Parabens are a group of the alkyl esters ofp-hydroxybenzoic acid and typically include methylparaben, ethylparaben, propylparaben, butylparaben, isobutylparaben, isopropylparaben, and benzylparaben. Parabens (or their salts) are widely used as preservatives in cosmetics, toiletries, and pharmaceuticals due to their relatively low toxicity profile and a long history of safe use. Testing of parabens has revealed to varying degrees that individual paraben compounds have weakly estrogenic activity in some in vitro screening tests, such as ligand binding to the estrogen receptor, regulation of CAT gene expression, and proliferation of MCF-7 cells. Reported in vivo effects include increased uterine weight (i.e., butyl-, isobutyl-, and benzylparaben) and male reproductive-tract effects (i.e., butyl- and propylparaben). However, in relation to estrogen as a control during in vivo studies, the parabens with activity are many orders of magnitude less active than estrogen. While exposure to sufficient doses of exogenous estrogen can increase the risk of certain adverse effects, the presumption that similar risks might also result from exposure to endocrine-active chemicals (EACs) with far weaker activity is still speculative. In assessing the likelihood that exposure to weakly active EACs might be etiologically associated with adverse effects due to an endocrine- mediated mode of action, it is paramount to consider both the doses and the potency of such compounds in comparison with estrogen. In this review, a comparative approach involving both dose and potency is used to assess whether in utero or adult exposure to parabens might be associated with adverse effects mediated via an estrogen- modulating mode of action. In utilizing this approach, the paraben doses required to produce estrogenic effects in vivo are compared with the doses of either 17 β-estradiol or diethylstilbestrol (DES) that are well established in their ability to affect endocrine activity. Where possible and appropriate, emphasis is placed on direct comparisons with human data with either 17 β-estradiol or DES, since this does not require extrapolation from animal data with the uncertainties inherent in such comparisons. Based on these comparisons using worst-case assumptions pertaining to total daily exposures to parabens and dose/potency comparisons with both human and animal no-observed-effect levels (NOELs) and lowest-observed- effect levels (LOELs) for estrogen or DES, it is biologically implausible that parabens could increase the risk of any estrogen- mediated endpoint, including effects on the male reproductive tract or breast cancer. Additional analysis based on the concept of a hygiene-based margin of safety (HBMOS), a comparative approach for assessing the estrogen activities of weakly active EACs, demonstrates that worst-case daily exposure to parabens would present substantially less risk relative to exposure to naturally occurring EACs in the diet such as the phytoestrogen daidzein.
Keywords Breast Cancer, Endocrine Active Chemicals (EACs), Estrogen, Male Reproductive-Tract Effects, Parabens, Risk Assessment
INTRODUCTION
The objective of this review is to assess the potential endocrine activity of paraben compounds that are commonly used as preserving agents and whether they pose any risk to human health. The issue of environmental endocrine-active chemicals (EACs) and whether exposure to such chemicals might adversely affect human health is a subject of intense debate. The basis for the discussion hinges on the issue that while chemicals may mimic some types of activity demonstrated by the naturally occurring hormone estrogen, the magnitude of activity (i.e., potency) is substantially lower and the potential to result in an adverse effect mediated via an estrogen mode of action has not been established in humans. There is no question that exposure to sufficient doses of high-potency exogenous estrogen, either in utero (e.g., diethylstilbestrol, DES) or from adolescence into adulthood (e.g., oral contraceptives, hormone replacement therapy), can increase the risk of deleterious effects. However, the presumption that similar risks might also result from exposure to other chemicals such as EACs with far weaker estrogenic activity is still speculative. In particular, it has been hypothesized that in utero exposure to weakly EACs might be associated with a range of adverse effects on the developing fetus, including reproductive- tract abnormalities in males and increased risk of breast, testicular, or prostate cancer later in life. In reviewing this issue, the U.S. Environmental Protection Agency (U.S. EPA, 1997) noted that “With few exceptions (e.g., DES), a causal relationship between exposure to a specific environmental agent and an adverse effect on human health operating via an endocrine disruption mechanism has not been established” (p. 6).
Due to the fact that both endogenous estrogen (e.g., 17β- estradiol) and exogenous estrogen (e.g., DES) are established risk factors for a range of adverse outcomes, it is prudent to consider whether exposure to exogenous chemicals with substantially weaker estrogenic activity might also have similar effects. In attempting to address this complex issue, it is not sufficient to simply demonstrate that a particular chemical has either in vitro or in vivo estrogenic activity, but rather to consider how this activity compares to the estrogenic activity (i.e., potency) in humans of 17β-estradiol, DES, or phytoestrogens. In addition, the absorption and metabolism of these compounds are also important. Because the effects resulting from in utero and postnatal exposure to estrogen have been well studied in both animals and humans, there is a broad database from which to judge the likelihood that weak EACs might have similar effects in humans.
In this review, a comparative approach involving both dose and potency is used to assess whether in utero or adult exposure to parabens might be associated with adverse effects mediated via an estrogenic mode of action. In utilizing this approach, the paraben doses required to produce effects in vivo are compared with the doses of a known estrogenic compound (e.g., 17β-estradiol or DES) that are required to produce similar in vivo effects. In addition, exposure to parabens is also contrasted with exposure to weakly estrogenic phytoestrogens commonly found in foods. By comparing the potency of parabens with the established estrogenic potency of DES, 17β-estradiol, and several ubiquitous phytoestrogens, it is possible to assess the likelihood that exposure to parabens at typically used concentrations might be capable of causing adverse effects mediated via an estrogen mode of action.
FIG. 1. Structures of paraben compounds.
As illustrated in Figure 1, parabens are a group of alkyl esters of p-hydroxybenzoic acid and typically include methylparaben, ethylparaben, propylparaben, butylparaben, isobutylparaben, isopropylparaben, and benzylparaben, which are considered in this review. Parabens (or their salts) are widely used as preservatives in cosmetics, toiletries, and pharmaceuticals due to their relatively low toxicity (CIR, 1994, 1995; Soni et al., 2001, 2002).
Parabens are lipophilic compounds with increasing octanol/water partition coefficients (K^sub ow^, usually expressed as log K^sub ow^) and antimicrobial activity with increasing molecular weight and length of the alkyl side chain. The molecular weights and log octanol/water partition coefficients of the parabens are summarized in Table 1.
This review summarizes the in vitro data on the interaction of parabens with estrogen receptors in various screening assay cell culture systems, followed by a summary of in vivo data. This is followed by a critical assessment of whether parabens might be associated with increased risk of estrogen-mediated endpoints as a consequence of either in utero or adult exposure. The endpoints considered include adverse effects on the male reproductive tract from in utero exposure and breast cancer in women as a consequence of adult exposure.
