A prognostic signature consisting of metabolism-related genes and SLC17A4 serves as a potential biomarker of immunotherapeutic prediction in prostate cancer

Abstract

Background: Prostate cancer (PCa), a prevalent malignant cancer in males worldwide, screening for patients might benefit more from immuno-/chemo-therapy remained inadequate and challenging due to the heterogeneity of PCa patients. Thus, the study aimed to explore the metabolic (Meta) characteristics and develop a metabolism-based signature to predict the prognosis and immuno-/chemo-therapy response for PCa patients.

Methods: Differentially expressed genes were screened among 2577 metabolism-associated genes. Univariate Cox analysis and random forest algorithms was used for features screening. Multivariate Cox regression analysis was conducted to construct a prognostic Meta-model based on all combinations of metabolism-related features. Then the correlation between MetaScore and tumor was deeply explored from prognostic, genomic variant, functional and immunological perspectives, and chemo-/immuno-therapy response. Multiple algorithms were applied to estimate the immunotherapeutic responses of two MeteScore groups. Further in vitro functional experiments were performed using PCa cells to validate the association between the expression of hub gene SLC17A4 which is one of the model component genes and tumor progression. GDSC database was employed to determine the sensitivity of chemotherapy drugs.

Results: Two metabolism-related clusters presented different features in overall survival (OS). A metabolic model was developed weighted by the estimated regression coefficients in the multivariate Cox regression analysis (0.5154*GAS2 + 0.395*SLC17A4 - 0.1211*NTM + 0.2939*GC). This Meta-scoring system highlights the relationship between the metabolic profiles and genomic alterations, gene pathways, functional annotation, and tumor microenvironment including stromal, immune cells, and immune checkpoint in PCa. Low MetaScore is correlated with increased mutation burden and microsatellite instability, indicating a superior response to immunotherapy. Several medications that might improve patients` prognosis in the MetaScore group were identified. Additionally, our cellular experiments suggested knock-down of SLC17A4 contributes to inhibiting invasion, colony formation, and proliferation in PCa cells in vitro.

Conclusions: Our study supports the metabolism-based four-gene signature as a novel and robust model for predicting prognosis, and chemo-/immuno-therapy response in PCa patients. The potential mechanisms for metabolism-associated genes in PCa oncogenesis and progression were further determined.

Keywords: immune infiltration; immuno-/chemotherapy response; metabolism; prognostic model; prostate cancer.