, clinical measures (HbA1C, BP and lipids) are measured at baseline and every three months in the initial six months of the program. Thirty patients with previous DSME participated in the GDV. Complete pre and post A1c data was able to be collected for 16 patients (72% African American, 11% Hispanic, and 17% White; 50% male and 50% female; average age of 53.6 SD 12.4; average A1c of 9.10 SD 2.17). 14 GDV were held over a 2-year period. Attendance: 63% attended 2+ visits, and 36% attended 3+ visits. Change in HbA1c: Pre HbA1c 9.10 SD 2.17; Post HbA1c 7.98 SD 1.78 (p=.085). GDVs were investigated as a novel method for delivering DM care in an urban, resource-poor community. GDV allow for medical care, peer-to-peer support and continued emphasis on behavior change. Although the decrease in HbA1c was not statistically significant many patients had a reduction in their HbA1c. Combining education and medical components of a visit improves glycemic control in an underserved patient population. Clinical Diabetes/ Therapeutics JOSHUA VALGARDSON, MARICRUZ MERINO, JAMIE REDGRAVE, JAMES I. HUDSON, MARGO S. HUDSON, Gallup, NM, Boston, MA, Belmont, MA STEVEN KAUFMAN, FRANCINE GRABOWSKI, NADIA ALI, ANDREW KATZ, Cherry Hill, NJ, Camden, NJ HEALTH CARE DELIVERY—ECONOMICS SMBG requires effective patient-provider communication and patient education. Baseline characteristics of two groups were not different in age, sex, body mass index (BMI), and other metabolic parameters. At baseline, mean A1c of control group vs test group were 7.6± 0.5 vs 7.4±0.9%. After 6 months intervention using smart phone application based DSMP, there were increasing trend in control group (from 7.6±0.5 to 7.6±0.7%) and decreasing trend in intervention group (from 7.4±0.9 to 7.2±1.0%) for A1C. But it was not statistically different. Target A1C (below 7.0%) achievement were significantly different between two group (9.1% in control group vs 47.8% in intervention group, P<0.027). We divided the intervention group into good or poor user. The A1c level was significantly improved in good user-intervention group than poor user-intervention and control group. (-0.59±0.26, +0.10±0.13 and +0.07±0.18%, P<0.028). Smart phone application based DSMP was resulted in significant reduction of A1c during 6 months intervention, especially in the good user of this system. Supported by: R01DK-081796, R01DK-080726, R01DK-065664, R01HD-46113, P3DK092924, P30DK09 1277-P Do Food Costs Increase for Participants in a Diabetes Prevention Program? GERALD L. SCHAFER, THOMAS J. SONGER, M. KAYE KRAMER, VINCENT C. ARENA, RACHEL G. MILLER, KARL K. VANDERWOOD, ANDREA M. KRISKA, Pittsburgh, PA Given the high number of individuals at risk of developing type 2 diabetes, it is imperative that perceived obstacles to successful participation in diabetes prevention programs be thoroughly investigated. One perception is that food expenses increase for individuals making the healthy lifestyle changes recommended in these intervention efforts, i.e. it costs more to eat healthfully. This project assessed household food expenses over the course of a year in overweight adults (BMI ≥ 24kg/m2) with pre diabetes and/or the metabolic syndrome, participating in the Group Lifestyle Balance ™ Program, a diabetes prevention intervention adapted from the Diabetes Prevention Program. Data were obtained from 28 participants (a subgroup of a larger study) who completed baseline expense surveys, with 22 and 21 surveys completed at 6 and 12 months, respectively. Compared to baseline, significant decreases were found at 6 months in mean weight (-11.6 lbs, -5.4%, p<0.001), median triglycerides (-27.5 mg/dl, p=0.02), and mean diastolic (-6.8 mmHg, p<0.001), and systolic (-7.6 mmHg, p<0.0001) blood pressure. Weight loss, diastolic, and systolic blood pressure remained significantly lower at 12 months. The median of reported monthly food expenses at grocery and other food stores was unchanged from baseline ($450) to 6 months ($450, p=0.42), and baseline to 12 months ($463, p=0.37). This finding of no change held true for the median of total reported food expenses (spending at grocery and other stores, take-out food, and restaurant purchases) from baseline ($728) to 6 months ($735, p=0.85), and baseline to 12 months ($741, p=0.81). These results demonstrate that healthy lifestyle changes can take place without a significant increase in household food expenses. Such a finding may help dispel perceptions that prevent some individuals from participating in type 2 diabetes prevention programs. 1275-P GANG CHEN, PEIYING HUANG, MINGZHU LIN, JUNPING WEN, HUIBIN HUANG, XUEJUN LI, SHUYU YANG, Fuzhou, China, Xiamen, China POSTERS Clinical Diabetes/ Therapeutics Correlation of Peri-Operative Blood Glucose Level With Post-Operative Complications, Hospital Costs, and Length of Hospital Stay in Patients With Gastrointestinal Malignancies Objective: To investigate the effect of peri-operative hyperglycemia on post-operative complications, hospital costs, and length of hospital stay in non-diabetic patients with gastrointestinal malignancies. Methods: A total of 1015 non-diabetic patients with gastrointestinal malignancies who had complete clinical information and underwent surgical intervention in our hospital between January 2010 and December 2010 were retrospectively evaluated. Records on fasting plasma glucose (FPG), liver function, and kidney function were collected before and 1 d after surgery. The correlation of peri-operative blood glucose level with post-operative complications, hospital costs, and length of hospital stay was further assessed in non-diabetic patients with gastrointestinal malignancies. Results: One day after surgery, a blood glucose >6.1 mmol/L and > 7.0 mmol/L existed in 79.0% and 60.4% of the patients, respectively, which was significantly higher than before surgery. A blood glucose >9.13 mmol/L significantly increased the length of hospital stay, hospital costs, and the incidence of post-operative complications. Conclusion: Non-diabetic patients with gastrointestinal malignancies should be monitored when the blood glucose is > 9.13 mmol/L 1 d after surgery. Glucose-lowering measures should be carried out if necessary to reduce the incidence of post-operative complications, length of hospital stay, and medical costs. Supported by: NIH 1278-P Real-World Retrospective Study of Treatment Intensification After Failing 2 Oral Antidiabetic Drugs (OADs) Among Elderly With Type 2 Diabetes Mellitus (T2DM) MAYANK AJMERA, WENHUI WEI, STEVE ZHOU, RITUPARNA BHATTACHARYA, CHUNSHEN PAN, USHA SAMBAMOORTHI, Morgantown, WV, Bridgewater, NJ, Boca Raton, FL 1276-P Self-Monitoring Blood Glucose Without Using the Results: Findings from the Diabetes & Aging Study Among elderly patients, the management of T2DM is complicated by population heterogeneity and elderly-specific complexities. Few studies have been done to understand treatment intensification among elderly patients failing multiple OADs. Using the Humana database, this retrospective cohort study analyzed elderly (≥65 y) Medicare patients with T2DM and previously treated with 2 OADs, but having poor glycemic control (A1C ≥8%). Treatment intensification was defined as addition of a third OAD, a glucagon-like peptide-1 (GLP-1) receptor agonist, or insulin during the observation period. Factors associated with time to treatment intensification and the choice of intensification were identified. Of the 16,802 included patients (female 48.6%, white 69.6%, mean age 73.2 y, baseline A1C [where available] 8.9%, mean follow-up 22 months), 49.1% (n=8,252) intensified treatment during the follow-up period, with median time to intensification of 18 months using product-limit estimates. Of those patients who intensified treatment, 57.6 % used a third OAD, 40.6% used insulin, and 1.8% used a GLP-1 receptor agonist, and 76.0% received these within 1 y after the index date. Independent factors associated with higher likelihood of intensification were: polypharmacy (adjusted hazard ratio [aHR] 1.10; 95% CI: 1.05-1.15); 3 or more office visits (aHR 1.09; 95% CI: 1.031.15); baseline A1C >9% (aHR 1.51; 95% CI: 1.44-1.58); and higher diabetes complication burden (aHR 1.09; 95% CI: 1.01-1.16). The aHR was lower for patients of African-American race; of higher age; living in the South; with HMO coverage; and before/after a Medicare coverage gap. Among elderly T2DM patients failing multiple OADs, only half received treatment intensification, and patient complexities play a significant role in its timing, which should be a critical component when designing effective intervention programs. RICHARD W. GRANT, ELBERT HUANG, DEBORAH J. WEXLER, NEDA LAITEERAPONG, MARGARET WARTON, HOWARD H. MOFFET, ANDREW J. KARTER, Oakland, CA, Chicago, IL, Boston, MA We investigated the prevalence, predictors, and costs associated with unused results from self-monitored blood glucose (SMBG) in type 2 diabetes. We studied 7320 patients not prescribed insulin enrolled in the Diabetes Study of Northern California (DISTANCE) survey cohort. Patients were 58.7 (9.9) years old and had 8.3 (7.1) years diabetes duration. Patients reported whether they used SMBG results to make adjustments to diet, exercise or medicines; and whether their doctor reviewed their SMBG results. We categorized SMBG results as “used” (by patient or provider) or “unused” (not used by either patient or provider). SMBG results were unused by patient and provider in 15.2% of patients. Patients with unused results had lower HbA1c (7.1 ± 1.5% vs. 7.3 ± 1.4%, p < 0.001) than patients who used results. In separate models adjusted for demographic and clinical differences, major predictors of unused results included: patient report that diabetes was not a high priority (RR 1.81; 95% CI: 1.58-2.07), that his/her PCP did not engage in shared decision making (RR 1.66; 1.46-1.90), and that no health care provider had taught them to adjust their diet or medicines based on SMBG results in the past year (RR 2.27; 2.00-2.57). Patients with unused results were dispensed 171 ± 191 test strips/year in the prior two years at an estimated cost of $168 per year. The nearly 1 in 6 non-insulin treated patients practicing SMBG without either the patient or provider using the results represents a wasteful and ineffective practice for patients and health systems alike. Our results suggest that the decision to initiate and continue SMBG must be made in concert with the patient’s own priorities, and, if prescribed, Supported by: Sanofi U.S., Inc. & For author disclosure information, see page 829. A334 Guided Audio Tour poster ADA-Funded Research YOSHIKO ONDA, AYA MORIMOTO, RIMEI NISHIMURA, HIRONARI SANO, KAZUNORI UTSUNOMIYA, NAOKO TAJIMA, Tokyo, Japan Background and aims: We previously reported that about 1 out of 10 patients with childhood-onset type 1 diabetes was still presenting to pediatricians, even after the age of 30 years or older. The objective of this study is to investigate the factors determining the behavior of these adult patients still presenting to pediatricians. Materials and methods: Of the 1,385 patients diagnosed as having type 1 diabetes at < 18 years of age between 1965 and 1979 from two nationwide surveys, the current study included a total of 857 patients for whom information on their attending physicians was available as of January 1, 2000. We sent these physicians a questionnaire to investigate the size of their practice (hospital/ clinic-based), their office locations, the population size of their city (≥ 500,000 or < 500,000 or government-designated cities [GDC] or non-governmentdesignated cities [NGDC]), and their board certification status (BCS). Results: 52.8% of their physicians’ offices were located in GDC, with the 47.2% located in NGDC. 72.5% of the pediatricians and 76.6% of the internists were found to have BCS with no difference seen in the rate of specialist visits between those presenting to pediatricians and those presenting to internists (p = 0.32; chi-square test). Logistic regression analysis showed that the younger the age at diagnosis and the more the physicians’ offices tended to be in GDC, the significantly more likely that they still presented to pediatricians, but that their physicians’ BCS had no correlation with the rate of their visits to pediatricians (p = 0.003/0.002/0.271). Conclusion: Not physician BCS but physician office location was the factor determining the behavior of adult patients still presenting to pediatricians. 1282-P Metabolic Control in Adults With Type 2 Diabetes (T2D) in the General Population and a Diabetes Center 1280-P SANJEEV N. MEHTA, ALLISON B. GOLDFINE, MARTIN J. ABRAHAMSON, LORI M. LAFFEL, Boston, MA A Cross-National Comparison of Safety and Efficacy Information in Insulin Glargine Drug Labels In the US, <20% of T2D adults achieve control of A1c, BP, and cholesterol (ABC). Effectively managing the T2D epidemic may require timely referral of suboptimally controlled patients to diabetes specialty providers. We assessed ABC control from 2005-10 in T2D adults (age ≥20 yrs) in the general population (3 2-year NHANES cycles) and in T2D patients newly referred to an outpatient diabetes center and after 1 year of specialty care. We used 2005 ABC goals: A1c<7%, BP<130/80, LDL<100 mg/dL. At baseline, referrals (n=3298, 57% male) were younger (58.0 vs. 61.4, p<.001) and had shorter T2D duration (7.6 vs. 7.9 yrs, p<.001) than NHANES adults (n=1569, 50% male). At baseline, referrals were less likely to be at target for A1c (referrals vs. NHANES: 33 vs. 56%, 22 vs. 53%, 22 vs. 51%, p<.001 for all) or BP (37 vs. 43%, 37 vs. 51%, 38 vs. 54%, p<.001 for all) compared to NHANES adults; those at LDL goal differed only in 2009-10 (50 vs. 51%, p=.9; 52 vs. 56%, p=.2; 61 vs. 52%, p=.009). After 1 year, more referrals were at goal for A1c (43%, 38%, 41%, p<.001 compared to baseline), BP (46%, 42%, 45%, p<.001 compared to baseline), and LDL (61%, p=.01; 69%; p<.001; 69%, p.08 (trend) compared to baseline). Within 1 year, the proportion of referrals achieving ABC control increased 240% (Figure). By 2009-10, 1 in 15 T2D adults presenting to the diabetes center had ABC control, compared to 1 in 4 in the general population, suggesting appropriate referral of sicker T2D adults and measurable benefit for this group. JENNIFER M. POLINSKI, AARON S. KESSELHEIM, JOHN D. SEEGER, JOHN G. CONNOLLY, NITEESH K. CHOUDHRY, WILLIAM H. SHRANK, Boston, MA Type 2 diabetes mellitus (T2DM) has reached epidemic proportions worldwide. Many patients with T2DM will require insulin, and the evidencebased use of insulin is described in the prescription label. However, labels in different settings may provide inconsistent or contradictory information. We examined label content for insulin glargine across 17 settings with high T2DM prevalence: Abu Dhabi, Argentina, Brazil, Canada, China, Germany, Israel, Italy, Japan, Mexico, Russia, Saudi Arabia, South Korea, Spain, Turkey, United Kingdom, and the US. We assessed heterogeneity across 7 label sections pertinent to effective and safe use: indications and usage, contraindications, dosage and administration, warnings and precautions, adverse drug reactions (ADRs), use in specific populations, and overdose. We compared label characteristics in settings where drug labels were governed by a local regulatory authority, versus settings where labels were administered by a regional body or adopted from another locale. All 17 labels cautioned that providers should consider age, illness, diet, and exercise when prescribing. Only 2 (12%) described care of the fasting patient. Caution was urged for patients with renal or hepatic impairment in 16 (94%) labels. Four (24%) did not describe responses to missed doses and 5 (29%) failed to recommend that patients be counseled about the risk of hypoglycemia. ADR counts ranged from 4 (Brazil) to 41 (Argentina). Labels emerging from regional or adopted regulatory bodies reported fewer patients in efficacy studies than did labels from settings with their own drug regulatory agencies (365±0 patients vs. 3,560±2,938, p=0.04). There is substantial variation in the content of drug labels for glargine. Information describing the risks or prevention of hypoglycemia was often missing. Settings with local regulatory agencies based labels on the experiences of a larger number of patients. Such variation may lead to international inconsistency in T2DM quality of care. Supported by: Eli Lilly and Company 1281-P The Effect of Comorbidity on the Costs of Recent-Onset Type 2 Diabetes Mellitus in Veterans MANGALA RAJAN, ANUSHUA SINHA, HEATHER FRANKLIN, LEONARD POGACH, East Orange, NJ Recent epidemiological studies suggest that the cardiovascular benefit of tighter control in recent onset diabetes (DM) may be limited to individuals with a lower illness burden. The goal of this study is to compare total costs ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A335 POSTERS of treating a Veteran with recent-onset DM with and without comorbidities in the Veterans Health Administration (VHA). VHA data for FY2004 from the Health Economic Resource Center were used to assess costs. Chronic comorbidity was categorized as either dominant (limited life expectancy and extensive medical treatment likely to minimize attention to diabetes), discordant (not related to diabetes in terms of pathophysiology, treatment plans, or self-management strategies), or concordant (related to diabetes with synergies in treatment plans and management) based on ICD-9 and CPT codes, using a validated protocol. There were 53,513 veterans with recent-onset DM, of whom 7% had microvascular illnesses, 25% had macrovascular disease, 32% had discordant illness and 7% dominant illnesses. There were 3,816,487 veterans with no DM who were used as the comparison group with a slightly lower comorbidity burden. The inpatient and outpatient costs associated with treating patients with no comorbidities and DM was $2,416 [CI $2,088 - $2,203]; an incremental cost of $743 compared to equivalent patients with no DM. For veterans with discordant disease (such as mental illness), this cost was $6,867 [CI $6,645 - $7,087]with an incremental cost of $2,361. For veterans with concordant disease the cost was $5,338 [CI $5,070 - $5,605] with an incremental cost of $2,508. For those veterans with dominant disease the cost was $19,907[CI $18,810- $21,000]; an incremental cost of $11,181. Chronic comorbidity is associated with higher incremental costs for DM. These results can be used to project long-term costs of Veterans with DM and co-morbid conditions. These results support the recent ADA guidelines on individualizing goals for patients with DM and comorbidity from the perspective of costs. 1279-P What Are the Factors Determining the Behavior of Adult Patients With Childhood-Onset Type 1 Diabetes Still Presenting to Pediatricians? Results from the DERI Mortality Study Clinical Diabetes/ Therapeutics HEALTH CARE DELIVERY—ECONOMICS HEALTH CARE DELIVERY—ECONOMICS 6.5%, and no dementia. Prior to a physician encounter, intervention patients interacted with the tool and generated a summary for their physician that included individual patient’s LE estimates, geriatric screening results and treatment preferences. Control patients received an A1C pamphlet. Physicians and patients were surveyed before and after the encounter. Results: Intervention (N=68) and control patients (N=25) were similar by gender (77% female), age (mean 74 years), ethnicity/race (89% black) and duration of diabetes (mean 16 years). Intervention patients were more likely to have their physician report an A1C discussion versus controls (90% vs. 76%, p=0.20). Intervention patients were slightly more likely to report an A1C goal (53% vs. 44%, p=0.44) and have a goal that matched their physician’s goal (35% vs. 28%, p=0.51). Decisional conflict scores tended to decline more for intervention (53.8→24.7) than control patients (51.6→36.6) (p=0.09). Intervention patients with LE estimates ≤ 5 years had a shift towards higher A1C goals, 7.1-8.0% (55%→65%) (p<0.01). Patients with LE estimates > 5 years had a shift towards lower A1C goals, ≤7.0% (49%→60%) (p<0.01). Control patients did not have A1C goal shifts of the same size and/ or direction. Conclusion: A personalized decision support tool incorporating prognostic information and patient preferences encouraged active discussion regarding A1C goal selection, decreased patients’ decisional conflict regarding A1C goal choice, and increased personalization of A1C goals based on LE estimates. 