GenX (HFPO Dimer Acid) Risk Assessment and Liver and Immune Toxicity

RoseyFields conducts GenX risk assessments using the following methodology.

Chemical Identity, Regulatory Background, and Exposure Sources

“GenX chemicals” is the common designation for hexafluoropropylene oxide dimer acid (HFPO-DA; CASRN 13252-13-6) and its ammonium salt (CASRN 62037-80-3), a short-chain perfluoroether carboxylic acid within the broader per- and polyfluoroalkyl substances (PFAS) class. HFPO-DA is the principal chemical associated with the “GenX” processing-aid technology, which manufacturers began using in 2009 as a replacement for perfluorooctanoic acid (PFOA) in the production of high-performance fluoropolymers, including certain nonstick coatings. GenX chemicals may also be generated as an unintended byproduct of fluoromonomer production.

The regulatory history of GenX is unusually compressed and still evolving relative to legacy PFAS such as PFOA and PFOS. EPA released draft human health toxicity assessments for GenX chemicals for public comment in November 2018, then finalized the assessment in October 2021 following two rounds of independent external peer review and public comment (EPA, 2021). RoseyFields’ assessors rely on the October 2021 final values rather than the November 2018 draft values, because EPA revised its points of departure, uncertainty factors, and resulting reference doses materially between the draft and final versions in response to peer-review and public comment, and EPA itself cautioned that the 2018 draft values were not intended for use in risk management decisions. EPA subsequently used the 2021 final chronic RfD, together with default exposure assumptions and a relative source contribution factor, to derive a lifetime drinking-water Health Advisory for GenX chemicals in 2022.

GenX chemicals reach the environment primarily through manufacturing and industrial discharge; the best-documented and most heavily litigated example is discharge from the Chemours Fayetteville Works facility into the Cape Fear River in North Carolina, which contaminated downstream public and private drinking-water supplies and, to a lesser and more localized extent, ambient air in the surrounding region through stack emissions and fugitive releases. GenX chemicals have also been detected in surface water, groundwater, finished drinking water, and rainwater in other areas with fluoropolymer manufacturing or processing activity, and EPA’s 2018 fact sheet characterized ambient-air and drinking-water exposure to GenX chemicals as generally localized or regional in nature rather than reflecting the broad, diffuse national background exposure historically associated with PFOA and PFOS. Unlike PFOA and PFOS, GenX chemicals have a substantially shorter serum elimination half-life and lower bioaccumulation potential in humans, a toxicokinetic property that materially affects both hazard characterization (Section 3) and practical dose reconstruction in litigation (Section 8). RoseyFields’ assessors also note that EPA’s toxicity assessment, consistent with the scope described in EPA’s own 2018 fact sheet, addresses only the potential health effects of oral exposure to GenX chemicals in isolation; it does not evaluate cumulative or mixture effects from co-occurring PFAS, an important scoping limitation given that GenX contamination sites frequently involve detectable co-occurring PFAS such as PFOA, PFOS, and other perfluoroether compounds associated with the same manufacturing process.

Point-of-Departure Selection

EPA’s oral database for HFPO-DA and its ammonium salt includes acute, short-term, subchronic, and chronic-duration studies in rats and mice, submitted principally by DuPont/Chemours under Toxic Substances Control Act reporting obligations, supplemented by EPA’s own review of the publicly available peer-reviewed literature. Across this database, the liver is consistently identified as the most sensitive target organ in both sexes and both species, across exposure durations ranging from 15 days to two years and doses ranging from 0.5 to 1,000 mg/kg-day; corroborating findings include hepatocellular hypertrophy and altered lipid and immune parameters in a 28-day mouse study (Rushing et al., 2017), and hepatocellular adenomas and carcinomas in a chronic 2-year rat bioassay (Caverly Rae et al., 2015), discussed further in Section 7.

