Low dose effects of bisphenol A

Abstract

In 2007, a group of experts critically analyzed hundreds of publications on bisphenol A (BPA), including the evidence for low dose effects. Here, we have updated these evaluations to determine the strength of the evidence for low dose effects of BPA. Based on the cut-offs for “low doses” established previously (i.e., the lowest observed adverse effect level [LOAEL], or 50 mg/kg/day for mammalian studies), we identified more than 450 low dose studies. Using an integrative approach, we examined five endpoints in depth that had evidence from two or more study types (in vitro, in vivo laboratory animal, and human). Based on all available studies, we are confident that consistent, reproducible, low dose effects have been demonstrated for BPA. We conclude that the doses that reliably produce effects in animals are 1–4 magnitudes of order lower than the current LOAEL of 50 mg/kg/day and many should be considered adverse.

Keywords: :

• cell culture
• endocrine disruptor
• epidemiology
• LOAEL
• non-monotonic dose response
• aquatic animal
• xenoestrogen

Introduction

Bisphenol A (BPA; CAS# 80-05-7) is a high-volume chemical that is widely used in the production of consumer products.¹ BPA has estrogenic properties in vitro and in vivo, binds the membrane-associated estrogen receptor (ER), and may also act via the androgen receptor, thyroid hormone receptor, peroxisome proliferator-activated receptor-γ, and other endocrine-relevant signaling pathways.² A recent study by the US Environmental Protection Agency (EPA) and US National Toxicology Program (NTP) assigned BPA the third highest Toxicological Priority Index (ToxPi) score of the 309 chemicals examined based on its ability to interact with a number of signaling pathways.³

This review focuses on examining one aspect of the action of BPA, effects at low doses. We provide a brief overview of previous assessments of low dose effects and an updated review of low dose studies published since 2007. Further, we use an integrative biology approach to identify endpoints consistently affected by BPA exposures, examining five endpoints that have been observed in multiple studies from several laboratories, assessing the effects of low doses on multiple levels of biological organization (cells, animals, and human populations).

Defining “Low Dose”

In 2001, an NTP panel of experts assessing low dose effects for a number of endocrine-disrupting chemicals (EDCs) defined “low dose” as any biological change occurring in the range of typical human exposures or at doses lower than those tested in traditional toxicology assessments.⁴ At that time, the NTP panel concluded that there were low dose effects for a number of EDCs. It should be noted that the NTP’s definitions for low dose (human exposure levels vs. doses used in traditional toxicology assays) produce very different cut-offs for many EDCs, including BPA.⁵

Defining “low dose” cut-offs for BPA studies

Using the NTP’s definition of doses lower than those tested in traditional toxicology assessments,⁴ a cut-off for low dose BPA studies was set at 50 mg/kg/day for laboratory animals;⁶ this was identified as the LOAEL in traditional toxicology studies. This cut-off was selected because no adverse effects should ever be observed below this dose.

Most aquatic animals are exposed to EDCs via the water or soil in which they live. For this reason, the low dose cut-off for aquatic animals was set at 17.2 mg/L, the maximum BPA concentration detected in an environmental sample.⁷ The low dose cut-off for cell culture experiments was set at 1 × 10⁻⁷ M based on calculations of circulating BPA concentrations in animals administered the LOAEL.⁸ Finally, establishing a low dose cut-off for epidemiology studies is simple, because one definition of “low dose” is a dose in the range of typical human exposures.⁵ Thus, any study of non-occupationally exposed individuals meets this criterion.

“Low dose effects” make no assumptions about the shape of the dose response curve. Therefore, “low dose” is not synonymous with non-monotonicity. Non-monotonic dose responses have recently received significant attention (for example, ref. 5) and will be discussed briefly in this review.

Previous Reviews of the BPA Low Dose Literature

Following the 2001 NTP report on low dose effects for EDCs, several reviews focused on the ability of BPA to cause harmful effects at doses below the US EPA reference dose of 50 μg/kg/day or the LOAEL of 50 mg/kg/day. In 2005, vom Saal and Hughes quantified the number of studies showing harmful effects from low dose BPA.⁹ In 2006, vom Saal and Welshons identified more than 100 studies that reported significant effects of BPA below the LOAEL; 40 studies reported effects at doses below 50 μg/kg/day.¹⁰ In 2007, a NIH-sponsored panel of experts working on EDCs critically analyzed hundreds of publications on BPA, publishing the Chapel Hill consensus statement;¹¹ the accompanying detailed reviews addressed low dose effects in laboratory animals,⁶ wildlife including amphibians and fish,⁷ in vitro cell and tissue culture,⁸ cancer,¹² and human populations.¹ In 2008, the National Toxicology Program reviewed the entirety of the BPA literature, including studies relevant to low doses.¹³ Finally, in 2012, Vandenberg and colleagues performed weight-of-evidence evaluations and assessed low dose effects of BPA on the prostate and mammary gland,⁵ and in 2013, J Rochester published a comprehensive review of all BPA epidemiology studies published to date.¹⁴

Importance of Route of Administration?

