Bisphenol S and F: A Systematic Review and Comparison of the Hormonal Activity of Bisphenol A Substitutes

1There is increasing evidence that bisphenol A (BPA)—used in plastics, receipts, food packaging, and other products—might be harmful to human health due to its actions as an endocrine-disrupting chemical (EDC). Scientists, regulators, and the general public have raised concerns about the use of BPA, especially because of its ubiquitous nature and potential for continuous exposure. This has prompted industry to seek alternative chemicals. As manufacturers have begun to remove BPA from their products as a result of consumer concern, there has been a gradual shift to using bisphenol analogs. For the purpose of our review, we chose to evaluate two of these analogs—bisphenol S (BPS) and bisphenol F (BPF)—because of their widespread consumer and commercial use. BPS is used for a variety of industrial applications, for example, as a wash fastening agent in cleaning products, an electroplating solvent, and a constituent of phenolic resin. BPS is also used as a developer in thermal paper, including products marketed as “BPA-free paper”. BPF is used to make epoxy resins and coatings, especially for systems needing increased thickness and durability (i.e., high-solid/high-build systems), such as tank and pipe linings, industrial floors, road and bridge deck toppings, structural adhesives, grouts, coatings, and electrical varnishes. BPF epoxy resins are also used for several consumer products such as lacquers, varnishes, liners, adhesives, plastics, water pipes, dental sealants, and food packaging. BPS and BPF have been detected in many everyday products, such as personal care products (e.g., body wash, hair care products, makeup, lotions, toothpaste), paper products (e.g., currency, flyers, tickets, mailing envelopes, airplane boarding passes), and food (e.g., dairy products, meat and meat products, vegetables, canned foods, cereals). BPS, BPF, and BPA have been detected in indoor dust at the following concentrations: BPS, 0.34 μg/g; BPF, 0.054 μg/g; BPA, 1.33 μg/g. BPS and BPF have also been detected in surface water, sediment, and sewage effluent, generally at lower concentrations than BPA, but in the same order of magnitude. In humans, BPS and BPF have been detected in urine at concentrations and frequencies comparable to BPA. In urine samples from 100 American, nonoccupationally exposed adults, found BPF in 55% of samples at concentrations up to 212 ng/mL, and BPS in 78% of samples at concentrations up to 12.3 ng/mL. BPA was found in 95% of the samples, with concentrations up to 37.7 ng/mL.

2Ideally, substitutes used to replace a chemical of concern would be inert, or at least far less toxic than the original chemical(s). Unfortunately, many chemical replacements are untested before being placed on the market, and in some cases are similar enough to the original chemical to cause concern. For that reason, such chemical analogs should be evaluated before they are used as replacements for toxic chemicals. These chemicals may be just as harmful as the originals—or more so—and have been described as “regrettable substitutions,” as is the case with several perfluorinated chemicals, pesticides, and flame retardants. In the case of BPS and BPF, these chemicals are structural analogs to BPA (Figure 1); thus their effects in physiological systems may be similar. BPA is a known endocrine disruptor based on in vitro and animal laboratory studies, and exposures to environmental levels of BPA have been associated with adverse health outcomes in children and adults in more than 75 human studies. To evaluate the endocrine-disrupting properties of the BPA substitutes BPS and BPF, we conducted a systematic review of the literature using the National Institute of Environmental Health Sciences’ Office of Health Assessment and Translation (OHAT) systematic review protocol. In this analysis we summarize in vivo and in vitro literature and compare the hormonal potency of BPS and BPF to BPA using the in vitro studies.

3Literature search and review. We performed a comprehensive literature search in order to identify studies describing endocrine and other physiological effects of exposure to BPF and BPS. The search included all articles published and indexed for all years to June 2014. Electronic searches were performed in Web of Science (https://webofknowledge.com/) and PubMed (http://www.ncbi.nlm.nih.gov/pubmed) using CAS registry numbers and common names. Our search logic is summarized in Table 1.

4For inclusion, the studies had to be primary literature and assess any in vitro or in vivo physiological effects of BPS or BPF exposure. Two independent reviewers (J.R.R. and A.L.B.) screened all titles and abstracts for relevancy, using Distiller SR® software (Evidence Partners), and resolved any conflicts or discrepancies. Data from the studies were extracted, and were cross-checked by the two reviewers. When needed, data were extracted from figures or graphs using Universal Desktop Ruler® software (version 3.6; AVPSoft), with measurements taken in triplicate by a single reviewer.

