Multivariate gene expression-based survival predictor model in esophageal adenocarcinoma

Abstract

Background: Despite the recent development of molecular-targeted treatment and immunotherapy, survival of patients with esophageal adenocarcinoma (EAC) with poor prognosis is still poor due to lack of an effective biomarker. In this study, we aimed to explore the ceRNA and construct a multivariate gene expression predictor model using data from The Cancer Genome Atlas (TCGA) to predict the prognosis of EAC patients.

Methods: We conducted differential expression analysis using mRNA, miRNA and lncRNA transciptome data from EAC and normal patients as well as corresponding clinical information from TCGA database, and gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of those unique differentially expressed mRNAs using the Integrate Discovery Database (DAVID) database. We then constructed the lncRNA-miRNA-mRNA competing endogenous RNA (ceRNA) network of EAC and used Cox proportional hazard analysis to generate a multivariate gene expression predictor model. We finally performed survival analysis to determine the effect of differentially expressed mRNA on patients' overall survival and discover the hub gene.

Results: We identified a total of 488 lncRNAs, 33 miRNAs, and 1207 mRNAs with differentially expressed profiles. Cox proportional hazard analysis and survival analysis using the ceRNA network revealed four genes (IL-11, PDGFD, NPTX1, ITPR1) as potential biomarkers of EAC prognosis in our predictor model, and IL-11 was identified as an independent prognostic factor.

Conclusions: In conclusion, we identified differences in the ceRNA regulatory networks and constructed a four-gene expression-based survival predictor model, which could be referential for future clinical research.

Keywords: Cox analysis; Kaplan-Meier analysis; differential expression genes; esophageal adenocarcinoma; functional enrichment analysis.