Update on the Health Effects of Bisphenol A: Overwhelming Evidence of Harm

Abstract

In 1997, the first in vivo bisphenol A (BPA) study by endocrinologists reported that feeding BPA to pregnant mice induced adverse reproductive effects in male offspring at the low dose of 2 µg/kg/day. Since then, thousands of studies have reported adverse effects in animals administered low doses of BPA. Despite more than 100 epidemiological studies suggesting associations between BPA and disease/dysfunction also reported in animal studies, regulatory agencies continue to assert that BPA exposures are safe. To address this disagreement, the CLARITY-BPA study was designed to evaluate traditional endpoints of toxicity and modern hypothesis-driven, disease-relevant outcomes in the same set of animals. A wide range of adverse effects was reported in both the toxicity and the mechanistic endpoints at the lowest dose tested (2.5 µg/kg/day), leading independent experts to call for the lowest observed adverse effect level (LOAEL) to be dropped 20 000-fold from the current outdated LOAEL of 50 000 µg/kg/day. Despite criticism by members of the Endocrine Society that the Food and Drug Administration (FDA)’s assumptions violate basic principles of endocrinology, the FDA rejected all low-dose data as not biologically plausible. Their decisions rely on 4 incorrect assumptions: dose responses must be monotonic, there exists a threshold below which there are no effects, both sexes must respond similarly, and only toxicological guideline studies are valid. This review details more than 20 years of BPA studies and addresses the divide that exists between regulatory approaches and endocrine science. Ultimately, CLARITY-BPA has shed light on why traditional methods of evaluating toxicity are insufficient to evaluate endocrine disrupting chemicals.

Over the last 20 years, bisphenol A (BPA; CAS# 80-05-7) has become one of the most studied endocrine disrupting chemicals (EDCs), because it is one of the highest volume chemicals in worldwide production, it is used in a wide variety of products, and exposure is documented in virtually everyone in the United States and elsewhere (1). The first objective of this mini-review is to present a brief history of the illusion of controversy about the myriad health hazards posed by low doses of BPA, and the prediction that it would be impossible that BPA could cause effects within the range of predicted human exposure based on exposure models rather than on data from biomonitoring studies. This manufactured controversy was initiated by the plastic industry, and subsequently has been perpetuated by regulatory agencies that have ignored findings from thousands of mechanistic in vitro and low-dose animal experiments, as well as significant findings from more than 100 epidemiological studies.

In this review “low doses” of BPA refer to any doses below the previous Environmental Protection Agency (EPA)/Food and Drug Administration (FDA) determined lowest observed adverse effect level (LOAEL) (2); for BPA, the LOAEL of orally administered BPA is currently 50 000 µg/kg/day. Any significant adverse effect at a dose below this LOAEL would lower the calculation of the “safe” amount for daily human exposure, which is referred to as the acceptable daily intake (ADI) dose by the FDA and reference dose by the EPA; this assumed “safe” dose is currently set at 50 µg/kg/day in the United States. The estimated “safe” dose is calculated based on the assumption that a dose 1000-fold lower than the LOAEL should be without effect if oral exposure occurred each day over the lifetime (3).

Our second objective is to review the current state of the evidence regarding the health hazards posed by low doses of BPA that are relevant to human exposure, including results from the recent government–academic collaborative program (the Consortium Linking Academic and Regulatory Insights on BPA Toxicity; CLARITY-BPA). In 2012, the CLARITY-BPA study was created specifically to address the divide concerning BPA safety, and directly compare the results from a guideline study (eg, a regulatory study conducted by FDA scientists) and hypothesis-driven endpoints conducted by academic investigators using tissues from randomly selected animals from the same cohort. The collaboration involved the National Institute of Environmental Health Sciences (NIEHS), which funded 14 academic programs to examine gene and protein expression, epigenetic markers, morphometric analyses, and neuroendocrine and behavioral responses using tissues from rats produced by the FDA’s National Center for Toxicological Research (NCTR). The study was conducted using good laboratory practice (GLP)-compliant protocols to ensure quality control and detailed record keeping, making it acceptable for use in regulatory decisions according to FDA standards (4).

At the end of the CLARITY-BPA program, the FDA’s contract with the NIEHS-funded academic investigators called for a consensus publication integrating all findings from the FDA guideline study and the academic investigators for use in a new BPA risk assessment. Instead, immediately after finishing its toxicological guideline portion of the CLARITY-BPA study, and prior to peer review, the FDA declared BPA “safe” (5). Yet, a majority of the academic scientists that participated in the CLARITY-BPA study disagreed with the FDA’s published conclusions (6). Based on an integration of the academic investigator studies and data from the FDA’s guideline study, a majority of the independent academic investigators concluded that the FDA’s acceptable daily intake (safe) dose of BPA should be 20 000-fold lower than the current estimated “safe” dose (7). Unfortunately, there is no expectation that the FDA will complete the collaboration with the academic investigators to produce a consensus publication that was an agreed upon outcome at the inception of CLARITY-BPA.

The dispute over the safety of BPA has laid bare a deep divide between regulatory toxicologists working for federal agencies or chemical corporations in contrast to scientists trained in the principles of endocrinology. These principles include the understanding of “low-dose” effects and nonmonotonic dose responses, as well as the presence of sex-specific and tissue-specific effects of hormones. It is these endocrine principles that are required to understand hormone action and disruption of endocrine function by environmental EDCs. Multiple Endocrine Society scientific statements and position papers have reviewed the EDC literature and controversies that we will discuss in this review (8-10). Other reviews by endocrinologists have also attempted to provide the rationale for the field of endocrine disruption to be included as a subdiscipline in the field of endocrinology (11-13).

BPA, an Endocrine Disrupting Chemical

BPA was discovered to have estrogenic activity in 1936, when it was demonstrated to induce cornification of vaginal epithelium when injected into female rats; at that time, chemists were trying to determine the structural characteristics of estrogenic compounds (14), which is an ongoing endeavor (15). With its 2 benzene rings and 2 (4,4′)-OH substituents, BPA fits within the binding pockets of both estrogen receptor (ER)α and ERβ (16). Although the most studied mechanism of action of BPA is its estrogenic activity (11, 17), its endocrine mode of action is much more complex (18). Not only is BPA a nuclear ER agonist, it is also an agonist of the membrane ERs (mERs), G protein–coupled receptor 30 (GPR30/GPER), and can activate nongenomic rapid signaling cascades through these receptors at very low concentrations (Fig. 1) (23-25). BPA is an antagonist of thyroid hormone receptor, interfering with the normal binding of 3,5,3′-triiodothyronine (26, 27), and modest antiandrogenic activity has been reported based on data from a yeast reporter assay (28). BPA also binds to several orphan receptors, including estrogen-related receptor γ (29) as well as the aryl hydrocarbon receptor (30). The large number of receptors and signaling pathways affected by BPA led the US National Toxicology Program (NTP) to assign it the third highest Toxicological Priority Index score of more than 300 chemicals examined (31). A more recent application of the Key Characteristics approach found evidence that BPA triggers 9 of the 10 characteristics of an EDC (32). Furthermore, although the primary phase 2 metabolites of BPA (BPA-glucuronide and BPA-sulfate) are not ER agonists, a metabolite produced in small amounts (4-methyl-2,4-bis(4-hydroxyphenyl)pent-1-ene; MBP) is about 1000-times more biologically active than BPA. Metabolism of BPA to MBP increases the spacing between the 2 phenolic rings, resulting in contacts between MBP and ERα and ERβ that more closely mimic those of estradiol-17β relative to BPA (33).

Figure 1.

BPA interaction with estrogen receptor subtypes. The bioactive concentration of BPA, acting via each of the subtypes, GPER (19), estrogen related receptor γ (ERRγ) (20), ERα (21), and ERβ (22), is indicated, along with the cell type in which the effect has been observed.

Because BPA has a Kd value of ~370 nM for ERα (11), approximately 3700-fold lower than the Kd of estradiol-17β for this receptor, BPA is often referred to as a “weak” estrogen. This is a flawed description of BPA (or any hormone receptor agonist) because potency is not equal to affinity of a ligand for a specific receptor subtype in a specific tissue. For example, the physiological range of estradiol-17β is up to 100-fold lower (1-10 pM, where there is high proportionality between estradiol concentration and receptor occupancy) than the Kd for ERα (~100 pM), and the 50% effective concentration (EC50) for MCF-7 cell proliferation is at approximately 1% to 3% receptor occupancy (11).

