Evaluation of a New CA15-3 Protein Assay Method: Optical Protein

Transcription

Technical Briefs
Use of Low Concentrations of Human IgA Anti-Tissue
Transglutaminase to Rule Out Selective IgA Deficiency
in Patients with Suspected Celiac Disease, Eloy Fernández,1* Carlos Blanco,1 Sara Garcı́a,1 Angeles Dieguez,2 Sabino
Riestra,3 and Luis Rodrigo3 (1 Biochemistry Department,
Hospital Cabueñes, Gijón, Spain; 2 Immunology and
3
Gastroenterology Departments, Hospital Central Asturias, Oviedo, Spain; * address correspondence to this
author at: Biochemistry Department, Hospital de Cabueñes, Cabueñes s/n, 33394 Gijón, Asturias, Spain; fax
34-985185022, e-mail eloy.fernandez@sespa.princast.es)
Selective IgA deficiency (IgAD) is the most common
well-defined primary immunodeficiency disorder in humans (1, 2 ). Patients with IgAD frequently share the
haplotype HLA-DQ2, which is also associated with celiac
disease (CD) (3 ), and therefore have a 10- to 20-fold
increased risk of CD (4 ).
High concentrations of anti-tissue transglutaminase (htTG) IgA antibody are used to diagnose CD (5, 6 ), but
antibodies are not increased in IgAD (7, 8 ). This has led to
the use of assays for total IgA when testing for CD and/or
testing for IgG-class antibodies against h-tTG (9 ).
The aim of our study was to assess whether a secondgeneration IgA anti-h-tTG assay can detect IgAD, as the
concentrations of IgA antibodies would be expected to be
very low. This could eliminate the expense for additional
tests in many individuals.
We studied 4 groups of patients. The disease group
included 28 patients with IgAD [18 females (median age,
38 years; range, 8 –79 years) and 10 males (median age, 24
years; range, 5–75 years)] diagnosed between June 2001
and May 2003. All had total IgA concentrations ⬍0.05 g/L
and normal concentrations of IgG and IgM and had a
clinical diagnosis of IgAD. The diseased control group
consisted of 63 patients [32 males (median age, 56 years;
range, 1–92 years) and 31 females (median age, 31 years;
range, 1– 82 years)] in whom total IgA was ⬎0.05 g/L but
below the lower limit of the reference interval (0.70 g/L;
median IgA, 0.39 g/L; range, 0.07– 0.69 g/L). The final
diagnoses in the adult diseased controls were multiple
myeloma (30 patients), chronic lymphoid leukemia (4),
anemia (4), chronic kidney failure (4), Waldestrom disease
(3), acute pulmonary edema (2), scleroderma (1), and
acute pericarditis (1). The 14 pediatric patients of this
group presented with complaints in relation to a febrile
syndrome, diarrhea, and pneumonia. The healthy control
group included 82 consecutive blood donors [48 males
(median age, 42 years; range, 25–54 years) and 34 females
(median age, 28 years; range, 21– 48 years)] with total IgA
above the lower reference limit (median IgA, 1.74 g/L;
range, 0.72–3.95 g/L). Finally, we studied sera from 773
consecutive pregnant women enrolled in a study of CD.
The study was performed according to the principles of
the Helsinki Declaration, and oral informed consent was
obtained from each participant.
We obtained 5 mL of blood from each individual and
measured IgA anti-tTG antibodies in serum with a commercially available sandwich ELISA with human recom-
1014
Clinical Chemistry 51, No. 6, 2005
binant tTG from eukaryotic cells of Lepidoptera (Baculovirus/Sf9 system; Celikey; Pharmacia Diagnostics GmbH).
Results are reported as absorbance values. All measurements were made in a single batch on a Triturus ELISA
automated analyzer (Grifols) by a single operator following the manufacturer’s instructions. The intraassay imprecision (CV) of the h-tTG ELISA was 7.2% at 0.07 g/L, 8.2%
at 0.12 g/L, and 5.7% at 0.46 g/L (n ⫽ 20). The interassay
CV was 12% at a serum IgA concentration of 0.07 g/L,
11% at 0.12 g/L, and 7.9% at 0.46 g/L (n ⫽ 12).
To detect IgAD, total serum IgA was also measured in
all IgAD patients and controls by nephelometry (BN II;
Dade-Behring). IgA ⬍0.05 g/L was considered to be
indicative of selective IgAD.
The Mann–Whitney U-test was used to estimate differences in the anti-h-tTG absorbance readings between
groups, and the Spearman rank method was used to
calculate the correlation between anti-h-tTG and total IgA.
ROC analysis was performed with MedCalc®, Ver. 7.4.4.1
(MedCalc Software). For all statistical analyses, a twotailed P ⬍0.05 was considered significant.
Anti-tTG absorbance increased with total IgA serum
concentration (Fig. 1A) and was lower in IgAD patients
than in both diseased and healthy controls (P ⬍0.0001). In
27 of the 28 IgAD patients (96%), the anti-h-tTG absorbance was ⬍0.013. In 173 individuals of the 3 groups
studied, the total IgA concentrations and anti-h-tTG absorbances were correlated [rs ⫽ 0.926; 95% confidence
interval (95% CI), 0.901– 0.944]. ROC curve analysis for
distinguishing IgAD patients from all diseased and
healthy controls provided an optimal cutoff (minimum
sum of false-positive and false-negative rates) of 0.013 for
the anti-h-tTG assay; at this cutoff, the sensitivity, specificity, and area under the curve were 96% (95% CI,
82%–99%), 83% (76%– 89%), and 0.94 (0.89 – 0.97), respectively. At a cutoff of 0.022, the sensitivity and specificity
were 100% and 63%, respectively. Anti-h-tTG absorbances
of IgAD patients and diseased controls overlapped (Fig.
1A): In 24 (38%) and 50 (79%) of 63 diseased controls, the
anti-h-tTG absorbance values were ⬍0.013 and ⬍0.022,
respectively. As expected, only 3 of 82 (3.7%) healthy
controls had anti-tTG values ⬍0.022, and none had an
absorbance reading ⬍0.013.
Of the 773 pregnant women (median total IgA, 1.7 g/L;
range, 0 –5.65 g/L), 6 (0.77%) had a total IgA serum
concentration ⬍0.05 g/L. An anti-tTG absorbance cutoff
of 0.013 provided the highest sum of sensitivity (100%)
and specificity (98.3%) in detecting IgAD in this group of
pregnant women (Fig. 1B). Thus, when the anti-h-tTG
result is known and is ⬎0.013, IgAD can be excluded with
some confidence, and total IgA would need to be measured to exclude IgAD in only 13 of 767 (1.7%) patients
with absorbances ⬍0.013. Furthermore, a cutoff of 0.022,
as described above, also detected all IgAD cases but with
a lower specificity (94%), thus leading to additional 46
total IgA determinations. At the IgAD prevalence of
0.77% found in our study, the positive and negative
predictive values at an anti-h-tTG cutoff of 0.013 were
31% (95% CI, 14%–57%) and 100% (99%–100%), respectively.
The IgA anti-h-tTG assay was positive in 2 of the 773
pregnant women tested [0.26%; ⬎100 kilounits/L in both
cases, expressed as Celikey arbitrary units calculated
according to calibration curve (ROC-based cutoff, 2.6
kilounits/L)]. Informed consent for intestinal biopsy was
obtained in the 2 cases, and biopsy specimens obtained
from the second duodenal portion during gastroduodenoscopy showed classic subtotal (stage 3b) and total
(stage 3c) villous atrophy, respectively, according to a
modified Marsh classification (10 ). The prevalence of
biopsy-confirmed CD in these women was 1 in 387 (2.6
per 1000; 95% CI, 0.4 –10.4) and is in accordance with that
reported previously in the general population of our area
(11 ). Given the high positive predictive value of the
1015
copresence of anti-endomysium antibodies, anti-tTG, and
HLA DQ2– 8 haplotype (12 ), patients with both CDrelated antibodies and this HLA type should undergo
intestinal biopsy. However, given the relatively high
prevalence of the HLA DQ2 marker in the general population (20%–30%) (13 ), the presence of this CD genetic
marker alone should not be an indication for intestinal
biopsy in asymptomatic patients without abnormal findings in the biochemical studies and with negativity for
CD-related serology. None of the 6 women with IgAD
tested positive for either IgG anti-tTG or IgG anti-endomysium antibodies, and thus none was classified as
having CD.
There is an increased prevalence of CD in IgAD patients, and this condition is generally unknown at the time
of CD diagnosis. To avoid false-negative results for measurements of IgA-class immunoglobulins, two diagnostic
approaches have been proposed. The first approach uses
a 2-step strategy with the quantification of serum IgA and
assay of IgA anti-h-tTG; in those cases with IgAD, a
serologic test of IgG class is undertaken. The second
approach consists of the performance of a serologic test
for CD based on IgG as part of the first step.
Specific IgG testing can be useful for identification of
CD patients with IgAD (14 –16 ) and those who have
normal concentrations of total IgA but produce only IgG
anti-h-tTG (17 ). The use of IgG anti-tTG antibodies in
diagnostic and screening strategies has recently been
proposed to ensure the detection of CD in IgAD individuals (9, 15, 16 ), but limited sensitivity of different IgG
anti-h-tTG tests has been reported in patients with IgAD
(6 ).
The present data suggest that isolated measurement of
IgA anti-h-tTG can detect IgAD at a cutoff that produces
relatively few false positives. We suggest that testing of all
patients for serum total IgA is not necessary when using
IgA anti-h-tTG testing for CD. Low IgA anti-h-tTG absorbance readings should be investigated by measurement of
total IgA.
Although the cost-effectiveness of this approach has not
been established, in our study, the use of an anti-h-tTG
cutoff of 0.013 would have avoided as many as 97.5% of
IgA tests and still detected all pregnant women with
IgAD.
In conclusion, our data suggest that a human IgA
anti-tTG ELISA is useful not only to detect cases of CD,
but also to screen for IgAD.
Fig. 1. Individual IgA anti-tTG absorbance values according to the
serum total IgA concentration.
(A), box-plots of anti-tTG absorbance values in selective IgA-deficient patients
(IgAD), diseased controls with low IgA, and healthy controls with normal IgA. IgAD
patients (median, 0.006; range, 0.000 – 0.022) vs diseased controls (median,
0.015; range, 0.002– 0.060), P ⬍0.0001; IgAD patients vs healthy controls
(median, 0.053; range, 0.016 – 0.262), P ⬍0.0001; diseased controls vs healthy
controls, P ⬍0.0001. The central box represents the values from the lower to the
upper quartile (25th–75th percentiles). The middle line represents the median.
The error bars represent the range from the minimum to the maximum value,
excluding outlying values, which are displayed as separate points. (B), anti-tTG
absorbance values in pregnant women with IgAD and in those with detectable
serum IgA. Horizontal dashed line represents the threshold value providing the
best sum of sensitivity (Sens) and specificity (Spec; minimizing the sum of the
false-negative and -positive rates) according to ROC curve analysis.
We are grateful to David H. Wallace for help in preparation of this manuscript. This work was supported by a
grant from the Fondo de Investigaciones Sanitarias (FIS:
02/1422).
References
1. Cunningham-Rundles C. Physiology of IgA and IgA deficiency. J Clin Immunol
2001;21:303–9.
2. Koskinen S. Long-term follow-up of health in blood donors with primary
selective IgA deficiency. J Clin Immunol 1996;16:165–70.
3. Cataldo F, Marino V, Ventura A, Bottaro G, Corazza GR. Prevalence and
1016
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
Technical Briefs
clinical features of selective immunoglobulin A deficiency in coeliac disease:
an Italian multicentre study. Italian Society of Paediatric Gastroenterology
and Hepatology (SIGEP) and “Club del Tenue” Working Groups on Coeliac
Disease. Gut 1998;42:362–5.
Collin P, Maki M, Keyrilainen O, Hallstrom O, Reunala T, Pasternack A.
Selective IgA deficiency and coeliac disease. Scand J Gastroenterol 1992;
27:367–71.
Hill PG, Forsyth JM, Semeraro D, Holmes GK. IgA antibodies to human tissue
transglutaminase: audit of routine practice confirms high diagnostic accuracy. Scand J Gastroenterol 2004;39:1078 – 82.
Van Meensel B, Hiele M, Hoffman I, Vermeire S, Rutgeerts P, Geboes K, et
al. Diagnostic accuracy of ten second-generation (human) tissue transglutaminase antibody assays in celiac disease. Clin Chem 2004;50:2125–35.
Rittmeyer C, Rhoads JM. IgA deficiency causes false-negative endomysial
antibody results in celiac disease. J Pediatr Gastroenterol Nutr 1996;23:
504 – 6.
Bazzigaluppi E, Lampasona V, Barera G, Venerando A, Bianchi C, Chiumello
G, et al. Comparison of tissue transglutaminase-specific antibody assays
with established antibody measurements for coeliac disease. J Autoimmun
1999;12:51– 6.
Korponay-Szabo IR, Dahlbom I, Laurila K, Koskinen S, Woolley N, Partanen
J, et al. Elevation of IgG antibodies against tissue transglutaminase as a
diagnostic tool for coeliac disease in selective IgA deficiency. Gut 2003;52:
1567–71.
Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac
disease: time for a standardized report scheme for pathologists. Eur J
Gastroenterol Hepatol 1999;11:1185–94.
Riestra S, Fernandez E, Rodrigo L, Garcia S, Ocio G. Prevalence of coeliac
disease in the general population of northern Spain. Strategies of serologic
screening. Scand J Gastroenterol 2000;35:398 – 402.
Farrell RJ, Kelly CP. Celiac sprue. N Engl J Med 2002;346:180 – 8.
Sollid LM. Molecular basis of celiac disease. Annu Rev Immunol 2000;18:
53– 81.
Korponay-Szabo IR, Kovacs JB, Czinner A, Goracz G, Vamos A, Szabo T. High
prevalence of silent celiac disease in preschool children screened with
IgA/IgG antiendomysium antibodies. J Pediatr Gastroenterol Nutr 1999;28:
26 –30.
Lenhardt A, Plebani A, Marchetti F, Gerarduzzi T, Not T, Meini A, et al. Role
of human-tissue transglutaminase IgG and anti-gliadin IgG antibodies in the
diagnosis of coeliac disease in patients with selective immunoglobulin A
deficiency. Dig Liver Dis 2004;36:730 – 4.
Tommasini A, Not T, Kiren V, Baldas V, Santon D, Trevisiol C, et al. Mass
screening for coeliac disease using antihuman transglutaminase antibody
assay. Arch Dis Child 2004;89:512–5.
Picarelli A, Sabbatella L, Di Tola M, Gabrielli F, Greco R, Di Cello T, et al.
Celiac disease diagnosis in misdiagnosed children. Pediatr Res 2000;48:
590 –2.
Previously published online at DOI: 10.1373/clinchem.2004.047233
Sensitive Automated ELISA for Measurement of Vitamin D-Binding Protein (Gc) in Human Urine, Anna Lis
Lauridsen,1* Michael Aarup,2 Anna Lisa Christensen,1 Bente
Jespersen,2 Kim Brixen,3 and Ebba Nexo1 (1 Department of
Clinical Biochemistry, Norrebrogade, Aarhus University
Hospital, Aarhus, Denmark; Departments of 2 Nephrology and 3 Endocrinology, Odense University Hospital,
Odense, Denmark; * address correspondence to this author at: Department of Clinical Biochemistry, Aarhus
University Hospital, Norrebrogade 44, DK-8000 Aarhus
C, Denmark; fax 45-8949-3060, e-mail all@dadlnet.dk)
We report a sensitive automated ELISA for measurement
of group-specific component (Gc; also known as Gc
globulin and vitamin D-binding protein) that detects
lower concentrations than any other published method.
This new ELISA enables measurement of Gc in human
urine.
Gc is a 50- to 58-kDa multifunctional plasma protein
synthesized mainly by hepatocytes and usually present in
plasma in concentrations between 4 and 6 ␮mol/L. The
functions of Gc are diverse. Gc is an important player in
the actin-scavenger system, which prevents the harmful
consequences of actin in the blood stream during tissue
injury. It binds actin monomers with high affinity, and
Gc–actin complexes are rapidly cleared from the circulation (1 ). Gc also has functions in the immune system,
acting as a co-chemotactic factor together with complement C5a (2 ) and, after deglycosylation, as a very potent
macrophage-activating factor (Gc-MAF) also capable of
activating osteoclasts (3, 4 ). The name “vitamin D-binding protein” is derived from its ability to bind and
transport vitamin D metabolites. Usually, ⬍0.1% of 25hydroxyvitamin D (25OHD) and ⬍1% of 1,25-dihydroxyvitamin D [1,25(OH)2D] circulate in their free forms
(5 ). Gc is probably also important for the renal activation
of 25OHD to 1,25(OH)2D. Mice lacking the multifunctional receptor megalin lose Gc–25OHD complexes in the
urine and are deficient in 1,25(OH)2D (6 ). Thus, there is
evidence that, generally, Gc and Gc–25OHD complexes
are filtered in the glomerulus and reabsorbed by megalinmediated endocytosis into the proximal tubular cells,
where vitamin D activation takes place. This theory
postulates that, under healthy conditions, only trace
amounts of Gc should be excreted in the urine, whereas
urinary loss of Gc is expected to increase with decreases in
the capacity for reabsorption in the proximal tubules. To
evaluate the fate of Gc in humans with various kidney
diseases, there is a need for an assay sensitive enough to
measure the minute amounts of Gc excreted in urine. This
prompted us to develop a sensitive ELISA.
We used the principles described by Engbaek (7 ) to
optimize the ELISA for Gc. The assay is based on an
immobilized polyclonal antibody that captures the Gc,
which subsequently binds a biotinylated monoclonal detection antibody that reacts with peroxidase–avidin–tetramethylbenzidine, producing a color reaction.
As capture antibody, we added to each well 8 ␮g of
rabbit anti-human Gc globulin (DakoCytomation Denmark A/S) in 100 ␮L of 50 mmol/L sodium carbonate (pH
9.6). We incubated the ELISA plates (F96-Maxisorp Nunc
immunoplates; Nunc A/S) at 4 °C for 20 h before emptying the wells and adding 200 ␮L of 1 mol/L ethanolamine
(pH 8 –9). After another 20 h at 4 °C, the plates were
stored at ⫺20 °C. We biotinylated the monoclonal mouse
anti-human Gc globulin (AntibodyShop; SSI) after overnight dialysis of 200 ␮L against 0.1 mol/L sodium bicarbonate (pH 8.3) by mixing gently for 4 h in the dark at
room temperature with 10 ␮L of 4.4 mmol/L biotin–
amidocaproate–N-hydroxysuccinimide ester (Sigma).
Subsequently, we added 10 ␮L of 100 mmol/L lysine
monohydrochloride (Fluka), waited for 15 min, and
added 10 ␮L of rabbit ␥-globulin (50 g/L; Calbiochem)
and bovine ␥-globulin (100 g/L; Sigma) in 10 mmol/L
sodium phosphate (pH 7.4). We then dialyzed the mixture
for 48 h against 10 mmol/L sodium phosphate (pH 7.4)
and for 24 h after addition of 1 g/L sodium azide (Merck).
1017
Fig. 1. Effect of freezer storage.
Relative Gc concentrations (%) in 3 urine samples after storage for 3 and 8
months at ⫺20 °C (䡺) or at ⫺80 °C (f, samples left undisturbed until
measurement; F, samples frozen and thawed twice before measurement).
The biotinylated antibody was stored at ⫺20 °C and
before use was diluted 1:10 000 in the assay buffer, 100
mmol/L sodium phosphate (pH 8.0) containing 1 g/L
bovine albumin (Sigma). We used an 8-point calibration
curve with purified human Gc-globulin, mixed type (Calbiochem), diluted in assay buffer (0, 1.25, 2.5, 5, 10, 20, 40,
and 80 pmol/L) and obtained the calibration curve by
cubic regression.
For the ELISA, we used an automated analyzer (BEP2000; Dade Behring) at 37 °C. Between each step, we
incubated the plates for 30 min and washed the wells
three times with a washing buffer consisting of 10
mmol/L sodium phosphate (pH 7.4), 145 mmol/L sodium chloride, and 1 g/L Tween 20 (Merck). On the
BEP-2000, controls and samples were diluted 1:10 in assay
buffer, and 100-␮L duplicates of the calibrators and diluted controls and samples were added. The biotinylated
detection antibody (100 ␮L) was then added, followed by
100 ␮L of peroxidase–avidin (DakoCytomation Denmark
A/S) diluted 1:2000 in 10 mmol/L sodium phosphate (pH
7.4) containing 400 mmol/L sodium chloride and 0.2 g/L
lysozyme (Sigma). The color reaction was started by addition of 100 ␮L of TMB-One ready-to-use substrate (Kem-EnTec Diagnostics A/S) and stopped after 9 min by addition of
100 ␮L of 1 mol/L phosphoric acid to each well. Color
development was measured photometrically at 620 nm.
We checked for linearity by measuring 5 dilutions of
each of 10 urine samples. The linear regression lines
between measured and expected Gc concentrations all
had intercepts not differing from 0, and 7 of 10 slopes did
not differ from 1, whereas 3 slopes differed slightly (95%
confidence intervals, 0.94 – 0.99, 1.02–1.36, and 1.01–1.10).
Imprecision was estimated at 4 concentrations by 4 measurements on each plate and 2 plates a day for 6 days. At
urinary Gc concentrations of 14, 84, 198, and 521 pmol/L,
the within-plate, within-day, between-day, and total imprecisions calculated as recommended by Krouwer and
Rabinowitz (8 ) were 1.8%–3.6%, 2.0%– 4.1%, 1.4%– 4.2%,
and 2.6%–5.8%, respectively. Recovery was 95%–108% for
221 and 426 pmol/L added to five different urines and
each measured four times. The detection limit, defined as
the concentration corresponding to a signal 3 SD above
the mean for the calibrator free of Gc, was 2.5 pmol/L, but
in practice we used 10 pmol/L. The effect of storage for 3
and 8 months at ⫺20 and ⫺80 °C is shown in Fig. 1.
Contrary to our experiences with Gc in plasma (unpublished data), urine samples for Gc measurement cannot be
stored at ⫺20 °C. However, when 2 vials of each of 15
urine samples were stored, one at ⫺80 °C for 1.5 months
followed by 3 months at ⫺20 °C, the other at ⫺80 °C for
4.5 months, the Gc concentration did not differ. This
indicates that the initial freezing temperature is crucial.
We recommend either measurement of Gc in fresh urine
or storage of urine samples at ⫺80 °C until measurement.
To examine the urinary loss of Gc as a function of
kidney function, we enrolled 99 patients with various
kidney diseases [dialysis (n ⫽ 0), glomerulonephritis (35),
Table 1. Urinary and plasma concentrations of Gc, albumin, and creatinine in 10 healthy controls (28 – 66 years of age) and
in 99 patients with various kidney diseases (20 – 88 years of age).
Median (range)
Healthy controls
Urinary Gc, nmol/24 h
Gc in second-void morning urine, nmol/L
Gc/Creatinine in second-void morning urine,
nmol/mmol
Urinary albumin, mg/24 h
Urinary creatinine, mmol/24 h
Plasma Gc, ␮mol/L
Plasma albumin, g/L
Plasma creatinine, ␮mol/L
Creatinine clearance, mL/min
a
0.62 (0.03–1.49)
0.70 (0.07–1.11)
0.05 (0.01–0.09)
17 (10–27)
13 (10–22)
3.9–6.4e
36–51e
44–134e
72–138e
Spearman nonparametric correlation coefficients between 24-h urinary Gc and the other variables.
24-h urinary Gc compared with the other variables: bP ⬍0.001; cP ⬍0.01; dP ⬍0.05.
e
95% central reference interval.
b– d
Patients
27 (0.03–897)
13 (0.04–375)
2.8 (0.01–62)
734 (14–8204)
11 (4–20)
4.8 (3.4–7.7)
41 (22–47)
224 (59–712)
37 (6–186)
Correlationa with urinary
Gc in patients
0.94b
0.93b
0.86b
0.05
0.12
⫺0.38b
0.32c
⫺0.24d
1018
Technical Briefs
diabetic nephropathy (17), polycystic kidney disease (9), and
other or nonspecified kidney disease (38)] and 12 healthy
individuals (of those, 2 were excluded because of urinary
albumin ⬎30 mg/24-h) in a cross-sectional study approved
by the local ethics committee (No. 20020055). Plasma and
urine samples (24-h urine and second-void morning urine)
were stored immediately at ⫺80 °C and, without thawing,
moved to a ⫺20 °C freezer before measurement. We measured plasma Gc by our immunonephelometric method on
a Behring Nephelometer 2 (Dade Behring) (9 ) and albumin
and creatinine on an Integra 700 (Roche). We used Kruskal–
Wallis and Mann–Whitney tests for comparisons between
groups and the Spearman nonparametric method to look for
correlations. We performed the calculations with SPSS 10.0.5
and set 0.05 as the significance level.
In the 99 patients, the median 24-h excretions of Gc and
albumin did not differ significantly between patients in
the various kidney disease groups but were significantly
(P ⬍0.001) higher in patients than in healthy controls
(Table 1). Measurement of Gc in second-void morning
urine gave a good estimate of the 24-h urinary excretion,
equivalent to that given by the Gc/creatinine ratio, as
seen from the high correlations in Table 1. Urinary Gc did
not correlate with plasma Gc, but correlated with markers
of kidney disease, particularly with urinary albumin
excretion (Table 1). Linear regression with Gc as the
dependent and albumin as the independent variable (after
ln transformation to obtain approximate gaussian distributions of residuals) showed a highly significant relationship in urine [r ⫽ 0.86; P ⬍0.001; mean (95% confidence
interval) slope, 1.2 (1.04 –1.3); mean (95% confidence interval) intercept, ⫺4.6 (⫺5.5 to ⫺3.7)], but not in plasma
(P ⫽ 0.23). The median urinary Gc/albumin ratio was
significantly lower in patients with glomerulonephritis
than in patients with polycystic, other, or nonspecified
kidney diseases. The correlation between creatinine clearance and the Gc/albumin ratio in 24-h urine was high (r ⫽
⫺0.54; P ⬍0.001) as was the correlation in second-void
morning urine (r ⫽ ⫺0.59; P ⬍0.001). Longitudinal studies, however, are needed to test the utility of the urinary
Gc/albumin ratio as a marker of kidney function.
In the kidneys, both Gc and albumin are believed to be
filtered in the glomerulus and subsequently reabsorbed
by megalin-cubilin–mediated endocytosis in the proximal
tubule (10 ). Because of this shared fate, the high correlation between urinary Gc and urinary albumin was expected. However, contrary to the negative correlation
between plasma and urinary albumin, we found no
significant correlation between plasma and urinary Gc.
The missing decrease in plasma Gc concentration is in
agreement with findings in megalin-knockout mice (6 )
and in some studies of patients with kidney diseases
(11, 12 ), but not in others (13, 14 ).
In conclusion, we have developed a sensitive automated ELISA capable of measuring very low Gc concentrations with low imprecision (functional detection limit,
0.01 nmol/L). In 99 patients, the urinary loss of Gc
increased with increasing severity of kidney disease, but
had no relationship with plasma Gc concentration.
We thank Nyreforeningen, the Institute of Clinical Medicine at Aarhus University, the County of Funen Research
Foundation, and the Aarhus University Research Foundation for financial support and Lene Dabelstein Petersen
for technical assistance.
References
1. Lee WM, Galbraith RM. The extracellular actin-scavenger system and actin
toxicity. N Engl J Med 1992;326:1335– 41.
2. Kew RR, Webster RO. Gc-globulin (vitamin D-binding protein) enhances the
neutrophil chemotactic activity of C5a and C5a des Arg. J Clin Invest
1988;82:364 –9.
3. Schneider GB, Benis KA, Flay NW, Ireland RA, Popoff SN. Effects of vitamin
D binding protein-macrophage activating factor (DBP-MAF) infusion on bone
resorption in two osteopetrotic mutations. Bone 1995;16:657– 62.
4. Yamamoto N, Kumashiro R. Conversion of vitamin D3 binding protein
(group-specific component) to a macrophage activating factor by the stepwise action of ␤-galactosidase of B cells and sialidase of T cells. J Immunol
1993;151:2794 – 802.
5. Bikle DD, Siiteri PK, Ryzen E, Haddad JG. Serum protein binding of
1,25-dihydroxyvitamin D: a reevaluation by direct measurement of free
metabolite levels. J Clin Endocrinol Metab 1985;61:969 –75.
6. Nykjaer A, Dragun D, Walther D, Vorum H, Jacobsen C, Herz J, et al. An
endocytic pathway essential for renal uptake and activation of the steroid
25-(OH) vitamin D3. Cell 1999;96:507–15.
7. Engbaek F. A general procedure for optimizing concentrations of capture
antibody, biotinylated detecting antibody, and enzyme-labeled avidin in
ELISAs—application to assays for ␣-fetoprotein, prolactin, FSH and LH in
serum. J Clin Immunoassay 1994;17:151–9.
8. Krouwer JS, Rabinowitz R. How to improve estimates of imprecision. Clin
Chem 1984;30:290 –2.
9. Lauridsen AL, Vestergaard P, Nexo E. Mean serum concentration of vitamin
D-binding protein (Gc globulin) is related to the Gc phenotype in women. Clin
Chem 2001;47:753– 6.
10. Christensen EI, Birn H. Megalin and cubilin: synergistic endocytic receptors
in renal proximal tubule. Am J Physiol Renal Physiol 2001;280:F562–F573.
11. St. John A, Thomas M, Dick I, Young P, Prince RL. Parathyroid function in
mild to moderate renal failure: evaluation by oral calcium suppression test.
J Clin Endocrinol Metab 1994;78:1436 – 8.
12. van Hoof HJ, de Sevaux RG, Van Baelen H, Swinkels LM, Klipping C, Ross
HA, et al. Relationship between free and total 1,25-dihydroxyvitamin D in
conditions of modified binding. Eur J Endocrinol 2001;144:391– 6.
13. Auwerx J, De Keyser L, Bouillon R, De Moor P. Decreased free 1,25dihydroxycholecalciferol index in patients with the nephrotic syndrome.
Nephron 1986;42:231–5.
14. Grymonprez A, Proesmans W, Van Dyck M, Jans I, Goos G, Bouillon R.
Vitamin D metabolites in childhood nephrotic syndrome. Pediatr Nephrol
1995;9:278 – 81.
DOI: 10.1373/clinchem.2004.045658
Serum Erythropoietin Measured by Chemiluminescent
Immunometric Assay: An Accurate Diagnostic Test for
Absolute Erythrocytosis, Pascal Mossuz,1†* François Girodon,2† Sylvie Hermouet,3 Irène Dobo,4 Eric Lippert,5 Magali
Donnard,6 Veronique Latger-Cannard,7 Nathalie Boiret,8 Vincent Praloran,5 and Jean Claude Lecron9 (1 Laboratory of Hematology, CHU Grenoble, France; 2 Laboratory of Hematology, CHU Dijon, France; 3 Laboratory of Hematology, CHU,
Nantes, France; 4 Laboratory of Hematology, CHU Angers,
France; 5 Laboratory of Hematology, CHU Bordeaux,
France; 6 Laboratory of Hematology, CHU Limoges, France;
7
Laboratory of Hematology, CHU Nancy, France, 8 Laboratory of Hematology, CHU Clermont Ferrand, France; 9 Laboratory of Protein Chemistry, CHU, and University of Poitiers, Poitiers, France; † Pascal Mossuz and François Girodon
1019
contributed equally to this work; * address correspondence to this author at: Laboratoire d’Hématologie, CHU
Grenoble, BP217x, 38043 Grenoble Cedex, France; fax 33476-765-935, e-mail PMossuz@chu-grenoble.fr)
Absolute erythrocytosis (AE), suspected from a high
hemoglobin concentration and/or hematocrit, can be confirmed by an increased red cell mass (RCM) (1 ). Schematically, one distinguishes three major mechanisms of AE:
(a) erythropoietin (Epo)-independent proliferation of
clonal erythroid precursors as found in polycythemia vera
(PV) and other myeloproliferative disorders; (b) Epodependent polyclonal proliferation of erythroid precursors as found in secondary erythrocytoses that are secondary to production of Epo as a consequence of either a
physiologic response to tissue hypoxia or of tumoral
production; (c) idiopathic erythrocytoses (IEs) in patients
without evidence of PV or secondary erythrocytoses (2 ).
The serum Epo concentration reflects its oxygen-regulated production by kidney. Thus, serum Epo is decreased
in PV and increased in secondary erythrocytoses. Use of
serum Epo as a diagnostic test for PV (3–5 ) is controversial (6 – 8 ). Indeed, until recently, the lack of standardization of the reagents and methods impeded identification
of reliable thresholds. As a consequence, the diagnosis of
PV is still largely based on exclusion and/or indirect
clinical and biological criteria initially proposed by the
Polycythemia Vera Study Group (PVSG) (9 ). However,
the WHO guidelines (10 ), which are based on major
criteria (e.g., splenomegaly, lack of secondary erythrocytosis) and minor criteria (e.g., modification in blood cell
count, bone marrow histology), recently classified the
endogenous erythroid colony assay and serum Epo measurements as major and minor PV diagnostic criteria,
respectively.
We recently demonstrated in a large multicenter study
(n ⫽ 241) that a commercial ELISA for serum Epo was a
reliable and accurate biological diagnostic test in patients
with AE (11 ). In this study, we determined a low Epo
threshold with 65% sensitivity and 100% specificity for
the diagnosis of PV and a high Epo threshold with 19.7%
sensitivity and 100% specificity for secondary erythrocytoses. However, this ELISA, which is not automated,
could be negatively impacted by technical and/or interindividual bias limiting interlaboratory reproducibility; in
addition, it is not suitable for a short series of samples.
A fully automated chemiluminescent enzyme-labeled
immunometric assay (Immulite; DPC) has been developed that is suitable for routine assay of serum Epo in
individual samples as well as in small or large series of
samples (12, 13 ). We compared the sensitivity, specificity,
and predictive values of this automated method with
those of the ELISA for the diagnosis of PV and secondary
erythrocytoses.
From 2001 to 2004, 193 samples from patients with
suspected AE (hematocrit ⬎50% for males and ⬎45%
for females) were collected in 8 university hospitals.
Recruitment was in agreement with standards of the
ethics committee “Comité Consultatif de Protection des
Personnes dans la Recherche Biomédicale” (CCPPRB). AE
was established by isotopic determination of RCM in 137
patients before any treatment. Patients were classified
according to WHO guidelines (10 ) as follows: 81 PV, 53
secondary erythrocytoses, and 3 IE. The demographic and
clinical characteristics of the patients are summarized in
Table 1. Blood was harvested at diagnosis in dry tubes
and centrifuged 15 min at 1400g after blood clotting, and
the sera were frozen at ⫺80 °C.
The sera from the 137 AE patients were assayed for Epo
by trained technicians in a blind manner in two indepen-
Table 1. Characteristics of patients.a
PV
No. of patients
M/F
Median (range) age, years
Mean (SD) RBC, ⫻ 1012/L
Mean (SD) Hb, g/L
Mean (SD) Ht, %
Mean (SD) platelets, ⫻ 109/L
Mean (SD) WBC, ⫻ 109/L
ELISA Epo, IU/L
Mean (SD)
Median (range)
Immulite Epo, IU/L
Mean (SD)
Median (range)
Correlation coefficientc (r)/slope
81
52/29
67 (34–91)
6.8 (0.97)
18.8 (1.6)
58 (6)
446 (198)
12.8 (1.7)
SEb
53
45/8
59 (25–88)
5.7 (0.55)
17.7 (1.6)
53 (5)
209 (56)
9.4 (1.4)
IE
3
2/1
55 (33–82)
5.4 (0.88)
16.4 (2.9)
50 (9)
286 (107)
8.0 (0.5)
1.69 (2.19)
0.8 (0.6–13.7)
10.40 (6.80)
8.2 (1.45–33.9)
8.33 (2.7)
7.7 (6–11.3)
2.38 (1.60)
1.6 (0.25–13.0)
0.79/1.11
13.07 (7.95)
11.7 (2.8–40.1)
0.87/1.14
10.86 (2.63)
10.1 (8.7–13.8)
0.99/1.07
a
Patients with AE were classified into 3 groups according to the WHO classification of tumors of hematopoietic and lymphoid tissues as PV, secondary erythrocytosis,
and IE. Serum Epo concentrations were measured by ELISA and chemiluminescent immunometric (Immulite) methods according to the procedures described in the text
and in the manufacturers’ instructions. Regression analysis was performed using Deming regression method.
b
SE, secondary erythrocytosis; WBC, leukocytes; RBC, erythrocytes; Hb, hemoglobin; Ht, hematocrit.
c
Pearson correlation.
