Chronic hepatitis B virus (HBV) infection can

Transcription

1 Distinct Evolution and Predictive Value of Hepatitis B Virus Precore and Basal Core Promoter Mutations in Interferon-Induced Hepatitis B e Antigen Seroconversion Hung-Chih Yang, 1,2,3 Chi-Ling Chen, 2 Yueh-Chi Shen, 1 Cheng-Yuan Peng, 4 Chun-Jen Liu, 2,3 Tai-Chung Tseng, 5 Tung-Hung Su, 2,3 Wan-Long Chuang, 6 Ming-Lung Yu, 6 Chia-Yen Dai, 6 Chen-Hua Liu, 2,3 Pei-Jer Chen, 2,3,7 Ding-Shinn Chen, 2,3,7 and Jia-Horng Kao 2,3,5,8 Precore (PC) (G1896A) and basal core promoter (BCP) (A1762T/G1764A) mutations of the hepatitis B virus (HBV) genome often emerge in chronic hepatitis B (CHB) patients. Their roles in hepatitis B e antigen (HBeAg) seroconversion induced by interferon (IFN) therapy remain controversial, partly because quantitative analysis for these mutants is lacking. This study aimed to develop a new assay to accurately quantify the PC and BCP mutant percentages and correlate their dynamic changes with IFN-induced HBeAg seroconversion in HBeAg-positive CHB patients. The PC and BCP mutant percentages were analyzed by polymerase chain reaction (PCR)-pyrosequencing. Our results showed that this quantitative assay for PC and BCP mutants achieved high accuracy (R 2 > 0.99) within a range between 10% and 90% mutants. We examined dynamic changes of the PC and BCP mutant percentages following IFN treatment in 203 HBeAg-positive CHB patients. By multiple logistic regression analysis, we found that the chance of HBeAg seroconversion increased by 2.2% (odds ratio [OR] , 95% confidence interval [CI]: , P ) and 2.3% (OR , 95% CI: , P ) per 1% increase of the pretreatment PC and BCP mutant percentages, respectively, after adjustment for other predictors. However, only the pretreatment PC mutation percentage was significantly associated with HBeAg seroconversion with HBV DNA < 2,000 IU/mL (OR , 95% CI: , P < 0.001). Furthermore, the mutant percentage of PC, but not BCP, in patients achieving HBeAg seroclearance with HBV DNA < 20,000 IU/mL increased significantly during IFN treatment (P ). Interestingly, patients with HBeAg seroconversion who had a high PC mutant percentage at the end of IFN treatment tended to exhibit high viremia after seroconversion. Conclusion: Quantitative analysis of PC and BCP mutants can predict IFN-induced HBeAg seroconversion and demonstrate their distinct evolution patterns during HBeAg seroconversion. (HEPATOLOGY 2013;57: ) Chronic hepatitis B virus (HBV) infection can lead to a broad spectrum of clinical outcomes, ranging from inactive carriers to severe hepatic complications, such as liver cirrhosis and hepatocellular carcinoma. 1 The natural course of individuals with chronic hepatitis B (CHB), particularly those acquiring HBV infection in the perinatal period or early childhood, consists of three distinctive phases of disease: Abbreviations: BCP, basal core promoter; CHB, chronic hepatitis B; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus; IFN, interferon; PC, precore. From the 1 Department of Microbiology, National Taiwan University College of Medicine, Taipei, Taiwan; 2 Graduate Institute of Clinical Medicine, National Taiwan University College of Medicine, Taipei, Taiwan; 3 Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan; 4 Department of Internal Medicine, China Medical University Hospital, Taichung, Taiwan; 5 Department of Internal Medicine, Buddhist Tzu Chi General Hospital Taipei Branch, Taipei, Taiwan; 6 Faculty of Internal Medicine, College of Medicine Kaohsiung Medical University, Kaohsiung, Taiwan; 7 Department of Medical Research, National Taiwan University Hospital, Taipei, Taiwan; 8 Hepatitis Research Center, National Taiwan University Hospital, Taipei, Taiwan. Received May 6, 2012; accepted October 14, Supported by grants from National Taiwan University Hospital (101-S1878), National Science Council (NSC B MY3), Liver Disease Prevention & Treatment Research Foundation, National Research Program for Biopharmaceuticals (NRPB), and the Taiwan-LiverNet Consortium (NSC B ). 934

