Evaluation Of An Immune-Related Gene Prediction Model On The Prognosis Of Gastric Adenocarcinoma

Purpose: Gastric adenocarcinoma(GAC)is the most common malignant tumor and one of the main causes of death caused by cancer in the world.As a new type of tumor therapy,immunotherapy has great potential in clinical application.At present,some factors related to tumorigenesis and prognosis have been reported,but the specific mechanism is not clear.There is a real hunger that we need to find new molecular biomarkers that can accurately indicate the stage of disease progression and predict clinical outcomes.The purpose of this paper is to establish an accurate prediction model based on immune-related genes(IRGs)to provide new benefits for GAC immunotherapy.Method:1.Download GAC expression profile data and clinical data from The Cancer Genome Atlas(TCGA)and Gene expression Database(GEO).Combined with pathologically confirmed GAC patients(n=145)in the Department of Colorectal surgery of the National Cancer Center from June 2012 to August 2014 and IRGs in Imm Port database,differentially expressed genes are screened.Unsupervised cluster analysis is used to divide the patients in the training set into two groups.The difference in prognosis between the two groups is evaluated by Kaplan-Meier(K-M)analysis.2.The difference of IRGs expression between GAC and paracancerous tissues in the training set is analyzed,and the P value is corrected by Benjamini Hochberg method.Only genes with the value of False Positive Rate(FDR)less than 0.05 and the change of expression more than 2 times are identified as significantly differentially expressed genes.Nine IRGs are used to construct a prediction model,and GAC patients in the training set are divided into high-risk group and low-risk group(based on the median risk score).K-M analysis is applied to evaluate the prognosis of the two groups,and the effectiveness of the predictive model is evaluated by calculating the area under(AUC)the Receiver Operating Characteristic(ROC)curve.3.K-M analysis and AUC of ROC curve are used to evaluate the effectiveness of the predictive model in different clinical types of the verification set.4.The relationship between IRGs in the prediction model and immune checkpoint,proportion of immune cells is analyzed visually by maftool package of R.Use K-M analysis to evaluate the difference of tumor mutational burden(TMB)between high risk group and low risk group of GAC patients.5.The effect of elevated risk score on GAC prognosis is evaluated by using univariate Cox regression analysis,multivariate Cox regression analysis and meta-analysis for various factors affecting GAC.6.The method of GSEA analysis is applied to explore the important KEGG pathways related to the prediction model.Result:1.Through the comprehensive analysis of the training set,it is found that 9 IRGs are independently related to the prognosis of GAC,including ADM,APOD,CXCR4,ITGAV,Nrp1,RFX5,STC1,TAP1 and ZC3HAV1.A predictive model is constructed based on 9 IRGs,and the training set of GAC patients is divided into two groups of high and low risk(bounded by the median value of the risk score).K-M survival analysis shows that the overall survival time(OS)of high-risk patients is well below that of low-risk patients(P< 0.0001).At the same time,the AUC of 3-years survival time and 5-years survival time ROC curve of the predictive model is larger than that of TNM pathological stage.2.In the validation set,the validation of GAC patients grouped by predictive model can also conclude that the OS of the high-risk group is well below that of the low-risk group.In the quantitative reverse transcription polymerase chain reaction(q RT-PCR)group,the AUC of 3-years survival time and 4-years survival time ROC curve is higher than that of TNM pathological stage.At the same time,it is verified that there is a significant difference in the expression of IRGs in the predictive model between GAC tissues and adjacent tissues.There is also significant difference in the expression of IRGs in the predictive model between the high risk group and the low risk group.In different differentiation degree,pathological stage,sex,age and TCGA classification,the overall survival time of the high risk group is worse than that of the low risk group.In patients with IV stage GAC,the sample size may be too small to draw the same conclusion.3.In the GSE26253 data set,the Recurrence Free Survival(RFS)of patients with advanced disease is significantly shorter than that of patients with early disease(P < 0.0001),and the RFS of patients with low risk is significantly superior to that of patients with high risk(P < 0.0025).4.The analysis of GAC patients in TCGA dataset and GSE84437 shows that the OS of patients with high stromal score is superior to that of patients with low stromal score.The OS of patients with high score of immune cells is superior to that of patients with low score of immune cells.There are significant differences in the proportion of immune cells between the high risk group and the low risk group,such as M1 macrophages,monocytes,CD8 T cells,follicular helper T cells,memory CD4 T cells and M2 macrophages.At the same time,there is a correlation between IRGs of the predictive model and these six kinds of cells.In the high risk group,the expression of VTCN1,ENTPD1 and FGL1 is significantly up-regulated,while the expression of LGALS9 is significantly down-regulated.There is a good correlation between IRGs of the predictive model and differentially expressed immune checkpoints,especially TAP1 and CXCR4.The OS of patients with high TMB is significantly superior to that of patients with low TMB(P< 0.036).The TMB of high risk score group is well below that of low risk score group(P < 0.0001).The risk score of high TMB group is well below that of low TMB group(P < 0.0001).The low mutation rate of 9 IRGs in the predictive model supports them as biomarkers of diagnosis or prognosis.5.In GAC patients from TCGA,age(> 62 years old,HR=1.89,P < 0.001),MUC type(HR=0.25,P =0.02),stage IV(HR=3.86,P =0.0005),stage III(HR=2.28,P =0.017)and increasing risk score(HR=2.31,P < 0.0001)are independent risk factors for GAC prognosis.The same conclusion can be obtained in patients with GSE84437(P < 0.0037)and q RT-PCR(P < 0.0001).Through Meta analysis of the data of all GAC patients in the literature,it is also proved that the increasing risk score is an independent risk factor for GAC prognosis(HR=2.218,95%CI=1.804-2.727,P < 0.0001).6.The pathways related to the prediction model include cell adhesion molecules(CAMs),the MAPK signaling pathway,DNA replication,nucleotide excision repair,the cell cycle,cytokine-receptor interaction,the P53 signaling pathway,mismatch repair,ECM-receptor interaction.Conclusion:1.The prediction model based on 9 IRGs can accurately achieve risk stratification for GAC patients.2.This predictive model can be applied to predict the prognosis of OS in different clinical subgroups,as well as RFS.3.Low immune cell score and low TMB may indicate poor overall survival in patients with GAC.The OS of GAC patients with significantly up-regulated VTCN1,ENTPD1 and FGL1 expression and significantly down-regulated LGALS9 expression at immune checkpoints may be poor.4.The independent risk factor for GAC prognosis is the increasing risk score.