One issue that has received significant attention in the study of BPA is the route of administration, because the diet is considered the major, and sometimes the sole, route of human exposure.¹⁵ Because information is still lacking about all of the products containing BPA and non-dietary sources of exposure have been and continue to be identified (reviewed in ref. 16), the requirement that animal studies use oral administration is not appropriate. Newborn rodents and monkeys have limited capacity to metabolize BPA and route of administration has little effect on average serum BPA levels.^17-20 In addition, in adult animals, the proportion of ingested BPA circulating in serum that remains unconjugated (bioactive) differs dramatically depending on whether absorption occurs via the oral mucosa, which bypasses first-pass metabolism, or only in the gastrointestinal tract (for example, due to gavage administration), which results in a higher proportion of BPA being metabolized.²¹

New Efforts to Focus on Low Dose

Over the past few years, new attention has focused on addressing the issue of low dose exposures. In 2009, the National Institute of Environmental Health Sciences (NIEHS) launched a program to fill data gaps related to the toxicity of BPA.²² Extramural grants were funded, and a BPA grantees consortium was established; the study of low doses was an important issue that was considered during the design and execution of the funded studies. Additional efforts being undertaken by NIEHS involve a chronic toxicity study of BPA being conducted in collaboration with the US FDA. This guideline compliant study will also include low doses, and will examine endpoints used in traditional toxicology testing as well as hypothesis-driven mechanistic endpoints developed in academic laboratories. The results from this study will be available over the next few years.

This review is focused on summarizing the published results of these studies, as well as other peer-reviewed studies from around the world. Below, we summarize new low dose findings for in vitro, aquatic animal, mammalian laboratory animal, and epidemiology studies. We also examine a few examples in more depth and conclude with consensus statements regarding the strength of the overall evidence for low dose effects of BPA.

Brief Review of New Low Dose Findings

In vitro

We identified more than 100 new in vitro studies that included doses at or below the cut-off of 1 × 10⁻⁷ M. Similar to what was reported previously,⁸ these studies have largely focused on cell lines and primary cells isolated from the male or female reproductive tracts, cells and tissue explants involved in energy balance, neuronal and glial cells, and pituitary cells (Table S1).

Studies of cells isolated from the male reproductive tract identified several tissues that are sensitive to BPA. Effects were observed on cell proliferation, gene expression, and hormone production in human prostate cancer cell lines, as well as spermatogonia, Leydig cells, whole testes, and testicular seminoma cells from a number of species (see refs. 23–34 and Table S1). Cells isolated from the female reproductive tract, including ovarian granulosa or theca-interstitial cells, oocyte-cumulous complexes, and placental cytotrophoblasts, have also been studied (see refs. 35–44 and Table S1). Low dose effects were observed in many of these cells on endpoints ranging from altered hormone secretion^36,39 to altered gene expression^36,40,42 and altered proliferation.⁴³

A large number of studies have also examined the effects of BPA on breast cancer cell lines and primary cells. Although many classical endpoints (i.e., gene expression, cell proliferation)^45-50 are affected by low doses of BPA in these cells, newer endpoints have been assessed as well. For example, several studies demonstrate that BPA concentrations below 1 × 10⁻⁷ M interfere with chemotherapeutic agents, conferring resistance in cell lines and primary cells isolated from human patients.^51-53

In vitro studies have also focused on cells relevant to energy balance. Two studies have shown that low doses affect the release of insulin and ion channel activity in isolated pancreatic islets.^54,55 Others show altered release of adiponectin and inflammatory cytokines, changes to the expression of adipocyte proteins, and lipid accumulation in adipose tissue explants or isolated adipocytes and pre-adipocytes.^56-59 BPA has also been shown to affect muscle cells, including cardiac myocytes, with low dose effects on arrhythmogenic activities and physiological endpoints related to calcium cycling and contractility.^60-62

Overall, these in vitro studies have examined a large number of cell types and biological endpoints that may be relevant to cellular function in vivo. The low dose cut-off of 1 × 10⁻⁷ M has been debated and it may be reasonable to shift the low dose cut-off for in vitro studies to 10⁻⁹ M or lower. Many studies examining doses at or below 10⁻⁹ M report significant biological effects,^{23,24,26-28,30,33,43,49,52,55,57,60,61,63-78} challenging the hypothesis that BPA only has “weak” endocrine disrupting activity.

Non-mammalian animal models

We identified more than 75 studies of non-mammalian models that examined doses of BPA below the low dose cut-off of 17.2 mg/L (Table S2). Most examined aquatic animals including invertebrates such as daphnia and mollusks, fish such as medaka, trout and zebrafish, and the amphibian Xenopus laevis.

One well-studied endpoint is the expression of vitellogenin, an egg yolk precursor protein that is expressed in females of many egg-laying species and is a sensitive biomarker for estrogen-like responses. BPA exposures below 12.2 mg/L induced vitellogenin expression in males of various fish species, including fathead minnows,⁷⁹ goldfish,⁸⁰ medaka,⁸¹ and zebrafish.⁸² Other studies have focused on the effect of BPA on reproductive endpoints. In some cases, reproductive success was shown to be decreased based on measures of hatchability,⁸³ sperm motility,⁸⁰ fertilization rates,⁸⁴ and the cumulative number of offspring produced per animal.⁸⁵ Mortality is also a common endpoint that has been assessed in both invertebrate and vertebrate species, with significant decreases in survival noted in a number of invertebrates^84-89 and fish.^87,90-94 Studies have also reported altered gene expression (often of hormone receptors and enzymes, as well as classical liver markers) following low dose BPA exposure in zebrafish,^95-97 a well-established genetic model, and in other less well characterized organisms including aquatic midge,⁹⁸ Atlantic cod,⁹⁹ carp,¹⁰⁰ Kryptolebias marmoratus,^101-106 rainbow trout,¹⁰⁷ rare minnow,^108,109 and Xenopus.^110,111