BPA in the 1 to 10 nM range (0.23-2.3 ng/mL) stimulates MCF-7 cell proliferation, and 1 to 10 nM BPA concentrations initiate transcriptional activity of both Esr1 (ERα) and androgen receptor Ar mRNA in fetal urogenital sinus mesenchyme in primary culture (21, 34). Critically, there are a number of biomonitoring studies showing that humans are exposed to bioactive (unconjugated) BPA in the 1 to 10 nM range in serum (1), and recent findings based on use of new direct assay methods indicate that actual exposures to BPA are significantly higher than previous reports using indirect assay methods to assess total BPA in urine (35). Furthermore, much lower serum BPA levels are bioactive when exposure occurs during fetal life via maternal exposure in mice and other species (36-38). Importantly, the toxicokinetics of BPA are similar in mice, rhesus monkeys, and humans after oral administration (39).

Hormones regulate their own receptors and thus modulate sensitivity to themselves (eg, receptor up- and downregulation); low doses typically increase receptor number whereas high doses are inhibitory (34, 38). Because estrogens modulate other hormonal systems and enzymes, estrogenic chemicals have pleiotropic effects. For example, BPA alters the response of tissues to estrogens (by changing expression of ERs) as well as to other hormones such as progesterone, androgens, and oxytocin (34, 38, 40). BPA also modulates the activity of aromatase, the enzyme that converts androgens to estrogens, creating a double hit of BPA and additional intracellular estradiol (eg, BPA increases aromatase activity and intracellular estradiol in the fetal prostate mesenchyme (41, 42)). Low doses of BPA increase androgen receptor gene expression in fetal mouse prostate mesenchyme, which leads to a permanent upregulation of prostate androgen receptor and thus increased sensitivity to androgens (34, 43). BPA and other estrogenic chemicals also appear to induce basal cell hyperplasia in the fetal mouse prostate (44), and in rats BPA treatment shifts dorsolateral prostate stem cell commitment toward basal cell progenitors, increasing the risk for carcinogenesis (45). In addition, the BPA metabolite MBP binds to sex hormone binding globulin, and thus can disrupt the percent of androgens and estrogens that circulate in blood in the free (bioactive) versus bound state. MBP also binds to androgen receptors and progesterone receptors, disrupting their activity (46).

Without apparent concern for the estrogenic activity of the BPA monomer, in the 1950s, chemists at Bayer in Germany and General Electric in the United States succeeded in creating the BPA-based polymer polycarbonate, which first appeared on the market in 1958 (47). The FDA approved BPA for use in food contact materials in 1963 in the absence of any published safety information (48). Today, BPA remains a building block for polycarbonate plastics and resins used in many consumer products, including reusable food storage containers and epoxies used to line food/beverage containers (49, 50). BPA is also routinely added to other types of plastics (eg, plastics made from polyacrylate, polyarylate, polyetherimide, polyester, polyester–styrene, polysulfone, polyethylene tetraphthalate, and polyvinyl chloride) to act as an antioxidant, hardening agent, or a stabilizer (51). Unpolymerized (free) BPA is used as a developer on the surface of thermal paper, in cosmetics and personal care products, medical and sports equipment, toys, clothing, and other goods (52, 53). Human exposures to BPA have been documented around the world (54-56) and indicate that the vast majority of people are exposed multiple times per day to this chemical from multiple sources (57, 58).

Given the very high volume of BPA produced per year (estimated at about 10 billion kilograms) it is not surprising that BPA is also a major environmental contaminant. The United States Geological Survey reported that the maximum amount of BPA detected in leachates from landfills in the United States was 6.4 mg/L (59). In another study of landfill leachates, BPA accounted for 84% of the estrogenic activity that was detected in Japan (60), which was consistent with a US study in which BPA was found to account for virtually all estrogenic activity in landfill leachates (61).

Thousands of Studies of BPA Show Harm at Low Doses

In 1997, Nagel et al. (62) found that BPA showed limited binding to sex hormone binding globulin, the high-affinity plasma binding protein involved in the transport of estradiol and testosterone in blood, thus increasing the percent free BPA in blood (63). This finding, together with information from ERα binding studies in MCF7 breast cancer cells, led to the first “low dose” BPA finding that oral administration of 2 µg of BPA/kg/day to pregnant female CD-1 mice induced enlarged prostates in male offspring (62, 63), as well as adverse effects on other accessory reproductive organs in male offspring in adulthood (64). Thus, the first academic studies of BPA by endocrinologists revealed that adverse effects of BPA occurred at a dose of 2 µg/kg/day, 25 000-fold lower than the EPA’s and FDA’s presumed LOAEL of 50 000 µg/kg/day (3).

After these initial studies on BPA, thousands of studies on the chemical were published between 2000 and 2020 describing its effects on a wide range of health hazards. There has been an exponential increase in publications each year reporting adverse effects of BPA, from 115 publications before 2005 (65) to over 500 publications in 2011 and then over 1000 publications in 2017 (Fig. 2). Scientists from a wide range of disciplines, many that would not typically examine environmental chemicals, evaluated whether BPA could impact a range of organs and systems. Overall, this extensive literature shows that BPA has a broad impact on many of the most common chronic noncommunicable diseases, with disruption of normal immune function likely a mediating factor (67, 68).

Figure 2.

Data on the number of bisphenol A (BPA) citations from animal experiments (thousands) and human epidemiological studies (>100) as of September 2018. The numbers do not include in vitro mechanistic studies. The figure is from an FDA webinar about the findings from the guideline portion of the Consortium Linking Academic and Regulatory Insights on BPA Toxicity (CLARITY‐BPA), from (66): FDA 2018 Bisphenol A (BPA): Toxicology and pharmacokinetic data to inform on-going safety assessments. FDA Grand Rounds, K.B. Delclos, September 13, 2018. https://www.fda.gov/science-research/about-science-research-fda/bisphenol-toxicology-and-pharmacokenetic-data-inform-ongoing-safety-assessments (public domain).

Many comprehensive reviews have examined the breadth and depth of the BPA literature concerning health hazards and exposure estimates at different times: in 2007, a series of 5 literature reviews (69-73), as well as the Chapel Hill Consensus Statement from a NIEHS-hosted workshop, concluded that there was strong evidence that “prenatal and/or neonatal exposure to low doses of BPA results in organizational changes in the prostate, breast, testis, mammary glands, body size, brain structure and chemistry, and behavior of laboratory animals” (74). Other reviews covering both the human health and experimental animal findings that have drawn similar conclusions were published after 2007 (13, 38, 67, 75-79).

We provide a few examples that highlight the ability of low levels of BPA exposure to induce adverse health effects in both animal and human studies (Table 1). The effects on the prostate, discussed above, have been extended to include studies demonstrating that perinatal BPA exposures increase the sensitivity of the prostate to adult hormone exposures, increasing the incidence of urethral obstruction associated with hydronephrosis (Fig. 3A and 3B) (80, 81), prostatitis, and prostatic epithelial hyperplasia (Fig. 3C), as well as prostatic intraepithelial neoplasia (82). As discussed further below, effects of BPA on prostate stem cells have also been reported by Prins and colleagues (45).

Table 1.

List of adverse effects reported in laboratory animal experiments and in human epidemiological studies

Outcome	Observed in laboratory animals	Associations in epidemiology studies
Metabolic disease
Obesity	√	√
Impaired glucose tolerance	√	√
Type 2 diabetes	√	√
Hypertension	√	√
Cardiovascular disease/dysfunction	√	√
Altered liver function	√	√
Adrenal hyperplasia	√
Neural and behavioral effects
Sex-specific changes in brain structure	√
Altered estrogen action	√
Aggression/aggressive behaviors	√	√
Impaired learning/memory	√	√
Male reproduction
Reduced libido		√
Altered sperm count, sperm quality	√	√
Abnormalities of spermatogonia	√
Urethra–bladder obstruction and hydronephrosis	√
Reduced serum testosterone	√
Prostate hyperplasia, metaplasia, neoplasia	√
Female reproduction
Uterine fibroids		√
Ovarian cysts, polycystic ovarian syndrome	√	√
Chromosomal abnormalities, oocytes	√
Early onset of puberty	√	√
Mammary gland development, hyperplasia	√
Mammary cancer	√
Estrous cycle/menstrual cycle disruption	√
Reduced serum estradiol	√
Risk of miscarriage		√

Figure 3.

