“What is abnormal ALT? ” Prof Seng Gee Lim

Transcription

“What is abnormal ALT? ” Prof Seng Gee Lim
“What is abnormal ALT?”
Prof Seng Gee Lim
Co-Chairman
Singapore Hepatitis Conference
ALT: Biologic Variability
• Intra-individual variation
• Diurnal variation: up to 45%
• Highest in the afternoon
• Lowest at night
• Day-to-day variation: 10-30%
• Variability other than liver disease
• Gender
• Age
• Exercise: Moderate exercise↓
• BMI- correlation not clearly understood, but may be due
to:
• ALT isoenzyme made in adipose tissue
• NAFLD
Dufour DR, et al. Clin Chem. 2000;46:2027-2049.
Yang RZ, et al. Genomics. 2002;79: 445-450.
Gender and Age Affect Upper
Reference Limits for ALT
Males
Females
Dufour DR, et al. Clin Chem. 2000;46:2027-2049.
ALT Measurement Limitations
•ALT correlated with numerous clinical
factors
• 1033 blood donors with negative serology
Variable
Pearson Correlation
P Value
Age
0.126
.0001
BMI
0.345
.0001
Triglycerides
0.223
.0001
Cholesterol
0.167
.0001
Bilirubin
0.115
.0002
Iron
0.084
.006
Uric Acid
0.355
.0001
GGT
0.549
.0001
Piton A, et al. Hepatology.1998;27:1213-1219.
ALT Measurement Limitations
• ALT associated with several demographic
factors
Variable
Regression
Coefficient
P Value
Male
0.118
.0001
BMI
0.177
.0001
Age
0.012
.02
Smoker
-0.242
.03
Piton A, et al. Hepatology.1998;27:1213-1219.
ALT Measurement Limitations
• Definition of normal values
• Statistical definition
• Standard practice to define normal lab values
• Middle 95% of healthy volunteers
• Influenced by reference population
• Abnormal: >97.5th percentile
• Biological definition
• Based on risk of developing disease/complication
• Cholesterol, blood sugar
• Abnormal: ALT level associated with disease
• Requires standardization of laboratory assays
ALT Measurement Limitations:
What Type of Variability Exists Between Labs?
• In 1997 College of American Pathologists survey,
average variation in aminotransferase values between
laboratories using the same methods,1 4–9%
• Between laboratories using different assays, range
was 39–85 U/L for the same specimen
• Multi-enzyme control sera (eg, Seraclear-HE) designed
as calibrator to be used in different labs2
• Study in Japan demonstrated <10% inter-laboratory
variation for ALT3
• Majority of Japanese laboratories use same
reference method (Japanese Society of Clinical
Chemistry method)
1. Dufour DR, et al. Clin Chem. 2000;46:2027-2049; 2. Eto A, et al. Clin Chem. 1995;41:872-880; 3.
Tatsumi N, et al. Southeast Asian J Trop Med Public Health. 1999;30(suppl 3):99-104.
ALT Measurement Limitations:
Conclusions
• Intra-individual variability in ALT
• Inter-individual variability other than liver disease
• Age, gender, BMI
• Debate about normal ALT
• Statistical approach
• Prati (bottom 95%): 30 (M) and 19 (F) U/L
• Piton (bottom 95%): 42–66 (M) and 31–44 (F) U/L
• Biological approach
• Increased liver mortality ?
• Problem of laboratory standardization
• Best evidence
• Prati method most reasonable to define absence of liver
disease
How good is ALT in diagnosing
liver damage?
Other Markers
Author
Hasanjani et al.
Int J Clin Pract. 2005;
59:791–794.
Myers et al.
J Hepatol. 2003;
39:222–230.
ter Borg et al.
N=
Test
HBeAg
status
91
AST
Negative
223
AST
GGT
actitest
Mixed
AST
Mixed
233
Lancet. 1998;
351:1914–1918.
AURO
C
0.82
p-value
NA
0.005
AST≈ALT
<0.0001 predicting
activity but NS
between tests
AST>ALT
<0.005
0.82
0.807
ROC curves (ter Borg et al.)
ROC curves (Myers et al.)
