Baseline interpatient hepatitis B viral diversity differentiates HBsAg outcomes in patients treated with tenofovir disoproxil fumarate

Accepted Manuscript Baseline Interpatient Hepatitis B Viral Diversity Differentiates HBsAg Outcomes in Patients Treated With Tenofovir Disoproxil Fumarate Prista Charuworn, Paul N. Hengen, Raul Aguilar Schall, Phillip Dinh, Dongliang Ge, Amoreena Corsa, Hendrik W. Reesink, Fabien Zoulim, Kathryn M. Kitrinos PII: DOI: Reference:

S0168-8278(14)00926-X http://dx.doi.org/10.1016/j.jhep.2014.12.008 JHEPAT 5472

To appear in:

Journal of Hepatology

Received Date: Revised Date: Accepted Date:

9 April 2014 2 December 2014 2 December 2014

Please cite this article as: Charuworn, P., Hengen, P.N., Schall, R.A., Dinh, P., Ge, D., Corsa, A., Reesink, H.W., Zoulim, F., Kitrinos, K.M., Baseline Interpatient Hepatitis B Viral Diversity Differentiates HBsAg Outcomes in Patients Treated With Tenofovir Disoproxil Fumarate, Journal of Hepatology (2014), doi: http://dx.doi.org/10.1016/ j.jhep.2014.12.008

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Title 119/130

Baseline Interpatient Hepatitis B Viral Diversity Differentiates HBsAg Outcomes in Patients Treated With Tenofovir Disoproxil Fumarate

Prista Charuworn 1, Paul N. Hengen1, Raul Aguilar Schall 1, Phillip Dinh 1, Dongliang Ge 1, Amoreena Corsa 1, Hendrik W. Reesink2, Fabien Zoulim3, and Kathryn M. Kitrinos 1 1

Gilead Sciences, Inc, Foster City, CA, USA

2

Department of Gastroenterology and Hepatology, Academic Medical Center, Amsterdam, The Netherlands 3

INSERM U1052, Hospices Civils de Lyon, Lyon University, Lyon, France

Keywords: HBsAg loss; genetic distance; nucleotide treatment outcome; HBV genotype differences; HBV coding regions

Address for correspondence: Prista Charuworn 333 Lakeside Dr. Foster City, CA 94404 Phone: (650)372-7961 Fax: 650-522-5854 [email protected]

1

Abbreviations: CHB, chronic hepatitis B; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; NA, nucleos(t)ide analogues; TDF, tenofovir disoproxil fumarate; HBV, hepatitis B virus; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CP, core promoter; PC, precore; BCP, basal core promoter; URR, upper regulatory region; NRE, negative regulatory element; CURS, core upstream regulatory sequence; Enh2, enhancer II; ORF, open reading frame.

Financial Support: Funding for this study was provided by Gilead Sciences, Inc.

Financial Disclosures: Prista Charuworn, Paul N. Hengen, Raul Aguilar Schall, Phillip

Dinh, Dongliang Ge, Amoreena Corsa, and Kathryn M. Kitrinos are employees of Gilead Sciences, Inc., and may own stocks and shares in the company. Hendrik W. Reesink has served as a consultant for Gilead Sciences, Inc., GlaxoSmithKline, Janssen-Cilag, Merck, PRA-International, Roche, R-Pharm, Abbvie, Santaris, and Regulus; and has received research funding from Bristol-Myers Squibb, Boehringer Ingelheim, Chugai, Gilead, Idenix, Janssen-Cilag, Japan Tobacco, Phenomix, PRA-International, Roche, Santaris, and Vertex. Fabien Zoulim has served as a consultant for Gilead Sciences, JanssenCilag, Roche, Transgene, Novira; and has received research funding from Roche, Gilead, Scynexis, Novira, Novartis.

Author’s contributions: PC and KMK were involved in the concept, design, analysis, and writing of the manuscript; PNH, RAS, PD and DG were involved in the design, analysis and critical revision of the manuscript; AC was involved in the input data analysis and critical revision of the manuscript; FZ and HWR were involved in the

2

analysis and critical revision of the manuscript. All authors approved the submitted manuscript.

