Intraprostatic steroid hormones and endocrine disruptors in prostate cancer.

Prostate cancer (PCa) is a hormone-dependent malignancy governed by steroid signaling. Despite characterization of androgen pathways, the broader intraprostatic steroid environment remains underexplored. Endocrine-disrupting chemicals (EDCs) may modulate these processes, yet their tissue distribution is largely unknown. We aimed to comprehensively profile intraprostatic steroids and EDCs and assess their associations with tumor aggressiveness, prognosis, and risk stratification. Thirty steroids and 16 EDCs were quantified using liquid chromatography-tandem mass spectrometry in cancerous tissue, non-cancerous tissues (from peripheral and transition zones), and plasma from 173 patients with localized PCa undergoing robot-assisted radical prostatectomy. Analysis of variance was used to compare analyte levels across biological matrices and according to disease aggressiveness, multivariate statistical models evaluated associations with patient prognosis and risk categories, and machine learning (ML) models were applied for risk stratification. Steroid profiles differed markedly between plasma and prostate tissue, whereas EDCs accumulated in the prostate and were generally not associated with plasma concentrations. Cancerous tissue showed lower conjugated dehydroepiandrosterone (DHEA) and higher pregnenolone, corticoids, and testosterone compared to non-cancerous tissue. Elevated intratumoral cortisol, cortisone, androstenedione, conjugated testosterone, 7-ketoDHEA, and estrone were associated with greater tumor aggressiveness. Higher intratumoral parabens correlated with increased postoperative PSA, while conjugated dihydrotestosterone (DHT) and phytoestrogen daidzein were inversely associated with PSA levels. Higher EDC burden and intratumoral steroid levels were associated with higher risk categories. In contrast, DHT showed a consistent protective profile, being negatively associated with both prognosis and risk classification. Integrating steroids and EDCs improved ML-based risk classification beyond preoperative clinical variables.