Long-term outcome of hepatitis B e antigen–negative hepatitis B surface antigen carriers in relation to changes of alanine aminotransferase levels over time

Authors


  • Potential conflict of interest: Y. F. Liaw has been involved in clinical trials for or served as a global advisory board member of Roche, Bristol-Myer Squibb, GlaxoSmithKline, Novartis, and Gilead Sciences.

Abstract

The baseline alanine aminotransferase (ALT) level was reported to have prognostic value in chronic hepatitis B virus (HBV) infection, during which ALT may change over time. Instead of baseline ALT, this study aimed to study the prognostic value of the height of ALT during the course of chronic HBV infection. A total of 4376 asymptomatic hepatitis B e antigen (HBeAg) negative, surface antigen (HBsAg) carriers with baseline ALT less than 2 times the upper limit of normal (ULN) were monitored with ALT measurement and ultrasonography every 3 to 12 month for over 3 years. Maximal ALT levels during follow-up were correlated with long-term outcomes using morbidity and mortality data from hospital records, cancer registration, and national mortality database. Baseline ALT level was normal in 3673 subjects and increased to abnormal level in 1720 (46.8%) during a mean follow-up period of 13.4 ± 5.2 (3.0-28.7) years. The incidence of liver cirrhosis, hepatocellular carcinoma (HCC), and mortality increased with increasing maximal ALT level during follow-up, especially in those with maximal ALT of at least 2 times ULN, as compared with those who maintained normal ALT. Cox regression analysis indicated that age at entry, sex, and maximal ALT level during follow-up were significant independent factors associated with the development of cirrhosis, HCC, and mortality whereas cirrhosis was also an independent factor for HCC development and mortality. Conclusion: Persistently normal ALT was associated with excellent long-term prognosis, whereas increasing ALT levels of at least 2 times ULN during follow-up was associated with increasing morbidity and mortality. ALT of at least 2 times ULN is therefore an appropriate threshold for anti-HBV therapy, whereas those with ALT 1 to 2 times ULN require liver biopsy for decision. (HEPATOLOGY 2009.)

Chronic hepatitis B virus (HBV) infection is one of the major causes of liver morbidity and mortality.1, 2 Recent community-based studies involving a large number of hepatitis B surface antigen (HBsAg) positive subjects, mainly (>85%) hepatitis B e antigen (HBeAg) negative, have shown that baseline age, sex, HBeAg serostatus, and serum HBV-DNA level are factors associated with hepatocellular carcinoma (HCC) development and liver-related mortality.3, 4 Abnormal alanine aminotransferase (ALT) at entry was also found to be associated with cirrhosis development and liver-related mortality.5, 6 All HBV treatment guidelines of major liver associations also recommend that patients with active HBV replication and an ALT level greater than 2 times the upper limit of normal (ULN) are candidates for antiviral therapy.7–9 The main reason for choosing this threshold was that patients with ALT less than 2 times the ULN responded poorly to current drugs in terms of HBeAg seroconversion during short-term therapy, not because of the prognostic value of this ALT level.9, 10

Instead, large long-term mortality risk studies showed that subjects with a baseline ALT level of 1 to 2 times the ULN were also associated with higher mortality from liver diseases, compared with ALT less than 1 times the ULN.11, 12 Furthermore, one study showed that baseline ALT of 0.5 to 1 times the ULN was associated with increased mortality, compared with ALT less than 0.5 times the ULN.11 However, these studies did not include the assessment of HBsAg or hepatitis C virus (HCV) serostatus. Whether these findings are also applicable to “inactive” HBsAg carriers remains unclear. We therefore conducted the current study to clarify the association of different ALT levels with outcomes of HBeAg-negative HBsAg carriers, using long-term follow-up data in the HBsAg carrier clinic of Chang Gung Memorial Hospital. Because chronic HBV infection is a dynamic state of interactions among HBV, hepatocytes, and immune cells, serum ALT levels usually reflect the degree of HBV-related liver cell injuries and may change over time during the course.13–16 In particular, “inactive” HBeAg-negative HBsAg carriers may develop HBeAg-negative hepatitis with ALT elevation even after 10 to 20 years' remission.17–20 Therefore, we chose to assess the prognostic value of the height of ALT levels over time instead of a single baseline ALT level.