Word Count: 4687/5000 (including abstract, references, table, and figure legends) Figure and table count: 1 table, 4 figures Supplementary figure and table count: 3 tables, 1 figure

3

Abstract (245/250) Background and Aims: HBsAg loss is a desired, but rare, treatment-induced clinical endpoint in chronic hepatitis B (CHB). Few studies have evaluated viral factors contributing to HBsAg loss. Methods: This study evaluated baseline interpatient sequence diversity across the HBV genome in tenofovir disoproxil fumarate-treated patients who lost HBsAg and compared it to that of control patients with high HBsAg levels throughout therapy. Twenty-one HBeAg+ patients (14 genotype[GT] A and 7 GT D) who achieved HBsAg loss and 27 controls (17 GT A and 10 GT D), were analyzed. Population sequencing was performed on baseline samples and pairwise genetic distances were calculated for 17 overlapping regions across the HBV genome as a measure of interpatient viral diversity. Results: Overall, viral diversity was up to 10-fold higher across GT D patients compared to GT A patients throughout the HBV genome. Within the pol/RT and HBs genes, interpatient viral diversity was significantly lower among HBsAg loss patients for both GT A and D, with the difference driven largely by a reduction in diversity in the small S gene. Conversely, interpatient viral diversity was generally higher in HBsAg loss patients across the HBx gene regulatory elements and pre-core region. Conclusion: In HBsAg loss patients, less interpatient viral diversity was observed within structural-coding regions while specific regions across the HBx and pre-core genes encoding nonstructural regulatory elements generally displayed higher interpatient viral diversity. These distinct patterns may reflect different responses to adaptive pressure for HBV genomic structural and nonstructural elements.

4

Introduction Chronic hepatitis B (CHB) is an infectious disease affecting more than 240 million people worldwide (1). Current treatments, using either nucleos(t)ide analogs (NA) for direct inhibition of the viral polymerase or pegylated interferon (PEG) for immune-mediated viral control, have been effective in achieving HBV DNA suppression and HBeAg seroconversion. However, these endpoints are usually not durable once treatment is discontinued. A more desirable treatment endpoint, as highlighted in international CHB treatment guidelines, is HBsAg loss (2-4). Patients who achieve HBsAg loss have improved clinical outcomes, including decreased risk of hepatocellular carcinoma, progression to cirrhosis, and death (5). The rate of spontaneous HBsAg loss is low (~1-3% of patients per year), and while treatment-induced HBsAg loss rates are higher, they remain low with 0-7% of patients becoming HBsAg seronegative after 48 weeks of NA or PEG treatment (3). Several studies have assessed clinical and patient factors associated with treatmentinduced HBsAg loss, mostly focusing on HBeAg+ patient populations (6- 8). While the rate of HBsAg loss varies depending on treatment, similar clinical and patient factors have been observed to predict HBsAg loss, such as early HBsAg decline (> 1 log10 decline by week 12), HBeAg loss, Caucasian race, and genotype (GT) (6-9). HBsAg loss is associated with GT A and GT D for patients treated with NAs (3,7,9). While these studies have identified baseline and ontreatment parameters associated with HBsAg loss, further understanding of the baseline parameters could lead to the identification of patients with a higher chance of clearing HBsAg on treatment. To date, little is known about the predictive nature of any viral factor(s) for loss of HBsAg. The objective of this study was to determine, of the patients who were enrolled in the tenofovir disoproxil fumarate (TDF) phase 3 pivotal clinical trial (GS-US-174-0103), if pretreatment interpatient viral diversity within the major transcribed regions of the HBV genome

5

differs across GT A and D patients who achieve HBsAg loss compared to GT A and D patients that maintain high levels of HBsAg during treatment with TDF for up to 192 weeks. Additionally, we sought to compare the interpatient viral diversity across the HBV genome between GT A and D. Methods Patient population Patients included in these analyses were enrolled in study GS-US-174-0103. The protocol, study design, and patient enrollment criteria have been previously reported (6). Briefly, HBeAg+ CHB patients were randomized 2:1 to receive either TDF or adefovir dipivoxil (ADV) for 48 weeks followed by open-label TDF. After up to 192 weeks of treatment, 23 of 266 patients experienced HBsAg loss, including 14 harboring GT A HBV and 7 harboring GT D HBV. Seventeen GT A and 10 GT D patients who maintained high levels of HBsAg (defined as >10,000 IU/mL for > 70% of visits over 192 weeks of treatment) were selected as controls. Control patients were matched to HBsAg loss patients by genotype and HBeAg status. Controls were selected to be similar to HBsAg loss subjects on HBV DNA (>1x 106 copies/ml) and ALT (>1.5 ULN). Overall, 8/27 control subjects (1/10 GT D and 7/17 GT A) and 21/21 HBsAg loss subjects experienced persistent HBeAg loss over 192 weeks of treatment. Laboratory assays used in this study have been previous published (10). All patients enrolled in GS-US-174-0103 provided written informed consent, and the study protocol was approved by the institutional review boards of the participating institutions. Direct Sequencing Population sequencing of the HBV surface (HBs), polymerase/ reverse transcriptase (pol/RT), core, and HBx genes was performed on baseline (pre-treatment) samples. As HBV exists within an individual as quasispecies, population sequencing captured the dominant viral variants contributing to the intrapatient viral diversity pool. This method has been used previously to assess the role of interpatient genetic variation on treatment outcome (11). HBV DNA from

