Predictive Value of ALT Levels for Nonalcoholic Steatohepatitis (NASH) and
Advanced Fibrosis in Non-alcoholic Fatty
Liver Disease (NAFLD)
Siddharth Verma, Donald Jensen, John Hart, Smruti R. Mohanty
Abstract and Introduction
Abstract
Background: Non-alcoholic fatty liver disease (NAFLD) patients with elevated serum
alanine aminotransferase (ALT) generally undergo a liver biopsy to evaluate for
possible non-alcoholic steatohepatitis (NASH) or advanced fibrosis. However, patients
with normal ALT could also have advanced stages of NAFLD.
Aim: To determine ALT value that will accurately predict NASH and advanced fibrosis
using area under the receiver operating characteristics curve (AUROC) analysis.
Methods: Demographic, clinical and laboratory data of an ethnically diverse cohort of
biopsy proven NAFLD patients were retrospectively analysed under univariate and
multivariate analyses. Liver biopsies were scored using NASH clinical research network
(NASH CRN) system. AUROC were performed for NAFLD Activity Score ≥5 (NASH)
and fibrosis score ≥2 (advanced fibrosis).
Results: Two hundred and twenty-two patients were analysed. Fifty six (23%) had
normal ALT. There was no difference in the rate of advanced fibrosis between normal
and elevated ALT (26.8% vs. 18.1%, P = 0.19). However, significantly lower
percentage of normal ALT group had NASH compared with elevated ALT group
(10.7% vs. 28.9%, P < 0.01). Overall, 37.5% of normal ALT group had NASH or
advanced fibrosis, whereas 53% of elevated ALT had no NASH or advanced fibrosis.
Higher ALT values correlated with higher specificity, but lower sensitivity for both
NASH and advanced fibrosis. AUROC for ALT level correlating NASH and advanced
fibrosis were 0.62 and 0.46 respectively.
Conclusion: There is no optimal ALT level to predict NASH and advanced fibrosis.
Metabolic risk factors should be evaluated to select patients for a liver biopsy to
confirm NASH and advanced fibrosis.
Introduction
Non-alcoholic fatty liver disease (NAFLD) encompasses a broad spectrum of clinical
and histological manifestation, ranging from mild steatosis, NASH, progressive fibrosis,
cirrhosis and ultimately to hepatocellular carcinoma.[1] Non-alcoholic steatohepatitis
(NAFLD) is considered to be the hepatic manifestation of metabolic syndrome (MS),[2–
4]
which has been shown to be a strong predictor for hepatic steatosis[1, 5–8] and advanced
liver disease.[2–4, 8] In general, elevated serum alanine aminotransferase (ALT) levels are
further evaluated with a liver biopsy to determine whether patients with NAFLD have
progressed to NASH and advanced fibrosis.[8] Prior studies have suggested that elevated
ALT levels correlate with NASH and advanced fibrosis.[9, 10] However, patients with
normal ALT may also have NASH and advanced fibrosis as well.[5, 8, 11–13] The lack of a
systematic association between ALT level and advanced form of NAFLD has important
clinical relevance, as it raises the issue of whether NAFLD patients with normal ALT
should undergo a liver biopsy to confirm NASH or advanced fibrosis.[8] Therefore,
predictive values of ALT levels are needed to determine a level that could assess for
advanced liver injury, including NASH and advanced fibrosis, while potentially
avoiding the need for a liver biopsy.
Although studies have included NAFLD patients with normal and elevated ALT
levels,[8–13] investigations into the best cut-off value of ALT to predict NASH and
advanced fibrosis using AUROC are lacking. In addition, previous studies comparing
clinical characteristics of NAFLD in patients with normal and elevated ALT levels were
limited because of a lack of multi-ethnic patient population, small sample size or for not
utilizing the updated NASH clinical research network (CRN) grading system[14] for
histological evaluation. The aim of our study was to identify a critical threshold of ALT
by performing AUROC analysis that would predict NASH or advanced fibrosis in an
ethnically diverse patient population. In addition, we compared demographic,
biochemical, metabolic and histological variables between patients with normal and
elevated ALT levels using the updated NASH CRN grading system.
