Identification of Potential Predictor of Biochemical Recurrence in Prostate Cancer

Prostate cancer is a common malignancy in men. Radical prostatectomy is one of the primary treatment modalities for patients with prostate cancer. However, early identification of biochemical recurrence is a major challenge for post-radical prostatectomy surveillance. There is a lack of reliable predictors of biochemical recurrence. The purpose of this study was to explore potential biochemical recurrence indicators for prostate cancer.

We analyzed transcriptomic data of cases with biochemical recurrence in The Cancer Genome Atlas (TCGA). Then, we performed integrative bioinformatics analyses to establish a biochemical recurrence predictor model of prostate cancer.

There were 146 differentially expressed genes (DEGs) between prostate cancer and normal prostate, including 12 upregulated and 134 downregulated genes. Comprehensive pathway enrichment analyses revealed that these DEGs were associated with multiple cellular metabolic pathways. Subsequently, according to the random assignment principle, 208 patients were assigned to the training cohort and 205 patients to the validation cohort. Univariate Cox regression analysis showed that 7 genes were significantly associated with the biochemical recurrence of prostate cancer. A model consisting of 5 genes was constructed using LASSO regression and multivariate Cox regression to predict biochemical recurrence of prostate cancer. Expression of PAH and AOC1 decreased with an increasing incidence of prostate cancer, whereas expression of DDC, LINC01436 and ORM1 increased with increasing incidence of prostate cancer. Kaplan-Meier curves and receiver operator characteristic (ROC) curves indicated that the 5-gene model had reliable utility in identifying the risk of biochemical recurrence of prostate cancer.

This study provides a model for predicting prostate cancer recurrence after surgery, which may be an optional indicator for postoperative follow-up.