Abstract
Background
Diversity in the HLA genes might be associated with disease outcomes—the heterozygote advantage hypothesis. We tested this hypothesis in relation to hepatitis B virus (HBV)-associated hepatocellular carcinoma (HCC).
Methods
We utilized DNA from > 10 000 Taiwanese individuals with current or past HBV infection to examine the association between HLA diversity and critical natural history steps in the progression from HBV infection to HCC. Individuals were classified as homozygotes at a given locus when imputed to carry the same 4-digit allele for the 2 HLA alleles at that locus.
Results
Increase in number of homozygous HLA class II loci was associated with an increased risk of chronic HBV infection (Ptrend = 1.18 × 10–7). Among chronic HBV carriers, increase in number of homozygous HLA class II loci was also associated with an increased risk of HBV-associated HCC (Ptrend = .031). For individual HLA loci, HLA-DQB1 homozygosity was significantly associated with HCC risk (adjusted hazard ratio = 1.40; 95% confidence interval, 1.06–1.84). We also found that zygosity affects risk of HCC through its ability to affect viral control.
Conclusions
Homozygosity at HLA class II loci, particularly HLA-DQB1, is associated with a higher risk of HBV-associated HCC.
Keywords: hepatocellular carcinoma, hepatitis B virus, human leukocyte antigen, cirrhosis, Taiwan, zygosity
Studying nearly 10 000 individuals in Taiwan, we found homozygosity at HLA class II loci is associated with a higher risk of HCC among chronic HBV carriers. HLA homozygosity also reduces the likelihood of early HBV clearance and viral load control.
Genes encoding the human leukocyte antigen (HLA) complex are highly polymorphic. Doherty and Zinkernagel hypothesized that greater diversity in HLA loci, which are responsible for antigen presentation to immune cells, could protect against infections (heterozygosity advantage hypothesis), by resulting in a higher probability of a robust cytotoxic T lymphocyte response to pathogens [1]. This hypothesis has been confirmed by reports showing that heterozygosity at HLA loci is linked to better control of pathogenic infections [2–5]. For example, heterozygosity at HLA class I loci has been observed to be associated with a slower progression to acquired immunodeficiency syndrome (AIDS) after infection with human immunodeficiency virus (HIV) [3].
A corollary of the heterozygote advantage hypothesis postulates that individuals who are heterozygous at HLA loci will have a lower risk of cancer development due to an increased ability to recognize exogenous antigens and de novo endogenous mutations in cancer and its precursors. In support of this hypothesis, some studies have reported on the association between zygosity at HLA loci and cancers of hematopoietic origin (eg, subtypes of non-Hodgkin lymphoma) [6–8]. Whether heterozygosity advantage is evident for cancers caused by infectious agents is less well understood. In particular, investigation of the role of zygosity at HLA loci in progression to cancer and its precursors given chronic carriage of an oncogenic infection has never been reported.
Motivated by an initial suggestion that heterozygosity at the HLA-DR-DQ regions increases the likelihood of hepatitis B virus (HBV) infection clearance [2], the present study evaluated whether zygosity at HLA loci is associated with progression to hepatocellular carcinoma (HCC), given chronic HBV infection. In addition, we were able, for the first time, to investigate whether zygosity associations with HCC are explained by viral control or other factors that affect progression to cancer.
METHODS
Study Design and Participants
DNA was obtained from 6187 individuals negative for hepatitis B surface antigen (HBsAg) but positive for antibodies against hepatitis B core antigen (anti-HBc, indicating past infection) from the Taiwan Biobank (TWB); 3416 chronic HBV carriers (306 HCC cases and 3110 chronic HBV carriers without HCC) from a longitudinal cohort (Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer-Hepatitis B Virus [REVEAL-HBV] study); and 400 patients with HBV-related HCC from the Taiwan Liver Cancer Network (TLCN) (Table 1). Detailed information has been published previously [9–11] and is provided in Supplementary Materials and Supplementary Table 1. This study received both Taiwan and National Cancer Institute Institutional Review Board approval. Written informed consent was obtained from all participants. Patients or the public were not involved in the design, conduct, reporting, or dissemination plans of our research.
Table 1.
Study Population
| Study Set | Study No. | Study Type | HBsAg | Anti-HBc | Clinical Implication | Group | Follow-up HBV-HCC Status | Follow-up HBV-Liver Cirrhosis Status |
|---|---|---|---|---|---|---|---|---|
| Taiwan Biobank | 6187 | Biobank | Negative | Positive | Past infection | Healthy individuals who cleared HBV | NA | NA |
| REVEAL-HBV | 3416 | Longitudinal cohort | Positive | Positive | Chronic hepatitis B infection | Chronic hepatitis B carriers | Incident HCC (n = 306) | Incident liver cirrhosis (n = 339) |
| Taiwan Liver Cancer Network | 400 | Biobank | Positive | Positive | Chronic hepatitis B infection | Chronic hepatitis B carriers (HCC) | NA | NA |
Abbreviations: HBc, hepatitis B core; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; NA, not applicable.