1020
Technical Briefs
dent laboratories. The assays used, the Epo ELISA (Quantikine IVD Erythropoietin ELISA; R & D Systems Inc.) and
a two-site sandwich immunoassay with chemiluminescent detection on an automated random access immunoassay analyzer (Immulite; DPC), were performed according to the manufacturers’ instructions. Values were
expressed as IU/L. Manufacturer reference intervals for
serum Epo were 3.3–16.6 IU/L for the ELISA and 4.1–20.1
IU/L for the chemiluminescence immunoassay. We performed statistical analysis of the sensitivity and specificity
of the Epo ELISA and chemiluminescence assay with
Statview software, Deming regression and correlation
analysis with Analyze-It software, and ROC curve analysis and predictive values with Stata, Ver. 7.0.
We first evaluated the performance characteristics of
the automated chemiluminescent immunoassay for Epo
(Immulite). The within-run imprecision (CV) was 6.8%
and 10% for sera at 28.2 and 0.95 IU/L Epo, respectively
(n ⫽ 8). The interassay CV was 9.2%, as determined by 12
independent measurements of a sample containing 28.2
IU/L Epo. The apparent recovery of Epo added in a 1:1
ratio to sera containing 13 and 0.5 IU/L Epo was 105%. To
assess assay linearity, a serum containing 124 IU/L Epo
was diluted 1:5, 1:10, 1:100, and 1:1000 in the sample
diluent. Compared with theoretical values, the measured
concentrations were, respectively, 107%, 106%, 110%, and
undetectable (for the 1:1000-diluted sample; theoretical
value ⫽ 0.124 IU/L) in 1 representative experiment of 2.
As reported previously (12, 13 ), all of these tests confirmed that the Immulite Epo assay is reproducible and
accurate. We tested Epo stability in a serum containing 9.6
IU/L. Measured concentrations after 1, 2, and 7 days
were, respectively, 104%, 96%, and 93% of initial values in
samples stored at 4 °C and 105%, 95%, and 98% in
samples stored at 20 °C (1 representative experiment of 3).
After 1– 4 freeze–thaw cycles, values for 1 sample were
100%, 103%, 98%, and 89% of initial values, respectively.
Thus, sera can be stored up to 1 week at 20 °C or
frozen/thawed up to 3 times before being assayed. In 17
paired serum and EDTA-plasma (Vacutainer; Becton
Dickinson) samples, Epo concentrations were 33% lower
in plasma, with a correlation coefficient of 0.99, and SD of
the residuals of 2.25 IU/L. Taking into account this
difference and according to manufacturer’s instructions,
we recommend avoiding EDTA-plasma for measurements of Epo by the Immulite method.
The mean (SD) difference between the Immulite assay
and the ELISA was 1.54 (2.61) IU/L with one major outlier
(13.6 IU/L for the ELISA and 36.9 IU/L for the Immulite
assay). As shown in Table 1, the measured Epo concentrations for all groups of patients were significantly higher
with the Immulite assay than with the ELISA (P ⫽ 0.017),
but the results were highly correlated as evidenced by
regression analysis [n ⫽ 137; r ⫽ 0.93; slope ⫽ 1.17 (95%
confidence interval, 1.09 –1.25); intercept ⫽ 0.44 (0.23–
1.13) IU/L]. Correlations between the two methods remained excellent and highly significant when results for
each group of patients were considered separately: PV,
n ⫽ 81, r ⫽ 0.79 (P ⬍0.001); secondary erythrocytoses, n ⫽
53, r ⫽ 0.87 (P ⫽ 0.001); IE, n ⫽ 3, r ⫽ 0.99 (P ⬍0.001).
A cutoff of 2.8 IU/L provided 100% specificity (95%
confidence interval, 95%–100%) and 78% sensitivity (68%–
85%) for the diagnosis of PV, and a cutoff of 13.8 IU/L
provided 100% specificity (95%–100%) and 34% sensitivity (23%– 47%) for the diagnosis of secondary erythrocytoses. For 59% of the 137 untreated AE patients, the results
were outside these 2 thresholds. By comparison, the Epo
ELISA thresholds of 1.4 and 13.7 IU/L (also defined by
ROC curve analysis as giving 100% specificity and 100%
positive predictive value) allowed the direct diagnosis of
49% of the 137 untreated patients with AE. Sera with Epo
concentrations lower than the detection limit of the ELISA
(⬍0.6 IU/L) were measurable by the Immulite assay
[mean (SD), 1.37 (0.75) IU/L; median (range), 1.1 (0.25–
3.3) IU/L]. The Epo concentration was below the detection limit of the Immulite assay (0.25 IU/L) in only 1
patient. The better differentiation of low Epo concentrations by the Immulite assay could be of prognostic interest
in PV patients because low serum Epo values correlate
with a high risk of thrombosis (5, 14 ). Indeed, we found
significant inverse correlations between Epo concentrations and RCM, hematocrit, and hemoglobin (n ⫽ 81). For
RCM, r ⫽ ⫺0.33 (P ⫽ 0.01); for hematocrit, r ⫽ ⫺0.31 (P ⫽
0.004); and for hemoglobin, r ⫽ ⫺0.37 (P ⫽ 0.006).
In conclusion, this study performed with an automated
chemiluminescent immunometric method on a large
group of polycythemic patients defined low and high
serum Epo thresholds (2.8 and 13.8 IU/L) that allowed the
presumptive diagnosis of 78% of PV and 34% of secondary erythrocytoses, without further investigation. A low
Epo concentration having been validated as a criterion for
PV diagnosis by the WHO classification (10 ), our study
demonstrates that serum Epo is a major biological criterion for the diagnosis of PV and secondary erythrocytoses
in patients with confirmed AE and supports development
of a new diagnostic strategy based on serum Epo. The
ability to measure serum Epo in small series and in
individual samples (because of a random-access immunoassay analyzer) should facilitate and increase use of this
assay as a first-line test.
Hence, in accordance with the new WHO PV diagnostic
criteria, we propose use of the Epo assay in a first step
with determination of the RCM in patients with suspicion
of AE. We suggest that more complex, time-consuming, or
costly tests (e.g., clonogenic cultures, PRV-1 determination, and/or bone marrow histology) should be performed in a second step for patients whose serum Epo
results are inconclusive.
We are grateful to our colleagues in the various clinical
departments for providing clinical data. This work was
financed by a grant from the French Ministry of Health
(PHRC région Bourgogne). We also greatly thank Dr. J.L.
Bosson from the clinical center of investigation (CHUGrenoble, France) for expert assistance in the statistical
study.
References
1. Pearson TC, Messinezy M, Westwood N, Green AR, Bench AJ, Huntly BJ, et
al. Polycythemia vera updated: diagnosis, pathobiology, and treatment.
Hematology (Am Soc Hematol Educ Program) 2000:51– 68.
2. Tefferi A. Polycythemia vera: a comprehensive review and clinical recommendations. Mayo Clin Proc 2003;78:174 –94.
3. Birgegard G, Wide L. Serum erythropoietin in the diagnosis of polycythaemia
and after phlebotomy treatment. Br J Haematol 1992;81:603– 6.
4. Remacha AF, Montserrat I, Santamaria A, Oliver A, Barcelo MJ, Parellada M.
Serum erythropoietin in the diagnosis of polycythemia vera. A follow-up
study. Haematologica 1997;82:406 –10.
5. Messinezy M, Westwood NB, El-Hemaidi I, Mardsen JT, Sherwood RS,
Pearson TC. Serum erythropoietin values in erythrocytoses and in primary
thrombocythaemia. Br J Haematol 2002;117:47–53.
6. Casadevall N. Determination of serum erythropoietin. Its value in the
differential diagnosis of polycythemias. Nouv Rev Fr Hematol 1994;36:
173– 6.
7. Lindstedt G, Lundberg PA. Are current methods of measurement of erythropoietin (EPO) in human plasma or serum adequate for the diagnosis of
polycythaemia vera and the assessment of EPO deficiency. Scand J Clin Lab
Invest 1998;58:441–58.
8. Spivak JL. Polycythemia vera: myths, mechanisms, and management. Blood
2002;100:4272–90.
9. Pearson TC. Evaluation of diagnostic criteria in polycythemia vera. Semin
Hematol 2001;38:21– 4.
10. Jaffe ES, Lee Harris N, Stein H, Vardinan JW, eds. World Health Organization
classification of tumours. Pathology and genetics of tumours of haematopoietic and lymphoid tissues. Lyon, France: IARC Press, 2001:32– 4.
11. Mossuz P, Girodon F, Donnard M, Latger-Cannard V, Dobo I, Boiret N, et al.
Diagnostic value of serum erythropoietin level in patients with absolute
erythrocytosis. Haematologica 2004;89:1194 – 8.
12. Benson EW, Hardy R, Chaffin C, Robinson CA, Konrad RJ. New automated
chemiluminescent assay for erythropoietin. J Clin Lab Anal 2000;14:271–3.
13. Owen WE, Roberts WL. Performance characteristics of the IMMULITE 2000
erythropoietin assay. Clin Chim Acta 2004;340:213–7.
14. Andreasson B, Carneskog J, Lindstedt G, Lundberg PA, Swolin B, Wadenvik
H, et al. Plasma erythropoietin concentrations in polycythaemia vera with
special reference to myelosuppressive therapy. Leuk Lymphoma 2000;37:
189 –95.
New Enzymatic Assay Using Phospholipase D to Measure Total Calcium in Serum, Mitsutoshi Sugano,1 Kazuyoshi Yamauchi,1* Keiko Sugano,1 Kenji Kawasaki,1 Minoru
Tozuka,2 Tsutomu Katsuyama,3 Haruyo Soya,4 Tatsuhiko
Tanaka,4 Shigeyuki Imamura,5 and Shozo Nomoto3 (1 Department of Laboratory Medicine, Shinshu University Hospital, Matsumoto, Japan; 2 Clinical Laboratory Center, The
University of Tokyo Hospital, Tokyo, Japan; 3 Department
of Laboratory Medicine, Shinshu University School of
Medicine, Matsumoto, Japan; 4 Shino-test Corporation,
Kanagawa, Japan; 5 Asahikasei Pharma Corporation, Shizuoka, Japan; * address correspondence to this author at:
Department of Laboratory Medicine, Shinshu University,
Hospital, 3-1-1 Asashi, Matsumoto 390-8621, Japan; fax
81-263-34-5316, e-mail yamauchi@hsp.md.shinshu-u.ac.jp)
Various methods have been used to measure calcium in
body fluids. Atomic absorption spectrophotometry (AAS)
is the most reliable (1 ), but it requires special instrumentation. The most widely used method involves colorimetric detection of calcium complexes by o-cresolphthalein
complexone (o-CPC) (2, 3 ) or arsenazo. With the o-CPC
method, reagent stability and recoveries at low concentrations are poor, and magnesium interferes in the reac-
1021
tion (4 ). Newer methods using o-CPC (5–7 ) or other
colorimetric agents (8 –10 ) are not entirely satisfactory.
Various enzymatic methods have been described, including methods using porcine pancreatic ␣-amylase (EC
3.2.1.1) (11 ), phospholipase D (PL-D; EC 3.1.4.4) (12, 13 ),
and urea amidolyase (14 ). The first two are based on
activation of enzymes by calcium, whereas the third is
based on inhibition of the enzyme by calcium. The ␣-amylase method is reportedly inaccurate for patients with
hyperamylasemia (11 ), and the other 2 methods each
require 2 reaction steps (12 ). In this report, we describe a
new, simple, specific enzymatic assay based on activation
of PL-D. We investigated the assay characteristics and its
suitability for use in routine laboratory tests.
We obtained 126 serum samples from patients admitted
to Shinshu University Hospital after receiving informed
consent from the patients and approval by our institutional ethics committee.
We obtained PL-D from Streptomyces chromofuscus [(15 );
Asahi Kasei Pharma], bis(p-nitrophenyl) phosphate
(BPNPP) from Kanto Chemical Co., Good’s buffer from
Doujin Laboratories, and SRM 915 and 909a from NIST.
The reagent sets for the o-CPC and ␣-amylase methods
were from Serotec Co. Ltd. and Ono Pharmaceutical Co.
Ltd., respectively. Other reagents were analytical grade
(Wako Pure Chemical).
The new method is based on increased PL-D-catalyzed
hydrolysis of BPNPP by calcium ions, as follows. The
p-nitrophenol released by the reaction is detected at 405
nm (Scheme 1).
The assay is a 2-point fixed-rate assay performed at
37 °C; for our experiments we used a Hitachi 7170 analyzer. Briefly, 9.5 ␮L of sample was mixed with 160 ␮L of
solution containing 750 U/L PL-D and 0.275 mmol/L
calcium acetate in 60 mmol/L Good’s buffer (pH 7.5).
After incubating the mixture for 5 min at 37 °C, we added
160 ␮L of reagent containing 1.5 mmol/L BPNPP in 60
mmol/L Good’s buffer (pH 7.5) and measured the absorbance at time points 23–34 (6.76 –10.00 min). We performed the o-CPC and ␣-amylase methods on the same
analyzer according to the manufacturer’s instructions.
For the AAS method (1 ), we used a 0.1-mL sample on a
Hitachi Z-5000 Atomic Absorption Spectrophotometer
(flame-type analyzer) equipped with an autosampler and
a multirange micropipetting device (Gilson Inc.) with
triple rinsing of the inside of the tip. We increased sample
Scheme 1.
1022
Technical Briefs
Table 1. Within- and between-run reproducibility of the new
enzymatic method.
Within-run (n ⫽ 20)
Between-run (n ⫽ 20)
Samplea
Meanb (SD), mmol/L
CV, %
L
M
H
L
M
H
1.322 (0.0152)
2.560 (0.020)
3.405 (0.027)
1.327 (0.014)
2.526 (0.025)
3.383 (0.014)
1.2
0.78
0.79
1.0
0.98
0.42
a
Samples were prepared with a pooled serum. L, M, and H indicate low,
medium, and high calcium concentrations, respectively.
b
Mean of duplicate measurements.
dilution from 50- to 75-fold to decrease viscosity and
increase linearity.
To prepare calibrators, we dissolved NIST SRM 915
(10.010 g) in 55 mL of 2 mol/L hydrochloric acid and
brought the volume to 1000 mL with water. We diluted
this stock solution with a diluent containing 140 mmol/L
NaCl and 5 mmol/L KCl to prepare calibrators (1.5, 2.5,
and 3.5 mmol/L). We assayed the calibrators included
with each reagent set by the AAS method and calculated
correction factors for the calibration of each method. We
performed linear regression analyses with the Medical
Communication Program (Ver. 5 for Windows 95; Sysmex
Co. Ltd.). Details of the method optimization experiments
are shown in the Data Supplement that accompanies the
online version of this Technical Brief at http://www.
clinchem.org/content/vol51/issue6/.
To inhibit the activity of endogenous phosphodiesterase, we used an anionic detergent at a final concentration
of 2.5 g/L. To increase the linear portion of the calibration
curve, we added sodium sulfate and tartaric acid (as
inhibitors of the PL-D reaction) to reagents 1 and 2,
respectively. We decided on final tartaric acid and sodium
sulfate concentrations of 37.5 and 200 mmol/L, respectively.
To remove the negative interference of EDTA in the
determination of serum calcium concentrations, we added
nickel ions [nickel(II) acetate tetrahydrate] to reagent 1. A
final concentration of 0.4 mmol/L or higher was necessary to prevent the influence of EDTA completely. The
concentration of nickel ion had no effect on either dilution
linearity or assay sensitivity. We decided on a final nickel
ion concentration of 0.5 mmol/L.
The absorbance increased linearly above time point 17,
and we decided to use the change in absorbance between
time points 23 and 34 to calculate calcium concentrations.
Ca2⫹ was added to reagent 1 to improve the linearity at
low concentrations. The calibration curve was linear from
0 to 7.0 mmol/L. When we used serial dilutions of SRM
915 in a pooled serum, the assay was linear from 0 to 6.0
mmol/L.
The results of the reproducibility study are summarized
in Table 1. When we assayed SRM 915 solutions and
samples of SRM 909a to which 0.252 or 1.26 mmol/L
calcium had been added, the recoveries were 97.0% and
99.5%, respectively.
The mean (SD) values for SRM 909a-I (labeled value,
2.225 mmol/L) and SRM 909a-II (labeled value, 3.540
mmol/L) obtained from triplicate determinations in 5
separate experiments were 2.218 (0.001) and 3.542 (0.002)
mmol/L, respectively.
We observed no interference when we added Mg2⫹,
Fe2⫹, Fe3⫹, Cu2⫹, and Zn2⫹ at maximum concentrations of
205.7, 89.5, 53.7, 31.5, and 30.6 ␮mol/L, respectively (the
calcium values obtained being 98.9%–100.3% of the expected values). Unconjugated bilirubin up to 196 mg/L
did not interfere, nor did conjugated bilirubin up to 205
mg/L, hemoglobin up to 4.6 g/L, turbidity up to 2000
hormadin turbidity units, or EDTA up to 1.00 g/L.
We compared the new method with the AAS method
(Fig. 1A) and the o-CPC method (Fig. 1B), and the
␣-amylase method with the AAS method (Fig. 1C). The
absolute value of the y-intercept obtained with the new
enzymatic method was the smallest among the 3 methods.
Moreover, the slope for the new method was closer to
unity than were the slopes of the other methods.
We assessed the stability of the reagents stored for 2
months at 5 °C and room temperature, using 3 serum
samples. The measured values, as a percentage of the
initial values, were 99%–101% at 5 °C and 96%–98% at
room temperature.
Various enzymatic assay methods for measuring calcium have been developed (11–13 ) in attempts to overcome the problems experienced with the o-CPC method,
the most widely used conventional method. PL-D is
specifically activated by calcium, not by magnesium,
which can affect the values obtained with the o-CPC
method. Although a workable method using PL-D was
developed, the actual method involved coupled enzyme
reactions using both PL-D and choline oxidase (12, 13 ).
The proposed method is a simple assay system consisting
of a 1-step reaction with BPNPP used as the substrate for
PL-D.
We carefully devised the composition of the reagent
mixture to establish the new method. It was necessary to
first inhibit the endogenous phosphodiesterase because
the absorbance obtained for the serum blank is increased
by the reaction of phosphodiesterase with BPNPP. The
addition of anionic detergent seemed to suppress this
influence efficiently. Partial inhibition of PL-D was also
necessary to make the assay adequately quantitative. The
Km of PL-D for calcium is comparatively low, 7.5 ⫻ 10⫺5
mol/L (16 ), and the absorbance did not increase in a
calcium-concentration– dependent manner. We therefore
added tartaric acid and sodium sulfate to the reagent
mixture to inhibit the PL-D activity. Tartaric acid effectively inhibited PL-D activity by its chelating action, and
sodium sulfate acted as a competitive inhibitor. These
agents made the Km value higher, and under those
conditions the absorbance increased in proportion to the
calcium concentration. Consequently, assay linearity was
improved. We also added nickel ion to the reagent mixture to eliminate the influence of the chelating action of
EDTA. Because the chelating stability constant of EDTA
for nickel ions (18.56) (17 ) is much larger than that for
1023
Fig. 1. Correlations of the new enzymatic method (A), the o-CPC
method (B), and the ␣-amylase method (C) with the AAS method.
For all 4 methods, n ⫽ 126. The results of the regression analyses were as
follows: for the new method (A), y ⫽ 1.014x ⫺ 0.023 mmol/L (r ⫽ 0.992; Sy兩x ⫽
0.215 mmol/L); for the o-CPC method (B), y ⫽ 1.051x ⫺ 0.161 mmol/L (r ⫽
0.994; Sy兩x ⫽ 0.202 mg/L); for the ␣-amylase method (C), y ⫽ 0.966x ⫹ 0.067
mg/L (r ⫽ 0.992; Sy兩x ⫽ 0.197 mmol/L).
calcium (10.96) (18 ), nickel ions could be completely
substituted for calcium ions, in terms of chelation by
EDTA.
The o-CPC method does not measure albumin-bound
calcium completely and is insufficient in terms of stoichiometry because the affinity of calcium for albumin is
greater than that of calcium for o-CPC. This problem is
made apparent by the lower recoveries at low calcium
concentrations. In our study, this influence was revealed
by the y-intercepts obtained in the correlation between the
o-CPC and the AAS methods (⫺0.161 mmol/L). The
affinity of calcium for PL-D is much greater than that of
calcium for albumin; therefore, the reagents in the new
enzymatic method should react with albumin-bound calcium. Indeed, the y-intercept (⫺0.023 mmol/L) was improved vs that seen for the o-CPC method.
Reagent stability is also a major problem with the
o-CPC method, the stability being adversely affected by
CO2 absorption (8 ). The reagent used in the new enzymatic method was stable when stored for 2 months at 5 °C
and room temperature.
In conclusion, we have developed a new, simple enzymatic assay for the measurement of serum calcium. The
method has good precision, is specific for calcium, being
free from influences by metal ions and EDTA, and may be
suitable for clinical laboratory tests.
1024
Technical Briefs
References
1. Cali JP, Bowers GN Jr, Young DS. A referee method for the determination of
total calcium in serum. Clin Chem 1973;19:1208 –13.
2. Connerty HV, Briggs AR. Determination of serum calcium by means of
orthocresolphthalein complexone. Am J Clin Pathol 1966;45:290 – 6.
3. Gitelman HJ. An improved automated procedure for the determination of
calcium in biological specimens. Anal Biochem 1967;18:521–31.
4. Corns CM, Ludman CJ. Some observations on the nature of the calciumcresolphthalein complexone reaction and its relevance to the clinical
laboratory. Ann Clin Biochem 1987;24:345–51.
5. Cohen SA, Sideman L. Modification of the o-cresolphthalein complexone
method for determining calcium. Clin Chem 1979;25:1519 –20.
6. Lorentz K. Improved determination of serum calcium with 2-cresolphthalein
complexone. Clin Chim Acta 1982;126:327–34.
7. Cowley DM, Mottram BM, Haling NB, Sinton TJ. Improved linearity of the
calcium-cresolphthalein complexone reaction with sodium acetate. Clin
Chem 1986;32:894 –5.
8. Corns CM. A new colorimetric method for the measurement of serum
calcium using a zinc-zincon indicator. Ann Clin Biochem 1987;24:591–7.
9. Leary NO, Pembroke A, Duggan PF. Single stable reagent (arsenazo III) for
optically robust measurement of calcium in serum and plasma. Clin Chem
1992;38:904 – 8.
10. Chapoteau E, Czech BP, Zazulak W, Kumar A. New reagent for colorimetric
assay of calcium in serum. Clin Chem 1993;39:1820 – 4.
11. Kayamori Y, Katayama Y. Enzymatic method for assaying calcium in serum
and urine with porcine pancreatic ␣-amylase. Clin Chem 1994;40:781– 4.
12. Tabata M, Kido T, Ikemoto M, Totani M, Murachi T. Use of phospholipase D
and choline oxidase for the enzymatic determination of calcium ion in
serum. J Clin Biochem Nutr 1986;1:11–9.
13. Morishita Y, Iinuma Y, Nakashima N, Kadota A, Miike A, Tadano T.
Enzymatic assay of calcium in serum with phospholipase D. Clin Chem
1999;45:2280 –3.
14. Kimura S, Iyama S, Yamaguchi Y, Hayashi S, Fushimi R, Amino N. New
enzymatic assay for calcium in serum. Clin Chem 1996;42:1202–5.
15. Imamura S, Horiuti Y. Purification of Streptomyces chromofuscus phospholipase D by hydrophobic affinity chromatography on palmitoyl cellulose.
J Biochem 1979;85:79 –95.
16. Geng D, Baker DP, Foley SF, Zhou C, Stieglitz K, Roberts MF. A 20-kDa
domain is required for phosphatidic acid-induced allosteric activation of
phospholipase D from Streptomyces chromofuscus. Biochim Biophys Acta
1999;1430:234 – 44.
17. Schwarzenbach G, Gut R, Anderegg G. Komplexone XXV. Die polarographische untersuchung von austauschgleichgewichten. Neue daten der bildungskonstanten von metallkomplexen der äthylendiamin-tetraessigsäure
und der 1,2-diaminocyclohexan-tetraessigsäure. Helv Chim Acta 1954;37:
937–57.
18. Schwarzenbach G, Anderegg G. Die verwendung der quecksilberelektrode
zur bestimmung der stabilitätskonstanten von metallkomplexen. Helv Chim
Acta 1957;40:1773–92.
Presence of Filterable and Nonfilterable Cell-Free
mRNA in Amniotic Fluid, Paige B. Larrabee,1 Kirby L.
Johnson,2 Inga Peter,3 and Diana W. Bianchi2* (Divisions of
1
Newborn Medicine and 2 Genetics, Department of Pediatrics, Floating Hospital for Children, and 3 Institute of
Clinical Research and Health Policy Studies, Tufts-New
England Medical Center, Boston, MA; * address correspondence to this author at: Tufts-New England Medical
Center, 750 Washington St., Box 394, Boston MA 02111;
fax 617-636-1469, e-mail dbianchi@tufts-nemc.org)
Much current research focuses on the properties and
clinical applications of circulating nucleic acids (1 ). The
recent discovery of cell-free RNA in the plasma and
serum of cancer patients has generated considerable interest (2– 6 ). Circulating RNA is surprisingly stable, and
Ng et al. (7 ) recently showed that a considerable proportion of plasma mRNA species is particle associated and
thus possibly protected from nuclease degradation (8 ).
Fetal-derived mRNA has also been found in the plasma
and serum of pregnant women (9 –11 ) and in amniotic
fluid (12 ), and has many potential clinical applications
(13, 14 ). Amniotic fluid is routinely collected during amniocentesis for fetal chromosome analysis or fetal lung
maturity studies. However, little is known regarding the
biology of circulating fetal mRNA or fetal mRNA in
amniotic fluid.
In this report, we explore whether cell-free fetal nucleic
acids in amniotic fluid have properties similar to circulating DNA and mRNA. Expanding on the work of Ng et al.
(7 ), we hypothesized that cell-free fetal mRNA in amniotic fluid might be present in a particle-associated form
and could thus be filterable, whereas the non-particle–
associated form of DNA would be present in such high
concentrations that there would be no significant reduction in its quantity by filtration. Additionally, we wished
to compare the quantities of nucleic acids in amniotic
fluid containing cells with the quantities in filtered and
unfiltered cell-free supernatant. We hypothesized that
whole amniotic fluid containing amniocytes would contain a significantly larger amount of DNA and RNA than
cell-free amniotic fluid.
This study was performed with Institutional Review
Board approval from Tufts-New England Medical Center.
Seven amniotic fluid samples with a minimum volume of
3 mL each were obtained from women undergoing scheduled amniocenteses. One sample originated from a
woman with polyhydramnios undergoing therapeutic
amnioreduction. From 5 of the 7 samples, two 200-␮L
portions of uncentrifuged, unfiltered amniotic fluid were
set aside at ⫺80 °C. Uncentrifuged fluid was not available
for the other 2 samples because the amniocytes were
needed for clinical studies. For all 7 samples, the remaining amniotic fluid was aliquoted into 1.5-mL microcentrifuge tubes and centrifuged at 1600g for 10 min at 4 °C. The
supernatant was then carefully removed and subjected to
a second centrifugation at 16 000g for 10 min at 4 °C. The
supernatant was again carefully removed and then divided into 4 additional aliquots: 3 portions were individually passed through filters (Millex-GV; Millipore) with
pore sizes of 0.22, 0.45, and 5 ␮m. The remaining aliquot
was not subjected to filtration. All aliquots were then
divided into 2 portions and stored at ⫺80 °C until RNA
and DNA extractions were performed.
For each sample, RNA was extracted from 200 ␮L of
each of the 5 amniotic fluid aliquots (uncentrifuged,
cell-free unfiltered, and portions passed through filters
with pore sizes of 0.22, 0.45, and 5 ␮m) with the QIAamp
Viral RNA Mini Kit (Qiagen), according to the “Viral
RNA Mini Spin Protocol” as recommended by the manufacturer. The buffer volumes were adjusted proportionally for sample size. A 15-min incubation at room temperature with RNase-free DNase (Qiagen) was used between
buffers AW1 and AW2. RNA was stored at ⫺80 °C until
analysis.
For each sample, DNA was extracted from 200 ␮L of
each of the 5 amniotic fluid aliquots described above with
the QIAamp DNA Mini Kit (Qiagen), according to the
“Blood and Body Fluid Spin Protocol” as recommended
by the manufacturer. DNA was stored at ⫺80 °C until
analysis.
Real-time quantitative reverse transcription-PCR was
used to measure the mRNA concentration in amniotic
fluid, with transcript quantification verified by parallel
amplification of the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene, as described previously (7 ).
Real-time quantitative PCR was used to measure the
DNA concentration in amniotic fluid, with transcript
quantification verified by parallel amplification of the
␤-globin gene, as described previously by Lo et al. (15 )
except that each primer was used at 100 nM and the probe
was used at 50 nM.
Amplification data were collected by the 7700 Sequence
Detector and analyzed with the Sequence Detection System software, Ver. 1.6.3 (PE-ABI). Each sample was run in
triplicate with the mean results of the 3 reactions used for
further calculations.
Descriptive statistics, including medians and 25th and
75th percentile ranges, were generated for GAPDH
mRNA and ␤-globin DNA in amniotic fluid separately in
5 aliquots: uncentrifuged, cell-free unfiltered, and portions passed through filters with pore sizes 0.22, 0.45, and
5 ␮m. A nonparametric Friedman 2-way ANOVA test
was carried out to compare differences in GAPDH mRNA
and ␤-globin DNA concentrations between aliquots. As a
follow-up procedure to compare effects of filter sizes in a
pairwise manner with an adjustment for multiple comparisons, we used the Student–Newman–Keuls test with
prior logarithmic transformation of the studied outcomes.
The threshold for significance was set at ␣ ⫽ 0.05. All
statistical tests were performed with SAS/STAT software
(SAS Institute, Inc.).
The decrease in GAPDH mRNA concentration in amniotic fluid samples with cell removal and filtration with
decreasing pore size is shown in Fig. 1A (Friedman test,
P ⫽ 0.0001). Pairwise analysis showed a statistically
significant difference between the samples filtered with
0.22 ␮m pore size and the rest (Student–Newman–Keuls
test, P ⬍0.05). Overall, the GAPDH mRNA concentration
decreased by a median of more than 60-fold (interquartile
range, 27- to 100-fold) in comparisons of paired samples
from the uncentrifuged samples and the portions passed
through the 0.22 ␮m filters. The GAPDH mRNA concentration decreased by a median of more than 17-fold
(interquartile range, 12- to 19-fold) in comparisons of
paired samples from the centrifuged, unfiltered samples
and the portions passed through the 0.22 ␮m filters. In
comparison, there was no statistically significant difference in ␤-globin DNA concentrations (Friedman test, P ⫽
0.98) except for the uncentrifuged fraction vs the rest (Fig.
1B). Pairwise analysis confirmed the statistically significant difference between the uncentrifuged fraction vs the
rest (Student–Newman–Keuls test, P ⬍0.05). The ␤-globin
DNA concentration decreased by a median of more than
1025
Fig. 1. Amniotic fluid mRNA (A) and DNA (B) concentrations before and
after centrifugation to remove cells and after filtration through different
pore sizes.
(A), amniotic fluid GAPDH mRNA concentrations (ng/L), as determined by
real-time quantitative reverse transcription-PCR (y axis), plotted against filter
pore size and cell-free vs uncentrifuged (unspun) fractions (x axis). (B), amniotic
fluid ␤-globin DNA concentrations (genome-equivalents/mL), as determined by
real-time quantitative PCR (y axis), plotted against filter pore size and cell-free vs
uncentrifuged fractions (x axis). The lines inside the boxes denote medians. The
⫹ denote means. The boxes mark the interval between the 25th and 75th
percentiles. The whiskers denote the interval between the 10th and 90th
percentiles. 䡺 indicate data points outside the 10th and 90th percentiles. The
ⴱ in panel B denotes a single outlier at ⬃116 000 genome-equivalents/mL in the
unspun data set; this outlier was removed to allow for clarity of data presentation.
32-fold (interquartile range, 10- to 50-fold) in comparisons
of the paired samples from the uncentrifuged samples
and the portions passed through the 0.22 ␮m filters.
The study of cell-free RNA, particularly fetal RNA in
the maternal circulation, and RNA in amniotic fluid is a
new field of interest with many potential clinical applications (12–14 ). However, very little is known about the
kinetics and origin of cell-free mRNA. Apoptosis has been
suggested as a source of these nucleic acids (16 ) and could
explain the remarkable stability of cell-free mRNA as a
result of packaging into protected apoptotic bodies (17 ).
Ng et al. (7 ) recently explored the properties of nucleic
1026
Technical Briefs
acids in plasma and showed that filterable GAPDH
mRNA species are present, and therefore likely to be
particle bound, whereas the majority of ␤-globin DNA is
not filterable and thus is not particle bound.
Our study analyzed cell-free nucleic acids in amniotic
fluid for the presence of particle-associated mRNA species, and like Ng et al. (7 ), we found the greatest decrease
in GAPDH mRNA after filtration through a 0.22 ␮m filter,
whereas filtration did not significantly reduce the amount
of cell-free ␤-globin DNA present. Additionally, this study
evaluated the difference in amounts of nucleic acids
present in whole amniotic fluid containing cells and the
cell-free fraction. Much more ␤-globin DNA could be
isolated from whole amniotic fluid than from the cell-free
fraction, but there was no statistically significant difference in the amount of GAPDH mRNA that could be
isolated from the two fractions.
This study demonstrates similar properties of cell-free
nucleic acids in amniotic fluid and in plasma. Ng et al. (7 )
suggested that circulating DNA and RNA in plasma are
both protected from degradation by associated particles,
but the non-particle–associated form of DNA is less
efficiently degraded than the non-particle–associated
form of RNA and thus is present in much greater quantities relative to the particle-associated form. We extend
this hypothesis to the properties of cell-free nucleic acids
in amniotic fluid. These data suggest that in different
body fluids, there is a universal mechanism of cell-free
nucleic acid processing, possibly via packaging during
apoptosis.
To our knowledge, this is the first study to evaluate the
properties of cell-free mRNA in amniotic fluid. We have
also studied fetal gene expression in amniotic fluid (12 ).
More research is needed to further evaluate the origin and
kinetics of cell-free nucleic acids in amniotic fluid as this
material has significant clinical potential for the study of
health and development in living fetuses.
References
1. Lo YMD. Circulating nucleic acids in plasma and serum: an overview. Ann
N Y Acad Sci 2001;945:1–7.
2. Kopreski MS, Benko FA, Kwak LW, Gocke CD. Detection of tumor messenger
RNA in the serum of patients with malignant melanoma. Clin Cancer Res
1999;5:1961–5.
3. Lo KW, Lo YMD, Leung SF, Tsang YS, Chan LY, Johnson PJ, et al. Analysis
of cell-free Epstein-Barr virus associated RNA in the plasma of patients with
nasopharyngeal carcinoma. Clin Chem 1999;45:1292– 4.
4. Chen XQ, Bonnefoi H, Pelte MF, Lyautey J, Lederrey C, Movarekhi S, et al.
Telomerase RNA as a detection marker in the serum of breast cancer
patients. Clin Cancer Res 2000;6:3823– 6.
5. Silva JM, Dominguez G, Silva J, Garcia JM, Sanchez A, Rodriguez O, et al.
Detection of epithelial messenger RNA in the plasma of breast cancer
patients is associated with poor prognosis tumor characteristics. Clin
Cancer Res 2001;7:2821–5.
