LL2-10 hindsight check: Bisphenol A: contested science, divergent safety evaluations (Gies and Soto), Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Ch. 10, pp. 215–239#

Checked 25 September 2026. The check covers what happened between publication (January 2013) and September 2026 that bears on the chapter’s claims, evidence, predictions and recommendations. Page numbers are report pages (PDF page minus 2). All web sources were accessed on 25 September 2026.

Author context. The authors are Andreas Gies (German Federal Environment Agency, UBA) and Ana Soto (Tufts). Soto stayed an active party to the dispute after 2013: - She was one of the university grantees in the US CLARITY-BPA programme (listed as a collaborator in the NTP compendium, NTP RR-18, Oct 2021). Her laboratory’s mammary-gland study (Montévil et al. 2020) was examined in detail by EFSA in 2023 (Claim 2). - She co-authored the 2020 integration of eight CLARITY academic studies (Heindel et al., Reprod Toxicol, Dec 2020). - She is one of 41 co-authors of a 2024 commentary that “strongly endorse[s]” EFSA’s 2023 TDI and criticises the German Federal Institute for Risk Assessment (BfR) and the US FDA (vom Saal et al., EHP, April 2024).

None of this makes the chapter wrong. It does mean that much of the later literature “vindicating” it comes from the same scientific network that wrote it. Below, independent checks carry more weight: EFSA’s own re-analyses and the formal dissents of other agencies.

Annex 3 note. Not applicable. BPA is a new case in the 2013 report, not one of the 2001 cases updated in Annex 3.

Access note. The web-search quota for this session had run out before this check began. Everything below was verified by fetching primary sources directly: - PubMed/PMC E-utilities, including the full texts of EFSA’s 2023 BPA opinion, EFSA’s 2021 opinion on non-monotonic dose responses, the 2018 ECHA/EFSA endocrine-disruptor guidance and vom Saal et al. 2024; - Crossref; - the EU Publications Office (Cellar), for the texts of EU legislation and Court of Justice judgments; - the US Federal Register API; - web pages and PDFs from EFSA, EMA, the EEA, the FDA, the French Conseil constitutionnel and PARC.

Several sources could not be reached: - the ECHA website (403); - OECD test-guideline pages (403); - Wiley’s EFSA Journal pages (403); - the EUR-Lex front end (bot challenge).

BfR’s own Opinion 018/2023 was not fetched. Its position is taken from the joint EFSA–BfR divergence report and from Kortenkamp et al. (2024). The FDA’s February 2018 statement on CLARITY-BPA returned 404 and is cited via Vandenberg et al. (2019).

Overview#

In the EU, the chapter’s central regulatory argument has been vindicated more fully than most Late Lessons chapters can claim. Its scientific scaffolding has fared unevenly, and the dispute it described has not closed. If anything it has widened.

What has been vindicated or strengthened - EFSA reversed itself, by orders of magnitude. The tolerable daily intake (TDI) went: - from 50 µg/kg bw/day (2006, reaffirmed 2010); - to a temporary 4 µg/kg bw/day in January 2015 (EFSA news, 21 Jan 2015); - to 0.2 ng/kg bw/day in April 2023 (EFSA CEP Panel, EFSA J, 19 April 2023).

That is a 20,000-fold cut in 2023 and a 250,000-fold cut overall. EFSA found that mean and high-percentile dietary exposure in all age groups exceeds the new TDI “by two to three orders of magnitude”. - The critical effect came from the kind of study the chapter said was being wrongly excluded (pp. 221, 223). It was an academic, non-guideline, single-laboratory mouse study (Luo et al. 2016, splenic Th17 cells after gestational and lactational exposure). - The EU acted on it: - BPA was reclassified in 2016 as toxic to reproduction category 1B, applying from 1 March 2018. This is the CLP equivalent of the old Category 2 that the Rapporteur recommended before the 2004 Category 3 outcome the chapter blames on industry advocacy (p. 225). - It was listed as a substance of very high concern (SVHC) three times in 2017–2018, and the courts upheld all three listings. - Its use in thermal paper was restricted from 2020, and in toys from 2018 (with a full ban applying from 2030). - It was banned from food contact materials by Regulation (EU) 2024/3190. The main transition period ended on 20 July 2026.

Together these substantially implement the chapter’s precautionary recommendation (p. 226), about 12 years later and justified by a risk finding, not by precaution. - The EEA itself reported in September 2023 that 92% of adults sampled in 11 European countries had urinary BPA above the health-based biomonitoring value derived from EFSA’s new TDI (EEA briefing, 14 Sept 2023). - Test guidelines moved as predicted (p. 229). OECD reproductive screening guidelines 421/422 (2016) and the prenatal developmental guideline 414 (2018) added endocrine-sensitive endpoints; the 2018 update to 414 explicitly added anogenital distance and thyroid hormones. - Evidence-selection rules proved to be the battleground (p. 223, fn 2). The divergence is now documented in formal inter-agency reports: EFSA–BfR (2023) and EFSA–EMA (2023).

What was wrong, overstated or overtaken - The dispute did not resolve. BfR proposed an alternative TDI of 0.2 µg/kg bw/day, 1,000 times EFSA’s. EMA formally dissented. The FDA’s position is unchanged. The German Research Foundation’s (DFG) Senate Commission on Food Safety (SKLM) judged EFSA’s critical study unsuitable. The SKLM counts five health-based guidance values proposed since 2006, spanning a factor of 250,000 (Leist et al., Arch Toxicol, July 2024). - The joint study meant to settle it reproduced the split. CLARITY-BPA was government-designed and government-funded in both arms: - its guideline (GLP) core study found “no BPA-related effects” in in-life data and no consistent lesion pattern below 25,000 µg/kg (Camacho et al., FCT, Oct 2019); - the university grantees, working on animals from the same exposed litters, reported effects at 2.5 µg/kg.

The divide therefore tracks study paradigm, not only funder. This weakens funding as the explanation (Table 10.2) and strengthens the “different languages” diagnosis (p. 229). - Non-monotonic dose responses did not carry the case. EFSA’s Scientific Committee (2021) accepted non-monotonic dose responses as real, especially for receptor-mediated effects. But its BPA case study found “no clear indications” of them. EFSA’s 2023 TDI rests on a dose-related, monotonic effect, and EFSA judged the non-monotonic claim from Soto’s own laboratory “weak and inconclusive”. - The free-BPA serum values (p. 224) did not survive. A controlled NIEHS/FDA study found peak unconjugated BPA of 1.5 ng/ml after a 100 µg/kg oral dose (Thayer et al., Environ Int, Oct 2015). The chapter’s 4–6 ng/ml in unexposed mothers is therefore implausible without contamination. - Human epidemiology stayed at the level of “warning signs” (p. 226). EFSA 2023 rated no human outcome “Likely”. It rated neurodevelopment, fetal growth and adult male fertility “Not Likely”. - Substitution by other bisphenols happened; the chapter did not anticipate it. BPS and BPF rose, and the EU now regulates hazardous bisphenols as a group. - The US has not moved. The FDA still says BPA is safe at current levels. A 2022 petition to revoke its food-contact authorisations has had no Federal Register decision.

How to read the lessons now. The most durable lessons are institutional: - the rules deciding which evidence and which endpoints count can move a “safe” level by orders of magnitude on the same body of evidence (pp. 221–223); - guideline tests have blind spots (pp. 222–223, 229); - markets can move before regulators (p. 225); - agencies with shared evidence still diverge (pp. 222–223).

All four gained strong independent support after 2013. The chapter’s specific scientific mechanisms are weaker guides: non-monotonicity (pp. 217–219), measured free-BPA serum levels (p. 224), and funding source as the cause of the evidence split (pp. 228–229). The eventual regulatory turn came by a different route from the one the chapter pointed to: - systematic-review protocols; - lower tested doses; - acceptance of intermediate (non-apical) endpoints as adverse; - toxicokinetic scaling.

