- Email: [email protected]
Testing for endocrine disruption post-EDSTAC: extrapolation of low dose rodent effects to humans
Toxicology Letters 120 (2001) 233– 242 www.elsevier.com/locate/toxlet
Testing for endocrine disruption post-EDSTAC: extrapolation of low dose rodent effects to humans John Ashby * Syngenta Central Toxicology Laboratory, Alderley Park, Macclesfield, Cheshire SK10 4TJ, UK
Abstract The study of chemically-induced endocrine disruption in mammals is a relatively new field of endeavour, and it has been assailed by an unusual level of disagreement among investigators regarding the developmental effects produced by chemicals in animals. This article discusses the several sources of uncertainty in endocrine toxicity studies, and the intrinsic variability of many of the key experimental parameters. It is concluded that current uncertainties regarding extrapolation of rodent effects to humans are due to the absence of an extensive agreed rodent control database for the developmental parameters under study, coupled to the established intrinsic variability of these parameters between strains/species of test animals and test protocols. Only when these factors are generally accepted and well studied will it be possible to relate effects seen in rodents to humans. © 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Endocrine disruption; EDSTAC; Bisphenol A; Diethylstilboestrol; Phytoestrogens
1. Endocrine disruption as a new toxicological concern
2. Sources of uncertainty and variability in rodent endocrine disruption studies
Validation of the several assays discussed by the US EDSTAC initiative is now well in train. However, before data derived from such rodent assays can be interpreted with confidence, it will be necessary to focus a range of problems, as discussed herein.
The study of chemicals for endocrine disrupting (ED) properties is complicated by the many variables, as shown in Fig. 1. Currently, most routine rodent toxicological assessments are conducted using oral gavage of the chemical to young adult animals, with clear adverse effects being the assay endpoint. However, the lateral trends shown in Fig. 1 are increasingly followed, usually without an explanation of why the particular constellation of experimental conditions chosen were selected. Some of these trends are justifiable, while others are not. Lack of attention to such methodological details makes it difficult to reconcile different ED assay outcomes for the same chemical. This po-
* Tel.: +44-1625-582828; fax: +44-1625-590249. E-mail address: [email protected] (J. Ashby).
0378-4274/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 3 7 8 - 4 2 7 4 ( 0 1 ) 0 0 2 9 9 - 5
234
J. Ashby / Toxicology Letters 120 (2001) 233–242
Fig. 1. Transitions facing toxicology as studies move from the bold words along the four lateral arms. Fig. 4. Prostate weights at post-natal day 31 for control and NP-treated SD rats. Data taken from Lee (1998) * Statistical significance for NP versus C (PB 0.05) taken from Lee (1998). †† Statistical significance for C (13– 20) versus all other controls (P B 0.01), as reported by Odum and Ashby (2000).
Fig. 2. Dose– response curves for genistein in immature AP mouse uterotrophic assays following exposure by either s.c. injection or oral gavage in corn oil. The studies were conducted as described earlier with RM1 as diet (Tinwell et al., 2000). ** P B0.01.
tential source of confusion could escalate with the growing use of gene microarray DNA technology coupled to the use of stable cell lines. An example of the importance of some of these experimental
variables is shown by the different uterotrophic assay responses to the phyto-oestrogen genistein produced by changing the route of administration to the rodents (Fig. 2). The relevance of the oral route of exposure to consideration of the properties of phyto-oestrogens in diet is obvious, but it is difficult to relate the assay response derived using the subcutaneous (s.c.) injection route to humans, despite it being the stronger of the two responses. The strongest assay response may not always be the most relevant response for human or wildlife risk assessment purposes. An equally important consideration is realisation of the hierarchy of assay sensitivities. Another source of genuine variability is illustrated by the data of Gray et al. (1999), who have
Fig. 3. Effect of three different commercial rodent diets on uterine weight in immature AP rats (Ashby et al., 2001; unpublished observations), and on the sexual maturation of the female animals. Inhibition of the uterotrophic response was observed following concomitant exposure to the GnRH antagonist antarelix (300 mg/kg daily; Deghenghi et al., 1993). * P B 0.05; ** P B 0.01.
J. Ashby / Toxicology Letters 120 (2001) 233–242
235
3. Diet and endocrine disruption
Fig. 5. Response of male rats at pnd 90 following their exposure in utero to OP or BBP in diet (1 p.p.m.). The control testes weight and the varying responses to OP and BBP are discussed in the text.
shown that the induction of adverse effects in neonatal rats, whose mothers were treated with an anti-androgen, obey the following sensitivity hierarchy: changes in ano-genital distance\induction of hypospadias\ induction of ectopic testes. The establishment of a range of sensitivity hierarchies for individual assays will do much to reduce the incidence of apparent disagreements in the literature regarding the ED activities of individual chemicals.
The possible role of diet in the aetiology of changes in the onset of human puberty has been long recognised. However, the role of diet in modulating the outcome of rodent endocrine disruption studies has yet to be studied. The potential importance of this topic is illustrated by the data shown in Fig. 3. This illustrates that three different rodent diets have different oestrogenic properties as assayed using the immature rat uterotrophic assay or in assays to determine the time of puberty in the rats (age of vaginal opening and first oestrus). It is pertinent to note here that the test protocol used by Nagel et al. (1997) to demonstrate the effects of bisphenol A (BPA) on the mouse prostate utilised Purina 5001 diet, and that of Ashby et al. (1999), who failed to confirm the effects on the mouse prostate, utilised RM1 diet. An additional aspect of the data shown in Fig. 3 is that Purina 5001 is among the richest in phyto-oestrogens, while AIN-76A is devoid of phyto-oestrogens (Thigpen et al., 1999). The apparent uterotrophic activity of Purina 5001 cannot therefore be ascribed to its constituent phyto-oestrogens, and this was confirmed by the loss of uterotrophic activity for all three diets when the GnRH antagonist Antarelix was co-administered to the animals. This blockade by
Fig. 6. Percentage of males in F1 and F2 litters following in utero exposure of F0 pregnant females to BBP. The F1 animals (male and female shown) were mated to virgin animals on pnd 90, following assessment of testes in a parallel group of animals. The values in bold represent failure to confirm the initial change in male F2 sex ratios (unpublished data; experiments described in Ashby et al., 1997). The lowest historical control male sex ratio for our laboratory was 43% prior to these data.