6

plasma or serum samples was extracted, PCR amplified, and sequenced. (DDL, Rijswijk, The Netherlands). HBV regions were selected for viral diversity analysis based on published reports describing the role of the specific region as an important immune epitope, a regulatory element or an open reading frame (Figure 1a,b,c) (12-14). The URR, NRE, CURS, CP, BCP Enh2, and S promoter are regulatory regions that impact transcriptional control of HBV genes. The PC region encodes for a nonstructural signaling element of the core protein that is important in cellular trafficking and is cleaved prior to protein extracellular release. The HBx gene, which contains or overlaps with the URR, NRE, CURS, Enh2, CP, BCP, and PC, is involved in initiating cccDNA transcription, though its other functions are not well understood (15). The core, pre-S1, preS2, and small S (including the a-determinant) encode the structural components of the HBV virion. Lastly, the pol/RT region, including the RT tail, encodes for the RNA polymerase that is needed for viral replication. DNA Sequence Analyses Alignment of DNA sequences from each region was done using MUltiple Sequence Comparison by Log-Expectation (MUSCLE) (16). Distance matrices were created for the pairwise comparison of sequences using the program PhyML (17), estimating the phylogenetic distance by the Maximum Likelihood method under a general time-reversible model of nucleotide evolution. Viral diversity is defined as the mean pairwise genetic distance, an estimate of the genetic divergence between populations within a viral species (18). The variance is a measure of the deviation of individual pairwise genetic distances with respect to the mean within each group. For analysis of the small HBs gene, a sliding window of 19 bases was used to span the length of the gene sequence using a step size of 10 bases. Statistical Analyses The Mann-Whitney-Wilcoxon test was used to evaluate equality of means between the HBsAg loss and the control groups within genotypes, and the means across genotypes within groups. The non-parametric Levene’s test was implemented to assess homogeneity of variances

7

(19). We used a Hochberg procedure to control the family-wise type I error rate because of the multiple tests performed for the different regions. The unadjusted significance level was the nominal alpha=0.05. The analyses were generated using SAS software, Version 9.2 of the SAS System for Windows. Copyright © 2002-2008 by SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA. Results Baseline Characteristics In Study GS-US-174-0103, 23 HBeAg+ patients achieved HBsAg loss on TDF treatment through week 192 (9). Of these, 21/23 patients (91.3%) harbored GT A (n=14) or GT D (n=7) HBV and were selected to assess interpatient viral diversity. Of these 21 patients, 16 (76%) achieved HBsAb seroconversion (Supplelemental Table 3). Twenty-seven patients (17 GT A, 10 GT D) who maintained high levels of HBsAg (>10,000 IU/mL for >70% of visits across 192 weeks of treatment) were selected as controls (Supplemental Figure 1 and Supplemental Table 3). Within each genotype, the HBsAg loss and control patients were not statistically different in gender, past interferon experience, previous exposure to HBV therapy (including interferon), ALT >1.5 ULN, age >40 years, median ALT, mean HBV DNA, or mean Knodell necroinflammatory score (p >0.05) (Table 1). As previously reported, patients who achieved HBsAg loss had significantly higher mean quantitative HBsAg levels at baseline compared to control patients (p = 0.02 for GT A, p = 0.01 for GT D) (9). HBsAg loss patients have less interpatient viral diversity within the pol/RT and HBs genes Eight regions across the HBs and pol/RT genes were assessed for interpatient viral diversity using mean pairwise genetic distance (Figure 1a). For the majority of regions analyzed, GT A and D HBsAg loss patients had significantly lower mean pairwise genetic distance compared to control patients (Figure 2). For GT D patients, significantly lower mean pairwise genetic distance between HBsAg loss patients compared to control patients was observed in all