TABLE 1
Molecular weights and log k^sub ow^ of parabens
Despite almost 10 years of concern about the potential effects to humans from exposure to EACs, there has been no official guidance on how to systematically assess potential risks. Several expert authoritative bodies have, however, concluded that the use of existing screening level assays is inappropriate for use in the hazard assessment characterization of endocrine modulating chemicals (CSTEE, 1999; SCCNFP, 2001; IPCS, 2002). Therefore, a methodology to assess potential risks from exposure to exogenous EACs must rely substantially on a comparative approach where effects are scaled in proportion to either estrogen or DES. While in vitro studies are rapid, relatively easy to conduct, and offer clear advantages in terms of their ability to screen large numbers of compounds, they suffer from an inability to account for critical in vivo metabolic processes. An obvious analogy would be with mutagenicity testing and carcinogenicity testing, with the former providing important insights into possible outcomes or modes of action and the later critical in confirming the correctness of preliminary observations.
The dilemma concerning how to assess the likelihood that exposure to EAC\s might be etiologically associated with adverse effects has been questioned. In 1993, Sharpe and Skakkebaek put forth the “estrogen hypothesis” as a biologically plausible explanation for an apparent increase in male reproductive tract disorders. They speculated that in utero exposure to a number of weakly estrogenic compounds could explain this phenomenon. In revisiting this hypothesis, Sharpe (2003) noted that “all of the identified ‘environmental estrogens’ possess weak or very weak intrinsic estrogenic activity when measured by conventional in vitro and in vivo assays for estrogenicity…. Based on estrogenic potency, human exposure to the most potent environmental estrogens would need to be at least 1000-fold higher than this level for adverse effects relevant to the human male to be induced, and such levels of exposure are remote” (p. 9). As demonstrated in the following review, because of their extraordinarily weak activity as measured by either in vitro or in vivo estrogen assays, it appears that parabens fall precisely into the category of compounds where extremely weak estrogenic activity and the lack of substantial exposure would lead to negligible potential for adverse effects.
POTENTIAL FOR ENDOCRINE ACTIVITY OF PARABENS
Parabens have been studied in a number of in vitro and in vivo systems that examine the estrogenic activity of these compounds. Both in vitro and in vivo data are valuable in assessing the likelihood that exogenous weakly estrogenic compounds might pose a risk to humans. In vitro screening data provide useful order-of- magnitude approximations between tested compounds and estrogen and also can determine the nature and extent to which substances interact with a hormone receptor, as well as solve mechanistic issues. On the other hand, only in vivo data can provide evidence of how hormonally active substances are influenced by absorption, distribution, metabolism and excretion in an intact animal. Consequently, a combined in vitro/in vivo approach is warranted to gain a complete understanding of the activities of the compound in question. Throughout this evaluation, whether derived from in vitro or in vivo data, potency comparisons between a tested paraben and the estrogen reference compound are noted. For in vitro data, magnitude comparisons are typically based on a comparison of the molar ratio of a tested paraben required to produce an equivalent effect (e.g., competitive binding to ERa in human MCF-7 cells) as the reference compound (e.g., estradiol). In the in vivo studies (e.g., Routledge et al., 1998) potency comparisons are based on the lowest dose of a tested paraben that produced a statistically significant effect (e.g., increased uterine weight) compared to the reference compound used in that study (e.g., 600 mg/kg/day butylparaben produced a significant uterotrophic effect compared to 0.04 mg/kg/day estradiol to yield a potency difference of 15,000). It should be noted that in making potency comparisons, particularly for in vivo effects, the evaluation must also consider the magnitude of the effect. For example, in the preceding illustration, 600 mg/ kg/day significantly increased uterine weight to 38.8 mg compared to 27.1 mg in controls. However, in the same experiment, uterine weight from an estradiol dose of 0.04 mg/kg/day increased uterine weight to 108.7 mg compared to 27.1 mg in controls. While not explicitly considered in the various comparisons presented in this review, it is important to keep in mind that this phenomenon actually provides an additional margin of safety over and above the wide margins derived later.
In Vitro Estrogenicity Data
Byford et al. (2002) studied the estrogenic activity of four parabens (methyl, ethyl, propyl, and butyl) and Darbre et al. (2003) the estrogenic-activity of benzylparaben in estrogensensitive MCF-7 human breast cancer cells. In addition to determining ligand ability to bind estrogen receptor (ER) a in a competitive binding assay, both studies also examined other aspects of estrogenicity including ligand ability to regulate an estrogen-responsive gene (ERE-CAT) transfected into MCF-7 cells, and ligand ability to regulate estrogen-dependent proliferation of MCF-7 cells. In the competitive binding assay, all parabens studied were at least 10,000 to 100,000 times less potent than 17β-estradiol. In MCF-7 cells with a stably transfected estrogen-regulated ERE-CAT reporter gene, the tested parabens showed effects similar to 17β-estradiol, although at 1000-10,000 times greater concentrations. These data are summarized in Table 2, which shows that, depending on the end-point measured and the specific paraben under study, these compounds are from 1000 to 1,000,000 times less potent than 17β-estradiol.
TABLE 2
Summary of in vitro potency data of parabens in MCF-7 cells
Routledge et al. (1998) studied the in vitro estrogen receptor (ER) binding activity of methyl-, ethyl-, propyl-, and butylparaben. In a competitive binding assay using rat ER, butylparaben was able to compete with [3H]estradiol with an affinity approximately five orders of magnitude less than diethylstilbestrol (DES). All of the parabens were positive (i.e., estrogenic) in a yeast estrogen screen (E-screen) with potency in the order butyl > propyl > ethyl > methyl; the main paraben metabolite/>hydroxybenzoic acid (PHBA) was inactive in this assay. Methylparaben was the weakest, with a potency approximately 2,500,000 less than 17/S-estradiol, while the other parabens were several orders of magnitude more potent than methylparaben.
In an extensive assessment of the estrogenic properties of parabens, Okubo et al. (2001) studied the ER-dependent activity of methyl-, ethyl-, propyl-, butyl-, isopropyl-, and isobutylparaben in human breast cancer MCF-7 cells and binding affinity for human ERa and ERβ. In MCF-7 cells, the relative proliferative potency (RPP) of isopropyl- and isobutylparaben (the most potent of the compounds tested) was approximately 170,000 times lower than 17β-estradiol; the RPP of the other parabens were one to two orders of magnitude less than isopropyl- or isobutylparaben. The relative binding affinity (RBA) of parabens for human ERa and ERβ was determined with DES as a positive control. RBA values were greater for parabens with longer or branched side chains, although all were at least 1000 times less potent than DES. The RBAs for methylparaben were not calculated because the binding was so weak. The relative binding affinities to the two estrogen receptors are apparently similar (Okubo et al., 2001; Satoh et al., 2000). All of these reported effects were demonstrated to be related to the ability of parabens to interact with the ER, since the anti- estrogen ICI 182,780 inhibited their proliferative effects in a dose- dependent manner.
In a recombinant yeast assay, cells transfected with the human estrogen receptor (ERα) gene together with expression plasmids (containing estrogen-responsive elements) were incubated in medium containing benzyl-, butyl-, propyl-, ethyl-, or methylparaben to determine their estrogenic activity in comparison with estrogen (17β-estradiol) (Miller et al., 2001). There was a progressive increase in the estrogenic activity with increasing molecular weight with methyl-, ethyl-, propyl-butyl-, and benzylparaben, demonstrating relative responses of 1/3,000,000, 1/200,000, 1/ 30,000, 1/8000, and 1/4000 the potency of estrogen, respectively.
