Article Text

Download PDFPDF
Hepatitis B viraemia: its heritability and association with common genetic variation in the interferon γ signalling pathway
  1. Hsuan-Hao Huang1,
  2. Wei-Liang Shih1,
  3. Yi-Hsiu Li1,
  4. Chih-Feng Wu1,
  5. Pei-Jer Chen2,
  6. Chih-Lin Lin3,
  7. Chun-Jen Liu2,
  8. Yun-Fan Liaw4,
  9. Shi-Ming Lin4,
  10. Shou-Dong Lee5,
  11. Ming-Whei Yu1
  1. 1Graduate Institute of Epidemiology, College of Public Health, National Taiwan University, Taipei, Taiwan
  2. 2Division of Gastroenterology, Department of Internal Medicine, National Taiwan University Hospital and National Taiwan University College of Medicine, Taipei, Taiwan
  3. 3Department of Gastroenterology, Ren-Ai Branch, Taipei City Hospital, Taipei, Taiwan
  4. 4Liver Research Unit, Chang Gung University, Taoyuan, Taiwan
  5. 5Division of Gastroenterology, Department of Medicine, Taipei Veterans General Hospital, Taipei, Taiwan
  1. Correspondence to Dr Ming-Whei Yu, Graduate Institute of Epidemiology, College of Public Health, National Taiwan University, Room 522 No.17, Xuzhou Road Zhongzheng District, Taipei City 10055, Taiwan; yumw{at}


Objective High viraemia of hepatitis B virus (HBV) influences all phases in the development of hepatocellular carcinoma (HCC). This study was designed to estimate the overall contribution of host genetics to HBV viraemia, and investigate the influence of common single-nucleotide polymorphisms (SNPs) in the interferon γ (IFNγ) signalling pathway, which is pivotal in the non-cytolytic clearance of HBV.

Methods We first determined familial correlations and heritability (ie, proportion of phenotypic variation that is attributable to additive genetic factors) for HBV viraemia using 280 HCC families, including 766 adult HBV carriers. Then family-based association analysis was conducted for viraemia with a panel of 40 SNPs across ten IFNγ-related genes. For replication, seven tagging SNPs in identified candidate regions were also tested in a further 1011 unrelated individuals with longitudinal data on HBV viraemia over 16 years.

Results After adjustment for HBV genotype and sex, significant correlations for viraemia were detected among both siblings and mother–child pairs. Heritability accounted for approximately 30% (p<0.0002) of the variance of viral load, whereas HBV genotype and sex together explained less than 3%. Heritability estimates increased up to 74.0% after further exclusion of subjects with episodes of liver biochemical abnormalities. Our initial family-based association analysis identified two SNPs (rs2284553 (intronic SNP) and rs9808753 (Q64R)) on the IFNγ receptor 2 (IFNGR2) gene that were robustly associated with viraemia after multitest correction (all p<0.02). The SNPs were also associated with the longitudinal levels of viraemia and the persistence of a high viraemia of ≥4.39 log copies/ml (all p<0.0001) in unrelated individuals.

Conclusions HBV viraemia appears to have substantial heritability. Polymorphisms in the IFNGR2 gene appear to be associated with the variability of viraemia.

  • Genotype
  • hepatitis B
  • hepatocellular carcinoma
  • single nucleotide polymorphism
  • viral load
  • cancer epidemiology
  • cancer prevention

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Significance of this study

What is already known about this subject?

  • High viraemia of hepatitis B virus (HBV) influences all phases in hepatocellular carcinoma (HCC) development.

  • The mean steady-state viraemia level of HBV is highly variable between chronic HBV carriers.

  • While certain viral factors of HBV have been associated with HBV replication, the impact of host genetics on the levels of viraemia remains unknown.

What are the new findings?

  • There are significant correlations for HBV viraemia among both siblings and mother–child pairs in HCC families after adjustment for HBV genotype and sex.

  • Heritability accounted for around 30% of the total variability in HBV viral load, whereas HBV genotype and sex together explained less than 3%.

  • Common genetic variants in the interferon γ receptor 2 gene are associated with both the levels of HBV viraemia and the likelihood of persistence for high viraemia.

How might it impact on clinical practice in the foreseeable future?

  • Our findings imply a role of HBV viraemia in familial aggregation of HCC and may have importance in public health or clinical applications.