8

regions except PreS1/S2.Futhermore, the a-determinant, which is the coding region for the primary neutralizing antibody epitope of the small HBs gene, exhibited the largest difference (3fold) in mean pairwise genetic distance between the two groups. For GT A patients, significantly lower viral diversity was observed between HBsAg loss patients compared to controls for most regions, except PreS1/S2, the S promoter, and the a-determinant. The lack of difference observed in the a-determinant for GT A patients is a reflection of the highly conserved nature of this region within GT A; 28/31 GT A patients had identical sequences across the a-determinant. For both GT A and D, the largest mean pairwise genetic distance was observed in the RT tail, a nonoverlapping region in pol/RT, which is consistent with previous reports that mutations occur more frequently in regions with less constraint (20). There were no differences in interpatient viral diversity in the PreS1/S2 region between the controls and HBsAg loss patients in both GT A and D. Therefore, the main driver for the observed difference in interpatient pairwise mean genetic distance of HBs is the small HBs and its overlapping pol/RT gene. Overall, across the HBs and pol/RT genes, the mean pairwise genetic distance between GT A patients was approximately 10fold less than what was observed between GT D patients, regardless of HBsAg outcome. Sliding window analysis of small HBs gene reveals multiple areas account for interpatient viral diversity differences in HBsAg loss patients compared to controls. As the mean interpatient pairwise genetic distance results for GT A and D demonstrated that the small HBs gene contributed to the differences between HBsAg loss and control groups, further analysis was conducted to determine if the mean interpatient pairwise genetic distance difference was driven by the entire small HBs gene or specific sub-regions (Figure 3a,3b). A sliding window analysis was conducted using overlapping 19-base pair nucleotide length segments migrating every 10 base pairs spanning the small HBs gene. Several regions that were individually highly significant (p-unadjusted <0.005) failed to maintain significance after adjusting for 67 simultaneous hypothesis testings. For GT D patients, multiple areas across the small HBs gene had lower mean interpatient pairwise genetic distance in HBsAg loss patients

9

compared to controls. Interestingly, a significant difference observed between groups was a segment within the main immunodominant loop that includes the a-determinant. Higher mean interpatient pairwise genetic distance was observed for HBsAg loss patients compared to controls in 4 regions, including nucleotides 331-349, a segment within the immunodominant loop located between transmembrane 2 and the a-determinant (Figure 3a). Transmembrane region 2, which is involved in directing small HBs protein orientation and folding, is highly conserved across all GT A and D patients. Similar to what was observed with the mean interpatient pairwise genetic distance analysis, the sliding window analysis of the small HBs gene for GT A patients revealed less viral diversity compared to GT D, both in magnitude and location between HBsAg loss patients and controls. Little or no viral diversity was observed across the a-determinant for both groups. Within small HBs, GT A HBsAg loss patients had higher levels of diversity compared to controls in seven 19-bp segments; whereas there were 14 segments where controls showed higher viral diversity compared to HBsAg loss patients (Figure 3b). Core, precore, basal core promoter and core promoter regions show genotype-specific differences Nine overlapping regions across the HBx and core genes were assessed for interpatient viral diversity using mean pairwise genetic distance (Figure 1b). For GT A patients, 6/9 regions had significant differences between HBsAg loss and control patients (Figure 4). Of these 6 regions, only the core gene had lower mean interpatient pairwise genetic distance in the HBsAg loss patients compared to controls. For the remaining 5 regions (PC, URR, NRE, CURS, and EnhII) the HBsAg loss patients had significantly higher mean interpatient pairwise genetic distance compared to controls, with the pre-core region having the largest difference (22-fold increase in mean pairwise genetic distance in HBsAg loss patients compared to controls). The full length HBx gene did not show statistically significant differences in the mean interpatient genetic distance between the HBsAg loss and control groups, suggesting that sequence