In a similar study, Blair et al. (2000) determined the in vitro rat uterine estrogen receptor RBA for methyl-, ethyl-, propyl-, butyl-, and benzylparaben in comparison with 17β-estradiol (RBA =1). The results of this study demonstrated increasing RBA values with increasing alkyl side-chain length with values of 0.0004, 0.0006, 0.0006, 0.0009, and 0.003, for methyl-, ethyl-, propyl-, butyl-, and benzylparaben, respectively. All of the preceding in vitro data demonstrate that whether compared to DES or 17β- estradiol, parabens are many orders of magnitude less potent based on several measures of estrogenic activity (i.e., ligand binding to or proliferation of MCF-7 cells, regulation of CAT gene expression or ligand binding to ERa or ERβ).
In Vivo Data in Female Animals
Many of the parabens that are the subject of this review have been tested for their in vivo effects in the immature or ovariectomized mouse or rat uterotrophic assay. The results of these assays more closely reflect the full spectrum of in vivo absorption, distribution, metabolism, and excretion. However, as described later, even with in vivo systems, there is some question concerning the biological relevance of subcutaneous dosing studies, since this exposure route may bypass important dermal metabolic processes. Ideally, the results of uterotrophic assays are presented as uterine weight and uterine weight normalized to body weight, and doses are provided as milligrams per kilogram per day. Because doses in the studies presented later were sometimes provided only as dose per animal (rather than per body weight), without providing body weights for each dose group, potency estimates are calculated by dividing the paraben dose/animal (or mg/kg/day dose) that caused an effect by the dose/animal (or mg/kg/day dose) of the study’s reference compound that caused an effect.
The topical application of benzylparaben to immature CD1 mice at a dose of 33 mg (equivalent to approximately 2500 mg/kg/day) produced a significant increase in normalized uterine weight compared to control, although this increase was only one third of that produced by 17β-estradiol at a dose of 100 ng (a potency difference of approximately 330,000) (Darbre et al., 2003). Similarly, subcutaneous injection of isobutylparaben in CDl mice at a dose of 12.0 mg (equivalent to approximately 720 mg/kg/day)* increased normalized uterine weight comparable to a dose of approximately 5 n\g of 17β-estradiol (a potency difference of approximately 2,400,000) (Darbre et al., 2002).
Lemini et al. (1997) reported that subcutaneously administered p- hydroxybenzoic acid (PHBA; the major metabolite of parabens) was estrogenic in both immature and ovariectomized mice as measured by increased cornification of vaginal epithelial cells and increased uterine weight. Compared with 17β-estradiol (1 g/100 g), PHBA demonstrated dose-dependent effects, with a dose of 500/g/100 g producing cornification that was almost comparable in both intact and ovariectomized animals. With respect to effects on uterine weight, in intact animals doses of PHBA of 5 and 50 g/100 g were essentially without effect while a dose of 500/ug/100 g produced an effect comparable to estrogen. The relative uterotrophic potency of PHBA with respect to 17β-estradiol was 1000 times less, as calculated by the authors using molar doses.
In contrast to the positive uterotrophic findings reported by Lemini et al. (1997) for PHBA, similar results were not confirmed by Hossaini et al. (2000) in either immature B6D2F1 mice or Wistar rats exposed to test compounds for 3 days. In this study, PHBA, methyl-, ethyl-, propyl-, and butylparaben were subcutaneously administered to mice at doses of 1, 10, and 100 mg/kg/day for 3 days.[dagger] No dose of PHBA or any of the parabens produced an estrogeniclike response in the mouse uterotrophic assay. Methyl-, ethyl-, and propylparaben also did not result in significant uterotrophic effects following oral exposure at doses up to 100 mg/kg/day for methyl- and propylparaben and 1000 mg/kg/day for ethylparaben. In the rat uterotrophic assay a subcutaneous dose of 5 mg/kg/day of PHBA (the only dose tested) produced no response, while 100-, 400-, and 600mg/kg/day doses of butylparaben produced increased uterine wet weights, with only the highest dose (i.e., 600 mg/kg) showing a significant effect after normalizing for body weight. A subcutaneous dose of 0.1 mg/kg/day estradiol benzoate consistently caused significantly increased normalized uterine weights (Hossaini et al., 2000). The inability of Hossaini et al. (2000) to reproduce the positive uterotrophic findings in mice at a subcutaneous PHBA dose 200 times greater calls into question the validity of positive findings reported by Lemini et al. (1997).
Routledge et al. (1998) examined methyl- and butylparaben (representing the extremes of estrogen receptor binding in the E- screen assay) in the immature rat uterotrophic assay (intact and ovariectomized) following oral and subcutaneous administration over 3 successive days. Uterine weights were not normalized to body weights; the authors provided the results as wet and dry uterine weights. Methylparaben administered orally up to 800 mg/kg/day or subcutaneously at up to 80 mg/kg/day did not increase uterine weights in immature rats. There were no significant effects on uterine weight following oral administration of butylparaben at up to 1200 mg/kg/day. Following subcutaneous administration of butylparaben, the lowest dose that produced a positive effect on wet and dry uterine weight was 600 mg/kg/day. Butylparaben doses of 200 and 400 mg/kg/day significantly increased dry but not wet uterine weights. Subcutaneous doses of 0.04-0.4 mg/kg/day estradiol increased uterine wet and dry weight. A possible explanation for the disparity in results for butylparaben following oral or subcutaneous administration is discussed later in the section on pharmacokinetics.
Depending on the assay (i.e., MCF-7 cells, yeast E-screen, rat or mouse uterotrophic assay, etc.), all of the testing conducted on the parabens, whether in vitro or in vivo, indicates that these compounds are orders of magnitude less potent than estrogens. Parabens with shorter side chains are less potent (i.e., less estrogenic) than those with longer or branched side chains, with the following potency pattern prevailing: isobutylparaben > isopropylparaben > benzylparaben > butylparaben > propylparaben > ethylparaben > methylparaben. As illustrated in Table 3, methyl-, ethyl-, and propylparaben have no in vivo uterotrophic activity; therefore, estimates of potency differences compared to estradiol were not calculated. While high subcutaneous doses of butylparaben and isobutylparaben showed a positive uterotrophic response, both compounds were at least three orders of magnitude less potent than 17β-estradiol.