EPA selected an oral gavage reproduction/developmental toxicity screening study in CD-1 mice (conducted to OECD Test Guideline 421; internally designated DuPont-18405-1037, 2010) as the critical study, and a “constellation” of related liver lesions in parental (F0) female mice — cytoplasmic alteration consistent with hepatocellular hypertrophy, hepatocellular single-cell necrosis, focal necrosis, and hepatocellular apoptosis — as the critical effect. EPA’s selection of the constellation, rather than any single lesion in isolation, followed the conclusion of the National Toxicology Program’s Pathology Working Group that the dose-related pattern across this group of related lesions, considered together, represents the biologically meaningful marker of adversity in this study. This study and endpoint were selected because they produced the lowest point of departure among the available dose-response datasets, making them the most health-protective basis available, and because the liver constellation of lesions is independently corroborated across the broader toxicological database, including subsequent mechanistic work specifically examining these lesions’ pathogenesis (Heintz et al., 2023; Thompson et al., 2023). RoseyFields’ assessors note as an important limitation that the critical study itself (DuPont-18405-1037, 2010) is an unpublished, industry-submitted Good Laboratory Practice study rather than an independently published, peer-reviewed journal article; this point is discussed further in Section 9.

Because sufficient dose-response data were available across multiple dose groups, EPA used benchmark dose (BMD) modeling rather than a simple NOAEL/LOAEL approach, deriving a BMDL10 (the 95% lower confidence bound on the dose associated with a 10% extra risk of the constellation of liver lesions) as the point of departure in the animal dose metric.

RoseyFields’ assessors also weigh the fact that this critical effect occurred in female mice specifically; while liver toxicity is observed in both sexes and both tested species across the broader database, the female mouse constellation-of-lesions endpoint produced the lowest, most health-protective point of departure among the candidate endpoints EPA evaluated, and EPA’s selection of it is consistent with the general risk-assessment principle of basing reference values on the most sensitive endpoint and subgroup identified in an adequate database, rather than on a central tendency across sexes or species.

Pharmacokinetic Modeling and Dose Conversion

EPA’s default hierarchy for converting an animal point of departure to a human-equivalent dose favors, in order of preference, a validated physiologically based toxicokinetic (PBTK) model, then chemical-specific data-informed dosimetric adjustment, then default body-weight scaling. No validated PBTK model exists for HFPO-DA. One human serum half-life dataset exists: twenty-five Chemours facility workers in the Netherlands provided paired blood samples before and after a 72-to-96-hour off-work period, yielding an average estimated half-life of 81 ± 55 hours (range 42–333 hours), according to an unpublished industry dataset submitted to EPA that has not itself been peer reviewed. EPA concluded this dataset was insufficient to support a data-informed dosimetric adjustment factor because it had not undergone independent peer review, was limited to eighteen individuals with measurable paired concentrations, and — critically — was not stratified by sex, even though the critical mouse effect is more severe in females (DuPont-18405-1037, 2010). EPA therefore applied its default procedure of scaling by body weight raised to the three-quarters power (BW^3/4) (EPA, 2011) to extrapolate from the animal point of departure to a human equivalent dose (HED). As a check on this default approach, EPA separately calculated what the POD(HED) would have been using the Chemours human half-life data in place of the default scaling, and found the two approaches produced reasonably similar POD(HED) values, which EPA cited as evidence the default BW^3/4 approach was not producing an implausible result. This default scaling approach is also understood to capture much of the interspecies difference in basal metabolic rate and toxicokinetics, which is why, as discussed in Section 5, the interspecies uncertainty factor applied on top of this scaling is reduced below its full default value. Applying BW^3/4 scaling to the animal BMDL10 yielded a POD(HED) of 0.01 mg/kg-day for the female-mice endpoint, the value carried forward into the reference dose derivation.

Uncertainty Factor Derivation

EPA applied a composite uncertainty factor to the POD(HED) of 0.01 mg/kg-day, built from five component factors:

An intraspecies (human variability) factor, UFH, of 10 was applied because the critical study was conducted in a genetically homogeneous laboratory animal population and provides no chemical-specific data on the range of human sensitivity, including potentially more susceptible subgroups such as pregnant women, infants, and individuals with pre-existing liver disease.

