PFOA Toxicology and Cancer Toxicity

1. Methodology Overview

Dr. Rosenfeld of RoseyFields conducts PFOA risk assessments using the following methodology: we characterize hazard by synthesizing the peer-reviewed human epidemiological and animal toxicological literature on PFOA across its priority non-cancer health outcomes and its carcinogenicity evidence base; we identify the points of departure (PODs) EPA has derived for the most sensitive, well-supported endpoints and apply standard uncertainty-factor methodology to select a single overall oral reference dose (RfD); we evaluate carcinogenic potential using EPA’s weight-of-evidence cancer classification and apply the resulting cancer slope factor (CSF); and we document the toxicokinetic behavior and sensitive-population considerations that are essential to interpreting these values correctly, given PFOA’s unusually long human elimination half-life. The summary below documents the toxicological and epidemiological evidence base for PFOA and the derivation of its RfD and CSF, as reflected in EPA’s April 2024 toxicity assessment.

2. Chemical Identity and Background

Perfluorooctanoic acid (PFOA, CASRN 335-67-1, and its related salts and ammonium/sodium forms) is a fully fluorinated eight-carbon carboxylic acid belonging to the per- and polyfluoroalkyl substances (PFAS) class. Its carbon-fluorine backbone is exceptionally resistant to environmental and metabolic degradation — PFOA is not metabolized in mammals and does not break down appreciably in the environment, earning PFAS compounds the common description “forever chemicals.” PFOA was historically used in the manufacture of fluoropolymers (e.g., non-stick and stain-resistant coatings) and in a range of industrial and consumer applications; direct manufacture and use in the United States has substantially declined following voluntary phase-outs, but PFOA remains a widespread and persistent contaminant in drinking water, soil, and human serum due to its historical use and environmental persistence.

Ingestion of contaminated drinking water is a significant PFOA exposure pathway for the general population; national biomonitoring and drinking water surveys cited in EPA’s assessment document PFOA in a substantial share of U.S. public water systems. This background informs EPA’s April 2024 toxicity assessment, which was prepared specifically to support the PFAS National Primary Drinking Water Regulation (NPDWR), described further in RoseyFields’ companion PFOA Risk Assessment Memorandum.

3. Toxicokinetics (ADME)

PFOA’s toxicokinetic profile differs fundamentally from that of conventional organic solvents such as TCE and PCE, and this difference is central to how its toxicity values were derived.

Absorption and Distribution

PFOA is well absorbed following oral exposure and distributes throughout the body, with the highest concentrations generally found in serum, liver, and kidney rather than in adipose tissue (unlike lipophilic solvents). PFOA binds to serum proteins (notably albumin) rather than partitioning into fat, and it crosses the placenta and is present in breast milk, making prenatal and lactational exposure a well-documented pathway for infants.

Metabolism

PFOA is not metabolized in mammals — it is eliminated as the parent compound, unlike TCE and PCE, which undergo extensive hepatic biotransformation via cytochrome P450 oxidation and glutathione conjugation. This absence of metabolism simplifies certain aspects of PFOA’s toxicokinetics (there are no reactive metabolites to characterize) but means that elimination is governed entirely by renal and other excretory clearance of the parent compound, discussed below.

Elimination and the Central Role of Serum Half-Life

PFOA elimination in humans is remarkably slow compared with laboratory rodents: human serum half-life estimates from multiple independent studies range from approximately 1.8 to 4.7 years, compared with half-lives measured in days (rather than years) in mice and rats. Because of this cross-species difference, EPA’s dose-response modeling for PFOA relies on pharmacokinetic modeling to convert both animal and human study PODs into human-equivalent doses (PODHEDs) calibrated to human serum clearance, rather than relying on simple applied-dose extrapolation.

PFOA elimination also shows notable sex-related and lifestage-related variability in both animals and humans. In rats, females eliminate PFOA substantially faster than males — attributable to sex-hormone-regulated renal organic anion transporters (OATs/OATPs) that favor PFOA reabsorption in males and excretion in females — though this pattern is less pronounced in mice and is actually reversed in cynomolgus monkeys and hamsters, indicating the rat-specific mechanism cannot be assumed to translate directly to humans. In humans, postmenopausal women and adult men show longer PFOA elimination half-lives than premenopausal women, plausibly reflecting the contribution of menstruation, pregnancy/placental transfer, and lactation as elimination routes specific to reproductive-age women — meaning a woman’s reproductive status at the time of exposure can materially affect how long PFOA persists in her body.