Abbreviations

AFP, alpha-fetoprotein; ALT, alanine aminotransferase; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HDV, hepatitis D virus; maxALT, maximal alanine aminotransferase; PNALT, persistently normal alanine aminotransferase; ULN, upper limit of normal; US, ultrasonography.

Patients and Methods

Patients

The “HBsAg carrier clinics” of Chang Gung Memorial Hospital in Taipei and Linkou medical centers have been in operation since 1980 and have provided an easily accessible service for chronic HBsAg carriers. Most subjects who visited “HBsAg carrier clinics” were asymptomatic at entry. They visited this clinic because of incidental detection of HBsAg on blood donation, general checkup, workup for non-liver disease, or referral from our outpatient department as a stable HBsAg carrier with normal ALT. After entry, these HBsAg carriers were followed every 3 to 12 months with ALT, alpha-fetoprotein (AFP), and ultrasonography (US) as the basic tools. Subjects were recruited into this study if they fulfilled the following criteria: (1) HBsAg positive for at least 1 year; (2) HBeAg negative, anti-HBe-positive; (3) ALT level less than 2 times the ULN, no evidence of cirrhosis, and no concomitant hepatitis C or D virus infection at entry; (4) no antiviral or immunomodulatory therapy before entry; (5) regular follow-up for at least 3 years, and more than five visits. Patients who had alcohol, drugs, or other possible causes of hepatitis were excluded. No patient in this cohort admitted intravenous drug abuse or homosexuality.

The long-term follow-up study on chronic HBV infection was approved by the human research committee of Chang Gung Memorial Hospital.

Assessment at Baseline and During Follow-Up

At entry, the subjects were tested for liver biochemical tests, virological markers [HBsAg, HBeAg, anti-HBe, antibodies against HCV [anti-HCV], and antibody against hepatitis D virus (HDV; anti-HDV)], AFP, and real-time abdominal US. Liver biochemical tests were assayed every 6 months if ALT levels were normal, and every 1 to 3 months if ALT levels were elevated, or more often when indicated. HBsAg was assayed 6 months after entry and then every 3 to 5 years. Assays for anti-HCV and anti-HDV were repeated if ALT levels were elevated to 2 times ULN or greater. Serum HBV DNA was not routinely assayed but was measured in patients when their ALT levels were elevated at least 2 times the ULN.

ALT was measured by commercially available kits with automatic analyzers in clinical laboratory of this hospital, a laboratory certified by The College of American Pathologists. Different kits with slightly different normal range values (generally within 5-40 IU/mL) were used during a span of 24 years. Our laboratory has chosen 36 U/L as the ULN for both sexes in recent 10 years. HBsAg, HBeAg, anti-HBe, and anti-HDV were assayed by radioimmunoassay (Abbott Diagnostics, North Chicago, IL). Anti-HCV was tested by a second-generation or third-generation enzyme immunoassay (Abbott Diagnostics). HBV DNA was tested by sandwich molecular hybridization assays using a Digene Hybrid Capture System (Digene Diagnostics Inc., Beltsville, MD). The lower detection limit was 1.4 × 105 copies/mL. Conventional abdominal US was performed with a high-resolution real-time scanner, using a 3.5-MHz or 5.0-MHz rectilinear or convex transducer. AFP and US were monitored every 3 to 12 months in patients older than age 35, patients with liver cirrhosis, or patients with a familial history of HCC.

Major Events During Follow-Up

Maximal ALT Level.

The maximal ALT (maxALT) level was defined as the highest ALT level documented during the follow-up period of an individual subject. To avoid the influence of ALT level by the status of disease progression, data of ALT within 2 years before mortality and around the time of HCC diagnosis were not included in the determination of maxALT level.

HBsAg Clearance.

For those with HBsAg persistent for more than 12 months, HBsAg seroclearance was defined as loss of serum HBsAg on two occasions at least 6 months apart and maintained to the end of the study.19

Liver Cirrhosis and Hepatic Steatosis.