236
J. Ashby / Toxicology Letters 120 (2001) 233–242
Fig. 7. Dose– response curve for DES in immature mouse uterotrophic assays following exposure by s.c. injection. † First dose at which DES was significantly different (PB 0.01) from controls. The data are derived from Shelby et al. (1996), Tinwell et al. (2000).
Fig. 8. Data derived from Long et al. (2000) for the effects of BPA on the vagina of two strains of ovariectomised rats.
Fig. 9. Mean day of vaginal opening in CF1 mice and wild mice according to intra-uterine position: 0M, adjacent to two females, etc. The sources of the data are shown above.
J. Ashby / Toxicology Letters 120 (2001) 233–242
237
pituitary gland. The full potential of diet to modulate both rodent sexual development and rodent ED assay outcomes has yet to be recognised.
4. Intrinsic animal tissue weight variability in ED assays
Fig. 10. Uterotrophic activity of 1 mg/kg (grey) and 10 mg/kg (black) DES in the three strains of mice shown. Weanling CD-1 and C57Bl6 mice were available only at the starting ages shown, so the AP mice acted as aged matched controls for these two strains. DES was administered by s.c. injection in arachis oil for 3 days (Ashby et al., unpublished observations, 2000). * PB 0.05; ** P B0.01 by a one-sided Student’s t-test.
Antarelix indicates that undefined components of the diets are eliciting oestrogenic effects, of different magnitudes, via a common action on the
Some current problems in ED research studies are caused by insufficient knowledge of the intrinsic variability of some assay parameters. One such example is provided by the data of Lee (1998) for the effect of nonylphenol (NP) on the prostate gland of Sprague –Dawley rats (Fig. 4). A variety of dosing protocols was used, but all animals were killed on postnatal day (pnd) 31. NP consistently reduced relative prostate weight compared with the control weights, except when the agent was injected between pnd 13 and 30, when no difference to control prostate weights was seen. This led
Fig. 11. Comparison of CF1 mouse prostate weights (from published sources), and Alpk (AP) mouse uterotrophic responses following exposure to either BPA or DES. The uterotrophic assays used the s.c. injection route of exposure and the prostate studies involved oral exposure in utero. The uterotrophic data are from Tinwell et al. (2000). * P B0.05.
238
J. Ashby / Toxicology Letters 120 (2001) 233–242
Lee to conclude that the dosing of NP should take place before pnd 13 for it to be active in the assay. However, equally interesting was the sudden drop in relative prostate weight in the pnd 13 – 30 test group, a fact not remarked upon by Lee. The reduction was significant compared with the other control groups (P B 0.01) and could not be explained by changes in animal body weight; absolute prostate weights were not provided in the paper. Thus, until this sudden change in control relative prostate weight is accounted for, it is difficult to interpret the effects induced by NP. A similar situation is shown in Fig. 5. The initial report by Sharpe et al. (1995) of reduced testes weight induced in rats by low dose levels (1 ppm in drinking water) of octylphenol (OP) or butyl benzyl phthalate (BBP) was followed by a negative repeat study with BBP from Ashby et al. (1997). Sharpe et al. (1998) subsequently reported temporal changes in their control rat testes weights and that, while they were low, OP elicited a trophic effect on the testes, as compared with the earlier reported atrophic effect. The reason for the temporal changes in testes weights in the laboratory of Sharpe, and for the different testes weights between the two laboratories, has no current explanation; however, Sharpe et al. (1998) suggested that the situation shown in Fig. 5 should give us pause for thought. Subsequent to these studies, OP was reported to be negative in a rat multigeneration assay that involved testing both low and high dose levels of OP (Tyl et al., 1999). Both of these examples illustrate the need for detailed control studies before major conclusions on chemically-induced ED can be drawn. The intrinsic variability of ED endpoints can be further illustrated by the already presented example of our study on BBP. Groups of males and females from our study were not killed at pnd 90, but were mated with virgin animals to produce an F2 generation. Although the sex ratios of litters derived from the mating of females (exposed in utero) with virgin males fell within our normal ranges, an unexpected change in the percentage of male offspring derived from the mating of males (exposed in utero) with virgin females was observed, an effect of a magnitude not seen before in our laboratory (Fig. 6). We discussed the effect
with Richard Sharpe, and decided to repeat the study before the effect could be accepted as related to administration of BBP. In that repeat study, we again measured testes weights and again saw no effect caused by BBP (Fig. 5, study 2). Groups of F1 males and females were again mated, but a normal sex ratio in the resultant litters was observed (Fig. 6). We concluded that the first effect was due to chance, even though its publication when first seen would have been provocative and exciting.