Claim-by-claim assessment#

Claim 1: Low-dose developmental effects of BPA in rodents are established, with “far-reaching agreement” in the scientific community; at least 46 peer-reviewed studies report effects at oral doses of 50 µg/kg bw/day or less (pp. 219, 221; Table 10.1, pp. 227–228)#

Original claim. Since Colerangle and Roy (1997), “hundreds of papers on low-dose effects” have shown developmental sensitivity, latent effects and non-monotonic curves. “Within the scientific community, there is far-reaching agreement on these concepts and findings” (p. 219). At least 46 studies report effects at or below the then EFSA ADI of 50 µg/kg (p. 221). The chapter also concedes that “the existence and plausibility of such effects are still disputed by some scientists and members of regulatory bodies” (pp. 217–218).

Subsequent developments

Regulatory evidence syntheses moved decisively toward the chapter’s position: - EFSA 2015 (EFSA J 13(1):3978, Jan 2015, as summarised in the 2023 opinion): - judged mammary-gland proliferation “Likely”; - judged reproductive, neurobehavioural, immune, metabolic and cardiovascular effects “as likely as not” (ALAN); - took these uncertainties into account through an extra uncertainty factor on a kidney-weight reference point; - found that the aneuploidy BPA induces in vitro (compare Hunt et al. 2003, Table 10.1) “was not expressed in vivo”. - EFSA 2023 (EFSA J 21(4):6857, 19 April 2023) was a systematic review of literature from January 2013 to 15 October 2018, run under a protocol that had undergone public consultation. It judged many animal clusters “Likely”, including: - cellular immunity and allergic lung inflammation; - all three neurotoxicity clusters (neuromorphology, nervous-system function, behaviour); - several reproductive endpoints.

Endpoints other than Th17 cells (the ratio of primordial to total ovarian follicles, sperm motility and uric acid) had benchmark-dose lower bounds “only slightly higher (up to sevenfold)” than the critical one. All were “several orders of magnitude lower” than the 2015 reference point. - ECHA’s Member State Committee agreed unanimously in 2017 that BPA is an endocrine disruptor with “probable serious effects” on human health. The General Court upheld this (T-636/17, 20 Sept 2019), and the Court of Justice dismissed the appeal (C-876/19 P, 21 Dec 2021).

Replication, the specific test the chapter’s critics demanded (p. 219), gave a split answer: - The FDA’s own work found high-dose effects only. Its subchronic dose-finding study (gestation day 6 to postnatal day 90) found “clear adverse effects” only at 100,000 and 300,000 µg/kg, across seven doses from 2.5 to 2,700 µg/kg (Delclos et al., Toxicol Sci, May 2014). - The CLARITY-BPA core study found nothing consistent at low doses. This two-year guideline study (FDA NCTR) reported “No BPA-related effects… in the in-life and non-histopathology data”. Lesions were “variable across control and BPA-treated groups”, with a possible relationship only at 25,000 µg/kg (NTP RR-9, Sept 2018; Camacho et al. 2019). - The FDA read the core study as reassurance. An FDA interim statement of February 2018 said currently authorised uses “continue to be safe” (reported by Vandenberg, Hunt and Gore, Nat Rev Endocrinol, June 2019, who “disagree”). - The CLARITY grantee studies found low-dose effects. Using animals from the same dosed litters, they reported effects in brain, prostate, urinary tract, ovary, mammary gland and heart, “many… at the lowest dose tested, 2.5 µg/kg/day”. Because these effects appeared “in the same animals across organs evaluated in different labs”, the grantees conclude they are “biologically – and toxicologically – relevant” (Heindel et al. 2020; Prins et al., Basic Clin Pharmacol Toxicol, Aug 2019). The NTP’s compendium collates both arms but “does not attempt to integrate the findings” (NTP RR-18, Oct 2021). - Professional-society consensus was reaffirmed by one side. The Endocrine Society’s second statement concluded that recent work had given “a much fuller understanding” of low-dose effects, non-monotonicity and developmental vulnerability (Gore et al., Endocr Rev, Dec 2015). - Opposing statements appeared in parallel. A 2013 open letter from toxicology journal editors, published in several journals, called the European Commission’s endocrine-disruptor approach “scientifically unfounded precaution” (Dietrich et al., Food Chem Toxicol, Dec 2013).

Dissent persisted after EFSA 2023: - BfR “acknowledged that there is evidence that BPA can have this and other effects on the immune system”. But it rejected the Th17 endpoint as a basis for a guidance value (EFSA–BfR divergence report, April 2023). - EMA “does not dispute” that Th17 increases and reduced ovarian follicle counts were observed in mice. It disputes their relevance to humans (EFSA–EMA divergence report, EMA/150385/2023).

Verdict: partly held up. The substantive claim has strong later support from the EU’s lead food-safety assessor and from ECHA’s committee, upheld in court. That claim is that BPA causes effects in rodents at doses far below the old 5 mg/kg no-observed-adverse-effect level (NOAEL) and the 50 µg/kg ADI, especially after developmental exposure. Even BfR’s dissenting TDI (0.2 µg/kg) is 250 times below the ADI the chapter attacked. But “far-reaching agreement” overstated the position in 2013 and still does. The one purpose-built replication attempt, the CLARITY core study, did not reproduce low-dose effects under guideline conditions. EFSA’s 2023 conclusions also rest mainly on post-2013 studies, not on the specific studies listed in Table 10.1.

Implication for weight. Cite the chapter as correct in direction on low-dose developmental toxicity, with EFSA 2023 as the authority, not Table 10.1. Do not cite it for the existence of scientific consensus; the record since 2013 is of entrenched disagreement.

Claim 2: Non-monotonic dose-response curves (NMDRs) are common for hormones and endocrine disruptors, so extrapolating from high to low doses is invalid for them (pp. 217–219)#

Original claim. BPA “frequently produces dose-response curves that are non-linear”, often inverted-U shaped (p. 217, Figure 10.1). Deviations from monotonicity “are very frequent” for hormones (p. 218). The paradigm “if a high dose of a chemical does not cause harm, then a low dose will not either… does not hold true” for endocrine disruptors, and Paracelsus’s paradigms, applied naively, “do not contribute to the protection of human health” (p. 219).

Subsequent developments - EFSA’s Scientific Committee (EFSA J 19(10):6877, 20 Oct 2021) concluded that: - “Observations of NMDR have been confirmed in certain studies and are particularly relevant for receptor-mediated effects”; - there is “no gold standard” for detecting them statistically; - benchmark-dose methods should be used “with caution” where they occur.

It also reported that of 35 BPA in vivo data sets flagged as potentially non-monotonic in a 2016 external review, none met all six checkpoints. Its own BPA case study found “no clear indications of NMDR”. It called for international guidance. - EFSA 2023 set explicit criteria for “indications for an NMDR”: - at least three doses; - adjacent doses showing an effect in the same direction; - biological plausibility or replication.

It re-analysed the one low-risk-of-bias study making the strongest claim, Montévil et al. (2020) from Soto’s laboratory on mammary-gland development in CLARITY animals. It concluded that “the evidence for NMDR in the Montévil et al. study appears weak and inconclusive”, and that the step function the authors used is “neither common nor conventional” for biological processes. The Th17 effect chosen as critical showed “a consistent dose-related increase”. - CLARITY analyses divided along the familiar line: - A checkpoint analysis of the 81 statistically significant core-study findings found only 2 meeting at least five of six checkpoints, and “none of the endocrine-related or reproductive endpoints” (Badding et al., FCT, Sept 2019; authors from the consultancy Exponent; funding not checked here). - The academic integration reported that “many of the responses were non-monotonic” (Heindel et al. 2020). - Critiques existed from the start. A consultancy critique of the main review the chapter cites (Vandenberg et al. 2012) argued that it presented “anecdotes” without systematic evaluation (Rhomberg and Goodman, Regul Toxicol Pharmacol, Oct 2012). It was published before the chapter, which does not engage it. - Other syntheses: - The US National Research Council reviewed EPA’s 2013 draft state-of-the-science evaluation of NMDRs (NRC 2014; report not read here). - The National Academies recommended systematic review as the way to evaluate low-dose endocrine effects (NASEM, July 2017; summary page only). - The Endocrine Society restated non-monotonicity as an established endocrine principle (Gore et al. 2015).

