The Development of a Gleason Score-Related Gene Signature for Predicting the Prognosis of Prostate Cancer.

The recurrence of prostate cancer (PCa) is intrinsically linked to increased mortality. The goal of this study was to develop an efficient and reliable prognosis prediction signature for PCa patients. The training cohort was acquired from The Cancer Genome Atlas (TCGA) dataset, while the validation cohort was obtained from the Gene Expression Omnibus (GEO) dataset (GSE70769). To explore the Gleason score (GS)-based prediction signature, we screened the differentially expressed genes (DEGs) between low- and high-GS groups, and then univariate Cox regression survival analysis and multiple Cox analyses were performed sequentially using the training cohort. The testing cohort was used to evaluate and validate the prognostic model's effectiveness, accuracy, and clinical practicability. In addition, the correlation analyses between the risk score and clinical features, as well as immune infiltration, were performed. We constructed and optimized a valid and credible model for predicting the prognosis of PCa recurrence using four GS-associated genes (SFRP4, FEV, COL1A1, SULF1). Furthermore, ROC and Kaplan-Meier analysis revealed a higher predictive efficiency for biochemical recurrence (BCR). The results showed that the risk model was an independent prognostic factor. Moreover, the risk score was associated with clinical features and immune infiltration. Finally, the risk model was validated in a testing cohort. Our data support that the GS-based four-gene signature acts as a novel signature for predicting BCR in PCa patients.