5Study quality for in vivo studies was assessed using a protocol developed by OHAT. Briefly, risk of bias (RoB) in experimental methodology was assessed by answering 14 questions. The RoB questions covered biases in subject selection, protocol performance, attrition/exclusion of subjects, detection of outcomes, selective reporting of outcomes, and statistical methodology. Questions were rated as “definitely low RoB,” “probably low RoB,” “probably high RoB,” or “definitely high RoB” depending on standardized responses. The individual RoB questions are provided in Figure 2. Next, “key” study quality questions, identified a priori, were used to determine the initial quality of each study, then ratings of the remaining questions were used to determine the overall study quality: “low,” “moderate,” or “high.” If any study received a “low” rating, it was removed from analysis. This protocol has been described in detail elsewhere.

6As specified in the OHAT protocol, in vitro studies were not assessed for quality, but were used to support specific in vivo end points. For example, estrogen receptor (ER) binding or activation studies support the biological plausibility of increased uterine growth, an in vivo estrogenic response. Where there were at least three in vitro studies, the strength of support was rated on the following factors: relevance of biological process or pathway to human disease, consistency across model systems (where there were more than two systems), physiological relevance of the dose concentration, potency (magnitude of response compared with positive control), dose response (monotonic or nonmonotonic), and publication bias. These factors were integrated for a final rating of “weak,” “moderate,” or “strong” in vitro support of the biological plausibility of in vivo observations, but they were not used to exclude studies. In vitro observations that had fewer than three studies per end point, or did not relate to any observed in vivo end points, are described in the text.

7Our search identified 1,370 studies; of these, 32 studies (25 in vitro only and 7 in vivo) were identified as relevant for inclusion. Figure 2 shows the study quality ratings for the in vivo studies. All studies were rated moderate quality or better; therefore, no in vivo studies were removed because of low quality.

8BPS. The literature reporting the physiological effects of BPS exposure consisted of 4 in vivo studies and 18 in vitro studies. The in vivo studies are presented in Table 2. BPS exposure caused acute toxicity in Daphnia magna. found that postnatal BPS exposure in rats caused an induction of uterine growth, a marker of estrogen exposure, at the lowest and highest doses. The authors also found that BPS bound to the nuclear ER at 0.0055% relative binding affinity. studied BPS exposure in zebrafish (Danio rerio) and found decreases in gonad weight, alterations in plasma estrogen and testosterone, and disrupted reproduction (i.e., decreased egg production and hatchability, increased time to hatch, increased embryo malformations). Another study in zebrafish showed that BPS exposure increased female to male sex ratio; decreased body length; altered testosterone, estradiol, and vitellogenin concentrations; and led to reproductive disruption (i.e., decreased egg production, increased time to hatch, decreased sperm count).

9In vitro data from 12 studies assessing estrogenicity provided strong evidence supporting the estrogenic responses observed in in vivo studies (Table 3), based on relevance of the end point to human health [e.g., interaction with human ERα and G-protein coupled receptor 30 (GPR30)], consistent response across eight cell lines, and physiologically relevant concentrations assessed (micromolar range). Several of these studies showed that BPS had weaker estrogenic potency than estradiol (E2) when assayed in nuclear receptor models. However, two studies showed that BPS had equivalent or greater estrogenic potency to E2 when assayed in membrane receptor models; BPS induced membrane receptor–mediated pathways typically up-regulated by E2. Four studies showed that BPS bound to the ER in competitive binding assays. There was also one study showing androgenic activity of BPS and one study showing antiandrogenic activity. In addition, in other in vitro experiments BPS exposure induced caspase 8 production, which indicates that BPS may alter cellular apoptotic and survival signaling. BPS also had effects on hepatic cells; it bound to serum albumins, and it caused DNA damage.

10BPF. Of the five in vivo studies, four showed that BPF was estrogenic, androgenic, and thyroidogenic (Table 2). Nineteen in vitro studies showed estrogenic, androgenic, and other physiological/biochemical effects (Table 3). BPF was acutely toxic in Daphnia magna. Two studies showed that BPF exposure induced uterine growth in rats, indicating estrogenic activity. There were also two studies that showed evidence of androgenic activity: One study indicated that BPF increased the weight of the testes, and the other showed a cumulative effect of BPF when co-administered with testosterone propionate that increased Cowper’s gland weight. The cumulative effect indicates that BPF may augment other androgens, if indeed it acts synergistically. BPF exposure also increased thyroid weight and altered thyroid hormone concentrations, as well as caused changes to hematological parameters and enzyme expression.