Procedures
Genotyping experiments and subsequent analyses for TWB samples were conducted by the National Center for Genome Medicine, Academia Sinica, on a custom Axiom Genome-Wide Array Plate (the TWB chip) based on technology developed by Affymetrix. The TWB chip contains a selection of approximately 653 000 single-nucleotide polymorphism (SNP) sites. For individuals recruited in REVEAL-HBV and TLCN, human genomic DNA was extracted from peripheral blood leukocytes, using a standard method, and was genotyped using the Axiom-CHB1 genome-wide array comprising 642 832 SNPs. All coordinates refer to genome build HG19/GRCh37. Genotypes were called using the Axiom Analysis Suite software. SNP& Variation Suite Version 8.0 (Golden Helix) was used to conduct systematic quality-control steps on genotype calls as detailed in Supplementary Materials. Plots of the first 2 principal components (PCs) for subjects included in TWB, REVEAL-HBV, and TLCN are shown in Supplementary Figure 1. Strong population substructure was not observed.
After stringent quality-control steps, we used the genotyping data from 25 759 242 to 33 534 827 bp on chromosome 6 based on hg19 positions to impute 4-digit HLA alleles at 6 HLA loci, including classical HLA class I genes (HLA-A, HLA-B, and HLA-C) and classical HLA class II genes (HLA-DRB1, HLA-DQB1, and HLA-DPB1). Imputation was conducted based on the platform-specific Asian reference (Affymetrix Axiom Genome-Wide CHB 1 Array Plate) provided in HLA genotype imputation with attribute bagging (HIBAG), an R package [12]. Posterior probability for each possible HLA allele was also calculated and shown in Supplementary Figure 2. No meaningful difference on posterior probability was observed across our 3 study sets. Individuals were classified as homozygotes at a given locus when imputed to carry the same 4-digit allele for the 2 HLA alleles at that locus.
To validate the imputation results, we selected samples from 295 subjects in the REVEAL-HBV cohort for high-resolution HLA typing by next-generation sequencing at 6 HLA loci. We compared imputed HLA alleles to 4-digit HLA sequencing data and found high concordance between actual genotype and imputation results at all loci (Supplementary Table 2). Importantly, accuracy of >93% was observed for each of the 6 HLA loci evaluated. More detailed information is provided in Supplementary Materials.
Statistical Analysis
Detailed information for analyses conducted is provided in Supplementary Materials. Briefly, we included age, sex, and eigenvectors to adjust for genetic/ethnic origin in all models. Other variables considered included known HCC risk factors, such as positivity of HBeAg, serum alanine transaminase level, smoking, alcohol consumption, body mass index, and diabetes. However, those risk factors did not change the magnitudes of the associations with HLA zygosity and were included in the primary models.
Analyses of chronic HBV infection were conducted on data in the REVEAL-HBV cohort and TWB using unconditional logistic regression to estimate adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs). Although there was no strong population substructure (Supplementary Figure 1), we additionally included ethnicity (Minnan, Hakka, and others: Aborigines and Mainlanders) in the model to account for residual population substructure. Controls included individuals with HBsAg negative and anti-HBc positive (n = 6187).
To assess the association between zygosity at HLA loci and HCC/liver cirrhosis among chronic HBV carriers, and to fully utilize the time-to-event data for REVEAL-HBV cohort, we used proportional hazards regression models to estimate hazard ratios (HRs) and corresponding 95% CIs based on an attained-age time scale. Models were additionally adjusted for HBV DNA load (ordinal variables: <300 [undetectable], 300–9999, 10 000–99 999, 100 000–999 999, and ≥1 million copies/mL) [9] In addition, to assess whether there were potential independent effects of outside of the main associated zygosity (due to specific HLA alleles or SNPs in the major histocompatibility complex [MHC] region), we additionally adjusted models for HLA allele (DQB*03:01) and SNP (rs9270649), which appeared to be mostly strongly associated with HCC in this study. Likelihood ratio tests were used to compare models with and without an interaction term between zygosity at HLA loci and HBV genotype (B or BC vs C). In analyses including data from both REVEAL-HBV and TLCN, we used unconditional logistic regression to estimate ORs and corresponding 95% CIs. This was done because time-to-event data were not available in the TLCN. All HCC cases from the REVEAL-HBV cohort and TLCN were treated as 1 case group.
Comparison of baseline HBV DNA load between HLA homozygotes and heterozygotes was performed with the Wilcoxon-rank sum test among individuals with information on baseline HBV DNA load from the REVEAL-HBV cohort. To adjust for potential confounders, ordinal logistic regression model was used, using categories of HBV DNA load as ordinal outcome (<300 [undetectable], 300–9999, 10 000–99 999, 100 000–999 999, and ≥1 million copies/mL) [9]. Long-term changes in HBV DNA levels for participants with HBV DNA levels ≥300 copies/mL at enrollment by HLA homozygotes and heterozygotes were plotted using the “geom_smooth” function in the “ggplot2” R package, fitting a generalized additive model (method “gam”). Direct and indirect effects of the HLA diversity on HCC risk in relation to HBV replication (ie, HBV DNA level at baseline) were evaluated using mediation analyses (R package “mediation” 4.4.7). Proportion of mediation was estimated based on coefficients for zygosity at HLA-DQB1 in regression models on linear scale.
Analyses of serological outcomes (ie, seropersistence of HBV DNA and HBsAg) were conducted among REVEAL-HBV cohort. HRs and 95% CIs were estimated from Cox proportional hazards regression using time-in-study as the time scale.
Finally, we conducted sensitivity analyses that were restricted to a postimputation quality control call threshold >0.5 (recommended by HIBAG) [12] to remove poorly imputed HLA alleles and to reduce potential bias caused by imputation error. We also conducted analysis stratified by HLA-DQB*03:01 to account for confounding effect due to specific HLA alleles. Results were largely consistent with those from the primary analyses (data not shown).