  • The identified genetic variants associated with control of HBV viraemia could potentially be used in combination with clinical variables to optimise clinical practice in the treatment of hepatitis B.


Chronic infection with hepatitis B virus (HBV) is a main cause of end-stage liver disease, accounting for 30% of cirrhosis and 53% of hepatocellular carcinoma (HCC) worldwide.1 The rates of progression to cirrhosis and HCC vary widely between carriers of HBV, depending on the interplay between viral and host factors.2–4

HBV carriers show remarkable variation in viral load as measured by blood HBV-DNA levels and duration of persistence for high viraemia.4 5 High viraemia as a risk factor for both incidence and mortality of HCC among HBV carriers has been demonstrated by several large prospective studies.4–8 Its role in HCC development was further strengthened by results from our recent longitudinal study, which indicated that prolonged maintenance of high viraemia was more important than transient increases of viraemia in predicting risk of HCC.4 5 Viral factors of HBV, including the X protein and HBV drug resistant mutations, have been associated with HBV replication by in vitro and in vivo studies.9–12 However, the influences of host genetic factors on the levels of viraemia is unclear.

There is a substantial amount of data, predominantly experimental, showing that both innate and adaptive immunity contribute to the control of HBV replication.13 Interferon γ (IFNγ), produced by activated cytotoxic T lymphocytes (CTLs) and natural killer cells, is pivotal in immunity to viral infection, and has been proposed as the main mediator in the non-cytolytic inhibition of HBV replication.13–18 IFNγ binds to a pair of receptor subunits, the IFNγ receptor (IFNGR)-1 and IFNGR-2, resulting in activation of the JAK-STAT signalling pathway.19 20 In the IFNγ signalling pathway, highly penetrant mutations or rare variants in the IFNGR and STAT1 gene have been implicated in the susceptibility to mycobacterial diseases.21–24 A recent report showed a possible role for a functional polymorphism within the IFNγ gene in viral clearance and treatment response of hepatitis C virus (HCV).25 No such studies have been conducted for hepatitis B.

In this study, we first used a collection of families with HCC to measure familial correlations of HBV viraemia and also the narrow-sense heritability, which estimates the overall contribution of host additive genetic factors to the inter-individual variation in the viraemia. Next, we focused on a wide range of genes related to IFNγ signalling, with the aim of identifying common single-nucleotide polymorphisms (SNPs) involved in the host control of HBV viraemia and related conditions. According to our preliminary results, HBV viraemia has a substantial heritability in biological relatives from the HCC families. We thus conducted a first-stage genetic association study in such families. Tagging-SNPs in the implicated regions were also evaluated using an existing longitudinal study with unique data on the long-term dynamics of viraemia in carriers of the HBV surface antigen (HBsAg).4 5

Subjects and methods


Two independent samples were collected for this study.

Family study

Families used in this study were ascertained via a comprehensive recruitment of HCC patients in whom at least 70% can be attributable to hepatitis B in Taiwan.26 Since 1997, these patients have been consecutively recruited from three major teaching hospitals.27 In Taiwan where perinatal transmission is the predominant route of HBV transmission, HBV carriers in the immune tolerance phase are usually young (before age 20) and have high HBV-DNA levels (usually >108 copies/ml). For most HBV carriers, the immune-clearance phase occurs at ages of 25–35 years.2 Thus relatives were recruited through the index cases if they were aged 25–75 years. Participating relatives were interviewed in person by a structured questionnaire (including items on history of antiviral therapy and chronic disease, demographic characteristics, and lifestyle habits), and underwent an abdominal ultrasonography, liver biochemical tests (including alanine aminotransferase (ALT)), serological tests for HBsAg and antibodies to HCV (anti-HCV), and biometric measurements. Participants with abnormal liver biochemical tests or ultrasonographic features were informed in a report of results and advised to receive further clinical evaluation or treatment. Families with at least two biological relatives who were positive for HBsAg and negative for anti-HCV were eligible for inclusion in the analysis. Subjects who were on antiviral therapy were excluded. In total, 280 different families containing 904 individuals were available. Because about one third of HBV carriers with unresectable HCC may develop reactivation of HBV characterised by a ≥10-fold rise in circulating levels of HBV-DNA during chemotherapy, and high pre-chemotherapy HBV viral load is associated with poor survival,28 29 to eliminate the influence of treatment and/or selective survival due to tumour status at enrollment on HBV viraemia, we further excluded 138 HCC probands with tumour size >5 cm at entry, thus leaving a total of 160 HCC patients (142 of whom were probands) and 606 unaffected individuals for analysis. Among these families, 256 were sibships or nuclear families and the remaining 24 were more complex.

