Interaction of signal transducer and activator of transcription 3 polymorphisms with hepatitis B virus mutations in hepatocellular carcinoma

Authors


  • Potential conflict of interest: Nothing to report.

  • This study was supported by grants from the National Outstanding Young Scholar Fund (81025015), the Key Project (91129301), and the Creative Research Group (81221061) of the National Natural Scientific Foundation of China.

Abstract

Hepatitis B virus (HBV) mutations and signal transducer and activator of transcription 3 (STAT3) activation are closely associated with hepatocellular carcinoma (HCC). However, single nucleotide polymorphisms (SNPs) of STAT3 have not been implicated in HCC susceptibility. This study was designed to evaluate the effect of STAT3 SNPs and their interactions with HBV mutations on HCC risk. A total of 2,011 HBV-infected subjects (including 1,021 HCC patients) and 1,012 healthy controls were involved in this study. SNPs rs4796793 (−1697, C>G), rs2293152 (intron 11, C>G), and rs1053004 (3′ untranslated region, T>C) were genotyped using quantitative polymerase chain reaction. HBV mutations were determined via direct sequencing. It was found that rs2293152 (GG versus CC) was significantly associated with HCC risk compared with the subjects without HCC, adjusting for age and sex (adjusted odds ratio [AOR], 1.30; 95% confidence interval [CI], 1.04-1.62). The impact of rs2293152 was greater in women compared with men. Compared with HCC-free HBV-infected subjects, rs2293152 GG was solely associated with HCC in women (AOR, 2.04; 95% CI, 1.15-3.61). rs2293152 GG was significantly associated with high viral load (≥1 × 104 copies/mL) (AOR, 1.37; 95%, CI 1.01-1.88) and increased frequencies of T1674C/G (AOR, 1.61; 95% CI, 1.06-2.46) and A1762T/G1764A (AOR, 1.64; 95% CI, 1.14-2.35). In multivariate regression analyses, multiplicative interaction of rs1053004 with T1674C/G significantly increased HCC risk, whereas rs2293152 and A1726C interaction reduced it, adjusting for covariates including HBV mutations in the enhancer II/basal core promoter/precore region; the interaction of rs4796793 with preS2 start codon mutation significantly increased HCC risk, adjusting for covariates including HBV mutations in the preS region. Conclusion: STAT3 SNPs appear to predispose the host with HBV mutations to hepatocarcinogenesis, and this effect may differ in men versus women. STAT3 SNPs may have applicability in future HCC surveillance algorithms. (Hepatology 2013;57:2369–2377)

Liver cancer in males is the fifth most frequently diagnosed cancer and the second most frequent cause of cancer-related death; in females, it is the seventh most commonly diagnosed cancer and the sixth leading cause of cancer death worldwide.1 Hepatocellular carcinoma (HCC) accounts for 70% to 85% of the total liver cancer burden.1 The highest HCC incidence rates are found in East and Southeast Asia and in sub-Saharan Africa. The rate is three-fold higher in males than in females. Chronic infection with hepatitis B virus (HBV) or hepatitis C virus (HCV), aflatoxin B1 exposure, alcohol drinking, cigarette smoking, diabetes, and some genetic factors are risk factors of HCC.2 More than half of global HCC cases occur in China, a country with about 94 million people who are seropositive for hepatitis B surface antigen (HBsAg).1, 3

In East and Southeast Asia, where HBV genotypes B and C are endemic, genotype C, hepatitis B e antigen (HBeAg) expression, viral load (>1 × 104 copies/mL), viral mutations in the enhancer II/basal core promoter/precore (EnhII/BCP/PC) and the preS regions of HBV have been shown to be significantly associated with HCC.4-10 The HBV mutations are gradually accumulated during HBV-induced hepatocarcinogenesis.4, 9 The HCC-associated HBV mutations are probably generated via an evolutionary process in inflammatory microenvironment and in turn promote hepatocarcinogenesis.11, 12 Persistent inflammation is significantly associated with HBV-induced carcinogenesis and late recurrence of HCC.6, 13