All statistical analysis was performed in the R Statistical Computing environment version 3.3.1 (http://www.rproject.org). All statistical tests were 2-sided, and P values less than .05 were considered statistically significant.
RESULTS
Homozygosity at HLA Class II Loci Is Associated With Increased Risk of Chronic HBV Infection
As a first step, we wanted to confirm the report by Thursz et al [2] that used low-resolution HLA typing among 632 Ghanaian subjects to demonstrate that persistent HBV infection is associated with homozygosity at HLA-DR-DQ [2]. Because over 90% of the adult population in Taiwan is infected with HBV at a young age prior to implementation of a nationwide newborn HBV vaccination starting in the late 1980s [13], we were able to evaluate this question in a large series of 9603 adult Taiwanese who were either long-term chronic HBV carriers and free of liver cirrhosis/HCC at study entry (3416 individuals from the REVEAL-HBV cohort) or who had evidence of HBV infection that successfully cleared (6187 individuals from TWB who were HBsAg negative but anti-HBc positive [14]). Comparing these 2 groups, we found that HLA class II homozygosity (but not HLA class I) for at least 1 locus was associated with an increased risk of chronic HBV infection (adjusted OR = 1.36; 95% CI, 1.20–1.54; P = 1.71 × 10–6; Table 2), after adjustment for sex, age, PCs, and ethnic group. Furthermore, increase in number of homozygous HLA class II loci was associated with an increased risk of chronic HBV infection (Ptrend = 1.18 × 10–7). The effect of homozygosity was observed for each individual HLA class II locus (ie, HLA-DPB1, -DQB1, and -DRB1), with the adjusted ORs associated with homozygosity at HLA-DPB1, -DQB1, and -DRB1 of 1.37 (95% CI, 1.20–1.56; P = 3.97 × 10–6), 1.30 (95% CI, 1.10–1.55; P = 2.74 × 10–3), and 1.40 (95% CI, 1.14–1.73; P = 1.58 × 10–3), respectively. Therefore, our findings, using high-resolution HLA calling of nearly 10 000 individuals, are in line with the previous report and suggest that homozygosity at HLA class II loci reduces the likelihood of early clearance of HBV infection.
Table 2.
Association Between Zygosity at HLA Class I Loci and Class II Loci and Chronic HBV Infection Among People With a History of HBV Infection in Taiwan
| HLA Zygosity | HBsAg Positive, REVEAL-HBV cohort (n = 3416) | HBsAg Negative/Anti-HBc Positive, TWB (n = 6187) | Adjusted OR (95% CI)a |
|---|---|---|---|
| Class I loci | |||
| Heterozygous at all loci | 2309 (67.6) | 4408 (71.2) | Ref. |
| Homozygous in at least 1 locus | 1107 (32.4) | 1779 (28.8) | 1.00 (.88–1.14) |
| Number of homozygous | |||
| 1 | 786 (23.0) | 1290 (20.9) | 0.96 (.83–1.12) |
| 2 | 218 (6.4) | 324 (5.2) | 1.09 (.84–1.41) |
| 3 | 103 (3.0) | 165 (2.7) | 1.14 (.78–1.68) |
| P trend | .59 | ||
| HLA-A | |||
| Heterozygote | 2777 (81.3) | 5128 (82.9) | Ref. |
| Homozygote | 639 (18.7) | 1059 (17.1) | 0.95 (.81–1.11) |
| HLA-B | |||
| Heterozygote | 6059 (89.5) | 5598 (90.5) | Ref. |
| Homozygote | 357 (10.5) | 589 (9.5) | 1.08 (.88–1.31) |
| HLA-C | |||
| Heterozygote | 2881 (84.3) | 5402 (87.3) | Ref. |
| Homozygote | 535 (15.7) | 785 (12.7) | 1.11 (.93–1.32) |
| Class II loci | |||
| Heterozygous at all loci | 1865 (54.6) | 4012 (64.8) | Ref. |
| Homozygous in at least 1 locus | 1551 (45.4) | 2175 (35.2) | 1.36 (1.20–1.54) |
| Number of homozygous | |||
| 1 | 1102 (32.3) | 1653 (26.7) | 1.27 (1.11–1.46) |
| 2 | 268 (7.8) | 355 (5.7) | 1.56 (1.22–1.99) |
| 3 | 181 (5.3) | 167 (2.7) | 1.78 (1.28–2.48) |
| P trend | 1.18 × 10–7 | ||
| HLA-DPB1 | |||
| Heterozygote | 2203 (64.5) | 4563 (73.8) | Ref. |
| Homozygote | 1213 (35.5) | 1624 (26.2) | 1.37 (1.20–1.56) |
| HLA-DQB1 | |||
| Heterozygote | 2835 (83.0) | 5432 (87.8) | Ref. |
| Homozygote | 581 (17.0) | 755 (12.2) | 1.30 (1.10–1.55) |
| HLA-DRB1 | |||
| Heterozygote | 3029 (88.7) | 5702 (92.2) | Ref. |
| Homozygote | 387 (11.3) | 485 (7.8) | 1.40 (1.14–1.73) |
Data are No. (%).
Abbreviations: CI, confidence interval; HBc, hepatitis B core; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; OR, odds ratio; Ref., reference; REVEAL-HBV, Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer-Hepatitis B Virus study; TWB, Taiwan Biobank.
aAdjusted for sex, age (in categories, 30–39, 40–49, 50–59, 60–79 years), ethnicity, and principal components.
