Hepatitis B virus (HBV) infection is a global

Transcription

1 VIRAL HEPATITIS Serum Hepatitis B Surface Antigen Levels Help Predict Disease Progression in Patients With Low Hepatitis B Virus Loads Tai-Chung Tseng, 1,3,8 Chun-Jen Liu, 2,3 Hung-Chih Yang, 2,6 Tung-Hung Su, 2,3 Chia-Chi Wang, 1,8 Chi-Ling Chen, 3 Cheng-An Hsu, 7 Stephanie Fang-Tzu Kuo, 9 Chen-Hua Liu, 2,3 Pei-Jer Chen, 2,3 Ding-Shinn Chen, 2,3 and Jia-Horng Kao 2,3,4,5 Chronic hepatitis B patients with high viral loads are at increased risk of cirrhosis and hepatocellular carcinoma (HCC). In patients with low viral loads, higher hepatitis B surface antigen (HBsAg) levels have been shown to predict HCC development. However, little is known about the difference in risk for other hepatitis B virus (HBV)-related adverse outcomes with varying HBsAg levels. A total of 1,068 Taiwanese hepatitis B e antigen (HBeAg)-negative HBV carriers with serum HBV DNA level <2,000 IU/mL at baseline were followed for a mean duration of 13.0 years. Patients were categorized based on their HBsAg levels, and the relationships between HBsAg level and development of HBeAg-negative hepatitis, hepatitis flare, and cirrhosis were investigated. Of the 1068 patients with low viral loads, 280 developed HBeAg-negative hepatitis, with an annual incidence rate of 2.0%. HBsAg level, but not HBV DNA level, was found to be a risk factor for HBeAg-negative hepatitis. Multivariate analysis showed that the adjusted hazard ratio in patients with an HBsAg level 1,000 versus <1000 IU/mL was 1.5 (95% confidence interval, ). The positive correlation was present when evaluating other endpoints, including hepatitis flare and cirrhosis, and remained consistent when the study population was restricted to those with normal alanine aminotransferase (ALT) level at baseline. The annual incidence rate of HBeAg-negative hepatitis was lowered to 1.1% in patients with low levels of HBV DNA, HBsAg, and ALT. Conclusion: In HBeAg-negative patients with low viral loads and genotype B or C virus infection, a higher HBsAg level can predict disease progression. HBsAg <1,000 IU/mL in combination with low levels of HBV DNA and ALT help define minimal-risk HBV carriers. (HEPATOLOGY 2013;57: ) Hepatitis B virus (HBV) infection is a global health problem, resulting in more than 1 million deaths per year. 1 Patients with chronic HBV infection are at risk of developing cirrhosis, hepatic decompensation, and hepatocellular carcinoma (HCC), with an estimated lifetime risk of 25% 40% in carriers who acquire the virus early in life. 1 4 Several cohort studies have indicated that serum HBV DNA is the major driving force of disease progression in patients with chronic HBV infection. 5 7 For instance, risk of HCC in HBV carriers starts to increase when the viral load is higher than 2,000 IU/ ml at study entry. 6 In contrast, patients with low viral loads (HBV DNA <2,000 IU/mL) are usually defined as low-risk HBV carriers. 3,5,6,8 However, the results Abbreviations: ALT, alanine aminotransferase; CI, confidence interval; ENH, HBeAg-negative hepatitis; 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; HR, hazard ratio; PCR, polymerase chain reaction. From the 1 Division of Gastroenterology, Department of Internal Medicine, Buddhist Tzu Chi General Hospital Taipei Branch, Taipei, Taiwan; 2 Division of Gastroenterology, Department of Internal Medicine, Taipei, Taiwan; 3 Graduate Institute of Clinical Medicine, Taipei, Taiwan; 4 Hepatitis Research Center, and the Departments of Medical Research, Taipei, Taiwan; 5 Departments of Medical Research, Taipei, Taiwan; 6 Departments of Microbiology, Taipei, Taiwan; 7 Laboratory Medicine, National Taiwan University College of Medicine and National Taiwan University Hospital, Taipei, Taiwan; 8 School of Medicine, Tzu Chi University, Hualien, Taiwan; and 9 St. Vincent s Hospital, Melbourne, Victoria, Australia. Received March 5, 2012; Revised July 31, 2012; accepted August 5, Supported by grants from the Buddhist Tzu-Chi General Hospital Taipei Branch (TCRD-TPE-100-C1-3), the National Taiwan University Hospital (NTUH100-S1534), the Department of Health (DOH99-DC-1001 and DOH100-DC-1019), and the National Science Council, Executive Yuan, Taiwan (NSC B , NSC B , and NSC B ). 441

