Association Between Common Genetic Variations And Early-onset Gastric Cancer Risk

Gastric cancer(GC)is one of the most common malignant tumors in the world.Although its morbidity and mortality have been decreasing,the incidence of early-onset gastric cancer(EOGC)is increasing.EOGC is defined as any GC diagnosed at the age of 50 or earlier.Compared with non-EOGC,genetic factors may play a more important role in the development of EOGC.Previous studies have found many gene mutations associated with EOGC,including CDH1 missense mutation or deletion,loss-of-function of CTNNA1,TP53 mutation,which indicated that functional mutations of key genes may lead to the development of EOGC.In recent years,genome-wide association studies(GWAS)have discovered more than 10 susceptible regions associated with GC risk.These genetic variants have high frequency in the population,often locate in the non-coding regions and affect the development of GC through complex regulatory mechanisms.However,whether they affect the age at onset of GC or EOGC risk is still unknown.Based on the largest GWAS of GC in Chinese populations,we analyzed the association of susceptible variants and polygenic risk score of GC with EOGC risk.Then,we conducted a genome-wide association analysis to identify the specific common genetic variants susceptible to EOGC and constructed a polygenic risk score for EOGC to assess EOGC risk.Therefore,we expect to add reliable genetic markers for predicting risk and identifying high-risk populations of GC and for better implementation of GC tertiary prevention.Part I:Association between polygenic risk score and age at onset of gastric cancer[Background]In our previous study,we constructed a polygenic risk score(PRS)derived from 112 single nucleotide polymorphisms(SNPs)assocaited with GC risk in Chinese population,and found that the PRS can predict the risk of GC based on a prospective cohort and can be used as a measurable indicator for revealing the etiology and identify the population at high genetic risk of GC.However,the association between PRS of conventional GC and EOGC has not been clear.Therefore,we explored the association between PRS and age at onset and early-onset risk of GC in this study.[Methods]Gastric cancer cases from existing GWAS were included.Based on co-dominant,dominant and additive genetic models,the association between 12 SNPs susceptible to GC(genome-wide association level P<5×10-8)and age at onset of GC were analyzed.Meanwhile,a PRS was constructed based on 112 SNPs with the strongest cumulative effect associated with GC risk.The PRS was categorized into low(quintile 1),intermediate(quintile2-4)and high genetic risk(quintile 5)and used to depict the relationship between PRS and age of GC onset.The method of t test was used in the comparison of the average numbers of two groups,while analysis of variance was used to compare the differences between multiple groups.Pearson correlation test were used to analyzed the correlation between two variates.In addition,cases diagnosed before 50 years old were defined as EOGC.cox proportional hazard model was used to test the association between 12 SNPs and EOGC risk with early-onset age as the timescale,And Also,the relationship between PRS and EOGC risk were analyzed with low genetic risk as the reference group[Results]At single genetic variants level,the results showed that variants in 1q22(rs7366775,rs72706172,rs138554234)and 8q24.3(rs10105842,rs2585177)were negatively correlated with the age of GC onset,and the risk of EOGC increased with the increase of the number of risk alleles.Besides,rs7366775 in 1q22 not only had the strongest and most significant correlation with the age of GC onset(r=-0.06,P=2.54×10-08),but also had the strongest correlation with EOGC risk(HR=1.28,95%CI:1.14～1.44,P=3.09×l0-05).At PRS level,the PRS was also negatively correlated with age of GC onset(r=-0.05,P<0.001),and significantly associated with the risk of EOGC in a dose-response manner(intermediate genetic risk:HR=1.19,95%CI:1.03～1.39,P=0.022;high genetic risk:HR=1.44,95%CI:1.20-1.71,P<0.001;Ptrend=4.59×10-05).