Between 2003 and 2009, 50,884 women in the United States and Puerto Rico joined a federal study for a reason none of them would have chosen: each had a sister, full or half, who had been diagnosed with breast cancer. None had the disease themselves when they enrolled. The National Institute of Environmental Health Sciences, which runs the Sister Study, banked their blood. In the years that followed, some of the women were diagnosed too.
Some of that blood has now been tested for forever chemicals, and the results, published this week in the Journal of Clinical Oncology, reach Washington while the Environmental Protection Agency is deciding whether to give some water systems until 2031 to get two of those chemicals out of the tap. Women with the most PFOA in their blood had a 67 percent higher rate of estrogen receptor-negative breast cancer than women with the least, and for PFOS the gap was 93 percent, according to the NIH release. Neither chemical showed a clear association with breast cancer as a whole.
What the freezer held
The team, led by NIH Stadtman Investigator Alexandra White with Che-Jung Chang as first author, used high-resolution mass spectrometry to measure 17 PFAS in samples from 1,609 participants later diagnosed with breast cancer and from a comparison sample of 2,269 women. They deliberately oversampled ER-negative cases, ending with 324 of them, so that the rarer tumor type would have enough numbers to support a conclusion. Every family in the study already had a breast cancer diagnosis in it, which makes this a higher-risk group than the country at large, and its proportions may not carry over neatly to women without that history.
Comparing the top exposure quartile with the bottom, Healio reports a hazard ratio of 1.67 for PFOA (95% CI, 1.1 to 2.53) and 1.93 for PFOS (95% CI, 1.15 to 3.22) for ER-negative disease, from sampling-weighted Cox models adjusted for confounders and for the other PFAS in the blood. For ER-positive tumors, the hormone-driven majority, there was no consistent association.
| Result | Value (HR) | Lower bound | Upper bound |
|---|---|---|---|
| PFOA, highest vs lowest quartile | 1.67 | 1.1 | 2.53 |
| PFOS, highest vs lowest quartile | 1.93 | 1.15 | 3.22 |
Because the blood was drawn years before anyone was diagnosed, the design goes a long way toward answering the obvious objection that a tumor, or the treatment for it, changed what the lab found. It does not prove the chemicals caused the cancers, and an undiagnosed disease can’t be fully ruled out. But no one is going to randomize women to drink PFOA, and a prospective cohort with blood frozen before diagnosis is the strongest design anyone will actually run.
The subtype is where the stakes sit for patients. ER-negative tumors lack the receptor that hormone-blocking drugs target, so those drugs don’t work against them, and a 24-year analysis of breast cancer trials found that ER-negative disease carried a higher hazard of recurrence in the first five years after diagnosis than ER-positive disease. “We know very little about risk factors for ER-negative tumors,” White told Healio. A measurable, reducible exposure is a lead that field rarely gets.
The researchers then pooled seven prospective studies that had measured PFAS before diagnosis. For ER-negative disease, PFOA carried a relative risk of 1.3 per log increase in exposure (95% CI, 1.05 to 1.62), while PFOS came in at 1.2 (95% CI, 0.91 to 1.58), an interval that points upward but still includes no effect at all. The authors, as GEN reported, described the PFOS evidence as “suggestive.” PFOA cleared that bar. PFOS did not, quite.
| Result | Value (RR) | Lower bound | Upper bound |
|---|---|---|---|
| PFOA, per log increase | 1.3 | 1.05 | 1.62 |
| PFOS, per log increase | 1.2 | 0.91 | 1.58 |
The study did not test how the chemicals might act, and White said so: “We did not study mechanisms as part of this paper, and that is an important next step.” She noted that PFAS “are well established to impact epigenetics, inflammation and immunity.” Those are close to the grounds the World Health Organization’s cancer agency used in 2023 when it classified PFOA as carcinogenic to humans: sufficient evidence in animals, strong mechanistic evidence in exposed people for epigenetic alterations and immunosuppression, and only “limited” human evidence, for kidney and testicular cancer. Breast cancer was not on that list, and one cohort does not put it there. A tumor type that ignores estrogen and still tracks with these chemicals does point researchers toward the nonhormonal routes IARC had already flagged.
The arithmetic a headline leaves out
White refused to oversell her own numbers. “These associations are quite strong, but these are relative hazards,” she said. “On an absolute scale, the risk contributed here is going to be lower.”
Susan Neuhausen of City of Hope, who was not involved in the study, sketched what that might look like. Lifetime risk of any breast cancer is about 13 percent, and 20 to 30 percent of breast cancers are ER-negative, which puts the average woman’s lifetime ER-negative risk at about 3 percent. Applying the top PFOA group’s hazard, she estimated, raises it to about 5.1 percent. That is an illustration, a hazard ratio from a family-history cohort laid over a national average, not a count of extra cases the study observed. Neuhausen called it “a well conducted study” all the same, and added that “any increased risk is of concern.”
The women enrolled when PFOA use had only just begun the sharp decline EPA dates to after 2002, and years before PFOS fell off after 2015, so their blood reflects an exposure era many younger women missed. Body burdens of both chemicals have dropped. That is less reassuring than it sounds, because, as Neuhausen put it, “they have been replaced with other PFAS, and we don’t know what effects they have.” The authors’ first request is for studies of exactly those replacements.
Four parts per trillion, and two more years
Health and Human Services Secretary Robert F. Kennedy Jr. said the study “links two forever chemicals to a higher risk of ER-negative breast cancer and demands further investigation.” The health department put an inconvenient finding about two industrial chemicals in front of the public in plain language, which is what a department answerable to patients rather than manufacturers ought to do.
The Trump administration’s environmental agency has been moving at a different speed. On May 18, 2026, EPA proposed a rule, under docket EPA-HQ-OW-2025-1742, that keeps the 4.0-parts-per-trillion drinking water limits for PFOA and PFOS set in 2024 but would offer two-year federal exemptions, moving compliance from April 2029 to April 2031, to systems in states, territories and Tribes that have not obtained primacy for those limits. Systems testing at or above 12 parts per trillion would have to take short-term steps to cut exposure in the meantime. A companion proposal would rescind the regulatory determinations for PFHxS, PFNA, HFPO-DA and a hazard-index mixture of those three plus PFBS.
Holding the 4.0 line on the two best-studied chemicals deserves credit. Offering two more years to meet it does not. Neuhausen called filtering PFAS from drinking water “imperative”; the proposal treats it as something some utilities can push past the end of the decade. Chang’s line from the NIH release is the hopeful one: exposure “is still potentially modifiable.” For a household, that means a filter. For a country, it means the tap, which is the one piece of this the government controls outright.
EPA stopped taking public comment on July 20, nearly four months before the Sister Study’s blood had its say. The extension is still a proposal. The agency can still read the paper.