Adverse effects due to exposure to BPA during development or in adulthood. (A) Adult control (CTL) vehicle-only exposed adult male mouse. (B) Adult male mouse treated perinatally with 20 µg/kg/day BPA and then in adulthood administered physiologically relevant doses of testosterone and estradiol (second hit) via Silastic capsules that together caused obstructive voiding disorder and hydronephrosis. (C) The dorsolateral prostate from the same BPA-treated male in (B) showing epithelial hyperplasia and prostatitis (arrows); from: Taylor JA, Jones MB, Besch-Williford CL, Berendzen AF, Ricke WA, vom Saal FS 2020 Interactive Effects of Perinatal BPA or diethylstilbestrol and Adult Testosterone and Estradiol Exposure on Adult Urethral Obstruction and Bladder, Kidney, and Prostate Pathology in Male Mice. Int J Mol Sci 21:3902 (open access). (D) Whole-mount mammary glands from adult female mice showing ducts (D), terminal ducts (TD) and a significant increase in alveolar buds (AB) due to perinatal exposure to 0.25 µg BPA/kg/day from Alzet osmotic pumps; from: Vandenberg LN, Maffini MV, Schaeberle CM, et al. 2008 Perinatal exposure to the xenoestrogen bisphenol-A induces mammary intraductal hyperplasias in adult CD-1 mice. Reprod Toxicol 26:210–219, (open access). (E) Confocal images of intact mouse oocytes immunostained to visualize the meiotic spindle (green) and the chromosomes (red). Normal metaphase I (CTL) configuration and representative meiotic abnormality from female mice exposed to BPA leaching from polycarbonate cages and water bottles; from: Hunt PA, Koehler KE, Susiarjo M, et al. 2003 Bisphenol A exposure causes meiotic aneuploidy in the female mouse. Curr Biol 13:546–53, (open access). (F) Decreased daily sperm production (significant effect [*] at and above 20 µg BPA/kg/day) in adult male rats fed BPA for 6 days; from: Sakaue M, Ohsako S, Ishimura R, et al. 2001 Bisphenol A affects spermatogenesis in the adult rat even at a low dose. J Occupational Health 43:185–190, with permission from the publisher. (G) Inhibition by BPA (at and above 40 µg/kg/day) of estrogen-stimulated hippocampal synaptic spine formation in adult female rats; from: MacLusky NJ, Hajszan T, Leranth C 2005 The environmental estrogen bisphenol A inhibits estradiol-induced hippocampal synaptogenesis. Environ Health Perspect 113:675–679, (public domain). (H) Rapid phosphorylation of extracellular signal-regulated kinases (pERK1/2) following administration of doses of BPA ranging from 10^–14 to 10^–7 M or in combination with 1 nM estradiol-17β in rat pituitary cells showing oscillating and agonist/antagonist effects across a wide range of doses beginning with a significant stimulating effect (*) of BPA alone at 10^–14 M (0.01 pM = 2.3 fg/mL BPA), and also antiestrogenic effects when BPA was administered together with 1 nM (0.27 pg/mL 17β-estradiol); from: Jeng YJ, Watson CS 2011 Combinations of physiologic estrogens with xenoestrogens alter ERK phosphorylation profiles in rat pituitary cells. Environ Health Perspect 119:104–112, (public domain).

Another example comes from the mammary gland, where numerous studies have demonstrated that BPA exposure during perinatal development can alter rodent mammary gland morphology (Fig. 3D), and enhance the gland’s responsiveness to hormones and carcinogens (83-85), at μg/kg/day or even ng/kg/day doses (reviewed in (75, 86)). Mechanistic studies have revealed that BPA acts on the mammary gland via ER activation in the fetal mesenchyme, altering gene expression and development of adipose and other cells of the stromal compartment and promoting growth of the fetal mammary epithelium (87-90). Furthermore, at least 1 study suggests that BPA can induce carcinoma (including adenocarcinoma) of the mammary gland in sensitive rat strains, providing evidence that it is a complete carcinogen (91).

A third example is the organs of the female reproductive tract, where BPA induces adverse effects on the uterine epithelium (92) and ovary. In fact, numerous studies have shown that BPA alters follicle maturation and meiosis in oocytes, disrupts steroidogenesis, and reduces oocyte quality (77, 93, 94). Remarkably, much of the work on the ovary was stimulated by a study showing that BPA leached from polycarbonate animal cages and bottles at levels sufficient to alter reproductive outcomes in female mice (95-97), challenging the old assumption that BPA leaching from plastic products would always be “too low” to affect health outcomes (Fig. 3E). Furthermore, studies demonstrate that exposures to very low doses of BPA during perinatal development (25 ng/kg/day) decreased lifetime reproductive success, including both fertility and fecundity in mice, clear adverse outcomes (98).

The effects of BPA on oocytes during early development and then again during maturation is mirrored by disruption of male germ cells as well. BPA disrupts spermatogonial stem cell differentiation during fetal life in mice (99), resulting in reduced sperm production in adulthood (64). In addition, similar to the effects of estradiol and estrogenic drugs, BPA exposure in adulthood results in a marked reduction in spermatogenesis (Fig. 3F) (100) associated with inhibition of the brain–pituitary–testicular axis and reduced testosterone (101).

There are also myriad effects of BPA on brain structure, function, and behavior as well as on pituitary function. Importantly, the effects of BPA are typically different (and in a number of cases opposite) for males and females (102). This is not unexpected because males and females have markedly different hormonal profiles, and hormones have well-characterized sexually dimorphic roles in brain development (103). In particular, BPA, along with most other environmental chemicals with estrogenic activity, is referred to as a selective estrogen receptor modulator, because similar to other selective estrogen receptor modulators, such as the well-studied drug tamoxifen, effects are different based on both the dose and the tissue being examined (13). For example, in the mammary gland and prostate, low doses of BPA act as an ER agonist, whereas in the rat or monkey hippocampus and prefrontal cortex, BPA has antiestrogenic activity (104, 105), resulting in inhibition of estrogen stimulation of synaptic spines (Fig. 3G), which is associated with dementia and loss of short-term memory (106). Rapid effects of BPA have also been shown to be mediated by G protein–related receptors (GPR30/GPER) activating ERK1/2 pathways, and a number of studies have shown that via rapid signaling pathways BPA is equipotent to estradiol (107). BPA has also shown mixed agonist/antagonist estrogenic activity in rat pituitary cells (Fig. 3H), with significant effects occurring at BPA concentrations as low as 10^–14 M or 2.3 fg/mL (19).

Finally, a large number of studies show effects of BPA on metabolic diseases (67), including effects on the liver (108), adipose tissue, and pancreas, including glucose intolerance (37, 67, 109). BPA also induces changes in hormones that regulate adipocyte function and body weight (37). As discussed more below, extensive research has revealed the intracellular pathways that mediate the effects of BPA on insulin homeostasis (110, 111). Many studies show sex differences in the effects of BPA on metabolic outcomes (67).

Creating the Controversy: Early Chemical Industry Studies of BPA

The earliest studies of BPA describing effects on the developing male reproductive system (62) were quickly disputed by 2 studies funded by the plastic industry (112, 113). However, these studies were subsequently discounted by a NTP review panel because they not only failed to show any effects of BPA, they also failed to show any effects of the positive control, diethylstilbestrol, at a dose previously shown to cause adverse effects (114), possibly due to laboratory contamination (115).

When another academic investigator with expertise in the male reproductive system repeated the findings of significant effects on the mouse prostate following prenatal exposure to low doses of BPA and diethylstilbestrol (43), the independent replication was thought to have ended the controversy over BPA (116). Subsequent studies by Prins and colleagues have shown effects of BPA on prostate epigenetic programming and on prostate progenitor cells, and have concluded that early life BPA exposure increases the risk of prostate cancer (45). Yet, chemical industry and US FDA representatives continue to refer to studies that have been discredited by stating that BPA has no effects at low doses; meanwhile, BPA sales have been estimated to exceed $US 20 billion in 2020 (117, 118).

