TOXICOLOGY AND APPLIED PHARMACOLOGY 153, 12–19 (1998)
`ARTICLE NO. TO988544
`
`Some Alkyl Hydroxy Benzoate Preservatives (Parabens) Are Estrogenic
`
`Edwin J. Routledge,*,1 Joanne Parker,* Jenny Odum,† John Ashby,† and John P. Sumpter*
`
`*Department of Biology & Biochemistry, Brunel University, Uxbridge, Middlesex, UB8 3PH, United Kingdom, and †Zeneca Central Toxicology Laboratory,
`Alderley Park, Macclesfield, Cheshire, SK10 4TJ, United Kingdom
`
`Received May 5, 1998; accepted August 6, 1998
`
`Some Alkyl Hydroxy Benzoate Preservatives (Parabens) Are
`Estrogenic. Routledge, E. J., Parker, J., Odum, J., Ashby, J., and
`Sumpter, J. P. (1998). Toxicol. Appl. Pharmacol. 153, 12–19.
`
`The inadvertent estrogenicity of certain synthetic chemicals,
`and their subsequent effects on the endocrine system of humans
`and wildlife, is of concern. In this paper we report findings from in
`vitro and in vivo (uterotrophic) studies which confirm that a range
`of alkyl hydroxy benzoate preservatives (parabens) are weakly
`estrogenic. In a receptor-binding assay, butylparaben was able to
`compete with 3H-estradiol for binding to the rat estrogen receptor
`with an affinity approximately 5 orders of magnitude lower than
`that of diethylstilboestrol, and between 1 and 2 orders of magni-
`tude less than nonylphenol. In an in vitro yeast-based estrogen
`assay, the four most widely used parabens (namely methyl-, ethyl-,
`propyl-, and butylparaben) were all found to be weakly estrogenic
`with the most potent (butylparaben) being 10,000-fold less potent
`than 17␤-estradiol. The estrogenic activity of parabens was inhib-
`ited by 4-hydroxy tamoxifen in vitro, illustrating the requirement
`of these chemicals to interact with the estrogen receptor in order to
`activate the yeast. When administered orally to immature rats, the
`parabens were inactive. However, subcutaneous administration of
`butylparaben produced a positive uterotrophic response in vivo,
`although it was approximately 100,000 times less potent than
`17␤-estradiol. Given their use in a wide range of commercially
`available topical preparations, it is suggested that the safety in use
`of these chemicals should be reassessed, with particular attention
`being paid to estimation of the actual levels of systemic exposure
`of humans exposed to these chemicals. The acquisition of such
`data is a prerequisite to the derivation of reliable estimates of the
`possible human risk of exposure to parabens. © 1998 Academic Press
`
`It is now realized that certain synthetic compounds, used in
`a wide range of products, can mimic the effects of the main
`natural estrogen, 17␤-estradiol, by binding to the estrogen
`receptor and influencing the expression of estrogen-dependent
`genes. This realization has coincided with epidemiological
`data, suggesting a progressive decline in human male repro-
`ductive health and fertility (Carlsen et al., 1992, 1995; Auger
`et al., 1995; and references therein) which, if real, may be
`
`associated with exposure to endocrine-disrupting chemicals
`(EDCs) in the environment, and in particular, those which
`mimic the action of natural estrogens (Sharpe et al., 1995). The
`importance of estrogen action on male reproduction is illus-
`trated by the fact that the ␣ form of the estrogen receptor
`(ER-␣), and the more recently discovered ER-␤, are expressed,
`alone or together, throughout the male reproductive tract (Hess
`et al., 1997; Fisher et al., 1997; Kuiper et al., 1996). Moreover,
`new evidence shows that estrogen regulates reabsorption of
`luminal fluid in the head of the epididymis and may therefore
`be an important determinant of sperm concentration and hence
`fertility (Hess et al., 1997). Although the reports on humans
`remain conjecture, a field of research embracing endocrinology
`and toxicology was created to resolve this issue.
`Although many natural and synthetic chemicals are in wide
`use and are ubiquitous in the environment, little is known about
`the potential risks to humans following exposure to known
`xenoestrogens, or their potential routes and levels of exposure.
