High-Throughput Transcriptomics for Mechanism-Informed Mixture Risk Assessment of Perfluoroalkyl Substances Using...

Widespread human exposure to mixtures of perfluoroalkyl substances (PFAS) presents a challenge for risk assessment, as toxicological data for many PFAS are limited and mixture risk assessment requires insight into their shared toxicological mechanisms. In this study, we reanalyzed publicly available high-throughput transcriptomic (HTTr) datasets from human liver spheroids by integrating benchmark dose modeling with pathway enrichment analysis to identify shared pathway perturbations that served as the basis for deriving mechanistically informed transcriptomic points of departure (tPODs) for nine PFAS, including legacy, long-chain, and short-chain substitute compounds. Three shared pathways, namely PPAR signaling, PPARα-regulated lipid metabolism, and fatty acid, triglyceride, and ketone body metabolism, were identified as consistently perturbed across all nine PFAS. Using these shared pathways, PFAS were further prioritized using the Toxicological Prioritization Index, and long-chain PFAS showed greater overall perturbations than short-chain PFAS. For PFAS with available animal-derived PODs, the resulting tPODs were generally consistent with or more conservative than apical PODs. The combined margin of exposure estimates, calculated by integrating tPODs with population-level serum concentrations from different countries/regions, indicated potential concern for some populations under central-tendency conditions, whereas conservative lower-bound estimates suggested potential concern for all populations and high concern for a small fraction of individuals. Overall, the findings indicate that mechanistically informed tPODs derived from HTTr data, together with human biomonitoring measurements, can support early-tier mixture risk assessment of chemicals.