1283-P Patient Reported Perspectives on Diabetes Disease, Comorbidities, Glucose Control, and Adherence POSTERS Clinical Diabetes/ Therapeutics JANICE LOPEZ, ROBERT A. BAILEY, KATHY ANNUNZIATA, MARCIA F.T. RUPNOW, SILAS MARTIN, Raritan, NJ, Princeton, NJ The ADA/EASD guidelines suggest the management and treatment of type 2 diabetes mellitus (T2DM) can be impacted by many factors such as age, comorbidities, risk of hypoglycemia, and treatment adherence. Understanding the burden of these important factors from a patient perspective is important in determining treatment approaches. Using the 2011 US National Health and Wellness Survey (NHWS) data, we characterized the T2DM population as a whole, with added focus on the elderly. NHWS results were stratified by age and gender, then weighted and projected to reflect the demographic composition of the total adult population (based on Census data). Of 75,000 US adults aged ≥18 y who responded to the survey, 7,828 reported a diagnosis of T2DM (mean age, 58.5 ± 13.6 y; 56% male). Of this population, 3,241 (34.5%) were ≥65 y and 636 (14.8%) were ≥75 y (54% and 49% were male, respectively). These respondents represent the projected US population ≥65 y and ≥75 y of 7.6 M and 3.2 M, respectively. In the ≥65 y and ≥75 y subgroups, the most common comorbidities were high cholesterol (69%, 66%), hypertension (69%, 70%), and arthritis (53%, 57%). In both subgroups ≥65 y and ≥75 y, nearly 56% used statins; 49% and 47%, respectively, reported ever experiencing hypoglycemia. 63% of participants reported being adherent to treatment. These data show that compared to the overall population, older adults have a greater burden of comorbidities, but a trend of better glucose control based on reported HbA1c. Supported by: Retirement Research Foundation (2007-250); NIDDK (P30DK092949-01) 1285-P Factors Associated With Adherence to Oral Antihyperglycemic Monotherapy in Patients With Type 2 Diabetes KAAN TUNCELI, CHANGGENG ZHAO, MICHAEL J. DAVIES, KIMBERLY G. BRODOVICZ, CHARLES M. ALEXANDER, LARRY RADICAN, Whitehouse Station, NJ Medication adherence in type 2 diabetes (T2DM) is low, leading to suboptimal outcomes. This study focused on factors associated with adherence in patients (pts) with T2DM on oral antihyperglycemic monotherapy (OAM). This retrospective cohort analysis used a US claims database (MarketScan). The index period was 01/01/07 to 03/31/10 and the index date was the date of first OAM prescription (Rx) in this period. Eligible pts with T2DM were ≥18 yrs old at T2DM diagnosis, were continuously enrolled in the database for 1 yr before and after the index date, and had an OAM Rx. Adherence was assessed using the proportion of days covered (PDC). Logistic regression was used to assess factors (age, gender, OAM status [new to or previously on OAM], OAM dosing regimen [once- or twice-daily], average daily drug cost, and number of concomitant Rxs) associated with adherence (PDC ≥80%). Eligible pts (N=133,449; 51% men) had a mean age of 61 yrs (35% ≥65 yrs), 39% were on a once-daily OAM, and 49% on ≥3 concomitant Rxs. Overall, 59% of pts were considered adherent to OAM. Table shows results from the logistic regression analysis examining factors associated with adherence to OAM. In conclusion, adherence to OAM is multifactorial and is influenced by pt characteristics, treatment regimens, and drug cost. Table. Odds Ratios (OR) for adherence to OAM (PDC ≥80%)* Parameter** OR*** 95% CI Male 1.14 1.11, 1.17 Age <45 years 0.35 0.33, 0.36 Age 45 and <65 years 0.60 0.58, 0.61 New to OAM treatment 0.83 0.81, 0.85 Twice-daily dose for OAM 0.82 0.80, 0.84 Average Daily Cost/Copay for OAM 0.72 0.69, 0.75 No other concomitant Rxs 0.42 0.41, 0.44 1 concomitant Rxs 0.67 0.65, 0.70 2 concomitant Rxs 0.87 0.84, 0.89 * Adjusted for baseline characteristics and comorbid conditions. ** Reference categories: female, age ≥65 years, previously on OAM treatment, once-daily dose for OAM, and ≥3 concomitant Rxs. *** An OR >1 indicates a positive association with adherence and an OR <1 indicated a negative association with adherence. 1284-P A Pilot RCT of Personalized Decision Support for Older Patients With Diabetes ELBERT S. HUANG, AVIVA G. NATHAN, JENNIFER COOPER, PRIYA M. JOHN, WILLIAM DALE, NANANDA COL, DAVID O. MELTZER, MARSHALL H. CHIN, Chicago, IL, Biddeford, ME Background: Geriatric diabetes care guidelines encourage individualized glycemic targets (i.e., A1C goal) for older patients based on patient life expectancy (LE) and preferences. We pilot tested a web-based decision support tool which provides individualized prognostic information and elicits patient preferences. Methods: We randomized physicians and their patients to the decision support tool, with 3:1 recruitment. Patients were > 65 years, had A1C ≥ Supported by: Merck, Sharp & Dohme & For author disclosure information, see page 829. A336 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—OBESITY AND TYPE 2 DIABETES 5TT) and were pair-matched for age, gender, BMI, and whole body insulin sensitivity index (WBISI). Within all subjects, fasting %DNL was associated positively with BMI (r=0.62; P=0.05) and negatively with WBISI (r = -0.66; P=0.03). Interestingly, when the two genotypes were compared, the TT subjects showed a lower %DNL peak (P=0.01) and a lower delta increase of %DNL (P=0.05), compared to CC after the drink. By contrast, AbsDNL tended to be higher in TT subjects at fasting (P=0.16) and was significantly higher after the drink (P=0.05). These are the first data of de novo lipogenesis in children and show that the rs1260326 variant in the GCKR gene modulates the lipogenic pathway and leads to metabolic inflexibility. Also, these data suggest that higher fasting lipogenic rates, that are inappropriately resistant to stimulation by consumption of sugars, are associated with the higher triglycerides levels and hepatic fat accumulation. 1286-P Quality of Care for Type 2 Diabetes Mellitus Patients in Dubai: A Retrospective Cohort Study KATHERINE M. OSENENKO, BONNIE M. KORENBLAT DONATO, SHELAGH M. SZABO, ELLEN E. KOROL, LARA QATAMI, FATHEYA AL AWADI, JABER AL ANSARI, ROSS MACLEAN, ADRIAN R. LEVY, ABDULRAZZAQ A. AL MADANI, Vancouver, BC, Canada, Wallingford, CT, Dubai, United Arab Emirates, Princeton, NJ Despite the high prevalence (25%) of type 2 diabetes mellitus (T2DM) in the United Arab Emirates (UAE), few data are available on the types and quality of care administered. The objective was to assess quality of T2DM care in Dubai, UAE. T2DM patients were systematically sampled from the Diabetes Clinic database at a large hospital in Dubai. Inclusion criteria were: visit between October 2009 and March 2010 (accrual period), age ≥18 years and UAE nationality. Patient and disease characteristics and care patterns were synthesized. Quality of care was assessed for the 12 months after the accrual period, based on the United States (US) Healthcare Effectiveness Data and Information Set (HEDIS) Comprehensive Diabetes Care measures, including: glycosylated hemoglobin (HbA1c) screening, control (<8%) and poor control (>9%); low-density lipoprotein (LDL) screening and control (<100 mg/dL); blood pressure (BP) control (<140/90 mmHg); and screening for retinopathy or nephropathy. Of 250 eligible patients, mean age at enrolment was 58 years (standard deviation (SD): 12 years), 33% were male, and mean T2DM duration was 14 years (SD: 8 years). Quality of care measures are presented in Table 1. Although guideline monitoring of clinical measures was conducted for most, higher rates of screening for complications and meeting clinical targets would reduce the burden of T2DM in the UAE. These measures would be useful for informal comparison to other local, or US, HEDIS estimates. Supported by: AHA (11CRP5620013) & Supported by: Bristol-Myers Squibb PEDIATRICS—OBESITY AND TYPE 2 DIABETES & Guided Audio Tour: Pediatric Obesity and Type 2 Diabetes—Insights and Outcomes (Posters: 1287-P to 1293-P), see page 19. CHERIL L. CLARSON, HILARY BROWN, STEPHANIE DEJESUS, MICHELLE JACKMAN, FARID H. MAHMUD, HARRY PRAPAVESSIS, KEVIN J. SHOEMAKER, JUSTINE WILSON, DAVID J. HILL, London, ON, Canada, Toronto, ON, Canada 1287-P Hepatic De Novo Lipogenesis in Youth: Role of the GCKR rs1260326 Variant The REACH study (ClinicalTrials.gov #NCT00934570) assessed a structured lifestyle intervention with metformin extended release (MXR) or placebo on BMI, metabolic risk factors, and adipokines in obese adolescents in a 2-year trial. Sixty-nine obese adolescents (10-16 years, BMI >95th centile) were randomized to MXR or placebo, and subsequently to moderate or vigorous intensity exercise for 12 weeks. Group exercise continued weekly and nutritional and behavioral counseling monthly. Sixty-one subjects (88%) completed 6 months, 47 (68%) 12 months, and 29 (42%) 24 months. Mixed factor ANOVA with an interaction term between treatment condition and time was used to compare the patterns of change over time, depending on treatment condition. The medication (ignoring exercise) time interaction was significant for BMI z score (p=.01) and % body fat (p=.03). In the MXR group, there were significant differences in BMI z score at baseline (2.22±037) and 6 months (2.08±0.48, p<.001), 12 months (2.05±0.49, p=.002) and 24 months (2.10±0.46, p=.04); The % body fat also decreased in the MXR group between baseline (45.81±5.03) and 6 months (42.65±7.13, p<.001), 12 months (43.47±06.17, p=.0002), and 24 months (44.90±07.19, p=.03). No significant improvements were seen in placebo assigned subjects. The NICOLA SANTORO, ELIZABETH PARKS, BRIDGET PIERPONT, SONIA CAPRIO, New Haven, CT, Dallas, TX The rs1260326 single nucleotide polymorphism (SNP) in the glucokinase regulatory protein (GCKR) gene variant has been recently associated with an increased hepatic fat content, elevated concentrations of plasma large VLDL and triglycerides (TG) in obese children and adolescents. The rs1260326 is a functionally-relevant SNP consisting of a C (common allele) to T (risk allele) substitution coding for a proline to leucine substitution at the position 446 (P446L). Recent studies have suggested that the mechanism underlying increased HHF% may be an increased hepatic de novo lipogenesis (%DNL). To test this hypothesis, 10 adolescents (mean age of 15.8±2.3, mean BMI of 35.8±7.57) underwent the assessment of lipogenic stimulation, an oral glucose tolerance test (OGTT), and fast-MRI to assess hepatic fat content (HFF%). Lipogenesis was assessed by measuring D2O incorporation into VLDL-TG palmitate after consumption of glucose (75g) and fructose (25g). Absolute lipogenesis (AbsDNL) was calculated as %DNL x large VLDLTG content. Subjects were selected according to the genotype (5CC and ADA-Funded Research & 1289-P Structured Lifestyle Intervention With Metformin Extended Release or Placebo in Obese Adolescents For author disclosure information, see page 829. Guided Audio Tour poster A337 POSTERS Hyperinsulinemia is commonly associated with obesity in cross section studies. However, it is debated if hyperinsulinemia predicts weight gain in adults and children. Aim of the present study was to investigate the effect of fasting insulin on longitudinal changes in body weight and the development of cardiovascular disease risk factors in early childhood. 424 Children (211 boys and 213 girls) from Da Qing city, China, were recruited at 5 years old and followed-up for 5 years. Blood pressure, anthropometric measurements, fasting plasma insulin, glucose and triglycerides (TG) levels were measured at baseline and follow up, while birth weight and television (TV) viewing time only at baseline. The relationships among these parameters were investigated in boys and girls. Multivariate regression analysis was used to test if the fasting insulin predicts weight change (ΔWeight) during the 5-years follow-up. Compared with children in the lowest quartile of fasting insulin at 5 years old, the children in the top quartile had significant higher ΔWeight 5 years later in both boys (18.39±0.86 vs. 13.08±0.73 kg, p<0.01) and girls (15.80±0.60 vs. 12.03±0.71 kg, p<0.01).Similar trend was found in body mass index (BMI). Fasting insulin at entry significantly associated with ΔWeight over the 5years follow-up period after the adjustment of gender, birth weight, TV viewing time and body weight at 5 years old (p<0.0001). In addition, Systolic blood pressures, fasting plasma glucose, TG and the HOMA_IR at 10 years old were also significantly higher in those children who had higher fasting insulin 5 years before. Of note, body weight at baseline did not significantly associated with change of fasting insulin during the follow-up period (p>0.05). Fasting insulin predicts weight gain and the increase of cardiovascular risk factors in Chinese children of early childhood, but the body weight dose not predict increasing of fasting insulin over 5 year follow-up. Clinical Diabetes/ Therapeutics YANYAN CHEN, JINPING WANG, YAYUN JIANG, HUI LI, YINGHUA HU, GUANGWEI LI, Beijing, China, Daqing, China Table 1. Quality of care assessment among T2DM patients in Dubai, April 1, 2010 to March 31, 2011 T2DM Patients (N=250) Diabetes Care Measure n (%) HbA1c screening 245 (98.0) HbA1c poor control (>9.0%) 65 (26.0) HbA1c control (<8.0%) 135 (54.0) LDL screening 225 (90.0) LDL control (<100 mg/dL) 160 (64.0) BP control (<140/90 mmHg) 108 (43.2) Retinopathy attention/screening 80 (32.0) Nephropathy attention/screening 39 (15.6) & 1288-P Fasting Insulin Predicts Weight Gain and Cardiovascular Risk Factors in Early Childhood over a 5-Year Period: The Da Qing Children Cohort Study PEDIATRICS—OBESITY AND TYPE 2 DIABETES at least 1 major event, and by 2.4 (95% CI 1.4-4.3) for those with at least 4 major events. The odds for clinically elevated depressive symptoms or impaired QOL were associated with a 2-fold increase among those reporting at least 1 major event (95% CIs 1.1-4.7 and 1.1-3.6) and with a 4-fold increase among those with at least 4 major events (95% CIs 2.1-10.2 and 1.9-7.9). Assessing exposure to stressful events among youth with T2D and identifying supportive interventions may be important in mitigating risk of diminished self-management and psychosocial dysfunction. following changes were significant but did not differ by medication group: insulin resistance (HOMA) (p=.03) and fasting glucose (p<.001) decreased at 6 months; leptin decreased at 6 (p=.003) and 12 months (p=.02); The following changes were significant but did not differ by intensity of exercise at 6 months: BMI z score (p=.0004), % body fat (p=.001), HOMA (p=.007), fasting glucose (p<.001), and leptin (p=.005) decreased at 6 months, while the adiponectin to leptin ratio (p=.01) increased. A structured lifestyle intervention was successful in arresting the rise in BMI z score and % body fat in obese adolescents and the addition of MXR to lifestyle intervention led to a decline in BMI z score and % body fat. Supported by: NIDDK/NIH (U01-DK61212) & Supported by: Children’s Health Foundation; CIHR & 1290-P SILVA A. ARSLANIAN, LAURA PYLE, NEIL H. WHITE, FIDA BACHA, SONIA CAPRIO, MOREY W. HAYMOND, LYNNE LEVITSKY, ROBIN S. GOLAND, STEVEN M. WILLI, Pittsburgh, PA, Rockville, MD, St. Louis, MO, Houston, TX, New Haven, CT, Boston, MA, New York, NY, Philadelphia, PA Effects of Sex and Race on the Change in Total and Abdominal Adipose Tissue and Intrahepatic Lipid in Response to Exercise in Obese Adolescents The TODAY trial enrolled 699 obese youth (10-17 years) with type 2 diabetes (T2D) of < 2 years, and found that, regardless of random assignment to 1 of 3 treatments (metformin alone, metformin + rosiglitazone, or metformin + lifestyle), glycemic failure rates (HbA1c ≥8% for 6 months or sustained metabolic decompensation) were higher in non-Hispanic blacks (NHB, 52.8%) vs. whites (NHW, 36.6%) and Hispanics (H, 45.0%). We investigated whether differences in insulin sensitivity (IS, 1/fasting insulin [1/IF]), insulinogenic index (IGI, ΔI30/ΔG30), or β-cell function (βCF) relative to IS (oral disposition index [oDI], 1/IF x ΔI30/ΔG30) explained the racial disparity in response to treatment. The table shows baseline characteristics in NHB, H and NHW. ANTHONY R. DELDIN, JENNIFER L. KUK, FIDA BACHA, INGRID LIBMAN, TAMARA S. HANNON, CHRIS BOESCH, SOJUNG LEE, Pittsburgh, PA, Toronto, ON, Canada, Houston, TX, Indianapolis, IN, Bern, Switzerland POSTERS Clinical Diabetes/ Therapeutics 1292-P Racial Disparity in Maintenance of Glycemic Control in the TODAY Trial: Does Insulin Sensitivity, β-Cell Function, or Both Play a Role? We examined the influence of sex and race on the change in total (TAT), abdominal subcutaneous (ASAT) and visceral adipose tissue (VAT) and ectopic fat in the liver and skeletal muscle in response to aerobic (AE) versus resistance exercise (RE) training without calorie restriction in obese youth. Fifty-nine obese adolescent boys and girls (BMI>95th, 12-18 yrs) were randomized to 3-months of AE (n=29) or RE (n =30). Total and regional AT were measured by whole-body MRI and intrahepatic lipid by proton magnetic resonance spectroscopy. Mid-thigh skeletal muscle density was measured by computed tomography. Independent of sex, body weight did not change (P>0.1) in response to AE and RE training. In both exercise groups, there were no ethnic differences in changes in skeletal muscle density, TAT or ASAT. There were also no sex differences in skeletal muscle density. With respect to intrahepatic lipid, white girls had greater (P<0.05) reductions in intrahepatic lipid than black girls and white boys with RE, but the differences were diminished with increasing intrahepatic lipid loss. For TAT and ASAT, girls had significantly greater decreases in TAT with AE than boys, wherein the differences diminished with increasing fat loss. For VAT, there were sex and race main effects in that girls and blacks lost more VAT in response to the same RE training (P<0.05), after adjusting for baseline differences in VAT. In terms of the fitness measures, boys gained 25.5% more strength with RE and 28% more aerobic fitness with AE than girls (P<0.05). Our findings suggest that in response to aerobic or resistance training, there are sex and race differences in adiposity and fitness changes in obese adolescent boys and girls. In general, there are more sex differences than racial differences and that females have greater improvements in adiposity, but boys have greater improvements in fitness. NHB had higher BMI, HbA1c, IGI, and oDI, and lower IS compared with H and NHW. Within each race, the trends were similar between those who failed vs. those who did not fail to maintain glycemic control on assigned treatment; baseline HbA1c was higher and βCF (IGI and oDI) was lower in those who failed vs. those who did not; BMI and IS were the same (data not shown). These results indicate that, regardless of race, impaired βCF at baseline but not IS, was associated with failure of glycemic control over time. The higher failure rate in NHB was not associated with greater impairment in βCF. Other factors, e.g. behavioral/psychosocial, may