Registration Dossier

Toxicological information

Specific investigations: other studies

Currently viewing:

Administrative data

Link to relevant study record(s)

Description of key information

TBBPA does not adversely mediate androgen homeostasis and does not modulate the androgen pathway in exposed wildlife.
TTBPA does not bind and activate the estrogen receptor and does not bind to the ER in vitro or significantly modulate the estrogen pathway in exposed wildlife.
TBBPA causes decreased serum levels of T4 without concomitant changes in T3 or TSH in repeated-dose toxicity assays as well as developmental, prepubertal and reproduction toxicity studies. Multiple lines of evidence in vivo demonstrate that TBBPA can modulate the thyroid pathway in wildlife; however, the observed responses are near the maximal aqueous environmental concentration of 4.87 µg/L.

Additional information

Thorough literature assessments were carried out in which the potential effects of TBBPA on androgen-mediated endocrine disruption, endocrine-mediated toxicities associated with the estrogen pathway and endocrine-mediated toxicities associated with the thyroid pathway were evaluated from both a human health and wildlife perspective. These assessments are attached in full.

In evaluating information for endocrine-mediated responses, the widely accepted definition “An endocrine disruptor is an exogenous substance or mixture that alters function(s) of the endocrine system and consequently causes adverse effects in an intact organism, or its progeny” (WHO/IPCS 2002) was utilized; the definition of adversity was based on that provided by WHO/IPCS 2004. Endocrine-mediated toxicities from repeated-dose toxicity studies as well as reproductive and developmental studies were considered. In concert with these in vivo studies, evaluations of endocrine-mediated responses included assessment of in vitro data were considered per the following:

1. Postulated downstream effects of the biochemical changes observed in vitro require confirmation in appropriately performed and documented animals studies or in well-performed human epidemiology investigations.

2. Concentrations resulting in biochemical effects in in vitro systems need to be related to human exposures to the chemical assessed and its toxicokinetics, i.e., blood and tissue levels (Judson et al. 2011).

Summary and Conclusions (Androgen Human Health)

There are a number of in vivo studies (toxicity and developmental and reproductive studies of ≤14 weeks’ duration) where animals were exposed to TBBPA without any indications of adverse effects mediated through the androgen pathway (Table included in attachment; Van der Ven et al., 2008; NTP, 2014; Cope et al., 2014). For example, in the toxicity studies there were no changes observed in androgen sensitive organ (e.g., prostrate, testes, epididymis) weights or histopathology. Also, in the developmental and reproductive studies conducted (Van der Ven et al., 2008; Cope et al., 2014) there were no changes in endpoints sensitive to either androgen agonists or antagonists (e.g., changes in male and female reproductive organ weights and histopathology, measures of reproductive performance (e.g., mating indices, fertility indices) or endpoints such as balano-preputial separation (F1), nipple retention, etc. that provide lines of evidence of androgen activity. In a study conducted by Saegusa et al. (2009) rats were exposed to TBBPA in the diet at concentrations up to 10 000 ppm (estimated to be 2100 mg/kg) from gestation day 10 to postnatal day 20. No changes in number of implantation sites, number of live offspring; no changes in reproductive outcome parameters in F1 animals, no changes in androgen sensitive organ weights or histopathology changes were reported. These data collectively demonstrate that TBBPA does not show androgenic activity in either repeated dose toxicity studies or following exposure during early development and across generations in a guideline reproductive study. 

 

There are a number of in vitro studies in the literature that evaluate the ability of TBBPA to bind and activate the AR (Table included in the attachment). In a number of AR reporter gene assays, TBBPA showed none to weak AR agonist or antagonist activity (Hamers et al., 2001; Molina-Molina et al., 2013; Kitamura et al., 2005; Harju et al., 2007; Christen et al., 2010). This was also the case in yeast assays with AR (Li et al., 2013; Huang et al., 2013). In a more recent paper in yeast transformed with human AR, TBBPA was reported to have AR antagonist activity with an IC50 of 982 nM (Roelofs et al., 2015). A number of assays were conducted to evaluate the ability of TBBPA to interfere with steroidogenesis by measuring CYP19 (aromatase activity) and steroid hormones levels. In human placental microsomes or H295R cells, TBBPA did not show any activity as reported by Roelof et al. (2013). This was also true in the studies reported by Canton et al. (2005) and Song et al. (2008) where exposures of H295R human adrenocortical carcinoma cells to TBBPA did not change aromatase activity or impact levels of hormones measured. In a more recent report using murine Ma-10 Leydig cells, TBBPA increased the secretion of testosterone and upregulated 5aRed1 gene during a 48 hour incubation (Roelof et al., 2015). This increase in vitro does not appear to be associated with any androgen mediated effects in animal studies, nor or findings consistent with other in vitro studies.

