Determination of species- and sex-dependent PFAS pharmacokinetics (PK) through Bayesian hierarchical modeling
On this page:
The carbon chain length and degree of fluorination of per- and polyfluoroalkyl substances (PFAS) greatly influence its physio- and biochemical properties, resulting in environmental persistence for some compounds. For humans, exposure to numerous PFAS occurs through a variety of environmental sources such as food and water, and protein binding coupled with possible renal resorption allows some PFAS to remain in the body for years. Laboratory animal studies provide a means for interspecies extrapolation to inform human PFAS PK. However, physiological variation, sex-specific differences in PFAS clearance, disparate numerical methodologies, and compartmental assumptions for fitting PK models can result in large differences among PK parameters, making cross-species extrapolation difficult. To address these differences, we developed a Bayesian inference hierarchical model to estimate PFAS PK for multiple species across numerous data sets. After reviewing the existing literature to obtain time course blood and tissue concentration data, we used fifty studies to estimate PK for six PFAS: PFHxA, PFHxS, PFNA, PFDA, PFBS, and PFBA in rats, mice, and non-human primates of both sexes. Using the Bayesian approach, we obtained distributional estimates of animal PK parameters of interest such as beta-phase elimination half-life, volume of distribution (Vd), and clearance. The resulting half-life estimates ranged from two hours for shorter-chained (C) PFAS in female rats up to eighty days for longer chained (C) PFAS across both sexes. Additional sex-dependent differences in Vd demonstrated higher Vd estimates in males than females for longer chain PFAS. Ultimately, this study provides a large collection of in vivo derived PFAS PK parameters across multiple species and sexes. These estimates, which include quantitative information about uncertainty, are useful for the development of more robust and physiologically relevant PK models. Finally, they can provide a foundation for quantitative structural activity relationship models to predict half-lives for PFAS when PK data are unavailable.