Homozygosity at HLA Class II Loci, Particularly HLA-DQB1, But Not HLA Class I Loci, Is Associated With a Higher Risk of HCC
Next, to investigate whether zygosity at HLA loci is associated with progression to HCC among chronic HBV carriers, we evaluated the 3416 HBV chronic carriers recruited in the REVEAL-HBV cohort in 1991–1992 and followed through the end of 2016. By the end of follow-up, a total of 306 incident HCC cases were identified in this cohort. We found that increase in number of homozygous HLA class II loci was associated with an increased risk of HCC after adjustment for sex, age, and PCs (Ptrend = .031; Table 3). The adjusted HRs associated with having 1, 2, or 3 homozygous loci at the 3 HLA class II loci versus being heterozygous at all 3 of these HLA class II loci were 1.15 (95% CI, .89–1.47), 1.21 (95% CI, .80–1.81), and 1.64 (95% CI, 1.04–2.59), respectively. When individual HLA class II loci were evaluated, only homozygosity at HLA-DQB1 was significantly associated with an increased risk of HCC (adjusted HR = 1.40; 95% CI, 1.06–1.84; P = .016). These results were consistent in an analysis that included HCC cases from REVEAL-HBV and TLCN combined (Supplementary Table 3). For example, the adjusted OR for homozygosity at HLA-DQB1 was 1.39 (95% CI, 1.13–1.72; P = .002) in analysis that incorporated HCC cases from TLCN. Associations remained significant when analyses were adjusted for the SNP (rs9270649, HR for homozygosity at HLA-DQB1 = 1.35, P = .033) or HLA allele (HLA-DQB1*03:01, HR = 1.34, P = .035) most strongly associated directly with HCC in our dataset (Supplementary Table 4), indicating that the observed zygosity associated is independent of the potential drivers in the MHC region of that association. We found no statistically significant heterogeneity in the association with HLA class II homozygosity and HBV genotype (B or BC vs C, Pheterogeneity > .05).
Table 3.
Association Between Zygosity at HLA Class I Loci and Class II Loci and HCC Among REVEAL-HBV Cohort in Taiwan
| HLA Zygosity | Incident HCC Cases (n = 306) | Chronic HBV Carriers Without HCC (n = 3110) | HR Adjusted for Age, Sex, and Principal Components (95% CI)a | HR Further Adjusted for Viral Load (95% CI)b |
|---|---|---|---|---|
| Class I loci | ||||
| Heterozygous at all loci | 191 (62.4) | 2118 (68.1) | Ref. | Ref. |
| Homozygous in at least 1 locus | 115 (37.6) | 992 (31.9) | 1.31 (1.04–1.65) | 1.27 (1.00–1.60) |
| Number of homozygous loci | ||||
| 1 | 88 (28.8) | 698 (22.4) | 1.42 (1.10–1.83) | 1.31 (1.02–1.68) |
| 2 | 18 (5.9) | 200 (6.4) | 0.99 (.61–1.60) | 1.11 (.68–1.80) |
| 3 | 9 (2.9) | 94 (3.0) | 1.18 (.60–2.31) | 1.25 (.64–2.45) |
| P trend | .168 | .130 | ||
| HLA-A | ||||
| Heterozygote | 244 (79.7) | 2533 (81.4) | Ref. | Ref. |
| Homozygote | 62 (20.3) | 577 (18.6) | 1.13 (.85–1.49) | 1.13 (.86–1.50) |
| HLA-B | ||||
| Heterozygote | 271 (88.6) | 2788 (89.6) | Ref. | Ref. |
| Homozygote | 35 (11.4) | 322 (10.4) | 1.11 (.78–1.58) | 1.11 (.78–1.58) |
| HLA-C | ||||
| Heterozygote | 252 (82.4) | 2629 (84.5) | Ref. | Ref. |
| Homozygote | 54 (17.6) | 481 (15.5) | 1.24 (.93–1.67) | 1.27 (.95–1.71) |
| Class II loci | ||||
| Heterozygous at all loci | 157 (51.3) | 1708 (54.9) | Ref. | |
| Homozygous in at least 1 locus | 149 (48.7) | 1402 (45.1) | 1.21 (.97–1.51) | 1.17 (.93–1.46) |
| Number of homozygous | ||||
| 1 | 101 (33.0) | 1001 (32.2) | 1.15 (.89–1.47) | 1.13 (.88–1.46) |
| 2 | 27 (8.8) | 241 (7.7) | 1.21 (.80–1.81) | 1.04 (.69–1.56) |
| 3 | 21 (6.9) | 160 (5.1) | 1.64 (1.04–2.59) | 1.70 (1.08–2.70) |
| P trend | .031 | .068 | ||
| HLA-DPB1 | ||||
| Heterozygote | 193 (63.1) | 2010 (64.6) | Ref. | Ref. |
| Homozygote | 113 (36.9) | 1100 (35.4) | 1.16 (.92–1.46) | 1.20 (.95–1.51) |
| HLA-DQB1 | ||||
| Heterozygote | 240 (78.4) | 2595 (83.4) | Ref. | Ref. |
| Homozygote | 66 (21.6) | 515 (16.6) | 1.40 (1.06–1.84) | 1.21 (.92–1.60) |
| HLA-DRB1 | ||||
| Heterozygote | 367 (87.3) | 2762 (88.8) | Ref. | Ref. |
| Homozygote | 39 (12.7) | 348 (11.2) | 1.17 (.84–1.64) | 1.15 (.82–1.61) |
Data are No. (%).
