What is a “normal”value for serum ALT activity?

WHAT IS A “NORMAL”
VALUE FOR SERUM ALT
ACTIVITY?
D. Robert Dufour, MD, FACB, FCAP
Consultant Pathologist
Attending, Liver Clinic
VAMC, Washington DC
Emeritus Professor
of Pathology
DISCLOSURE
• I have no conflicts to disclose
REFERENCE INTERVALS
• The standard approach for laboratory test
reference intervals involves selecting
asample of the (apparently healthy)
population
• The central 95% of values (after excluding
statistical outliers) is considered the reference
interval (mean ± 1.96 sd if data has a
Gaussian distribution)
REFERENCE INTERVALS
• Minimum requirement for laboratories is to
validate that the range they use is appropriate
for their population (often use what is
suggested by manufacturer)
• Establishing reference interval requires many
apparently healthy volunteers (for tests such
as ALT, minimum of 400); validation only
requires 20
PROBLEMS
• There are several problems with this
approach
• How well did the manufacturer select
appropriate individuals?
• How well does “apparently healthy”
exclude clinically inaparent disease (a
major problem with liver disease)
• Is “typical” actually “healthy”?
ALT REFERENCE LIMITS
• Two surveys have evaluated upper reference
limits used by laboratories
• Neuschwander-Tetri (Arch Intern Med 2008)
surveyed 11 labs, found ULN varied from 35
to 79 U/L in men and 31 to 55 in women
• Dutta (Hepatology 2009) surveyed 67 labs in
Indiana; found ULN varied from 31-70 (with
most labs having same limits for men and
women)
ALT REFERENCE LIMITS
• In both surveys, found little difference based
on instrument or method used
• In Dutta survey, 79% based reference limits
on manufacturer’s suggesetions; in only half
were “normal” volunteers also tested (only an
average of 25 individuals tested)
HOW COMPARABLE ARE
ALT METHODS?
• Assays measure enzyme activity; changes in
reaction conditions change actual values
• Main difference is reaction temperature, not
always standardized in different labs
• Although there is a standardized IFCC
method using pyridoxal-5´-PO4, most labs do
not use this because of stability, measurement
issues; in our studies, little difference in
results
SELECTING RIGHT GROUP
Prati Ann Intern Med 2002;137:1
• Studied 6835 blood donors; excluded those
with HCV or HBV markers, high glucose or
lipids, or high BMI
• Found upper reference limit of 30 in men and
19 in women
• Found that this cutoff detected more persons
with HCV viremia (76 vs 55%), mostly
detecting milder histologic damage
SELECTING RIGHT GROUP
Ruhl Hepatology 2012;55:447
• Used data from NHANES
• Excluded persons with HCV, HBV, alcohol >
2 drinks/d (in men; 1 drink/d in women), high
BMI or waist circumference, diagnosed
diabetes or pre-diabetes, or A1c > 6.0%
• Found upper reference limit of 29 U/L in
men, 22 in women to have best performance
SELECTING RIGHT GROUP
Lee Hepatology 2010;51:1577
• Evaluated ALT in 665 prospective liver
donors with normal liver biopsy and no
clinical features to exclude from reference
interval (high cholesterol, TG, glucose, BMI,
or HCV, HBV, ANA positivity)
• Mean age 25 in men, 30 in women
• Found ULN of 33 U/L in men and 25 in
women
DOES TYPICAL = HEALTHY?
• For a number of parameters, values that are
typical of the population have been found to
be unhelathy:
• Blood pressure
• Weight
• Cholesterol
• Glucose
• “Reference limits” for these are based on
health outcome studies
WHAT DATA EXIST
ON HEALTHBASED LIMITS
FOR ALT?
HEALTH-BASED LIMITS
Tai Hepatology 2009;49:1859
• Studied 4,376 HBeAg negative “carriers”
• Most had ALT < 36 U/L (ULN), and 40% had
ALT < 18 U/L
• Of those with lowest ALT, only 11% had
steatosis, compared to 27% of those with
ALT 19-36 and about 50% of those with
higher levels
HEALTH-BASED LIMITS
Kim BMJ 2004;386:983
• Used death certificate data to determine risk
of death from liver disease in almost 150,000
Korean persons with baseline ALT
• Compared to those with ALT < 20 U/L (ULN
35-40), relative risk of liver death was 2.5 in
men with ALT 20-29, and 9.5 with ALT 3039; in women, RR was 3.8 if ALT 20-29 and
6.6 if 30-39
HEALTH-BASED LIMITS
Yuen Gut 2005;54:1610
• Evaluated outcomes in 3223 HBV patients
followed a median of 47 months; ULN ALT
53 U/L in men, 31 in women
• Compared to those with baseline ALT < 0.5x
ULN (22% of total), those with ALT 0.5-1.0x
ULN (31%) had significantly higher risk of
liver-related complications; risk highest in
those with ALT 1.0-2.0 x ULN, and
decreased with higher ALT
HEALTH-BASED LIMITS
Burgert JCEM 2006;91:4287
• Studied 392 obese adolescents; ALT ULN 35
U/L
• Compared to those with ALT < 17, those with
ALT 17-35 had worse glucose tolerance,
insulin sensitivity, and higher triglycerides
• Only 48% of those with fatty liver by MRI
had elevated ALT, those with fat had mean
ALT of 34, compared to 15 in those without
HEALTH-BASED LIMITS
Chang Clin Chem 2007;53:686
• Evaluated 5237 men with normal (< 35 U/L)
ALT and no evidence NAFLD on US
• Followed for median 2.5 yrs, 984 (19%)
developed NAFLD on repeat US
• Compared to those in lowest quintile of ALT
values (< 16), those with higher ALT had
progressively higher RR of developing
NAFLD on follow-up (p < 0.001 for trend)
HEALTH-BASED LIMITS
Fracanzani Hepatology 2008;48:792
• Evaluated liver biopsy findings in 458
persons with NAFLD, most biopsied for high
ALT (> 40 U/L)
• NASH present in 59% with normal ALT
versus 74% with high (p < 0.01)
• Only 27% of those with NASH and “normal”
ALT had values > 30 in men or 19 in women
SUMMARY
• Currently, labs use very different ULN,
making comparison difficult, even though
there is not much difference in actual values
• If persons with low likelihood of liver disease
excluded, ULN around 30 in men and 20 in
women
• In those at risk for liver disease, risk of
complications increases at much lower levels
in many studies (15-20)
SUMMARY
• Based on experiences with cholesterol,
glucose, A1c, change to health-based
reference limits requires support of
professional societies of clinicians AND
laboratorians
• Together can address both health implications
and laboratory procedures needed to assure
comparable results between labs