underlie the racial disparity in response to treatment in TODAY. Supported by: NIDDK & Supported by: NIDDK 1291-P Correlates of Exposure to Stressful Life Events in Youth With Type 2 Diabetes (T2D) in the TODAY Clinical Trial & NATALIE WALDERS-ABRAMSON, ELIZABETH M. VENDITTI, BARBARA J. ANDERSON, LAURE EL GHORMLI, MITCHELL GEFFNER, JOAN KAPLAN, MICHAELA KOONTZ, CAROLYN LANDIS, MARISA PAYAN, RONALD SALETSKY, PATRICE YASUDA, Aurora, CO, Pittsburgh, PA, Houston, TX, Rockville, MD, Los Angeles, CA, Philadelphia, PA, Cleveland, OH, Syracuse, NY 1293-P Minimum Prevalence of Persistent Albuminuria in Youth <18 Years of Age With Type 2 Diabetes Mellitus: A Canadian Paediatric Surveillance Program Study ELIZABETH A.C. SELLERS, STASIA HADJIYANNAKIS, SHAZHAN AMED, ALLISON DART, ROLAND DYCK, JILL HAMILTON, VALERIE LANGLOIS, CONSTADINA PANAGIOTOPOULOS, ANNE-MARIE UGNAT, HEATHER J. DEAN, Winnipeg, MB, Canada, Ottawa, ON, Canada, Vancouver, BC, Canada, Saskatoon, SK, Canada, Toronto, ON, Canada The NIDDK-sponsored TODAY clinical trial of youth-onset T2D provided the opportunity to examine relationships between stressful life events, physiological measures (BMI, metabolic control, hypertension, triglycerides), medication adherence, and psychosocial factors (depression, quality of life [QOL]). During the final TODAY visit, participants completed the Yeaworth (1980) Adolescent Life Change Event Scale. Respondents indicated whether or not 33 events had occurred over the past year and, if so, rated the associated level of distress (0=not at all upsetting to 4=extremely upsetting). Data were available from 497 participants at mean age 18.4 years and 66% female, 31% Black non-Hispanic, 40% Hispanic, and 22% White non-Hispanic. Total number of events, number of major stressful events (considering only those events that were somewhat, very, or extremely upsetting), and sum of distress levels were calculated for each respondent. Participants reported an average 5.2 events (SD 2.9, range 0-14), 1.9 major events (SD 2.1, range 0-13), and 11.7 cumulative event distress (SD 9.1, range 0-61). Major stressful events were not related to physiological measures. Relationships between stress exposure and psychosocial factors were found. The odds of nonadherence to medication increased by 1.6 (95% CI 1.0-2.3) for those with The natural history of type 2 diabetes diagnosed in childhood is still largely unknown. Evidence suggests that microvascular complications occur at an earlier age with a shorter duration of diabetes in youth-onset compared to adult-onset disease. The prevalence of albuminuria, the first sign of nephropathy, is not known in this population. This observational study used data from the Canadian Pediatric Surveillance Program (CPSP) that reported cases of persistent albuminuria in youth (< 18 years of age) with type 2 diabetes. Case reporting occurred during a 24 month period from March 2010-2012. Type 2 diabetes was defined using national guidelines. Persistent albuminuria was defined as a positive albumin-to-creatinine ratio in a minimum of 2 out of 3 urine samples over a minimum of 3-6 months and confirmed with at least one first morning sample. Descriptive statistics were used to illustrate demographic and clinical features of the population. Data from a previous national surveillance study & For author disclosure information, see page 829. A338 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—OBESITY AND TYPE 2 DIABETES Methylation of CpG loci at 5-7y showed temporal stability, and predicted future adiposity. Thus, longitudinal tracking co-efficients (p< 0.001 for all loci) were such that methylation at 5-7y predicted 77-88% of the variation in methylation at 14y, and for each unit of methylation at 5-7y, adiposity (% fat) differed by 1.25 (95% CI: 0.47-2.03) at CpG locus -841; 0.75 (0.26-1.24) at CpG -816; 0.63 (0.09-1.17) at CpG -78 and 0.70 (0.01-1.40) at CpG -521 (all p≤0.03). Whilst there were also associations between methylation and insulin resistance, these were attenuated by the inclusion of adiposity in the models. PCG1α is central to energy homeostasis through regulation of mitochondrial function, pancreatic β-cell function and adipogenesis. These findings are novel, and suggest a functional role for early methylation of PCG1α in the later development of obesity and such epigenetic marks in childhood may have predictive value. for type 2 diabetes in youth was used to calculate the minimum prevalence of persistent albuminuria. 50 cases were reported over the 24 month study period. Mean age at diagnosis of diabetes was 12.3 years (SD 2.1). Mean duration of diabetes at diagnosis of albuminuria was 0.76 years (SD 1.36). 64% were female. 80% were of Canadian First Nation heritage, 10% Caucasian and 10% other ethnicities. 38/50 (76%) were from Manitoba, 10/50 (20%) from Ontario and one each from Alberta (2%) and Nova Scotia (2%). 65% were exposed to either gestational or pre-gestational diabetes in utero. Minimum prevalence of persistent albuminuria in youth with type 2 diabetes was 7.4%. In conclusion, persistent albuminuria occurs in youth with type 2 diabetes in the first year after diagnosis, demonstrates regional variation and is associated with First Nation heritage and exposure to maternal diabetes in pregnancy. Supported by: Public Health Agency of Canada 1296-P MING LI, JINHUA YIN, LU XU, BCAMS GROUP, XINHUA XIAO, Beijing, China CLAUDIA TOLEDO-CORRAL, BRIDGETTE BLEBU, PRAJAKTA PARAB, PRASANNA MOHANTY, LAUREN GYLLENHAMMER, TING LIU, MICHAEL I. GORAN, MARC J. WEIGENSBERG, Los Angeles, CA Abnormal intrauterine growth resulting in a birth weight (BW) lower or greater was an independent risk factor for future metabolic disorders. The aim of this study was to clarify the association of BW with insulin resistance (IR), adverse adipokine profile and abnormal metabolism in a population of Chinese children and adolescence. This is a retrospective study including 3091 children and adolescents aged 6 to 18 in the Beijing area from the Beijing Child and Adolescent Metabolic Syndrome (BCAMS) study. The subjects were evaluated with respect to anthropometric measurements, pubertal development, blood pressure, glucose, insulin, lipid profiles, and adipokines including leptin, adiponectin, RBP-4 and resistin. When compared with normal birth weight (NBW) peers, low birth weight (LBW) was a strong determinant for children’s IR (OR=1.79, P=0.006), but high birth weight (HBW) predicted current central obesity (OR=1.712,P=0.002) in our study population of Chinese children and adolescents. BW negatively correlated with children’s serum leptin level (P=0.008), leptin/adiponectin ratio (P<0.001) and RBP-4(P=0.005), but positively correlated with serum adiponectin level (P=0.006) after adjusting for age, gender, Tanner stage, parental DM history, current body mass index (BMI), family annual income, parental education and children’s physical activity. BW negatively correlated with current HOMA-IR (P<0.001) and MetS components including systolic blood pressure (P<0.001), diastolic blood pressure (P=0.007), triglycerides (P=0.003) and fasting plasma glucose (p=0.014) after adjustment for above covariates. In conclusions, a lower BW was a strong predictor for future glucose and lipid metabolic disorder including adipokine changes early in childhood, whereas higher BW predicted childhood central obesity. This study extends previous findings showing associations between elevated serum cortisol and metabolic syndrome (MS) in Latino children ages 8-13 yrs, by examining associations between salivary cortisol patterns and MS, visceral adipose tissue (VAT), and insulin sensitivity (Si) in older OLAs. 107 healthy OLAs (51M/46F; age 15.8±1.2y; BMI-z: 2.0±0.4) were admitted to hospital overnight to measure nocturnal cortisol rise (NCR=salivary cortisol rise from 2200hrs to awakening at 0530hrs), cortisol awakening response (CAR=salivary cortisol from awakening to 30 min later); blood pressure (BP), fasting serum cortisol, glucose, insulin, and lipids; % body fat by DXA; VAT using 1.5T MRI. Si was determined by FSIVGTT and minimal modeling. Children with MS had higher CAR vs children without MS (1.10±0.09 vs. 0.82±0.08 µg/dl, p=0.02), after adjusting for age, sex, %body fat. Serum cortisol and NCR were not associated with MS (p>0.05). Among individual features of MS, children with either a high systolic or diastolic BP had a lower NCR than those with normal BP (0.57±0.08 vs 0.81±0.06 µg/dl, p=0.02) when adjusted for age, sex and total %body fat. Waist circumference, triglycerides, and HDL were not associated with any cortisol measures (all p>0.05). Fasting serum cortisol, but not CAR or NCR, was associated with VAT (r=0.20, p=0.04). Si was negatively correlated to CAR and serum cortisol (r=-0.23 & -0.22, p<0.05) and marginally associated with NCR (r=0.18, p=0.06) after adjusting for age, sex and %body fat. Adjusting for VAT did not alter the association between CAR and MS (p=0.04). However adjusting for Si did weaken this relationship (p=0.18). In OLAs, multiple markers of hypothalamic-pituitary-adrenal axis activity (HPA) were associated with cardio-metabolic risk factors. The specific link between MS and altered HPA axis activity may be at least partly mediated by Si, but is independent of VAT or total body fat. Supported by: Beijing Municipal Science Technology Commission 1297-P Supported by: NCMHD (P60MD00254); NCRR (M01RR00043-46) Lower Soluble Receptor for Advanced Glycation End Products Independently Predicts Metabolic Syndrome in Youth 1295-P SARAH R. BRICKEY, JUSTIN RYDER, DONALD R. MCCLELLAN, GABRIEL Q. SHAIBI, Phoenix, AZ Methylation of PGC1α at 5-7y Predicts Adiposity Throughout Childhood: A Longitudinal Study The soluble receptor for advanced glycation end products (sRAGE) has antiatherogenic properties in adults. Despite the increasing prevalence of obesity, metabolic syndrome and type 2 diabetes in youth, little is known about the relationship between sRAGE and cardiometabolic disease risk in the pediatric population. Therefore, the purpose of this investigation is to examine the relationship between sRAGE and cardiometabolic risk factors in Latino youth. Data from 133 Latino youth (58% female; age 15.3±3.3 years) enrolled in the Arizona Insulin Resistance Registry were analyzed. Metabolic syndrome was determined using a continuous risk score calculated from the individual components normalized to the population adjusting for age and gender. Components included waist circumference, high-density lipoprotein cholesterol (HDL-c), triglycerides, mean arterial pressure (MAP), and HOMAIR. sRAGE levels were determined from fasting serum using an enzymelinked immunosorbent assay. In univariate analysis, sRAGE was inversely associated with waist circumference (r=-0.22, p= 0.01), MAP (r=-0.15, p= 0.09), and HOMA-IR (r=-0.29, p<0.01) and positively associated with HDL-c (r=0.19, p<0.05). In multiple regression analysis that included age, gender, and BMI, as covariates, sRAGE was an independent protective predictor of metabolic syndrome risk score (p=0.03). This model accounted for approximately 55% of the total variance in metabolic syndrome (total R2=0.55, p<0.001). JOANNE HOSKING, REBECCA CLARKE-HARRIS, TERENCE J. WILKIN, ALISON N. JEFFERY, BRAD S. METCALF, JONATHAN PINKNEY, KAREN A. LILLYCROP, GRAHAM C. BURDGE, Plymouth, United Kingdom, Southampton, United Kingdom, Exeter, United Kingdom Obesity is considered the most important cause of the increase in type 2 diabetes in childhood. Early environmental exposures modify the risk of obesity, but epigenetic association has so far been based on single measurements at time points that are disparate from either the health outcome, or the environmental exposure. Crucial evidence from longitudinal studies linking events over time is lacking. We undertook longitudinal analysis of DNA methylation in the peroxisomal proliferator-γ-co-activator-1α promoter (PCG1α) in blood taken from 40 children (20 boys) annually over nine years. Seven CpG loci were sequenced which have been shown previously to be hypermethylated and associated with decreased PGC1α expression in overweight subjects. The association between methylation of PGC1α at 5-7y, adiposity (DEXA % fat) and insulin resistance (HOMA2-IR) between 9-14y was modelled using generalised estimating equations, taking into account gender, pubertal timing, and physical activity. ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A339 POSTERS Associations between Salivary Cortisol Patterns and Cardio-Metabolic Risk Factors in Overweight Latino Adolescents (OLAs) Clinical Diabetes/ Therapeutics Birth Weight Predicts Adverse Adipokine Profile, Insulin Resistance and Abnormal Metabolism in Chinese Children and Adolescents: Results from BACMS Study 1294-P PEDIATRICS—OBESITY AND TYPE 2 DIABETES These data suggest that sRAGE may be an early independent biomarker of cardiometabolic disease risk in youth. Prospective studies are needed to establish the predictive utility of sRAGE for long-term disease outcomes. Distinct HDL Subspecies are Associated With Increased Arterial Stiffness in Youth With Type 2 Diabetes 1298-P AMY S. SHAH, SCOTT M. GORDON, L. JASON LU, LAWRENCE M. DOLAN, ELAINE M. URBINA, W. SEAN DAVIDSON, Cincinnati, OH 1300-P Human epidemiological studies demonstrate that increased HDL cholesterol (HDL-C) concentrations are associated with a decrease risk in cardiovascular disease. However, recent drug therapies designed to raise plasma HDL-C concentrations have failed to reduce cardiovascular events. These findings suggest that the cardio-protective effects of HDL may lie in specific HDL subspecies and not in the HDL-C number. Thus, we sought to define the composition of HDL in adolescents with T2D and determine whether specific HDL subspecies are associated with early markers of arterial disease. Youth with T2D (n=10) and two control groups lean (n=9) and obese (n=11) adolescents were recruited in a cross sectional study. Plasma was separated into 18 fractions using gel filtration chromatography. Lipid associated proteins were then isolated and identified using mass spectrometry. Concurrently, arterial thickness and stiffness measures including common/ bulb/internal carotid intima media thickness (IMT) and pulse wave velocity (PWV) were measured. Compared to lean and obese controls, youth with T2D exhibited decreased phospholipid content in fractions containing large HDL subspecies that was inversely associated with pulse wave velocity (p<0.001). No association between HDL-C and pulse wave velocity was seen. Proteomic analysis of the HDL sized fractions demonstrated decreases in 17 of 45 identified proteins in the T2D group compared to lean youth. The most striking protein decreases occurred in apolipoprotein (apo) A-1, apoC-I, apoE and paraoxonase-1 (p<0.05) in fractions associated with large HDL particles. Our data demonstrate early changes in the lipid and protein compositions of specific HDL subspecies in adolescents with T2D that are related to early markers of arterial disease. These findings suggest that analyzing the composition of HDL, rather than HDL-C, may be useful in assessing cardiovascular risk in this population. Overweight/Obese Girls With Type 1 Diabetes (T1D) at High Risk for Adverse Behavioral Health Outcomes POSTERS Clinical Diabetes/ Therapeutics KARL E. MINGES, ROBIN WHITTEMORE, ARIANA CHAO, KATHRYN M. MURPHY, MARGARET GREY, New Haven, CT, Philadelphia, PA Overweight/obese youth and youth with T1D report more health-related concerns relative to their normal weight peers and youth without T1D. However, the relationship between weight status and behavioral health outcomes in youth with T1D is not known. A secondary analysis of baseline data from a multi-site trial of an internet behavioral program was undertaken to examine differences of normal weight (BMI ≥ 5th and < 85th percentile) and overweight/obese (BMI ≥ 85th percentile) boys and girls with T1D in clinical (HbA1c, duration), psychosocial, self-management, quality of life, and family factors. Youth (n=318, mean age=12.3 ± 1.1 years, 55% female, 62.7% white) completed validated self-report measures of perceived stress, self-worth, social relationships, depression, self-management (activities, adherence to treatment), quality of life, and family conflict and support. Parents reported socioeconomic factors, and HbA1c was determined by DCA2000. T-tests and chi-square analyses were used. Overall, overweight/obesity was prevalent (39.0%) and was high in girls (42.6%) and boys (33.1%). Overweight/obese youth had a significantly higher HbA1c (8.6 ± 1.6) relative to normal weight youth (8.2 ± 1.4) (p=0.03), but this was attenuated when stratified by gender. Overweight/obese girls had parents with lower education (p<0.01), but not lower income (p=0.09), and had longer diabetes duration (p=0.02) compared to normal weight girls. Further, overweight/obese girls exhibited significantly lower psychosocial functioning (p=0.001-0.04), self-management (p=0.01-0.02), quality of life (p=0.01) and family functioning (p=0.01-0.04) compared to normal weight girls. No significant differences were exhibited between normal weight and overweight/obese boys. Thus, overweight/obesity is prevalent among youth with T1D and overweight/obese girls with T1D comprise a group at high risk of adverse health outcomes. Greater attention to weight status in youth with T1D is warranted. 1301-P The Relationship between A1c, 2hr Plasma Glucose, and Continuous Glucose Monitoring-Determined Glycemic Patterns in Obese Adolescents Supported by: NIH/NINR (2R01NR04009); NIH/NIDDK (1T32DK097718) CHRISTINE L. CHAN, KIM MCFANN, KRISTEN J. NADEAU, LINDSEY NEWNES, PHILIP S. ZEITLER, MEGAN M. KELSEY, Aurora, CO 1299-P Studies comparing A1c in diagnosing diabetes to the “screening gold standard” 2hr plasma glucose after 75 gram glucose load (2hr PG) or fasting plasma glucose (FPG) have reported discrepancy of the A1c vs. 2hr PG. To better understand these discrepancies, continuous glucose monitoring (CGM) was utilized to directly study free-living glycemic patterns in obese adolescents. Relationships among A1c, FPG, 2hr PG and CGM variables were examined. 111 adolescents ages 10-18 yrs, BMI ≥85th%ile, A1c ≤7.5% and not on pharmacologic diabetes treatment participated. Linear regression was used to examine the association of A1c, 2hr PG, and FPG with the following outcomes from 24 hours of CGM data: % time spent ≥120, ≥140, and ≥200 mg/dl; day area under the curve (AUC), night AUC, and total AUC. Subjects average age was 13.9 ± 2.2 yrs, A1c 5.7 ± 0.4%, 2hrPG 134 ± 34 mg/dL, BMI 97.9 ± 2.6%.; 37% male, 59% Hispanic, 23% white, 17% black. While OGTT and A1c each correlated significantly with multiple CGM outcomes, r2 did not differ significantly between A1c and OGTT on any CGM outcome (Table 1). FPG failed to predict any CGM outcome. This is the first study to compare A1c and 2hr PG to direct measures of glycemia with CGM in overweight/obese adolescents. 