New Susceptibility and Resistance HLA-DP Alleles to HBV-Related Diseases Identified by a Trans-Ethnic Association Study in Asia

Previous studies have revealed the association between SNPs located on human leukocyte antigen (HLA) class II genes, including HLA-DP and HLA-DQ, and chronic hepatitis B virus (HBV) infection, mainly in Asian populations. HLA-DP alleles or haplotypes associated with chronic HBV infection or disease progression have not been fully identified in Asian populations. We performed trans-ethnic association analyses of HLA-DPA1, HLA-DPB1 alleles and haplotypes with hepatitis B virus infection and disease progression among Asian populations comprising Japanese, Korean, Hong Kong, and Thai subjects. To assess the association between HLA-DP and chronic HBV infection and disease progression, we conducted high-resolution (4-digit) HLA-DPA1 and HLA-DPB1 genotyping in a total of 3,167 samples, including HBV patients, HBV-resolved individuals and healthy controls. Trans-ethnic association analyses among Asian populations identified a new risk allele HLA-DPB1*09∶01 (P = 1.36×10−6; OR = 1.97; 95% CI, 1.50–2.59) and a new protective allele DPB1*02∶01 (P = 5.22×10−6; OR = 0.68; 95% CI, 0.58–0.81) to chronic HBV infection, in addition to the previously reported alleles. Moreover, DPB1*02∶01 was also associated with a decreased risk of disease progression in chronic HBV patients among Asian populations (P = 1.55×10−7; OR = 0.50; 95% CI, 0.39–0.65). Trans-ethnic association analyses identified Asian-specific associations of HLA-DP alleles and haplotypes with HBV infection or disease progression. The present findings will serve as a base for future functional studies of HLA-DP molecules in order to understand the pathogenesis of HBV infection and the development of hepatocellular carcinoma.


Introduction
Hepatitis B virus (HBV) infection is a major global health problem, resulting in 0.5-1.0 million deaths per year [1]. The prevalence of chronic HBV infection varies. About 75% of the chronic carriers in the world live in Southeast Asia and East Pacific [2]. Due to the introduction of vaccination programs, the prevalence of HBV infection in many countries has gradually been decreasing with consequent decreases in HBV-related hepatocellular carcinoma (HCC) [3]. Although some HBV carriers spontaneously eliminate the virus, about 10-15% of carriers develop liver cirrhosis (LC), liver failure and HCC [4]. Moreover, the progression of liver disease was revealed to be associated with the presence of several distinct mutations in HBV infections [5]. Genetic variations in STAT4 and HLA-DQ genes were recently identified as host genetic factors in a large-scale genome-wide association study (GWAS) for HBV-related HCC in China [6].
With regard to the genes associated with susceptibility to chronic HBV infection, HLA-DP and HLA-DQ genes were identified by GWAS in Japanese and Thai populations in 2009 [7] and 2011 [8], respectively. In addition, our previous GWAS confirmed and identified the association of SNP markers located on HLA-DPA1 (rs3077) and HLA-DPB1 (rs9277535) genes with susceptibility to chronic hepatitis B (CHB) and HBV clearance in Japanese and Korean subjects [9]. The significant associations of HLA-DP with CHB and HBV clearance have mainly been detected in Asian populations, such as Japanese [8,9], Thai [7], Chinese [10][11][12], and Korean [9]. In 2012, the association between HLA-DPA1 gene SNPs and persistent HBV infection was replicated in a Germany non-Asian population for the first time; however, this showed no association with HBV infection [13]. These results seem to be explained by the fact that allele frequencies of both rs3077 (0.155, 0.587 and 0.743 for C allele, on HapMap CEU, JPT, and YRI) and rs9277535 (0.261, 0.558 and 0.103 for G allele, on HapMap CEU, JPT, and YRI) are markedly different between populations. Moreover, the previous study showed that HBsAg seropositivity rates were higher in Thailand and China (5-12%) than in North America and Europe (0.2-0.5%) [2]. These results suggest that comparative analyses of HLA-DP alleles and haplotypes in Asian populations would clarify key host factors of the susceptible and protective HLA-DP alleles and haplotypes for CHB and HBV clearance. Here, we performed trans-ethnic analyses of HLA-DP alleles and haplotypes in Asian populations comprising Japanese, Korean, Hong Kong and Thai individuals. The findings from this study will serve as a base for future functional studies of HLA-DP molecules.