6. Dasi F, Lledo S, Garcia-Granero E, Ripoll R, Marugan M, Tormo M, et al.
Real-time quantification in plasma of human telomerase reverse transcriptase (hTERT) mRNA: a simple blood test to monitor disease in cancer
patients. Lab Invest 2001;81:767–9.
7. Ng EKO, Tsui NBY, Lam NYL, Chiu RWK, Yu SCH, Wong SCC, et al. Presence
of filterable and nonfilterable mRNA in the plasma of cancer patients and
healthy individuals. Clin Chem 2002;48:1212–7.
8. Hasselmann DO, Rappl G, Tilgen W, Reinhold U. Extracellular tyrosinase
mRNA within apoptotic bodies is protected from degradation in human
serum. Clin Chem 2001;47:1488 –9.
9. Poon LL, Leung TN, Lau TK, Lo YMD. Presence of fetal RNA in maternal
plasma. Clin Chem 2000;46:1832– 4.
10. Ng EK, Tsui NB, Lau TK, Leung TN, Chiu RW, Panesar NS, et al. mRNA of
placental origin is readily detectable in maternal plasma. Proc Natl Acad Sci
U S A 2003;100:4748 –53.
11. Wataganara T, LeShane ES, Chen AY, Borgatta L, Peter I, Johnson KL, et al.
Plasma gamma-globin gene expression suggests that fetal hematopoietic
cells contribute to the pool of circulating cell-free fetal nucleic acids during
pregnancy. Clin Chem 2004;50:689 –93.
12. Larrabee PB, Johnson KL, Lai C, Ordovas J, Cowan JM, Tantravahi U, et al.
Global gene expression analysis of the living human fetus using amniotic
fluid: a feasibility study. JAMA 2005;293:836 – 42.
13. Wataganara T, Bianchi DW. Fetal cell-free nucleic acids in the maternal
circulation: new clinical applications. Ann N Y Acad Sci 2004;1022:90 –9.
14. Wataganara T, LeShane ES, Chen AY, Sullivan LM, Peter I, Borgatta L, et al.
Circulating cell-free fetal nucleic acid analysis may be a novel marker of
fetomaternal hemorrhage after elective first-trimester termination of pregnancy. Ann N Y Acad Sci 2004;1022:129 –34.
15. Lo YM, Tein MS, Lau TK, Haines CJ, Leung TN, Poon PM, et al. Quantitative
analysis of fetal DNA in maternal plasma and serum: implications for
noninvasive prenatal diagnosis. Am J Hum Genet 1998;62:768 –75.
16. Hahr S, Hentze H, Englisch S, Hardt D, Fackelmayer FO, Hesch RD, et al.
DNA fragments in the blood plasma of cancer patients: quantitations and
evidence for their origin from apoptotic and necrotic cells. Cancer Res
2001;61:1659 – 65.
17. Halicka HD, Bedner E, Darzynkiewicz Z. Segregation of RNA and separate
packaging of DNA and RNA in apoptotic bodies during apoptosis. Exp Cell
Res 2000;260:248 –56.
DOI: 10.1373/clinchem.2004.047670
Stability of Nucleosomal DNA Fragments in Serum,
Stefan Holdenrieder,* Susanne Mueller, and Petra Stieber
(Institute of Clinical Chemistry, University of Munich,
Munich, Germany; * address correspondence to this author at: University Hospital of Munich-Grosshadern, Institute of Clinical Chemistry, Marchioninistrasse 15,
D-81377 Munich, Germany; fax 49-89-7095-6298, e-mail
Stefan.Holdenrieder@.med.uni-muenchen.de)
Circulating DNA is increased in various benign and
malignant pathologic conditions, including cancers, sepsis, and graft-vs-host and autoimmune diseases as well as
after trauma or ischemia (1– 8 ). Changes in circulating
DNA correlate with the response to antitumor therapy
and with tumor recurrence (9 –11 ). Furthermore, DNA
concentration reportedly has predictive and prognostic
relevance in cancer (11, 12 ). Despite the nonspecific nature of circulating DNA, it might have considerable
potential for monitoring cancer and management of therapy (9 –12 ).
In serum and plasma, DNA is thought to exist predominantly as mono- and oligonucleosomes (13, 14 ), which
are formed by a core particle of a double set of the
histones H2A, H2B, H3, and H4 wrapped by 146 bp of
DNA on the outside (15 ). By this composition they seem
to be protected against rapid digestion by endonucleases
(16 ). Circulating nucleosomes can be quantified by realtime PCR of the DNA but also by immunologic assays,
which are particularly well suited for serial measurements
(17 ).
Achieving reliable results in these immunochemical
assays requires adherence to a strict preanalytical protocol
that includes careful venipuncture, centrifugation of the
1027
Fig. 1. Results of the experiments on preanalytical conditions, shown in absolute nucleosome concentrations [measured in arbitrary units (AU)]
and deviations (%) from the concentrations measured under standard conditions.
(A–F), serum samples were stabilized with 10 mmol/L EDTA, aliquoted, and incubated for 1, 2, 3, 4, 6, and 24 h and 2, 3, and 7 days, respectively, at 4, 25, and 37 °C,
respectively, before storage at ⫺70 °C and nucleosome measurement. Samples showed only minor changes when incubated at 4 and 25 °C, whereas a considerable
decrease was observed after incubation at 37 °C. (G and H), stabilized sera were vortex-mixed for 5, 10, and 30 s or, alternatively, rolled in an overhead roller for 15
and 30 min. They were then stored at 25 or at 37 °C for 4 h before being frozen at ⫺70 °C. Neither shaking nor rolling influenced the measured nucleosome
concentrations, but after incubation at 37 °C, the measured concentrations were lower. (I), stabilized sera underwent several freeze–thaw cycles before nucleosome
measurements. (J), deep-frozen sera were thawed 2 and 12 h before measurement and were, meanwhile, stored at 4 and 25 °C. Freezing–thawing had no impact on
nucleosome concentrations.
1028
Technical Briefs
sample within 1–2 h after venipuncture, addition of EDTA
for stabilization of nucleosomes, and storage at ⫺70 °C if
measurement is to be delayed. This procedure is based on
our earlier studies on preanalytical factors that could
influence the nucleosome concentrations between venipuncture and centrifugation, between centrifugation and
EDTA addition, between EDTA addition and freezing,
during long-term storage, and between thawing and test
performance (17 ). Our results indicated that a delay
between venipuncture and centrifugation can lead to a
time-dependent increase in nucleosome concentrations,
which was most pronounced at 37 °C, whereas a delay in
EDTA addition after centrifugation was associated with a
time-dependent decrease in results (17 ).
In many instances, blood samples are transported to the
laboratory by mail; we therefore investigated various
additional preanalytical conditions that might influence
the stability of nucleosomes in serum during shipping:
Sera from 5 volunteers were exposed to prolonged time of
transportation, different temperatures, shaking and rolling, several freeze–thaw cycles, and measurements with
various delays after thawing. In all experiments, the
samples were centrifuged, within 30 min after venipuncture, at 3000g for 15 min and stabilized with 10 mmol/L
EDTA (pH 8) immediately after centrifugation. Subsequently, they were aliquoted, and methodical experiments were performed. They were then stored at ⫺70 °C
and analyzed in batches containing all samples from a
single patient.
The nucleosome ELISA (Cell Death Detection ELISAplus;
Roche Diagnostics) is based on a quantitative sandwich
enzyme immunoassay principle: Monoclonal mouse antibodies directed against DNA (single- and doublestranded DNA) and histones (H1, H2A, H2B, H3, and H4)
detect specifically mono- and oligonucleosomes. The antihistone antibody is bound to the microtiter plate, whereas
the anti-DNA antibody labeled with peroxidase reacts
with 2,2⬘-azino-di(3-ethylbenzthiazoline-sulfonate). The
amount of captured nucleosomes is proportional to the
resulting color development and enables spectrophotometric quantification in arbitrary units (17 ).
In the first experiment, we varied the time between
stabilization of the sera with EDTA and storage at ⫺70 °C
(1, 2, 3, 4, 6, and 24 h and 2, 3, and 7 days), and samples
were stored at various temperatures during that time (4,
25, and 37 °C) to simulate potential stressful transportation conditions. Prolonged “transportation time” clearly
did not influence the values in those stabilized sera that
were stored at 4 and 25 °C. The median SD for all time
points was ⬍10%. At 37 °C, however, the values decreased continually, and after 24 h, only approximately
one-half of the initial concentration remained (median SD,
50.2%; Fig. 1, A–F).
In the second experiment, we investigated the influence
of agitation by vortex-mixing the samples for 5, 10, and
30 s or, alternatively, by rolling them in a slow overhead
roller for 15 and 30 min. Subsequently, all of the samples
were incubated at 25 or 37 °C for 4 h before they were
frozen at ⫺70 °C. Neither after shaking nor after rolling
the samples did we observe major changes in the concentrations of stabilized sera, particularly if they were incubated at 25 °C for 4 h (median SD ⬍5%). However, after
additional incubation at 37 °C for 4 h, the values tended to
be lower (median SD up to 20%; Fig. 1, G and H).
We then analyzed the influence of freezing and thawing
on the stabilized sera. Repeated refreezing (up to 3 times
at ⫺70 °C) led to only minor changes in the concentrations
(median SD ⬍10%). Thawing the samples at various time
points before measurements (2 and 12 h) and storage at
various temperatures (4 and 25 °C) also had no impact on
the values (median SD ⬍10%; Fig. 1, I and J).
Our results indicate that the concentration of nucleosomes in sera stabilized with 10 mmol/L EDTA is not
influenced by preanalytical conditions such as time of
transportation, moderate temperature (4 –25 °C), shaking,
rolling, and several freeze–thaw cycles. However, longterm exposure to high temperatures (37 °C) should be
avoided as it can cause a notable decrease in the measured
nucleosome concentration. This might be attributable to
enhanced activation of serum nucleases or by direct
thermal damage of the nucleosomes.
When these precautions are taken and the preanalytical
protocol is followed, including early centrifugation and
subsequent stabilization of the sera with EDTA, samples
can be shipped by mail without adverse effects on the
results of nucleosome measurements.
The nucleosome assays were provided by Roche Diagnostics, Germany.
References
1. Leon SA, Shapiro B, Sklaroff DM, Yaros MJ. Free DNA in the serum of cancer
patients and the effect of therapy. Cancer Res 1977;37:646 –50.
2. Shapiro B, Chakrabarty M, Cohn EM, Leon SA. Determination of circulating
DNA levels in patients with benign or malignant gastrointestinal disease.
Cancer 1983;51:2116 –20.
3. Holdenrieder S, Stieber P, Bodenmueller H, Busch M, Fertig G, Fuerst H, et
al. Nucleosomes in serum of patients with benign and malignant diseases.
Int J Cancer 2001;95:114 –20.
4. Zeerleder S, Zwart B, Wuillemin WA, Aarden LA, Groeneveld AB, Caliezi C, et
al. Elevated nucleosome levels in systemic inflammation and sepsis. Crit
Care Med 2003;31:1947–51.
5. Lo YMD, Tein MSC, Pang CCP, Yeung Ck, Tong KL, Hjelm NM. Presence of
donor-specific DNA in plasma of kidney and liver transplant recipients.
Lancet 1998;351:1329 –30.
6. Lo YMD, Rainer TH, Chan LYS, Hjelm NM, Cocks RA. Plasma DNA as a
prognostic marker in trauma patients. Clin Chem 2000;46:319 –23.
7. Rainer TH, Wong LKS, Lam W, Yuen E, Lam NYL, Metrewel C, et al.
Prognostic use of circulating plasma nucleic acid concentrations in patients
with acute stroke. Clin Chem 2003;49:562–9.
8. Ziegler A, Zangemeister-Wittke U, Stahel RA. Circulating DNA: a new
diagnostic gold mine? Cancer Treatment Rev 2002;28:255–71.
9. Holdenrieder S, Stieber P. Therapy control in oncology by circulating nucleosomes. Ann N Y Acad Sci 2004;1022:211– 6.
10. Lo YM, Chan LY, Chan AT, Leung SF, Lo KW, Zhang J, et al. Quantitative and
temporal correlation between circulating cell-free Epstein-Barr virus DNA and
tumor recurrence in nasopharyngeal carcinoma. Cancer Res 1999;59:
5452–5.
11. Holdenrieder S, Stieber P, von Pawel J, Raith H, Nagel D, Feldmann K, et al.
Circulating nucleosomes predict the response to chemotherapy in patients
with advanced non small cell lung cancer. Clin Cancer Res 2004;10:
5981–7.
12. Gautschi O, Bigosch C, Huegli B, Jermann M, Marx A, Chasse E, et al.
Circulating deoxyribonucleic acid as prognostic marker in non-small-cell lung
cancer patients undergoing chemotherapy. J Clin Oncol 2004;22:4157– 64.
13. Rumore PM, Steinman CR. Endogenous circulating DNA in systemic lupus
14.
15.
16.
17.
erythematosus. Occurrence as multimeric complexes bound to histone.
J Clin Invest 1990;86:69 –74.
Chan KC, Zhang J, Chan AT, Lei KI, Leung SF, Chan LY, et al. Molecular
characterization of circulating EBV-DNA in the plasma of nasopharyngeal
carcinoma and lymphoma patients. Cancer Res 2003;63:2028 –32.
Luger K. Structure and dynamic behavior of nucleosomes. Curr Opin Genet
Dev 2003;13:127–35.
Ng EK, Tsui NB, Lam NY, Chiu RW, Yu SC, Wong SC, et al. Presence of
filterable and nonfilterable mRNA in the plasma of cancer patients and
healthy individuals. Clin Chem 2002;48:1212–7.
Holdenrieder S, Stieber P, Bodenmueller H, Fertig G, Fuerst H, Schmeller N,
et al. Nucleosomes in serum as a marker for cell death. Clin Chem Lab Med
2001;39:596 – 605.
DOI: 10.1373/clinchem.2005.048454
Relative Quantification of Experimental Data from Antigen Particle Arrays, Susan Pang,1* Julie Reeve,2 Michael
Walker,2 and Carole Foy1 (1 LGC Ltd, Teddington, Middlesex, United Kingdom; 2 Genesis Diagnostics Ltd.,
Littleport, Cambridgeshire, United Kingdom; * address
correspondence to this author at: LGC Ltd, Queens Road,
Teddington, Middlesex, TW11 0LY, United Kingdom;
e-mail Susan.Pang@lgc.co.uk)
Protein arrays typically consist of a capture protein on a
matrix, e.g., glass slide, silicon chip, or coded microparticle (1 ). The latter minimizes steric constraints and enhances reaction kinetics (2 ). Microarray technologies have
been used for detecting allergens (3 ) and cytokines (4 ). A
primary advantage of microarray technologies over conventional immunoassays is the ability to multiplex assays.
Currently, ELISA is the method of choice for autoantibody detection (5 ). This method, however, is labor-intensive and requires comparatively large sample and reaction volumes. Nonetheless, ELISAs are currently
performed to aid differential diagnosis of certain autoimmune diseases. Dermatomyositis (6 ) is characterized by
detection of anti-Jo-1 IgG in patient sera. Both anti-Sm and
anti-RNP/Sm IgGs are indicative of systemic lupus erythematosus (7 ). The presence of anti-Scl70 IgGs aids
diagnosis of systemic sclerosis (8 ). Anti-SSB and -SSA
IgGs are present in systemic lupus erythematosus (9 ) or
Sjögren syndrome (10 ). With microarray technologies, all
these autoantibodies can be screened simultaneously.
The Luminex particle array platform comprises a hundred microparticles, each possessing a distinct fluorescent
signature generated by a blend of two internal fluorescent
dyes. Capture protein is conjugated to the bead surface, to
assay for the cognate entity within a single reaction vessel.
The instrument includes a microfluidics system and two
lasers. A 635 nm laser excites the red and infrared
classifier fluorophores that form the particular signature
of each bead set. The second laser (523 nm) excites
phycoerythrin dye used as a molecular tag for detection.
Detection of cytokines (11 ), cancer markers, and allergens
(12 ) has been reported.
Use of internal calibration curves for cytokine quantification has been documented (13 ), but with antigen arrays,
quantification is more difficult because of the time and
1029
expense required for antibody synthesis. Where quantification of antibodies has been cited, competitive immunoassays are described that use labeled forms of the target
analyte specifically tailored for the particular assays described (14 ). Assays for quantifying total immunoglobulin content have also been described (15 ), but not in the
context of creating a reference material for a specific
antibody within the same class of immunoglobulins.
Commercial Luminex-based assays for autoantibodies
are available, e.g., from Linco Research and Zeus Scientific. These tests are qualitative, however, and do not
allow for direct comparisons between assays. In this
report we describe a set of internal standards for antigen
arrays that enable interassay comparisons by creating a
point of reference for the detection of human IgG. Relative
quantification would enable monitoring of treatment administered to combat disease.
We illustrate the use of an internal IgG calibration curve
and the detection of six autoantibodies: Jo-1 IgG, Sm IgG,
Scl-70 IgG, RNP/Sm IgG, SSB/La IgG and SSA/Ro IgG in
patient serum samples. Recombinant forms of the cognate
antigens (5 mg/L in phosphate-buffered saline, pH 7.4;
AroTec Diagnostics Ltd.) were coupled to the surface of
Luminex xMapTM carboxylated microspheres according
to the manufacturer’s instructions. Within the multiplex,
⬃10 000 of each antigen-coupled bead set were challenged with serum diluted 1 in 300 (50 ␮L), obtained from
Genesis Diagnostics. To assess the scope for quantification, we used 22 serum samples. Each reaction mixture
was agitated at room temperature for 1 h. Tubes were
microcentrifuged for 1 min, and the resulting supernatant
was discarded. Beads were washed three times with
protein array wash buffer [50 ␮L; phosphate-buffered
saline (pH 7.4), containing 10 g/L bovine serum albumin,
0.2 g/L Tween 20, and 0.2 g/L sodium azide; Sigma]. The
beads were incubated with biotinylated sheep anti-human IgG antibody (Amersham Pharmacia Biotech UK)
diluted 1 in 10 000 (100 ␮L) and mixed for 1 h at room
temperature. Beads were washed before incubation with
streptavidin-conjugated phycoerythrin (400 ng/100 ␮L;
Molecular Probes) for 30 min at room temperature. Tubes
were foil-wrapped to prevent photobleaching of beads.
Beads were washed before injection into the Luminex
instrument, in which a minimum of 100 events per bead
set were analyzed. Serum was designated as positive if
the fluorescent output was greater than the upper 95%
confidence interval of the single “normal” (nondisease
state) serum sample included in each assay. The negative
control contained antigen-conjugated bead sets treated
with protein array buffer.
To quantify the analytes relative to a reference point, 11
sets of calibration beads were synthesized and incorporated
into the assay. These comprise microspheres conjugated to
known quantities of purified human IgG, ranging from 10
ng/L to 250 mg/L, to construct the calibration curve.
Three experiments were performed, each with triplicate
determinations. Mean values, 95% confidence intervals,
and CVs were determined with Microsoft Excel 97. Twoway ANOVAs (Statistica, Ver. 6; StatSoft) were applied.
1030
Technical Briefs
Fig. 1. IgG calibration curve.
Data points represent mean values of triplicate determinations, and the error
bars indicate 95% confidence intervals. AU, arbitrary units.
Plotting the logarithm of known IgG concentrations
against observed fluorescence output (Fig. 1) produced a
highly robust sigmoidal trendline with a correlation coefficient exceeding 0.95 (n ⫽ 12). The antibody–antigen interaction is known to exhibit this trend, as demonstrated by other
immunodetection methods (16 ). Median fluorescent intensities (MFIs) from multiplexed antigen arrays were interpolated from the IgG calibration curve constructed from a
distinct multiplexed antibody array assay with all 11 concentrations of IgG-coupled bead sets within a separate
reaction vessel. This enabled conversion of MFIs (ranging
from 0 to 15 000 arbitrary units) to conventional units of
measure (␮g/L) relative to the known concentration of IgG
coupled to the calibration bead sets.
Interpolation of each experimental data point from the
IgG calibration curve within the linear portion of the
curve lowered the majority of CVs for the assay of each
sample (Table 1). The MFI from the SSA/Ro IgG assay of
serum sample 7 exceeded the range of the calibration
curve; therefore, no relative concentration could be determined for this sample. CVs in Table 1 highlighted in bold
denote values that increased on interpolation. Increases in
the CVs ⬎5.73% were because the MFIs lay within the
plateau of the curve. Pairwise analysis of the platform and
samples showed the extent of the variation when these
high-scoring positives were interpolated (Fig. 1 in the
Data Supplement that accompanies the online version of
this Technical Brief at http://www. clinchem.org/content/
vol51/issue6/).
Trends in the observed MFIs for the three experiments
were consistent for all six IgG assays. For data presented
as MFI, the highest values were observed during the first
experiment, 4 days after coupling of antigen to the beads.
A significant decrease in fluorescent emission was apparent by the time experiment 2 was performed 7 days after
coupling. An additional slight decrease in fluorescent
output was seen between experiments 2 and 3; the latter
Table 1. Differences (%) between the CVs obtained by subtracting the CVs of the fluorescent output in MFI from the data
interpolated from the IgG calibration curve.a
Assay
Sample
Jo IgG
Sm IgG
Scl70 IgG
RNP/Sm IgG
SSB/La IgG
SSA/Ro IgG
Blank
Negative
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
0
⫺0.42
⫺0.33
⫺0.75
⫺0.99
⫺1.32
⫺0.46
⫺0.54
⫺0.56
⫺0.18
⫺3.12
⫺2.36
⫺0.84
⫺0.57
⫺0.94
⫺0.39
⫺29.53
⫺0.53
⫺0.60
⫺0.61
8.86
⫺0.43
⫺0.14
⫺2.03
⫺0.99
⫺0.06
⫺0.90
1.57
⫺0.75
1.00
⫺0.39
1.95
⫺0.52
⫺0.25
⫺2.27
⫺1.42
⫺1.37
1.34
⫺0.73
⫺0.61
⫺14.03
1.43
⫺1.15
⫺0.76
⫺2.52
1.43
⫺1.74
⫺0.82
⫺2.09
⫺1.21
⫺1.36
⫺0.88
⫺0.69
⫺0.80
⫺0.19
⫺1.13
⫺0.83
⫺0.89
⫺1.53
⫺0.14
⫺5.90
⫺0.18
⫺0.09
⫺0.39
11.53
⫺0.79
⫺0.46
⫺0.63
⫺3.66
⫺1.44
2.18
⫺1.80
⫺1.26
⫺0.44
⫺1.05
0.23
⫺0.90
0.23
⫺0.70
2.04
⫺1.05
⫺0.59
⫺2.24
⫺1.57
⫺1.85
0.41
⫺0.96
⫺0.60
⫺26.63
⫺0.07
⫺1.54
⫺1.41
⫺3.41
1.33
⫺0.56
⫺2.23
0
⫺0.59
⫺1.32
⫺0.44
⫺0.78
⫺0.89
⫺0.20
⫺0.29
4.58
⫺0.70
5.73
⫺1.39
4.53
⫺0.58
⫺0.97
⫺0.33
⫺29.83
⫺0.25
⫺0.39
⫺0.49
⫺2.62
⫺0.68
⫺0.75
10.16
⫺0.75
⫺0.17
⫺0.75
0.16
⫺0.90
⫺0.25
⫺0.47
0.70
NAb
⫺0.58
12.28
⫺0.42
2.41
4.12
⫺1.25
⫺0.05
⫺20.58
0.88
⫺0.57
19.30
⫺2.43
1.62
⫺0.42
24.82
a
b
Values in bold increased on interpolation.
NA, no relative concentration could be determined because the MFI in the assay exceeded the range of the calibration curve.
was performed 8 days after coupling (Fig. 2 in the online
Data Supplement). However, on interpolation of data
points from the calibration curve, the trend was reversed
such that the highest signal was observed for experiment
3 and the lowest for experiment 1.
In parallel studies, antigen-conjugated beads stored at
4 °C gave consistent fluorescent emission within a period
of 1 month, whereas antibody-conjugated beads showed
diminished fluorescent emission (data not shown).
We identified two limitations of this methodology: The
inherent instability of antibody-coupled beads and the
occurrence of data points from test samples outside the
linear portion of the semilogarithmic calibration curve. To
resolve the latter issue, serum samples exhibiting fluorescent output within the plateau of the trendline should be
reassayed after further dilution. Problems with long-term
stability of protein-conjugated bead sets were evident
when the beads were stored at 4 °C. On the Luminex
platform, antibody-conjugated beads were viable for approximately 3 weeks. Antigen-conjugated beads exhibited
slightly greater longevity, although decoding of the fluorescent signatures was problematic after storage at 4 °C
beyond 1 month. The constituents of the storage buffer
may have a detrimental effect on the fluorescent dyes
within the microspheres.
The reversal of the signal output profile suggests that
antibody-bound beads were more liable to degradation
than antigen-coupled bead sets within the same timescale.
The more elaborate structural complexity of antibodies
compared with antigens may account for the greater
instability of the former. Rapid freezing and lyophilization were procedures explored as alternative methods to
prolong the shelf-life of protein-coupled beads, and both
approaches appeared to be feasible (17 ). This provides the
possibility of developing calibration bead sets as reference
materials, thus enabling Luminex assay standardization.
This study illustrated the complexity of quantifying target
analytes within antigen arrays. Production of purified antibodies is laborious and expensive. Methods that can be used
for antibody purification, e.g., affinity chromatography,
could theoretically be used to obtain material comparable to
the target analyte of an antigen array. However, consistent
antibody purity is paramount for quantification.
Although this approach has broad application for the
comparison of any IgG, it will not measure absolute
concentrations of target analyte. This is largely because of
the presence of factors (e.g., soluble receptors, heterophilic antibodies, serum binding proteins, hemoglobin,
and lipids) in sera that can interfere with antibody-based
immunoassays (18 ). Nonetheless, this method has reduced intraassay variability and enables interassay comparisons for a wide range of antigen arrays.
This work was supported by a grant from the Department
of Trade and Industry, under the Measurements for
Biotechnology program. We thank Dr. Malcolm Burns
(LGC) and Dr. Steve Ellison (LGC) for statistical advice
and Dr. Lyndsey Birch (LGC) for reading this manuscript.
1031
References
1. Zhou H, Roy S, Schulman H, Natan MJ. Solution and chip arrays in protein
profiling. Trends Biotechnol 2001;19:S34 –S39.
2. Cutler P. Protein arrays: the current state-of-the-art. Proteomics 2003;3:3–18.
3. Fall BI, Eberlein-Konig B, Behrendt H, Niessner R, Ring J, Weller MG.
Microarrays for the screening of allergen-specific IgE in human serum. Anal
Chem 2003;75:556 – 62.
4. Wiese R, Belosludtsev Y, Powdrill T, Thompson P, Hogan M. Simultaneous
multianalyte ELISA performed on a microarray platform. Clin Chem 2001;
47:1451–7.
5. Alem M, Moghadam S, Malki J, Zaidi A, Nayak N, Li TM. Detection of
autoantibodies to nuclear antigens by EIA and IF techniques. Allerg Immunol
(Paris) 1997;188:191– 4.
6. Arnett FC, Hirsch TJ, Bias WB, Nishikai M, Reichlin M. The Jo-1 antibody
system in myositis: relationships to clinical features and HLA. J Rheumatol
1981;8:925–30.
7. Hildebrandt S, Weiner ES, Senecal JL, Noell GS, Earnshaw WC, Rothfield NF.
Autoantibodies to topoisomerase I (Scl-70): analysis by gel diffusion,
immunoblot, and enzyme-linked immunosorbent assay. Clin Immunol Immunopathol 1990;57:399 – 410.
8. Nakamura RM, Tan EM. Recent advances in laboratory tests and the
significance of autoantibodies to nuclear antigens in systemic rheumatic
diseases. Clin Lab Med 1986;6:41–53.
9. Chan EY, Mok TM, Lawton JW, Ko OK, Ho L, Lau CS. Comparison of counter
immunoelectrophoresis with immunoblotting for detection of anti-extractable
nuclear antigen antibodies in systemic lupus erythematosus. Asian Pac J
Allergy Immunol 1999;17:275–9.
10. Ben-Chetrit E, Fischel R, Rubinow A. Anti-SSA/Ro and anti-SSB/La antibodies in serum and saliva of patients with Sjogren’s syndrome. Clin Rheumatol
1993;12:471– 4.
11. Carson RT, Vignali DA. Simultaneous quantitation of 15 cytokines using a
multiplexed flow cytometric assay. J Immunol Methods 1999;227:41–52.
12. Bacarese-Hamilton T, Mezzasoma L, Ingham C, Ardizzoni A, Rossi R, Bistoni
F, et al. Detection of allergen-specific IgE on microarrays by use of signal
amplification techniques. Clin Chem 2002;48:1367–70.
13. de Jager W, te Velthuis H, Prakken BJ, Kuis W, Rijkers GT. Simultaneous
detection of 15 human cytokines in a single sample of stimulated peripheral
blood mononuclear cells. Clin Diagn Lab Immunol 2003;10:133–9.
14. Martins TB. Development of internal controls for the Luminex instrument as
part of a multiplex seven-analyte viral respiratory antibody profile. Clin Diagn
Lab Immunol 2002;9:41–5.
15. Gordon RF, McDade RL. Multiplexed quantification of human IgG, IgA, and
IgM with the FlowMetrix system. Clin Chem 1997;43:1799 – 801.
16. Koertge TE, Butler JE. The relationship between the binding of primary
antibody to solid-phase antigen in microtitre plates and its detection by the
ELISA. J Immunol Chem 1985;83:283–99.
17. Pang S, Smith J, Onley D, Reeve J, Walker M, Foy C. A comparability study of the
emerging protein array platforms with ELISAs. J Immunol Methods; in press.
18. Pantanowitz L, Horowitz GL, Upalakalin JN, Beckwith BA. Artifactual hyperbilirubinemia due to paraprotein interference. Arch Pathol Lab Med 2003;
127:55–9.
DOI: 10.1373/clinchem.2005.048512
Lupus Anticoagulant: Performance of a New, Fully
Automated Commercial Screening and Confirmation
Assay, Barbara Montaruli,1* Antonella Vaccarino,2 Cristina
Foli,2 Cecilia Rus,2 Cecilia Agnes,2 Maddalena Saitta,1 and
Mario Bazzan2 (1 Laboratory Analysis and 2 Department of
Haematology and Thrombotic Disorders, Ospedale Evangelico Valdese–ASL-1, Turin, Italy; * address correspondence to this author at: Laboratorio Analisi, OspedaleEvangelico Valdese–ASL-1, Via Silvio Pellico, 28, 10125
Turin, Italy; fax 39-11-6688640, e-mail b.montaruli@tin.it)
Lupus anticoagulants (LAs) are acquired circulating anticoagulants that interfere with phospholipid (PL)-dependent coagulation tests and are frequently associated with
thromboembolic disorders and obstetric complications.
1032
Technical Briefs
Detection of LAs is of major importance in patients with
these conditions (1, 2 ). LAs are diagnosed according to
the criteria proposed by the Scientific and Standardization
Committee on LAS of the International Society of Thrombosis and Hemostasis (3 ). According to these criteria, a
diagnosis of LA should follow a 4-step procedure respecting the following principles: (a) prolongation of 1 (or
more) PL-dependent clotting test (screening test); (b)
evidence of inhibition demonstrated after mixing equal
amounts of patient and normal plasma (mixing test); (c)
evidence that the inhibitor is PL dependent, as demonstrated by correction of the clotting defect in the presence
of excessively high PL concentrations (confirmatory test);
and (d) lack of specific inhibition of any coagulation factor
(distinction from other coagulopathies by specific factor
assays).
Despite these criteria, diagnosis of LA remains a problem for the clinical laboratory. Contributing to these
problems are the marked differences in sensitivity and
specificity for the various LA screening assays that have
been proposed, the lack of a universally accepted definition of a positive mixing test, technical variables affecting
the various assays for LA, the difficulty with result
interpretation, and the heterogeneous nature of LA itself
(4, 5 ). At present, the most used screening tests for
detecting LA are a dilute activated partial thromboplastin
time, kaolin clotting time, and the diluted Russell venom
time. Silica clotting time (SCT) has been described as a
specific and sensitive alternative to kaolin clotting time
for detecting LA (6 –9 ). We have evaluated a new automated “SCT screening and confirmatory” assay (not commercially available at this time) that has been proposed
for the detection of LA.
The SCT Screen (HemosIL; IL) was run on an ACL 9000
automated coagulometer (IL). Prolongation of the SCT
Screen tests was expressed as the ratio of patient coagulation time to the clotting time of the control (normal
pool). Mixing studies were carried out on 1:1 and 4:1
mixtures of patient and normal plasmas on all samples
that had prolonged SCT Screen times; failure to correct
the clotting time was considered evidence of an inhibitor.
For a confirmatory test of SCT, we used this new commercial assay (SCT Confirm, HemosIL; IL). LA was diagnosed when the confirmatory procedure was positive.
Whereas the results of SCT Screen coagulation tests were
expressed as the (sample clotting time/normal pool clotting time) ratio, the results of SCT Confirm tests were
expressed as a “normalized LA ratio”: the ratio result
from the LA screen test divided by the ratio result from
the LA confirmation test (patient confirm clotting time/
normal pool confirm clotting time).
Mean (SD) and the parametric 95th percentile of clotting time ratios in 30 healthy controls (voluntary blood
donors; 15 males and 15 females; age range, 16 – 81 years)
were 1.04 (0.12) and 1.21 for SCT Screen and 1.02 (0.10)
and 1.22 for the SCT Confirm test. Confirm ratios above
the 95th percentile were regarded as positive.
For quality control, we used a normal pool and a
LA-positive sample (LA⫹). Within-run imprecisions
(CVs; n ⫽ 10) for the SCT Screen were 1.0% for the normal
pool and 3.1% for the LA⫹ sample; for the SCT Confirm,
the CVs were 2.0% for the normal pool and 2.6% for the
LA⫹ sample. The CVs over 10 separate days for the SCT
Screen were 3.9% for the normal pool and 5.2% for the
LA⫹ sample; for the SCT confirm, the CVs were 4.8% for
the normal pool and 5.7% for the LA⫹ sample.
We investigated the diagnostic specificity in 41 patients
with known coagulation abnormalities: 12 patients on
low–molecular-weight heparin therapy (LMWH), 23 patients on oral anticoagulant therapy [OAT LA⫹ (n ⫽ 7)
and OAT LA⫺ (n ⫽ 16)], 3 patients with factor deficiencies of the intrinsic coagulation system (1 with a factor
VIII:C activity of 37%, 1 with a factor IX:C activity of 39%,
and 1 with a factor XI:C activity of 41%), and 3 patients
with type 1 von Willebrand disease (vWD; Table 1). Six of
12 patients on LMHW, all of the patients with defects of
intrinsic coagulation factors, and 1 of 3 patients with vWD
had prolonged SCT Screen times, but all of them were
identified as LA-negative by the SCT Confirm assay.
Nineteen of 23 patients on OAT had prolonged SCT
Screen times: 7 of these 19 previously identified as having
a LA were confirmed as LA-positive by the SCT Confirm
assay, whereas the other 12 were identified as LA-negative (see Fig. 1).
To evaluate the screening performance (sensitivity and
specificity) of the SCT Screen and Confirm assays, we
collected and studied consecutive plasmas from 136 “anticoagulant-free” patients (54 males and 82 females; age
range, 16 – 84 years) for whom a LA determination was
requested by the Department of Thrombosis and Hemorrhagic Diseases. All 136 plasma samples were further
analyzed for the presence of LA by our laboratory’s
routine LA (SCT in-house method) screening and confirmation tests (Fig. 1). Forty-six of 136 patients were identified as LA-positive by our routine LA tests. Of these
samples, a prolonged SCT Screen test was found in 40.
