Recommended Toxicity Reference Values for PFOS and PFHxS In Mammals and PFOS in Birds

Abstract Perfluorooctane sulfonate (PFOS) and perfluorohexane sulfonate (PFHxS) are frequently detected in environmental media at and near locations where per- and polyfluoroalkyl substances (PFAS) have been released. Because of their persistence, relatively high bioaccumulation potential, and the frequency and magnitude of detection at these locations, PFOS and PFHxS are likely to be a key focus of site-specific ecological risk assessments (ERAs) used to assess the potential exposures of avian and mammalian wildlife to PFAS. Toxicity reference values (TRVs) are a critical parameter of the collective modeling of exposure and effects to wildlife in food chain models used in site-specific ERAs and in deriving ecological screening values (ESVs) for abiotic media. This work reviewed published mammalian and avian laboratory toxicity studies with the goal of recommending TRVs for use in site-specific ERAs and ESV derivation. Selection criteria prioritized oral exposure studies in birds and mammals evaluating effects on population-level apical endpoints, such as reproduction, growth, and survival. In all cases, reproductive effects were the most sensitive apical endpoint. The no-effect and low-effect avian TRVs for PFOS were 0.55 and 1.1 mg/kg-bw/d, respectively, as derived from a chronic dietary exposure study with Japanese quail (Cortunix japonica). For mammals, TRVs were derived from chronic dietary exposure studies in deer mice (Peromyscus spp.). The no-effect and low-effect mammalian TRVs for PFOS were 0.25 and 0.5 mg/kg-bw/d, respectively. Perfluorohexane sulfonate was an order of magnitude less toxic than PFOS, with no-effect and low-effect mammalian TRVs of 7 and 14 mg/kg-bw/d, respectively. Given these TRVs are bracketed by other doses within relatively straightforward dose-response relationships and are based on chronic exposures that evaluated a variety of apical endpoints clearly linked to population-level ecological health, we believe these TRVs are considered robust and are expected to be useful for site-specific ERAs and ESV calculation.