Characteristics of studied subjects
The characteristics of a total of 3,167 samples, including Japanese, Korean, Hong Kong and Thai subjects, are shown in Table 1. Each population included three groups of HBV patients, resolved individuals and healthy controls. The clinical definitions of HBV patients and resolved individuals are summarized in Materials and Methods. Some of the Japanese and all of the Korean samples overlapped with the subjects in our previous study [9,14].
We performed genotyping for HLA-DPA1 and HLA-DPB1 in all 3,167 samples, and a total of 2,895 samples were successfully genotyped. The characteristics of successfully genotyped samples are shown in Table S1.
Association of HLA-DPA1 and HLA-DPB1 alleles in Asian populations As for a general Asian population, including 464 Japanese, 140 Korean, 156 Hong Kong, and 122 Thai subjects, five HLA-DPA1 alleles and twenty-four HLA-DPB1 alleles were observed (Table  S2). The frequencies of HLA-DPA1 and HLA-DPB1 alleles were similar between Japanese and Korean subjects. On the other hand, the number of alleles with frequencies of 1-2% was larger in Hong Kong and Thai populations, despite the small sample size. Although the frequencies of HLA-DP alleles varied in Asian populations, HLA-DPB1*05:01 was the most prevalent with over 30% in all populations.
The associations of HLA-DPA1 and HLA-DPB1 alleles with chronic HBV infection (i.e., comparison between HBV patients and healthy controls) are shown in Table S2. To avoid false positives caused by multiple testing, the significance levels were corrected based on the numbers of HLA-DPA1 and HLA-DPB1  (Table S2). The pooled OR and 95% CI were 0.51 and 0.41-0.63, respectively in a meta-analysis (P = 3.15610 210 ) (Fig. S1A). As shown in Table S2, HLA-DPB1 shows higher degree of polymorphism than HLA-DPA1. The most common allele in Asian populations, HLA-DPB1*05:01, was significantly associated with HBV susceptibility in both Japanese and Korean subjects. Although HLA-DPB1*05:01 showed no significant association in the Hong Kong and Thai populations, the same direction of association (i.e., HBV susceptibility) was observed. Meta-analysis of the four populations revealed a significant association between HLA-DPB1*05:01 and susceptibility to HBV infection (P = 1.51610 24 ; OR = 1.45; 95% CI, 1.19-1.75) (Fig. S1B). The frequency of HLA-DPB1*09:01 was significantly elevated in Japanese HBV patients (15.7%) as compared with healthy controls (8.7%) (P = 3.70610 26 ; OR = 1.94; 95% CI, 1.45-2.62), and this association was most significant (i.e., the smallest P value) in the Japanese population. Because of lower allele frequencies of HLA-DPB1*09:01 or lack of statistical power in the other populations, no significant associations were observed. A common allele in Thai subjects, HLA-DPB1*13:01, was significantly associated with susceptibility to HBV infection (P = 2.49610 24 ; OR = 2.17; 95% CI, 1.40-3.47) with the same direction of associations in Japanese and Hong Kong (OR = 1.52 and 1.40, respectively).
The frequencies of associated HLA-DP alleles in a comparison of HBV patients with healthy controls (Table S2) or with HBVresolved individuals (Table S3) were similar in all four Asian populations. In the Japanese population, the associations of susceptible and protective HLA-DPB1 alleles to chronic HBV infection seem weaker in the comparison of HBV patients with HBV-resolved individuals than in the comparison of HBV patients with healthy controls. Moreover, the results of association analyses showed no difference in the comparison of HBV patients with HBV-resolved individuals, including or excluding HCV positive individuals (Table S3). In contrast, the association became stronger in the comparison of HBV patients with HBV-resolved individuals among the Korean subjects. The protective allele HLA-DPB1*04:01 was also identified to have a strong association with HBV clearance in Hong Kong subjects (Table S3). Moreover, in Hong Kong subjects, the HLA-DPB1*05:01 associated with the risk for HBV infection showed lower frequency in HBV-resolved individuals (42.9%) than in the healthy controls (48.1%), which accounts for a strong association in the comparison of HBV patients with HBV-resolved individuals (P = 6.24610 23 ; OR = 1.64; 95% CI, 1.14-2.36). Although the number of samples was insufficient, HLA-DP*100:01 showed a significant association with protection against HBV infection in the Hong Kong population (P = 3.05610 26 ; OR = 0.03; 95% CI, 0.0007-0.20).