 

Also, within the US Environmental Protection Agency (EPA) high throughput screening (HTS) initiative (ToxCast/Tox21) androgen receptor bioactivity of TBBPA was evaluated in 9 selected assays (Table included in the attachment). The activities reported in these ToxCast/Tox21 assays were either inactive (6 assays) with AR bioactivity reported (as of February 9, 2016) for 3 assays. For AR binding assay (NVS_NR_rAR) the AC50 for TBBPA was >10 times that for testosterone proprionate (TP). In an agonist transcriptional assay (OT_AR_ARSRC1_0960) the AC50 for TBBPA was 43 µM versus 0.0074 µM for TP with the AC50 for androgen antagonist activity (Tox21_AR_LUC_MDAKB2_ Antagonist) 89 µM compared to 23 µM for hydroxyflutamide. These data support the overall none to weak AR bioactivity reported in the literature studies for TBBPA.

 

In conclusion, multiple lines of evidence demonstrate that TBBPA has none to weak AR bioactivity in vitro, does not inhibit aromatase activity or result in changes in steroidogenesis that results in any lines of evidence in repeated-dose toxicity assays, or developmental and reproduction toxicity studies, that suggest androgen-mediated responses. Taken together, both the in vivo studies in animal models and the in vitro data support the conclusion that TBBPA does not adversely mediate androgen homeostasis.

Summary and Conclusions (Androgen Wildlife)

There are a number of in vivo studies focused on growth and reproduction in invertebrates and fish (as presented in the Table in the attachment). Growth and reproduction are often used as individual and population level indicators of endocrine modulation (Dang et al. 2012, Arcand-Hoy and Benson 1998), but only when combined with molecular, biochemical and physiological endocrine biomarkers. While effects on growth and reproduction were noted in these studies, no androgen specific endpoints were evaluated. Of particular interest, Kuiper et al. (2007) conducted an evaluation of zebrafish reproduction following aqueous exposure to TBBPA. Fecundity and fertilization were reported to be decreased following TBBPA, with the variability of data within and across exposure levels confounding the interpretation of these findings. Histological changes were also observed in TBBPA exposed female fish, though a high incidence of findings in the controls confounds the analysis. At the highest exposure level, sexual development was skewed towards females.

 

Huang et al. (2013) evaluated ARα and ARβ mRNA, as well as liver and testis histology following TBBPA exposure in the mosquito fish (i.e. Gambusia sp.). In the livers of adult fish, no changes in ARα and ARβ were noted. In adult testes, mRNA for these receptors were found to be both upregulated and downregulated. In juveniles, ARα was found to be upregulated. Even though the ARα and ARβ mRNA expression levels changed relative to controls, the reported fold induction was small (<2). In addition, no histological changes were noted in the liver or testes of TBBPA exposed fish. The Huang et al. (2013) study does not follow a standard, regulatory accepted protocol (e.g. OECD 210, Fish Early life Stage Toxicity Study), therefore making regulatory interpretation difficult. Gambusia sp. are not commonly used as reproductive or developmental models as they are live-bearers as opposed to egg-bearers. Fish commonly used in reproduction and development studies (i.e. fathead minnows, zebrafish, medaka) are egg-bearers. The reproductive physiologies of the live-bearers are different from egg-bearers, making interpretation of findings difficult. Due to the complexities of working with live-bearers, the testing laboratory resorted to artificial fertilization methods to obtain juvenile fish. Gambusia sp. were captured from the field and transported to the lab for use after 3 wk acclimation period. Response patterns in field collected fish may be different than traditional laboratory reared fish. Given the complexities of described, the authors should have utilized a positive control in this study to be able to fully interpret the findings. For these reasons, the reliability of this study is low. 

 

Two studies in birds focused on gonadal and reproductive parameters (e.g. plasma testosterone, histology) that could provide specific information on androgen modulation. No effects were reported either in changes in plasma testosterone levels, gonadal-somatic index, testis weight asymmetry, ovotestis formation or seminiferous tubule diameter following a single 15 µg/g dose egg injection (Berg et al. 2001, Halldin et al. 2001). 

 

In vitro assay data are often used to evaluate the potential for chemicals to modulate androgen receptors (AR) in wildlife, even if the assays use human receptors. The yeast recombinant and AR-CALUX assays (Table included in the attachment) are commonly used illustrate that TBBPA does not strongly interact with the AR (Harju et al. 2007, Hamers et al. 2016, Li et al. 2010, Huang et al. 2013). In three of the studies, no interaction of TBBPA with the AR was noted up to 20 µM TBBPA. In only one study was an interaction reported, with a TBBPA EC50 of 110 µM (Huang et al. 2013). This TBBPA concentration was 5-10x the maximum concentration used in the other studies and the dose-response curves were bi-phasic. Hamers et al. (2006) classified a chemical as non-potent if the observed response was < 20% of control at 10 µM. In addition, Huang et al. (2013) study utilized a transfected human AR, which may have different binding characteristics compared to fish AR.