One final study deserves attention due to its use of Organization for Economic Cooperation and Development (OECD) guidelines to test the effects of BPA on histopathological changes in the ovary.¹¹² Zebrafish were exposed to very low doses of BPA for 2 wk starting at 16 wk of age; histological abnormalities were observed at 1, 10, 100, and 1000 µg/L, providing strong evidence for low dose adverse effects on the fish ovary. OECD guidelines examine endpoints that are easily assessed, are reproducible, and focus on adverse effects,¹¹³ although there is debate regarding whether they examine appropriate and sensitive endpoints for endocrine disruption.^114-116

Mammalian laboratory animals

Previous assessments found evidence for low dose effects of BPA on a number of endpoints, including brain development, behaviors, development of the male reproductive tract, growth and metabolism, and adult male reproductive health using the LOAEL (50 µg/kg body weight/day) as the low dose cut-off.^6,13 Additional evidence for low dose effects has continued to accumulate from mammalian studies focused on those endpoints and others (Table S3). In this section, we will focus on systems and endpoints for which strong evidence did not exist in 2007.

Female reproductive system

Prior to 2007, six studies reported effects of early life BPA exposure on gene expression and histology of the female reproductive tract as well as disturbances in chromosome behavior in oocytes.⁶ Since that time, seven additional studies have revealed low dose effects on abnormalities (including hyperplasias) in the female reproductive tract,^117-119 uterine gene expression,^120,121 and measures of fertility.^122,123 Numerous studies have also shown effects on the formation of ovarian follicles, the presence of polycystic ovaries, follicle growth, and the genetic quality of the egg; these are discussed in a later section of this review.

Fewer studies have examined the effects of adult exposures on female reproduction. These studies have identified some effects on oocyte health^124-126 and uterine gene expression,^127,128 but many endpoints have not been replicated (Table S3). Thus, while the evidence for low dose effects on female reproductive endpoints following developmental exposures is strong, the evidence for effects following adult exposures remains inconclusive.

Mammary gland

Since 2007, 11 studies, including one examining the non-human primate,¹²⁹ have examined the effects of developmental exposures to BPA on the morphology of the mammary gland^130-136 and altered responses of this organ to carcinogenic challenges.^137-139 Another recent study revealed the presence of carcinomas in the mammary glands of rats exposed to low doses of BPA only during early development in the absence of any additional carcinogen treatment, providing further evidence that the mammary gland is affected by BPA.¹⁴⁰ A recent weight-of-evidence analysis found strong support for low dose effects and high levels of consistency between studies for both endpoints;⁵ low doses were found to alter mammary gland morphology at many life stages, and numerous studies reported that low doses increase the incidence of preneoplastic and neoplastic lesions, as well as increase the response to chemical carcinogens. Some of these studies are discussed in more detail below.

Immune system

Five recent studies suggest that developmental BPA exposures alter the response of rodents to infectious agents and allergens.^141-145 These studies typically show effects of BPA on some aspects of immune response, but not all (i.e., altered innate but not adaptive immune responses, effects in some organs but not others, etc.). Because these studies used different agents to invoke immune responses and assessed different organs’ responses, there is not yet enough evidence to determine whether the effects of BPA on the immune system are consistent. However, these studies collectively provide weak support for low dose effects of BPA on the response of the immune system to specific agents.

Brain morphology

Since 2007, five studies have demonstrated low dose effects of adult BPA exposures on brain endpoints such as synapse formation and remodeling.^146-150 Some of these effects were observed only in response to additional hormone treatments (i.e., BPA attenuates the effects of estrogens on synapse formation/remodeling). Importantly, these studies collectively demonstrate consistent low dose effects on neuronal synapses in rodents and non-human primates.

Metabolism

Finally, although the endpoints examined are variable, three recent studies report effects of adult BPA exposures on metabolic endpoints. Exposures lasting 8 d to 10 wk in length altered blood glucose and insulin levels, glucose and insulin tolerance, food intake, triglyceride levels, and responses to sugar challenges.^151-153 The variability in affected endpoints may be evidence of the diverse effects of adult BPA exposures and/or differences in exposure periods.

We used a cut-off of 50 mg/kg/day, the LOAEL for BPA, as a cut-off to determine which studies should be considered “low dose.” This cut-off is consistent with the NTP’s definition of “low dose,”⁴ and is the same cut-off that was used previously by the Chapel Hill panel.⁶ Yet, because studies estimate human exposures from dietary sources in the range of 0.1–5 µg/kg body weight/day,¹⁵⁴ there have been suggestions that only studies examining exposures in this range should be considered “low dose.”¹⁵⁵

Importantly, many significant effects are observed following administration of doses below the EPA reference dose of 50 µg/kg body weight/day, and some of these studies show effects in the range of estimated human exposures (≤ 5 µg/kg body weight/day).^{118,124,126,130,132,133,135,136,141,143,150,152,156-185} Previous reviews have also noted a relatively large number of studies that demonstrated effects in the range of human exposures from dietary sources.^6,9 Studying actual or suspected human exposure levels in rodents may not be appropriate for risk assessment purposes as this leaves no margin of error for differences in metabolism between species or individuals, the presence of vulnerable human subpopulations, or the possibility that animal models are less sensitive than humans. These factors are traditionally accounted for by calculating a “safe” reference dose from the NOAEL using safety factors.⁵ Thus, if adverse effects are observed in animals at doses ≤ 50 µg/kg/day and these safety factors are applied, the “safe” dose for humans would be calculated as 50 ng/kg/day or lower.