Patients and Methods
Study Design
We examined a retrospective cohort of patients with biopsy proven NAFLD from the
University of Chicago Medical Center (UCMC) pathology database between 1 June
1995 and 30 June 2005. Biopsy reports containing the terms 'steatosis,' 'steatohepatitis,'
and/or 'fat' were included. All biopsy reports were kept computerized. Furthermore, all
biopsies performed for abnormal liver appearance on imaging studies, or abnormal
intra-operative findings during bariatric surgery or cholecystectomy were included
irrespective of ALT levels (normal or abnormal). Furthermore, all patients with
abnormal imaging results were offered advice on dietary modifications irrespective of
ALT levels prior to a liver biopsy. Relevant demographic, clinical and laboratory data
were obtained from UCMC's computerized medical records. Data on diabetes mellitus
(DM) and MS were collected based on American Diabetes Association[15] and National
Cholesterol Education Program ATP III criteria.[16] Obesity was defined as body mass
index ≥30 kg/m2, as waist circumference measures were unavailable.[17] The research
study design has been well described in our previous study.[18] This study was approved
by the institutional review board.
Case Definitions and Liver Histology
All liver biopsies were examined and scored by a single pathologist. The pathologist
was blinded to the liver function tests, and all liver biopsies were rereviewed by the
pathologist. Patients with other chronic liver diseases including Hepatitis B and C, iron
over load, medication-related steatosis, significant alcohol use (daily alcohol
consumption of ≥40 g/d in males and ≥20 g/d in females) and liver transplant were
excluded. NAFLD patients were divided according to normal (≤35 IU/L) and elevated
(>35 IU/L) ALT levels. Values for normal and elevated ALT levels were institution
based, and were standardized for all patients, irrespective of gender or ethnicity. Cases
were defined histologically using the updated NASH CRN grading system.[14] Steatosis
was scored as 1, 2 or 3 for 5%–33%, 34%–66% and >66% steatosis respectively.
Fibrosis was scored as 0 (no fibrosis), 1 (perisinusoidal or periportal), 2 (perisinusoidal
and portal/periportal), 3 (bridging) and 4 (cirrhosis). Lobular inflammation was scored
as 1, 2 or 3 for <2 foci per 200× field, 2–4 foci per 200× field and >4 foci per 200× field
respectively. Ballooning was scored as 0, 1 and 2 for no balloon cell, few balloon cells
and many cells/prominent ballooning cells respectively. Lastly, Mallory's hyaline was
scored as 0 for 'none to rare' or 1 for 'many.' According to the NASH CRN study, a
NAFLD Activity Score (NAS) of ≥5 was considered NASH,[14] while a fibrosis score
≥2 was considered a marker of advanced fibrosis.
Statistical Analysis
Statistical analysis was performed using StataCorp, LP STATA software, version 11.
Results were expressed as mean ± standard deviation for continuous variables, and as
frequency and per cent for categorical variables. For unequal variance, t-tests were
performed to compare the means of continuous variables. Categorical variables were
compared using chi-square test. P values ≤ 0.05 were considered statistically significant.
Logistic regression analysis was performed to analyse variables independently
associated with the presence of NASH and advanced fibrosis in normal ALT group,
elevated ALT group and overall group. Variables significant with univariate analyses
were entered into a multivariate analysis model from which independent variables with
P > 0.1 were excluded sequentially. The odds ratio and associated P values were
reported. Lastly, AUROC were created to determine a critical ALT threshold associated
with NASH and advanced fibrosis.
Results
Demographic, Comorbid and Biochemical Features
Two hundred and thirty-eight patients met the criteria of NAFLD of which 16 patients
were excluded as data on baseline ALT levels were lacking. Of the remaining 222
patients, 56 (23%) had normal ALT values. Mean ALT value was 24.9 IU/L in the
normal ALT group compared to 90.4 IU/L in the elevated ALT group. The ethnic
distribution of NAFLD patients were as follows: 64.7% Caucasians, 15.1% African
Americans (AA), 13.4% Hispanics and 6.7% Asians. Patients with normal ALT were
older, female, higher BMI, and had higher rates of type II diabetes and hypertension
compared with elevated ALT group (Table 1). Patients with elevated ALT had higher
levels of AST, total and low-density cholesterol, iron and ferritin compared with normal
ALT group (Table 1). There were no significant differences in ethnicity, metabolic
syndrome (MS), fasting glucose or HDL between the two groups (Table 1).
Histological Features
Percentages of patients with steatosis ≥2 (P < 0.01) and inflammation ≥2 (P < 0.04)
were significantly lower in normal ALT group compared with elevated ALT group.