In Vivo Data in Male Animals
Butylparaben, DES, and a number of other compounds (i.e., bisphenol A, octylphenol, and genistein) were studied in neonatal Wistar rats to determine effects on the male reproductive tract (Fisher et al., 1999). Butylparaben was administered subcutaneously at a dose* of 2 mg/kg on postnatal day (PND) 2-18; DES was administered subcutaneously at doses of 10, 1, or 0.1 g/animal, equivalent to 0.37, 0.037, or 0.0037 mg/kg/day, respectively, while ethinyl estradiol was administered at a dose of 10 g/animal (equivalent to 0.37 mg/kg). Testes weight was determined in treated animals at 18, 25, 35, and 75 days of age; other endpoints considered were effects on the rete testis and efferent ducts and epithelial cell height in the efferent ducts. Both estradiol and DES (at all dose levels, although with decreasing severity) produced significant reductions in testes weight at all days at which this endpoint was assessed, while butylparaben had no effects on testes weight on any day or on any other endpoints studied. Table 4 summarizes the data on the effects of parabens on the male reproductive tract.
TABLE 3
Summary of paraben uterotrophic assay results and potency comparison with estradiol
In a study of the potential reproductive effects in male Wistar rats, 3-week-old animals were exposed to butylparaben at dietary levels of 0.00%, 0.01%, 0.10%, and 1.00% for 8 weeks (Oishi, 2001).[dagger] While there were no significant effects on body weight, there were significant reductions in absolute epididymides and seminal vesicle weights at the highest dose and a significant effect on relative epididymides weight at the two highest doses (i.e., 0.10 and 1.0%); there were no treatment-related effects on weight of the testes, ventral prostate, and preputial gland (either absolute or relative to body weight). All doses showed significant decreased sperm reserves in the cauda epididymides, daily sperm count (DSP) in the testes, and efficiency of sperm production (DSP/ g testes); reduction in sperm concentration in the cauda epididymides was only significant at the highest dose. Serum testosterone levels decreased with dose with significance at the two highest doses (0.10 and 1.0%), although it should be noted that the biological significance of changes in a variable parameter like serum testosterone is perhaps questionable. The author noted that the decrease in sperm number could have been a direct toxic effect of butylparaben consistent with the spermicidal properties reported by Song et al. (1989, 1991). Consequently, it remains to be determined if butylparaben exerts the reported effects on sperm- related parameters as a result of direct toxicity or through an estrogen-mediated mode of action.
TABLE 4
Summary of data on effects of parabens on rodent male reproductive tract
Oishi (2002b) also conducted a study on the effect of butylparaben on reproductive function in male ICR (Crj:CD-1) mice. The mice, aged 27-29 days, were exposed to butylparaben at dietary levels of 0.00%, 0.01%, 0.10%, and 1.00% for 10 weeks.* There was no effect on body weight at any dose; the only significant effect on potential target organs was an increase in absolute and relative epididymides weight at the highest dose. There were no significant reductions in sperm counts, but significant decreases in elongated spermatids at the lowest dose (0.01%) and round spermatids at the highest dose were observed. Serum testosterone levels were significantly decreased at the highest dose (1.0%). No lesions were noted in Sertoli or Leydig cells at any dose. The reason for the paradoxical increase in epididymides weight and decreased testosterone at the highest dose could not be explained.
Oishi (2002a) also conducted a study on the effect of propylparaben on reproductive function in male Wistar rats. In this study, 3-week-old animals were exposed to propylparaben at dietary levels of 0.00%, 0.01%, 0.10%, and 1.00% for 4 weeks.[dagger] While final body weights were significantly lower at the highest dose, there were no significant reductions in absolute or relative weights of testes, epididymides, ventral prostate, seminal vesicles, or preputial gland at any dose. The two highest doses produced significant reductions in sperm reserves and concentration in the cauda epididymides; all doses produced decreases in daily sperm count (DSP) in the testes and efficiency of sperm production (DSP/g testes). Serum testosterone levels appeared to decline with dose (no trend test was done); however, a significant decrease was only observed at the highest dose (1.0%).*
In another study to investigate the effects of in utero exposure to offspring, butylparaben was injected subcutaneously into pregnant Sprague-Dawley rats at doses of 100 or 200 mg/kg/day[dagger] from gestation day 6 to postnatal day (PND) 20, with a 2-day interruption during parturition (Kang et al., 2002). The proportion of pups born alive was significantly reduced at both doses, while the proportion of pups surviving to weaning was significantly reduced at the highest dose. Male reproductive organ weights showed a confusing pattern of significant effects (i.e., increases and decreases), with no consistent relationship with dose or time. Sperm counts in the caudal epididymis and sperm motility were significantly decreased at both doses (see earlier footnote on average propylparaben intake). In female offspring there were no effects on uterus or ovary weights from any dose atany PND. There were no effects on anogenital distance in male or female offspring and a significant effect on vaginal opening at 100 mg/kg, but not at 200 mg/kg.
Finally, a recent study with methyl- and ethylparaben demonstrated no effects on the secretion of sex hormones or male reproductive function following dietary administration to 25-to 27- day-old rats at doses of 0.1 % and 1.0% (Oishi, 2004). Neither compound produced adverse effects on sperm or the weights of the testes, epididymis prostate, seminal vesicles, testosterone, luteinizing hormons (LH), or follicle-stimulating hormons (FSH) at a dose of approximately 1000 mg/kg/day.
As noted in most of the studies just cited that reported a paraben effect on sperm production or a sperm-related parameter, parabens (methyl, ethyl, propyl, and butyl) have been reported to have potent in vitro spermicidal activity (Song et al., 1989, 1991). This activity (i.e., butyl > propyl > ethyl > methyl) has been studied in conjunction with the development of parabens as possible vaginal contraceptive agents, with butylparaben six times more potent than methylparaben in studies with human sperm. Butylparaben is the most potent of the parabens and the most well studied. It appears that butylparaben produces spermicidal effects primarily as a consequence of impairment of sperm membrane function. It is, therefore, biologically plausible that the paraben doses used in the studies reviewed could produce similar in vivo reductions in the various sperm-related parameters as a result of direct toxicity and not as a consequence of an estrogen-mediated mode of action.
As shown in Table 4, notwithstanding potential effects on sperm or sperm-related parameters, to the extent that other effects are reported, virtually all occurred in conjunction with the highest administered butylparaben dose. With the exception of the study by Fisher et al. (1999), none of the studies included an estrogen “control” group. This would have permitted a direct comparison of the in vivo potency of a tested paraben with estrogen (e.g., 17β-estradiol or DES).[double dagger] As noted earlier, the effects on sperm might have been due to direct toxicity and not from an estrogen-mediated mode of action. However, the other reported effects (i.e., changes in some reproductive organ weights and testosterone) could have been the result of an estrogen-mediated mode of action. Consequently, these data (i.e., the lowest doses that produced a nonsperm effect) in conjunction with human paraben exposure data are used later in this review in the section that compares the dose/potency of parabens with either DES or 17β- estradiol. As stressed throughout this review, the only way to assess the potential effects of parabens that might be due to an estrogen-mediated mode of action is to account for both the dose and potency in comparison to the established effects of estrogen.
PHARMACOKINETICS
The pharmacokinetics (i.e., absorption, distribution, metabolism, and excretion) of parabens may influence whether exposure to these compounds might have the capability of acting as EACs in humans. Notwithstanding their substantially lower potency than estrogen, determining the extent to which these compounds can reach potential target organs following exposure will have critical implications in a risk evaluation.
