No association between IFNL3 (IL28B) genotype and response to peginterferon alfa-2a in HBeAg-positive or -negative chronic hepatitis B

Background & aims It has yet to be firmly established whether host IFNL3 (IL28B) genotype influences interferon responsiveness in patients with chronic hepatitis B. We investigated associations between single-nucleotide polymorphisms (SNPs) in the IFNL3 region and response to peginterferon alfa-2a in 701 patients enrolled in three large, randomized, international studies. Methods Responses were defined as hepatitis B surface antigen (HBsAg) loss and/or hepatitis B e antigen (HBeAg) seroconversion plus hepatitis B virus (HBV) DNA <2000 IU/ml in HBeAg-positive patients, and HBsAg loss and/or HBV DNA <2000 IU/ml in HBeAg-negative patients (24 weeks after end of treatment). Associations between treatment response and the number of copies of the poor-response allele at three SNPs (rs8099917, rs12980275, rs12979860) were explored with logistic regression models in Asian and white patients. Results The HBeAg-positive and -negative populations comprised 465 (92% Asian, 50% HBV genotype C) and 236 (79% Asian, 41% HBV genotype C) patients, respectively, and had respective response rates of 26% and 47%. The IFNL3 genotype was strongly associated with ethnicity. There was no association between IFNL3 genotype and treatment response in HBeAg-positive or -negative patients. Independent predictors of treatment response were: sex, HBV DNA level and alanine aminotransferase level in HBeAg-positive Asian patients; age in HBeAg-negative Asian patients; and HBV DNA in HBeAg-negative white patients. Conclusions This is the largest analysis to date of associations between IFNL3 genotype and peginterferon response in patients with chronic hepatitis B. The data suggest that IFNL3 polymorphism is not a major determinant of the response to peginterferon alfa-2a in either HBeAg-positive or HBeAg-negative patients.


Methods
Responses were defined as hepatitis B surface antigen (HBsAg) loss and/or hepatitis B e antigen (HBeAg) seroconversion plus hepatitis B virus (HBV) DNA <2000 IU/ml in HBeAg-positive patients, and HBsAg loss and/or HBV DNA <2000 IU/ml in HBeAgnegative patients (24 weeks after end of treatment). Associations between treatment response and the number of copies of the poor-response allele at three SNPs (rs8099917, rs12980275, rs12979860) were explored with logistic regression models in Asian and white patients.

