Environmental PET-microplastic exposure and risk of non-alcoholic fatty liver disease: An integrated computational toxicology and multi-omics study.

The escalating severity of global microplastic pollution has triggered significant public health concerns. Polyethylene terephthalate (PET), a ubiquitous plastic constituent, extensively permeates aquatic systems, food chains, and daily living environments. However, its potential long-term health impacts, particularly its association with Non-alcoholic Fatty Liver Disease (NAFLD), remain poorly understood. In this study, we adopted an integrative network toxicology approach combined with multi-omics data and machine learning to systematically elucidate the mechanistic relationship between exposure to PET nanoplastics and the pathogenesis of NAFLD. Through comprehensive interrogation of multi-source databases, we identified 20 overlapping targets common to both PET nanoplastic exposure and NAFLD. By employing a comprehensive integrative machine learning framework comprising eleven distinct algorithms, we further identified six core candidate genes: CCL2, GRIA3, JUN, PFKFB3, PIM1, and PPARA. The resulting diagnostic model achieving a maximum Area Under the Curve (AUC) of 0.94 in the training set and demonstrating generalizability in an independent validation cohort (AUC > 0.6). Shapley Additive Explanations (SHAP) analysis identified PFKFB3 and PPARA as the most influential predictors. Single-cell transcriptome analysis revealed cell-type-specific expression patterns of these core genes within hepatocytes, macrophages, and endothelial cells, highlighting their pivotal roles in key intercellular communication pathways, such as the chemokine and macrophage migration inhibitory factor (MIF) signaling pathways. Furthermore, molecular docking and molecular dynamics simulations suggest that PET-derived oligomers or surface-associated chemical functional groups may form specific interactions with the active sites of core proteins. Given the current scarcity of clinical cohorts in public databases that concurrently incorporate measures of microplastic exposure and transcriptomic profiles, this study employs a computational toxicology framework to elucidate the interaction networks between these factors through systematic bioinformatic analysis. By integrating PET-microplastic-related targets with NAFLD-associated transcriptomic data via exploratory systems toxicology modeling, we identified a potential molecular nexus linking PET exposure to NAFLD pathogenesis. These findings establish a theoretical foundation for future mechanistic investigations into microplastic exposure and its toxicological implications.