The inhibitor activity observed in SCT Screen-positive
patients was confirmed to be of the LA type by the SCT
Confirm assay. Six of 46 samples identified as having a LA
were SCT-negative; in these patients, the only test positive
was the diluted Russell venom time. Six of 91 LA-negative
patients were positive by SCT Confirm assay. Of these
patients, 2 were positive for anti-cardiolipin IgM (18.0 and
12.0 MPL, respectively; normal values ⬍7.0 kMPL/L), 1
for anti-prothrombin IgG (12.5 kIU/L; normal values ⬍9.0
kIU/L), 1 for anti-protein S IgM (21.0 kIU/L; normal
Table 1. Mean (SD) SCT Screen and SCT Confirm clotting
time ratios in patients during LMWH or OAT therapy and in
patients with factor deficiencies or vWD.
Mean (SD) ratio
LMWH (n ⫽ 12)
OAT (n ⫽ 23)
Factor deficiencies (n ⫽ 3)
vWD (n ⫽ 3)
SCT Screen
SCT Confirm
1.29 (0.13)
2.12 (1.32)
1.31 (0.19)
1.21 (0.17)
1.04 (0.08)
1.41 (0.80)
1.07 (0.05)
1.03 (0.04)
1033
8. Tripodi A, Chantarangkul V, Clerici M, Mannucci PM. Laboratory diagnosis of
lupus anticoagulant for patients on oral anticoagulant treatment. Performance of dilute Russell venom test and silica clotting time in comparison with
Staclot LA. Thromb Haemost 2002;88:583– 6.
9. Chantarangkul V, Tripodi A, Clerici M, Bressi C, Mannucci PM. Laboratory
diagnosis of lupus anticoagulants effect of residual platelets in plasma,
assessed by Staclot LA and silica clotting time. Thromb Haemost 2002;87:
854 – 8.
Fig. 1. SCT Confirm ratios in individual patients.
The dashed horizontal line indicates the upper limit of normal (95th percentile).
The boxes indicate the median (line inside each box) and 25th and 75th
percentiles (limits of each box); the whiskers indicate the range of values
between the 10th and 90th percentiles. fact def, factor deficiencies.
values ⬍15.0 kIU/L), and 1 for anti-protein C IgM (28.4
kIU/L; normal values ⬍18 kIU/L) autoantibodies, and 2
were negative for all anti-phospholipid antibodies investigated by the ELISA method. Thus, the new SCT assay
was positive in 87% of those who had a positive result by
our LA test and was negative in 93% of those whose
results were negative by our LA assay.
All of these patients had histories of thromboembolic
disease (4 with venous thrombosis and 2 with arterial
thrombosis). These findings can suggest, at least in some
patients, a role of SCT as an independent risk factor for
thrombosis. Furthermore, the presence of SCT positivity
in patients with thromboembolic events reduces, at least
in part, the number of patients who are otherwise seronegative for anti-phospholipid autoantibodies. More
data and prospective studies are needed to confirm this
hypothesis.
References
1. Greaves M, Cohen H, Machin SJ, Mackie I. Guidelines on the investigation
and management of the antiphospholipid syndrome. Br J Haematol 2000;
109:704 –15.
2. Arnout J. Antiphospholipid syndrome: diagnostic aspects of lupus anticoagulants. Thromb Haemost 2001;86:83–91.
3. Brandt JT, Triplett DA, Aving B, Scharrer I. Criteria for the diagnosis of lupus
anticoagulants: an update. Thromb Haemost 1995;74:1185–90.
4. Jennings I, Greaves M, Mackie IJ, Kitchen S, Woods TA, Preston FE. Lupus
anticoagulant testing: improvements in performance in a UK NEQAS proficiency exercise after dissemination of national guidelines on laboratory
methods. Br J Haematol 2002;119:364 –9.
5. Tripodi A, Biasiolo A, Chantarangkul V, Pengo V. Lupus anticoagulant testing:
performance of clinical laboratories assessed by a national survey lyophilized
affinity purified immunoglobulin with LA activity. Clin Chem 2003;49:1608 –
14.
6. Chantarangkul V, Tripodi A, Arbini A, Mannucci PM. Silica clotting time (SCT)
as a screening and confirmatory test for the detection of the lupus anticoagulants. Thromb Res 1992;67:355– 65.
7. Dragoni F, Minotti C, Palumbo G, Faillace F, Redi R, Borganzoni V, et al. As
compared to kaolin clotting time, silica clotting time is a specific and
sensitive automated method for detecting lupus anticoagulant. Thromb Res
2001;101:45–51.
Immunochemical Quantification of Free Light Chains in
Urine, Ileana Herzum,* Harald Renz, and Hans Günther Wahl
(Department of Clinical Chemistry and Molecular Diagnostics, Philipps University of Marburg, Marburg, Germany; * address correspondence to this author at: Department of Clinical Chemistry and Molecular Diagnostics,
Philipps University of Marburg, 35033 Marburg, Germany; e-mail herzumi@med.uni-marburg.de)
Quantitative measurements of plasma and urinary paraprotein concentrations play a major role in the monitoring
of patients with multiple myeloma. The concentrations
are routinely estimated from the size of the M-spike on
protein electrophoresis (PEL) or by automated immunologic assays for IgG, IgA, IgM, IgE, or IgD. In the case of
light chain myeloma and intact immunoglobulin myeloma with predominant light chain production, light
chain concentrations could, until recently, be measured
only by the size of the urinary light chain M-spike on PEL
or by the measurement of the total (free and bound) light
chain concentrations.
A latex-enhanced assay (Freelite; The Binding Site, Ltd.)
measuring free light chains (FLCs) in serum and urine has
recently become available for the BNII (Dade Behring)
analyzer. The Myeloma Management Guidelines (1 ) recommend the Freelite test for serial monitoring of the FLCs
in serum, but periodic 24-h urine collection is still required for Bence Jones proteinuria (BJP) and total urinary
protein (TUP) quantification. Depending on the glomerular and tubular function, serum and urine FLC concentrations may not change to the same degree (2 ), so that
monitoring of serum FLC alone is questionable for revealing the actual degree of disease in patients with BJP and
tubular dysfunction.
We evaluated the analytical performance of the immunochemical test for serum and urine with the BNII analyzer. The test uses antibodies that specifically recognize
an epitope of the common region of ␬ and ␭ light chains
that is “hidden” when the light chains are attached to the
immunoglobulin heavy chain (3 ).
Intra- (within) and interassay (day-to-day total) imprecision (CV) was determined with control material and
with patient samples containing high and low concentrations of polyclonal or monoclonal FLCs (Table 1). The
high CV observed for the serum sample with a high
monoclonal ␭ FLC concentration may reflect the variable
1034
Technical Briefs
degree of polymerization, which is common at high FLC
concentrations. This phenomenon has been described
previously, most notably for ␭ FLCs (4 ). The linearity of
urine samples from patients with BJP, evaluated as the
correlation coefficient of the linear regression line of the
measured vs expected values in serial linear dilutions,
was good for ␬ FLC (9 dilutions; range, 118 – 865 mg/L;
slope, 0.885; intercept, 81 mg/L; r ⫽ 0.90) and ␭ FLC (17
dilutions; range, 17–14 900 mg/L; r ⫽ 0.97). Linearity of
TUP measured simultaneously by the benzethonium chloride method (Roche Diagnostics) was very good for the
sample with ␬ FLC (9 dilutions; range, 0.03– 0.58 g/L;
slope, 1.0145; intercept, ⫺0.06 g/L; r ⫽ 0.995) and ␭ FLC
(17 dilutions; range, 0 –3.01 g/L; slope, 0.945; intercept,
⫺0.01 g/L; r ⫽ 0.96). The linearity of serum samples was
also determined for both FLCs: ␬ (9 dilutions; range,
88 –984 mg/L; slope, 1.0754; intercept, 151.4 mg/L; r ⫽
0.6984); ␭ (8 dilutions; range, 384-6600 mg/L; slope, 1.022;
intercept, ⫺50.29 mg/L; r ⫽ 0.9912). Because we observed
extremely high concentrations of FLCs, much higher than
the TUP, in some urine samples with BJP, we studied the
reliability of the Freelite test in urine.
We measured FLCs and TUP in 105 urine samples (87
patients) on which immunofixation electrophoresis (IFE;
SEBIA) had been performed. We measured TUP by the
benzethonium chloride and biuret method with the Hitachi 917 analyzer and by the modified biuret and pyrocatechol violet dry-chemistry method with the Vitros 250
analyzer (Johnson & Johnson). Urine IFE showed 63
samples with monoclonal bands; 20 were ␬ FLC, 15 were
␭ FLC, 21 were intact immunoglobulins plus ␭ (n ⫽ 10) or
␬ (n ⫽ 11) FLCs, and 7 were intact immunoglobulins
without FLC.
The FLC concentration ranges were 1– 4800 mg/L for ␬
and 1–14 200 mg/L for ␭. The lowest FLC concentration
with an associated monoclonal band by IFE was 4 mg/L.
Considering a ␬/␭ ratio outside the interval 1:2.71–1:0.25
(3 ) to be abnormal, we identified the FLC type of the BJP,
as shown by the monoclonal bands in the urine IFE, with
a sensitivity of 87% and a specificity of 53%.
We determined the imprecision (CV) of the TUP methods, using the same urine sample with ␭ BJP. The CVs
were 7.2% (0.06 g/L) for the benzethonium chloride, 12%
(0.56 g/L) for the biuret, 4.2% (0.48 g/L) for the modified
biuret, and 6.7% (0.05 g/L) for the pyrocatechol violet
method. The total urinary FLC concentration exceeded
the benzethonium chloride TUP in 54 of 105 cases, the
biuret TUP in 26 of 105 cases, the modified biuret TUP in
17 of 105 cases, and the pyrocatechol violet TUP in 46 of
105 cases, with maximum differences of 11, 9.2, 7.8, and 14
g/L, respectively.
To assess recovery of FLC by TUP methods, we measured FLC and TUP in a normal urine sample without
bands by IFE (␬ ⫽ 17.60 mg/L; ␭ ⫽ 7.44 mg/L) to which
we had added purified ␬ and ␭ light chains. The solutions
of purified material were provided and quantified by
radial immunodiffusion by The Binding Site. The final
concentrations of ␬ and ␭ FLCs added were 1240 and 930
mg/L, respectively. The linearity for the urine sample
with added FLCs, evaluated as the linear regression line
of the measured vs expected values in serial linear dilutions, was good for both ␬ (5 dilutions; range, 816 –9530
mg/L; R ⫽ 0.945) and ␭ (6 dilutions; range, 834 –1600
mg/L; R ⫽ 0.9441). TUP measurements of the samples
showed good linearity for all methods. Recovery of the
purified FLCs, however, differed among the 4 TUP methods. For ␬, the TUP results were 0.12, 3.9, 2.39, and 0.39
g/L for the benzethonium chloride, biuret, modified
biuret, and pyrocatechol violet, respectively, and for ␭, the
TUP results were 0.61, 3.2, 2.65, and 0.57 g/L, respectively.
Previous authors have emphasized the difficulty of
measuring clones of FLCs, as their structures are heterogeneous and can be modified through pH, polymerization, and oligomerization (5– 8 ). Both of the routinely
used methods for monitoring BJP, urine PEL and TUP, are
unspecific for FLCs and have several drawbacks. Urine
PEL is time-consuming and insensitive, requires previous
concentration, and is subject to interference from other
small urinary proteins in a tubular proteinuria pattern,
which frequently occurs in such patients (9 –11 ). The TUP
methods show variable, partial recovery of the FLCs
(12, 13 ). The Freelite assay provides specific quantification of BJP and has acceptable analytical performance.
Table 1. Imprecision of the Freelite assay on the BN II analyzer.
␬ FLC
Intraassay imprecision
Controls (n ⫽ 10)
Patient samples
Polyclonal FLCs
Serum (n ⫽ 10)
Urine (n ⫽ 10)
Monoclonal FLCs
Serum (n ⫽ 10)
Urine (n ⫽ 10)
3.3% (20 mg/L)
6.1% (42 mg/L)
␭ FLC
Interassay imprecision
2.7% (20 mg/L)
6.1% (42 mg/L)
Intraassay imprecision
3.0% (30 mg/L)
3.7% (66 mg/L)
2.8% (95 mg/L)
6.0% (90 mg/L)
2.0% (93 mg/L)
2.1% (37 mg/L)
2.7% (121 mg/L)
4.3% (172 mg/L)
4.3% (15 mg/L)
9.2% (1.7 mg/L)
13% (1090 mg/L)
8.0% (4490 mg/L)
4.4% (31 mg/L)
12% (1.2 mg/L)
Interassay imprecision
5.6% (30 mg/L)
5.4% (66 mg/L)
Because the diagnostic performance is poor, monoclonality needs to be confirmed by IFE (14 ).
We conclude that monitoring of renal involvement and
BJP in patients with FLC myeloma can be improved by
measuring both TUP and FLC in urine. Monitoring of the
TUP concentration should be performed with the same
assay.
We thank I. Pietrek and R. Heinz for excellent technical
support and The Binding Site, Ltd., for the purified light
chains.
References
1. Durie BG, Kyle RA, Belch A, Bensinger W, Blade J, Boccadoro M, et al. Myeloma
management guidelines: a consensus report from the Scientific Advisors of the
International Myeloma Foundation. Hematol J 2003;4:379 –98.
2. Waldmann TA, Strober W, Mogielnicki RP. The renal handling of low
molecular weight proteins. II. Disorders of serum protein catabolism in
patients with tubular proteinuria, the nephrotic syndrome, or uremia. J Clin
Invest 1972;51:2162–74.
3. Bradwell AR, Carr-Smith HD, Mead GP, Tang LX, Showell PJ, Drayson MT, et
al. Highly sensitive, automated immunoassay for immunoglobulin free light
chains in serum and urine. Clin Chem 2001;47:673– 80.
4. Abraham RS, Charlesworth MC, Owen BA, Benson LM, Katzmann JA, Reeder
CB, et al. Trimolecular complexes of lambda light chain dimers in serum of
a patient with multiple myeloma. Clin Chem 2002;48:1805–11.
5. Heino J, Rajamaki A, Irjala K. Turbidimetric measurement of Bence-Jones
proteins using antibodies against free light chains of immunoglobulins. An
artifact caused by different polymeric forms of light chains. Scand J Clin Lab
Invest 1984;44:173– 6.
6. Le Bricon T, Bengoufa D, Benlakehal M, Bousquet B, Erlich D. Urinary free
light chain analysis by the Freelite immunoassay: a preliminary study in
multiple myeloma. Clin Biochem 2002;35:565–7.
7. Solling K. Polymeric forms of free light chains in serum from normal
individuals and from patients with renal diseases. Scand J Clin Lab Invest
1976;36:447–52.
8. Solomon A, Schmidt W, Havemann K. Bence Jones proteins and light chains
of immunoglobulins. XIII. Effect of elastase-like and chymotrypsin-like neutral proteases derived from human granulocytes on Bence Jones proteins.
J Immunol 1976;117:1010 – 4.
9. Kyle RA. The monoclonal gammopathies. Clin Chem 1994;40:2154 – 61.
10. Levinson SS, Keren DF. Free light chains of immunoglobulins: clinical
laboratory analysis. Clin Chem 1994;40:1869 –78.
11. Handy BC. Urinary ␤2-microglobulin masquerading as a Bence Jones
protein. Arch Pathol Lab Med 2001;125:555–7.
12. Boege F, Koehler B, Liebermann F. Identification and quantification of
Bence-Jones proteinuria by automated nephelometric screening. J Clin
Chem Clin Biochem 1990;28:37– 42.
13. Watanabe N, Kamei S, Ohkubo A, Yamanaka M, Ohsawa S, Makino K, et al.
Urinary protein as measured with a pyrogallol red-molybdate complex,
manually and in a Hitachi 726 automated analyzer. Clin Chem 1986;32:
1551– 4.
14. Tate JR, Gill D, Cobcroft R, Hickman PE. Practical considerations for the
measurement of free light chains in serum. Clin Chem 2003;49:1252–7.
DOI: 10.1373/clinchem.2004.045435
Human N-Terminal proBNP Is a Monomer, Dan L.
Crimmins (Department of Pathology and Immunology,
Division of Laboratory Medicine, Washington University
School of Medicine, 660 South Euclid Ave., Box 8118, St.
Louis, MO 63110; fax 314-454-5208, e-mail crimmins@
pathology.wustl.edu)
The cardiac hormone B-type natriuretic peptide (BNP) is
synthesized in myocytes as a prepro 134-amino acid
1035
residue molecule. The 108-residue proBNP mature form
of the hormone is proteolytically cleaved to a biologically
active form of 32 amino acids (residues 77–108) and an
N-terminal fragment (residues 1–76; NT-proBNP) with an
as yet undefined biological function (1 ). Clinically, both
BNP and NT-proBNP have shown great promise as
secreted, bloodborne diagnostic markers of left ventricle
dysfunction. Measurement of each is based on immunoassays; it therefore is likely that changes in the molecular
form, e.g., posttranslational modifications, further proteolytic processing, and an oligomeric state for either analyte,
could affect their measurements (2–5 ). These types of
confounding molecular issues are likely for many analytes, with the myocyte damage marker cardiac troponin
I one of the better studied. In this case, the commercial
assays use antibodies directed to different epitopes, making “universal” calibration and determination of absolute
analyte concentration difficult (6 ).
A previous report has indicated that NT-proBNP exists
as a coiled-coil trimer, based on size-exclusion HPLC
(SE-HPLC) of human, plasma-extracted material and a
computer algorithm that predicts coiled-coils (7 ). I reinvestigated this claim on synthetic NT-proBNP, using the
physicochemical techniques of analytical sedimentation,
equilibrium ultracentrifugation, and circular dichroism
(CD), and demonstrate that NT-proBNP is a monomer
and not a trimer.
NT-proBNP was produced by solid-phase peptide synthesis (AnaSpec, Inc.) and obtained as a gift from DadeBehring (Newark, DE). I used N-terminal sequencing and
mass spectrometry as quality assurance procedures. Edman sequencing (8 ) was performed by Midwest Analytical, Inc. on 2 Coomassie-stained Immobilon-PSQ (Sigma)
NT-proBNP– containing membrane sections. The first 52
residues were positively identified, with no preview,
before the signal was not discernable from background
(data not shown). Matrix-assisted laser desorption/ionization mass spectrometry (8 ) gave an experimental mass
of 8457.0 compared with a calculated value of 8457.6 (data
not shown). Lyophilized NT-proBNP was dissolved in
and exhaustively dialyzed vs phosphate-buffered saline
(pH 7.2) at 4 °C. Analyte concentration was estimated
gravimetrically and based on a molar absorptivity at 280
nm (⑀280) of 0.82 L 䡠 g⫺1 䡠 cm⫺1; the 2 different techniques
yielded better than 95% agreement.
The synthetic peptide in a neutral pH physiologic salt
solution was run at 0.5 mL/min on SE-HPLC, and the
resulting elution profile is plotted in Fig. 1A as A214nm vs
time in minutes. The chromatographic process is monitored at 214 nm, which measures “peptide bond” absorbance; there thus is no bias in analyte detection under
these analysis conditions. This is in contrast to the antibody-based, postcolumn analysis of plasma-extracted
peptide (7 ), where detection is strictly a function of
antibody reactivity. Furthermore, it is unclear what effect,
if any, the C18 solid-phase plasma extraction procedure
used in that study has on the molecular state of NTproBNP before chromatography. Fig. 1A shows NTproBNP eluting well before cytochrome C (12.4 kDa) and
1036
Technical Briefs
just after myoglobin (17 kDa). One might be tempted to
interpret this result to imply that NT-proBNP is a dimer,
i.e., 8.5 kDa ⫻ 2 ⫽ 17 kDa; this is incorrect, however, as
discussed below. The analyte profile observed here is not
identical to that of the previous data (7 ), and is likely a
result of combined use of a different SE-HPLC column
packing material, different detection procedures, different
protein markers, and different sample preparation methods. Nonetheless, one common attribute is that the elution
of NT-proBNP is earlier than what would be expected of
a typical globular protein of ⬃8.5 kDa. It is invalid here,
and in general, on SE-HPLC performed under benign
conditions to use the elution position as a surrogate for
analyte molecular mass. The elution position on SE-HPLC
is dictated by hydrodynamic volume, which is a function
of the degree of hydration, molecular asymmetry, and the
polar/nonpolar nature of the analyte and not on molecular mass (9 ). It becomes possible to estimate molecular
mass for single-chain species only when the protein
calibrators and analytes possess the same tertiary structure, as occurs when denaturing/disulfide-reducing solvents are used, for example (10 ). In the neutral-pH
phosphate-buffered saline solution used here, all one can
conclude is that NT-proBNP elutes unexpectedly with a
larger molecular volume than the corresponding globular
protein of ⬃8.5 kDa.
Sedimentation equilibrium ultracentrifugation was used
to assess the oligomeric state of synthetic NT-proBNP.
This analysis was performed at Iowa State University
Protein Facility (ISUPF) on a Beckman Optima XL-A rotor
at 4 °C. A concentration of 0.26 g/L was used at rotor
speeds of 20 000, 30 000, and 40 000 rpm for various run
Fig. 1. SE-HPLC (A), sedimentation equilibrium (B), and CD (C) of NT-proBNP.
(A), SE-HPLC on a TosoHaas G2000SW column [600 ⫻ 7.5 mm (i.d)]. (Top), Bio-Rad Gel Filtration Protein Mixture (left to right: thyroglobin, IgG, ovalbumin, myoglobin,
and vitamin B12) with indicated molecular masses; (middle), horse heart cytochrome C (12.4 kDa); (bottom), NT-proBNP. (B), analytical sedimentation equilibrium
ultracentrifugation. (Top), residuals; (bottom), A280 vs radius. (C), CD analysis in a 1-mm pathlength cell at room temperature. Curve 1, 0.04375 g/L; curve 2, 0.0875
g/L; curve 3, 0.175 g/L. CD units are millidegrees.
1037
times with optical scanning at 280 nm, which monitors
aromatic residues in proteins/peptides. Data were plotted
as A280nm vs radius in centimeters. This plot is shown in
Fig. 1B, with the corresponding residuals for the 40 000
rpm rotor speed. For a single sedimenting ideal species,
the instrument software (Ver. 2.01) transforms the absorbance vs radius data into a molecular weight of 8351, i.e.,
a monomer. The residuals, which are a measure of the
goodness-of-fit of the curved line through the data points,
are randomly yet narrowly distributed around 0. This
attests to the high quality and lack of bias of the data. I
attempted to fit the 40 000 rotor speed data to a 2-idealspecies monomer– dimer and a monomer–trimer equilibrium. After 11–13 computer program iterations, the results, given as concentrations of each species, were as
follows: 0.262 g/L monomer with 1.4 ⫻ 10⫺4 g/L dimer,
and 0.262 g/L monomer with 5.7 ⫻ 10⫺7 g/L trimer.
Clearly, the only species present during ultracentrifugation was a monomer. It would not be possible to analyze
the serum-generated sample (7 ) by this physicochemical
technique because of the extremely low (ng/L) sample
concentration.
NT-proBNP was reported to be a trimer containing a
coiled-coil motif of repeating heptad units (7 ). Specifically, residues 17–38 were predicted to form a trimeric
coiled-coil in a pattern represented as a-b-c-d-e-f-g. Positions a and d in this 7-residue repeat are almost invariantly hydrophobic residues, e and g are usually charged
residues of opposite sign, and the remaining 3 residues
are usually hydrophilic. The molecular forces, including
sequence position and specific amino acids along the
7-residue motif, that determine coiled-coil formation have
been studied extensively (11 ). Typically a 4- or 5-heptad
repeat or greater is necessary to produce stable coiledcoils in benign neutral-pH buffer depending on the exact
amino acid sequence. Thus, the predicted 3-heptad coiledcoil would have to be extraordinarily stable to exist as a
trimer in benign medium, a point also discussed by
Seilder et al. (7 ). This 22-residue stretch represents ⬃30%
of the sequence; it therefore is reasonable to expect that
the helix content of this putative trimer would be at least
⬃30%. CD provides an excellent physicochemical measurement of protein helix content because the helix spectrum has a large diagnostically distinct negative doublet
minima pair at 222/208 nm (12 ). The CD run performed
at ISUPF on a Jasco J-710 instrument (Fig. 1C) showed no
such minima pair for 3 different NT-proBNP concentra-
tions. However, the spectra did show a minimum at ⬍200
nm, which is indicative of a random coil (12 ), i.e., an
unordered, possibly extended-like tertiary structure. The
CD results showing no helix implies the absence of
coiled-coils because such a quaternary structure requires
association of slightly left-hand–twisted helices of 3.5
residues per turn.
The solution structure of NT-proBNP inferred from the
respective results of the 3 experimental techniques is
summarized in Table 1. The experimental techniques so
chosen allow for solution structural assignment of the
peptide. Collectively, these data convincingly indicate
that NT-proBNP is not a coiled-coil trimer and in fact is a
monomer. This is a consequence of essentially no helix as
assessed by CD, which implies no coiled-coil and therefore no quaternary association, i.e., oligomerization, of
individual molecules. Finally, the ultracentrifuge data
conclusively show that at moderate concentrations and in
a benign medium, synthetic human NT-proBNP is monomeric.
It is not intuitively obvious how to reconcile the results
from this study and previous work (1, 7, 13 ) regarding the
oligomeric nature of human NT-proBNP. In the earlier
work, the sample was prepared by hydrophobic solidphase extraction and elution with organic solvent. It is
unclear how this procedure could affect either association
or disassociation of the analyte. The extractant was then
chromatographed by SE-HPLC in benign buffer, and the
column eluate was measured by immunoreactivity. The
identified fraction was of “high molecular weight”, the
inference being oligomeric NT-proBNP. Another possibility involves a putative non-NT-proBNP binding component partner in serum, stable to SE-HPLC, that would
produce an immunoreactive high–molecular-weight complex that collapses to “normal-eluting” (1, 7, 12 ) NTproBNP after SE-HPLC run under denaturing conditions.
This would be expected because the putative non-NTproBNP– binding component partner is silent, i.e., unobservable, by immunodetection.
The actual solution structure of NT-proBNP must await
high-resolution nuclear magnetic resonance or x-ray studies. One can speculate from the data presented here,
however, that synthetic NT-proBNP is likely an unordered random coil with an extended-like structure. Whatever the case, human synthetic NT-proBNP is a monomer,
and the potential confounding issue of analyte oligomerization is not a problem for this analyte.
Table 1. Structural assignment of synthetic NT-proBNP from results of the study.
Structural assignmenta
Experimental technique
2° structure
3° structure
4° structure
Coiled-coil
SE-HPLC
Sedimentation equilibrium ultracentrifugation
CD
NAb
NA
Random coil, no helix
Extended, nonglobular
NA
NA
NA
Monomer, not trimer
NA
NA
No
No
2° structure refers to helix, ␤-sheet, or random coil content; 3° structure refers to the overall three-dimensional shape of the molecule; and 4° structure refers to
the putative state of association of individual molecules.
b
NA, not available from experimental technique.
a
1038
Technical Briefs
I thank Vonnie Landt, Jitka Olander, and Jack Ladenson,
in whose laboratory this work was performed, for suggestions and critical reading of the manuscript. The Mass
Spectrometry Facility kindly provided instrument time
and is supported by NIH Grants P41-RR00954, P60DK20579, and P30-DK56341.
References
1. Goetze JP. Biochemistry of pro-B-type natriuretic peptide-derived peptides:
the endocrine heart revisited. Clin Chem 2004;50:1503–10.
2. Clerico A, Emdin M. Diagnostic accuracy and prognostic relevance of the
measurement of cardiac natriuretic peptides: a review. Clin Chem 2004;50:
33–50.
3. Hammerer-Lercher A, Ludwig W, Falkensammer G, Müller S, Neubauer E,
Puschendorf B, et al. Natriuretic peptides as markers of mild forms of left
ventricular dysfunction: effects of assays on diagnostic performance of
markers. Clin Chem 2004;50:1174 – 83.
4. Ala-Kopsala M, Magga J, Peuhkurinen K, Leipälä Ruskoaho H, Leppäluoto J,
et al. Molecular heterogeneity has major impact on the measurement of
circulating N-terminal fragments of A- and B-type natriuretic peptides. Clin
Chem 2004;50:1576 – 88.
5. Doust JA, Glasziou PP, Pietrazk E, Dobson AJ. A systematic review of the
diagnostic accuracy of natriuretic peptides for heart failure. Arch Intern Med
2004;164:1978 – 84.
6. Christenson RH, Duh SH, Apple FE, Bodor GS, Bunk DM, Dalluge J, et al.
Standardization of cardiac troponin I assays: round robin of ten candidate
reference materials. Clin Chem 2001;47:431–7.
7. Seilder T, Pemberton C, Yandle T, Espiner E, Nicholls G, Richards M. The
amino terminal regions of proBNP and proANP oligomerize through leucine
zipper-like coiled-coil motifs. Biochem Biophys Res Commun 1999;255:
495–501.
8. Dieckgraefe BD, Crimmins DL, Landt V, Houchen S, Anant R, Porche-Sorbet
R, et al. Expression of the regenerating gene family in inflammatory bowel
disease mucosa: Reg I␣ upregulation, processing, and antiapoptotic activity.
J Invest Med 2002;50:421–34.
9. Gooding KM, Regnier FE. Size exclusion chromatography. In: Gooding KM,
Regnier FE, eds. HPLC of biological macromolecules. New York: Marcel
Dekker, 1990:47–75.
10. Fish WW, Mann KG, Tanford C. The estimation of polypeptide chain
molecular weights by gel filtration in 6 M guanidine hydrochloride. J Biol
Chem 1969;244:4989 –94.
11. Lau SYM, Taneja AK, Hodges RS. Synthesis of a model protein of defined
secondary and quaternary structure: effect of chain length on the stabilization and formation of two-stranded ␣-helical coiled-coils. J Biol Chem
1984;259:13253– 61.
12. Greenfield N, Fasman GD. Computed circular dichroism spectra for the
evaluation of protein conformation. Biochemistry 1969;8:4108 –16.
13. Shimizu H, Masuta K, Asada H, Sugita K, Sairenji T. Characterization of
molecular forms of probrain natriuretic peptide in human plasma. Clin Chim
Acta 2003;334:233–9.
DOI: 10.1373/clinchem.2004.047324
Evaluation of a New CA15-3 Protein Assay Method:
Optical Protein-Chip System for Clinical Application,
Hong-Gang Zhang,1* Cai Qi,2 Zhan-Hui Wang,2 Gang Jin,2
and Rui-Juan Xiu1 (1 Institute of Microcirculation, Peking
Union Medical College & Chinese Academy of Medical
Science, Beijing, Peoples Republic of China; 2 Institute of
Mechanics, Chinese Academy of Sciences, Beijing, China;
* address correspondence to this author at: Institute of
Microcirculation, Peking Union Medical College & Chinese Academy of Medical Science, 5 DongDanSanTiao,
Beijing 100005, China; e-mail Zhanghg1966126@yahoo.
com.cn)
Carbohydrate antigen 15-3 (CA15-3) is frequently measured as a breast cancer marker test. Here we describe a
novel type of optical biosensor system, the optical protein
chip (OPC), to detect CA15-3 in serum.
The complex formed by interaction between an antibody molecule and its corresponding antigen can be
detected on a silicon substrate by an optical sensor, as
described in previous reports (1, 2 ). For processing and
modification of the silicon substrate surface, silicon wafers were cut into ⬃2 ⫻ 0.7 cm rectangles and made
hydrophilic by immersion in an acidic peroxide solution
(300 mL/L H2O2–980 mL/L H2SO4; 1:3 by volume) and
light shaking in a shaker for 30 min. The solution not only
removed contaminants from the silicon surface but also
increased the number of silanol groups on the surface.
The hydrophilic surfaces were rinsed in distilled water 3
times and in absolute ethanol 3 times, then incubated in a
mixture of 3-aminopropyltriethoxysilane and ethanol
(1:15 by volume) and shaken lightly in a shaker for 2 h.
The liquid was then removed, and the silicon wafers were
rinsed in absolute ethanol 3 times and in phosphatebuffered saline (PBS) buffer 3 times. The wafers were then
placed in a mixture of glutaraldehyde and PBS (1:10 by
volume), shaken lightly in a shaker for 1 h, and finally,
washed in PBS buffer 3 times and left in a beaker with PBS
buffer until use. Through the reaction of glutaraldehyde
with 3-aminopropyltriethoxysilane, Fc regions of the antibody molecules were covalently immobilized on the
chip surfaces.
Protein chip preparation and detection included the
following steps: (a) CA15-3-specific monoclonal antibody
(Biodezign) was concentrated to 0.1 g/L, and then 20 ␮L
of CA15-3 solution was delivered individually to each
analytical spot on the chip by a microfluidics system
(MFS) at a flow rate of 2 ␮L/min for 10 min. (b) After the
entire volume of solution flowed onto each analytical spot
on the silicon surface, 40 ␮L of distilled water was
delivered individually to each spot on the chip by the
MFS at a flow rate of 8 ␮L/min for 5 min to remove all
nonadsorbed CA15-3 monoclonal antibody molecules on
the analytical spot surface. (c) After the entire volume of
distilled water flowed onto the analytical spots, 20 ␮L of
a 1 g/L bovine serum albumin solution was delivered in
the same way at a flow rate of 2 ␮L/min for 10 min to
block nonspecific binding. (d) The chip was rinsed with 50
␮L of distilled water in the same way at a flow rate of 10
␮L/min for 5 min. (e) Serum samples were diluted with
equal volumes of Tween 20 (20 mL/L) to a final volume of
50 ␮L, then the diluted samples were delivered individually to each analytical spot on the chip by the MFS at a
flow rate of 2 ␮L/min for 25 min until the entire serum
solution had flowed onto the analytical areas. (f) The chip
was rinsed with 100 ␮L of distilled water in the same way
at a flow rate of 20 ␮L/min for more than 5 min. (g) The
chip was removed from the MFS and dried under a
stream of nitrogen. The thicknesses of layers in the
analytical areas were measured with biosensor imaging
ellipsometry, which produced an ellipsometric image of a
surface of each chip with a lateral resolution of 2 ␮m. The
biosensor system used here was developed to visualize
antigen–antibody binding on the surface, as described in
the literature (3 ).
The OPC detection procedure was performed at least
twice for each sample. Quantitative analysis was performed with use of a calibration curve, which was constructed with a serum sample with a known concentration
of CA15-3 that had been determined by an electrochemiluminescence immunoassay (ECLIA). CA15-3 was undetectable in 30 serum samples from healthy blood donors.
ROC plot analysis (4 ) was used to assess the accuracy of
the OPC test and to compare it with ECLIA detection in 60
serum samples from patients.
The CA15-3 image format determined by the OPC test
is shown in Fig. 1. The calibration curve was approximated by the equation: y ⫽ 1 ⫺ e⫺x, which was usable
up to ⬃ 20 kIU/L. Test samples need to be diluted when
their concentrations are ⱖ20 kIU/L (Fig. 1 in the Data
Supplement that accompanies the online version of
this article at http://www.clinchem.org/content/vol51/
issue6/). The within-run imprecision (CV) values were
5.2%, 2.5%, and 4.6% at 5, 10, and 18 kIU/L, respectively
(n ⫽ 10), and the interassay CVs were 7.5%, 3.8%, and
6.3%. The lower limit of detection was 1 kIU/L at a
signal-to-noise ratio of 3. The limit of quantification,
defined as the lowest amount detectable with imprecision
(CV) ⬍20% (n ⫽ 10), was 4 kIU/L.
Because CA15-3 is most useful for monitoring advanced breast cancer (5, 6 ), we collected 60 serum samples from women with breast cancer and other breast
diseases for a preliminary clinical study of our test. The
median patient age was 48.5 years (range, 22–75 years).
Study patients included 24 women with intraductal carcinoma, 15 women with mucinous carcinoma, 5 women
with in situ lobular carcinoma, 2 women with medullary
carcinoma, and 14 women with breast diseases but no
evidence of cancer. We also collected 30 serum samples
from healthy blood donors. Serum was separated from
the blood cells and stored at ⫺70 °C until analysis. OPC
tests were performed with an optical biosensor system;
this immunosensor system is based on imaging ellipsometry developed at the Institute of Mechanics, China Acad-
1039
emy of Sciences. For comparison, we measured CA 15-3
by an ECLIA on an Elecsys 2010 system (Roche Diagnostics). Both tests were done without knowledge of the
clinical status of the patients or knowledge of the results
of the other test. The results obtained by ECLIA detection
(kIU/L) and the OPC method (kIU/L) were compared by
use of Bland–Altman plots with Analyze-it Software
(General⫹Clinical Laboratory statistics, Ver. 1.71; Fig. 2 in
the online Data Supplement). The areas under the ROC
curves for differentiating women with breast cancer from
healthy women and women with other breast diseases
were 0.807 (95% confidence interval, 0.695– 0.919) for the
OPC test and 0.882 (95% confidence interval, 0.776 – 0.998)
for the ECLIA test (Fig. 3 in the online Data Supplement).
