Integrating network toxicology and in vitro validation to elucidate PET microplastic-induced osteoarthritis pathogenesis.

Objective: This study aims to elucidate the molecular mechanisms through which PET microplastics (PET-MP) influence osteoarthritis (OA) pathogenesis by integrating network toxicology, machine learning, and Methods: Differential gene expression analysis and WGCNA were applied to multiple OA datasets to identify disease-related targets. PET-MP biological targets were predicted via ChEMBL, SwissTargetPrediction, and PharmMapper. Overlapping targets were screened using machine learning algorithms, and molecular docking was performed to assess binding interactions. Results: A total of 452 PET-associated targets were identified, with 12 core PET-MP-OA genes established through intersection analysis. Functional enrichment implicated the NF-κB and IL-17 signaling pathways. Machine learning screening based on feature importance and SHAP values prioritized six hub genes: AKR1A1, INSR, KIF11, MMP1, KCNN4, and TK1. Molecular docking generated predicted AutoDock Vina scores ranging from -3.893 to -7.434 kcal/mol. Conclusion: PET-MP may promote OA progression by disrupting molecular pathways related to inflammation, oxidative stress, and cartilage degradation. The identified hub genes offer new insights into microplastic toxicology in joint disease and represent potential therapeutic targets and biomarkers for PET-MP-induced OA.