If you’d asked me which forever chemicals to worry about in a pregnancy, I’d have named the two everybody names: PFOA and PFOS, the stars of the lawsuits, the contaminated-water maps and the documentaries. A prospective cohort out of Henan Province, China made me put that assumption down. Across the whole cohort, the chemicals that tracked with gestational diabetes were three lesser-studied cousins, PFHpS, PFPeS and PFHxA, and none of them falls under a single US federal drinking-water limit.

The team followed 746 pregnant women, measured eight PFAS in their blood and diagnosed gestational diabetes with an oral glucose tolerance test. Of those women, 183 developed it, or 24.5 percent, which is a lot of sick pancreases for one hospital cohort.

DEVELOPED GESTATIONAL DIABETES
of the Henan cohort
Figure data
MeasureShare
of the Henan cohort24.5 percent
183 of 746 women were diagnosed by oral glucose tolerance test. Source: Environmental Health: PFAS exposure, gestational diabetes and vitamin D status in a Henan cohort (2026)

Then they checked which of the eight chemicals moved with the diagnosis. Three did: PFHpS at an odds ratio of 1.19 (95% CI 1.07–1.32), PFPeS at 1.23 (1.09–1.38) and PFHxA at 1.13 (1.02–1.25). Those are modest odds ratios, but every interval stays clear of 1.0, and none of the three is a famous compound.

ODDS OF GESTATIONAL DIABETES, FULL COHORT (OR)
Figure data
ResultValue (OR)Lower boundUpper bound
PFHpS1.191.071.32
PFPeS1.231.091.38
PFHxA1.131.021.25
The three PFAS, out of eight measured, that tracked with a diagnosis across all 746 women. Source: Environmental Health: PFAS exposure, gestational diabetes and vitamin D status in a Henan cohort (2026)

That obscurity matters, because regulation in this country follows fame. The EPA’s 2024 drinking-water rule set limits for PFOA and PFOS, plus PFHxS, PFNA, GenX and a mixture index that pulls in PFBS. In May 2026 the agency proposed rescinding everything except the famous two, and letting water systems request an extra two years, to 2031, to meet even those. This is an administration that promised to put families ahead of the chemical lobby, and on this it’s shrinking the list. None of the three Henan compounds was on it to begin with.

They don’t get a pass on toxicity, either. In a 2020 zebrafish screen published in Environmental Health Perspectives, PFPeS and PFHpS produced the same developmental damage as PFOS: body-axis defects, swim-bladder defects and hyperactivity. PFHxA has its own EPA toxicological review, finalized in 2023, which names developmental, endocrine, blood and liver effects as its critical hazards. The agency has done the homework on PFHxA, and the compound still has no federal water limit. Somehow the burden of proof keeps landing on pregnant women carrying these chemicals instead of on the industry that put them into the water.

The researchers also did what I wish every exposure study did. Instead of pretending people get dosed with one chemical at a time, they modeled the whole mixture at once, and the combined PFAS burden carried an odds ratio of 1.50 (1.04–2.16) for gestational diabetes, with PFHpS the biggest single contributor. Nobody carries just PFPeS. You carry the soup.

The vitamin D split surprised me most. When the authors sorted women by vitamin D status, the associations clustered in the deficient group, where PFOS (OR 1.12, 1.01–1.25), PFHpS (1.22, 1.09–1.37) and PFPeS (1.26, 1.12–1.42) all tracked with diabetes. PFOS hadn’t registered across the full cohort and only showed up here. In women who weren’t deficient, none of those associations was positive. Same chemicals, same city, same pregnancies, and whether the exposure seemed to turn into disease hinged on one fat-soluble vitamin.