The mechanisms by which hepatic inflammation drives HCC development include increased expression of proinflammatory transcription factors such as signal transducer and activator of transcription 3 (STAT3).14 STAT3, whose gene is located on chromosome 17, is a key molecule of the Janus kinase/STAT signaling pathway. Some cytokines and growth factors, including interleukin-6 and hepatocyte growth factor, can activate STAT3. STAT3 activation requires phosphorylation of a critical tyrosine residue (Tyr705), which mediates its dimerization, which in turn is a prerequisite for nucleus entry and DNA binding. Activation of STAT3, a major kinase-independent target of sorafenib, is a principal pathway implicated in promoting tumorigenesis.15, 16 HBV X protein (HBx) constitutively activates STAT3.17, 18 Moreover, HBx mutants significantly increase STAT3 activation compared with wild-type HBx.19 Activated STAT3 specifically binds HBV enhancer 1, a region containing an androgen-responsive element site, leading to an overall stimulation of HBV gene expression.20, 21STAT3 single nucleotide polymorphisms (SNPs), which might affect STAT3 expression and activation upon stimulation, have a substantial effect on genetic predisposition to inflammatory diseases and cancers.22-24 We therefore hypothesize that STAT3 SNPs might contribute to dysregulation of Janus kinase/STAT pathway and immune balance upon HBV infection and facilitate the generation of HBV mutations, thus contributing to HBV-induced carcinogenesis.

The roles of STAT3 SNPs and their interactions with HBV mutations on HBV-induced carcinogenesis have not been reported. In this study, we conducted a large epidemiological study to investigate the associations of STAT3 SNPs, HBV mutations, and their interactions with the risk of HCC. This study may be helpful in determining the HBV-infected subjects who are more likely to develop HCC and therefore need special interventions.

Abbreviations

ALT, alanine aminotransferase; AOR, adjusted odds ratio; ASC, asymptomatic HBsAg carrier; CHB, chronic hepatitis B; CI, confidence interval; EnhII/BCP/PC, the enhancer II/basal core promoter/precore; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HBx, HBV X protein; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HWE, Hardy-Weinberg equilibrium; MGB, Minor Groove Binder; PCR, polymerase chain reaction; SNP, single nucleotide polymorphism; STAT3, signal transducer and activator of transcription 3.

Patients and Methods

Study Population.

Healthy controls were those who received annual physical examinations at the 1st Affiliated Hospital, Second Military Medical University, Shanghai, China, from September 2009 to June 2010. The controls were free of HBV and/or HCV infection and had no history of liver disease. The asymptomatic HBsAg carriers (ASCs) were from our community-based HBV-infected cohort established in Shanghai and the health examination center at the 1st Affiliated Hospital. Patients with chronic hepatitis B (CHB), patients with liver cirrhosis, and patients with HCC were recruited from the affiliated hospitals of the Second Military Medical University, Shanghai, China; the 88th Hospital in Taian City, Shandong, China; and Southwest Hospital, Chongqing, China. Patients were newly diagnosed from October 2009 to September 2011. ASC status, CHB, cirrhosis, and HCC were diagnosed according to criteria that have been described.6 In total, 1,012 healthy controls and 2,011 HBV-infected subjects, including 1,021 HCC patients, were involved in this study. None of the study subjects had been included in any of our previous studies. The study protocol conformed to the 1975 Declaration of Helsinki and was approved by the ethics committee of the Second Military Medical University. All participants provided written informed consent.

Examination of Serological Viral Markers, Liver Function, and Viral Load.

The sera and genomic DNA of each subject were prepared and stored as described.25 Serological testing for HBV markers, α-fetoprotein, alanine aminotransferase (ALT), and viral load were performed as described.26 Antibodies to HCV and human immunodeficiency virus were examined in the hospitals from which the HBV-infected patients were recruited. Patients who were seropositive for either virus were not included. Antibody to hepatitis delta virus was examined using commercial kits (Wantai Bio-Pharm, Beijing, China), and the seropositive patients (about 1% in the HBV-infected patients with and without HCC) were also excluded.

HBV Genotyping, DNA Sequencing, and Mutation Analysis.

HBV was genotyped by multiplex polymerase chain reaction (PCR) and nested multiplex PCR as described.3, 27 The HBV EnhII/BCP/PC region and preS region were amplified using nested PCR as described.6, 7 The amplicons were directly sequenced using ABI PRISM BigDye sequencing kits and an ABI 3730 Genetic Analyzer (Applied Biosystems, Foster City, CA) in both forward and reverse directions. The HBV sequences were aligned and analyzed using MEGA 5.0 and Bioedit 7.0 software packages. Wild-type viral nucleotides were determined as described.6 A site with a frequency of mutations in combination >20%, either in genotype B or in genotype C from all HBV-infected subjects, was termed as a hotspot. HBV preS deletion was defined as reported.7

Selection of STAT3 SNPs and Genotyping.