`Furthermore, in the absence of a systematic study of all the
`major production chemicals, it is possible that humans are
`exposed to a number of, as yet, unidentified estrogenic chem-
`icals. This situation is further complicated by the fact that
`many xenoestrogens apparently bear little structural resem-
`blance to natural steroidal estrogens, although recent computer
`modeling studies demonstrate specific steric and electrostatic
`properties which may explain their ability to interact with the
`estrogen receptor (Waller et al., 1996). We therefore decided to
`test a group of alkyl hydroxy benzoate preservatives (parabens)
`for estrogenicity, due to their extensive use in a wide range of
`products to which humans are exposed, and also because of
`their structural similarity to alkylphenols (Fig. 1), which are
`known to be estrogenic (Soto et al., 1991; Routledge and
`Sumpter, 1997). We now report the findings from ensuing in
`vitro and in vivo studies in which a range of alkyl hydroxy
`benzoate preservatives (namely methyl-, ethyl-, propyl-, and
`butylparaben) were assessed for estrogenic activity.
`
`MATERIALS AND METHODS
`
`1 To whom correspondence should be addressed. Fax: 01895 274 348;
`E-mail: edwin.routledge@brunel.ac.uk
`
`Chemicals. For the in vitro studies, methylparaben (MP), ethylparaben
`(EP), propylparaben (PP), butylparaben (BP), para-hydroxy benzoic acid
`(pHBA), 17␤-estradiol, and 4-hydroxy tamoxifen were all purchased from
`
`0041-008X/98 $25.00
`Copyright © 1998 by Academic Press
`All rights of reproduction in any form reserved.
`
`12
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2006
`Page 1
`
`

`

`PARABENS ARE ESTROGENIC
`
`13
`
`this stage, control wells (blanks) appeared light orange in color, due to
`background expression of ␤-galactosidase, and turbid, due to growth of the
`yeast. Estrogenic chemicals were indicated by turbid wells in which a dose-
`dependent change in the color of the medium from light orange to red was
`detected.
`To demonstrate that the chemicals tested were acting through the estrogen
`receptor, the ability of 4-hydroxy tamoxifen, an estrogen antagonist known to
`act via the estrogen receptor, to inhibit the activity of PP and BP was
`investigated. A 2 mM stock solution of 4-hydroxy tamoxifen was serially
`diluted in ethanol, and 10-␮l aliquots of each concentration were transferred in
`duplicate to optically flat 96-well microtiter plates and allowed to evaporate to
`dryness (final concentration range 100 ␮M to 50 nM). Aliquots (200 ␮l) of the
`assay medium containing the recombinant yeast, the chromogenic substrate
`CPRG, and 17␤-estradiol (or the parabens) were then dispensed to the sample
`wells containing the 4-hydroxy tamoxifen. The concentration of 17␤-estradiol
`(or paraben) added to the medium was sufficient to produce an obvious, but not
`maximal, response. Controls were assay medium containing estradiol or para-
`bens (largest attainable response) and assay medium without the addition of
`17␤-estradiol or parabens (normal background expression). The plates were
`shaken, sealed, and incubated as before. The antiestrogenic activity of 4-hy-
`droxy tamoxifen was observed as a dose-dependent inhibition of the color
`change of the medium from yellow to red.
`
`Animals.
`Immature female Alpk:AP rats (21–22 days old; body weights
`38 –55 g) were obtained from the Zeneca breeding unit (Alderley Park).
`Ovariectomized (OVX) rats of the same strain were obtained from the same
`source. The ovariectomy was performed on 6 – 8-week-old animals. The latter
`were maintained for 2 weeks after the operation, during which time daily
`vaginal smears were taken, and only noncycling animals were used for the
`uterotrophic assays. Animal studies were performed in accordance with the UK
`“Animals (Scientific Procedures) Act.” Animal care and procedures were
`carried out according to in-house standards. All animals were housed in wire
`mesh cages; solid bottoms were provided for the immature animals. Temper-
`ature was controlled at 22° ⫾ 3, humidity was controlled at 30 –70%, and a 12
`h/12 h light dark cycle was maintained. Animals were weaned on R&M No. 3
`(Special Diet Services Ltd., Witham, Essex, UK) at postnatal Days 19 –21 and
`maintained on R&M No. 1 (Special Diet Services Ltd.) from postnatal Day 21
`onward. Diet and water were available ad libitum. Immature animals were
`acclimatized for 24 h before being dosed.