Abbreviations: CI, confidence interval; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HR, hazard ratio; REVEAL-HBV, Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer-Hepatitis B Virus study.
aHRs and 95% CIs were estimated from Cox proportional hazards regression models with attained age as the time scale and were adjusted for sex and principal components. Censored at death, diagnosis of HCC, and end of follow-up (2016), whichever occurred first.
bAdditionally adjusted for baseline HBV DNA load (in categories, <300, 300–9999, 10 000–99 999, 100 000–999 999, ≥1 000 000 copies/mL). Subjects with missing on baseline HBV DNA load were retained in the analysis (n = 176).
It is known that serum HBV DNA levels are one of the strongest predictors of HBV-related liver disease risk [9, 15–17]. We therefore evaluated the impact of additional adjustment for baseline HBV DNA load on the association between HLA zygosity and HCC risk (Table 3). In doing so, we noted continued evidence that homozygosity at all 3 HLA class II loci was associated with a higher risk of developing HCC (adjusted HR = 1.70; 95% CI, 1.08–2.70) but also noted an attenuated association between homozygosity at HLA-DQB1 and HCC risk (adjusted HR = 1.21; 95% CI, .92–1.60), suggesting that zygosity associations might be partially explained by the effect of HLA zygosity on control of viral replication.
Homozygosity at HLA-DQB1, But Not Other HLA Class II Loci, Is Associated With Viral Replication
We next wanted to examine more directly whether zygosity at HLA loci affects risk of HCC through its ability to affect viral control among chronic HBV carriers and/or through other mechanisms. To do so, we first investigated the association between homozygosity at HLA class II loci and viral load at both baseline and during follow-up in chronic HBV carriers who were free of liver cirrhosis and HCC at entry into the REVEAL-HBV cohort. We found significantly higher serum HBV DNA levels at baseline for individuals homozygous at HLA-DQB1 (P = .00077; Figure 1A). The association remained statistically significant after adjustment for sex, age, PCs, and significant SNP (P = .0020) or HLA-DQB1*0301 (P = .0019). Furthermore, we observed that homozygosity at HLA-DQB1 was associated with consistently higher HBV DNA levels throughout more than a decade of follow-up among individuals with detectable serum HBV DNA level (≥300 copies/mL) at study entry (Figure 1B). Such associations were consistently observed in sensitivity analyses restricted to subjects who did not develop HCC during follow-up or to subjects who did not clear HBsAg during follow-up (Supplementary Figure 3). Other HLA class II loci were not associated with serum HBV DNA levels among chronic HBV carriers (Supplementary Figure 4). To further clarify the roles of homozygosity at HLA-DQB1 and serum HBV DNA level in the development of HCC, we conducted a mediation analysis, and found that 46.9% of effect due to homozygosity at HLA-DQB1 on HCC risk was mediated through the serum HBV DNA load at baseline. These results support control of HBV viral replication as one mechanism by which HLA zygosity affects the risk of developing HCC.
Figure 1.
Distribution of serum hepatitis B virus (HBV) DNA level according to zygosity at HLA-DQB1 for (A) baseline and (B) during follow-up. P value was obtained from Wilcoxon rank sum test. Generalized additive model was used to draw the regression lines.
Homozygosity at HLA Class II Loci Was Not Associated With Long-term Seropersistence of HBV Infection
While all study subjects in the REVEAL-HBV cohort were long-term chronic HBV carriers infected with HBV early in life, we had the opportunity to evaluate whether homozygosity at HLA-DQB1 was associated with continued persistence versus late clearance of HBV infection, as indicated by seropersistence of HBV DNA or HBsAg during follow-up. Of the cohort participants, 17.6% cleared infection during an average of 12 years of follow-up (average of 1.5% per year). Among chronic HBV carriers with detectable serum HBV DNA level (≥300 copies/mL) at entry into the REVEAL-HBV cohort, homozygosity at HLA-DQB1 was not associated with seropersistence of HBV DNA (Supplementary Table 5). Homozygosity at HLA-DQB1 was also not associated with seropersistence of HBsAg. Taken together with findings summarized above, these results suggest that while zygosity is associated with the initial control over HBV infection (persistence vs clearance; Table 2) and with viral load among chronic carriers (Figure 1), zygosity is not associated with late persistence among long-term HBV carriers (Supplementary Table 5).
No Statistically Significant Association Between Homozygosity at HLA-DQB1 and Liver Cirrhosis
Next, we examined whether homozygosity was associated with an increased risk of liver cirrhosis, an important HCC precursor, among those in the REVEAL-HBV cohort who were not able to adequately control HBV viral load. For this analysis, we restricted analyses to subjects who had >10 000 virus copies/mL in circulation at baseline and seropersistence of HBV DNA during follow-up, suggestive of a lack of viral control. The association between homozygosity at HLA-DQB1 and risk of liver cirrhosis (adjusted HR = 1.27; 95% CI, .92–1.74; P = .15; Table 4) was not statistically significant. Nevertheless, the point estimate was in the positive direction and consistent with other findings. This leaves open the possibility that mechanisms other than those associated with viral control might partially explain the association between HLA-DQB1 zygosity and clinical manifestations of chronic HBV infection. No associations were observed for other HLA class II loci. Sensitivity analyses that evaluated the association between HLA-DQB1 zygosity and risk of liver cirrhosis among all subjects from the REVEAL-HBV cohort were consistent with those presented in Table 4 (data not shown).
Table 4.
