Mindfully.org

Final Report of the Endocrine Disruptors Low-Dose Peer Review

National Institute of Environmental Health Sciences, NIH, National Toxicology Program 14may01

TABLE OF CONTENTS

EXECUTIVE SUMMARY
Peer Review Organizing Committee
Subpanels: Chairs, Rapporteurs, Panelists
Selected Studies from Principal Investigators
Selected Studies: Requested Parameters
Issues Relative to the Evaluation of Endocrine Low-Dose Studies
Subpanel Questions and Issues

Chapter 1:
Report of the Bisphenol A Subpanel

Background

Chair
George Stancel, University of Texas Health Science Center at Houston

Rapporteur
Gail Prins, University of Illinois at Chicago

Facilitator
Penelope Fenner-Crisp, U.S. Environmental Protection Agency

Panelists
Ralph Cooper, U.S. Environmental Protection Agency
Warren Foster, Bureau of Chemical Hazards, Health Canada
Jun Kanno, National Institute of Health Sciences-Japan
John Faust, California Office of Environmental Health Hazard Assessment

Statistics and Dose-Response Modeling Subpanel Representatives
Joseph Haseman, NIEHS (statistics)
Robert Delongchamp, National Center for Toxicological Research (modeling)

Prior to the meeting the members of the Subpanel were provided with a selected set of background references and a set of "selected studies" which had been selected by the members of the organizing committee. The raw data for some, but not all, of the selected studies was provided in advance to the Statistics and Dose-Response Modeling Subpanels who independently analyzed the data. Drs. Joseph Baseman (Statistics Subpanel) of NIEHS and Robert Delongchamp (Modeling Subpanel) of the National Center for Toxicological Research then served as members of the Bisphenol A Subpanel at the meeting.

Based on an analysis of the selected studies, the Subpanel was then asked to address a set of 7 questions.

1. What is the extent of empirical evidence demonstrating low-dose effects of bisphenol A (BPA) on reproductive and developmental endpoints from studies in mammalian species?

For this meeting, "low dose effects" refer to biological changes that occur at environmentally relevant exposure levels or a doses that are lower than those typically used in EPA's standard toxicity testing paradigm. Within and across studies, describe the specificity, consistency, and strength of the evidence with consideration of the timing of exposure, when the endpoint was measured, and sensitivity of the endpoint. Are conclusions supported by appropriate statistical analyses?

2. What is the extent of empirical evidence demonstrating the lack of low-dose effects of BPA? Within and across studies, describe the specificity, consistency, and strength of the evidence with consideration of the timing of exposure, when the endpoint was measured, and sensitivity of the endpoint. Are conclusions supported by appropriate statistical analyses?

3. If possible, identify differences in study design or biological factors that might account for the observed differences in study outcomes.

4. How do the findings from studies of low-dose effects on reproductive and developmental outcomes using chemical X compare with those for other endocrine active chemicals?

Describe the specificity and consistency of the evidence, and if possible, identify the similarities and/or differences in study design, chemical activity, species or strain, etc. that might explain the observed outcomes.

5. Describe the available and relevant pharmacokinetic, biologic, and other mechanistic information that strengthen or weaken the plausibility of low-dose effects of BPA. How did this information impact the Subpanel's overall conclusions?

Describe the shape of the dose-response curves for BPA in the low-dose region using empirical data as well as biologically based dose-response models.

6. Based on the totality of available knowledge, what is the Subpanel's overall conclusion regarding whether BPA can cause hormone-related effects on reproductive and developmental endpoints at doses lower than those typically used in the standard toxicological dose-setting paradigm?

7. Are there specific knowledge gaps for which additional research relative to the low-dose question for BPA is needed? If possible, suggest ways to address those gaps.

(NOTE: Because the Subpanel was not provided with an explicit level to be considered "low dose", considerable time was spent for an on site discussion about what dose should be used as a cut-off for this level to address questions (1) - (7) above. In the initial discussion, it was noted that a NOAEL had not been found in rodent studies used for setting a reference dose for BPA. A 1982 NTP Technical Report (CAS no. 80-05-7) found "no convincing evidence that bisphenol A was carcinogenic for F344 rats or B6C3Fl mice of either sex". Nevertheless, since the incidence of testis tumors was significantly elevated in the low dose (1000 ppm) male rat group, the EPA apparently used these data to support their conclusion that this dose, which corresponds to an oral dose of 50 mg/kg/day based on typical food consumption rates, represents a LOAEL for BPA Applying the standard use of uncertainty factors of 10 for interspecies variability, intraspecies variability, and subchronic to chronic comparison, the Subpanel originally decided to use the oral reference dose of 50 µg/kg/day as the "low dose" cut-off.

However, after a lengthy Subpanel discussion on this point, members of the Organizing Committee instructed the Subpanel to consider 5 mg/kg/day or less as representing a "low dose" of BPA, and during the course of the meeting instructions were similarly given to consider 1 µg/kg/day or less of diethylstilbestrol (DES) as a low dose of that chemical. [Note added in proof. The Organizing Committee based this directive on the definition of "low dose" used in this peer review as "doses that are lower than those typically used in EPA's standard toxicity testing paradigm". This would generally be interpreted as doses in the range of a NOAEL, or in the absence of a NOAEL, the use of - ~ LOAEL/10. Given the oral LOAEL of 50 mg/kg/day in rats as noted above, this approximation corresponds to a value of 5 mg/kg/day, and this was the rationale used by the Organizing Committee.])

It should be emphasized that the Subpanel used the 5 mg/kg/day dose level as instructed by the Organizing Committee to define the low-dose cutoff for BPA, regardless of the route or duration of administration or the age of the animal used for a particular protocol. This is an important point, since the concentration of a chemical reached at a tissue site, can vary widely following administration of identical doses by different routes and over different durations. This is especially true when one route is oral and the other is parenteral. The Subpanel also did not distinguish the duration of administration of BPA, i.e., the number of days for which a dose of 5 mg/kg/day was administered, the age at which exposure occurred, or the developmental stage for in utero exposure. Thus, in some cases BPA was administered by injection or silastic implants, while in others it was given via feed, and in others it was given by gavage. Additionally, some exposures were in utero, while others were during neonatal or adult life.)

Question 1. What is the extent of empirical evidence demonstrating low-dose effects of BPA on reproductive and developmental endpoints from studies in mammalian species? Within and across studies, describe the specificity, consistency, and strength of the evidence with considerations of the timing of exposure, when the endpoints were measured, and sensitivity of the endpoint. Are conclusions supported by appropriate statistical analyses?

Studies from vom Saal and Colleagues.

Several studies in mice provided evidence for a low dose effect(s) of BPA. These include the following. The report by Nagel et al (1997), EHP 105:70-76, demonstrates an increase in absolute prostate weight of CF-1 male mouse offspring at 6 months of age following administration of 2 and 20 µg/kg/day BPA to pregnant mothers. The statistics Subpanel reanalyzed this data and found it to be significant at the level of p < 0.05, and the BPA Subpanel found this data to be credible. However, the changes in body weight reported in this study were found to be unusual in that the body weight appears decreased in the low dose group, but this finding is not replicated in other studies from the same group, where in fact, a low dose of BPA increases body weight (Howdeshell et al, vide infra). In addition, a study from another group (Ashby, d., H. Tinwell, et al., "Lack of effects of low dose levels of bisphenol A and diethylstilbestrol on the prostate gland of CF-1 mice exposed in utero", Reg. Toxicol. Pharmacol. 30:156166, 1999) did not observe BPA-related effects on body weight.

A subsequent paper from the same laboratory (vom Saal et al., Tox. Industrial Hlth 14:239-260) using other measurements of the same group of 6 month old animals, found a small (approximately 20%) decrease in sperm efficiency (daily sperm production per gram of testis weight) in the 20 µg/kg/day dose group (but not the 2 µg/kg/day group). The authors reported this decrease as significant (p < 0.05), but this level of significance was not confirmed by the Statistics Subpanel's revaluation which found only p < 0.10 (see final report of Statistics Subpanel). The Subpanel did not consider the issue of whether or not a 20% decrease in daily sperm production efficiency, even if real, is likely to have biological or toxicological significance. It was also noted that statistically significant changes in preputial gland weight, seminal vesicle weight, and testis weight were not observed in this study.

The above 2 studies were performed with CF-1 mice obtained from the colony at the University of Missouri. That colony is no longer available, but in an oral presentation at the Low Dose Peer Review, data were presented showing an effect of in utero exposure of 10 µg/kg/day to pregnant CD-1 mice on enlargement of the prostate in male offspring. Since the information from this study was not provided prior to the peer review, neither the statistics nor BPA Subpanels had the opportunity to independently analyze the raw data.

