The clinical significance of persistently normal ALT in chronic hepatitis B infection
Journal of Hepatology Dec 2007
Michelle Lai, Benjamin J. Hyatt, Imad Nasser, Michael Curry, Nezam H. Afdhal
Beth Israel Deaconess Medical Center, Division of Hepatology, Harvard Medical School, Boston, MA, USA
"There is significant fibrosis and inflammation in 37% of patients with PNALT (persistently normal ALT) and a liver biopsy should be considered in patients older than 40 with high normal ALT.... We know that in hepatitis C infection, there are up to 25% of patients with persistently normal transaminase levels who have been found to have significant fibrosis....
.... Increasing ALT was associated with more fibrosis and inflammation such that significant fibrosis (stage 2-4) was found in 18%, 34%, and 62% of the patients in the normal ALT, ALT 1-1.5X ULN, and ALT>1.5X ULN group, respectively. Significant inflammation (grade 2-3) was seen in 34%, 54% and 78% of the patients in the normal ALT, ALT 1-1.5X ULN, and ALT>1.5X ULN group, respectively. Overall 37% of patients in the normal ALT group had either significant inflammation or fibrosis....
....Greater age and histological grade are the strongest predictors of significant fibrosis (p=0.0005 and <0.0001, respectively). Controlling for all other covariates, it was revealed that an increase in age by a decade approximately doubles the risk of having significant fibrosis. An increase by one grade increases the risk of significant fibrosis by more than 6 times. Higher ALT and HBeAg+ are also significant predictors of significant fibrosis (p=0.037 and 0.046, respectively)."
Chronic hepatitis B virus (HBV) disease is caused by both necroinflammation and active viral replication. The role of ALT levels as a predictor of liver injury has recently been questioned. The aim of the study was to determine whether normal ALT is associated with liver injury in a cohort of HBV patients undergoing liver biopsy.
This is a retrospective review of chronic HBV patients divided into 3 groups; (1) persistently normal ALT (PNALT); (2) ALT 1-1.5X ULN and (3) ALT>1.5X ULN. Multiple clinical, biochemical, virological variables were evaluated.
One hundred and ninety-two patients met the inclusion criteria, 59 with PNALT, 26 with ALT 1-1.5X ULN, and 107 with ALT>1.5X ULN. Increasing age, higher ALT, higher grade of inflammation on biopsy, and HBeAg positivity predicted fibrosis. 18% of patients with PNALT had stage 2+ fibrosis and 34% had grade 2 or 3 inflammation. Overall 37% of patients with PNALT had significant fibrosis or inflammation. Subgroup analysis showed the majority with fibrosis belonged to the high normal ALT group and that only a minority who were young and immune tolerant had significant findings on biopsy.
There is significant fibrosis and inflammation in 37% of patients with PNALT and a liver biopsy should be considered in patients older than 40 with high normal ALT.
Hepatitis B virus (HBV) infection is a serious public health problem with more than 350 million people estimated to be chronically infected worldwide, of whom one million die annually from HBV-related liver disease . Chronic hepatitis B develops in 90% of newborns infected with HBV, approximately 30% of infants, and less than 5% of immunocompetent adults . Death from chronic liver disease occurs in 15-25% of chronically infected persons .
In chronic hepatitis B infection, chronic inflammation can lead to an increasing degree of fibrosis and ultimately cirrhosis and its complications. Chronic hepatitis B is a common cause of cirrhosis in the United States and is an important cause of liver cancer . The five-year rate of progression from chronic hepatitis B to cirrhosis is estimated to be 12-20% , , .
However, disease progression is variable and not everyone with chronic hepatitis B infection progresses to severe fibrosis or cirrhosis. The rate of progression to cirrhosis is influenced by the degree of inflammation which, in turn, is determined by factors such as the replicative activity of the virus, superinfection by hepatitis delta virus (HDV), and concurrent liver injury from other causes . More recently the level of HBV DNA at baseline has also been associated with progression to cirrhosis and HCC . A persistent and important question is whether patients with active viremia but normal ALT can progress to liver disease.
We hypothesized that patients with persistently normal ALT but active viral replication (HBV DNA>10,000copies/mL) may have clinically significant histological disease.