2 442 TSENG ET AL. HEPATOLOGY, February 2013 from our recent cohort indicated that the prognosis of low-viremic patients is variable, and hepatitis B surface antigen (HBsAg) level can serve as a marker to stratify HCC risk. 9 More specifically, HCC risk in low-viremic patients with HBsAg level 1,000 IU/mL is higher than those with HBsAg level <1,000 IU/mL. 9 Nevertheless, the underlying mechanisms remain unclear and warrant further investigation.it is generally believed that, in HBV carriers, cirrhosis usually results from the accumulation of extracellular matrix from liver cell injury, and HCC may subsequently emerge in the setting of cirrhosis. 10,11 The major driving force of all these adverse events has been shown to be serum HBV DNA level. 5 7 However, HBV DNA level fluctuates easily, whereas HBsAg level remains relatively stable. 9,12 In fact, in a recent cohort study enrolling patients infected with genotype D virus, HBsAg level <1,000 IU/mL was shown to be an effective criterion for defining a 3- year inactive carrier state in patients with HBV DNA level <2,000 IU/mL. 13 Based on these findings, we hypothesized that in low-viremic patients infected with genotype B or C virus, a higher HBsAg level indicates a less sustained viral suppression and therefore induces higher risks of hepatitis B e antigen (HBeAg)-negative hepatitis and cirrhosis. If this hypothesis holds true, we may partially explain the positive correlation between HBsAg level and HCC in low-viremic patients. Furthermore, we may claim that a higher HBsAg level may serve as an indicator for active liver disease in low-viremic patients with genotype B or C virus infection. To address this important issue, we enrolled a large cohort of 2,688 treatment-naïve patients who were diagnosed with chronic HBV infection and received long-term follow-up at the National Taiwan University Hospital. The primary aim of our study was to explore whether a higher HBsAg level is associated with increased risks of HBeAg-negative hepatitis (ENH), multiple ENH, hepatitis flare, and cirrhosis in patients with HBV DNA level <2,000 IU/mL. Patients and Methods Patient Cohort. Figure 1 shows the inclusion and exclusion criteria for the ERADICATE-B (Elucidation Fig. 1. Flow of study participants. of Risk factors for DIsease Control or Advancement in Taiwanese hepatitis B carriers) cohort. 9 A total of 3,947 HBsAg-positive patients aged above 28 years were enrolled consecutively between 1985 and All of the patients had been HBsAg-positive for more than 6 months and received over 3 years of regular follow-up at the National Taiwan University Hospital. After excluding patients with evidence of hepatitis C virus (HCV) or hepatitis D virus (HDV) coinfection and patients without adequate serum samples for analysis, 3,489 patients remained. We further excluded 411 patients who were diagnosed with cirrhosis at baseline and 390 patients who received antiviral therapy before the end of follow-up. After these exclusions, a total of 2,688 HBV carriers were included. Among these, 1,068 HBeAg-negative patients who had HBV DNA <2,000 IU/mL at baseline were our main target in the study. All enrolled patients gave informed Address reprint requests to: Jia-Horng Kao, Director, Graduate Institute of Clinical Medicine, National Taiwan University College of Medicine, 1 Chang-Te Street, Taipei 10002, Taiwan. kaojh@ntu.edu.tw; fax: (886) Copyright VC 2012 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI /hep Potential conflict of interest: Nothing to report. Additional Supporting Information may be found in the online version of this article.

3 HEPATOLOGY, Vol. 57, No. 2, 2013 TSENG ET AL. 443 consent as approved by the National Taiwan University Hospital ethical committee. Data Collection. Patients were tested for serological markers (HBsAg, HBeAg, anti-hbe, antibodies against HCV [anti-hcv], antibodies against HDV [anti- HDV]) and underwent liver function tests and alphafetoprotein measurements at baseline. Throughout follow-up, if alanine aminotransferase (ALT) levels remained within normal limits, liver function and alpha-fetoprotein levels were assayed every 6 months; if ALT levels were elevated, the assays were conducted at least every 3 months. Serum samples collected at each visit were stored at 20 C until analysis. Serum alpha-fetoprotein and abdominal ultrasonography using a high-resolution real-time scanner were checked for HCC surveillance every 3 to 6 months from enrollment. Diagnosis of HBeAg-Negative Hepatitis, Multiple HBeAg-Negative Hepatitis, Hepatitis Flare, Cirrhosis, and HCC. ENH was defined as serum ALT more than twice the upper limit of normal with a concomitant serum HBV DNA level 2,000 IU/mL 5,12,14 (the upper limit of normal for serum ALT is 40 U/L). Multiple ENH was defined as having at least two episodes of ENH during follow-up. Person-years for patients who achieved multiple ENH were censored when their first episode of ENH occurred. Hepatitis flare was defined as ALT elevation more than five times the upper limit of normal with a concomitant serum HBV DNA level 2,000 IU/mL. 15 Apart from the aforementioned criteria, exclusion of drug or alcohol use and absence of serological evidence suggestive of other viral hepatitis infection or autoimmune liver disease were essential to ensure the correct diagnosis. In addition, because real-time polymerase chain reaction (PCR) for HBV quantification was not available until 2004 in Taiwan, most HBV DNA levels were determined retrospectively using stored sera. Cirrhosis was diagnosed using histology or ultrasonographic findings together with clinical features such as thrombocytopenia, gastroesophageal varices, or ascites. 