Following oral exposure, all parabens that have been studied are rapidly and almost completely absorbed from the gastrointestinal tract in rats, rabbits, and dogs (CIR, 1994, 1995; Soni et al., 2001, 2002; Jones et al., 1956). Following absorption, parabens are easily hydrolyzed by liver esterases, with phydroxybenzoic acid (PHBA) as the primary metabolite. Other metabolites include p- hydroxyhippuric acid and ester and a variety of conjugated glucuronides. Metabolism is rapid with metabolites appearing in the urine within 0.5 h after ingestion.
The metabolism of parabens (methyl, ethyl, propyl, and butyl) and PHBA was studied in fasted dogs following intravenous (iv) administration of 50 mg/kg or oral administration of 1 g/kg (Jones et al., 1956). Depending on the paraben, recovery in the urine of total administered dose (parent + metabolites) ranged from 40 to 95% suggesting that the majority of these compounds retained their basic phenolic character. In this study, a fasted human was given an oral dose (70 mg/kg) of methylparaben and no detectable parent compound was found in the plasma or urine. Finally, no accumulation was noted in dogs orally administered either methyl- or propylparaben at 1.0 g/ kg/day for 1 year. Following oral administration to rabbits of either 400 or 800 mg/kg isobutylparaben, between 32 and 50% of the parent compound was recovered in the urine as free or conjugated PHBA after 24 h. Only about 0.2-0.9% of the parent compound was detected in the urine as unmetabolized isobutylparaben (CIR, 1995).
The metabolism and excretion of methylparaben (and to a lesser extent propylparaben) has been studied in six preterm infants (four males and two females) as a consequence of their use as preservatives in gentamicin, an antibiotic used to treat sepsis (Hindmarsh et al., 1983). Following intramuscular administration, urine was analyzed by high-performance liquid chromatography (HPLC) to determine the percent of dose excreted over 18-24 h. The percentage of administered methylparaben accounted for in urine was from 13.2 to 88.1%, with the majority in conjugated form. The only metabolite detected was PHBA.
As summarized in Table 3, there is no evidence that methyl-, ethyl-, or propylparaben is estrogenic in vivo in uterotrophic assays following either subcutaneous or oral administration. This is consistent with the observation, from in vitro assays that screen for estrogenic activity, that parabens with shorter or unbranched side chains have less activity than those with longer or branched side chains. While in vitro screening data are useful in assessing activity relative to estrogen, in taking a weight-of-evidence approach, primary emphasis must be placed on in vivo data. Butylparaben and isobutylparaben have been reported to be estrogenic in uterotrophic assays following subcutaneous administration. It is unknown if a similar effect would occur following topical administration. Benzylparaben was positive following topical application albeit with a potency >300,000 times less than estradiol. Unfortunately, there are insufficient data on the potential in vivo uterotrophic effects of butyl-, benzyl-, isopropyl- , and isobutylparaben following both oral and topical exposure to permit a definitive determination of the effect of route of administration on this endpoint. While liver esterases clearly come into play in the metabolism of parabens following oral exposure with first-pass hepatic activity leading to substantial conversion to PHBA and other metabolites, as discussed later, the situation with respect to subcutaneous administration or topical application may not be equivalent to oral administration.
Because most paraben-containing products are topically applied to the skin, dermal absorption is particularly important with respect to estimating doses and potential effects. As described in the following paragraphs, the skin is not an impenetrable barrier, and depending on the specific paraben (i.e., K^sub ow^) and the type of formulation (e.g., aqueous or lipid), some fraction of an applied dose may be absorbed and enter the systemic circulation, although not necessarily as the parent compound.
In an in vitro study using human epidermal membranes, the ability of different vehicles (acetone, ethanol or ointment) to affect the permeability of four parabens (methyl, ethyl, propyl, and butyl) with or without occlusion was determined (Cross and Roberts, 2000). While the skin permeability of the parabens alone was not measured, the results demonstrated that different vehicles, particularly in conjunction with occlusion, enhanced skin permeability. For example, the percentages of butylparaben in ointment, acetone, or ethanol penetrating the skin over a period of 10 h were 39%, 44%, and 57%. These results suggest that parabens formulated into certain skin- care products can penetrate the skin. The determinants of actual dermal permeability are likely to be a complex interaction of individual parabens, partition coefficient (i.e., log K^sub ow^), and the presence of other ingredients in a particular product. The skin permeability of four parabens (methyl, ethyl, propyl and butyl) was studied in an in vitro system using guinea pig skin (Kitagawa et al., 1997). Permeability increased as a function of n-octanol/water partition coefficients (K^sub ow^) in the order butyl > propyl > ethyl > methyl. This was interpreted as demonstrating that these parabens penetrate both human and animal skin via nonpolar stratum corneum lipid lamella. However, neither of these in vitro studies could account for dermal metabolism of parabens as described next.
Butylparaben has also been studied in an in vivo absorption experiment in Wistar rats. After 4 h, almost 50% of the ^sup 14^C- labeled administered dose was recovered in the urine, and 10% was recovered in the skin (Yamashita et al., 1994). Similarly, in an in vitro study with butylparaben, 88% of the applied dose was recovered at the end of the skin diffusion experiments (Koyama et al., 1994). Because both of these studies were based on recovery of ^sup 14^C- radioactivity, it is not possible to determine how much of the recovered compound was parent compound or metabolites.
An important aspect of metabolism pertaining to dermal absorption is provided by several in vitro studies demonstrating substantial enzymatic hydrolysis of parabens in skin by cutaneous esterases. In an in vitro study with rat skin to determine percutaneous penetration of propyl- and butylparaben, fullthickness abdominal skin from male Wistar rats was mounted in a flo\w-through diffusion cell (Bando et al., 1997). Experiments were conducted with and without pretreatment with diisopropyl fluorophosphate (DFP), an irreversible lipophillic inhibitor of cholinesterase. For butylparaben and propylparaben, respectively, without pretreatment with DFP, approximately 96% and 30% from either compound of the total penetrated amount was the primary metabolite PHBA. No PHBA from either compound was detected in receptor fluid following pretreatment with DFP. This study demonstrates the importance of dermal metabolism in reducing the amount of parabens that might enter the systemic circulation.
The enzymatic hydrolysis of parabens (methyl, ethyl, propyl, and butyl) has also been studied in extracts from different layers of human skin, where it was possible to distinguish four carboxyesterases capable of hydrolyzing 4-hydroxybenzoic acid esters (Lobemeier et al., 1996). The anatomic site from which the various esterases were derived (i.e., cutis, keratinocytes, subcutaneous fat, blood) determined their hydrolytic activity. Parabens with shorter side chains (i.e., methylparaben) were the preferred substrate for the subcutaneous esterase I, with decreasing activity as chain length increased. Subcutaneous esterase II demonstrated preferential activity for butylparaben over methylparaben. Butylparaben was the preferred substrate of the esterase III from keratinocytes with less activity as chain length decreased. As noted by Lobemeier et al. (1996), “We conclude from our results that the above mentioned paraben esterase III of keratinocytes is sufficient to completely hydrolyze the traces of parabens that may enter the skin from topically applied ointments” (p. 650). The ultimate determination of how these findings might apply in an in vivo situation must await an appropriate study to confirm these results.
