Integrating multi-omics and machine learning to decipher the role of GSTP1 in endocrine-disrupting chemical-induced prostate cancer pathogenesis

Prostate cancer (PCa) pathogenesis involves complex interactions between genetic susceptibility and exposure to endocrine-disrupting chemicals (EDCs). This study aimed to systematically identify key genes linking EDC exposure to PCa using an integrated bioinformatics and machine learning (ML) approach. We analyzed four transcriptomic datasets (GSE32571, GSE71016, GSE46602, GSE200879) and identified 437 differentially expressed genes (DEGs). By integrating these with high-confidence EDC-interacting genes from the Comparative Toxicogenomics Database (CTD), we pinpointed 169 core candidates. An ensemble ML framework, evaluating over 70 algorithm combinations, identified an optimal model (glmBoost + RF) that refined this list to eight core genes, including Glutathione S-Transferase Pi 1 (GSTP1). Mendelian randomization (MR) analysis established a causal, protective role for GSTP1 against PCa risk (OR = 0.880, 95 % CI = 0.777-0.998, P = 0.046). Molecular docking and dynamics simulations revealed stable binding between GSTP1 and high-priority EDCs, such as Benzo[a]pyrene (binding energy: -9.8 kcal/mol), indicating a direct interaction mechanism. Functional enrichment analyses implicated these genes in oxidative stress response and xenobiotic metabolism. Furthermore, single-cell RNA sequencing and immune infiltration analysis suggested a role for GSTP1 in modulating the tumor microenvironment. Our findings elucidate a critical "EDC-GSTP1-PCa" axis, highlighting GSTP1's potential as a therapeutic target and providing mechanistic insights into environmental chemical-induced prostate carcinogenesis.