2hr PG did not outperform A1c in predicting CGM outcomes. Thus, under free-living circumstances, 2hr PG and A1c may be equally valid in predicting glycemia. Predicting Progression to Prediabetes in Obese Latino Youth JOON YOUNG KIM, MICHAEL I. GORAN, CLAUDIA M. TOLEDO-CORRAL, MARC J. WEIGENSBERG, MYUNGHAN CHOI, GABRIEL Q. SHAIBI, Phoenix, AZ, Los Angeles, CA We have previously shown that 1-hr glucose concentration during an Oral Glucose Tolerance Test (OGTT) independently predicts the development of prediabetes and β-cell dysfunction among obese Latino youth. However, whether 1-hr glucose is a more powerful predictor than other glycemic indicators is unknown. Therefore, the purpose of this study was to compare the predictive power of 1-hr glucose to HbA1c, fasting, and 2-hr glucose for identifying future prediabetes in youth. Obese normoglycemic Latino youth with a family history of type 2 diabetes (67 M / 49 F; 11.5 ± 1.9 yrs) were assessed at baseline for HbA1c, fasting, 1-hr, and 2-hr glucose during an OGTT and were subsequently followed for up to 8 years. Receiver operating characteristic (ROC) curves were used to estimate the predictive power of each glycemic indicator at baseline for identifying progression to prediabetes (impaired fasting glucose and/or impaired glucose tolerance). In addition, a multivariable prediction model combining all glycemic indicators was assessed for predictive power. During follow-up, 52.9% of participants developed prediabetes. The area under the 1-hr glucose ROC curve was significantly greater than the area under the HbA1c ROC curve (73.4% vs. 58.1%; p<0.05). When compared to fasting and 2-hr glucose, predictive power of 1-hr glucose was modestly but not significantly greater than that of either fasting (66.9%; p=0.35) or 2-hr glucose (63.6%; p=0.12). The multivariable model that included HbA1c, fasting, 1-hr, and 2-hr, glucose was the most powerful model and was more predictive than the model that did not include 1-hr glucose (77.4% and 70.7%, respectively; p=0.07). These results provide support for the clinical utility of 1-hr glucose during a standard OGTT as a prospective marker of diabetes risk. Specifically, combining 1-hr glucose concentrations with traditional glycemic indicators may help to identify normoglycemic youth at highest risk for progressing to prediabetes. Table 1. R2, P-value for LnA1c and Ln2hrPG LnA1c Ln2hrPG FPG R-square p-value R-square p-value R-square p-value %≥120mg/dl 0.18 <0.0001 0.21 <0.0001 0.04 0.02 %≥140mg/dl 0.13 0.0004 0.26 <0.0001 0.03 0.04 %≥200mg/dl 0.11 0.10 0.08 0.09 0.03 0.06 Day AUC 0.18 <0.0001 0.25 <0.0001 0.06 0.01 Night AUC 0.20 0.0002 0.13 0.0002 0.03 0.10 24hr AUC 0.21 <0.0001 0.25 <0.0001 0.06 0.02 Supported by: NIH (R01DK59211) Supported by: NIH/NCATS (UL1TR000154); Genentech, Inc. (1210-F05); Medtronic Diabetes & For author disclosure information, see page 829. A340 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—OBESITY AND TYPE 2 DIABETES 1302-P DOSE DIFFERENCE (95% CI) OF LEAST-SQUARE MEAN FROM PLACEBO Obstructive Sleep Apnea in Obese Adolescents: Associations With Cardiometabolic Risk Markers SARA WATSON, ZHUOKAI LI, WANZHU TU, HASNAA JALOU, JAMIE BRUBAKER, SANDEEP GUPTA, JORDAN HUBER, AARON CARROLL, TAMARA S. HANNON, Indianapolis, IN 2-hr blood glucose (mg/dL) Pediatric studies examining the association between obstructive sleep apnea (OSA) and insulin sensitivity/cardiometabolic risk are limited and conflicting. We examined cardiometabolic risk markers among obese youth with OSA and their obese peers without OSA. We performed a retrospective analysis of 162 patients (age 13.5±2.4 y) who underwent polysomnography for suspected obesity-related OSA. Fasting lipids, glucose, insulin, and glycosylated hemoglobin (HbA1c) were performed as part of clinical evaluation. Patients were categorized into three groups by degree of OSA as measured by the apnea hypopnea index (AHI): none or mild (AHI <5), moderate (AHI 5-9.9), and severe (AHI >9.9). Despite similar degrees of obesity, patients with moderate or severe OSA had higher fasting insulin (p = 0.0031) and non-HDL cholesterol (p = 0.0061), as compared with those without OSA. After controlling for body mass index (BMI), there was a positive association between the AHI and log HOMAIR (homeostasis model assessment-insulin resistance) (p = 0.002). The association of AHI with HOMA-IR was greatest among Hispanic patients (p= 0.018). Obese youth with OSA show evidence of worse cardiometabolic profiles (including higher fasting insulin, HOMA-IR, and non-HDL cholesterol) compared with their equally obese peers who do not have clinically significant OSA. A novel finding is that fasting insulin and HOMA-IR are more strongly linked with OSA among male Hispanic adolescents. OGTT MTT OGTT -15.7 -53.5 1.05 (-79.7, 48.3) (-133.0, 26.0) (-13.3, 15.4) OGTT 75.2 (-49.2, 199.7) OGTT 2.0 (-1.4, 5.5) OGTT MTT 2.8 2.8 (1.7, 4.5) (1.9, 4.1) 100 -48.2 -78.01 ( -112.2, 15.8) (-157.5, 1.5) -11.5 (-25.8, 2.9) 26.4 (-101.4, 152.2) 1.2 (-2.4, 4.7) 3.4 2.6 (2.1, 5.4) (1.8, 3.8) 200 -42.5 -71.39 -18.3 (-108.3, 23.4) (-153.2,10.4) (-33.1, -3.6) 47.0 (-90.4, 184.4) 1.4 (-2.4, 5.2) 2.8 2.8 (1.7, 4.6) (1.9, 4.1) mean ratio; bweighted average EIRIN CAROLAN, DECLAN CODY, JEAN O’CONNELL, ANDREW HOGAN, DONAL O’SHEA, Dublin, Ireland Childhood obesity is an epidemic that threatens the health of future generations.The development of insulin resistance and Type 2 Diabetes Mellitus in children is a major concern. We investigated the inflammation and immune profile in obese versus lean children. In total 49 subjects were recruited from a pediatric tertiary center, categorized as obese (n=29) and non-obese (n=20). The obese cohort had a mean BMI z score of 3.40; lean cohort had a score of 0.18. Albeit none of the study participants had developed Type 2 Diabetes Mellitus, there is a significant level of insulin resistance in the obese cohort, with a mean HOMA-IR of 4.8 and mean fasting insulin of 149 pmol/L. A pro-inflammatory environment was found in the obese cohort with significantly increased levels of circulating TNF-alpha, leptin and soluble CD 163 (mean 105ng/ml in non obese vs 150ng/ml in obese cohort). This is an important marker linking the pro-inflammatory “M1” macrophage function to insulin resistance. On a cellular level TLR-4 stimulation of obese peripheral blood mononuclear cell (PBMC) resulted in high levels of IL-1beta (mean 1500pg/ml in non obese vs 2100 pg/ml in obese cohort), a cytokine well described as pathogenic in metabolic disease. Micro RNA 33a, 33b and 107 have been associated with metabolic disease in murine models and adults, these small sequences post-transcriptional modify the target gene expression. We measured circulating PBMC microRNA and found a significantly increased expression of micro RNA 33a and 33b in the obese cohort. The significant alterations in the immune profile of the children from our study population clearly depict the need for action in both preventing and reversing childhood obesity from a young age. 1303-P Metabolic Effects and Safety of Single Rising-Dose Sitagliptin in Adolescents With Type 2 Diabetes LARRY A. FOX, IAIN P. FRASER, NAOMI NEUFIELD, MARK S. KIPNES, TRACIE L. MILLER, PHILLIP S. ZEITLER, HENRY RODRIGUEZ, JOCELYN GILMARTIN, SUSI LEE, JACLYN K. PATTERSON, XIUJIANG LI, LATA MAGANTI, WEN-LIN LUO, DANIEL TATOSIAN, EDWARD A. O’NEILL, S. AUBREY STOCH, Jacksonville, FL, Whitehouse Station, NJ, West Hollywood, CA, San Antonio, TX, Miami, FL, Aurora, CO, Indianapolis, IN Supported by: National Children's Research Centre (Ireland) The incidence of type 2 diabetes (T2DM) in adolescents is rising, but limited antihyperglycemic treatments are available to them. In a double-blind study, 35 patients (10 to 17 yrs old) were randomized to single oral doses of sitagliptin (SITA) or placebo (PBO) in a 3:1 ratio. SITA doses (50, 100, or 200 mg) were tested in ascending order in up to 12 patients/dose. Pharmacodynamic parameters were measured during an oral glucose tolerance test (OGTT) and a meal tolerance test (MTT). Glucose concentrations 2 hours after OGTT or MTT were reduced in patients receiving SITA compared to those receiving PBO, but changes did not reach statistical significance. The 100 and 200 mg doses demonstrated similar effects, but were larger compared to the 50 mg dose. The ratio of active to total GLP-1 was significantly increased for all SITA doses in both tests. Glucagon concentrations were numerically lower for the 100 and 200 mg doses relative to PBO, and insulin and C-peptide levels were greater for all three doses of SITA during the OGTT. Trends toward changes observed in these parameters are consistent with known mechanistic effects of SITA. A total of 18 adverse experiences, 13 with SITA, 5 with PBO, were reported by 8 patients. With the exception of 1 report of IV site pain of moderate intensity, all were considered mild in intensity by the investigator. In conclusion, a 100 mg/day dose of SITA is an appropriate dose for further evaluation in adolescent patients with T2DM. 1305-P Natural History of Elevated Alanine Aminotransferase (ALT) Levels in Youth With Type 2 Diabetes (T2D) NORMA VAN WALLEGHEM, HEATHER J. DEAN, ELIZABETH SELLERS, DIABETES EDUCATION RESOURCE FOR CHILDREN, Winnipeg, MB, Canada Non-alcoholic fatty liver disease (NAFLD) is the most common cause of liver disease among children, paralleling the rise of obesity. Unexplained ALT elevation is a frequently used surrogate marker for NAFLD in children and adults. At the Diabetes Education Resource for Children and Adolescents (DER-CA) in Winnipeg, Canada, ALT levels are done at diagnosis of T2D and annually if ALT <30 IU/L. If ALT >30 IU/L, levels are repeated at every subsequent clinic visit (every 3 to 4 months). We target ALT >100 IU/L for further evaluation for NAFLD; these children are tested for hepatitis B and C and undergo liver ultrasound to confirm steatosis. The aim of this clinic audit was to assess the frequency of NAFLD identified by elevated ALT in our clinical population and describe the natural history of elevated ALT over 5 years. Between 2007 and 2011, 69/395 youth with T2D (30 males, 39 females) have had an ALT >100 IU/L on one or more occasions. NAFLD was confirmed by ultrasound in >85% of youth. During this 5 year period For author disclosure information, see page 829. Guided Audio Tour poster A341 POSTERS 1304-P The Development of Metabolic Disease in Obese Children: The Role of Immune Cell Dysregulation via Increased Micro-RNA 33a and 33b, Soluble CD163 and IL-1beta Clinical Diabetes/ Therapeutics Supported by: Merck, Sharp & Dohme Supported by: T32DK065549-08 (to S.W.); NIH (R03HD057532 to T.S.H.); IUPUI (to T.S.H., W.T., A.C.) & 50 aweighted Demographic and clinical characteristics associated with HOMA-IR Regression Coefficient 95% CI p-value AHI 0.008 0.003 – 0.012 0.002 Female sex 0.104 -0.089 – 0.296 0.289 Black race 0.153 -0.039 – 0.345 0.117 Hispanic Race 0.502 0.141 – 0.863 0.007 Other Race 0.401 -0.177 – 0.980 0.173 Age -0.069 -0.113 – -0.024 0.003 BMI 0.032 0.019 – 0.044 <0.001 *log HOMA-IR was used to accommodate the skewness in the HOMA-IR ADA-Funded Research GEOMETRIC MEAN RATIO (SITAGLIPTIN/ PLACEBO) 2-hr glucagon 2-hr insulin 2-hr C-peptide GLP-1a (pg/mL) (microIU/mL) (ng/mL) 2-hr Activeb/Totalb PEDIATRICS—OBESITY AND TYPE 2 DIABETES the prevalence of elevated ALT has ranged from 10-15% of our clinical population. Three year data shows persistence of ALT >100 IU/L in <25% of youth. Four year data shows normalization of liver enzymes in all youth, despite no significant improvement in A1c or BMI z-score. All youth are of Canadian First Nations heritage. The majority (71%) presented with elevated ALT at diagnosis of T2D. The mean age at first ALT elevation >100 IU/L was 12.2 years (median 12 years; range 8-17 years). In these youth, the mean ALT was 126 IU/L, mean A1c was 9.6%, and mean BMI z-score was 2.2. Of the 29 youth who had genotyping for the G319S polymorphism of the HNF-1alpha gene, 69% (20/29) have one or two copies of the S allelle. In summary, many youth with T2D have resolution of the biochemical signs of NAFLD overtime suggesting no need for invasive testing or repeat imaging. diabetes were compared to a group of youth with type 2 diabetes without albuminuria reported in a previous CPSP Study conducted from 2006-2008. Compared to youth with type 2 diabetes and no albuminuria, those with albuminuria were younger at diagnosis of diabetes (12.2 vs. 13.6 years; p<0.001); more likely to be of Canadian First Nation heritage (80 vs. 47%); p<0.001), more likely to have been exposed to maternal pre-gestational diabetes (39 vs. 21 %, p=0.03) and more likely to be hypertensive (56 vs. 24%, p<0.001) at the time of diagnosis of diabetes. Gender distribution, BMI z-score, exposure to gestational diabetes, symptomatic presentation, and presence of dyslipidemia or elevated liver transaminases did not differ significantly between the 2 groups at diagnosis of type 2 diabetes. In conclusion, in youth with type 2 diabetes of First Nation heritage, those exposed to maternal pre-gestational diabetes and those with hypertension at diagnosis of diabetes are at increased risk for the development of persistent albuminuria. 1306-P Racial Differences in Intima Media Thickness (IMT) in Obese Adolescents: No Relationship to Insulin Sensitivity Supported by: Public Health Agency of Canada POSTERS Clinical Diabetes/ Therapeutics FIDA BACHA, KIM SUTTON-TYRRELL, SOJUNG LEE, HALA TFAYLI, SILVA A. ARSLANIAN, Houston, TX, Pittsburgh, PA, Beirut, Lebanon 1308-P African Americans (AA) men and women have twice the age adjusted rate of fatal coronary heart disease (CHD) compared with whites. Carotid IMT is a marker of subclinical atherosclerosis (ScA) and is predictive of the risk of CHD and stroke. We investigated whether racial differences in ScA, IMT and pulse wave velocity (PWV), are present in high-risk obese adolescents, and examined its determinants. 85 adolescents, 35 with normal glucose tolerance (NGT), 26 impaired glucose regulation (IGR), and 24 type 2 diabetes (T2DM) underwent carotid ultrasound (IMT and PWV), DEXA, BP, and lipids. A subset (27AA, 36AW) underwent a 3-h hyperinsulinemic (80 mu/min/m2)-euglycemic clamp to assess the relationship of insulin sensitivity to ScA. Age (years) BMI (kg/m2) Visceral adipose tissue (VAT) (cm2) Intima media thickness (mm) Pulse Wave Velocity (cm/msec) Insulin sensitivity adjusted for VAT (mg/kg/min per µu/mL) HbA1c (%) SBP (mmHg) AA (13 male, 27 female; 16 NGT, 10 IGR, 14 T2DM) 14.8 ± 0.3 34.8 ± 0.8 61.4 ± 5.1 0.54±0.007 674.5± 49.2 1.9±0.3 6.0±0.2 125.6±2.3 Mobile Health (mHealth) Intervention Called BodiMojo Using Text Messaging Aimed at Healthy Lifestyles for Youth With Diabetes (DM): A Pilot Randomized Controlled Trial (RCT) ALAN T. SCHULTZ, JESSICA T. MARKOWITZ, TARA M. COUSINEAU, DEBRA L. FRANKO, LORI M. LAFFEL, Boston, MA Teens and young adults with type 1 and type 2 diabetes (T1D/T2D) are frequently overweight or obese, increasing their risk for comorbidities. mHealth modalities provide opportunities for lifestyle interventions in such populations that almost universally use mobile technologies. We compared an mHealth lifestyle intervention, BodiMojo, with a pamphlet about healthy lifestyle activities in a 1-month pilot RCT in 90 youth with DM. Participants set physical activity and nutrition goals following randomization to either the mHealth or pamphlet group. We measured self-efficacy and A1c at baseline and after 1 month and also assessed satisfaction with the mHealth intervention. The mHealth group received daily text messages with general health tips plus check-in text messages inquiring about goals 3x/week and logistical check-ins 2x/month (total of 14 texts that were programmed for text responses from mHealth participants). Eligibility required a cell phone with texting capacity; 79% had a smartphone. Participants (47% male, 87% white, 91% T1D) were 18.7±1.6 years old with DM duration of 10.0±4.6 years; A1c of 8.5±1.7%. Of the 45 in the mHealth group, 40 (89%) responded to ≥1 text; range of responses 0-13. Of the 40 who responded, mean response rate was 10.3±3.5; median 12. There were no differences in self-efficacy or A1c between groups at baseline or follow-up. Of those in the mHealth group, 82% believed BodiMojo helped them achieve health goals and 58% would recommend BodiMojo to others. Although self-efficacy and A1c were unchanged, it is encouraging that 89% of mHealth participants responded to texts and ~75% (33/45) responded to ≥50% of the texts. More intensive or longer duration mHealth interventions can likely offer greater opportunity to engage young persons with DM in healthy lifestyle activities as well as initiatives aimed at improving A1c in order to preserve health and prevent complications. AW P (12 male, 33 female; 19 NGT, 16 IGR, 10 T2DM) 15.3 ± 0.3 ns 35.3 ± 0.8 ns 85.6 ± 5.1 0.001 0.50±0.007 0.001 704.7 ± 42.7 ns 2.4±0.2 0.02 5.6±0.1 126.2±1.8 0.07 ns Triglycerides were lower (95.9±9.2 vs. 125.6±8.8 mg/dl) and HDL higher (41.9±1.6 vs. 37.5±1.4 mg/dl), p<0.05 in AA vs. AW with no differences in cholesterol, BP or hs-CRP. In a multiple regression analysis, race independently contributed to the variance in IMT (β= -0.41, p<0.001) besides age (β= 0.30, p=0.009), SBP (β= 0.26, p=0.016) and HbA1c (β= 0.28, p=0.013), while sex, insulin sensitivity and BMI did not (R2= 0.35, p<0.001). Racial difference in IMT persisted when the NGT and dysglycemia groups were analyzed separately. The racial disparity in ScA is present in high risk youth, is not explained by traditional CVD risk factors, insulin resistance or CRP. It remains to be determined what factors are driving these racial differences. Supported by: NIH/NIDDK (1R43DK85748-1A1) 1309-P Uric Acid: Is There a Link With Cardiovascular Risk Factors also in Youth? MARCELE SCHETTINI, LORENA VENEZA, JAQUELINE CAIRES, PAULA PERAZZO, ATILA OLIVEIRA, ANA LUISA OLIVEIRA, ANA MAYRA OLIVEIRA, Salvador, Brazil, Feira de Santana, Brazil, Campinas, Brazil 1307-P Clinical Factors Associated With Persistent Albuminuria in YouthOnset Type 2 Diabetes Mellitus: A Canadian Paediatric Surveillance Program Study There are evidences that recognize uric acid (UA) as cardiovascular risk factor and link it with insulin resistance (IR) in adults but comprehensive studies on this issue are sparse in youth. We aimed to evaluate the frequency and characteristics of hyperuricemia in a sample of 543 Brazilian youth (251 boys; 11.1±3.2y; BMI_zs 2.3±2.1). The measurements included: anthropometry, blood pressure, uric acid, insulin, lipid profile, standard oral glucose tolerance test. IR was defined by Homeostasis model assessment of IR (HOMA-IR), excessive weight by body mass index z-score (BMI zs), abdominal obesity by waist circumference (WC) and metabolic syndrome by International Diabetes Federation criteria. Hyperuricemia was defined as the values over the median value for each age and the subjects were classified as with hyperuricemgia (Group 1) and normal uricemia (Group 2). Hyperuricemia was seen in 348 (46.4%) individuals. The number of subjects with hyperuricemia was higher in males than in females (p=0.001). There were significant association between Group 1 and BMIzs, abdominal obesity, high systolic and diastolic blood pressure, total cholesterol, ELIZABETH A.C. SELLERS, STASIA HADJIYANNAKIS, HEATHER J. DEAN, ALLISON DART, ROLAND DYCK, JILL K. HAMILTON, VALERIE LANGLOIS, CONSTADINA PANAGIOTOPOULOS, ANNE-MARIE UGNAT, SHAZHAN AMED, Winnipeg, MB, Canada, Ottawa, ON, Canada, Saskatoon, SK, Canada, Toronto, ON, Canada, Vancouver, BC, Canada Evidence is compelling that microvascular complications of type 2 diabetes occur at an earlier age with a shorter duration of diabetes in youthonset type 2 diabetes mellitus compared to adult-onset disease. The aim of this study was to identify the clinical features at the time of diagnosis of diabetes associated with persistent albuminuria in youth with type 2 diabetes (<18 years of age). This observational case-control study used data from the Canadian Pediatric Surveillance Program (CPSP) that reported cases of persistent albuminuria in youth with type 2 diabetes. Clinical features at diagnosis of & For author disclosure information, see page 829. A342 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—OBESITY AND TYPE 2 DIABETES group. In contrast, mean HDL-C (37 vs 44 mg/dL) and vitamin D levels (12.4 vs 14.5 IU/L) significantly decreased in the control group. Separate linear regression models, adjusted for sex, age and z-BMI, showed that children