In several reports (Howdeshell et al., Nature 401:763-764 and Howdeshell and vom Saal, Am. Zoologist 40: in press) in utero exposure to 2.4 µg/kg/day BPA advanced puberty in female CF-1 mice (measured as the number of days between vaginal opening and first vaginal oestrus), although there was no change in the age of vaginal opening). This effect was observed only for females positioned between two females during intrauterine development. In the same study, increases in body weight at weaning of both male and female CF-1 were observed following in utero exposure in animals located between 2 females or between 1 male and 1 female, but not

between animals located between 2 males, during intrauterine growth. It was noted at the Peer Review Meeting that the authors did not provide the raw data to the Statistics Subpanel for reanalysis for either of these studies, so there was no way for that Subpanel to independently confirm the reported positional effects.

These observations seemed counterintuitive to the Subpanel since animals positioned between 2 females would be expected to be exposed in vivo to higher levels of endogenous estrogens from neighboring fetuses than animals positioned between 2 males. While counterintuitive, the Subpanel could not rule out the possibility of endocrine signaling loops, e.g., some type of "feed forward" loop, not currently documented. [Note added in proof. The Statistics Subpanel found in data provided in advance of a paper to be presented by vom Saal's group in Berlin in November, 2000, that no consistent positional effects on body weight were observed in castrated mice treated at 3 months of age with testosterone or 5alphadihydrotestosterone. See the Statistics Subpanel report for further details.] The Bisphenol A Subpanel did not discuss whether or not these reported changes in body weight at weaning or time between vaginal opening and vaginal oestrus are likely to have biologically or toxicologically significant implications.

Studies from Ben-Jonathan and Colleagues.

In a set of studies using Fisher 344 rats, another group demonstrated (Khurana et al., Endocrinology in press) that sc injections of 5-10 mg/kg/day of BPA on neonatal days 1-5 altered plasma levels of prolactin and developmental patterns of this hormone in the plasma of both male and female animals between 15 and 30 days of age. BPA also produced more modest changes in hypothalamic and pituitary levels of estrogen receptor mRNA levels measured by RT-PCR.

In other studies, this same group investigated the effects of administering BPA to young adult rats (7-8 weeks of age) and observed the following. 1) silastic implants yielding BPA release estimated to be approximately 0.5 mg/kg/day for 3 days increase uterine epithelial cell height and uterine weight in one strain of rats (F344) but not another (SD), and 2) silastic implants releasing BPA at an estimated daily rate of 0.3 - 0.5 mg/kg/day increase serum prolactin levels, again in F344 but not SD rats. In another study with 7-8 week old rats, this group reported increases in uterine DNA replication and c-fos gene expression in F344 rats - although no statistically significant effects were noted in the low dose range of 5 mg/kg/day or less used by the Subpanel-there was a trend toward a response over the entire dose range, including the low doses. Interestingly, this latter study displayed an apparent monotonic dose response curve over the entire dose range studied. It is also noteworthy, that collectively these studies illustrated a clear difference in sensitivity to the effects of BPA in the two strains of rats.

On balance, the Subpanel found the set of studies provided from Dr. Ben-Jonathan's group to be very credible, and consistent. At the same time, however, it should be stressed that these studies did NOT find any low dose effects in SD animals, and that the "low dose" effects of BPA seen in F344 animals were observed at what would be considered the "high edge" of the low dose range. Furthermore, BPA administration in all of these studies was via sc injection or release from silastic implants, and both these routes of administration would be expected to have far higher levels of bioavailability than oral administration of BPA.

Note. It should be mentioned that while the Bisphenol A Subpanel found the data from Dr. Ben-Jonathan's group to be credible and consistent, these data were not provided to the Statistics Subpanel for their independent re-analysis.]

Low dose effects observed in other studies.

Several other reports contained data that reported effects in the low dose range of BPA. In one study from Welsch's group an effect of BPA on ventral prostate weight was observed in male SD rat offspring exposed in utero, although a subsequent study with a different sampling strategy with larger N values did not repeat this observation. Furthermore, the effect that was observed did not show a clear dose-response relationship.

In multigenerational studies in SD rats with a very large number of endpoints, statistically significant (p < 0.05) increases in ovarian weight were found in certain BPA-treated groups in a study by Tyl and significantly decreased anogenital distances were found some BPA-treated groups in a study by Ema. However, the Subpanel felt these observations displayed erratic dose response relationships, and some changes were eliminated when the values in question were subject to corrections for body weight differences between groups. Given the large number of endpoints in these studies, the Subpanel felt these miscellaneous observations might simply represent coincidental effects.

Summary. There are several reports of low dose effects of BPA which the Subpanel finds credible as outlined above, especially in studies with CF- I mice from vom Saal's group, and this group presented similar data for CD-1 mice in an oral presentation at the meeting. Data from Ben-Jonathan's group was considered very credible by the Subpanel, and this data is also potentially important because it provides very strong evidence for strain differences in the sensitivity to BPA which is consistent with known differences in SD and F344 rats with regard to sensitivity to endpoints of estrogenic action. These latter studies, however, were performed at the very high end of the low dose range. Also, they all utilized either sc injection or release from silastic implants as the route of administration of BPA, and the Subpanel questions whether similar results would be obtained if this chemical was administered by the oral route which would be more toxicologically relevant. These findings would thus provide greater support for a low-dose effect if they could be repeated at lower doses and/or following oral administration. In conclusion, there is credible evidence for low dose effects of BPA, and the conclusions reported in the above several studies are supported by appropriate statistical analyses. This evidence is limited to one dose level in a small number of reports and it is thus difficult to generalize about specificity, consistency, or strength of the evidence relating to the timing of exposure, biological endpoints measured or their functional significance, or their sensitivity.

Question 2. What is the extent of empirical evidence demonstrating the lack of low-dose effects of BPA? Within and across studies, describe the specificity, consistency, and strength of evidence with consideration of timing of exposure, when the endpoint was measured, and sensitivity of the endpoint. Are conclusions supported by appropriate statistical analyses?

A number of studies have provided evidence demonstrating the lack of low dose effects of BPA. Since low dose effects were NOT observed in these studies, the experimental details are not reviewed as extensively as those in Question #1 above-rather, interested readers may refer directly to the Selected Studies chosen by the Organizing Committee. This lack of discussion here is simply due to the lack of any observed effects. The Subpanel felt that the selected studies provided fell into 3 categories.

First there were 3 very large studies, conducted as GLP studies, that failed to show low dose effects of BPA. These included several multigenerational studies (by Tyl and Ema) in rats that examined a large number of endpoints, and a large study (by the Cagen group) using CF-1 mice specifically designed to be conducted exactly as one of the vom Saal studies.

Second, a large study from the Welsch group that used multiple pups per litter exposed during pregnancy found no BPA effects on prostate weight or on other endpoints.

Finally, a number of studies from Ashby's group using both mice and rats, including some that used the same CF-1 strain of mice and were designed to replicate the vom Saal studies, did not observe low dose effects of BPA.

The Subpanel explicitly noted that this collection of studies covered the reported window of sensitivity of exposure to BPA reported in studies referred to in Question #1, that they involved very long duration exposures over multiple generations, that they used both mice and rats, and that they administered BPA by several routes. Several of these studies also went so far as to include DES as a known estrogen so as to repeat the experimental design of some of the vom Saal studies. (Note: these studies did not observe an effect of DES on the endpoints measured as reported in studies from vom Saal's group, but the Subpanel did not address the question of whether one would or would not expect to see an effect of DES on the endpoints measured in either set of studies. The Subpanel did note, however, that if DES was included in a given study, its effect generally "mirrored" the effect of BPA, i.e., either the two produced a similar effect or neither produced any effect.) As a group these studies are very consistent, the conclusions are supported by appropriate statistical analyses, and the Statistics Subpanel confirmed the lack of BPA effects for the studies noted above, except for the second Welsch designed to investigate sampling design for which that Subpanel was not provided the raw data. Collectively, these studies found no evidence for a low dose effect of BPA, despite the considerable strength and statistical power they represent, which the Subpanel considered especially noteworthy.

Question 3. If possible, identify differences in study design or biological factors that might account for the observed differences in study outcomes.

There were a number of differences between the studies that provided evidence demonstrating either a low dose effect of BPA or the lack of such an effect.

1. Some of the no effect studies were multigenerational, and animals were thus chronically exposed to BPA. Animals in these studies may have adapted so that they did not show a response to BPA at some "critical" time. However, there were other studies that failed to observe a low dose effect of BPA which utilized an exposure paradigm designed to reproduce the short-term exposure studies from the vom Saal group and these also failed to show a low dose effect of BPA.

2. Diets were not identical in studies that observed and did not observe low dose effects. In particular, the Subpanel wishes to note that the background level of estrogens, e.g., from dietary sources, may have been different and this could have contributed to the positive effect of the BPA via previously unrecognized mechanisms (vide infra). For example, vom Saal studies used a diet (Purina 5001) different from that used by the Cagen and Tyl studies (Purina 5002), and Thigpen reported (background data) that the 5002 diet had soy/phytoestrogen levels approximately half of those found in the 5001 diet. The Subpanel thought the possible contribution of dietary estrogens should be considered in light of the report that intrauterine position plays an important role (presumably due to small differences in exposure to endogenous estrogens) in determining whether developmental exposure to BPA in utero produces biological effects in the adult animal. These studies reported that IUP effects played a role in the response of both female and male animals for low dose effects of BPA, i.e., 2F animals (animals between 2 females) show the greatest response following developmental exposure to BPA.