We selected three representative STAT3 SNPs which had the minor allele frequency of >5% in Chinese Han population according to the International HapMap Project (www.hapmap.org). rs2293152 (C>G, in intron 11) was selected because it had been linked to inflammatory diseases.28-30 rs4796793 (C>G, in the promoter −1697) and rs1053004 (T>C, in the 3′ untranslated region) were selected because they were the representatives of two haplotype blocks as determined using online Haploview 4.2 software (http://hapmap.ncbi.nlm.nih.gov) (Supporting Fig. 1). The SNPs were genotyped using fluorescent-probe real-time quantitative PCR in a Light Cycler 480 (Roche, Basel, Switzerland). Three reduplicative samples were run with one template-free control. Primers and probes (Taqman or Minor Groove Binder) were designed and synthesized by GeneCore BioTechnologies (Shanghai, China). Supporting Table 1 shows the sequences of primers/probes and PCR program.

Statistical Analysis.

Differences in categorical variables were evaluated using chi-square test. Levels of HBV DNA and ALT with skewed distribution were adjusted to normal distribution by transformation into logarithmic function, and then compared by Student t test or analysis of variance. Hardy-Weinberg equilibrium (HWE) was examined online (http://ihg.gsf.de/ihg/snps.html). For the main effect of SNPs, unconditional logistic regression model was conducted to calculate odds ratios (ORs) and their 95% confidence intervals (CIs), adjusting for age and sex. Because HCC is more frequent in males than in females, sex may be a major confounder. We therefore stratified our study population into females and males and evaluated the associations of SNPs with HBV-related HCC (HBV-HCC) within each stratum. Contributions of SNPs and their multiplicative interactions with sex to HCC in all study subjects or in the HBV-infected subjects were assessed using multivariate regression analyses, adjusting for age. Phi coefficient was used to evaluate the possible correlations between the HBV mutations. Contributions of SNPs and their multiplicative interactions with the HBV mutations to HCC in those with HBV sequencing data were evaluated by multivariate regression analyses, adjusting for covariates including HBV DNA level and HBV mutations. The contributions of three SNPs and their interactions with sex or HBV mutations were calculated by adding each SNP and the interaction term to the same multivariate regression model. Due to low viral load or mutations in the primer binding sites, the HBV fragments were not successfully amplified from fractions of the HBsAg-positive subjects. The HBV mutations in the EnhII/BCP/PC region and those in the preS region were separately evaluated in the multivariate regression analyses. All statistical tests were two-sided and conducted with SPSS 16.0 for Windows (SPSS, Chicago, IL). P < 0.05 was considered statistically significant. The Bonferroni correction was employed to accommodate the comparison of HBV-HCC patients with multiple control groups, but does not correct for the multiple SNPs.

Results

Baseline Characteristics of the Study Participants.

Table 1 summarizes the characteristics of the HBV-infected patients and healthy controls. Healthy controls were 10 years older than the HBV-infected patients on average. There was no significant difference in sex distribution between healthy controls and the HBV-infected patients (P = 0.490). The proportion of males in the HCC patients was significantly higher than that in healthy controls and the HBV-infected patients without HCC. The HCC patients were older than HBV-infected patients without HCC. HBV genotype C and HBeAg negativity were more frequent in the HCC patients than in the HBV-infected patients without HCC.

Table 1. Patient Characteristics
CharacteristicsHealthy Controls (n = 1,012)HBV-Infected Patients Without HCC (n = 990)HBV-Infected Patients with HCC (n = 1,021)P
ASC (n = 316)CHB (n = 316)Cirrhosis (n = 358)
  • For multiple comparisons, statistically significant levels were corrected using the Bonferroni correction (P = 0.017).

  • Abbreviation: ND, no data.

  • *

    Between HBV-infected patients with HCC and healthy controls.

  • Between HBV-infected patients with HCC and HBV-infected patients without HCC.

  • Between HBV-infected patients with HCC and all of the controls, including healthy controls.