Verdict: partly held up. The general principle is now mainstream in EU assessment practice: non-monotonicity exists, is most plausible for receptor-mediated effects, and requires caution before extrapolating downward. But its application to BPA, which is the chapter’s core use of it, has been weakened. EFSA’s Scientific Committee and the 2023 panel found no convincing BPA non-monotonic curve. The 2023 TDI was driven by a monotonic effect found by testing low doses directly. The chapter’s pairing of Paracelsus with the dose–response assumption also conflates two claims: low doses can have effects, and effects must rise with dose. The regulatory turn vindicated the first, not the second.

Implication for weight. Use the mechanism in its narrow form: methods that assume effects rise steadily with dose can miss effects that appear only at untested low doses, so test the doses of concern directly. Do not present BPA as a demonstrated case of an inverted-U curve overturning risk assessment.

Claim 3: EFSA’s TDI of 50 µg/kg bw/day, based on the industry-sponsored Tyl et al. (2002) GLP study, is not protective; alternative derivations from Nordic-favoured studies and NTP data imply ADIs 380 and 6,250 times lower (0.13 and 0.008 µg/kg bw/day) (pp. 221–222)#

Original claim. EFSA derived its ADI from a NOAEL of 5 mg/kg in Tyl et al. (2002), using “arbitrarily selected data” (pp. 221–222). The chapter’s own arithmetic: - the Nordic minority’s neurobehavioural studies (LOAEL 40 µg/kg) imply 0.13 µg/kg; - the NTP’s 2.4 µg/kg developmental effect level implies 0.008 µg/kg (p. 222).

Subsequent developments

EFSA’s guidance values:

Other assessors:

Comparison with the chapter’s own numbers: - EFSA’s 2023 TDI (0.0002 µg/kg) is about 650 times below the chapter’s Nordic-based figure (0.13) and 40 times below its NTP-based figure (0.008). EFSA went further than the chapter’s most precautionary number. - BfR’s alternative (0.2 µg/kg) sits close to the chapter’s Nordic-based figure. - The chapter’s two “alternative ADIs” therefore roughly bracket where European expert opinion divided a decade later.

Verdict: strengthened. Every EU assessor that revisited BPA after 2013 concluded that 50 µg/kg was far too high. This includes EFSA twice and the dissenting BfR. EFSA’s own reassessment went far beyond the chapter’s figures. The claim remains contested outside the EU (FDA) and on the choice of critical endpoint within it (BfR, EMA, SKLM). But the specific point, that the Tyl-based value was not protective, has no current EU defender.

Implication for weight. This is the chapter’s strongest vindicated claim. Its wider lesson is the one most worth carrying forward: on one heavily studied substance, choices about which studies count, what counts as adverse and how uncertainty is handled produced guidance values spanning a factor of about 250,000 between 2006 and 2023. The chapter predicted such spread and later events confirm it (pp. 220–223).

Claim 4: Human internal free-BPA exposure is higher than EFSA assumes. Measured maternal serum is 4–6 ng/ml; conjugates can be deconjugated in tissues; EFSA’s reliance on rapid conjugation (Völkel 2002) “is far from being precautionary”; effective rodent doses overlap human intake (pp. 224–226)#

Original claim. Two biomonitoring studies measured 4–6 ng/ml free BPA in mothers’ blood. That is “around three orders of magnitude” above modelled estimates of 0.1–10 pg/ml (p. 224). EFSA “ignores consistent results from peer-reviewed scientific work”. Glucuronidases in the placenta and other tissues can reactivate conjugated BPA (p. 224). “The effective doses in these studies overlap the doses of current human intake” (p. 225). “EFSA’s assumption that internal doses of free BPA are lower in humans than in rodents at comparable doses is unproven” (p. 226).

Subsequent developments

Measured serum levels (weakened): - CDC scientists documented the contamination problem. Contamination with BPA during biomonitoring analysis can come from “solvents and reagents, the experimental apparatus used, the laboratory environment, and/or even the analyst” (Ye et al., EHP, March 2013). - A pooled analysis found typical serum levels far below measurability. Teeguarden and colleagues analysed 93 studies (more than 30,000 people). They concluded that typical serum BPA is “orders of magnitude lower than levels measurable by modern analytical methods”, and they “question reports of measurable BPA in human serum” (Teeguarden et al., FCT, Dec 2013; authors include Pacific Northwest National Laboratory and FDA NCTR scientists). - The decisive independent test came from government scientists. An NIEHS/FDA controlled study gave 14 volunteers 100 µg/kg deuterated BPA, twice EFSA’s old TDI. Unconjugated BPA peaked at a mean of 1.5 ng/ml and was “less than 1% of the total… at all times” (Thayer et al., Environ Int, Oct 2015; no competing interests declared). Sustaining the chapter’s 4–6 ng/ml in unexposed mothers would require exposures far above anything in the general population. - Accurate measurement is possible; high levels are not shown. A four-laboratory round robin showed that unconjugated BPA can be measured in serum without contamination (Vandenberg et al., Environ Health, April 2014). It did not show that typical free-BPA levels are in the ng/ml range. - Unusual routes can produce high levels. High transient serum levels (Cmax about 7 ng/ml unconjugated BPA) were produced by handling receipt paper after using a hand sanitiser containing penetration enhancers, then eating with the hands (Hormann et al., PLoS One, Oct 2014). Sublingual absorption, which bypasses first-pass metabolism, was high in dogs (Gayrard et al., EHP, Aug 2013). A human study then found “evidence against sublingual absorption” of BPA eaten in soup (Teeguarden et al., TAAP, Oct 2015). - A related dispute about urine methods has been largely answered. Gerona, vom Saal and Hunt argued that indirect urinary methods underestimate exposure (Lancet Diabetes Endocrinol, Jan 2020); CDC and German biomonitoring scientists replied (Calafat, Koch et al., April 2020; letter not read here). The largest comparison since, in 1,879 pregnant women, found the direct method on average only 8.6% higher, which is “unlikely to alter the interpretation of health outcome data” (Ashley-Martin et al., Environ Int, Dec 2021).

Internal dose in humans versus rodents (strengthened): - EFSA’s 2015 and 2023 opinions reversed the assumption the chapter criticised (p. 226). They converted animal doses to human-equivalent doses using BPA-specific toxicokinetic factors based on blood exposure (area under the curve). The factors are 0.0155 for mice and 0.1656 for rats. In other words, a human is estimated to reach the same internal exposure as a mouse at about 1/65 of the mouse’s oral dose. - This scaling explains much of the 2023 TDI. It is how a mouse benchmark dose of about 0.53 µg/kg (Leist et al. 2024) becomes a human reference point of 8.2 ng/kg (EFSA 2023). - EFSA 2023 cites work on tissue deconjugation. It cites Gauderat et al. (2016), in pregnant sheep, as showing that deconjugation of BPA glucuronide “is a determining factor of fetal exposure”, consistent with the chapter’s deconjugation point (p. 224).

Overlap of effect doses and human intake (strengthened within EFSA’s framework): - EFSA’s reference point is at the bottom of the chapter’s intake range. The 8.2 ng/kg human-equivalent reference point lies at the bottom of the chapter’s adult intake range (0.008–1.5 µg/kg bw/day, p. 224). EFSA found exposure exceeds the TDI by two to three orders of magnitude. - Biomonitoring points the same way. HBM4EU biomonitoring found that in 14 of 15 European data collections, 100% of women exceeded the guidance value derived from the new TDI (Tagne-Fotso et al., Environ Int, Aug 2024). The EEA reported 92% of adults exceeding it (EEA, 14 Sept 2023).

Verdict: partly held up. The headline evidence did not survive: measured free BPA of 4–6 ng/ml in unexposed mothers is now best explained by contamination or unusual exposures. The chapter itself warned of contamination (p. 216) but did not apply the warning here. The chapter’s broader toxicokinetic point, that rodent-to-human extrapolation had understated human internal exposure, was taken up by EFSA in the form of human-equivalent-dose factors. Within EFSA’s framework, animal effect doses now overlap human intake.