5. Influence of animal strain on the outcome of ED assays of chemicals Some endocrine activities of chemicals can be reproduced across laboratories with remarkable success, as illustrated by the uterotrophic activity of diethylstilbestrol (DES) in the mouse, conducted using different strains of mice, by different investigators in different laboratories, at different times (Fig. 7). However, some marked instances of strain specificity in ED data have been recorded. Probably the first was the sensitivity of F344 rats, and the complete insensitivity of Sprague –Dawley (SD) rats, to prolactinaemia induction by both oestradiol and BPA (Steinmetz et al., 1997). However, other studies have shown SD rats to be sensitive to estrogens (nonylphenol on the uterus (Odum et al., 1999) and BPA on the testes (Fialkowski et al., 2000)). A similar rat strain conflict is evident with the finding by Long et al. (2000) that BPA initiates DNA synthesis in the vaginal epithelium of the F344 rat, but not in the SD rat (Fig. 8). That contrasts with the induction of pathological changes in the vagina of SD pups treated with BPA in utero (Fialkowski et al., 2000). Such differences in strain sensitivity can be subtle, as shown by the data in Fig. 8. Those data demonstrate that the insensitivity of SD rats to the vaginal effects of BPA must be associated with differences in intermediate or late gene expressions, because both strains of rat show similar sensitivities to BPA up to and including the expression of the immediate/early cell-proliferation gene c-fos. Two other sets of observations confirm the
J. Ashby / Toxicology Letters 120 (2001) 233–242
seriousness of strain-specific ED responses. The first relates to the presence of a marked relationship between the mean day of vaginal opening and intra-uterine position in wild mice (Mus musculis) (Zelinski and Vandenbergh, 1991; Fig. 9) and the absence of such a dependence in CF1 mice (Howdeshell et al., 1999). The effect of intra-uterine position in the wild mice was most marked in the females adjacent to two females in utero (0M; Zelinski and Vandenbergh, 1991), and this same 0M intra-uterine position also showed the greatest change in first oestrus in the CF1 mice exposed in utero to BPA (Howdeshell et al., 1999). Mysteries such as this require urgent study if the effects reported by Howdeshell et al. (1999) for BPA are to be extrapolated usefully to humans. The second example is that Spearow et al. (1999) have defined CD-1 mice as substantially less sensitive than other strains (e.g. C57Bl6) to the testicular effects of oestradiol, yet these (presumably) same CD-1 mice are now used by the group of vom Saal to study effects on the mouse prostate (vom Saal, 2000), the group having observed reduced oestrogen sensitivity over time for their in-house colony of CF1 mice. The word ‘presumably’ was used in the previous sentence because the alarming situation is presenting that rodent strain identification terms may not adequately define the genetic makeup of the strain. This may provide an explanation for why Ashby et al. (1999) were unable to confirm the mouse prostate effects for BPA and DES reported earlier by Nagel et al. (1997). Both laboratories used CF1 mice, but those used by Ashby et al. (1999) were obtained from Charles River in 1996, while those used by Nagel et al. (1997) had been bred at the University of Missouri since 1979. Charles River culled their stock of CF1 mice in the mid-1980s and re-established the present colony from a brother/sister mating (personal communication; vom Saal, 2000). The two strains of CF1 mice used in these two recent experiments may therefore be genetically divergent, despite their possession of the same strain identifier. Whatever the critical differences between strains, they are not evidenced at all levels of oestrogen sensitivity, as illustrated by the similar uterotrophic activity of DES in CD-1, C57Bl6 and AP mice strains (Fig. 10).
239
6. Low dose effects of BPA and DES If the observation that oestrogens can elicit effects orders of magnitude below current perceptions of their toxicological no effect level (vom Saal, 2000) becomes to be accepted as a general phenomenon, chemical toxicity testing protocols will have to be redesigned, ab initio. The claims for such low dose effects originated with the effects seen for BBP and OP by Sharpe et al. (1995), as already discussed. However, the data of particular current interest are those of Nagel et al. (1997), vom Saal et al. (1997), for BPA and DES, respectively. The original reports of the activity of BPA on the mouse prostate involved three assay parameters: increased prostate gland weights, increased prostate androgen receptor levels, and increased prostate budding at pnd 1. The increases in prostate gland weights observed following in utero exposure of CF1 mice to DES (vom Saal et al., 1997) and to BPA (Nagel et al., 1997) are shown in Fig. 11. Ashby et al. (1999), Cagen et al. (1999) were unable to confirm those effects using the currently available Charles River CF1 mice. Possible reasons for this conflict of findings have been discussed in previous sections; the most obvious variables to eliminate in future studies being the different diets used in the different studies (Purina 5001 and RM1; see Fig. 3) and the possibility that the epithet ‘CF1 mouse’ covers a critically different range of genotypes (see preceding section). These considerations apart, a deeper mystery is evident, as follows. The reported activity of DES on the mouse prostate gland is essentially consistent with the known ED activities of DES, as epitomised in Fig. 11 by the immature mouse uterotrophic assay data shown in the lower half of the figure (also consistent with the developmental toxicities and growth promoting activities of DES; see Ashby, 1999). Thus, although effects were evident for DES on the mouse prostate gland at doses lower than in the uterotrophic assays, this could speculatively be accounted for by the exposures in the prostate gland studies occurring in utero. However, the activity of BPA on the mouse prostate gland occurred at dose levels unheralded by its uterotrophic activity in the immature mouse (Fig. 11).
Fig. 12. Dam and pup body weights of SD or AP rats exposed to BPA during gestation days 6 –21 of pregnancy, as described by Fialkowski et al. (2000), Talsness and Chahoud (2000). No significant weight changes were observed (litter as the statistical unit). These are preliminary data from an ongoing study conducted in these laboratories (Ashby et al., unpublished observations, 2000).
240 J. Ashby / Toxicology Letters 120 (2001) 233–242
J. Ashby / Toxicology Letters 120 (2001) 233–242
Those uterotrophic assay data for BPA (Tinwell et al., 2000) showed sporadic signs of weak activity at some dose levels between 10 and 200 mg/kg body weight, but in the absence of any indication of a dose –response relationship, and no activity was evident in the dose range covered by the prostate effects. In fact, the uterotrophic assay data shown in Fig. 11, which are the composite of eight separate experiments, involving assessments made on several hundred animals, show no reproducible statistically significant increases in uterine weight. Therefore, the question is posed as to how two oestrogens of such grossly dissimilar intrinsic oestrogenic activities, as are DES and BPA, could induce such similar low dose effects on the mouse prostate gland The low dose prostate gland effects reported for BPA (Nagel et al., 1997; vom Saal, 2000) are echoed by the low dose oestrogenic effects reported for this chemical in SD rats (Fialkowski et al., 2000; Talsness and Chahoud, 2000). However, the negative low dose findings for BPA described by Cagen et al. (1999) and Ashby et al. (1999) are worthy of equal respect — they may hold the key to the effective extrapolation of the data for BPA to humans. We are presently in the middle of a repeat study of the effects of BPA in SD rats, as described by Chahoud and coworkers (Fialkowski et al., 2000; Talsness and Chahoud, 2000). We are also studying the sensitivity of Alderley Park (Wistar-derived) rats in the same study. The pups from those studies had normal anogenital distances and are now approaching sexual maturity Fig. 12.
7. Overall conclusions The study of chemical endocrine disruption is a valid and important field of endeavour. However, in order to make progress in discerning the true hazards posed to wildlife and humans by endocrine active chemicals, it will be necessary to reduce the incidence of premature conclusions and to increase the rate of data acquisition and data integration.