2 HEPATOLOGY, Vol. 57, No. 3, 2013 YANG ET AL. 935 immune tolerant, immune clearance, and inactive residual phases. 2 In some patients with inactive disease, reactivation of hepatitis B can occur. Hepatitis B e antigen (HBeAg) is an important serological marker of the disease status of CHB. The seroconversion of HBeAg, defined by the loss of HBeAg and appearance of anti-hbe antibody, signifies the transition from the immune clearance to the inactive residual phase and is often accompanied by the reduction of HBV replication and remission of hepatocyte injury. 3 CHB patients who undergo HBeAg seroconversion usually exhibit a favorable prognosis. 4 Although HBeAg seroconversion has long been recognized as a critical event in the natural history of CHB, the underlying mechanisms remain largely elusive. Several virological factors have been demonstrated to impact the progression of CHB, including viral genotypes, serum HBV DNA levels, and certain naturally occurring mutants. 5 Because of the error-prone nature of the HBV reverse transcriptase, HBV mutants often emerge during the long-term HBV infection period. 6,7 Among these mutants, the precore stop codon (PC) and basal core promoter (BCP) mutants are best characterized. The PC mutant exhibits a G-to-A mutation in the precore region of HBV genome of nucleotide 1896 (codon 28; TGG to TAG), which abolishes the synthesis of HBeAg. 8 The double mutations of BCP often occur at nucleotide 1762 and 1764 with A to T and G to A substitution, respectively. Both PC and BCP mutations have been reported to be associated with HBeAg seroconversion. 9,10 These mutants are more prevalent in HBeAg-negative patients than in those with HBeAg positivity. Of note, most of the prior studies analyzed these mutants using qualitative assays, like direct sequencing. Few studies attempted to quantify the percentages of PC and/or BCP mutants in a population of HBV within a single individual Besides, longitudinal analysis of the dynamic change of PC and BCP mutants is scarce. Therefore, the exact roles of these mutants during HBeAg seroconversion remain unclear. Recently, Nie et al. 12 quantitatively and extensively analyzed the dynamic evolution of PC and BCP mutants in the longitudinal serum samples from 18 CHB patients. They found a temporal correlation between the increase of PC and/or BCP mutant frequency and HBeAg seroconversion, although the conclusion is limited by the small sample size. Interestingly, a previous study also revealed that PC mutation decreases the risk of HCC, whereas BCP mutation is associated with the development of HCC, 14 indicating the distinctive roles of these mutants in the long-term outcome of CHB patients. Interferon (IFN) has both an antiviral and immunomodulatory effect. IFN-based therapy with conventional or pegylated IFN has been shown to enhance HBeAg seroconversion in CHB patients. 15 Around 30%-40% of the HBeAg-positive CHB patients receiving a finite course of IFN therapy achieve HBeAg seroconversion, whereas HBeAg seroconversion occurs in less than 15% of untreated patients. 3,15 Several prior studies using qualitative assays investigated the roles of PC and BCP mutants in IFN-induced HBeAg seroconversion, but did not achieve consistent results. Therefore, the exact role of these two mutations in HBeAg seroconversion remains controversial. Pyrosequencing is a new sequencing technology based on the principle of sequencing-on-synthesis. It is a cheap, high throughput, and convenient method for analysis of the frequency of a particular nucleotide position among a heterogeneous population, and is suitable to quantify the viral mutants in a large number of samples. It has been shown to have the ability to accurately determine the frequency of influenza and HBV mutants in a mixture of wildtype and mutant viruses. 21,22 In this study, we developed a novel quantification assay using a pyrosequencing technique to measure the percentages of PC and BCP mutants in patients receiving IFN therapy and correlate them with the likelihood of HBeAg seroconversion. Patients and Methods Patients. A total of 203 HBeAg-positive CHB patients who had a pretreatment serum alanine aminotransferase (ALT) level over two times the upper limit of normal (ULN) and received a finite course of IFN therapy were retrospectively enrolled. Among them, 42 patients participated in a controlled trial comparing the efficacy of IFN-a-2b with and without ribavirin between 1998 and 1999 at the National Taiwan Address reprint requests to: Prof. Jia-Horng Kao, Distinguished Professor and Director, Graduate Institute of Clinical Medicine, National Taiwan University College of Medicine, 7 Chung-Shan South Road, Taipei 10002, Taiwan. kaojh@ntu.edu.tw; fax: 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: Dr. Yu consults for and received grants from Abbott. He is on the speakers bureau of and received grants from Roche. He also received grants from Bristol-Myers Squibb. Dr. Chen consults for, advises, is on the speakers bureau of, and received grants from Bristol-Myers Squibb. He consults for Roche and advises Abbott. Additional Supporting Information may be found in the online version of this article.