`
`Uterotrophic assays. The test protocol for both the immature rat and the
`OVX rat uterotrophic assays was as described previously (Odum et al., 1997;
`Ashby et al., 1997). Each experiment was accompanied by a vehicle control
`group (arachis oil) and a positive control group [17␤-estradiol (0.4 mg/kg/day
`body wt) by oral gavage or 17␤-estradiol (0.04 mg/kg/day body wt) by
`subcutaneous injection (sc) depending on the nature of the experiment (see
`Table 1]. Animals were dosed on three successive days with the appropriate
`chemical dissolved or suspended in arachis oil. A total dosing volume of 5
`ml/kg body wt was used initially (Table 1, Experiments 1 and 2), but this was
`increased to 10 ml/kg for the remaining experiments due to the low solubility
`of MP and BP. The maximum doses used were determined by the turbidity of
`the dosing solutions and restriction of dosing volume. In-house standard
`operating procedures allow a maximum sc dosing volume of 5 ml/kg, but
`special veterinary permission was obtained for the use of 10 ml/kg. The
`animals showed no signs of distress with these dosing volumes. Animals were
`killed by an overdose of halothane 24 h after the final dose. Vaginal opening
`(or otherwise) was recorded for immature animals at the time of death. Vaginal
`smears were also taken from OVX animals at the time of death. Uteri were
`excised, trimmed free of fat, pierced, and blotted to remove excess fluid. The
`body of the uterus was cut just above its junction with the cervix and at the
`junction of the uterine horns with the ovaries. The uterus was then weighed
`(wet weight). Uterine dry weight was also determined by drying the uteri at 70°
`for 24 h before reweighing. A quantitative estimation of vaginal cytology was
`performed by counting the number of fully cornified cells in at least 2 fields
`containing a total of at least 100 cells (⫻200 magnification).
`
`FIG. 1. Diagram illustrating the structural similarity of parabens and
`alkylphenolic chemicals. R ⫽ CH3 (A ⫽ methylparaben; B ⫽ methylphenol),
`R ⫽ C2H5 (A ⫽ ethylparaben; B ⫽ ethylphenol), R ⫽ C3H7 (A ⫽ propy-
`lparaben; B ⫽ propylphenol), and R ⫽ C4H9 (A ⫽ butylparaben; B ⫽
`butylphenol).
`
`Sigma (Poole, Dorset, UK). 4-Nonylphenol (95% pure) was supplied by
`Schenectady International, Inc. (Schenectady, NY). For the in vivo studies, MP
`was supplied by Aldrich Chemical Company (Gillingham, Dorset, UK), BP
`was supplied by Fluka Chemie AG (Switzerland), and arachis oil and 17␤-
`estradiol were purchased from Sigma. The purity of the all the chemicals used
`(unless stated otherwise) was greater than 99%.
`4-n-Dodecylparaben was prepared as described by Dymicky and Huhtanen
`(1979) and had mp 54 –56° (Dymicky and Huhtanen mp 37–37.5°). The carbon
`and hydrogen analysis was within 0.1% of theory. The required mass ion (306)
`was observed. The NMR spectrum (250 Mhz, CDCl3) was as predicted, and
`the IR spectrum showed a strong carbonyl absorption at 1678 cm⫺1.
`Competitive binding assay. The 105,000g supernatant, isolated as de-
`scribed by Shelby et al. (1996), from the uteri of immature rats was used as a
`source of cytosolic estrogen receptors in a competitive binding assay. The
`assay was conducted as previously described (Korach, 1979; Shelby et al.,
`1996). Aliquots of 100 ␮l cytosol, diluted to 500 ␮l, were incubated with
`[2,4,6,7-3H]-estradiol (5 nM) and different concentrations of unlabeled com-
`petitors (0.5 nM-500 ␮⌴) at 4°C for 18 h. The receptors were precipitated with
`hydroxyapatite in buffer (Shelby et al., 1996), washed with TEGM buffer,
`resuspended in Optiphase-MP scintillant (LKB Scintillation Products), and
`counted using a Packard 2000CA Tri-carb analyzer.
`The recombinant yeast estrogen screen. Details of the estrogen-inducible
`expression system in yeast (including assay validation and preparation of the
`medium components) have been described previously (Routledge and
`Sumpter, 1996). In brief, the DNA sequence of the human estrogen receptor
`was integrated into the yeast genome, which also contained expression plas-
`mids carrying estrogen-responsive sequences controlling the expression of the
`reporter gene Lac-Z (encoding the enzyme ␤-galactosidase). Thus, in the
`presence of estrogens, ␤-galactosidase is synthesized and secreted into the
`medium, where it breaks down the chromogenic substrate chlorophenol red-
`␤-D-galactopyranoside (CPRG), which is initially yellow, into a red product
`that can be measured spectroscopically.
