A peer-reviewed paper in Environmental Health Perspectives has identified at least 485 chemicals in US pesticide products linked to breast cancer, according to researchers at the Silent Spring Institute. The finding surfaces a structural question about how the Environmental Protection Agency’s review architecture intersects with the toxicology of the products it regulates.

The EPA’s framework “considers each pesticide ingredient’s individual health risks but does not consider the full pesticide formula or exposure to multiple pesticides,” the paper states. Researchers identified roughly 145 of the 485 chemicals as EPA-registered active ingredients; the remainder fall under the “inert” classification, a category covering approximately 4,000 EPA-approved substances used as surfactants or penetrating agents, including PFAS compounds. Co-author Jenny Kay, of the Silent Spring Institute, characterized the inert classification as a regulatory label rather than a toxicological one: “They are not biologically inert, they are just called inert because they do not target the pest. It is a regulatory blind spot.”

Federal biomonitoring through the National Health and Nutrition Examination Survey detected 84 of the identified chemicals in the general US population, a figure researchers describe as an undercount because the survey does not track most of the 485 compounds. Breast cancer incidence has risen by 1% annually overall and 1.4% annually among women under 50, according to the American Cancer Society. Kay pointed to an additional mechanism the current review framework does not capture: non-monotonic dose responses, in which hormone-disrupting chemicals produce effects at low exposures that disappear at higher exposures. She said this phenomenon has led the EPA both to dismiss certain chemical risks and to disregard the possibility that those chemicals cause cancer at low doses.

Causal structure. Pesticide chemical exposures (active plus inert) function as causes; biological mechanisms including endocrine disruption operate as mediators; breast cancer incidence is the effect; the EPA review framework moderates exposure levels by determining which chemicals reach market; individual susceptibility and other environmental and lifestyle exposures are confounders; industry revolving-door dynamics confound regulatory stringency. Two arrows require explicit absent-arrow assumptions to make identifiability tractable. The framework variable is assumed not to directly cause breast cancer independent of its effect on exposure. The revolving-door variable is assumed not to directly affect breast cancer independent of regulatory outcomes. NHANES biomonitoring functions as an instrument for measuring exposure presence in the population.

Under this structure, the back-door criterion requires conditioning on individual susceptibility, other exposures, and industry-influence variables to identify the causal effect of cumulative pesticide exposure on breast cancer. The current EPA framework does not measure or model the cumulative-exposure variable, so the criterion cannot be satisfied within its architecture. The estimable intervention quantity is the difference in population-level breast cancer incidence under a cumulative-assessment regime versus the current regime; the relevant counterfactual is what incidence would have been had inert ingredients been subject to the same review stringency as active ingredients. The 2023 EPA rejection of the legal petition to begin more stringent reviews of inert ingredients, as documented by Kay, forecloses the counterfactual within the existing regulatory pathway.

Framing tension. Two definitions of “inert” pull in opposite directions: a regulatory definition tied to function (does not target the pest) and a biological definition tied to toxicity (does not produce physiological effects). The EPA’s framework operationalizes the former; the findings on non-monotonic dose responses that Kay cites support the latter. The structural gap is in the assessment architecture itself, not in any single chemical’s listing status. Kay called for expanding California’s Proposition 65 to include the 16 pesticides the study identified as not currently listed, alongside broader cumulative-assessment reform. Prop 65 expansion functions as a labeling intervention rather than an exposure intervention: it changes information flow to consumers without altering the regulatory test applied to chemical approvals. Within that framing, adding chemicals to Prop 65 operates inside the existing ingredient-by-ingredient framework rather than changing it.

Forward trajectories. Under a regulatory reform scenario, cumulative-assessment requirements are added to EPA review and the 2023 petition’s recommendations are implemented. The 2023 EPA rejection of that same petition, according to Kay, marks the recorded resistance such reform has already encountered. Under a status-quo scenario, breast cancer incidence continues rising at the reported rates, with the under-50 trend at 1.4% annually, while the EPA review framework remains unchanged. Under a partial-reform scenario, Prop 65 or state-level measures expand disclosure while federal review architecture remains intact; this addresses information gaps but leaves the assessment-structure question unresolved.

Continued incidence growth under the status-quo trajectory, combined with a regulatory framework structurally unable to detect the relevant exposure variable, would produce a widening gap between measured regulatory risk and population-level biological risk over time. The Silent Spring Institute’s broader systematic effort has identified approximately 500 breast cancer-linked chemicals in consumer goods and 462 in drinking water in prior work; the pesticide count of 485 represents a parallel layer of the same exposure architecture rather than a separate domain. A separate report on atrazine contamination, affecting approximately 25% of the nation’s water intake and roughly 675,000 wells, sits within this same layer.

The paper’s characterization of top EPA pesticide and chemical safety office officials as former industry lobbyists, and its description of the pesticide office as having been viewed as heavily infiltrated by industry, provides context for the 2023 petition rejection. Asked about industry influence, Kay said: “I’m not in the head of EPA scientists, but I do worry a lot about industry pressure and the revolving door between EPA and industry employees because all of that factors into the way the EPA considers and regulates chemicals.” The paper treats this pattern as a factor in regulatory outcomes without establishing causation between specific personnel moves and specific regulatory decisions.

The evidence reported in the paper and the broader breast-cancer-chemicals inventory supports the following mismatch: the EPA’s review framework models a flat causal structure (single ingredient → single risk), while the underlying toxicology operates through an interactive structure (multiple ingredients, formula effects, non-monotonic dose responses). No single chemical listing or Prop 65 expansion closes that gap; closing it requires changing the variable the regulatory test measures.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Causal DAG
Maps cause and effect as an explicit directed graph, exposing confounders and mediators (Pearl).
Wicked Futures
Explores a long-horizon, deeply entangled future with no clean resolution.
Wicked Problems
Treats a problem as wicked — no stopping rule, no clean test of success, every attempt consequential.