Detection of liver cirrhosis was mainly based on US study supplemented with liver histology, thrombocytopenia, or endoscopic findings of varices. The detection of liver cirrhosis by US was according to a scoring system based on the US features of liver parenchyma, liver surface, hepatic vessel, and spleen size.21 The diagnosis of liver cirrhosis was made when the score was higher than or equal to 8 on at least two US studies more than 1 year apart and persisted up to the last US study, as described elesewhere.17

Hepatic steatosis was evaluated by US with a scoring system, based on the discrepancy of echogenicity between liver and kidney, the degree of posterior attenuation, and visibility of vessel.22 The diagnosis of hepatic steatosis was made when hepatic steatosis was detected by at least two US studies more than 1 year apart.

The accuracy of US diagnosis of cirrhosis and hepatic steatosis was assessed in a side study involving 100 patients who had received liver histology study between 2000 and 2005. The histological diagnosis of liver cirrhosis was made according to Ishak classification.23 Hepatic steatosis is defined as fat-containing cells constituting more than 5% of total hepatocytes. The correlation study showed that histological findings of cirrhosis and hepatosteatosis were closely associated with the US findings (r = 0.571, P < 0.001 for cirrhosis; r = 0.693, P < 0.001 for steatosis; Tai DI et al., unpublished data).

Cancer and Mortality.

In addition to their medical records, the study subjects were linked with the database of Cancer Registration, Chang Gung Memorial Hospital. This database has recorded information of all cancers found in this hospital since 1987.

National citizen identification numbers of the study subjects were also linked with our national mortality database, which has been established by the Statistics Office, Department of Health, Taiwan, since 1985. Cause of death was classified using the International Classifications of Diseases, Injuries and Causes of Death (ICD-9, World Health Organization, 1977).24

Statistic Analysis

To compare characteristics between groups, the chi-squared test, Fisher's exact test, student t test, linear-by-linear association, Pearson's correlation, and analysis of variance were used when appropriate. Survival analysis was done by Kaplan-Meier actuarial curve method with the log rank test. Univariate and multivariate analyses were performed to identify factors associated with mortality and HCC or cirrhosis development. Variables found to be significant in the univariate models were tested in a multivariate setting using the Cox proportional hazards regression models. All statistical analyses were performed using SPSS (version 11.5; SPSS Inc., Chicago, IL), and P < 0.05 was considered significant.

Results

A total of 4376 HBeAg-negative, HBsAg-positive subjects meeting our inclusion criteria (3673 with normal baseline ALT and 703 with ALT 1-2× ULN) were included in this analysis. They were aged 15 to 78 years (mean, 36 ± 9.6 years) at entry, and 59.6% of them were males. These subjects were followed for a total of 96,710 visits (average, 22.1 visits/case) and 66,953 US studies (average, 15.3 studies/case) during a mean follow-up period of 13.4 ± 5.2 years (median, 13.0 years; range, 3.0-28.7 years). Nineteen patients with ALT level increased over 2 times ULN in association with active HBV replication during follow-up received anti-HBV therapy (2 received thymosin-α1, 6 received interferon-α, and 11 received lamivudine). The follow-up data for these 19 patients were censored on the day of starting therapy. At entry, ALT level was less than 0.5 times ULN (low normal) in 1747 (39.9%), 0.5 to 1 times ULN (high normal) in 1926 (44.0%), and 1 to 2 times ULN in 703 (16.1%) subjects (Table 1).

Table 1. Changes of ALT Level Over Time During Follow-Up
Maximal ALT During Follow-upBaseline ALT
<0.5 × ULN0.5-1 × ULN1-2 × ULN
  1. Data expressed as number of patients (%).

<0.5 × ULN N = 314 (7.2%)314 (18%)
0.5-1 × ULN N = 1639 (37.5%)880 (50.4%)759 (39.4%)
1-2 × ULN N = 1579 (36.1%)391 (22.4%)756 (39.3%)432 (61.5%)
2-5 × ULN N = 584 (13.3%)117 (6.7%)288 (15.0%)179 (25.5%)
>5 × ULN N = 260 (5.9%)45 (2.6%)123 (6.4%)92 (13.1%)
Total cases 4376 (100%)1747 (100%)1926 (100%)703 (100%)