`Yeast screen procedure. Stock solutions of MP, EP, PP, BP (10 mM), and
`17␤-estradiol (200 nM) were serially diluted in ethanol, and 10-␮l aliquots of
`each concentration were transferred, in duplicate, to optically flat 96-well
`microtiter plates and allowed to evaporate to dryness. Aliquots (200 ␮l) of the
`assay medium containing the recombinant yeast and the chromogenic substrate
`CPRG were then dispensed to each sample well. Thus, the parabens were
`diluted in twofold steps from 500 ␮M to 250 nM, and 17␤-estradiol was
`diluted from 10 nM to 5 pM in the medium, respectively. Appropriate solvent
`controls (blanks) were also included. The plates were sealed with autoclave
`tape and shaken vigorously for 2 min on a titer plate shaker before incubation
`at 32°C. After an 84-h incubation, the color of the medium was read at an
`absorbance of 540 nm (using a Titertek Multiskan MCC/340 plate reader). At
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2006
`Page 2
`
`

`

`14
`
`ROUTLEDGE ET AL.
`
`FIG. 2. Competitive binding to isolated estrogen receptors of diethylstilbestrol (DES), 4-nonylphenol, and butylparaben. Data are derived from duplicate
`samples taken from a representative experiment.
`
`Statistics. Variation within experiments for uterus wet and dry weights
`was determined by analysis of variance using the GLM procedure described in
`SAS (1989). Percentages of cornified cells were also considered by analysis of
`variance, following a double arcsine transformation (Freeman and Tukey,
`1950). Differences from control values were assessed statistically using a
`two-sided Student’s t test based on the error mean square from the analysis of
`variance.
`
`RESULTS
`
`Competitive Binding Assay
`
`Figure 2 shows that butylparaben was able to compete with
`3H-estradiol for binding to the rat estrogen receptor with an
`affinity approximately five orders of magnitude lower than that
`of diethylstilboestrol and between one and two orders of mag-
`nitude less than 4-nonylphenol. In this test system, receptor
`binding occured in the absence of metabolic conversion, indi-
`cating that the parent compound itself (as opposed to a metab-
`olite) may be the active form. The receptor-binding affinities of
`diethylstilboestrol and 4-nonylphenol are comparable with
`those reported by Shelby et al. (1996).
`
`Estrogenic Activity of Parabens in the Yeast Screen
`
`Figure 3 shows the response of the yeast estrogen screen to
`varying concentrations of p-hydroxybenzoic acid (pHBA), MP,
`EP, PP, and BP compared to 17␤-estradiol. All of the parabens
`tested were found to be estrogenic in vitro, whereas the parent
`compound (and a main metabolite) pHBA was inactive (Fig.
`3B). MP (R ⫽ CH3) produced the weakest response in the
`screen and was approximately 2,500,000-fold less potent than
`17␤-estradiol (Fig. 3A). The magnitude of the estrogenic re-
`
`sponse increased with alkyl group size, as shown by the fact
`that EP (R ⫽ C2H5), PP (R ⫽ C3H7), and BP (R ⫽ C4H9) were
`approximately 150,000-fold, 30,000-fold, and 10,000-fold less
`potent than 17␤-estradiol, respectively (Fig. 3A). Propylpara-
`ben was found to be equivalent in potency to 4-nonylphenol,
`whereas butylparaben was 3-fold more potent. 4-n-Dode-
`cylparaben (containing an unbranched linear alkyl group com-
`posed of 12 carbons) was also tested in order to investigate the
`effect of longer alkyl chain derivatives, which have greater
`antimicrobial activity (Dymicky and Huhtanen, 1979). 4-n-
`Dodecylparaben was inactive over the concentration range
`tested, and there was no evidence of toxic effects on the yeast
`even at a concentration of 500 ␮M (Fig. 3A).
`
`Inhibition of the Estrogenic Activity of Parabens by 4-
`Hydroxy Tamoxifen
`
`The ability of the antiestrogen 4-hydroxy tamoxifen to in-
`hibit the effect of PP and BP was used to demonstrate the
`requirement of these compounds to interact with the estrogen
`receptor in order to show estrogenic activity. Yeast cells were
`incubated in the presence and absence of either PP or BP at
`concentrations required to produce a distinct, but not maximal,
`response and concurrently with varying concentrations of 4-hy-
`droxy tamoxifen. Figure 4 illustrates that in the absence of PP
`or BP, the final background absorbance of the medium (A540)
`remained approximately 0.9. In the presence of PP or BP, the
`increased synthesis of ␤-galactosidase elevated the absorbance
`of the medium to approximately 2.6. Wells in which the
`chemicals were incubated concurrently with varying doses of
`4-hydroxy tamoxifen all showed a dose-dependent inhibition
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2006
`Page 3
`
`

`

`PARABENS ARE ESTROGENIC
`
`15
`
`FIG. 3. Response of the yeast estrogen screen to a range of parabens of varying alkyl group size, para-hydroxybenzoic acid (the parent compound and a main
`⬃ 2.7) at the highest concentrations are not shown. (A) The graph depicts
`metabolite), and 4-nonylphenol (an established xenoestrogen). Maximal responses (A540
`the log concentration of 17␤-estradiol (E2) serially diluted from 10 nM to 5 pM, and methylparaben, ethylparaben, propylparaben, butylparaben, 4-n-
`dodecylparaben diluted from 500 ␮M and 250 nM, plotted against the absorbance of the medium after a 3-day incubation. (B) Comparison of the estrogenic
`activity of butylparaben with para-hydroxybenzoic acid (diluted from 500 ␮M down to 250 nM) and 4-nonylphenol (diluted from 2 mM down to 1 ␮M). The
`blank depicts the basal level of expression in the assay.