Longitudinal study of HBV viraemia

Subjects were chosen from a cohort study of 2903 unrelated, originally healthy, male HBsAg-positive government employees,30 who were recruited in 1989–1992 and followed by medical examinations (including blood testing and ultrasonography) approximately every year and by computerised data linkage with the profiles on the national registry of cancer and death in Taiwan. In 2004, following the case–cohort sampling design, we selected a random subcohort (n=1084; including 53 HCC cases) from all cohort members meeting the following criteria: (1) anti-HCV negative, (2) no prior history of antiviral therapy, and (3) at least two blood samples collected after study entry. All cases (n=59) not already included who developed HCC by 2005 and met the above criteria were added to complete the case–cohort population (n=1143).5 The present study was performed on a total of 1011 HBsAg carriers (115 of whom were ascertained with incident HCC after extended follow-up by 1 year) with white blood cell DNA and detectable HBV DNA in plasma at baseline, which corresponds to 88.5% of the 1143 subjects in the case–cohort study. Baseline and all follow-up plasma samples from subjects were used for measurement of HBV viraemia. The 1011 subjects had contributed 7574 plasma samples for the analysis of longitudinal series of HBV viral load assayed every about 1.5 years.

Laboratory analysis

SNP selection and genotyping

SNPs were identified from the National Center for Biotechnology Information dbSNP database, and were genotyped using the ABI 7900HT TaqMan SNP genotyping system (Applied Biosystems, Foster City, California, USA). The power of the transmission disequilibrium test (TDT) is influenced by both the marker- and trait-allele frequencies. Common trait alleles generally offer the most power.31 Given our sample size, we aimed to identify common variants. We initially screened 27 SNPs with a minor-allele frequency (MAF) greater than ∼0.2 in the intragenic sequences of the IFNG, IFNGR1, IFNGR2, JAK1, JAK2, STAT1, IRF1, IRF2, IRF8 and IRF9 genes for association with HBV viraemia. Two SNPs (rs17127114 and rs793801) were then eliminated because of showing deviation (p<0.05) from Hardy–Weinberg equilibrium. When associations were identified for SNPs in IFNGR2 and STAT1, tagging-SNPs were selected within and 5-kb flanking each gene using the program tagger with a cut-off of 0.8 for r2 and a MAF >0.1.32 An additional 15 SNPs in the IFNGR2 and STAT1 regions were chosen to achieve effective coverage (83.0% and 78.0% for the IFNGR2 and STAT1 regions, respectively) of the common variants (MAF>0.1) observed in the HapMap data of Han Chinese ( All SNPs included for analysis had genotype call rates >93%.

Assay of viral factors

Serum HBsAg, HBeAg and anti-HCV were assayed using commercially available immunoassay (Abbott Laboratories, Chicago, Illinois, USA; Roche Diagnostics, Indianapolis, Indiana, USA; General Biologicals Corporation, Hsin-Chu, Taiwan). Plasma HBV-DNA levels and HBV genotype were determined in-house by PCR-based methods as described previously.5 6 The PCR assay for HBV-DNA showed a broad linear range of quantification (102–1010 copies/ml), with a detection limit of 215–250 copies/ml. The within-run and between-run coefficients of variance were below 9% and 15%, respectively.