Male sex, n (%)763 (75.40)186 (58.86)230 (72.78)264 (73.74)864 (84.13)<0.001*,,
Age, years, mean ± SD59.56 ± 15.1045.08 ± 10.6144.18 ± 14.5150.68 ± 11.3452.92 ± 11.17<0.001*,,
HBV genotype, n (%)      
 BND97 (34.28)52 (25.00)56 (22.86)107 (16.39)<0.001
 CND186 (65.72)156 (75.00)189 (77.14)546 (83.61) 
HBeAg, n (%)      
 PositiveND130 (41.14)132 (45.36)107 (35.55)241 (25.08)<0.001
 NegativeND186 (58.86)159 (54.64)194 (64.45)720 (74.92) 
HBV DNA, log10 copies/mL, mean ± SDND3.88 ± 1.804.43 ± 1.674.13 ± 1.373.83 ± 1.18<0.001
ALT, log10 U/L, mean ± SDND1.36 ± 0.211.97 ± 0.541.75 ± 0.441.66 ± 0.35<0.001

Associations of the STAT3 SNPs With HBV-Related HCC.

Genotyping accuracy of rs4796793, rs2293152, and rs1053004 was ascertained by sample success rates and call rates of 99.8%, 99.9%, and 98.3% in healthy controls and 98.6%, 99.9%, and 98.2% in the HBV-infected patients, respectively. In healthy controls, rs4796793 and rs2293152 were conformed to HWE (P > 0.05 for each), whereas rs1053004 was out of HWE (P = 0.001). We amplified and sequenced a DNA fragment covering rs1053004 from 40 randomly selected healthy controls (GenBank No. JX296640-JX296679) and the genotyping results were 100% consistent with those of quantitative PCR.

Table 2 presents the associations of the SNPs and their multiplicative interactions with sex with HCC risk. rs2293152 GG genotype was significantly associated with an increased risk of HCC as compared with all subjects without HCC, and this association was exclusively evident in females. Multiplicative interaction of rs2293152 (GG versus CC) with sex (male versus female) was significantly associated with a reduced risk of HCC. rs1053004 CC genotype was associated with a reduced risk of HCC in females (adjusted odds ratio [AOR], 0.49; 95% CI, 0.25-0.97) although the P value did not reach the significance level after Bonferroni correction. Multiplicative interaction of rs1053004 (CC versus TT) with sex (male versus female) was significantly associated with an increased risk of HCC. No significant differences in the distributions of the three SNPs were found between HCC patients and cirrhosis patients (data not shown). The associations of the three SNPs with HCC risks were evaluated in the HBV-infected subjects stratified by HBV genotypes. It was found that rs2293152 GG genotype was significantly associated with an increased risk of HCC in the genotype C HBV-infected subjects, adjusting for age and sex (AOR, 1.76; 95% CI, 1.21-2.56).

Table 2. Associations of STAT3 SNPs and Their Multiplicative Interactions With Sex With HCC Risk
SNP (Accession Code)SexGenotypeHealthy ControlsHBV-Infected Patients Without HCCHBV-HCC PatientsHBV-HCC Patients Versus HBV-Infected Patients Without HCCHBV-HCC Patients Versus All Controls, Including Healthy Controls
AOR (95% CI)PAOR (95% CI)P
  • *

    The difference was still significant after Bonferroni correction (cutoff: P = 0.025 [two comparisons]).

  • Multiplicative interactions of STAT3 SNPs with sex.