Implication for weight. Do not cite the serum values. The durable lesson is methodological: when a contaminant is everywhere, including in laboratories, measurement artefacts can drive a scientific dispute for a decade. Resolving them needs controlled, labelled-dose studies, and those came from government laboratories, not from either side’s routine biomonitoring.

Claim 5: Industry-funded BPA studies systematically find no low-dose effects (0/11 industry-funded versus 94/104 government-funded, Table 10.2, p. 228); independent science and regulatory toxicology “seem to speak different languages” (p. 229)#

Original claim. The tally is from Hughes and vom Saal (2005), covering studies to 2004. “The results of industry-sponsored studies and independent scientific studies deviate strongly” (p. 229). Contract laboratories are “not economically independent”, and academic laboratories “may be better qualified for testing for subtle changes” (p. 229).

Subsequent developments - No updated BPA-specific funding tally was found in this check (PubMed searches returned nothing comparable). The underlying data source is vom Saal and Hughes, EHP, Aug 2005. - CLARITY-BPA was the natural experiment on this claim. Both arms were publicly funded: FDA for the core study, NIEHS for the grantees (vom Saal et al. 2024; NTP RR-9). The guideline arm, run in a government GLP laboratory, found no consistent low-dose effects. The academic arm, working on animals from the same dosed litters, reported many. - The split therefore survived the removal of industry funding. It tracked study design, endpoints and statistical conventions, which is the chapter’s “different languages” point. - Vandenberg, Hunt and Gore judged that CLARITY, “despite its flaws”, provides “compelling evidence” of low-dose effects (Nat Rev Endocrinol 2019). - The FDA read the same programme as confirming safety. - The “different languages” are now written down in official documents: - EMA’s divergence report states that for quantifying risk it relies on GLP guideline studies as “pivotal studies”, with “robust non-GLP studies” only “to support the evaluation”, and requires “apical endpoints to minimise and avoid uncertainty” (EMA/150385/2023). - BfR calls EFSA’s intermediate-endpoint approach a “paradigm shift” “not in line with general practices” (EFSA–BfR report). - Both reports conclude that convergence is “not possible” (BfR) or that there is “no agreement” on what counts as an adverse effect (EMA). - The general funding-effect mechanism has independent support elsewhere. A Cochrane methodology review found industry-sponsored drug and device studies more often report favourable results and conclusions (Lundh et al., Cochrane, Feb 2017; abstract not re-read here). That is evidence about sponsorship in general, not about BPA.

Verdict: partly held up. The “different languages” diagnosis is strengthened: it is now the explicit, documented basis of formal disagreement between EU agencies. Funding source as the explanation is weakened. When industry money was removed and the two paradigms were run on the same animals, the split reappeared. Table 10.2 remains an accurate historical tally, but it confounds funder with study design, and the chapter reads it causally.

Implication for weight. Frame the mechanism as a clash of evidential paradigms (guideline versus hypothesis-driven studies, apical versus intermediate endpoints), with funding as one reinforcing factor. Do not frame it as a pure conflict-of-interest story. The CLARITY result is itself a strong transferable lesson: a jointly designed study cannot settle a dispute if the parties have not first agreed what counts as an adverse effect.

Claim 6: BPA industry advocacy through the Weinberg Group led the EU to classify BPA as a Category 3 reproductive toxicant rather than the Rapporteur’s Category 2, avoiding toxic labelling and authorisation (p. 225)#

Original claim. The chapter quotes the Weinberg Group’s own website (2005): its advocacy “proved very effective, as ultimately the C&L working group did not follow the recommendation of the Rapporteur Member State to classify BPA as a Category 2 reproductive toxicant”. Category 2 would have meant a skull-and-crossbones label, and “every use of BPA would have required a formal authorisation” (p. 225).

Subsequent developments - The historical outcome is confirmed. Commission Directive 2004/73/EC (29th adaptation to technical progress) lists BPA as “Repr. Cat. 3; R62” (Directive 2004/73/EC, OJ L 152, 30 April 2004). - The substantive judgment was later reversed. Commission Regulation (EU) 2016/1179 classified BPA as Repr. 1B (H360F, “may damage fertility”), applying from 1 March 2018 (Reg. 2016/1179, OJ L 195, 20 July 2016). Under the CLP Regulation, Repr. 1B corresponds to the old Category 2. - SVHC identification followed in three steps: - reproductive toxicity, 12 January 2017 (decision ED/01/2017); - endocrine disruption with effects on human health, 6 July 2017 (ED/30/2017, on a French dossier); - endocrine disruption with effects on the environment, 3 January 2018 (ED/01/2018). - The industry association PlasticsEurope challenged all three listings and lost each time: - T-185/17, 11 July 2019; - T-636/17, 20 September 2019; appeal C-876/19 P dismissed 21 December 2021; - T-207/18, 16 December 2020; appeal C-119/21 P dismissed 9 March 2023. - The authorisation consequence was overstated. SVHC status does not by itself trigger authorisation: a substance must also be placed on Annex XIV. BPA “is mainly used as a monomer… It is thus used as an intermediate” (T-207/18, para. 1), and REACH exempts intermediates from authorisation (Art. 2(8)(b), from knowledge of the Regulation). This check found no evidence that BPA has been added to Annex XIV; ECHA’s site was unreachable. EU risk management has run instead through restrictions (thermal paper, toys) and food-contact law (Claim 8). - The causal influence claim remains unverified. No independent documentary evidence was found (for example lobbying records or working-group minutes) that the consultancy’s advocacy caused the Category 3 outcome, beyond the firm’s own marketing claim.

Verdict: partly held up. The documented facts are accurate: the Category 3 outcome, and the consultancy publicly claiming credit. The Rapporteur’s original hazard judgment was restored in 2016–2018 and has withstood three General Court actions and two appeals. The causal attribution rests on a firm’s self-promotion and remains unproven. The claim that Category 2 would have required authorisation for “every use” was wrong in law.

Implication for weight. The mechanism (advocacy aimed at classification decisions, which act as chokepoints for later regulation) is plausible and consistent with how the eventual reclassification cascaded into SVHC listing and food-contact law. Treat the BPA instance as suggestive, not demonstrated, and correct the authorisation point when citing.

Claim 7: Recommendation: the EU should “dare to start again” with a transparent BPA risk assessment “conducted by the scientists authoring the papers with high scientific impact in this field”, with stakeholder conferences to expose the interests and influence of industry and NGOs (p. 226)#

Original recommendation. Quoted in the heading (p. 226).

Subsequent developments - EFSA restarted twice: - 2015 opinion. A weight-of-evidence assessment that included non-guideline studies, preceded by a two-phase public consultation (EFSA 2015 consultation report). - 2016 mandate. The European Commission asked EFSA for a full re-evaluation. - 2017 protocol. A pre-established hazard-assessment protocol, put out for public consultation (EFSA protocol, Dec 2017; consultation report) and piloted (EFSA, Nov 2019). - Call for data from interested parties. - Draft opinion in public consultation from 15 December 2021 to 22 February 2022 (EMA/150385/2023). - Final opinion in April 2023, with responses to comments in its Annex N. - Guideline and academic studies were appraised on the same footing. Both were tiered by risk of bias, and the critical study was academic (EFSA 2023). - The assessment was not led by the high-impact BPA researchers. EFSA’s Food Contact Materials (CEP) Panel and working group carried it out, with named hearing experts. The leading academic BPA researchers were not on the panel. Many of them later endorsed the result publicly: 41 co-signed the 2024 endorsement (vom Saal et al. 2024). - Transparency increased structurally under the Transparency Regulation (Claim 9). Disagreements among EU bodies were made public through joint divergence reports, as EU law requires (EFSA–BfR; EFSA–EMA). - The outcome converged with the chapter’s position, and it did so through a protocol-driven systematic review, not through handing the assessment to the field’s lead authors.

Verdict: partly held up. The substance of the recommendation was implemented, and its premise was vindicated: that a fresh, transparent assessment would change the answer. The specific form was not adopted, and in hindsight it was the weaker part of the recommendation. Assessment by the authors of the contested papers would have compounded the positionality problem noted above. EFSA’s protocol route gave a result that is harder to dismiss as partisan, though it was still contested by other agencies.