3 936 YANG ET AL. HEPATOLOGY, March 2013 University Hospital. 23 A detailed description of this study has been published. 23 Briefly, all patients received 5 million units (MU) IFN-a-2b daily for 4 weeks and then 5MU IFN-a-2b thrice a week for another 28 weeks with and without ribavirin. The remaining 161 patients received pegylated IFN-a-2a 180 lg weekly for 24 or 48 weeks. Among these 161 patients, 112 were consecutively enrolled in a study investigating host and viral factors in HBeAg-positive CHB patients receiving the treatment with pegylated IFN-a-2a between December 2005 and December The other 49 patients were consecutively enrolled at the China Medical University Hospital from Nov 2004 to May All serum samples were stored at 80 until used. All enrolled patients signed informed consent. This study was approved by the Ethical Committee of the National Taiwan University Hospital. In Taiwan, the Bureau of National Health Insurance launched a reimbursement program for treatment of CHB since October According to the Asian-Pacific consensus statement on the management of CHB, 24 the HBeAg-positive CHB patients with pretreatment serum ALT levels over two times the ULN were eligible for treatment with pegylated IFN-a-2a for 24 weeks. Therefore, most of our patients receiving pegylated IFN-a-2a were treated for 24 weeks, and only 30 patients received 48 weeks of pegylated IFNa-2a. Definition of Treatment Responses. HBeAg seroclearance was defined by the loss of HBeAg in an HBeAg-positive CHB patient. 15 The primary therapeutic endpoint was HBeAg seroclearance or seroconversion at 24 weeks posttherapy. The secondary therapeutic endpoint was HBeAg seroclearance or seroconversion with HBV DNA < 2,000 IU/mL and 20,000 IU/mL at 24 weeks posttherapy. HBV DNA Extraction. HBV viral DNA was extracted from 200 ll frozen serum samples using the QIAamp DNA Blood Mini kit (Qiagen, Germany) and eluted into 50 ll AE buffer according to the manufacturer s instructions. Quantification of the Percentage of PC and BCP Mutants by PCR Pyrosequencing. The detailed procedures are described in the Supporting Material. Briefly, the regions encoding the PC (nt1896) and BCP (nt1762 and nt1764) mutation sites were amplified by polymerase chain reaction (PCR) with three primers, a 5 0 -biotinylated universal M13 primers (5b-M13) and a pair of amplification primers, including the pairs for either PC mutation (F1825 and R1931-M13) or BCP mutation (F1676-M13 and R1819) (Supporting Table 1). The primers were mixed at a ratio of 10 (the primer without M13 tag) to 9 (5b-M13) to 1 (the primer with M13 tag) 25 to generate the M13-tagged biotinylated-pcr product for subsequent singlestranded DNA (ssdna) purification and pyrosequencing. For the pyrosequencing reaction, we first isolated biotin-labeled ssdna with the PyroMark Vacuum Prep Workstation (Qiagen, Germany) by following the manufacturer s protocol. The beads capturing ssdna were released into a PSQ HS 96 plate prefilled with the sequencing primer (F1875-seq for the PC mutation and R1783-seq for the BCP mutation) (Supporting Table 1) in the annealing buffer. Only the immobilized strand was used for pyrosequencing. Real-time pyrosequencing of the immobilized strand was performed with up to 96 samples in parallel at 28 C with the PyroMark Q96 ID Instrument (Qiagen, Germany) using 96 PyroMark Gold Q96 reagents, including the enzyme and substrate mixtures, following the manufacturer s instructions. All the samples were independently measured at least twice and the average of PC and BCP mutant percentages were calculated. Statistical Analysis. All factors collected are presented as the means and standard deviations (SDs) for continuous variables and frequency distribution for categorical variables. Unconditional logistic regression was conducted to estimate the odds ratios (OR) and 95% confidence intervals (CI) for the association between PC and BCP mutant percentages and the chance of HBeAg seroclearance/seroconversion. Among 91 subjects with more than one measurement of PC and BCP mutant percentages, the slopes of their changes over the five timepoints were estimated by the general linear regression model. Using these slopes for each individual as an independent variable, we conducted further unconditional logistic regression model to estimate the association between these slopes of change over time and HBeAg seroclearance/ seroconversion. Results Development and Validation of a Novel Pyrosequencing Assay for Quantification of PC and BCP Mutations. The DNA template was amplified by PCR with a pair of amplification primers, one of which contained M13-tagged sequence, and a 5 0 -biotinylated universal M13 primer. The resulting PCR product was subjected to ssdna purification and subsequent pyrosequencing. The detailed procedures are described in the Materials and Methods and the Supporting Material. The primer design and PCR amplification strategy