241
References Ashby, J., 1999. Dose levels of 0.02– 0.2 mg/kg/day DES are not suitable for use as a positive control in endocrine toxicity studies. Editorial. Regul. Toxicol. Pharmacol. 29, 235– 237. Ashby, J., Tinwell, H., Lefevre, P.A., Odum, J., Paton, D., Millward, S.W., Tittensor, S., Brooks, A.N., 1997. Normal sexual development of rats exposed to butyl benzyl phthalate from conception to weaning. Regul. Toxicol. Pharmacol. 26, 102– 118. Ashby, J., Tinwell, H., Haseman, J., 1999. Lack of effects for low dose levels of BPA and DES on the prostate gland of CF1 mice exposed in utero. Regul. Toxicol. Pharmacol. 30, 156– 166. Ashby, J., Tinwell, H., Odum, J., 2001. DNA adducts, estrogenicity and rodent diets. Mutat. Res. (in press). Cagen, S.Z., Waechter, J.M., Jr, Dimond, S.S., Breslin, W.J., Butala, J.H., Jekat, F.W., Joiner, R.L., Shiotsuka, R.N., Veenstra, G.E., Harris, L.R., 1999. Normal reproductive organ development in CF-1 mice following prenatal exposure to Bisphenol A. Toxicol. Sci. 50, 36 – 44. Deghenghi, R., Boutignon, F., Wuthrich, P., Lenaerts, V., 1993. Antarelix (EP24332) a novel water soluble LHRH antagonist. Biomed. Pharmacother. 47, 107– 110. Fialkowski, O., Merker, H-J., Talsness, C.E., Chahoud, I., 2000. Histopathological findings in the testes of male rat offspring following prenatal exposure to a low and a high dose of bisphenol A. Proceedings of the RH Workshop, Copenhagen, May 2000, Abstract 55. Gray, L.E., Jr, Ostby, J., Monosson, E., Kelce, W.R., 1999. Environmental antiandrogens: low doses of the fungicide vinclozolin alter sexual differentiation of the male rat. Toxicol. Ind. Health 15, 48 – 64. Howdeshell, K.L., Hotchkiss, A.K., Thayer, K.A., Vandenbergh, J.G., vom Saal, F.S., 1999. Exposure to bisphenol A advances puberty. Nature 401, 763– 764. Lee, P.C., 1998. Disruption of male reproductive tract development by administration of the xenoestrogen, nonylphenol, to male newborn rats. Endocrine 9, 105– 111. Long, X., Steinmetz, R., Ben-Jonathan, N., Caperell-Grant, A., Young, P.C.M., Nephew, K.P., Bigsby, R.M., 2000. Strain differences in vaginal responses to the xenoestrogen bisphenol A. Environ. Health Perspect. 108, 243– 247. Nagel, S., vom Saal, F.S., Thayer, K., Dhar, M., Boechler, M., Welshons, W., 1997. Relative binding affinity-serum modified access (RBA-SMA) assay predicts the relative in vivo bioactivity of the xenoestrogens bisphenol A and octylphenol. Environ. Health Perspect. 105, 70 – 76. Odum, J., Ashby, J., 2000. NP Neonatal exposure of male rats to nonylphenol has no effect on the reproductive tract. Toxicol. Sci. 56, 400– 404. Odum, J., Pyrah, I., Soames, A., Foster, J., Van Miller, J., Joiner, R., Ashby, J., 1999. Effects of p-nonylphenol (NP) and diethylstilbestrol (DES) on the Alderley Park (Alpk) rat: comparison of mammary gland and uterus sensitivity following administration by either oral gavage or subcuta-
242
J. Ashby / Toxicology Letters 120 (2001) 233–242
neously implanted mini-pumps. J. Appl. Toxicol. 19, 367– 378. Sharpe, R.M., Fisher, J.S., Millar, M.M., Jobling, S., Sumpter, J.P., 1995. Gestational and lactational exposure of rats to xenoestrogens results in reduced testicular size and sperm production. Environ. Health Perspect. 103, 1136–1143. Sharpe, R.M., Turner, K.J., Sumpter, J.P., 1998. Endocrine disruptors and testis development. Environ. Health Perspect. 106, A220– A221. Shelby, M.D., Newbold, R.R., Tully, D.B., Chae, K., Davis, V.L., 1996. Assessing environmental chemicals for estrogenicity using a combination of in vitro and in vivo assays. Environ. Health Perspect. 104, 1296–1300. Spearow, J.L., Doemeny, P., Sera, R., Leffler, R., Barkley, M., 1999. Genetic variation in susceptibility to endocrine disruption by estrogen in mice. Science 285, 1259–1261. Steinmetz, R., Brown, N.G., Allen, D.L., Bigsby, R.M., BenJonathan, N., 1997. The environmental estrogen bisphenol A stimulates prolactin release in vitro and in vivo. Endocrinology 138, 1780–1786. Talsness, C.E., Chahoud, I., 2000. Histopathological findings in the vagina of rat female offspring following prenatal exposure to a low and a high dose of bisphenol A. Proceedings of the RH Workshop, Copenhagen, May 2000, Abstract 56.
Thigpen, J.E., Setchell, K.D.R., Goelz, M.F., Forsythe, D.B., 1999. The phytoestrogen content of rodent diets. Environ. Health Perspect. 107, 182– 183. Tinwell, H., Joiner, R., Pate, I., Ashby, J., 2000. Uterotrophic activity of Bisphenol A (BPA) in the immature mouse. Regul. Toxicol. Pharmacol. 32, 118– 126. Tyl, R.W., Myers, C.B., Marr, M.C., Brine, D.R., Fail, P.A., Seely, J.C., Van Miller, J.P., 1999. Two-generation reproduction study with para-tert-octylphenol in rats. Regul. Toxicol. Pharmacol. 30, 81 – 95. vom Saal, F.S., Timms, B.G., Montano, M.M., Palanza, P., Thayer, K.A., Nagel, S.C., Dhar, M.D., Ganjam, V.K., Parmigiani, S., Welshons, W.V., 1997. Prostate enlargement in mice due to fetal exposure to low doses of estradiol or diethylstilbestrol and opposite effects at high doses. Proc. Natl. Acad. Sci. USA 94, 2056– 2061. vom Saal, F.S., 2000. Low dose effects of bisphenol A and estrogenic drugs. Proceedings of the RH Workshop, Copenhagen, May 2000, Session 5 Abstract. Zelinski, W.J., Vandenbergh, J.G., 1991. Effect of intrauterine position and social density on age of first reproduction in the female house mouse (Mus musculus). J. Comp. Psychol. 105, 134– 139.