There is a major debate as to the appropriate approach to the patient with HBV, active viremia and normal ALT. Keefe et al. recommend individualized care and performance of a liver biopsy . The new AASLD guidelines recommend the consideration of liver biopsy and treatment in patients with persistent borderline normal or slightly elevated ALT levels, particularly if the patient is above age 40 . Our findings of significant disease in 37% of PNALT patients suggest that some of the patients with persistent normal ALT > 40 years should be considered for liver biopsy and treatment. Which of the PNALT patients should be biopsied? Our subgroup analysis, which demonstrates that 46% of patients with high normal ALT have significant disease as compared to 20% of patients with low normal ALT, would prompt liver biopsy in the high normal subgroup. This is consistent with the new AASLD guideline suggestion of lowering the upper limits of normal for ALT and recent data from Lin et al. that correlate virological parameters of progressive disease with high normal ALT .
When classifying the clinical scenarios of HBV, immunotolerant patients are commonly defined as young, active viral replication (HBV DNA >105copies/mL, usually >108copies/mL) with HBeAg+, but normal ALT. Many of the normal ALT patients in this study did not really meet the criteria for immunotolerance due to either older age or lower viral loads. In the 25 young immunotolerant patients, significant fibrosis was only seen in only 3 patients. In a recent paper by Andreani et al., no or minimal fibrosis was found in all of their immunotolerant patients . We need to differentiate immunotolerance from normal ALT, and the term immunotolerance should be reserved for young patients with HBV DNA >108copies/mL and low normal ALT (<25IU).
In the normal ALT patients the decision to perform a liver biopsy should balance the cost and risks of liver biopsy against the chance of not identifying patients whose disease will progress without treatment. With the development on the horizon of non-invasive techniques of staging liver disease, the balance may swing towards evaluating all HBV patients including those with normal ALT. In addition, even though we looked at ALT over time, it is impossible to really know whether ALT is persistently normal for years or that we are seeing the product of flares of inflammation over time. Clinical evaluation of HBV should always take into account duration of disease since persistent disease with viremia has been shown to be associated with risk of cirrhosis development irrespective of ALT level .
Age is likely a surrogate marker of the duration of disease. The majority of our subjects were unable to report how long they had chronic hepatitis B. Since the majority of our subjects were from Asia where vertical transmission or infection early in childhood was likely, age was probably representative of duration in most of these subjects. It is not surprising that age is a strong predictor of significant fibrosis in this chronic disease in which damage develops over time. In fact, we consider age above 40 to be a critical determinant in the decision to biopsy and treat.
Given that active inflammation is the driving force leading to fibrosis, it is also not surprising that increasing grade is associated with more fibrosis. In fact, one could surmise that older age, more inflammation and active viremia are all potential drivers of the fibrotic process. Higher ALT, as a biomarker of inflammation, also predicts significant fibrosis, in HBeAg+ patients, but not in anti-HBe+ patients. This may be due to the decreased power from the smaller sample after stratification.
One of the shortcomings of our study was the inability to characterize the upper limit of viral replication in all subjects. The PCR assays used in this study did not dilute serum to quantify the upper end and so the range stopped at >200,000copies/mL for many patients. We cannot then comment on the absolute association between viral load and fibrosis. Interestingly, HBeAg+ may be an indicator of more active viral replication and was associated with a minor increased risk of fibrosis on biopsy. We only biopsied patients with HBV DNA >10,000copies/mL and so cannot comment on patients with even lower level HBV DNA replication. The correlation between HBV DNA and level of liver injury on biopsy is not well characterized but there are some data on the effect of HBV DNA reduction on improvement in liver injury. In an analysis of clinical trial data using oral nucleoside and nucleotide analogs, Mommeja-Marin showed that there was a strong linear correlation between log reduction in HBV DNA and improvement in inflammation and fibrosis on liver biopsy . Recent prospective follow-up studies of large cohorts of carriers from Asia found that high levels of HBV DNA were independent risk factors for the subsequent development of cirrhosis and HCC , , , , .
Are these findings of active inflammation and fibrosis in normal ALT patients really that surprising? We know that in hepatitis C infection, there are up to 25% of patients with persistently normal transaminase levels who have been found to have significant fibrosis . In prior small preliminary studies, there have also been reports of histological injury in patients with normal ALT similar to the findings in this study , . We cannot exclude a referral bias in this study population but several factors suggest that this may be a typical sample of U.S. HBV patients. First, biopsy was not selected but was a standardized protocol for all patients with HBV DNA >10,000copies/mL. Second, 90% of the referral base of HBV patients is from affiliated Asian Health Centers and although we cannot exclude that only the sicker population was being referred, the demographics of the study populaton is highly representative of a typical U.S. HBV cohort.
Our findings strongly suggest that clinicians need to evaluate all patients with HBV DNA >10,000copies/mL carefully for liver fibrosis and inflammation and that age >40 years and an ALT above 25 may trigger evaluation with a liver biopsy. This population represents an important number of patients with HBV at risk for disease and further studies with non-invasive biomarkers for liver injury and fibrosis should include this group of patients.