post-vitamin D intake had a 2 mg/dL higher HDL-C (p=0.02); a 6 mg/dL lower LDL-C ( p<0.001); a 5 mg/dL lower glucose ( p<0.001); and a 3.2 IU/L higher vitamin D level (p<0.001) than at baseline. In contrast, children who did not receive vitamin D had a 6 mg/dL lower HDL-C (p<0.01), and a 1.8 IU/L lower vitamin D level (p<0.01) than at baseline. These results suggest that vitamin D supplementation among Indian children with inadequate vitamin D levels could improve lipid and glucose levels. triglyceride, fasting insulin, HOMA-IR, number of metabolic syndrome (MS) components, C-protein reactive (CRP) (p<0.001 for all) and HDL-c (p=0.018) and LDL-c (p=0.011). After adjustment for age and sex, BMI zs (OR, 3.1; CI, 1.8 to 5.5; p=0.000), arterial hypertension (OR, 2.1; CI, 1.2 to 3.9; p=0.007), HOMA-IR (OR, 2.0; CI, 1.3 to 3.1; p=0.000) and dyslipidemia (OR, 2.3; CI, 1.5 to 3.7; p=0.001) were independently associated with hyperuricemia. These observations suggest that AU in youth is associated with abdominal obesity, IR and CRP, a marker of inflammation, supporting the hypothesis that AU is probably a real risk factor for cardiometabolic disease even in early stages of life. However the significance of this factor in trigger or accelerate atherosclerosis process ask for follow-up studies. Supported by: FAPESB 1312-P Effect of a Resistance Training Program on Body Composition and Metabolic Profile in Obese Non-Diabetic Adolescents 1310-P Effects of a Short-Term Resistance Training Program on Blood Pressure and on Emerging Cardiovascular Risk Factors in Obese NonDiabetic Adolescents Aerobic exercise is considered an effective tool in the improvement of body composition and metabolic profile thus able to preventing cardiovascular diseases. However the effects of an isolated resistance training (RT) program have not been well established. We aimed to evaluate the effects of an isolated RT program on body composition and metabolic profile in obese nondiabetic adolescents. The intervention was performed three times/week solely with resistance exercises during three months, including 24 sedentary obese non-diabetic adolescents (17 girls/7 boys; 14.1±1.0 y; 87.8±11.3 Kg; 32.1±3.6 kg/m2; Z-IMC 2.6±0.3). Body fat, lean and fat mass were measured by dual-energy X-ray absorptiometry. Fasting (FPG) and 2-h post-load PG (2hPG) after 75 g glucose anhydrous, insulin, HOMA-IR, leptin, adiponectin and non-esterified fatty acids (NEFA), triglycerides (TG), total (TC), high-density (HDL-c) and low-density lipopotrein (LDL-c) cholesterol were analyzed. After written informed consent all volunteers were evaluated before and after RT. After the intervention period of an isolated and regular RT program, we noticed a significant reduction on percent body fat (44.6±4.3 vs. 44.0±4.1 %, P<0.01), while body mass (87.8±11.3 vs. 87.2±11.5 Kg, NS), lean (46.849±7.344 vs. 46.666±7.98 g, NS) and fat mass (37.427±6.426 vs. 36.921±6.700 g, NS) were kept unchanged. Instead of it, insulin (28.5±13.6 vs. 23.3±10.2 mU/ ml, P<0.01) and HOMA-IR (6.0±3.2 vs. 4.9±2.2, P<0.05) were significantly reduced after intervention while no differences on FPG, 2-hPG, TC, HDL-c, LDL-c, TG, leptin, adiponectin and NEFA levels were observed. Our results suggest that a 3-month program solely of RT was able to reduce body fat, insulin levels and HOMA-IR, independently of weight loss in obese nondiabetic adolescents. The effects of isolated resistance training exercises (RT) on blood pressure (BP), endothelial function and low-grade inflammation are still unclear in obese adolescents. We aimed to evaluate the effects of RT on BP, endothelial function, and inflammatory markers in obese non-diabetic adolescents. RT was performed three times/week solely with resistance exercises during three months, including 24 obese non-diabetic adolescents (17 girls/7 boys; 14.1±1.0y; 87.8±11.3Kg; 32.1±3.6kg/m2; Z-IMC 2.6±0.3). BP was measured by casual and 24h ambulatory method. Microvascular reactivity was performed at skin using laser-Doppler flowmetry and vasomotion analyses was assessed by fast-fourier transform analysis to determine the contribution of the five frequency components (i.e., endothelial, 0.01-0.02Hz; neurogenic, 0.02-0.06Hz; myogenic, 0.06-0.15Hz; respiratory, 0.15-0.40Hz; and cardiac, 0.40-1.60Hz) on the variability of the signal. Endothelial-dependent and -independent vasodilation was tested respectively after acetylcholine and sodium nitroprusside iontophoresis. Endothelin-1 (ET-1), C-reactive protein (CRP), fibrinogen, interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) were measured. All volunteers were evaluated before and after RT. No change in body mass (32.1±3.6 vs. 31.7±3.7kg/m2; P>0.05) was observed after the RT program, but we noted a reduction in casual and 24h ambulatory systolic (P<0.01), diastolic (P<0.01) and mean BPs (P<0.01). RT led to significant reductions on fibrinogen (P<0.05) and ET-1 (P<0.05) levels while TNF-α and CRP were kept unchanged. We observed increments on cardiac component of the vasomotion (P<0.05) and on endothelium-dependent response (P<0.05) after RT. An isolated RT program reduced casual and 24h ambulatory BPs and also some of the emerging cardiovascular risk factors studied, independently of weight loss in obese non-diabetic adolescents. Supported by: FAPERJ 1313-P Glucose Screening and Identification of Pre-Diabetes in Children and Adolescents 2007–2011 Supported by: FAPERJ PATRICK J. O’CONNOR, NICOLE TROWER, ALAN SINAIKO, KAREN L. MARGOLIS, ELYSE KHARBANDA, EMILY PARKER, NANCY SHERWOOD, KENNETH ADAMS, JOAN LO, LOUISE C. GREENSPAN, DAVID MAGID, MATTHEW DALEY, Minneapolis, MN, Oakland, CA, Denver, CO 1311-P Improvement in Cardiovascular Risk in Argentine Indian School Children after Vitamin D Supplementation This cohort study quantifies glucose screening rates in children and adolescents in defined age, gender, BMI, and race strata. Subjects were 68,322 individuals ages 3-17 at cohort entry followed for a median of 37 months from 1/1/2007 to 12/31/2010, and with at least one clinic visit. Subjects’ laboratory data were examined for date and results of any outpatient fasting or random glucose, glycated hemoglobin (A1c), or oral glucose tolerance tests (oGTT). We report descriptive statistics on glucose testing rates and results indicating pre-diabetes by age, gender, BMI, race/ ethnicity, and year. Prediabetes was defined as fasting plasma glucose of 100-125 mg/dL, 2-hour oGTT glucose of 140-199 mg/dL, or glycated hemoglobin (A1c) of 5.7%-6.4% without a diabetes diagnosis. Overall, the rate of glucose screening was 10.7% (7,278/68,322). It was similar in males and females but higher in those with older age, obesity, minority race, or Hispanic ethnicity. The screening rate increased from 3.9% (1,132/28,948) per year in 2007 to 7.2% (2,301/32,167) in 2010. Glucose screening was most often done with fasting or random glucose, but 7.7% (560/7,278) of tests were A1c. About 13.9% (1,013/7,278) of tests indicated pre-diabetes. Of 1,013 subjects with pre-diabetes, 79.1% (801/1,013) were age 12 or older, 60.0% (608/1,013) had minority race or Hispanic ethnicity, and 30.9% (313/1,013) were obese. A diabetes diagnosis (250.xx) was present in 2.2% (161/7,278) of subjects with one or more glucose/A1c tests. We concluded that glucose screening rates in children and adolescents are VALERIA HIRSCHLER, GUSTAVO MACCALLINI, MILVA SANCHEZ, CLAUDIO ARANDA, MARIANA HIDALGO, LUIS CASTANO, CLAUDIA MOLINARI, Buenos Aires, Argentina, Barakaldo, Argentina Low vitamin D levels correlate with measures of cardiovascular risk. The objective was to determine whether vitamin D supplementation reduces cardiovascular risk among Argentine Indian school children. In a prospective one-year study, 215 (98M) children aged 10.1±2.6 years, were evaluated in November 2011 (spring season) at baseline. A treated cohort of 194 (106 M) children aged 10.9±2.3 years who received 5000 units of vitamin D weekly during 8 weeks; and a control group of 21 (7M) children aged 10.8±1.6 years who did not receive vitamin D were evaluated 1 y later (November, 2012). Anthropometry, lipids, glucose, and vitamin D levels ([25(OH)D]) were measured at baseline and one year. The prevalence of overweight and obesity was 26/215 (12.1%) at baseline; 16/194(8.2%) in the treated; and 2/21 (9.5%) in the control . 31(14.4%) at baseline, 2 (1%) in the treated, and 7 (33.3%) in the control had severe vitamin D deficiency ( <10 IU/L); 165 (77.1%) at baseline, 147(76.6%) in the treated, and 14 (66.7%) in the control had deficiency (10- <20 IU/L); 18 (8.4%) at baseline, 43 (22.2%) in the treated, and none in the control had insufficiency (20- <30 IU/L). Mean values of glucose (73 vs 78 mg/dL), LDL ( 83 vs 89 mg/dL), HDL-C (46 vs 44 mg/dL), and vitamin D (17.7 vs14.5 IU/L) significantly improved in the treated ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A343 POSTERS INGRID BARBARA FERREIRA DIAS, PAULO DE TARSO VERAS FARINATTI, MARIA DAS GRAÇAS C. SOUZA, DIOGO G. PANAZZOLO, DIOGO P. MANHANINI, ERICK BALTHAZAR, DIEGO M. DANTAS, PRISCILA A. MARANHÃO, EDUARDO H.A. PINTO, ELIETE BOUSKELA, LUIZ GUILHERME KRAEMER-AGUIAR, Rio de Janeiro, Brazil Clinical Diabetes/ Therapeutics INGRID B.F. DIAS, PAULO DE TARSO VERAS FARINATTI, MARIA DAS GRAÇAS C. SOUZA, ERICK BALTHAZAR, DIOGO G. PANAZZOLO, DIOGO P. MANHANINI, DIEGO M. DANTAS, PRISCILA A. MARANHÃO, ELIETE BOUSKELA, LUIZ GUILHERME KRAEMER-AGUIAR, Rio de Janeiro, Brazil PEDIATRICS—TYPE 1 DIABETES low but increasing, and the proportion of those screened who meet criteria for pre-diabetes is nearly 14%. Opportunistic screening rates are higher in demographic subgroups with the highest risk of pre-diabetes. Little is known about the overall health status, comorbidities, care, or subsequent metabolic status of children and adolescents who meet criteria for pre-diabetes. Parameter* Sitagliptin 50 mg Sitagliptin 100 mg Sitagliptin 200 mg (n=9) (n=9) (n=8) 3438 (2881, 4103) 5869 (4918, 7003) 12965 (10749, 15638) AUC0-∞ (nM·hr) Cmax (nM) 366 (288, 464) 666 (526, 845) 1876 (1458, 2413) C24hr (nM) 32 (25, 41) 43 (34, 55) 78 (60, 101) Tmax (hr) 3.0 (1.5, 5.0) 3.0 (2.0, 4.5) 2.5 (1.0, 3.1) Apparent t1/2 (hr) 12.1 (1.7) 11.2 (2.1) 11.7 (1.8) Between Population Comparisons Parameter# Adolescent Adult GMR (90% CI) AUC0-∞ (nM·hr) 6424 (5740, 7190)‡ 7851 (6360, 9691)§ 0.82 (0.66, 1.01) Cmax (nM) 1876 (1435, 2453)¥ 1803 (1354, 2401)† 1.04 (0.75, 1.44) C24hr (nM) 78 (54, 112) ¥ 106 (72, 156)† 0.74 (0.48, 1.14) *Values shown for AUC and Cmax and C24hr are back-transformed geometric least square mean (95% CI) from ANOVA model performed on natural logtransformed values; for Tmax the median (min, max) and for apparent terminal t½ the harmonic mean with jack-knife standard deviation are shown. #Between population comparison was made for dose-adjusted (to 100 mg) sitagliptin AUC across all available doses in adolescent and adult patients with T2DM; between population comparison for sitagliptin Cmax and C24hr were made at 200 mg dose in adolescent and adult patients. Population values shown are back-transformed geometric least-squares means (95% CI) from linear mixed effect model (for AUC0-∞) or ANOVA model (for Cmax and C24hr) performed on natural log-transformed values. Herman et al, JCEM 91:4612 (2006); ‡N = 25; §N=14 observations from 7 subjects; ¥N = 8; †N = 7. Abbreviations: GMR, geometric least square mean ratio for Adolescent / Adult; CI, confidence interval. 1314-P Longitudinal Correlates of Youth Health Risk Behaviors among Children and Adolescents With Type 2 Diabetes in the TODAY Study POSTERS Clinical Diabetes/ Therapeutics CAROLYN E. IEVERS-LANDIS, PATRICE YASUDA, NATALIE WALDERS-ABRAMSON, NANCY AMODEI, KIMBERLY L. DREWS, JOAN KAPLAN, LORRAINE KATZ, SYLVIA LAVIETES, RON SALETSKY, DANIEL SEIDMAN, Cleveland, OH, Los Angeles, CA, Aurora, CO, San Antonio, TX, Washington, DC, Philadelphia, PA, New Haven, CT, Syracuse, NY, New York, NY The purpose is to report predictors of health risk behaviors (cigarette and alcohol use) among children and adolescents with type 2 diabetes (T2D) enrolled in the Treatment Options for type 2 Diabetes in Adolescents and Youth (TODAY) clinical trial. Hypothesized predictors included key demographics, child psychological and health factors (depressive symptomatology, academic achievement, BMI z-scores, HbA1c), and parent/ family and environmental factors (stressors, neighborhood safety). Data were obtained from TODAY participants at baseline (N=599), 6- (N=569) and 24-month (N=507) assessments. At baseline, participants were on average 14.2 years old (SD=2.1), 22.0% were Non-Hispanic White, 43.0% Hispanic, and 34.9% Non-Hispanic Black. Mean BMI-for-age percentile was 97.7 (SD=3.3). A notable percentage was from a disadvantaged background with 26.7% of parents having less than a high school degree. At baseline 10.0% had tried smoking cigarettes and drinking alcohol (4.0% smoking only; 18.0% drinking only), at 6 months 13.4% smoked/drank (4.0% smoking only; 21.4% drinking only), and at 24 months 17.8% smoked/drank (3.6% smoking only; 26.4% drinking only). Optimal multivariate linear regressions at each assessment generally indicated that older White males with worse grades and higher levels of depression and stress were more likely to report engaging in health risk behaviors, similar to findings with youth without diabetes. A demographic variable (SES), environmental factor (neighborhood safety), and both biological measures (BMIz and HbA1c) were not significant correlates at any assessment. These findings may assist health care providers in identifying youth with T2D who are at greatest risk for engagement in these health risk behaviors. Supported by: Merck, Sharp & Dohme PEDIATRICS—TYPE 1 DIABETES Guided Audio Tour: Pediatric Type 1 Diabetes—Highlights from Multicenter Studies (Posters: 1316-P to 1322-P), see page 19. Supported by: NIDDK/NIH (U01-DK61212), (U01-DK61230), (U01-DK61239), (U01DK61242), (U01-DK61254) A Potential New Diagnostic Index for Type 1 Diabetes & 1316-P JAY M. SOSENKO, JAY S. SKYLER, LINDA A. DIMEGLIO, JEFFREY P. KRISCHER, CRAIG A. BEAM, LISA RAFKIN, DELLA MATHESON, JERRY P. PALMER, DIABETES PREVENTION TRIAL-TYPE 1 AND TRIALNET, Miami, FL, Indianapolis, IN, Tampa, FL, Seattle, WA 1315-P Pharmacokinetics and Pharmacodynamics of Sitagliptin in Adolescents With Type 2 Diabetes We assessed the use of a 60-minute index (Index60) for diagnosing type 1 diabetes (T1D) among those who underwent serial 2-hr OGTTs for diagnostic surveillance in the Diabetes Prevention Trial-Type 1 (DPT-1). Index60 is derived from a proportional hazards model that includes the fasting C-peptide, 60-minute glucose, and 60-minute C-peptide of baseline OGTTs in DPT-1. Since repeat testing is needed for the standard 2-hr glucose value ≥200 mg/dl (2-hr≥200) in the absence of symptoms or unequivocal hyperglycemia (ADA guideline), we particularly examined confirmation at the next OGTT (at or before the 6-month visit). When an Index60 value ≥2.30 (Index60≥2.30) was first exceeded, the frequency of diagnosis of T1D was comparable to the frequency of diagnosis when 2-hr≥200 was first exceeded [153/170 (90%) for Index60≥2.30 vs. 166/188 (88%) for 2-hr≥200]. Of 61 with Index60≥2.30 and a 2-hr glucose value <200 mg/dl on the same OGTT, 47 (77%) were diagnosed. Of 64 with 2-hr≥200 and an Index60 value <2.30 on the same OGTT, 45 (70%) were diagnosed. A higher proportion was confirmed for Index60≥2.30 (80/124; 65%) than for 2-hr≥200 (74/148; 50%). Of the 80 confirmed with Index60≥2.30, 32 could not also be confirmed with 2-hr≥200 on the same OGTTs. All 32 (confirmed solely with Index60≥2.30) were later diagnosed by standard criteria. If Index60≥2.30 had been used as a criterion in those 32, the diagnosis would have been made1.0±0.2 years (mean±SEM; p<0.001) earlier. The use of Index60≥2.30 and 2-hr≥200 together as criteria (whichever was first exceeded) instead of 2-hr≥200 alone, would have resulted in more confirmed (92 vs. 74; 24% increase). In summary, a diagnosis of T1D was at least as likely for Index60≥2.30 as for 2-hr≥200. Moreover, the confirmation rate was higher for Index60≥2.30 than for 2-hr≥200, and confirmation with Index60≥2.30 often preceded the standard diagnosis of T1D. Index60≥2.30 and 2-hr≥200 used together resulted in a higher number confirmed. As an adjunct to the 2-hr glucose, Index60 should contribute to the earlier diagnosis of T1D. IAIN P. FRASER, NAOMI NEUFELD, LARRY A. FOX, MARK S. KIPNES, TRACIE L. MILLER, PHILLIP S. ZEITLER, HENRY RODRIGUEZ, JOCELYN GILMARTIN, SUZI LEE, JACLYN K. PATTERSON, XIUJIANG LI, LATA MAGANTI, WEN-LIN LUO, DANIEL TATOSIAN, EDWARD A. O’NEILL, S. AUBREY STOCH, Whitehouse Station, NJ, West Hollywood, CA, Jacksonville, FL, San Antonio, TX, Miami, FL, Aurora, CO, Indianapolis, IN The increase in obesity has led to an increased incidence of type 2 diabetes (T2DM) in adolescents. There are limited antihyperglycemic treatment options in this population. This study evaluated the pharmacokinetic (PK) and pharmacodynamic (PD) properties of sitagliptin (SITA), a DPP-4 inhibitor, in adolescents with T2DM. In a double-blind study, 35 patients (10 to 17 yrs old) were randomized to single oral doses of SITA or placebo (PBO) in a 3:1 ratio. SITA doses (50, 100, or 200 mg) were tested in ascending order in up to 12 patients/dose. SITA AUC0-∞ increased in a dose-proportional manner, whereas SITA Cmax increased in a modestly greater than dose-proportional manner and C24hr increased in a modestly less than dose proportional manner (Table). Pooled across available doses, adolescents had an ~18% lower AUC0-∞ as compared to adults with T2DM, but this difference was contained within the prespecified bounds (0.50, 2.00) for similarity between adolescents and adults. Mean weighted average inhibitions (WAI) of plasma DPP-4 activity over 24 hours were significantly different between SITA and PBO (PBO-adjusted DPP-4 WAI = 67.2% for 50 mg, 73.8% for 100 mg, and 81.2% for 200 mg). The PK/PD relationship between plasma SITA concentration and DPP-4 inhibition was similar between adolescents (IC50 value = ~25.4 nM) and adults (25.7 nM). Given the observed PK and PK/PD data in this study, SITA 100 mg q.d. is an appropriate dose for further clinical evaluation in adolescents with T2DM. Supported by: NIDDK & For author disclosure information, see page 829. A344 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—TYPE 1 DIABETES ~2 Weeks ~6 Months~12 Months N=561 N=490 N=547 A1c mean ± SD(%) 11.5 7.3 ± 7.8 ± 2.3 1.3 ± 1.4 Injections NA 7.3 8.0 ± 1.3 ± 1.5 Pump 7.3 7.4 ± 1.3 ± 1.1 IDAA1c 50% 26% ≤ 9.0% Pump Use <1% 14% 33% 1317-P Treatment Patterns in Asymptomatic Patients With Type 1 Diabetes Diagnosed by Research Screening RACHELLE GANDICA, MARY PAT GALLAGHER, ELLEN GREENBERG, SARAH POLLAK, STEVEN COOK, ELIZABETH LEVINE, ROBIN GOLAND, TYPE 1 DIABETES TRIALNET STUDY GROUP, New York, NY We present the 2 year clinical follow up of subjects with asymptomatic type 1 diabetes (T1D) diagnosed by research screening, a unique population since new-onset T1D is generally diagnosed with symptoms of hyperglycemia. Fifteen subjects with asymptomatic new-onset T1D were referred to the Berrie Center for routine clinical care because of an abnormal oral glucose tolerance test done in the NIH TrialNet study. Subjects were followed for at least 24 months. Subjects’ mean age at T1D diagnosis was 12 ± 3 years (range 1-42): 8 males and 7 females. At T1D diagnosis, A1C was < 6% in 13 subjects; 14 had normal fasting blood sugars, and mean A1C was 5.7 ± 0.1% (range 4.9 to 6.6%). Insulin treatment was not recommended at T1D diagnosis in 9 subjects because self-monitored blood sugars and A1C were in the normal range. Insulin was initiated at a mean of 11 months after diagnosis (range 1-20 months) in this group for elevated self-monitored blood sugars. Six subjects were started on insulin immediately at T1D diagnosis because of elevated self-monitored blood sugars. (Table 1). No subject had DKA or was hospitalized for hyperglycemia. Clinical management of asymptomatic new-onset T1D is variable. Initiation of insulin at biochemical T1D diagnosis occurred in only 6 of 15 subjects. The clinical outcome based on A1C after 2 years appears similar regardless of timing of insulin start but more information is needed. A randomized trial should be done to guide therapy of asymptomatic new-onset T1D. Age Age range Time from diagnosis to insulin start Time from diagnosis to A1c > 6% A1C at 0 months A1C at 12 months A1C at 24 months BMI Z score at 0 months BMI Z score at 24 months 44% & 1319-P Hypoglycemia is one of the most important barriers to the achievement of the metabolic target in young patients. We evaluated the incidence of severe hypoglycemia (SH) in children and adolescents with T1DM cared for by 29 Italian diabetes Centers, and identified the main risk factors related to the patients and their parents. Information on clinical and sociodemographic patient characteristics and their family was collected during a routine visit. Overall, 2025 consecutive patients were evaluated (mean age 12.4±3.8 years; 53% males; mean diabetes duration 5.6±3.5 years; 21.1% treated with CSII; mean Hba1c levels 7.9±1.1). During 12 months 102 (5.0%) patients experienced one or more SH episodes (incidence of 7.7 events/100 patients-years). At multivariate Poisson regression analysis the risk of SH was 44% higher in male children (OR 1.44; 95%CI 1.04-1.99), 2 times higher when the patient used short acting insulin analogues (OR 2.11; 95%CI 1.173.79) compared to regular insulin and 54% lower for patients treated with insulin glargine (OR 0.46; 95%CI 0.22-0.95) compared with NPH insulin. Mother’s age was inversely associated with the risk of SH (OR 0.95; 95%CI 0.92-0.97). After adjusting for clustering (center effect) the magnitude of the risk associated with short acting insulin analogues was decreased and no longer statistically significant (OR 1.61; 95%CI 0.87-2.97). Similarly, the protective effect of insulin glargine as compared with NPH insulin was no longer significant, although a lower risk associated with insulin glargine was still present (OR 0.51; 95%CI 0.19-1.44). In conclusion, SH still represents an important problem in children and adolescents with T1DM. The experience of the specialist in managing insulin therapy can play an important role, and the incidence of SH could be reduced through better use of existing therapeutic options and better education to individuals with diabetes and their families. 