ODDS OF GESTATIONAL DIABETES, VITAMIN D-DEFICIENT WOMEN (OR)
Figure data
ResultValue (OR)Lower boundUpper bound
PFOS1.121.011.25
PFHpS1.221.091.37
PFPeS1.261.121.42
In women who were not deficient, none of these associations was positive. Source: Environmental Health: PFAS exposure, gestational diabetes and vitamin D status in a Henan cohort (2026)

Wait, why would a vitamin stand between a pollutant and a pancreas? I couldn’t let that go, and the biology turned out to be more satisfying than I expected. PFAS are thought to wreck glucose control by driving oxidative stress, inflammation and endocrine disruption, three processes that grind down the insulin-making beta cells. Vitamin D pushes back on all three: it helps beta cells release insulin, it turns down inflammatory signaling and it blunts oxidative stress. So a woman with plenty of it has a buffer sitting right on the pathways the chemicals attack, and a deficient woman has had that buffer pulled out from under her. The vitamin isn’t neutralizing the PFAS so much as reinforcing the tissue the PFAS is trying to wear down.

I’d hold the vitamin D finding loosely, and the authors say so themselves, because the non-deficient group was small and exposure came from a single blood draw. But it doesn’t stand alone. An earlier birth-cohort study found that vitamin status modified the PFAS–diabetes link, and a prospective cohort in Beijing reported that vitamin D did the same, which makes three Chinese populations with the same shape. The broader association carries weight too. A Mount Sinai-led review published in January 2026 found that higher PFAS exposure was consistently associated with gestational diabetes across 129 studies and more than one million participants, and its authors called pregnancy a particularly sensitive window. A 2026 meta-analysis in the Journal of Hazardous Materials pooled the same question.

The loudest dissent came from the US. In March 2026, the ECHO consortium pooled 5,229 pregnancies from 16 American cohorts and found no link between PFAS and gestational diabetes, and its six-PFAS mixture actually ran slightly inverse, with a prevalence ratio of 0.75. I take that seriously, and I also looked at how it was built: one blood sample drawn at some point between 1999 and 2021, only the six PFAS common enough to show up in at least 60 percent of women, and diagnoses pulled from self-report or medical records. The abstract reports no split by vitamin D. If the signal lives in lesser-tracked compounds and in deficient women, could a pooled design like that average it away? I think it could, and I’d love to see the ECHO team rerun their freezer samples to find out.

What I most want answered is the question no cohort can settle: would raising a deficient woman’s vitamin D before or during pregnancy actually lower her risk, or does low vitamin D just travel with some other exposure? That takes a trial, not another observational study, and it’s the one I’d fund first.

If I were pregnant or planning to be, I’d ask for a vitamin D level at my first prenatal visit and correct it if it came back low, and I’d stop assuming the forever chemicals worth fearing are only the ones famous enough to have a lawsuit.

Sources

  1. Environmental Health: PFAS exposure, gestational diabetes and vitamin D status in a Henan cohort (2026)
  2. US EPA: PFAS national primary drinking water regulation and 2026 proposed changes
  3. Environmental Health Perspectives: Developmental toxicity of GenX and other PFAS in zebrafish, Gaballah et al. (2020)
  4. US EPA IRIS: Toxicological review of perfluorohexanoic acid (PFHxA) and related salts (2023)
  5. Environmental Health: Early-pregnancy PFAS, vitamin status and GDM, prospective birth cohort (2026)
  6. Environment International: Vitamin D modifies the PFAS–GDM association, Beijing cohort (2025)
  7. Mount Sinai: Exposure to forever chemicals linked to higher risk of gestational diabetes (January 2026)
  8. EMJ: Gestational diabetes tied to widespread PFAS chemicals (eClinicalMedicine review summary)
  9. Journal of Hazardous Materials: Prenatal PFAS, glucose homeostasis and GDM, systematic review and meta-analysis (2026)
  10. Diabetes Care: PFAS during pregnancy and gestational diabetes in the ECHO Cohort (2026)
  11. Science of the Total Environment: PFAS as a GDM risk factor through interference with glucose homeostasis (2022)