Changes of ALT Level over Time During Follow-up

Of the 703 subjects with a baseline ALT level between 1 and 2 times ULN, 156 (22.3%) showed ALT normalization during follow-up. They were included in the maxALT 1-2× ULN group. Of the 3673 subjects with normal baseline ALT level, ALT increased to 1 to 2 times ULN in 1147 (31.2%) and to 2 or more times ULN in 573 (15.6%) during follow-up. Survival analysis showed linear decline in the cumulative probability of persistently normal ALT (PNALT) in female subjects. Men showed a faster decrease in the cumulative probability of PNALT during the first 3 years of follow-up (65.7% versus 87% in women; log rank test, <0.001). ALT increased to greater than 1 times ULN in an additional 14.5% of men and 9.3% of women between the third and sixth years of follow-up. After that, the probability of ALT changes became similar between sexes.

MaxALT remained at low normal (<0.5 × ULN) in 314 (7.2%) subjects, was high normal (0.5-1 × ULN) in 1639 (37.5%), between 1 and 2 times ULN in 1579 (36.1%), between 2 and 5 times ULN in 584 (13.3%), and more than 5 times ULN in 260 (5.9%) subjects (Table 1). The follow-up period for ALT and survival were shortest in the maxALT less than 0.5 times ULN group and longest in the maxALT greater than 5 times ULN group (P < 0.001; Table 2).

Table 2. Demography in HBeAg-Negative Carriers with Different maxALT During Follow-Up
CategorySubcategoryMaximal ALT Level*P Value
<0.5 × ULN0.5-1 × ULN1-2 × ULN2-5 × ULN>5 × ULN
N = 314N = 1639N = 1579N = 584N = 260
  • Data in the same row with different letters (for example, a versus b; a versus c) indicate statistically significant difference between maxALT groups.

  • *

    Linear-by-linear association.

AgeAt entry33.3 ± 10.2a35.9 ± 10.1b35.8 ± 9.0b37.4 ± 9.5c37.9 ± 8.7c<0.001
Follow-upBy ALT8.9 ± 4.5a10.8 ± 5.0b12.3 ± 5.5c13.6 ± 5.4d14.2 ± 5.6d<0.001
(Year)By survival11.1 ± 4.5a12.6 ± 5.0b13.9 ± 5.3c15.0 ± 5.2d15.7 ± 5.0d<0.001
SexFemale83.4%51.4%29.6%25.0%19.2%<0.001
 Male16.6%48.6%70.4%75.0%80.8% 
ALT (U/L)At entry11.8 ± 3.3a18.4 ± 6.4b29.0 ± 13.7c31.2 ± 14.4d32.3 ± 14.3d<0.001
Hepatic steatosisPersisted over 1 year10.5%27.1%47.5%50.3%39.6%<0.001

Age at entry was younger (33.3 ± 10.2 years; range, 15-78 years) in subjects with maxALT less than 0.5 times ULN than that of other groups (mean, 35.8-37.9 years; range, 15-77 years). Female was predominant (83.4%) in the maxALT less than 0.5 times ULN group, probably because our laboratory chose 36 U/L as ULN for both males and females. Male was predominant (80.8%) in the maxALT greater than 5 times ULN group (P < 0.001),

Long-Term Outcomes

HBsAg seroclearance occurred in 565 (12.9%) patients at a mean age of 46.5 ± 10.6 years (median, 46.3 years; range, 21.9-86.8 years). HBsAg seroclearance rate was 9% to 17% without linear by linear association among maxALT groups (Table 3). Patients with hepatic steatosis tended to have a significantly higher incidence of HBsAg seroclearance (P < 0.001).

Table 3. Outcome in HBeAg-Negative Carriers with Different maxALT and Hepatic Steatosis During Follow-Up*
CategoryHepatic SteatosisMaximal ALT LevelP Value
<0.5 × ULN0.5-1 × ULN1-2 × ULN2-5 × ULN>5 × ULN
YesN = 33N = 444N = 750N = 294N = 103
NoN = 281N = 1195N = 829N = 290N = 157
  • *

    Linear-by-linear association.

  • Exclude 43 patients with cirrhosis whose maxALT appeared after their cirrhosis developments.