4. Non-Cancer Toxicological and Epidemiological Evidence Base

EPA’s toxicity assessment prioritized four non-cancer health outcomes — hepatic, immune, cardiovascular, and developmental effects — as having the strongest overall weight of evidence across human epidemiological studies, animal toxicological studies, and mechanistic data. For each outcome, EPA reached a formal evidence integration judgment; all four judgments concluded PFOA exposure is likely to cause the effect in question in humans under relevant exposure circumstances.

Hepatic Effects

EPA found moderate evidence for an association between PFOA exposure and hepatic effects in humans, based primarily on consistent positive associations with serum alanine aminotransferase (ALT) — a standard liver injury biomarker — across multiple medium-confidence studies spanning general population, high-exposure community, and occupational cohorts (Jain, 2019; Nian et al., 2019; Darrow et al., 2016; Gallo et al., 2012; Lin et al., 2010, among others), further supported by a meta-analysis of 25 studies in adolescents and adults (Costello et al., 2022). The animal evidence is robust, based on 27 high- or medium-confidence toxicological studies; EPA specifically distinguished adaptive hepatocellular hypertrophy alone from adverse hepatotoxicity accompanied by necrosis, inflammation, or clinically significant enzyme changes, and identified a chronic dietary study in Sprague-Dawley rats (NTP, 2020), a developmental study in CD-1 mice (Cope et al., 2021), and a 29-day gavage study (Loveless et al., 2008) as providing the most comprehensive dose-dependent evidence of hepatotoxicity. Mechanistically, PFOA activates multiple hepatic nuclear receptors — most notably PPARα, but also CAR, PXR, PPARγ, ERα, and HNFα — leading to dysregulated lipid metabolism, triglyceride and cholesterol accumulation, and ultimately steatosis; EPA notes an important species-specificity caveat here, since PPARα expression and activity differ substantially between rodents and humans, and a human-specific cholesterol transfer protein (CETP) absent in rodents further complicates direct extrapolation of rodent lipid findings to humans. EPA’s evidence integration judgment: PFOA exposure is likely to cause hepatotoxicity in humans, based on coherent animal effects at doses as low as 0.3 mg/kg-day and human ALT associations at median serum PFOA levels as low as 1.3 ng/mL.

Immune Effects

EPA’s evidence integration judgment concluded PFOA exposure is likely to cause adverse immune effects, specifically immunosuppression, in humans, driven primarily by consistent epidemiological evidence of reduced vaccine antibody response at median serum PFOA levels as low as 1.1 ng/mL. Animal studies showed coherent immunomodulatory responses at doses as low as 1 mg/kg-day, consistent with (though not identical in magnitude to) the human findings, though EPA notes some animal findings co-occurred with overt systemic toxicity, raising questions about their independent biological significance. Evidence for PFOA’s potential effect on allergic sensitization was judged more uncertain, with limited animal or mechanistic support. EPA specifically flags children and young individuals exposed during critical developmental windows as a potentially susceptible population for this endpoint, given the antibody-response data, and notes the absence of long-term/chronic animal studies examining immune effects across different developmental lifestages as a remaining source of uncertainty.

Cardiovascular Effects

EPA concluded PFOA exposure is likely to cause adverse cardiovascular effects, specifically serum lipid effects, driven primarily by consistent epidemiological evidence of elevated total cholesterol at median PFOA exposure levels representative of the general (NHANES) population (median 3.7 ng/mL), and further supported by associations with blood pressure in high- and medium-confidence adult studies. Animal evidence showed coherent lipid homeostasis perturbations across developmental, subchronic, and chronic rodent studies at doses as low as 0.3 mg/kg-day, mechanistically linked to the same nuclear-receptor-mediated pathways implicated in PFOA hepatotoxicity.