4 HEPATOLOGY, Vol. 57, No. 3, 2013 YANG ET AL. 937 Fig. 1. Illustration and validation of the quantitative PCR pyrosequencing assay for the PC and BCP mutations. (A) Schematic illustration of the threeprimer strategy for amplification of precore-encoding region and the resultant 5 0 -biotinylated and M13-tagged PCR product for the subsequent ssdna purification. (B,C) Validation of the quantitative PCR-pyrosequencing assay for (B) PC and (C) BCP mutations using the reference plasmids with a predefined mutant percentage from 0 to 100%. Each sample was measured independently in triplicate and the data are shown as mean 6 SE. are illustrated in Fig. 1A. To validate the accuracy of this assay, we generated four reference plasmids containing wildtype PC, mutant PC, wildtype BCP, and mutant BCP sequences, respectively, which serve as standards. We then mixed the wildtype and mutant PC and BCP plasmids with a range of predefined ratios, from 0% to 100% mutant in increments of 10%. Our data showed that the measured percentages of PC and BCP mutants were quite similar to the predefined ratios (Fig. 1B,C). The standard error of triplicate measurement of an individual sample was within 5% (mostly, within 3%). We concluded that this assay could achieve high accuracy with R 2 of 0.99 within a dynamic range between 10% and 90%. Therefore, we further took advantage of this new assay to analyze the percentages of PC and BCP mutants in the serial samples of 203 HBeAg-positive CHB patients who received IFN-based therapy. All the samples were measured twice using the same assay, and the results showed high reproducibility for both PC and BCP mutation percentages (the correlation coefficient r ¼ 0.99, data not shown). Baseline Characteristics of Patients. The demographic data of the 203 HBeAg-positive CHB patients are shown in Table 1. Among them, there were 42 patients receiving 32 weeks of standard IFN with (21 patients) or without ribavirin (21 patients) and the remaining 161 patients receiving pegylated IFN for 24

5 938 YANG ET AL. HEPATOLOGY, March 2013 Table 1. Baseline Characteristics of 203 HBeAg-Positive CHB Patients Receiving IFN-Based Therapy IFN (n ¼ 21) IFNþRBV (n ¼ 21) PegIFN (24wk) (n ¼ 131) PegIFN (48wk) (n ¼ 30) Total (n ¼ 203) Gender (M/F) 20/1 17/4 92/39 20/10 149/54 Age (yr) ALT (U/mL) Log-HBV-DNA (cps/ml) Genotype (B/C/unknown) 10/10/1 15/6/0 85/46/0 25/5/0 135/67/1 PC G1896A (%) BCP A1762T (%) BCP G1764A (%) HBeAg seroconversion (Yes/No) 5/16 5/16 33/98 12/18 55/148 HBeAg seroconversion with HBV 2/19 2/19 20/111 5/25 29/174 DNA < 2,000 IU/mL (Yes/No) HBeAg seroconversion with HBV 3/18 4/17 22/109 7/23 36/167 DNA < 20,000 IU/mL (Yes/No) HBeAg loss (Yes/No) 6/15 6/15 40/91 12/18 64/139 HBeAg loss with HBV DNA < 2,000 IU/mL (Yes/No) 3/18 3/18 23/108 5/25 34/169 HBeAg loss with HBV DNA < 20,000 IU/mL (Yes/No) 4/17 5/16 26/105 7/23 42/161 weeks (131 patients) or 48 weeks (30 patients). The previous study has reported that adding ribavirin does not increase the efficacy of IFN on HBeAg seroconversion. 23 The percentages of PC and BCP mutation represented the frequencies of these two mutants among the serum HBV population within a single individual. In these IFN-treated cohorts, there were in total 55 patients and 64 patients achieved HBeAg seroconversion and HBeAg seroclearance, respectively, at 24 weeks post-ifn therapy. Predictive Value of Pretreatment PC and BCP Mutant Percentages in HBeAg Seroconversion of CHB Patients Receiving IFN Therapy. To determine the role of PC and BCP mutant percentages in IFNinduced HBeAg seroconversion, we analyzed them using this new quantitative PCR pyrosequencing assay. We measured the PC and BCP mutant percentages at different timepoints from pretreatment to 24 weeks after treatment. The comparisons of the mutant percentages of PC, BCP (A1762T), and BCP (G1764A) at baseline (pretreatment) between the patients with and without IFN-induced HBeAg seroconversion were versus , versus , and versus , respectively (Table 2). Using univariate logistic regression analysis, we found that the PC G1896A mutation at baseline was significantly associated with HBeAg seroconversion (Table 2). The association between the BCP mutation G1764A at baseline and HBeAg seroconversion was only borderline significant (P ¼ 0.079). The chance of HBeAg seroconversion was increased by 1.7% (OR ¼ 1.017, 95% CI: , P ¼ 0.002) per 1% increase of PC mutant frequency at baseline. Using HBeAg seroclearance as the therapeutic endpoint, we found that, similarly, the chance of HBeAg seroclearance was significantly associated with both PC and BCP mutant frequencies (G1764A) at baseline (Table S2). However, when HBeAg seroconversion with low HBV DNA (<2000 IU/mL) was used as the therapeutic endpoint, only PC mutation at baseline (OR ¼ 1.026, 95% CI: , P < 0.001), but not BCP mutations, was significantly associated with the therapeutic response. Table 2. Univariate Analysis of the Association Between PC and BCP Mutant Percentages and HBeAg Seroconversion (left)/hbeag Seroconversion With HBV DNA < 2000 IU/mL (right) at 6 Months Off Therapy % of Mutant (mean 6 S.D) HBeAg Seroconversion HBeAg Seroconversion with HBV DNA < 2000 IU/mL % of Mutant (mean 6 S.D) Responder Nonresponder Odds Ratio (95% CI) P Responder Nonresponder Odds Ratio (95% CI) P Baseline (per 1% increase)* PC G1896A ( ) ( ) <0.001 BCP A1762T ( ) ( ) BCP G1764A ( ) ( ) *The odds ratio of each individual mutant at baseline represents the odds ratio of HBeAg seroconversion and HBeAg seroconversion with HBV DNA <2000 IU/ ml per 1% increment of the indicated mutant percentage at baseline.