Implication for weight. The lesson that survives is “restart with pre-registered, public evidence-appraisal rules”, not “hand the assessment to the leading researchers”. The divergence reports also show that transparency makes disagreement visible; it does not resolve it.

Claim 8: Recommendation: until final decisions, take precautionary measures to cut exposure well below rodent-effect levels by terminating BPA uses involving close contact with humans via food or the environment (p. 226)#

Original recommendation. Quoted in the heading (p. 226).

Subsequent developments

European Union: - Food contact materials. Commission Regulation (EU) 2024/3190 (adopted 19 December 2024, published 31 December 2024, in force 20 January 2025): - prohibits the use of BPA in manufacturing food contact materials, and the placing on the market of such materials (Art. 3); - forbids residual BPA in materials made with other bisphenols (Art. 4); - prohibits “hazardous” bisphenols unless authorised (Art. 5). These are bisphenols with harmonised classification as carcinogenic, mutagenic or reprotoxic (category 1A/1B) or as endocrine disruptors (category 1). BPS is named as classified Repr. 1B (recital 9). - Derogations and phase-in: - Derogations cover polysulfone filtration membranes and epoxy coatings on large tanks and vessels, where “no alternatives currently exist” or replacement would bring “disproportionate costs” (recitals 6–7). - Most single-use and repeat-use articles could be placed on the market until 20 July 2026. Packaging for fruit, vegetables and fishery products, and exterior metal coatings, has until 20 January 2028. - Single-use articles may then be filled for a further 12 months, and packaged food sold “until exhaustion of stocks” (Arts. 11–12). - Regulation (EU) 2026/250 (2 February 2026) corrected drafting errors. It added a 20 July 2027 end date for repeat-use articles first placed on the market by 20 July 2026, and did not postpone the ban. - The 2024 Regulation’s stated rationale is EFSA’s 2023 risk finding, plus the absence of analytical methods able to enforce a limit at the new TDI (recitals 3–4). It is not presented as precaution. - Earlier EU steps: - polycarbonate infant bottles banned from 2011 (Implementing Regulation 321/2011); - specific migration limit cut to 0.05 mg/kg, with bans for infant-food packaging and cups (Regulation 2018/213); - thermal paper ≥0.02% by weight banned after 2 January 2020 (Reg. 2016/2235); - toys limited to 0.04 mg/l migration from 26 November 2018 (Directive 2017/898). From 1 August 2030, the new Toy Safety Regulation (EU) 2025/2509 places BPA under the generic ban on reprotoxic substances and bans “34 bisphenols identified by ECHA”. - The cost-benefit record is instructive. For the thermal-paper restriction, ECHA’s socio-economic committee (SEAC) “concluded that overall the estimated costs outweigh the potential health benefits”. It could not quantify benefits because RAC found the dose–response unquantifiable. SEAC also noted the cost was “a very small proportion” of personnel costs, and the Commission adopted the restriction anyway (Reg. 2016/2235, recital 7).

France: - Law 2012-1442 suspended BPA in all food-contact packaging (as the chapter anticipated). - The Conseil constitutionnel struck out the bans on manufacture and export as disproportionate, and upheld the ban on import and placing on the French market (Décision 2015-480 QPC, 17 Sept 2015, on a PlasticsEurope challenge).

United States: - The FDA’s only actions were amendments based on uses having been abandoned: polycarbonate baby bottles and sippy cups (77 FR, 17 July 2012, on an American Chemistry Council petition) and BPA epoxy coatings in infant-formula packaging (78 FR, 12 July 2013). The FDA states these were “not based on safety”. - A petition from the Environmental Defense Fund, the Endocrine Society and others to revoke or restrict BPA authorisations was filed on 2 May 2022 (87 FR 41079, 11 July 2022). The FDA says it is “reviewing” it (FDA BPA page). No FDA Federal Register action on BPA appeared between August 2022 and September 2026 (Federal Register API search). - EPA included BPA in a December 2024 rule under section 8(d) of the Toxic Substances Control Act (TSCA) requiring submission of health and safety data (89 FR, 13 Dec 2024). It listed BPA among Work Plan chemicals not currently being prioritised (18 Dec 2024). Reporting deadlines were extended in 2025 and 2026.

Substitution: - BPA substitutes are nearly universal in the US. In the US survey NHANES 2013–2014, BPS and BPF were detected in 89.4% and 66.5% of urine samples (Lehmler et al., ACS Omega, June 2018). - BPS and BPF look hormonally similar to BPA. A systematic review found them “as hormonally active as BPA” (Rochester and Bolden, EHP, July 2015). - Exposure has fallen in some regions and risen in others. A 2025 meta-analysis reports falling BPA in the US, Canada and Europe but rising BPA and BPS in parts of Asia and the Middle East (Acevedo et al., Environ Res, Jan 2025). - Exposure in Europe remains above the new guidance value. European biomonitoring shows most adults exceed the value derived from EFSA’s 2023 TDI (EEA 2023).

Verdict: held up. The EU has essentially done what the chapter recommended, and its own risk assessment now says exposure is far above tolerable levels. That retrospectively supports the call for earlier action. The delay was about 12 years from the recommendation to the main compliance date, with a tail into 2029 or later. Implementation came through the ordinary risk route, not an explicit precautionary one, and not at all in the US. The chapter did not foresee that bans on one substance would drive substitution by structural analogues, which the EU has only lately addressed with group-based rules.

Implication for weight. Strong support for the lesson that exposure reduction was justified before scientific closure. Add two later lessons: - substance-by-substance action invites regrettable substitution, so action should target structural groups; - the costs of phase-out are handled through derogations and long transitions, and these extend exposure well past the decision date.

Claim 9: Recommendation: uncouple testing from producer funding through an industry-financed, government-managed fund; contract-laboratory results “must not outweigh those from independent academic laboratories”; strengthen adviser independence, including distance from bodies like ILSI (p. 229)#

Original recommendation. “Research laboratories could be paid by a fund that is financed by the industry, over which industry has no control and which is managed by governments”. Adequate pay and time for independent experts, financed by fees (p. 229).

Subsequent developments - No industry-financed, government-managed testing fund has been created in the EU or the US (none found in this check). - Partial institutional analogues, all publicly funded: - Transparency Regulation (EU) 2019/1381, applying from 27 March 2021 (OJ L 231, 6 Sept 2019). It set up an EFSA database of studies commissioned by business operators, which laboratories must also notify (Art. 32b). It also created “verification studies”: “in exceptional circumstances of serious controversies or conflicting results”, the Commission may ask EFSA to commission independent studies (Art. 32d). - Regulation (EU) 2025/2455 of 26 November 2025, the “one substance, one assessment” data platform (OJ L, 12 Dec 2025). It gives ECHA a “data generation mechanism” to commission studies “using the best independent resources available”, including verification studies in “serious controversy” (Art. 24). It creates an ECHA study-notification database, with obligations applying from 2 November 2027. It also requires the EEA to establish an EU early-warning system for emerging chemical risks by 2 January 2027 (Art. 22). - PARC (Partnership for the Assessment of Risks from Chemicals), launched 11 May 2022 for seven years with €400 million, half from the EU and half from member states. It involves close to 200 institutions plus ECHA, EFSA and the EEA, and aims at monitoring, hazard data generation and new methods (PARC). - Weighting of contract versus academic studies has diverged by agency. EFSA 2023 applied the same appraisal to both, and an academic study set the TDI. EMA still treats GLP studies as “pivotal” for quantifying risk (EMA/150385/2023). - Adviser independence rules tightened. EFSA’s Policy on Independence (adopted 21 June 2017) imposes a two-year cooling-off period for employment, consultancy, advisory-body membership and research funding from “legal entities pursuing private or commercial interests”. It also caps private-sector research funding for experts at 25% of their team’s budget (EFSA policy). This check found no EFSA rule naming ILSI specifically. - Whether these rules changed outcomes cannot be shown. The more precautionary 2023 opinion came after them, but it also used a new protocol, new studies and a new panel. No study isolating the effect of independence rules on BPA outcomes was found.