Statistical analysis

The values of HBV-DNA levels were log10 transformed to approximate normality. Hardy–Weinberg equilibrium was evaluated using the PowerMarker software ( We used PedCheck v1.0 to detect Mendelian inconsistencies33 and Haploview v4.134 to calculate linkage disequilibrium (LD) (r2). The familial clustering of viraemia was evaluated in two ways: in method 1 familial correlations were calculated using the Pearson's correlation coefficients (for mother–child pairs) or the intraclass correlation coefficients (ICCs) with a random effect model (eg, for sibships)35 as appropriate; in method 2 narrow sense heritability that estimates the additive genetic effects due to allelic variation was calculated by using a variance–component approach for assessing the polygenic function with the SOLAR software.36 Family-based associations for SNPs with viraemia were analysed by PBAT v3.5, and by a second method, the quantitative TDT (QTDT v.2.6.1).37 38 These analyses were carried out using sex and HBV genotype and/or serum ALT as covariates. To account for multiple testing, false discovery rate (FDR) q values were calculated with the ranked p values from QTDT. No conventional q value cut-off has been established to categorise a discovery as significant in candidate gene association studies. We chose a FDR q-value cut-off of 0.2, as has been used previously.39 In the family sample, a generalised estimating equations approach was used to provide estimates of the effect size for each SNP. In the longitudinal study with unrelated individuals, unconditional logistic regression modelling was used to evaluate the association between genotype and cross-sectional measure of phenotype. We assessed the influence of SNPs on longitudinal viraemia by using random intercept models with the PROC MIXED and PROC GLIMMIX procedures in SAS v.9 software (SAS Institute). Time, age and HBV genotype were considered as covariates in the model because they are determinants of viral load.4–6 8 In analyses that assessed the associations between SNPs and HBV genotype, we used viral load as a covariate to differentiate that the finding is not secondary to the association between SNPs and viraemia.


As shown in table 1, subjects with HCC appear to have higher HBV viraemia and were more likely to carry HBV genotype C than same-sex non-HCC relatives in the family sample. Among non-HCC relatives of the family sample, men demonstrated increased viral load compared with women (p=0.0195). The non-HCC male relatives also showed higher viraemia than those who remained HCC free throughout follow-up from the longitudinal study (p<0.0001). Additionally, the proportions of carriers of HBV genotype C and elevated ALT were much higher in the former group than in the latter group (all p<0.0001).

Table 1

Characteristics of study subjects

Familial resemblance of viraemia

Of the 280 families, 69 (24.6%) were families of HBV genotype C, 168 (60%) were families of HBV genotype B, and 43 (15.4%) were families with discordant HBV genotype between relatives. Table 2 shows selected characteristics of various probands' relatives. HBV viraemia was higher in relatives with elevated ALT than in those with normal ALT across various kinships, as found among HBV carriers of general Taiwanese population.5

Table 2

Selected characteristics of hepatocellular carcinoma proband patients' relatives by serum alanine aminotransferase (ALT) levels

After adjustment for sex and HBV genotype, there were significant positive correlations among both siblings (ICC=0.22; 95% CI, 0.06 to 0.36) and mother–child pairs (Pearson's r=0.24; 95% CI, 0.08 to 0.39). Stronger evidence of familial clustering of the viraemia was observed when subjects with episodes of ALT elevation were excluded from calculations (siblings: ICC=0.50 (95% CI, 0.34 to 0.63); mother–child pairs: Pearson's r=0.40 (95% CI, 0.18 to 0.58)). The additive genetic variance component for HBV viraemia was significantly different from zero (all p<0.0002), with an estimated heritability ranging from 26.6% to 74.0% in different conditions. As expected, serum ALT was the strongest covariate, explaining 15.8% of the total variation in viraemia. HBV genotype and sex were also significantly correlated with viraemia in the variance–component analysis (p<0.01). However, they together explained only less than 3% of the total variation. The viral load was not associated with age (figure 1).

Figure 1

Familial correlations and heritability for HBV viraemia. (A, B) Intraclass correlation coefficients (for all kinships and sibships) or Pearson's correlation coefficients (for mother–child pairs) and 95% CIs (error bars). In calculation of correlations, adjustment was performed by regressing the value of viral load on sex and HBV genotype and taking the residual as the working variable in analysis of the correlation between subjects within a pedigree. n: number of pedigrees (all kinships), mother–child pairs, or sibships. (C, D) Heritability estimates (h2) and SE of mean. *p <0.05; **p<0.01; ***p<0.0001.

Association studies of IFNγ-related SNPs and viraemia

Family-based association analysis

We began by analysing the family data with family-based association methods, which are less sensitive to population stratification.40 In an analysis of the first 25 SNPs across ten genes, two SNPs (rs9808753 in IFNGR2 and rs11894425 in STAT1) showed associations with HBV viraemia in both QTDT (p=0.0113 and 0.007, respectively) and PBAT (p=0.0356 and 0.0083, respectively). When three additional tagging SNPs in IFNGR2 were also typed, two (rs2284553, r2=0.51; rs2268241, r2=0.96) of which in LD with rs9808753 achieved statistical significance. We further analysed 12 tagging-SNPs in STAT1 and found an additional SNP (rs2280235), which was in LD (r2=0.79) with the rs11894425 SNP, approached significance (p=0.0549 in QTDT and p=0.0484 in PBAT). After adjusting for multiple testing, four SNPs (rs2284553, rs2268241, and rs9808753 in IFNGR2; rs11894425 in STAT1) were associated with variation in viral load with FDR q between 0.02 and 0.14 (table 3 and figure 2). We also re-analysed the family data after exclusion of proband patients. A total of 203 families remained for analysis, and similar results were obtained (eg, from QTDT: rs2284553, p=0.0016; rs2268241, p=0.0157; rs9808753, p=0.0155; rs11894425, p=0.0414).