Introduction
Single-nucleotide polymorphisms (SNPs) in the region of the gene for interferon lambda 3 (IFNL3, also known as IL28B; Gene ID: 282617) on chromosome 19 are strongly associated with clinical outcomes in patients infected with hepatitis C virus (HCV). IFNL3 genotype has been shown to be the strongest baseline predictor of a sustained virologic response to dual therapy with pegylated interferon (PegIFN) alfa plus ribavirin [1][2][3][4][5][6]. Differences in IFNL3 allele frequencies between different population groups largely explain the well-established ethnic differences in response rates to interferon (IFN)-based therapies [1,7,8]. IFNL3 genotype is also associated with the rate of spontaneous clearance of acute HCV infection [9][10][11][12]. The biological mechanism responsible for the IFNL3 genotype associations remains poorly defined, but responsiveness to exogenous/endogenous IFN is believed to be central, reflected by the observation that patterns of intrahepatic expression of IFN-stimulated genes differ according to IFNL3 genotype in patients with chronic HCV infection [13,14].
In contrast, it has yet to be firmly established whether, and to what extent, host IFNL3 genotype might influence the response to IFN-based therapy in patients with chronic hepatitis B (CHB). Several retrospective analyses with discordant results have examined the association between the response to IFN-based treatment and IFNL3 genotype [15][16][17][18][19][20][21][22][23]. Some studies have reported significant associations between SNPs in the IFNL3 region and treatment outcomes in hepatitis B e antigen (HBeAg)-positive [18] and HBeAg-negative [22] patients, while other studies have reported no association between these genetic polymorphisms and treatment outcomes [15][16][17]19,21,23]. However, between-study comparisons have been hampered by important differences in patient populations, such as HBeAg-antigen status, ethnic background, and hepatitis B virus (HBV) genotype, as well as treatment regimens and outcome definitions. HBV genotype is particularly important. Patients infected with HBV genotypes A or B are more likely to experience HBeAg seroconversion and hepatitis B surface antigen (HBsAg) loss after treatment with IFN [24]. The prevalence of HBV genotypes varies geographically, with genotypes A and D being more prevalent in the West, and genotypes B and C being more prevalent in Asia. Thus, the present study was designed with these deficiencies in mind. We examined associations between three markers of host IFNL3 genotype (rs12979860, in the form of salaries for authors HH, YZ, CW and BS, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the 'Contributions to the manuscript' section. Support for third-party writing assistance for this manuscript, furnished by Blair Jarvis Health Interactions, was provided by Roche Products Ltd, Welwyn, UK. LW has received fees for serving as a speaker, consultant, and advisory board member for AbbVie, Bristol-Myers Squibb (BMS), Gilead, Johnson & Johnson (J&J), and GlaxoSmithKline (GSK) within the last 3 years. Additional funding for the study was provided to LW by the China National Science and Technology Major Project for Infectious Diseases Control during the 12th Five-Year Plan Period. HW has received grant/research support from Abbott, BMS, Gilead, Merck, Novartis, Roche, Roche Diagnostics, and Siemens, and has received consulting fees and has been an advisory board and speakers' bureau member for Abbott, AbbVie, Biolex, BMS, Boehringer Ingelheim, Eiger Pharmaceuticals, Falk Foundation, Gilead, IST, J&J/Janssen-Cilag/Janssen TE, Medgenics, Merck/Schering-Plough, Novartis, Novira, Roche, Roche Diagnostics, Siemens, Transgene, and ViiV. Y-FL has served as an advisory board member for Roche. HL-YC has received fees for serving as a speaker and advisory board member for Roche, BMS, Gilead, Novartis, and MSD, and as a speaker for Echosens and GSK, and has received an unrestricted grant for HBV research from Roche. TP has received grant/ research support from BMS, Roche, MSD, Novartis, Fibrogen, and Bayer, and has been a member of advisory boards or speakers' bureaux for BMS, Roche, MSD, Novartis, and GSK. JJ has received fees for serving as a speaker, consultant, and advisory board member for BMS, MSD, Novartis, and Roche within the last 3 years. MRB has been a member of speakers' bureaux for BMS, Gilead, Roche, and Janssen, and has been a member of advisory boards for AbbVie, Roche, Gilead, and MSD. MD has been a member of speakers' bureaux for Roche, MSD, AbbVie, BMS, Gilead, and Janssen. SG has received fees as a member of advisory boards and speakers' bureaux for Roche, BMS, Gilead, Janssen, and MSD. YZ has been a Genentech/Roche employee and has owned Roche stock and options. HH and CW are employees of Roche and declare stock ownership. BS has received fees for serving as a consultant/ contract statistician for Roche, through BStats Solutions Ltd, from March 2013 to the present. AJT has received grant funding from AbbVie, Gilead Sciences, BMS, and Merck, has received rs8099917, and rs12980275) and response to 48 weeks of treatment with PegIFN alfa-2a using consistent outcome definitions in a large cohort of patients who were enrolled in three international randomized studies. The dataset included a large number of Asian and white patients with HBeAg-positive and HBeAg-negative CHB [25][26][27].

Subjects
Patients with CHB who were included in this analysis had received 48 weeks of treatment with PegIFN alfa-2a 180 μg/week with or without oral lamivudine in three large, randomized, international clinical studies [25][26][27]. The rationale for combining data from patients treated with and without lamivudine was that lamivudine therapy did not influence long-term outcomes with PegIFN alfa-2a in two large trials [25,26]. The complete inclusion and exclusion criteria, study designs, and primary results of these studies have been published elsewhere [25][26][27]; thus, only a brief overview will be provided. In the first study, HBeAg-negative patients with an HBV DNA level >100,000 copies/ml and elevated alanine aminotransferase (ALT) levels were randomized to 48 weeks of therapy with one of three treatments: (1) PegIFN alfa-2a plus oral lamivudine, (2) PegIFN alfa-2a plus matched placebo, or (3) oral lamivudine alone [25]. In the second study, the same three treatment regimens were evaluated in HBeAg-positive patients with an HBV DNA level >500,000 copies/ml and elevated ALT levels [26]. In the third study (NEPTUNE), HBeAg-positive patients with an HBV DNA level >100,000 copies/ ml and elevated ALT levels were randomized to either 24 or 48 weeks of treatment with PegIFN alfa-2a at a dosage of either 90 or 180 μg/week [27]. Patients co-infected with HCV or human immunodeficiency virus and individuals with a history or evidence of decompensated cirrhosis were excluded from each of the three trials.
This pooled retrospective analysis included only patients who received PegIFN alfa-2a at a dose of 180 μg/week and for a duration of 48 weeks, had provided informed consent for IFNL3 genotyping, and had an adequate sample available for testing.