Humans

Since 2007, more than 50 epidemiology studies have examined various health outcomes in relation to BPA exposure. Eight studies examining neurobehavioral outcomes have been published from four prospective US cohorts and a cross-sectional Korean cohort.^186-194 These studies focused on pre- and postnatal BPA exposure and neurobehavioral outcomes measured through either parent report or standardized testing. Outcomes included anxious and depressive behavior as well as social and interpersonal problems in children from several age groups. Collectively, these studies raise concern over pre- and postnatal exposure to BPA and negative effects on neurobehavioral outcomes in children.

Fifteen recent studies with various designs have examined the association between BPA and various aspects of metabolic syndrome.^195-210 In several, increased urinary BPA concentration was associated with higher odds of increased body mass index, obesity, cardiovascular disease, insulin resistance, and type 2 diabetes (Table S4). Cross-sectional and longitudinal studies have also demonstrated associations between BPA exposure levels and coronary artery disease and peripheral arterial disease.^211-213

Other studies have investigated associations between BPA and thyroid dysfunction. This is of particular concern because thyroid hormones play an important role in brain development. Prenatal BPA exposure was associated with decreased serum thyroid stimulating hormone (TSH) in male neonates.^214,215 Several cohorts suggest negative associations between urinary BPA and TSH and/or total thyroxine (T4) in adolescents and adults.^216,217 Collectively, these studies indicate that BPA may affect thyroid function at different life stages.

Finally, since 2007, multiple studies have examined BPA and human reproduction. Several prospective IVF cohort studies examined associations between BPA and early IVF outcomes in women.^218-222 Significant negative associations between BPA and peak ovarian estradiol were observed in multiple cohorts, although other associations were only observed in a single cohort (i.e., oocyte yield, and other downstream IVF outcomes). In one cohort, later reproductive endpoints (i.e., embryo implantation) were negatively associated with urinary BPA concentrations.²²²

In sum, the epidemiology literature indicates there are reproducible effects of environmental BPA exposures on a number of endpoints including behaviors, metabolic syndrome, and thyroid hormone signaling. Although we did not assess the quality of each of these studies, and some are certainly limited due to their size and/or design, these findings taken together lead to the conclusion that current levels of BPA exposure are related to a number of disease outcomes in humans that are consistent with results from controlled animal studies and mechanistic in vitro studies.

Non-Monotonicity

As mentioned earlier, non-monotonicity is distinct from “low dose.” A number of studies examining low doses of BPA have revealed non-monotonic dose response curves (NMDRCs), which often manifest as U- or inverted U-shaped curves. Non-monotonicity is easily revealed when lower doses induce significant differences from untreated controls and higher doses do not. It is common for these findings to be presented as not showing a dose–response relationship because the relationship is not linear.⁵

In the in vitro literature, studies with five or more groups are quite common, and thus, a number have demonstrated non-monotonicity (Table 1 and ref. 223). Fewer in vivo studies have examined the large number of doses necessary to demonstrate non-monotonicity (Table 2). Additional analyses of dose response shapes are needed to determine the frequency of NMDRCs for BPA studies, the types of endpoints that manifest NMDRCs, and whether studies of these endpoints in different laboratories produce consistent dose response shapes.²²⁴

Table 1. Endpoints demonstrating non-monotonicity in in vitro and in vivo low dose studies of BPA

Study type	Model system	Endpoint	Doses tested
In vitro	Primary prostate mesenchymal cells	Total DNA content, Total RNA content	10 doses between 10^−13 and 10^−4 M
In vitro	mLTC-1 Leydig tumor cells	Candidate gene expression	5 doses between 10^−8 and 10^−4 M
In vitro	Primary pancreatic islets (mouse)	Insulin content of islets	5 doses between 10^−10 and 10^−6 M
In vitro	Adipose tissue (primary, human)	Adiponectin release	4 doses between 10^−10 and 10^−7 M
In vitro	Primary ovarian follicles (mouse)	Meiosis II abnormalities	5 doses between 3x10^−9 and 3 × 10^−5 M
In vitro	Primary hypothalamic neurons (rat)	Synaptic density	4 doses between 10^−9 and 10^−6 M
In vitro	Primary granulosa cells (rat)	Progesterone production	4 doses between 10^−7 and 10^−4 M
In vitro	GH3/B6/F10 pituitary tumor cells	Prolactin release	10 doses between 10^−15 and 10^−6 M
In vitro	Umbilical vein endothelial cells	Tube morphology	3 doses between 10^−9 and 10^−6 M
In vitro	Primary ventricular myocytes (rat)	Fractional shortening	5 doses between 10^−12 and 10^−6 M
In vitro	3T3-L1 preadipocytes	Lipid accumulation	4 doses between 10^−9 and 10^−6 M
In vitro	HepG2 hepatocellular cancer cells	Production of mitochondrial superoxide anions	7 doses between 10^−12 and 10^−4 M
In vitro	Fetal testes (human)	Testosterone secretion	4 doses between 10^−12 and 10^−5 M
In vitro	HBL-100 breast epithelial cells	Cell proliferation	7 doses between 10^−11 and 10^−5 M
In vivo	Mouse, developmental exposure	Body weight at weaning	Subcutaneous: 5000, 10 000, 20 000 µg/kg/day
In vivo	Mouse, developmental exposure	Body weight in adulthood	Oral: 50, 500, 5000, 50 000 µg/kg/day
In vivo	Mouse, developmental exposure	Mammary gland morphology	Oral: 0.6, 3, 6, 12, 120, 600, 1200 µg/kg/day
In vivo	Mouse, adult exposure	Mammary tumors, tumor volume, number of metastases	Oral: 0.5, 5, 50, 500 µg/kg/day
In vivo	Mouse, adult exposure	Adipose tissue weight, plasma insulin, blood triglyceride levels, liver gene expression	Oral: 5, 50, 500, 5000 µg/kg/day