1.0
1.0
0.8
0.75
0.6
0.4
ALT (AUC=0.757)
AST (AUC=0.807)
AST, HBV DNA (AUC=0.832)
0.2
0
Sensitivity
Sensitivity
comparison
0.50
Actitest (0.82 ± 0.04)
AST (0.82 ± 0.04)
ALT (0.81 ± 0.04)
0.25
0
0
0.2
0.4
0.6
1-specificity
0.8
1.0
0
0.2
5
0.50
0.75
1-specificity
1.0
ALT and Prognosis
Baseline ALT and Liver Mortality
• n=141,108 (men=94,533, women=46,575) Korean Insurance clients
• Baseline variables: demographics, BP, glucose, BMI, smoking,
alcohol, self reported chronic illnesses and FHx
Baseline
ALT (IU/L)
Number
Death from liver
disease (n)
RR (95% CI)
<20
37,425
45
1 (0.7–1.4)
20-29
36,589
113
2.9 (2.4–3.5)
30-39
11,975
110
9.5 (7.9–11.5)
40-49
4,068
74
19.2 (15.3–24.2)
50-99
3,887
117
30.0 (25.0–36.1)
100
589
42
59.0 (43.4–80.1)
Multivariate Analysis: Men
•ALT
8.93 (7.49 to 10.64)
•Hypertension
1.91 (1.46 to 2.49)
•F. Glucose≥7
3.18 (2.29 to 4.42)
•Alcohol≥200g/d
3.33 (1.64 to 6.76)
•FHx liver disease 2.62 (1.93 to 3.56)
Women
27.32 (11.24 to 66.40)
3.63 (0.75 to 17.45)
NA
NA
Kim HC, et al. BMJ. 2004;328:983-988.
Is this a good study?
• ALT is not a
disease, it’s a
surrogate marker
• What is the cause
of raised ALT?
• Is raised ALT a
predictor of
prognosis in those
with known liver
disease?
Concerns with study:
• Limited baseline
variables
• No disease or
cause of raised
ALT
• No multivariate
analysis of
important (missing)
baseline variables
US National Health and Nutrition
Examination Survey (NHANES III)
• n=14,950 adults was examined 1988-1994
prospectively and mortality determined by
death certificate
• HBV and HCV were excluded serologically
• Abn ALT defined as M>30 U/L, F>19 U/L
Liver Disease
Mortality
Normal ALT
(n=12,794)
Elevated ALT
(n=2156)
n
HR (age adj)
HR (multivariate adj)
18
1.0
1.0
16
11.2 (3.1–40.5)
8.2 (2.1–319)
Ruhl, Gastroenterol 2009
Issues with NHANES III
• Small numbers of liver disease mortality
• Strong HR after multivariate adjustment but
wide confidence intervals
• The cause of liver death not ascertained
• However, abn ALT is a signal to investigate
further
Natural History of ALT for Patient Management:
The R.E.V.E.A.L. Study
• The R.E.V.E.A.L. study was a large prospective cohort study that
examined the risk of cirrhosis and HCC
• Elevated ALT (≥45 IU/L) at study entry was a significant univariate
risk factor for HCC development (HR 4.1 (2.8-6.0); P < 0.001)
• Elevated ALT levels were not an independent risk factor, unlike HBV
DNA levels, which were the strongest predictor of HCC risk
• ALT U/L ≥45 adjusted HR = 1.1 (0.7-1.7)
No. (%) of
Participants
(N=3653)
PersonYears of
Follow-up
No. of HCC
cases
Incidence Rate
Per 100,000
Person-Years
Crude HR (95% CI)*
3435 (94)
39,469
133
337
1.0
218 (6)
2310
31
1342
4.1 (2.8–6.0)
873 (24)
10,154
11
108
1.0
300–9999
1161 (32)
13,518
15
111
1.0 (0.5–2.2)
0.96
10,000–99,999
643 (18)
7404
22
297
2.7 (1.3–5.6)
0.006
100,000–999,999
349 (9)
3845
37
962
8.9 (4.6–17.5)
<0.001
≥1 million
627 (17)
6858
79
1152
10.7 (5.7–20.1)
<0.001
Level of ALT,U/L
<45
≥45
HBV DNA, copies/mL
<300 (Undetectable)
* Cox proportional hazard models were used
Chen CJ, et al. JAMA 2006; 295:65–73.
P-value
<0.001
Natural History of ALT for Patient Management:
The R.E.V.E.A.L. Study
• Elevated ALT (≥45 IU/L) levels at study entry were an independent
predictor of risk of cirrhosis
• Multivariate adjusted RR = 1.5 (P < 0.05)
• HBV DNA levels >106 copies/mL at study entry were associated with
the greatest risk of cirrhosis
Multivariate-adjusted relative risk (95% CI)*
All patients (N=3582)
HBeAg-negative patients (N=3037)
1.0 (ref)
1.0 (ref)
1.5 (1.1–2.1)†
1.6 (1.0–2.6)
1.0 (ref)
1.0 (ref)
300–9999
1.4 (0.9–2.2)
1.4 (0.9–2.1)
10,000–99,999
2.5 (1.6–3.8)‡
2.4 (1.5–3.7)‡
100,000–999,999
5.6 (3.7–8.5)‡
5.4 (3.5–8.3)‡
≥1 million
6.5 (4.1–10.2)‡
6.7 (4.1–11.0)‡
Level of ALT,U/L
<45
≥45
HBV DNA, copies/mL
<300 (Undetectable)
* Multiple Cox proportional hazard regression analysis - adjusted for age, sex, cigarette smoking, alcohol consumption,
HBeAg status
† P<0.05; ‡ P<0.001
Iloeje UH, et al. Gastroenterol. 2006; 130:678–686.