 

Multiple lines of evidence demonstrate that TBBPA does not have AR bioactivity or modulate the androgen pathway in exposed wildlife. In avian, fish and invertebrate studies, there is no conclusive evidence that suggests TBBPA causes an androgen-mediated response. While one study using a non-traditional testing species, Gambusia sp., reported changes in the expression of ARα and ARβ mRNA following TBBPA exposure, no histological changes were noted in the liver or testis. The lack of histological findings indicates that the mRNA measures did not correspond to an adverse consequence in exposed fish. The reliability of this study is low due to concerns regarding the species used, its physiology, fish being collected from a field environment prior to laboratory use, and the lack of adherence to an accepted testing protocol. Given no positive control was used in this atypical testing species, interpretation of the findings is difficult.

Summary and Conclusions (Estrogen Human Health)

There are a number of in vivo studies (toxicity and developmental and reproductive studies of ≤14 weeks’ duration) where animals were exposed to TBBPA without any indications of adverse effects mediated through the estrogen pathway (Table included in the attachment; Van der Ven et al., 2008; NTP, 2014; Cope et al., 2014; Ositimz, 2016). For example, in the toxicity studies there were no changes observed in estrogen sensitive organ (e.g., uterus, ovary, testes, epididymis) weights or histopathology. Also, in the developmental and reproductive studies conducted (Van der Ven et al., 2008; Cope et al., 2014) there were no changes in endpoints (e.g., estrous cyclicity, reproductive performance (mating indices, fertility indices, fecundity indices, gestation index), gestation/lactation body weights, gestation length, litter data (pup sex ratios, live birth index, stillborn index, pup survival indices), sperm evaluations (% abnormal, % motile), primordial follicular counts, etc.) that provide lines of evidence of estrogenic activity. These data collectively demonstrate that no estrogenic activity was reported when TBBPA was orally administered to rats for up to 14 weeks at dose levels up to 1000 mg/kg/day.

 

There are two studies reported for TBBPA that show conflicting results in uterotrophic assays, a screen used to evaluate estrogenic activity in vivo. In an early study (Kitamura et al. 2005), which was not conducted according to OECD guidelines (440), TBBPA was reported to increase relative uterine weights at all dose levels compared to control without demonstrations of a dose response. This study is not considered to be highly reliable, however, as it was conducted in an ovariectomized mouse model for 3 days vs. the 7 days recommended in the guideline, did not specify the use of a low phytoestrogen diet, and did not report changes in absolute wet or blotted uterine weights, the measure used to evaluate a uterotrophic response. In a subsequent study designed according to OECD guideline 440 (i.e., a K1 study for reliability), TBBPA was reported to be negative for both estrogen agonist and antagonist activity (Ohta et al. 2012). Both absolute uterine wet and blotted weights were evaluated in this mouse ovariectomized model with dosing conducted daily for 7 days.

 

Although there were no adverse effects in the subchronic toxicology studies or developmental or reproduction studies to suggest TBBPA interferes with estrogen homeostasis, in a 2-year cancer bioassay conducted by the National Toxicology Program, TBBPA increased the incidence of uterine tumors in female Wistar Han rats with administration of oral dose levels of 500 and 1000 mg/kg (NTP, 2014; Dunnick et al., 2014). Since TBBPA is non-genotoxic, the mode of action (MoA) for this response could either be due to its ability to interact with the estrogen receptor, essentially mimicking estrogens or interfering in the metabolism of estrogens with chronic high dose administration.

As such the MoA of TBBPA and its ability to mimic estrogen was evaluated in vitro. There are a number of in vitro studies in the literature that evaluate the ability of TBBPA to bind and activate the estrogen receptor (Table included in the attachment). Also, within the US Environmental Protection Agency (EPA) high throughput screening (HTS) initiative (ToxCast/Tox21) the estrogen receptor bioactivity of TBBPA was evaluated in 18 selected assays within the EDSP21 Dashboard (http://actor.epa.gov/edsp21/) (Table included in the attachment). A model described by Judson et al. (2015) used to integrate the results from these 18 assays was conducted which incorporates information on TBBPA concentrations, cytotoxicity, and activity in each estrogen receptor assay to calculate an AUC value. An AUC value represents an integration of the activity among all the assays and results in a value ranging from 0 to 1, with 1 equal to 17α-ethinylestradiol activity. As reported within the EDSP21 dashboard, TBBPA was modelled to have an AUC value of 0 for estrogen receptor agonist and antagonist activity, and thus is given an overall designation of being inactive for the estrogen receptor (Table included in the attachment). The results from the ToxCast™ /Tox21 assays, along with the weak and/or negative results in the peer-reviewed literature described above, together provide evidence that TBBPA does not have estrogen receptor bioactivity. 