10

differences within HBx reside at the 3’ end of the gene, a region that is largely composed of regulatory elements. In contrast, for GT D patients, 4/9 regions (core, BCP, NRE, and Enh2) had significant differences between HBsAg loss and control patients. Similar to GT A, the core gene had lower mean interpatient pairwise genetic distance in the HBsAg loss patients compared to controls. Likewise, the NRE exhibited a pattern similar to GT A whereas HBsAg loss patients displayed higher levels of interpatient viral diversity compared to controls. Comparable to what was observed within HBs and pol/RT regions, GT A patients had lower mean interpatient pairwise genetic distance in the majority of HBx and core gene regions evaluated compared with GT D patients within the HBsAg loss and control groups. Discussion This study evaluated the association between baseline interpatient viral diversity across the HBV genome with on-treatment HBsAg loss in HBeAg + patients treated with TDF in Study GS-US-174-0103. When comparing HBV genetic regions that encode structural proteins within viral genotype, lower levels of interpatient viral diversity were observed in HBsAg loss patients compared to controls. Conversely, in known regulatory regions encoding nonstructural proteins, increased interpatient viral diversity within HBsAg loss patients was generally observed relative to controls. In addition, the location and magnitude of interpatient viral diversity appear to be HBV genotype-dependent, with GT A patients exhibiting significantly less diversity across the entire genome relative to GT D patients. Previous work on HBV intrapatient viral diversity and treatment outcome has focused on different regions of the HBV genome, particularly structural-coding regions that are rich in T and B cell immune epitopes (21-24). Sequence variations within these regions are hypothesized to be a result of immune selection of escape variants. Hence, viral diversity is thought to be a reflection of viral evolution occurring in the face of host immune or cellular pressures, and frequently, antiviral agents. Off therapy, it is the virus-host interactions that drive the diversification of the viral population and, over time, these interactions frequently select for viral

11

escape mutants. This hypothesis is supported by our interpatient viral diversity results in the HBs and core regions, as significantly greater interpatient viral diversity was observed between patients that maintained high levels of HBsAg compared to those that achieved HBsAg loss. One explanation for this is that increased viral diversity in the structural regions perturbs host immune control, resulting in survival of more viral variants in these patients with persistently elevated HBsAg levels. Conversely, tighter immune control in patients who eventually lose HBsAg limits the survival and propagation of viral variants. Similar results have also been observed in patients infected with HCV. Increased intrapatient diversity within the hypervariable region (HVR) of the structural protein, E2, contributes to poor host antibody neutralizing response (25). Furthermore, multiple studies have observed that baseline intrapatient HCV diversity correlates with response to interferon-based therapy in combination with ribavirin, although results did vary depending on the region(s) studied and the clinical endpoint evaluated (18, 26-28). Our data revealed the opposite pattern was generally observed within HBV regions that encode nonstructural proteins, particularly those that were localized in areas with known regulatory function. In these regions, within HBV genotype, there was more conservation among control patients compared to HBsAg loss patients. One explanation is that selective pressures on regulatory regions encoding nonstructural proteins drive a different genetic response as compared to regions that encode structural proteins. It has been observed that natural selection drives the sequence conservation of important transcriptional regulatory elements (29). In the absence of strong, effective immune pressure, the virus may utilize the conservation of these regions to take advantage of an optimized regulatory system. The results observed in this study are in contrast to another study evaluating HBV viral diversity and the clinical endpoint of HBeAg seroconversion, where both treatment-naive subjects as well as those receiving interferon-based therapy, had increased intrapatient viral diversity within the precore/core region in patients prior to HBeAg seroconversion, compared to patients who did not seroconvert (21). However, there are several differences between this study and ours

12

that make comparisons challenging. First, in the aforementioned study, the precore (nonstructural) and core (structural) regions were combined for the analysis, whereas these regions were evaluated separately in our study. Second, their study evaluated genotype B patients, whereas this study evaluated genotype A and D patients. Indeed, the viral diversity across the core and HBx regions was not identical in genotype A and D patients, suggesting that genotype differences occur in these regions.