3. The Subpanel also felt that differences in the strains of mice used could in theory have contributed to different responses to low doses of BPA. For example, while the studies of both Cagen's and Ashby's groups used CF-1 mice in attempts to replicate the experimental format used in the vom Saal studies, the CF-1 mice used by the Missouri group had been raised in a closed colony since 1979. While the studies of both Cagen and Ashby also used CF-1 mice, these were from true outbred colonies. 

4. Careful examination of the raw data indicates that certain parameters in the control animals were different in studies that observed and did not observe low dose effects of BPA. In particular, the control BW and prostate weights differ between some studies, e.g., some of the Ashby studies and the vom Saal studies. This raises the theoretical possibility that tissues may have already been maximally stimulated by estrogens and /or that the differences in body and prostate weights could be indicative of different levels of maturation in the animals used in the two studies.

5. The routes of administration of BPA varied across studies, and this was felt to be potentially most significant for studies using sc injection or release from silastic implants versus those studies using oral administration. Even within the oral dosing groups there were differences, e.g., in different studies BPA was given in drinking water, by gavage, or in oil via micropipettes.

6. There may have been some differences in housing of the animals between different studies. For example, housing of animals singly versus group housing for different periods of time. This could, in theory, affect the outcome of studies since there are known effects of housing due to the phenomenon of male dominance.

7. Different bedding was used by Ashby, vs that used by vom Saal, although Cagen used the same bedding as the Missouri group. Thus, while bedding could be a potential factor (as a source of possible exposure to estrogenic substances) the Subpanel thinks this is less likely to be a potential factor than other differences.

8. The sample size was significantly different between studies that did and did not report low dose effects of BPA. However, the Subpanel did not thinly this particular concern was likely to be the principle basis for the differences because the negative studies had the larger number of animals. This further emphasizes that the different studies reviewed by the Subpanel indeed observed different outcomes.

9. There were differences between studies with respect to whether or not there was an analysis of BPA purity in the starting material, and in the concentrations actually present in dosing solutions. The Cagen study included an analysis of the BPA used as well as an analysis of the dosing solutions for their actual BPA content. The Ashby studies did not perform chemical analyses of BPA during the actual study, but did determine the stability of BPA solutions (stable). Vom Saal did not perform chemical analyses of BPA preparations used, but did determine the estrogenic potency. Ashby and vom Saal obtained BPA from the same supplier (Aldrich), although the Ashby sample was obtained from an Aldrich distribution source in the UK while the vom Saal material was from a source in the U.S. In contrast, Cagen used BPA obtained from a different supplier (Dow).

The Subpanel found no specific reason to suspect that there were differences in the estrogenic activity, impurities, or other properties between the BPA batches used in these different studies, or in the preparation of dosing solutions. Nevertheless, the studies that reported a positive low dose effect did not specifically analyze their dosing solutions or starting material, and one has to recognize this is always a potential confounder. Thus, without analyses done at the time of the actual study, one cannot unequivocally rule out potential effects of contaminants or errors in the preparation of dosing solutions.

Question 4. How do the findings from studies of low-dose effects on reproductive and developmental outcomes using BPA compare with those for other endocrine active chemicals? Describe the specificity and consistency of the evidence, and if possible, identify the similarities and/or differences in study design, chemical activity, species or strain, etc. that might explain the observed outcomes.

Due to the large amount of material on BPA assigned by the Organizing Committee, and the level of discussion of that material, the Subpanel did not have sufficient time for an in depth analysis of studies on other endocrine active chemicals. The Subpanel felt that it could only briefly consider DES, since some of the BPA (both those reporting and not reporting low dose effects of BPA) also studied DES. As noted previously, a representative of the Organizing Committee instructed the Subpanel to consider a dose of 1 µg/kg/day as a "low dose" of DES.

There are credible reports that DES may produce low dose effects, including studies from both the vom Saal and Newbold laboratories. Similar to the vom Saal findings in the male reproductive tract, Newbold's lab found low-dose stimulatory effects on female reproductive tract endpoints. Studies from the Welsch group also showed a decreased body weight at 10 µg/kg/day of DES in drinking water and an effect on the vaginal opening (advanced) in the female pups. Data from another study provided by the Organizing Committee as background information (Gupta) also showed a low-dose BPA effect on prostate, which was blocked by the antiestrogen 1C1 182,780. However, other credible studies have not observed such low dose effects of DES or other estrogens. Thus the limited time available to discuss this question, as well as the conflicting results, did not really allow the Subpanel's discussion of this point to contribute much to the central issue of low dose effects of BPA.

(As a point of reference, several members of the Subpanel, who are familiar with uterotrophic assays, expressed the opinion during discussions that an oral dose of 0.2 µg/kg/day DES is approximately the lowest dose at which they would expect to observe an effect of the synthetic estrogen following this route of administration.)

Question 5. Describe the available and relevant pharmacokinetic, biologic, and other mechanistic information that strengthen or weaken the plausibility of low dose effects of BPA. How did this information impact on the Subpanel's overall conclusions? Describe the shape of the dose response curves in the low-dose region using empirical data as well as biologically based dose-response models.

The selected studies included some in vitro data which indicate that BPA is not bound as extensively to serum proteins as estradiol. This data alone would suggest that BPA might preferentially exit the plasma space (relative to the endogenous hormone) to enter target cells. However, this hypothetical possibility has not been established in vivo, which is the critical issue since decreased plasma binding might also be expected to enhance both renal and hepatic clearance and such an effect would be expected to decrease BPA concentrations at cellular receptor sites.

There is also data available that the bioavailability of BPA is likely to be significantly less following oral vs parenteral administration because of first pass hepatic metabolism, especially since the glucuronides generated by metabolism do not have appreciable affinity for the classical estrogen receptor.

A very important issue for thorough analyses of developmental effects is fetal uptake via transplacental transfer, of non-metabolized BPA, as well as fetal biotransformation and accumulation of BPA and metabolites. Extensive information on this important point was not available in the selected studies, but from comments offered to the Subpanel by audience members, these issues may be contentious.

An in-depth knowledge of directly measured pharmacokinetic parameters such as bioavailability, bioaccumulation, transplacental transfer and fetal accumulation, clearance, volume of distribution, half-life, and the complete spectrum of metabolites formed, is essential to understand the toxicology of BPA and is particularly important because of the low affinity, as measured in vitro, of this chemical for estrogen receptors. This is also important because BPA and other endocrine active chemicals could in theory affect the metabolism of endogenous steroid hormones via induction or inhibition of P450s or glucuronyl transferases, or other mechanisms.

With respect to the shape of the dose curve, there are not sufficient doses of BPA that have been reported to elicit low dose effects to establish the shape of the dose response curve for this chemical. Parenthetically, the Subpanel noted that in the studies with F344 rats, which contained 1-2 doses of BPA in the low dose range (5 mg/kg/day or less), showed a monotonic dose response curve (Ben-Jonathan studies).

In addition, the Subpanel wishes to emphasize that the large number of negative data points in the literature make it impossible to perform any sensible dose response modeling for BPA in the low dose range at this time.

The paucity of BPA pharmacokinetic data led to the Subpanel's view that this is an area that represents a critical data gap. Extensive pharmacokinetic information was not provided in the selected studies and background information. This, coupled with the available time, precluded the Subpanel from undertaking a rigorous consideration of pharmacokinetic issues. Thus, if definitive and reproducible pharmacokinetic information on BPA is available, it should be thoroughly analyzed in the context of low dose effects, and if such data is not available, it should be a high priority for future work. This was emphasized by the striking difference in response of two rat strains (F344 and SD) to BPA in the selected studies, and because of the clear evidence of genetic effects on hormonal responsiveness reported in the literature and described in part during the oral sessions at the meeting.

Question 6. Based on the totality of available knowledge, what is the Subpanel's overall conclusion regarding whether BPA can cause hormone-related effects on reproductive and developmental endpoints at doses lower than those typically used in the standard toxicological dose-setting paradigm?

There is credible evidence that low doses of BPA can cause effects on specific endpoints. However, due to the inability of other credible studies in several different laboratories to observe low dose effects of BPA, and the consistency of these negative studies, the Subpanel is not persuaded that a low dose effect of BPA has been conclusively established as a general or reproducible finding. In addition, for those studies in which low dose effects have been observed, the mechanism(s) is uncertain (i.e., hormone related or otherwise) and the biological relevance is unclear.

(Note: The Subpanel wishes to emphasize that the above is a consensus statement. The Subpanel expended a considerable amount of time and effort developing the above statement to answer the question posed by the Organizing Committee about our "overall conclusion", and the true sense of our overall conclusion is accurately presented from the entirety of the above statement. Thus, the presentation of only a portion of the above statement would not accurately represent the content or spirit of our conclusion. Consequently, the Subpanel will not endorse anything but the above statement in its complete form as presented here.)

