Integrative multi-omics analysis and machine learning refine global histone modification features in prostate cancer

Abstract

Background: Prostate cancer (PCa) is a major cause of cancer-related mortality in men, characterized by significant heterogeneity in clinical behavior and treatment response. Histone modifications play key roles in tumor progression and treatment resistance, but their regulatory effects in PCa remain poorly understood. Methods: We utilized integrative multi-omics analysis and machine learning to explore histone modification-driven heterogeneity in PCa. The Comprehensive Machine Learning Histone Modification Score (CMLHMS) was developed to classify PCa into two distinct subtypes based on histone modification patterns. Single-cell RNA sequencing was performed, and drug sensitivity analysis identified potential therapeutic vulnerabilities. Results: High-CMLHMS tumors exhibited elevated histone modification activity, enriched proliferative and metabolic pathways, and were strongly associated with progression to castration-resistant prostate cancer (CRPC). Low-CMLHMS tumors showed stress-adaptive and immune-regulatory phenotypes. Single-cell RNA sequencing revealed distinct differentiation trajectories related to tumor aggressiveness and histone modification patterns. Drug sensitivity analysis showed that high-CMLHMS tumors were more responsive to growth factor and kinase inhibitors (e.g., PI3K, EGFR inhibitors), while low-CMLHMS tumors demonstrated greater sensitivity to cytoskeletal and DNA damage repair-targeting agents (e.g., Paclitaxel, Gemcitabine). Conclusion: The CMLHMS model effectively stratifies PCa into distinct subtypes with unique biological and clinical characteristics. This study provides new insights into histone modification-driven heterogeneity in PCa and suggests potential therapeutic targets, contributing to precision oncology strategies for advanced PCa.