In contrast, our data are similar to preliminary results

presented by Lada et al. While making sequence comparisons across all HBV genotypes in patients receiving pegylated interferon and TDF combination therapy, this study showed an increase in viral sequence diversity within the HBs gene at baseline when comparing responders (reduction in viral load) with non-responders (24). There are certain limitations associated with our study. Population sequencing was used to analyze the HBV genome, which only detects mutations present in 20% or more of the population and does not allow for linkage analysis. However, as described previously, the population or consensus sequence approach reflects the majority of viral genetic variants within a single patient (11). Analyses using more sensitive approaches such as deep sequencing analyses are needed to confirm these results. Additionally, by using a population sequencing approach, bias may be attributed to tracts of high G +C content. However, comparisons within and between genotypes were conducted to identify any such sequencing artifacts in order to control for this kind of bias. Finally, this study is also constrained by the small number of patients evaluated from a single clinical study, that all subjects are HBeAg-positive, and only genotype A and D patients. Thus, interpretation of the study results is constrained to these two HBV genotypes and HBeAgpositive status for those receiving treatment with TDF. However, HBsAg loss in CHB patients, whether spontaneous or treatment-induced, is still a rare event and the current forefront of HBV research and drug development is still attempting to elucidate the basic mechanisms of virus-host interaction, as was the goal and main contribution of this study. Furthermore, study GS-US-1740103 was a global study with good representation of genotypes A and D. Therefore, we believe

13

the patients evaluated in this study represent the patient population likely to achieve HBsAg loss on TDF. However, larger studies of interpatient HBV viral diversity evaluating more genotypes are needed to verify the finding of this study and its applicability to other patient populations. This study is the first to evaluate interpatient HBV viral diversity across the entire HBV coding region and determine its impact on HBsAg loss. Overall, lower interpatient viral diversity was observed among HBsAg loss patients compared to controls primarily within genetic regions that encode for structural elements, while regions that encode nonstructural regulatory or signaling elements generally displayed higher interpatient viral diversity among HBsAg loss subjects compared to controls. This comparative analysis within GT A and D CHB patients suggests that, regardless of how each individual viral population evolved over time in response to host immune pressures or past drug exposure, certain patterns in interpatient viral diversity observed at the time of treatment initiation were associated with HBsAg loss on TDF therapy. Whether these results can be used to generate a predictive model of HBsAg response to antiviral therapy remains to be determined. Further work incorporating larger populations of patients with HBsAg loss on current or emerging therapies and using the latest DNA sequencing technology might further elaborate and validate this approach of using interpatient viral diversity to understand the host-virus evolution, guide patient selection for certain therapeutic intervention and outcome, and predict likelihood of targeted treatment response.

14

References 1. World Health Organization. Hepatitis B. Fact Sheet No. 204. July 2012. Available

at: http://www.who.int/mediacentre/factsheets/fs204/en. Accessed May 30, 2012. 2. Lok ASF and McMahon BJ. Chronic Hepatitis B: update 2009. Hepatology 2009; 50:135. 3. EASL. EASL Clinical Practice Guidelines: Management of chronic hepatitis B infection. J Hepatol 2012; 57:167-185. 4. Liaw YF, Kao JH, Piratvisuth T, Chan HLK, Chien RN, Liu CJ, Gane E et al. Asianpacific consensus statement on the management of chronic hepatitis B: a 2012 update. Hepatol Int 2012; 6:531-561. 5. Fattovich G, Giustina G, Sanchez-Tapias J, Quero C, Mas A, Olivotto PG, Solinas A, et al. Delayed clearance of serum HBsAg in compensated cirrhosis B: relation to interferon alpha therapy and disease prognosis. European Concerted Action on Viral Hepatitis (EUROHEP). Am J Gastroenterol. 1998; 93:896-900 6. Heathcote EJ, Marcellin P, Buti M, Gane E, De Man RA, Krastev Z, Germanidis G et al. Three-year efficacy and safety of tenofovir disoproxil fumarate treatment for chronic hepatitis B. Gastroenterology 2011;140(1):132-43. 7. Zoutendijk R, Hansen BE, van Vuuren AJ, Boucher CA, Janssen HL. Serum HBsAg decline during long-term potent nucleos(t)ide analogue therapy for chronic hepatitis B and prediction of HBsAg loss. J Infect Dis. 2011; 204(3):415-8. 8. Takkenberg RB, Lansen L, de Niet A, Zaaijer HL, Weegink CJ, Terpstra V et al. Baseline hepatitis B surface antigen (HBsAg) as predictor of sustained HBsAg loss in

