Abstract
BACKGROUND: Nearly half of patients with prostate cancer will develop metastasis. Immunotherapy is currently a promising strategy for treating metastatic prostate cancer. This study aimed to construct an immune subtyping system and provide a more comprehensive understanding of tumor microenvironment. METHODS: Data were downloaded from TCGA database and cBioPortal database. Consensus clustering was used to identify immune subtypes. Immune features were scored by ESTIMATE and CIBERSORT. Efficacy of different subtypes in immunotherapy was predicted by TIDE tool. Immune landscape was delineated through "monocle." Coexpressed gene modules were identified by weighted correlation network analysis. Univariate Cox regression analysis and LASSO analysis were applied to construct a prognostic model. RESULTS: Four immune subtypes (IS1 to IS4) were identified. Prognosis, mutation patterns, expression of immune genes, immune biomarkers, immunohistochemical biomarkers, and prediction efficacy of immunotherapy were significantly different among four immune subtypes. Five coexpressed gene modules were identified and an 11-gene prognostic model was constructed based on the modules. CONCLUSIONS: The study developed a novel immune subtyping system and an 11-gene prognostic model of prostate cancer, which could guide personalized treatment and immunotherapy for patients with prostate cancer.