Integrative Multi-omics Analysis and Machine Learning Refine Global Histone Modification Features in Prostate Cancer

Xiaofeng He ¹ Qintao Ge ² Wenyang Zhao ¹ Yu Chao ¹ Huiming Bai ¹ Xiaotong Wu ¹ Jing Tao ¹ Wen-Hao Xu ^2* Yunhua Qiu ^1* Lei Chen ^1* Jianfeng Yang ^1*

• ¹ Department of Urology, Longhua Hospital, Shanghai University of Traditional Chinese Medicine, Shanghai, China., Shanghai, China
• ² Shanghai Cancer Center, Fudan University, Shanghai, China

Prostate cancer (PCa) is a leading cause of cancer-related mortality in men, with significant heterogeneity in clinical behavior and treatment response. While histone modifications are known to play critical roles in tumor progression and treatment resistance, their global regulatory effects in PCa remain poorly understood. This study aimed to comprehensively explore histone modification-driven heterogeneity in PCa using integrative multi-omics analysis and machine learning. We developed the Comprehensive Machine Learning Histone Modification Score (CMLHMS), which stratified PCa into two distinct subtypes with divergent biological and clinical characteristics. High-CMLHMS tumors displayed elevated histone modification activity, enriched proliferative and metabolic pathways, and were strongly associated with progression to castration-resistant prostate cancer (CRPC). In contrast, low-CMLHMS tumors exhibited stress-adaptive and immune-regulatory phenotypes.Single-cell RNA sequencing further validated these subtypes, revealing distinct differentiation trajectories linked to tumor aggressiveness and histone modification patterns. Drug sensitivity analysis identified therapeutic vulnerabilities between subtypes. High-CMLHMS tumors were more responsive to growth factor and kinase inhibitors (e.g., PI3K, EGFR inhibitors), reflecting their reliance on proliferative and metabolic pathways. Conversely, low-CMLHMS tumors demonstrated heightened sensitivity to cytoskeletal and DNA damage repair-targeting agents (e.g., Paclitaxel, Gemcitabine), consistent with their stress-adaptive phenotype. In conclusion, this study provides novel insights into the role of histone modifications in driving PCa heterogeneity and progression. The CMLHMS model offers a powerful tool for stratifying PCa subtypes, identifying therapeutic targets, and informing precision oncology strategies. Targeting histone modification-driven vulnerabilities holds promise for improving outcomes in advanced PCa.