HBsAg clearanceYes9.1%13.5%16.7%17.0%13.6%NS
 No12.8%11.1%12.2%9.0%10.8%NS
CirrhosisYes00.2%0.3%1.7%3.9%<0.001
 No0.7%0.9%1.8%8.6%6.4%<0.001
HCCYes000.1%0.3%1.9%0.001
 No0.4%0.1%0.5%2.4%3.8%<0.001
Non-HCCYes03.6%1.1%2.4%2.9%NS
 CancersNo2.5%1.3%1.4%2.1%1.9%NS
Mortality,Yes00.5%0.7%0.3%1.9%NS
 TotalNo0.7%1.3%2.1%3.1%7.6%<0.001
Mortality, LiverYes000.3%0.3%1.0%0.071
 No000.5%2.1%3.8%<0.001
Mortality,Yes00.5%0.4%01.0%NS
 Non-liverNo0.7%1.3%1.6%1.0%3.8%0.025

Liver cirrhosis was detected in 118 patients (2.7%) during follow-up. Among them, 75 patients had their maxALT appear before cirrhosis development. The diagnosis of cirrhosis was confirmed histologically in 26 patients (including 13 patients with HCC). An additional seven patients had endoscopic evidence of varices. The mean age at diagnosis of liver cirrhosis was 48.4 ± 11.4 years (median, 47.7 years; range, 20.5-80.1 years). There was a trend of lower incidence of liver cirrhosis in patients with lower maxALT, especially when the effect of steatosis was excluded (P < 0.001, Table 3).

HCC was documented in 23 (0.5%) patients. Four HCC-related mortalities were reported by other hospitals to the national mortality database. For the rest of the patients, the diagnosis of HCC was confirmed by histology in 14, by cytology in two, and by images and AFP studies in three patients. Eleven patients received surgical treatment, and eight patients received local ablation or transhepatic arterial chemoembolization in this hospital. The mean age at diagnosis of HCC was 56.3 ± 10.8 years (median, 52.4 years; range, 40.9-80.2 years). There was a trend of higher incidence of HCC in patients with higher maxALT, especially when the effect of steatosis was excluded (P < 0.001). Non-HCC cancers were documented in 77 (1.8%) subjects. There was no association between the incidence of non-HCC cancers and maxALT of the subjects (Table 3).

Mortality was documented in 65 patients (1.5%) at a mean age of 58.2 ± 12.7 years (median, 57.1 years; range, 31.5-90.2 years). Twenty (0.5%) patients died of liver diseases at a mean age of 61.3 ± 8.8 years (median, 62.3 years; range, 48.9-80.6 years), 45 patients (1.0%) died of nonliver diseases. Liver-related mortality rate was zero in subjects with maxALT less than 1 times ULN, increased progressively to 3.8% in subjects with maxALT of at least 5 times ULN who had no evidence of hepatic steatosis (<0.001) but was not associated with maxALT level in the steatosis group (Table 3).

Subjects with steatosis had higher rates of HBsAg seroclearance andcancers other than HCC. Otherwise, all rates tended to be higher in subjects without steatosis.

Survival Analysis

Survival was higher in subjects with lower maxALT than in those with higher maxALT (P = 0.0031, Fig. 1). A statistically significant difference was observed between maxALT greater than 5 times ULN and other respective groups (P = 0.0578-0.0006). Liver-related mortality among subjects with maxALT less than 1 times ULN, maxALT 1 to 2 times ULN, and maxALT of 2 times ULN or higher are compared in Fig. 2. There was a trend toward higher liver-related mortality rates in subjects with higher maxALT (P = 0.0215 to <0.0001).

Figure 1.

Cumulative survival according to maxALT groups. A trend of higher cumulative survival in patients with lower maxALT group was found (P = 0.0031). Statistical significant differences were observed between maxALT greater than 5 × ULN with other groups (P = 0.0578-0.0006).

Figure 2.

Cumulative liver-related mortality in three maxALT groups. A trend of higher cumulative probability of liver-related mortality in patients with higher maxALT groups (P < 0.001). A higher cumulative liver-related mortality is poor in the maxALT 1 to 2 times ULN group than in the maxALT less than 1 times ULN group (P = 0.010). The cumulative liver-related mortality in the maxALT 1 to 2 times ULN group was lower than the maxALT greater than 2 times ULN groups (P = 0.0215).