However, there was no significant difference in ballooning between two groups (Table
2). Rates of NASH and advanced fibrosis are shown in Fig. 1. Both the mean NAS
score (2.4 vs. 3.4; P < 0.05) and percentage of patients with NASH (10.7% vs. 28.9%; P
< 0.01) were significantly lower in the normal ALT group compared with elevated ALT
group. Although the mean fibrosis score between these two groups did not differ (Table
2), percentage of patients with advanced fibrosis was slightly higher in the normal ALT
group compared with elevated ALT group (26.8% vs. 18.1%; P = 0.19). In the normal
ALT group, 37.5% had evidence of either NASH or advanced fibrosis, and an
additional 7.1% had evidence of both NASH and fibrosis. Contrarily, 53% of patients
with elevated ALT had normal liver biopsies, without evidence of either NASH or
advanced fibrosis. Trends in sensitivity and specificity of ALT levels are listed in Table
3. In general, sensitivity for NASH or advanced fibrosis decreased with progressively
higher ALT levels, while specificity for NASH or advanced fibrosis increased with
progressively higher ALT levels.
(Enlarge Image)
Figure 1.
Histological findings in patients with normal and elevated alanine aminotransferase
(ALT).
At the univariate analysis, Mallory bodies, an increase in AST per 10 units, an increase
in ferritin per 100 ng/ml, steatosis, ballooning and inflammation were significantly
associated with NASH in the overall group. Steatosis and ballooning were significant
within the normal ALT group, whereas ballooning was only significant in elevated ALT
group for NASH (Table 4). On multivariate analysis, ferritin per 100 ng/ml increase
(P= 0.01; OR, 1.2; 95% CI, 1.03–1.34) was independently associated with NASH in all
patients. However, an increase in ferritin per 100 ng/ml showed a weak association (P =
0.05; OR, 1.14; 95% CI, 1.00–1.32) with NASH in the elevated ALT group.
Factors significantly associated with advanced fibrosis under univariate analysis are
listed in Table 5. Although an increase in ALT per 10 units was significant in the
normal ALT group (P = <0.01), it was not significant in the elevated ALT group or in
the overall population. In the multivariate analysis, Mallory bodies (P = 0.04; OR, 2.6;
95% CI, 1.03–6.42), ballooning (P < 0.01; OR, 17.3; 95% CI, 6.8–43.8) and
inflammation (P < 0.01; OR, 36.4; 95% CI, 4.0–330.7) were significantly associated
with advanced fibrosis in all patients. Although ballooning (P < 0.01; OR, 49.7; 95%
CI, 3.8–644.3) was the only finding significantly associated in the normal ALT group,
both ballooning (P < 0.01; OR, 18.7; 95% CI, 5.9–58.7) and inflammation (P < 0.01;
OR, 27.4; 95% CI, 3.0–251.2) were significantly associated with advanced fibrosis in
the elevated ALT group.
AUROC and Logistic Regression Analysis
Area under the receiver operating characteristics curves were established for various
ALT levels to determine an ALT cut-off value, which would most accurately predict
NASH (Fig. 2) and advanced fibrosis (Fig. 3). The AUROC analysis included patients
of all ethnicity. AUROC curve for ALT level was 0.62 for NASH and 0.46 for
advanced fibrosis.
(Enlarge Image)
Figure 2.
Area under the receiver operating characteristics (AUROC) for alanine
aminotransferase (ALT) values and non-alcoholic steatohepatitis NASH.
(Enlarge Image)
Figure 3.
Area under the receiver operating characteristics (AUROC) for alanine
aminotransferase (ALT) values and advanced fibrosis.
Discussion
The progression of NAFLD to its advanced stages is associated with significant
morbidity in approximately 20% of patients, including complications such as gastrooesophageal varices, ascites, liver failure, hepatopulmonary syndrome and
encephalopathy.[19] Furthermore, greater than 20% of NAFLD patients may develop
cirrhosis over their lifetime according to a study by Matteoni et al.[1] Of the patients
who develop cirrhosis, 30–40% may suffer liver-related mortality within a 10-year
period.[20] Therefore, recognizing patients with NASH and advanced fibrosis early in the
disease spectrum is essential not only in managing but also in preventing further
progression to cirrhosis and HCC, and its related complications. Although elevated ALT
levels generally lead to further abdominal imaging or histological evaluation, there is
the potential for delay in the diagnosis of advanced stage of NAFLD in those with
normal ALT levels. This may explain why patients with normal ALT levels in our study
were significantly older than those with elevated ALT levels, a finding which was
consistent with a recent study by Fracanzani et al.[8]
Unfortunately, a vast majority of patients with normal ALT and underlying NAFLD are
rarely investigated early in the disease or undergo a liver biopsy. This may be possibly
caused by a lower index of suspicion for underlying advanced liver disease associated
with NAFLD.[8, 12, 13] A recent decision tree analysis demonstrated the importance of
early liver biopsy in diagnosing NAFLD in early stages, amenable to lifestyle changes,
while reducing disease progression to cirrhosis-related mortality or transplant eligible
disease.[21] Studies instead have focused on non-invasive, serological markers to assess
NAFLD disease severity[22–26] to avert the need for a liver biopsy and its potential
complications. However, these studies have been limited to a single or undefined ethnic
group,[8, 11, 22–24, 26] baseline elevated ALT level,[22, 26] obese patients undergoing
bariatric surgery[24, 25] or did not use the revised NASH CRN scoring system.[22, 24]
Given that NASH or advanced fibrosis may be present in patients with normal or
elevated ALT levels, to our knowledge, this is the first investigation to examine the
AUROC for ALT and its utility in predicting NASH and advanced fibrosis by utilizing
the NASH CRN scoring system in an ethnically diverse patient population.