Verdict: partly held up. The diagnosis has been partly acted on: producer funding and control of evidence is a structural problem, and it needs transparency and a public capacity to commission tests. The EU now has study-notification duties, verification-study powers and a publicly funded research partnership. The specific remedy has not been adopted: an industry-financed, government-managed fund, and a rule that academic results should outweigh contract-laboratory results.

Implication for weight. Treat the recommendation as broadly prescient in direction. Note that institutions preferred transparency plus public verification capacity over redirecting industry money. The claimed link between independence rules and better outcomes remains untested.

Claim 10: Prediction and claims: test strategies are improving, with OECD incorporating hormone-sensitive endpoints such as timing of vaginal opening and anogenital distance (p. 229); human epidemiological associations (child behaviour, obesity, IVF outcomes, birth weight) are warning signs, not proof (p. 226)#

Original claim. “OECD is currently modifying its guidelines and incorporating many new endpoints that are sensitive to hormonal perturbation, such as timing of vaginal opening, and anogenital distance” (p. 229). Sensitive endpoints such as mammary-gland development and neurobehaviour “are not commonly assessed” (p. 221). The listed human associations “are not a proof of causation but should be regarded as additional warning signs” (p. 226).

Subsequent developments

Test guidelines: - OECD guidelines added endocrine-sensitive endpoints. Per the ECHA/EFSA endocrine-disruptor guidance (EFSA J 16(6):5311, June 2018): - OECD TG 421 and 422 “were updated in 2016 to incorporate parameters suitable to detect ‘EATS-mediated’ parameters”. EATS means oestrogen, androgen, thyroid and steroidogenesis pathways. From knowledge, not re-checked on OECD pages, the 2016 additions include anogenital distance, nipple retention and thyroid hormones. - TG 414 (prenatal developmental toxicity) gained “AGD measurement or thyroid hormones measurement” in 2018. - Crossref metadata for TGs 421, 422 and 443 carry a 25 June 2025 date, suggesting further revision (content not checked). - Mammary-gland endpoints remain provisional. TG 443 still recommends that “parameters involving pup mammary glands of both sexes be included, when validated”. - Endocrine disruption became a formal hazard class. The EU added endocrine-disruptor hazard classes to CLP by Delegated Regulation (EU) 2023/707, which Regulation 2024/3190 now uses as a trigger for bans (recital 9).

Human epidemiology: - EFSA 2023 found the human evidence weak throughout. Its assessment of human studies, covering literature to 2018, concluded (EFSA 2023): - neurodevelopment: “Not Likely”, despite the Braun et al. child-behaviour studies the chapter cites; - impaired pre- and postnatal growth (compare the chapter’s Miao et al. 2011 on birth weight), preterm delivery, and male fertility after adult exposure (compare Li et al. 2010): “Not Likely”; - obesity and type 2 diabetes: ALAN; - female fertility (compare the IVF studies) and pre-eclampsia: ALAN; - asthma and allergy: ALAN; - “none of the clusters” of reproductive or metabolic outcomes reached “Likely”. - Umbrella reviews report associations of moderate quality. They list associations with type 2 diabetes, insulin resistance, polycystic ovary syndrome, obesity, hypertension and cardiovascular disease (Symeonides et al., Ann Glob Health, Aug 2024, Minderoo-funded; Lin et al., Environ Res, Nov 2023). Most underlying meta-analyses were of moderate methodological quality. - A structural limit on epidemiology has been quantified. Repeated urine samples in pregnancy showed an intraclass correlation of only 0.32. This “high degree of within-person variability” hampers well-powered studies (Jusko et al., JESEE, Sept–Oct 2014).

Verdict: held up. The prediction about OECD guidelines was borne out, although the endpoints the chapter considered most sensitive (mammary gland, neurobehaviour) are still not routine. The chapter’s caution about human data was well judged: 13 years on, no human outcome has reached “Likely” in EFSA’s assessment. Several of the specific associations it listed were judged “Not Likely”. The case for action has rested on animal evidence plus exposure data, not on human proof, much as the chapter’s DES analogy anticipated (p. 219).

Implication for weight. Supports the mechanism that when exposure is transient and effects are delayed, human proof may never arrive, so demanding it amounts to inaction (p. 219). It also warns against citing the chapter’s list of human associations as evidence of harm.

Cross-cutting observations for using this section as a lens#

1. A “safe” level can be an artefact of the rules used to judge evidence. The same substance had guidance values that moved about 250,000-fold between 2006 and 2023 (SKLM 2024). The EFSA–BfR and EFSA–EMA reports name the drivers: - the definition of adversity (intermediate versus apical endpoints); - which studies are admitted; - how uncertainty analysis is done; - which toxicokinetic scaling factors are used.

In technology-neutral terms: when evidence is plentiful but heterogeneous, the procedural rules for weighing it determine the answer more than any single new study. This strongly confirms pp. 220–223 and fn. 2. 2. Joint studies do not settle disputes about what counts. CLARITY-BPA put regulatory and academic paradigms on the same animals, and each side read the outcome as confirming its view (FDA 2018 versus Vandenberg et al. 2019). Resolving such disputes needs prior agreement on adversity criteria and evidence integration, not just more data. The chapter’s “different languages” (p. 229) is the durable insight; its funding explanation (Table 10.2) is the weaker one. 3. Vindication came by a different route than the one predicted. The chapter staked its case on non-monotonicity, measured serum levels and funding bias. The turn actually came from: - pre-registered systematic-review protocols; - direct testing at low doses; - accepting intermediate endpoints as adverse; - toxicokinetic scaling.

A lesson for using the chapter: it can be right about the policy conclusion while partly wrong about why. 4. Delay is built into the response. It ran from the chapter (2013) to a TDI (2023), a ban (in force 2025), main compliance (July 2026), extended compliance (January 2028), filling (to 2029) and stock exhaustion. The derogations for “critical” uses and long-lived packaging show how the costs of phase-out are managed by stretching exposure over time (Reg. 2024/3190, recitals 5–7, 16–20). 5. Substitution is part of the hazard. “BPA-free” market responses (p. 225, markets moving first) led to widespread exposure to BPS and BPF, analogues with similar hormonal activity. Group-based regulation came about a decade later: Reg. 2024/3190, Art. 5, and the 34 bisphenols in the Toy Safety Regulation. The chapter was silent on alternatives. 6. Costs and who bears them. ECHA’s socio-economic committee could not show that benefits exceeded costs for the thermal-paper restriction, because the dose–response could not be quantified. The restriction went ahead because costs were small relative to the sector. When harm cannot be quantified, cost–benefit tests tend toward inaction unless decision-makers accept precaution explicitly. 7. Jurisdictions and agencies diverge on shared evidence (pp. 222–223 confirmed). By 2026 the EU had banned BPA in food contact while the FDA maintained it is safe. Within the EU, EFSA, BfR and EMA formally disagree, and EU law requires them to publish that disagreement. 8. The institutions of “late lessons” are being formalised. The Transparency Regulation (verification studies) and Regulation 2025/2455 (ECHA data generation, and an EEA early-warning system due by January 2027) directly echo pp. 226–229. Whether they change outcomes remains untested. 9. Author positionality. Soto’s later work (CLARITY grantee, Heindel 2020, vom Saal 2024) makes her both a source for the chapter and a party to its later vindication. EFSA’s independent re-analysis of her lab’s non-monotonic claim (weak) and its adoption of low-dose effects (strong) show why independent adjudication matters. 10. Slips to correct when citing: - Category 2 would not have made “every use” subject to authorisation (p. 225). Authorisation requires Annex XIV listing, and intermediate uses are exempt. - Measured free-BPA serum levels of 4–6 ng/ml (p. 224) should not be cited. - “Far-reaching agreement” (p. 219) and “irrefutable” (p. 220) overstate the position. - “Over 800” studies showing toxicity (p. 223) is inconsistent with “at least 46” (p. 221). - Ashby et al. (1999) appears in Table 10.1 as a positive finding; per the working digest, it is usually cited as a failed replication.