11,14,16,17 Detection of liver cirrhosis via abdominal ultrasonography was performed according to a scoring system based on the features of liver parenchyma, liver surface, hepatic vessel, and spleen size. For ultrasonographic diagnosis of cirrhosis to be made, the score had to be 8 on at least two ultrasonographic studies more than 6 months apart. 7,16,17 HCC was diagnosed either via histology/cytology or via typical image findings (arterial enhancement and venous wash-out by contrast-enhanced computed tomography or magnetic resonance imaging) in hepatic nodules >1 cm. 18 Serological Assays. Serum HBsAg, HBeAg, anti- HBe, anti-hcv, and anti-hdv were tested using commercial assays (Abbott Laboratories, Abbott Park, IL). Quantification of HBV DNA and HBsAg Levels. Serum samples at enrollment and the third year of follow-up were tested for both HBV DNA and HBsAg levels. HBV DNA levels were quantified using the Abbott RealTime HBV assay, 0.2 ml protocol (Abbott Laboratories, Abbott Park, IL) with the lower detection limit being 15 IU/mL. HBsAg levels were quantified using the Architect HBsAg QT assay (Abbott Laboratories, Abbott Park, IL) according to the manufacturer s instructions. 12,19,20 The detection range of the Architect assay is IU/mL; if the HBsAg level was found to be higher than 250 IU/mL, the samples were diluted to 1:100 to 1:1,000 to obtain a reading within the calibration curve range. Extraction of Viral DNA. Viral DNA in the serum was extracted using commercial kits (QIAamp DNA Blood and Tissue Mini Kit; QIAGEN Inc., Valencia, CA). The extracted DNA was used to genotype HBV. Determining HBV Genotype. HBV genotype was determined by real-time PCR-based single-tube assay as described. 21 This method consists of two consecutive steps. The first step uses PCR to amplify the region (nucleotides ), and the second step uses melting curve analysis to genotype HBV. Statistical Analysis. Mean and SD were calculated for continuous variables and percentages were used for categorical variables. The clinical follow-up started at the time of enrollment. Person-years were censored on the date that each endpoint was identified, death, the last date of follow-up, or June 30, 2011, whichever came first. The cumulative incidence of each endpoint by different variables was derived using the Kaplan- Meier curve analysis, and a log-rank test was used to test for the statistical difference. In this study, ENH, multiple ENH, hepatitis flare, and cirrhosis were adopted as endpoints. Both HBV DNA and HBsAg levels were categorized using a log 10 scale. HBV DNA levels were categorized into <200 IU/mL and 200-1,999 IU/mL. HBsAg levels were first categorized by three different cutoffs: 10, 100, and 1,000 IU/mL to determine which cutoff is more viable. Cox proportional hazards regression model was used to calculate the crude and multivariate-adjusted hazard ratio (HR) and 95% confidence interval (CI) of each factor for ENH, multiple ENH, hepatitis flare, and cirrhosis development. Age, sex, and serum levels of HBV DNA, ALT, and HBsAg were included as covariates.

4 444 TSENG ET AL. HEPATOLOGY, February 2013 In addition, we also performed two subgroup analyses. First, we restricted our study population to those who received follow-up for at least 14 years. Receiver operating characteristic curve analysis was used to evaluate the performance of HBsAg level in predicting 14-year ENH. The sensitivities and specificities of different HBsAg cutoffs were calculated. The optimal cutoff level of HBsAg was determined using the Youden index, which was defined as sensitivity þ specificity 1. Second, we analyzed the relationship between HBsAg level and ENH in patients with HBV DNA level <2,000 IU/mL plus ALT level <40 U/L at baseline (normal ALT) using the Cox proportional hazards regression model. We also investigated the relationships between ENH and hepatitis flare during follow-up and cirrhosis development. First, we designated ENH/hepatitis flare as time-dependent variables, because both events occurred during the follow-up. Second, the person-years of follow-up in this analysis were censored when cirrhosis was diagnosed or at the end of follow-up. The Cox proportional hazard regression model was adopted for the following analysis. All statistical tests were 2-tailed and P < 0.05 was considered as statistical difference. All analyses were performed using Stata statistical software (version 10.0; Stata Corp, College Station, TX). Results Baseline Characteristics and Follow-up Results. Table 1 shows the baseline characteristics of the 1,068 HBeAg-negative patients with HBV