 

Dunnick et al. (2014) and Lai et al. (2015) proposed that the inhibition of estrogen sulfotransferase by TBBPA may be involved in mode of action for the TBBPA-induced uterine tumor response. Similar to TBBPA, estrogen and its metabolites are conjugated to both sulfate and glucuronic acid, with sulfation serving as the main pathway for the inactivation of estrogen (Xu et al., 2012). Though information is limited to in vitro studies, data indicate that TBBPA can inhibit estrogen sulfotransferase (ES) (Table included in the attachment). When these data are considered relative to the metabolic pathways for both TBBPA and estrogen, findings support that TBBPA could compete in vivo with the same enzyme systems as estrogens. This hypothesis was recently investigated; Borghoff et al. (2016) reported on a study that was specifically designed to evaluate the role of metabolic parameters in the development of uterine tumors following repeated, high dose exposures to TBBPA in Wistar Han rats. Data from this study demonstrated that administration of the same doses of TBBPA associated with the induction of uterine tumors result in a disruption in the balance of conjugates, reflected by a decrease in the TBBPA sulfate to glucuronide ratio; this dose-dependent decrease in this ratio supports a limitation in this sulfation pathway at dose levels of 250 mg/kg and above. These data also provide critical information regarding the feasibility for TBBPA to influence estrogenic potential through a disruption of its metabolism. As reported by Wikoff et al. (2016), which provides a comprehensive evaluation of the MoA for TBBPA-induced uterine tumors in an adverse outcome pathway framework, available data support that inhibition of ES by TBBPA would result in an increased bioavailability of estrogens, but only at dose levels of TBBPA that exceed the capacity of estrogen sulfation pathway. Based on the demonstration that the TBBPA metabolic profile is altered at high doses (such as those used in the cancer bioassay), and thus this initial event that is only operative under chronic high dose administration in a likely sensitive strain of rat, this MoA for the development of TBBPA-induced uterine tumors would then not be feasible in humans given differences in the kinetic and dynamic factors associated with high dose exposures in rats relative to human exposure levels to TBBPA. Human exposure levels are estimated to range from 3.2 E^-7 to 8.4 E^-5 mg/kg-day, and lifetime average daily dose estimates are associated with a margin of exposure that is >32 000 000 for cancer-based endpoints (Wikoff et al., 2015), supporting this mode of action; inhibition of ES is not feasible in humans at current exposure levels.

 

In conclusion, multiple lines of evidence demonstrate that TBBPA does not bind and activate the estrogen receptor. In repeated-dose toxicity assays, as well as developmental and reproduction toxicity studies, TBBPA did not demonstrate a significant estrogen-mediated response (e.g., changes in estrogen sensitive tissue weights and histology, accelerated vaginal opening, etc.) consistent with either estrogen receptor agonist activity or increased circulating estrogen levels due to inhibition of estradiol sulfation. Rather, estrogen-related responses are limited to effects observed following high-dose, chronic exposure in which the homeostatic conditions of estrogen metabolism are altered. Such conditions are not predicted to be feasible in humans. As reported by Wikoff et al. (2016), there are orders of magnitude differences between TBBPA plasma concentrations in rats associated with a limitation in the sulfation of TBBPA relative to serum concentrations of TBBPA measured in humans. Such comparisons are even more stark when one considers that the majority of samples evaluated in human studies reported in the literature have levels below the limit of detection. Similar margins of exposure are observed with external dosimetry comparisons (Wikoff et al., 2015).

Summary and Conclusions (Estrogen Wildlife)

There are a number of in vivo studies focused on growth and reproduction in invertebrates and fish (Wollenberger et al. 1999, Suprenant 1989, Suprenant 1989b, Brown et al. 2005, Kuiper et al. 2007). Growth and reproduction are often used as individual and population level indicators of endocrine modulation, but only when combined with molecular, biochemical and physiological endocrine biomarkers. In many of these studies, effects on growth and reproduction were noted, but estrogen specific endpoints (Vtg and aromatase) suggested no modulation of the estrogen pathway (Kuiper et al. 2007b, Song et al. 2014, Wang et al. 2011, Ronisz et al. 2004, Wollenberger et al. 2005). 