.
Testing for endocrine disruption post-EDSTAC: extrapolation of low dose rodent effects to humans John Ashby * Syngenta Central Toxicology Laboratory, Alderley Park, Macclesfield, Cheshire SK10 4TJ, UK
Abstract The study of chemically-induced endocrine disruption in mammals is a relatively new field of endeavour, and it has been assailed by an unusual level of disagreement among investigators regarding the developmental effects produced by chemicals in animals. This article discusses the several sources of uncertainty in endocrine toxicity studies, and the intrinsic variability of many of the key experimental parameters. It is concluded that current uncertainties regarding extrapolation of rodent effects to humans are due to the absence of an extensive agreed rodent control database for the developmental parameters under study, coupled to the established intrinsic variability of these parameters between strains/species of test animals and test protocols. Only when these factors are generally accepted and well studied will it be possible to relate effects seen in rodents to humans. © 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Endocrine disruption; EDSTAC; Bisphenol A; Diethylstilboestrol; Phytoestrogens
1. Endocrine disruption as a new toxicological concern
2. Sources of uncertainty and variability in rodent endocrine disruption studies
Validation of the several assays discussed by the US EDSTAC initiative is now well in train. However, before data derived from such rodent assays can be interpreted with confidence, it will be necessary to focus a range of problems, as discussed herein.
The study of chemicals for endocrine disrupting (ED) properties is complicated by the many variables, as shown in Fig. 1. Currently, most routine rodent toxicological assessments are conducted using oral gavage of the chemical to young adult animals, with clear adverse effects being the assay endpoint. However, the lateral trends shown in Fig. 1 are increasingly followed, usually without an explanation of why the particular constellation of experimental conditions chosen were selected. Some of these trends are justifiable, while others are not. Lack of attention to such methodological details makes it difficult to reconcile different ED assay outcomes for the same chemical. This po-
* Tel.: +44-1625-582828; fax: +44-1625-590249. E-mail address: [email protected] (J. Ashby).
0378-4274/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 3 7 8 - 4 2 7 4 ( 0 1 ) 0 0 2 9 9 - 5
234
J. Ashby / Toxicology Letters 120 (2001) 233–242
Fig. 1. Transitions facing toxicology as studies move from the bold words along the four lateral arms. Fig. 4. Prostate weights at post-natal day 31 for control and NP-treated SD rats. Data taken from Lee (1998) * Statistical significance for NP versus C (PB 0.05) taken from Lee (1998). †† Statistical significance for C (13– 20) versus all other controls (P B 0.01), as reported by Odum and Ashby (2000).
Fig. 2. Dose– response curves for genistein in immature AP mouse uterotrophic assays following exposure by either s.c. injection or oral gavage in corn oil. The studies were conducted as described earlier with RM1 as diet (Tinwell et al., 2000). ** P B0.01.
tential source of confusion could escalate with the growing use of gene microarray DNA technology coupled to the use of stable cell lines. An example of the importance of some of these experimental
variables is shown by the different uterotrophic assay responses to the phyto-oestrogen genistein produced by changing the route of administration to the rodents (Fig. 2). The relevance of the oral route of exposure to consideration of the properties of phyto-oestrogens in diet is obvious, but it is difficult to relate the assay response derived using the subcutaneous (s.c.) injection route to humans, despite it being the stronger of the two responses. The strongest assay response may not always be the most relevant response for human or wildlife risk assessment purposes. An equally important consideration is realisation of the hierarchy of assay sensitivities. Another source of genuine variability is illustrated by the data of Gray et al. (1999), who have
Fig. 3. Effect of three different commercial rodent diets on uterine weight in immature AP rats (Ashby et al., 2001; unpublished observations), and on the sexual maturation of the female animals. Inhibition of the uterotrophic response was observed following concomitant exposure to the GnRH antagonist antarelix (300 mg/kg daily; Deghenghi et al., 1993). * P B 0.05; ** P B 0.01.
J. Ashby / Toxicology Letters 120 (2001) 233–242
235
3. Diet and endocrine disruption
Fig. 5. Response of male rats at pnd 90 following their exposure in utero to OP or BBP in diet (1 p.p.m.). The control testes weight and the varying responses to OP and BBP are discussed in the text.
shown that the induction of adverse effects in neonatal rats, whose mothers were treated with an anti-androgen, obey the following sensitivity hierarchy: changes in ano-genital distance\induction of hypospadias\ induction of ectopic testes. The establishment of a range of sensitivity hierarchies for individual assays will do much to reduce the incidence of apparent disagreements in the literature regarding the ED activities of individual chemicals.
The possible role of diet in the aetiology of changes in the onset of human puberty has been long recognised. However, the role of diet in modulating the outcome of rodent endocrine disruption studies has yet to be studied. The potential importance of this topic is illustrated by the data shown in Fig. 3. This illustrates that three different rodent diets have different oestrogenic properties as assayed using the immature rat uterotrophic assay or in assays to determine the time of puberty in the rats (age of vaginal opening and first oestrus). It is pertinent to note here that the test protocol used by Nagel et al. (1997) to demonstrate the effects of bisphenol A (BPA) on the mouse prostate utilised Purina 5001 diet, and that of Ashby et al. (1999), who failed to confirm the effects on the mouse prostate, utilised RM1 diet. An additional aspect of the data shown in Fig. 3 is that Purina 5001 is among the richest in phyto-oestrogens, while AIN-76A is devoid of phyto-oestrogens (Thigpen et al., 1999). The apparent uterotrophic activity of Purina 5001 cannot therefore be ascribed to its constituent phyto-oestrogens, and this was confirmed by the loss of uterotrophic activity for all three diets when the GnRH antagonist Antarelix was co-administered to the animals. This blockade by
Fig. 6. Percentage of males in F1 and F2 litters following in utero exposure of F0 pregnant females to BBP. The F1 animals (male and female shown) were mated to virgin animals on pnd 90, following assessment of testes in a parallel group of animals. The values in bold represent failure to confirm the initial change in male F2 sex ratios (unpublished data; experiments described in Ashby et al., 1997). The lowest historical control male sex ratio for our laboratory was 43% prior to these data.