Additional studies funded by groups with a vested interest in BPA production contributed to the controversy about BPA safety. One was conducted in the late 1990s at an organization called the Chemical Industry Institute of Toxicology (CIIT), which was funded by chemical corporations. The authors (119) administered different doses of BPA to pregnant and lactating rats via drinking water, using a rat strain, CD-SD, that is less sensitive to estrogenic chemicals than other rodent models (65). The study was first presented at a Society of Toxicology meeting with the title: “‘Effects of perinatal exposure to low doses of BPA in male offspring of Sprague–Dawley rats’” (120), but the title had shifted to: “Effect of different sampling designs on outcome of endocrine disruptor studies” by the time the CIIT study was published in 2000 (119).

In 2000, an NTP/NIEHS-sponsored meeting was held concerning the dispute over the issue of “low dose” effects of BPA and other endocrine disrupting chemicals, which chemical corporations insisted were not possible (121). Scientists who participated in this meeting submitted their raw data for review by the NTP Low-Dose Peer Review Statistics Subpanel. With regard to the CIIT study, the NTP panel stated that the authors’ conclusions of no effects of BPA were “‘flawed,’” ‘illogical,’ and ‘misleading’,” and that there was a statistically significant increase in adult prostate size with perinatal exposure to 10, 1000, and 10 000 µg/kg/day doses of BPA (121); see pp. A-86, A-89–A-91). To justify discounting their positive findings, Elswick et al. (119) proposed that their results showing effects of BPA might be “false positives” due to different approaches regarding the number of pups per litter that were examined (based on modeling). Specifically, Elswick et al. proposed that by sampling 1 male per litter, they had produced a false-positive finding. The NTP statistics panel stated: “To suggest that using fewer pups per litter (thereby increasing the variability) would lead to increased findings of statistical significance [false positive findings] is illogical.” (121), p. A-90).

The NTP Low Dose report contained a public comment section, in which the American Chemistry Council made the false claim that the Elswick et al. study had not shown any statistically significant effects (121), p. C-89), in direct contradiction of the conclusion by NTP statisticians. This is an example of the chemical industry creating a controversy over data from a study it funded and is a reminder that a detailed examination of data is required to determine the validity of the conclusions drawn by study authors (122).

Continuing the Controversy: Additional Studies from the Chemical Industry and Regulatory Agencies

Additional industry-funded studies reveal other issues around the evaluation of BPA safety, including how studies are conducted by chemical manufacturers and evaluated by agencies such as the EPA and FDA. Around the globe, regulators typically rely on the results of “guideline” studies when determining whether a chemical is hazardous. Guideline studies using GLP protocols follow well-established methods, many developed in the early 1900s, with few deviations, to administer chemicals and then quantify effects on pre-established endpoints (123, 124). The outcomes typically evaluated in a guideline, GLP-compliant study are acknowledged to be “adverse effects” from the perspective of regulatory toxicology (eg, altered organ weight, histopathology, hematology parameters), but they do not examine for signs relevant to many disease or endocrine-related dysfunctions (125, 126).

The first guideline GLP-complaint study was conducted at RTI, a commercial animal testing firm, and was funded by the plastics industry lobbying arm (127). A significant criticism of this study was the use of the CD-SD rat (65) and the lack of inclusion of a positive control, an issue affecting many regulatory toxicology studies that has been the subject of previous reviews and critiques (115, 128). After publication of this study, a report revealed that the RTI facility experienced an arson attack involving the burning of polycarbonate cages (polymerized BPA), contaminating the rooms and food used in the study. Although a review panel assembled by the chemical industry determined the contamination was of no concern, it complicates interpretation of the study.

The second study conducted by Tyl et al. (129), also funded by the plastics industry, examined the effects of a wider range of BPA doses in CD-1 mice. In this study, they included a positive control, orally administered 100 µg/kg/day 17β-estradiol; not only is this a very high dose, 17β-estradiol has limited oral bioactivity, which is why 17α-ethinylestradiol, with over 10-fold greater oral activity relative to 17β-estradiol, is typically used as a positive control when administration is oral (124). This guideline, GLP-compliant study was also reported to show no low-dose effects of BPA. Yet, there were several strange aspects of this study (129). First, the prostate weights in 14-week-old naïve control males were ~72 mg, almost twice as large as a normal 14-week-old CD-1 male prostate (~42 mg (34)), and larger than a typical 9-month-old CD-1 mouse prostate (34). The study author would later state that the reason the prostate was so large was that the males were examined not at 14 weeks, but in older males (123), although this discrepancy has never been resolved. A second strange approach used in this study was that the weight of the anterior prostate, which is considered to have a very high sensitivity to estrogenic chemicals and is enlarged by developmental exposure to BPA (80), was weighed together with the seminal vesicle, which is reduced in size by BPA and other estrogenic chemicals due to a permanent reduction in 5α-reductase activity following fetal exposure to estrogenic chemicals such as BPA or estradiol (64, 130). Finally, even though body weight, body fat, and other metabolic outcomes have been shown to display nonmonotonic dose–response relationships in response to estrogenic chemicals (37), the Tyl et al. (129) study authors did not examine the apparent low-dose increase in body weight and high-dose reduction in body weight in male offspring with methods appropriate for nonlinear data. Instead, they stated there was no dose-related effect of BPA on body weight.

These cases highlight the limitations in relying on compliance with test guidelines or GLP regulations as a determination of study quality. Yet, regulatory agencies, such as the FDA, have excluded from their regulatory consideration any studies that do not follow GLP protocols, instead using the 2 flawed RTI studies to conclude that BPA is safe (131). This over-reliance on GLP, which was originally designed to prevent fraud and data manipulation, was subjected to criticism in a publication coauthored by 36 scientists (124).

Similar questionable conclusions have been documented in publications from scientists working at regulatory agencies. In a 2008 decision of the European Food Safety Authority, the agency stated that: “there is sufficient capacity for biotransformation of BPA to hormonally inactive conjugates in neonatal humans…” (132). This is a statement for which there is no actual data from the scientific literature, and the European Food Safety Authority statement that “exposure of a human fetus to free BPA would be negligible” has been shown to not be correct (35, 133). Further, scientists at the US FDA’s NCTR published numerous studies focused on the toxicokinetics of BPA (eg, how it is distributed and excreted from the body following uptake via different routes of exposure). In 2010, an FDA publication claimed in the abstract: “No age-related changes were seen in internal exposure metrics for aglycone [unconjugated] BPA in monkeys” (134), yet the data in the publication show a clear ~4-fold reduction in BPA metabolism in newborn rhesus monkeys relative to adults; this age-related change in metabolism of BPA is similar to what has been shown in mice and rats (39, 135, 136).

In another study, conducted to collect preliminary data for the CLARITY-BPA study, the FDA authors reported that their negative control rats were contaminated with BPA; unconjugated BPA in serum was not different between untreated negative controls and animals treated with up to 80 µg BPA/kg/day (137). Yet, the authors concluded that BPA contamination was unimportant because of the absence of low-dose effects of BPA in these animals (138).

The disconnect between the results of thousands of studies, largely conducted by scientists in academic or government-affiliated laboratories (such as NIH in the USA and INRA in France), in comparison to the results of a few guideline studies conducted by industry and regulatory agency laboratories, became a flashpoint in the evaluation of BPA. The possibility of a “funder effect” (eg, where source of funding is a major determinant of outcome in science (139-141)) could not be dismissed; in an evaluation of 115 BPA studies published before 2005, 100% of industry-funded studies concluded that BPA was safe, whereas over 90% of academic studies showed that low doses of BPA caused harm (65). Concerns were also raised over the kinds of endpoints evaluated in guideline studies compared with the more disease-relevant outcomes evaluated in many hypothesis-driven academic studies; the outcomes evaluated in guideline studies were often many orders of magnitude less sensitive than the disease-relevant endpoints included in academic investigator mechanistic studies (eg, (44, 142)).

