Bioinformatics Analysis and Identification of Potential Genes Associated with Pathogenesis and Prognosis of Gastric Cancer.

Abstract

Gastric cancer (GC) is a deadly cancer and a challenging public health problem globally. This study aimed to analyze potential genes associated with pathogenesis and prognosis of gastric cancer. This work selected the overlapping differentially expressed genes (DEGs) in GC from four datasets, the GSE29272, GSE29998, GSE54129 and GSE118916 Gene Expression Omnibus databases. These DEGs were used to carry out comprehensive bioinformatic analysis to analyze the related functions and pathways enriched, the relative expression levels and immune infiltrates, the prognostic characteristics and the interaction network. In total, 55 DEGs increased while 98 decreased in their expression levels. For those DEGs with increased expression, they were mostly concentrated on "focal adhesion" and "ECM-receptor interaction", whereas DEGs with decreased expression were mostly associated with "gastric acid secretion" and "drug metabolism cytochrome P450". MCODE and ClueGO results were then integrated to screen 10 hub genes, which were FN1, COL1A1, COL3A1, BGN, TIMP1, COL1A2, LUM, VCAN, COL5A2 and SPP1. Survival analysis revealed that higher expression of the ten hub genes significantly predicted lower overall survival of GC patients. TIMP1 was most significantly related to neutrophils, CD8+ T cells, as well as dendritic cells, while LUM was most significantly related to macrophages. Immunohistochemistry results and functional testing showed that the expression of COL5A2 was elevated in GC and that it might be a key gene in GC tumorigenesis.