Materials and methods
We screened the charts of all patients seen in the Beth Israel Deaconess Medical Center (BIDMC) Liver Center between the dates of January 1, 2000 and April 19, 2005 with the ICD-9 code of 070.32. This time period was chosen because it coincided with the clinical practice guideline at BIDMC of biopsying all patients seen with HBV DNA>10,000copies/mL and was felt to exclude bias in biopsy decision. We included patients with a positive hepatitis B surface antigen, HBV DNA>10,000copies/mL or equivalent, a liver biopsy or clinical cirrhosis. For this study, clinical cirrhosis was defined as having either one of the following: ascites, esophageal varices, or encephalopathy; or two of the following: thrombocytopenia <130,000, hypoalbuminemia <3.5mg/dL, portal hypertensive gastropathy, radiographic splenomegaly, and radiographic nodular liver. Patients were excluded if they had hepatocellular carcinoma, immunosuppression, HIV, history of positive HCV RNA, hemachromatosis or other chronic liver diseases or treatment with oral antiviral nucleosides or nucleotides therapy prior to biopsy, but included if their therapy was limited to interferon more than 1 year prior to biopsy.
Data collected were age, sex, race, e antigen and antibody status, viral load, mode and date of transmission if known, all ALT and AST values known, weight, current or prior heavy alcohol use, co-infection with HIV, HCV, or HDV, comorbidities, prior to interferon therapy, biopsy date, stage, grade, presence of steatosis and iron and biopsy length. The date of the liver biopsy was used as the reference point with all data collected prior to biopsy. Liver biopsies were scored using the Metavir scoring system for both inflammation and fibrosis. Age was defined as age at biopsy. Prior or current heavy alcohol use was defined as >50g/day for 5 years or more.
The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the BIDMC Institutional Review Board. Since it was a retrospective chart review with all identifying patient information removed, informed consent was waived.
Twenty-two patients had viral loads that were measured by non-PCR methods, given in picograms/mL, with the lower limits of detection at 5pg/mL. These values were converted to copies/mL using the conversion factor of 1pg/mL=283,000copies/mL. Viral load in these 22 patients ranged from 15 to 11,820pg/mL with five viral loads not quantified above the upper limit of 2000pg/mL. Values given as IU/mL were converted to copies/mL by multiplying by a factor of 5. Viral loads measured by PCR methods ranged from 10,000 to 1,700,000,000copies/mL, with multiple values not quantified above different upper limits which ranged from 200,000 to 500,000,000copies/mL. All analyses were done on log transformation of the viral load. Many of the viral loads were not quantified above the upper limit of the assays which limited significant analyses involving viral load. For our analyses, the upper limit of the assay was used in these cases with an additional variable created to note that the viral load was not quantified.
Based on pre-biopsy ALT values, patients were classified into one of three categories: normal ALT, ALT 1-1.5 times upper limits of normal (ULN), and ALT>1.5 times ULN. The maximal ALT level over at least a minimum of 6 months determined the group selection. Normal ALT was defined by having at least two ALT values equal to or less than 40IU/L at least six months apart and no elevated ALT at any time point prior to biopsy. The normal cutoff of 40IU/L was used by the BIDMC Laboratory for both men and women and for female adults by LabCorp and Quest Diagnostics. However, for male adults, it is 55 and 60IU/L for LabCorp and Quest Diagnostics, respectively. Stage of fibrosis and grade of inflammation were also converted into binomial variables of significant fibrosis (stage 2-4) vs. no significant fibrosis (stage 0-1) and significant inflammation (grade 2-3) vs. no significant inflammation (grade 0-1).
Univariate analyses, correlation matrices, χ2 test, and multivariate regression analyses were performed using SAS statistical software 9.1 for Windows. Multiple logistic regression analysis was performed to identify predictors of significant fibrosis. The variables found to correlate with stage in the correlation matrix and in univariate analysis were included in the multivariate regression analysis.
Further subgroup analyses were performed. One stratified patients with persistent normal ALT into low normal (ALT 0-25) and high normal (ALT 26-40), which is consistent with the recommendation of normal ALT in males according to the new AASLD guidelines for HBV. The second subgroup analysis stratified the patients by HBeAg status. Distribution of stage and grade was also looked at in subgroups of high ALT patients. They were categorized into subgroups of patients with median ALT value >1.5X ULN, >2XULN, >3XULN, or >5XULN.