Question 7. Are there specific knowledge saps for which additional research relative to the lowdose question is needed? If possible, suggest ways to address those saps.

There are numerous knowledge gaps which limit the ability to assess low dose effects of BPA. Some of the specific items suggested during the Subpanel discussion are listed below, although it should be noted that this listing is not intended to be all-inclusive.

1. Studies should be performed with multiple doses of BPA in the low dose range, especially following oral administration during in utero or early neonatal development. If experimental paradigms can be developed to conclusively establish low dose effects of the chemical as a general, reproducible phenomenon, these should be used to obtain sufficient data points to perform credible physiologically based pharmacokinetic modeling.

2. Extensive pharmacokinetic data should be obtained in multiple species including CF-1 and CD-1 mice and in F344 and SD rats. This data is intrinsically important and is also required for modeling studies.

3. There is no meaningful data on the occupancy of estrogen receptors following exposure of animals to BPA in the low dose range, especially during critical periods of development. Data on the occupancy of receptors, in the reproductive tract, pituitary, and brain, following exposure of animals to low and high doses of BPA would be very valuable and is essential to rigorously address mechanistic questions.

4. The use of pharmacological (e.g., specific receptor antagonists) and genetic (e.g., knock out animals) approaches would provide important information about the mechanism of BPA effects, especially the role (if any) of estrogen receptors in any observed effects.

5. Further studies on the intrauterine position effect are suggested to fill existing knowledge gaps. It would be valuable to establish the generality and reproducibility of this effect, as well as establishing unequivocally the endogenous hormone levels as a function of intrauterine position and the site of their production. In related areas, it may also be important to carefully examine the effect(s) of minor differences in background levels of estrogens (e.g., provided by different feeds, due to genetic variation and species differences, etc.).

6. Genetic and epigenetic factors that affect responses to BPA and hormones in general are important areas that deserve further study. These include not only factors that affect hormone and receptor levels, but also factors in "intermediate" steps in hormone action which could lead to observed differences in sensitivity (e.g., such differences in "intermediate" or "down stream" effects have been suggested from studies comparing F344 and SD rats).

7. Mechanistic studies of BPA action that span the full course of in utero development, neonatal life, puberty, and adulthood would provide important data not currently available.

8. Given recent advances in understanding the basic mechanisms of steroid receptor actions, ligand specific effects of BPA on the transcriptional activity of receptors, recruitment and activation of co-activators and co-repressors, regulation of transcription by protein- protein vs DNA-binding mechanisms, and non-genomic actions of BPA might aid our understanding of the actions of this chemical at all dose levels. These studies would be especially important to determine if effects of BPA are mediated through classical hormone regulated pathways, or whether other mechanisms are operable.

9. One of the most critical needs is to search for other possible markers and specific endpoints (e.g., in addition to, or instead of, gross measures such as organ weights) that can be used to reproducibly investigate low dose effects of BPA. The development of easily measured and sensitive molecular endpoints, especially endpoints that can be assessed shortly after exposure, are critical needs which would greatly aid our ability to resolve current questions about low dose effects of BPA. \

Chapter 2:
Report of the Other Environmental Estrogens and Estradiol Subpanel

Introduction:

Chair
Michael Gallo, UMDNJ-Robert Wood Johnson Medical School

Rapporteur
Kenneth Reuhl, Rutgers University

Facilitator
Lynn Goldman, Johns Hopkins University School of Public Health

Panelists
Mari Golub, California-EPA
Claude Hughes, UCLA School of Medicine
Richard Lyttle, Wyeth-Ayerst Research
Lynne McGrath, Schering-Plough Research Institute
Patricia Whitten, Emory University

Statistics and Dose-Response Modeling Subpanel Representatives
John Bailer, Miami University of Ohio (statistics)
Michael Kohn, NIEHS (modeling)

Estrogens are classically defined as those compounds (endogenous and exogenous) that induce the state of estrus in the immature or ovariectomized female rat or mouse. Variations on this theme have been used since the discovery and characterization of the estrogen specific binding protein called the estrogen receptor. In elegant experiments it has been shown that the estrogen action of estradiol and other estrogens is mediated by the estrogen receptor (ER). Hence, many compounds are being classified as estrogens because they bind to the estrogen receptor and turn on estrogen responsive genes. In addition, the recent discovery of a second high affinity estrogen binding protein (ERß) has added new complexities to the general principle of induction of estrus. The ER and ERß are not distributed equally in the same tissues or cells. Hence, even the definition of a classic ER agonist, antagonist and partial agonist is in flux.

The subpanel addressed the question of what is an estrogen and how the definition should flavor the deliberations. The members also addressed the question of potency of a compound as an estrogen when compared to estradiol. The area of relative potency and mechanisms of action were beyond the scope of this meeting but are clearly areas for further research. The discussions focused on three major areas: 1) criteria to evaluate effects, 2) relevant parameters to be considered, and 3) gaps in the databases.

The criteria to evaluate effects centered on the question of the ability to lead to disease, or do the effects lead to a persistent and detectable change in cells, tissues or organs. Importantly, the question of whether the effect(s) were part of a continuum of toxicity, or simply a manifestation of a physiologic response that rapidly reversed with no permanent effect, was addressed. The overarching question of what is a low dose is imbedded in the above questions.

The subpanel considered several key parameters in their deliberations. The first and foremost parameter was the compound in question. What was known about the chemistry and biological effects of the compound as tested. Secondly, the questions focused on the specifics of the protocol such as the age and time of exposure and/or examination of the animals, the length of exposure of the animals, the dose range tested, what was the lowest dose tested, what was the lowest effect level, was there an effect at the lowest dose tested, and the species, strain, sex and number of animals tested. An important question that remained in front of the subpanel was what is a low dose. The consensus was that a low dose might be compound specific based on background exposure, body burden, or chemical class. It was generally agreed that a single number could not be used as the low dose for all compounds. The robustness of each study was considered using the above criteria. Dose-response curve models (See Figure 1) were discussed that allowed the subpanel members to characterize the general biological effects of the compounds. The subpanel developed an operational definition for "low-dose effects" that was based on the dose-response data for the selected endpoints for each agent under evaluation. Lowdose effects were considered to be occurring when a nonmonotonic dose-response resulted in significant effects below the presumed NOEL (no-observed-effect level) expected by the traditional testing paradigm.

Qualitative and quantitative data gaps were identified for all the compounds examined.

Compounds Examined:

The compounds selected for examination were considered to be representative of several classes of estrogenic xenobiotics, and estradiol. These compounds include: genistein (an isoflavone soy derivative), diethylstilbestrol [DES] (the prototypical non-steroidal estrogen), octyl-, and nonylphenols (environmental chemicals), methoxychlor [an estrogenic methoxylated derivative of p,p'-DDT] (an organochlorine insecticide) and estradiol (the primary ovarian estrogen). For each compound the subpanel made a judgement on five criteria: 1) empirical evidence for a low dose effect, 2) empirical evidence for lack of a low dose effect, 3) relevant PB/PK and/or mechanistic information, 4) overall evaluation of hormonal activity at "low doses", and 5) data gaps.

Genistein is an isoflavone derived from soy. As such genistein is a general dietary component of humans, including newborns and children, and animals. There is a large body of experimental and clinical literature relating to the biological actions of genistein. A low dose effect at 25ppm in the diet is achievable in human mammary tissue, and in mammary tissue, brain and lymphocytes of CD rats. The latter effect involved an increase in proliferation of splenic T-lymphocytes stimulated with anti-CD3. At 5ppm in the diet there was a trend toward a decrease in volume of sexually dimorphic nuclei (SDN) of the medial preoptic area (POA) of the hypothalamus in male rats that returns toward normal at 625ppm. The volume of SDN-POA is approximately 5-10 times larger in male rats than in female rats. However, in Fl male rats that had been exposed to genistein the SDN volume was intermediate between control male and control female rats. The physiological consequence of a change in volume of SDN-POA in male rat pups has to be more fully examined and elucidated. The lack of data below the 5ppm dietary concentration of genistein was considered as evidence that an effect at the lowest dose tested had been demonstrated. Little mechanistic data and/or physiologically based-pharmacokinetic (PB/PK) information were made available to the panel. However, there is a great deal of information in the open literature. The overall evaluation of genistein is that hormonal activity was demonstrated at low doses (525ppm) and the effects were seen in the CNS, mammary tissue and WBCs. There are several data gaps that should be addressed regarding genistein. There should be clarification of the activity at the lower end of the dose-response curve (<25ppm in the diet), what mechanisms are involved in genistein action (test in the ERKO and ß, and test in studies using estrogen antagonists. The subpanel was unanimous in its recommendation that the "low dose" studies must be replicated.