1318-P EDA CENGIZ, CRYSTAL CONNOR, KATRINA RUEDY, WILLIAM V. TAMBORLANE, CRAIG KOLLMAN, ROY BECK, JOYCE M. LEE, GEORGEANNA KLINGENSMITH, MARIA J. REDONDO, JANET SILVERSTEIN, PEDIATRIC DIABETES CONSORTIUM, New Haven, CT, Tampa, FL, Ann Arbor, MI, Aurora, CO, Houston, TX, Gainesville, FL & 1320-P SWEET: Securing Appropriate Services and Infrastructure for Pediatric and Adolescent Diabetes in Europe The PDC T1D NeOn study was designed to assess the natural history and clinical outcomes in children with T1D. Clinical measures were analyzed for 590 participants with an A1c result 18-27 mos after diagnosis (mean age 9.1 yrs, 51% female, 67% non-Hispanic White). A1c levels rose substantially 6-12 mos after T1D onset in association with waning of residual endogenous insulin secretion, as estimated by the % of participants with Insulin Dose Adjusted A1C (IDAA1c) values <9.0%(Table). A modest rise in A1c levels was observed from 12-24 mos, despite a further reduction in the % of participants with IDAA1c values <9.0%. The proportion of participants who switched from injection to pump therapy increased steadily and A1c levels were lower in pump vs injection therapy participants during the 2 years (Table). Moreover, a repeated measures regression model adjusting for clinical site, socioeconomic factors, race and age group showed that switching to a pump was associated with a mean 0.41% decrease in subsequent A1c levels (p<0.001). The substantial loss of residual endogenous insulin during the 2nd yr of T1D suggests that future β-cell preservation studies should target differences in c-peptide responsiveness between experimental and control groups in the 2nd rather than the 1st yr of T1D. The metabolic consequences of waning of the honeymoon period may have been blunted, in part, by a concomitant increase in use of insulin pump therapy. MICHAEL WITSCH, OLGA KORDONOURI, BAERBEL ASCHEMEIER, THOMAS DANNE, SWEET GROUP, Luxembourg City, Luxembourg, Hannover, Germany “SWEET” is an acronym derived from “Better control in Pediatric and Adolescent diabeteS : Working to CrEat CEnTers of Reference” and is based on a partnership of established national and European diabetes organizations (www.sweet-project.eu) led by the International Society for Pediatric and Adolescent Diabetes (ISPAD) with valuable contributions of IDF Europe, FEND, and PCDE. Peer audited Centers of Reference (COR´s) with a continuous electronic documentation of at least 150 pediatric patients with diabetes treated by a multidisciplinary team based on the ISPAD Clinical Practice recommendations have been created in 12 European countries. Although these clinics should not be regarded as representative for the whole country, the acknowledgement as COR includes a common objective of targets and guidelines as well as recognition of expertise in treatment and education at the center. In a first step, the SWEET Online platform allows twelve countries in eleven languages to connect to one unified diabetes database. Aggregate data is de-identified and exported for longitudinal health and economic data analysis and includes now more than 92,000 patient visits. In 2012, data was documented of 5,766 youth with type 1 diabetes, 58 with type 2 and 110 For author disclosure information, see page 829. Guided Audio Tour poster A345 POSTERS VALENTINO CHERUBINI, RICCARDO BONFANTI, DONATELLA LO PRESTI, GIUSEPPE LUCISANO, CLAUDIO MAFFEIS, CARLAMARIA MONCIOTTI, IPPOLITA P. PATERA, FABIO PELLEGRINI, BASILIO PINTAUDI, IVANA RABBONE, MARIA CHIARA ROSSI, STEFANO ZUCCHINI, ANTONIO NICOLUCCI, SHIP-D STUDY GROUP, Ancona, Italy, Milan, Italy, Catania, Italy, Santa Maria Imbaro, Italy, Verona, Italy, Padova, Italy, Passoscuro, Italy, Turin, Italy, Bologna, Italy Clinical Outcomes in Youth Within the 2nd Year of Type 1 Diabetes: Results of the Pediatric Diabetes Consortium (PDC) T1D New Onset (NeOn) Study & 39% Severe Hypoglycemia in Italian Pediatric Population With Type 1 Diabetes Mellitus: A Multicenter Retrospective Observational Study Supported by: NIH; TrialNet ADA-Funded Research ~24 Months N=508 8.1 ± 1.5 8.4 ± 1.7 7.7 ± 1.0 10% Supported by: Novo Nordisk, Inc.; DK020572 Insulin Treatment Insulin Treatment Not Recommended Initiated at T1D at T1D Diagnosis Diagnosis N=9 N=6 17 ± 4 years 5 ± 1 years 3 to 42 years 1 to 7 years 11 ± 7 months 0 months 8 ± 1 months 0 months 5.5 ± 0.1% 6.1 ± 0.1% 6.2 ± 0.4% 6.3 ± 0.2% 7.1 ± 0.3% 6.9 ± 0.2% 0.2 0.6 0.0 0.9 & ~18 Months N=566 8.0 ± 1.5 8.2 ± 1.6 7.6 ± 1.1 16% Clinical Diabetes/ Therapeutics & PEDIATRICS—TYPE 1 DIABETES with other diabetes forms. The median age was 14.04 years and diabetes duration was 5.07 years. Among centers notable differences in the average HbA1c was present ranging between 7.5 and 9.4%, respectively. This benchmarking data will now be used in quality control circles to exchange best practices. Through their network, the COR´s wish to obtain political power on a national and international level and to facilitate dissemination of new approaches and techniques. The SWEET Group hopes to extend from the initial group of centers within countries, throughout Europe and beyond. OB, respectively (p=0.59). Frequency of HTN, dyslipidemia and/or MA was relatively low (Fig. 1). By multivariate analysis, OB youth were more likely to have HTN (p<0.001) and dyslipidemia (p<0.001), but slightly less likely to have MA (p=0.05), than NW youth. In OW youth, frequency of HTN and MA were intermediate but not statistically different from NW youth, while dyslipidemia was significantly higher in OW than NW youth (p=0.006). In conclusion, OB youth with T1D are more likely to have concomitant HTN and dyslipidemia than NW youth, which may increase their cardiovascular disease risk. The slightly lower rate of MA in the OB group merits further investigation. Supported by: Bayer; DexCom, Inc.; Eli Lilly and Company; Medtronic, Inc.; Novo Nordisk, Inc.; sanofi-aventis & 1321-P Polyunsaturated Fatty Acids (PUFA) and Cardiovascular Risk Factors in Youth With Type 1 Diabetes (T1D): SEARCH Nutrition Ancillary Study POSTERS Clinical Diabetes/ Therapeutics SARAH C. COUCH, JAMIE L. CRANDELL, ABIGAIL PEAIRS, ANGELA D. LIESE, AMY S. SHAH, LAWRENCE M. DOLAN, JANET TOOZE, TESSA L. CRUME, IRENA B. KING, ELIZABETH MAYER-DAVIS, Cincinnati, OH, Chapel Hill, NC, Columbia, SC, Winston-Salem, NC, Aurora, CO, Albuquerque, NM Individual plasma n-3 and n-6 PUFAs and estimates of fatty acid desaturase activity may influence plasma lipids and thereby cardiovascular disease (CVD) risk. In youth with T1D, altered PUFA synthesis has been reported, but associations with CVD risk factors have not been ascertained. We investigated whether individual plasma phospholipid (PL) n-6 and n-3 PUFAs and delta-5 desaturase (D5D) and delta-6 desaturase (D6D) activities were associated with fasting triglycerides (TG), total cholesterol (TC), low-density lipoprotein (LDL)-C and high-density lipoprotein (HDL)-C in 881 youth with T1D. In multivariate analyses adjusted for demographics (age, race, gender, parental education, and clinical site), disease factors (disease duration, insulin regimen and dose, HbA1c), and body mass index (BMI) the following associations (expressed as beta coefficients, * =p<.05; **p<.01) were observed: Supported by: Leona M. and Harry B. Helmsley Charitable Trust Guided Audio Tour: Pediatric Type 1 Diabetes—Adolescence and Beyond (Posters: 1323-P to 1329-P), see page 19. Plasma PL n-6 PUFA Plasma PL n-3 PUFA Desaturase activitya 1 2 3 4 5 6 Independent LA DGLA AA ALA EPA DHA D5D D6D outcomes Log-TG 0.00 0.04 -0.04* 0.44* 0.01 -0.02 -0.24* 0.07 TC 0.01 1.46 -1.26* 27.19 7.96** -0.25 -8.84* 3.58 LDL-C 0.12 -0.36 -0.09 12.57 3.28** -2.47 -0.10 -0.73 HDL-C -0.15 1.54 -0.60 6.59 4.57* 2.38 -5.93 4.03 a Estimated D5D activity = AA/DGLA; Estimate D6D activity = DGLA/LA; 1LA = Linoleic Acid; 2DGLA = Dihomo-γ-linolenic acid; 3AA = Arachidonic acid; 4ALA = Alpha linolenic acid; 5EPA = Eicosapentaenoic acid; 6DHA = Docosahexaenoic acid & ARIANA M. CHAO, ROBIN WHITTEMORE, KARL E. MINGES, KATHRYN M. MURPHY, MARGARET GREY, New Haven, CT, Philadelphia, PA Though it is recognized that duration of type 1 diabetes impacts HbA1c, psychosocial adjustment, and family functioning, few studies have explored these relationships among young adolescents. Further, the impact of developmental stage at diagnosis is not well established. The purpose of this study was to explore differences in HbA1c, psychosocial adjustment, and family functioning in young adolescents by duration of diabetes and developmental stage at diagnosis. We used baseline data from 320 participants (11-14 years; 55% female; 37% minority) enrolled in a multi-site clinical trial of a coping skills training program. Data on participant and clinical characteristics were obtained by self-report. Duration of diabetes was categorized into 5 years. Developmental stage at diagnosis was categorized into infant/toddler/preschool (diagnosed between 1-5 years), school age (diagnosed between 5-10 years), and early adolescence (diagnosed after 10 years). HbA1c was determined by DCA2000. Scales were used to measure self-management, perceived stress, response to stress, self-efficacy for diabetes, depression, quality of life, diabetesspecific family support, and diabetes-related family conflict. One-way ANOVA with Tukey’s post-hoc comparisons revealed that youth diagnosed for <2 years had lower HbA1c (p<.01) as well as higher self-management diabetes care activities and self-management diabetes communication (p=.02, .03). There were no significant differences by developmental stage except for HbA1c, where youth diagnosed in early adolescence had lower HbA1c (p<.01). Although psychosocial adjustment and family functioning did not differ by developmental stage at diagnosis, selfmanagement was higher among adolescents diagnosed for a shorter time. This suggests that young adolescents with longer diabetes duration need additional self-management support during the transition to adolescence to optimize metabolic control. In this sample, n-6 PUFAs and D5D activity were related to beneficial effects on plasma lipids, whereas findings were mixed relative to n-3 PUFAs. Further research is needed to determine whether associations observed are related to long-term health implications in individuals with T1D. Studies of genetic variants of desaturase enzymes in T1D and associations with PUFA status may help to understand the link between dietary fatty acids and blood lipids and the impact of PUFA metabolism on health. Supported by: NIH/NIDDK & 1323-P Developmental Stage at Diagnosis and Duration of Diabetes: Implications for Young Adolescents With Type 1 Diabetes 1322-P Prevalence of Cardiovascular Risk Factors in Obese Youth With Type 1 Diabetes (T1D) MARIA J. REDONDO, DONGYUAN XING, INGRID M. LIBMAN, SANJEEV MEHTA, JOANNE HATHWAY, KATHLEEN BETHIN, BRANDON NATHAN, MICHELLE ECKER, AVNI SHAH, KELLEE M. MILLER, VINCE CHEN, WILLIAM V. TAMBORLANE, JENISE WONG, ROY BECK, Houston, TX, Tampa, FL, Pittsburgh, PA, Boston, MA, Buffalo, NY, Minneapolis, MN, Stanford, CA, New Haven, CT, San Francisco, CA We studied the relationship between body mass index (BMI) and coexistence of hypertension (HTN), dyslipidemia and micro-/macroalbuminuria (MA) in youth <18 yrs with T1D≥1 yr enrolled in the T1D Exchange clinic registry. Analysis included 11,285 participants (median age 12.8 yrs, median duration 4.0 yrs, 49% female, 78% non-Hispanic white, 56% pump users) from 59 centers, classified as obese (OB, BMI ≥95th percentile), overweight (OW, 85-<95th) or normal weight (NW, 5-<85th). Underweight (<5th) were excluded. Current diagnoses of HTN, dyslipidemia and MA were collected from medical charts. The three groups were compared in logistic regression models for complications and linear regression model for A1c. We found that 64% of youth were NW, 22% OW and 14% OB. Mean HbA1c at enrollment was 8.5±1.5%, 8.6±1.4% and 8.6±1.4% in NW, OW and Supported by: NIH/NINR (2R01NR04009); Jonas Center for Nursing Excellence & For author disclosure information, see page 829. A346 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—TYPE 1 DIABETES Table. Change between Baseline and 2-year follow-up within T1D and non-DM T1D Non-DM Baseline 2-YR F/U Change Baseline 2-YR F/U Change Breakfast# 3.3±1.2 3.8±1.6 0.5±1.7** 3.5±1.0 3.7±1.8 0.2±1.7 Vegetables# 2.8±1.2* 3.6±1.4 0.9±1.5** 3.1±1.2* 3.8±1.3 0.7±1.3** Fruit# 2.9±1.2 3.8±1.3 0.9±1.4** 2.9±1.3 3.8±1.5 -0.9±1.6** PA d/wk 5.0±1.8 4.8±1.7 -0.3±2.2 4.9±1.7 4.5±1.9 -0.4±2.3 PA hrs/d 1.8±1.2* 2.2±1.4 0.4±1.6** 1.5±0.8* 1.9±1.3 0.4±1.2** PA level† 1.9±0.6 2.0±0.5 0.1±0.7 2.0±0.5 2.1±0.6 0.1±0.7 #Times per week †PA level: 1=mild, 2=moderate, 3=strenuous *p<0.05 comparing T1D and non-DM at baseline **p<0.05 comparing change between baseline and 2-year follow-up within T1D and non-DM 1324-P KELLY K. ZINN, BARBARA VELSOR-FRIEDRICH, SUE PENCKOFER, MEG GULANICK, Huntsville, TX, Chicago, IL The purpose of this study was to explore diabetes self-care practices among older adolescents, 15 to 18 years old, with type 1 diabetes (T1DM) and determine if there was a difference in self-care practices between adolescents who reported good (HbA1c < 9%) versus poor (HbA1c > 9%) glycemic control (GC). Six focus groups were conducted (3 good GC, 3 poor GC) by the same moderators. Sessions were digitally recorded and transcripts were analyzed using an inductive approach. Thirteen adolescents with good GC (mean age = 16.2 yrs, mean HbA1c = 7.7%, mean duration = 5.8 yrs) and 8 adolescents with poor GC (mean age = 16.5 yrs, mean HbA1c = 11.4%, mean duration = 7.8 yrs) participated. Fifty-two percent of the sample was female and 85% were White. Themes generated directly from the focus group data were categorized as 1) life as a teen with T1DM, 2) diabetes self-care, or 3) interactions with the healthcare team. Teens with good GC described life with T1DM as a struggle, but that they were stronger than the disease. In contrast, teens with poor GC felt that diabetes interfered with their life and that diabetes self-care was a burden for which they had to adjust their lifestyle. Teens with good GC have integrated diabetes self-care into their identity and are able to adjust diabetes self-care to fit their lifestyle. Teens with good GC felt physicians were impersonal where those with poor GC described physicians as condescending. Teens with good and poor GC felt nurses treated them like a person. Adolescents with T1DM struggle with the complex treatment regimen but those with good GC demonstrated acceptance of the disease as part of who they were as a person thus diabetes self-care was part of their routine. The findings suggest that healthcare professionals must develop a trusting relationship with older adolescents with T1DM and include them in treatment decisions. By treating teens with respect and as a valuable partner, healthcare professionals can facilitate acceptance of the disease and its treatment regimen. & & RACHEL LIEBERMAN, R. PAUL WADWA, FRANZISKA K. BISHOP, JANET K. SNELLBERGEON, CHRISTINA REINICK, DAVID MAAHS, St. Louis, MO, Aurora, CO Vitamin D deficiency is common and associated with increased cardiovascular disease (CVD) risk. Pulse wave velocity (PWV) is a marker of vascular stiffness associated with CVD. We hypothesized that Vitamin D (25(OH)D) levels would be inversely associated with PWV. Comparisons were made between adolescents with T1D (n= 178; age= 17.6 + 2.2 y; duration= 10.9 + 3.1 y; A1c= 9.1 + 1.7%) and non-DM controls (n= 46; age= 16.8 + 1.6 y). Vitamin D was categorized (deficiency < 20; insufficiency 21-29; and sufficiency > 30 ng/ml) and PWV measured in the carotid-femoral segment (Sphygmocor Vx,AtCor Medical, Lisle, IL). Vitamin D was slightly higher in adolescents with T1D v. controls (29.0 + 0.8 v. 25.5 + 1.6 ng/ml; p = 0.052). Vitamin D was significantly associated with PWV after adjusting for age, sex, quarter of the year, and ethnicity in adolescents with T1D (beta = -0.01 + 0.004, p = 0.02) but not in the non-DM adolescents (beta = -0.01 + 0.008, p = 0.17). Vitamin D remained significantly associated with PWV after additionally adjusting for hs-CRP in adolescents with T1D (-0.01 + 0.004, p = 0.009). After adjusting for BMI z-score, lipids, and BP, the relationship of Vitamin D with PWV was not significant (Table). Low Vitamin D levels are significantly associated with increased PWV in adolescents with T1D, but not independent of BMI, lipids, or BP. Further research is needed to determine if Vitamin D supplementation would be beneficial to lower CVD risk in adolescents with T1D. 1325-P Changes in Diet and Physical Activity in Adolescents With and Without Type 1 Diabetes Over 2 Years FRANZISKA K. BISHOP, R. PAUL WADWA, JANET K. SNELL-BERGEON, NHUNG NGUYEN, JENNA REYNOLDS, DAVID MAAHS, Aurora, CO Diet and physical activity (PA) are fundamental aspects of care in type 1 diabetes (T1D), but scant data exist on diet and PA behaviors of T1D adolescents over time, especially compared to non-diabetic (non-DM) controls. Data in 210 T1D (baseline age=15.4±2.1 yrs, T1D duration=8.8±3.0 