Compared with the Biacore system, a fairly widely
applied optical detection method based on surface plasmon resonance, the OPC technology used in this study
also allows label-free samples and crude samples to be
used directly without previous purification. Both technologies are based on the optical sensor principle, but OPC is
a direct optical visualization method based on imaging
ellipsometry that offers biomolecular layer visualization
with a distinct graph and qualitative and quantitative
result analysis. Compared with the Biacore method, the
OPC technology has advantages such as (a) optical sampling without disturbance; (b) identification, detection,
and purification of biomolecules not only by antigen–
antibody interactions but also by receptor–ligand interactions; and (c) real-time detection and monitoring of biomolecular interactions between carbohydrates, proteins,
and nucleic acids. The OPC setup used in this study has
some unique advantages. The multibioprobe analysis for
1 analyte allows up to 24 bioprobes to be arrayed on a
chip at the same time, or multianalyte analysis for 1
bioprobe can be arrayed on a chip allowing up to 24
different samples at the same time. Compared with the
Biacore system, a disadvantage of the OPC system is that
it is not easy to use because of the complicated physical
requirements of the current system.
The power and flexibility of proteomic analysis techniques, which facilitate protein separation, identification,
Fig. 1. Results of OPC test based on imaging
ellipsometry for detection of CA15-3.
(Left), image in grayscale of anti-CA15-3 antibody
immobilized after reaction with patient serum containing CA15-3 antigen. (Right), image in 3 dimensions deduced from the data on the left according
to the principle that the intensity in the image is
proportional to the square of the thin layer thickness. 1a and 4c, spots containing anti-CA15-3 IgG
as a control; the mean thickness of the anti-CA15-3
IgG layer is 6.4 nm. 2a–3c, spots containing CA153/anti-CA15-3 complex formed by different samples
from patients with breast cancer.
1040
Technical Briefs
and characterization, should hasten our understanding of
processes at the protein level (7 ). The combination of
imaging ellipsometry and protein chip technology provides a new potential biosensor system for detection and
monitoring of biomolecular interaction events for the
fields of proteomics, clinical laboratory testing, and biomolecular interaction research.
This study was supported by the China Academy of
Sciences as a scientific program of the National Project of
China.
References
1. Jin G, Tengvall P, Lundström I, Arwin H. A biosensor concept based imaging
ellipsometry for visualization of biomolecular interactions. Anal Biochem
1995;232:69 –72.
2. Jin G, Jansson R, Arwin H. Imaging ellipsometry revisited: developments for
visulization of thin transparent layers on silicon substrates. Rev Sci Instrum
1996;67:2930 – 6.
3. Wang Z, Jin G. A label-free multisensing immunosensor based on imaging
ellipsometry. Anal Chem 2003;75:6119 –23.
4. Handley JA, McNeil BJ. The meaning and use of the area under a receiver
operating characteristic (ROC) curve. Radiology 1982;143:29 –36.
5. Kurebayashi J, Yamamoto Y, Tanaka K, Kohno N, Kurosumi M, Moriya T, et al.
Significance of serum carcinoembryonic antigen and CA15-3 in monitoring
advanced breast cancer patients treated with systemic therapy: a large-scale
retrospective study. Breast Cancer 2003;10:38 – 44.
6. Clinton SR, Beason KL, Johnson JT, Jackson M, Wilson C, Holifield K, et al. A
comparative study of four serological tumor markers for the detection of
breast cancer. Biomed Sci Instrum 2003;39:408 –14.
7. Zhang HG, Xiu RJ. Micro-vascular medicine and proteomics [Review]. Clin
Hemorheol Microcirc 2003;29:189 –92.
Detection of Mutated Angiotensin I-Converting Enzyme by Serum/Plasma Analysis Using a Pair of Monoclonal Antibodies, Sergei M. Danilov,1* Jaap Deinum,2 Irina
V. Balyasnikova,1 Zhu-Li Sun,1 Cornelis Kramers,2,3,4 Carla
E.M. Hollak,5 and Ronald F. Albrecht1 (1 Department of
Anesthesiology, University of Illinois at Chicago, Chicago, IL; Departments of 2 Medicine, 3 Pharmacology/
Toxicology, and 4 Internal Medicine, University Medical
Center, Nijmegen, The Netherlands; 5 Department of Hematology, Academic Medical Center, Amsterdam, The
Netherlands; * address correspondence to this author at:
Anesthesiology Research Center, University of Illinois at
Chicago, 1819 W. Polk St. (M/C 519), Chicago, IL 60612;
fax 312-996-9680, e-mail danilov@uic.edu)
Angiotensin I-converting enzyme (ACE; CD143) is a Zn2⫹
carboxydipeptidase that plays a key role in the regulation
of blood pressure and in the development of vascular
pathology and remodeling (1–3 ). ACE is constitutively
expressed on the surface of endothelial cells, macrophages, dendritic cells, and various other cell types (4, 5 ).
Somatic ACE contains two homologous domains, N- and
C-terminal, each with a catalytic center (2, 6 ). ACE has
been accepted as a CD marker, CD143 (4, 6 ).
Soluble serum ACE originates from endothelial cells by
proteolytic cleavage by an unidentified protease of the
Arg1203–Ser1204 peptide bond in the stalk region near the
C-terminal transmembrane sequence of the ACE molecule
(7–11 ). At physiologic conditions, the concentration of
ACE in blood is very stable (12 ), whereas the ACE
concentration in serum is often significantly increased in
granulomatous diseases (in particular, sarcoidosis) or
Gaucher disease (13–18 ).
We described a Pro1199Leu mutation, located in the
juxtamembrane stalk region of ACE (19, 20 ), that explained a considerable familial increase in blood ACE
activity in individuals from several Dutch families (19 ).
The same phenotype and autosomal-dominant inheritance pattern have been described in Japan (21 ) and Italy
(22 ). Despite the fact that patients with this mutation at
first scrutiny do not have clinical abnormalities (19 ), the
finding of increased ACE has led to confusion for treating
physicians (23, 24 ).
We recently observed reduced binding of soluble ACE
in Dutch patients with a Pro1199Leu substitution detected
by a new monoclonal antibody (mAb), 1B3, which recognizes a Pro1199-containing epitope in the C-terminal region
of soluble ACE (25 ). We therefore set out to develop a
method that would use mAb 1B3 in combination with
mAb 9B9 to the central part of the N-domain of ACE
(26 –28 ), which would enable us to distinguish persons
with the Pro1199Leu mutation from patients with increased ACE attributable to other diseases, such as sarcoidosis and Gaucher disease.
We used sera from 7 persons with an ⬃5-fold increase
(Fig. 1 in the Data Supplement that accompanies the
clinchem.org/content/vol51/issue6/) in blood ACE
attributable to the Pro1199Leu mutation, designated
“hyperACE” (19 ), and sera from 10 first-degree relatives
without the mutation. All individuals gave permission to
use their blood and samples. The Medical Ethical Committee of the University Medical Center in Nijmegen, The
Netherlands, approved the sampling protocol. As controls, we used sera and citrated plasmas from 32 members
of the Department of Anesthesiology, University of Illinois at Chicago (5 women and 27 men; age range, 30 –75
years). All were apparently healthy and not on medication.
We also obtained sera from 7 patients with active
sarcoidosis (4 women and 3 men, age range, 29 –54 years).
The serum samples had been kept at ⫺80 °C for 1– 6 years.
The diagnosis of sarcoidosis was based on clinical and
radiographic findings and was supported by a tissue
biopsy showing characteristic histologic features.
Sera from 17 patients with Gaucher disease (10 men and
7 women; age range, 30 – 62 years) had been stored at
⫺20 °C at the Department of Hematology, Academic
Medical Center (Amsterdam, The Netherlands) for 8 –13
years. These sera had been obtained just after the start of
treatment with enzyme supplementation (placental or
recombinant glucocerebrosidase). In all patients with
Gaucher disease, the diagnosis was confirmed by defi-
cient glucocerebrosidase activity in leukocytes (29 ) and by
genotyping.
For immunocapture studies, the following mAbs to
human ACE were used: mAb 1B3, which recognizes a
C-terminal part of soluble ACE (25 ), and mAbs 9B9
(26 –28 ) and 2B11, which recognize epitopes in the N- and
C-domains of ACE, respectively. ACE activity in human
serum or plasma was measured by a fluorometric assay
(30, 31 ).
For the immunocapture enzyme assay (ICEA), we
coated 96-well plates (Corning) with anti-ACE mAbs (5
mg/L) via a bridge of affinity-purified goat anti-mouse
IgG (26 ). We then incubated the wells with 50 ␮L of
diluted (1:10) serum/plasma and measured plate-bound
ACE activity by adding a substrate for ACE directly into
the wells (26 ).
Shown in Fig. 1 is the ACE activity captured from
plasma of affected individuals vs healthy controls by mAb
1B3 (directed to a C-terminal epitope; Fig. 1A), mAb 9B9
(directed to the central part of the N-domain of ACE;
Fig. 1B), and mAb 2B11 (directed to the central part of the
C-domain of ACE; Fig. 1C). Because of variations in ACE
concentrations in the tested patients, the absolute
amounts of ACE captured by mAb 1B3 were not visibly
lower in individuals with hyperACE compared with
healthy controls. However, hyperACE individuals could
be clearly separated from healthy controls and from
patients with increased plasma ACE by calculation of the
ratio of the amounts of ACE captured by mAbs 1B3 and
9B9 (or mAb 2B11). The 1B3/9B9 binding ratio was not
influenced by ACE concentration in individuals with
low-normal or high-normal ACE concentrations (for details, see the online Data Supplement). Dilution of samples also did not significantly affect the ratio (see the
online Data Supplement). The intra- and interassay CVs
for the 1B3/1B9 binding ratio were 4.3% and 5.6%,
respectively, in healthy controls and 5.4% and 7.4%,
respectively, in patients with high ACE activity (details
provided in the online Data Supplement).
To validate this assay for use in clinical practice, we
simultaneously determined the serum 1B3/9B9 binding
ratio of patients with sarcoidosis and Gaucher disease vs
that of healthy individuals and patients with the
Pro1199Leu mutation. Patients with sarcoidosis (n ⫽ 7) or
Gaucher disease (n ⫽ 17), and hyperACE patients (n ⫽ 5)
had 3- to 4-fold increased serum ACE activity compared
with healthy individuals. In the individuals with the
Pro1199Leu mutation, however, the 1B3/9B9 binding
ratio was 3-fold lower than in healthy individuals or
patients with sarcoidosis (Fig. 2 in the online Data Supplement). We observed no overlap of 1B3/9B9 binding
ratio between carriers of the Pro1199Leu mutation and
other patients. We should note that despite the fact that
the mean (SD) 1B3/9B9 binding ratio of patients with
Gaucher disease [0.435 (0.065)] was dramatically higher
(2.7-fold) than in carriers of the ACE mutation, the absolute value of the ratio was significantly lower than in
healthy individuals or patients with sarcoidosis.
The absolute value of the 1B3/9B9 binding ratio de-
1041
Fig. 1. Capture of ACE activity from plasma by mAbs to ACE (ICEA).
Plasma from the indicated patients was diluted with phosphate-buffered saline
(1:10 for patients with normal ACE activity and 1:50 for patients with high ACE
activity) and incubated in wells of microtiter plates coated with mAb 1B3 (A), 9B9
(B), or 2B11 (C) via goat anti-mouse IgG (25 ). Immunocaptured ACE activity was
quantified by spectrofluorometric assay with Hip-His-Leu as a substrate. Data are
the mean (SD; error bars) of triplicates. (D), ACE-binding ratios for patients with
normal (within the interval for the general population; F) and high ACE activity
(hyper; ), obtained with 3 different pairs of mAbs: 1B3/9B, 1B3/2B11, and
9B9/2B11.
pends to some extent on assay configuration (ICEA or
ELISA, duration of incubation of serum/plasma samples
with mAb-coated plate, source of bridge antibodies) and
ranged between 0.4 and 0.7, as described by Balyasnikova
et al. (25 ) and in the online Data Supplement.
The absolute value of the 1B3/9B9 binding ratio also
1042
Technical Briefs
depends on duration of storage. The ratio was somewhat
decreased in sera from patients with Gaucher disease that
had been stored for 8 –13 years at ⫺20 °C compared with
sera from sarcoidosis patients that had been stored for
shorter times. This suggests that the decrease in 1B3/9B9
binding ratio during long-term storage reflects proteolytic
cleavage of the C-terminal end of soluble ACE. Nevertheless, even with different storage times, the binding ratio
still allows differentiation of genetically increased ACE.
From the reduced binding of mAb 1B3 to the soluble
ACE from heterozygous individuals with the Pro1199Leu
mutation (25 ), we conclude that the epitope that is recognized by mAb 1B3 is either disrupted or severely altered
by the mutation. We do not know whether mAb 1B3 binds
at all to mutated ACE because at this time we do not have
pure, mutated ACE or sera from homozygous carriers.
The differential binding characteristics of mAbs 9B9
and 1B3 allowed us to develop an easily applied ELISA to
clearly differentiate individuals with increased ACE attributable to the Pro1199Leu mutation from persons with
increased ACE attributable to other causes. This ELISA is
described in detail in Fig. 3 of the online Data Supplement.
The issue of the impact of mutated ACE on the assay is
relevant for the following reasons. The occurrence of
increased ACE activity attributable to the Pro1199Leu
mutation (hyperACE) is fairly common; more than 30
apparently unrelated index patients are currently known
in The Netherlands, with one-half of their family members harboring the mutation (C. Kramers and J. Deinum,
personal observations). These persons have come to light
in almost all instances because their physicians ordered
ACE tests when the probands presented with nonspecific
complaints. In none of the patients could the diagnosis of
sarcoidosis (and other granulomatous disease) or Gaucher
disease be made. Often these patients had undergone
extensive diagnostic evaluation (23 ). The genetically determined increase in blood ACE may lead to incorrect
diagnosis of (neuro)sarcoidosis and unwarranted treatment with immunosuppressants (24 ). With the assays we
propose here, it is possible that, in the case of increased
ACE activity, hyperACE can be diagnosed straightforwardly, without need for further evaluation. If the 1B3/
9B9 ratio is within the value for a reference population,
further evaluation is necessary. We are not certain
whether the strategy we propose will apply to the situation elsewhere in the world, but the mutation has been
described recently in Germany (24 ) as well, and previous
reports from Italy and Japan (21, 22 ) suggest that hyperACE may occur worldwide.
For clinical practice, we propose that a sizeable increase
in ACE activity (more than 2-fold higher than the mean
ACE activity in the general population) should lead to a
request for 1B3/9B9 ICEA or ELISA testing (see the online
Data Supplement). For higher sensitivity, we would recommend diluting samples with high ACE activities to the
mean value of a control sample cohort.
In summary, we have developed an immunoassaybased strategy to detect the presence of mutated ACE in
plasma. The assay could be a valuable tool in the exploration of the differential diagnosis of increased ACE.
We are grateful to Drs. H.J.T. Ruven and J.C. Grutters
from the St. Antonius Hospital (Nieuwegein, The Netherlands), and to Dr. A. Groener from the Academic
Medical Center (Amsterdam, The Netherlands) for help
with collecting the patient sera. We thank Dr. R. Minshall
(University of Illinois at Chicago, Chicago, IL) for critical
reading of the manuscript.
References
1. Ehlers MRW, Riordan JF. Angiotensin-converting enzyme: new concepts
concerning its biological role. Biochemistry 1989;28:5311– 8.
2. Corvol P, Williams TA, Soubrier F. Dipeptidyl dipeptidase: angiotensinconverting enzyme. Methods Enzymol 1995;248:283–305.
3. Dzau VJ, Bernstein K, Celermajer D, Cohen J, Dahlof B, Deanfield J, et al.
The relevance of tissue angiotensin-converting enzyme: manifestations in
mechanistic and endpoint data. Am J Cardiol 2001;88 (Suppl):1L–20L.
4. Franke FE, Metzger R, Bohle R-M, Kerkman L, Alhenc-Gelas F, Danilov SM.
Angiotensin I-converting enzyme (CD 143) on endothelial cells in normal and
in pathological conditions. In: Kishimoto T, Kikutari H, van dem Borne AEG,
Goyert SM, Mason DY, Miyasaka M, et al., eds. Leucocyte typing VI: white
cell differentiation antigens. New York: Garland Publishing, 1997:749 –51.
5. Danilov SM, Sadovnikova E, Scharenborg N, Balysnikova IV, Svinareva DA,
Semikina EL, et al. Angiotensin-converting enzyme (CD143) is abundantly ed
by dendritic cells and discriminates human monocytes-derived dendritic
cells from acute myeloid leukemia-derived dendritic cells. Exp Hem 2003;
31:1301–9.
6. Danilov SM, Franke FE, Erdos EG. Angiotensin-converting enzyme (CD143).
In: Kishimoto T, Kikutari H, van dem Borne AEG, Goyert SM, Mason DY,
Miyasaka M, et al., eds. Leucocyte typing VI: white cell differentiation
antigens. New York: Garland Publishing, 1997:746 –9.
7. Ching SF, Hayes LW, Slakey LL. Angiotensin-converting enzyme in cultured
endothelial cells. Synthesis, degradation and transfer to culture medium.
Arteriosclerosis 1983;3:581– 8.
8. Hooper NM, Keen J, Pappin DJC, Turner AJ. Pig kidney angiotensin converting enzyme. Purification and characterization of amphipathic and hydrophilic
forms of the enzyme establishes C-terminal anchorage to the plasma
membrane. Biochem J 1987;247:85–93.
9. Wei L, Alhenc-Gelas F, Corvol P, Clauser E. The two homologous domains of
human angiotensin I-converting enzyme are both catalytically active. J Biol
Chem 1991;266:9002– 8.
10. Woodman ZL, Oppong SY, Cook S, Hooper NM, Schwager SL, Brandt WF, et
al. Shedding of somatic angiotensin-converting enzyme (ACE) is inefficient
compared with testis ACE despite cleavage at identical stalk sites. Biochem
J 2000;347:711– 8.
11. Hooper NM, Karran EH, Turner AJ. Membrane protein secretases. Biochem
J 1997;321:265–79.
12. Alhenc-Gelas F, Richard J, Courbon D, Warnet JM, Corvol P. Distribution of
plasma angiotensin I-converting enzyme levels in healthy men: relationship
to environmental and hormonal parameters. J Lab Clin Med 1991;117:
33–9.
13. Lieberman J. Elevation of serum angiotensin-converting enzyme level in
sarcoidosis. Am J Med 1975;59:365–72.
14. Lieberman J, Beutler E. Elevation of angiotensin-converting enzyme in
Gaucher’s disease. N Engl J Med 1976;294:1442– 4.
15. Silverstein E. Friedland J, Lyons HA, Gourin A. Elevation of angiotensinconverting enzyme in granulomatous lymph nodes and serum in sarcoidosis:
clinical and possible pathological significance. Ann NY Acad Sci 1976;278:
498 –513.
16. Ainslie GM, Benatar SR. Serum angiotensin converting enzyme in sarcoidosis: sensitivity and specificity in diagnosis: correlations with disease
activity, duration, extra-thoracic involvement, radiographic type and therapy.
Q J Med 1985;55:253–70.
17. Beneteau-Burnat B, Baudin B. Angiotensin-converting enzyme: clinical applications and laboratory investigation n serum and other biological fluids. Crit
Rev Clin Lab Sci 1991;28:337–56.
18. Beutler E, Kay A, Saven A, Garver P, Thurston D, Dawson A, et al. Enzyme
replacement therapy for Gaucher disease. Blood 1991;78:1183–9.
19. Kramers C, Danilov SM, Deinum J, Balyasnikova IV, Scharenborg N, Looman
M, et al. Point mutation in the stalk of angiotensin-converting enzyme
causes a dramatic increase in serum angiotensin-converting enzyme but no
cardiovascular disease. Circulation 2001;104:1236 – 40.
20. Eyries M, Michaud A, Deinum J, Agrapart M, Chomilier J, Kramers C, et al.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
Increased shedding of angiotensin-converting enzyme by a mutation identified in the stalk region. J Biol Chem 2001;276:5525–32.
Okabe T, Fujisawa M, Yotsumoto H, Watanabe J, Takaku F, Lanzillo JJ, et al.
Familial elevation of serum angiotensin-converting enzyme. Q J Med 1985;
55:55– 61.
Luisetti M, Martinetti M, Cuccia M, Dugoujon M, De Rose V, Peona V, et al.
Familial elevation of serum angiotensin-converting enzyme activity. Eur Resp
J 1990;3:441– 6.
Kramers C, Deinum J. Increased serum activity of angiotensin-converting
enzyme (ACE): indication of sarcoidosis? A Bayesian approach. Ned Tijdschr
Geneeskd 2003;147:473– 6.
Linneback M, Kesper K, Jeub M, Urbach H, Wullner U, Klockgether T, et al.
Hereditary elevation of angiotensin-converting enzyme suggesting neurosarcoidosis. Neurology 2003;61:1819 –20.
Balyasnikova IV, Metzger R, Sun Z-L, Franke FE, Berestetskaya YV, Chubb
AJ, et al. Monoclonal antibodies 1B3 and 5C8 as probes for monitoring the
nativity of the soluble angiotensin-converting enzyme. Hybridoma 2005;24:
14 –26.
Danilov S, Jaspard E, Churakova T, Towbin H, Savoie F, Wei L, et al.
Structure-function analysis of angiotensin I-converting enzyme using monoclonal antibodies. J Biol Chem 1994;269:26806 –14.
Kost OA, Balyasnikova IV, Chemodanova EE, Nikolskaya II, Albrecht RFII,
Danilov SM. Epitope-dependent blocking of the angiotensin-converting enzyme dimerization by monoclonal antibodies to the N-terminal domain of
ACE: possible link of ACE dimerization and shedding from the cell surface.
Biochemistry 2003;42:6965–76.
Balyasnikova IV, Woodman ZL, Albrecht RFII, Natesh R, Acharya KR,
Sturrock ED, et al. The localization of an N-domain region of angiotensinconverting enzyme involved in the regulation of ectodomain shedding using
monoclonal antibodies. J Proteome Res 2005;4:258 – 67.
Daniels LB, Glew RH. ␤-Glucosidase assays in the diagnosis of Gaucher’s
disease. Clin Chem 1982;28(Pt 1):569 –77.
Pilliquod Y, Reinharz A, Roth M. Studies on the angiotensin-converting
enzyme with different substrates. Biochim Biophys Acta 1970;206:136 –
42.
Friedland J, Silverstein E. A sensitive fluorometric assay for serum angiotensin-converting enzyme. Am J Pathol 1976;66:416 –24.
Danilov SM, Savoie F, Lenoir B, Jeunemaitre X, Azizi M, Tarnow L, et al.
Development of enzyme-linked immunoassays for human angiotensin I-converting enzyme suitable for large-scale studies. J Hypertens 1996;14:719 –
27.
DOI: 10.1373/clinchem.2004.045633
Combination of His-Tagged T4 Endonuclease VII with
Microplate Array Diagonal Gel Electrophoresis for
High-Throughput Mutation Scanning, Matt J. Smith,1
Gabriella Pante-de-Sousa,1,2 Khalid K. Alharbi,1 Xiao-he Chen,1
Ian N.M. Day,1* and Keith R. Fox3 [1 Human Genetics
Division, School of Medicine, Southampton University
Hospital, Southampton, UK; 2 Department of Physiology,
Federal University of Para, Belem-Para, Brazil; 3 School of
Biological Sciences, University of Southampton, Southampton, UK; * address correspondence to this author
at: Human Genetics Division, Duthie Building (Mp808),
School of Medicine, Southampton University Hospital,
Tremona Road, Southampton SO16 6YD, UK; fax 44-(0)2380794264, e-mail inmd@soton.ac.uk]
Various physical mutation-scanning methods have been
developed to avoid unnecessary resequencing of long
stretches of DNA (1– 6 ). Protein-based mutation-scanning
techniques include enzymatic digestion [reviewed in Ref.
(7 )], protein binding to a DNA duplex, and direct analyses of the in vivo or in vitro gene product. One such
enzyme is T4 endonuclease VII (endoVII), the product of
1043
gene 49 of bacteriophage T4 (8 ). Radiolabel replacement
with fluorescent tags has facilitated automated analysis
(9 ). EndoVII recognizes heteroduplex structural distortions, nicking 2– 6 bp 3⬘ to the distortion, with efficiency
dependent on sequence context (10 ) and mismatch type
(11 ). Perfectly matched DNA undergoes some background digestion, which produces a highly reproducible
pattern (12 ). Mutation detection sensitivity obtained with
endoVII digestion was found to be similar to that for
denaturing HPLC and direct sequencing (13 ).
Microplate array diagonal gel electrophoresis
(MADGE) (14 ) provides an open-faced 96-well gel format
for polyacrylamide gels. Recently, nondenaturing 192-,
384-, and 768-well formats of MADGE for high-throughput checking of PCR and post-PCR reactions (15 ) have
been developed. We have combined, in proof-of-principle
experiments, the mismatch digestion properties of endoVII with the high-throughput capabilities of MADGE and
a newly developed denaturing MADGE format to create a
simple mutation-scanning technique that can screen
⬃1000 PCR samples during a single 35-min electrophoretic run.
Plasmid pRB210 (T4 endonuclease VII in pET11a) was a
kind gift from Professor B. Kemper (Institute for Genetics,
University of Cologne, Germany). The PCR primers used
to amplify the endoVII gene from pRB210 were as follows:
forward, 5⬘-GCGCCATATGATGTTATTGAC-3⬘; reverse,
5⬘-CAGCGGATCCTCATTTTAAACT-3⬘. After trimming
was performed with BamHI and NdeI (New England
Biolabs), pETendoVII was generated by ligation into
pET15b (Novagen). Expressed N-terminal His-tagged endoVII was then purified by affinity chromatography.
We used a single colony from pETendoVII-transfected
BL21 (DE3) Gold cells (Stratagene) to inoculate a 1-L Luria
broth culture containing 100 ␮g/L carbenicillin. After
overnight culture at 30 °C, an identical fresh 500-mL
passage was made, and at mid-log phase of growth
(absorbance at 600 nm, 0.6 – 0.8), protein expression was
induced by 1 mmol/L isopropyl-␤-d-thiogalactopyranoside. Cells were harvested after 2 h by centrifugation at
5000g for 10 min, and then lysed by sonication (10 cycles
of 30 s on and 30 s off at a probe amplitude of 10 –15 ␮m
in a MSE Soniprep 150). Cell debris and intact cells were
removed by centrifugation at 10 000g for 40 min. All steps
were carried out at 4 °C. The cell lysate was passed
through a Schleicher & Schuell 0.2 ␮m single-use filter.
EndoVII was purified by use of 1-mL HiTrap columns
in conjunction with the ⌬KTATM FPLCTM chromatography system (Amersham Bioscience), according to the
manufacturer’s instructions. Protein purity was assessed
by sodium dodecyl sulfate gel electrophoresis (Fig. 1 in
the Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/
content/vol51/issue6/), enzyme activity was confirmed
(without His-tag removal) with digests of synthetic heteroduplex substrates (data not shown), and protein quantification was by Bradford assay. Storage was in 50
mmol/L Tris-HCl (pH 8) with 1 mmol/L dithiothreitol
and 500 mL/L glycerol at ⫺80 °C.
1044
Technical Briefs
All primers were from MWG-Biotech. Exon 3 from
wild-type LDLR (GenBank accession no. Nm_000527) was
PCR-amplified using primers LDLR-F (5⬘-GCCTCAGTGGGTCTTTCCTT-3⬘) and LDLR-R (5⬘-CCAGGACTCAGATAGGCTCAA-3⬘), respectively, with 6-carboxyfluorescein (FAM) and hexachloro-6-carboxyfluorescein
(HEX) 5⬘ end labels for probe generation or without end
labels for generating amplicon from genomic DNAs for
testing. Jumpstart Taq polymerase (Sigma-Aldrich) was
used to ensure the highest quality probe generation, but
the thermal and ionic conditions for probe and test sample
amplifications were otherwise identical and essentially as
given by Whittall et al. (16 ). Probe PCR parallel reactions
from microplate wells were pooled and purified with
Wizard PCR prep reagents (Promega). The same 220-bp
PCR amplicon of LDLR exon 3 was generated (with
unlabeled primers) from samples from 330 unrelated
familial hypercholesterolemic individuals previously mutation scanned by single-strand conformation polymorphism (SSCP) analysis (16 ) and by meltMADGE (17 ). Six
previously defined heterozygotes, c.259T⬎G (p.W66G),
c.266G⬎A (p.C68Y), c.269A⬎G (p.D69G), c.301G⬎A
(p.E80K), c.301delG (p.E80fs), and c.313 ⫹ 1G⬎A (splice
site), were examined. Initially samples known to contain 1
of the 6 mutations were used to test endoVII digestion and
were analyzed by capillary electrophoresis on an ABI-310
instrument. Subsequently, 330 amplicons were screened
blind with denaturing MADGE (below) as the analytical
platform. All protocols were developed by M.J. Smith and
were validated by independent use by G. Pante-de-Sousa
and X. Chen.
To form the heteroduplexes, we mixed 2.5 ␮L of purified fluorescently labeled probe (representing an equivalent volume of original PCR) and 5.5 ␮L of unpurified test
PCR amplicon, heated the mixture to 95 °C, and allowed
it to cool to reform duplex DNA. For endoVII digestion,
we used a 10-␮L reaction volume containing 8 ␮L of
probe/test mixture and 2 ␮L of 5⫻ endoVII reaction
mixture [250 mmol/L K2HPO4 (pH 6.5), 25 mmol/L
MgCl2, 5 mmol/L dithiothreitol, and 0.1 g/L endoVII].
Phosphate ions have been shown to improve the efficiency of endoVII (18 ). Digestions were for 20 min at
37 °C.
EndoVII reaction mixture (2.5 ␮L) was mixed with 12
␮L of deionized formamide, denatured at 95 °C for 5 min,
and then chilled on ice before capillary electrophoresis
(Applied Biosystems 310 Genetic Analyzer).
For endoVII-MADGE, the reaction was terminated by
addition of 3 ␮L of loading dye (10 mmol/L NaOH, 50
mmol/L EDTA, 800 mL/L formamide, 2.5 g/L bromphenol blue, and 2.5 g/L xylene cyanole FF). Samples were
denatured by heating at 95 °C for 5 min and placed on ice
until gel loading.
EndoVII digestion fragments were resolved on a 10%
polyacrylamide denaturing MADGE gel containing 7
mol/L urea and 1⫻ Tris-borate-EDTA buffer [90 mmol/L
Tris-HCl (pH 8.3), 90 mmol/L boric acid, 2 mmol/L
EDTA]. After sample loading, the gel was covered by a
second glass plate. This plate– gel–plate sandwich was
secured by rubber bands, and silicon rubber tubing was
inserted along the long edge of the sandwich to prevent
electrophoretic edge artifacts. The assembly was placed in
a purpose-built 2-L gel tank (19 ) (with capacity for 10
gels) containing 1⫻ Tris-borate-EDTA buffer at 65 °C for
electrophoresis at 10 V/cm for 35min. EndoVII-MADGE
gels were scanned and analyzed with either a FluorImagerTM 595 or a Typhoon Trio⫹ (Molecular Dynamics,
Amersham Biosciences) and ImageQuant fragment analysis software (Molecular Dynamics).
LDLR mutations c.259T⬎G, c.301delG, c.301G⬎A, and
c.313 ⫹ 1G⬎A generated a strong digest fragment for at
least 1 probe strand, whereas c.266G⬎A generated a
lower yield of digest fragment on 1 of the probe strands.
c.269A⬎G displayed cleavage of the A䡠C heteroduplex
when the label was on the C strand (mutant as probe). A
typical example of the digestion pattern of the LDLR
mutants can be seen in Fig. 2 of the online Data Supplement. The extra peaks observed corresponded to expected
digest fragment sizes. These same products were trialed
under various conditions in denaturing MADGE gels
followed by fluoroimaging: the protocol described above
was efficient.
A typical 96-well endoVII-MADGE gel from blind
scanning of 330 familial hypercholesterolemic individuals
is shown in Fig. 1 (also shown, with dual label, in Fig. 3 of
the online Data Supplement). Previous mutation scanning
of this sample set had identified 47 heterozygous individuals with 1 of the 6 mutations: c.259T⬎G, c.266G⬎A,
c.269A⬎G, c.301G⬎A, c.301delG, or c.313 ⫹ 1G⬎A (Table
1). When we used only wild-type probe, endoVIIMADGE identified 51 samples containing additional digest fragments; 46 of these corresponded to the previously
identified mutations covering 5 of the 6 known LDLR
mutations (c.259T⬎G, c.266G⬎A, c.301G⬎A, c.301delG,
and c.313 ⫹ 1G⬎A). The c.269A⬎G mutation remained
undetected (see above). Of the 5 additional samples, 3
displayed digestion patterns matching those for positively
identified known LDLR mutations: 1 with the pattern for
c.259T⬎G and 2 with the pattern for c.301G⬎A. The
remaining 2 samples displayed unique digest patterns
that did not correspond to digest patterns for the 5 known
mutations (Fig. 4A in the online Data Supplement). One
digest pattern was similar to that for c.313 ⫹ 1G⬎A, but
c.313 ⫹ 1G⬎A was characterized by a strong digestion
fragment, whereas the unidentified mutation produced a
significantly weaker fragment (Fig. 4B in the online Data
Supplement). Sequencing showed that the sample was
heterozygous for the base change c.311G⬎T. The second
sample produced a digestion fragment close to the undigested amplicon. Sequencing showed a 2-base deletion,
c.196_197delGT. c.311G⬎T has been reported previously
(www.ucl.ac.uk/fh/genebook.html), whereas c.196_
197delGT appears to be a novel mutation. Of the 7
mutations detected, 2 displayed detectable mismatchspecific digestion patterns in both the sense and antisense
strands, c.196_197delGT and c.259T⬎G, whereas the remainder were identified by digestion of 1 strand.
This study suggests the feasibility of combining the
1045
that the reduced resolution and increased relative background associated with short-track electrophoresis did
not decrease the rate of mutation detection. EndoVIIMADGE also identified 2 previously unrecognized mutations in the sample set. EndoVII-MADGE consistently
compared favorably with SSCP analysis of the same
region (Table 1) in many heterozygotes, detecting 7 of 8
different sequence variations (8 of 8 when test samples
were end labeled). This approach could potentially add to
strategies for the investigation of unknown mutations at
the population level.
M.J. Smith was the recipient of a University of Southampton Faculty of Health Medicine and Life Science crossschool PhD studentship. We thank Professor Borries Kemper for clone pRB210. This work was also supported by
the UK Department of Health, National Genetics Reference Laboratory (Wessex), and HOPE.
References
Fig. 1. EndoVII-MADGE analysis.
Shown is a typical endoVII-MADGE gel image for the LDLR exon 3 amplicon. The
8 ⫻ 12 array set at a 71.6-degree angle allows tracks to pass through 2
successive rows, allowing 96 samples to be run on a single gel. Samples
containing mutations (tracks indicated by arrows) were identified by the presence
of an extra band or by an increase in intensity of a background band. In this
example, the 5⬘ fluorescent label was on the antisense strand.
mismatch digestion properties of endoVII with the highthroughput capabilities of MADGE to create a simple
high-throughput mutation-scanning method. We found
Table 1. Number of separate cases detected for a set of
mutations distributed through LDLR exon 3.