The ability of dermal esterases* to substantially degrade (hydrolyze) topically applied parabens may have important implications with respect to decreasing the potential for endocrine activity for the three parabens that have demonstrated in vivo activity. As already described, PHBA, the major paraben metabolite, is devoid of estrogenic activity as demonstrated by Hossaini et al. (2000), who was unable to confirm the positive uterotropic findings of Lemini et al. (1997) in rats or mice.t To the extent that parabens might be absorbed through the skin suggests that they would be metabolized to PHBA, a nonestrogenic substance. The activity of dermal esterase may also be important in interpreting the results of studies in which paraben test substances are administered subcutaneously, thereby bypassing much (if not all) of likely dermal metabolism. In the only study in which a paraben (i.e., butylparaben) was administered both orally and subcutaneously (Routledge et al., 1998), there were differential effects. Following oral exposure (at doses up to 1200 mg/kg/day), butylparaben was inactive in the immature rat uterotrophic assay; however, a subcutaneous dose of 600 mg/kg/day produced a positive uterotrophic response (increased wet and dry weights). These data suggest that the rapid absorption and metabolism of orally administered butylparaben could account for its lack of uterotrophic activity, while subcutaneous doses may have bypassed or overwhelmed (to an unknown extent) the metabolic capabilities of the skin.
EXPOSURE TO PARABENS
In determining the safety of certain parabens, it is first necessary to ascertain the extent of their use. This information, along with the amounts of various parabens that are formulated into consumer products, can provide an approximation of potential exposures. Based on voluntary registration with the U.S. FDA, Steinberg (2004) compiled the most recent tabulation of preservative use in cosmetic formulations. This tabulation demonstrates that methyl-, propyl-, and ethylparaben are used in 82% of the 17,907 cosmetic formulations containing preservatives listed in the database in 2003. While very weakly estrogenic in some in vitro assays, these three parabens have not demonstrated any estrogenic activity in in vivo assays, suggesting there is no biological basis for including them in an assessment of potential estrogen-mediated risks. The other parabens, butyl-, isobutyl-, isopropyl-, and benzylparaben, were contained in 13%, 1.3%, 0.2%, and 0.02%, respectively, of the cosmetic formulations containing preservatives.
Because of their very limited use in cosmetic formulations, and therefore a limited opportunity for exposure, isobutyl-, isopropyl- , and benzylparaben are not considered in this analysis. However, with butylparaben present in 13% of cosmetic formulations (i.e., >2300 products), there is an opportunity for exposure to this paraben. In a recent survey, the Cosmetic, Toiletry, and Fragrance Association (CTFA) requested its member companies provide the maximum concentration of butylparaben used in FDA cosmetic product categories. The maximum concentrations of use for butylparaben ranged from 0.00004 to 0.5%. Parabens can not be used at higher concentrations as their use is limited by solubility.
Exposure of a 60-kg woman using multiple product categories (bath products, colognes and toilet waters, powders, hairsprays, shampoos, tonics and other hair grooming aids, blushers, foundations, lipstick, makeup bases, bath soaps and detergents, deodorants, skin cleansers, depilatories, face preparations, body moisturizers, skin fresheners, and sun products) was estimated by multiplying estimated product exposure in milligrams per kilogram per day by the reported maximum butylparaben concentration in the product category. This resulted in a total composite butylparaben exposure of 0.307-1.02 mg/ kg/day.
Based on Cross and Roberts (2000), it is reasonable to assume that approximately 40% of butylparaben could be dermally absorbed, which results in a butylparaben dose of approximately 0.12-0.41 mg/ kg/day. It is important to note that this estimate assumes that an individual uses every product in the above list every day (i.e., worst-case assumption). The calculated butylparaben dose of 0.12- 0.41 mg/kg/day also assumes that dermal esterases do not further reduce or completely metabolize this dose to p-hydroxybenzoic acid (PHBA). In the following sections, potential risks from parabens as a consequence of their possible weak estrogenic activity are assessed by comparing the paraben doses required to produce an in vivo effect with total paraben exposure levels (i.e., 0.12-0.41 mg/ kg/day) and contrasting these levels and their estrogenic potency with amounts of estrogen known to produce adverse effects.
EXPOSURE TO PHYTOESTROCENS
A useful approach for providing a perspective on this issue is to compare the estrogen receptor binding affinity of parabens with that of dietary phytoestrogens (i.e., naturally occurring plant estrogens). As a class of structurally diverse and ubiquitous dietary compounds, phytoestrogens can have significant biological activity based primarily on their ability to act as weak estrogens by binding to the estrogen receptor (ER) (Burroughs et al., 1985). More than 300 foods have been shown to contain phytoestrogens. Coumestrol, found in soy protein, sunflower, and many legumes, is the most estrogenic phytoestrogen studied and binds to estrogen receptors with less affinity than 17β-estradiol but with more affinity than other phytoestrogens. Zearalenone is a fungal contaminant of grains such as corn, wheat, barley, and oats, while genistein, which is less potent than coumestrol, is also found in various grains and soybean products. Phytoestrogens are capable of producing a spectrum of biological activity and, depending on the dose, may have estrogenic or antiestrogenic effects (Setchell et al, 1984).
In a comprehensive assessment of the interactions of a number of weakly estrogenic chemicals, including phytoestrogens, with both estrogen receptor a (ERa) and estrogen receptor β (ERβ), Kuiper et al. (1998) compared the relative potency of 17β- estradiol and other natural estrogens with several dietary phytoestrogens (e.g., genistein, coumestrol, zearalenone). These in vitro studies with ERa and ERβ proteins and a transient gene expression assay demonstrate that naturally occurring dietary phytoestrogens have far less binding affinity for either ERa or ERβ than 17β-estradiol. Safford et al. (2003) also assessed the relative estrogen receptor binding affinity of these compounds and 17β-estradiol with ERα and ERβ* Since neither of these studies included parabens in their evaluations, the data on essentially the same endpoint (i.e., binding affinity to ERa and ERβ) from Okubo et al. (2001) are used for comparison. While these three approaches are slightly different, the results illustrate that in vitro, many commonly consumed phytoestrogens have greater binding affinity to estrogen receptors than parabens.
In the case of phytoestrogens, their ubiquitous (and growing) presence in the food supply, and their likely daily intake suggests widespread exposure. A number of studies have demonstrated that peak serum concentrations of daidzein and genistein of 500-1000 nM can be achieved after meals rich in soybeans or soybean protein (Kurzer and Xu, 1997;Lapciketal., 1997). As illustrated in Table 5, phytoestrogens are more potent than parabens with respect to their ability to bind to the ER.