rs4796793 (C>G)CombinedCC4313914141.00 1.00 
CG4494484530.98 (0.80-1.19)0.8010.97 (0.82-1.15)0.716
GG1301341421.04 (0.78-1.39)0.7831.04 (0.81-1.32)0.781
G (CG+GG)5795825950.99 (0.82-1.20)0.9310.99 (0.84-1.15)0.847
rs4796793 (GG versus CC) × sex (male versus female)1.31 (0.64-2.72)0.4621.54 (0.81-2.92)0.190
FemaleCC102128691.00 1.00 
CG101139740.93 (0.61-1.42)0.7441.00 (0.69-1.47)0.984
GG4438180.85 (0.45-1.62)0.6260.72 (0.40-1.29)0.269
G (CG+GG)145177920.91 (0.61-1.36)0.6420.93 (0.65-1.33)0.676
MaleCC3292633451.00 1.00 
CG3483093790.98 (0.78-1.23)0.8850.95 (0.79-1.15)0.606
GG86961241.11 (0.80-1.53)0.5371.12 (0.85-1.46)0.421
G (CG+GG)4344055031.01 (0.82-1.26)0.9040.99 (0.83-1.18)0.895
rs2293152 (C>G)CombinedCC2942452651.00 1.00 
CG5004664960.96 (0.76-1.20)0.6961.06 (0.88-1.28)0.531
GG2151982521.28 (0.98-1.68)0.0701.30 (1.04-1.62)0.019*
G (CG+GG)7156647481.05 (0.85-1.30)0.6451.13 (0.95-1.35)0.164
rs2293152 (GG versus CC) × sex (male versus female)0.56 (0.30-1.06)0.0760.57 (0.32-0.99)0.046
FemaleCC6672321.00 1.00 
CG132140701.10 (0.65-1.86)0.7231.12 (0.70-1.79)0.651
GG5069592.04 (1.15-3.61)0.015*2.24 (1.34-3.73)0.002*
G (CG+GG)1822091291.39 (0.86-2.27)0.1821.44 (0.93-2.24)0.104
MaleCC2281732331.00 1.00 
CG3683264260.92 (0.72-1.19)0.5441.05 (0.86-1.29)0.615
GG1651291931.12 (0.82-1.52)0.4731.16 (0.91-1.49)0.239
G (CG+GG)5334556190.98 (0.77-1.25)0.8731.09 (0.90-1.32)0.404
rs1053004 (T>C)CombinedTT4533854111.00 1.00 
TC4004514580.98 (0.80-1.20)0.8571.08 (0.91-1.27)0.392
CC1421301401.05 (0.78-1.40)0.7531.03 (0.81-1.31)0.832
C (TC+CC)5425815981.00 (0.83-1.21)0.9821.07 (0.91-1.25)0.435
rs1053004 (CC versus TT) × sex (male versus female)2.37 (1.06-5.28)0.0362.26 (1.10-4.65)0.027
FemaleTT108122701.00 1.00 
TC96141760.99 (0.65-1.50)0.9431.07 (0.73-1.56)0.732
CC3939120.51 (0.25-1.07)0.0760.49 (0.25-0.97)0.040
C (TC+CC)135180880.88 (0.59-1.32)0.5290.92 (0.64-1.33)0.668
MaleTT3452633411.00 1.00 
TC3043103820.98 (0.78-1.23)0.8691.08 (0.89-1.30)0.435
CC103911281.22 (0.88-1.69)0.2331.15 (0.89-1.50)0.286
C (TC+CC)4074015101.03 (0.83-1.28)0.7611.10 (0.92-1.31)0.303

Associations of the STAT3 SNPs With Chronic HBV Infection, Cirrhosis, HBeAg Seroconversion, and Viral Replication.

rs4796793 G(CG+GG) allele was significantly associated with HBeAg seroconversion and inversely associated with high viral load (≥1 × 104 copies/mL) (AOR, 0.69; 95% CI, 0.56-0.86); rs2293152 CG genotype was significantly associated with cirrhosis (AOR, 2.41; 95% CI, 1.13-5.13) while its G(CG+GG) allele and GG genotype were significantly associated with high viral load (G allele: AOR, 2.28; 95% CI, 1.19-4.37; GG genotype: AOR, 2.73; 95% CI, 1.32-5.65, respectively) in females; rs1053004 TC genotype was significantly associated with HCC-free chronic HBV infection (AOR, 1.25; 95% CI, 1.01-1.56), whereas its variant genotypes were inversely associated with high viral load (Supporting Table 2).

The HCC-Related HBV Mutations and Their Correlations.

We successfully amplified and sequenced the EnhII/BCP/PC region from 252 (79.75%) ASCs, 172 (54.43%) CHB patients, 224 (62.57%) cirrhosis patients, and 512 (50.15%) HCC patients as well as the preS region from 130 (41.14%) ASCs, 164 (51.90%) CHB patients, 194 (54.19%) cirrhosis patients, and 460 (45.05%) HCC patients (GenBank No. JX556943-JX559050).