15

chronic hepatitis B patients treated with peginterferon alfa-2a and adefovir Antivir Ther. 2013: 18(7):895-904. 9. Marcellin P, Buti M, Krastev Z, de Man R, Zeuzem S, Lou L, Gaggar A, et al. Kinetics of hepatitis B surface antigen loss in patients with HBeAg-positive chronic hepatitis B treated with tenofovir disoproxil fumarate J Hep 2014 [ePub ahead of print]. 10. Marcellin P, Gane E, Buti M, Afdal N, Sievert W, Jacobson IM et al. Regression of cirrhosis during tenofovir disoproxil fumarate treatment for chronic hepatitis B. Lancet 2013; 381(9865):468-75. 11. Donlin MJ, Cannon NA, Aurora R, Li K, Wahed AS, Di Bisceglie AM et al. Contribution of genome-wide genetic differences to outcome of interferon-based therapy in Caucasian American and African American patients. PLOS One 2010; 5(2): e9032. 12. Kramvis A and Kew MC. The core promoter of hepatitis B virus. J Viral Hep 1999; 6:415-427. 13. Chen BF. Liu CJ, Jow GM, Chen PJ, Kao JH, Chen DS. High prevalence and mapping of pre-S deletion in hepatitis B virus carriers with progressive liver diseases. Gastroenterology 2006; 130:1153-1168. 14. Desmond CP, Bartholomeusz A, Gaudieri S, Revil PA, Lewin SR. A systematic review of T-cell epitopes in hepatitis B virus: identification, genotypic variation and relevance to antiviral therapeutics. Antivir Ther 2008; 13:161-175. 15. Lucifora J, Arzberger S, Durantel D, Belloni L, Strubin M, Levrero M et al. Hepatitis B virus X protein is essential to initiate and maintain virus replication after infection. J Hepatology; 2011;55(5):996-1003. 16. Edgar, R. C. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 2004;32(5):1792–1797.

16

17. Guindon S, Delsuc F, Dufayard JF, Gascuel O. Estimating maximum likelihood phylogenies with PhyML. Methods Mol Biol. 2009;537:113-137. 18. Zeuzem S. Hepatitis C virus:kinetics and quasispecies evolution during anti-viral therapy. Forum 2000:10(1):32-42. 19. Levene, H. Robust tests for equality of variances. In: Olkin I, Ghurye SG, Hoeffding W, Madow WG, Mann HB,eds. Contributions to Probability and Statistics: Essays in Honor of Harold Hotelling, Menlo Park, CA: Stanford University Press, 1960:278–292. 20. Mizokami M, Orito E, Ohba K, Ikeo K, Lau JY, Gojobori T. Constrained evolution with respect to gene overlap of hepatitis B virus. J Mol Evol 1997;44(S1):S83-90. 21. Lim SG, Cheng Y, Guindon S, Seet BL, Lee LY, Hu P, Wasser S et al. Viral quasispecies evolution during hepatitis Be Antigen seroconversion. Gastroenterology 2007:133:951958. 22. Wu S, Imazeki F, Kurbanov F, Fukai K, Arai M, Kanda T, Yonemitshu Y et al. Evolution of hepatitis B genotype C viral quasi-species during hepatitis B e antigen seroconversion. J Hepatol 2011:54:19-25. 23. Cheng Y, Guindon S, Rodrigo A, Lim SG. Increased viral quasispecies evolution in HBeAg seroconverter patients treated with oral nucleoside therapy. J Hepatol 2013; 58:217-224. 24. Lada O, Zhang Q, Lapalus M, Martinot-Peignoux M, Laouenan C, Estrabaud E, De Muynk S et al. Quasispecies analysis of HBV strains isolated from chronic hepatitis B patients treated with peg-interferon + tenofovir therapy. J Hepatol 2013; 58(S1):S306. 25. Von Hahn T, Yoon JC, Alter H, Rice CM, Rehermann B, Balfe P, McKeating JA. Hepatitis C virus continuous escapes from neutralizing antibody and T cell responses during chronic infection in vivo. Gastroenterology 2007; 132:667-678.

