Combined Effects of Six Cytokine Gene Polymorphisms and SNP-SNP Interactions on Hepatocellular Carcinoma Risk in Southern Guangxi, China

Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide, representing more than 85% of all primary liver cancers (Parikh et al., 2007). About half of all liver cancer deaths were reported to have occurred in China (Jemal et al., 2011). Southern Guangxi has one of the highest occurrences of liver cancer in China, which ranked first in all cancer-related deaths (Zhang et al., 2010). Chronic infection with hepatitis B virus (HBV) that is perinatally acquired from carrier mothers is a primary cause in China (Yeh et al., 1989). After initial HBV infection, 90% to 95% of adults can rely on their own immune systems to clear the virus, causing self-limiting hepatitis. Only 5% to 10% of infected individuals develop chronic hepatitis, and 10% to 25% of these patients eventually progress to HCC (Lok et al., 2001; Iino et al., 2002). This phenomenon suggests that the body’s scavenging ability for HBV is based on individual differences. That is, although HBV infection is the major risk factor for HCC, only a fraction of HBV-infected individuals develops HCC. This development behavior suggests that genetic factors may also have important functions in HCC etiology. Cytokines are a family of proteins that mediate


Introduction
Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide, representing more than 85% of all primary liver cancers (Parikh et al., 2007).About half of all liver cancer deaths were reported to have occurred in China (Jemal et al., 2011).Southern Guangxi has one of the highest occurrences of liver cancer in China, which ranked first in all cancer-related deaths (Zhang et al., 2010).Chronic infection with hepatitis B virus (HBV) that is perinatally acquired from carrier mothers is a primary cause in China (Yeh et al., 1989).After initial HBV infection, 90% to 95% of adults can rely on their own immune systems to clear the virus, causing self-limiting hepatitis.Only 5% to 10% of infected individuals develop chronic hepatitis, and 10% to 25% of these patients eventually progress to HCC (Lok et al., 2001;Iino et al., 2002).This phenomenon suggests that the body's scavenging ability for HBV is based on individual differences.That is, although HBV infection is the major risk factor for HCC, only a fraction of HBV-infected individuals develops HCC.This development behavior suggests that genetic factors may also have important functions in HCC etiology.
Cytokines are a family of proteins that mediate activated T-cells (NFAT).Thus, this site may influence the activity of transcription factors and subsequently affect the transcription and translation levels of IL-2 (Crabtree et al., 1994;Pyo et al., 2003).Several studies have found that IL-2-330A/C is associated with the risk of genetic susceptibility to nasopharyngeal carcinoma (Wei et al., 2010) and gastric cancer (Sugimoto et al., 2010).Interferon-gamma (IFN-γ), a major cytokine secreted by T cells, is crucial for anti-viral immune response.The constitutive expression of IFN-γ in the liver causes chronic active hepatitis through the recruitment of lymphomononuclear cells (Toyonaga et al., 1994).
The expression of IFN-γ is slightly higher in HCC tissue than in adjacent normal tissue (Chia et al., 2002).Su et al. (2012) found that IFN-γ-1615 GA/AA genotypes are significantly related to an elevated risk of breast cancer.IL-10 is a potent immunosuppressive anti-inflammatory cytokine that is highly expressed in HCC tissue and adjacent normal tissue (Chia et al., 2002).IL-10 is the most studied anti-inflammatory cytokine involved in cancer research.The promoter region is highly polymorphic and is associated with gastric cancer and HCC (Shin et al., 2003;Pan et al., 2013).IL-6 is a multifunctional cytokine that serves pro-and anti-inflammatory functions.The level of IL-6 is higher in HCC patients than in healthy individuals (Giannitrapani et al., 2011).Numerous cytokines participate in the construction of a highly complex and coordinated network in which they modulate their own synthesis and that of other cytokines (Balkwill et al., 1989;Elias et al., 1992).Recent studies have evaluated the effects of individual SNPs but failed to consider the potential combined effects and SNP-SNP interactions in two or more loci.Therefore, one-on-one association cannot sufficiently explain the complexity of disease causality (Onay et al., 2006;Lin et al., 2013).Therefore, we hypothesized that not only the individual effects but also the combined effects of SNPs and SNP-SNP interactions in cytokine genes may increase HCC risk.Cytokine gene functional SNPs (rs2069762, rs2069705, rs1143623, rs16944, rs1800769, and rs1800872) were selected to test our hypothesis in this study.