DNA level <2,000 IU/mL. Of these patients, 600 (56.2%) were men; 910 (85.2%) had ALT levels <40 U/L; 438 (41.0%) had HBV DNA levels <200 IU/mL; and 585 (54.8%) had HBsAg levels <1,000 IU/mL. The follow-up results are summarized in Supporting Table 1. Our study had 13,904.4 person-years of follow-up, with an average follow-up duration of years (mean 6 SD [median, 13.3 years; range, years]). Throughout the follow up period, 280 patients developed ENH, with the incidence rate being 2.0%. Among them, 205 (73.2%) patients had more than one episode of ENH during the follow up and met the criteria of multiple ENH. The annual incidence rates of multiple ENH, hepatitis flare, and cirrhosis were 1.4%, 0.6%, and 0.3%, respectively. In terms of diagnosing endpoints, 55 (19.6%) patients with ENH, 33 (33.7%) patients with hepatitis flare, and 16 (40.0%) patients with cirrhosis were confirmed on liver biopsy. We further evaluated the relationship between hepatitis activity during follow-up and cirrhosis Table 1. Baseline Characteristics of 1,068 HBeAg-Negative Patients with HBV DNA Level <2,000 IU/mL Characteristic n (%) Sex Women 468 (43.8) Men 600 (56.2) Age at enrollment, years (52.9) (29.7) (12.4) (5.1) Serum ALT level, U/L < (54.5) (30.7) (14.8) Serum HBV DNA level, IU/mL < (41.0) 200 1, (59.0) Serum HBsAg level, IU/mL < (11.0) (15.6) (28.2) 1,000 9, (42.4) 10, (2.8) HBV genotype* B 517 (82.1) C 103 (16.4) Undetermined 10 (1.6) *Available only in patients with an HBV DNA level between 200 and 1,999 IU/mL (n ¼ 630) because of the detection limit of a PCR-based assay. development. We found that both ENH and hepatitis flare during follow-up were associated with subsequent cirrhosis development (Supporting Table 2). Determining Optimal HBsAg Cutoff in Predicting HBeAg-Negative Hepatitis. We first investigated the relationship between HBsAg level and ENH and found that the HR was 1.1 (95% CI, ; P ¼ 0.025) for every log 10 increase of HBsAg level. After identifying a positive correlation between HBsAg and ENH, we correlated the cumulative incidence of ENH with three different HBsAg level cutoffs: 10 IU/mL (Fig. 2A), 100 IU/mL (Fig. 2B), and 1,000 IU/mL (Fig. 2C) and analyzed the data using the Kaplan- Meier curve analysis. Difference in cumulative incidence of ENH was found only when patients were categorized by HBsAg level of 1,000 IU/mL (P ¼ 0.004), but not 10 IU/mL (P ¼ 0.260) or 100 IU/mL (P ¼.165). In addition, we adopted receiver operating characteristic curve analysis to evaluate using HBsAg level to predict ENH development in the subcohort of 14 years of follow-up (n ¼ 718, 243 develop ENH). This time frame was chosen because the median period of follow-up was 13.3 years. The area under the receiver operating characteristic curve for HBsAg level was 0.56 (95% CI, ). We calculated the

5 HEPATOLOGY, Vol. 57, No. 2, 2013 TSENG ET AL. 445 Fig. 2. Cumulative incidence of HBeAg-negative hepatitis was analyzed by three HBsAg cutoffs of (A) 10 IU/mL, (B) 100 IU/mL, and (C) 1,000 IU/mL in 1,068 HBeAg-negative patients with HBV DNA <2,000 IU/mL. Cumulative incidence of (D) multiple HBeAg-negative hepatitis, (E) hepatitis flare, and (F) cirrhosis development was analyzed using a cutoff level of 1,000 IU/mL.

6 446 TSENG ET AL. HEPATOLOGY, February 2013 Table 2. Cox Regression Analysis of Factors Associated with HBeAg-Negative Hepatitis Characteristic Person-years of Follow-up n Annual Incidence Rate* Crude HR (95% CI) P Adjusted HR (95% CI) P Sex Women 6, , Men 7, , ( ) < ( ) Age at enrollment, years , , , , ( ) ( ) , , ( ) ( ) , ( ) ( ) Serum ALT level, U/L <20 8, , , , ( ) < ( ) , , ( ) < ( ) <0.001 Serum HBV DNA level, IU/mL <200 5, , , , ( ) ( ) Serum HBsAg level, IU/mL <1,000 7, , ,000 6, , ( ) ( ) HBV genotype B 6, , C 1, , ( ) Undetermined , ( ) *Per 100,000 person-years. Available only in patients with an HBV DNA level between 200 and 1,999 IU/mL (n ¼ 630) because of the detection limit of a PCR-based assay. HBV genotype was thus not included as an adjusting variable in multivariate analysis. sensitivities and specificities of different cutoffs and determined the optimal cutoff using the Youden index (Supporting Table 3). The optimal cutoff was IU/mL, which was quite close to 1,000 IU/mL. Because a round number is easier to memorize and to use clinically, we decided to adopt an HBsAg level of 1,000 IU/mL as the cutoff in our analysis. Factors Affecting ENH Risk in HBeAg-Negative Patients with Low Viral Loads. In HBeAg-negative patients with HBV DNA levels <2,000 IU/mL, advanced age, male sex, and elevated ALT level (but not elevated HBV DNA level) were found to be independent risk factors for ENH development (Table 2). In terms of HBsAg level, univariate analysis revealed that the HR of ENH development was 1.4 (95% CI, ) for HBsAg levels 1,000 IU/mL when compared with HBsAg levels <1,000 