Construction and validation of a novel diagnostic model with palmitoylation-related genes for prostate cancer

Prostate cancer (PC) continues to represent a significant contributor to male cancer mortality worldwide, necessitating the discovery of innovative diagnostic indicators and molecular targets. Our investigation utilized computational biology approaches combining multi-omics analysis with machine intelligence to elucidate the role of palmitoylation-related genes (PRGs) in PC pathogenesis and prognosis. By harmonizing transcriptomic datasets from TCGA and GEO repositories, we identified dysregulated PRGs and stratified PC into two molecularly distinct subtypes via unsupervised clustering. These subtypes exhibited divergent clinical outcomes, immune microenvironment heterogeneity (e.g., Dendritic cells, CD8⁺ T cells, and Macrophages infiltration), and distinct drug sensitivity profiles. Single-cell RNA sequencing further localized key PRGs-ZDHHC2, ZDHHC5, ZDHHC15, ZDHHC9, and LYPLA1-within tumor cell populations, linking their expression to immune evasion and metabolic reprogramming. A robust diagnostic model, integrating 101 machine learning algorithms, demonstrated high predictive accuracy for survival and immunotherapy response. Functional validation in DU145 cells confirmed that modulating these PRGs significantly suppressed proliferation and colony formation, highlighting their pathobiological relevance. Collectively, this multidimensional analysis delineates a comprehensive framework for understanding palmitoylation-driven oncogenesis and establishes a precision medicine toolkit for risk stratification and treatment optimization in PC.

The online version contains supplementary material available at 10.1007/s10616-025-00842-3.