Study population
All HCC patients and cancer-free controls were recruited from First Affiliated Hospital of Guangxi Medical University, First Affiliated Hospital of Guangxi College of Traditional Chinese Medicine, and First Affiliated Hospital of Guilin Medical University from August 2007 to November 2010.The subjects received a detailed description of the study protocol and signed informed consent forms that were approved by the institutional review board for each medical center.The patients were all newly diagnosed with HCC but had not been treated with radiation, chemotherapy, or surgical therapy before enrollment in the study.The cancer-free controls were recruited in the same period from the Department of Orthopedics and Ophthalmology of the same hospitals.The controls were genetically unrelated visitors or companions of patients and were frequency matched to the cases by age (±5 years), gender, and nationality (Han, Zhuang, or others).The study protocol was approved by the institutional review board of the Guangxi Cancer Institute.

Data collection
After informed consent was obtained, a structured interview was conducted by trained interviewers.Data were collected from interviews and reviews of medical records on the demographic characteristics, recent and prior tobacco use, alcohol consumption, and HBV virus markers.Upon consent, 5mL of nonheparinized blood was collected from each participant for cytogenetic and molecular genetic analyses.All laboratory and questionnaire data were coded, entered by two investigators, and verified using EpiData 3.1 (www.epidata.dk/download.php).Neither the laboratory nor the data entry personnel had any knowledge of the subjects' case-control status.

Genotyping
Genomic DNA was extracted from leukocytes using the phenol-chloroform method and stored at -80°C.Four SNPs (Table 1) were selected through the SNP info platform (http://snpinfo.niehs.nih.gov/) with the following criteria: the genes were the main cytokine genes involved in the pro-or anti-inflammatory pathways of liver disease, the SNPs were in the transcription factor binding site of the 5' promoter region, and the pairwise linkage disequilibrium (LD) had a r 2 threshold of 0.8 and a minor allele frequency of 5% or higher in Han Chinese in Beijing (CHB).Considering these criteria, we selected six loci SNPs for genotyping (Table 1).The polymorphisms were genotyped using high-flux TaqMan MGB, a fluorescent quantitative real-time polymerase chain reaction (PCR) method.Each PCR reaction mixture (25 µL) contained 1.00 µL of DNA template, 10.25 µL of H 2 O, 12.50 µL of 2×TaqMan universal PCR mix, and 1.25 µL of 20×SNP genotyping assay mix.The genomic DNA was amplified at 95°C for 10min, followed by 40cycles of 92°C for 15s and 60°C for one min.All genotyping reagents and analytical software were purchased from Applied Biosystems.Genotyping was performed according to the manufacturer's instructions.Approximately 5% of the samples were randomly repeated to validate the genotyping procedures, and the concordance rate was 100%.

Statistical analysis
The differences in the distributions of categorical variables, including demographic characteristics, tobacco smoking, and alcohol drinking, and genotype/allele frequencies of selected SNPs between the patients and controls were evaluated using chi-square tests.Crude or adjusted (for age, gender, nationality, smoking, drinking, and HBsAg status) odd ratios (ORs) and 95% confidence intervals (CIs) were obtained from unconditional univariate and multivariable logistic regression analyses to evaluate the associations between SNPs and HCC risk in the case-control analysis.Data were further stratified by age, gender, nationality, smoking, drinking, and DOI:http://dx.doi.org/10.7314/APJCP.2014.15.16.6961 Six Cytokine Gene Polymorphisms and SNP-SNP Interactions and HCC Risk in Southern Guangxi, China HBsAg status.Continuous variables were analyzed using Student's t-test and presented as mean±SD.HaploView 4.2 was used to test Hardy-Weinberg equilibrium in the control subjects and LD between SNPs in the same gene.The interactions between environmental factors and genomic variants, as well as those between SNPs were qualitatively analyzed with a logistic regression model.All statistical tests were two-sided with a 0.05 significance level, and all data were analyzed with SPSS 13.0 for Windows.

Characteristics of the study population
This study included 720 HCC patients and 784 controls with available DNA samples.Table 2 presents the frequency distribution of the selected patient and control characteristics.No statistical differences in age and gender distributions were detected between the cases and controls because of the frequency matching by design.The mean age was 47.72 years (±11.16,range: 19 years to 81 years) for the controls and 48.65 years (±11.03,range: 18 years to 78 years, P=0.111) for the HCC patients.Approximately 83.3% and 16.7% of the controls and 86.0% and 14.0% of the patients were male and women, respectively (P=0.150).No significant difference in ethnicity distribution was detected between the cases and controls (P=0.911).However, the prevalence of smoking, drinking, and HBsAg positivity was significantly higher in the patients than in the controls (P<0.001for each).Therefore, smoking, drinking, and HBsAg status were adjusted in the subsequent multivariate logistic regression analyses.