An interspecies factor, UFA, of 3 (rather than the full default of 10) was applied because the BW^3/4 scaling already performed in Section 4 is understood to account for the toxicokinetic component of interspecies variability, leaving only the toxicodynamic component — differences in how animal and human tissue actually responds to a given internal dose — to be addressed by this residual factor.

A LOAEL-to-NOAEL factor, UFL, of 1 was applied because the point of departure is a BMDL derived from formal dose-response modeling rather than a LOAEL identified by simple inspection of the dose groups; the benchmark-dose approach is understood to already address the uncertainty this factor would otherwise cover.

A subchronic-to-chronic duration factor, UFS, was applied differently for the two reference doses. For the subchronic RfD, UFS was set to 1, because the critical study is itself of subchronic-comparable duration and directly informs a subchronic-duration reference value. For the chronic RfD, UFS was increased to 10, to extrapolate from the subchronic critical study to a full lifetime chronic exposure duration, given the absence of a chronic-duration study demonstrating the same low-dose liver constellation (the chronic rat bioassay of Caverly Rae et al. (2015), discussed in Section 7, examined different, generally higher-dose endpoints, including tumors, rather than replicating the low-dose constellation of non-neoplastic liver lesions used as the critical effect here).

A database uncertainty factor, UFD, of 10 was applied to both reference doses to account for deficiencies in the overall toxicological database, including the absence of a full multigenerational reproductive toxicity study, limited chronic-duration data corroborating the critical low-dose non-neoplastic liver endpoint, and residual uncertainty regarding the relevance to humans of the mode of action underlying the observed liver effects (Section 9).

The resulting composite uncertainty factor is 10 × 3 × 1 × 1 × 10 = 300 for the subchronic RfD, and 10 × 3 × 1 × 10 × 10 = 3,000 for the chronic RfD.

Derivation of the Oral Reference Dose

Dividing the POD(HED) of 0.01 mg/kg-day by the applicable composite uncertainty factor yields:

Subchronic RfD = 0.01 mg/kg-day ÷ 300 = 3 × 10⁻⁵ mg/kg-day.

Chronic RfD = 0.01 mg/kg-day ÷ 3,000 = 3 × 10⁻⁶ mg/kg-day (equivalent to 3 nanograms per kilogram body weight per day).

EPA has not derived an inhalation reference concentration (RfC) for GenX chemicals; the final 2021 toxicity assessment, like the 2018 draft before it, addresses oral exposure only. RoseyFields’ assessors treat the absence of an inhalation value as a significant limitation for any matter involving air-emission or vapor-intrusion exposure pathways, discussed further in Section 9.

Carcinogenicity Classification

Under EPA’s 2005 Guidelines for Carcinogen Risk Assessment descriptor framework, EPA classified GenX chemicals as having “Suggestive Evidence of Carcinogenic Potential.” This classification rests on findings from the chronic 2-year oral bioassay in Sprague-Dawley rats (Caverly Rae et al., 2015), which reported hepatocellular adenomas and carcinomas in high-dose female rats, together with pancreatic acinar cell carcinoma and Leydig cell adenoma in high-dose male rats, all occurring principally at the highest tested dose (500 mg/kg-day). EPA’s confidence in this tumor evidence is tempered by several factors evident in the underlying study: a high incidence of premature mortality among high-dose animals that complicates interpretation of the chronic tumor data, and the recognized low relevance to humans of rodent Leydig cell tumors, which typically arise through a rodent-specific hormonal mechanism.