[Conclusion]In conclusion,common variants of GC also have a great effect on the risk of EOGC.PRS may contribute to the risk of both GC and EOGC.PRS can be used as a measurable indicator for risk prediction for occurrence and early-onset of GC,and to guide the primary and secondary prevention of GC.Part Ⅱ:Genome-wide association study of early-onset gastric cancer[Background]EOGC usually has no warning symptoms or diagnostic indicators at the early stage.Additionally,young people are not included in the target population of GC screening and the medical examination rate and subjective visiting rate of them are low,which lead to the diagnosis of GC at an advance stage.Therefore,there are still great obstacles in the prevention and treatment of EOGC.Current studies mainly focus on CDH1 genetic mutation in hereditary EOGC,while very few studies focused on the genetic susceptibility of sporadic EOGC at the genome-wide level.Therefore,we systematically evaluated the association between genetic loci and EOGC risk to identify common genetic variations susceptible to EOGC.[Methods]Based on case-control design,all samples from 6 existed GWASs of GC were included.Among them,cases aged<50 years were defined as EOGC while those aged≥50 years were defined as late-onset gastric cancer(LOGC),including East-GWAS(577 EOGCs,4295 LOGCs and 3397 controls),North-GWAS(292 EOGCs,1319 LOGCs and 2091 controls),BJ-GWAS(120 EOGCs,336 LOGCs and 1118 controls),NJ-GWAS(127 EOGCs,423 LOGCs and 1155 controls),Onco-GWAS(155 EOGCs,985 LOGCs and 1053 controls)and SX-GWAS(241 EOGCs,1384 LOGCs and 2100 controls).Firstly,East-GWAS,North-GWAS,BJ-GWAS,NJ-GWAS and Onco-GWAS were used as discovery datasets.On the basis of additive genetic model,cox proportional hazards models were used to conduct GWASs on EOGC risk in EOGCs and controls,and GWASs on early-onset risk of GC in EOGCs and LOGCs with 50 years as cutoff for observation and early-onset age as timescale,and multivariate logistics regression was used to conduct GWASs on LOGC risk in LOGCs and controls.Then,a meta-analysis was performed and independently susceptible variants associated with EOGC risk and early-onset risk of GC at P<1×10-4 and in no association with LOGC risk at P>0.05 were indentified.Next,a PRS for EOGC was constructed based on candidate genetic variants,and the associations between PRS and EOGC risk were analyzed in all single and combined discovery datasets based on cox proportional hazards models,and validated in SX-GWAS based on multivariate logistics regression.Meanwhile,functional annotation was carried out for regions susceptible to EOGC,where candidate genetic variants located,to identify functional regions with activity of transcriptional regulation.And we identified functional genes within these regions based on GTEx data.Furthermore,we conducted gene-based analysis with MAGMA to validated the association between functional genes and EOGC and explore the potential mechanism of EOGC tumorigenesis.[Results]The results of GWASs showed that 14 independent SNPs were associated with the risk of EOGC.Among them,rs58003373 in 2q12.1 was in the strongest association with EOGC risk compared with LOGCs and normal population.The frequency of T allele of rs58003373 in EOGCs,LOGCs and normal population was 0.355,0.306 and 0.319,respectively.T allele of rs58003373 could increase the risk of EOGC by 25%(HR:1.25,95%CI:1.15～1.36,P=2.06×10-7).A EOGC-specific PRS was constructed based on the 14 genetic loci found above,and participants were divided into low,medium and high genetic risk groups according to the PRS quintiles.The results showed that the risk of EOGC increased with the increase of genetic risk in any discovery dataset or combined dataset(all P<0.05).A similar dose-response relationship was also observed in SX-GWAS(intermediate genetic risk:OR=1.03,95%CI=1.01～1.05,P=0.003;high genetic risk:OR=1.03,95%CI=1.00～1.05,P=0.047;Ptrend=0.047).After multi-stage annotation and validation of the regions susceptible to EOGC,we finally found that genetic variations in 1p34.1 and 10q26.13 may affect the development of EOGC by regulating the gene expression of HYI-AS1,PLK3 and LHPP.[Conclusion]The higher PRS,which constructed with 14 specific variants susceptible to EOGC,was associated with higher EOGC risk.This score can be combined with polygenic risk score of conventional GC to indentify populations at high risk of GC and guide individualized precise prevention,early diagnosis and early treatment.Meanwhile,this study revealed the important role of common genetic variations in the development of EOGC,which provide evidence for further exploration of the pathogenesis of EOGC.