IU/mL. Multivariate analysis showed that HBsAg level 1,000 IU/mL remained an independent risk factor of ENH development with an HR of 1.5 (95% CI, ). Factors Affecting Risks of Multiple HBeAg-Negative Hepatitis, Hepatitis Flare, and Cirrhosis in HBeAg-Negative Patients with Low Viral Loads. To test whether the positive correlation between HBsAg and ENH is robust, we investigated the relationships between HBsAg level and other endpoints, including multiple ENH, hepatitis flare, and cirrhosis. When performing Kaplan-Meier curve analysis, we consistently found that HBsAg 1,000 IU/mL, compared with HBsAg <1,000 IU/mL, was associated with higher risks of multiple ENH (Fig. 2D, P ¼ 0.003), hepatitis flare (Fig. 2E, P ¼ 0.004), and cirrhosis development (Fig. 2F, P ¼ 0.017). After adjustment for age, sex, and ALT and HBV DNA levels, HBsAg 1,000 IU/mL remained a risk factor for multiple ENH (HR, 1.7; 95% CI, ), hepatitis flare (HR, 2.3; 95% CI, ), and cirrhosis development (HR, 4.1; 95% CI, ) (Supporting Tables 4, 5, and 6). Factors Affecting ENH Risk in Patients with Low Viral Loads Plus ALT Level <40 U/L at Baseline. In order to determine whether HBsAg level can help identify minimal-risk HBV carriers, the role of HBsAg was investigated in the subcohort of patients with HBV DNA <2,000 IU/mL plus ALT <40 U/L at baseline (n ¼ 910). We consistently found that HBsAg 1,000 IU/mL, compared with HBsAg <1,000 IU/mL, was associated with a higher ENH risk with the HR of 1.8 (95% CI, ) using multivariate analysis (Table 3). Relationship Between Baseline HBsAg Level and Change of HBV DNA Level Between Baseline and Year 3 of Follow-up. Because a nonsustained viral suppression is believed to be associated with subsequent hepatitis activity, we determined HBV DNA levels at year 3 of follow-up in those who had available serum samples (n ¼ 980) and investigated whether a higher HBsAg level at study entry was associated with

7 HEPATOLOGY, Vol. 57, No. 2, 2013 TSENG ET AL. 447 Table 3. Cox Regression Analysis of Risk Factors Associated with HBeAg-negative Hepatitis in 910 Patients with HBV DNA Level <2,000 IU/mL Plus ALT level <40 U/L at Baseline Characteristic Person-years of follow-up n Annual Incidence Rate* Crude HR (95% CI) P Adjusted HR (95% CI) P Sex Women 6, , Men 6, , ( ) < ( ) Age at enrollment, years , , , , ( ) ( ) , , ( ) ( ) , ( ) ( ) Serum ALT level, U/L <20 8, , , , ( ) < ( ) Serum HBV DNA level, IU/mL <200 5, , ,999 7, , ( ) ( ) Serum HBsAg level, IU/mL <1,000 6, , ,000 5, , ( ) < ( ) <0.001 HBV genotype B 6, , C 1, ( ) Undetermined ( ) *Per 100,000 person-years. Available only in patients with an HBV DNA level between 200 and 1,999 IU/mL (n ¼ 630) because of the detection limit of a PCR-based assay. HBV genotype was thus not included as an adjusting variable on multivariate analysis. a subsequent surge of HBV DNA. It was found that baseline HBsAg level 1,000 IU/mL was associated with an upsurge of HBV DNA level 2,000 IU/mL at year 3 of follow-up in different clinical settings, including the overall cohort (18.3% versus 10.5%, P < 0.001) and the subcohort with ALT level <40 U/L at baseline (17.2% versus 9.9%, P ¼ 0.002) (Fig. 3). Effect of the Changes of HBV DNA, HbsAg, and ALT Levels on ENH Risk. After determining HBsAg and HBV DNA levels in patients who had available serum at year 3 of follow-up, we enrolled only those patients who had virological data and did not develop ENH within the first 3 years of follow-up (n ¼ 905) and a subcohort with above criteria plus ALT level <40 U/L at baseline (n ¼ 797). We aimed to use this subgroup analysis to investigate the impact of changes of viral factors on ENH development, and the results are summarized in Table 4. Using univariate analysis, compared with patients with persistently low levels of HBV DNA, HBsAg, or ALT, patients with persistently high levels of any of these three factors were shown to have a higher risk of ENH (Table 4). For example, compared with patients with HBsAg level <1,000 IU/ ml at baseline and year 3 of follow-up, the HR of ENH was 1.7 (95% CI, ) for patients with HBsAg levels 1,000 IU/mL at baseline and year 3. In addition, ENH risk also became greater when patients had increased HBV DNA levels, with an HR of 2.6 (95% CI, ). Multivariate analysis revealed that increased levels of HBV DNA and ALT and persistently high levels of HBsAg and ALT remained independent risk factors for ENH development in different clinical settings (Table 4). Discussion For patients with chronic HBV infection, serum HBV DNA level 2,000 IU/mL has been widely used as an indicator for active viral replication, which is said to be the driving force of disease progression. 6,7 On the other hand, patients with HBV DNA <2000 Fig. 3. Relationship between baseline HBsAg level and relapse of HBV DNA (2,000 IU/mL) at year 3 of follow-up in different clinical settings.

