Study On The Associations Between The Genetic Variants Of Potential Susceptibility Gene Identified Based On GWAS And The Risk Of Hepatocellular Carcinoma In Shunde Population

Research background and objective:Hepatocellular carcinoma ranks the worlds most malignant tumor in which China is "liver cancer superpower",and interact with genetic and environmental factors,as everyone knows.According to the latest statistics in 2015,an estimated 782,500 new liver cancercases and 745,500 deaths occurred worldwide during 2012,with China alone accounting for about 50%of the total number of cases and deaths.In recent years,basic and clinical studies of liver cancer continue to make new progress,but statistics show that the incidence and mortality of liver cancer were not decreased,continued to show a rising trend year by year.Liver cancer has become a major public health problem seriously harm to life and health of residents in china,but also brings heavy social and economic burden to some countries and regions in the world.A large number of epidemiological and experimental studies have shown that the development of liver cancer is a multi-factor,multi-stage process of accumulation by environmental factors and genetic factors.The environmental risk factors of liver cancer have been confirmed by researches,including hepatitis B virus or hepatitis C virus infection,aflatoxin intake,alcohol,smoking,and so on.Since 40 years ago(1970s),forming a high incidence area of liver cancer,including Qidong,Shunde,Fusui Tongan,there are relatively few studies about liver cancer in Shunde compared with the other three,especially in the last ten years.In China,more than 80%of liver cancer associated with chronie hepatitis B,but only a small portion of hepatitis B patients lead to liver cancer at last.And liver cancer shows a familial aggregation phenomenon.Thus,the genetic factors determined the susceptibility to liver cancer,however,the environmental factors were just induced factors.The complexity of the mechanism of liver disease were compose by gene expression in different liver cells,interaction between genes,the biological effects of genes mutated,but the priority was identification of genes associated with liver disease.With the Human Genome Project and the international haplotype Programme(HapMap)in-depth,as well as genome research technology,genome-wide association studies has become more sophisticated,and is widely used in cancer genetic research.Since 2010,research institutions in China and Japan have been studied liver cancer by GWAS in different populations,as well as the United States and Korea,finding some variants related in genes.In short,the environmental factors were just induced factors,but genetic factors determined the susceptibility to liver cancer.It was important to find specific genetic markers for early prediction and diagnosis of liver cancer,in order to prevention and control of liver cancer.Thus,our study was selected shunde as a research field,while studied regional environmental factors of liver cancer by case control study,and was selected susceptibility genes GRIK1,STAT4,HLA-DQA1 and HLA-DQB1,by fine mapping of susceptibility region which found by analyzing the latest domestic and foreign GWAS studies.Potential functional variants,which were selected by bioinformatics,were identified using logistic interaction model including both gene variants and environmental factors logistic interaction model including both gene variants and environmental factors.The three risk prediction model,including genetic model,environmental factors model,genetic and environmental factors model,was established by the two interaction patterns(gene-gene model and gene-environment model).Our studis may be hoped to provide a theoretical reference for prevention and treatment of liver cancer.Materials and Methods:The new cases in case-control study,were diagnosed in the department of Oncology integrating with clinical manifestations,imaging examination,clinical pathology,serum markers,and so on,while the controls were recruited from ones who came frome ENT,Neurology,Endocrinology,in the Shunde First People’s Hospital from November 2010 to November 2014.All the hepatocellular carcinoma cases and controls were matched at the beginning of the study,according to gender and age(±3 years).The epidemiological questionnaire were investigated by face-to-face manner,and were divided into six parts,detailed twenty items demographies,twenty-eight items personal habits,seven items occupational history,fifteen items residential history,seven items family history,and so on.The nurse help to take the peripheral blood in the morning in order to extract DNA.14 potential function of genetic variants in GRIK1 STAT4,HLA-DQ,HLA-DQAUHLA-DQB1(rs9273651,rs2832499,rs455804,rs2647073,rs3104369,rs3997872,rs9272219,rs1063355,rs9273363,rs9275319,rs363504,rs1031507,rs7574865,rs897200)were screened by bioinformatics,and were genotyped using Sequenom MassArray flux.Questionnaire information,checked and corrected,were matched with SNP data,at last,formed a complete database which would used to statistical analysis.Quantitative variables in HCC group and control group,was descripted by (?),and were eompared with t test,while nominal variables in the two groups,was descripted by proportion,and were compared with chi-square test.The risk factors of hepatocellular carcinoma were analyzed by logistic model.The association between SNPs gene variants and HCC was selected several kind of statistical methods,such as logistic regression,Pearson chi-square test,as well as Cochran-Armitage trend testThe interactions between gene,in which we selected GRIK1,STAT4,HLA-DQA1 and HLA-DQB1,and risk factors were estimated by logistic model in which interactive items were included.At last,risk prediction models of hepatocellular carcinoma were established by MDR,in which the best combination of factors was screened,and by logistic regression,in which the accuracy of model were evaluated by ROC curve.All the data were analyzed by several statistical packages,such as IBM SPSS19.0,STAT12.0,PHASE 