`
`of ␤-galactosidase expression, indicating that all the com-
`pounds are acting via the estrogen receptor. The activity of
`17␤-estradiol was also inhibited by 4-hydroxy tamoxifen in a
`dose-dependent manner, as expected (results not shown).
`
`Uterotrophic Responses to MP and BP
`
`As the MP and BP used in the uterotrophic assays were
`obtained from different sources from those used in the in vitro
`tests, a sample of each chemical was also tested in the yeast
`screen, as described previously. Both the MP and BP used in
`the in vivo studies were found to be estrogenic in vitro, and
`their relative potencies were consistent with those illustrated in
`Fig. 3A (results not shown).
`MP and BP (which were, respectively, the least and most
`potent parabens in vitro) were tested in immature and OVX rat
`
`uterotrophic assays by oral and subcutaneous routes. The im-
`mature and OVX uterotrophic assays are generally considered
`to give comparable results (Zacharewski, 1998), but some
`weak estrogens may be more active in one or other of the
`assays (Ashby et al., 1997). The results of the immature rat
`uterotrophic assays on MP and BP are shown in Table 1. MP
`(administered orally or subcutaneously at dose levels of up to
`800 mg MP/kg/day) failed to increase uterus weights in im-
`mature rats (Table 1, Experiment 3). However, BP produced a
`small (but statistically insignificant) increase in uterus wet and
`dry weights in immature rats when given orally at dose levels
`from 800 mg BP/kg/day up to 1200 mg BP/kg/day (Table 1).
`Subcutaneous administration of BP significantly increased
`uterus wet weights in animals given doses between 400 and
`800 mg BP/kg/day, and a dose of 1200 mg BP/kg/day in-
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2006
`Page 4
`
`

`

`16
`
`ROUTLEDGE ET AL.
`
`tance, biodegradability, and cost (inexpensive), also because
`they are considered to be safe (Elder, 1984). Moreover, their
`excellent chemical stability in relation to pH (effective between
`pH 4.5–7.5) and temperature means that products can be safely
`autoclaved without significant loss of antimicrobial activity as
`a result of hydrolysis (Maddox, 1982).
`The antimicrobial activity of parabens increases concur-
`rently with alkyl chain length, but their water solubility de-
`creases and oil solubility increases. Since microbial replication
`generally occurs in the water phase of oil/water bases, the
`amount of paraben dissolved in the water phase generally
`determines the preservative efficacy (Dal Pozzo and Pastori,
`1996). For this reason, the lower alkyl chain members of the
`homologous series are often preferred in formulations. How-
`ever, it
`is common to find combinations of two or more
`paraben homologues (with different solubility characteristics)
`within a single product, in order to increase the ability of the
`system to withstand microbial contaminations (Dal Pozzo and
`Pastori, 1996).
`In recent years, the toxicology of parabens has been thoroughly
`reviewed in the EEC and US, which has led to p-hydroxybenzo-
`ates being more widely permitted in foods in the UK and to their
`GRAS (generally recognised as safe) status being reconfirmed by
`the US FDA (use limits of 0.1% w/w for MP and PP in food). The
`average daily intake of parabens in food is estimated to be 1–16
`mg/kg for infants and 4–6 mg/kg for persons aged 2 years or
`older (Elder, 1984). Maximum levels of parabens in pharmaceu-
`tical products seldom exceed 1% w/w, and the EEC directive
`76/768/EEC and the Danish cosmetic regulations permit the pres-
`ervation of cosmetic products with MP, EP, PP, and BP up to a
`maximum combined concentration of 0.8% w/w. Parabens pre-
`serve fats, proteins, oils, and gums in hair shampoos and condi-
`tioners, skin care products (cleansers and moisturisers), colognes,
`perfumes, toothpastes, soaps, and a range of make-up and sun-
`care products (Elder, 1984; Sherma and Zorn, 1982; Rastogi et al.,
`1995). In a recent survey of 215 cosmetic products, 77% of the
`rinse-off products and 99% of the leave-on products contained
`parabens (Rastogi et al., 1995). Therefore, products containing
`parabens may be applied to the skin, hair, scalp, lips, mucosae
`(oral, ocular, vaginal), and nails, either occasionally or on a daily
`basis, and their use may extend over a period of years. Thus, the
`frequency and duration of application (and therefore exposure of
`humans to parabens) both orally and topically may be continuous
`(Elder, 1984).