 

For instance, Kuiper et al. (2007) conducted an evaluation of zebrafish reproduction following aqueous exposure to TBBPA. Fecundity and fertilization were reported to be decreased following TBBPA exposure, the variability of data within and across exposure levels confounds the interpretation of these findings. Histological changes were also observed in TBBPA exposed female fish, though a high incidence of findings in the controls confounds the analysis. At the highest exposure level, sexual development was skewed towards females. However, no specific markers are available to directly link the observed responses to estrogen pathway modulation. In a second study by the same research group investigating specific markers of estrogenic activity (Vtg, aromatase, gonado-somatic index or GSI, liver somatic index or LSI), no increase in estrogenic activity was noted following exposure to TBBPA aqueous concentrations ≤ 0.193 mg/L (Kuiper et al. 2007b). Collectively, these data suggest that the observed reproductive responses are not associated with modulation of the estrogen pathway.

 

Huang et al. (2013) did evaluate ERα and ERβ mRNA, as well as liver and testis histology following TBBPA exposure in mosquitofish. In the livers and testis of adult fish, as well as in juvenile fish, changes in the expression of ERα and ERβ were noted. Even though the ERα and ERβ mRNA expression levels significantly changed relative to controls, the reported fold induction was small (<2). Importantly, no histological changes were noted in the liver or testes. Therefore, while small changes in ER receptor expression were noted, these changes were not associated with any toxicological findings. The reliability of this study is low due to concerns regarding the species used, its physiology, fish being collected from a field environment prior to laboratory use, and the lack of adherence to an accepted testing protocol. Given no positive control was used in this atypical testing species, interpretation of the findings is difficult.

 

Chow et al. (2011) was the only study where Vtg was measured that reported an actual increase in Vtg. Vitellogenin is an egg yolk precursor that is linked to estrogen receptor modulation. Other studies evaluated in this analysis utilized similar exposure concentrations and observed no changes in Vtg levels. The study by Chow et al. (2011) compared the TBBPA response to bisphenol A (BPA) and reported that TBBPA was approximately 2.7x less potent in inducing Vtg relative to BPA. Data suggest a weak estrogenic response, though concentrations that cause Vtg response are also those where systemic toxicity/mortality were observed. Therefore, the Vtg response was likely secondary to other toxicological responses.

 

Two in vivo studies in birds focused on gonadal and reproductive parameters (e.g. plasma testosterone, histology) that could provide specific information on estrogen pathway modulation. No effects following a 15 µg/g dose were reported (Berg et al. 2001, Halldin et al. 2001). Using embryonic chicken hepatocytes, no induction of Vtg was observed following exposure to ≤ 100 µM TBBPA (Ma et al. 2015).

 

In vitro assay data are often used to evaluate the potential for chemicals to modulate estrogen receptors (ER), even if the assays use human receptors. The yeast recombinant and ER-CALUX assays (Table included in the attachment) are commonly used to illustrate that TBBPA does not strongly interact with the estrogen receptor (Harju 2007, Hamers 2016, Huang 2013, Meerts 2001, Miller 2001). In five of the studies, no interaction of TBBPA with the estrogen receptor was noted. In only one study was an interaction reported, with a very weak TBBPA ERα potency value being reported (REC20> 1x10^-5 mol/L) (Li 2001). Hamers (2006) classified a chemical as non-potent if the observed response was < 20% of control at 10 µM. In addition, Li (2010) utilized a transfected human ER, which may have different biding characteristics compared to a fish ER. As reported within the US Environmental Protection Agency (EPA) EDSP21 dashboard, TBBPA was given an overall designation of being inactive for the estrogen receptor (http://actor.epa.gov/edsp21/).

 

Multiple lines of evidence demonstrate that TBBPA does not bind to the ER in vitro or significantly modulate the estrogen pathway in exposed wildlife. In avian and invertebrate studies, there is no evidence that suggests TBBPA causes an estrogen-mediated response. In fish, the overwhelming majority of studies demonstrated that TBBPA does not modulate the estrogen pathway. Most of these studies used estrogen specific measures of activity (vitellogenin or Vtg or aromatase activity) and reported that no positive estrogen modulation findings were observed. While one study using a non-traditional testing species (i.e. Gambusia sp.) reported changes in ERα and ERβ mRNA following TBBPA exposure, no histological changes were noted in the liver or testis (Huang et al. 2013). The lack of histological findings indicates that the mRNA measures did not correspond to an adverse consequence in exposed fish. The reliability of this study is low due to concerns regarding the species used, its physiology, fish being collected from a field environment prior to laboratory use, and the lack of adherence to an accepted testing protocol. Given no positive control was used in this atypical testing species, interpretation of the findings is difficult. Chow et al. (2011) reported weak estrogenic activity in zebrafish exposed to TBBPA, however, the study used a high level of carrier solvent (i.e. 1% DMSO) which could increase TBBPA availability in the organism or increase toxicity. In addition, the levels at which Vtg was increased were also associated with increased fish mortality, suggesting systemic toxicity was occurring at the same concentrations at which the weak estrogenic responses were observed.

