Immune-related gene signature for predicting biochemical recurrence after RP in prostate cancer subtypes.

The prognostic role of molecular subtypes of prostate cancer (PCa) patients is still unclear. And we speculated whether the change of immunologic and hallmark gene sets of adjacent notumor tissues could promote the BCR of PCa patients. We intended to identify PCa subtypes and find significant gene sets based on the activity changes of immunologic and hallmark gene sets in tumor and nontumor tissues to predict biochemical recurrence (BCR)-free survival. In addition, cell-based assays were utilized to validate the function of the gene in PCa. We calculated and analyzed these gene sets related to BCR in tumor and notumor tissues through gene set variation analysis (GSVA) and then found three clinically relevant subtypes of PCa by nonnegative matrix factorization method (NMF) in immunologic and hallmark gene sets. Patients with subtype 2 had better BCR-free survival than other subtypes. Using the least absolute shrinkage and selection operator method (LASSO), two prognostic gene sets in tumor but not in notumor tissue were identified frequently. Functional enrichment analysis revealed that genes from these two significant gene sets closely correlated with cell cycle in tumor tissue. Then, we applied random forest (RF) and artificial neural network (ANN), to identify gene In review biomarkers in predicting BCR and we found 36 genes related to the high risk of BCR. CENPE is the only common gene and can be used as a prognostic gene of BCR. Cell-based assays validated that CENPF promoted tumor proliferation in PCa. Three clinically relevant subtypes of PCa were identified. Two significant genes sets were related to the cell cycle and ATPase activity. And CENPF promotes tumor proliferation as an oncogene.