Association Between Zygosity at HLA Class II Loci and Liver Cirrhosis in the REVEAL-HBV Cohort in Taiwan Among Chronic HBV Carriers With Inadequate Control of HBV Viral Load
| Among Baseline HBV DNA ≥ 10 000 copies/mL and HBV DNA Persistence | |||
|---|---|---|---|
| HLA Zygosity | Liver Cirrhosis (n = 229) | Chronic HBV Carriers Without Liver Cirrhosis (n = 988) | HR Adjusted for Age, Sex, Principal Components, and Viral Load (95% CI)a |
| Class II loci | |||
| Heterozygous at all loci | 119 (52.0) | 522 (52.8) | Ref. |
| Homozygous in at least 1 locus | 110 (48.0) | 466 (47.2) | 1.07 (.82–1.39) |
| Number of homozygous | |||
| 1 | 77 (33.6) | 329 (33.3) | 1.05 (.79–1.40) |
| 2 | 22 (9.6) | 81 (8.2) | 1.08 (.69–1.71) |
| 3 | 11 (4.8) | 56 (5.7) | 1.16 (.62–2.16) |
| P trend | .56 | ||
| HLA-DPB1 | |||
| Heterozygote | 151 (65.9) | 616 (62.3) | Ref. |
| Homozygote | 78 (34.1) | 372 (37.7) | 1.07 (.71–1.23) |
| HLA-DQB1 | |||
| Heterozygote | 179 (78.2) | 810 (82.0) | Ref. |
| Homozygote | 50 (21.8) | 178 (18.0) | 1.27 (.92–1.74) |
| HLA-DRB1 | |||
| Heterozygote | 203 (88.6) | 879 (89.0) | Ref. |
| Homozygote | 26 (11.4) | 109 (11.0) | 1.10 (.73–1.66) |
Data are No. (%).
Abbreviations: CI, confidence interval; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HR, hazard ratio; REVEAL-HBV, Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer-Hepatitis B Virus study.
aHRs and 95% CIs were estimated from Cox proportional hazards regression models with attained age as the time scale and were adjusted for sex, principal components, and baseline HBV DNA load (in categories, 10 000–99 999, 100 000–999 999, ≥1 000 000 copies/mL). Censored at diagnosis of liver cirrhosis, diagnosis of HCC, death, and end of follow-up (2004), whichever occurred first.
No Association Between Homozygosity at HLA-DQB1 and HCC Risk Among Individuals With Liver Cirrhosis
Finally, among liver cirrhotic patients in the REVEAL-HBV cohort, we did not observe an increased risk of HCC among liver cirrhotic patients who were homozygous at HLA class II loci (Table 5), indicating that zygosity at HLA does not contribute to the final transition step in the natural history of cirrhotic HCC among liver cirrhotic patients.
Table 5.
Association Between Zygosity at HLA Class II Loci and HCC Among Liver Cirrhotic Patients in the REVEAL-HBV Cohort in Taiwan
| HLA Zygosity | Cirrhotic HCC (n = 138) | Liver Cirrhosis (n = 201) | HR Adjusted for Age, Sex, and Principal Components (95% CI)a |
|---|---|---|---|
| Class II loci | |||
| Heterozygous at all loci | 69 (50.0) | 107 (53.2) | Ref. |
| Homozygous in at least 1 locus | 69 (50.0) | 94 (46.8) | 1.09 (.78–1.52) |
| Number of homozygous | |||
| 1 | 46 (33.3) | 70 (34.8) | 1.08 (.74–1.58) |
| 2 | 12 (8.7) | 14 (7.0) | 0.92 (.49–1.72) |
| 3 | 11 (8.0) | 10 (5.0) | 1.37 (.72–2.60) |
| P trend | .52 | ||
| HLA-DPB1 | |||
| Heterozygote | 87 (63.0) | 128 (63.7) | Ref. |
| Homozygote | 51 (37.0) | 73 (36.3) | 1.08 (.76–1.53) |
| HLA-DQB1 | |||
| Heterozygote | 106 (76.8) | 166 (82.6) | Ref. |
| Homozygote | 32 (23.2) | 35 (17.4) | 1.13 (.76–1.68) |
| HLA-DRB1 | |||
| Heterozygote | 118 (85.5) | 181 (90.0) | Ref. |
| Homozygote | 20 (14.5) | 20 (10.0) | 1.08 (.66–1.76) |
Data are No. (%).
Abbreviations: CI, confidence interval; HCC, hepatocellular carcinoma; HR, hazard ratio; REVEAL-HBV, Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer-Hepatitis B Virus study.
aHRs and 95% CIs were estimated from Cox proportional hazards regression models with time-in-study as time scale and were adjusted for sex, age (in categories, 30–39, 40–49, 50–59, 60–79 years), and principal components. Follow-up started from the date of diagnosis of liver cirrhosis and censored at diagnosis of HCC, death, and end of follow-up (2016), whichever occurred first.
DISCUSSION
This large-scale population-based investigation demonstrates that homozygosity at HLA class II loci, particularly HLA-DQB1, is associated with a higher risk of HBV-associated HCC among chronic HBV carriers. Our results support the hypothesis that HLA diversity is an important factor in immunosurveillance and associated with multiple key steps in the development of HCC, including persistence of HBV among those infected and control of viral replication among long-term carriers. Our study adds important new information to previous efforts that have focused on the identification of specific SNPs or alleles within the HLA region associated with HCC risk and provides insights into how overall HLA diversity might contribute to cancer development.