Summary and Conclusions (Thyroid Human Health)

A number of well-performed animal studies with TBBPA are available and do not show any indications for adverse effects that would be mediated by perturbations in thyroid hormone homeostasis. In guideline in vivo studies (e.g., repeated dose toxicity (≤14 weeks’ duration) and developmental and reproductive studies) TBBPA exposure did not show indications of adverse effects in endpoints that would be associated with perturbations in thyroid hormone homeostasis such as changes in thyroid weights or thyroid histopathology (Table included in attachment; Van der Ven et al., 2008; NTP, 2014; Cope et al., 2015; Ositimz et al., 2016). Although these studies reported decreases in T4, corresponding decreases in T3 and/or TSH were not observed. Rats are typically susceptible to thyroid tumors when there is a sustained decrease in T4 and T3 and a corresponding increase in TSH, however, in two cancer studies in which mice and rats were administered TBBPA for 2-years there were no thyroid gland non-neoplastic or neoplastic lesions reported (NTP, 2014).

Similar to the repeated dose toxicity studies, in a guideline multigeneration reproduction toxicity study (Cope et al., 2015), no changes in selected endpoints influenced by perturbations in thyroid hormone modulation (e.g., thyroid weight, histopathology, fetal weight and survival, pup growth and survival, etc.) were observed although there were decreased serum T4 levels measured without corresponding changes in T3 or TSH. In a single generation reproduction study reported by Van der ven et al. (2008), besides a decrease in T4 and a slight sex specific change in T3, endpoints associated with perturbations in thyroid hormone homeostasis. In this study hearing deficits were not reported. Lilenthal et al. (2008) however did report that animals exposed to the same dose levels and conditions as Van der Van et al. (2008) there were changes in auditory responses such as brainstem auditory evoked potentials (BAEPs) at 50-110 days of age in female but not male rats. This change was stated to be associated with decreased in T4 serum levels; however there were no direct correlations carried out to show this relationship with dose, change in T4 serum levels, and change in BAEP. 

Changes in thyroid hormones without corresponding changes in thyroid gland weight or histology were also observed in a number of studies in which dosing of TBBPA extended from gestation to early life (Saegusa et al., 2009; Hass et al., 2003; Choi et al., 2011). Although both Hass et al. 2003 and Choi et al. 2011 demonstrated a decrease in T4, Saegusa et al. 2009 showed a decrease in T3 without corresponding changes in T4 or TSH. In all three studies no corresponding changes in thyroid histopathology were observed with changes in either T3 or T4 in TBBPA exposed animals. 

 

Although a number of studies measured a decrease in T4 serum levels in rats exposed to high dose levels of TBBPA, as noted by the EU (2006), these decreases were not of sufficient magnitude to induce perturbations in thyroid function including changes in T3, TSH and thyroid histopathology. Such a change in serum T4, especially at non- biologically relevant doses of TBBPA, constitutes a lack of concern regarding human exposures and potential adverse consequences on thyroid function.  In fact, in several human studies there is no association between TBBPA levels in serum and changes in thyroid hormones. In one study the concentration of TBBPA in serum of mothers of normal infants did not show any significant correlations with thyroid hormones (Kim and Oh, 2014). Kicinski et al. (2012) did not see any consistent associations between serum TBBPA levels, T4, T3, or TSH levels in Flemish adolescents; additionally, no correlations with performance in a battery of neurobehavioral tests were observed. These two studies support that there are no changes in perturbations of thyroid hormone homeostasis or neurobehavior (i.e., effect potentially associated with thyroid hormone modulation during development) associated with TBBPA exposure in humans.

 