Because the overall weight of evidence is “suggestive” rather than “likely to be carcinogenic” or “carcinogenic to humans,” EPA’s own cancer risk assessment guidelines direct that a quantitative cancer slope factor or unit risk generally not be derived unless a well-conducted study permits reasonably confident characterization of the dose-response relationship and its uncertainty. EPA accordingly did not derive a cancer slope factor, an inhalation unit risk, or a cancer-risk-based drinking-water concentration for GenX chemicals, and did not incorporate a cancer-based benchmark into its 2022 lifetime drinking-water Health Advisory. RoseyFields’ assessors treat this as a meaningful and litigation-relevant distinction from PFOA and PFOS, both of which carry stronger cancer weight-of-evidence characterizations and, in the case of PFOA, an EPA-derived quantitative cancer slope factor.

Practical Application: HQ/HI, Exposure Reconstruction, and Litigation Support

For non-cancer risk characterization, RoseyFields’ assessors calculate a Hazard Quotient (HQ) as the estimated or reconstructed chronic daily oral dose divided by the chronic RfD of 3 × 10⁻⁶ mg/kg-day; an HQ exceeding 1.0 indicates the estimated exposure exceeds the level EPA considers likely to be without appreciable risk of the critical liver effect over a lifetime. Because the chronic RfD is extremely low relative to many other environmental contaminants, comparatively small absolute exposure doses can yield an HQ greater than 1.0, a feature RoseyFields’ assessors explain carefully to retaining counsel and testifying experts, since it reflects the size of the applicable uncertainty factors and the sensitivity of the critical endpoint rather than necessarily indicating a proportionally severe hazard.

Because EPA did not derive a cancer slope factor (Section 7), RoseyFields does not calculate an excess lifetime cancer risk (ELCR) for GenX chemicals using the standard multiplicative CSF approach used for PFOA and many other regulated carcinogens. Instead, RoseyFields’ risk characterizations address the “suggestive” cancer evidence qualitatively, alongside the quantitative non-cancer HQ analysis, disclosing explicitly that the animal tumor data exist but were not considered by EPA to be of sufficient quality or consistency to support quantitative extrapolation.

For drinking-water matters — the dominant real-world exposure scenario for GenX litigation to date, exemplified by the Cape Fear River basin contamination associated with the Fayetteville Works facility — RoseyFields’ assessors use EPA’s 2022 lifetime Health Advisory of 10 parts per trillion (0.01 micrograms per liter) as a risk-management benchmark for comparison against measured or modeled drinking-water concentrations, while independently reconstructing individual dose using site-specific water concentration data, default or individualized water-intake assumptions, and body-weight data appropriate to the individual or exposure period in question. A methodologically important limitation specific to GenX, discussed further below, is that its comparatively short serum half-life and lower bioaccumulation potential mean that serum biomonitoring — the primary individual exposure-reconstruction tool RoseyFields uses for legacy long-chain PFAS such as PFOA and PFOS — is considerably less reliable for GenX, since an individual’s current serum concentration may substantially understate a historical peak exposure that has since been eliminated from the body; RoseyFields’ assessors therefore place correspondingly greater reliance on environmental monitoring data, water-utility records, and exposure modeling to reconstruct historical dose in GenX matters than in comparable PFOA or PFOS matters.

In regulatory and permitting contexts, RoseyFields’ assessors also use the chronic and subchronic RfDs, together with site-specific exposure assumptions, to evaluate proposed remediation targets, to comment on state-level groundwater or surface-water standards (several states, including North Carolina, have separately promulgated or proposed their own GenX standards informed by, but not always identical to, EPA’s federal values), and to evaluate the adequacy of a responsible party’s proposed cleanup levels at contaminated sites under state remediation programs or the Comprehensive Environmental Response, Compensation, and Liability Act. Because EPA’s GenX values are calculated at the federal level using default national exposure assumptions, RoseyFields’ assessors routinely adjust the underlying HQ/HI calculations to reflect site-specific or individual-specific water consumption, body weight, and exposure duration data where such data are available, consistent with EPA’s own guidance that default assumptions are intended to be refined with case-specific information whenever it exists.