6 HEPATOLOGY, Vol. 57, No. 3, 2013 YANG ET AL. 939 Table 3. Multivariate Analysis of Factors Associated With HBeAg Seroconversion (left) and HBeAg Seroconversion With HBV DNA < 2000 IU/mL at 6 Months Off Therapy HBeAg Seroconversion HBeAg Seroconversion with HBV DNA < 2000 IU/mL Characteristics OR (95% Cl) P-value OR (95% Cl) P-value Age ( per 1 year increase) ( ) ( ) Gender Male Female ( ) ( ) ALT (per 1 U/L increase) ( ) ( ) Treatment group IFN IFNþRBV ( ) ( ) PegIFN (24wk) ( ) ( ) PegIFN (48wk) ( ) ( ) Genotype B C ( ) ( ) HBV titer (per 1 log 10 HBV DNA increase) ( ) ( ) Baseline (per 1% increase)* PC G1896A mutant ( ) ( ) <0.001 BCP A1762T mutant ( ) ( ) *The odds ratio of each individual mutant at baseline represents the odds ratio of HBeAg seroconversion and HBeAg seroconversion with HBV DNA < 2000 IU/ ml per 1% increment of the indicated mutant percentage at baseline. We further took into consideration other baseline host and viral factors, including age, gender, ALT level, HBV viral load, and HBV genotype. Previous studies have shown that the dual mutations, A1762T and G1764A, of BCP mutants are highly associated, and our own data also showed the high correlation (r ¼ 0.991) between these two BCP mutations among all the samples we measured Supporting Fig. 1. We thus took either A1762T or G1764A mutation as an indicator of BCP mutation for the multivariate analysis. Actually, the results using either mutation for calculation came to the same conclusion (data not shown). Therefore, without specific mention, the percentage of A1762T was used for the following analysis regarding BCP mutation. Using multivariate logistic regression analysis we further showed that, at baseline, higher pretreatment PC and BCP mutant percentages were independently associated with higher rate of HBeAg seroconversion (Table 3). The chance of HBeAg seroconversion was increased by 2.2% (OR ¼ 1.022, 95% CI: , P ¼ 0.001) and 2.3% (OR ¼ 1.023, 95% CI: , P ¼ 0.001) per 1% increase of PC and BCP mutants, respectively. We also demonstrated the similar effects of PC and BCP mutant percentages on the treatment response when HBeAg seroclearance was adopted as a therapeutic endpoint (Table S3). However, using the HBeAg seroconversion with low HBV DNA (<2,000 IU/mL) as the therapeutic endpoint, we demonstrated that only the PC mutation percentage was significantly associated with the treatment response (OR ¼ 1.030, 95% CI: , P < 0.001) (Table 3). HBsAg loss is the ultimate therapeutic endpoint in patients receiving IFN treatment. However, in these cohorts we did not find any patients exhibiting HBsAg loss. To further investigate the relationship between the percentages of PC and BCP mutants and HBsAg loss, we decided to adopt alternative endpoints, such as HBsAg < 1,000 IU/mL at 24 weeks posttherapy and HBsAg decrease 1 or 2 logs during IFN treatment. However, we could not demonstrate the predictive role of PC and BCP mutation percentages in these therapeutic endpoints. Differential Evolution Patterns of PC and BCP Mutants During HBeAg Seroconversion. To investigate the dynamic changes of PC and BCP mutants during IFN treatment, we carried out an extensive analysis of the percentages of PC and BCP mutations in 91 patients (42 with IFN with/without ribavirin and 49 with 24 or 48 weeks of pegylated IFN) who had serial serum samples from the start of treatment to 24 weeks after the end of treatment. The evolution patterns of the PC and BCP mutants in patients grouped by HBeAg seroconversion with/without low HBV DNA are illustrated using boxplots (Fig. 2). We found that PC mutation, but not BCP mutations, increased in patients with HBeAg seroconversion. In addition, compared to the patients with HBeAg seroconversion and low HBV DNA (<2,000 IU/mL), those with HBeAg seroconversion and high HBV

7 940 YANG ET AL. HEPATOLOGY, March 2013 Fig. 2. The evolution patterns of PC and BCP mutants along the course of IFN therapy. The distribution of PC (A) and BCP (B) mutant percentages at different timepoints are illustrated using a boxplot. Patients were grouped based on occurrence of HBeAg seroconversion with/without low HBV DNA (<2,000 IU/mL). The borders of each box represents the 25 percentile (lower) and 75 percentile (upper) and the line within the box indicates the median mutant percentage. The top and bottom whiskers indicate the maximum and minimum of the data. DNA (>2,000 IU/mL) tended to have higher PC mutation frequency during treatment. We also used HBeAg seroclearance for patient grouping and found that the PC mutation, but not BCP mutation, accumulated in patients with HBeAg seroclearance along the course of IFN treatment Supporting Fig. 2. By using the slopes over time indicating the dynamic changes of PC and BCP mutations, we further investigated the relationship between the rate of change of PC and BCP mutations and HBeAg seroconversion/seroclearance with low HBV DNA. The results revealed that the slope of PC mutation, but not that of BCP mutation, was significantly associated with HBeAg seroclearance with HBV DNA < 20,000 