Table 2. Examples of adverse endpoints in laboratory animals

Species/sex	Exposure type	Endpoint	Effective doses (µg/kg/day)
Mouse/both	Developmental	Anxiety behaviors	10, 20, 500, 8000
Mouse/female	Developmental	Measures of adiposity	50, 260
Rat/Female	Developmental	Altered brain physiology	50, 20 000
Mouse/Female	Developmental	Response of mammary gland to hormones	0.25
Mouse/Male	Adult	Sperm count	2
Mouse/Male	Development + Adult	Sperm motility	500, 5000
Rat/Male	Developmental	Prostatic PIN lesions after hormone challenge	10
Mouse/Male	Developmental	Immune response to infection	0.3, 3
Mouse/Male	Adult	Immune response to infection	22 800, 45 600
Rat/Female	Developmental	Response of mammary gland to chemical carcinogens	250
Mouse/Female	Developmental	Response of mammary gland to chemical carcinogens	25, 250
Mouse/Female	Pubertal	Chromosomal abnormalities in oocytes	200
Mouse/Female	Developmental	Ovarian cysts	1, 100, 1000
Rat/Male	Developmental	Fetal resorptions in unexposed female mates	100, 200, 400, 800, 1600
Rat/both	Developmental	Body weight on high fat diet	50, 70
Mouse/Female	Pregnancy	Glucose tolerance, fasting insulin levels during pregnancy	10
Mouse/Female	Pregnancy	Glucose and insulin sensitivity months after pregnancy	100
Mouse/Male	Developmental	Blood glucose, fasting insulin levels	10
Mouse/Female	Developmental	Lifetime fecundity	0.025, 25
Mouse/Female	Developmental	Lifetime fertility	25
Sheep/Female	Developmental	Follicular reserve	50
Rat/both	Developmental	Glucose tolerance, insulin tolerance	50
Rat/Male	Adult	Fasting blood glucose, insulin levels	0.005, 0.5, 50, 500
Rhesus monkey/ female	Developmental	Abnormal meiosis	400
Rat/Female	Developmental	Milk yield, milk protein content	0.7, 64
Mouse/Male	Adult	Plasma glucose	5, 50, 500

For additional details about the study designs and citations for these various studies, see Table S3.

Integration of Endpoints Across Levels of Biological Organization

This review and the data presented in supplementary tables, coupled with the 2007 reviews of the BPA literature,^1,6-8 indicate that there are several hundred studies showing biological effects at or below the low dose cut-offs that were established by the Chapel Hill expert panels; the results of more than 50 epidemiology studies indicate that these effects could be occurring at current levels of human exposure.¹⁴ Using the entire body of literature available, we next asked whether similar endpoints are affected across multiple levels of biology and/or in multiple species. Here, we examine five examples where data are available from two or more study types: in vitro, in vivo animals, and human populations. For each example, all published studies (from both pre- and post- Chapel Hill [2007]) were examined, regardless of whether the study in question showed effects of BPA. Details about individual studies are presented in the supplementary tables. Note that we focused on specific endpoints (i.e., insulin regulation), and not all studies in a specific category (i.e., “metabolism”).

These five examples were selected in part based on the expertise available within the BPA grantees consortium. There are a number of additional examples of endpoints that could be examined with an integrative approach. These include conserved effects of BPA on cells and tissues in the male reproductive tract; the response of cultured immune cells and the immune system; measures of adiposity; production of and responses to thyroid hormone; and many others. Due to space constraints, we will not discuss these examples in detail. However, these examples coupled with the five discussed below provide strong evidence for conserved low dose effects.

Example 1: Low dose effects of BPA on liver enzymes

We identified two in vitro studies that have examined the effects of low doses on hepatocytes (Fig. 1A).^73,225 These studies showed effects on liver enzymes and markers of oxidative stress, potential markers of liver function, at or below 10⁻⁹ M; two other in vitro studies found effects of higher doses on similar endpoints.^226,227 In laboratory animals, four studies showed low dose effects on the expression of liver enzymes and hepatic gene expression (Fig. 1B).^152,228-230 This is in contrast to other endpoints such as liver weight and liver histopathology that revealed few low dose effects (Table S3); organ weight is an insensitive measure of pathology and likely reflects only gross toxic effects of an agent. Finally, three human epidemiology studies suggest a relationship between urinary BPA concentrations and changes in liver enzymes (Fig. 1C).^195,199,231 Collectively, these results indicate that BPA alters hepatic function in multiple species across multiple levels of biological organization; additional analyses are needed to determine whether these alterations should be deemed adverse.