The cumulative probability of HCC development was significantly associated with maxALT level (P < 0.0001). A higher cumulative probability of HCC was found in patients with maxALT 2 to 5 times ULN or maxALT of at least 5 times ULN than in those with maxALT 0.5 to 1.0 times ULN or maxALT 1 to 2 times ULN (P = 0.0133 to less than 0.0001, Fig. 3).

Figure 3.

Cumulative probability of HCC during follow-up. A trend of higher cumulative probability of HCC in patients with higher maxALT groups (P < 0.001). Higher cumulative probability of HCC is seen in maxALT greater than 2 times ULN groups than in those with maxALT 0.5 to 1 times ULN or 1 to 2 times ULN groups (P = 0.0133 to <0.0001).

Multivariate Analysis

Factors associated with the development of liver cirrhosis in long-term follow-up were examined by Cox regression analysis. Sex, age at entry, maxALT level, hepatic steatosis, and HBsAg clearance were included in the analysis. Sex, hepatic steatosis, age at entry, and maxALT levels were found to be independent factors associated with liver cirrhosis development (Table 4).

Table 4. Cox Regression Analysis
Factors Associated with Development of Liver Cirrhosis
FactorHazard Ratio95.0% Confidence intervalP Value
LowerUpper
Female sex0.3620.2140.6130.000
Hepatic steatosis0.1940.1180.3180.000
Age at entry1.0631.0461.0800.000
MaxALT   0.000
 >5 × ULN6.1151.41326.4560.015
 2-5 × ULN5.5851.32623.5290.019
 1-2 × ULN2.0350.4818.6030.334
 0.5-1 × ULN0.8320.1893.6660.808
 <0.5 × ULN1.0   
Factors Associated with Development of HCC
FactorHazard Ratio95.0% Confidence IntervalP Value
LowerUpper
Liver cirrhosis21.0458.28053.4890.000
Age at entry1.0651.0271.1050.001
MaxALT   0.057
 >5 × ULN2.1680.26617.6610.470
 2-5 × ULN1.0540.1308.5370.961
 1-2 × ULN0.5420.0634.6540.577
 0.5-1 × ULN0.1660.0102.6640.205
 <0.5 × ULN1.0   
Factors Associated with Total Mortality
FactorHazard ratio95.0% Confidence intervalP Value
LowerUpper
HBsAg clearance0.4400.1990.9720.042
Liver cirrhosis3.6061.9566.6480.000
Hepatic steatosis0.3170.1600.6270.001
Age at entry1.0931.0701.1160.000

Factors associated with HCC and mortality were also examined by Cox regression. Sex, age at entry, maxALT level, hepatic steatosis, liver cirrhosis, and HBsAg clearance were included in the analysis. Liver cirrhosis, age at entry, and maxALT levels were found to be independent factors associated with HCC development (Table 4). Liver cirrhosis, hepatic steatosis, HBsAg clearance, and age at entry were independent factors associated with mortality (Table 4).

Discussion

ALT is the most convenient and cheapest surrogate marker of liver injury.14 Because chronic HBV infection is a dynamic state of interactions between HBV, the liver cells, and the immune cells of the host, ALT level may fluctuate with greatly variable patterns during the course of chronic HBV infection.13 It is not easy to classify ALT changes over time properly. Using maximal ALT level over time to classify patients is relatively simple and is especially useful in identifying patients with PNALT level over a long period. This classification is meaningful, as shown in a recent study involving 105 HBsAg-positive patients, in that patients with at least one ALT approximately 1 to 2 times ULN in 12 months had a much higher chance of significant liver histological changes than those with PNALT.15 Our results further show that HBeAg-negative HBsAg carriers who maintained their ALT at less than 1 times ULN during long-term follow-up had excellent prognosis. They had higher survival and lower incidence of liver cirrhosis and HCC compared with those with maxALT greater than 1 times ULN, particularly those with greater than 2 times ULN.