IU/mL (OR ¼ 8.897, 95% CI: , P ¼ 0.039), and was also marginally associated with HBeAg seroclearance with HBV DNA < 2,000 IU/mL (OR ¼ 9.999, 95% CI: , P ¼ 0.088) (Table 4). However, when HBeAg seroconversion was adopted as the endpoint, the slope of PC mutation was only marginally associated with HBeAg seroconversion with HBV DNA < 20,000 IU/mL (OR ¼ 8.736, 95% CI: , P ¼ 0.097), but was not associated with HBeAg seroconversion with HBV DNA < 2,000 IU/mL (Table S4). This is probably due to fewer patients having HBeAg seroconversion with low HBV DNA. Taken together, these results indicated that the percentage of PC mutation, but not that of BCP mutation, did increase along the course of the IFN-based therapy in patients achieving HBeAg seroclearance. Association Between the PC Mutant Percentages at the End of Treatment (EOT) and the Viral Load After HBeAg Seroconversion. Although HBeAg seroconversion is often accompanied by a favorable prognosis, HBeAg-negative hepatitis sometimes occurs after HBeAg seroconversion. HBeAg-negative hepatitis is often associated with higher HBV DNA levels. Table 4. Multivariate Analysis of Factors Associated With HBeAg Loss at 6 Months Off Therapy Based on the Slope of the PC and BCP Mutant Percentages Over Time HBeAg Seroclearance with HBV DNA < 2000 IU/mL HBeAg Seroclearance with HBV DNA < IU/mL Characteristics OR (95% Cl) P-value OR (95% Cl) P-value Age (pr 1 year increase) ( ) ( ) Gender Male 1.00 Female ( ) ALT (U/L, c) ( ) ( ) Treatment group IFN IFNþRBV ( ) ( ) PegIFN (24wk) ( ) ( ) PegIFN (48wk) ( ) ( ) Genotype B C ( ) ( ) HBV titer (per 1 log 10 HBV DNA increase) ( ) ( ) PC G1896A mutant percentage slopes over time ( ) ( ) BCP A1762T mutant percentage slopes over time ( ) ( ) 0.648

8 HEPATOLOGY, Vol. 57, No. 3, 2013 YANG ET AL. 941 Fig. 3. The association between the PC mutant percentage at EOT and the probability of high viremia after HBeAg seroconversion. The patients with IFN-induced HBeAg seroconversion were divided into four groups according to the levels of PC mutant percentage at EOT. The numbers shown above the bar graph indicate the number of the patients with high viremia (HBV DNA level >200,000 IU/mL) over the total number of patients in each group with the indicated level of PC mutation. Therefore, we investigated whether the percentages of PC and BCP mutants affected the levels of viral replication after HBeAg seroclearance. Among the 55 patients who achieved HBeAg seroconversion, 28 patients had stored sera for comprehensive analysis of PC and BCP mutations at different timepoints. Interestingly, in these 28 patients with HBeAg seroconversion, a higher percentage of PC mutants at EOT was associated with a higher chance of having high HBV viral load after HBeAg seroconversion. Four out of six (66.7%) individuals with the highest PC mutants (>75% of PC mutation) at EOT exhibited high viremia (>200,000 IU/mL), whereas none of the 10 patients with the lowest PC mutants (<25% of PC mutation) had a high viral load (Fig. 3). In addition, the PC mutant percentages at EOT in patients with high viremia after HBeAg seroconversion was significantly higher than those in patients without high viremia (73.7% versus 35.8%, P ¼ 0.002). Analyzing the 35 patients who had HBeAg seroclearance and available serum samples at EOT for analysis, we also found that the patients with a higher percentage of PC mutation at EOT were more likely to have high viremia after HBeAg seroclearance Supporting Fig. 2. Discussion In this study we comprehensively analyzed the percentages of PC and BCP mutations at baseline and during and after IFN treatment in a large number of HBeAg-positive CHB patients with HBV genotype B or C infection. Our results clearly demonstrated the value of both PC and BCP mutation percentages at baseline in predicting HBeAg seroconversion. Although previous studies suggested the association of baseline BCP mutation with IFN-induced HBeAg seroconversion, 16,17,20 the role of pretreatment PC mutation in HBeAg seroconversion is somewhat controversial. Lok et al. 18 showed that HBeAg-positive CHB patients with PC mutation is more likely to clear HBeAg, but other studies failed to demonstrated this association. 16,17,19,20 These discrepant findings probably resulted from the use of less sensitive and informative qualitative assays, although the influence of the different ethnic and geographic origins in these studies cannot be excluded. By quantitative analysis of PC and BCP mutations, we found that the percentages of both PC and BCP mutants at baseline could predict IFN-induced HBeAg seroconversion, suggesting the importance and utility of the quantitative analysis of HBV mutants in evaluating the prognosis of CHB. Interestingly, although both PC and BCP mutant percentages at baseline were associated with HBeAg seroconversion, only the percentage of PC mutation, but not that of BCP mutation, was associated with HBeAg seroconversion plus low HBV DNA. Detailed analysis of the viral evolution from baseline to 24 weeks posttherapy confirmed the distinct evolution patterns