Elucidating the pathogenic mechanism underlying plasticizer-induced chronic kidney disease via integrated network toxicology and machine learning.

Plasticizers are ubiquitous plastic additives that have been linked to multiple adverse health outcomes and elevated risks of chronic kidney disease (CKD). However, the precise molecular mechanisms underlying plasticizer-induced CKD remain poorly understood. In this study, we screened five representative plasticizers and employed an integrated strategy of network toxicology and machine learning to dissect the pathogenic links between plasticizer exposure and CKD. We identified 88 overlapping targets between plasticizer exposure and CKD pathogenesis. Functional enrichment analyses revealed that these targets were mainly enriched in inflammatory responses, metabolic dysregulation, and apoptotic processes, which are critical for CKD development. Through protein-protein interaction (PPI) network analysis and systematic evaluation of 107 machine learning models, the Stepwise Logistic Regression (Stepglm) [both] + Random Forest (RF) model was determined as the optimal classifier, which prioritized CTSS, CASP1, ALB, and MCL1 as core hub genes. Crucially, SHAP interpretability analysis further revealed that CTSS and CASP1 function as pro-injury drivers, whereas ALB and MCL1 act as putative protective factors in plasticizer-related renal damage. Molecular docking simulations exhibited strong and stable binding interactions between the five plasticizers and these four core proteins. Collectively, our findings provide novel computational mechanistic insights into plasticizer-associated CKD, and offer a theoretical framework to guide future experimental validation and the development of targeted interventions against plasticizer-related nephrotoxicity.