The “hotspots” in the EnhII/BCP/PC region and the preS region of HBV genotype C and their associations with HCC are listed in Supporting Tables 3 and 4, respectively. Of those, C1653T, T1674C/G, G1719T, A1727G, T1753C, A1762T/G1764A, A1846T, G1896A, C2875A, A1C/T, C7A, C10A, A31C/T, T49A, A52C/T, C76A, G105C/T, C109A/T, A135C, G147C, preS deletion, and preS2 start codon mutation were significantly associated with an increased risk of HCC, whereas A1652G, C1673T, A1726C, A1727T, C1730G, and C1799G were significantly associated with a reduced risk of HCC, compared with the HBV-infected subjects without HCC (after the Bonferroni correction for multiple comparison). However, preS1 deletion, preS2 deletion, and T53C (F141L), the mutations reported to be related to HCC,4, 5, 7, 12 were not significantly associated with HCC. Of those HCC-related HBV mutations, A1762T/G1764A, G1719T, preS deletion, C10A, T49A, A135C, A1C/T, A31C/T, A52C/T, and C109A/T were more frequent in males than in females in HBV-infected patients without HCC (Supporting Table 5).

Correlation analyses indicated that the HBV mutations in the preS2 region including C10A, C31C/T, T49A, A52C/T, C109A/T, and A135C correlated with each other (phi > 0.800). Supporting Table 6 shows the correlations between the selected HCC-related mutations. In addition, G1896A, T1674C/G, C2875A, and C76A were significantly associated with HBeAg seroconversion; A1762T/G1764A was significantly associated with high viral load (AOR, 1.64; 95% CI, 1.18-2.28) and abnormal ALT (AOR, 1.94; 95% CI, 1.37-2.74).

Interactions of the STAT3 SNPs With the HBV Mutations and Their Associations With HCC.

The associations of the STAT3 SNPs with the HCC-related HBV mutations including T1674C/G, A1762T/G1764A, T1753V, preS deletion, preS2 start codon mutation, C2875A, C7A, and C76A were assessed using the data of the HBV-infected patients, including the HCC patients. It was found that rs2293152 variant genotypes were significantly associated with increased frequencies of T1674C/G and A1762T/G1764A (Table 3), rather than other mutations.

Table 3. Associations of STAT3 SNPs With Frequencies of T1674C/G and A1762T/G1764A Using Data of HBV-Infected Patients, Including HCC Patients
SNP (Accession Code)Genotype/AlleleT1674C/GA1762T/G1764A
TC/GAOR (95% CI)AG/AA/TGTAAOR (95% CI)
rs4796793 (C>G)CC3221111.001692871.00
CG3281050.92 (0.67-1.25)1912750.83 (0.63-1.09)
GG105341.00 (0.64-1.57)50911.11 (0.74-1.66)
G (CG+GG)4331390.93 (0.70-1.24)2413660.88 (0.68-1.14)
rs2293152 (C>G)CC208511.001271531.00
CG3411291.49 (1.03-2.16)1823161.41 (1.04-1.91)
GG171661.61 (1.06-2.46)881601.64 (1.14-2.35)
G (CG+GG)5121951.53 (1.08-2.17)2704761.48 (1.11-1.97)
rs1053004 (T>C)TT3051081.001622741.00
TC3401090.91 (0.67-1.24)1932870.88 (0.67-1.16)
CC106270.79 (0.48-1.28)53840.98 (0.65-1.48)
C (TC+CC)4461360.88 (0.66-1.18)2463710.90 (0.69-1.16)

Contributions of the three SNPs and their multiplicative interactions with the HCC-related HBV mutations to HCC were assessed using multivariate regression analyses, adjusting for covariates, including the HBV mutations. The HBV mutations included in each equation were not strongly correlated (phi < 0.300) (Supporting Table 6). rs2293152 GG genotype was significantly associated with an increased risk of HCC, but the interaction of rs2293152 GG genotype with A1726C was associated significantly with a reduced risk of HCC; the interaction of rs1053004 TC genotype with T1674C/G was significantly associated with an increased risk of HCC in the subjects with HBV sequencing data of the EnhII/BCP/PC region (Table 4). The interaction of rs4796793 GG genotype with preS2 start codon mutation was significantly associated with an increased risk of HCC in the patients with the preS region sequencing data (Table 5).

