Identifying a 6-Gene Prognostic Signature for Lung Adenocarcinoma Based on Copy Number Variation and Gene Expression Data

Abstract

The occurrence of lung adenocarcinoma (LUAD) is a complicated process, involving the genetic and epigenetic changes of proto-oncogenes and oncogenes. The objective of this study was to establish new predictive signatures of lung adenocarcinoma based on copy number variations (CNVs) and gene expression data. Next-generation sequencing was implemented to obtain gene expression and CNV information. According to univariate, multivariate survival Cox regression analysis, and LASSO analysis, the expression profiles of lung adenocarcinoma patients were screened and a risk score formula was established and experimentally validated in a local cohort. The model was evaluated by three independent cohorts (TCGA-LUAD, GSE31210, and GSE30219), and then validated by clinical samples from LUAD patients. A total of 844 CNV-related differentially expressed genes (CNV-related DEGs) were identified. These genes are significantly associated with the imbalance of various oxidative stress pathways. A CNV-associated-six gene signature was dramatically linked to overall survival in lung adenocarcinoma samples from both training and validation groups. Functional enrichment analysis further revealed involvement of genes in p53 signaling pathway and cell cycle as well as the mismatch repair pathway. Risk score is an independent marker considering clinical parameters and had better prediction in clinical subpopulation. The same signature also classified tumor tissues of clinical patients with CNV detected from their corresponding nontumorous tissues with an accuracy of 0.92. In conclusion, we identified a new class of 6 CNV-related gene markers that may act as efficient prognostic predictors of lung adenocarcinoma, thus contributing to individualized treatment decisions in patients.