Are replacement PFAS chemicals actually safer?
A walk-through of how the unified MCP server answers a cross-domain public-health question by federating environmental monitoring, adverse outcome pathways, gene-disease associations, and transcriptomic datasets across seven knowledge graphs.
GenX, the chemical that replaced PFOA, is detected at 73% the frequency of legacy PFOS in U.S. water systems and was the most potent PPARα activator of 16 PFAS tested by the EPA — convergent evidence across seven graphs that compound-by-compound substitution is not reducing risk.
The scenario
PFAS (per- and polyfluoroalkyl substances) are “forever chemicals” used since the 1950s in non-stick coatings, waterproof textiles, firefighting foams, and food packaging. As evidence of harm from PFOS and PFOA mounted, manufacturers introduced replacements like GenX (HFPO-DA) and ADONA, marketed as safer. Whether the replacements are actually safer, or whether they hit the same biological targets, is the question.
How the assistant approached it
The answer requires linking four different kinds of evidence. The MCP server federated seven Proto-OKN knowledge graphs plus three external tools to produce it:
- Environmental contamination —
sawgraphfor water-system monitoring of monitored PFAS compounds. - Mechanistic toxicology —
biobricks-aopwikifor adverse outcome pathways, plus PubMed for the EPA's 16-PFAS receptor-binding study. - Disease and pathway evidence —
spoke-okn,ubergraph, andwikidatafor gene-disease associations and functional annotation; g:Profiler for pathway enrichment of the 10 most-implicated PFAS genes. - Experimental datasets —
ndefor transcriptomic studies, plusspoke-genelabfor differential-expression validation across spaceflight and other stress contexts.
Cross-graph joins used gene symbols, Ensembl IDs, and chemical compound identities as shared identifiers. The same chemical (GenX) and the same gene (PPARA) can be referenced consistently across SAWGraph, NDE, SPOKE-OKN, and PubMed because of that shared identifier substrate.
Findings
Replacement PFAS are already ubiquitous
SAWGraph tracks 25 distinct PFAS compounds across U.S. water systems. Replacement chemicals are not trace contaminants — they are already present at frequencies comparable to the legacy compounds they were meant to replace.
| Compound | Observations | Status |
|---|---|---|
| PFOS | 23,086 | legacy — restricted |
| PFHxS | 22,132 | legacy |
| PFBA | 21,470 | legacy |
| PFOA | 20,399 | legacy — restricted |
| PFPE-diacid | 17,694 | emerging — in current use |
| GenX (HFPO-DA) | 16,954 | replacement — in current use |
| ADONA | 15,804 | replacement — in current use |
GenX appears at 73% the frequency of PFOS, ADONA at 68%. Exposed populations encounter complex PFAS mixtures, not individual chemicals.
The replacement is more potent at the same target
The strongest mechanistic finding draws from PubMed literature and NDE transcriptomic datasets together:
- 16-PFAS comparison (Evans et al., EPA 2022). In PPARα receptor-binding assays across 16 PFAS, GenX was the most potent PPARα activator — lowest effective concentration, highest fold induction, highest area under the curve.
- PPARα knockout proof (NDE GSE212294). In PPARα-KO mice, GenX's hepatic effects disappear completely; PFOA retains some effects through alternative receptors. GenX is a purer PPARα activator than the legacy compound it replaced.
- Transcriptomic concordance (Heintz et al. 2024; NDE GSE248251). GenX's transcriptomic profile most closely matches the prototypical PPARα agonist drug GW7647 across mouse, rat, and human hepatocytes — not a general toxicant signature.
- Effects at drinking-water concentrations (Shi et al. 2023). GenX disrupts hepatic lipid metabolism via PPARα signaling at 0.1 and 10 µg/L — concentrations already documented in SAWGraph.
- The pattern repeats (Jackson et al. 2024). Next-generation perfluoroether acids PFO4DA and PFO5DoA also activate PPARα. The substitution cycle continues.
NDE returned 11 GenX-specific transcriptomic studies spanning mouse, rat, human hepatocytes, marsupial blood, Drosophila brain, and zebrafish embryos — consistent PPARα signaling and lipid-metabolism enrichment across systems.
AOPs and pathway enrichment confirm the mechanism
AOP-Wiki encodes 40 liver-related adverse outcome pathways; three are directly PFAS-relevant:
- AOP 166 — PPARα activation → hepatocellular adenomas/carcinomas
- AOP 220 — CYP2E1 activation → liver cancer via oxidative stress
- AOP 213 — inhibition of β-oxidation → non-alcoholic steatohepatitis (NASH)
g:Profiler enrichment of the 10 most-implicated PFAS genes (PPARA, CYP2E1, ACOX1, FABP1, SREBF1, NR1I2, CYP1A1, CYP3A4, NR1H4, ABCB11) returned 168 significantly enriched terms, converging sharply on lipid metabolism and PPAR signaling:
| Pathway (source) | p-value |
|---|---|
| Metabolism of lipids (Reactome) | 1.0e-13 |
| PPARα activates gene expression (Reactome) | 6.9e-12 |
| PPAR signaling pathway (KEGG) | 3.5e-9 |
| Alcoholic liver disease (KEGG) | 3.3e-7 |
| Chemical carcinogenesis — DNA adducts (KEGG) | 9.3e-6 |
| Non-alcoholic fatty liver disease (KEGG) | 2.4e-3 |
| Fatty acid metabolic process (GO:BP) | 1.2e-15 |
The 10 PFAS target genes form a coherent functional module. Any chemical that activates PPARα — legacy or replacement — enters this same molecular network.
Two genes, two mechanisms, converging on the same diseases
SPOKE-OKN gene-disease associations for the two key PFAS targets:
- PPARA → liver disease (direct), kidney cancer, hypertension, diabetes mellitus, obesity
- CYP2E1 → liver cancer (direct DE), mitochondrial disease, steroid metabolism disease, kidney cancer (8 independent paths via shared mitochondrial pathways)
PPARα (lipid metabolism) and CYP2E1 (oxidative stress) converge on liver cancer and kidney cancer through distinct mechanistic routes confirmed in separate knowledge graphs — exactly the kind of independent-paths convergence that supports causal inference.
Cross-species datasets validate the picture
A broader NDE query returned 20+ PFAS transcriptomic datasets spanning carp, zebrafish, mouse, rat, and human models — including the PFOA-induced NAFLD mouse model, PFAS human liver spheroids (TempO-Seq), and the GenX hepatic effects dataset that experimentally confirms a replacement PFAS produces the same liver-tissue transcriptomic signature as legacy compounds.
Wikidata independently annotates PPARA with cholesterol homeostasis, lipid response, and fatty-acid metabolic process. SPOKE-GeneLab shows PPARA differential expression across 10 NASA spaceflight assays, confirming it as a bona fide stress-responsive gene rather than an artifact of PFAS-specific experimental design.
Synthesis: convergence across independent lines
The case against compound-by-compound regulation rests on independent lines of evidence pointing at the same conclusion:
| Evidence | Source |
|---|---|
| GenX at 73% the detection frequency of PFOS | SAWGraph |
| GenX is the most potent PPARα activator of 16 PFAS tested | PubMed (Evans et al.) |
| GenX effects entirely PPARα-dependent | NDE (GSE212294) |
| GenX profile matches PPARα agonist across 3 species | PubMed + NDE |
| GenX disrupts lipid metabolism at drinking-water concentrations | PubMed (Shi et al.) |
| PPARα activation → liver tumors | AOP-Wiki |
| 10 PFAS genes enriched for PPAR signaling (p=3.5e-9) | g:Profiler |
| PPARA and CYP2E1 both link to liver and kidney cancer | SPOKE-OKN, Ubergraph |
| 20+ PFAS datasets spanning fish to human hepatocytes | NDE |
| PPARα annotated for lipid/cholesterol metabolism | Wikidata |
| Next-gen replacements (PFO4DA, PFO5DoA) also activate PPARα | PubMed (Jackson et al.) |
No single database holds this picture. SAWGraph knows contamination but not biology; PubMed has the potency comparison but not the prevalence; NDE provides transcriptomic proof but not disease outcomes; AOP-Wiki maps the pathway but not which chemicals trigger it; SPOKE-OKN knows disease associations but not exposure. Only by querying across the federation can the full argument come together.
Bottom line
The federated answer supports class-based regulation of PFAS rather than compound-by-compound substitution: the replacement is more potent at the same molecular target, effects occur at existing environmental concentrations, the next generation already repeats the pattern, and biomarker candidates from the PPARα gene network would detect harm from any PPARα-activating PFAS. Each of those claims rests on a different graph, but the conclusion is the convergence.