Probabilistic Concentration-Response Modeling and Risk Prioritization of Defined Mixtures of PFAS Using Human In...

Per- and polyfluoroalkyl substances (PFAS) are persistent environmental contaminants. Some individual PFAS are linked to adverse health effects, but little data is available to evaluate PFAS mixtures. We tested the biological effects of 20 defined PFAS mixtures (containing 4-56 PFAS) across eight human cell types, including iPSC-derived cardiomyocytes, neurons, and hepatocytes; primary hepatocytes; and HepG2s, endothelial, and renal proximal tubule epithelial cell lines. Mixtures were designed to reflect realistic exposures based on drinking/surface water data, regulatory limits, biomonitoring results, and prior in vitro studies. Cells were exposed to five 10-fold dilutions spanning environmentally relevant concentrations, and cell-specific viability and functional endpoints were measured. Concentration-response was modeled to derive mixture-specific points of departure (PODs). Mixtures based on prior in vitro bioactivity and highly contaminated water samples were the most potent. HepG2s, hepatocytes, and iPSC-derived neurons were the most sensitive cells. Experimental mixture PODs were compared with predictions from concentration addition (CA) models based on individual PFAS data. CA showed low accuracy in distinguishing active from inactive mixtures and underestimated potency (∼300-fold). Overall, our findings demonstrate that direct testing of PFAS mixtures provides more health-protective estimates than component-based models and supports a tiered testing strategy prioritizing sensitive human liver cell models.