of PC and BCP mutations during IFNinduced HBeAg seroconversion. Chu et al. 11 consistently showed that the PC mutant ratio increased during spontaneous HBeAg seroconversion. Recently, Nie et al. 12 also quantitatively analyzed the dynamic changes of PC and BCP mutants in 13 CHB patients who had spontaneous HBeAg seroconversion. They found that PC and/or BCP mutant ratios increased during HBeAg seroconversion. However, different from Nie et al. s study, our results showed that the PC mutant percentage increased, but the BCP mutant percentage was relatively unchanged during HBeAg seroconversion. The reasons for the different observations between Nie et al. s study and ours remain unclear, although they might be due to the difference between spontaneous and IFN-induced HBeAg seroconversion, or due to sampling bias between studies. The distinct evolution patterns of PC and BCP mutations during HBeAg seroconversion might suggest different biological mechanisms involved in selection of PC and BCP mutants. Consistently, a prior longterm longitudinal follow-up study on a large CHB cohort revealed that BCP mutation enhances the development of HCC, but PC mutation is associated with the decreased risk of HCC, 14 indicating the distinct biological effects of PC and BCP mutations on the disease status of CHB. It has been suggested that

9 942 YANG ET AL. HEPATOLOGY, March 2013 cytosolic HBeAg is an immunogen targeted by the host immune response for elimination of infected hepatocytes during the immune clearance phase. 26 Along the course of HBeAg seroclearance, the host immune pressure tends to select HBeAg-negative viral strains, like PC mutants, probably through immune-mediated hepatocytolysis. This might explain the accumulation of PC mutation during HBeAg seroconversion. However, the percentage of BCP mutation at baseline was also associated with HBeAg seroconversion, although it was not associated with HBeAg seroconversion plus low HBV DNA level and did not significantly increase during HBeAg seroconversion. This fact implies that the percentage of BCP mutant probably reflects the strength of host immune response against HBV, but this immune pressure is not the direct driving force to cause HBeAg seroconversion. Therefore, the accumulation of BCP mutants may not be observed during a short time frame in which HBeAg seroconversion occurs, but can be seen in a long period of follow-up, like in Nie et al. s study. 12 It would be interesting to examine this phenomenon in patients undergoing spontaneous HBeAg seroconversion and correlate the quantities of PC or BCP mutants with their long-term clinical outcomes. Interestingly, in contrast to the above observation that a higher pretreatment PC mutant percentage and its increase during IFN treatment had a favorable outcome, HBeAg seroconversion, in the HBeAg-positive patients receiving IFN therapy, we found a high percentage of PC mutant at the end of IFN therapy seemed to be associated with high viremia after HBeAg seroconversion. Consistently, several previous studies demonstrated that PC mutant prevailed in patients with HBeAg-negative hepatitis. 27,28 In addition, by quantitative analysis of the PC mutant ratios, Chu et al. 11 reported that all patients with high viremia and ALT elevation after HBeAg seroconversion exclusively exhibited the PC mutant. Although the underlying mechanisms are not known, we speculated that a very high percentage of PC mutation at EOT probably implies the efficient escape of viral mutants from the immune control, and the failure of the host immune response to contain viral replication. Alternatively, a high percentage of PC mutation may be associated with some other HBV mutations that cause high viremia. Because only 28 patients with HBeAg seroconversion had available sera at EOT for analysis in this study, a larger cohort with longitudinal and comprehensive analysis of the entire HBV genome may be required to solve this issue. One drawback of this study was the heterogeneous treatment populations, which might compromise our conclusion to some extent. Because the reimbursement policy of the Bureau of National Insurance in Taiwan, most of the patients receiving pegylated IFN-a-2a in this cohort were treated for 24 weeks, not for 48 weeks. The latter is the current preferred regimen demonstrated by the NEPTUNE study. 