8 448 TSENG ET AL. HEPATOLOGY, February 2013 Table 4. Risk for HBeAg-Negative Hepatitis in Different Clinical Settings. Patients with HBV DNA Level <2,000 IU/mL at Baseline (n ¼ 905) Patients with HBV DNA Level <2,000 IU/mL Plus ALT Level <40 U/L at Baseline (n ¼ 797) At Baseline At Year 3 Annual Incidence Rate* Crude HR (95% CI) Adjusted HR (95% CI) Annual Incidence Rate* Crude HR (95% CI) Adjusted HR (95% CI) Serum HBV DNA level, IU/mL <2,000 <2,000 1, <2,000 2,000 3, ( ) 2.2 ( ) 2, ( ) 2.2 ( ) Serum HBsAg level, IU/mL <1,000 <1,000 1, <1,000 1,000 1, ( ) 1.5 ( ) 1, ( ) 1.6 ( ) 1,000 <1,000 1, ( ) 0.8 ( ) ( ) 0.8 ( ) 1,000 1,000 1, ( ) 1.6 ( ) 1, ( ) 1.9 ( ) Serum ALT level, U/L <40 < < , ( ) 4.0 ( ) 4, ( ) 3.8 ( ) 40 <40 3, ( ) 3.0 ( ) , ( ) 9.5 ( ) *Per 100,000 person-years Variables include age, sex, and dynamics of HBV DNA, HBsAg, and ALT levels. P < P < IU/mL are regarded as low-risk patients. Our prior study showed that combining HBsAg level <1,000 IU/mL and HBV DNA <2,000 IU/mL can define patients to have even lower risk for HCC. 9 In this study, we consistently found that, in patients with HBV DNA <2,000 IU/mL, HBsAg level <1,000 IU/ ml was associated with a lower risk of ENH, hepatitis flare, and cirrhosis development. These lines of evidence suggest that HBsAg level <1,000 IU/mL can better define HBV carriers with extremely low risk for HBV-related adverse outcomes. ALT level is usually regarded as a result of the interactions among host immunity, hepatocytes, and viral replication and is always an important factor in predicting patient s prognosis. 6,9,11 In fact, ample evidence has shown that viral load is transiently suppressed when hepatitis activity occurs. 22 Because our primary criterion for enrollment was HBeAg-negative status with an HBV DNA level <2,000 IU/mL, we believe that some of the study participants, especially those with elevated ALT levels, were not really at an inactive state but had transient viral suppression. Therefore, it was not surprising to observe that this subgroup of patients had a higher risk of adverse outcomes. To validate the independent role of HBsAg level, we performed a subgroup analysis, which enrolled only patients with low viral loads and normal ALT levels at baseline. HBsAg level was shown again as an independent risk factor of ENH. Combining all the data derived from this cohort, the different incidence rates of ENH and HCC categorized by different biomarkers at enrollment are shown in Fig. 4A and 4B, respectively. The annual incidence rates of ENH and HCC for the normal ALT group were 1.4% (95% CI, 1.2% 1.7%) and 0.1% (95% CI, 0.06% 0.17%), respectively, and these rates are comparable to those of previous reports concerning inactive HBV carriers in Taiwan. 10,23 Of particular note, both incidence rates of ENH and HCC decreased to 1.1% (95% CI, 0.9% 1.4%) and 0.03% (95% CI, 0.01% 0.11%), respectively, when HBsAg level <1,000 IU/mL was added to the criteria in the normal ALT group. To be noted, irrespective of how strict the criteria are selected, minimal risk for disease progression still exists. Inactive carrier state has been adopted to describe HBsAg carriers who have persistently normal ALT levels plus HBV DNA levels that are persistently <2,000 IU/ ml. 8,24 Nevertheless, it is very difficult to define real inactive carriers, because in theory they need to remain in the inactive carrier state indefinitely, thus they cannot be identified using a single point data. In addition, even with very stringent criteria, there is no guarantee that inactive carriers are free from cirrhosis and HCC development in their lifetime. Therefore, we decided to use the term minimal-risk HBV carriers to describe HBeAg-negative patients with HBV DNA levels <2,000 IU/mL plus ALT levels <40 U/L and HBsAg levels <1,000 IU/mL at baseline. These patients have been documented to have a similar HCC risk to subjects negative for HBsAg and anti-hcv. 9,25 In addition, they have a significantly lower risk for HBV-related hepatitis and cirrhosis. Interestingly, the notion of using HBV DNA <2,000 IU/mL plus HBsAg <1,000 IU/ ml to predict a 3-year inactive carrier state has also been proposed in patients with genotype D infection, which is consistent with our results. 13 In other words, this is the first study to show that, in patients with

