Plasma metabolomics identifies key metabolites and improves the prediction of benign prostatic hyperplasia

Background: The role of metabolic biomarkers in the pathogenesis of benign prostatic hyperplasia (BPH) remains unclear. This study aimed to assess the association between nuclear magnetic resonance (NMR)-based metabolomics and BPH risk and explore the utility of integrating metabolites into risk prediction models. Materials and methods: This study analyzed 78 724 participants with complete NMR-based plasma metabolomics data from the UK Biobank. Cox proportional hazards (CPH) models were used to analyze the association between 143 metabolites and BPH risk. Elastic Net (ENet) regularization and stepwise regression were applied to select key metabolites and reduce dimensionality. An eXtreme Gradient Boosting (XGBoost) model was constructed for risk prediction, with SHapley Additive exPlanations values determining feature importance. Results: Over a median follow-up of 13.6 years, 7668 participants developed BPH. CPH models identified 46 metabolites significantly associated with time-to-BPH incidence. ENet regularization further refined these to nine key metabolites. The integration of metabolites with established risk factors – including age, testosterone, waist-to-hip ratio, dietary score, diabetes, and hypertension – modestly improved prediction accuracy (concordance index: 0.688 vs. 0.685; net reclassification improvement: 0.081; integrated discrimination improvement: 0.003). Conclusions: These findings highlight the associations between circulating metabolites and the hazard of BPH, supporting the potential of NMR-based metabolomics to enhance risk prediction and inform clinical management strategies.