Table 3

p values for single-nucleotide polymorphisms (SNPs) in candidate genes of the interferon γ (IFNγ) signalling pathway from family-based association analysis

Figure 2

Linkage disequilibrium block structure in IFNGR2 and STAT1. (A, C) Structure of the gene of interest. Exons are depicted by black boxes, and untranslated regions by grey boxes; the arrow indicates the direction of transcription. (B, D) Linkage disequilibrium (r2) among SNPs in HapMap Han Chinese samples (HapMap Public Release 24/phase II), derived from the Haploview software. SNPs typed in the family sample are indicated by black rectangles. SNPs significantly associated with HBV viraemia in family-based association analysis are indicated with an asterisk (*).

Cross-sectional analysis in unrelated individuals

In addition to the family-based study, we performed association analysis with a second sample of 1011 unrelated individuals from the general Taiwanese population in a longitudinal viral-load study.4 5 First, we tested associations with HBV viraemia determined at baseline for seven tagging SNPs that fell into two LD blocks containing significant SNPs in the family sample (figure 2). STAT1 SNP rs6718902, which was in LD (r2=0.66) with the most significant SNP rs11894425 in the family sample, exhibited significant association, irrespective of viral load treated as a continuous or a binary outcome according to a cut-off of 4.39 log copies/ml, the threshold associated with increased risk of HCC previously described.5 6 In IFNGR2, when treating viral load as a continuous variable, no SNPs revealed associations. However, SNPs rs9808753 (GG vs AA, p=0.0344) and rs2284553 (AA vs GG, p=0.0147), which have shown association with viraemia in the family sample, were both associated with a high viral load of ≥4.39 log copies/ml. Furthermore, the results of this analysis closely mimicked that of the family sample in regard to the genetic effect size (figure 3).

Figure 3

Associations of selected SNPs in IFNGR2 and STAT1 with HBV viraemia, by study design. In the cross-sectional analysis of the 1011 unrelated individuals, HBV viraemia determined at baseline was used for analysis. (A) Estimated differences (β estimates; regression coefficients) in plasma viral load (log copies/ml) between genotypes of SNPs. (B) Estimated ORs and 95% CIs for having a high viral load defined as ≥4.39 log copies/ml. IFNGR2 SNPs rs9808753 and rs2284553 influence viraemia both at baseline and longitudinally in unrelated individuals. Analyses were adjusted for age, sex (family study only), HBV genotype, and the time of assay (longitudinal study only). Significant associations are shown in bold.

Longitudinal analysis in unrelated individuals

In the longitudinal analysis with unrelated individuals, the two IFNGR2 SNPs, rs9808753 and rs2284553, showed similar effects for cross-sectional and longitudinal viraemia measures, and were both significantly associated with longitudinal viraemia measures (rs9808753 GG vs AA, p=0.0124; rs2284553 AA vs GG, p=0.0095) and the likelihood of maintenance of a high viral load of ≥4.39 log copies/ml (rs9808753 GG vs AA, p<0.0001; rs2284553 AA vs GG, p<0.0001) over time. Only 10% of HBV carrier subjects remained HBeAg positive. Because positivity for HBeAg is a marker of higher viral load (>105 copes/ml),2–6 we next re-analysed the data after exclusion of HBeAg-positive subjects. As expected, the results were similar to those obtained using the entire sample (eg, for longitudinal viraemia measure: rs9808753 GG vs AA, p=0.0116; rs2284553 AA vs GG, p=0.0213; for maintenance of viral load ≥4.39 log copies/ml: rs9808753 GG vs AA, p<0.0001; rs2284553 AA vs GG, p<0.0001). Of the four genotyped STAT1 SNPs in LD block 1, however, only rs6718902 showed significant association (CT vs CC: p=0.0285 and 0.0166, respectively, for β coefficient and OR of ≥4.39 log copies/ml) (figure 3). In re-analysis of the data with different cut-points (4 or 5 log copies/ml) to define high viraemia, the effect estimates for SNPs rs9808753, rs2284553, and rs6718902 remained significant, and STAT1 rs11894425 also became significant when using 5 log copies/ml as the cut-point to define high viraemia (data not shown).