17

26. De Queiroz ATL, Maracaja-Coutinho V, Jardim ACG, Rahal P, de Carvalho-Mello MVG, Matioli M. Relationship of pretreatment sequence diversity in NS5A region of HCV genotype 1 with immune response between pegylated INF/ribavirin therapy outcomes. J Viral Hepat 2011;18:142-148. 27. Fan Z, Mao Q, Zhou D, Lu Y, Xing J, Xu Y, Ray SC, Di Bisceglie AM. High diversity of hepatitis C viral quasispecies is associated with early virological response in patients undergoing antiviral therapy. Hepatology 2009; 50 (6):1765-1772. 28. Cannon NA, Donlin MJ, Fan X, Aurora R, Tavis JE. Hepatitis C virus diversity and evolution in the full open-reading frame during antiviral therapy. Plos One 2008;3(5):e2123. 29. Loots GG. Genomic identification of regulatory elements by evolutionary sequence comparison and functional analysis. In: van Heyningen V and Hill RE, eds. Advances in Genetics. Elsevier Inc. 2008; 61:269-293.

18

Acknowledgement The authors would like to acknowledge Eduardo B. Martins and Michael Miller for their editorial review of the manuscript.

Figure 1. Diagrams showing the regions analyzed, including nucleotide length. nt= nucleotides Panel a. Diagram showing the regions analyzed, including nucleotide length, within HBs and pol/RT genes. PanelCbb. Diagram showing PCthe regions analyzed, including nucleotide length,

C within HBx and Core genes. Segment nucleotide lengths are as follows: CP 87 nt, BCP 108, PC 87 nt, NRE 24 nt, CURS 107 nt, URR 131 nt. Panel c. Diagram of the HBV genome showing the relative position of the major regions. nt= nucleotides

Figure 2. Mean pairwise genetic distance for regions within HBs and pol/RT genes. The variance URR of the mean genetic distance for each region is summarized in Supplemental Table 1.

Figure 3. A sliding window analysis of the small HBs gene using overlapping 19 nucleotides segments migrating every 10 nucleotides spanning the HBs. The x-axis represents the center of each segment and y-axis represents the weighted mean genetic distance. Weights were defined as the ratio between the number of observed distances different from zero and the total number of possible positive distances within each group; down weighting regions with little variability. The approximate location of the a-determinant is nt 369-440 and the four transmembrane (TM) regions are nt 21-84; nt 240-291, nt 510-573 and nt 606-669 for TM regions 1,2,3 and 4, respectively. Grayed out segments correspond to statistically significant differences after accounting for multiple testing. Nt = nucleotides. Panel 3a. Sliding window analysis of HBs for GT D patients. Panel 3b. Sliding window analysis for GT A patients.

19

Figure 1a,b

Figure 1c

Figure 2

Figure 4. Mean pairwise genetic distance for regions within core and HBx genes. The variance of the mean genetic distance for each region is summarized in Supplemental Table 2.

20

Figure 3a

Figure 3b

Figure 4

Table 1 Genotype A Characteristics

Genotype D

Controls (n=17)

HBsAg loss (n=14)

p-value

Controls (n=10)

HBsAg loss (n=7)

p-value

Age > 40 years, n (%)

4 (24)

8 (57)

0.08

1 (10)

0 (0)

1.00

ALT >1.5 ULN, n (%) Mean or median values

17 (100)

13 (93)

0.45

9 (90)

7 (100)

1.00

Median ALT, U/L (range)

160 (60-670)

131 (50-264)

0.32

116 (62-421)

140 (75-425)

0.59

Male sex, n (%)

12 (71)

13 (93)

0.19

8 (80)

4 (57)

0.59

Self-reported years positive for HBV- no. (%) ≥4 yrs

9 (53)

1(7)

0.01

5 (50)

1 (14)

0.30

Previous Exposure to HBV Therapyc, n (%)

8 (47)

3 (21)

0.26

5 (50)

2 (29)

0.62

Previous Interferon Experience, n (%)

7 (41)

2 (14)

0.13

4 (40)

1 (14)

0.34

Mean HBV DNA, log10 copies/mL (SD)

9.01 (0.54)

9.19 (0.43)

0.20

8.95 (1.21)

9.32 (0.36)

0.81

Mean HBsAg, IU/mL (SD)

61,758 (61,124)

122,049 (72,976)

0.02

48,915 (40,348)

127,594 (66,624)

0.01

8.9a (1.6)

9.4a (1.3)

0.36

7.3b (2.0)

9.1 (2.2)

0.13

Mean Knodell necroinflammatory score (SD) a

N=16 (controls) and N=13 (HBsAg loss); bN=9; cPast treatment includes exposure to IFN, LAM.

Within genotype- Fisher’s exact test for categorical variables and Wilcoxon rank-sum test for continuous variables.