11As shown in Table 3, in vitro data from 12 studies provided strong evidence that BPF had estrogenic activity, supporting in vivo observations. This rating was based on relevance to human health (MCF-7 cell and human ER), consistency across five cell models, and the use of relevant concentrations (micromolar range). One study showed that BPF was not estrogenic in a yeast two-hybrid assay. One study indicated that BPF was antiestrogenic. Moderate evidence from 6 studies showed that BPF was antiandrogenic based on relevance to human health [i.e., human androgen receptor (AR)], consistency across four cell models, and potency (i.e., within 100 orders of magnitude of positive control). BPF also showed other in vitro effects such as cytotoxicity, cellular dysfunction, DNA damage, and chromosomal aberrations, and decreased adiponectin production and secretion in vitro.

12Potency of BPS and BPF compared with BPA. BPS and BPF are already being used as alternatives for BPA; thus, it is important to understand whether these substitutes possess endocrine-disruptive/active properties similar to those of BPA. Seventeen studies tested BPS and/or BPF along with BPA in the same assays, allowing the potencies and mechanisms of action to be directly compared. Table 4 presents these results, comparing the hormonal potencies of BPF and/or BPS to BPA. The average estrogenic potency (mean ± SD) for BPF compared with BPA was 1.07 ± 1.20, with a range of 0.10–4.83. The average estrogenic potency for BPS compared with BPA was 0.32 ± 0.28, with a range of 0.01–0.90. These results indicate that the potencies of BPS and BPF are in the same order of magnitude as the potency of BPA, and BPF may be just as potent (or more potent) than BPA. Further, BPS and BPF have potencies in the same order of magnitude as BPA in regard to androgenic, antiandrogenic, antiestrogenic, and aryl hydrocarbon activity and inhibitory hormonal signaling in adipocytes (Table 4).

13used several assays to assess steroidogenic activity, as well as teratogenicity, genotoxicity, carcinogenicity, and metabolic effects. Similar to the present evaluation, they found that BPS and BPF had estrogen receptor binding, estrogenic activity, and antiandrogenic activity similar to those of BPA, with BPS being the least potent. However, BPS and BPF exhibited the greatest steroidogenic (i.e., progesterone) activity, increasing levels of 17α-hydroxyprogesterone and progesterone levels, whereas BPA did not. Although the authors did not examine the mechanism of action of progesterone up-regulation, previous work suggested a direct inhibition of the CYP17 (cytochrome P450 17A1) lyase reaction, independent of ER action. Thus, BPA analogs may have additional disruptive effects that have not been detected with BPA.

14Although relatively few studies have examined the hormonal actions of BPS and BPF (especially in vivo), the in vitro literature indicates that BPS and BPF have actions and potencies similar to those of BPA and supports the biological plausibility of their hormonal activity in vivo. This is not surprising because BPF and BPS are structural analogs of BPA and thus mechanisms of action would be expected to be similar. For example, BPF showed cumulative, possibly synergistic, actions in vivo when co-administered with an androgen, and BPA has also been shown to have these types of effects when combined with other hormones or xenoestrogens. Particularly interesting is the fact that BPS seems to have actions on nongenomic signaling similar to those of BPA. BPA is sometimes called a “weak” estrogen because of its relatively weak binding/activation of the nuclear receptors compared with E2, although this is not always the case (Table 3;;; ). However, when the nongenomic estrogenic activity of BPA was measured, it was comparable, if not more potent, than E2. This potent, nongenomic estrogenic activity of BPA has been described in several experimental models. The potency of BPS in a nongenomic signaling assay was similar to that of BPA. In femtomolar to picomolar concentrations, BPS induced membrane ERα-mediated pathways and actions: MAPK (mitogen-activated protein kinase) signaling, cell proliferation, and activation of caspase 8. These rapid, nongenomic pathways are important for optimal cell function, mediating proliferation and apoptosis, as well as other actions such as pancreatic cell function and estrogen-mediated brain function and behavior.