yrs, A1c=9.0±1.6%, 52% male) and 67 non-DM (age=15.4±2.1 yrs, 51% male) adolescents were collected at baseline and again at 2-year follow-up (mean follow up=2.2±0.4 yrs). Diet data (meals/day, snacks/day, and weekly consumption of breakfast, fruit, and vegetables), and PA were collected using interviewer administered questionnaires. T-tests and chi-squared tests were used for comparisons. Both T1D and non-DM adolescents reported increased vegetable and fruit intake and increased PA (hrs/d on PA days) from the baseline to the 2-year follow-up and these changes were significant (Table). T1D adolescents reported eating breakfast more often from baseline to the 2-year follow-up while meals and snacks per day remained unchanged for both groups. Over 2 years, T1D and non-DM adolescents had a healthier diet with increased fruit and vegetable intake and increased PA. However, both groups are far from meeting the guidelines of daily fruit and vegetables intake Overall, T1D and non-DM adolescents have similarly poor dietary patterns despite T1D adolescents receiving dietary education and support through their diabetes care. Late adolescence could be an age range to target diet interventions. ADA-Funded Research & 1326-P The Association Between Vitamin D and Pulse Wave Velocity in Adolescents With and Without Type 1 Diabetes Supported by: NIDDK; JDRF & 1327-P The Impact of a Multidisciplinary Program using Motivational Interviewing to Improve Glycemic Control in Adolescents With Diabetes AMANDA G. PERKINS, LINDA GRAHAM, ANNA L. CASS, ELAINE C. MORELAND, JOYCE W. ADAMS, BRYCE A. NELSON, Greenville, SC, Columbia, SC Adolescents in the Diabetes Control and Complications Trial had significantly higher hemoglobin A1cs (A1c) than adults. They have poorer treatment compliance compared to other pediatric populations and thus have increased risk for long term complications. This retrospective study evaluates the efficacy of a behavior change program, “Get Real About Diabetes” (GRAD), in improving A1c in adolescents with type 1 diabetes. GRAD was developed at our pediatric endocrinology practice and is ongoing. It combines traditional medical/educational visits with a Pediatric Nurse Practitioner with motivational interviewing sessions with a Licensed Clinical Social Worker. Data from ninety-four patients with type 1 diabetes were analyzed. Median age was 13 years and median duration of participation was 13 For author disclosure information, see page 829. Guided Audio Tour poster A347 POSTERS Supported by: JDRF (11-2007-694); NIDDK (DK075630); CTSI (UL1RR025780) Clinical Diabetes/ Therapeutics & A Comparison of Type 1 Diabetes Self-Care Among Older Adolescents With Good and Poor Glycemic Control PEDIATRICS—TYPE 1 DIABETES months. Patients entered GRAD with a median A1c of 9.75%. This value was rising in the 6 visits prior to participation with a median increase of 0.4% (p<0.01). By the 10 th GRAD visit, median A1c decrease was 0.7%. Upon completion of GRAD participants had a median A1c increase of 0.1% from the A1c at program entry. Median A1c changes were not statistically significant for the overall population during GRAD participation. Subgroup analysis showed that patients with the highest baseline A1cs (4th quartile, median A1c 11.9%) saw the largest initial improvement with a median A1c decrease of 2.4% (p=0.02) by the 10 th GRAD visit. Further, patients demonstrated A1c stabilization for 6 visits after completing the program (median time 50 weeks). To our knowledge, this is the largest population studied using motivational interviewing in youth with diabetes. We concluded that participation in GRAD prevented deterioration in glycemic control in a population in which glycemic control was declining. While more research is needed in this area, this study emphasizes the feasibility of incorporating this unique approach in routine clinic care. 1328-P Development of a Parent Proxy Diabetes-Specific Transition Readiness Questionnaire POSTERS Clinical Diabetes/ Therapeutics & At baseline, LCs reported lapses in care of 11.6 ± 9.7 months. Compared to CCs, LCs at baseline had higher A1C levels (%) (10.86 vs. 9.43; p = 0.009), a greater incidence of severe hypoglycemia in the past 3 months (29% vs. 10%; p = 0.031), and more ED visits in the past 6 months (0.63 vs. 0.12; p = 0.02). They were older than CCs (21.0 vs 19.6 yrs; p = 0.000) and less likely to have health insurance (p = 0.000) or live with family (p = 0.007). At 12 months, 78% of CCs and 54% of LCs completed the program. The most common reasons for loss to follow-up included being unreachable, finding care elsewhere, and scheduling conflicts. Among completers at 12 months, both LCs and CCs had lower A1C levels (change in %) compared to baseline (LC -0.87 ± 1.14, p = 0.029; CC -0.37 ± 1.14, p = 0.055). Between-group differences were no longer significant for A1C levels (p = 0.136), hypoglycemia (no cases; cannot calculate p-value), and ED visits (p = 0.296). LCs and CCs did not differ on number of clinic visits in the study period (LC 2.72 ± 1.46, CC 3.04 ± 1.12, p = 0.341). Our study suggests that, for those who complete it, a transition program is effective in reducing A1C, hypoglycemia, and ED use among YAs with a history of lapsed diabetes care to levels comparable to those in continuous care. Loss to follow-up remains a major challenge in this population. Results of this pilot study need validation in a larger population due to the importance of reconnecting these vulnerable YA to medical care. JESSICA T. MARKOWITZ, LISA K. VOLKENING, LORI M. LAFFEL, Boston, MA 1330-P The Transition Readiness Assessment Questionnaire (TRAQ) is a patient self-report of preparation to transfer from pediatric to adult care for pediatric chronic disease. The TRAQ, modeled in stages of change theory, examines acquisition of self-care skills in 16-26 year olds with special healthcare needs. We adapted the 29-item TRAQ to a diabetes (DM)-specific parent report of youth’s self-care skills and assessed it in 187 parents (82% mothers; 92% white) of youth aged 5-21 yrs with DM (99% T1D). The DM-specific survey has 30 items with 5 response options, scored by averaging items for a total score of 1-5 (higher score = greater readiness). The revised survey had good internal consistency (Cronbach’s α=0.93) and demonstrated the TRAQ’s 2 factor structure (Self-Management α=0.86; Self-Advocacy α=0.88). Scores were normally distributed and increased with youth age (Figure). At all ages, self-advocacy scores were higher than self-management scores. In 21/30 questions, response variability was low, ≥50% of responses were identical. In 16/21 in which parents responded child does not do task, youth were younger than when parents chose other responses (12.7 vs 16.5 yrs; p<0.03). In 5/21 in which parents responded child always does task, youth were older (14.8 vs 10.5 yrs; p<0.0001). Acquisition of self-advocacy skills develops earlier than self-management skills; thus, one should focus on DM self-management to prepare youth for optimal transition. Early Elevation of AER Associated With Poor Gluten Free Diet Adherence in Young People With Diabetes and Coeliac Disease ANNA PHAM-SHORT, KIM C. DONAGHUE, GEOFFREY R. AMBLER, ALBERT K. CHAN, STEPHEN HING, JANINE CUSUMANO, MARIA E. CRAIG, Sydney, Australia Consensus guidelines from the ADA and ISPAD recommend screening for coeliac disease (CD) in type 1 diabetes (T1D), however there are limited data demonstrating adverse effects of untreated CD or non-adherence to the gluten free diet (GFD). In particular, it is unclear whether the risk of microvascular complications is different in people with both T1D and CD. We therefore examined (i) the complication rates in youth aged < 20 yrs with T1D, stratified by biopsy confirmed CD (CD+, n=129, 43% M) or not (CD-, n=2510, 48% M) and (ii) the association between GFD adherence and complications. Outcomes examined were retinopathy, early elevation of albumin excretion rate (AER, > 7.5µg/min) and peripheral/autonomic neuropathy. Explanatory variables included HbA1c, diabetes duration and GFD adherence (defined as normal TTG titres (n=69) or elevated (n=60). Early elevation of AER was associated with GFD non-adherence vs adherence (40% vs 23%, p=0.04). In multivariate logistic regression, early elevation of AER was associated with GFD non-adherence (odds ratio 2.4, 95% CI 1.0-5.4, P=0.04) and diabetes duration (1.1, 1.0-1.3, P=0.03). HbA1c was not significant in the model (1.2, 1.0-1.6, P=0.10). There was no association between GFD adherence and neuropathy or retinopathy. Since early elevation of AER is a recognised predictor of diabetic nephropathy, our data support the importance of GFD adherence in CD. Characteristics at most recent visit CDCD+ p-value (n=2510) (n=129) Current age 16.5 [14.8 - 17.9] 16.1 [14.3 - 17.8] 0.13 GFD adherent GFD non-adherent p- value (n=69) (n=60) 15.8 [14.1 - 17.8] 16.4 [14.3 - 17.9] 0.43 Diabetes duration (yrs) 9.3 [6.3-12.2] < 0.001 9.2 [6.6 - 12.2] 9.4 [6.3 - 12.3] 0.81 6.6 [3.0 - 10.3] 6.2 [3.1 - 9.6] 6.7 [2.8 - 11.5] 0.54 7.2 [4.9 - 10.4] Coeliac disease duration (yrs) & 1329-P Outcomes of a 12-Month Transition Program (LEAP: Let’s Empower and Prepare) among Young Adults (YA) With T1D and a History of Continuous (CC) versus Lapsed (LC) Care HbA1c (%) 8.6 [7.7 - 9.6] 0.04 8.2 [7.6 - 9.0] 8.7 [7.8-10.0] 0.003 Elevated AER (> 7.5 µg/min) 600/2118 (28%) 36/117 (31%) 0.60 15/65 (23%) 21/52 (40%) 0.04 Peripheral nerve abnormality 627/2425 (26%) 36/128 (28%) 0.61 17/68 (25%) 19/60 (32%) 0.40 Retinopathy 556/2381 (23%) 27/125 (22%) 0.75 12/67 (18%) 15/58 (26%) 0.38 8.3 [7.6 - 9.3] ELIZABETH PYATAK, ANNE L. PETERS, PAOLA A. SEQUEIRA, JAMIE R. WOOD, LUCY MONTOYA, MARC J. WEIGENSBERG, Los Angeles, CA Individuals with T1D are often lost to care during transition from pediatric to adult care. In this study we attempted to identify and treat YA who had been lost to medical care. The LC group was given the one-year LEAP program given to those in their last year of pediatric care (the CC group). CC (n = 51) and LC (n = 24) participants in LEAP completed demographic, psychosocial, and A1C measures at baseline and 12 months. LCs were interviewed at baseline regarding their life circumstances and diabetes care. & For author disclosure information, see page 829. A348 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—TYPE 1 DIABETES 1331-P 1333-P Circulating Biomarkers of Adiposity and Inflammation in Obese and Lean Children With New Onset Autoimmune Type 1 Diabetes Skin Advanced Glycation Endproducts (AGEs) are Increased at Onset in Children With Type 1 Diabetes (T1D) MARIA J. REDONDO, LUISA M. RODRIGUEZ, DINAKAR IYER, MOREY W. HAYMOND, CHRISTIANE S. HAMPE, ASHOK BALASUBRAMANYAM, SRIDEVI DEVARAJ, Houston, TX, Seattle, WA EILEEN BAEZ, SHREEPAL SHAH, DANIA FELIPE, JOHN D. MAYNARD, JAMES HEMPE, STUART CHALEW, New Orleans, LA, Albuquerque, NM No DM 108 10.6±4.8 2.08±0.44 (2.17) 1.53±0.34 (1.60) New Onset T1D 30 9.3±4.4 0.007±0.002 2.24±0.44 (2.44) 1.65±0.35 (1.80) Established T1D 126 13.8±3.8 6.1±3.6 2.60±0.77 (2.50) 1.96±0.60 (1.87) Adjusted SIFs for 405 [kx0.5, km0.5] (p=0.018) and 420 [kx0.5, km0.5] (p=0.030) were higher in children with new onset T1D compared to those without diabetes. There was no difference between new onset and nondiabetics for SIF at 375nm or corrections using (kx=1.0, km=0). Measurement of SIF at 405 and 420 nm with (kx=0.5, km=0.5) correction suggest that skin AGEs may already be increased at onset of T1D in youth. Further study will be needed to determine whether these changes are associated with later development of diabetes complications. Supported by: Texas Children's Hospital 1332-P “Let’s Empower and Prepare (LEAP)”: Efficacy of a Young Adult (YA) Type 1 Diabetes (T1D) Transition Program PAOLA SEQUEIRA, MARC J. WEIGENSBERG, JAMIE R. WOOD, SUSAN CLARK, LUCY MONTOYA, HEATHER SPEER, MARISA COHEN, VALERIE RUELAS, ELIZABETH A. PYATAK, ANNE L. PETERS, Los Angeles, CA, Orange, CA 1334-P Social Networking for Care Improvement and Panel Management ELISSA R. WEITZMAN, SKYLER KELEMEN, KATHARINE C. GARVEY, Boston, MA YA with T1D transitioning from pediatric to adult care are often lost to follow-up and have deterioration in glycemic control. This study was done to evaluate the efficacy of a transition program (LEAP) which consisted of structured education, transition coordinator, and a designated YA clinic in an urban public hospital led by a med-peds endocrinologist. Reported here are findings from a 1-yr prospective pilot study of 19-25 yr olds with T1D in their last year of pediatric care before leaving to adult care. We recruited: 1) Intervention group (IG, n=51), receiving routine diabetes care at 2 pediatric diabetes clinics plus access to the LEAP program and transfer to the YA T1D Clinic at age 21; and 2) Control Group (CG, n=30), receiving routine care at a 3rd pediatric diabetes clinic. Demographics, surveys, number of clinic visits, and A1C were obtained at baseline and 12mo. At baseline, groups had same A1C (9.4% vs 9.2%, p=0.58), but differed by ethnicity, gender, socioeconomic status (SES), and % with private insurance (IG 2% vs CG 60%, p=0.000). 75% in IG and 83% in CG completed the study. More completers who transitioned out of pediatric care in IG successfully achieved adult care follow-up compared to CG (88% vs 33%, p=0.000). At 12mo, IG had a lower 3 mo incidence of severe hypoglycemia (0% vs 16%, p=0.021), improved in Global Well Being (p=0.008) and Perceived Stress (p=0.023) compared to CG. A1C was significantly reduced across the year intervention in IG vs CG (-0.36% vs +0.49%, p=0.012), adjusting for SES, ethnicity, baseline A1C. Our results demonstrate that a structured comprehensive diabetes transition program with accessible adult care increases adult follow-up care and lowers A1C levels in lower SES patients even when compared to a higher SES group. Developing and studying similar transition programs may be useful in transitioning YA in a planned manner and maintaining diabetes control. More research is needed to investigate long term outcomes for this critical period for YA with T1D. Detailed self-care and behavioral health information about adolescents with Type 1 diabetes (T1D) is scarce in clinical settings but vital to healthcare quality. Using a social networking application (app) we collected structured reports from youth receiving care for T1D in an ambulatory clinic, aiming to: 1) validate patient-reported data from the app against the electronic medical record (EMR); 2) measure information gain on health risk behaviors; and 3) test acceptability of returning aggregated app data to clinicians for panel management. Of 88 enrolled patients ages 13-25 years, 51% were female; mean age was 18 years. Mean HbA1c% was higher in the general clinic population (8.4%) than among participants; groups did not differ by age or gender. App data were current with EMR data (Table 1); 84% were entered within 30 days of the referent visit. Nearly twice as many severe hypoglycemic events were identified using the app than EMR. App data substantially enhanced EMR reports of health risk behaviors. Of 32 patients reporting alcohol use via the app, 27 (84%) had no mention or a negative report of alcohol use at the concurrent visit. Similar results were seen for tobacco (4/5 reporting use in the app had no EMR mention) and marijuana (10/11 reporting use had no EMR mention). Clinicians reported no workflow interruptions and high utility of app data. Table 1. Concordance of patient-reported app and EMR data Disease Factors Patient App Report N (%) EMR N (%) Concordance* Insulin pump use 66 (75.9) 67 (76.1) 96.6% Continuous Glucose Monitor (CGM) use 9 (10.2) 4 (4.6) 93.2% Any past-year severe hypoglycemia 13 (14.8) 7 (8.0) 88.6% HbA1c% (average, SD) 7.7 (0.9) 7.8 (1.1) 74.0% Age at diabetes diagnosis (average, SD) 10.4 (4.8) 10.3 (4.7) 85.2% * Calculated as the percentage of participants with an exact match between app report and EMR Complementary patient-centered data sources can inform diabetes care in a model that yields valid and novel data without burdening clinicians or workflows. Supported by: CIMIT (220376); Harvard Catalyst (027343.386541.05230); CDC (PO1HK000088-01) ADA-Funded Research & For author disclosure information, see page 829. Guided Audio Tour poster A349 POSTERS Group n Age (yrs) Duration (yrs) SIF405[kx0.5,km0.5] (LSM) SIF420[kx0.5,km0.5] (LSM) Clinical Diabetes/ Therapeutics AGEs mediate complications of aging and diabetes. Skin AGEs can be estimated from skin intrinsic fluorescence (SIF) which corrects for melanin, hemoglobin and light scattering. We previously reported that SIF is higher in children with established diabetes compared to nondiabetics. However there have been no prior studies evaluating SIF of children with new onset T1D. Children without T1D and patients with new onset and established T1D were evaluated. New onset patients were studied within 1 week of diagnosis. Skin fluorescence (440 - 655 nm) was excited by LEDs at 375, 405 and 420 nm using a Scout DS (VeraLight, Albuquerque, NM). SIFs were generated by correcting for excitation and white LED reflectance using (kx=1.0, km=0) or (kx=0.5, km=0.5). SIFs between groups were compared by least squares means (LSM) adjusted for the effects of age, gender, and race using PROC GLM of SAS. Results reported as unadjusted mean±1SD. Inflammation plays a pivotal role in the development of type 1 diabetes (T1D) and its complications. However, the contribution of obesity-induced inflammation to pediatric T1D is poorly understood. We aimed to study biomarkers of adiposity and inflammation in children with new onset autoimmune T1D. We prospectively studied 32 lean (BMI<85th percentile) and 18 obese (BMI≥95th percentile) children with newly diagnosed autoimmune T1D (age range 2-18 yrs) for up to 2 yrs after diagnosis. Serum adipokines (leptin, total and high molecular weight [HMW] adiponectin, omentin, resistin, chemerin, visfatin) and cytokines (interferon [IFN]-gamma, interleukin [IL]-10, IL-12, IL-6, IL-8 and tumor necrosis factor [TNF]) were measured 7 wks (median) (range 3-16 wks) after diagnosis. Lean children were 71.9% non-Hispanic White, 21.9% Hispanic and 6.3% African-American, compared with 27.8%, 55.6% and 16.7%, respectively, among obese children (p=0.01). Obese T1D children, compared with lean T1D children, had higher serum levels of leptin (p=0.0001), visfatin (p=0.0001), chemerin (p=0.001) and TNF (p=0.002), and lower total adiponectin (p=0.04) and omentin (p=0.003), after adjustment for race/ethnicity and Tanner pubertal stage. Among children with follow-up ≥1 yr, higher leptin levels at 7 wks of diagnosis predicted higher HbA1c 1.5 yrs later (range 1-2 yrs) independently of race/ethnicity and diabetes duration (p=0.0006, r2=0.41, n=41). Higher TNF was associated with obesity and female gender, after adjusting for race/ethnicity (p=0.0003). In conclusion, obese children with autoimmune T1D have a circulating pro-inflammatory adipokine and cytokine profile that may contribute to pathogenesis and thus represent a therapeutic target in this population. PEDIATRICS—TYPE 1 DIABETES 1335-P 1337-P Moderate Income as a Risk Factor for Youth With Type 1 Diabetes Vascular Function Is Not Significantly Impaired in Physically Fit Type 1 Diabetic Youth Compared With Healthy Peers POSTERS Clinical Diabetes/ Therapeutics SARAH JASER, KAITLYN RECHENBERG, ROBIN WHITTEMORE, SANGCHOON JEON, MARGARET GREY, Nashville, TN, New Haven, CT BRENT J. LOGAN, LINDA A. JAHN, ARTHUR L. WELTMAN, EUGENE J. BARRETT, Charlottesville, VA Low income has been established as a risk factor for poorer outcomes in youth with type 1 diabetes, but the effect of moderate income has not been studied. The purpose of this secondary analysis of data from a multi-site study was