236
J. Ashby / Toxicology Letters 120 (2001) 233–242
Fig. 7. Dose– response curve for DES in immature mouse uterotrophic assays following exposure by s.c. injection. † First dose at which DES was significantly different (PB 0.01) from controls. The data are derived from Shelby et al. (1996), Tinwell et al. (2000).
Fig. 8. Data derived from Long et al. (2000) for the effects of BPA on the vagina of two strains of ovariectomised rats.
Fig. 9. Mean day of vaginal opening in CF1 mice and wild mice according to intra-uterine position: 0M, adjacent to two females, etc. The sources of the data are shown above.
J. Ashby / Toxicology Letters 120 (2001) 233–242
237
pituitary gland. The full potential of diet to modulate both rodent sexual development and rodent ED assay outcomes has yet to be recognised.
4. Intrinsic animal tissue weight variability in ED assays
Fig. 10. Uterotrophic activity of 1 mg/kg (grey) and 10 mg/kg (black) DES in the three strains of mice shown. Weanling CD-1 and C57Bl6 mice were available only at the starting ages shown, so the AP mice acted as aged matched controls for these two strains. DES was administered by s.c. injection in arachis oil for 3 days (Ashby et al., unpublished observations, 2000). * PB 0.05; ** P B0.01 by a one-sided Student’s t-test.
Antarelix indicates that undefined components of the diets are eliciting oestrogenic effects, of different magnitudes, via a common action on the
Some current problems in ED research studies are caused by insufficient knowledge of the intrinsic variability of some assay parameters. One such example is provided by the data of Lee (1998) for the effect of nonylphenol (NP) on the prostate gland of Sprague –Dawley rats (Fig. 4). A variety of dosing protocols was used, but all animals were killed on postnatal day (pnd) 31. NP consistently reduced relative prostate weight compared with the control weights, except when the agent was injected between pnd 13 and 30, when no difference to control prostate weights was seen. This led
Fig. 11. Comparison of CF1 mouse prostate weights (from published sources), and Alpk (AP) mouse uterotrophic responses following exposure to either BPA or DES. The uterotrophic assays used the s.c. injection route of exposure and the prostate studies involved oral exposure in utero. The uterotrophic data are from Tinwell et al. (2000). * P B0.05.
238
J. Ashby / Toxicology Letters 120 (2001) 233–242
Lee to conclude that the dosing of NP should take place before pnd 13 for it to be active in the assay. However, equally interesting was the sudden drop in relative prostate weight in the pnd 13 – 30 test group, a fact not remarked upon by Lee. The reduction was significant compared with the other control groups (P B 0.01) and could not be explained by changes in animal body weight; absolute prostate weights were not provided in the paper. Thus, until this sudden change in control relative prostate weight is accounted for, it is difficult to interpret the effects induced by NP. A similar situation is shown in Fig. 5. The initial report by Sharpe et al. (1995) of reduced testes weight induced in rats by low dose levels (1 ppm in drinking water) of octylphenol (OP) or butyl benzyl phthalate (BBP) was followed by a negative repeat study with BBP from Ashby et al. (1997). Sharpe et al. (1998) subsequently reported temporal changes in their control rat testes weights and that, while they were low, OP elicited a trophic effect on the testes, as compared with the earlier reported atrophic effect. The reason for the temporal changes in testes weights in the laboratory of Sharpe, and for the different testes weights between the two laboratories, has no current explanation; however, Sharpe et al. (1998) suggested that the situation shown in Fig. 5 should give us pause for thought. Subsequent to these studies, OP was reported to be negative in a rat multigeneration assay that involved testing both low and high dose levels of OP (Tyl et al., 1999). Both of these examples illustrate the need for detailed control studies before major conclusions on chemically-induced ED can be drawn. The intrinsic variability of ED endpoints can be further illustrated by the already presented example of our study on BBP. Groups of males and females from our study were not killed at pnd 90, but were mated with virgin animals to produce an F2 generation. Although the sex ratios of litters derived from the mating of females (exposed in utero) with virgin males fell within our normal ranges, an unexpected change in the percentage of male offspring derived from the mating of males (exposed in utero) with virgin females was observed, an effect of a magnitude not seen before in our laboratory (Fig. 6). We discussed the effect
with Richard Sharpe, and decided to repeat the study before the effect could be accepted as related to administration of BBP. In that repeat study, we again measured testes weights and again saw no effect caused by BBP (Fig. 5, study 2). Groups of F1 males and females were again mated, but a normal sex ratio in the resultant litters was observed (Fig. 6). We concluded that the first effect was due to chance, even though its publication when first seen would have been provocative and exciting.