Plastic manufacturers used several additional tactics to deny that BPA posed health hazards, many of which are consistent with approaches used by tobacco, lead paint, and other polluting industries to protect harmful chemicals and industrial processes from regulations (143, 144). In one approach, the chemical industry paid the Harvard Center for Risk Analysis, a group funded by a variety of industry sponsors, to conduct a review of a small subset of the published articles about BPA. The Harvard Center for Risk Analysis panel concluded that BPA exposure was negligible and did not pose any hazard to health, but the report’s release was delayed for a few years. This led to criticism from some panel members that so much additional research had been published by the time the report was released that the conclusions were no longer relevant (65, 145).

Human Exposures: BPA Biomonitoring and Epidemiology

As the controversy over the health hazards of BPA exposure was ongoing, another dispute emerged: Are humans actually exposed to sufficient amounts of BPA to be concerning? Although a part of this debate revolved around whether low versus high doses cause harm in animals, another part centered on quantifying human exposures. This raised new challenges, because even though BPA and its major metabolites (BPA-glucuronide and BPA-sulfate) are well documented in urine samples from around the world (1), industry-sponsored researchers and then FDA scientists argued that unconjugated (bioactive) BPA must be measured in blood to accept that it could be causing harm (134, 146). When high enough levels of unconjugated BPA were measured in human serum samples to warrant concern, FDA and industry-funded researchers suggested that the samples must be contaminated due to the ubiquitous presence of BPA in products; the levels reported in serum were incongruent with industry and FDA exposure estimates to BPA of less than 1 μg/kg/day based on oral exposure models (134, 147). However, this “contamination” hypothesis was refuted as the basis for findings of biologically active levels of unconjugated BPA in human biomonitoring studies (122).

To directly examine the contamination controversy, a blinded round-robin analysis of BPA in human serum was funded and coordinated by NIEHS. This approach allowed 3 groups, including the vom Saal laboratory and the Gerona laboratory at UCSF, to demonstrate that they could accurately quantify unconjugated BPA in human serum without contamination (148). The FDA laboratory initially was part of the round-robin, but left the program; subsequent publications revealed that the FDA laboratory (as well as their animal facility) was experiencing problems with BPA contamination (137). Meanwhile, the contamination hypothesis for human biomonitoring samples continued to be promoted. Researchers at the US Centers for Disease Control and Prevention (CDC) published a paper stating that BPA contamination was “an elusive laboratory challenge” for biomonitoring studies, even though the CDC established that they had achieved a contamination-free assay for BPA and other chemicals (149).

Why was there such a disconnect between measurements of BPA in human samples and the estimates of BPA intake? Since the majority of epidemiological studies involve measuring total BPA (unconjugated plus BPA metabolites) in urine, the amount of biologically active BPA in blood was modeled based on assumptions about route of administration, with the FDA focusing on BPA exposures from food and food packaging. Yet, human exposure to BPA should not be modeled by animal studies evaluating rodents administered BPA by gavage one time per day, because an analysis of CDC data from the National Health and Nutrition Examination Survey (NHANES) (57), as well as an analysis of all urine samples collected from volunteers over a week (58) indicated multiple routes of exposure to BPA multiple times per day. The use of gavage administration has formed the basis for the FDA’s contention that little bioavailable BPA should be found in blood, but there are numerous problems with gavage that have been discussed in depth elsewhere (150). Briefly, administration of BPA by intragastric gavage results in very low bioavailability (39, 151). Gavage increases stress hormones, and intragastric gavage bypasses absorption in the mouth (152). A single gavage administration per day does not take into account the markedly different toxicokinetics of BPA after transdermal absorption (153), which is expected from exposures due to handling of BPA-containing thermal receipt paper (154, 155) and use of cosmetics (156). Metabolism of BPA is limited compared with gavage administration when BPA is absorbed through the skin (153), which may be a contributing factor in elevated BPA exposure in cashiers who handle these receipts all day (157, 158).

The controversy over BPA concentrations in human samples continues. Recent work (35), which involved use of newly available authentic standards for direct analysis of BPA and its metabolites, showed that the indirect assay used by the CDC in NHANES dramatically underestimated BPA levels (in urine) by as much as 170-fold. So far, the CDC has not indicated a willingness to reassess the use of its indirect assay methods for BPA or other phenolic compounds for which authentic standards now exist. However, the qualitative approach used by epidemiologists of ranking people from low to high exposures using data based on indirect assay methods was shown to still be accurate using the direct analysis methods (159). Importantly, the members of populations that have been examined in epidemiological studies that involved use of the indirect assay method who are in the highest exposure range may actually have been experiencing dramatically higher exposures than had been reported based on prior indirect assay methods.

Furthermore, the question over whether harm can be expected from current human exposure levels can be addressed by the more than 100 epidemiological studies that are available for BPA (10, 75, 76), including a recent study suggesting that higher BPA exposures are associated with increased mortality over the following decade in an NHANES cohort (160). Although many of the epidemiology studies on BPA were cross-sectional in design, making it difficult to draw conclusions about the causal relationship between BPA exposure and disease, dozens of prospective cohort studies now support the conclusion that maternal BPA exposures are associated with anxiety, depression, hyperactivity, and aggression in children (161); metabolic diseases in adults (162); and fertility outcomes in women utilizing artificial reproductive technologies, among others (75, 163).

The BPA Dispute: Is Any Amount Safe?

Based on the thousands of rodent and hundreds of primate and human studies showing that BPA causes (in animals) or is associated with (in humans) diseases, it has perhaps become clear why BPA has become the “flash-point” for disputes about the regulation of chemical safety. How can regulatory agencies conclude that BPA is safe at current levels of human exposure when so many studies suggest otherwise? The FDA’s insistence that BPA exposure is too low to be of concern is related to the fact that in chemical risk assessments, risk = hazard × exposure, and by declaring exposure to be so low as to be of no concern, the FDA can justify ignoring the huge BPA literature identifying hazards described above, and thus declare risk to be negligible.

Problem 1: Reliance on guideline studies, while ignoring academic low-dose studies

First, it helps to know how regulatory agencies like the US FDA, EPA, and their counterparts around the globe, make determinations of safety. As discussed above, in their evaluation of chemicals, regulatory agencies rely almost exclusively on the results of GLP-compliant guideline studies; nonguideline studies like those conducted in the vast majority of academic laboratories are considered of little value, and are thus excluded from many risk decisions. The reasons for this are multifold, and can differ between jurisdictions, but the overarching philosophy that guides this decision is an assumption that data collected in guideline studies are reproducible and reliable, and thus would be interpreted similarly between different regulators (125, 126, 164). Importantly, guideline studies typically examine very high, toxic doses that are completely unrelated to levels that are relevant to human exposures. This is why the lowest dose that had been examined in a guideline study was 50 000 µg/kg/day (3) prior to the advent of research focusing on BPA as an endocrine disruptor (62). It did not matter that the 50 000 µg/kg/day dose caused adverse effects; the assumption by regulators was that a dose 1000-fold lower than this LOAEL would be safe (based on linear extrapolation of high-to-low dose effects), although there were never studies conducted to determine whether this assumption was invalid for BPA or any other EDC that disrupts hormonal processes that are already operating at concentrations that cause effects (18, 165).

Problem 2: Linear dose–response models ignore nonmonotonic dose–response data

The use of linear dose models to estimate “safe” exposure doses is an issue that has emerged as a major source of controversy between endocrinologists and regulatory toxicologists. The assumption that a linear model is appropriate is based on the reliance on dogma formulated in the early 1500s that dose–response relationships are expected to be monotonic. This is often translated as “the dose makes the poison,” for example, higher doses should cause an increasing response (166). This core risk assessment assumption is false for hormones, hormonally active drugs, and EDCs such as BPA (13).