to compare glycemic control, self-management, and psychosocial outcomes (depression, stress, and quality of life) at different income levels in a diverse sample of early adolescents with type 1 diabetes. Youth (n = 320, mean age = 12.3 + 1.1, 45% male, 62% white, mean A1C = 8.3 + 1.4) completed established measures of self-management and psychosocial outcomes. A1C levels were collected from medical records, and caregivers reported on annual family income, which was categorized as high (> $80K), moderate ($40-80K) or low (< $40K). Youth from high-income families had significantly lower A1C values (mean = 7.9) than those from the moderateincome group (mean A1C = 8.9, p<.01) or the low-income group (mean A1C = 8.6, p<.01). Youth from the high-income group also reported significantly greater diabetes problem-solving and more self-management goals than those from the moderate- or low- income groups (both p <.01). In terms of psychosocial outcomes, youth from the high-income group reported significantly fewer symptoms of depression, lower levels of perceived stress, and better quality of life than those in the moderate- or low-income groups (all p < .05). After controlling for race/ethnicity and parental education in multivariate analysis, youth from the high-income group had lower A1C and reported better diabetes problem-solving, less depressive symptoms, and less perceived stress than those from the low- and moderate-income groups (all p<.05). While previous interventions have targeted youth from low-income families as being most vulnerable, results suggest that early adolescents from moderate-income families are also at risk for poorer outcomes and require greater attention. Cardiovascular disease (CVD) is a major cause of morbidity and mortality in Type 1 Diabetes (DM1). In their third to fifth decades risk ratios for CVD related deaths in DM1 are 8-20 fold higher than the general population. Current data also indicate that DM 1 youth may be less physically fit compared to agegender matched peers. Decreased fitness has been associated with vascular dysfunction, though data is limited. We evaluated DM1 (n=11) and non-DM1 (n=9) adolescents for fitness (measured by VO2max), body composition, and multiple measures of vascular function. The latter include tonometry to evaluate arterial stiffness, by augmentation index (AI) and pulse wave velocity (PWV), and ultrasound to assess flow-mediated dilation nitric oxide dependent vascular reactivity (FMD) and contrast-enhanced ultrasound of forearm muscle microvasculature (CEU). We found no differences in VO2max (40±3 vs 39.6±3 mL/min/kg). Our subjects were similar in age, BMI, and percent body fat. HgbA1C (5.4 vs 8.7%) and fasting glucose (88 vs 131 mg/dL) on the day of study were both lower in the controls (p=<0.01, each). FMD was slightly, but not significantly, lower in DM1 (10.9±2.2 vs 8.1±1.2%). Vascular stiffness was comparable (PWV = 4.9±0.2 vs 5.3±0.3 m/s. AI = 1.1±4 vs 0.2±5%, in controls and DM1, respectively) and muscle microvascular volume by CEU was similar at baseline and increased in response to insulin similarly in both groups. Our data suggest that when compared to age and fitness matched controls, adolescents with DM1 of > 1yr duration have similar vascular stiffness and exhibit normal microvascular responses to insulin despite inadequate glucose control. Nitric oxide mediated FMD may be modestly decreased. These findings suggest that adolescents with DM 1 who maintain fitness may be at a lower risk of developing early findings of vascular dysfunction. Supported by: NIH/NINR (2R01NR04009) Supported by: University of Virginia 1336-P 1338-P Cyclical Variation in HbA1c Values During the Year—Clinical and Research Implications Data Mining Methodologies for Detecting Insulin Omission Eating Disorder in Girls With Type 1 Diabetes NATHAN R. HILL, CATHERINE J. PETERS, REBECCA J. THOMPSON, DAVID R. MATTHEWS, PETER C. HINDMARSH, Oxford, United Kingdom, London, United Kingdom URI HAMIEL, YUVAL GREENFIELD, VALENTINA BOYKO, CHANA GRAPH-BAREL, LIAT LERNER-GEVA, BRIAN REICHMAN, ORIT PINHAS-HAMIEL, Ramat Gan, Israel Glycosylated haemoglobin (HbA1c) is a key process and outcome measure in diabetes. Reporting of this indicator varies Some schemes use single measures in the year and others an average of several taken during the year. Single measure reporting may be influenced by seasonal variation as may intervention studies in clinical trials. To quantify variation in HbA1c we subjected 5140 paediatric HbA1c measurements obtained from 1999-2012 to Time Series Analysis. Fourier Transform analysis described a waveform of maximal power/amplitude at a periodicity of 6 months that indicated a greater than random tendency for peaks and troughs in the time series to recur at this frequency and indicated a distinct rhythm and the decimal precision was determined using Autocorrelation analysis. This revealed regular oscillations within the dataset occurring every 6.9 months so that on average HbA1c values were 0.3% lower during the summer quarter compared to the winter quarter. Median clinic HbA1c for 2012 was 7.9% with 41% of children attaining a HbA1c < 7.5%. Children had a lower median HbA1c (7.7%) than adolescents (8.4%). Insulin pump therapy was associated with a lower median HbA1c (7.7%) than multiple daily injections (8.5%). These data demonstrate regular cyclicity in HbA1c values during the year. Submission of data to quality improvement schemes should be based on the average for the individual over the whole year. Clinical trial design should include estimates of HbA1c seasonal variation to avoid Type 1 and 2 statistical errors. Insulin omission is a specific eating disorder that occurs in patients with type 1 diabetes mellitus (T1DM), mostly in females, who omit or restrict their required insulin doses in order to lose weight. Diagnosis of this specific eating disorder is difficult. We aimed to use multiple clinical criteria to create an algorithm for the detection of insulin omission eating disorder. Data from 287 (181 females and 165 males) patients with T1DM and 26 patients with T1DM and eating disorders were used. The Weka (Waikato Environment for Knowledge Analysis) machine learning software, decision tree classifier based on the J48 algorithm with 10 fold cross validation was used to developed prediction models. Model performance was assessed by cross-validation in a further 43 patients. The insulin omission eating disorder and non-eating disorder populations were discriminated by: female sex, HbA1c>9.2%, more than 20% of HbA1c measurement above the 90th percentile, the mean of 3 highest delta HbA1c z-score >1.28, current age and age at diagnosis. The models developed showed good discrimination (sensitivity and specificity 0.88 and 0.74, respectively). The external test dataset revealed good performance of the model with a sensitivity and specificity of 1.00 and 0.97, respectively. In summary, using data mining methods we developed a clinical prediction model to determine an individual’s probability of having eating disorders. This model provides a decision support system for the detection of insulin omission eating disorders in adolescent females with T1DM. & For author disclosure information, see page 829. A350 Guided Audio Tour poster ADA-Funded Research PEDIATRICS—TYPE CATEGORY 1 DIABETES 1339-P 1341-P A Novel Mutation in GATA6 Causes Pancreatic Agenesis Excess Weight Gain during Insulin Pump Therapy Is Associated With Higher Total Daily and Basal Insulin Doses DIANA E. STANESCU, NKECHA HUGHES, PUJA PATEL, DIVA D. DE LEON, Philadelphia, PA 1340-P Transitioning from Pediatric to Adult T1D Care: Knowledge Alone Is Insufficient ANIA M. JASTREBOFF, DARRYLL CAPPIELLO, JODIE M. AMBROSINO, SYLVIA LAVIETES, PAULINA ROSE, MELINDA ZGORSKI, PATRICIA GATCOMB, WILLIAM V. TAMBORLANE, STUART WEINZIMER, KATE WEYMAN, New Haven, CT The transition from pediatric to adult diabetes care is a critical step in maintaining long-term health in patients with type 1 diabetes (T1D) but there is a paucity of data regarding the challenges that emerging young adult patients face in making the transition. In establishing the Yale T1D Bridge Clinic, we developed a 20 item quiz to assess diabetes knowledge and enable targeted education for our transitioning patients. The quiz encompassed basic diabetes pathophysiology and case-scenario management-based questions specific to treatment modality. The first 19 patients (11 female) who completed the quiz on enrollment in the clinic were aged 16-24 yrs with mean A1c 7.7% (range 6.7-10.5%). As shown in the figure, knowledge quiz score did not correlate with A1c (r=-0.26, p=0.3); patients with quiz scores of >85% were just as likely to have A1c levels >7.5% as they were to have values <7.0%. Thus, although many patients transitioning to adult care may have the knowledge necessary for successful diabetes self-care, this knowledge alone is insufficient to achieve glycemic control during the transition process. These data underscore the need for clinicians to carefully assess other factors that may impede glycemic control during the transition process, such as diabetes distress/burden of care, fear of hypoglycemia, depression, and disordered eating, and to develop better motivational strategies for patients with T1D during the transition process. 1342-P Carotid Intima-Media Thickness Is Increased in Youth With Type 1 Diabetes RACHEL M. SIPPL, FRANZISKA BISHOP, DAVID MAAHS, PAUL WADWA, JANET K. SNELL-BERGEON, Aurora, CO Cardiovascular disease (CVD) is increased in type 1 diabetes (T1D), but it is unknown how early in life these risks begin. This study examined carotid intima-media thickness (cIMT) as a surrogate CVD risk marker among adolescents with T1D and without diabetes (non-DM). Study participants were 29 T1D youths (age 18 ± 3 yrs, duration 11 ± 3 yrs) and 27 non-DM youths (age 17 ± 2 yrs) from the Determinants of Macrovascular Disease in Adolescents with Type 1 Diabetes. cIMT was measured in the common carotid artery using the Panasonic CardioHealth ® Station bilaterally, and the mean cIMT from the largest side was used. CVD risk factors examined included gender, age, body mass index (BMI), blood pressure and fasting lipids. Two sample t-tests and multivariable linear regression were used to examine the relationship between diabetes status and cIMT. Youths with T1D had increased cIMT when compared to non-DM youths (mean±SD=0.54±0.10 vs 0.48±0.05 mm, p = 0.01). Gender, age, body mass index (BMI), and SBP were associated with cIMT, and so were included in linear regression models (Table). cIMT remained significantly higher in youth with T1D when adjusting for age and gender, and when further adjusted for BMI and SBP. Youths with T1D have increased cIMT when compared to non-DM youth of the same age and gender, indicating that the higher risk for subclinical atherosclerosis in type 1 diabetes is present as early as adolescence and is independent of obesity and hypertension. cIMT Least Square Means and 95% Confidence Intervals Model covariates T1D (n=29) Non-DM (n=27) P-value Age, Diabetes, Gender 0.53 (0.51-0.56) 0.48 (0.45-0.52) 0.03 Age, Diabetes, Gender, BMI 0.53 (0.50-0.56) 0.49 (0.46-0.52) 0.048 Age, Diabetes, Gender, SBP 0.53 (0.50-0.56) 0.49 (0.46-0.52) 0.04 Age, Diabetes, Gender, BMI, SBP 0.53 (0.51-0.56) 0.48 (0.46-0.51) 0.02 Supported by: K12DK094714 ADA-Funded Research & Supported by: JDRF (11-2007-694); NIDDK; Panasonic For author disclosure information, see page 829. Guided Audio Tour poster A351 POSTERS While higher total daily dose (TDD) of insulin has been associated with excess weight gain on insulin pump therapy, the role of higher total basal dose (TBD) of insulin on weight gain has not been studied. Therefore, we evaluated the impact of higher TBD on weight gain in relationship to glycosylated hemoglobin (HbA1c), hypoglycemic episodes, and change in body mass index (BMI) z score in a group of children and adolescents with type 1 diabetes mellitus (T1DM). One-year data from 91 (54 female and 37 male) patients (2.3 - 17.8 years of age), transitioned from multiple daily insulin regimen (glargine and premeal rapid acting insulin) to insulin pump therapy were reviewed retrospectively. The patients were divided into two groups based on change in BMI z score from baseline: Group 1 (no change or decrease) and Group 2 (increase). There were 33 patients (60.6% female) in Group 1 and 58 patients (58.6% female) in Group 2. Patients in Group 1 were younger than Group 2 (10.7 + 4.2 vs. 12.5 + 3.6 years; p<0.025) but had similar duration of diabetes (4.9 + 2.6 vs. 4.0 + 3.8 years; p=0.92) and pubertal status (2.5 + 1.7 vs. 3.1 + 1.4; p=0.09). Group 2 had a significantly higher TDD (0.79 + 0.23 vs. 0.89 + 0.19 U/ kg/day; p<0.025) and TBD (0.38 + 0.1 U/kg vs. 0.28 +0.1 U/kg; p<0.025) than Group 1. Indeed, Group 1 had higher bolus: basal insulin ratio (1.82 + 0.6 vs. 1.46 + 0.6, p<0.01) than Group 2. Groups 1 and 2 had similar HbA1c values (7.67 + 0.71 vs. 7.70 + 0.64%; p=0.79), hypoglycemic episodes (1.01 + 0.4 vs. 1.5 + 0.9 episodes/patient/year; p=0.28) and activity levels (2.3 + 0.5 vs. 2.1 + 0.7; p=0.15). Excess weight gain was associated with not only higher TDD but also higher TBD during insulin pump therapy independent of glycemic control, frequency of hypoglycemic episodes and activity level. Therefore, frequent evaluations of bolus and basal insulin doses during insulin pump therapy is warranted in order to avoid excess weight gain. Clinical Diabetes/ Therapeutics CLAUDIA C. BOUCHER-BERRY, ELAINE PARTON, RAMIN ALEMZADEH, Chicago, IL, Milwaukee, WI Haploinsufficiency of GATA6 has been recently linked to pancreatic agenesis and cardiac malformations. Our aim is to describe a novel mutation in GATA6 in a female infant diagnosed shortly after birth with truncus arteriosus, absent gallbladder, unilateral hydronephrosis and hydroureter, and neonatal diabetes. The child developed hyperglycemia in the first 24 hrs of life and required insulin therapy. An abdominal CT at 1 month of life showed the pancreas was reduced to a small portion of the pancreatic head, consistent with pancreatic agenesis. The child expired at 3 months of age after cardiorespiratory failure due to her severe cardiac malformation. Analysis of the child’s genomic DNA reveled a novel duplication causing a frameshift at amino acid position 204 resulting in stop codon downstream (pHis204ValfsX97). The truncated protein is predicted to lack DNA binding because of the absent zinc finger domains. Plasmids containing wild type and mutated GATA6 were transfected into HEK 293T cells and Western Blotting of cell extracts confirmed the small molecular weight of the mutant protein. Mutations in GATA6 have been demonstrated to be the most common cause of pancreatic agenesis, and have been found even in cases of diabetes, diagnosed outside the neonatal period. The mutation described in this case expands the list of GATA6 mutations causing pancreatic agenesis and cardiac malformations. The severe phenotype in this case underlines the importance of GATA6 on the development of multiple organs. PEDIATRICS—TYPE 1 DIABETES systems (one blinded, and the other display) on either abdomen and/or upper buttocks for seven days home use. The OneTouch® Verio IQ meter (SMBG) was used for CGM calibrations, diabetes management and served as the comparative reference glucose values. All subject completed the study. 17 subjects were male; 29 subjects were white. The mean age of subjects was 12 ± 4; 3 subjects were under age 6. Duration of diabetes was 4 ± 3 yrs. 24 subjects used CSII for insulin therapy; 6 subjects were on MDI therapy. The baseline HbA1C was 8.6 ± 1.3%. There were no serious adverse events, sensor fractures, or infection complications. Except for 2 (3%) infrequent mild skin irritations at the sensor adhesive site, no other device related adverse events were reported. A total of 1882 comparative paired CGM and SMBG values were evaluated. CGM readings were within 20% (or 20 mg/dL for SMBG value ≤ 80 mg/dL) of the SMBG values 77% of the time (95% CI: 75% to 79%). The Mean Absolute Relative Difference (MARD) vs. SMBG was 15%. The clinically accurate CEG A zone was 76%, and the combined A+B zones was 98%. The system performed consistently between day (6 AM to 6 PM) 76% (N=1260, 95% CI: 74% to 79%), and night 79% (N=622, 95% CI: 75% to 82%), at different sensor wear locations, abdomen 78% (N=821, 95% CI: 76% to 81%), and buttocks 76% (N=1061, 95% CI: 74% to 79%), all p values >0.10. The duration of diabetes, age, sex, BMI and baseline A1C showed no significant impacts on the performance. The system provided glucose readings on average 92% of time with 81% of the sensors lasting to day 7, with an average 6% CV of the concurrently worn sensors. This is the first report of the new G4 PLATINUM CGM used in youths. The system was safe to use and performed similarly in abdomen and upper buttocks, and at different time of the day. The clinical accuracy, on-time and sensor life of the system all compared favorably to the CGM system currently approved for pediatric use. 1343-P An Independent Inverse Association between Soluble CD25 and Beta-Cell Function in Young People With Type 1 Diabetes POSTERS Clinical Diabetes/ Therapeutics LOREDANA MARCOVECCHIO, KATE DOWNES, PAMELA CLARKE, JASON COOPER, JOHN TODD, DAVID DUNGER, Cambridge, United Kingdom To assess the potential association between a marker of the immune function (soluble CD25 (sCD25)) and circulating levels of C-peptide in young people with childhood-onset type 1 diabetes (T1D) and a variable diabetes duration. C-peptide and sCD25 levels were measured in non-fasting serum samples collected from 231 young people (136 males; median age (interquartile range): 14.5 [12.9-16.3] years, T1D duration: 5.0 [2.3-7.0] years, age at T1D onset: 10.4 [8.2-12.4] years) followed in the Nephropathy Family Study. Median C-peptide levels were 24.7 [9.6-67.7] pmol/l and median sCD25 levels were 3179.2 [2536.5-3936.1] pg/ml. C-peptide levels (log transformed) were positively associated with age at T1D onset (β=0.346, p<0.001), whereas a negative association was found with T1D duration (β=-0.390, p<0.001), log HbA1c levels (β=-0.158, p=0.021), and log sCD25 levels (β=0.223, p=0.001). In a multiple regression model, the association between log C-peptide and log sCD25 persisted after adjusting for age at T1D onset, duration and HbA1c levels (β=-0.158, p=0.018). In 195 subjects, where C-peptide and sCD25 were re-assessed in a second sample collected after a median interval of 1.2 [range: 0.8-3.6] years from the first one, an inverse association between changes in C-peptide and sCD25 was found (β=-0.176; p=0.014), after adjusting for the effect of duration of diabetes. In this study sCD25 was significantly and inversely associated with C-peptide levels, even in subjects with long-duration T1D, likely reflecting the effect of the immune system on beta-cells. These preliminary findings need to be confirmed in larger studies, including a better assessment of beta-cell function. 1346-P T-Cell Auto Reactivity and C-Peptide at Onset of Pediatric Type 1 Diabetes (T1D) 1344-P MELISSA A. BURYK, H-MICHAEL DOSCH, INGRID LIBMAN, ROY K. CHEUNG, VINCENT C. ARENA, YIHE HUANG, MASSIMO PIETROPAOLO, DOROTHY J. BECKER, Pittsburgh, PA, Toronto, ON, Canada, Ann Arbor, MI Low Vitamin D Levels Observed in Over Half the Children and Adolescents in San Diego at Diagnosis of Type 1 Diabetes HIBA AL-ZUBEIDI, LUCERO LEON-CHI, RON S. NEWFIELD, San Diego, CA We previously reported that residual c-peptide at onset and during remission of T1D is associated with BMI z-score, age and HbA1c, but not islet autoantibody number. A relationship between islet T-cell autoimmunity and low c-peptide levels has been reported in adults with phenotyp