Developmental Effects

EPA concluded PFOA exposure is likely to cause developmental toxicity in humans, based primarily on epidemiological evidence of decreased birth weight in studies where maternal PFOA was measured during pregnancy (median levels ranging from 1.1 to 5.2 ng/mL), supported by coherent evidence for biologically related effects (decreased postnatal growth, decreased birth length) and by consistent, dose-dependent decreases in fetal weight and other developmental effects in animal models gestationally exposed to PFOA at doses as low as 0.5 mg/kg-day. EPA notes that while mechanistic data support the biological plausibility of the animal findings, the data are currently too limited to fully establish human relevance of the specific underlying mechanism.

5. Carcinogenicity Evidence Base

Human Epidemiological Evidence

Two medium-confidence independent studies provide evidence of an association between elevated PFOA serum concentration and kidney cancer (Shearer et al., 2021; Vieira et al., 2013), and two studies in the same underlying cohort provide evidence of an association with testicular cancer (Barry et al., 2013; Vieira et al., 2013). EPA also identified some evidence of breast cancer in one susceptible subpopulation studied. EPA describes this human evidence as supporting a plausible, though not definitively causal, association, and specifically notes that the PFOA cancer database would benefit from additional large, high-confidence cohort studies in independent populations.

Animal Toxicological Evidence

Three chronic bioassay studies, all conducted in the same rat strain, together demonstrate increased incidence of three distinct tumor types in male rats fed PFOA-contaminated diets: Leydig cell tumors (LCTs), pancreatic acinar cell tumors (PACTs), and hepatocellular adenomas/carcinomas. EPA specifically notes that pancreatic acinar cell adenocarcinoma is a rare tumor type in this rat strain (historical control incidence of 0/340 in females and 2/340 in males), which increases confidence that its occurrence in PFOA-treated animals is treatment-related rather than incidental. EPA also notes that site concordance between the human testicular cancer findings and the animal LCT findings is present, strengthening that particular line of evidence, while acknowledging that tumor site concordance between species is not a requirement for a carcinogenicity determination under its guidelines.

Mode of Action

Available mechanistic data suggest that multiple, tissue-specific modes of action likely contribute to PFOA’s renal, testicular, pancreatic, and hepatic tumorigenicity, rather than a single unifying mechanism. EPA found a lack of information supporting a mutagenic mode of action for PFOA and no evidence of increased cancer susceptibility specifically from early-life exposure, a finding directly relevant to how the cancer slope factor was applied (see Section 6).

Cancer Classification

Applying its Guidelines for Carcinogen Risk Assessment, EPA classified PFOA as Likely to Be Carcinogenic to Humans via the oral route — a determination that the evidence is “adequate to demonstrate carcinogenic potential to humans but does not reach the weight of evidence for the descriptor Carcinogenic to Humans.” EPA’s own document explicitly notes that other agencies have reached different classifications using different criteria: IARC (as reported in Zahm et al., 2023) and CalEPA (2021) have both classified PFOA as carcinogenic to humans, and EPA attributes this difference to the fact that IARC’s and CalEPA’s classification criteria are not identical to EPA’s own Guidelines for Carcinogen Risk Assessment — a distinction RoseyFields flags as relevant for any client operating under California or international regulatory frameworks, which may apply a more stringent classification than the federal EPA determination used as the basis for this page.

6. Derivation of the Oral Reference Dose (RfD)

Unlike TCE, PCE, and 1,4-dioxane — each of which EPA assessed using endpoint-specific human-equivalent concentrations/doses within a margin-of-exposure framework — PFOA’s toxicity assessment derives and selects a single overall oral RfD, of the classical IRIS-style type, applicable to both short-term and chronic exposure scenarios.

Points of Departure and Candidate RfDs

For each of the four priority non-cancer health outcomes, EPA derived multiple candidate RfDs from both epidemiological and animal toxicological studies, using benchmark dose (BMD) modeling where feasible (a hybrid modeling approach for developmental, hepatic, and cardiovascular effects capable of estimating response probability at specified exposure levels; multivariate-model-derived benchmark responses for immune effects). Candidate RfDs based on epidemiological data clustered within a single order of magnitude of one another (10⁻⁷ to 10⁻⁸ mg/kg-day) regardless of endpoint or study population, while candidate RfDs based on animal toxicological data were generally two to three orders of magnitude higher — a divergence EPA attributes to the substantial interspecies toxicokinetic differences (particularly half-life) between rodents and humans, and to the general principle that human epidemiological data eliminate interspecies extrapolation uncertainty entirely. For this reason, EPA excluded animal-derived candidate RfDs from further consideration in favor of epidemiology-based candidates for both health-outcome-specific and overall RfD selection.