Two large studies involving subjects who received health examinations showed that baseline serum ALT level was associated with liver-related mortality during a follow-up period of 8 to11 years.11, 12 Although viral markers were not tested and only ALT level at entry was used, Kim et al.11 suggested that the risk of liver-related mortality increased even in patients with high normal baseline ALT level and suggested revising the ULN of ALT from 35 to 40 U/L to 30 U/L for men. A large series with subjects who attempted to donate blood also revealed that using conventional ULN of ALT may miss the patient with liver injury in chronic hepatitis C or nonalcoholic liver disease and suggested lowering the ULN of ALT to 19 U/L for women.25 Another study in HBeAg-negative HBsAg carriers with normal ALT for 2 years showed a higher HBV DNA level in patients with high normal ALT (0.5-1 × ULN) than patients with low normal ALT (<0.5 × ULN).26 All these studies showed differences between subjects with low normal ALT (<0.5 × ULN) level and those with high normal ALT level (0.5-1.0 × ULN). Of note is that these studies depended on a single ALT level. In contrast, the study using PNALT up to 2 years as baseline showed no difference between subjects with low normal ALT and high normal ALT, in that none of them encountered ALT elevation greater than 1 times ULN during 12 to 77 months' follow-up.26 However, as shown in Table 1, ALT levels of subjects with chronic HBV infection were changing over a long period, and more than 45% of our subjects with low or high normal ALT level at entry had their ALT levels increased to abnormal level (31.2% to 1-2 × ULN, 11.0% to 2-5 × ULN and 4.6% to >5 × ULN) during a mean follow-up period of 9 to 10 years. Given the findings that patients with higher maxALT had a significantly longer duration of follow-up (Table 2), it is possible that even more patients with normal baseline ALT will encounter ALT elevation greater than 1 × ULN if they were followed for longer amount of time. Obviously, ALT level at entry is not a reliable marker of a dynamic changing disease process, such as chronic HBV infection.

Notably, our results showed no difference in the long-term outcomes between subjects with low normal maxALT (<0.5 × ULN) and those with high normal maxALT (0.5-1 × ULN). A recent study in 35 HBeAg-negative HBsAg carriers with PNALT for more than 1 year also showed that there was no difference in liver histology between those with ALT less than 0.5 times ULN and those with ALT of 0.5 to 1 times ULN.27 Our results have shown that 50.4% of the patients with baseline ALT less than 0.5 × ULN had a maxALT of 0.5 to 1 times ULN during long-term follow-up (Table 1) and had similar prognosis to those who maintained their ALT at less than 0.5 times ULN. These results are contrary to the previous observation based on a single ALT at entry and clearly indicate that there is no need to lower the ULN of ALT.11, 28

Several recent studies evaluated liver histology in HBsAg carriers with PNALT.14, 29–32 The definition of so-called “PNALT” in these studies varies from a single ALT level to normal ALT in a mean of 3 years. Patients with PNALT level identified in the study with a mean follow-up of 3.2 years (range, 0.5-11 years) were associated with a milder fibrosis or inflammation,30 as compared with those identified in a single ALT level or PNALT for 6 to 12 months.29, 31, 32 Another study with sequential HBV DNA measurements suggested that 3 years' observation was needed for determining long-term outcome in HBeAg-negative patients.33 With survival analysis in the current study, a rapid decline of the cumulative probability of PNALT was observed in men in the initial 3 years of follow-up. Our observations add support to the notion that at least 3 years are needed to decide whether a HBsAg carrier has PNALT, and that assessment of prognosis based on data at a single time point is inappropriate.

The results of the current study have also shown that the incidences of cirrhosis and HCC tend to increase in patients with higher maxALT and are definitely higher in groups with maxALT of at least 2 times ULN than in groups with maxALT less than 1 times ULN (Tables 3, 4). The differences were confirmed by survival analysis (Figs. 1–3). These data lend support to the recommendation that viremic patients with ALT level greater than 2 times ULN require antiviral therapy.7–9 In areas of the world in which HBV-DNA testing is not available or not affordable, ALT of at least 2 times ULN might be useful in selecting patients for treatment as long as other potential contributing liver conditions are ruled out. For patients with maxALT of 1 to 2 times ULN, their liver-related mortality was only slightly higher than for those with maxALT less than 1 times ULN (Fig. 2). This is contradictory to the finding of Yuen et al.34 that patients with ALT level of 1 to 2 times ULN at entry had highest liver disease–related complications. The only possible explanation for the surprising finding of Yuen et al.34 is that the worst prognosis was the results of subsequent dynamic changes during the course, rather than ALT 1 to 2 times ULN per se. This discrepancy again suggests that ALT at a single point is not reliable in a dynamic changing disease such as chronic HBV infection.