Figure 1. Low dose effects of BPA on liver enzymes. (A) In vitro studies examining the effects of BPA on hepatocyte enzyme production and production of reactive oxygen species (ROS). (B) Four in vivo laboratory rodent studies examined the effects of low dose BPA exposures on production of liver enzymes. (C) Summary of epidemiology studies examining effects of BPA on liver enzymes. Open circles indicate applied doses that did not induce significant effects (A and B) or populations that were unaffected (C). Red circles indicate doses that significantly increased the measure of interest (A and B) or populations that were significantly affected (C). Blue circles indicate doses that significantly decreased the measure of interest (A).

Example 2: Low dose effects of BPA on insulin regulation

Three studies have reported that isolated pancreatic islets exposed to BPA in vitro have altered insulin responses to glucose (Fig. 2A).^55,63,232 All three studies reported increases in insulin content following relatively short-term (1–48 h) incubations with low doses of BPA. In rodents, six studies demonstrated that developmental or adult BPA exposures can affect fasting plasma insulin levels as well as insulin levels in response to feeding or glucose challenges (Fig. 2B).^{151,152,168,233-235} In all these studies, plasma insulin levels were increased following BPA exposures, consistent with what was observed in vitro. Finally, as mentioned above, eight epidemiology studies examined relationships between urinary BPA concentrations and type 2 diabetes, a condition that involves abnormal insulin response (Fig. 2C).^{195,199,201,202,204,209,236-238} These studies overwhelmingly report increased odds of diabetes and/or insulin resistance associated with increased BPA exposure metrics. Collectively, these results indicate that low doses of BPA can adversely affect metabolic pathways that are revealed in vitro in cultured pancreatic islets and in vivo in rodents, and suggest associations between BPA exposures and type 2 diabetes in some human populations, an obvious adverse endpoint.

Figure 2. Low dose effects of BPA on insulin signaling. (A) Insulin responses in cultured pancreatic islets. (B) Low dose effects of BPA on insulin levels in lab rodents. (C) Summary of epidemiology studies examining relationships between type II diabetes and urinary concentrations of BPA. Open circles indicate applied doses that did not induce significant effects (A and B) or populations that were unaffected (C). Red circles indicate doses that significantly increased the measure of interest (A and B) or populations that were significantly affected (C). Blue circles indicate doses that significantly decreased the measure of interest (B).

Example 3: Low dose effects of developmental BPA exposures on behavior

Several dozen studies are available examining the effects of low doses of BPA on behaviors in rodents. We focused on studies examining developmental exposures, although the actual exposure periods vary in important ways (Table S3). Examining only those studies that included two specific behavioral tests (the open field test, an assessment for anxiety, and the Morris water maze, a test for spatial learning and memory) reveals the complexity of these behavioral tests which must take into account the sexually dimorphic nature of these behaviors (Fig. 3A and B).^{169,171,174,239-247} Considering factors such as age, sex, and hormonal status, numerous studies report similar effects in neurobehavioral assessments examining low doses of BPA. Furthermore, previous reports were convincing enough in 2008 for the NTP to conclude that based on experimental data, there was “some concern” that low doses of BPA could affect development of the brain and behaviors in humans.²⁴⁸ Finally, as noted above, eight epidemiology studies have examined relationships between BPA exposure levels—most often measured during gestation—and altered behaviors in children (Fig. 3C).^186-194 Consistent with a number of rodent studies, several of these studies suggest that low doses of BPA affect behavioral endpoints in children.

Figure 3. Low dose effects of BPA on select neurobehaviors. (A) Effects of BPA on anxiety behaviors in the open field test. (B) Effects of BPA on spatial learning and memory in rodents in the Morris water maze. (C) Epidemiology studies examining relationships between maternal urinary BPA concentrations and neurodevelopmental parameters in children. Open circles indicate applied doses that did not induce significant effects (A and B) or populations that were unaffected (C). Red circles indicate doses that significantly increased the measure of interest (A) or populations that were significantly affected (C). Blue circles indicate doses that significantly decreased the measure of interest (A and B).

Example 4: Low dose effects of BPA on the response of the mammary gland to subsequent environmental challenges

Three studies have examined the response of cultured mammary cancer cells to chemotherapeutic agents in the presence of low doses of BPA (Fig. 4A).^51-53 These studies indicate that BPA alters the response of mammary epithelial cells to environmental challenges. In rodents, five studies have shown that BPA induces preneoplastic or neoplastic lesions,^{132,134,157,163,249} and four additional studies have demonstrated that low dose BPA exposures increase the response of the mammary gland to carcinogenic challenges administered later in life (Fig. 4B).^137-139,250 The consistency of these results across laboratories is indicative of a highly reproducible low dose effect and provides strong evidence for adverse effects of BPA on this organ.