MHC class I and II genetic diversity confers a selective advantage against multiple-strain infections in mice [18]. It is expected that heterozygotes are able to recognize and bind a more diverse set of peptides than homozygotes, leading to more T-cell clonal expansion and more efficient and specific cytotoxic T lymphocyte responses against infections [19–21]. In the case of HBV, a report that examined the number of HBV envelope and nucleocapsid core derived proteins presented to HLA class II molecules found a substantially greater breadth of response in heterozygous compared to homozygous individuals [19]. The less consistent association between HLA class I loci and several key steps in the progression from initial infection to end-stage liver diseases further suggest that CD4+ T-cell recognition of antigens (binding with HLA class II loci) is essential to controlling HBV infection [22]. This evidence is in line with several studies showing that HLA alleles or SNPs within the HLA class II region are associated with persistent infection resulting in chronic HBV infection, treatment response, and HCC progression [23, 24].
Our results demonstrate that the association between homozygosity at HLA class II loci and HCC is largely explained by viral control. However, effects driven by mechanisms other than viral control cannot be ruled out because of the modest power and nonsignificant increased risk of liver cirrhosis was observed for homozygosity at HLA-DQB1. This is a particularly important question to follow up in the future. Should the association between HLA-DQB1 zygosity and liver cirrhosis among HBV chronic carriers be replicated, it would suggest that mechanisms other than those related to viral control (eg, recognition of de novo tumor mutations) might be important.
Strengths of this study include its large sample size and setting in a region that is ethnically homogeneous. Multiple outcomes in the progression of HCC were observed in 1 cohort conducted in 1 population, which allays potential concerns about confounding due to population substructure. Rich epidemiological data and virologic factors were collected in samples that were collected prior to the development of HCC in anti-HBV treatment-naive subjects. However, there are some limitations to this study. First, information on HLA alleles was based on imputation. Therefore, residual confounding due to misclassified HLA alleles cannot be ruled out. Given high concordance of zygosity call between genotype and imputed data has been observed for all 6 HLA loci evaluated (>95% for 5 HLA loci), however, the impact of residual misclassification is likely to be small, if any. Second, among homozygous individuals in our study, the nonsignificant increased risk of liver cirrhosis could potentially be due to incomplete ascertainment of cirrhosis given that the diagnosis was primarily based on ultrasound results. In addition, we lacked information on the diagnosis of liver cirrhosis after 2004; some controls might have been diagnosed with liver cirrhosis after 2004. Both of these sources of misclassification of cirrhosis status might have led to an attenuation of the observed effect in our study. Finally, our results in individuals of Asian ancestry require replication in other racial and ethnic groups.
In summary, our study is the first to report evidence linking HLA zygosity with HBV-related liver cancer. Our findings offer new evidence in support of an important role for diversity in proteins involved in immunosurveillance in the development of HBV-related HCC. Whether HLA zygosity is also important for other infection-related cancers and solid tumors not caused by infection merits further investigation.
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Acknowledgments. We thank the Taiwan Liver Cancer Network and Taiwan Biobank for providing the DNA samples and clinical data. We also thank Dr Anna E. Coghill, National Institutes of Health and Division of Cancer Epidemiology and Genetics, and National Cancer Institute Fellows Editorial Boards for providing editorial comments.
Author contributions. H. I. Y. is the guarantor of the article. Z. L., C. J. C., A. H., and H. I. Y. designed and oversaw the study. H. I. Y. supervised genome-wide association studies. M. C. supervised HLA typing. M. V., X. G., and Y. Y. contributed to result interpretations of HLA imputation. Z. L., C. J. H., Y. H. H., R. M. P., and H. I. Y. contributed to data analysis. H. I. Y. and C. J. C. contributed to sample collection and sample preparation. M. H. P., M. H. L., and K. J. Y. provided important material support. The manuscript was drafted by Z. L. and H. I. Y. All authors critically reviewed the article and approved the final manuscript.
Disclaimer . The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. The study sponsors were not involved in the study design and collection, analysis, or interpretation of data, the writing of the article, or the decision to submit it for publication.
Financial support. This work was supported by the Taiwan Academia Sinica and Ministry of Science and Technology (grant number 101-2314-B-001-005-MY3); Bristol-Myers Squibb Company (assay of serum HBV DNA); Roche Diagnostics International (assay of serum hepatitis B surface antigen); National Institutes of Health, Frederick National Laboratory, Center for Cancer Research (contract number HHSN261200800001E); and National Cancer Institute Intramural Research Program.
Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
Contributor Information
Zhiwei Liu, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, USA.
Chih-Jen Huang, Genomics Research Center, Academia Sinica, Taipei, Taiwan.
Yu-Han Huang, Institute of Clinical Medicine, National Yang-Ming University, Taipei, Taiwan.
Mei-Hung Pan, Genomics Research Center, Academia Sinica, Taipei, Taiwan.
Mei-Hsuan Lee, Institute of Clinical Medicine, National Yang-Ming University, Taipei, Taiwan.
Kelly J Yu, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, USA.
Ruth M Pfeiffer, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, USA.
Mathias Viard, Basic Science Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Yuko Yuki, Basic Science Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Xiaojiang Gao, Laboratory of Integrative Cancer Immunology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA.
Mary Carrington, Basic Science Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA; Ragon Institute of Massachusetts General Hospital, Massachusetts Institute of Technology, and Harvard, Cambridge, Massachusetts, USA.
Chien-Jen Chen, Genomics Research Center, Academia Sinica, Taipei, Taiwan.
Allan Hildesheim, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, USA.
Hwai-I Yang, Genomics Research Center, Academia Sinica, Taipei, Taiwan; Institute of Clinical Medicine, National Yang-Ming University, Taipei, Taiwan; Graduate Institute of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan.