No. of cases detected
Mutation
c.313 ⫹ 1G⬎A
c.311G⬎T
c.301G⬎A
c.301delG
c.269A⬎G
c.266G⬎A
c.259T⬎G
c.196_197delGT
Wild type
Amino acid
change
SSCP
analysis
EndoVII-MADGE with
wild-type probe
Splice site
p.C83F
p.E80K
p.E80fs
p.D69G
p.C68Y
p.W66G
p.V45fs
21
0
17
3
1
2
3
0
283
21
1
19
3
0
2
4
1
279
1. Orita M, Iwahana H, Kanazawa H, Hayashi K, Sekiya T. Detection of
polymorphisms of human DNA by gel electrophoresis as single-strand
conformation polymorphisms. Proc Natl Acad Sci U S A 1989;86:2766 –70.
2. Fischer SG, Lerman LS. Length-independent separation of DNA restriction
fragments in two-dimensional gel electrophoresis. Cell 1979;16:191–200.
3. Underhill PA, Jin L, Lin AA, Mehdi SQ, Jenkins T, Vollrath D, et al. Detection
of numerous Y chromosome biallelic polymorphisms by denaturing highperformance liquid chromatography. Genome Res 1997;7:996 –1005.
4. Cotton RG, Rodrigues NR, Campbell RD. Reactivity of cytosine and thymine
in single-base-pair mismatches with hydroxylamine and osmium tetroxide
and its application to the study of mutations. Proc Natl Acad Sci U S A
1988;85:4397– 401.
5. Li Q, Liu Z, Monroe H, Culiat CT. Integrated platform for detection of DNA
sequence variants using capillary array electrophoresis. Electrophoresis
2002;23:1499 –511.
6. Reed GH, Wittwer CT. Sensitivity and specificity of single-nucleotide polymorphism scanning by high-resolution melting analysis. Clin Chem 2004;
50:1748 –54.
7. Taylor GR. Enzymatic and chemical cleavage methods. Electrophoresis
1999;20:1125–30.
8. Youil R, Kemper BW, Cotton RG. Screening for mutations by enzyme
mismatch cleavage with T4 endonuclease VII. Proc Natl Acad Sci U S A
1995;92:87–91.
9. Del Tito BJ, Poff HE 3rd, Novotny MA, Cartledge DM, Walker RI, Earl CD, et
al. Automated fluorescent analysis procedure for enzymatic mutation detection. Clin Chem 1998;44:731–9.
10. Pottmeyer S, Kemper B. T4 endonuclease VII resolves cruciform DNA with
nick and counter-nick and its activity is directed by local nucleotide
sequence. J Mol Biol 1992;223:607–15.
11. Solaro PC, Birkenkamp K, Pfeiffer P, Kemper B. Endonuclease VII of phage
T4 triggers mismatch correction in vitro. J Mol Biol 1993;230:868 –77.
12. Youil R, Kemper B, Cotton RG. Detection of 81 of 81 known mouse ␤-globin
promoter mutations with T4 endonuclease VII—the EMC method. Genomics
1996;32:431–5.
13. Andrulis IL, Anton-Culver H, Beck J, Bove B, Boyd J, Buys S, et al.
Comparison of DNA- and RNA-based methods for detection of truncating
BRCA1 mutations. Hum Mutat 2002;20:65–73.
14. Day IN, Humphries SE. Electrophoresis for genotyping: microtiter array
diagonal gel electrophoresis on horizontal polyacrylamide gels, hydrolink, or
agarose. Anal Biochem 1994;222:389 –95.
15. Gaunt TR, Hinks LJ, Rassoulian H, Day IN. Manual 768 or 384 well
microplate gel ‘dry’ electrophoresis for PCR checking and SNP genotyping.
Nucleic Acids Res 2003;31:e48.
16. Whittall R, Gudnason V, Weavind GP, Day LB, Humphries SE, Day IN. Utilities
for high throughput use of the single strand conformational polymorphism
method: screening of 791 patients with familial hypercholesterolaemia for
mutations in exon 3 of the low density lipoprotein receptor gene. J Med
Genet 1995;32:509 –15.
17. Day IN, Alharbi KK, Haddad I, Ye S, Lawlor DA, Whittal RA, et al. Mutation
scanning of LDLR in the whole population: severe, moderate and silent
mutations, paucimorphisms and cholesterol level. Eur J Hum Genet 2004;
12:302.
1046
Technical Briefs
18. Golz S, Greger B, Kemper B. Enzymatic mutation detection. Phosphate ions
increase incision efficiency of endonuclease VII at a variety of damage sites
in DNA. Mutat Res 1998;382:85–92.
19. Day IN, O’Dell SD, Cash ID, Humphries SE, Weavind GP. Electrophoresis for
genotyping: temporal thermal gradient gel electrophoresis for profiling of
oligonucleotide dissociation. Nucleic Acids Res 1995;23:2404 –12.
Carbohydrate-Deficient Transferrin Measured by Capillary Zone Electrophoresis and by Turbidimetric Immunoassay for Identification of Young Heavy Drinkers,
Jean-Bernard Daeppen,1* Frederic Anex,1 Bernard Favrat,2
Alvine Bissery,1 Joelle Leutwyler,1 Roland Gammeter,1 Patrice
Mangin,2 and Marc Augsburger2 (1 Alcohol Treatment Center, CHUV, Lausanne, Switzerland; 2 Institute of Forensic
Medicine, CHUV, Lausanne, Switzerland; * address correspondence to this author at: Alcohol Treatment Center,
Mont-Paisible 16, CHUV, 1011 Lausanne, Switzerland;
e-mail jean-bernard.daeppen@inst.hospvd.ch)
Carbohydrate-deficient transferrin (CDT) measured by
capillary zone electrophoresis (CZE), particularly asialotransferrin (Tf), is purported to better differentiate between excessive and moderate drinkers than does CDT
measured by turbidimetric immunoassay (TIA) (1, 2 ). The
use of biological markers such as CDT is of particular
interest for identifying young heavy drinkers because
other clinical signs of heavy drinking are generally absent
and heavy drinking is a leading cause of morbidity and
mortality in this age group (3, 4 ). Several authors have
shown interest in the ability of CDT to identify nondependent heavy drinkers (5, 6 ); we therefore describe here
the performance of CZE measurements of asialo- and
disialo-Tf and TIA analysis of CDT in a large community
sample of 19-year-old men, of whom 21% were heavy
drinkers.
From a sample of 1018 men attending a mandatory
1-day army recruitment process for all Swiss males at age
19 years, 1004 (98.6%) agreed to complete a research
questionnaire. Of these, 581 young men (57.9%) consented
to give blood for the measurement of asialo-Tf (CZE),
disialo-Tf (CZE), and CDT (TIA). The Ethics Committee of
the Lausanne University Medical School approved the
study protocol. Volunteers were compensated for participation in the study.
Volunters gave written informed consent and then
completed an instrument entitled “Health and Lifestyle
Questionnaire”, which included questions assessing the
typical quantity and frequency of alcohol consumption
during the 12 months preceding the survey and the
frequency of drunkenness over the last 30 days. One drink
was defined as a 250-mL can or bottle of beer, a 120-mL
glass of wine, or a 40-mL shot of liquor straight or in a
mixed drink, and corresponded to ⬃12 g of pure ethanol.
A study investigator was present during administration
of the questionnaire to verify that participants answered
all items. Serum samples were obtained by centrifugation
of peripheral blood collected in 10-mL tubes. Samples
were stored at ⫺20 °C before analysis.
Total CDT was measured by anion-exchange chromatography and TIA with the Axis-Shield CDT (TIA) reagent
set (7 ). To separate and measure Tf isoforms, we used a
previously described and validated CZE method (8, 9 )
with the Ceofix CDT reagent set (Analis) on a Hewlett
Packard (HP) 3D-CE instrument. The CZE conditions are
described in Table 1 of the Data Supplement that accompanies the online version of this Technical Brief at http://
www.clinchem.org/content/vol51/issue6/. CZE electropherograms showing the serum Tf profiles for a heavy
drinker before and after addition of anti-Tf polyclonal
antibody to the serum are shown in Fig. 1 of the online
Data Supplement, and CZE electropherograms showing
the Tf profiles of a teetotaler and of 2 heavy drinkers are
shown in Fig. 2 of the online Data Supplement.
Peaks representing the different Tf isoforms were quantified as the amounts of the asialo-, disialo-, trisialo-,
tetrasialo-, pentasialo-, and hexasialo-Tf (CZE) as a percentage of the total Tf content, in terms of valley-to-valley
areas under the curve. The intraday CV values (n ⫽ 6) for
“low” (0.6% by CZE) and “high” disialo-Tf (4.8% by CZE)
were 9.8% and 1.2%, respectively, and the interday CVs
(n ⫽ 5) for low (0.6% by CZE) and high disialo-Tf (4.8% by
CZE) were 11% and 2.3%, respectively. The intra- and
interday CVs for asialo-Tf (0.5% by CZE; n ⫽ 6) were 6.8%
and 11%, respectively. The limit of quantification of each
Tf (CZE) isoform was 0.1%, expressed a percentage of
total Tf isoforms.
Continuous data are reported as the mean (SD) and the
median (interquartile range). We used a ␹2 test to compare
categorical variables and Mann–Whitney U-tests to compare continuous variables because this nonparametric
statistic makes no assumption about the distributional
properties of variables. We also determined the areas
under the ROC curves (AUROC), the sensitivity, and the
specificity for disialo-Tf (measured by CZE) and CDT
(measured by TIA) in identifying heavy drinkers.
There were 121 (20.8%) heavy drinkers in the sample:
31 (5.3%) who reported typical alcohol consumption of
⬎21 drinks/week over the last 12 months, 52 (8.9%) who
said they had been drunk at least 3 times over the last
month; and 38 (6.5%) who reported both. Mean (SD)
alcohol consumption in heavy drinkers was 26.4 (8.4)
drinks (⬃300 g of ethanol) per week. Among the remaining participants, 435 (74.9%) were categorized as moderate drinkers, reporting, on average, 6.0 (4.7) drinks (⬃65 g
of ethanol) per week, and 25 (4.3%) were considered
abstinent (mean reported quantity and frequency ⫽ 0).
The abstaining participants were retained as part of the
moderate-drinker group.
Our results indicate that asialo-Tf (CZE) could not
differentiate between moderate and heavy drinkers because 574 (98.8%) of the participants had a asialo-Tf (CZE)
value of 0% and only 3 moderate drinkers and 4 heavy
drinkers had positive values. We did, however, find
significant differences between heavy and moderate
drinkers for disialo-Tf as measured by CZE {mean (SD),
0.8 (0.6)% [median (interquartile range), 0.7 (0.5– 0.9)%] vs
0.6 (0.2)% [0.6 (0.5– 0.8)%]; P ⬍0.01} and for CDT as measured by TIA {2.5 (0.8)% [2.3 (2.0 –2.8)%] vs 2.1 (0.5)%
[2.0 (1.8 –2.3)%]; P ⬍0.001}. The areas under the ROC
curves for disialo-Tf (CZE) and CDT (TIA) are shown in
Fig. 1. ROC curve analysis indicated low sensitivities and
specificities for disialo-Tf measured by CZE (AUROC ⫽
0.58) and for CDT measured by TIA (AUROC ⫽ 0.66) in
identifying heavy drinkers, with optimal cutoffs (inflection point on ROC curve) of 0.62% (sensitivity, 60.3%;
specificity, 52.1%) and 2.2% (sensitivity, 58.7%; specificity,
63.2%), respectively. As also shown in Fig. 1, the sensitivities and specificities at the usual cutoffs of 0.7% for
disialo-Tf measured by CZE (sensitivity, 47.1%; specificity, 64.4%) (1 ) and 2.6% for CDT measured by TIA
(sensitivity, 34.7%; specificity, 87.2%) (7 ) were not optimal.
We also explored the sensitivities and specificities of
disialo-Tf (CZE) and CDT (TIA) for differentiating the 121
heavy drinkers from the 25 abstainers after excluding the
435 moderate drinkers (not reported in Fig. 1). At the
optimal cutoff of 0.59%, the sensitivity and specificity of
disialo-Tf (CZE) in identifying heavy drinkers were 64.5%
and 68.0%, respectively (AUROC ⫽ 0.66); at the optimal
cutoff of 2.0% for CDT (TIA), the sensitivity and specificity of CDT (TIA) were 76.9% and 64.0% (AUROC ⫽ 0.75),
respectively.
Generalizing the relationship between alcohol use and
asialo-Tf (CZE), disialo-Tf (CZE), and CDT (TIA) concentrations to the total sample of 1004 men would be possible
only if the participants who refused to give blood had
alcohol use and alcohol-related problems that were similar to those who consented to give blood. Although not
reported in Fig. 1 or the supplemental data, the results
Fig. 1. Areas under the ROC curves for identification of heavy drinkers
by disialo-Tf measured by CZE (dashed line) and CDT measured by TIA
(solid line).
The dotted line represents the line of identity.
1047
also indicated that the 581 participants who agreed to give
blood were consuming significantly more alcohol [mean
(SD), 10.0 (12.61) vs 8.1 (10.39) drinks per week; P ⬍0.05]
and were more likely to report having been drunk at least
3 times over the last 30 days (15.5% vs 11.4%; P ⬍0.01)
than were the 423 participants who refused to give blood.
Our results indicate that asialo-Tf (as measured by
CZE) is of no diagnostic value for young men. These
results contrast with previous work suggesting that
asialo-Tf (CZE) was highly efficient in identifying patients
reporting a mean alcohol consumption ⬎50 g/day (AUROC ⫽ 0.91) (1 ). Two factors might help to explain the
differences observed between that earlier study and the
present one: In the earlier study, participants were older,
consumed more alcohol, and were more likely to have
altered hepatic function related to alcohol (1, 8 ). These
conditions apparently increase the sensitivity of CDT (10 ).
The other factor, as suggested by Legros et al. (8 ), was that
in the earlier study (1 ), the performance of the biological
markers was amplified because the participants were
compared after exclusion of those who were abstinent or
alcohol dependent and those who reported drinking
30 –50 g of ethanol/day. These findings suggest that the
superiority of asialo-Tf (CZE) over CDT (TIA) is evident
only in individuals who consume excessive amounts of
alcohol, such as very heavy drinkers, who are predominantly alcohol dependent.
Our results confirm that either heavy alcohol consumption or regular drunkenness significantly increases disialo-Tf (CZE) and CDT (TIA), but does not do so sufficiently to differentiate heavy from moderate drinkers.
Disialo-Tf (CZE) and CDT (TIA) had relatively similar
sensitivities and specificities for identifying young heavy
drinkers. According to previous findings (1, 8 ), the exclusion of moderate drinkers amplified the sensitivities and
specificities of disialo-Tf (CZE) and CDT (TIA) for correctly classifying heavy drinkers and teetotalers.
Our data also confirm the predictive limitations of
asialo-Tf (CZE) and disialo-Tf (CZE), as described earlier
for CDT (TIA), i.e., that the sensitivity of CDT is decreased
at younger ages (10, 11 ) and that CDT performs poorly for
the identification of heavy alcohol consumption in college
students (12, 13 ).
Although these results may broadly apply to young
men, it is important to recognize several limitations when
generalizing these findings to other populations. These
results may not hold true for samples of other individuals,
such as women, older persons, or those recruited within
medical settings. Our study sample consisted mostly of
Caucasians; thus, the findings may not apply to other
ethnic groups. The differences we observed in drinking
patterns between those who agreed to give blood and
those who refused preclude generalizing the findings to
the overall sample. Finally, although great effort was
made to optimize the accuracy of these data, the information obtained regarding alcohol use and alcohol-related
problems was based solely on the estimates and recollections of the participants.
1048
Technical Briefs
We are grateful to Magali Dovat for skillful technical
assistance with asialo-Tf (CZE) and disialo-Tf (CZE) measurements and to George Danko, PhD, for careful help in
the editing of the manuscript.
References
1. Legros FJ, Nuyens V, Baudoux M, Zouaoui Boudjeltia K, Ruelle JL, Colicis J,
et al. Use of capillary zone electrophoresis for differentiating excessive from
moderate alcohol consumption. Clin Chem 2003;49:440 –9.
2. Arndt T. Asialotransferrin—an alternative to carbohydrate-deficient transferrin? [Letter]. Clin Chem 2003;49:1022–3.
3. Pearson H. The demon drink. Nature 2004;428:598 – 600.
4. Room R, Graham K, Rehm J, Monteiro M. Drinking and its burden in a global
perspective: policy considerations and options. Eur Addiction Res 2003;9:
165–75.
5. Reynaud M, Schellenberg F, Loiseaux-Meunier MN, Schwan R, Maradeix B,
Planche F, et al. Objective diagnosis of alcohol abuse: compared values of
carbohydrate-deficient transferrin (CDT), ␥-glutamyl transferase (GGT), and
mean corpuscular volume (MCV). Alcohol Clin Exp Res 2000;24:1414 –9.
6. Aithal GP, Thornes H, Dwarakanath AD, Tanner AR. Measurement of
carbohydrate-deficient transferrin (CDT) in general medical clinic: is this test
useful in assessing alcohol consumption? Alcohol Alcohol 1998;33:304 –9.
7. Bio-Rad. %CDT TIA: instruction manual. Hercules, CA: Bio-Rad Laboratories,
2002.
8. Legros FJ, Nuyens V, Minet E, Emonts P, Zouaoui Boudjeltia K, Courbe A, et
al. Carbohydrate-deficient transferrin isoforms measured by capillary zone
electrophoresis for detection of alcohol abuse. Clin Chem 2002;48:2177–
86.
9. Lanz C, Marti U, Thormann W. Capillary zone electrophoresis with a dynamic
double coating for analysis of carbohydrate-deficient transferrin in human
serum. Precision performance and pattern recognition. J Chromatogr A
2003;1013:131– 47.
10. Arndt T. Carbohydrate-deficient transferrin in serum: a new marker of chronic
alcohol abuse: a critical review of pre-analysis, analysis and interpretation
[Review]. Clin Chem 2001;47:13–27.
11. Agelink NW, Dirkes-Kersting A, Zeit T, Bertling R, Malessa R, Klieser E.
Sensitivity of carbohydrate-deficient transferring (CDT) in relation to age and
duration of abstinence. Alcohol Clin Exp Res 1998;33:164 –7.
12. Yeastedt J, La Grange L, Anton RF. Female alcoholic outpatients and female
college students: a correlational study of self-reported alcohol consumption
and carbohydrate-deficient transferrin levels. J Stud Alcohol 1998;59:
555–9.
13. Nystrom M, Perasalo J, Salaspuro M. Carbohydrate-deficient transferrin
(CDT) in serum as a possible indicator of heavy drinking in young university
students. Alcohol Clin Exp Res 1992;16:93–7.
DOI: 10.1373/clinchem.2004.044461
Comparison of the Unsaturated Iron-Binding Capacity
with Transferrin Saturation as a Screening Test to Detect C282Y Homozygotes for Hemochromatosis in
101 168 Participants in the Hemochromatosis and Iron
Overload Screening (HEIRS) Study, Paul C. Adams,1*
David M. Reboussin,2 Cathie Leiendecker-Foster,3 Godfrey C.
Moses,4 Gordon D. McLaren,5 Christine E. McLaren,6 Fitzroy
W. Dawkins,7 Ishmael Kasvosve,7 Ron T. Acton,8 James C.
Barton,9 Dan Zaccaro,2 Emily L. Harris,10 Richard Press,11
Henry Chang,12 and John H. Eckfeldt3 (1 Department of
Medicine, London Health Sciences Center, London, Ontario, Canada; 2 Department of Public Health Sciences,
Wake Forest University School of Medicine, WinstonSalem, NC; 3 Department of Laboratory Medicine and
Pathology, University of Minnesota, Minneapolis, MN;
4
MDS Laboratories, Toronto, Ontario, Canada; 5 Division
of Hematology/Oncology, Department of Medicine, Uni-
versity of California, Irvine, CA, and Veterans Affairs
Long Beach Healthcare System, Long Beach, CA; 6 Epidemiology Division, Department of Medicine, University of
California, Irvine, CA; 7 Department of Medicine, Howard
University, Washington, DC; 8 Departments of Microbiology, Medicine, and Epidemiology and International
Health, University of Alabama at Birmingham, Birmingham, AL; 9 Southern Iron Disorders Center, Birmingham,
AL; 10 Kaiser Permanente Center for Health Research,
Portland, OR; 11 Department of Pathology, Oregon Health
& Science University, Portland, OR; 12 Division of Blood
Diseases and Resources, National Heart Lung and Blood
Institute, NIH, US Department of Health and Human
Services, Bethesda, MD; * address correspondence to this
author at: Department of Medicine, London Health Sciences Centre, 339 Windermere Rd., London, ON N6A
5A5, Canada; fax 519-858-5114, e-mail padams@uwo.ca)
The diagnosis of hemochromatosis was previously based
on a combination of clinical and laboratory assessments
that included history and physical examination, increased
transferrin saturation (TS) and serum ferritin, liver biopsy, the amount of iron removed by phlebotomy, and
pedigree studies identifying other family members with
iron overload (1 ). Since the discovery of the hemochromatosis gene (HFE) in 1996 (2 ), most studies from referral
centers have shown that ⬎90% of typical hemochromatosis patients are homozygous for the C282Y mutation of
the HFE gene (3 ). Before the availability of DNA-based
testing, it was assumed that most hemochromatosis patients have increased TS. However, recent population
screening studies incorporating HFE genotyping have
now shown that many C282Y homozygotes will have a
normal TS and may never develop clinical signs and
symptoms related to iron overload (4 – 8 ). TS has been
recommended in many studies as the most clinically
useful screening test for hemochromatosis because it is
widely available and may be increased even in young
adults with a genetic predisposition to hemochromatosis.
Another potential advantage over DNA-based testing as
an initial screening test is that TS may detect many types
of iron overload other than those associated with HFE
mutations. In addition, screening for iron overload instead of performing DNA-based testing may reduce the
risks of potential genetic discrimination that some authors
suggest is associated with identification of a C282Y homozygote with normal serum iron tests (9 –11 ). The TS is
a 2-step assay in which serum iron is the numerator and
the denominator is either total iron-binding capacity
(TIBC), [serum iron ⫹ unsaturated iron-bonding capacity
(UIBC)] or an adjusted serum transferrin. The UIBC is a
1-step automated colorimetric assay that has been reported to have similar or better operating characteristics
than TS for the detection of C282Y homozygotes (12–15 ).
In this study, UIBC is compared directly with the TS (as
measured by serum iron/serum iron ⫹ UIBC) for the
detection of C282Y homozygotes in a large primary care
population.
The study design and overall results of the Hemochro-
1049
Fig. 1. ROC curves comparing UIBC (dashed line) with TS (solid line) in undiagnosed male (A) and female (B) C282Y homozygotes.
(A), the AUC was significantly greater for UIBC (0.96) than TS (0.94; P ⬍0.001). (B), the AUC was significantly greater for UIBC (0.93) than TS (0.90; P ⬍0.001).
matosis and Iron Overload Screening (HEIRS) Study have
been reported previously (16, 17 ). Participants were recruited from 5 field centers that serve ethnically and
socioeconomically diverse populations. The study recruited all participants ⱖ25 years of age who gave informed consent, and it was approved by all local Institutional Review Boards. All participants had nonfasting
testing for serum UIBC, serum iron, and serum ferritin
and were genotyped for the C282Y and H63D mutations
of the HFE gene. All C282Y homozygotes were notified
and offered genetic counseling and advice on treatment
options. In this analysis, participants who reported a
previous diagnosis of hemochromatosis or iron overload,
whether reported to be treated or untreated, were excluded because phlebotomy therapy or other interventions potentially could affect the serum TS and UIBC.
ROC curves were generated for males and females to
compare TS and UIBC as tests for the detection of C282Y
homozygotes. Baseline cut points were determined from
the intersection of the sensitivity and specificity plots
from the ROC curves. Men and women were analyzed
separately.
TS was calculated from the ratio serum iron/(serum
iron ⫹ UIBC) and expressed as a percentage. Samples
from field centers located in the United States were tested
at the Fairview-University Medical Center at the University of Minnesota in Minneapolis, MN, and those from
Canada were tested at MDS Laboratory Services, Toronto,
Canada. Serum iron and UIBC were measured by a
ferrozine-based colorimetric assay on a Hitachi 911 (Fairview-University) or 917 (MDS) with reagents supplied by
Roche (Iron Prod #1970743 and UIBC Prod #1030600;
Roche Diagnostics Corp.). Internal quality-control pool
results from both laboratories are shown in Table 1 of the
Data Supplement that accompanies the online version
of this Technical Brief at http://www.clinchem.org/
content/vol51/issue6/. Method biases were assessed 3
times yearly by use of external proficiency testing samples
provided by the College of American Pathologists Surveys (Northfield, IL) and by use of blind replicate samples
that were collected from 2% of all participants and analyzed in both laboratories. In addition, comparisons between MDS Laboratory Services and the Central Laboratory were done before the start of testing, and 2% of the
MDS samples were repeated at the Central Laboratory
throughout the study.
Testing for HFE C282Y and H63D alleles was performed with DNA obtained from EDTA–whole-blood
samples by a modification of the Invader assay (Third
Wave Technologies) that increases the allele-specific fluorescent signal by including 12 cycles of locus-specific
PCR before the cleavase reaction (16 ).
ROC curves and 95% confidence intervals (CIs) were
generated with S-PLUS (Insightful Inc.). The nonparametric method of Delong for correlated ROC curves was used
to compare the curves for UIBC and TS (18 ).
The HEIRS study recruited 101 168 participants from
February 2001 through February 2003. A total of 1216
participants were excluded from this analysis because
they reported a previous diagnosis of hemochromatosis
or iron overload (including 97 C282Y homozygotes). In
addition, 47 participants had a missing UIBC. Among the
remaining 99 905 participants included in the analysis,
236 undiagnosed C282Y homozygotes were detected (91
1050
Technical Briefs
Table 1. Cut points for UIBC and TS.
Group/Test
Men
UIBC
TS
Women
UIBC
TS
Cut pointsa
C282Y/C282Y
homozygotes
detected, n
Sensitivity,
%
Specificity,
%
False positives,
n
C282Y/C282Y homozygotes
missed with an
increased ferritin,b n
PPV,c %
LRⴙ
⬍26 ␮mol/L
⬍24 ␮mol/L
⬍22 ␮mol/L
⬍19 ␮mol/L
ⱖ48%
ⱖ50%
ⱖ54%
ⱖ60%
82
80
76
71
75
75
74
72
90.1
87.9
83.5
78.0
82.4
82.4
81.3
79.1
90.0
92.5
95.0
97.6
90.5
92.5
95.3
97.5
3691
2763
1853
877
3521
2788
1736
915
6/9
6/11
10/15
14/20
10/16
10/16
11/17
13/19
2.2
2.8
3.9
7.5
2.1
2.6
4.1
7.3
9.0
11.8
16.7
33.0
8.7
10.9
17.3
32.0
⬍30 ␮mol/L
⬍29 ␮mol/L
⬍27 ␮mol/L
⬍24 ␮mol/L
ⱖ41%
ⱖ44%
ⱖ47%
ⱖ52%
122
118
110
98
111
107
99
87
84.1
81.4
75.9
67.6
76.6
73.8
68.3
60.0
90.1
92.5
95.0
97.5
90.1
93.1
95.4
97.5
6203
4685
3161
1573
6191
4301
2886
1547
8/23
8/27
12/35
18/47
11/34
13/38
17/46
26/58
1.9
2.5
3.4
5.9
1.8
2.4
3.3
5.3
8.5
10.9
15.0
26.9
7.7
10.8
14.826
24.305
a
The cut points represent specificities greater than or equal to 90%, 92.5%, 95%, and 97.5%, respectively, for each group and test.
Number of C282Y homozygotes not detected by the TS or UIBC tests who also had increased ferritin. These data are provided to illustrate how many C282Y
homozygotes with possible iron overload may have been missed at the various cut points.
c
PPV, positive predictive value; LR⫹, positive likelihood ratio.
b
men and 145 women). Non-C282Y homozygotes included
37 002 men and 62 667 women. The median age of all
participants in this study was 50 years (range, 25–100
years). By self-identified race/ethnicity, the participants
included 44% Caucasian, 27% African-American, 13%
Asian, 13% Hispanic, 0.7% Pacific Islander, 0.6% Native
American, and 2% mixed or unknown race; 97% of the
C282Y homozygotes were Caucasian. There were 4 Hispanic and 2 African-American C282Y homozygotes. An
increased serum ferritin (⬎300 ␮g/L in men, ⬎200 ␮g/L
in women) was found in 88% of the male C282Y homozygotes and in 57% of the female homozygotes. In men, the
area under the ROC curve (AUC) was significantly
greater for UIBC (0.96; 95% CI, 0.94 – 0.98) than TS (0.94;
95% CI, 0.90 – 0.97; P ⬍0.001; Fig. 1A). In women, the AUC
for the ROC was also significantly greater for UIBC (0.93;
95% CI, 0.91– 0.96) than for TS (0.90; 95% CI, 0.86 – 0.93; P
⬍0.001; Fig. 1B).
In this study we demonstrated that UIBC has a greater
AUC than TS for the detection of C282Y homozygotes.
The cut points shown in Table 1 were selected as a balance
between sensitivity and specificity (Figs. 1 and 2 in the
online Data Supplement). These cut points can be raised
or lowered to reduce false-positive and false-negative
results as appropriate (19 –21 ). Both TS and UIBC are
better for the detection of C282Y homozygotes with an
increased serum ferritin than the detection of all unselected C282Y homozygotes. In screening for hemochromatosis, adjusting the cut points to increase the sensitivity
(lower TS, higher UIBC) increases the number of C282Y
homozygotes detected but also increases the number of
non-C282Y homozygotes requiring further evaluation.
This follow-up evaluation may include DNA-based testing for HFE mutations. Adjusting the cut points in the
other direction (higher TS, lower UIBC) will detect fewer
C282Y homozygotes. This may be less important because
many such undetected C282Y homozygotes with a normal serum ferritin never develop iron overload (Table 1)
(8 ).
Both TS and UIBC likely are efficacious in the detection
of non-HFE iron overload. However, because confirmation of iron overload in such cases requires liver biopsy or
quantitative phlebotomy, it was beyond the scope of this
study to determine the operating characteristics of TS and
UIBC in the detection of non-HFE iron overload. Preliminary results of other studies of pedigrees with ferroportin
mutations have suggested that serum TS is less commonly
increased than serum ferritin (22 ).
Possible explanations for the improved performance of
UIBC over TS include the reduced analytical error in 1
assay vs 2 assays or the possibility that a circadian rhythm
in serum iron affects TS more than UIBC (23 ). Information
collected from both laboratories used in this study has
estimated the cost of UIBC testing to be slightly less than
TS testing, probably less than US $1.00, depending on
how one cost accounts the preanalytical (e.g., specimen
collection, processing, and loading of the serum specimen
on the automated analyzer) and postanalytical (e.g., results-reporting) steps. The reagent costs are also slightly
higher for the 2-step TS analysis compared with the 1-step
UIBC analysis. These analytical cost estimates are similar
or slightly lower than those reported for previous studies
(12–15 ). The ROC curves for UIBC and TS have only small
differences and could be considered to be almost equivalent, but the advantages of a single test at a lower cost
would make UIBC the preferred test.
In summary, this study has demonstrated in a large
primary care population that the UIBC is a useful test
for the detection of C282Y homozygotes. The UIBC has a
greater AUC for the ROC curve compared with TS.
Because it is a 1-step automated test, which is somewhat
less expensive to perform than TS testing, UIBC may be
the preferred biochemical screening test for C282Y-linked
hemochromatosis.
Acknowledgments
field centers
Birmingham, AL—University of Alabama at Birmingham
—Dr. Ronald T. Acton (Principal Investigator)—Dr. James
C. Barton (Co-Principal Investigator)—Ms. Deborah
Dixon—Dr. Susan Ferguson—Dr. Richard Jones—Dr.
Jerry McKnight—Dr. Charles A. Rivers—Dr. Diane Tucker—Ms. Janice C. Ware.
Irvine, CA—University of California, Irvine—Dr. Christine
E. McLaren (Principal Investigator)—Dr. Gordon D.
McLaren (Co-Principal Investigator)—Dr. Hoda AntonCulver—Ms. Jo Ann A. Baca—Dr. Thomas C. Bent—Dr.
Lance C. Brunner—Dr. Michael M. Dao—Dr. Korey S.
Jorgensen—Dr. Julie Kuniyoshi—Dr. Huan D. Le—Dr.
Miles K. Masatsugu—Dr. Frank L. Meyskens— Dr. David
Morohashi—Dr. Huan P. Nguyen—Dr. Sophocles N. Panagon—Dr. Chi Phung—Dr. Virgil Raymundo—Dr.
Thomas Ton—Professor Ann P. Walker—Dr. Lari B. Wenzel—Dr. Argyrios Ziogas.
London, Ontario, Canada—London Health Sciences Center
—Dr. Paul C. Adams (Principal Investigator)—Ms. Erin
Bloch—Dr. Subrata Chakrabarti—Ms. Arlene Fleischhauer—Ms. Helen Harrison—Ms. Bonnie Hogan—Ms.
Kelly Jia—Dr. John Jordan—Ms. Sheila Larson—Dr. Edward Lin—Ms. Melissa Lopez—MDS Laboratories—Dr.
Godfrey Moses—Ms. Lien Nguyen—Ms. Corry Pepper—
Dr. Tara Power—Dr. Mark Speechley—Dr. Donald Sun—
Ms. Diane Woelfle.
Portland, OR and Honolulu, HI—Kaiser Permanente Center
for Health Research, Northwest and Hawaii, and Oregon
Health and Science University—Dr. Emily L. Harris (Principal Investigator)—Dr. Mikel Aickin—Dr. Elaine Baker—
Ms. Marjorie Erwin—Ms. Joan Holup—Ms. Carol Lloyd—
Dr. Nancy Press—Dr. Richard D. Press—Dr. Jacob Reiss—
Dr. Cheryl Ritenbaugh—Ms. Aileen Uchida—Dr. Thomas
Vogt—Dr. Dwight Yim.
Washington, D.C.—Howard University—Dr. Victor R.
Gordeuk (Principal Investigator)—Dr. Fitzroy W.
Dawkins (Co-Principal Investigator)—Ms. Margaret
Fadojutimi-Akinsiku—Dr. Oswaldo Castro—Dr. Debra
White-Coleman—Dr. Melvin Gerald—Ms. Barbara W.
Harrison—Dr. Ometha Lewis-Jack—Dr. Robert F. Murray—Dr. Shelley McDonald-Pinkett—Ms. Angela Rock—
Dr. Juan Romagoza—Dr. Robert Williams.
1051
central laboratory
Minneapolis, MN—University of Minnesota and FairviewUniversity Medical Center—Dr. John H. Eckfeldt (Principal Investigator and Steering Committee Chair)—Ms.
Catherine Leiendecker-Foster—Dr. Ronald C. McGlennen—Mr. Greg Rynders—Dr. Michael Y. Tsai.
coordinating center
Winston-Salem, NC—Wake Forest University—Dr. David
M. Reboussin (Principal Investigator)—Dr. Beverly M.
Snively (Co-Principal Investigator)—Dr. Roger Anderson—Ms. Elease Bostic—Ms. Brenda L. Craven—Ms.
Shellie Ellis—Dr. Curt Furberg—Mr. Jason Griffin—Dr.
Mark Hall—Mr. Darrin Harris—Ms. Leora Henkin—Dr.
Sharon Jackson—Dr. Tamison Jewett—Mr. Mark D.
King—Mr. Kurt Lohman—Ms. Laura Lovato—Dr. Joe
Michaleckyj—Ms. Shana Palla—Ms. Tina Parks—Ms.
Leah Passmore—Dr. Pradyumna D. Phatak—Dr. Stephen
Rich—Ms. Andrea Ruggiero—Dr. Mara Vitolins—Mr.
Gary Wolgast—Mr. Daniel Zaccaro.
nhlbi project office
Bethesda, MD—Ms. Phyliss Sholinsky (Project Officer)—
Dr. Ebony Bookman—Dr. Henry Chang—Dr. Richard
Fabsitz—Dr. Cashell Jaquish—Dr. Teri Manolio—Ms. Lisa
O’Neill.
nhgri project office
Bethesda, MD—Ms. Elizabeth Thomson.
