A Multigene Prognostic Model Of Colorectal Cancer Was Constructed Based On M6A Methylation And Epithelial Interstitial Transformation

Objective:Colorectal cancer(CRC)is a commonly diagnosed digestive malignancy worldwide,with its incidence ranking the second in China among all the cancers.Despite of the fact that precancerous screening and improved therapeutic techniques have significantly prolonged the survival of CRC patients,a considerable proportion was initially diagnosed as advanced stage and suffered from poor prognosis.Previous studies have demonstrated that CRC development is a complicated process involving various molecular events,therefore identification and validation of molecular biomarkers is crucial for the early diagnosis and prognosis evaluation of CRC.N6-methyladenosine(m6A)methylation modification is defined as the methylation at the N6 position of adenosine and suggested to exert its biological roles through regulating mRNAs and non-coding RNAs.Epithelial-mesenchymal transition(EMT)is a biological process where epithelial cells acquire mesenchymal characteristics.Recent studies have suggested both the events play a crucial role in the occurrence and development of CRC.In this study,using bioinformatics methods,we aimed to construct and validate a prognostic risk model based on selected m6A methylation modification and EMT related molecules,which is hoped to provide novel insights into the biomarker-based CRC prognosis evaluation.Methods:The mRNA expression profile data and Copy number variation of clinical samples were obtained from the Cancer Genome Atlas(TCGA)database.The gene expression information of four cohorts(GSE39582,GSE103479,GSE14333 and GSE38832)were downloaded from GEO database(https://www.ncbi.nlm.nih.gov/geo/).The data of 1184 EMT related genes were obtained from dbEMT database(http://www.dbemt.bioinfo-minzhao.org).The clinical information for immunotherapy-treated bladder cancer were obtained from IMvigor210CoreBiologies dataset(http://researchpub.gene.com/IMvigor210CoreBiologies/IMvigor210.).Weighted gene co-expression network analysis(WGCNA)was used to select the co-expression modules related with m6A methylation modification.Cox analysis was used to identify hub genes related with m6A methylation modification.Using the nine hub genes,the CRC samples were classified into two clusters and the transcriptome,immune infiltration and mutations etc.were compared.Moreover,a prognostic risk model was constructed using these hub genes and validated in GEO cohorts.The correlations of the model with clinical features,immune infiltration,stem cell phenotype and immunotherapy etc.were also analyzed.Results:First,based on 1184 EMT related genes,9 hub genes were selected using the WGCNA method as follows:YAP1,FAM3C,NUBPL,GLO1,JARID2,NFKB1,CDKN1B,HOOK1 and GIPC2.The enriched pathway analysis revealed these genes were correlated with several pathways such as hypoxia-inducible factor and cellular proliferation pathways.Then,consistent clustering was performed using the expression patterns of these genes.As a result,the TCGA-GSE39582 samples were divided into Cluster 1 and 2.Significant differences of the proportions of immune infiltration and clinical efficacy of immune checkpoint inhibitors(ICIs)were observed between the clusters,while the opposite was for mutation counts.Then,we performed principal component analysis for the expression of 9 hub genes in the TCGA+GSE39582 cohort,and defined the first principal component as the risk scores.Based the scores,the samples were divided into high and low risk groups.The survival analysis suggested patients with high risk scores had a significantly worse outcome than those with low risk scores.This prognostic stratifying performances were well validated in subgroup analysis for colorectal adenocarcinoma,colon adenocarcinoma,Ⅰ-Ⅱ stage,Ⅲ-Ⅳ stage,patients older than 60 years,males and females.The immune and stroma scores of high risk group were significantly less than that of low risk group.Significant differences of the proportions of immune infiltration could be observed between the groups.With regard to T stage,CRC patients within T3/4 stage had a significantly more risk scores than those within T1/2 stage.Finally,the prognostic stratifying performance of the model was well validated in the GSE103479 and GSE14333 cohort.Conclusion:In this study,nine hub genes related with m6A methylation and EMT phenotype were selected to establish a prognostic risk model for CRC patients.The performance of this risk model was well validated in GEO cohorts.This model will not only effectively identify high risk CRC subpopulations with poor outcome,but also have the potential to help predict the clinical efficacy of ICI therapy.However,considering the inherent limitations,multicentre clinical validations based on large samples and mechanism investigations for hub genes are still of great necessity.