The Science of Safe Levels: How PFAS Toxicology Informs Regulatory Limits
Setting regulatory limits for PFAS in drinking water, food, soil, or air requires translating complex toxicological data—from animal studies, human epidemiology, and mechanistic research—into specific numerical values that define “acceptable” exposure. This translation process involves choices about which health endpoints to protect, how to account for sensitive subpopulations, what safety factors to apply, and how to handle scientific uncertainty—choices that are both scientifically informed and policy-determined. Understanding the toxicological basis for PFAS health-based values helps interpret their significance, appreciate their limitations, and evaluate claims about PFAS risk with appropriate scientific literacy.
Key Toxicological Endpoints Driving PFAS Risk Values
PFAS health effects span multiple organ systems and disease endpoints—cancer, immune effects, thyroid disruption, developmental effects, and others. Regulatory risk assessments select the most sensitive health endpoint (the one showing effects at the lowest dose) as the basis for health-based value derivation, on the principle that protecting against the most sensitive effect protects against all others. For PFOA and PFOS, the most sensitive endpoints identified in recent EPA assessments include neonatal body weight reduction (a developmental effect) and vaccine antibody response reduction (an immune effect)—both observed at doses consistent with general population blood levels. These sensitive endpoints drove the EPA’s extraordinarily low 2022 health advisory values of 0.004 and 0.02 ppt for PFOA and PFOS respectively.
The choice of sensitive endpoint is scientifically defensible but not value-free—different stakeholders may argue for weighting different endpoints based on their views about which health effects are most consequential. The immune endpoint (vaccine response) is particularly significant because it was derived from studies in children—a lifestage protection standard that effectively sets the reference dose to protect the most vulnerable population. Industry stakeholders have questioned the methodological reliability of some endpoint studies; health advocates have argued that even the sensitive endpoints may understate harm given evidence of effects at lower doses than those used in reference value derivation.
From NOAEL/LOAEL to Reference Doses: The Derivation Process
Traditional toxicological risk assessment derives reference doses (RfDs) from animal studies by identifying the no-observed-adverse-effect level (NOAEL) or lowest-observed-adverse-effect level (LOAEL) and applying uncertainty factors to account for animal-to-human extrapolation, human variability, study quality limitations, and subchronic-to-chronic extrapolation. For PFAS, this process is complicated by the availability of human epidemiological data that often shows effects at lower doses than animal studies—enabling human data to serve as the basis for RfD derivation without relying on animal-to-human extrapolation factors. EPA’s PFOA and PFOS reference dose development used human epidemiological data as the primary basis, with smaller uncertainty factors applied compared to animal-study-based derivations.
Benchmark dose (BMD) modeling—a more sophisticated approach that fits mathematical dose-response models to the available data and identifies doses associated with specific response magnitudes—has become standard practice for PFAS RfD derivation, replacing the cruder NOAEL/LOAEL approach that discards information about the shape of the dose-response curve. BMD modeling provides both more precise central estimates and quantitative characterization of dose-response uncertainty—important inputs for risk characterization that communicates both the point estimate of risk and the uncertainty bounds around it.
Mixture Toxicity: A Challenge for Single-Chemical Risk Assessment
Humans are never exposed to a single PFAS compound in isolation—real-world exposures involve complex mixtures of dozens of PFAS compounds, each present at concentrations below their individual reference values but collectively producing a mixture exposure that may exceed health-protective thresholds. Current PFAS risk assessment frameworks are beginning to address mixture toxicity: the EPA’s MCL for PFAS in drinking water includes a combined MCL for four PFAS (PFNA, PFHxS, HFPO-DA, and PFBS) based on hazard index calculation that accounts for mixture effects. Methodological development for PFAS mixture risk assessment—including guidance on combining individual PFAS contributions using dose-addition models appropriate for chemicals with similar mechanisms of action—represents an active area of regulatory science that will shape future PFAS health-based value development.