PFAS in pregnancy: a hypertension signal in Puerto Rico, smaller babies in China

Hypertensive disorders of pregnancy are among the outcomes the C8 Science Panel flagged as a probable link with PFOA exposure, and human studies since have been inconsistent about whether the association holds in populations with ordinary exposure levels [s1]. Two cohort analyses published in August add data from very different settings and arrive at findings that are worth reading side by side, mostly because of how differently they are hedged.

Puerto Rico: blood pressure

The first analysed 548 participants in the PROTECT cohort in Puerto Rico [s1]. PFAS concentrations were measured in maternal serum at 16 to 20 weeks' gestation, and blood pressure at two time points: 16 to 20 weeks and 24 to 28 weeks [s1].

Six compounds were included. PFOA, PFOS, PFHxS and PFNA were detected in more than 50% of samples and modelled continuously; perfluorodecanoic acid and perfluoroundecanoic acid were detected in 25% to 50% and modelled categorically as above or below the detection limit [s1].

For individual compounds against continuous blood pressure measures, the study found positive linear relationships for several PFAS, but describes the estimates as imprecise, with mixture effects that varied [s1]. The clearer result is for the mixture against binary outcomes. The six-PFAS exposure mixture was associated with higher odds of hypertension at 24 to 28 weeks, at an odds ratio of 2.14 (95% CI 1.32 to 3.46) [s1]. High mean arterial pressure at the same visit came in at 1.50 (0.96 to 2.36), and high pulse pressure at 16 to 20 weeks at 1.54 (0.82 to 2.92) [s1].

Only the first of those three excludes the null. The other two are consistent with a real effect of similar size and equally consistent with no effect at all, which is what a confidence interval spanning 1.00 means. The paper is careful to say the second two hold "to a lesser extent" [s1].

The odds of hypertension were higher among nulliparous women and among those carrying female fetuses [s1]. The authors read the timing pattern as suggesting that 24 to 28 weeks may be a more sensitive period for high blood pressure and mean arterial pressure, while 16 to 20 weeks may be more sensitive for pulse pressure [s1]. That is an inference from two measurement points, not from a designed comparison of windows.

China: fetal growth

The second study drew on the Ma'anshan Birth Cohort, measuring maternal serum PFAS in early pregnancy and profiling the metabolome of cord serum [s2]. Outcomes were birth weight, birth length and head circumference, each as a Z-score [s2].

Higher early-pregnancy exposure to PFNA, PFDA, PFOS, PFOA, 6:2 Cl-PFESA and 8:2 Cl-PFESA was associated with lower birth weight Z-score, with individual coefficients between -0.112 and -0.092 [s2]. The PFAS mixture gave a mean coefficient of -0.056 (95% CI -0.103 to -0.009) [s2]. Within the mixture, 6:2 Cl-PFESA carried the highest average relative weight at 14.9% — notable because it is one of the emerging replacement compounds rather than a legacy one [s2].

The sex interaction was significant, at -0.122 (95% CI -0.227 to -0.017), with negative associations in females (mean -0.115, 95% CI -0.201 to -0.031) and not in males [s2].

For mechanism, the meet-in-the-middle analysis identified phenylalanine metabolism and steroid hormone biosynthesis as pathways jointly related to exposure and fetal growth in the overall population, with arginine and purine metabolism identified among females [s2]. The high-dimensional mediation analysis produced what the authors call exploratory candidate intermediate metabolite features [s2]. That phrasing is doing real work: these are candidates, not demonstrated intermediates.

Reading them together

The two studies do not test the same hypothesis, and neither replicates the other. What they share is a design — serum PFAS measured once in pregnancy, outcomes measured later in the same pregnancy — and its consequences.

A single serum measurement stands in for exposure across gestation, and PFAS concentrations change during pregnancy as plasma volume expands. Both studies also face confounding by the same variables that determine diet, water source and housing. Neither is randomised, and neither could be.

The effect sizes in the fetal growth study are small in absolute terms: a coefficient of -0.056 on a birth weight Z-score is a fraction of a standard deviation, well inside normal variation for any individual baby [s2]. Its relevance is population-scale, not clinical.

The sex-specific finding is the most likely of these results to be over-read. It appears in both papers in different forms — female fetuses in the Puerto Rico hypertension analysis, female infants in the Chinese birth weight analysis [s1] [s2] — and that convergence is interesting. It is also exactly the kind of subgroup result that appears by chance often enough to demand replication before it becomes a claim.

What to watch

Both cohorts are ongoing. The questions that would firm this up are whether the hypertension association holds with repeated PFAS measurement across trimesters, and whether the sex interaction in birth weight reproduces in an independent cohort. Neither paper claims to have settled either.

Sources

• [s1] Association of PFAS exposure with maternal blood pressure and hypertension during pregnancy in the PROTECT cohort. Environmental Pollution, 4 August 2026. https://doi.org/10.1016/j.envpol.2026.128902
• [s2] Metabolomic signatures and sex-specific associations of prenatal PFAS exposure with fetal growth: A meet-in-the-middle and high-dimensional mediation analysis. Environment International, 14 August 2026. https://doi.org/10.1016/j.envint.2026.110461