Construction Of Individualized Risk Prediction Model For Gastric Cancer Of Chinese Han Population Based On Quantitative System Evaluation

BackgroundAs a common malignant tumor worldwide,gastric cancer（GC）is a serious public health problem,which ranks the fifth in incidence and the third in mortality of malignant tumor,and finally contribute to heavy burden.At present,the main way to reduce the mortality rate of GC is individualized prevention and treatment.While the development of predictive strategies and treatment relied on the knowledge of individual GC risk,which containing genetic basis,production environment,life style,as well as the medical and health services on the occurrence and development of gastric cancer.While the occurrence and development of GC was a result of many factors.Helicobacter pylori（Hp）infection,smoking,alcohol drinking,high-salt diet and so on are the identified risk factors for GC.In addition,other diseases and heredity also play important roles.Single nucleotide polymorphism（SNP）is the most studied genetic risk factor.Therefore,screening the risk factors of GC in Chinese Han population,making a summary analysis and quantitative evaluation,and finally constructing an individual risk assessment model of GC to predict the risk of individual GC could be a good way to reduce the incidence of GC,improve early diagnosis and lighten the cost of medical treatment.ObjectiveBy means of systematic evaluation and meta analysis,the comprehensive risk fraction of biological,environmental and genetic predisposing factors as well as the incidence probability of GC in Chinese Han population were collected and quantified,then the individualized prediction model of GC risk was constructed.Methods（1）The data of the study on the risk factors of gastric cancer published in Chinese Han population during 2006 to 2017 were collected by NoteExpress,then meta analysis was conducted with review manager 5.3 software to obtain the characteristics of biological、environmental and genetic factors and calculated the Odds Ratio（OR）and95%Confidence Interval（95%CI）.Publication bias was assessed by Stata 14.0 software.（2）The polygeneic risk model constructed by Demchuk was used to evaluate the joint contribution of multiple genetic factors to GC.The frequency and combined OR values were converted into the relative risk and the baseline incidence ratio.Then the risk fraction and combined risk fraction were determined.The risk assessment model of gastric cancer was established with the DataWindow function of PowerBuilder 9.0software to realize individual prediction of the GC risk.（3）The case-control study including 405 hospital confirmed GC patients and 405normal controls matched by age and sex was conducted.Polymerase chain reaction-restriction fragment length polymorphism（PCR-RFLP）was used for genotyping of rs751402 and rs17655 in ERCC5.Goodness-of fit chi-square（χ²）test was utilized to exam Hardy-Weinberg Equilibrium（HWE）in control group.Unconditional logistic regression was performed to consider the relationship of the two SNPs with GC risk with SPSS 21.0 software.Results（1）The biological risk factor of GC was Hp infection.The environmental risk factors included smoking,alcohol drinking,family history of GC,preserved food,irregular diet,hot food,high salt diet,smoked and fried food,history of stomach disease,mental depression.PSCA（rs2294008）CT、PSCA（rs2976392）AG,PRKAA1（rs13361707）CT+CC、PLCE1（rs2274223）GG+AG、GSTT1 deletion type、GSTM1deletion type、NAT2M1 CT、MTHFR 677 CT、TNF-β-252 AG、ERCC2-312（rs1799793）AA、IL-1β-511 TT、IL10-1082 AG、XRCC1-194（rs1799782）CT+TT、COX-2 765（rs20417）CG、COX-2 1195 AA、mir-146a（rs2910164）GG、CDH1 160 CA、ADPRT762 CC、IL17（rs2275913）AA+AG were concluded in genetic risk factors.（2）It was observed that the risk and frequency of multiple susceptibility genes were closely related to the OR value and the frequency of single gene in the population in the polygeneic risk model.It can be found that the risk distribution of combined genes moved towards the direction of greater risk degree,and the influence of a gene on the genetic risk of polygene can be evaluated effectively.The standard incidence ratio,risk fraction and combined risk fraction of the risk factors of GC were obtained,and the calculation method of the combined risk fraction was united with PowerBuilder software.Finally,an individual risk prediction model for gastric cancer in Chinese Han population was constructed.（3）The genotyping results showed that rs17655 in ERCC5 was associated with GC risk.Comparing with CC genotype,GG genotype could increase the risk of GC,while there was no significant relationship between rs751402 and GC.Stratified analysis showed that rs751402 could increase the risk of GC in people younger than 50 years of age(OR_adj:1.95,95%CI:1.09-3.49);rs17655 increased the risk of GC in persons over50(OR_adj:2.21,95%CI:1.35-3.61),males(OR_adj:1.78,95%CI:1.10-2.87),smokers(OR_adj:2.13,95%CI:1.19-3.84),drinkers(OR_adj:2.78,95%CI:1.37-5.65),as well as thoese without family history of cancer(OR_adj:1.81,95%CI:1.15-2.86).Conclusions（1）Hp infection,smoking,alcohol drinking,family history of GC,preserved food,irregular diet,hot food,high salt diet,smoked and fried food,history of stomach diseases,mental depression,PSCA（rs2294008）CT,PSCA（rs2976392）AG,PRKAA1（rs13361707）CT+CC,PLCE1（rs2274223）GG+AG,GSTT1 deletion genotype,GSTM1 deletion genotype,NAT2M1 CT,MTHFR 677 CT,TNF-β-252 AG,ERCC2-312（rs1799793）AA,IL-1β-511 TT,IL10-1082 AG,XRCC1-194（rs1799782）CT+TT,COX-2 765（rs20417）CG,COX-2 1195 AA,mir-146a（rs2910164）GG,CDH1 160CA,ADPRT 762 CC,IL17（rs2275913）AA+AG were all GC risk factors in Chinese Han population.（2）In this study,30 risk factors related to GC were taken into account.An individual risk assessment model of GC based on Chinese Han population was constructed to predict the risk of GC in populations with different risk factors.It was beneficial to the primary screening of high risk population and early diagnosis of GC.（3）The relationship between two SNPs and the risk of GC was found in the case-control study,which provided a basis for further improvement of the risk prediction model.