15BPS and BPF had potencies in the same order of magnitude as BPA. The issue of potency is complicated because of the fact that lowest observed effect levels depend on end point, receptor type, pathway, tissues, windows of exposure, and so on. In general, BPS was slightly less potent than BPA. The average BPF potency was similar to BPA, with a fairly wide range of potencies. However, the implications of these differences are not clear. In regard to potency, it is not known whether a compound that is, for example, half as potent as BPA in vitro would have half the effect in vivo, especially because very little is known about the exposure and metabolism of BPS and BPF. Further, even if potencies of BPS and BPF are slightly less than that of BPA, it is unclear if these compounds are safer; many scientists have advocated a “no-threshold” approach to endocrine disruption because thresholds may change during development or may be very difficult to assess.

16The metabolism and biological fate of BPS and BPF have not been well studied, but in vitro and in vivo experiments indicated that BPF metabolism and distribution are similar to those of BPA. In vitro, BPA was metabolized by human and rat hepatic cells to many different metabolites, including non-bioactive sulfate and glucuronide conjugates. In vivo, BPF administered to pregnant rats via gavage resulted in the excretion of BPF and several metabolites in the urine, including the nonactive sulfate-conjugated BPF. Active BPF was also distributed to many tissues, including the uterus, placenta, amniotic fluid, and fetuses. The ratio of the active parent compound to the metabolites/conjugates was similar to that of BPA. The primary route of excretion for BPF appeared to be through the sulfatase conjugate, rather than the glucuronide conjugate (as with BPA). suggested that this may be due to the fact that BPF glucuronide may be more easily deconjugated to its bioactive state and reabsorbed in large quantities, which also appears to occur with BPA. No studies have assessed the metabolism of BPS or the bioactivity of the metabolites. Studies determining the metabolism of BPS and the bioactivity of metabolites from BPF and BPS are warranted.

17The body of literature on the in vivo effects of BPS and BPF is scant, but it points to these chemicals as endocrine disruptors and reproductive toxicants. BPS induced uterine growth in rodents (indicative of estrogenic action) and disrupted reproduction in fish, and BPF also had uterotropic (estrogenic) effects in female rodents and gonadotropic (androgenic) effects in male rodents. Although most of the in vitro data support estrogenic, and to some extent, antiandrogenic, actions of BPS and BPF (Table 3), one in vitro study showed that BPS has androgenic activity similar to BPA. Thus, the in vitro data support the in vivo observations of hormonal and endocrine disruptive activity of these compounds.

18Concern over the endocrine-disruptive effects of BPA has resulted in hundreds of laboratory studies, including in vitro and in vivo studies, identifying estrogenic and other effects. Although some regulators have rejected this body of literature because of a lack of standardized protocols, reviews of these studies have indicated strong methodologies and stringent laboratory practices, often of higher quality than studies employing Good Laboratory Practices. Many in vivo BPA studies have demonstrated adverse outcomes at “low” (i.e., environmentally or physiologically relevant) doses. Many studies also report that BPA has a nonlinear, or nonmonotonic, dose–response curve. Nonmonotonic dose responses are indicative of an endocrine-mediated response and are consistent with natural hormone responses. Further, nearly 100 human studies described the relationship between BPA and several endocrine-related health impacts on reproduction, neurodevelopment, thyroid function, and metabolic health. Although epidemiological studies are less controlled than laboratory animal experiments, making it difficult to show causation, they are important indicators of potential health effects. Further, although BPA is quickly metabolized and excreted from the body [with a half life of about 6 hr], the fact that it is found in almost all humans sampled at any one time suggests the ubiquitous and constant nature of BPA exposure, which is disconcerting in light of the animal and human evidence of health effects. Many researchers have raised concern over this overwhelming evidence and have called for stricter regulation of BPA. Although this concern has prompted BPA to be phased out of certain products, the structural analog replacements may not be any safer.

19Because BPS and BPF appear to have metabolism, potencies, and mechanisms of action in vitro similar to BPA, including hormonal actions beyond that of BPA, they may pose similar potential health hazards as BPA. Therefore, when evaluating the safety of compounds for consumer use, it may be prudent to consider entire classes instead of individual compounds. In addition, as other researchers have suggested, future research efforts should focus on designing chemical substitutes that do not have biological or hormonal activity similar to those of BPA. Further, this review demonstrates that systematic reviews may be useful in the process of conducting safety evaluations of chemical classes. The use of the bisphenol class of compounds as replacements for BPA in consumer products with high human contact should be implemented with caution.