One problem is that for regulators to accept a phenomenon such as nonmonotonicity, they often insist that the underlying mechanisms are known; this is a criterion that no issue in science can fully meet, even though much is known, generally, about mechanisms mediating nonmonotonic dose–responses (Table 2). Work from Nadal and colleagues (25, 109, 110) published over the past 15 years revealed the molecular mechanisms by which a low (1 nM) dose of BPA alters glucose-induced insulin secretion. Through an ERβ-mediated action, low-dose BPA (1 nM) decreases transcription of genes encoding calcium channels, reducing calcium influx from mouse pancreatic β cells in primary culture (Fig. 4A). Yet, a high (100 nM) dose of BPA acts via an ERα-mediated activation of PI3K, which, in turn, activates calcium channels and has the opposite effect on calcium influx relative to the low BPA dose (Fig. 4B) (25). Thus, at 100 nM BPA, 2 mechanisms are active at the same time: an ERβ-mediated decrease in calcium influx and an ERα-mediated increase in calcium influx, while at 1 nM BPA only the ERβ-mediated mechanism is activated. Therefore, the nonmonotonic response for calcium flux is the outcome of 2 different molecular mechanisms involving 2 different ER subtypes, each of which shows a monotonic response to BPA (Fig. 4C). For a single outcome (calcium entry into the β cell), the lack of difference from negative control levels at a high dose of BPA may lead to the conclusion that there is no effect (that it is a no adverse effect dose), but at the mechanistic level, there is significant disruption of cell activity at both low and high doses of BPA.

Table 2.

Examples of mechanisms responsible for nonmonotonic dose responses

Type of Mechanism	Example	References
Overlapping monotonic responses acting on a common endpoint	An ERβ-mediated decrease in calcium influx and an ERα-mediated increase in calcium influx in pancreatic β-cells in response to BPA	(25, 167)
	An ERα-mediated suppression and an Erβ-mediated stimulation of calcium handling in myocytes
Low dose stimulation, high dose toxicity	At low doses, breast cancer cells proliferate in response to ER agonists, but at high doses, these ER agonists are cytotoxic. If the outcome is cell number, there is an inverted-U shaped relationship	(11)
Mixed cell populations with different expression of receptors	Endocrine tissues like the uterus and mammary gland have mixed populations of cells, some of which express ERα, others express ERβ, others express both. In the uterus, estrogen action via ERα promotes cell proliferation but action via ERβ promotes apoptosis.	(168)
Compounds that bind more than one receptor, with different affinities	BPA can bind mER and GPER at very low concentrations, nuclear ERs at somewhat higher concentrations, and other nuclear receptors at still higher concentrations. Thus, effects at high doses can illustrate the integration of signaling from multiple receptors	(169)
Altered receptor expression	Hormones (and EDCs) can alter expression of their own receptors. If the hormone down-regulates expression of the receptor, with higher doses, there are fewer receptors that can respond	(170)
Receptor desensitization	G-protein coupled receptors can be desensitized when continually bound by peptide hormones (eg, FSH, gonadotropins, glucagon)	(171)

Figure 4.

Mechanisms of nonmonotonic dose–response (NMDR) relationships. (A) BPA at the low (1 nM) dose acts via ERβ to decrease Cacna1e expression and decrease calcium entry into mouse pancreatic β cells in primary culture in the presence of 11 mM glucose, (B) At 100 nM BPA acts via ERα to enhances the calcium currents in a PI3K-dependent manner; Panels A and B from: Villar-Pazos S, Martinez-Pinna J, Castellano-Mu.oz M, et al. Molecular mechanisms involved in the non-monotonic effect of bisphenol-a on Ca2+ entry in mouse pancreatic β-cells. Sci Rep. 2017;7(1):11770, (open access). (C) The consequence of BPA acting via these two pathways at high doses versus just via ERβ at low doses is the appearance of a non-monotonic dose-response curve for calcium entry into β cells. Panel C was supplied by Dr. Angel Nadal with permission. (D) Up- and downregulation of estrogen receptors in the rat uterus in response to increasing doses (from low to high) of 17β-estradiol administered to ovariectomized females via Silastic capsules; from: Medlock KL, Lyttle CR, Kelepouris N, Newman ED, Sheehan DM 1991 Estradiol down-regulation of the rat uterine estrogen receptor. Proc Soc Exp Biol Med 196:293–300, with permission from the publisher. (E) Up- and downregulation of 8 genes involved in glucose metabolism in fetal mouse prostate mesenchyme cells in primary culture as the dose of estradiol increases from a physiological range (10 pM) to above a physiological range (100 pM); from: Taylor JA, Richter CA, Suzuki A, et al. 2012 Dose-Related Estrogen Effects on Gene Expression in Fetal Mouse Prostate Mesenchymal Cells. PLoS One 7(10) e48311, (open access). (F) Receptor specificity for BPA at low but not high doses: BPA binds to different receptors as dose increases, referred to as receptor crosstalk, which can result in nonmonotonic dose–response relationships.

Other mechanisms for nonmonotonic responses have been revealed. For example, in the rat uterus, low doses of estradiol upregulate ER expression whereas high doses downregulate ER expression (Fig. 4D) (172). Nonmonotonic responses of the fetal mouse urogenital sinus mesenchyme, with upregulation and downregulation of gene activity in response to estradiol, have also been shown in primary cultures (Fig. 4E) (173). Finally, nonmonotonic dose responses have been observed as a result of receptor crosstalk based on data from different studies with a wide range of concentrations. BPA binds at very low concentrations to GPR30/GPER, resulting in rapid intracellular signaling, then as the concentration of BPA increases, it binds to nuclear ERs, then at higher doses BPA binds to androgen and finally thyroid receptors, and alters their transcriptional activity, although this will vary based on the tissue being examined (Fig. 4F).

BPA has thus become the battleground chemical that threatens the validity of the assumptions underlying chemical risk assessments, which presume that toxicants such as BPA have a threshold below which there is no effect. Yet, no threshold exists for disruption of a hormonal system already operating above any putative threshold (174, 175). As expected, neither the regulatory agencies nor chemical corporations have accepted this possibility.

Problem 3: Study size requirements are not justified

Another significant impediment to the use of academic research in risk assessments is that the FDA sets standards for group/sample sizes for a study to be included in a risk assessment. The NIH Guide for the Care and Use of Animals in NIH-funded research ensures that researchers use the minimum number of animals required to achieve statistical significance, which must be justified by power analysis. In contrast, the guideline studies used in regulatory decision-making often have large sample sizes (n = 20-50 per dose, per sex) because much of the data collected are categorical (yes/no presence of a condition); the sample sizes in guideline studies are preset, and not calculated using power analysis.

In setting arbitrary group size numbers that are not selected based on power analysis, the approaches used by regulatory agencies violate ethical standards followed by federally funded academic researchers. Further, the use of preset sample sizes in guideline studies, predetermined without being subject to power analysis, means that these studies could be either over- or underpowered. Thus, requiring that all studies must have a large arbitrary number of animals per group ultimately biases decisions towards studies conducted by, or funded by, industry, that is not constrained by NIH regulations concerning animal research; strangely, animal experiments conducted by the FDA also violate NIH animal use guidelines.

Failure to Get CLARITY on BPA

This mini-review describes more than 20 years of controversy over the effects of BPA at low doses relevant to human exposure. Recently, large initiatives, such as the CLARITY-BPA study, have been implemented to specifically address some of these issues. As noted at the beginning of this review, the CLARITY-BPA study was designed to combine a guideline study with additional outcomes that had previously been shown to be sensitive to BPA or other estrogenic EDCs.

Designing the shared methods to be used in the CLARITY-BPA study was challenging. For example, the academic scientists expressed deep concern over the use of a NCTR rat strain that was less sensitive to estrogenic hormones than the NTP’s CD-1 mouse model (65). Surprisingly, the NTP and the FDA initially resisted the demand to include a positive control for estrogenic activity; however, after much debate, 17α-ethinylestradiol (2 low doses) was selected for this purpose. There was disagreement over the use of gavage administration of BPA, particularly because the FDA would not include a naïve negative control (ie, rats that were not restrained or gavaged). As described above, partway through the study, it became clear that the FDA laboratory had struggled to control BPA contamination in their rodent colony (137), so there were concerns that the CLARITY-BPA animals may have been contaminated as well. This remained largely unanswered as the FDA abandoned its biomonitoring protocol and only reported testing a few CLARITY-BPA animals for their BPA levels (4, 6)

Ultimately, the CLARITY-BPA study included animals that were exposed via gavage to BPA (first through the pregnant mother, and then after birth directly to the offspring) from gestational day 6 through just up to postnatal day 21 or throughout postnatal life up to the time of tissue collection. The BPA dose groups were administered 2.5, 25, 250, 2500, or 25 000 μg/kg/day, and the positive control 17α-ethinylestradiol dose groups were administered 0.05 or 0.5 μg/kg/day. These 17α-ethinylestradiol doses are lower than the BPA administered doses because it is generally considered a more “potent” estrogen: in oral contraceptives, the dose of 17α-ethinylestradiol that is sufficient to inhibit women’s fertility is approximately 0.5 µg/kg/day, and in studies with mice, effects have been observed at and below the 0.05 µg/kg/day dose of 17α-ethinylestradiol (176-178). The CLARITY-BPA animals were euthanized at either 1 year or 2 years of age, with a smaller subset euthanized at earlier ages for select endpoints evaluated by academic investigators. Typical outcomes measured at 1 and 2 years of age included guideline study endpoints conducted by the FDA (eg, organ weight, histopathology, clinical chemistry measurements). Also measured at these ages, as well as younger ages, were outcomes selected by the academic investigators, including modern molecular/genetic, histochemical, epigenetic, neuroendocrine and behavioral experimental methods (for additional reviews see (7, 179-181)).