Of the 593 patients with an ICD-9 code of 070.32, 192 patients met our inclusion and none of the exclusion criteria. Of these 192 patients, there were 59 patients with PNALT, 26 patients with ALT 1-1.5X ULN, and 107 patients with ALT>1.5X ULN (see Fig. 1). Of the 59 patients with normal ALT, 20 had a mean ALT< 25IU/L while 39 patients had a mean ALT level of 26-40IU/L. Eight patients with clinical cirrhosis were included without a liver biopsy. The number of patients excluded and the reasons for exclusions are shown in Fig. 1.
Table 1 shows the demographic data. The normal ALT group was predominantly female while those with elevated ALT were predominantly male (p<0.0001). While the difference in age between the groups was not statistically significant, there was a trend for normal ALT group to include younger patients. Of the 25 patients under age 40 with high viral load and normal ALT, who represent probable immunotolerance, only 3 had significant fibrosis. A higher ALT was associated with high alcohol intake but not with either HBV DNA level or HBeAg status.
Fig. 2 illustrates the stage and grade on liver biopsy in the different ALT groups. Increasing ALT was associated with more fibrosis and inflammation such that significant fibrosis (stage 2-4) was found in 18%, 34%, and 62% of the patients in the normal ALT, ALT 1-1.5X ULN, and ALT>1.5X ULN group, respectively. Significant inflammation (grade 2-3) was seen in 34%, 54% and 78% of the patients in the normal ALT, ALT 1-1.5X ULN, and ALT>1.5X ULN group, respectively. Overall 37% of patients in the normal ALT group had either significant inflammation or fibrosis.
Table 2 shows the characteristics of the normal ALT subgroups. We found a significant disparity in gender, with the low normal group being 85% female and the high normal group 54% male (p<0.0001). The racial breakdown and age were similar. Though not statistically significant, there was a trend for higher mean weight, stage and grade in the high normal group as compared to the low normal group. There was also a higher proportion of patients with steatosis in the high normal patients. Log of viral load, history of alcohol use, HBeAg status, and age were similar between the two subgroups.
Fig. 3 shows the distribution of stage and grade within each of the normal ALT subgroups. Significant fibrosis in the high normal group was five times as prevalent as in the low normal group (25% vs. 5%). The prevalence of significant inflammation in the high normal group (41%) was double that of the low normal group (20%). Overall 20% of patients with low normal ALT had either significant inflammation or fibrosis compared to 46% with high normal ALT.
Fig. 4 shows the distribution of stage and grade within each of the ALT groups including the high ALT subgroups. The distribution of stage and grade was not statistically significantly different between the different high ALT subgroups.
Stratifying for HBeAg status revealed that anti-HBe+ patients were older, heavier and had lower viral load (Table 3, Table 4).
The correlation matrix revealed that increased fibrosis was correlated with older age, more significant inflammation, higher ALT and male sex. Increased inflammation was correlated with higher ALT. Older and heavier patients were more likely to have steatosis. Asian-Americans and women had lower weight. Women also had lower ALT.
Given the association observed between many of the covariates, multivariate logistic regression analysis was performed, controlling for all covariates simultaneously. The analysis revealed that increasing age, histological grade and ALT group as well as HBeAg positivity are significant independent predictors of fibrosis (Table 5). Greater age and histological grade are the strongest predictors of significant fibrosis (p=0.0005 and <0.0001, respectively). Controlling for all other covariates, it was revealed that an increase in age by a decade approximately doubles the risk of having significant fibrosis. An increase by one grade increases the risk of significant fibrosis by more than 6 times. Higher ALT and HBeAg+ are also significant predictors of significant fibrosis (p=0.037 and 0.046, respectively). With movement from one ALT category to the next, one's risk of significant fibrosis is increased by 0.58 times. HBeAg+ subjects are almost 3 times as likely to have significant fibrosis as subjects who are anti-HBe+.
Table 6 shows the results of the multiple logistic regression analysis performed after stratifying for HBeAg status. After stratification, older age and grade still remained significant predictors of significant fibrosis. In the HBeAg+ cohort, an increase in age by 10 years doubled the risk of significant fibrosis (p=0.017) while the risk was 2.5 times in the anti-HBe+ cohort (p=0.016). With every increase in grade, there was a 5.4 times increased risk of significant fibrosis in HBeAg positive patients and 12.4 times increased risk of significant fibrosis in anti-HBe+ patients. Increasing ALT was a predictor of significant fibrosis and inflammation in HBeAg+ patients but not in anti-HBe+ patients (OR 1.767 and 1.885; p=0.042 and 0.026, respectively). Significant alcohol use was found to be a predictor only in anti-HBe+ (OR 30.724; p=0.033). Stage was a predictor of significant inflammation in the anti-HBe+ cohort (OR 10.171, p=0.008).