Sample collection, processing, and analysis
Three tag SNPs in the IFNL3 region of chromosome 19 were investigated (rs12980275, rs8099917, and rs12979860). These SNPs were selected because they have been shown to influence the response to IFN-based therapies in patients with chronic HCV infection [1][2][3]9,27].
DNA was extracted from stored serum samples using the QIAamp DNA Blood Mini QIAcube Kit (Qiagen, Hilden, Germany). The concentration and presence of extracted DNA was measured and confirmed using a PicoGreen assay with a DNA/RNA spectrophotometer. Samples from China were genotyped using the MassARRAY System (Agena Bioscience, San Diego, CA, USA) at CapitalBio Corp., Beijing, China. Samples from regions outside of China were genotyped via TaqMan PCR (Roche Molecular Diagnostics, Pleasanton, CA, USA) at Quintiles-Singapore, Singapore.

Definition of response
The pre-specified primary outcome in all three studies was treatment response at 24 weeks after the end of treatment. For the purposes of this analysis, treatment response was defined as follows: (1) HBeAg seroconversion plus HBV DNA <2000 IU/ml or HBsAg loss in HBeAgpositive patients; or (2) HBV DNA <2000 IU/ml or HBsAg loss in HBeAg-negative patients.

Statistics
Separate statistical analyses were conducted for Asian and white patients within the HBeAgpositive and HBeAg-negative populations because of differences in HBV genotype distribution among ethnic groups. Baseline characteristics (age, sex, ethnic origin, HBV genotype, baseline ALT, baseline HBV DNA) were summarized overall and by ethnicity for the HBeAg-positive and HBeAg-negative populations. The Wald chi-square test was used to test for significant associations between treatment response and IFNL3 genotype for each SNP. For each of the three SNPs, unadjusted Wald odds ratios, 95% confidence intervals, and associated p-values were estimated from univariate logistic regression models to assess the quantitative effect of SNP genotype on outcome, fitting the IFNL3 SNP genotype as a fixed binary effect and treatment response as the binary outcome variable.
To identify other significant baseline predictors of treatment response, baseline covariates (sex, age, race, HBV genotype, HBV DNA level, and ALT level) were explored separately using univariate logistic regression models in HBeAg-positive Asian and white patients and in HBeAg-negative Asian and white patients. Significant associations were defined as p 0.05. The final analysis for each SNP was an additive logistic regression model, modelling the effect of genotype, adjusted for significant baseline covariates, on treatment response. All statistical analyses were performed using SAS 9.2 (SAS Institute Inc., Cary, NC, USA).
Deviation from Hardy-Weinberg equilibrium was assessed by Fisher's exact test, and linkage disequilibrium was assessed by the squared Pearson correlation coefficient (r 2 ) for each SNP using Plink (v1.07) (http://pngu.mgh.harvard.edu/purcell/plink/) [28]. Markers yielding Fisher's exact p-values of <0.001 were investigated for genotyping errors but were not excluded from further analysis.

Ethics
The clinical studies included in this analysis were conducted in accordance with the Declaration of Helsinki and the laws and regulations in force in the countries in which the research was conducted. In the original studies the protocols and all amendments were approved by the independent Institutional Review Boards of relevant institutions listed below. All patients provided informed written consent prior to participating in the studies. Genetic analysis of samples in China was approved by the Administration Office of National Human Genetic Resources of China.