IFNγ-related SNPs and other clinical features

None of the selected SNPs demonstrated convincing evidence of association with elevated ALT, HCC or liver cirrhosis detected by ultrasonography (data not shown). HBV genotype C is known to be associated with HCC, and was correlated with both cross-sectional and longitudinal viraemia measures.4–6 8 We thus also investigated association with HBV genotype for seven tagging SNPs available for both family and longitudinal sample. Four STAT1 SNPs (rs11894425, rs2280235, rs6718902 and rs4853533), which belonged to LD block 1, were associated with HBV genotype in at least one of the two samples even after adjustment for viral load. Additionally, the SNP rs2280235 exhibited significant concordant association with HBV genotype in both the family and the longitudinal study sample (GG vs AA: p=0.0382 and 0.0342, respectively) (table 4).

Table 4

Relationships between single nucleotide polymorphisms (SNPs) within the LD Block 1 of the STAT1 gene and infection with genotype C HBV*


HBV is a prevalent infectious agent in humans, and it causes a spectrum of liver disease, from simple fatty liver at the most benign end to chronic hepatitis, cirrhosis and HCC at the opposite end.1–8 27 30 HBV replication persists throughout the natural course of HBV infection, and high viraemia contributes to all phases in HCC development.2–8 Through epidemiological findings, evidence has been accumulated in favour of a significant but heterogeneous genetic component in HCC susceptibility.27 41–46 Until now, the elucidation of the genetic factors for HCC has been accomplished typically through studies focussing on the disorder as the phenotype of interest.41–46 To our knowledge, no work has been undertaken to investigate whether host genetic background influences inter-individual variation in the levels of HBV viraemia.

Here we showed the familial correlations among both siblings and mother–child pairs, as well as a significant heritability for HBV viraemia in HCC families after adjustment for significant covariates including HBV genotype and sex. Moreover, when subjects with episodes of ALT elevation, which often occurs immediately after an upsurge of viral load, were excluded, clustering of HBV viraemia shifted toward a high heritability. This indicates that the variability observed in the steady-state viral load has a strong familial effect. In Taiwan, chronic HBV infection is mainly through intra-familial transmission.2 Accordingly, a concordant HBV genotype among family members was found in most of the families analysed. As viral genotype is a determinant of HBV viraemia,4–6 8 it might be an important source of familial effect for viral load. Thus, this study attempted to elucidate the hypothesised genetic component in HBV viraemia after controlling for viral genotype.

We also estimated the contribution of HBV genotype to the familial clustering of HBV viraemia. Surprisingly variance component calculation indicates that only less than 3% of the variation in the viraemia was attributable to HBV genotype. There is thus rationale to better define the genetic contribution to HBV viraemia. This study evaluated whether genetic variation in SNPs of IFNγ-related genes were associated with HBV viraemia control. IFNγ plays a critical role in modulating immune system function, and experimental studies have indicated that the replication of HBV is strongly inhibited in response to IFNγ, which eliminates pregenomic RNA-containing capsids from hepatocytes.13–15 17 18 47 Besides, there is evidence showing that IFNγ is essential in terminating persistent viral infection.48

Using a family-based association study and a longitudinal viral-load study of an independent series of unrelated HBsAg carriers, we found compelling evidence that HBV viraemia is linked with two correlated IFNGR2 SNPs (rs9808753 and rs2284553), located within a 207.6-kb region containing genes encoding both subunits of the type I IFN receptor (IFNAR1 and IFNAR2) and a receptor chain for interleukin 10 (IL-10RB). Notably, SNPs in IFNAR2 and IL-10RB have been associated with persistent HBV infection in Gambian families.49 To demonstrate that the association of HBV viraemia with IFNGR2 is independent of the previously reported SNPs, we searched the HapMap database of Han Chinese to identify SNPs in LD with IFNGR2 SNPs. This analysis confirmed the independence of the IFNGR2 association we identified because none of the SNPs in the LD blocks of IFNAR1, IFNAR2 and IL-10RB showed an r2 >0.1 with rs9808753 or rs2284553. We evaluated any aspects of HBV viraemia, including both a quantitative trait analysis and stratified analyses according to a predetermined cut-off. The most robust result regarding IFNGR2 SNPs rs9808753 and rs2284553 appeared to be the consistently significant association with a high viral load of ≥4.39 log copies/ml, a threshold value of HBV viraemia associated with increased risk of HCC in our previous studies.5 6