Our results demonstrate that reliance on ALT levels may not accurately assess the
extent of hepatic injury in NAFLD patients. Liver biopsies of those with elevated ALT
demonstrated advanced fibrosis in a similar rate compared to patients with normal ALT.
Furthermore, more than one in three patients with normal ALT levels had evidence of
NASH or advanced fibrosis and almost 10% had evidence of both NASH and fibrosis.
Contrarily, more than half of the patients with elevated ALT did not have either NASH
or advanced fibrosis. Prior studies in patients with normal ALT levels have
demonstrated wide variations in the rates of NASH or advanced fibrosis. NASH was
present in 59% of patients with normal ALT in a recent study by Fracanzani et al.,[8]
whereas the rate of NASH in patients with normal ALT in this study by Lee et al.[11]
was 2.9%. In our study, the rate was 10.7%. This difference in rate of NASH in normal
ALT patients may likely be multifactorial. In the studies by Fracanzani et al. and Lee et
al., the patient population consisted of a single ethnic background, elevated ALT was
defined as ≥40 IU/L and the Brunt criteria[27] for histological diagnosis of NAFLD was
used. Furthermore, interobserver variability and selection bias may have confounded the
results in Lee et al.'s study as multiple pathologists evaluated the liver biopsies and
patients were young, healthy male liver donors.[11] In our study, the patient sample
consisted of a mixed ethnic group, including AA, Hispanics and Asians, reflective of
the general population and a stricter criterion for normal ALT was applied. In addition,
all biopsies were reviewed by a single pathologist, eliminating interobserver bias, and
the pathologist was blinded to liver function tests, thus eliminating the potential for
intra-observer bias. Biopsies were also evaluated using the updated NASH CRN
grading system. The NASH CRN established a validated scoring system (NAS) with
significant interrater reproducibility.[14] Specifically, NAS refined the scoring system by
Brunt et al.[27] to not only encompass the entire spectrum of NAFLD, instead of only
NASH, but also to assess histologically distinct lesions that comprise NALFD, instead
of an aggregate gradation of severity. This study concluded that a NAS of ≥5 correlated
with NASH, whereas NAS of ≤2 was non-diagnostic of NASH.[14] Our results not only
underscore the importance that a substantial number of patients with normal ALT levels
may have underlying NASH or advanced fibrosis, whereas those with elevated ALT
levels may have mild NAFLD only, but also expand the results of Lee et al. and
Fracanzani et al.'s study to include a multi-ethnic patient population and provide results
based on the updated scoring system.
Nonetheless, ALT level is often considered by many clinicians as an easily accessible
surrogate marker for evaluating underlying liver disease activity and severity of liver
injury.[28] While our results demonstrate that normal and elevated ALT levels do not
correlate with the severity of NAFLD, we further performed an AUROC analysis to
evaluate whether a specific value of ALT would best predict NASH or advanced
fibrosis. The AUROC curve in our study failed to demonstrate an optimal ALT level
that would best predict NASH or advanced fibrosis. With respect to ALT and fibrosis
score ≥2, the AUROC curve lies below the line of no-discrimination indicating that
predicting degree of fibrosis can be achieved by chance as often as it is achieved by
ALT levels alone. Wong et al. recently demonstrated that the AUROC for ALT in
predicting NASH was 0.61 and 0.57 for advanced fibrosis.[29] Although these results are
similar to our findings, this study population was limited to only Chinese patients,
biopsies were evaluated by two pathologists and the sample size was considerably
smaller.[29] However, clinical practice often involves patients of various ethnicities.