Ethics committees and institutional review boards
The following ethics committees/institutional review boards, listed by country, considered and approved the study protocol.
Australia   Table 1, stratified by ethnicity. HBV genotypes B and C predominated in Asian patients, whereas genotypes A and D were more common in white patients ( Table 1). The genotype distribution of IFNL3 genotypes at rs12979860 in HBeAg-positive and HBeAg-negative patients is shown in Table 1. The overall frequency of CC and non-CC genotypes was similar in HBeAg-positive and HBeAg-negative patients. In the HBeAg-positive population, the CC genotype was more common in Asian (84%) than in white (36%) patients (p<0.0001). Similarly, in the HBeAg-negative population more Asian than white patients (86% versus 55%, p<0.0001) had a CC genotype. Similar patterns in genotype distribution were observed at rs12980275 and rs8099917 (Table 1). Linkage disequilibrium r 2 results indicated the strongest linkage disequilibrium was between rs12979860 and rs12980275, with r 2 values 0.68-1.0 across the subgroups of Asian and white HBeAg-positive and -negative patients. Scores were similar for rs12979860 and rs8099917 (r 2 0.43-0.65) and for rs12980275 and rs8099917 (r 2 0.47-0.79) (Table A in S1 File). Fisher's exact p-values, to test for deviation from Hardy-Weinberg equilibrium for rs12980275, rs12979860, and rs8099917, respectively, were 0.296, 0.011, and 0.188 (HBeAg-positive Asian patients), 1.0, 0.212, and 1.0 (HBeAg-negative Asian patients), 0.420, 0.242, and 1.0 (HBeAg-positive white patients) and 0.705, 0.705, and 0.578 (HBeAg-negative white patients). The deviation from Hardy-Weinberg equilibrium for rs12979860 in HBeAg-positive Asian patients could indicate either violation of one or more assumptions of Hardy-Weinberg equilibrium, could be an indication of genotyping errors [29], or could be by chance. The potential consequence of such a deviation is a biased estimate of the association's magnitude for rs12979860 in HBeAg-positive Asian patients. The overall response rate was 26% (119/465) among HBeAg-positive patients and 47% (110/236) among HBeAg-negative patients.

Results
Asian and white patients were analyzed separately, given the difference in frequency of HBV genotypes between the two groups ( Table 1). Results of the univariate logistic regression analysis indicated that there were no statistically significant associations between treatment response and genotype at any of the three SNPs in the HBeAg-positive Asian (Fig 1A) or white (Fig 1B) populations, or in the HBeAg-negative Asian (Fig 1C) or white populations (Fig 1D). The results did not change when HBeAg seroconversion alone was considered as a treatment outcome in HBeAg-positive patients ( Table 2). In both the HBeAg-positive and -negative  populations, numerically lower response rates were observed among white carriers of non-CC rs12979860 genotypes compared with carriers of CC rs12979860 genotypes (Fig 1B and 1D); however, as both analyses included only a very small number of samples, these differences failed to reach statistical significance.

Exploration of baseline covariates
Associations between baseline covariates and treatment response were explored separately by univariate logistic regression analyses in HBeAg-positive Asian and white patients (Fig 2A and   After adjusting for significant baseline covariates, there were no statistically significant associations between IFNL3 genotypes and response in the final logistic regression models (Fig 3). 2B) and in HBeAg-negative Asian and white patients (Fig 2C and 2D). For HBeAg-positive Asian patients, significant covariates (p<0.05) were sex, HBV DNA, and ALT. For HBeAgnegative Asian patients, age was the only significant baseline predictor of response. For HBeAg-negative white patients, HBV DNA was found to be significant. No significant associations were found for HBeAg positive white patients.

HBsAg loss
A total of 18 of 465 (3.9%) HBeAg-positive patients (including 12/428 Asian and 6/31 white patients) and 10 of 236 (4.2%) HBeAg-negative patients (including 8/186 Asian and 2/47 white patients) included in the analysis experienced HBsAg loss. Given these low event rates, it was not possible to test for an association between IFNL3 genotype and HBsAg loss.