Table 4. Contributions of STAT3 SNPs and Their Interactions With HBV Mutations in the EnhII/BCP/PC Region to HCC Risk
VariablesAOR (95% CI)P
rs4796793 (C>G)  
 Age, years1.03 (1.01-1.05)0.001
 Sex, male versus female1.81 (1.08-3.05)0.025
 Viral load, ≥104 versus <104 copies/mL0.88 (0.57-1.36)0.567
 T1674C/G1.90 (1.09-3.31)0.024
 A1762T/G1764A1.32 (0.80-2.15)0.275
 A1726C1.40 (0.77-2.56)0.268
 rs4796793, GG versus CC0.90 (0.51-1.59)0.707
 rs4796793, (GG versus CC) × T1674C/G0.97 (0.31-3.05)0.962
rs2293152 (C>G)  
 Age, years1.04 (1.02-1.06)2.94E-04
 Sex, male versus female1.86 (1.06-3.26)0.031
 Viral load, ≥104 versus <104 copies/mL1.34 (0.83-2.16)0.232
 T1674C/G2.06 (1.14-3.70)0.016
 A1762T/G1764A2.10 (1.24-3.56)0.006
 A1726C2.39 (1.00-5.69)0.050
 rs2293152, GG versus CC1.98 (1.18-3.34)0.010
 rs2293152, (GG versus CC) × A1726C0.20 (0.06-0.69)0.011
rs1053004 (T>C)  
 Age, years1.04 (1.02-1.05)4.47E-06
 Sex, male versus female1.92 (1.27-2.92)0.002
 Viral load, ≥104 versus <104 copies/mL1.11 (0.78-1.58)0.570
 T1674C/G1.33 (0.77-2.29)0.315
 A1762T/G1764A1.45 (0.98-2.16)0.065
 A1726C1.28 (0.78-2.10)0.337
 rs1053004, TC versus TT0.71 (0.47-1.06)0.093
 rs1053004, (TC versus TT) × T1674C/G2.25 (1.02-4.95)0.044
Table 5. Contributions of STAT3 SNPs and Their Interactions With HBV Mutations in the PreS Region to HCC Risk
VariablesAOR (95% CI)P
rs4796793 (C>G)  
 Age, years1.04 (1.02-1.06)8.90E-05
 Sex, male versus female2.91 (1.65-5.14)2.21E-04
 Viral load, ≥104 versus <104 copies/mL0.46 (0.28-0.76)0.002
 preS deletion1.32 (0.78-2.22)0.307
 preS2 start codon mutation1.28 (0.68-2.39)0.445
 rs4796793, GG versus CC0.95 (0.50-1.79)0.873
 rs4796793, (GG versus CC) × preS2 start codon mutation5.59 (1.01-30.83)0.048
rs2293152 (C>G)  
 Age, years1.05 (1.03-1.07)2.35E-05
 Sex, male versus female2.08 (1.13-3.84)0.019
 Viral load, ≥104 versus <104 copies/mL0.53 (0.31-0.90)0.019
 preS deletion0.93 (0.53-1.65)0.805
 preS2 start codon mutation3.53 (1.33-9.42)0.012
 rs2293152, GG versus CC1.27 (0.71-2.27)0.424
 rs2293152 (GG versus CC) × preS2 start codon mutation0.34 (0.09-1.23)0.100
rs1053004 (T>C)  
 Age, years1.05 (1.03-1.06)1.09E-07
 Sex, male versus female2.76 (1.75-4.35)1.18E-05
 Viral load, ≥104 versus <104 copies/mL0.53 (0.36-0.79)0.002
 preS deletion1.56 (1.02-2.40)0.040
 preS2 start codon mutation1.38 (0.73-2.59)0.321
 rs1053004, TC versus TT1.34 (0.87-2.08)0.187
 rs1053004, (TC versus TT) × preS2 start codon mutation0.91 (0.35-2.40)0.855

Discussion

In this study, we found that STAT3 rs2293152 GG genotype was significantly associated with an increased risk of HCC. This effect was exclusively evident in females. Furthermore, the interaction of rs2293152 GG genotype with females was significantly associated with HCC risk. HCC is more frequent in males than in females. This sex disparity might be related to sex hormone signaling, increased exposure to environmental risk factors, and genetic predisposition such as polysomy of chromosome 7.2, 31-33 Cigarette smoking and heavy alcohol consumption have been proven to increase HCC risk in males.34 In China, exposures to alcohol and smoking are more common in males than in females. Some of the HBV mutations were more frequent in males than in females (Supporting Table 5). The rs2293152 effect on genetic susceptibility to HCC might be overwhelmed by strong effects of these risk factors in males. In contrast, the rs2293152 effect in females might reflect a less biased association of the SNP with HCC. Additionally, rs2293152 GG genotype was significantly associated with high viral load in females, rather than in males (Supporting Table 2). High viral load has been proven to be a major risk factor of HCC in prospective study.5 Interestingly, the interaction of rs1053004 with males was significantly associated with HCC risk. The reason remains to be clarified. Some data have linked STAT3 signaling to sex hormones. For example, estrogen can activate STAT3 signaling, whereas interleukin-6/STAT3 signaling activates androgen receptor-mediated gene expression.35, 36 It is biologically plausible that the interplay between sex hormones and SNPs-affected STAT3 functions might play a direct or indirect role in the mediation of sex differences in the susceptibility to HCC. Such speculation needs to be tested in in-depth molecular studies.