More importantly, our longitudinal study facilitated the discovery of the effect on the long-term dynamics of HBV viraemia and the persistence for high viraemia. SNP rs9808753 is a missense SNP (Q64R) located in the IFNGR2 extracellular domain, where several rare missense and in-frame mutations have been associated with IFNGR2 deficiency,21 while another significant SNP rs2284553 (+846G>A) resides in intron 1 with no known function. Both SNPs demonstrated association with the cross-sectional and the longitudinal measures of HBV viraemia. Since only a small fraction of the study population remained HBeAg positive at recruitment, the identified common genetic variants may modulate the control of viraemia during and/or after the immune-clearance phase. Although we have not yet performed functional studies, this result indicates that variation in the IFNGR2 gene is a reliable predictor for long-term tracking of HBV viraemia.

We found weak evidence for association of the STAT1 gene with HBV viraemia, in which no overlap in the SNP associations across two study samples, although the associated SNPs in either sample are totally located in the LD block 1. Multiple reasons might explain the non-replication results, including sampling artifacts resulting from the difference in sample size and composition of study population. In the same LD block, we also found that four correlated SNPs were associated with infection of HBV genotype C independent of viraemia in either sample, and one of which (rs2280235) achieved statistical significance in both samples. Furthermore, the effect sizes for these four SNPs in both samples generally overlap each other. These results highlight a region of 7059 bp spanning portion of the tail segment and the trans-activator domain of the STAT1 gene.

The STAT1 gene is a transcription factor that is activated upon tyrosine phosphorylation in response to stimulation by IFNα/β and IFNγ.19 20 STAT1 is also involved in cellular responses to a newly described cytokine IFNλ1, which has been implicated in immunity to HBV.50 Partial STAT1 deficiency due to mutations in the trans-activator domain, which disrupt STAT1 phosphorylation, is associated with predisposition to various bacterial and viral infections.21 24 It has also been documented that viruses can evade IFN-stimulated antiviral responses by blocking STAT1 phosphorylation.51–53 Some of viral proteins exert their effects via binding directly to STAT1.54 Expression of HBV polymerase can antagonise IFN response by inhibiting STAT1 nuclear translocation.55 Although the detailed mechanisms underlying the observed association between STAT1 alleles and infection of HBV genotype C remain unknown, this association suggests that certain HBV genotypes might act only in specific hosts. The interaction between virus and host gene in HBV control is more complex than was initially anticipated, emphasising the strength of conducting replication study with different samples in this study.

In conclusion, the degree of familial correlations highlights familial clustering of HBV viraemia among relatives in HCC families, which showed a higher viraemia and a much higher prevalence of liver biochemical abnormality than the HBsAg carriers selected among general Taiwanese population. The estimated heritability suggests that there is a substantial genetic component in control of HBV viraemia. These findings imply a role of HBV viraemia beside other factors in familial clustering of HCC found previously.27 Through family-based study and longitudinal viral-load study, we have provided evidence that inter-individual variation in HBV viral load can be attributable to variability in genes involved in IFNγ signalling. It has been widely accepted that CTL-induced intracellular inactivation of HBV is the central mechanism of viral clearance during HBV infection.14 Long-term persistence of viral DNA is associated with long-term activity of HBV specific CTL, which secrets IFNγ to exert antiviral function.13–18 The results regarding the IFNγ SNPs and HBV viraemia may be generalisable to other populations in which, unlike Taiwan, perinatal infection is infrequent. HBV viral load is a strong predictor of end-stage liver disease2–8 and has been shown to be an essential parameter for a decision to begin antiviral therapy and management of hepatitis B. Thus, our findings are useful in predicting outcome of chronic HBV infection and in optimising clinical practice in the treatment of hepatitis B.



  • Funding This study was supported by grants NSC 97-3112-B-002-028 and NSC 98-3112-B-002-021 from the National Science Council, Taiwan.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Research Ethics Committee at the College of Public Health, National Taiwan University.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • Digest
    Emad El-Omar Severine Vermeire Alexander Gerbes