A number of in vitro studies were conducted to evaluate the ability of TBBPA to interfere with the thyroid hormone pathway as presented in a Table in the attachment. Several studies reported that TBBPA interacts with human transthyretin (TTR, thyroid hormone binding transport protein) and displaced T4 in competitive binding assays (Meerts et al., 2000; Hamers et al., 2006; Harju et al., 2007); reported IC50 values ranged from 31 nM to 0.31 µM. Butt et al. (2011) reported that TBBPA inhibited deiodase activity. Although Hofmann et al., 2009 demonstrated that TBBPA (µM range) was a weak agonist and antagonist in the activation of TR in HepG2 cells, Oka et al., 2013 found this not to be the case for human TRα1 or TRβ1 in HXK293 cells. In the majority of in vitro assays presented in the Table either in yeast, human or animal cells lines with various TR constructs transfected, there is conflicting information concerning the binding and activation of the thyroid receptor hormone pathway, none of which provides any clarity to the changes that have been observed in the animal studies where decreased serum T4 is the only measure of the thyroid pathway decreased. In fact, depending on the cell line system, the TR receptor transfected in the system, and/or endpoint measured, there were none to weak TR agonist activity reported and mixed reports of antagonist TR activity (Kitamura et al., 2005; Freitas et al., 2011; Sun et al., 2009; Terasake et al., 2011; Shiizaki et al., 2010; Guyot et al., 2014). Also, in a Table in the attachment, two of three assays reported within the EDSP21 Dashboard show inactive TR bioactivity with one assay listed as active for antagonist activity with an AC50 of 42.4 µM compared to fulvestrant, a known TR antagonist with an AC50 of 0.001 µM.

In summary, although changes in T4, and in a few cases T3 were measured in TBBPA exposed animals under different protocols (i.e. routes of administration and duration, life stage, etc.), these changes were not reflected with changes in TSH or endpoints such as thyroid organ weights and histopathology that would be considered adverse, suggesting that these fluctuations are not of biological significance. This conclusion is similar to that determined by the EU (2006), UK COT (2004), and Health Canada (2013). Further, there was no evidence that the decreases in serum T4 were associated with hypothyroidism. Such evidence comes from a lack of effects in repeated dose toxicology or carcinogenicity studies; there were also no consistent effects reflected of a perturbation in thyroid homeostasis observed in early life and reproductive studies. Lastly, as reviewed by Lai et al. (2015), significant changes in T4 production in humans during development is reported to be associated with developmental delay, low body mass, brain developmental abnormalities and neurobehavioral development disorders. Although evidence of thinning of the parietal cortex occurred at postnatal day 11 by a crude screening level morphometric method in an animal study (Cope 2015), there were no neurological, neurodevelopmental or neuroperformance effects observed in this guideline study.

 

Lai et al. (2015) evaluated four possible hypotheses that could account for the decreases in serum T4 levels in the general absence of accompanying alterations in serum T3 and in complete absence of concurrent compensatory increases in serum TSH and evidence of histopathologic alterations. The first was the ability of TBBPA to interfere with thyroid hormone binding transport protein (TTR). TBBPA has been shown to compete with binding of T4 to TTR in vitro (Meerts et al., 2008; Hamer et al., 2004; Harju et al., 2007). If this type of competition occurred in vivo one would expect there to be evidence of overt hypothyroidism and elevated TSH in these repeated dose studies; however, there was no evidence to support this hypothesis. The second hypothesis proposed, which was determined to be the most likely, involved TBBPA stimulation of uridine diphosphate (UDT) up-regulation and increased metabolism of T4 in the liver, citing other research to support this hypothesis that selected hepatic UGT1A activity induced by some chemicals caused a reduction of serum T4 without compensatory increase in serum TSH and decrease in T3. As discussed by Lai et al. (2015), T4 is selectively inactivated by UGT1A isoform, whereas T3 is inactivated by UGT2B2. This may in part explain the change in T4 if TBBPA selectively stimulates UGT1A and not UGT2B2 activity. Additionally, Van der ven et al. (2008) reported an increase in liver weight which could reflect an induction in liver metabolizing enzymes, however an increase in liver weight was not consistently observed across repeated dosing studies (Borghoff et al., 2016; NTP, 2014).  The third hypothesis proposed was that of altered sequestration of T4 in body stores; this was considered by Lai et al. (2015) to be the most unlikely hypothesis based on the evaluation of evidence. Similarly, the fourth hypothesis, direct effects of TBBPA on the thyroid gland, lacked the evidence of hypothyroid effects including thyroid gland weight increases and histopathology. As such it is not clear how T4 levels are decreased, but the change does not appear to be at a significant magnitude to result in changes in T3 or TSH.

With respect to human exposure, it is notable that the alterations in T4, though not adverse, occur following repeated, high dose exposures in experimental systems. When compared to human exposure levels, very large margins of exposure exist (EU, 2006; EFSA, 2011; Health Canada, 2013; Wikoff et al, 2015). For example, very high margins of exposure were calculated by EFSA (2011) using changes in thyroid hormones as the point of comparison for all populations evaluated. As described above, doses in the range of 250 to 1000 mg/kg-day are required to achieve changes in T4, where as human exposure levels are estimated to range 3.2E^-7 to 8.4E^-5 mg/kg-day (Wikoff et al., 2015), thus suggesting that it is unlikely for there to be effects on thyroid hormones in humans.

