Construct A MiRNA Model To Predict Prognosis In Esophageal Cancer Patients Based On TCGA Database

Objective: Bioinformatics method is used to construct a miRNA model related to the prognosis of patients with esophageal cancer,and its application value is discussed.Methods: The retrospective cohort study was conducted.The miRNA sequencing data and clinical data of esophageal cancer patients were downloaded from TCGA ranging from establishment of the database to September 2020.197 samples were collected including 184 esophageal cancer tissues and 13 adjacent tissues.We conducted differentially expressed miRNA between esophageal cancer and adjacent tissues and matched miRNA sequencing data and clinical data of esophageal cancer samples,which were randomly divided into training set(n=100)and test set(n=84).The training set was used to construct a model and the test set was used to evaluate performance of the model.We conducted LASSO-COX analysis based on the training set to construct a miRNA model.The performance of the model was validated in the training set and test set using AUC and C-index.Kaplan-meier method was used to draw survival curve and calculate survival rate.Log-rank test was used for survival analysis.Univariate analysis and multivariate analysis were applied between risk score of the model and clinical prognostic information described as HR and 95%CI.HR<1 indicated the factor as a protective factor,HR>1 indicated the factor as a risk factor and HR=1indicated no influence on survival.Results:1.Differentially expressed miRNAsA total of 132 differentially expressed miRNAs were screened,including86 up-regulated and 46 down-regulated in esophageal cancer tissues.2.Consruction of the miRNA modelThe miRNA model was expressed as,risk score=hsa-mir-1269 expression level×0.144+hsa-mir-3651 expression level×0.232+hsa-mir-641 expression level×0.320+hsa-mir-935 expression level×0.268.All these 4miRNAs were up-regulated miRNAs in esophageal cancer tissues,and the differential expression multiples in adjacent tissues were 14.84,2.92,3.27,and 3.65 times,respectively.3.Validation and performance evaluation of the miRNA modelIn the training set and the test set,the C-index of the miRNA model was0.759 and 0.623.ROC curve and AUC showed that the model predicted the1-year,3-year and 5-year survival rates of patients with esophageal cancer in TCGA database were 0.738,0.698 and 0.811.4.Comparison of miRNA model and TNM stageIn the training set and the test set,the prediction probability of miRNA model and TNM stage of the 8th edition for 2-year survival time of esophageal cancer patients were 0.823,0.756 and 0.753,0.736.In low TNM stage and high TNM stage,the miRNA model and the 8th edition TNM stage predicted the 2-year survival time of esophageal cancer patients were 0.72,0.574 and0.903,0.548.5.Analysis of relevant factors affecting the prognosis of patients with esophageal cancerUnivariate analysis showed that TNM stage,N stage,M stage,grade and miRNA model were related factors for the prognosis of patients with esophageal cancer.Multivariate analysis showed that TNM stage and miRNA model were independent risk factors for prognosis of patients with esophageal cancer.6.miRNA target gene prediction and enrichment analysisThe target genes of hsa-mi R-1269 a,hsa-mi R-1269 b,hsa-mi R-3651,hsa-mi R-641,hsa-mi R-935 were 98,129,442,343 and 78,respectively.Bioprocess analysis of GO enrichment showed that target genes were mainly involved in biological processes such as cell adhesion and various organs development.Cell composition analysis showed that the target genes were mainly concentrated in the ubiquitin-protease system,organelle and plasma membrane components.Molecular function showed that the target genes were mainly concentrated in binding to histone acetyltransferase and Smad protein.KEGG pathway enrichment analysis showed that target genes were mainly involved in PI3K/Akt,c AMP,catabolism autophagy and other signaling pathways.Conclusions: The miRNA model based on TCGA database,consisting of hsa-mir-1269,hsa-mir-3651,hsa-mir-641 and hsa-mir-935,has good sensitivity and specificity,and can be used as a reliable biomarker to predict the prognosis of esophageal cancer patients.It is superior to TNM stage in predicting the prognosis of patients with esophageal cancer,and this model can also stratify and predict the prognosis of patients with the same clinical stage.