5. Influence of animal strain on the outcome of ED assays of chemicals Some endocrine activities of chemicals can be reproduced across laboratories with remarkable success, as illustrated by the uterotrophic activity of diethylstilbestrol (DES) in the mouse, conducted using different strains of mice, by different investigators in different laboratories, at different times (Fig. 7). However, some marked instances of strain specificity in ED data have been recorded. Probably the first was the sensitivity of F344 rats, and the complete insensitivity of Sprague –Dawley (SD) rats, to prolactinaemia induction by both oestradiol and BPA (Steinmetz et al., 1997). However, other studies have shown SD rats to be sensitive to estrogens (nonylphenol on the uterus (Odum et al., 1999) and BPA on the testes (Fialkowski et al., 2000)). A similar rat strain conflict is evident with the finding by Long et al. (2000) that BPA initiates DNA synthesis in the vaginal epithelium of the F344 rat, but not in the SD rat (Fig. 8). That contrasts with the induction of pathological changes in the vagina of SD pups treated with BPA in utero (Fialkowski et al., 2000). Such differences in strain sensitivity can be subtle, as shown by the data in Fig. 8. Those data demonstrate that the insensitivity of SD rats to the vaginal effects of BPA must be associated with differences in intermediate or late gene expressions, because both strains of rat show similar sensitivities to BPA up to and including the expression of the immediate/early cell-proliferation gene c-fos. Two other sets of observations confirm the
J. Ashby / Toxicology Letters 120 (2001) 233–242
seriousness of strain-specific ED responses. The first relates to the presence of a marked relationship between the mean day of vaginal opening and intra-uterine position in wild mice (Mus musculis) (Zelinski and Vandenbergh, 1991; Fig. 9) and the absence of such a dependence in CF1 mice (Howdeshell et al., 1999). The effect of intra-uterine position in the wild mice was most marked in the females adjacent to two females in utero (0M; Zelinski and Vandenbergh, 1991), and this same 0M intra-uterine position also showed the greatest change in first oestrus in the CF1 mice exposed in utero to BPA (Howdeshell et al., 1999). Mysteries such as this require urgent study if the effects reported by Howdeshell et al. (1999) for BPA are to be extrapolated usefully to humans. The second example is that Spearow et al. (1999) have defined CD-1 mice as substantially less sensitive than other strains (e.g. C57Bl6) to the testicular effects of oestradiol, yet these (presumably) same CD-1 mice are now used by the group of vom Saal to study effects on the mouse prostate (vom Saal, 2000), the group having observed reduced oestrogen sensitivity over time for their in-house colony of CF1 mice. The word ‘presumably’ was used in the previous sentence because the alarming situation is presenting that rodent strain identification terms may not adequately define the genetic makeup of the strain. This may provide an explanation for why Ashby et al. (1999) were unable to confirm the mouse prostate effects for BPA and DES reported earlier by Nagel et al. (1997). Both laboratories used CF1 mice, but those used by Ashby et al. (1999) were obtained from Charles River in 1996, while those used by Nagel et al. (1997) had been bred at the University of Missouri since 1979. Charles River culled their stock of CF1 mice in the mid-1980s and re-established the present colony from a brother/sister mating (personal communication; vom Saal, 2000). The two strains of CF1 mice used in these two recent experiments may therefore be genetically divergent, despite their possession of the same strain identifier. Whatever the critical differences between strains, they are not evidenced at all levels of oestrogen sensitivity, as illustrated by the similar uterotrophic activity of DES in CD-1, C57Bl6 and AP mice strains (Fig. 10).
239
6. Low dose effects of BPA and DES If the observation that oestrogens can elicit effects orders of magnitude below current perceptions of their toxicological no effect level (vom Saal, 2000) becomes to be accepted as a general phenomenon, chemical toxicity testing protocols will have to be redesigned, ab initio. The claims for such low dose effects originated with the effects seen for BBP and OP by Sharpe et al. (1995), as already discussed. However, the data of particular current interest are those of Nagel et al. (1997), vom Saal et al. (1997), for BPA and DES, respectively. The original reports of the activity of BPA on the mouse prostate involved three assay parameters: increased prostate gland weights, increased prostate androgen receptor levels, and increased prostate budding at pnd 1. The increases in prostate gland weights observed following in utero exposure of CF1 mice to DES (vom Saal et al., 1997) and to BPA (Nagel et al., 1997) are shown in Fig. 11. Ashby et al. (1999), Cagen et al. (1999) were unable to confirm those effects using the currently available Charles River CF1 mice. Possible reasons for this conflict of findings have been discussed in previous sections; the most obvious variables to eliminate in future studies being the different diets used in the different studies (Purina 5001 and RM1; see Fig. 3) and the possibility that the epithet ‘CF1 mouse’ covers a critically different range of genotypes (see preceding section). These considerations apart, a deeper mystery is evident, as follows. The reported activity of DES on the mouse prostate gland is essentially consistent with the known ED activities of DES, as epitomised in Fig. 11 by the immature mouse uterotrophic assay data shown in the lower half of the figure (also consistent with the developmental toxicities and growth promoting activities of DES; see Ashby, 1999). Thus, although effects were evident for DES on the mouse prostate gland at doses lower than in the uterotrophic assays, this could speculatively be accounted for by the exposures in the prostate gland studies occurring in utero. However, the activity of BPA on the mouse prostate gland occurred at dose levels unheralded by its uterotrophic activity in the immature mouse (Fig. 11).
Fig. 12. Dam and pup body weights of SD or AP rats exposed to BPA during gestation days 6 –21 of pregnancy, as described by Fialkowski et al. (2000), Talsness and Chahoud (2000). No significant weight changes were observed (litter as the statistical unit). These are preliminary data from an ongoing study conducted in these laboratories (Ashby et al., unpublished observations, 2000).
240 J. Ashby / Toxicology Letters 120 (2001) 233–242
J. Ashby / Toxicology Letters 120 (2001) 233–242
Those uterotrophic assay data for BPA (Tinwell et al., 2000) showed sporadic signs of weak activity at some dose levels between 10 and 200 mg/kg body weight, but in the absence of any indication of a dose –response relationship, and no activity was evident in the dose range covered by the prostate effects. In fact, the uterotrophic assay data shown in Fig. 11, which are the composite of eight separate experiments, involving assessments made on several hundred animals, show no reproducible statistically significant increases in uterine weight. Therefore, the question is posed as to how two oestrogens of such grossly dissimilar intrinsic oestrogenic activities, as are DES and BPA, could induce such similar low dose effects on the mouse prostate gland The low dose prostate gland effects reported for BPA (Nagel et al., 1997; vom Saal, 2000) are echoed by the low dose oestrogenic effects reported for this chemical in SD rats (Fialkowski et al., 2000; Talsness and Chahoud, 2000). However, the negative low dose findings for BPA described by Cagen et al. (1999) and Ashby et al. (1999) are worthy of equal respect — they may hold the key to the effective extrapolation of the data for BPA to humans. We are presently in the middle of a repeat study of the effects of BPA in SD rats, as described by Chahoud and coworkers (Fialkowski et al., 2000; Talsness and Chahoud, 2000). We are also studying the sensitivity of Alderley Park (Wistar-derived) rats in the same study. The pups from those studies had normal anogenital distances and are now approaching sexual maturity Fig. 12.
7. Overall conclusions The study of chemical endocrine disruption is a valid and important field of endeavour. However, in order to make progress in discerning the true hazards posed to wildlife and humans by endocrine active chemicals, it will be necessary to reduce the incidence of premature conclusions and to increase the rate of data acquisition and data integration.