Key Uncertainties and Sensitive Populations

RoseyFields’ assessors identify several uncertainties that are particularly important to disclose when applying the GenX RfDs and cancer characterization to a specific case.

First, the critical study underlying both reference doses (DuPont-18405-1037, 2010) is an unpublished, industry-submitted regulatory study rather than an independently published, peer-reviewed journal article, even though it was conducted to recognized OECD test-guideline and Good Laboratory Practice standards and was reviewed by EPA scientists, external peer reviewers, and the National Toxicology Program’s Pathology Working Group. RoseyFields’ assessors disclose this distinction explicitly, while noting that the underlying liver-toxicity finding is independently corroborated by multiple published, peer-reviewed studies (Rushing et al., 2017; Caverly Rae et al., 2015; Heintz et al., 2023; Thompson et al., 2023).

Second, there is active, ongoing, and industry-funded scientific debate over both the interpretation of the critical study’s histopathology and the human relevance of the underlying mode of action, and RoseyFields’ assessors note that EPA has already directly confronted an earlier round of this debate. Before finalizing its 2021 assessment, EPA commissioned an independent National Toxicology Program Pathology Working Group (eight pathologists) to re-review the original liver-lesion slides from the critical study, specifically because a manufacturer-funded reanalysis (Thompson et al., 2019) had argued that individual liver lesions, considered separately, did not support a treatment-related adverse effect finding. The NTP Pathology Working Group’s independent consensus review confirmed the original study’s findings and explicitly disagreed with the Thompson et al. (2019) reanalysis, concluding that the constellation of lesions considered together, not any single lesion in isolation, represents the biologically meaningful marker of adversity. A related and more recent line of manufacturer-funded mechanistic work (Heintz et al., 2023; Thompson et al., 2023; Thompson et al., 2026) argues that the same liver lesions are mediated by activation of the rodent-specific peroxisome proliferator-activated receptor alpha (PPARα) pathway, a mechanism these authors contend has substantially reduced relevance to humans, and on this basis argues EPA’s RfD remains overly conservative. RoseyFields’ assessors track this literature actively, disclose its funding source when it is used to challenge EPA’s values, and note that EPA’s own final 2021 assessment — which was finalized after the Thompson et al. (2019) critique and considered it directly via the independent NTP re-review — did not adopt either the original or the more recent human-relevance arguments in deriving its final reference doses. This history is an important point when opposing experts rely on this body of literature to argue for a higher acceptable exposure level than EPA’s own values would support, since it shows EPA’s position was reached after, and in spite of, direct engagement with the manufacturer-funded critique rather than in ignorance of it.

Third, no validated PBTK model and no adequately peer-reviewed human half-life data exist for GenX chemicals, meaning the animal-to-human dose conversion relies on a generic default (BW^3/4 scaling) rather than chemical-specific human toxicokinetic information; this is a meaningfully different — and less chemical-specific — situation than RoseyFields encounters with PFOA and PFOS, for which extensive human serum half-life and toxicokinetic data exist.

Fourth, no inhalation toxicity value (RfC) exists for GenX chemicals, which is a significant gap for any matter involving a documented or suspected air-emission or vapor-phase exposure pathway near a fluoropolymer-manufacturing or processing facility; in such matters, RoseyFields’ assessors must rely on route-to-route extrapolation from the oral RfD, itself an additional source of uncertainty that must be disclosed.

Finally, sensitive populations warranting particular attention in GenX matters include pregnant women and the developing fetus, given developmental effects reported in gestational-exposure studies, including accelerated parturition, altered maternal and fetal glucose and lipid metabolism, reduced birth weight, and delayed genital development in offspring (Conley et al., 2019); and residents of communities with historically elevated drinking-water or, in more limited circumstances, air exposure near known GenX manufacturing or discharge sites, for whom cumulative historical exposure may be considerably higher, and considerably more difficult to reconstruct from current serum biomarkers alone, than a snapshot environmental or biomonitoring sample would suggest.

Bibliography