Diethylstilbestrol (DES) is the prototypical non-steroidal estrogen. The compound was synthesized in the mid-1930s, and introduced in the 1940s as a drug to prevent spontaneous abortions, and the late 1940s as a caponizing agent in chickens and a supplement in cattle to enhance weight gain. DES was banned for use in pregnancy and as an indirect food additive in the 1960s but is still available for some medical and veterinary uses. DES is an animal carcinogen, as well as a transplacental carcinogen in humans. There is an enormous literature on the biology and toxicology of DES. The transplacental carcinogenicity and the effects on the neonatal mouse urogenital system are unique. The most recent literature is addressing questions

of effects on the developing male reproductive system. The importance of DES is underscored in that it serves as the model for several other estrogenic chemicals. Understanding the mechanism of action of this potent drug and toxicant will lead to further insights into the actions of other like compounds. There is very clear empirical evidence of a low dose effect on prostate size at 0.02µg/kg bw in CF-1 mice and supportive evidences in CD-1 mice. Additionally there is evidence of behavioral changes in CD-1 nice at this dose. There is evidence from dose-response studies that a no effect level exists at 0.002µg/kg bw in mice. PB/PK and mechanistic information exists but was not reviewed by the subpanel. Overall, hormonal activity was observed at low doses (see above) but non-hormonal effects may exist at lower doses. Data gaps exist in several areas. The low dose effects should be remodeled. A plausible mechanism and the studies to support it should be developed to validate the findings of enlarged prostates in treated males. Additionally, a META analysis should be conducted on the four major studies that have been completed on DES.

Alkylphenols are industrial compounds that have been identified in drinking water supplies and wastewaters. Several investigators have demonstrated the estrogenic potential of this class of compounds in vitro and in vivo. Two compounds were evaluated by the subpanel; nonylphenol and octylphenol. The "low dose" of nonylphenol is questionable. Renal toxicity occurred at the 200ppm in the parent generation of SD-rats of a multi-generation study, and increased relative uterine weight occurred at the same concentration in Fl females. Higher doses (2000 ppm) induced changes in testes and prostate weight, decreased live births and prolonged estrus cyclity. Several other changes occurred in the F1 offspring of the 25ppm group including an increase in proliferation of splenic T-lymphocytes stimulated with anti-CD3, an increased relative thymus weight, a decreased volume of SDN-POA in males, and a prolonged estrus in females. A pattern of change in SDN-POA volume in males following treatment with nonylphenol was similar to that observed for genistein. As stated above the SDN-POA changes are difficult to interpret at the moment, as are the anti-CD3 findings. The lack of a low dose effect appears to be at approximately 5ppm, but the immune markers have not been tested at that level. There was no PB/PK information available to the subpanel on nonylphenol. The data gaps for nonylphenol are similar to those stated above. What is the meaning and mechanistic basis for the changes in SDN-POA? Immune-markers should be evaluated at levels less than 25 ppm. More data are needed at the level of ng/kg/day, the human exposure level.

Octylphenol is another member of the class of alkylphenols. The compound is primarily used as an intermediate for the production of surfactants. The literature for low dose evaluation of this compound is limited to one major study. There was no evidence of a low dose effect in a five dose multigeneration study in rats. Only the highest dose (2000 ppm) induced toxicological changes. At doses ranging from 0.02-200ppm no effects were observed. Little or no PB/PK or mechanistic data exists for this compound. The binding of octylphenol to the ER is 3 to 7 orders of magnitude less than estradiol to the same receptor. Hence, there is no evidence that octylphenol induces hormonal activity at low doses. The only data gap identified by the subpanel was the absence of a confirmatory study in another laboratory species.

Methoxychlor is a chlorinated diphenylethane insecticide that is chemically related to p,p'-DDT. The major difference between methoxychlor and DDT is the substitution of methoxy-groups for the chlorines in the para-positions of the phenyl rings. The estrogenicity of methoxychlor in rodents has been known for several decades. The compound binds to the ER with a relatively low affinity compared to estradiol, but can induce uterotrophism in immature rodents. Despite the many studies on the mechanism(s) of action of methoxychlor few studies detailed the dose-response relationship for estrogenic action. The primary multi-dose study with methoxychlor examined six doses ranging from 0.05 to 150mg/kg/d given shortly before birth and through neonatal day 7, at which time the pups were dosed directly. Effects were seen at all doses with the exception of the 0.5 and below. At the effective doses there were a wide range of changes in estrogen sensitive organs. The lowest effective dose was 5/mg/kg/d. Hence it appears that at levels less than 5mg/kg/d (0.05 and 0.5mg/kg/d) none of the effects seen at higher doses were reported. Mechanistic studies have been carried out in several laboratories. Serum concentrations and mills concentrations mimic the administered doses. Additionally, the active metabolites also mirror the parent compounds in relative concentrations. Overall the classic estrogenic activity is limited to doses greater than 5mg/kg/d, but some immune effects (increase in proliferation of splenic T-lymphocytes stimulated with anti-CD3 in Fl female SD rats) have been reported at 10ppm in the diet. Data gaps that should be addressed to complete the profile on methoxychlor are a further evaluation on the anti-CD3 alterations. Are these changes related to estrogen action or are other pathways affected. A second data gap, which may be more important, is the comparison between technical grade and pure methoxychlor.

Estradiol is the ovarian steroid with the greatest estrogenic activity. It binds the ER with the greatest affinity of the compounds evaluated and induces all the classic effects that are termed estrogenic. The basic biology of estradiol defines the feedback pathways of the steroid-driven endocrine system. Estradiol has been widely studied and the molecular mechanisms of action are very well understood. The studies evaluated by the Subpanel addressed the question of lowdose activity keeping in mind that this is an extremely potent steroid that is difficult to evaluate in an in vivo system because of the homeostatic mechanisms of the test systems. The ovariectomized CD rat in the Tier 1 (EDSTAC Protocol) studies had reproducible changes in serum prolactin at 3µg/kg/d with an associated increase with administered dose. Several other changes in hormonally active tissue were reported, as was changes in LH and FSH as a function of dose and blood level. An apparent no effect level was attained in a 90-day feeding study with no changes being observed at the three lowest doses tested (3, 170 and 700µg/kg/d). Mechanistic studies were not conducted in this bioassay. However, the best metric for endocrine changes was the blood estradiol level rather than administered dose. Overall, the TIER1 dietary study demonstrated the three types of dose-response relationships based on particular endpoints, tissue responses and time points. The remaining data gap for estradiol in this test system is to determine the shape of the dose-response curve at the low effect and high no effect levels.

GENERAL RECOMMENDATIONS

Based on the studies and compounds reviewed the subpanel has three major recommendations that cross most studies: 1) Model dose-response relationships - in multiple dose studies, modeling of doseresponse relationships should be done in addition to pair-wise comparisons; 2) Replicate and validate studies - low dose effects are difficult to ascertain, hence the studies must be replicated and validated across laboratories;

3) Examine and elucidate the physiological and toxicological consequences of SDN-POA and antiCD3 changes - determine the biological significance of the volume changes in the SDN-POA in male rodents, and the significance and relationship of the anti-CD3 changes and estrogen action.

Figure 1 - Dose Response Curves

Chapter 3:
Report of the Androgens and Antiandrogens Subpanel

Introduction

Chair
Shuk-Mei Ho, University of Massachusetts Medical School

Rapporteur
Terry Brown, Johns Hopkins University School of Public Health

Facilitator
Eisuke Murono, National Institute for Occupational Safety and Health

Panelists
George Daston, The Procter & Gamble Company
Mitch Eddy, National Institute of Environmental Health Sciences
Lorenz Rhomberg, Gradient Corporation
Elizabeth Wilson, University of North Carolina at Chapel Hill

Statistics and Dose-Response Modeling Subpanel Representative
Ralph Kodell, National Center for Toxicological Research (statistics)

This article highlights major issues discussed in the Androgens and Antiandrogens Subpanel at the 2000 Endocrine Disrupters-Low Doses Peer Review Workshop. The goal of the Workshop was to review the scientific evidence related to potential low-dose effects of endocrine active chemicals (EACs) on human health. The Subpanel was charged to examine data from a selected study on vinclozolin (V), a fungicide with antiandrogenic activity 1, and other supporting information 2-21 in order to determine whether the body of evidence demonstrated low-dose androgenic/antiandrogenic effects or the lack of such. Specifically, it was asked to evaluate the consistency and strength of the scientific evidence presented in the study, including parameters such as timing of exposure, sensitivity of the endpoints, sample sizes, sampling methods, appropriateness of the controls, strain differences, species sensitivity, number of doses in the lowdose range, and the vigor of the statistical analyses. Other main topics of discussion included the shape of the dose-response curve for V, and other relevant mechanistic data that might strengthen or weaken the plausibility of its low-dose effects. During the Subpanel's deliberation, opinions on whether the weight of evidence provided sufficient grounds to change the traditional dose-setting paradigm, particularly in the low-dose region, for mammalian toxicology studies on V and related compounds, were reflected. Finally, knowledge gaps regarding environmentally active androgenic/antiandrogenic agents were identified.