Sources#

EFSA opinions, reports and policies - EFSA CEF Panel (2015). Scientific Opinion on the risks to public health related to the presence of bisphenol A (BPA) in foodstuffs. EFSA Journal 13(1):3978, January 2015. https://doi.org/10.2903/j.efsa.2015.3978 - EFSA (21 January 2015). News release on the 2015 BPA assessment. https://www.efsa.europa.eu/en/press/news/150121 - EFSA (2015). Report on the two-phase public consultation on the draft opinion on BPA. https://doi.org/10.2903/sp.efsa.2015.en-740 - EFSA CEF Panel (2016). Statement on the developmental immunotoxicity of BPA. EFSA Journal 14:4580. https://doi.org/10.2903/j.efsa.2016.4580 (identified; not read) - EFSA (December 2017). Bisphenol A (BPA) hazard assessment protocol. https://doi.org/10.2903/sp.efsa.2017.en-1354 ; consultation report https://doi.org/10.2903/sp.efsa.2017.en-1355 - EFSA (November 2019). Testing the study appraisal methodology from the 2017 BPA protocol. https://doi.org/10.2903/sp.efsa.2019.en-1732 - EFSA Scientific Committee (20 October 2021). Opinion on the impact of non-monotonic dose responses on EFSA’s human health risk assessments. EFSA Journal 19(10):6877. https://doi.org/10.2903/j.efsa.2021.6877 (full text via PMC8528485) - EFSA CEP Panel (19 April 2023). Re-evaluation of the risks to public health related to the presence of BPA in foodstuffs. EFSA Journal 21(4):6857. https://doi.org/10.2903/j.efsa.2023.6857 (full text via PMC10113887) - EFSA (19 April 2023). News: “Bisphenol A in food is a health risk”. https://www.efsa.europa.eu/en/news/bisphenol-food-health-risk - EFSA and BfR (April 2023). Report on diverging views between EFSA and BfR on EFSA’s updated BPA assessment (meeting 11 January 2023). https://www.efsa.europa.eu/sites/default/files/2023-04/bfr-efsa-art-30.pdf - EFSA (adopted 21 June 2017). EFSA’s policy on independence. https://www.efsa.europa.eu/sites/default/files/corporate_publications/files/policy_independence.pdf - ECHA and EFSA with JRC (June 2018). Guidance for the identification of endocrine disruptors. EFSA Journal 16(6):5311. https://doi.org/10.2903/j.efsa.2018.5311

Other European agency documents - EMA and EFSA (2023). Report on divergent opinion between EFSA and EMA on bisphenol A, EMA/150385/2023 (meeting 29 November 2022). https://www.ema.europa.eu/en/documents/report/report-divergent-opinion-between-efsa-ema-bisphenol_en.pdf - European Environment Agency (14 September 2023). Human exposure to Bisphenol A in Europe (briefing). https://www.eea.europa.eu/publications/peoples-exposure-to-bisphenol-a - PARC, Partnership for the Assessment of Risks from Chemicals (launched 11 May 2022). https://www.eu-parc.eu/

EU legislation (texts retrieved via the Publications Office Cellar) - Commission Directive 2004/73/EC (29th ATP), 29 April 2004: BPA Repr. Cat. 3; R62. http://data.europa.eu/eli/dir/2004/73/oj - Commission Regulation (EU) 2016/1179 (9th ATP to CLP), 19 July 2016: BPA Repr. 1B, H360F, applying from 1 March 2018. http://data.europa.eu/eli/reg/2016/1179/oj - Commission Regulation (EU) 2016/2235, 12 December 2016: BPA in thermal paper (≥0.02%) prohibited after 2 January 2020. http://data.europa.eu/eli/reg/2016/2235/oj - Commission Directive (EU) 2017/898, 24 May 2017: BPA migration limit in toys of 0.04 mg/l, applying from 26 November 2018. http://data.europa.eu/eli/dir/2017/898/oj - Regulation (EU) 2019/1381, 20 June 2019: transparency and sustainability of EU risk assessment in the food chain; applies from 27 March 2021. http://data.europa.eu/eli/reg/2019/1381/oj - Commission Regulation (EU) 2024/3190, 19 December 2024: BPA and other hazardous bisphenols in food contact materials. http://data.europa.eu/eli/reg/2024/3190/oj - Commission Regulation (EU) 2026/250, 2 February 2026: correcting Regulation 2024/3190. http://data.europa.eu/eli/reg/2026/250/oj - Regulation (EU) 2025/2455, 26 November 2025: common data platform on chemicals; monitoring and outlook framework; EEA early-warning system. http://data.europa.eu/eli/reg/2025/2455/oj - Regulation (EU) 2025/2509, 26 November 2025: safety of toys (applies from 1 August 2030). http://data.europa.eu/eli/reg/2025/2509/oj - Commission Delegated Regulation (EU) 2023/707 (endocrine-disruptor hazard classes under CLP). Cited via Regulation 2024/3190, recital 9; not read directly.

Court judgments - General Court, T-185/17 PlasticsEurope v ECHA, 11 July 2019 (reprotoxic SVHC listing; action dismissed), as cited in T-636/17. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62017TJ0185 - General Court, T-636/17 PlasticsEurope v ECHA, 20 September 2019 (endocrine disruptor, human health; dismissed). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62017TJ0636 - Court of Justice, C-876/19 P PlasticsEurope v ECHA, 21 December 2021 (appeal against T-636/17; dismissed). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62019CJ0876 - General Court, T-207/18 PlasticsEurope v ECHA, 16 December 2020 (endocrine disruptor, environment; dismissed). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62018TJ0207 - Court of Justice, C-119/21 P PlasticsEurope v ECHA, 9 March 2023 (appeal against T-207/18; dismissed). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62021CJ0119 - Conseil constitutionnel, Décision n° 2015-480 QPC, 17 September 2015 (Association Plastics Europe). https://www.conseil-constitutionnel.fr/decision/2015/2015480QPC.htm

US federal documents - FDA. Indirect Food Additives: Polymers (baby bottles and sippy cups; abandonment). 17 July 2012. https://www.federalregister.gov/documents/2012/07/17/2012-17366/indirect-food-additives-polymers - FDA. Indirect Food Additives: Adhesives and Components of Coatings (infant formula packaging; abandonment). 12 July 2013. https://www.federalregister.gov/documents/2013/07/12/2013-16684/indirect-food-additives-adhesives-and-components-of-coatings - FDA. Environmental Defense Fund et al.; Filing of Food Additive Petition (filed 2 May 2022). 87 FR 41079, 11 July 2022. https://www.federalregister.gov/documents/2022/07/11/2022-14682/environmental-defense-fund-maricel-maffini-breast-cancer-prevention-partners-clean-water-actionclean - FDA. Bisphenol A (BPA) information page (content current as of 25 April 2023). https://www.fda.gov/food/food-packaging-other-substances-come-contact-food-information-consumers/bisphenol-bpa ; and “BPA: Use in Food Contact Application” (content current as of 20 April 2023). https://www.fda.gov/food/food-packaging-other-substances-come-contact-food-information-consumers/bisphenol-bpa-use-food-contact-application - EPA. TSCA section 8(d) health and safety data reporting rule (includes BPA). 13 December 2024. https://www.federalregister.gov/documents/2024/12/13/2024-29406/certain-existing-chemicals-request-to-submit-unpublished-health-and-safety-data-under-the-toxic - EPA. Initiation of Prioritization under TSCA (BPA listed among potential candidates not currently initiated). 18 December 2024. https://www.federalregister.gov/documents/2024/12/18/2024-29829/initiation-of-prioritization-under-the-toxic-substances-control-act-tsca-notice-of-availability