In spite of the difficulties in reaching agreement between the FDA and academic participants, and the challenges concerning whether contamination did or did not occur—as well as other challenges due to not meeting the group size requirements of the academic investigators (45) and skewed animal body weights observed in some academic studies (181)—the CLARITY-BPA study produced some important results (Fig. 5). As discussed in more detail elsewhere (7, 179), there were significant effects of BPA on guideline endpoints in the animals exposed to 2.5 μg/kg/day (mammary gland, liver, kidney, and male reproductive tract); 25 μg/kg/day (female reproductive tract, kidney, and thyroid); 250 μg/kg/day (thyroid, liver, kidney, pituitary and spleen); 2500 μg/kg/day (female reproductive tract, thyroid, testis, liver, kidney, adrenal, and pancreas); and 25 000 μg/kg/day (ovary, female reproductive tract, pituitary, and male reproductive tract). The academic investigators similarly reported effects of BPA in all dose groups, with differences depending on the organ and outcome, as well as sex-specific effects (7, 180).

Figure 5.

Significant effects of BPA reported in the CLARITY-BPA study. (A) Summary of effects from the core guideline study conducted by FDA, and (B) summary of effects from academic studies. The presence of the organ indicates significant differences for one or more measurement in that organ, at the dose indicated, compared to the concurrent negative controls.

Two important studies have used modern mathematical and statistical approaches to draw broader conclusions from the CLARITY-BPA study data. In the first, repeated measures of mammary gland growth parameters showed nonmonotonic dose responses, and Montevil and colleagues (182) demonstrated mathematically that the nonmonotonic responses were not a random effect. Rather, there was a consistent pattern that emerged, with low BPA doses increasing mammary growth parameters; the break between low dose and higher dose responses was repeatedly shown to fall between the 25 and 250 μg/kg/day dose groups (182). In the second study, statistical correlations between studies conducted by the academic investigators in CLARITY-BPA revealed patterns of effects across the organs evaluated blind to treatment group by different investigators. Even at the lowest dose (2.5 μg/kg/day), this mathematical analysis suggested that the effects of BPA were nonrandom, and therefore should not be dismissed by the FDA as biologically implausible (7).

The consistency of adverse effects at 2.5 µg/kg/day in both the guideline and academic studies in CLARITY-BPA, in addition to numerous other published findings of effects at this low BPA dose in animals, has led the academic collaborators to conclude that this dose should now be the LOAEL (7). Using the 2.5 µg/kg/day dose as the new LOAEL and using a 1000-fold linear safety factor correction to account for animal to human variability (10×), variability among people (10×), and the use of a LOAEL instead of an actual no adverse effect level or NOAEL (10×), the new acceptable daily intake level should be 2.5 ng/kg/day (Fig. 6) (7). This represents a shift in the current FDA’s ADI dose by 20 000-fold relative to the current ADI of 50 µg/kg/day that was established in the 1980s (3, 179). We also note that several prior studies have shown effects of BPA at doses below 2.5 µg/kg/day (eg, on brain and behavior (183), mammary gland development (184), and female reproduction (98), among others). Thus, it cannot be dismissed that even a shift in the ADI to 2.5 ng/kg/day may not be sufficiently protective for public health.

Figure 6.

Calculation of the reference dose (or acceptable daily intake dose) from the old and new LOAEL. (A) Prior to the CLARITY-BPA study, the LOAEL calculated from guideline studies was 50 000 µg/kg/day. Using adjustment factors to account for uncertainties in the data (also called safety factors), the reference dose/ADI was determined to be 50 μg/kg/day. (B) Because effects were observed in the CLARITY-BPA study, even in the guideline endpoints evaluated by FDA, at 2.5 μg/kg/day, the new reference dose/ADI would be 2.5 ng/kg/day.

Sadly, the significant adverse effects observed in the very low dose groups in both the guideline study and the studies conducted by academic investigators in CLARITY-BPA were rejected by the FDA as not biologically plausible. The FDA drew this conclusion because the dose–response relationships were not always monotonic, and the effects were not consistent across all tissues; neither of these assumptions are consistent with known mechanisms of hormone action. Further, FDA authors stated: “However, while there were statistically significant differences from the vehicle control, particularly with the low stringency tests applied with the histopathology endpoints, these effects did not show a coherent and plausible pattern consistent with BPA-induced lesions … Based on a weight of evidence approach, we conclude that the core study data do not suggest a plausible hazard of BPA exposure in the lower end of the dose range tested” (6). The FDA’s conclusion, that effects at low doses that are different from effects at high doses are “not plausible,” is clearly contradicted by the consistency of findings across a large number of endpoints and organs in both the FDA’s guideline study and academic investigator findings in CLARITY-BPA (7). Importantly, it is mathematically straightforward to demonstrate that a monotonic response has occurred, but more challenging to assert that a nonmonotonic dose response can be distinguished from “noise” in an experiment (37, 166, 185). Yet, other regulatory agencies including ANSES, the French Agency for Food, Environmental and Occupational Health and Safety, have acknowledged that nonmonotonic dose responses are biologically plausible for EDCs (186). Further, ANSES has proposed a framework by which observed nonmonotonic dose responses can be evaluated for their strength and plausibility.

There are still other flaws with the approaches taken by the FDA to address the observed effects of BPA in the CLARITY-BPA study, including in their own portion of the study (eg, the guideline study). For example, the FDA relied on comparisons with “historical controls,” for example, actual nongavaged negative controls from previous experiments, to ignore significant differences between BPA-exposed animals and the concurrent gavaged negative controls (7, 187). This unscientific practice is troubling, especially because further comparison of CLARITY-BPA data with “historical controls” indicated that the CLARITY-BPA study rats were very different in many aspects from prior studies with this strain of rats when gavage administration was not used. One glaring example is that 17α-ethinylestradiol delayed puberty in CLARITY-BPA female rats (6), while in prior studies puberty was accelerated by 17α-ethinylestradiol in rats from the same colony at the NCTR, consistent with numerous other studies showing that developmental exposure to exogenous estrogens at low doses accelerates puberty onset in females (188), which is also true for BPA (189).

BPA Safety: What Do We Know Now That We Did Not Know 20 Years Ago?

The CLARITY-BPA study was a novel approach to address important questions about the effects of low doses of BPA. The combined study design, including a traditional guideline study with additional hypothesis-driven outcomes, allowed investigators to address whether the endpoints evaluated by academic researchers are more sensitive, and perhaps more appropriate, to evaluate the safety of a wider range of environmental chemicals. Ultimately, the results of the guideline and academic studies confirmed what expert groups had concluded more than a decade prior, and research had shown in the decade prior to that. Specifically, there are effects of low doses of BPA on outcomes that are acknowledged to be adverse (74), and these effects occur at doses that are far below what regulatory toxicologists are willing to accept because they challenge the core assumptions underlying chemical risk assessments.