Defining "low-dose effects"

In order to assess whether low-dose effects exist, it is imperative to first define what constitutes low dose for each EAC under investigation. The organizers of the conference, NTP and NIEHS, had asked the Subpanels to consider "low dose effect as biological changes that occur at environmentally relevant exposure levels or at doses that are lower than those typically used in EPA's standard toxicity testing paradigm". With regard to V, information on environmentally relevant exposure levels in human populations is currently unavailable. However, the no-observable-adverse-effect level (NOAEL) and low-observable-adverse-effect-level (LOAEL) for acute dietary exposure to V has been established to be 6 mg/kg/day and 11.5 mg/kg/day, respectively, while the chronic dietary NOAEL and LOAEL are set at 1.2 mg/kg/day and 2.3 mg/kg/day, respectively (revised Human Health Risk Assessment 5-12-00, Office of Pesticide Program). These values have been established from rat studies, using developmental and ventral prostate (VP) weight changes as endpoints in acute exposure studies and histopathological lesions of lungs, liver, ovary, and eye as adverse effects in chronic exposure studies. Human risk assessment levels have been derived from these values after adjustment for human factors. In order to facilitate further discussion the Subpanel had arbitrarily defined "lowdose range" as one below the currently recognized NOAEL/LOAEL. However, the Subpanel was aware of the fact that no studies had been conducted in dose-ranges below NOAEL/LOAEL for V. The issue of whether studies conducted at NOAEL/LOAEL should be considered as lowdose studies or only those carried out at dose ranges substantially lower than NOAEL/LOAEL be counted was debated. At present, no information is available on environmentally relevant exposure such as exposure levels from crop residue. This data gap has hampered current and future hazard evaluation, risk assessment, toxicity testing and risk management.

Vinclozolin is an antiandrogen

Vinclozolin (V, 3-(3,5-dichlorophenyl)-5-methyl-5-vinyl-oxazolidine-2, 4-dione) is a dicarboximide fungicide widely used to control fungal growth on several fruits, vegetables, and turfgrass 22. Structurally, it resembles hydroxyflutamide and exerts antiandrogenic action 20, 23. Prenatal exposure to V causes reduced anogenital distance (AGD), hypospadias, ectopic testes, viginal pouch formation, agenesis of the ventral prostate, and nipple retention in male offspring. Exposure of pregnant rats to low dosages of V, 12 mg/kg/d or lower, results in female-like AGD, retained nipples, and permanently reduced VP weights in some male offspring, while a high dose of 100 mg/kg/d causes hypospadias, deformity, and infertility in all male offspring. In contrast, fertility and reproductive functions are unaffected when male rats are exposed as adults to high doses of Vin a chronic manner 24. Mechanistically, the antiandrogenic action of V is mediated via metabolic conversion to two open-ring metabolites, Ml and M2, which have been shown to induce AR nuclear import, compete with androgen for AR binding, and block AR-DNA interaction 20. They do not bind estrogen receptor (ER)- nor inhibit 5a-reductase activity, yet do possess weak affinity for progesterone receptors. Biologically, M2 is more potent than M1, and has a Ki around 1 pM, as compared to hydroxyflutamide, which has a Ki of 0.1 nM, for the AR. In contrast, V, the parent chemical, is a poor AR antagonist 20. Dosimetry studies reveal that when M1 and M2 in maternal serum concentrations approach their respective Ki values for AR binding hypospadias and more severe infertility are noted in the exposed offspring 25.

The Vinclozolin data set

The study by Gray and associates1 was the only article selected for peer-review. The principal investigator had submitted raw data to the Statistics and Dose-Response Modeling Subpanel for re-evaluation and has responded to the 23 questions listed in the "Issues Relative to the Evaluation of Low-Dose Studies". In this study, pregnant rats were dosed (Pc) with V at 0, 3.125, 6.25, 12.5, 25, 50, or 100 mg/kg/day from gestational day (GD) 14 to postnatal day (PND) 3, and postnatal reproductive developmental abnormalities, including female-like AGD, retained nipples, cleft phallus with hypospadias, suprainguinal ectopic scrota/testes, a vaginal pouch, epididymal granulomas, and small to absent VP, in the male offspring were used as endpoints. The investigators reported that the AGD was significantly reduced in newborn male offspring and the incidence of areolas was increased. VP weight in one-year-old males was reduced and permanent nipples were observed in male offspring born to mothers treated with all tested doses. Males born to mothers exposed to the two higher doses (50 and 100 mg/kg/day) exhibited reproductive tract malformations and reduced ejaculated sperm numbers. Different endpoints displayed varied dose-response curves. For example, AGD, areolas, and VP weight displayed a continuous response to V treatment with no apparent threshold while hypospadias and ectopic testes exhibited threshold-responses.

Statistical re-evaluation revealed that statistical methods applied in the study were generally appropriate. The Statistic Subpanel agreed with the investigators' decision to use the litter as the basic experimental unit, given the significant "litter effects" that were present in the data. It also confirmed that at postnatal day2, AGD in male offspring was significantly reduced in the group exposed to the lowest test dose (3.125 mg/kg), but it did not find AGD reduction to be significant for the next lowest dose (6.25 mg/kg/day) group. Additionally, it agreed with the investigators that the 50 and 100 mg/kg/day V doses significantly reduced VP weight, the 100 mg/kg/day dose diminished seminal vesicle weight, and none of the tested doses affected body weights, testis weight, paired epididymis weight, or testicular and epididymal sperm counts.

However, the Subpanel's analyses disagreed with the investigators' finding that an effect of 6.25 mg/kg/day V on VP weight was significant.

Based on peer-review of the investigators' data and the Statistic Subpanel's re-evaluation, our Subpanel noted the following: 1) the study was well-designed and demonstrated V exerted antiandrogenic action developmentally, 2) the test doses were in the NOAEL/LOAEL range, 3) alteration in AGD in male offspring was the most sensitive endpoint and was significantly different at the lowest test dose (3.125 mg/kg), 4) the areolar/nipple persistence in males was a highly informative endpoint but was not subjected to statistical re-analysis, and 5) for some endpoints tested, the dose-response curves displayed no threshold and appeared linear to the lowest test dose, which simultaneously approached the limit of detection.

Other antiandrogenic EACs and their mechanisms of action

Other EACs with antiandrogenic activities have been identified10. According to their mechanisms of action, they could be broadly classified into AR antagonists, 5a-reductase inhibitors, and inhibitors of steroidogenesis.

In addition to V, several EACs are known to exert antiandrogenic activities through their action as AR antagonists. The p,p'-DDE, a persistent metabolite of the insecticide p,p'-DDT, binds AR with moderate affinity and inhibits androgen-induced transcription with a potency similar to that reported for the antiandrogen hydroxyflutamide 16. Prenatal exposure (gestational day 14-18) to 100/kg/day of p,p'-DDE induced AGD reduction and nipple retention in male offspring 10. Other metabolites of p,p'-DDT have also been reported to exhibit antiandrogenenic activities in a cell culture system 26. Similarly, 2,2-bis-phydroxyphenol-1, 1, 1-trichloroethane, the primary o-demethylated metabolite of the pesticide methoxychlor, is an effective AR antagonist 26. It exerts potent antiandrogenic action in vivo. The fungicide procymidone also acts as an AR antagonist 1'°1s. At a dose of 100 mg/kg/day, it produces in vivo antiandrogenic activities similar to those induced by V and p,p'-DDE 1°°1'°1s. Linuron, a herbicide, is a weak competitive inhibitor for the AR (Ki of 100 mM) although it is highly effective in altering sexual differentiation in vivo in an antiandrogenic manner 1°. Short-term treatment of castrated adult rats with linuron reduces testosterone- and DHT-dependent tissue weights in the Hershberger assay 13 at an oral dose of 100 mg/kg/day for 7 days. It has been suggested that the principal antiandrogenic action for linuron is mediated via inhibition of steroidogenesis 27. Likewise, the fungicide ketoconazole has been shown to exert antiandrogenic activity in vivo by lowering serum testosterone levels, and altering both gonadal synthesis and hepatic inactivation of testosterone Zs. Lastly, the phthalates, common plasticizers, are now believed to exert their antiandrogenic action via interference with the synthesis or metabolism of androgens 29.

Pharmaceuticals (e.g. flutamide and Casodex) developed to treat prostate cancers and benign prostatic hyperplasias (BPH) are potent antiandrogens. Flutamide, following in vivo hydroxylation to hydroxyflutamide acts as an AR antagonist. It binds to AR with high affinity but fails to initiate transcription 30. Casodex (1C1 176,334) binds to the AR with good affinity, but fails to induce receptor accessory protein dissociation, DNA binding, and transcriptional activation 31, 32. In a 5-day Hershberger assay 13, flutamide at 0.15, 0.6, 2.5 and 10 mg/kg/day effectively produces antiandrogenic responses 21. In contrast, finasteride, developed to treat BPH and hair loss, exerts its antiandrogenic action primarily by acting as a 5a-reductase inhibitor 33, 34. Gestational exposure to finasteride produces transient AGD reduction and nipple retention, and increases hypospadias incidence in male rat offspring 6, 7. It has been noted that the decrease in AGD apparently shows a linear response over the tested dose range. Yet, the hypospadias response does exhibit a threshold around 0.1 mg/kg/day and a 100% effect level at 100 mg/kg/day (with dosing through Day 20 of gestation) 6. Interestingly, unlike AR antogonists, finasteride does not completely block prostate differentiation or feminize the external genitalia despite high-dose exposure 15. Finally, the non-steroidal antiandrogen, nilutamide, has a weak binding affinity for the AR. Nonetheless, it has a long biological half-life and potent antiandrogenic activity in vivo. Its action is likely due to its inhibitory action on androgen synthesis 35-37.