Discussion
Among 2.2 billion individuals worldwide who are infected with HBV, 15% of these are chronic carriers. Of chronic carriers, 10-15% develops LC, liver failure and HCC, and the remaining individuals eventually achieve a state of nonreplicative infection, resulting in HBsAg negative and anti-HBc positive, i.e. HBVresolved individuals. To identify host genetic factors associated with HBV-related disease progression may lead HBV patients to discriminate individuals who need treatment.
Moreover, a significant association of HLA-DPB1*13:01 with risk of HBV infection (OR = 2.17; 95% CI, 1.40-3.47) was identified in the Thai subjects. However, the frequency of HLA-DPB1*13:01 in Thai healthy controls (11.5% in the present study) reportedly varies, ranging from 15.4% to 29.5%, due to the population diversity [15][16][17]. Therefore, a replication analysis is required to confirm the association of HLA-DPB1*13:01 with HBV infection in the Thai subjects. There were four other marginally associated HLA-DPB1 alleles with low allele frequencies below 5% in HBV patients and healthy controls, including HLA-DPB1*28:01, -DPB1*31:01, -DPB1*100:01, and -DPB1*105:01, in the Hong Kong and Thai subjects. Because these infrequent alleles may have resulted from false positive associations, the association needs to be validated in a large number of subjects.
HLA-DPB1*02:01 showed a significant association with protection against HBV infection in both Japanese and Hong Kong populations (Table S2); however, the HLA-DPB1*02:01 allele was not associated with HBV infection in the previous study [7]. Although HLA-DPB1*02:01 showed no association in either Korean or Thai populations, a significant association of HLA-DPB1*02:01 with protection against HBV infection among four Asian populations was detected in meta-analysis (P = 5.22610 26 ; OR = 0.68; 95% CI, 0.58-0.81) (Fig. S1B). We therefore conclude that the present finding is not a false positive.
We also identified a protective effect of HLA-DPB1*02:01 allele on disease progression in Asian populations. Previous studies identified the association of HLA class II genes including HLA-DQ and HLA-DR with development of HBV related hepatocellular carcinoma in the Chinese population [6,19,20]. In this study using Japanese and Korean samples, we identified significant associations between HLA-DPB1*02:01 and disease progression in CHB patients (P = 4.26610 25 ; OR = 0.45; 95% CI, 0.30-0.67, for Japanese and P = 8.74610 24 ; OR = 0.47; 95% CI, 0.29-0.75 for Korean) (Table S4). Although the association of HLA-DPB1*02:01 with disease progression was weaker after adjustment for age and gender in Korean subjects (P = 2.54610 22 ; OR = 0.55; 95% CI, 0.33-0.93), the same direction of association was observed (i.e. protective effect on disease progression) ( Table 2). The protective effects of HLA-DPB1*02:01 on disease progression showed a significant association after adjustment for age and gender in the Japanese population (P = 1.77610 24 ; OR = 0.47; 95% CI, 0.32-0.70); moreover, a significant association between HLA-DPB1*02:01 was observed among four Asian populations, under which population was adjusted by using dummy variables in a multivariate logistic regression analysis (P = 1.55610 27 ; OR = 0.50; 95% CI, 0.39-0.65) ( Table 2).
The HLA-DPA1 and HLA-DPB1 belong to the HLA class II alpha and beta chain paralogues, which make a heterodimer consisting of an alpha and a beta chain on the surface of antigen presenting cells. This HLA class II molecule plays a central role in the immune system by presenting peptides derived from extracellular proteins. We identified two susceptible haplotypes (DPA1*02:02-DPB1*05:01 and DPA1*02:01-DPB1*09:01) and three protective haplotypes (DPA1*01:03-DPB1*04:01, DPA1*01:03-DPB1*04:02, and HLA-DPA1*01:03-DPB1*02:01) to chronic hepatitis B infection, which may result in different binding affinities between HLA-DP subtypes and extracellular antigens. Although functional analyses of HLA-DP subtypes to identify HBV-related peptides are not fully completed, identification of susceptible and protective haplotypes as host genetic factors would lead us to understand the pathogenesis of HBV infection including viral factors.
