Integrating target, nontarget analysis with machine learning to illuminate PFAS characteristics and health risks in Chinese cosmetics

Per- and polyfluoroalkyl substances (PFAS) are extensively incorporated into cosmetics to enhance durability and water resistance. However, their pervasive application in cosmetic products has raised concerns regarding the potential health risks associated with dermal exposure of PFAS. To date, studies focusing on Chinese market cosmetics have primarily relied on target analysis of a limited array of common PFAS, which may lead to an underestimation of actual exposure risks. In this study, an integrated strategy, combining target analysis, nontarget screening, and machine learning modeling, was employed to evaluate PFAS concentrations and associated human health risks in 31 cosmetic products spanning 16 diverse product categories. Target analysis quantified 10 PFAS in 20 cosmetics, with total concentrations ranging from 0.189 to 143 ng/g. Nontarget screening identified 15 PFAS in 30 cosmetics, with total concentrations spanning from 4.72 to 263 ng/g. Notably, products marketed as “waterproof”, “sweatproof”, or “long-lasting” exhibited a higher propensity for PFAS presence. Subsequent risk assessment revealed that PFAS concentrations in two products (i.e., a lotion and a sunscreen) exceeded the acceptable daily intake dose, indicating potential health risks associated with dermal PFAS exposure. Collectively, these findings offer a comprehensive elucidation of PFAS concentrations and associated health risks in Chinese cosmetics, and establish a scientific foundation for forthcoming regulatory frameworks concerning PFAS application in cosmetic products.