`The ability of parabens to pass through the stratum corneum
`(the uppermost barrier layer of the skin) increases with ester chain
`length (Twist and Zatz, 1986; Lee and Kim, 1994). Skin metab-
`olism (esterase activity) itself may also contribute to increased
`passive diffusion of parabens as a consequence of enhanced
`diffusion gradients, the capacity of which may be homologue
`specific (Bando et al., 1997). However, the potential for percuta-
`neous absorption of parabens is also influenced by their relative
`distribution in the oil/water phases of formulations, and these are
`often altered by the addition of surfactants (solubilizers) or by
`
`FIG. 4.
`Inhibition of the estrogenic activity of parabens by 4-hydroxy
`tamoxifen. Yeast cells were incubated in the absence or presence of parabens
`and concurrently with varying doses of 4-hydroxy tamoxifen. The yeast cells
`were incubated with butylparaben (BP; 1 x 10-5 M) and propylparaben (PP: 4 x
`10-5 M) either alone (PP or BP) or concurrently with 4-hydroxytamoxifen
`(PP⫹T or BP⫹T) at various concentrations. The blank depicts the normal
`background absorbance of the medium (without the addition of parabens).
`
`creased uterus wet weights to approximately 170% that of
`control values (Table 1, Fig. 5). The lowest dose of BP which
`induced a significant uterotrophic response was 200 mg BP/
`kg/day. The increase in dry weights eliminates the possibility
`that the gain in uterine weight following exposure to butylpara-
`ben was due solely to water retention. Neither paraben caused
`premature vaginal opening in any experiment, unlike estradiol,
`where premature vaginal opening was routinely observed
`(Odum et al., 1997).
`In the OVX rat uterotrophic assay, subcutaneous adminis-
`tration of 800 mg MP/kg/day failed to increase uterus weight
`and vaginal cornification (Fig. 6). In contrast, subcutaneous
`administration of BP significantly increased uterus wet and dry
`weight up to 150% of control values at a dose of 1200 mg
`BP/kg/day (Fig. 6). Subcutaneous exposure to BP (1000 mg
`BP/kg/day) significantly increased vaginal cornification (Fig.
`6), but an increase in vaginal cornification was also observed in
`rats exposed subcutaneously to 800 mg BP/kg/day, although
`this response was not statistically significant due to large
`interanimal variation.
`
`DISCUSSION
`
`Parabens were firstly employed as preservatives in pharma-
`ceutical products in the mid 1920s (Sabalitschka, 1930) and
`were extensively used in foodstuffs and cosmetics shortly
`afterward. In 1981, the Food and Drug Administration reported
`the use of MP, EP, PP, and BP in over 13,200 formulations
`(Elder, 1984). The popular use of paraben preservatives in
`cosmetics and toiletries arises from their low toxicity, inert-
`ness, broad spectrum of activity, worldwide legislative accep-
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2006
`Page 5
`
`

`

`PARABENS ARE ESTROGENIC
`
`17
`
`TABLE 1
`The Activity of Methyl and Butylparaben in the Immature Rat Uterotrophic Assay
`
`Uterus weight (mg)
`
`Expt.
`number
`
`Compound
`
`Dose (mg/kg)
`
`Route
`
`Group
`size
`
`1
`
`2
`
`3
`
`4
`
`5
`
`Arachis oil
`Butyl paraben
`
`E2
`Arachis oil
`Butyl paraben
`
`E2
`Arachis oil
`Methyl paraben
`
`Butyl paraben
`
`E2
`Arachis oil
`Butyl paraben
`
`E2
`Arachis oil
`Butyl paraben
`
`E2
`
`5 ml
`4 mg/kg/day
`40 mg/kg/day
`400 mg/kg/day
`0.4 mg/kg/day
`5 ml/kg/day
`800 mg/kg/day
`1200 mg/kg/day
`0.4 mg/kg/day
`10 ml/kg/day
`40 mg/kg/day
`400 mg/kg/day
`800 mg/kg/day
`40 mg/kg/day
`80 mg/kg/day
`40 mg/kg/day
`400 mg/kg/day
`0.4 mg/kg/day
`10 ml/kg/day
`40 mg/kg/day
`200 mg/kg/day
`400 mg/kg/day
`600 mg/kg/day
`800 mg/kg/day
`0.04 mg/kg/day
`10 ml/kg/day
`600 mg/kg/day
`800 mg/kg/day
`1000 mg/kg/day
`1200 mg/kg/day
`0.04 mg/kg/day
`
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`Oral
`sc
`sc
`sc
`sc
`Oral
`sc
`sc
`sc
`sc
`sc
`sc
`sc
`sc
`sc
`sc
`sc
`sc
`sc
`
`5
`5
`5
`5
`5
`7
`5
`5
`5
`5
`5
`5
`5
`5
`5
`5
`5
`5
`10
`5
`5
`5
`5
`5
`5
`10
`5
`5
`5
`5
`5
`
`Wet
`
`34.0 ⫾ 4.6
`30.7 ⫾ 2.7
`32.0 ⫾ 5.6
`30.1 ⫾ 6.2
`115.1 ⫾ 12.5**
`28.2 ⫾ 5.2
`35.5 ⫾ 9.3
`33.2 ⫾ 6.4
`117.5 ⫾ 8.9**
`33.8 ⫾ 5.9
`34.5 ⫾ 3.2
`33.3 ⫾ 5.3
`31.0 ⫾ 4.3
`33.5 ⫾ 6.6
`32.8 ⫾ 1.5
`38.0 ⫾ 6.8
`41.6 ⫾ 6.5**
`122.7 ⫾ 9.4**
`30.3 ⫾ 3.8
`33.6 ⫾ 4.7
`32.6 ⫾ 2.8
`33.7 ⫾ 3.7
`42.3 ⫾ 3.5**
`47.6 ⫾ 4.9**
`116.2 ⫾ 10.2**
`27.1 ⫾ 4.7
`32.0 ⫾ 5.4
`38.8 ⫾ 7.6**
`42.6 ⫾ 3.9**
`47.1 ⫾ 2.6**
`108.7 ⫾ 16.4**
`
`Dry
`
`6.5 ⫾ 1.0
`6.2 ⫾ 0.2
`6.5 ⫾ 1.4
`6.5 ⫾ 1.4
`20.5 ⫾ 1.5**
`6.3 ⫾ 1.5
`7.5 ⫾ 1.8
`7.8 ⫾ 2.9
`22.5 ⫾ 1.8**
`6.4 ⫾ 0.9
`6.8 ⫾ 0.7
`6.3 ⫾ 0.9
`6.2 ⫾ 0.7
`6.8 ⫾ 1.1
`6.6 ⫾ 0.3
`7.6 ⫾ 1.3
`8.6 ⫾ 1.0*
`21.7 ⫾ 1.9**
`5.6 ⫾ 0.5
`6.1 ⫾ 0.8
`6.4 ⫾ 0.4*
`6.4 ⫾ 0.4*
`7.9 ⫾ 0.7**
`9.0 ⫾ 0.9**
`18.3 ⫾ 3.9**
`5.6 ⫾ 1.1
`6.6 ⫾ 1.1**
`7.6 ⫾ 1.5**
`8.5 ⫾ 0.7**
`9.1 ⫾ 0.7**
`18.9 ⫾ 2.8**
`
`Note. Estradiol (E2) was used as a positive control. Data are mean ⫾ SD. *denotes P ⬍ 0.05, **P ⬍ 0.01.
`
`their entrapment in micelles and liposomes (Komatsu, 1984;
`Komatsu et al., 1986), which adds another level of complexity to
`their behavior in topically applied products. Certain solubilizers
`may reduce absorption of parabens through the skin (Komatsu
`and Suzuki, 1979). However, even low penetration rates could
`result in considerable amounts of these chemicals entering the
`body from commercial preparations, which are often applied to
`large surface areas of skin (Dal Pozzo and Pastori, 1996). It is
`apparent that topical exposure of humans to parabens will depend
`largely on the product formulation within which they are con-
`tained and the pattern of product use.
