Machine learning-driven proteomic profiling reveals seminal plasma biomarkers linking trihalomethane exposure to impaired sperm quality.

The decline in sperm quality has emerged as a major public health concern with significant implications for male fertility and population sustainability. However, high-accuracy biomarkers for sperm quality are lacking, and the environmental drivers remain obscure. Trihalomethanes (THMs), ubiquitous disinfection by-products in drinking water, have been linked to declining semen quality worldwide, yet the molecular mechanisms underlying THM-induced reproductive toxicity remain underexplored. In this propensity score-matched case-control study designed to maximize internal validity, we identified SP biomarkers of oligospermia and teratospermia among Chinese sperm donors and investigated their associations with THM exposure. Using deep data-independent acquisition (DIA) proteomics, we quantified 1666 SP proteins, including 73 proteins enriched in male reproductive tissues. A robust XGBoost machine learning algorithm, validated to mitigate overfitting, determined the optimal biomarker combination, achieving an AUC of 0.95. Ten oligospermia biomarkers-including five testis- or epididymis-specific proteins-were inversely associated with blood THM concentrations. These proteins regulate interconnected pathways related to oxidative stress and energy metabolism. We propose that THM-induced disruption of these fundamental metabolic machineries not only impairs sperm motility but also arrests spermatogenesis, leading to reduced sperm count. This study provides the first evidence linking SP proteomic signatures to THM exposure, offering novel insights into the mechanistic pathways of environmental reproductive toxicity.