When the effects of V are compared to those induced by the aforementioned EACs it becomes apparent that these compounds all behave as antiandrogens. Thus far, there is little evidence for environmental chemicals that act as androgen mimics. Although examples of potential androgenic effects in wildlife have been reported 3s presently no evidence exists for effects in humans related to environmental exposures. If androgenic mimics were to be present in the environment, they most likely will be detected by their effects on female development and reproductive functions. In this regard, it has been reported that exposure of pregnant rats to testosterone adversely affects pregnancy and masculinizes female offspring 39.

Mechanistic models for screening environmental antiandrogens

In order to safeguard detection of low-dose effects caused by environmental antiandrogens, it is important to employ appropriate mechanism-based models with the most sensitive endpoints for detection. Existing multigenerational tests sample only a small number of pups for necropsy and have missed malformations and low-dose effects of EACs. The problem is more serious when the endpoints are low frequency events. For example, multigenerational studies with DDT have failed to detect the androgenic effects of p,p'-DDE in rats, mice and beagle dogs 40. With regard to detection of environmental androgen/antiandrogen three mechanism-based in vivo screening models, if utilized routinely and complementarily, should help to improve detectability. The gestational/postnatal exposure rat model system is based the ability of an EAC to interfere with low levels of endogenous androgen required for normal male development. This is recognized as a highly sensitive in vivo assay. The developmental endpoints such as AGD reduction, areolar/nipple persistence, preputial separation, hypospadias, testicular descent are highly sensitive in detecting antiandrogenic activities 29. Mechanism-wise, the in vivo antiandrogenic activity is, in general, supported by AR binding and transcriptional activity assays. Furthermore, the EAC's Ki for AR binding usually agrees with the relative potency of its antiandrogenic action 20. The Hershberger assay 13, a short-term in vivo assay, also has the sensitivity and specificity for detecting androgen/antiandrogen. Castrated mature rats are exposed to the chemical in the absence of presence of an androgen and accessory sex organ weights are used as detection endpoints. Androgenic activity is detected on the basis of stimulation of accessory sex organ weight gain in the absence of an androgen and antiandrogenic activity is measured as competitive inhibitory action on androgen-stimulated growth of these organs. Lastly, the peripubertal exposure model is another in vivo system that permits detection of androgenic/antiandrogenic activity based on preputial separation. In addition to in vivo assays, numerous in vitro or cell-free androgen/antiandrogen-screening assays have been developed. These assays are based on hormone specific mechanisms of action such as cell-free and whole-cell AR binding, androgen-dependent cell proliferation, and transcriptional activation of androgen-specific reporters or genes. Utilization of these assays as first-tier tests or as mechanism-finding assays should complement findings from in vivo assays 29, 41.

Comments and recommendation on current and future study design

Based upon existing knowledge of mechanisms, the current study designs are found to have the sensitivity and selectivity for detecting environmental antiandrogen, particularly if these compounds behave as competitive AR antagonists. However, protocol modifications may be required for the detection of EACs that act(s) as inhibitors of steroid biosynthesis and/or metabolism modulators. The Subpanel has found little additional benefit to extend the current standard toxicology tests to dose range below the currently set NOAEL/LOAEL. This conclusion is reached partially based on the low incidence of malformation detected by the current assays at NOAEL/LOAEL, suggesting that the detection limits of these assays may approximate these levels. Furthermore, based on our present understanding of the mechanisms of action for these detection assays we recognized that current data are obtained based upon environmental chemicals acting as antiandrogens, i.e. as agents that interfere with endogenous testosterone action. In intact adult animals, testosterone is present at high levels and therefore the effect of an antiandrogenic agent is only observed if it is present at high concentrations. Similarly, in castrated adult rats, detection of an antiandrogenic compound is based on its efficacy to block the action of an exogenously administered, potent androgen (testosterone or dihydrotestosterone). However, in gestational/prenatal exposure assays, antiandrogenic activity is detected as the efficacy of the EAC to antagonize the action of low levels of endogenous testosterone. Therefore, these assays are deemed to have higher sensitivities than the nondevelopmental assays. Yet, they are still dependent on the Kis of the EACs, which likely approach the exposure levels induced by the NOAEL/LOAEL.

Recognizing that knowledge gaps exist in this area of studies, the following recommendations have been made. First, since only one study has been conducted in the range of the NOAEL/LOAEL, verification and substantiation of data from this study with independent (unrelated and multiple) investigations should be a top priority for immediate future research. Furthermore, since no studies had been conducted at dose-ranges below the NOAEL/LOAEL for vinclozolin and other environmental antiandrogens a need exists to test the hypothesis that the dose-response for antiandrogens is linear to the NOAEL/LOAEL. Secondly, although so far no EACs have been identified as androgen mimics for the human, it is necessary to develop mechanism-based assays for their detection since this class of compounds apparently affects wildlife. Thirdly, research can be focused on further advancement of the basic knowledge on the mechanisms of androgenic/antiandrogenic action. These efforts should benefit future development of new detection methodologies. Likewise, as new molecular markers of tissue response are identified (by genomics or proteomics discovery platforms), it will be useful to include such molecular/biochemical biomarkers as endpoints since they may be more sensitive or specific than current biological endpoints. With the advent of bioinformatics and large-scale molecular modeling it becomes attractive and cost-effective to utilize these new approaches to analyze currently available and future biological data in order to formulate novel hypotheses to be tested under new experimental paradigms. Along this line of argument, since the AR ligand-binding domain (LBD) has been crystallized it offers new opportunities for structural modeling of the AR-LBD-ligand crystals to predict androgenic/antiandrogenic activities of various chemicals. One important issue that future research has to address is the dose-response relationships for androgenic/antiandrogenic EACs in different species and in multiple strains. This issue is critical to our understanding of sensitive populations, encompassing such factors as genetic predisposition, age, gender, past and current exposure history, dietary influences and multiple chemical sensitivity. Modeling in animal studies may prove to be fruitful endeavors preluding large-scale epidemiology studies. Lastly, it may be useful to develop credible dosimetry/mechanistic models for exposures occurring during in utero and early neonatal development since these are the most sensitive timepoints of detection.

References

1. Gray, L. E., Jr., Ostby, J., Monosson, E., and Kelce, W. R. Environmental antiandrogens: low doses of the fungicide vinclozolin alter sexual differentiation of the male rat. Toxicol. Ind.Health, 15: 48-64, 1999.

2. Health effects test guidelines OPPTS 870.3800 Reproduction and fertility effects. US EPA:Prevention, Pesticides and Toxics Substances (7101). EPA 712-C-98-208. 1998. Ref Type: Report

3. Anderson, C. A. and Clark, R. L. External genitalia of the rat: normal development and the histogenesis of 5 alpha-reductase inhibitor-induced abnormalities. Teratology, 42: 483-496, 1990.

4. Ashby, J. and Lefevre, P. A. The peripubertal male rat assay as an alternative to the Hershberger castrated male rat assay for the detection of antiandrogens, oestrogens and metabolic modulators. J.Appl.Toxicol., 20: 35-47, 2000.

5. Calabrese, E. J. and Baldwin, L. A. A general classification of U-shaped dose-response relationships in toxicology and their mechanistic foundations. Hum.Exp.Toxicol., 17: 353364, 1998.

6. Clark, R. L., Antonello, J. M., Grossman, S. J., Wise, L. D., Anderson, C., Bagdon, W. J., Prahalada, S., MacDonald, J. S., and Robertson, R. T. External genitalia abnormalities in male rats exposed in utero to finasteride, a 5 alpha-reductase inhibitor. Teratology, 42: 91100, 1990.

7. Clark, R. L., Anderson, C. A., Prahalada, S., Robertson, R. T., Lochry, E. A., Leonard, Y. M., Stevens, J. L., and Hoberman, A. M. Critical developmental periods for effects on male rat genitalia induced by finasteride, a 5 alpha-reductase inhibitor. Toxicol.Appl.Pharmacol., 119: 34-40, 1993.

8. Ewing, L. L., Desjardins, C., Irby, D. C., and Robaire, B. Synergistic interaction of testosterone and oestradiol inhibits spermatogenesis in rats. Nature, 269: 409-411, 1977.

9. Gray, L. E., Jr., Ostby, J. S., and Kelce, W. R. Developmental effects of an environmental antiandrogen: the fungicide vinclozolin alters sex differentiation of the male rat. Toxicol.Appl.Pharmacol., 129: 46-52, 1994.