29 Therefore, further studies are required to validate our observation in the HBeAg-positive CHB patients receiving 48 weeks of IFN-a-2a. In summary, by detailed quantitative analysis of longitudinal samples from a large number of HBeAg-positive CHB patients receiving IFN therapy, we have demonstrated different predictive values of PC and BCP mutants in HBeAg seroconversion. Particularly, our data revealed distinct and previously unnoticed evolution patterns of PC and BCP mutants during IFN-induced HBeAg seroconversion. Understanding the underlying mechanisms that regulate the dynamic evolution of PC and BCP mutants can help predict the prognosis of CHB patients as well as design better strategies of follow-up and antiviral therapy for them. Acknowledgment: We thank Drs. Hwei-Fang Tien, Wen-Chien Chou as well as Ming-Cheng Lee for help in pyrosequencing. References 1. Dienstag JL. Hepatitis B virus infection. N Engl J Med 2008;359: Liaw YF, Chu CM. Hepatitis B virus infection. Lancet 2009;373: Liaw YF, Lau GK, Kao JH, Gane E. Hepatitis B e antigen seroconversion: a critical event in chronic hepatitis B virus infection. Dig Dis Sci 2010;55: Yang HI, Lu SN, Liaw YF, You SL, Sun CA, Wang LY, et al. Hepatitis B e antigen and the risk of hepatocellular carcinoma. N Engl J Med 2002;347: Kao JH. Role of viral factors in the natural course and therapy of chronic hepatitis B. Hepatol Int 2007;1: Chotiyaputta W, Lok AS. Hepatitis B virus variants. Nat Rev Gastroenterol Hepatol 2009;6: Gunther S, Fischer L, Pult I, Sterneck M, Will H. Naturally occurring variants of hepatitis B virus. Adv Virus Res 1999;52: Carman WF, Jacyna MR, Hadziyannis S, Karayiannis P, McGarvey MJ, Makris A, et al. Mutation preventing formation of hepatitis B e antigen in patients with chronic hepatitis B infection. Lancet 1989;2: Chu CJ, Keeffe EB, Han SH, Perrillo RP, Min AD, Soldevila-Pico C, et al. Prevalence of HBV precore/core promoter variants in the United States. HEPATOLOGY 2003;38: Yuen MF, Sablon E, Yuan HJ, Hui CK, Wong DK, Doutreloigne J, et al. Relationship between the development of precore and core promoter mutations and hepatitis B e antigen seroconversion in patients with chronic hepatitis B virus. J Infect Dis 2002;186: Chu CM, Yeh CT, Lee CS, Sheen IS, Liaw YF. Precore stop mutant in HBeAg-positive patients with chronic hepatitis B: clinical characteristics and correlation with the course of HBeAg-to-anti-HBe seroconversion. J Clin Microbiol 2002;40: Nie H, Evans AA, London WT, Block TM, Ren XD. Quantitative dynamics of hepatitis B basal core promoter and precore mutants before and after HBeAg seroconversion. J Hepatol 2012;56:

10 HEPATOLOGY, Vol. 57, No. 3, 2013 YANG ET AL Pang A, Yuen MF, Yuan HJ, Lai CL, Kwong YL. Real-time quantification of hepatitis B virus core-promoter and pre-core mutants during hepatitis E antigen seroconversion. J Hepatol 2004;40: Yang HI, Yeh SH, Chen PJ, Iloeje UH, Jen CL, Su J, et al. Associations between hepatitis B virus genotype and mutants and the risk of hepatocellular carcinoma. J Natl Cancer Inst 2008;100: Lok AS, McMahon BJ. Chronic hepatitis B: update HEPATOLOGY 2009;50: Erhardt A, Reineke U, Blondin D, Gerlich WH, Adams O, Heintges T, et al. Mutations of the core promoter and response to interferon treatment in chronic replicative hepatitis B. HEPATOLOGY 2000;31: Fattovich G, McIntyre G, Thursz M, Colman K, Giuliano G, Alberti A, et al. Hepatitis B virus precore/core variation and interferon therapy. HEPATOLOGY 1995;22: Lok AS, Akarca US, Greene S. Predictive value of precore hepatitis B virus mutations in spontaneous and interferon-induced hepatitis B e antigen clearance. HEPATOLOGY 1995;21: Seo Y, Yoon S, Hamano K, Nakaji M, Yano Y, Katayama M, et al. Response to interferon-alpha in chronic hepatitis B with and without precore mutant strain detected by mutation site-specific assay. J Clin Gastroenterol 2004;38: Tseng TC, Yu ML, Liu CJ, Lin CL, Huang YW, Hsu CS, et al. Effect of host and viral factors on hepatitis B e antigen-positive chronic hepatitis B patients receiving pegylated interferon-alpha-2a therapy. Antivir Ther 2011;16: Duwe S, Schweiger B. A new and rapid genotypic assay for the detection of neuraminidase inhibitor resistant influenza A viruses of subtype H1N1, H3N2, and H5N1. J Virol Methods 2008;153: Lindstrom A, Odeberg J, Albert J. Pyrosequencing for detection of lamivudine-resistant hepatitis B virus. J Clin Microbiol 2004;42: Liu CJ, Lai MY, Chao YC, Liao LY, Yang SS, Hsiao TJ, et al. Interferon alpha-2b with and without ribavirin in the treatment of hepatitis B e antigen-positive chronic hepatitis B: a randomized study. HEPATOLOGY 2006;43: Liaw YF, Leung N, Kao JH, Piratvisuth T, Gane E, Han KH, et al. Asian-Pacific consensus statement on the management of chronic hepatitis B: a 2008 update. Hepatol Int 2008;2: Royo JL, Hidalgo M, Ruiz A. Pyrosequencing protocol using a universal biotinylated primer for mutation detection and SNP genotyping. Nat Protoc 2007;2: Milich D, Liang TJ. Exploring the biological basis of hepatitis B e antigen in hepatitis B virus infection. HEPATOLOGY 2003;38: Okamoto H, Yotsumoto S, Akahane Y, Yamanaka T, Miyazaki Y, Sugai Y, et al. Hepatitis B viruses with precore region defects prevail in persistently infected hosts along with seroconversion to the antibody against e antigen. J Virol 1990;64: Tur-Kaspa R, Klein A, Aharonson S. Hepatitis B virus precore mutants are identical in carriers from various ethnic origins and are associated with a range of liver disease severity. HEPATOLOGY 1992;16: Liaw YF, Jia JD, Chan HLY, Han KH, Tanwandee T, Chuang WL, et al. Shorter durations and lower doses of peginterferon alfa-2a are associated with inferior hepatitis B e antigen seroconversion rates in hepatitis B virus genotypes B or C. HEPATOLOGY 2011;54: