PFAS, blood lipids and the kidney: two studies on the same chain

Raised cholesterol is the oldest and least contested item on the list of PFAS-associated outcomes. It is also the least intuitive: a persistent industrial chemical that alters lipid handling does not fit the mental model most people have of environmental contamination. Two studies published within two weeks in August take that association and push on it from opposite directions — one asking what it leads to, the other asking what modifies it.

Lipids as a route to the kidney

The first study measured serum concentrations of legacy and emerging PFAS, renal function parameters and blood lipid profiles in 1,370 adult residents of Zhejiang Province, China, and quantified 452 lipid metabolites in a subset of 316 participants [s1].

PFAS mixture exposure was associated with reduced renal function, with estimated glomerular filtration rate decreases ranging from 2.2% to 2.9% per quartile increase across the models fitted, and PFOA emerging as a predominant contributor [s1]. The mediation analysis is the paper's point: elevated total cholesterol, low-density lipoprotein cholesterol, and disturbances in fatty acids and phospholipids mediated between 6.3% and 65.9% of the association between PFAS exposure and the decline in renal function [s1]. The authors name PFOA and PFHpA as the compounds most implicated [s1].

Two cautions belong immediately next to those numbers. A mediated proportion spanning 6.3% to 65.9% is not one estimate but a range across many mediators, and its width is itself the finding: some lipid measures accounted for very little, others for a great deal. And mediation analysis assumes a causal ordering — exposure, then lipids, then kidney function — that a study measuring all three at once cannot verify. Reverse causation is a live concern here in a specific way: reduced kidney filtration slows the clearance of PFAS, so impaired kidneys can raise serum PFAS rather than the other way round.

What modifies the lipid association

The second study examined six serum PFAS and a composite lifestyle score among 192 propensity score-matched case-control pairs, using conditional logistic regression, quantile g-computation and Bayesian kernel machine regression [s2]. The lifestyle score was built from smoking, alcohol drinking, physical activity, body shape and diet [s2].

Per interquartile-range increase in ln-transformed concentration, PFOA, PFHxS, PFNA, 6:2 chlorinated polyfluorinated ether sulfonate and perfluoroundecanoic acid were associated with 89%, 110%, 67%, 49% and 42% higher odds of hyperlipidemia respectively [s2]. Mixture analyses showed a joint positive association, at an odds ratio of 1.57 (95% CI 1.22 to 2.02), driven predominantly by PFNA, the emerging alternative 6:2 Cl-PFESA, and PFHxS [s2].

The interaction result is what the paper is built around. Among participants with an unhealthy lifestyle — defined as a composite score of 2 or below, based on the population median — higher levels of PFOA, PFHxS and PFNA were associated with significantly elevated odds of hyperlipidemia [s2]. For 6:2 Cl-PFESA, perfluorodecanoic acid and PFUnDA, associations were generally weaker among participants adhering to a relatively healthy lifestyle, supported by significant multiplicative interactions; negative additive interactions were also observed for 6:2 Cl-PFESA and PFNA [s2].

What this does and does not license

The authors' framing is that lifestyle modification may serve as a potential public health strategy to reduce environmental metabolic hazards [s2]. That is a hypothesis generated by a subgroup analysis, and it needs reading with the appropriate suspicion.

Effect modification analyses split an already small sample. With 192 matched pairs, the subgroups here are small enough that the difference between "significant in one stratum, not in the other" and "significantly different between strata" matters a great deal — and only the latter is real evidence of modification. The paper reports significant multiplicative interactions for three compounds, which is the stronger form of the claim, but only for three of the six [s2].

There is also a confounding structure worth naming. The lifestyle score includes body shape and diet [s2]. Both are related to PFAS exposure through food, and both are related to blood lipids directly. A score built from those components is not a clean modifier; it is partly a proxy for exposure and partly a determinant of the outcome.

Neither study is a trial, and neither shows that changing anything changes anything. This is not grounds for a reader to conclude that diet or exercise offsets a chemical exposure.

What the pair adds up to

The two studies are consistent on the base association: more PFAS in serum, worse lipid profile [s1] [s2]. Where they go beyond it is speculative in both cases — one proposing lipids as the route by which PFAS reach the kidney [s1], the other proposing lifestyle as a lever that dampens the lipid effect [s2]. Both are mechanistic stories layered on cross-sectional data.

What would test them is longitudinal measurement: PFAS and lipids at baseline, kidney function years later, in a cohort large enough to stratify. That study does not yet exist for these populations.