In summary, we identified a new risk allele HLA-DPB1*09:01, which was specifically observed in Northeast Asian populations, Japanese and Korean. Moreover, a new protective allele HLA-DPB1*02:01 was identified among four Asian populations: Japanese, Korean, Hong Kong and Thai. The protective allele HLA-DPB1*02:01 was associated with both chronic HBV infection and disease progression in chronic HBV patients. Identification of a total of five alleles, including two risk alleles (DPB1*09:01 and DPB1*05:01) and three protective alleles (DPB1*04:01, DPB1*04:02 and DPB1*02:01), would enable HBV-infected individuals to be classified into groups according to the treatment requirements. Moreover, the risk and protective alleles for HBV infection and disease progression, identified in this study by means of trans-ethnic association analyses, would be key host factors to recognize HBV-derived antigen peptides. The present results may lead to subsequent functional studies into HLA-DP molecules and viral factors in order to understand the pathogenesis of HBV infection and development of hepatocellular carcinoma.

Ethics Statement
All study protocols conform to the relevant ethical guidelines, as reflected in the a priori approval by the ethics committee of National Center for Global Health and Medicine, and by the ethics committees of all participating universities and hospitals, including The University of Tokyo, Japanese Red Cross Kanto-using statistical software R2.9. To avoid false-positive results due to multiple testing, significance levels were adjusted based on the number of observed alleles at each locus in each population. For HLA-DPA1 alleles, the number of observed alleles was 3 in Japanese, 4 in Korean, 5 in Hong Kong, and 5 in Thai subjects. Therefore, the significant levels for a were set at a = 0.05/3 in Japanese, a = 0.05/4 in Korean, a = 0.05/5 in Hong Kong, and a = 0.05/5 in Thai subjects. In the same way, significant levels for HLA-DPB1 alleles were a = 0.05/10, 0.05/11, 0.05/12, and 0.05/ 16, respectively. Multivariate logistic regression analysis adjusted for age and sex (used as independent variables) was applied to assess associations between the number of DPB1*02:01 alleles (i.e., 0, 1, or 2) and disease progression in CHB patients. To examine the effect of DPB1*02:01 allele on disease progression in all populations, population was further adjusted by using three dummy variables (i.e., (c1, c2, c3) = (0, 0, 0) for Japanese, (1, 0, 0) for Korean, (0, 1, 0) for Hong Kong, and (0, 0, 1) for Thai) in a multivariate logistic regression analysis. We obtained the following regression equation: logit(p) = 23.905+0.083*age+(20.929)*-sex+(20.684)*DPB1*02:01+1.814*c1+(20.478)*c2+0.782*c3. Significance levels in the analysis of disease progression in CHB patients were set as a = 0.05/10 in Japanese, a = 0.05/11 in Korean, a = 0.05/15 in Hong Kong, and a = 0.05/15 in Thai subjects. The phase of each individual (i.e., a combination of two DPA1-DPB1 haplotypes) was estimated using PHASE software [21], assuming samples are selected randomly from a general population. In comparison of the estimated DPA1-DPB1 haplotype frequencies, significant levels were set as a = 0.05/14 in Japanese, a = 0.05/17 in Korean, a = 0.05/17 in Hong Kong, and a = 0.05/ 18 in Thai subjects. Meta-analysis was performed using the DerSimonian-Laird method (random-effects model) in order to calculate pooled OR and its 95% confidence interval (95% CI). We applied meta-analysis for alleles with frequency.1% in all four Asian populations. The significance levels in meta-analysis were adjusted by the total number of statistical tests; a = 0.05/20 for DPA1 alleles, a = 0.05/57 for DPB1 alleles, and a = 0.05/74 for DPA1-DPB1 haplotypes. Figure S1 Comparison of odds ratios in association analyses for HLA-DP with chronic HBV infection among four Asian populations: (A) HLA-DPA1 alleles; (B) HLA-DPB1 alleles; and (C) HLA DPA1-DPB1 haplotypes. Meta-analysis was performed using the DerSimonian-Laird method (random-effects model) to calculate pooled OR and its 95% confidence interval (95% CI). Bold depicts a statistically significant association after correction of significance level. (DOCX)