CLARITY-BPA and related toxicology - National Toxicology Program (September 2018). NTP Research Report 9: CLARITY-BPA Core Study. https://doi.org/10.22427/NTP-RR-9 - CLARITY-BPA Research Program (October 2021). NTP Research Report 18: compendium of published findings. https://doi.org/10.22427/NTP-RR-18 - Camacho L et al. (2019). A two-year toxicology study of BPA in Sprague-Dawley rats: CLARITY-BPA core study results. Food Chem Toxicol 132:110728. https://doi.org/10.1016/j.fct.2019.110728 - Delclos KB et al. (2014). Toxicity evaluation of BPA administered by gavage to Sprague Dawley rats from GD6 through PND90. Toxicol Sci 139:174–197. https://doi.org/10.1093/toxsci/kfu022 - Prins GS et al. (2019). CLARITY-BPA academic laboratory studies identify consistent low-dose BPA effects on multiple organ systems. Basic Clin Pharmacol Toxicol 125(S3):14–31. https://doi.org/10.1111/bcpt.13125 - Heindel JJ et al. (2020). Data integration, analysis, and interpretation of eight academic CLARITY-BPA studies. Reprod Toxicol 98:29–60. https://doi.org/10.1016/j.reprotox.2020.05.014 - Badding MA et al. (2019). CLARITY-BPA Core Study: analysis for non-monotonic dose-responses and biological relevance. Food Chem Toxicol 131:110554. https://doi.org/10.1016/j.fct.2019.06.001 - Vandenberg LN, Hunt PA, Gore AC (2019). Endocrine disruptors and the future of toxicology testing: lessons from CLARITY-BPA. Nat Rev Endocrinol 15:366–374. https://doi.org/10.1038/s41574-019-0173-y

Commentary on the 2023 TDI and agency divergence - Zagorski JW, Kaminski NE (2023). When the weight of evidence does not weigh enough: EFSA’s draft scientific opinion on BPA. Toxicol Sci 193:115–118. https://doi.org/10.1093/toxsci/kfad034 - Kortenkamp A et al. (2024). Drivers of divergent assessments of BPA hazards to semen quality by various European agencies, regulators and scientists. Int J Hyg Environ Health 255:114293. https://doi.org/10.1016/j.ijheh.2023.114293 - vom Saal FS et al. (2024). The conflict between regulatory agencies over the 20,000-fold lowering of the TDI for BPA by EFSA. Environ Health Perspect 132:045001. https://doi.org/10.1289/EHP13812 - Leist M et al. (SKLM) (2024). Controversy on health-based guidance values for BPA: the need of criteria for studies that serve as a basis for risk assessment. Arch Toxicol 98:1967–1973. https://doi.org/10.1007/s00204-024-03778-3

Low-dose effects and non-monotonicity: consensus and critique - Rhomberg LR, Goodman JE (2012). Low-dose effects and nonmonotonic dose-responses of endocrine disrupting chemicals: has the case been made? Regul Toxicol Pharmacol 64:130–133. https://doi.org/10.1016/j.yrtph.2012.06.015 - Dietrich DR et al. (2013). Scientifically unfounded precaution drives European Commission’s recommendations on EDC regulation… Food Chem Toxicol (December 2013; also published in several other journals). https://doi.org/10.1016/j.fct.2013.07.005 - Gore AC et al. (2015). Executive Summary to EDC-2: The Endocrine Society’s second scientific statement on EDCs. Endocr Rev 36:593–602. https://doi.org/10.1210/er.2015-1093 - National Research Council (2014). Review of EPA’s State-of-the-Science Evaluation of Nonmonotonic Dose-Response Relationships as they Apply to Endocrine Disrupters. https://doi.org/10.17226/18608 (identified; not read) - NASEM (2017). Application of Systematic Review Methods in an Overall Strategy for Evaluating Low-Dose Toxicity from Endocrine Active Chemicals. https://doi.org/10.17226/24758 (summary page only)

Exposure, toxicokinetics and measurement - Ye X et al. (2013). Potential external contamination with BPA and other ubiquitous organic environmental chemicals during biomonitoring analysis. Environ Health Perspect 121:283–286. https://doi.org/10.1289/ehp.1206093 - Teeguarden J et al. (2013). Are typical human serum BPA concentrations measurable and sufficient to be estrogenic in the general population? Food Chem Toxicol 62:949–963. https://doi.org/10.1016/j.fct.2013.08.001 - Gayrard V et al. (2013). High bioavailability of BPA from sublingual exposure. Environ Health Perspect 121:951–956. https://doi.org/10.1289/ehp.1206339 - Vandenberg LN et al. (2014). A round robin approach to the analysis of BPA in human blood samples. Environ Health 13:25. https://doi.org/10.1186/1476-069X-13-25 - Hormann AM et al. (2014). Holding thermal receipt paper and eating food after using hand sanitizer results in high serum bioactive and urine total levels of BPA. PLoS One 9:e110509. https://doi.org/10.1371/journal.pone.0110509 - Thayer KA et al. (2015). Pharmacokinetics of BPA in humans following a single oral administration. Environ Int 83:107–115. https://doi.org/10.1016/j.envint.2015.06.008 - Teeguarden JG et al. (2015). 24-hour human urine and serum profiles of BPA: evidence against sublingual absorption following ingestion in soup. Toxicol Appl Pharmacol (October 2015). https://doi.org/10.1016/j.taap.2015.01.009 - Gerona RR et al. (2016). Direct measurement of BPA, BPA glucuronide and BPA sulfate in a diverse and low-income population of pregnant women. Environ Health 15:50. https://doi.org/10.1186/s12940-016-0131-2 - Gerona R, vom Saal FS, Hunt PA (2020). BPA: have flawed analytical techniques compromised risk assessments? Lancet Diabetes Endocrinol 8:11–13. https://doi.org/10.1016/S2213-8587(19)30381-X ; reply letter by Calafat AM, Koch HM et al., Lancet Diabetes Endocrinol 8:269–270, April 2020. https://doi.org/10.1016/S2213-8587(20)30070-X (not read) - Ashley-Martin J et al. (2021). Direct LC-MS/MS and indirect GC-MS/MS methods for measuring urinary BPA concentrations are comparable. Environ Int 157:106874. https://doi.org/10.1016/j.envint.2021.106874 - Jusko TA et al. (2014). Reproducibility of urinary BPA concentrations measured during pregnancy in the Generation R Study. J Expo Sci Environ Epidemiol 24:532–536. https://doi.org/10.1038/jes.2014.23 - Tagne-Fotso R et al. (2024). Exposure to BPA in European women from 2007 to 2014 using human biomonitoring data (HBM4EU). Environ Int 190:108912. https://doi.org/10.1016/j.envint.2024.108912

Substitutes and trends - Rochester JR, Bolden AL (2015). Bisphenol S and F: a systematic review and comparison of the hormonal activity of BPA substitutes. Environ Health Perspect 123:643–650. https://doi.org/10.1289/ehp.1408989 - Lehmler HJ et al. (2018). Exposure to BPA, BPF and BPS in U.S. adults and children: NHANES 2013–2014. ACS Omega 3:6523–6532. https://doi.org/10.1021/acsomega.8b00824 - Acevedo JM et al. (2025). Temporal and geographic variability of bisphenol levels in humans: systematic review and meta-analysis. Environ Res 264:120341. https://doi.org/10.1016/j.envres.2024.120341

Human epidemiology syntheses - Lin MH et al. (2023). Exposure to BPA associated with multiple health-related outcomes in humans: an umbrella review. Environ Res 237:116900. https://doi.org/10.1016/j.envres.2023.116900 - Symeonides C et al. (2024). An umbrella review of meta-analyses evaluating associations between human health and exposure to major classes of plastic-associated chemicals. Ann Glob Health 90:52. https://doi.org/10.5334/aogh.4459

Funding and research outcomes - vom Saal FS, Hughes C (2005). An extensive new literature concerning low-dose effects of BPA shows the need for a new risk assessment. Environ Health Perspect 113 (August 2005). https://doi.org/10.1289/ehp.7713 (source of Table 10.2) - Lundh A et al. (2017). Industry sponsorship and research outcome. Cochrane Database Syst Rev MR000033.pub3. https://doi.org/10.1002/14651858.MR000033.pub3 (identified; abstract not re-read)

Access notes. The following were not accessible, and claims that depend on them are flagged in the text: - ECHA web pages (403): candidate-list details, Annex XIV status and bisphenol group assessment. Annex XIV status was not directly verified. - OECD test-guideline pages (403). Guideline changes are taken from the 2018 ECHA/EFSA guidance and Crossref metadata. - Wiley EFSA Journal pages (403). EFSA opinions were read via PMC. - The BfR Opinion 018/2023 PDF (not fetched). - The FDA February 2018 CLARITY statement (404).