A Metabolism-Related Gene Landscape Predicts Prostate Cancer Recurrence and Treatment Response.

Background: Prostate cancer (PCa) is the most common malignant tumor in men. Although clinical treatments of PCa have made great progress in recent decades, once tolerance to treatments occurs, the disease progresses rapidly after recurrence. PCa exhibits a unique metabolic rewriting that changes from initial neoplasia to advanced neoplasia. However, systematic and comprehensive studies on the relationship of changes in the metabolic landscape of PCa with tumor recurrence and treatment response are lacking. We aimed to construct a metabolism-related gene landscape that predicts PCa recurrence and treatment response. Methods: In the present study, we used differentially expressed gene analysis, protein-protein interaction (PPI) networks, univariate and multivariate Cox regression, and least absolute shrinkage and selection operator (LASSO) regression to construct and verify a metabolism-related risk model (MRM) to predict the disease-free survival (DFS) and response to treatment for PCa patients. Results: The MRM predicted patient survival more accurately than the current clinical prognostic indicators. By using two independent PCa datasets (International Cancer Genome Consortium (ICGC) PCa and Taylor) and actual patients to test the model, we also confirmed that the metabolism-related risk score (MRS) was strongly related to PCa progression. Notably, patients in different MRS subgroups had significant differences in metabolic activity, mutant landscape, immune microenvironment, and drug sensitivity. Patients in the high-MRS group were more sensitive to immunotherapy and endocrine therapy, while patients in the low-MRS group were more sensitive to chemotherapy. Conclusions: We developed an MRM, which might act as a clinical feature to more accurately assess prognosis and guide the selection of appropriate treatment for PCa patients. It is promising for further application in clinical practice.