Figure 4. Low dose BPA alters the response of the mammary gland to environmental challenges. (A) In vitro studies reveal that BPA exposures confer chemotherapeutic resistance in mammary cells. (B) Low doses of BPA induce preneoplastic and neoplastic lesions in the rodent mammary gland, and alter adult challenges with carcinogenic agents. Open circles indicate applied doses that did not induce significant effects (A and B). Blue circles indicate doses that significantly decreased the measure of interest (A). Red circles indicate doses that significantly increased the measure of interest (B). Gray circles indicate doses that induced the adverse effect noted, but statistics were not performed.

Example 5: Low dose effects of BPA on the ovary

Oogenesis begins in the fetal ovary with the entry of germ cells into meiosis and the subsequent formation of follicles (an oocyte surrounded by somatic cells). In laboratory mammals, 11 studies have shown effects of low dose BPA exposures during ovarian development on three types of endpoints: (1) BPA exposures at the onset of meiosis affect meiotic prophase, increasing the likelihood that the exposed adult will ovulate chromosomally abnormal eggs,^251,252 (2) exposures occurring during follicle formation result in an increase in multi-oocyte follicles and a reduction in the total pool of oocytes,^251,253-256 and (3) exposures throughout gestation affect the adult ovary, leading to polycystic changes (Fig. 5A);^{118,119,122,257,258} these effects have been observed in numerous mammalian species including mouse, rat, sheep, and monkey as well as in nematodes²⁵⁹ and in human fetal ovaries exposed in vitro.²⁶⁰ Taken together, these studies suggest that exposure during ovary development can adversely affect several different aspects of early oogenesis.

Figure 5. Low dose effects of BPA on the ovary. (A) Effects of developmental exposures on ovarian endpoints, including adverse effects that manifest in adulthood. (B) Summary of in vitro studies showing effects of BPA on hormone production and enzyme activity in ovarian cells. (C) Effects of adult exposures on ovarian endpoints in rodents. (D) Summary of epidemiology studies examining relationships between BPA exposure metrics and ovarian response in women undergoing IVF. Open circles indicate applied doses that did not induce significant effects (A–C) or populations that were unaffected (D). Red circles indicate doses that significantly increased the measure of interest (A–C) or populations that were significantly affected (D). Blue circles indicate doses that significantly decreased the measure of interest (B). Gray circles indicate doses that induced the adverse effect noted, but statistics were not performed.

In the sexually mature female, groups of ovarian follicles initiate a complex growth process during which the oocyte undergoes growth and maturation and the theca and granulosa cells mature and provide an appropriate endocrine and paracrine environment. Each cell type plays an integral part in ovarian health and fertility. In vitro studies have shown effects of BPA on hormone production and enzyme expression/activity in granulosa and theca cells, but these effects were generally observed at doses above 10⁻⁸ M; lower doses were not typically examined (Fig. 5B).^{36,37,43,260-264} As mentioned, BPA alters zebrafish ovarian histopathology including the incidence of atretic follicles.¹¹² Four mammalian studies have also examined the effects of adult BPA exposure on meiotic maturation of the oocyte with variable results (Fig. 5C).^{124,125,263,265} However, other factors like dietary phytoestrogens are thought to influence the response to BPA at this stage.²⁶⁶ A number of studies of women undergoing assisted reproduction are consistent with the conclusion that BPA exposure is associated with poorer oocyte and embryo quality (Fig. 5D).^218-222,267 Thus, collectively, these studies suggest that low dose BPA exposure exerts adverse effects on growing follicles in the adult ovary.

Mechanisms of Low Dose Effects

Hormones are known to have effects at extremely low doses due to high affinity receptor–ligand interactions, the non-linear relationship between hormone concentration and number of bound receptors, and the non-linear relationship between the number of bound receptors and biological effects.^5,114,268 EDCs that interact with hormone receptors have been shown to follow these same “rules.” Low dose effects should also be considered in the context of circulating endogenous hormones (as well as phytoestrogens from the diet); thus, some low dose effects of EDCs may involve additive or synergistic effects.^269,270 Finally, it should be noted that many low dose effects can be attributed to the period of exposure. Exposures to hormones and EDCs during critical periods can irreversibly affect differentiation and organogenesis²⁷¹ and some developmental exposures have been shown to alter the epigenome, suggesting altered epigenetics may be one mechanism of BPA action.^272-274

Current State of Evidence

Several recent reviews have concluded that there are reproducible low dose effects for a number of EDCs, including BPA.^5,11,114,275 Additionally, in 2008, the NTP determined there was “some concern” about the effects of human exposure levels on prostate development, brain development, and some behaviors,²⁴⁸ suggesting that this group also concluded that low dose effects were reproducible for these endpoints. Thus, the question of whether low dose effects “exist” is no longer the correct one to ask; clearly they have been reported consistently for a number of endpoints, and the list of these endpoints continues to grow. In our analysis, the low dose cut-off for mammalian animal studies was set at 50 mg/kg/day, a dose identified in previous toxicology studies, consistent with the NTP’s definition of a low dose;⁴ this dose was also used by the Chapel Hill expert panel.⁶ Because of the number of studies reporting effects below this cut-off, these previous reviews and our current analysis challenge the setting of the LOAEL at 50 mg/kg/day as well as the “safe” reference dose of 50 µg/kg/day.