Genotype distribution of polymorphisms of cytokines between patients and controls
The allele and genotype distributions of the studied SNPs in the patients and controls and their associations with HCC risk are summarized in Table 3.All observed genotype distributions among the controls agreed with the Hardy-Weinberg equilibrium (P>0.05).No significant difference in the genotype frequencies of the studied SNPs was found between the patients and controls (P=0.874for IL-2-330A/C, P=0.606 for IFN-γ-1615G/A, P=0.643 for IL-1β-511C/T, P=0.272 for IL-1β-1464C/G, P=0.786 for IL-6-572C/G, and P=0.101 for IL-10-592T/G).Although none of the variant genotypes was individually associated with a significantly altered risk, the IL-2-330 AC and CC genotypes were associated with a non-significantly reduced HCC risk (adjusted OR=0.93; 95% CI=0.72 to 1.19 for AC, and adjusted OR=0.99; 95% CI=0.68 to 1.45 for CC).The AA genotype of IFN-γ-1615 was associated with a non-significantly increased HCC risk (OR=1.36;95% CI=0.92 to 2.01).The CT and TT genotypes of IL-1β -511 were associated with a non-significantly increased HCC risk (adjusted OR=1.31; 95% CI = 0.98 to 1.74 for CT, and adjusted OR=1.17; 95% CI=0.84 to 1.63 for TT).The GG genotype of IL-1β-1464G/C was associated with a non-significantly reduced HCC risk (adjusted OR = 0.84; 95% CI=0.60 to 1.17).The genotypes of IL-6-572C/G and IL-10-819T/G were associated with a non-significantly increased HCC risk (adjusted OR=1.05; 95% CI=0.82 to 1.34 for CG/GG, and adjusted OR=1.03; 95% CI=0.82 to 1.30 for TG/GG, respectively).

Evaluation of SNP-SNP interactions
Considering the possible interaction effects among the six SNPs and environmental factors on HCC risk, we further assessed the interactions of these SNPs  and environmental factors, such as smoking, drinking, and HBsAg positivity in HCC.However, no statistical evidence was found for interactions among these variables in the multivariate logistic regression models (data not shown).In addition, SNP-SNP interaction between IL-2-330 and IFN-γ-1615 was associated with an increased HCC risk (OR=1.078,95% CI=1.022 to 1.136).SNP-SNP interactions were not observed in any other SNPs (Table 6).