Multiple recent studies have demonstrated ethnic variations in prevalence of NAFLD,
including differences in ALT levels between ethnicities.[18, 30–35] Therefore, the results of
our study not only validate the results of Wong et al.'s study through a larger sample
size, but also include a mixed ethnic population representing the general population
demographics. Our results further reinforce the complex nature of NAFLD, including
differences between ethnicities. Further studies examining AUROC of serological
markers in an ethnically diverse NAFLD patient population with normal and elevated
ALT levels using the updated NASH CRN grading system are needed.
Our retrospective study has several limitations. Firstly, the number of patients,
including non-Caucasian patients with normal ALT levels was small in comparison with
elevated ALT group. However, our normal ALT sample size is similar to other studies
comparing ALT levels in NAFLD.[8, 12, 29] Furthermore, although the number of nonCaucasian patients with normal ALT was small, the primary aim of our study was to
evaluate the AUROC of ALT levels of all patients within an ethnically diverse cohort.
A comparison of histological differences in an ethnically diverse patient population with
normal or elevated ALT levels is outside the scope of our study. Studies examining
whether ethnicity affects histological changes in patients with normal or elevated ALT
levels are needed. Secondly, the sample size of African Americans, Hispanics and
Asians was small in comparison with Caucasians. However, our study contains the most
diverse patient population among other such studies to date and appears to be
representative of the US population. Thirdly, the lack of significant difference in the
rate of advanced fibrosis in the normal and elevated ALT groups may be owing to small
sample size leading to a type II error. Although patients with elevated ALT levels had
significantly higher NAS score, mean fibrosis score between these two groups did not
differ. In addition, patients with abnormal imaging or intra-operative findings were
selected for liver biopsies leading to potential selection bias. However, the selection
bias was applied to both normal and elevated ALT groups equally, thereby limiting its
effect. Further studies on a random sample of patients are warranted to confirm our
findings. Lastly, the cut-off value for abnormal ALT in our study was 35 IU/L. A 2002
study by Prati et al.[36] concluded that the upper limit of normal for ALT in patients with
chronic hepatitis C or NAFLD should be revised to 30 IU/L for men and 19 IU/L for
women. However, a recent study by Ruhl et al.[37] demonstrated that the upper limit of
normal ALT levels in the US population should be 29 IU/L and 22 IU/L for men and
women respectively. However, multiple studies examining ALT levels have used a cut-
off value for normal ALT substantially higher than our definition of normal ALT,[8, 11, 12,
24, 25, 29]
with some studies defining normal ALT as high as 75 IU/L.[12]
A few important conclusions can be drawn from our study. Firstly, ALT is a poor
marker for workup of NAFLD as a significant amount of patients with advanced liver
disease may have normal ALT levels. Consequentially, a normal ALT value may
provide false reassurance and potential delay in diagnosis of NAFLD, whereas patients
with elevated ALT value might be subjected to liver biopsy demonstrating mild
steatosis only. Secondly, there is no ideal ALT cut-off value using AUROC analysis
that would best predict underlying disease severity, including NASH and/or advanced
fibrosis. Although AUROC analysis in other liver diseases, such as hepatitis B[28, 38] and
C,[39–41] have demonstrated a poor correlation between ALT levels and disease severity,
our study is the first to show a similar lack of correlation in NAFLD patients of a
diverse ethnic background. Lastly, a liver biopsy should be considered for further
evaluation of NAFLD, including confirmation of NASH or advanced fibrosis, in
patients with additional risk factors for NAFLD, such as metabolic syndrome, diabetes,
obesity, hypertension and Asian/Hispanic background, irrespective of ALT levels. In
our study, the only significant factors correlating variations in ALT levels with degree
of NASH or fibrosis were histological, such as Mallory bodies, ballooning and
inflammation. Although liver biopsy is associated with limited morbidity and sampling
error,[42] it is the sole method of assessing histological changes associated with NAFLD,
and is the most accurate method of diagnosing NASH or any degree of fibrosis as of
now. Furthermore, a liver biopsy performed in the early stages of NAFLD may benefit
patients from non-pharmacological interventions, such as weight loss and lifestyle
modifications, while potentially reducing progression to advanced disease. Therefore,
we suggest that clinicians should have a higher index of suspicion for the potential of
underlying advanced stages of NAFLD, despite normal ALT levels, in patients with
predictive risk factors for advanced liver diseases, and a liver biopsy to confirm NASH
or advanced fibrosis should also be considered in these patients.