DO HUMAN EXPOSURES TO PARABENS POSE ANY POTENTIAL RISKS DUE TO ENDOCRINE-MODULATING EFFCECTS?
Extensive safety evaluations of parabens have previously been conducted, concluding that they are safe for use in numerous consumer products following either dermal or oral exposure (CIR, 1994, 1995; Soni et al., 2001, 2002). However, none of these assessments explicitly considered potential endocrine activity. Because the estrogen like properties of the parabens are the subject of this review\, this is the focus of the following evaluation of the potential risks to humans that might be associated with these compounds due to an estrogen-mediated mode of action.
TABLE 5
Comparative binding affinity of 17β-estradiol, des, dietary phytoestrogens. and parabens
Despite a lack of consistent regulatory guidelines on assessing potential risks from chemicals identified as endocrine-active compounds, there appears to be a consensus within the scientific community on the broad outlines of how such evaluations should be undertaken. In numerous reviews of this issue, there is general agreement that in evaluating potential endocrine-mediated effects of weakly estrogenic xenobiotic chemicals, potency and dose are critical aspects of a risk assessment process (Baker, 2001; Witorsch, 2002; Degen and Bolt, 2000; Bolt et al., 2001; Ashby, 2000a, 2000b, 2001; Williams et al., 2001). While in vitro screening data can provide useful information about comparative potency, molecular mechanisms involved and hormone receptor activity, in vivo data are afforded more weight than in vitro data because only in vivo data can account for critical pharmacokinetic considerations influencing target organ effects including absorption, distribution, metabolism and excretion. This concept is consistent with a review of this issue by IPCS (2002): “Other chemicals that have shown evidence in vitro of estrogenic activity have not shown similar evidence in in vivo systems, and caution is warranted in interpreting in vitro results without in vivo confirmation” (p. 2412). A common thread that runs through all of these discussions is that simply equating doses without consideration of potency is inappropriate, particularly in comparison with the well established effects of potent estrogens such as DES, ethinyl estradiol, or 17β-estradiol. For example, in reviewing the issue of endocrine active compounds and their effects on the male reproductive system, Williams et al. (2001) concluded that “weak environmental oestrogens are rather unlikely to pose a significant risk to the reproductive system of the developing male unless the compound in question also possesses some other biological activity of relevance” (p. 245).
TABLE 6
Comparative potency of xenoestrogens in the rodent uterotrophic assay
As illustrated in Table 6, comparing effective doses of compounds that produce a positive response in the rodent uterotrophic assay does not give a true picture of their comparative potency. Table 6 also illustrates that common foodstuffs containing naturally occurring phytoestrogens are more potent in producing effects in the rodent uterotrophic assay than parabens. Based on daily intakes of naturally occurring dietary phytoestrogens, it would be unwarranted to simply assume that exposure to less potent parabens would be associated with risks of estrogen-mediated effects.
The relative risk may even be lower considering the two phases of the uterotrophic response to estrogen as reported by Hewitt et al. (2004). In this study, early responses to estrogen included hyperemia and water imbibition, while later responses included DNA synthesis and mitosis of epithelial cells, resulting in real uterine growth. A time-resolved analysis of the uterotrophic response in ovariectomized rats to the weak estrogen genistein revealed that this phytoestrogen mimicked the water imbibition but did not stimulate cellular markers of proliferation (Diel et al., 2004). In other words, weak estrogens may be able to stimulate water imbibition, but are not able to trigger deleterious proliferative responses, which creates an addition margin of safety.
Effects of DES, 17β-Estradiol, and Ethinyl Estradiol in Humans and Animals
As described in this review, there are data showing that some individual paraben compounds have weak activity in a variety of in vitro and in vivo screening assays for estrogenic effects. However, in assessing the likelihood that in utero exposure to these compounds might be etiologically associated with any particular adverse effect, the biological plausibility of such effects must be considered in a comparative approach as part of the evaluation process. This is based on the extensive human and animal data on the effects of pre- and post natal exposure to potent estrogens. In this section, the data on DES, 17β-estradiol (E2), and ethinyl estradiol (EE) are summarized to illustrate the effects in humans and rodents from in utero exposure to these potent estrogens. The subsections compares these data on a dose and potency basis with the paraben potency data reviewed earlier.
DES
Considerable dose-response data, in both humans and experimental animals, demonstrate the effects of in utero exposure to DES on offspring. While the human DES data can be used to judge the likelihood that estrogen-mediated adverse effects might result from in utero exposure to exogenous estrogenic compounds, because of the manner in which DES was used clinically, these data are not useful for assessing the likelihood of adverse effects which might result from postnatal exposures.
DES is a synthetic estrogenic compound which is equal or more potent than 17β-estradiol and hundreds to thousands of times more potent (based primarily on results from in vitro tests) than compounds identified as EACs (McLachlan, 1981; Gaido et al., 1997; Golden et al., 1998). The large difference in potency between DES and parabens can be used to assess the likelihood that in utero exposure to these compounds might be a risk factor for certain adverse health endpoints. Based on extensive human experience, the DES database is widely recognized as the “gold standard” for assessing potential effects from exposure to weakly estrogenic EACs (U.S. EPA, 1997; IPCS, 2002; NAS, 1999). The maternal doses of DES that produce adverse effects in offspring exposed in utero and likely maternal intakes of weakly estrogenic exogenous compounds required to result in approximately comparable estrogenic activity can be directly compared. This approach is the most valid way to address this issue because of the enormous disparity between the estrogenic potency of 17β -estradiol and any EACs identified to date (including the parabens) and also because critical comparative data were derived from studies in humans.
Effects of DES in Humans
There are clinical studies that suggest that some maternal DES dosing regimens were not sufficient to result in certain adverse effects to offspring exposed in utero (Golden et al., 1998). A study conducted on a cohort of males from the Mayo Clinic, which appears to have employed the lowest DES dosing schedule of any clinical center in the United States, demonstrates a mean no effect maternal dose level for adverse effects attributable to in utero DES exposure (Leary et al., 1984). DES did not increase the risk of any urogenital abnormalities, including penile length or diameter, testis length or width, epididymal cysts, or varicocele. There were also no adverse effects on sperm volume, density, motility, morphology, or fertility. In this study, the median total administered maternal dose of DES of 720 mg (range 4-7405 mg; mean = 1429 mg) and a median exposure duration of 101 days (range 2-239 days) were not associated with any of the male reproductive-tract endpoints studied. Assuming a mean total DES intake of 1429 mg represents an approximate daily DES dose of 14 mg/day (approximately 0.3 mg/kg/day for a 50-kg woman). A potential problem with any such studies (whether in humans or animals) is if fetal exposure to DES occurs after the critical period of reproductive-tract development. Records of DES doses and durations of exposure in this cohort reveal that the 25th and 50th percentiles of the first gestational day of exposure were on days 54 and 89 with exposure durations of 28 and 100 days, respectively (Leary et al., 1984). Therefore, the timing of DES exposure is well within the window during which development of the reproductive tract is occurring.