The effect of rs2293152 was particularly evident in genotype C HBV–infected patients. Compared with HBV genotype B, genotype C is more prone to cause chronic HBV infection/inflammation and HCC.5, 26 According to the data reported here (Table 3 and Supporting Table 2) and elsewhere,28-30 rs2293152 might predispose the HBV-infected patients to dysregulation of STAT3-related inflammation pathway which affect viral replication and immunoselection of T1674C/G and A1762T/G1764A, thus contributing to HBV-induced hepatocarcinogenesis. rs1053004 and rs4796793 were significantly related to low viral load, while they were also related to persistent HBV infection and HBeAg seroconversion, respectively (Supporting Table 2). Thus, the two SNPs tend to be related to immune tolerance.

rs2293152 GG genotype was significantly associated with an increased risk of HCC; however, its interaction with A1726C, an HBV mutation inversely associated with HCC risk, was significantly associated with a reduced risk of HCC (Table 4). Thus, the rs2293152 effect could be strongly affected by the HBV mutations. This might be one of the reasons why rs2293152 has not been found as a susceptible locus of HBV-HCC in a recent genome-wide association study.37 In this study, we also found that the interaction of rs1053004 with T1674C/G was significantly associated with an increased risk of HCC, although rs1053004 and T1674C/G were not significantly associated with HCC risk in this equation (Table 4). This result indicates that the contribution of T1674C/G to HCC depends on rs1053004 genotype.

HBV mutations in the preS region affect HBsAg expression and are closely related to progressive liver diseases.4, 5, 7, 12, 38 The preS mutations have a high level of quasispecies. We defined the missing of three consecutive nucleotides or more in the preS region as “preS deletion”.7 “PreS start codon deletions” were mostly sorted into “preS start codon mutations.” Thus, HBV preS2 start codon mutations were significantly associated with HCC risk. We added the HBV mutations in the preS region along with other covariates into multivariate regression equations and found that the interaction of rs4796793 with preS2 start codon mutation was significantly associated with HCC risk (Table 5). This result indicates that rs4796793 might contribute to the effect of preS2 start codon mutation in hepatocarcinogenesis.

Our study has several limitations. First, we failed to amplify the two HBV fragments from partially overlapped fractions of HBV-infected populations, resulting in a possible preponderance of missing data and the inconsistence of the rs2293152 effect in the two multivariate analyses (Tables 4 and 5). Second, cases and controls were not matched for age and sex due to difficulty in recruiting older HBV-infected patients in hospitals. Third, other environmental exposures such as alcohol consumption and cigarette smoking in cases and controls were incomplete and thus not included in the analyses. Fourth, the study design is cross-sectional in nature. Future prospective studies are needed to validate the synergistic effect of these SNPs with the HBV mutations in hepatocarcinogenesis and define the HBV-infected subjects who are more likely to develop HCC.

In conclusion, rs2293152 is significantly associated with HCC risk, especially in females or in genotype C HBV-infected subjects. The interactions of rs1053004 with T1674C/G and rs4796793 with preS2 start codon mutation significantly increase HCC risk. The STAT3 polymorphisms might predispose the host to immune selection of the HBV mutations and contribute to the effect of the HBV mutations in hepatocarcinogenesis, and this effect may differ in men versus women. The present study provides important evidences for recognizing rs2293152 as a novel genetic marker of HBV-HCC and also presents a future direction of exploring genetic susceptibility to cancers whose occurrences are strongly affected by environmental factors in post–genome-wide association study era.

Acknowledgements

We thank Wu Ni and Xinyan Sun (2nd Affiliated Hospital, Second Military Medical University, Shanghai, China), Chengzhong Li and Qian Zhang (1st Affiliated Hospital, Second Military Medical University, Shanghai, China), Huafen Wang (88th Hospital, Taian City, Shandong, China), and Lei Han (Southwest Hospital, Chongqing, China) for help in the recruitment of the study subjects.

Ancillary