References
- 1. Doherty PC, Zinkernagel RM. Enhanced immunological surveillance in mice heterozygous at the H-2 gene complex. Nature 1975; 256:50–2. [DOI] [PubMed] [Google Scholar]
- 2. Thursz MR, Thomas HC, Greenwood BM, Hill AV. Heterozygote advantage for HLA class-II type in hepatitis B virus infection. Nat Genet 1997; 17:11–2. [DOI] [PubMed] [Google Scholar]
- 3. Carrington M, Nelson GW, Martin MP, et al. HLA and HIV-1: heterozygote advantage and B*35-Cw*04 disadvantage. Science 1999; 283:1748–52. [DOI] [PubMed] [Google Scholar]
- 4. Hraber P, Kuiken C, Yusim K. Evidence for human leukocyte antigen heterozygote advantage against hepatitis C virus infection. Hepatology 2007; 46:1713–21. [DOI] [PubMed] [Google Scholar]
- 5. Arora J, Pierini F, McLaren PJ, Carrington M, Fellay J, Lenz TL. HLA heterozygote advantage against HIV-1 is driven by quantitative and qualitative differences in HLA allele-specific peptide presentation. Mol Biol Evol 2020; 37:639–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Wang SS, Abdou AM, Morton LM, et al. Human leukocyte antigen class I and II alleles in non-Hodgkin lymphoma etiology. Blood 2010; 115:4820–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Shah N, Decker WK, Lapushin R, et al. HLA homozygosity and haplotype bias among patients with chronic lymphocytic leukemia: implications for disease control by physiological immune surveillance. Leukemia 2011; 25:1036–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Wang SS, Carrington M, Berndt SI, et al. HLA class I and II diversity contributes to the etiologic heterogeneity of non-Hodgkin lymphoma subtypes. Cancer Res 2018; 78:4086–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Chen CJ, Yang HI, Su J, et al. ; REVEAL-HBV Study Group. Risk of hepatocellular carcinoma across a biological gradient of serum hepatitis B virus DNA level. JAMA 2006; 295:65–73. [DOI] [PubMed] [Google Scholar]
- 10. Lin JC, Fan CT, Liao CC, Chen YS. Taiwan biobank: making cross-database convergence possible in the big data era. Gigascience 2018; 7:1–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Chang IC, Huang SF, Chen PJ, et al. The hepatitis viral status in patients with hepatocellular carcinoma: a study of 3843 patients from Taiwan liver cancer network. Medicine (Baltimore) 2016; 95:e3284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Zheng X, Shen J, Cox C, et al. HIBAG–HLA genotype imputation with attribute bagging. Pharmacogenomics J 2014; 14:192–200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Chang MH, Chen CJ, Lai MS, et al. Universal hepatitis B vaccination in Taiwan and the incidence of hepatocellular carcinoma in children. Taiwan childhood hepatoma study group. N Engl J Med 1997; 336:1855–9. [DOI] [PubMed] [Google Scholar]
- 14. Terrault NA, Lok ASF, McMahon BJ, et al. Update on prevention, diagnosis, and treatment of chronic hepatitis B: AASLD 2018 hepatitis B guidance. Hepatology 2018; 67:1560–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Yang HI, Sherman M, Su J, et al. Nomograms for risk of hepatocellular carcinoma in patients with chronic hepatitis B virus infection. J Clin Oncol 2010; 28: 2437–44. [DOI] [PubMed] [Google Scholar]
- 16. Yang HI, Yuen MF, Chan HL, et al. ; REACH-B Working Group. Risk estimation for hepatocellular carcinoma in chronic hepatitis B (REACH-B): development and validation of a predictive score. Lancet Oncol 2011; 12: 568–74. [DOI] [PubMed] [Google Scholar]
- 17. Iloeje UH, Yang HI, Su J, et al. Predicting cirrhosis risk based on the level of circulating hepatitis B viral load. Gastroenterology 2006; 130:678–86. [DOI] [PubMed] [Google Scholar]
- 18. Penn DJ, Damjanovich K, Potts WK. MHC heterozygosity confers a selective advantage against multiple-strain infections. Proc Natl Acad Sci U S A 2002; 99:11260–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Godkin A, Davenport M, Hill AV. Molecular analysis of HLA class II associations with hepatitis B virus clearance and vaccine nonresponsiveness. Hepatology 2005; 41:1383–90. [DOI] [PubMed] [Google Scholar]
- 20. Chowell D, Krishna C, Pierini F, et al. Evolutionary divergence of HLA class I genotype impacts efficacy of cancer immunotherapy. Nat Med 2019; 25:1715–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Pierini F, Lenz TL. Divergent allele advantage at human MHC genes: signatures of past and ongoing selection. Mol Biol Evol 2018; 35:2145–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Clifford GM, Rickenbach M, Polesel J, et al. ; Swiss HIV Cohort. Influence of HIV-related immunodeficiency on the risk of hepatocellular carcinoma. AIDS 2008; 22:2135–41. [DOI] [PubMed] [Google Scholar]
- 23. O’Brien TR, Yang HI, Groover S, Jeng WJ. Genetic factors that affect spontaneous clearance of hepatitis C or B virus, response to treatment, and disease progression. Gastroenterology 2019; 156:400–17. [DOI] [PubMed] [Google Scholar]
- 24. Thio CL, Carrington M, Marti D, et al. Class II HLA alleles and hepatitis B virus persistence in African Americans. J Infect Dis 1999; 179:1004–6. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.