Are these low dose effects adverse? There are two widely accepted definitions for “adverse effect”: (1) the US EPA’s definition is “a biochemical change, functional impairment, or pathologic lesion that affects the performance of the whole organism, or reduces an organism's ability to respond to an additional environmental challenge.”²⁷⁶ (2) the OECD definition is “a change in morphology, physiology, growth, development, or lifespan of an organism, which results in impairment of functional capacity or impairment of capacity to compensate for additional stress or increase in susceptibility to the harmful effects of other environmental influences.”²⁷⁷ Based on these definitions, we consider many of the endpoints affected by BPA to be adverse (Table 2). In sum, it is clear that consistent, reproducible, low dose effects are observed in several mammalian species and fall under the definition of adverse effects.

As for the issue of reproducibility, there are historical examples of specific endpoints shown to be sensitive to low doses of BPA by some scientists but not reproduced by other groups (reviewed in ref. 4). Subsequent analyses showed that many of the studies that were unable to detect low dose effects had problems including lack of concurrent positive controls, lack of a response in the positive control, or the requirement for inappropriately high doses of the positive control, as well as other experimental issues. Weight-of-evidence analyses using principles of endocrinology indicate that low dose effects of BPA are highly reproducible and consistent between many different studies.^5,275 The fact that there are any studies showing low dose effects of BPA—and there are several hundred such studies including more than 50 human studies—indicates that the current safety guidelines for BPA are not protective of human health.

Based on the available evidence, we are confident of the following:

(1) Consistent, reproducible low dose effects have been demonstrated for BPA in cell lines, primary cells and tissues, laboratory animals, and in some human populations.

(2) Many of these low dose effects should be considered adverse.

(3) Doses that reliably produce statistically significant effects in animals are 1–4 magnitudes of order lower than the current LOAEL of 50 mg/kg/day.

(4) EDCs, including BPA, often pose a greater threat when exposure occurs during early development, organogenesis, or during critical postnatal periods during which tissues are differentiating.

(5) Low dose effects have been demonstrated in rodents following developmental exposures to BPA, including effects on the male and female reproductive tract, brain development and behaviors, metabolism and growth endpoints, and development of preneoplastic and neoplastic lesions of the mammary gland and the mammary gland’s response to carcinogens.

(6) Low dose effects have been demonstrated in rodents following adult exposure to BPA on the brain and behaviors, metabolic parameters, and function of the male reproductive organs.

(7) In wildlife, consistent and reproducible low dose effects have been observed on a number of endpoints, including induction of vitellogenin expression in males of various fish species and feminization of the adult male reproductive tract, and other measures related to reproductive success.

(8) In cultured cells (both primary and immortalized cell lines) and tissue explants, levels in the range of 10⁻¹⁴–10⁻⁹ M induce effects on a number of endpoints. These endpoints provide important molecular and mechanistic understanding for some low dose effects observed in vivo.

(9) The epidemiology literature shows consistent relationships between current typical human exposures to BPA and a number of different health outcomes.

Future Directions for Low Dose Research

The major goal of this review was to address whether low dose effects exist for BPA, whether these effects are consistent and reproducible, and whether the endpoints affected represent adverse outcomes. Based on the hundreds of studies published to date, we conclude that there is sufficient evidence for low dose effects of BPA. In this review, we presented evidence for consistent, reproducible, adverse effects on numerous endpoints, including ovarian health (including histopathology and polycystic ovarian syndrome), some behaviors (anxiety, impaired learning), the mammary gland’s response to carcinogens, abnormal glucose/insulin homeostasis, and diabetes. Other adverse effects have been observed in endpoints such as altered immune responses to allergens, decreased fertility, and fecundity, altered brain function, and altered response of the prostate to hormone challenges; we were unable to elaborate on all of these examples. There do remain scientific data gaps and data needs that are worth pursuing. These include:

•The use of a consistent definition of “low dose” and “low dose effects” in the EDC literature.

•An acknowledgment that “low dose” is distinct from NMDRCs.

•Improved data on the sources of BPA exposure.

•Incorporation of low doses and additional sensitive endpoints in guideline studies.

•Improved longitudinal/life-course evaluation of animals and humans developmentally exposed to BPA.

•Better understanding of specific mechanisms of BPA action, including the determination of whether multiple mechanisms contribute to the effects of BPA on animals.

•Evaluation of additive and synergistic effects of BPA with other EDCs including phytoestrogens.

•Study of interactions between developmental BPA exposures and lifestyle or environmental challenges later life.

•To better define critical windows of vulnerability for various endpoints shown to be sensitive to low doses of BPA.

Abbreviations:
BPA	=	bisphenol A
CNS	=	central nervous system
EDC	=	endocrine disrupting chemical
EPA	=	US Environmental Protection Agency
ER	=	estrogen receptor
IVF	=	in vitro fertilization
LOAEL	=	lowest observed adverse effect level
NIEHS	=	National Institute of Environmental Health Sciences
NMDRC	=	non-monotonic dose response curve
NTP	=	US National Toxicology Program
OECD	=	Organization for Economic Cooperation and Development
T4	=	thyroxine
TSH	=	thyroid stimulating hormone

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

This manuscript was prepared as a product of the NIEHS BPA consortium.

Supplemental Materials

Supplemental materials may be found here: www.landesbioscience.com/journals/endo_dis/article/26490

10.4161/endo.26490