10. Gray, L. E., Jr., Wolf, C., Lambright, C., Mann, P., Price, M., Cooper, R. L., and Ostby, J. Administration of potentially antiandrogenic pesticides (procymidone, linuron, iprodione, chlozolinate, p,p'-DDE, and ketoconazole) and toxic substances (dib. Toxicol.lnd.Health, 15: 94-118, 1999.

11. Greco, T. L. and Payne, A. H. Ontogeny of expression of the genes for steroidogenic enzymes P450 side- chain cleavage, 3 beta-hydroxysteroid dehydrogenase, P450 17 alpha- hydroxylase/C 17-20 lyase, and P450 aromatase in fetal mouse gonads. Endocrinology, 135: 262-268, 1994.

12. Greene R.R and Burrill M.W., e. a. Experimental intersexuality. The effect of antenatal androgens on sexual development of female rats. The American Journal of Anatomy, 65(3): 415-469, 1938.

13. Hershberger LG and Shipley EG, e. a. Myotrophic activity of 19-Nortestosterone and other steroids determined by modified levator ani muscle method. Proceedings of the Society for Experimental Biology and Medicine, 83: 175-180, 1953.

14. Hosokawa, S., Murakami, M., Ineyama, M., Yamada, T., Yoshitake, A., Yamada, H., and Miyamoto, J. The affinity of procymidone to androgen receptor in rats and mice. J.Toxicol.Sci., 18: 8393, 1993.

15. Imperato-McGinley, J., Sanchez, R. S., Spencer, J. R., Yee, B., and Vaughan, E. D. Comparison of the effects of the 5 alpha-reductase inhibitor finasteride and the antiandrogen flutamide on prostate and genital differentiation: dose-response studies. Endocrinology, 131: 1149-1156, 1992.

16. Kelce, W. R., Stone, C. R., Laws, S. C., Gray, L. E., Kemppainen, J. A., and Wilson, E. M. Persistent DDT metabolite p,p'-DDE is a potent androgen receptor antagonist [see comments]. Nature, 375: 581-585, 1995.

17. Monosson, E., Kelce, W. R., Lambright, C., Ostby, J., and Gray, L. E., Jr. Peripubertal exposure to the antiandrogenic fungicide, vinclozolin, delays puberty, inhibits the development of androgen-dependent tissues, and alters androgen receptor function in the male rat. Toxicol. Ind.Health, 15: 6579, 1999.

18. Ostby, J., Kelce, W. R., Lambright, C., Wolf, C. J., Mann, P., and Gray, L. E., Jr. The fungicide procymidone alters sexual differentiation in the male rat by acting as an androgen-receptor antagonist in vivo and in vitro. Toxicol.lnd.Health, 15: 80-93, 1999.

19. Robaire, B., Ewing, L. L., Irby, D. C., and Desjardins, C. Interactions of testosterone and estradiol- 17 beta on the reproductive tract of the male rat. Biol.Reprod., 21: 455-463, 1979.

20. Wong, C., Kelce, W. R., Sar, M., and Wilson, E. M. Androgen receptor antagonist versus agonist activities of the fungicide vinclozolin relative to hydroxyflutamide. J.Biol.Chem., 270: 1999820003, 1995.

21. Yamada, T., Kunimatsu, T., Sake, H., Yabushita, S., Sukata, T., Okuno, Y., and Matsuo, M. Comparative evaluation of a 5-day Hershberger assay utilizing mature male rats and a pubertal male assay for detection of flutamide's antiandrogenic activity. Toxicol.Sci., 53: 289-296, 2000.

22. Sannino, A., Bandini, M., and Belzoni, L. Multiresidue determination of 19 fungicides in processed fruits and vegetables by capillary gas chromatography after gel permeation chromatography. J.AOAC Int., 82: 1229-1238, 1999.

23. Kelce, W. R., Lambright, C. R., Gray, L. E., Jr., and Roberts, K. P. Vinclozolin and p,p'-DDE alter androgen-dependent gene expression: in vivo confirmation of an androgen receptor-mediated mechanism. Toxicol.Appl.Pharmacol., 142: 192-200, 1997.

24. Fail, P. A., Pearce, S. W., Anderson, S. A., Tyl, R. W., and Gray, L. E. Endocrine and reproductive toxicity of vinclozolin (vin) in male Long-Evans Hooded rats. Toxicologist, 15:293, 1995.

25. Kelce, W. R., Monosson, E., Gamcsik, M. P., Laws, S. C., and Gray, L. E., Jr. Environmental hormone disruptors: evidence that vinclozolin developmental toxicity is mediated by antiandrogenic metabolites. Toxicol.Appl.Pharmacol., 126: 276-285, 1994.

26. Maness, S. C., McDonnell, D. P., and Gaido, K. W. Inhibition of androgen receptor-dependent transcriptional activity by DDT isomers and methoxychlor in HepG2 human hepatoma cells. Toxicol.Appl.Pharmacol., 151: 135-142, 1998.

27. Lambright, C., Ostby, J., Bobseine, K., Wilson, V., Hotchkiss, A. K., Mann, P. C., and Gray, L. E., Jr. Cellular and molecular mechanisms of action of linuron: an antiandrogenic herbicide that produces reproductive malformations in male rats. Toxicol.Sci., 56: 389-399, 2000.

28. Wilson, V. S. and LeBlanc, G. A. The contribution of hepatic inactivation of testosterone to the lowering of serum testosterone levels by ketoconazole. Toxicol.Sci., 54: 128-137, 2000.

29. Gray, L. E., Jr. Xenoendocrine disrupters: laboratory studies on male reproductive effects. Toxicol.Lett., 102-103: 331-335, 1998.

30. Kemppainen, J. A., Lane, M. V., Sar, M., and Wilson, E. M. Androgen receptor phosphorylation, turnover, nuclear transport, and transcriptional activation. Specificity for steroids and antihormones. J.Biol.Chem., 267: 968-974, 1992.

31. Wakeling, A. E. Steroid antagonists as nuclear receptor blockers. Cancer Surv., 14: 71-85, 1992.

32. Freeman, S. N., Mainwaring, W. 1., and Furr, B. J. A possible explanation for the peripheral selectivity of a novel non- steroidal pure antiandrogen, Casodex (1C1 176,334). Br.J.Cancer, 60: 664668, 1989.

33. Wilson, J. D., Griffin, J. E., and Russell, D. W. Steroid 5 alpha-reductase 2 deficiency. EndocrRev., 14: 577-593, 1993.

34. Li, X., Chen, C., Singh, S. M., Labrie, F., and Labire, F. The enzyme and inhibitors of 4ene-3oxosteroid 5 alpha-oxidoreductase [published erratum appears in Steroids 1995 Aug;60(8):540]. Steroids, 60: 430-441, 1995.

35. Gaillard-Moguilewsky, M. Pharmacology of antiandrogens and value of combining androgen suppression with antiandrogen therapy. Urology, 37: 5-12, 1991.

36. Harris, M. G., Coleman, S. G., Faulds, D., and Chrisp, P. Nilutamide. A review of its pharmacodynamic and pharmacokinetic properties, and therapeutic efficacy in prostate cancer. Drugs Aging, 3: 9-25, 1993.

37. Dole, E. J. and Holdsworth, M. T. Nilutamide: an antiandrogen for the treatment of prostate cancer [see comments]. Ann.Pharmacother., 31: 65-75, 1997.

38. Tyler CR., Jobling S, and Sumpter J.P. Endocrine disruption in wildlife: a critical review of the evidence. Crit.Rev.Toxicol. 28, 319-361. 1998. Ref Type: Abstract

39. Wolf, C., Ostby, J., Hotchkiss, A. K., and Gray, L. E. Effects of prenatal testosterone propionate on the development of male and female rats: A dose-response study. The 33rd Annual Meeting of the Society for the Study of Reproduction. Biology of Reproduction 62, 247. 2000. Ref Type: Abstract

40. Ware, G. W. Effects of DDT on reproduction in higher animals. ResidueRev., 59: 119140, 1975.

41. Gray, L. E., Jr., Kelce, W. R., Wiese, T., Tyl, R., Gaido, K., Cook, J., Klinefelter, G., Desaulniers, D., Wilson, E., Zacharewski, T., Waller, C., Foster, P., Laskey, J., Reel, J., Giesy, J., Laws, S., McLachlan, J., Breslin, W., Cooper, R., Di Giulio, R., Johnson, R., Purdy, R., Mihaich, E., Safe, S., and Colborn, T. Endocrine Screening Methods Workshop report: detection of estrogenic and androgenic hormonal and antihormonal activity for chemicals that act via receptor or steroidogenic enzyme mechanisms. Reprod.Toxicol., 11: 719-750, 1997.

source: http://ntp-server.niehs.nih.gov/htdocs/liason/LowDoseWebPage.html 24jul01

If you have come to this page from an outside location click here to get back to mindfully.org