Discussion
In the 720 HCC patients and 784 cancer-free controls, none of the six SNPs in cytokine genes (IL-2-330A/C, IFN-γ-1615G/A, IL-1β-511C/T, IL-1β-1464C/G, IL-6-572C/G, and IL-10-592T/G) was individually associated with HCC risk.Rather, the combined effects of these SNPs increased HCC risk among smokers, drinkers, and HBsAg-positive individuals.These findings support our hypothesis that the effect of an individual SNP is generally not significant and that the genetic combined effects of SNPs and SNP-SNP interactions increase HCC risk.
Numerous studies have investigated the relations between cytokine polymorphisms and HCC risk; however, their results are controversial (Shin et al., 2003;Hirankarn et al., 2006;Slattery et al., 2007).Two studies have consistently noted that IL-1β-511 polymorphisms are significant risk factors for HCC progression in the Thai population and in Japanese patients (Tanaka et al., 2003;Hirankarn et al., 2006).However, a recent meta-analysis has reported that IL-1β-511C/T is not significantly associated with HCC risk in an Asian population (Yang et al., 2011).And a recent study has suggested the lack of association of selected seven cytokine gene polymorphisms with the risk of HCC recurrence in Han Chinese population (Wu et al., 2013).Previous studies evaluated the effects of individual SNP or single one gene on the HCC risk and failed to consider the potential combined effects of these SNPs on different genes (Tanaka et al., 2003;Hirankarn et al., 2006;Pan et al., 2013).In the present study, no individual SNP was significantly associated with HCC risk.It is likely that the effect of single SNP on HCC risk may be modest and undetectable.Alternatively, the effect of a single SNP may be influenced by that of other gene SNPs.In this study, we found that individuals with four to six risk genotypes (i.e., IL-2-330AA, IFN-γ-1615AG/ AA, IL-1β-511CT/TT, IL-1β-1464CG/GG, IL-6-572GG/ CG, and IL-10-592TG/GG) may have an increased HCC risk.Consistent with the result of Nieters et al. (2005), the present study found that the synergistic rather than the individual effects of cytokine gene SNPs were associated with an increased HCC risk.
HCC is a multi-factorial disease that results from complex interactions between environmental and genetic factors.Therefore, we speculated that hereditary factors and environmental factors (smoking, drinking, and HBsAg positivity) increase HCC risk.Compared with individuals who never smoked or drank, smokers and drinkers had a significantly higher HCC risk, which was associated with the combined effects of the putative high-risk genotypes in cytokine genes (Table 5).Similar conclusions have been previously attained.Chen et al. (2005) have reported that smoking modifies the combined effects of multiple loci in the cytokine and DNA repair genes.Cigarette smoking is reportedly associated with an increased HCC risk in HBsAg carriers, and this association is independent of geography and ethnicity (El-Zayadi et al., 2006;Koh et al., 2011).Cigarette consists of over 4000 toxic substances, which cause adverse effects on the immune system (Zhang et al., 2014).Excessive alcohol consumption is a well-established risk factor for liver disease and HCC (Morgan et al., 2004).These results, together with our findings, suggest that smoking and drinking promote HCC progression.Previous epidemiological studies have found that chronic infection with HBV is a major etiological risk factor for HCC in China (Yeh et al., 1986;Groopman et al., 1996).Expectedly, the combined effects of cytokine gene SNPs with HBV infection increased HCC risk.However, we did not find statistical evidence on the interactions among these variables possibly because of the limited sample size.
The present study showed that the interaction between IL-2-330A/C and IFN-γ-1615G/A was significantly associated with HCC risk.This result indicates that the individual effect of IL-2-330A/C or IFN-γ-1615 G/A is generally minimal and SNP-SNP interaction increases HCC risk.Ikeguchi et al. (2005) evaluated the gene expression levels of IL-2 in HCC patients and found that IL-2 expression may be an important prognostic biomarker for HCC.However, it was reported that the distribution of IL-2 -330A/C genotypes between HCC patients and healthy people had no statistical difference (Ognjanovic   , 2009).IFN-γ was detected in HCC, and the authors concluded that IFN-γ may not play a large role in liver inflammation (Tangkijvanich et al., 2000).Similar to that, no significant differences were found in IL-2-330 and IFN-γ-1615 genotype distribution in Guangxi people of China between HCC patients and controls in this study.Although IL-2-330 and IFN-γ-1615 are located on different genes, their positions are considerably near (rs2069762 and rs2069705).Therefore, their locations may affect their biological functions through interactions.To date, SNP-SNP interactions between different cytokine genes and HCC risk have not yet been reported.Further studies with large sample sizes are necessary to confirm our findings.
To the best of our knowledge, this study is the first to investigate the association between cytokine gene SNP-SNP interactions and HCC risk.We found that SNP-SNP interactions between SNPs that did not exert individual effects on HCC risk were associated with HCC occurrence.Previous studies investigated SNP-SNP interactions during the development of breast cancer (Sapkota et al., 2013) and colorectal cancer (Goodman et al., 2006).These reports support our findings regarding the effects of SNP-SNP interactions on HCC risk.That is, SNPs with negligible effects may interact with other SNPs to increase HCC risk.In this study, potential confounding factors, such as age, gender, and nationality, were matched between the patients and controls to decrease possible biases.Furthermore, our results were based on adjusted estimates.The analyses were adjusted for other covariates, including smoking, alcohol drinking, and HBsAg status.However, several potential limitations in this study should be considered.Firstly, six SNPs were included in the analyses, and the possibility of false-positive associations could not be ruled out because of multiple tests.Secondly, this study excluded several other cytokine SNPs, such as TNF-A-308 and IL-10-819 polymorphisms (Budhu et al., 2006), which might affect HCC risk.Thirdly, the number of current studies was relatively small.Therefore, further investigations involving subjects of different races must be conducted to clarify this relationship.Despite these limitations, our findings are biologically plausible.Therefore, more risk polymorphisms of HCC should be induced as covariants to conclude an accurate effect.
In summary, none of the six SNPs of IL-2-330A/C, IFN-γ-1615G/A, IL-1β-511C/T, IL-1β-1464C/G, IL-6-572C/G, and IL-10-592T/G was individually associated with HCC risk.However, their combined effects and interactions may have important functions in HCC development in the high-risk regions in China.The cytokine-coordinated network system is so comprehensive that analyzing only a single pathway may be limiting.Therefore, further independent studies of multiple pathways with large sample sizes and relatively functional studies are needed to validate the current findings.

Table 1 . Candidate Genes and Target SNPs
Minor allele frequency in CHB according to HapMap database(HapMap Data Release 27 Phase I and II, February 2009, on NCBI B36 assembly, dbSNP126) * a

Table 2 . Demographic Characteristics of HCC Patients and Cancer-free Controls
*a P value for two-sided Chi-square test.

Table 3 . Genotype and Allele Frequency Polymorphisms Among Controls And Patients, and their Associations with HCC Risk
*a χ 2 and P value for two-sided Chi-square test.**b Adjusted for age, gender, nationality, smoking, alcohol drinking, and HBsAg status.