Uncertainty Factors Applied

For the epidemiology-based candidate RfDs, EPA applied the following uncertainty factors:

• UFA (interspecies) = 1: not applicable, since the critical studies are human epidemiological data.
• UFH (intraspecies/human variability) = 10: applied because no chemical-specific information was available to support a lower value; accounts for toxicokinetic, toxicodynamic, genetic, lifestage, health-status, and lifestyle-related variability across the human population.
• UFL (LOAEL-to-NOAEL) = 1: not needed, since a benchmark dose lower confidence limit (BMDL) — rather than a NOAEL or LOAEL — was used as the POD for every epidemiological endpoint carried forward.
• UFS (subchronic-to-chronic) = 1: applied because the critical hepatic and cardiovascular studies involved adults assumed to have long-term exposure, and because for the developmental and immune endpoints, EPA’s guidance treats the relevant exposure window (a developmental or immune-system-maturation critical period) as the biologically appropriate duration of concern rather than lifetime exposure — meaning a full subchronic-to-chronic adjustment was not scientifically appropriate for those endpoints.
• UFD (database) = 1: applied given a comprehensive PFOA toxicology database spanning short-term, subchronic, and chronic studies across three animal species, developmental and multigenerational reproductive toxicity studies, developmental neurotoxicity studies, and a large body of medium- and high-confidence epidemiological studies — the specific study types (developmental and multigenerational reproductive toxicity) that most commonly drive a higher UFD were determined to be adequately addressed in PFOA’s existing database.

The resulting composite uncertainty factor for all epidemiology-based candidate RfDs is 10 (UFH alone).

Health Outcome-Specific RfD Selection

Notably, three of the four health-outcome-specific RfDs (immune, cardiovascular, and developmental) converge on the identical value, 3 × 10⁻⁸ mg/kg-day, despite being derived from entirely independent studies, populations, and biological endpoints — a convergence EPA treats as materially strengthening confidence in the overall RfD, since it indicates PFOA produces adverse effects across multiple organ systems at approximately the same exposure level rather than one organ system driving the result in isolation.

Overall RfD

EPA selected an overall RfD of 3 × 10⁻⁸ mg/kg-day, with the immune, developmental, and cardiovascular endpoints (decreased anti-tetanus/anti-diphtheria antibody concentrations, decreased birth weight, and increased total cholesterol, respectively) serving jointly as co-critical effects. Because two of these three co-critical effects are developmental in nature and can result from a short-term exposure during a critical developmental window, EPA concluded the overall RfD is applicable to both short-term and chronic risk assessment scenarios — RoseyFields therefore applies a single RfD across acute and chronic PFOA exposure scenarios, unlike TCE, PCE, and 1,4-dioxane, each of which requires separate acute and chronic reference values.

7. Carcinogenicity Classification and Cancer Slope Factor

Consistent with the Section 5 classification of PFOA as Likely to Be Carcinogenic to Humans, EPA derived a quantitative cancer slope factor using the same general dose-response and pharmacokinetic modeling approach used for the non-cancer RfD.

Candidate CSFs were derived from both epidemiological data (renal cell carcinoma and kidney cancer, using weighted linear regression to calculate quartile-specific relative risks, converted to an internal serum-based CSF and then to an external dose-based CSF via the same clearance-based conversion used for the RfD) and animal toxicological data (multistage cancer models applied to BMDLs for LCTs, hepatocellular tumors, and pancreatic acinar cell tumors, converted to PODHEDs using the human clearance value). EPA selected the oral slope factor of 0.0293 per ng/kg-day for renal cell carcinoma in human males, from Shearer et al. (2021), as the basis of PFOA’s overall CSF.