Discussion
The identification of IFNL3 genotype as a biomarker predicting response to dual PegIFN alfa/ ribavirin therapy for chronic HCV infection was immediately useful in the clinic for identifying patients most likely to respond to therapy. A similar tool for stratifying IFN responsiveness in CHB patients would be extremely useful to predict, at baseline, which patients are most likely to respond to PegIFN alfa.
The underlying hypothesis for this study was that IFNL3 genotype would predict the outcome of treatment with PegIFN alfa in patients with CHB. However, our primary finding was that there was no association between IFNL3 genotype and treatment response. IFNL3 genotype was not associated with treatment response in HBeAg-positive or -negative patients of either Asian or white background.
Previous analyses that have examined relationships between these SNPs and response to IFN in patients with CHB have provided discordant results [15][16][17][18][19][20][21][22][23]. The largest of these analyses (n = 205) considered the relationship between IFNL3 genotype (rs12980275 and rs12979860) and treatment outcome in HBeAg-positive patients [18]. A significant association was reported between IFNL3 genotype and the specified primary outcome, which was HBeAg seroconversion alone. However, the relationship between response and IFNL3 genotype was no longer significant when a more stringent composite endpoint (i.e. HBeAg seroconversion plus HBV DNA level <2000 IU/ml at 24 weeks post-treatment) was considered [18]. Moreover, Sonneveld et al. [18] found that IFNL3 genotype was not a significant predictor of response to IFN when included in a previously calibrated PegIFN treatment index developed by Buster et al. [30] Similarly, IFNL3 genotype was not found to be associated with treatment response in a number of smaller cohorts of patients with HBeAg-positive CHB [15,17,21].
Lampertico et al. [22] found the rs12979860 CC genotype to be significantly associated with response to IFN in 101 HBeAg-negative patients with HBV genotype D infection who had received IFN-based therapies. IFNL3 genotype was associated with the primary response (HBsAg clearance) after a median of 11 years of follow-up. IFNL3 genotype was also associated with response defined as HBV DNA <2000 IU/ml at 6 months after treatment. It is important to note that the median duration of therapy was 23 months (range 10-48 months) [22]. Contrary to the results of Lampertico et al., a number of other analyses have not found a significant relationship between IFNL3 genotype and response to PegIFN alfa in patients with HBeAgnegative CHB [17,21], including a number of southern European cohorts, most of whom had HBV genotype D infection [20,23,31].
The explanation for these discordant results is most likely related to differences in sample size, patient characteristics (including ethnicity and HBV genotype), and the treatment regimen used. One of the strengths of the current study is the large sample size, as well as the very detailed patient phenotypes available from the PegIFN alfa-2a registration studies. The distribution of genotypes in Asian and white patients was consistent with previous reports of minor allele frequency in these populations [32][33][34][35][36][37]. The data are clear that, among Asian patients, there is no evidence for clinical utility of IFNL3 genotyping prior to consideration of PegIFN alfa-based therapy. Similarly, the data do not support a role for IFNL3 genotyping in white CHB patients prior to PegIFN alfa therapy, although we cannot entirely exclude a weak association with IFN responsiveness in this subgroup, given the smaller number of white patients, particularly with HBeAg-negative disease. In addition, it is also possible that subgroups within the heterogeneous Asian population with different minor allele frequencies may show some association.
The explanation for this difference in the relationship between IFNL3 genotype and IFN responsiveness between HBV and HCV is not clear. HBV and HCV are both hepatotropic viruses, and both respond to IFN therapy in vitro and in vivo. However, they have a number of fundamental differences in lifecycle and immune-evasion mechanisms. HBV is a DNA virus that has an episomal reservoir in the nucleus (cccDNA), and reverse transcription largely occurs within cytoplasmic capsids, protected from cytoplasmic RNA-sensing mechanisms. The strong association between intrahepatic IFN-stimulated gene expression and IFN responsiveness that is a feature of HCV infection, and that was recently linked to IFNL3 genotype, has not been described for HBV. Indeed, this seems to be a specific association between IFNL3 genotype and HCV, as it is not observed in normal liver [38]. Moreover, treatment with PegIFN alfa-2a has been observed to significantly reduce circulating and intrahepatic antigens in the absence of immune cell responses in a murine model of HBV infection [39]. Thus, fundamental differences in the host-virus relationship for HBV and HCV are clearly present, and remain an important focus for future research.
Prokunina-Olsson et al [40] have described a variant upstream of IFNL3 that creates a novel interferon gene, IFNL4, that is linked with HCV clearance in PegIFN-treated patients, possibly more so even than SNPs in IFNL3. The role of this variant in HBV response to PegIFN is currently unknown, however, and was not investigated in the current study.
In conclusion, the present study is the largest analysis to date of the association between IFNL3 genotype and response to IFN in patients with CHB. The data suggest that IFNL3 polymorphisms are not a major determinant of response to PegIFN alfa-2a in patients with HBeAg-positive or HBeAg-negative CHB. When selecting patients for treatment with PegIFN alfa-2a, clinicians should continue to use established baseline predictors of response.
Supporting information S1 File. Patients included from the original studies ( Fig A) and Linkage disequilibrium r2 values for each SNP combination in different subgroups (Table A). (DOCX)