Of note is that the differences in long-term outcomes between subjects with maxALT less than 1 × ULN and maxALT of 1 to 2 times ULN increased significantly when subjects with ultrasonographic evidence of hepatic steatosis were excluded (Table 3), confirming that hepatosteatosis may have confounded or exaggerated ALT elevation.35 These findings suggest that liver histology evaluation should be done in viremic patients with ALT levels between 1 and 2 times ULN. If histological study reveals more than mild liver inflammation and fibrosis and other causes of ALT elevation such as steatosis are excluded, anti-HBV therapy should be considered. Although patients with ALT 1 to 2 times ULN respond to 1-year therapy with direct antiviral agent poorly,10 they respond to therapy with pegylated interferon alpha 2a or long-term lamivudine therapy as well as patients with higher ALT levels.36–38

The original purpose of this HBsAg clinic was to detect and treat HCC in the early stages. Metabolic syndrome and fatty liver were not important issues at that time. However, body mass index and hepatic steatosis were found to be associated with HBsAg clearance in our recent case-control study.39 The US data in this study also suggest that hepatic steatosis is associated with a higher rate of HBsAg seroclearance and has a negative association with disease progression (Table 3). Perhaps increased HBsAg seroclearance in our patients with hepatic steatosis has overruled the adverse effect of hepatic steatosis. In addition, a recent study has shown that HBV is regulated by changes in the nutritional state through the metabolic regulator peroxisome proliferator-activated receptor gamma, coactivator 1 alpha.40 The conceivably improved nutrition status in patients with steatosis may have a positive effect in chronic HBV infection.

A large community-based cohort study in 2762 HBsAg carriers (81% HBeAg-negative; 94% ALT < ULN), also from Taiwan, has shown that serum HBV-DNA greater than 104 copies/mL, genotype C, and basal core promoter A1762T/G1764A double mutations are independent risk factors associated with HCC development.41 Our patients are similar to those participants of the above-mentioned study, and one of the limitations of our study is that we did not include information about HBV genotype and PC/BCP mutations. Nevertheless, the primary aim of this study was to assess the value of serum ALT alone because ALT assay is the cheapest and most accessible way to monitor patients and can be measured frequently in short intervals. In contrast, HBV assays are not widely available and are expensive. In addition, patients with active HBV replication have a higher chance of ALT elevation during long-term observation, whereas patients with low HBV replication have a higher chance of maintaining PNALT.42–44 Furthermore, serum ALT level associates with liver injury more directly than serum HBV-DNA level at the same time point, as shown in a study that most patients with PNALT and HBV DNA of 2000 to 19,999 IU/mL showed minimal histological changes.29 Therefore, our data support the idea that patients with ALT of at least 2 times ULN may be selected for anti-HBV therapy in an area in which HBV DNA assay is not available.

Another limitation of this study is that most of the diagnoses of cirrhosis and hepatic steatosis were made according to US findings. Patients with bridging fibrosis and early cirrhosis may not have been detected both at entry and during follow-up by US. However, the US criteria used in this study were based on a study involving our earlier patients, with early cirrhosis confirmed by follow-up biopsy.21 The sensitivity and specificity in detecting hepatic steatosis were also reported to be 89% to 94% and 84% to 93%, respectively.22, 45 In addition, the correlation of US and histological findings were confirmed in a side study involving 100 patients, as described in Patients and Methods.

In conclusion, the long-term prognosis of HBeAg-negative HBsAg carriers is excellent as long as they maintain PNALT, regardless of their HBV replication status; it seems appropriate to extend the follow-up interval from 6 months to 1 year in patients with PNALT over 3 years who have no other risk factors for HCC. In contrast, patients with ALT at least 2 × ULN may require antiviral therapy. For patients with ALT 1 to 2 × ULN, a liver biopsy is recommended, and antiviral therapy is considered if significant necroinflammation or fibrosis is evident. ALT level at a single time or during a short period is not a good prognostic tool for a changing disease such as chronic hepatitis B.

Acknowledgements

The authors thank Su-Chiung Chu for secretary assistance.

Ancillary