Per EPA’s Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens, age-dependent adjustment factors (ADAFs) — which EPA applies by default when a chemical has an affirmatively demonstrated mutagenic mode of action — were not applied to the PFOA CSF, because the evidence does not support a mutagenic mode of action for PFOA and does not indicate increased early-life cancer susceptibility. RoseyFields notes this as a meaningful distinction from carcinogens with an established mutagenic MOA, for which early-life or childhood PFOA exposure would otherwise warrant an upward risk adjustment beyond what the CSF alone captures.

8. Uncertainties and Sensitive Populations

Key Uncertainties

• Reliance on a single critical cancer study: the selected CSF is anchored in a single medium-confidence study (Shearer et al., 2021); while it is corroborated by an independent kidney cancer study (Vieira et al., 2013) and by site-concordant animal LCT data, EPA itself identifies additional large, high-confidence independent cohort studies as a priority research need for strengthening the PFOA cancer database.
• Uncharacterized carcinogenic mode(s) of action: EPA’s mechanistic evidence suggests multiple, tissue-specific modes of action across kidney, testicular, pancreatic, and hepatic tumorigenesis, but no single, fully characterized MOA has been established for any of these tumor types, introducing uncertainty into the linear low-dose extrapolation approach.
• Divergent cancer classifications across agencies: as noted in Section 5, IARC and CalEPA have classified PFOA as carcinogenic to humans (a stronger classification than EPA’s Likely to Be Carcinogenic to Humans), reflecting differences in classification criteria rather than differences in the underlying scientific data — a source of potential confusion for clients operating across multiple regulatory jurisdictions.
• Sex- and lifestage-dependent toxicokinetics: because PFOA elimination half-life differs by sex, reproductive status, and (in animal models) age, a given external dose does not translate into the same internal serum concentration across all individuals, and the mechanisms underlying human sex differences in PFOA half-life remain incompletely understood.
• Limited data on differential susceptibility by race, ethnicity, and socioeconomic status: EPA explicitly acknowledges that PFOA and PFAS exposure may disproportionately affect certain communities (low socioeconomic status communities, Tribal communities, minority communities, and communities near historical PFOA manufacturing or contamination sites), but describes the available data on differential health effects incidence across these populations as limited, identifying this as a high-priority research need rather than a fully characterized risk factor.

Sensitive and Susceptible Subpopulations

• Fetuses, infants, and children: the most well-studied sensitive population for PFOA, given epidemiological and animal evidence of developmental effects (decreased birth weight, decreased postnatal growth) and immune effects (decreased vaccine antibody response) at exposure levels relevant to the general population; children with autoimmune disease or on immunosuppressive medication may be particularly susceptible to the antibody-response endpoint specifically.
• Pregnant and lactating women, given placental transfer and breast milk as documented PFOA exposure/elimination pathways for the developing fetus and infant.
• Individuals by sex and reproductive status: premenopausal women eliminate PFOA faster than postmenopausal women and adult men, meaning internal dose (and cumulative body burden) can differ meaningfully between otherwise similarly exposed individuals depending on age, sex, and reproductive status.
• Communities with environmental justice concerns, including low-SES, Tribal, and minority communities, and communities near current or historical PFOA manufacturing or contamination sites, which EPA identifies as facing potentially disproportionate PFOA exposure.

9. Conclusion

PFOA presents a well-characterized, multi-organ non-cancer hazard profile — immune, cardiovascular, hepatic, and developmental effects each independently supported by coherent human epidemiological and animal toxicological evidence — with three of its four health-outcome-specific reference doses converging on an identical value (3 × 10⁻⁸ mg/kg-day), a notable strengthening feature of the overall RfD. PFOA is classified as Likely to Be Carcinogenic to Humans, with a quantitative cancer slope factor (0.0293 per ng/kg-day) anchored in human renal cell carcinoma data and corroborated by site-concordant animal tumor evidence, though EPA itself flags the cancer database as benefiting from additional independent cohort studies. PFOA’s toxicokinetics — no metabolism, and a multi-year human serum elimination half-life that differs by sex and reproductive status — distinguish it fundamentally from the volatile organic solvents in RoseyFields’ other chemical risk characterization pages, and are essential context for interpreting its toxicity values correctly. RoseyFields applies the RfD and CSF described here within the risk characterization framework detailed in its companion PFOA Risk Assessment Memorandum.

Bibliography