Identification of genome-wide copy number variation-driven subtypes for the treatment and prognostic prediction of esophageal carcinoma

Abstract

BackgroundEsophageal carcinoma (ESCA) is a frequently detected gastrointestinal cancer. Copy number variants (CNVs) have a dramatic impact on the screening, diagnosis and prognostic prediction of cancers. However, the mechanism of action of CNVs on ESCA occurrence and progression remains unclear. MethodsESCA samples from The Cancer Genome Atlas (TCGA) were typed by consensus clustering using CNV-associated genes. Weighted Gene Co-Expression Network Analysis (WGCNA) was used to section gene modules closely related to the two clusters, and sub-networks were constructed as hub genes. In addition, seven prognosis-correlated genes were further screened and retained by multivariate Cox regression analysis to develop a prognostic assessment model. The ssGSEA algorithm assessed energy metabolism levels in patients from different clusters and risk groups. Finally, quantitative real-time PCR (qRT-PCR) and live-dead cell staining verified the expression of genes associated with CNV risk scores. ResultsESCA was classified into two subtypes based on CNV values. Compared with cluster 1, cluster 2 had significantly higher level of immune score and tumor-associated immune cell infiltration as well as a noticeably better overall survival. The three modules most associated with the two clusters were identified by WGCNA, and a prognostic model with a strong prediction performance was constructed with their genes. Glycolysis, lactate metabolism, fatty acid synthesis, glutathione, methionine, and tryptophan metabolic pathway enrichment scores were remarkably higher in patients in cluster 1 and the high-risk group than in cluster 2 and the low-risk group. ConclusionThe current research maybe provides new understanding for the pathogenesis of ESCA based on CNV, providing an effective guidance for its clinical diagnosis and prognostic evaluation.