9 HEPATOLOGY, Vol. 57, No. 2, 2013 TSENG ET AL. 449 Fig. 4. Incidence rates of (A) HBeAg-negative hepatitis and (B) HCC in different clinical settings of 1,068 patients with HBV DNA level <2,000 IU/mL. The data are plotted and shown as incidence rates (95% CI). (C) Proposed algorithm to categorise risk levels of disease progression and corresponding management in Asian HBeAg-negative patients. genotype B or C infection, combining HBV DNA <2,000 IU/mL, normal ALT level, and HBsAg <1,000 IU/mL can define a minimal-risk HBV carrier. If the implications can be validated, it may provide a guide for optimal management of HBV carriers. For example, treatment-naïve patients with minimal risk for disease progression could receive less frequent follow-ups. In patients receiving antiviral treatment, this new marker may be integrated clinically to define a new treatment endpoint. However, more evidence is needed to confirm these findings and speculations. Based on these lines of evidence, 3,9,13,26 we proposed an algorithm to categorise risk levels of disease progression and corresponding management in HBeAg-negative patients with genotype B or C infection (Fig. 4C). Using single point levels of HBV DNA, ALT, and HBsAg, we could define patients as high, intermediate, low, and minimal risk for disease progression. In view of current guidelines on the management of chronic hepatitis B, 15,24,27 we recommended different follow-up intervals according to the risk levels. Patients are expected to receive better management of the disease once the usefulness of this algorithm is confirmed by prospective studies in the future. This study has a few limitations. First, checking a patient s HBV DNA level every 3 6 months is currently a recommended standard of care. 15,24,27 However, it should not be considered a clinical routine within the scope of this retrospective cohort study, in which the enrollment of patients dates back to In addition, real-time PCR was not available for HBV DNA measurement until 2004 in Taiwan; therefore, it is difficult to investigate the issue of maintaining long-term inactive carrier state in this retrospective cohort study. Second, liver cirrhosis was diagnosed based on clinical criteria, which are reliable for defining advanced cirrhosis but not early cirrhosis. Because patient enrollment dated back to 1985, and transient elastography was not available until 2003, using clinical criteria was the most feasible way to define cirrhosis for our study. In other words, we may have missed some cases of early cirrhosis, but the data are believed to be true for evaluating advanced

10 450 TSENG ET AL. HEPATOLOGY, February 2013 cirrhosis. Third, ENH was diagnosed mainly by ALT and HBV DNA elevation, which may be caused by other liver disease, such as nonalcoholic steatohepatitis. To overcome this limitation, we adopted three other endpoints: multiple ENH and HBeAg-negative hepatitis flare (which are more severe forms of hepatitis and are less likely caused by other etiologies) and cirrhosis (which results from repeated liver cell damage). Although the way in which we diagnosed these endpoints is imperfect, we believe that the consistent relationship between HBsAg level and these four endpoints in low-viremic patients indicates that a higher HBsAg level indeed correlates with higher risk for HBV-related adverse outcomes. However, we also noted that the increased risk of HBsAg level 1,000 IU/mL was modest when correlating with ENH. In other words, HBsAg level alone may not effectively determine which patients will and will not develop ENH. Therefore, combining more markers, such as different ALT levels, may achieve a better predictive value. The additive value of HBsAg level and predictive effect of combining other factors need to be validated in future prospective studies. In conclusion, the long-term prognosis of HBV carriers with HBV DNA level <2,000 IU/mL is still variable. In line with the relationship between HBsAg level and HCC risk, HBsAg level <1,000 IU/mL can serve as an indicator for lower risk of ENH, hepatitis flare, and cirrhosis. In clinical practice, combining HBsAg level <1,000 IU/mL with low or normal levels of HBV DNA and ALT at baseline may help define HBV carriers with minimal risk for HBV-related adverse outcomes. Acknowledgment: The authors thank Abbott Company for providing the quantitative HBsAg kits and Bristol-Myers Squibb Company for providing the unrestricted grant for viral load quantification. We also thank our colleagues at the National Taiwan University Hospital who enrolled and followed the patients, as well as the research assistants