The CLARITY-BPA study cost millions of dollars and required years of work, but did not change the view of regulators at the FDA based on old approaches and assumptions that violate basic principles of endocrinology. Yet, other studies using a similar design (a guideline study with additional “hypothesis-driven” endpoints) have drawn conclusions similar to the CLARITY-BPA academic participants: (190) A study conducted by researchers at the Danish National Food Institute found effects of low doses of BPA on neurobehaviors, male and female mammary glands, and sperm count, with relatively few effects on guideline endpoints (191, 192). In addition, another group collaborated by sharing tissues from a study examining the effects of BPA exposure during pregnancy in rhesus monkeys; numerous adverse effects in a variety of fetal tissues were identified (193-199) along with an analysis of BPA in both maternal and fetal serum (151).

Federal law requires that the FDA division in charge of food safety (CFSAN) assess the safety of food additives, which is defined in 21 CFR §170.3(i): “Safe or safety means that there is reasonable certainty in the minds of competent scientists that the substance is not harmful under the intended conditions of use.” The FDA continues to insist that BPA does not pose a threat to human health, even though “competent scientists” from a wide range of disciplines around the world have disputed this conclusion (7, 74, 114, 124, 187), including an FDA scientific advisory panel that rejected the agency’s 2008 risk assessment of BPA because of a lack of consideration of academic studies (200).

From BPA to other EDCs: Broader Lessons Learned

BPA has become over the last two decades one of the most studied EDC for understanding both low-dose effects and nonmonotonic dose responses. In fact, mechanistic studies of BPA have uncovered important molecular mechanisms responsible for nonmonotonic responses that not only inform scientists about the ways these responses can manifest (Table 2), they also counter the FDA’s assumptions that these responses are biologically implausible. Nonmonotonic dose responses have been reported for dozens of EDCs, and this work challenges scientists from a range of disciplines to use better mathematical and statistical approaches before dismissing non-monotonic responses (37).

Findings of adverse effects of BPA at doses far below the EPA/FDA’s LOAEL strongly challenge the approaches used by regulatory agencies to estimate “safe” levels of exposure. This is true not just for BPA, but potentially for all chemicals. The debate between regulatory toxicologists, who defend chemical risk assessment methods and assumptions, and endocrinologists with expertise in the mechanisms of hormone action and EDCs such as BPA, has been ongoing for more than 2 decades (165). The Endocrine Society has issued a number of position statements identifying the endocrine principles that should dictate the approaches to assess the safety of EDCs and the flaws in the approaches regulatory agencies have used to estimate the hazards of EDCs (8-10, 201).

Similarly, the CLARITY-BPA study, along with the work from the Danish National Food Institute, allows for comparisons of guideline endpoints and modern hypothesis-driven outcomes. What is clear across these studies is that the disease-relevant outcomes evaluated by academic laboratories are sensitive and represent adverse effects. Guideline studies would be improved by incorporating these outcomes, which would require collaborative approaches like those attempted in CLARITY-BPA, since no single laboratory could have the range of expertise of the CLARITY-BPA academic collaborators.

Importantly, an evaluation of the 17α-ethinylestradiol positive controls in the CLARITY-BPA study also challenge the typical claim, from regulators and industry scientists, that guideline studies are reliable because they are reproducible (123). The effects of 17α-ethinylestradiol in the CLARITY-BPA study were wildly divergent from the effects (or lack thereof) in a previous guideline study of the chemical (179).

The story of BPA also illustrates how outdated assumptions and approaches initially formulated in the 1500s are still used in chemical risk assessments, even though they violate basic principles of hormone action. Thus, 3 major issues are disputed concerning the FDA’s position regarding chemical risk assessments:

• 1) Regulatory agencies like the FDA have taken the position that: “For an observed effect to be toxicologically relevant (ie, potentially adverse), a clear dose response should be seen (eg, increasing the dose of a test substance causes an increase in the observed effect in the test subjects)” (202); that is, a “relevant” dose–response curve occurs only when the dose-response relationship is monotonic. However, nonmonotonic dose–response relationships are common for hormones, drugs that interact with receptors, and EDCs such as BPA (13, 203).

• 2) The core assumption in risk assessment is that there is a threshold below which there is no effect of a toxicant. The assumption of a threshold is disputed based on experimental data (174) and on theoretical grounds (175). This assumption by the FDA leads to the conclusion that there is a dose of BPA below which no effect occurs, which results in a large error in estimating the safety of BPA (11). Yet, even the lowest dose of BPA in the CLARITY-BPA study induced adverse effects, and even lower doses (ng/kg) have been shown to disrupt development of the mammary gland (184, 204). Position statements by members of the Endocrine Society identified above reiterate why the assumption of a threshold of adversity should be rejected for BPA as well as many other environmental pollutants that disrupt hormonal systems that are already operating above any putative threshold.

• 3) The FDA has asserted that, to consider an effect adverse, “the observed effect should occur in both sexes of test species” (202). The strange aspect of this assumption is that the endocrine system of males and females functions differently, which impacts the process of sexual differentiation during fetal life (130), the hormonal transitions during puberty, the establishment of menstrual cycles in women but not men throughout the fertile period of life, and postmenopausal changes in women and age-related changes in hormones in men that create hormonal differences during aging (205, 206). Because EDCs such as BPA impact different hormonal backgrounds in males and females, it is illogical to have as a core assumption that the response to an EDC has to be the same in males and females to accept statistically significant findings as biologically meaningful. In fact, the response observed in male and female rodents to fetal exposure to low doses of BPA is typically not only different, but in many cases opposite (38, 102).

Unfortunately, there is 1 more lesson to be learned from BPA: the issue of “regrettable substitutions” (207). As concerns have been raised by the public, and consumer patterns have shifted away from products containing BPA, the sale of “BPA-free” products has increased. Many of these products contain bisphenol analogues. Thus, there has been a “whack-a-mole” approach of substituting 1 hazardous chemical with equally hazardous related compounds, such as BPS and BPF (208, 209). Although the number of studies examining these analogues is much smaller, there is sufficient evidence that some of these chemicals also cause harm at low doses similar to BPA (210-213).

Conclusions

In more than 20 years, scientific knowledge of the effects of BPA has grown exponentially (Figs. 2 and 7). The amount of evidence that is available is overwhelming, and the conclusion is clear: low doses of BPA alter hormone-sensitive organs and are related to a wide range of human noncommunicable diseases based on evidence from human studies; the human data are supported by numerous findings from animal experiments and in vitro mechanistic studies. The lowering of the ADI by 20 000-fold might require the elimination of BPA in food contact items in the United States, which has already occurred in France. However, in 2010 the FDA acknowledged that by determining that BPA met the standard in 1963 of a reasonable certainty that the material would not be harmful under the intended conditions of use (21 CFR 177.1580), that they gave up the ability to even require industry to identify products that contain BPA (48). The FDA has abrogated its responsibility to ensure the safety of food and food/beverage packaging in the US], and the FDA continues to allow industry to declare chemicals such as BPA to be safe (214-216).

Figure 7.

A timeline of major developments in the history of BPA.

Glossary

Abbreviations

ADI

acceptable daily intake

BPA

bisphenol A

CDC

Centers for Disease Control and Prevention

CIIT

Chemical Industry Institute of Toxicology

EC50

50% effective concentration

EDC

endocrine disrupting chemical

ER

estrogen receptor

EPA

Environmental Protection Agency

FDA

Food and Drug Administration

GLP

good laboratory practice

GPR

G protein–coupled receptor

LOAEL

lowest observed adverse effect level

MBP

4-methyl-2,4-bis (4-hydroxyphenyl)pent-1-ene

NCTR

National Center for Toxicological Research

NHANES

National Health and Nutrition Examination Survey

NIEHS

National Institute of Environmental Health Sciences

NTP

National Toxicology Program

Additional Information

Disclosure Summary: L.N.V. has received travel reimbursements from universities, governments, NGOs, and industry to speak about endocrine-disrupting chemicals. She is a paid member of the Leadership Council at SUDOC LLC. Her EDC-related work has been supported by US federal agencies, the University of Massachusetts Amherst, and NGOs including the Cornell Douglas Foundation and the Great Neck Breast Cancer Coalition. F.V.S. has received travel reimbursements from universities, medical and scientific societies, governments, NGOs and industry to speak about endocrine-disrupting chemicals. He is a paid member of the Leadership Council at SUDOC LLC. His EDC-related research has been supported by US federal agencies, the University of Missouri, and NGOs.

Data Availability

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.