Dr. Jean MacCluer, Southwest Foundation for Biomedical Research, also contributed to the design of this study.
References
1. Pietrangelo A. Hereditary hemochromatosis: a new look at an old disease.
N Engl J Med 2004;350:2383–97.
2. Feder JN, Gnirke A, Thomas W, Tsuchihashi Z, Ruddy DA, Basava A, et al. A
novel MHC class I-like gene is mutated in patients with hereditary hemochromatosis. Nat Genet 1996;13:399 – 408.
3. Burke W, Thomson E, Khoury M, McDonnell S, Press N, Adams P, et al.
Hereditary hemochromatosis: gene discovery and its implications for population-based screening. JAMA 1998;280:172– 8.
4. Beutler E, Felitti V, Koziol J, Ho N, Gelbart T. Penetrance of the 845G to A
(C282Y) HFE hereditary haemochromatosis mutation in the U S A. Lancet
2002;359:211– 8.
5. Asberg A, Hveem K, Thorstensen K, Ellekjaer E, Kannelonning K, Fjosne U,
et al. Screening for hemochromatosis— high prevalence and low morbidity in
an unselected population of 65,238 persons. Scand J Gastroenterol 2001;
36:1108 –15.
6. Adams PC. Non-expressing C282Y homozygotes for hemochromatosis:
minority or majority of cases? Mol Genet Metab 2000;71:81– 6.
7. Andersen R, Tybjaerg-Hansen A, Appleyard M, Birgens H, Nordestgaard B.
Hemochromatosis mutations in the general population: iron overload progression rate. Blood 2004;103:2914 –9.
8. Yamashita C, Adams PC. Natural history of the C282Y homozygote of the
hemochromatosis gene (HFE) with a normal serum ferritin level. Clin
Gastroenterol Hepatol 2003;1:388 –91.
9. Tavill AS. Diagnosis and management of hemochromatosis. Hepatology
2001;33:1321– 8.
10. Shaheen N, Lawrence L, Bacon B, Barton J, Barton N, Galanko T, et al.
Insurance, employment, and psychosocial consequences of a diagnosis of
hereditary hemochromatosis in subjects without end-organ damage. Am J
Gastroenterol 2003;98:1175– 80.
11. Barash C. Genetic discrimination and screening for hemochromatosis: then
and now. Genet Test 2000;4:213– 8.
12. Adams PC, Kertesz AE, McLaren C, Barr R, Bamford A, Chakrabarti S.
Population screening for hemochromatosis: a comparison of unbound iron
binding capacity, transferrin saturation and C282Y genotyping in 5,211
voluntary blood donors. Hepatology 2000;31:1160 – 4.
1052
Technical Briefs
13. Hickman P, Hourigan L, Powell LW, Cordingley F, Dimenski G, Ormiston
BSJFW, et al. Automated measurement of unsaturated iron binding capacity
is an effective screening strategy for C282Y homozygous hemochromatosis.
Gut 2000;46:405–9.
14. Witte DL, Crosby WH, Edwards CQ, Fairbanks VF, Mitros FA. Hereditary
hemochromatosis. Clin Chim Acta 1996;245:139 –200.
15. Adams PC, Bhayana V. Unsaturated iron binding capacity: a screening test
for hemochromatosis? Clin Chem 2000;46:1870 –1.
16. McLaren C, Barton J, Adams P, Harris E, Acton R, Press N, et al.
Hemochromatosis and Iron Overload Screening (HEIRS) study design for an
evaluation of 100,000 primary care-based adults. Am J Med Sci 2003;325:
53– 62.
17. Adams PC, Reboussin DM, Barton JC, McLaren CE, Eckfeldt JH, McLaren
GD, et al. Hemochromatosis and iron-overload screening of a racially diverse
population. N Engl J Med 2005;352:1769 –78.
18. DeLong E, DeLong D, Clarke-Pearson D. Comparing the areas under two or
more correlated receiver operating characteristic curves: a non-parametric
approach. Biometrics 1988;44:837– 44.
19. Henderson AR. Assessing test accuracy and its clinical consequences: a
primer for receiver operating characteristic curve analysis. Ann Clin Biochem
1993;30:521–39.
20. Obuchowski N, Lieber M, Wians F. ROC curves in clinical chemistry: Uses,
misuses, and possible solutions. Clin Chem 2004;50:1118 –25.
21. Zweig M, Campbell G. Receiver-operating characteristic (ROC) plots: a
fundamental evaluation tool in clinical medicine. Clin Chem 1993;39:561–
77.
22. Pietrangelo A. Non-HFE hemochromatosis. Hepatology 2004;39:21–9.
23. Guillygomarch A, Jacquelinet C, Moirand R, Vincent Q, David V, Deugnier Y.
Circadian variations of transferrin saturation levels in iron-overloaded patients: implications for screening of C282Y-linked haemochromatosis. Br J
Haematol 2003;120:359 – 63.
Monoclonal versus Polyclonal ELISA for Assessment of
Fecal Elastase Concentration: Pitfalls of a New Assay,
Arne Schneider,1* Benjamin Funk,2 Wolfgang Caspary,1 and
Juergen Stein1 (1 Medical Department I and 2 Department
of Pediatrics, University Hospital, Frankfurt/Main, Germany; * address correspondence to this author at: Medical
Department I, Johann Wolfgang Goethe-University
Frankfurt, Theodor-Stern-Kai 7, D-60590 Frankfurt/Main,
Germany; fax 49-69-6301-6448, e-mail arne.schneider@
em.uni-frankfurt.de)
Exocrine pancreatic insufficiency is a frequent consequence of chronic or severe acute pancreatitis. Assessment of exocrine pancreatic function is commonly performed with stool tests, which have largely replaced the
need for invasive function tests. Elastase is excreted by the
pancreas and passes through the intestine without significant degradation or inactivation. Consequently, measurement of fecal elastase fulfills the requirements of an
almost ideal surrogate marker for pancreatic exocrine
function (1–5 ). The enzyme remains relatively stable in
vitro, allowing stool samples to be mailed to a laboratory
for analysis.
To date, the assay most widely used for measurement
of pancreatic elastase has been a commercially available
ELISA with a monoclonal antibody. The assay has been
reported to be pathologic in 93% of patients with severe
exocrine pancreatic insufficiency and in 63% of those with
mild chronic pancreatitis, and normal in 93%–96% of
those without pancreatic insufficiency. In contrast to
many other pancreatic function tests, the analysis does not
require interruption of oral porcine pancreatic replacement therapy because only the human form of elastase 1
is detected by the assay (1–5 ).
A new polyclonal antibody test using 2 different polyclonal antisera to human elastase has been reported to be
positive in 78% of patients compared with 69% positivity
for the monoclonal test in the same patients at a cutoff of
200 ␮g/g elastase (6 ). A cutoff of 100 ␮g/g gave similar
sensitivities and specificities for the 2 tests. Binding studies showed that the polyclonal test seems to detect antigens that partly differ from “classic” elastase 1 (7 ).
Simultaneous evaluation of stool specimens showed a
tendency for higher values in the polyclonal test, which
might cause a higher proportion of false-negative results.
We addressed 2 major questions: Does the polyclonal
test specifically detect a human antigen when used to test
samples from patients receiving oral porcine enzyme
replacement therapy? Does the corresponding fecal antigen of the polyclonal test remain as stable as fecal elastase
1 in stool samples stored under different conditions?
The microtiter plates for the monoclonal elastase test
(mET) are coated with a monoclonal antibody that specifically binds elastase 1 in stool specimens (Pancreatic
Elastase 1; Schebo Biotech AG). The test is based on an
immunoenzymatic method, which we performed according to the manufacturer’s guidelines. The polyclonal elastase test (pET; Elastase 1 ELISA) was purchased from
BioServ Diagnostics. This assay uses 2 polyclonal antisera
that recognize different antigenic epitopes of pancreatic
elastase. The test procedures are comparable to those of
the mET.
We measured fecal elastase in stool samples from 27
patients with cystic fibrosis (CF) and 10 patients with
normal exocrine pancreatic function. The patients with CF
had severely compromised exocrine pancreatic function,
and all were receiving oral pancreatic enzyme replacement therapy consisting of 5000 –10 000 U of porcine
lipase 䡠 kg⫺1 䡠 day⫺1.
Each patient with normal exocrine function provided a
stool specimen, which was divided into 4 parts: 1 was
tested directly after defecation with both the monoclonal
and the polyclonal assays; the other 3 were stored at 4 °C,
room temperature (22 °C), or 37 °C. We simultaneously
analyzed these 3 portions for elastase with both assays
after 24 and 48 h.
Statistical analyses were performed with GraphPad
Prism, Ver. 4.01. Comparisons between the elastase concentrations obtained with the mET and the pET were
performed with the Wilcoxon test. A P value ⱕ0.05 was
considered significant.
In CF patients receiving oral enzyme replacement therapy, the mET showed pathologically decreased fecal
elastase concentrations in all patients and the pET gave
normal or near-normal elastase concentrations (Fig. 1).
This difference was highly significant [mean (SD), 138 (69)
and 4.3 (9.0) ␮g/g with the pET and mET, respectively; P
⬍0.0001].
In addition, we analyzed a stool sample from a patient
1053
Table 1. Fecal elastase concentrations in 10 stool samples
assayed by mET and pET after 24 and 48 h of storage at
4 °C, room temperature (22 °C), or 37 °C.
Mean (SD) measured
elastase, ␮g/g
Assay
Mean (SD) of direct
measurement, ␮g/g
mET
434 (122)
pET
a
Fig. 1. Comparison of fecal elastase concentrations measured with the
pET and the mET in 27 patients with CF-associated severe exocrine
pancreatic insufficiency who were receiving continuous oral pancreatic
enzyme replacement therapy.
Solid lines indicate the mean value obtained with each assay; the dashed line
indicates the cutoff. The difference between the mean measured values was
significant (P ⬍0.0001).
with chronic pancreatitis and severely compromised exocrine pancreatic function with both the mET and pET
before and after addition of 100 and 300 U of commercially available porcine lipase (Kreon®; Solvay GmbH),
which had been dissolved in a NaHCO3 solution (84 g/L
NaHCO3). The elastase concentrations measured with the
mET and pET before addition of lipase were 2.2 and 23.2
␮g/g, respectively. After the addition of pancreatin, the
measured concentrations were 2.8 ␮g/g (100 U) and 2.9
␮g/g (300 U) in the mET vs 66 ␮g/g (100 U) and 244 ␮g/g
(300 U) in the pET. We also measured fecal elastase in
samples from 5 patients with chronic pancreatitis before
and during oral enzyme replacement therapy (500 –1000
U of porcine lipase 䡠 kg⫺1 䡠 meal⫺1). Before the initiation of
oral enzyme replacement therapy, measured elastase concentrations were 1.3 (0.9) ␮g/g in the mET and 18.2 (17.5)
␮g/g in the pET. Whereas the mET result remained low
[1.8 (0.9) ␮g/g] after the initiation of oral enzyme therapy,
the elastase concentrations measured by the pET assay
increased significantly [62.2 (31.1) ␮g/g; P ⫽ 0.03].
Compared with elastase concentrations assessed directly in fresh stool specimens, the measurements performed after 24 and 48 h of storage at different temperatures gave heterogeneous results. There was a tendency
for decreasing elastase concentrations in the samples
measured after 24 and 48 h of storage at 4 and 37 °C with
the mET (Table 1). The results of the pET were significantly higher after 48 h at 4 °C (P ⫽ 0.004), whereas the
other analyses seemed stable irrespective of whether the
samples were stored at room temperature or 37 °C.
Although fecal elastase testing has a low sensitivity for
the diagnosis of mild exocrine pancreatic insufficiency,
the assay reliably predicts clinically relevant degrees of
pancreatic insufficiency and is widely used in clinical
medicine (7, 8 ). In contrast to fecal chymotrypsin, the
Storage at
24 h
48 h
4 °C
22 °C
37 °C
428 (110)
454 (157)
436 (161)
387 (105)
435 (135)
391 (138)
4 °C
22 °C
37 °C
561 (118)
538 (110)
500 (146)
586 (94)a
564 (89)
501 (135)
544 (93)
P ⫽ 0.004 vs directly measured elastase.
monoclonal elastase 1 assay does not require interruption
of oral enzyme replacement therapy because commercially available enzyme compounds do not interfere with
the test (9 ). With regard to this point, the widely different
results obtained with the well-established mET compared
with the relatively new pET ELISA among our patients
with CF are a disappointing finding. We can proceed from
the assumption that the CF patients studied suffered from
severely compromised exocrine pancreatic insufficiency
because the disease inevitably leads to the destruction of
virtually all pancreatic tissue: ⬃60% of neonates diagnosed with CF already suffer from pancreatic insufficiency (10 ). This proportion increases to 92% during the
first year of life (11 ). To maintain adequate nutrition,
patients require oral pancreatic enzyme replacement therapy. As a consequence, an explanation for the substantial
difference between the elastase tests is that the polyclonal
assay is influenced by the oral pancreatic supplement,
which consists of capsules containing porcine pancreatin,
lipase, amylase, and proteases. The finding of measured
elastase concentrations ⬎100 ␮g/g in the pET in the
majority of CF patients may reflect adequate high-dose
oral enzyme substitution.
A recently published study concluded that the antibodies used in the pET assay detect antigens different from
elastase 1 (7 ). In a PubMed search, we found no additional protein-binding studies elucidating the interaction
of pET antibodies with fecal antigens. Accordingly, our
findings support the thesis that the pET assay detects
antigens that are different from elastase 1 but are probably
contained in oral pancreatic enzyme capsules. This has
also been shown in our own laboratory by simple addition of dissolved pancreatin from commercially available
enzyme capsules to fecal specimens, which produced a
proportional increase in the elastase concentration detected by the pET. The same was true for the comparative
analysis of stool samples from patients with chronic
pancreatitis before and during “in vivo” substitution of
enzymes. The fact that the pET generally gives higher
results than the mET further supports the concept of a
different antigenic specificity. For this reason, we cannot
1054
Technical Briefs
recommend that the pET be used to monitor patients
receiving oral enzyme replacement therapy, which is a
major drawback for the clinical application of the pET.
On the other hand, the antigenic substrate in the pET
seems to be equally stable in relation to elastase 1. Only
elastase concentrations determined with the pET in stools
stored at 4 °C for 48 h were significantly higher than
directly measured concentrations. Although both pET
and mET assessments were performed on stool samples
stored in continuously closed cups, partial evaporation
with consecutive concentration of the substrate cannot be
excluded.
If we take into consideration the different aspects of
fecal elastase testing highlighted in this study, the lack of
specificity of the pET assay remains the crucial problem.
We therefore support the proposal of renaming the polyclonal “elastase 1” ELISA because the test detects a
molecule different from elastase 1 (7 ). Further studies
should evaluate the effects of porcine enzymes on the
results of the pET assay in patients with normal and
mildly to moderately decreased exocrine pancreatic function. Additionally, the true antigen detected with the pET
assay should be characterized to reliably define the role of
this new assay in the diagnosis of exocrine pancreatic
insufficiency.
References
1. Loeser C, Moellgaard A, Foelsch UR. Fecal elastase-1: a novel, highly
sensitive and specific tubeless pancreatic function test. Gut 1996;39:
580 – 6.
2. Stein J, Spichez Z, Lembcke B, Caspary WF. Evaluation of fecal elastase as
a new non-invasive test for exocrine pancreatic insufficiency. Z Gastroenterol
1997;35(Suppl 1):122–9.
3. Siegmund E, Loehr JM, Schuff-Werner P. The diagnostic validity of noninvasive pancreatic function tests—a meta-analysis. Z Gastroenterol 2004;
42:1117–28.
4. Lüth S, Teyssen S, Forssmann K, Kolbel C, Krummenauer F, Singer MV.
Fecal elastase-1 determination: ‘gold standard’ of indirect pancreatic function tests? Scand J Gastroenterol 2001;36:1092–9.
5. Dominguez-Munoz JE, Hieronymus C, Sauerbruch T, Malfertheiner P. Fecal
elastase test: evaluation of a new noninvasive pancreatic function test. Am J
Gastroenterol 1995;90:1834 –7.
6. Keim V, Teich N, Moessner J. Clinical value of a new fecal elastase test for
detection of chronic pancreatitis. Clin Lab 2003;49:209 –15.
7. Hardt PD, Hauenschild A, Nalop J, Marzeion AM, Porsch-Ozcurumez M, Luley
C, et al. The commercially available ELISA for pancreatic elastase 1 based
on polyclonal antibodies does measure an as yet unknown antigen different
from purified elastase 1. Binding studies and clinical use in patients with
exocrine pancreatic insufficiency. Z Gastroenterol 2003;41:903– 6.
8. Lankisch PG. Now that fecal elastase is available in the United States,
should clinicians start using it? Curr Gastroenterol Rep 2004;6:126 –31.
9. Stein J, Jung M, Sziegoleit A, Zeuzem S, Caspary WF, Lembcke B. Immunoreactive elastase I: clinical evaluation of a new noninvasive test of pancreatic function. Clin Chem 1996;42:222– 6.
10. Waters DL, Dorney SF, Gaskin KJ, Gruca MA, O’Halloran M, Wilcken B.
Pancreatic function in infants identified as having cystic fibrosis in a
neonatal screening program. N Engl J Med 1990;322:303– 8.
11. Bronstein MN, Sokol RJ, Abman SH, Chatfield BA, Hammond KB, Hambidge
KM, et al. Pancreatic insufficiency, growth and nutrition in infants identified
by newborn screening as having cystic fibrosis. J Pediatr 1992;120:533–
40.
Soluble CD40 Ligand Measurement Inaccuracies Attributable to Specimen Type, Processing Time, and ELISA
Method, Anna Margrét Halldórsdóttir,1 Joshua Stoker,2
Rhonda Porche-Sorbet,1 and Charles S. Eby1* (Divisions of
1
Laboratory Medicine and 2 Cardiology, Washington
University School of Medicine, St. Louis, MO; * address
correspondence to this author at: Washington University
School of Medicine, St. Louis, MO 63110; fax 314-362-1461,
e-mail eby@pathbox.wustl.edu)
CD40 ligand (CD40L) is a member of the tumor necrosis
factor superfamily and is produced in a variety of cells,
including platelets. The soluble form (sCD40L) is a mediator of both inflammatory and hemostasis processes and
has been implicated in the pathogenesis of atherosclerosis.
Clinical studies have revealed increased sCD40L in patients with unstable angina (1 ) and identified an association between increased sCD40L and future risk for death
or nonfatal myocardial infarction (2, 3 ). While prospectively measuring sCD40L in a cohort of persons at risk for
cardiovascular complications, we identified both preanalytical and analytical sources of error. This report documents the effects of specimen type (serum and plasma),
processing (time and temperature), and commercial reagent selection on sCD40L ELISA results. These findings
raise concerns about the accuracy of sCD40L results
reported in recent clinical studies.
After obtaining informed consent, we enrolled 147
patients older than 60 years referred for diagnostic
cardiac catheterization in an Institutional Review
Board-approved study to evaluate the value of clinical,
echocardiographic, and biomarker variables for prediction of future cardiovascular complications. When combined with clinical predictors, B-type natriuretic peptide
and C-reactive protein, but not sCD40L, were independent predictors of death or cardiovascular hospitalization
at 6 months (data not shown). The unexpectedly poor
correlation between undiluted plasma sCD40L results
and clinical outcomes in this study motivated us to
perform the following investigations.
Venous blood from the 147 study participants [mean
(SD) age, 70.8 (6.9) years] was collected into plastic tubes
containing tripotassium EDTA (BD Vacutainer; Becton
Dickinson) before catheterization and placed on ice for
1– 4 h before processing. Blood from 10 control individuals [mean (SD) age, 38.7 (8.4) years] was collected into
both EDTA and plain glass tubes (Becton Dickinson) and
maintained at room temperature for 30 min before processing. Study and control samples were centrifuged
twice: first at 2790g for 5 min to separate cells from the
plasma/serum and then at 16 000g for 3 min to remove
any residual platelets. Supernatants were aliquoted and
stored at ⫺70 °C.
We assessed the effects of time and temperature on
measured sCD40L concentrations by collecting whole
blood from a single healthy individual into a syringe and
immediately aliquoting it into EDTA-containing and plain
glass tubes. For every time point analyzed, plasma and
serum tubes were kept on cells at room temperature, and
in 1 experiment they were also kept on ice. Samples were
then centrifuged and stored as described above. Selected
samples from the time–temperature experiment were
either filtered through a 0.2 ␮m syringe filter or were
ultracentrifuged at 200 000g for 4 h at 4 °C to remove
potential remaining platelet microparticles before repeat
sCD40L testing.
sCD40L concentrations were measured with an sCD40L
assay (Quantikine®; R&D Systems). According to the
package insert, the R&D ELISA is suitable for measuring
sCD40L in serum and plasma and is linear within the
analytical range of the assay (0.0625– 4 ␮g/L). The stated
lower limit of detection is 0.0042 ␮g/L. Reference intervals for serum (0.675–38.373 ␮g/L; mean, 8.273 ␮g/L; n ⫽
44) and platelet-poor EDTA plasma (0.106 –11.831 ␮g/L;
mean, 2.987 ␮g/L; n ⫽ 16) were provided by the manufacturer. The manufacturer’s protocols were followed.
We retested selected samples with 2 sCD40L assays
(BMS 239 and BMS 293) from Bender MedSystems. The
BMS 239 is suitable only for testing serum, whereas the
BMS 293 is a high-sensitivity assay designed both for
plasma and serum (package inserts). The lower limit of
detection for the BMS 239 is 0.095 ␮g/L, and the lower
limit of detection for the BMS 293 is 0.005 ␮g/L. The
manufacturer’s protocols were followed.
When establishing a central 95% interval for sCD40L
with 10 control plasma samples diluted 1:5, per the R&D
package insert, we found that all results were below the
lowest point on the calibration curve (0.0625 ␮g/L).
Following discussions with the manufacturer’s technical
consultants, we tested control and patient plasma samples
undiluted. The distribution of sCD40L concentrations in
undiluted plasma for the 147 patients is shown in Fig. 1A.
When the 2 control and 8 patient plasma samples with
sCD40L concentrations exceeding the upper limit of the
calibration curve were diluted 1:2 and 1:5 in calibration
diluent, the results were not linear (data not shown). To
determine whether the nonlinear dilution response was a
systematic analytical problem, we added recombinant
sCD40L to serum and plasma samples with sCD40L
concentrations ⬍0.2 ␮g/L to produce a predicted concentration of 2 ␮g/L. Serial 2-fold dilutions of these samples
in the calibration diluent also produced nonlinear results
(data not shown).
Interestingly, when we compared the values for undiluted serum and plasma samples from the control group,
the mean sCD40L concentration in undiluted serum (1.33
␮g/L) was 6-fold higher than in undiluted plasma (0.24
␮g/L). There was a weak correlation (r ⫽ 0.223) between
serum and plasma concentrations for undiluted samples.
To evaluate the correlation between sCD40L concentrations measured by ELISAs from different manufacturers,
we compared results obtained for selected serum and
plasma samples by the R&D ELISA and the Bender BMS
MedSystems ELISAs. The R&D and Bender BMS 239
assays showed good correlation for serum samples (Fig.
1B), but for plasma, the correlation was poor between the
1055
new Bender high-sensitivity BMS 293 assay and the R&D
ELISA (Fig. 1C).
Finally, we investigated the impact of the processing
variables time and temperature on sCD40L determinations. When serum from a healthy donor was stored on
cells at room temperature, there was a 6- to 7-fold increase
in sCD40L concentrations after 180 min (Fig. 1D). sCD40L
concentrations in similarly treated plasma samples did
not increase, and most values were below the analytical
range. When either serum or plasma was stored on ice, no
increase in sCD40L concentration was observed over 180
min (data not shown).
To examine whether the time-dependent increase in
serum sCD40L concentrations was attributable to release
of platelet microparticles producing membrane sCD40L,
we either filtered or ultracentrifuged specimens before
repeat testing. No difference was observed (data not
shown).
Accurate measurement of an analyte is essential for its
clinical diagnostic utility. Despite reports showing an
association between increased sCD40L and cardiovascular complications (2, 3 ), there is poor agreement among
studies regarding sCD40L ranges for controls or cases
with similar cardiovascular risk factors (1–7 ). In addition,
the methods sections in some reports fail to specify
whether serum or plasma was tested or whether plasma
was tested with an ELISA designed for testing serum, and
they provide few sample-processing details (2, 3, 6, 8 ).
After reviewing the literature and the manufacturers’
product specifications, we decided to measure plasma
sCD40L with the R&D ELISA. We were disappointed to
discover that the only sCD40L ELISA suitable for plasma
testing at that time lacked the sensitivity to measure
sCD40L in 100% of controls when plasma was diluted 1:5,
according to the manufacturer’s recommendations.
When we tested undiluted control and patient plasma
samples, 10 of 157 (6%) gave exceedingly high results, and
serial dilutions of these specimens produced nonlinear
results. After we shared these findings with the manufacturer, the R&D sCD40L ELISA was briefly withdrawn
from the market while changes were made in the assay
diluent to address the presence of heterophilic antibodies
in some samples. However, no changes were made to
increase the sensitivity of the assay.
Most clinical studies have measured sCD40L in serum
with either the R&D or Bender 293 ELISAs, reducing the
problem of analytical insensitivity. Thom et al. (9 ) reported that mean measured sCD40L concentrations were
9-fold higher in serum than in plasma when assayed with
a Bender sCD40L ELISA, which is consistent with our
results. In addition, we have shown that the agreement
between the R&D and Bender 239 ELISA methods for
measuring sCD40L in serum was excellent (Fig. 1B).
However, we have also shown that the serum sCD40L
concentration increases significantly with time in samples
stored at room temperature (Fig. 1D). This is in agreement
with previously published data (9, 10 ) and represents the
combination of in vivo sCD40L, which is likely to be the
1056
Technical Briefs
physiologically relevant component, and ex vivo-released
sCD40L.
Platelets are activated during the process of clot retraction, and sCD40L shedding from the platelet surface
probably accounts for the progressive increase in serum
concentrations. Shed sCD40L could be bound to platelet
microparticles (7 ). However, in our experiments, filtration
and ultracentrifugation did not lead to a decrease in
serum concentrations of sCD40L, suggesting that ex vivoreleased sCD40L is not bound to intact membrane. It may
therefore be impossible to distinguish between in vivo
and ex vivo release of sCD40L.
Fig. 1. sCD40L results depend on specimen type/processing and ELISA method.
(A), distribution of undiluted plasma sCD40L concentrations in 147 patients undergoing cardiac catheterization. Each data point represents 1 patient. Specimens were
tested undiluted. Dotted lines represent the dynamic range of the assay. (B and C), comparison of 2 commercial sCD40L ELISA tests, using plasma and serum
specimens from the time-temperature experiments. Dashed lines represent the lines of unity. Results have been corrected for dilution. (B), comparison of the R&D
assay with the Bender BMS 239 for serum samples. The results of the regression analysis were as follows: y ⫽ 0.96x ⫹ 0.42 ␮g/L. (C), comparison of the R&D ELISA
with the Bender BMS 293 for plasma samples. The results of the regression analysis were as follows: y ⫽ 0.96x ⫹ 6.08 ␮g/L. (D), effect of sample storage conditions
on measured sCD40L concentrations. Each data point represents the mean of 2 experiments, the error bars represent SD. Serum (E) and plasma (F) samples drawn
from a healthy donor were stored on cells at room temperature for different lengths of time before processing. The graph shows the change in measured sCD40L
concentrations with time. Results have been corrected for dilution.
The measurement of sCD40L concentrations in human
blood with the R&D ELISA is therefore problematic for
the following reasons: the assay lacks sensitivity for
measuring sCD40L concentrations in diluted plasma samples; testing of serum is problematic because of ex vivo
release of sCD40L; there is poor correlation between
plasma and serum samples; and the linearity of measurements obtained with the reformulated assay reagents has
not been evaluated.
Recently, Bender MedSystems began selling a highsensitivity sCD40L ELISA (Bender 293) suitable for
plasma and serum testing. A preliminary evaluation
confirmed that it is more sensitive than the R&D sCD40L
ELISA test for plasma, but no further studies have been
performed.
In summary, investigators should carefully consider the
choice of specimen type, specimen-handling procedures,
and properties of the commercial ELISA tests when measuring sCD40L concentrations in blood because each of
these variables can critically affect measured sCD40L
concentrations. The optimum strategy would be to measure sCD40L in platelet-free plasma by a sensitive analytical method.
We thank Bender MedSystems (Vienna, Austria) for supplying the high-sensitivity sCD40L ELISA.
References
1. Aukrust P, Müller F, Ueland T, Berget T, Aaser E, Brunsvig A, et al. Enhanced
levels of soluble and membrane-bound CD40 ligand in patients with
unstable angina. Circulation 1999;100:614 –20.
2. Heeschen C, Dimmeler S, Hamm C, van den Brand M, Boersma E, Zeiher A,
et al. Soluble CD40 ligand in acute coronary syndromes. N Engl J Med
2003;348:1104 –11.
3. Varo N, de Lemos J, Libby P, Morrow D, Murphy S, Nuzzo R, et al. Soluble
CD40L. Risk prediction after acute coronary syndromes. Circulation 2003;
108:1049 –52.
4. Schönbeck U, Varo N, Libby P, Buring J, Ridker P. Soluble CD40L and
cardiovascular risk in women. Circulation 2001;104:2266 – 8.
5. Viallard J, Solanilla A, Gauthier B, Contin C, Déchanet J, Grosset C, et al.
Increased soluble and platelet-associated CD40 ligand in essential thrombocythemia and reactive thrombocytosis. Blood 2002;99:2612– 4.
6. Aggarwahl A, Schneider D, Terrien E, Sobel B, Dauerman H. Increased
coronary arterial release of interleukin-1 receptor antagonist and soluble
CD40 ligand indicative of inflammation associated with culprit coronary
plaques. Am J Cardiol 2004;93:6 –9.
7. Jinchuan Y, Zonggui W, Jinming C, Li L, Xiantao K. Upregulation of CD40CD40 ligand system in patients with diabetes mellitus. Clin Chim Acta
2003;339:85–90.
8. Yan J, Zhu J, Gao L, Wu Z, Kong X, Zong R, et al. The effect of elevated serum
soluble CD40 ligand on the prognostic value in patients with acute coronary
syndromes. Clin Chim Acta 2004;343:155–9.
9. Thom J, Gilmore G, Yi Q, Hankey J, Eikelboom J. Measurement of soluble
P-selectin and soluble CD40 ligand in serum and plasma. J Thromb Haemost
2004;2:2067–9.
10. Nannizzi-Alaimo L, Rubenstein M, Alves V, Leong G, Phillips D, Gold H.
Cardiopulmonary bypass induces release of soluble CD40 ligand. Circulation 2002;105:2849 –54.
DOI: 10.1373/clinchem.2005.048199
1057
Differences and Similarities between Two Frequently
Used Assays for Amyloid ␤ 42 in Cerebrospinal Fluid,
Niki S.M. Schoonenboom,1,2†* Cees Mulder,2† Hugo Vanderstichele,3 Yolande A.L. Pijnenburg,1 Gerard J. Van Kamp,2
Philip Scheltens,1 Pankaj D. Mehta,4 and Marinus A. Blankenstein2 (1 Alzheimer Center and Department of Neurology,
and 2 Department of Clinical Chemistry, VU University
Medical Center, Amsterdam, The Netherlands; 3 Innogenetics NV, Ghent, Belgium; 4 Institute for Basic Research
in Developmental Disabilities, Department of Developmental Neurobiology, Division of Immunology, Staten
Island, NY; † these authors equally contributed to the
work; * address correspondence to this author at: Departments of Neurology and Clinical Chemistry, VU
University Medical Center, PO Box 7057, 1081 HV Amsterdam, The Netherlands; fax 31-(0)204440715, e-mail
niki.schoonenboom@vumc.nl)
Amyloid ␤ 42 (A␤ 42) concentrations in cerebrospinal
fluid (CSF) are used to identify Alzheimer disease (AD)
(1 ), but reported concentrations differ among studies, as
does diagnostic accuracy (2 ). These differences may relate
to the patient and control groups studied (3 ), processing
and storage methods (4 ), intra- and interassay variation of
the assays, or to the reagent antibodies used. A recent
metaanalysis (2 ) stressed the importance of standardizing
assays for A␤– 42 in CSF. In most studies, CSF A␤ 42 was
reported to be decreased, but in 2 studies, CSF A␤ 42 was
not significantly changed in AD (2 ), and in 1 study (5 )
even increased in the early stages of disease. These
dissimilarities might reflect the specificities of the antibodies incorporated in the assays.
The first aim of our study was to compare A␤ 42
concentrations measured by 2 different assays in the same
CSF samples. The first assay, widely used in Europe (6 ),
uses 2 monoclonal antibodies (mAbs) and detects the
full-length A␤ 42 peptide, A␤ (1– 42) (7 ). The second
assay [A␤ (N– 42)], used mainly in the United States (8 ),
detects both full-length A␤ 42 and A␤ peptides truncated
at the NH2 terminus (9 ).
The second aim of our study was to compare diagnostic
accuracies of the assays for patients with AD compared
with controls without dementia and patients with frontotemporal lobar degeneration (FTLD).
Finally, we investigated the relationship between CSF
A␤ (1– 42) and A␤ (N– 42) concentrations and albumin
ratio, age, disease duration, and disease severity.
Between October 2000 and December 2002, we recruited 39 AD patients, 24 FTLD patients, and 30 nondementia controls at the Alzheimer Center of the VU
University Medical Center (VUMC). All patients underwent a standardized investigative battery (3 ). A diagnosis
of “probable” AD was made according to the NINCDSADRDA criteria (10 ); the clinical picture of FTLD (including frontotemporal dementia, semantic dementia, and
progressive aphasia) was based on international clinical
diagnostic criteria (11 ). Disease duration was defined as
the time in years between the first symptoms by history
and the lumbar puncture.
1058
Technical Briefs
The control group (n ⫽ 30) consisted of 20 persons with
subjective memory complaints, who had undergone the
same battery of examinations as the patients; 5 spouses of
patients; 3 individuals with a positive family history for
AD, all without memory complaints; 1 patient with a
suspicion of intracranial hypertension; and 1 patient with
a possible neuritis vestibularis. No controls developed
dementia within 1 year. The Mini Mental State Examination (MMSE) score (12 ) was used as a measure of global
cognitive impairment. The study was approved by the
ethics review board of the VUMC. All patients and
controls gave written informed consent.
CSF was collected and stored as described previously
(4 ). The albumin ratio (serum albumin/CSF albumin) was
used as a measurement of the intactness of the blood–
brain barrier. Except for 1 FTLD patient and 2 controls, the
blood– brain barriers of the patients were intact (Table 1).
The INNOTESTTM ␤-AMYLOID(1– 42) (Innogenetics)
uses mAb 21F12, which binds the COOH terminus of the
A␤ 42 peptide (amino acids 36 – 42), as capture antibody
and biotinylated mAb 3D6, which binds the NH2 terminus (amino acids 1– 6), as detection antibody (6 ). A␤ (1–
42) peptides from Bachem were used for calibration. This
test was performed at the Department of Clinical Chemistry, VUMC, Amsterdam.
The sandwich ELISA for A␤ (N– 42) uses the commercially available mAb 6E10 (Signet Labs), specific to an
epitope covering N-terminal amino acid residues 1–17 of
A␤ 42, as capture antibody and the polyclonal antibody
R165 as detector antibody. R165 was made by immunizing rabbits with conjugated A␤ 33– 42 peptides (Ana
Spec). A␤ (1– 42) from Bachem was used for calibration,
although production procedures for the calibrators were
slightly different between the 2 laboratories. This test was
performed at the New York site according to an in-house
protocol.
For statistical analysis, we used SPSS (Ver. 11.0). Passing and Bablok regression analyses (13 ) were performed
with Medcalc, Ver. 4.30 (Medcalc Software), and we also
prepared a Bland–Altman plot (14 ). For group differences, we applied the Kruskal–Wallis test, followed by the
Mann–Whitney U-test applying the Bonferroni correction.