 

In conclusion, TBBPA causes decreased serum levels of T4 without concomitant changes in T3 or TSH in repeated-dose toxicity assays as well as developmental, prepubertal and reproduction toxicity studies. The decrease in T4 in serum, although statistically significant, appears not to be biologically sufficient to result in changes in T3, TSH or changes in endpoints that would result in hypothyroid activity such as thyroid weight or histopathology. The changes in early life exposure studies also are not reflected in changes in endpoints reflecting hypothyroid activity including neurological and neuroperformance function in a study conducted under regulatory guidelines. The changes in T4 observed in experimental studies (which do not result in an adverse biological response) appear to occur only following repeated, high dose exposures to TBBPA; such conditions are not predicted to be feasible in humans.

Summary and Conclusions (Thyroid Wildlife)

There are a number of in vivo studies focused on metamorphosis in frogs (Shi 2010, Jangnytsch 2014, Zhang 2014, Goto 2006, Veldhoen 2006, Fini 2007, 2012a, 2012b). Tadpole metamorphosis is a thyroid-mediated process and TBBPA was found to suppress T3 mediated tail regression or hind limb development in several of these studies at concentrations ≥ 10 nM. 

 

Of particular interest, Zhang (2014) conducted a tadpole metamorphosis assay in Xenopus laevis where TBBPA was administered at various developmental stages. TBBPA exhibited antagonistic activity at developmental stages where endogenous T3 was high and agonist activity when T3 was low. The lowest effective concentration was 5.4 µg/L (10 nM) which is at the higher end of concentrations observed in the environment (Zhang 2014); these non-monotonic dose response findings were corroborated by the same laboratory in another species Pelophylax nigromaculatus in 2015. Interestingly, 5.4 µg/L TBBPA concentration in Pseudocris regilla shows agonist rather than antagonist activity (Veldhoen et al. 2006). The next effective concentration in suppressing TR mediated metamorphosis was 1 µM as observed in Rana rugose and Xenopus laevis (Kitamura 2005: Fini 2007).

 

Two studies in fish focused on toxicity, eye development and mRNA expression of hypothalamic-pituitary thyroid axis (e.g. TRα & β, thyroid stimulating hormone, thyroid peroxidase, deiodinase type 1, 2, & 3) that could provide specific information on thyroid modulation. Increased expression of TRα was observed at 0.82 mg/L in zebrafish embryos (Chan and Chan 2012). At much higher concentrations (i.e. 300 µg/L), zebrafish larvae eye development and swimming ability was altered (Baumann 2016). Eye development is a thyroid mediated process. Thyroid disruption causes eye development impairments, decreased morphology and function.

 

In vitro assay data are often used to evaluate the potential for chemicals to modulate thyroid receptors (TR), even if the assays use human receptors. The yeast recombinant, thyroid hormone-responsive receptor (CHO-K1), and competitive binding experiments using Xenopus laevis, XL58-TRE-Luc cell line suggests that TBBPA is a weak modulator of the thyroid system (Shiizaki 2010, Kitamura 2005, Kudo 2006). Yang and Chang (2015) used a zebrafish liver cell line and reported that TBBPA does not strongly interact with the TR. TBBPA competitively inhibited 3,3’,5-triiodothyronine (T3) interaction with the TR, with a TBBPA IC50 of 3.5 µM (Kitamura 2005). This study utilized a transfected human TR, which may have different binding characteristics compared to a fish or frog TR.

 

Multiple lines of evidence in vivo demonstrate that TBBPA can modulate the thyroid pathway in wildlife; however the observed responses are near the maximal aqueous environmental concentration of 4.87 µg/L (Yang 2012). In vitro evidence in environmentally relevant species is inconclusive as one study with a zebrafish cell line reported a lack of TR modulation, while a study using a Xenopus laevis cell line reported weak modulation. Zhang (2014) reported the lowest effective TBBPA concentration for T3 associated development effects was 5.4 µg/L (10 nM). Maximal concentrations of TBBPA in water are ≤ 4.87 µg/L, with most levels detected in the pg/L range (Xu 2013, Yang 2012, Harrad 2009). With most TBBPA being rapidly metabolized, the duration of exposure to TBBPA is minimal. None of the in vivo studies strictly followed a standard OECD validated protocol (e.g. OECD 241, Larval Amphibian Growth and Development Assay; OECD 231, Amphibian Metamorphosis Assay), which would strengthen the ability to interpret these findings. Most of the studies measured changes in mRNA expression due to exposure to TBBPA; however, many of these studies did not perform histological examinations which could be used to link the mRNA changes to a critical endocrine finding.  Lastly, many of the studies identified did not measure TBBPA or monitor for iodine levels in the water. These two factors (i.e. concentration verification and aqueous iodine levels) can greatly affect the interpretation and/or outcome of the studies.