In contrast, at the University of Chicago where mean total maternal DES doses averaged over 11,000 mg with exposure also beginning during the first trimester, numerous adverse effects on the male reproductive tract have been reported in offspring exposed to DES in utero, including microphallus, epididymal cysts, cryptorchidism, testicular hypoplasia, decreased sperm count, and increased incidence of abnormal sperm (Labarthe et al., 1978; Bibbo and Gill, 1977). Most of the women in the treatment group received total DES doses between 11,500 mg and 12,600 mg (Wilcox et al., 1995), and in utero exposure to DES occurred during a critical time of reproductive-tract development. Assuming a total DES intake of 12,000 mg and 28 weeks (196 days) results in an approximate daily DES dose of 50 mg/day or approximately 1 mg/kg/day for a 50-kg woman. Even though this dosing regimen caused a number of male reproductive-tract abnormalities, there was no significant difference in fertility between men exposed in utero to DES and controls (Wilcox et al., 1995). The failure of high maternal doses of DES to adversely affect the fertility of male offspring exposed in utero challenges the notion that in utero exposure to weakly estrogenic compounds might impair male fertility or affect the development of the reproductive tract.
Effects of Estradiol and Ethinyl Estradiol in Rodents
In what is likely the definitive study of the effects of 17β- estradiol (E2) in rodents, Cook et al. (1998) conducted a 90-day, one-generation reproduction study in male and female Crl:CD Br rats with dietary levels of 0, 0.05, 2.5, 10, and 50 ppm. Numerous measurements were made in parental (P^sub 1^) and first-generation (F^sub 1^) male rats, including serum testosterone levels, testicular and epididymal weights, and a variety of sperm parameters. In P^sub 1^ rats, exposure at 10 and 50 ppm decreased testis and epi\didymis weights, while no effects were seen at 2.5 ppm. In F^sub 1^ males epididymis weight was significantly decreased at 2.5 ppm, which remained following a 26-day recovery period. No pups were produced at 10 or 50 ppm. While testes weight was not affected at any dose level, testicular spermatid numbers in F^sub 1^ rats were significantly increased in the 0.05 and 2.5 ppm groups to 135 and 140% of control values, respectively.* Epididymal sperm number was significantly decreased in the 2.5-ppm group, as was the percent sperm with normal morphology, although this later finding was judged biologically insignificant due to the small magnitude of the change which was also within the expected historical range of control values.
O’Connor et al. (1998) conducted a Tier I screening battery to compare responses from exposure to 17β-estradiol in these tests with the results from the study by Cook et al. (1998). The battery in this study included short-term in vivo tests in male and female (ovariectomized) adult rats and an in vitro yeast transactivation system to identify responses to receptor-specific ligands. Doses of 17β-estradiol used in the in vivo tests were 0, 1.0, 2.5, 7.5, and 50 g/kg/day administered by intraperitoneal injection. In the female battery uterine weight was significantly increased at a dose of 2.5 g/kg/day while uterine cell proliferation and cell height was significantly increased at all doses. In the male battery testes and epididymides weights were significantly increased at 50 g/kg/day, while seminal vesicle weight was increased at 7.5 g/kg/day. This study demonstrated that the screening battery in adult rats was as sensitive as the in utero 90-day, one-generation reproduction study in male and female rats described earlier.
Subcutaneous administration of 17β-estradiol to adult male rats at doses of 0.1, 10, 100, 200, 300,400, and 1000 g/kg/day produced a spectrum of effects on the reproductive tract (Gill- Sharma et al., 2001). While there was a significant effect on sperm motility at doses of 0.1-10 g/kg/day, there were no adverse effects on sperm counts or potency or any aspect of pregnancy outcome. At doses of 0.1-10 g/kg/day there were no effects on weights of the testes, seminal vesicles, ventral prostate, or epididymides, and no effects on serum levels of testosterone, LH, or FSH or any gross morphological changes in the testes compared to controls. In another study, subcutaneous administration of 17β-estradiol at a dose of 1 g/day (approximately equivalent to 5 g/kg/day based on a weight of 200 g) for 5 weeks had no effects on testes weight, but produced significant decreases in the weight of the epididymides, seminal vesicle, and ventral prostate, as well as sperm count and motility (Ganguly et al., 1992).
Oral ethinyl estradiol (EE) was administered to adult male Sprague-Dawley rats at doses of 0.1, 0.3, 3, or 10 mg/kg/day for 4 weeks prior to mating to determine potential effects on reproductive function (Iwase et al., 1995). The two highest doses of EE resulted in almost a complete absence of sperm in the epididymides of treated animals, while a dose of 0.3 mg/kg/day caused a significant reduction in sperm count, but had no effect on motility, while the 0.1-mg/kg/day dose had no effect on sperm count or motility. However, the lowest dose (0.1 mg/kg/day) produced a significant reduction in body weight as well as reductions in prostate and seminal vesicle weights relative to body weight and a reduction in absolute (but not relative) testes weight. Reproductive performance was not affected by either of the lowest doses. All of the preceding data illustrate that sufficient doses of estrogen can have deleterious effects on adult male reproductive function and provide a comparative basis for assessing whether exposure to weakly estrogenic substances might have similar effects.
Assessment of Potential Risks from Estrogen-Mediated Effects Following In Utero Exposure to Parabens
Comparisons Based on Human DES Data
Based on the Mayo Clinic data a DES dose of approximately 0.3 mg/ kg/day is a no-observed-effect level (NOEL) for adverse effects on the male reproductive tract. As summarized in Table 4, the two lowest doses of butyl- and propylparaben (which is not positive in the uterotrophic assay) associated with a significant effect on the male reproductive tract (i.e., decreased daily sperm production) were 10 and 12 mg/kg/day, respectively.* These doses are approximately 30-40 times greater than the human DES NOEL of 0.3 mg/ kg/day and approximately 10 times greater than the DES “effect” level of 1 mg/kg/day. As already reviewed, there is considerable uncertainty as to whether any paraben-induced effect on sperm parameters is estrogen-mediated or due to a direct cytotoxicity to developing spermatocytes (Oishi, 2001, 2002a, 2002b; Song et al., 1991). Therefore, it is inappropriate to use these data for comparing the effects of parabens on sperm parameters to the effects of DES on these same parameters. On the other hand, the 2-mg/kg/day subcutaneous dose of butylparaben reported by Fisher et al. (1999) was a NOEL for male reproductive effects, which is 6 times greater than the human DES NOEL of 0.3 mg/kg/day. However, this approximation does not account for the estrogenic potency differences between DES and any of the parabens with demonstrated in vivo estrogenic activity.
As summarized in Table 3, there are only three parabens with demonstrated in vivo activity in the rodent uterotrophic assay (i.e., butyl-, benzyl-, and isobutylparaben), with potencies from approximately 6000 to 3,300,000 less than estradiol. Assuming that DES is equal in potency to 17β-estradiol and applying the most conservative potency estimate to the human DES NOEL suggests that butylparaben is at least 6000-fold below the dose of DES that might cause effects on the male reproductive tract. Based on the estimated maximum total but
Comments