`Early animal studies generally suggest that parabens are rapidly
`metabolized and excreted. In these studies, the majority of the
`dose (up to 800 mg/kg) was recovered in the urine and feaces as
`p-hydroxybenzoic acid (pHBA), p-hydroxyhippuric acid (M1),
`p-hydroxybenzoyl glucuronide (M3), and p-carboxyphenyl sul-
`fate (M4) up to 72 h after exposure (Elder, 1984). However, as
`metabolism and elimination rates of parabens are believed to be
`dose-, route- (Kiwada et al., 1979), and probably species-depen-
`
`dent, and because the doses employed in the majority of animal
`studies (including this study) were short term and greatly ex-
`ceeded realistic doses in humans (Kiwada et al., 1979), the appli-
`cation of these findings in terms of an assessment of a possible
`estrogenic hazard are equivocal. However, when rats were ex-
`posed to 2 mg radiolabeled EP/kg (the presumed dose of practical
`usage in humans) by intravenous injection, approximately 170 ␮g
`was recovered as pHBA and 860, 600, and 140 ␮g were recovered
`as M1, M3, and M4, respectively. Excretion had almost ceased by
`5 h, after which time 180 ␮g of the total dose had appeared as
`unidentified metabolites, 50 ␮g was unrecovered, and about 600
`ng occurred in the urine as unmetabolized EP (Kiwada et al.,
`1979). Hydrolysis occured rapidly in the blood, with only 5%
`(100 ␮g) of the initial dose remaining as the parent EP 3 min after
`the injection, with levels decreasing rapidly (Kiwada et al., 1980)
`due to carboxyesterase activity in the blood, liver, and other
`tissues. This rapid metabolism of parabens may explain, in part,
`why butylparaben was approximately 2 to 3 orders of magnitude
`more potent in vitro (Fig. 3A) compared to subcutaneous admin-
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2006
`Page 6
`
`

`

`18
`
`ROUTLEDGE ET AL.
`
`FIG. 5. The effects of subcutaneous administration of butylparaben on uterus wet weights of immature rats. Estradiol (E2) (0.04 mg/kg, sc) was used a
`positive control. Controls received arachis oil only. The three studies (Experiments 3–5) shown in Table 1 were combined and statistically analyzed, as described
`in Materials and Methods. Data represent the adjusted means ⫾ SD of the consolidated data. The number above the bar represents the consolidated group size.
`Statistical significance is taken from Table 1. *P ⬍ 0.05, **P ⬍ 0.01. No open vaginas were observed in control and butylparaben-treated animals, whereas 5/10
`animals in the E2-treated group had open vaginas.
`
`istration in vivo (Figs. 5 and 6). Thus, it appears that parabens are
`rapidly absorbed, metabolized, and excreted, but the presence of
`unmetabolized paraben in the urine, together with incomplete
`recoveries, suggests that low levels of unmetabolized paraben
`(presumably the only estrogenic form) may remain in circulation
`in the body for extended periods of time.
`The results presented in Fig. 3A indicate that parabens
`possess intrinsic estrogenic activity with a potency between
`104- and 107-fold lower than that of the main natural estrogen
`
`FIG. 6. The effects of subcutaneous administration of methylparaben and
`butylparaben on uterus wet and dry weights (A and B) of OVX rats. Vaginal
`cornification (mean percentage of cornified cells) is indicated in the white
`boxes in A. Estradiol (E2) (0.04 mg/kg, sc) was used a positive control.
`Controls received arachis oil only. Data represent group means ⫾ SD. The
`number above the bar represents the group size, which was the same for both
`determinations. **P ⬍ 0.01, *P ⬍ 0.05.
`
`17␤-estradiol, and their ability to transactivate the estrogen
`receptor in vitro increases concurrently with alkyl group size.
`Similar findings have been reported with alkylphenolic chem-
`icals, in which estrogenic activity was dependent on alkyl
`group size and branching (Routledge and Sumpter, 1997). In
`view of this, parabens with longer and/or branched alkyl
`groups (e.g., heptylparaben,
`isopropylparaben, and isobu-
`tylparaben) and their metabolites (when known and available)
`should also be assessed for estrogenicity activity. Thus, the
`present data should stimulate further focused studies.
`Evaluation of the activity of MP and BP in rodent utero-
`trophic assays was complicated by the relative insolubility
`of the methyl compound, but
`the butyl compound was
`shown to be weakly estrogenic when administered via sub-
`cutaneous injection. Both compounds were inactive when
`administered to rats orally at the maximum practical dose
`level. The physicochemical properties of the parabens indi-
`cate that they can be absorbed across skin, and that suggests
`a possible estrogenic hazard to humans exposed topically to
`such chemicals. However, a more meaningful assessment of
`the effects of exposure to environmentally relevant concen-
`trations of estrogenic chemicals will requi