Natural History of ALT for Patient Management:
The R.E.V.E.A.L. Study
• ALT status at baseline is a stronger predictor of liver disease
mortality in HBsAg(-) participants than HBsAg(+) participants
ALT level, IU/L
Adjusted RR (95% CI)*
Among HBsAg(-) subjects
(N=19,665)
Among HBsAg(+) subjects
(N=3,653)
1–19
1.0
1.0
20–29
2.5 (1.4–4.4)†
1.2 (0.7–1.9)
30–39
3.4 (1.6–7.0)†
2.3 (1.4–4.0)‡
40–49
9.8 (4.9–19.6)†
1.6 (0.8–3.5)
50–99
8.5 (4.3–16.4)†
3.0 (1.8–5.1)†
≥100
31.4 (14.7–66.9)†
1.3 (0.5–3.6)
* In the HBsAg(-) group, gender, age, cigarette smoking, alcohol consumption and waist:hip ratio were adjusted.
In the HBsAg(+) group, gender, age, cigarette smoking, alcohol consumption, HBeAg status, liver cirrhosis at cohort
entry and family history of liver disease were adjusted
† P<0.001; ‡ P<0.01
• In HBsAg(+) participants HBV DNA and liver cirrhosis were stronger
predictors of liver disease
Jen CL, et al. J Hepatol. 2006; 44(S2):S173. Abstract 465.
Critique of the R.E.V.E.A.L. study
• Analysis of data and prediction based on baseline values
- only a single time point
• ALT can fluctuate especially in HBeAg-negative disease
• ALT as a risk factor for HCC development
• Univariate HR= 4.1 (95% CI, 2.8–6.0)
• Multivariate HR=HR = 1.1 (95% CI, 0.7-1.7)
• This means that after adjusting for other important
variables eg HBV DNA, ALT is no longer significant
• ALT seems to be a more important risk factor in non-HBV
cases…..but runs into same problems: no multivariate
analysis of other important risk factors
• Eg if we suspect NASH then we should evaluate
severity of DM, lipid levels, etc and include these
ALT and Hepatic Complications

Stratified:
ALT <0.5 x ULN
ALT >0.5-1 x ULN
ALT >1-2 x ULN
ALT >2-6 x ULN
ALT >6 x ULN
30
(%) Risk of Complications
3,233 Chinese CHB patients
Cox proportional hazards model:
•
20
•
10
For complications
– Male gender
– Stigmata of liver disease
– Increasing age
– Low albumin at baseline
– High AFP at baseline
For survival
– Male gender
– Increasing age
– Hepatitis symptoms
– Low albumin at baseline
0
0
30
60
90
120
150
180
Months of Follow-up
Yuen MF, et al. Gut. 2005; 54:1610–1614.
ALT and Hepatic Complications

Stratified:
ALT <0.5 x ULN
ALT >0.5-1 x ULN
ALT >1-2 x ULN
ALT >2-6 x ULN
ALT >6 x ULN
30
(%) Risk of Complications
3,233 Chinese CHB patients
Concerns about conclusions of
this study:
•
•
•
20
•
10
No baseline ultrasound
except in those with AFP
Liver complications based on
single ALT value at baseline
Not data on whether ALT was
persistently normal or
intermittently
Multivariate analysis does
not show ALT as an
independent risk factor
0
0
30
60
90
120
150
180
Months of Follow-up
Yuen MF, et al. Gut. 2005; 54:1610–1614.
REVEAL – HCV
• 89,293 community based persons
participated in a prospective study for
HCV.
• n=925 HCV+ were followed for >10y for
HCC
Lee, J Clin Oncol 2010
Conclusions
• The normal range for ALT is the subject of
considerable debate
• Normal ranges can be calculated by large cross
sectional population studies and generally has been
around 30 U/L for men and 19 U/L for women
• However, it has not been established that those
falling outside the normal range are at risk in a
properly conducted multivariate analysis
• In those with HBV, ALT≥45 were not found to be
associated with increased riskof HCC in 2 large
cohort studies, but there is a small risk of cirrhosis
• In HCV only levels≥45 are associated with HR~4.5
of HCC development
• ALT is not a disease, and as a surrogate marker, it
needs to be validated within that disease