241
References Ashby, J., 1999. Dose levels of 0.02– 0.2 mg/kg/day DES are not suitable for use as a positive control in endocrine toxicity studies. Editorial. Regul. Toxicol. Pharmacol. 29, 235– 237. Ashby, J., Tinwell, H., Lefevre, P.A., Odum, J., Paton, D., Millward, S.W., Tittensor, S., Brooks, A.N., 1997. Normal sexual development of rats exposed to butyl benzyl phthalate from conception to weaning. Regul. Toxicol. Pharmacol. 26, 102– 118. Ashby, J., Tinwell, H., Haseman, J., 1999. Lack of effects for low dose levels of BPA and DES on the prostate gland of CF1 mice exposed in utero. Regul. Toxicol. Pharmacol. 30, 156– 166. Ashby, J., Tinwell, H., Odum, J., 2001. DNA adducts, estrogenicity and rodent diets. Mutat. Res. (in press). Cagen, S.Z., Waechter, J.M., Jr, Dimond, S.S., Breslin, W.J., Butala, J.H., Jekat, F.W., Joiner, R.L., Shiotsuka, R.N., Veenstra, G.E., Harris, L.R., 1999. Normal reproductive organ development in CF-1 mice following prenatal exposure to Bisphenol A. Toxicol. Sci. 50, 36 – 44. Deghenghi, R., Boutignon, F., Wuthrich, P., Lenaerts, V., 1993. Antarelix (EP24332) a novel water soluble LHRH antagonist. Biomed. Pharmacother. 47, 107– 110. Fialkowski, O., Merker, H-J., Talsness, C.E., Chahoud, I., 2000. Histopathological findings in the testes of male rat offspring following prenatal exposure to a low and a high dose of bisphenol A. Proceedings of the RH Workshop, Copenhagen, May 2000, Abstract 55. Gray, L.E., Jr, Ostby, J., Monosson, E., Kelce, W.R., 1999. Environmental antiandrogens: low doses of the fungicide vinclozolin alter sexual differentiation of the male rat. Toxicol. Ind. Health 15, 48 – 64. Howdeshell, K.L., Hotchkiss, A.K., Thayer, K.A., Vandenbergh, J.G., vom Saal, F.S., 1999. Exposure to bisphenol A advances puberty. Nature 401, 763– 764. Lee, P.C., 1998. Disruption of male reproductive tract development by administration of the xenoestrogen, nonylphenol, to male newborn rats. Endocrine 9, 105– 111. Long, X., Steinmetz, R., Ben-Jonathan, N., Caperell-Grant, A., Young, P.C.M., Nephew, K.P., Bigsby, R.M., 2000. Strain differences in vaginal responses to the xenoestrogen bisphenol A. Environ. Health Perspect. 108, 243– 247. Nagel, S., vom Saal, F.S., Thayer, K., Dhar, M., Boechler, M., Welshons, W., 1997. Relative binding affinity-serum modified access (RBA-SMA) assay predicts the relative in vivo bioactivity of the xenoestrogens bisphenol A and octylphenol. Environ. Health Perspect. 105, 70 – 76. Odum, J., Ashby, J., 2000. NP Neonatal exposure of male rats to nonylphenol has no effect on the reproductive tract. Toxicol. Sci. 56, 400– 404. Odum, J., Pyrah, I., Soames, A., Foster, J., Van Miller, J., Joiner, R., Ashby, J., 1999. Effects of p-nonylphenol (NP) and diethylstilbestrol (DES) on the Alderley Park (Alpk) rat: comparison of mammary gland and uterus sensitivity following administration by either oral gavage or subcuta-
242
J. Ashby / Toxicology Letters 120 (2001) 233–242
neously implanted mini-pumps. J. Appl. Toxicol. 19, 367– 378. Sharpe, R.M., Fisher, J.S., Millar, M.M., Jobling, S., Sumpter, J.P., 1995. Gestational and lactational exposure of rats to xenoestrogens results in reduced testicular size and sperm production. Environ. Health Perspect. 103, 1136–1143. Sharpe, R.M., Turner, K.J., Sumpter, J.P., 1998. Endocrine disruptors and testis development. Environ. Health Perspect. 106, A220– A221. Shelby, M.D., Newbold, R.R., Tully, D.B., Chae, K., Davis, V.L., 1996. Assessing environmental chemicals for estrogenicity using a combination of in vitro and in vivo assays. Environ. Health Perspect. 104, 1296–1300. Spearow, J.L., Doemeny, P., Sera, R., Leffler, R., Barkley, M., 1999. Genetic variation in susceptibility to endocrine disruption by estrogen in mice. Science 285, 1259–1261. Steinmetz, R., Brown, N.G., Allen, D.L., Bigsby, R.M., BenJonathan, N., 1997. The environmental estrogen bisphenol A stimulates prolactin release in vitro and in vivo. Endocrinology 138, 1780–1786. Talsness, C.E., Chahoud, I., 2000. Histopathological findings in the vagina of rat female offspring following prenatal exposure to a low and a high dose of bisphenol A. Proceedings of the RH Workshop, Copenhagen, May 2000, Abstract 56.
Thigpen, J.E., Setchell, K.D.R., Goelz, M.F., Forsythe, D.B., 1999. The phytoestrogen content of rodent diets. Environ. Health Perspect. 107, 182– 183. Tinwell, H., Joiner, R., Pate, I., Ashby, J., 2000. Uterotrophic activity of Bisphenol A (BPA) in the immature mouse. Regul. Toxicol. Pharmacol. 32, 118– 126. Tyl, R.W., Myers, C.B., Marr, M.C., Brine, D.R., Fail, P.A., Seely, J.C., Van Miller, J.P., 1999. Two-generation reproduction study with para-tert-octylphenol in rats. Regul. Toxicol. Pharmacol. 30, 81 – 95. vom Saal, F.S., Timms, B.G., Montano, M.M., Palanza, P., Thayer, K.A., Nagel, S.C., Dhar, M.D., Ganjam, V.K., Parmigiani, S., Welshons, W.V., 1997. Prostate enlargement in mice due to fetal exposure to low doses of estradiol or diethylstilbestrol and opposite effects at high doses. Proc. Natl. Acad. Sci. USA 94, 2056– 2061. vom Saal, F.S., 2000. Low dose effects of bisphenol A and estrogenic drugs. Proceedings of the RH Workshop, Copenhagen, May 2000, Session 5 Abstract. Zelinski, W.J., Vandenbergh, J.G., 1991. Effect of intrauterine position and social density on age of first reproduction in the female house mouse (Mus musculus). J. Comp. Psychol. 105, 134– 139.
.