The ␹2 test with continuity correction was used to test
group differences within genders.
The sensitivities and specificities for CSF A␤ (1– 42) and
A␤ (N– 42) were calculated. Cut points corresponded to a
sensitivity ⱖ85% (15 ), but if a higher sensitivity was
obtained for a reasonable specificity, it was used. ROC
curves were constructed, and the areas under the curves
(AUCs) were calculated and compared (16 ). Spearman
correlations were calculated. A test was considered significant at P ⬍0.05. All reported tests are 2-tailed unless
stated otherwise.
The CSF A␤ (1– 42) and A␤ (N– 42) concentrations were
not statistically significantly different (Table 1 and Figs. 1
and 2 in the Data Supplement that accompanies the online
version of this Technical Brief at http://www.clinchem.
org/content/vol51/issue6/). Concentrations of both CSF
A␤ (1– 42) and A␤ (N– 42) were significantly lower in AD
patients than in patients with FTLD and in controls (Table
1). CSF A␤ (1– 42) concentrations differed significantly
between FTLD patients and controls, whereas CSF
A␤ (N– 42) concentrations did not differ significantly between the 2 groups (Table 1). The ratio of A␤ (1– 42) to
A␤ (N– 42) differed significantly only between the AD
and FTLD patient groups.
ROC curves for CSF A␤ (1– 42) and A␤ (N– 42) are
shown in Fig. 1. In AD patients vs controls, the sensitivity
and specificity for CSF A␤ (1– 42) were 90% and 93%,
respectively, at 473 ng/L and for CSF A␤ (N– 42), they
were 90% and 87%, respectively, at 383 ng/L. The AUCs
were not different (Fig. 1A) for A␤ (1– 42) and A␤ (N– 42)
[0.94 (95% confidence interval, 0.86 – 0.99) and 0.92 (0.83–
0.97), respectively; P ⫽ 0.47].
When we compared the AD and FTLD patient groups,
we obtained a specificity of 67% for CSF A␤ (1– 42) at a
sensitivity of 85% (448 ng/L). For CSF A␤ (N– 42), the
specificity was 75% at a sensitivity of 87% (373 ng/L). The
AUCs for CSF A␤ (N– 42) and CSF A␤ (1– 42) tended to be
different [Fig. 1B; 0.87 (76 – 0.97) and 0.77 (0.64 – 0.90); P ⫽
0.045].
The AUCs for CSF A␤ (1– 42) and CSF A␤ (N– 42) in
distinguishing FTLD patients from controls were significantly different [Fig. 1C; 0.69 (0.55– 0.81) and 0.54 (0.39 –
0.67); P ⫽ 0.007], but the discriminatory value was small
for A␤ (1– 42) and negligible for A␤ (N– 42), with the
confidence interval for the AUC including 0.5.
We found no significant correlation of either CSF
Table 1. Demographic data and CSF analyses for each diagnostic category.a
P
Age, years
Sex, M/F
Duration of disease, years
MMSE score
A␤ 1–42, ng/L
A␤ N–42, ng/L
A␤ 1–42/A␤ N–42
Albumin ratio
a
AD (n ⴝ 39)
FTLD (n ⴝ 24)
62 (52–79)
20/19
4 (1–11)
20 (3–28)
315 (140–626)
288 (116–674)
1.1 (0.5–1.7)
4.8 (2.0–10.6)
63 (49–85)
16/8
3 (1–11)
24 (3–29)
495 (202–1087)
588 (150–1324)
0.9 (0.4–1.3)
5.3 (1.5–17.3)
Controls (n ⴝ 30)
64 (32–79)
14/16
30 (25–30)
651 (337–1224)
629 (218–1075)
1.0 (0.6–2.6)
5.2 (2.8–18.5)
AD vs FTLD
AD vs controls
FTLD vs controls
0.58
0.26
0.054
0.02
⬍0.001
⬍0.001
0.001
0.6
0.14
0.90
0.66
0.41
⬍0.001
⬍0.001
⬍0.001
0.24
0.47
⬍0.001
0.02
0.66
0.07
0.99
Values are the median (minimum–maximum). P values refer to statistical difference between AD vs FTLD, AD vs controls, or FTLD vs controls.
1059
Fig. 1. ROC curves comparing A␤ (1– 42) (thick line) with A␤ (N– 42) (thin line) in AD vs controls (A), AD vs FTLD (B), and FTLD vs controls (C).
A␤ (1– 42) or A␤ (N– 42) with albumin ratio, MMSE score,
age, or disease duration (AD and FTLD) in either group.
The absolute concentrations of CSF A␤ (1– 42) and
A␤ (N– 42) were comparable. However, in earlier studies,
concentrations of CSF A␤ (N– 42) ranged from 36 to 623
ng/L in AD patients and from 111 to 629 ng/L in controls
(8, 17, 18 ). The reason for the low CSF A␤ (N– 42) concentrations measured in these studies could be a difference in
the affinity of the A␤ (N– 42) polyclonal antiserum samples or the purity and solubility of the peptides used as
calibrators (8 ). The sensitivity of an ELISA depends
largely on the binding characteristics of the antigen,
which may vary with temperature and buffer solutions, or
among different reagent lots (6 ). In addition, the affinity
of the antibodies used in the assays might vary for the
various A␤ 42 peptides involved in the pathogenesis of
AD, including oligomers of the A␤ 42 peptide. A future
study exchanging calibrators and antibodies among various ELISAs is necessary for harmonization.
ROC curve analysis revealed no difference in the ability
of the 2 assays to differentiate AD patients from controls.
In addition to the C-terminal heterogeneity, various Nterminally truncated peptides are found in the A␤ pools
of AD brains (19, 20 ). These peptides are considered to
play a role in the increased A␤ 42 production in developing AD. We speculate that A␤ (1– 42) and A␤ (N– 42)
concentrations go hand in hand at a certain stage of
disease, in mild to moderate AD as well as in controls.
Because N-terminally truncated A␤ 42 peptides can be
demonstrated early in the disease process (9 ), they might
be promising markers for the preclinical diagnosis of AD,
when used simultaneously with A␤ (1– 42) (21 ).
Several authors found decreased A␤ (1– 42) in CSF from
a subset of FTLD patients (3, 22 ). Very little information is
available about the CSF A␤ (N– 42) concentration in FTLD
(17 ). The reason for a decrease in CSF A␤ (1– 42) in FTLD
is unknown, although there might be a relationship with
the presence of an ⑀4 allele or with age (23 ). Interestingly,
a few studies have shown the involvement of 3 mutations
in the presenilin 1 gene (PSEN1) in familial forms of FTLD
(24 –26 ). These possible “loss of function” PSEN1 muta-
tions might act as inhibitors of the ␥-secretase cleavage of
amyloid precursor protein (27 ), leading to a decrease of
A␤ (1– 42) in the brain. Although most FTLD patients
included in our study had the sporadic form of FTLD, we
cannot exclude the possibility of a mutation in the PSEN1
gene in some of them.
References
1. Blennow K, Hampel H. CSF markers for incipient Alzheimer’s disease.
Lancet Neurol 2003;2:605–13.
2. Sunderland T, Linker G, Mirza N, Putnam KT, Friedman DL, Kimmel LH, et al.
Decreased ␤-amyloid1– 42 and increased tau levels in cerebrospinal fluid of
patients with Alzheimer disease. JAMA 2003;289:2094 –103.
3. Schoonenboom NS, Pijnenburg YA, Mulder C, Rosso SM, Van Elk EJ, Van
Kamp GJ, et al. Amyloid ␤(1– 42) and phosphorylated tau in CSF as markers
for early-onset Alzheimer disease. Neurology 2004;62:1580 – 4.
4. Schoonenboom NS, Mulder C, Vanderstichele H, Van Elk EJ, Kok A, Van
Kamp GJ, et al. Effects of processing and storage conditions on CSF amyloid
␤(1– 42) and tau concentrations: implications for use in clinical practice. Clin
Chem 2005;51:189 –95.
5. Jensen M, Schröder J, Blomberg M, Engvall B, Pantel J, Ida N, et al.
Cerebrospinal fluid A␤42 is increased early in sporadic Alzheimer’s disease
and declines with disease progression. Ann Neurol 1999;45:504 –11.
6. Vanderstichele H, Van Kerschaver E, Hesse C, Davidsson P, Buyse MA,
Andreasen N, et al. Standardization of measurement of ␤-amyloid (1– 42) in
cerebrospinal fluid and plasma. Amyloid 2000;7:245–58.
7. Olsson A, Vanderstichele H, Andreasen N, De Meyer G, Wallin A, Holmberg
B, et al. Simultaneous measurement of ␤-amyloid(1– 42), total tau, and
phosphorylated tau (Thr181) in cerebrospinal fluid by the xMAP technology.
Clin Chem 2005;51:336 – 45.
8. Mehta PD, Pirttila T, Mehta SP, Sersen EA, Aisen PS, Wisniewski HM.
Plasma and cerebrospinal fluid levels of amyloid ␤ proteins 1– 40 and 1– 42
in Alzheimer disease. Arch Neurol 2000;57:100 –5.
9. Sergeant N, Bombois S, Ghestem A, Drobecq H, Kostanjevecki V, Missiaen
C, et al. Truncated ␤-amyloid peptide species in pre-clinical Alzheimer’s
disease as new targets for the vaccination approach. J Neurochem 2003;
85:1581–91.
10. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM.
Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work
Group under the auspices of Department of Health and Human Services
Task Force on Alzheimer’s Disease. Neurology 1984;34:939 – 44.
11. Neary D, Snowden JS, Gustafson L, Passant U, Stuss D, Black S, et al.
Frontotemporal lobar degeneration: a consensus on clinical diagnostic
criteria. Neurology 1998;51:1546 –54.
12. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical
method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189 –98.
13. Passing H, Bablok W. A new biometrical procedure for testing the equality of
measurements from two different analytical methods. Application of linear
regression procedures for method comparison studies in clinical chemistry,
part I. J Clin Chem Clin Biochem 1983;21:709 –20.
14. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10.
1060
Technical Briefs
15. Consensus report of the working group on: “Molecular and biochemical
markers of Alzheimer’s Disease”. The Ronald and Nancy Reagan Research
Institute of the Alzheimer’s Association and the National Institute on Aging
Working Group. Neurobiol Aging 1998;19:109 –16.
16. Hanley JA, McNeil BJ. A method of comparing the areas under receiver
operating characteristic curves derived from the same cases. Radiology
1983;148:839 – 43.
17. Tapiola T, Pirttila T, Mehta PD, Alafuzoff I, Lehtovirta M, Soininen H.
Relationship between Apo E genotype and CSF ␤-amyloid (1– 42) and tau in
patients with probable and definite Alzheimer’s disease. Neurobiol Aging
2000;21:735– 40.
18. Mehta PD, Pirttila T, Patrick BA, Barshatzky M, Mehta SP. Amyloid ␤ protein
1– 40 and 1– 42 levels in matched cerebrospinal fluid and plasma from
patients with Alzheimer disease. Neurosci Lett 2001;304:102– 6.
19. Li R, Lindholm K, Yang LB, Yue X, Citron M, Yan R, et al. Amyloid ␤ peptide
load is correlated with increased ␤-secretase activity in sporadic Alzheimer’s
disease patients. Proc Natl Acad Sci U S A 2004;101:3632–7.
20. Lee EB, Skovronsky DM, Abtahian F, Doms RW, Lee VMY. Secretion and
intracellular generation of truncated A␤ in ␤-site amyloid-␤ precursor proteincleaving enzyme expressing human neurons. J Biol Chem 2003;278:4458 –
66.
21. Sergeant N, Kostanjevecki V, Casas K, Ghestem A, Grognet P, Drobecq H,
et al. Amino-truncated A␤2 species as early diagnostic and etiological
biomarkers of Alzheimer’s disease [Abstract]. Neurobiol Aging 2004;
25(Suppl 2):3.
22. Riemenschneider M, Wagenpfeil S, Diehl J, Lautenschlager N, Theml T,
Heldmann B, et al. Tau and A␤ 42 protein in CSF of patients with
frontotemporal degeneration. Neurology 2002;58:1622– 8.
23. Mann DM, McDonagh AM, Pickering-Brown SM, Kowa H, Iwatsubo T. Amyloid
␤ protein deposition in patients with frontotemporal lobar degeneration:
relationship to age and apolipoprotein E genotype. Neurosci Lett 2001;304:
161– 4.
24. Dermaut B, Kumar-Singh S, Engelborghs S, Theuns J, Rademakers R,
Saerens J, et al. A novel presenilin 1 mutation associated with Pick’s
disease but not ␤-amyloid plaques. Ann Neurol 2004;55:617–26.
25. Tang-Wai D, Lewis P, Boeve B, Hutton M, Golde T, Baker M, et al. Familial
frontotemporal dementia associated with a novel presenilin-1 mutation.
Dement Geriatr Cogn Disord 2002;14:13–21.
26. Raux G, Gantier R, Thomas-Anterion C, Boulliat J, Verpillat P, Hannequin D,
et al. Dementia with prominent frontotemporal features associated with
L113P presenilin 1 mutation. Neurology 2000;55:1577– 8.
27. Amtul Z, Lewis PA, Piper S, Crook R, Baker M, Findlay K, et al. A presenilin
1 mutation associated with familial frontotemporal dementia inhibits gamma-secretase cleavage of APP and notch. Neurobiol Dis 2002;9:269 –73.
Observations on Heat/Humidity Denaturation of Enzymes in Filter-Paper Blood Spots from Newborns,
Dennis E. Freer (Pediatrix Screening, Inc., 90 Emerson Ln.,
Suite 1403, Bridgeville, PA 15017; fax 412-220-0784, e-mail
Dennis_Freer@pediatrix.com)
Use of filter-paper blood spots from newborns for screening of inborn errors may include the assay of biotinidase
(EC 3.5.1.12; BIO), galactose-1-phosphate uridyltransferase (EC 2.7.7.12; UT), and glucose-6-phosphate dehydrogenase (EC 1.1.1.49; G6PD). There has been anecdotal
reference to heat and/or humidity denaturation of enzymes in filter-paper blood spots exposed to the elements
during storage or during transit to the laboratory (1 ), but
no quantitative description of the effects. Understanding
the phenomenon may lead to measures to identify denatured samples and prevent incorrect reporting of abnormal results.
We quantified filter-paper blood-spot enzyme values
for all samples collected during 3 months of the year,
February, July, and October, for a large region of Pennsylvania. The population means and SD for each enzyme
during each month were determined, and the data were
analyzed for seasonal effects.
We also performed a controlled experiment with bloodspot filter papers stored under different conditions of heat
and humidity to assess their relative influence on the
activities of the enzymes of interest. A blood sample (⬃15
mL) was drawn from an adult volunteer into a heparincontaining tube and mixed by inversion; blood was then
spotted on a series of Schleicher & Schuell 903 filter
papers to simulate newborn collections. Approximately
40 spots were applied, with occasional tube inversions, to
provide enough sample spots for serial testing, in duplicate, of a variety of environmental conditions. All samples
were dried for ⬃4 h at room temperature in air. Within
the next 2 h, time 0 samples were punched and then
assayed for BIO, UT, and G6PD activity. Filter papers
were then stored for 3 days under various conditions of
temperature, humidity, and exposure to air. The 4 temperature conditions used were as follows: freezing
(⫺20 °C), refrigeration (4 °C), room temperature (21 °C),
and 35 °C. Humidity conditions tested were ambient
humidity (⬃30%) and high humidity (samples stored in
containers with moisture present).
All 3 Astoria-Pacific SPOTCHECK procedures used
were modified from previous methods (2– 4 ). For each
assay, a 0.32-cm (1/8-inch) punch from each newborn
blood spot on Schleicher & Schuell 903 filter paper was
placed in a microtiter well, as were appropriate controls.
Spots were eluted according to the manufacturer’s instructions. The Astoria Pacific continuous flow system
was used for each enzyme, and manufacturer specifications were followed. Results for BIO are expressed as
enzyme response units (ERU), where 1 ERU ⫽ 1 ␮mol/dL
p-aminobenzoic acid produced over the course of 90 min
from the biotin–p-aminobenzoic acid substrate. For UT
and G6PD, the measured end product was NADPH
fluorescence, which was compared with diluted NADH
calibrators, and for both, results are expressed in ␮mol/L
NADH.
The changes in mean enzyme activities and percentage
changes from the February data for BIO, UT, and G6PD
for all results on samples received in February, October,
and July are shown in Table 1. The means for all 3
enzymes were highest in February, intermediate in October, and lowest in July. The percentage change from
February to July was largest for G6PD (⫺38%) and
smallest for UT (⫺23%). All changes except for the mean
UT values from February to October were statistically
significant (P ⫽ 0.005). The population mean data summarized in Table 1 show that for the 3 enzymes evaluated,
there are seasonal differences in activities, with the lowest
means for all 3 enzymes occurring in summer. There is no
documentation of seasonal biological variation for any of
these enzymes; therefore, the observed differences are
likely attributable to denaturation in summer.
The results for the single-sample in-house controlledenvironment study evaluating the effects of continuous
1061
Table 1. Seasonal changes in population mean activities of blood-spot BIO, UT, and G6PD in newborn samples for a large
region of Pennsylvania.
BIO
February
October
July
a
UT
G6PD
n
Mean, ERU
Change, %
n
Mean, ␮mol/L
Change, %
n
Mean, ␮mol/L
Change,a %
2131
3164
3121
51
44
37
⫺16
⫺26
2776
3165
3197
255
253
196
0
⫺23
2768
3164
3170
165
139
103
⫺16
⫺38
a
a
Percentage changes are relative to the February activity mean for each enzyme.
exposure of filter-paper blood spots to different conditions of heat and humidity are shown in Fig. 1. We
measured the enzyme activities for BIO (Fig. 1A), UT (Fig.
1B), and G6PD (Fig. 1C) on day 0, then for 3 consecutive
days after continuous exposure to various conditions of
temperature and humidity. The day 0 point for each
graph represents the mean of 4 data points. All other data
points are mean values of duplicate determinations and
are expressed as percentage change from the day 0 value.
Refrigerated and frozen samples for all 3 enzymes had the
least degradation (data not shown), with no data point for
any of the 3 enzymes on any day having an activity loss
⬎16%. For all samples under all other conditions, the
largest decrease in enzyme activities occurred on day 1,
followed by relatively minor changes. For samples stored
at room temperature, degradation was 7%–18% for BIO
and 18%–30% for UT and G6PD. Samples kept at 35 °C or
at room temperature under high humidity showed progressive losses of activity by day 3 of ⬎60% for UT and
G6PD, but only 30%–50% for BIO. For all 3 enzymes, the
greatest loss in activity occurred in samples stored at
35 °C under conditions of high humidity, with BIO activity lower by 60% on day 1 and residual activity only 10%
of the initial value by day 3. For UT, and particularly for
Fig. 1. Effect of storage under various conditions on BIO (A), UT (B),
and G6PD (C) activity in filter-paper blood spots.
Activity is shown relative to activity on day 0 (100%). Storage conditions: ⽧,
room temperature; ■, room temperature plus high humidity; Œ, 35 °C at
ambient humidity; F, 35 °C plus high humidity.
1062
Technical Briefs
G6PD, enzymatic activity was severely diminished on day
1 with an almost complete loss of activity by day 3. As
evident from the graphs in Fig. 1, the rate of denaturation
is dependent on temperature and humidity. The combined effect of 35 °C and high humidity for 3 days caused
a ⬎70% decrease in all enzyme activities. In all situations,
BIO appears to be the least denatured of the 3 enzymes.
Exposure of actual samples in transit to high heat and
humidity continuously for 3 days is unlikely, but in most
areas of the United States, samples awaiting pickup from
an outside mail deposit box could well be exposed to high
heat and humidity conditions for several hours on a July
afternoon, and perhaps for 2 afternoons over a weekend.
Obviously, different climates will produce varying but
predictable effects because of local seasonal weather variations.
Awareness that all of the enzymes are denatured to
some extent led us to establish cutoffs for each enzyme
below which heat denaturation might be a factor: For BIO,
we used the cutoff of 28 ERU; for UT, 125 ␮mol/L, and for
G6PD, 100 ␮mol/L. Thus, if a sample has a BIO value of
13 ERU (reference interval, 28 –90 ERU), a UT value of 104
␮mol/L (reference interval, 155–389 ␮mol/L), and a
G6PD value of 58 ␮mol/L (reference interval, 90 –350
␮mol/L), then there is a high degree of suspicion that the
sample was heat denatured in transit.
In practice, reviewing data for heat denaturation of
samples is more useful for BIO than for G6PD or UT.
Because G6PD is the most sensitive indicator, there is a
“canary effect”, i.e., G6PD is a good indicator of heat
denaturation of the other enzymes, but not vice versa. In
summer, however, we detect ⬃15 samples per month
with G6PD below the critical activity cutoff of 25 ␮mol/L
in which BIO and UT values are ⬍28 ERU and 125
␮mol/L, respectively. These results are not reported as
“positive” but rather as “unacceptable due to possible
heat denaturation”, and a repeat is requested. For UT, we
almost never see an activity value ⬍40 ␮mol/L, even in
summer, except in true galactosemia; therefore, evaluation for heat denaturation is rarely an issue.
The critical cutoff value for BIO (16 ERU for partial
deficiency) is close to the lower 2 SD limit of the reference
interval (28 –90 ERU). As a result, heat-denatured samples
often have values below the cutoff. During the winter, we
find fewer than 7 samples per month with a biotinidase
activity ⱕ16 ERU (true values ⬎8 but ⬍16 ERU are
reported as inconclusive, and a repeat is requested). In
summer, we see 40 –50 samples per month with BIO
values in the inconclusive range. In a limited review of
DNA mutations in a random selection of 18 heat-denatured cases, 4 had wild-type DNA, 4 had 2 copies of
D444H, 5 had 1 copy of D444H, and 5 had 1 copy of
complete deficiency mutations (data not shown). In winter, these samples would likely have BIO activities of
17–27 ERU. Thus, many heat-denatured samples in summer are clinically benign partial BIO deficiencies, which
because of a small activity loss fall into the inconclusive
range and could be misidentified as clinically significant.
The BIO result in these cases is reported as “unacceptable
due to possible heat denaturation”, and a repeat is requested.
Without review of all enzyme results, these samples
would have been reported as inconclusive with a request
for a repeat. The possible implication of an inconclusive
result may cause some anxiety for the parents of newborns. The more correct report, that the sample was
compromised and tests should be repeated, is less alarming. Frequent receipt of heat-denatured samples from 1
location may also suggest that sample handling procedures need to be examined.
References
1. Wolf B, Heard GS, Jefferson LG, Weissbecker KA, Secor McVoy JR, Nance WE,
et al. Newborn screening for biotinidase deficiency. In: Carter TP, Wiley AM,
eds. Genetic diseases: screening and management. New York: Alan R. Liss,
Inc., 1986:175– 82.
2. Wolf B, Heard GS, Weissbecker KA, Secor McVoy JR, Grier RE, Leshner RT.
Biotinidase deficiency: initial clinical features and rapid diagnosis. Ann Neurol
1985;18:614 –7.
3. Sturgeon P, Beutler E, McQuiston D. Automated method for screening
galactosemia. In: Skeggs LT Jr, ed. Automation in analytical chemistry,
Technicon Symposia, Vol. 1. White Plains, NY: Mediad, Inc., 1966:75–7.
4. Miwa S, Kanai M, Nomoto S. Use of the autoanalyzer for determination of
erythrocyte pyruvate kinase, glucose-6-phosphate dehydrogenase and cholinesterase. Br J Haematol 1967;13:54 – 60.
Improved HPLC Assay for Measuring Serum Vitamin C
with 1-Methyluric Acid Used as an Electrochemically
Active Internal Standard, Leslie F. McCoy, M. Bridgette
Bowen, Mary Xu, Huiping Chen, and Rosemary L. Schleicher*
(CDC, National Center for Environmental Health, Division of Laboratory Sciences, Mail Stop F18, Inorganic
Toxicology and Nutrition Branch, 4770 Buford Hwy, NE,
Atlanta, GA 30341-3724; * author for correspondence: fax
770-488-4139, e-mail RSchleicher@cdc.gov)
The National Health and Nutrition Examination Survey
(NHANES) laboratory at CDC has used a modification of
methods (1, 2 ) with electrochemical detection for measurement of serum vitamin C for the past 9 years. The
assay is relatively rapid, easy to perform, and gives good
precision. Quality-control (QC) materials have been kept
at ⫺70 °C for more than 10 years without substantial
degradation. A drawback of the method is the lack of an
internal standard to correct for analyte degradation, procedural errors, and detector drift. Significant vitamin C
degradation is intermittently encountered during the analytical process. Oxidation of ascorbic acid (AA) is accelerated by exposure to air, heat, light, and traces of copper
and iron (3 ) and may be introduced through contact with
unexpected sources, such as consumable supplies (4 ).
Detector drift is a characteristic of electrochemical detection and has been noted by others performing the serum
AA assay (5 ).
Our original HPLC assay used partition of largely
un-ionized AA and amperometric detection. A 25-cm C18
reversed-phase column is equilibrated with a mobile
phase consisting of monochloroacetic acid (pH 3.0) containing disodium EDTA and sodium octylsulfonate [originally used for ion pairing with catecholamines extracted
from adrenal chromaffin cells in a mixture containing AA
(1 )]. The AA in serum is stabilized by addition of metaphosphoric acid (MPA), reduced by addition of dithiothreitol (DTT), and then oxidized at ⫹650 mV referenced
to an Ag/AgCl electrode. The working electrode is a
thin-layer detector cell. When serum is treated as indicated, peaks for AA, uric acid (URIC), and DTT are
resolved and detected at the applied potential within ⬃18
min. The run time for each sample is shortened by
injecting the next sample before all peaks from the previous sample have eluted.
Several changes suggested themselves to modify this
method: (a) Improved column technology would allow
the use of a smaller column with smaller injection volumes and shorter retention times. (b) Sodium octylsulfonate could be eliminated because it does not effectively pair
with ascorbate. (c) A small amount of methanol in the
mobile phase would accelerate the elution and sharpen
peaks. (d) Calibrators could be prepared and frozen to
save daily preparation time and could be prepared in the
same fashion as the samples with the addition of internal
standard and other reagents to control for any errors in
handling and/or analyte degradation. (e) Longer runs
might be possible if an internal standard could be found
to adequately correct for analyte degradation and detector
drift.
We developed a revised method that uses an Agilent
1100 solvent delivery system connected in series to an
1100 diode array detector and a BAS electrochemical
detector set at ⫹650 mV. AA in serum was separated on
a YMC ODS-AQS-3 column [15 cm ⫻ 3 mm (i.d.); 3-␮m
particle size (120 Å)] with a 10-␮L injection volume of a
5-fold– diluted stabilized serum specimen. An Upchurch
0.5 mm stainless steel frit was used as a precolumn filter.
A mixture of 0.15 mmol/L monochloroacetic acid, 0.2
mmol/L disodium EDTA, and 150 mL/L methanol at pH
3.0 was used as a mobile phase at a flow rate of 0.3
mL/min. Stock solutions of AA were prepared gravimetrically. Three concentrations of calibrators (1.42–28.39
␮mol/L) representing 0.5, 3, and 10 ng on column were
diluted in 60 g/L MPA–2.5 mL/L DTT at pH 1.8 and
stored for up to 4 months at ⫺70 °C with minimal change
in values (0%–2% degradation). Once prepared, the highest calibrator was chromatographed, and a peak of 470
absorbance units ⫾ 2% at 245 nm was used as an
additional step to assess integrity. The internal standard,
1-methyluric acid (MURIC; Sigma Chemical Co), was
added to all samples and calibrators to achieve a final
concentration of 82.35 ␮mol/L. Assay calibration was
performed for each run. NHANES serum specimens were
prepared in the field as described previously (6 ). Peaks
were integrated by use of peak-area ratios of AA to
MURIC.
HPLC analysis using field-stabilized specimens showed
no interfering peaks. Blanks (reagents only) in each run
1063
showed no interfering peaks. AA, DTT, URIC, and
MURIC peaks in the specimens were identified by use of
calibrators and inspection of their retention times (Fig.
1A).
The optimum temperature to elute all constituents
within 12 min was 30 °C. All peaks were baseline separated. Increasing the mobile phase pH from 2.0 to 4.5
lengthened peak retention times, whereas increasing the
percentage of methanol shortened them. The retention
times of the analytes were stable with CVs ⬍12% over the
course of 6 months on a single column. At the time of
manuscript submission, 6 columns had been used in
routinely performing this assay over 14 months; fusion of
the AA peak with an earlier eluting peak was the primary
reason for retiring columns. The mean number of injections per column was 1401 (range, 676 –2196). On average,
all reagents for this assay were prepared monthly.
Calibration curves were linear up to 28.39 ␮mol/L,
which represents a final concentration of 141.95 ␮mol/L
in serum specimens (mean of daily calibrations: y ⫽
0.9897x ⫺ 0.0004; r2 ⫽ 1.0; SEregression ⫽ 0.0114; SEslope ⫽
0.0018; SEintercept ⫽ 0.0006). The limit of detection, estimated as 3 times the SD of a near-zero sample, was 0.68
␮mol/L, which represents 0.24 ng on column. The mean
Fig. 1. Typical chromatographic separation of AA from URIC, DTT
(reducing agent), and MURIC (internal standard; A), and Bland–Altman
difference plot for the 2 methods (B).
(A), the AA concentration in this patient’s serum sample was 57.3 ␮mol/L. (B),
the dashed line indicates the mean difference between methods [1.53 ␮mol/L
(2.6%)]; the dot-dashed lines indicate 2 SD. Conversion factor for AA: 1 mg/dL ⫽
56.78 ␮mol/L.
1064
Technical Briefs
(SD) recovery of AA added to serum at final concentrations ranging from 38.4 to 116.1 ␮mol/L was 97 (2)%
(Table 1 in the Data Supplement that accompanies the
clinchem.org/content/vol51/issue6/). Five specimens
with serum AA concentrations ⬎96.53 ␮mol/L were
diluted 0- to 7-fold with 60 g/L MPA–2.5 mL/L DTT. The
ratios of observed to expected results were 0.93–1.07.
The mean intraassay CVs for 5 samples (38.61– 60.76
␮mol/L) and 3 pools (24.42–55.08 ␮mol/L) processed in
10 –20 replicates in 1 run ranged from 0.6% to 3.6% (Table
2 in the online Data Supplement). The mean interassay
CVs for 5 samples (26.69 –57.92 ␮mol/L), each processed
in 1 replicate in 5 runs, ranged from 1.2% to 4.2% (Table 3
in the online Data Supplement). Over the course of 135
runs, 3 QC pools processed as duplicates in each run
showed CVs of the run means of 8% (low; 13.06 ␮mol/L),
4% (medium; 60.76 ␮mol/L), and 4% (high; 119.24 ␮mol/
L). The injection reproducibility was evaluated by use of
10 replicate injections using 3 separate QC pool preparations. The CV values of the means were ⱕ0.7% for all
pools.
The revised method was more accurate than our original method based on repetitive analysis of NIST standard
reference material SRM 970 Level I and II (recertified in
2004). The mean (SD) results for Level I [8.57 (0.84)
␮mol/L] were 102% of the target value and for Level II
[28.05 (1.34) ␮mol/L] were 100% of the target value
during 13–14 runs, compared with 93% and 92% of the
target values, respectively, obtained with the original
method in a similar number of runs. The CVs for these
results were also better with the revised method: 5%–10%
vs 7%–12% for the old method. Participation in 2 NIST
quality assurance exercises showed results in good agreement with consensus medians.
The mean number of injections per run (1 injection per
sample) during the evaluation period was 57 (range,
19 –94). Repeated analysis of low, medium, and high QC
pools gave CVs of the individual results of 6% for AA
when an internal standard was used. Quantification without an internal standard gave QC results with lower
values and slightly higher CVs (8%). The internal standard increased the mean values for the QC pools by
3%–5%. Although the internal standard compensated for
detector drift during runs, it was more stable than AA.
The mean (SD) decrease in peak area in QC pools measured at the beginning and end of each run was 5 (6)% for
AA and 0.5 (5)% for MURIC. Runs with significant drift
(⬎10% fall-off of AA signal) were more likely to be in
control when quantified with use of an internal standard.
Deming regression comparison between the original
and revised assays for 308 specimens in 10 separate assays
gave the following results: y ⫽ 1.06x ⫺ 1.9 ␮mol/L (Sy兩x ⫽
4.29 ␮mol/L; R2 ⫽ 1.00; n ⫽ 308). The 95% confidence
interval for the y-intercept was ⫺3.08 to ⫺0.79 ␮mol/L,
and the confidence interval for the slope was 1.04 –1.07.
Results in this data set spanned the reportable range
(3.41–194.76 ␮mol/L). Bland–Altman analysis showed a
small mean difference of 1.53 ␮mol/L (95% confidence
interval, 1.02–2.04 ␮mol/L) for the revised method (Fig.
1B). Regression analysis of the percentage difference
between methods as a function of AA concentration
showed that the difference increased with increasing
concentration (y ⫽ 0.055x ⫺ 0.03; R2 ⫽ 0.13; P ⬍0.0001 for
the slope). We anticipated a shift toward more positive
values with the revised method attributable to (a) the use
of an internal standard to correct for detector drift and (b)
processing of the calibrators as though they were unknowns. We did not expect a concentration-dependent
difference from these 2 changes and do not have an
explanation for the larger difference in the high concentration range.
Other assay conditions of interest were also investigated. A comparison set of 29 samples, selected because of
substantially different values obtained with the 2 methods
[mean (SD), 13 (1)% higher with the revised method],
were separated chromatographically with and without
the ion-pairing reagent in the mobile phase. The AA
results differed by a mean (SD) of only 1 (2)%, demonstrating that the ion-pairing reagent provided no added
specificity. Integration of the AA peak area gave more
accurate and precise results than did peak height (data
not shown). Because dilution of specimens led to losses of
up to 30%, smaller sample volumes are recommended
when re-measuring samples with unusually high AA
values, i.e., greater than the NHANES III 99th percentile
(123 ␮mol/L). Subjecting specimens to 1 freeze–thaw
cycle generally did not lead to AA degradation. Only 1 in
34 sets of QC pools showed significant degradation of AA
(⬎15% loss) after a single freeze–thaw cycle.
In summary, the development of an HPLC method that
includes an internal standard improves the precision and
accuracy of AA measurement and compensates for detector drift so that longer runs can be accommodated. Other
changes have enhanced performance of the new assay.
References
1. Herman HH, Wimalasena K, Fowler LC, Beard CA, May SW. Demonstration of
the ascorbate dependence of membrane-bound dopamine ␤-monooxygenase
in adrenal chromaffin granule ghosts. J Biol Chem 1988;263:666 –72.
2. Margolis SA, Davis TP. Stabilization of ascorbic acid in human plasma, and its
liquid-chromatographic measurement. Clin Chem 1988;34:2217–23.
3. Daubert TE, Danner RP. Physical and thermodynamic properties of pure
chemicals: data compilation. Washington, DC: Taylor and Francis, 1989.
4. Margolis SA, Park E. Stability of ascorbic acid in solutions stored in
autosampler vials. Clin Chem 2001;47:1463– 4.
5. Grun M, Loewus FA. Determination of ascorbic acid in algae by highperformance liquid chromatography on strong cation-exchange resin with
electrochemical detection. Anal Biochem 1983;130:191– 8.
6. Gunter EW, Lewis BG, Koncikowski SM. Laboratory procedures used for
the Third National Health and Nutrition Examination Survey (NHANES III),
1988 –1994 http://www.cdc.gov/nchs/data/nhanes/nhanes3/cdrom/nchs/
manuals/labman.pdf (accessed December 2004).

Evaluation of a New CA15-3 Protein Assay Method: Optical Protein

Transcription

Similar documents

PPCO Twist System

Plasma Thawer - kizlonmedical.com

The StethoSCOOP - University of Cincinnati College of Medicine

Solid-phase extraction and ion-pair reversed-phase - E

Zink_Spring 2014 (dragged)

here - Whole Foods Market

Here - BIOEFFECT

BioLife Plasma Center in American Fork “Apply Now”