2.1,MDR3.0.2 and MDR Permutation Testing Module(1.0 beta),ect.Results:The average age of people in the case group and the control group was 56.48±11.89,56.38±12.09 years,and here was no statistically significant difference(P=0.883).The gender ratio was 6.5:1,male 535 cases(86.71%),82 females(13.29%).The HCC case and control group in this case-conrol study included 617 people respectively from November 2010 to November 2014.Some findings were by analyzed the 617 HCC cases only in Shunde,where there is a higher incidence in Leliu,Xingtan Daliang,consistent with the 1970s’ fingdings.Six factors factors were included into the multi-factor model,finally,and then,the OR value and 95%Cl of all the six factors were caleulated after adjustment for gender and age.And most of all,hepatitis B virus infection was the major risk factor,where OR value and 95%Cl were 20.86(14.99-29.03),compared with other factors,including clonorchiasis 2.48(1.43-4.30),alcohol drinking 1.54(1.10-2.16),experience of drinking pond-ditch water 1.60(1.18-2.17),the family history of liver cancer 2.05(1.06-3.97)and the family history of carcinoma 1.56(1.04-2.35).Genetic variates of genotypes in control group were in line with HWE law,in addition to the two ones,rs9272219(P=0.000),and rsl063355(P=0.000).The distribution of genotype and allele in the two genetic variates(rs3997872 and rs9275319)in HCC case group statistically differed from ones in control group,in addition to the twelvegenetic variates,however,the two genetic variates in two groups had no statistical significance in the multivariate logistic regression analysis.The three genetic variates in GRIK1 generated five types haplotype of ACA,ACG,AAA,AAG GCA,in haplotype analysis,and the like,STAT4 included GTA,GCC,TTA,TCC,HLA-DQA1 included ATC,ATT,AAC and CTC,HLA-DQB1 included CCA,ACA,CTA,ACG ATG.The GRIK1 and STAT4 haplotypes in case group did not statistically differed from ones in control group,nor did in haplotype trend test.AAC haplotype of HLA-DQA1 in case group statistically differed from ones in control group,as a low risk compared to the haplotype ATC,OR of which was 0.43,and 95%confidence interval was 0.28-0.66,and then,the result of haplotype trend showed a statistically significant in HLA-DQA1(P=0.0041).ATG haplotype of HLA-DQB1 in the two groups statistically differed,as a low risk compared to the haplotype CCA,OR of which was 0.42,and 95%confidence interval was 0.25-0.72,the result of haplotype trend showed a statistically significant in HLA-DQB1(P=0.0065),as well as.Analyzed gene-environment interaction in logistic model,the results showed that,in GRIK1,a significant additive interaction was observed between rs363504 polymorphisms and family history of liver cance,and a ignificant multiplicative interaction between rs2832499 and HBV infection(P=0.0135),in STAT4,a significant additive interaction was observed between rs897200 and HBV infection(P=0.0360),and a ignificant additive interaction between rs7574865 and family history of carcinoma(P=0.0477),in HLA-DQA1,a significant additive interaction was observed between rs3997872 and alcohol drinking,and a ignificant multiplicative interaction,additive interaction between rs3997872 and HBV infection(P=0.0253,0.0128),in HLA-DQB1,a significant additive interaction was observed between rs9273651 and alcohol drinking,and a ignificant additive interaction between rs9275319 and HBV infection(P=0.0390),and a ignificant additive interaction between rs9275319 and family history of carcinoma(P=0.0240).Analyzed gene-gene interaction in software by MDR,9 SNPs in model produced 1-5 order different interaction models but all the combinations did not statistically differed.And then,in the gene-environment interaction,1-5 order different interaction models were produced with 14 factors in model together,which included 9 SNPs and 5 environmental factors,although each combination of factors were statistically significant.At last,we selected 5 factors model as the best model,which included three genetic variates of rs2832499,rs9273651,rs897200,and two environmental factors of HBV infection,clonorchiasis(TBA=0.7945,CVC=8/10,permutation P<0.001).Three kinds of liver cancer risk prediction model,that were genetic model,environmental factors model,and genetic and environmental factors model,were constructed using logistic regression,as well as,evaluated by the ROC curve.The area under the curve of three models were different,that was,0.5617,0.7842,0.8344,of which diagnosis results were statistically significant(P<0.01).The genetic and environmental factors risk prediction model was the best in the three ones by evaluated accuracy,and we found that the sensitivity was 74.96%,specificity was 87.71%,the correct index was 0.6267,and so on.Conclusions:(1)The incidence of HCC in Shunde is consistent with researchresults of the 1970s,so government of Shunde need lay much stress on monitoring HCC in Leliu,Xingtan,Daliang.Six environmental factors were included into our study as risk factors of HCC.And most of all 5 hepatitis B virus infection was the major risk factor,compared with five other factors,including clonorchiasis,alcohol drinking,drinking pond-ditch water,family history of liver cancer,as well as family history of carcinoma.(2)All the genetic variates in two groups had no statistical significance in the multivariate logistic regression analysis with additive,dominant and recessive model.Interaction between genetic variants and risk environmental factors only showed a statistical difference,reflecting HCC may be caused by the combined effect of environmental and genetic factors,by analysisrs363504、rs2832499、rs897200、rs7574865、rs3997872、rs9273651、rs9275319.(3)In the view of methodology,compared to only the environmental factors risk prediction model,the genetic and environmental factors risk prediction model perform well in predicting liver cancer,but again,this risk prediction model must include genetic factors and traditional risk factors.Therefor,more gene genetic variates need to be identified for improving the risk prediction models for liver cancer.