Identification of Potential Key Genes for Pathogenesis and Prognosis in Prostate Cancer by Integrated Analysis of Gene Expression Profiles and the Cancer Genome Atlas

<b>Background:</b> Prostate cancer (PCa)is a malignancy of the urinary system with a high incidence, which is the second most common male cancer in the world. There are still huge challenges in the treatment of prostate cancer. It is urgent to screen out potential key biomarkers for the pathogenesis and prognosis of PCa. <b>Methods:</b> Multiple gene differential expression profile datasets of PCa tissues and normal prostate tissues were integrated analysis by R software. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of the overlapping Differentially Expressed Genes (DEG) were performed. The STRING online database was used in conjunction with Cytospace software for protein-protein interaction (PPI) network analysis to define hub genes. The relative mRNA expression of hub genes was detected in Gene Expression Profiling Interactive Analysis (GEPIA) database. A prognostic gene signature was identified by Univariate and multivariate Cox regression analysis. <b>Results:</b> Three hundred twelve up-regulated genes and 85 down-regulated genes were identified from three gene expression profiles (GSE69223, GSE3325, GSE55945) and The Cancer Genome Atlas Prostate Adenocarcinoma (TCGA-PRAD) dataset. Seven hub genes (FGF2, FLNA, FLNC, VCL, CAV1, ACTC1, and MYLK) further were detected, which related to the pathogenesis of PCa. Seven prognostic genes (BCO1, BAIAP2L2, C7, AP000844.2, ASB9, MKI67P1, and TMEM272) were screened to construct a prognostic gene signature, which shows good predictive power for survival by the ROC curve analysis. <b>Conclusions:</b> We identified a robust set of new potential key genes in PCa, which would provide reliable biomarkers for early diagnosis and prognosis and would promote molecular targeting therapy for PCa.