who assisted in laboratory analyses and collection of clinical information. References 1. Chen DS. From hepatitis to hepatoma: lessons from type B viral hepatitis. Science 1993;262: Kao JH. Hepatitis B virus genotypes and hepatocellular carcinoma in Taiwan. Intervirology 2003;46: Liaw YF, Chu CM. Hepatitis B virus infection. Lancet 2009;373: Kao JH, Chen DS. Global control of hepatitis B virus infection. Lancet Infect Dis 2002;2: Tseng TC, Liu CJ, Chen CL, Wang CC, Su TH, Kuo SF, et al. Serum hepatitis B virus-dna levels correlate with long-term adverse outcomes in spontaneous hepatitis B e antigen seroconverters. J Infect Dis 2012; 205: Chen CJ, Yang HI, Su J, Jen CL, You SL, Lu SN, et al. Risk of hepatocellular carcinoma across a biological gradient of serum hepatitis B virus DNA level. JAMA 2006;295: Iloeje UH, Yang HI, Su J, Jen CL, You SL, Chen CJ, et al. Predicting cirrhosis risk based on the level of circulating hepatitis B viral load. Gastroenterology 2006;130: Villa E, Fattovich G, Mauro A, Pasino M. Natural history of chronic HBV infection: special emphasis on the prognostic implications of the inactive carrier state versus chronic hepatitis. Dig Liver Dis 2011; 43(suppl 1):S8 S Tseng TC, Liu CJ, Yang HC, Su TH, Wang CC, Chen CL, et al. High levels of hepatitis B surface antigen increase risk of hepatocellular carcinoma in patients with low HBV load. Gastroenterology 2012;142: Hsu YS, Chien RN, Yeh CT, Sheen IS, Chiou HY, Chu CM, et al. Long-term outcome after spontaneous HBeAg seroconversion in patients with chronic hepatitis B. HEPATOLOGY 2002;35: Tai DI, Lin SM, Sheen IS, Chu CM, Lin DY, Liaw YF. Long-term outcome of hepatitis B e antigen-negative hepatitis B surface antigen carriers in relation to changes of alanine aminotransferase levels over time. HEPATOLOGY 2009;49: Tseng TC, Liu CJ, Su TH, Wang CC, Chen CL, Chen PJ, et al. Serum hepatitis B surface antigen levels predict surface antigen loss in hepatitis b e antigen seroconverters. Gastroenterology 2011;141: Brunetto MR, Oliveri F, Colombatto P, Moriconi F, Ciccorossi P, Coco B, et al. Hepatitis B surface antigen serum levels help to distinguish active from inactive hepatitis B virus genotype D carriers. Gastroenterology 2010;139: Chu CM, Liaw YF. Predictive factors for reactivation of hepatitis B following hepatitis B e antigen seroconversion in chronic hepatitis B. Gastroenterology 2007;133: Liaw YF, Kao JH, Piratvisuth T, Chan HL, Chien RN, Liu CJ, et al. Asian-Pacific consensus statement on the management of chronic hepatitis B: a 2012 update. Hepatol Int 2012;6: Hung CH, Lu SN, Wang JH, Lee CM, Chen TM, Tung HD, et al. Correlation between ultrasonographic and pathologic diagnoses of hepatitis B and C virus-related cirrhosis. J Gastroenterol 2003;38: Lin DY, Sheen IS, Chiu CT, Lin SM, Kuo YC, Liaw YF. Ultrasonographic changes of early liver cirrhosis in chronic hepatitis B: a longitudinal study. J Clin Ultrasound 1993;21: Bruix J, Sherman M. Management of hepatocellular carcinoma: an update. HEPATOLOGY 2011;53: SuTH,HsuCS,ChenCL,LiuCH,HuangYW,TsengTC,etal.Serum hepatitis B surface antigen concentration correlates with HBV DNA level in patients with chronic hepatitis B. Antivir Ther 2010;15: Tseng TC, Liu CJ, Yang HC, Su TH, Wang CC, Chen CL, et al. Determinants of spontaneous surface antigen loss in hepatitis B e antigen-negative patients with a low viral load. HEPATOLOGY 2012;55: Yeh SH, Tsai CY, Kao JH, Liu CJ, Kuo TJ, Lin MW, et al. Quantification and genotyping of hepatitis B virus in a single reaction by realtime PCR and melting curve analysis. J Hepatol 2004;41: Liu CJ, Chen PJ, Lai MY, Kao JH, Chang CF, Wu HL, et al. A prospective study characterizing full-length hepatitis B virus genomes during acute exacerbation. Gastroenterology 2003;124: Chu CM, Liaw YF. Spontaneous relapse of hepatitis in inactive HBsAg carriers. Hepatol Int 2007;1: Lok AS, McMahon BJ. Chronic hepatitis B: update HEPATOLOGY 2009;50: Chen JD, Yang HI, Iloeje UH, You SL, Lu SN, Wang LY, et al. Carriers of inactive hepatitis B virus are still at risk for hepatocellular carcinoma and liver-related death. Gastroenterology 2010;138: Chan HL, Wong VW, Wong GL, Tse CH, Chan HY, Sung JJ. A longitudinal study on the natural history of serum hepatitis B surface antigen changes in chronic hepatitis B. HEPATOLOGY 2010;52: European Association for the Study of the Liver. EASL clinical practice guidelines: management of chronic hepatitis B virus infection. J Hepatol 2012;57: