- Email: [email protected]
Association between paraben exposure and menstrual cycle in female university students in Japan
Accepted Manuscript Title: Association between paraben exposure and menstrual cycle in female university students in Japan Author: Yukiko Nishihama Jun Yoshinaga Ayaka Iida Shoko Konishi Hideki Imai Miyuki Yoneyama Daisuke Nakajima Hiroaki Shiraishi PII: DOI: Reference:
S0890-6238(16)30116-2 http://dx.doi.org/doi:10.1016/j.reprotox.2016.05.010 RTX 7301
To appear in:
Reproductive Toxicology
Received date: Revised date: Accepted date:
19-10-2015 2-5-2016 13-5-2016
Please cite this article as: Nishihama Yukiko, Yoshinaga Jun, Iida Ayaka, Konishi Shoko, Imai Hideki, Yoneyama Miyuki, Nakajima Daisuke, Shiraishi Hiroaki.Association between paraben exposure and menstrual cycle in female university students in Japan.Reproductive Toxicology http://dx.doi.org/10.1016/j.reprotox.2016.05.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Title Association between Paraben Exposure and Menstrual Cycle in Female University Students in Japan
Author names and affiliations Yukiko Nishihamaa, Jun Yoshinagab, Ayaka Iidac, Shoko Konishid,e, Hideki Imaif, Miyuki Yoneyamag, Daisuke Nakajimag, and Hiroaki Shiraishig
a
Department of Environmental Studies, University of Tokyo, Kashiwanoha 5-1-5, Kashiwa,
Chiba 277-8563, Japan. b Faculty of Life Sciences, Toyo University, 1-1-1 Izumino, Itakura, Ora, Gunma 374-0193, Japan.
c
Department of Environmental, Occupational Health and
Toxicology, School of Allied Health Sciences, Kitasato University, Kitasato 1-15-1, Minami, Sagamihara, Kanagawa 252-0373, Japan. d Department of Human Ecology, Graduate School of Medicine, University of Tokyo, Hongou 7-3-1, Bunkyou, Tokyo 113-0033, Japan.
e
Department of Anthropology, University of Washington, Seattle WA 98195, USA. f Department of Nursing, Tokyo Healthcare University, Higashigaoka 2-5-1, Meguro, Tokyo 152-8558, Japan. g National Institute for Environmental Studies, Onogawa 16-2, Tsukuba, Ibaraki 3058506, Japan.
Corresponding author: Jun Yoshinaga, PhD Faculty of Life Sciences, Toyo University, 1-1-1 Izumino, Itakura, Ora, Gunma 374-0193, Japan. Tel/Fax: +81-276-82-9031. E-mail address: [email protected]
1
Highlights
Possible effect of paraben exposure on menstrual cycle was explored.
Urinary concentrations of parabens were used as biomarker of exposure.
Shorter menstrual cycle was associated with higher urinary paraben concentrations.
Parabens exposure may be one of environmental determinants of female fecundity.
Abstract
Parabens have been known to have estrogenic activity in many in vivo and in vitro studies and biomonitoring data indicated ubiquitous exposure in general populations. However, there are few human studies on reproductive effects of parabens. In this study, menstrual cycle length and its intra-individual variation were investigated by bleeding record over the period of 5 months and urinary concentrations of parabens were measured for 128 female Japanese university students. We found significantly negative relationships between menstrual cycle length and urinary estrogen-equivalent total paraben (odds=0.73, 95% CI 0.56-0.96) and butyl paraben concentrations (odds=0.83, 0.70-0.99), which indicated shortened cycle length in women with high urinary paraben concentrations. This study indicated that paraben exposure at not excessive levels is associated with menstrual cycle length or its variability among general female subjects. This results suggest a possibility that parabens exposure is one of the environmental causes of human reproductive problem.
Keywords paraben; urine; menstrual cycle; female fecundity.
2
1. Introduction
Alkyl esters of 4-hydroxy benzoic acid (parabens), used as a preservative in personal care products (PCPs) such as cosmetics and body creams, in medications, and in foods, are suspected endocrine disrupting chemicals (EDCs) [1, 2]. In Japan, the maximum permissible level of parabens in cosmetics is regulated by the Ministry of Health, Labor and Welfare (MHLW) [3] to be less than 1.0 g /100 g as a mixture of esters and/or sodium salt. Commonlyused parabens including methyl parabens (MP), ethyl paraben (EP), propyl paraben (PP) and butyl paraben (BP) are considered relatively safe because acute toxicities such as dermal and eye irritancy, and chronic toxicities are not seemed to be observed for these parabens [1]. However, many in vitro studies have reported estrogen receptor binding capacity of parabens [4-6]. Additionally, in vivo studies have demonstrated increased uterus weight in female rats administrated with parabens [7]. Meanwhile, a limited number of studies explored reproductive effects of exposure to parabens in humans. Buttke et al. [8] investigated the association of urinary paraben concentrations with the age of menarche for 12-16 years old female participants of the NHANES study (n=287) to find non-significant association. Smith et al. [9] examined the relationship between urinary paraben concentrations and marker of ovarian reserve (day-3 follicle-stimulating hormones (FSH), antral follicle count (AFC) and ovarian volume) in a prospective cohort study of 21.0-46.7 year-old women seeking fertility treatment in Boston, Massachusetts (n=192). They indicated possible reduction of ovarian reserve by paraben exposure. Apparently more epidemiologic studies are needed on a possible reproductive effects of parabens when we take into consideration their estrogenic activities observed in in vitro and in vivo studies. Ubiquitous exposure to this group of chemicals has been revealed by frequent detections in urine samples in a number of biomonitoring studies on general population [10-
3
12]. The purpose of this study is to investigate the relationship between paraben exposure and menstrual cycle characteristics of young Japanese female subjects. Menstrual cycle is considered to be regulated by endocrine actions and thus it could be affected by the exposure to EDCs such as parabens.
2. Material and methods
2.1. Subject
The subjects of this cross-sectional study were female university students. They were recruited at Tokyo Healthcare University located in Tokyo during 2012-2013 at the orientation of class for clinical laboratory practice (n=178). The purpose and procedure of the study was explained to students by survey staffs after distributing a brochure describing the same content. We approached 178 students who attended orientation for class of clinical laboratory practice, and obtained written consent from all of the students. The ethics committees of the Tokyo Healthcare University, the University of Tokyo and National Institute for Environmental Studies approved this study.
2.2. Sampling
A single spot urine sample was collected from subjects in the clinical laboratory practice typically 1-2 months after the orientation. Urine sample was taken by the subjects herself in a paper cup during the class time and was brought to class immediately. An aliquot was dispensed into polypropylene (PP) tubes by survey staffs. Travel blanks for urine samples were prepared.
4
All of the PP tubes for urine sampling were rinsed with ultrapure water and methanol (HPLC Grade, Kanto Chemical Co. Ltd., Japan) prior to use. The samples were stored at – 20°C until analysis.
2.3. Urinary parabens analysis
Analysis of MP, EP, PP, and BP in urine samples were carried out by liquid chromatography-tandem mass spectrometry (LC-MS/MS) after enzymatic deconjugation and solid phase extraction [13]. In brief, 1 mL of urine sample was added with d4-labelled internal standards, enzymatically deconjugated, and was subject to solid phase extraction (Sep-Pak ® Vac 6 cc (500 mg) tC18 Cartridges, Waters, Massachusetts, USA), which was followed by LC-MS/MS analysis (LC: Agilent 1200 series, MS/MS: Agilent 6460 Triple Quad LC/MS, Agilent Technologies, CA, USA). No parabens were detected in travel blanks. Recoveries of the internal standards were monitored for all of the samples and the result was 50±17% (MP), 44±15% (EP), 52±17% (PP) and 47±14% (BP) (n=128). The instrumental detection limit for the parabens was defined as the concentration corresponded to S/N=3 calculated from 7 repeated measurements of 2.0 ng/mL (MP) or 0.5 ng/mL (EP, PP, and BP) standard solutions. The recovery-corrected detection limits were 0.54, 0.25, 0.15 and 0.064 ng/mL in urine for MP, EP, PP, and BP, respectively. An in-house quality control urine sample was included in every batch (typically 20 samples/batch) of sample preparation and measurement. The repeatability and reproducibility of analysis were 1.0-2.4% and 1.4-3.3%, respectively, for the 4 parabens. The urinary concentrations of parabens (ng/mL) were adjusted with specific gravity (SG) of 1.020.
5
2.4. Menstrual cycle characteristics
Questionnaire survey on menstrual cycle in this study has been described in detail in our previous study [14]. The subjects were asked to keep a diary for recording their menstrual bleeding for 5 months to a calendar we prepared for this purpose from the day of their agreement of study participation (generally 1-2 months before urine sampling). Menstruation was defined as ≥ 3 consecutive days of bleeding by following the definition of the Japan Society of Obstetrics and Gynecology (JSOG) guideline [15]. Menstrual cycle length for each subject was defined as the mean of the intervals from the first day of an index menstruation to the day before the next menstruation during 5 months survey period. If the truncated days (days from the first day of survey to the first day of the first menstruation, or from the last day of an index menstruation to the last day of record) was longer than the mean menstrual cycle of the subject during 5 months survey period, we regarded the truncated days as one menstrual cycle and recalculated the mean menstrual cycles by adding this cycle. Intra-individual variability of menstrual cycle was defined as the standard deviation (SD) of menstrual cycle lengths recorded for each individual. The subjects were also asked to answer a brief questionnaire which asked about age at menarche, menstrual pain, oral contraceptive (OC) use in the past 6 months, and experience of pregnancy. The menstrual pain was asked by the following 4 categories: “almost none”; “feel pain, but no problem in daily activity”; “can go out but exercise is difficult”, and “too painful to get out of bed” [16].
6
2.5. Biological attributes and life style
The subjects were also asked to answer another questionnaire on biological attributes (age, height, and weight), lifestyle (active/passive smoking, alcohol and caffeine intake, physical activity), and food habit. Food habit of the subjects was asked by frequency of consumption of 6 foods, which were considered major sources of exogenous estrogen or estrogenic chemicals [17, 18]. For the consumption of meat, milk, milk products and egg, the frequency was asked by the following 7 scales: “every meal”, “daily”, “four and five times a week”, “two and three times a week”, “less than once a week”, “one and two times a month”, or “none”. For fish and soy products, the frequency was asked by 6 scales: “every meal”, “daily”, “four and five times a week”, “two and three times a week”, “less than once a week”, or “none”.
2.6. Statistical analyses
SG-adjusted urinary parabens concentrations were used for the following statistical analyses. The distribution of these concentrations skewed towards higher value; therefore, the concentrations were transformed to natural logarithm. When a paraben concentration was below the detection limit, 1/2 of detection limit value was substituted in statistical analysis. In addition to individual paraben concentrations, we used estrogen-equivalent total paraben (ETP) in our statistical analysis by summing the individual concentrations of the four parabens weighted by their relative estrogenic activity (eq. 1) [13]. The weighing factor was derived from in vitro yeast estrogen screen assays [6]. [ETP] = [1*MP + 16.7*EP + 83.3*PP + 250*BP] (μM) …… (eq. 1) The subjects were categorized into 3 groups by their menstrual cycle length in the quartiles: subjects whose mean menstrual cycle length was < 25 percentile, 25-75 percentile,
7
and > 75 percentile were categorized into “shorter cycle”, “moderate cycle” and “longer cycle”, respectively. With regard to intra-individual variability of menstrual cycles, the subjects were categorized into “small” and “large” by the median of SD of individuals’ menstrual cycle length. The menstrual pain was divided into two categories “no” (“almost none”) and “yes” (the rest) groups. With regard to frequencies of food consumption, subjects were re-categorized into 2 categories by one of original 6 or 7 scales which was mentioned in the 2.5. Biological attributes and life style so that number of subjects in the two was roughly equal. For the analyses of the relationship between the 3 categories of menstrual cycle length or 2 categories of intra-individual variability and their biological attributes and lifestyles, chisquared test, Kruskal-Wallis test, or Mann-Whitey’s U test were used. An ordinal logistic regression analysis was used to explore whether urinary paraben levels are associated with the 3 groups of menstrual cycle length (“shorter”, “moderate”, and “longer” cycle). Either one of the 4 urinary parabens concentration and ETP was used as an independent variable. A (binomial) logistic regression analysis was conducted to investigate if urinary parabens levels are associated with their intra-individual variability with 2 groups: classified intra-individual variability (“small” and “large”) as a dependent variable and each 4 urinary parabens and ETP concentration as an independent variable. In addition to biological attributes of the subjects (age, BMI, and age at menarche), other variables were included in the logistic regression analyses as independent variable when the variable was found to be statistically significantly associated with dependent variables in the preliminary bivariate analysis. We used forced entry method for logistic regression analyses. Statistical analyses were performed with SPSS for Windows (version 12.0; SPSS Japan Inc., Tokyo, Japan). Statistical significance was defined as probability of < 0.05.
8
3. Results
Of 178 subjects agreed to participate in this study, 145 (81%) completed bleeding diary and questionnaires. Of the 145, 15 OC users in the past 6 months and 2 subjects whose age was regarded as outlier by Smirnov-Grubbs test (25 and 36 years) were excluded, thus leaving 128 subjects for further analyses. No subject reported experience of pregnancy. Table 1 shows the characteristics of the 128 subjects. Mean ± SD (min-max) of age and BMI was 20.2±0.9 (19-22) years and 20.8±2.0 (16.6-27.2) kg/m2, respectively. Mean ± SD of menstrual cycle length was 35±12 days with the median (min-max) being 32 (20.9-98.0) days. The mean of intra-individual variability (SD of cycle lengths within individuals) was 10±13 days and the median (min-max) was 6.1 (0.66-70) days. Mean ± SD (min-max) of menstrual cycles per women were 4±1 (1-7) cycles. Preliminary bivariate analyses between length and intra-individual variability of menstrual cycles and biological, lifestyle factors and dietary habit found a suggestive association between menstrual cycle length and meat consumption (chi-squared test, p=0.090), and between intraindividual variability and menstrual pain (chi-squared test, p=0.060). These variables were used as covariates in the following logistic regression analysis. Table 2 shows unadjusted and SG- adjusted mean, geometric mean and median urinary concentrations of MP, EP, PP, and BP of the present subjects. MP, EP, PP and BP were detected in 100, 97, 98 and 70% of the subjects, respectively. One half of detection limit value, 0.13, 0.075, and 0.032 ng/mL for EP, PP, and BP, respectively, was substituted for undetectable subjects for the calculation of mean value (Table 2) and in the following statistical analyses. SG-adjusted mean ± SD concentrations were 340±366, 18.2±36.4, 29.7±45.8 and 4.58±10.6 ng/mL for MP, EP, PP and BP, respectively and 23.7±30.1 μM for ETP. The relative contribution of PP, MP, BP and EP to ETP was 46.1%, 21.7%, 19.9% and 12.3%, respectively.
9
Table 3 shows Spearman correlation coefficients between the SG-adjusted concentrations of 4 parabens in the urine samples. We found that all of the paraben pairs had a significantly positive correlation (r=0.26 to 0.48, p< 0.01). The results of logistic regression analyses using menstrual cycle length and intraindividual variability as dependent variable were shown in Tables 4 and 5. Only the results of models with a significant independent variable were shown in these tables. We found a significant relationship between menstrual cycle length and urinary ETP concentrations (adjusted odds= 0.73 (95% CI: 0.56, 0.96), p=0.027, Table 4) and BP concentrations (adjusted odds=0.83 (0.70, 0.99), p=0.037, Table 4). The urinary concentrations of ETP and BP were higher in subjects with shorter cycle lengths than those with moderate and longer cycle lengths (Figure 1 and Figure 2). Urinary MP, EP and PP concentration was not significantly associated with menstrual cycle length of the subjects. Variables used as covariates, i.e., age, BMI, age at menarche and meat consumption were not significant in all of the models. Urinary PP (adjusted odds=0.83 (0.67, 1.0), p=0.087, Table 5) and ETP concentrations (adjusted odds 0.77 (0.57, 1.0), p=0.093, Table 5) were selected as independent variables with suggestive association for intra-individual variability in the binary logistic regression analysis. Urinary concentrations of PP and ETP were higher in subjects with smaller intra-individual variability than in those with larger variability (Figure 3 and Figure 4).
4. Discussion
4.1. Menstrual cycle characteristics
The median cycle length of the present subjects (32 days, Table 1) was consistent with,
10
but intra-individual variability (median 6.1 days, Table 1) was larger than, that of the previous study on Japanese women of similar age (31 days and 3 days) [19]. However, the median cycle length of the present subjects seemed greater than those reported for women in US and Europe [20, 21] without known reasons.
4.2. Urinary parabens level
Median urinary concentrations of parabens of the present subjects (Table 2) were similar to those obtained for pregnant women (32.6 years, 108, 7.26 and 0.80 ng/mL for MP, EP and BP) in a previous study in Japan [13] except for PP for which higher median concentration than ours was reported (33.3 ng/mL). Median urinary concentrations were also similar to those for the U.S. female (> 6 years old), i.e., 137, 1.3, 29.1 and 0.50 ng/mL (unadjusted) for MP, EP, PP and BP, respectively [11]. Thus exposure levels of parabens of the present subjects were considered not apart from levels of general populations in Japan and the US. We found significant correlations between the concentrations of parabens (Spearman’s rank correlation coefficient: 0.26-0.48) (p< 0.01, Table 3). This result was probably due to the fact that more than one paraben are often added to a PCP in many cases to enhance preservative effect [1]. This means that co-exposure to multiple parabens takes place by the use of PCPs. Since the 4 parabens measured in this study are known to have estrogenic activity in common, but with varying relative potencies [6], additive effect is expected by the co-exposure. This was the reason of our usage of ETP in this study. The contribution of each of the 4 parabens to the median ETP concentration was 21.7%, 12.3%, 46.1% and 19.9% for MP, EP, PP and BP respectively. The greatest contribution of PP to ETP in the present subjects suggests that PP has the greatest risk potential for the present subjects among the 4 parabens. BP is almost same contribution of MP though the average concentration was the lowest (median: 0.690 ng/mL,
11
Table 2).
4.3. The relationship between menstrual cycle and urinary parabens concentration
As far as we know, this is the first epidemiologic study that explored a possible relationship between menstrual cycle and parabens exposure. In this study, significant relationship between menstrual cycle lengths and urinary ETP and BP concentrations were found (Table 4), suggesting that exposure to parabens shortens menstrual cycle length. A mechanism for the observed relationship between menstrual cycle length and exposure to parabens is not clear. When estrogen level rises, a negative-feedback system works to decrease FSH secretion leading to LH surge and subsequent ovulation [22]. It is well known that shorter menstrual cycle length is accompanied with shorter follicular phase length [23]. Higher estrogen-related metabolite excretion is reported for women with shorter menstrual cycle length [24]. Thus, elevated level of circulating estrogen results in earlier transition from follicular phase to luteal phase, and exogenous estrogen-like substances, such as parabens, may similarly works. Suggestive relationship between intra-individual variability and urinary PP and ETP concentrations was found in the present study (Table 5). This result indicated that the subjects with smaller intra-individual variability were exposed to higher levels of parabens. The underlying mechanism of the relationship is not clear either but we noticed that subjects with longer cycle length tended to have greater intra-individual variability (Spearman coefficient correlation: r=0.54, p< 0.001, Figure 5). Taking this positive relationship between menstrual cycle length and intra-individual variability into consideration, if parabens have an effect to shorten menstrual cycle length, then reduced intra-individual variability among higher urinary paraben levels, as observed in this study, would be apparently expected. Another explanation
12
is that parabens exposure reduced intra-individual variability, which resulted in apparently higher exposure levels among the subjects with shorter menstrual cycle length. The former explanation seems likely because the former relationship is significant but the later was not significant. However, it is not possible to definitely distinguish the two explanations from this study result alone. Epidemiologic studies reported that women who had shorter menstrual cycle [25] were prone to have prolonged time-to-pregnancy. This indicated reduced fecundity among women with shorter menstrual cycle length. If exposure to parabens has an effect to shorten human menstrual cycle length as indicated by the present study, then reduced fecundity would be expected for the exposed women. This may pose a serious concern as one of potential environmental causes of subfertility/infertility problem prevalent in many countries. Therefore, further studies are warranted to confirm the present result. Reproductive effect of paraben exposure in humans has been explored in two previous studies as far as we are aware; Buttke et al. [8] examined if there is a relationship between age of menarche and exposure to phenols including parabens among 12-16 year-old girls. In this study, the sum of urinary concentration of MP and PP was used as a biomarker of exposure to parabens. They found no significant association between the two. Smith et al. [9] explored the association between measures of ovarian reserve (day-3 follicle-stimulating hormone, antral follicle count (AFC), and ovarian volume) and urinary MP, PP and BP concentrations for 21.046.7 year-old women who underwent fertility treatment (n=192). They found a suggestive, but not a significant, negative trend between AFC and urinary PP levels (p=0.07). Thus studies to date did not find significant reproductive effects of paraben exposure in humans. The strength of our study included 1) menstrual cycles of subjects were assessed by a recording method, not by a recall method, for 5 months. 2) paraben exposure was assessed by biomarker of exposure. Urinary paraben levels in spot urine samples were reported to represent
13
exposure level of subject over longer period of time [12]. Moreover, 3) the fact that we limited our subjects to university student resulted in small inter-individual variabilities in age, one of the most influential factors of menstrual cycle length, and other attributes [14]; this reduced the possibility of potential confounding, though we included potential covariates of menstrual cycle length in the regression models. However, it is still possible that we failed to include some other potential covariates, e.g., psychological stress [26], and this might be a limitation of this study, in addition to a moderate sample size.
5. Conclusions
We found significant association between menstrual cycle length and urinary ETP and BP concentrations (p< 0.05) among female Japanese university students indicating that exposure to parabens shortens menstrual cycle length. This study result suggests that paraben exposure plays a role as environmental causes of infertility/subfertility, which is one of a public health concern across the world. Further epidemiologic studies using a larger sample size are needed.
Conflict of interest None declared.
Acknowledgement
The authors sincerely appreciate the cooperation of the subjects. They also appreciate Dr. Kouda, Department of Nursing, Tokyo Healthcare University, for her help in the sampling.
14
References
[1] Soni MG, Carabin IG, Burdock GA. Safety assessment of esters of p-hydroxybenzoic acid (parabens). Food and Chemical Toxicology 2005; 43: 985-1015. [2] Błędzka D, Gromadzińska J, Wąsowicz W. 2014. Parabens. From environmental studies to human health. Environment International 2014; 67: 27-42. [3] MHLW (Ministry of Health, Labor and Welfare). Standards for Cosmetics (Ministry of Health and Welfare Notification No.331 of 2000). Tokyo, Japan: MHLW 2000. [4] Nishihara T, Nishikawa J, Kanayama T, Dakeyama F, Saito K, Imagawa M, et al. Estrogenic Activities of 517 Chemicals by Yeast Two-Hybrid Assay. Journal of Health Science 2000; 46: 282-98. [5] Golden R, Gandy J, Vollmer G. A review of the endocrine activity of parabens and implications for potential risks to human health. Critical Reviews in Toxicology 2005; 35: 435-58. [6] Routledge EJ, Parker J, Odum J, Ashby J, Sumpter JP. Some alkyl hydroxy benzoate preservatives (parabens) are estrogenic. Toxicology and Applied Pharmacology 1998; 153: 12-9. [7] Lemini C, Jaimez R, Avila ME, Franco Y, Larrea F, Lemus AE. In vivo and in vitro estrogen bioactivities of alkyl parabens. Toxicol. Ind. Health 2003; 19: 69-79. [8] Buttke DE, Sircar K, Martin C. Exposures to endocrine-disrupting chemicals and age of menarche in adolescent girls in NHANES (2003-2008). Environmental Health Perspectives 2012; 120: 1613-18. [9] Smith KW, Souter I, Dimitriadis I, Ehrlich S, Williams PL, Calafat AM, et al. Urinary paraben concentrations and ovarian aging among women from a fertility center. Environmental Health Perspectives 2013; 121: 1299-305.
15
[10] Ye X, Bishop AM, Reidy JA, Needham LL, Calafat AM. Parabens as urinary biomarkers of exposure in humans. Environmental Health Perspectives 2006; 114: 1843-46. [11] Calafat AM, Ye X, Wong LY, Bishop AM, Needham LL. Urinary concentrations of four parabens in the U.S. population: NHANES 2005-2006. Environmental Health Perspectives 2010; 118: 679-85. [12] Smith KW, Braun JM, Williams PL, Ehrlich S, Correia KF, Calafat AM, et al. Predictors and variability of urinary paraben concentrations in men and women, including before and during pregnancy. Environmental Health Perspectives 2012; 120: 1538-43. [13] Shirai S, Suzuki Y, Yoshinaga J, Shiraishi H, Mizumoto Y. Urinary excretion of parabens in pregnant Japanese women. Reproductive Toxicology 2013; 35: 96-101. [14] Nishihama Y, Yoshinaga J, Iida A, Konishi S, Imai H. Menstrual cycle length and source of its variation in female university students majoring in nursing sciences. Nihon Eiseigaku Zasshi 2015; 70: 139-48 (in Japanese). [15] Mikuni M, Fujimoto S, Toda T, Kohno Y, Terakawa N, Ishikawa M, et al. Diagnosis, Treatment and Management of Gynecologic Diseases, Endocrine Diseases. Acta obstetrica et gynaecologica Japonica 2002; 54 (12): N-557 - 8. [16] Konishi S, Nishihama Y, Iida A, Yoshinaga J, Imai H. Association of antimüllerian hormone levels with menstrual-cycle type and dysmenorrhea in young asymptomatic women. Fertility and Sterility 2014; 102: 1439-43. [17] Mendola P, Buck GM, Sever LE, Zielezny M, Vena JE. 1997. Consumption of PCBcontaminated freshwater fish and shortened menstrual cycle length. American Journal of Epidemiology 1997; 146: 955-60. [18] Hartmann S, Lacorn M, Steinhart H. Natural occurrence of steroid hormones in food. Food Chemistry 1998; 62: 7-20. [19] Yano C, Tomiyasu T, Anai T. A general survey on what consitutes a "normal" menstrual
16
cycle. Bosei Eisei 2005; 45: 496-502 (in Japanese). [20] Gaskins AJ, Mumford SL, Zhang C, Wactawski-Wende J, Hovey KM, Whitcomb BW, et al. Effect of daily fiber intake on reproductive function: the BioCycle Study. The American Journal of Clinical Nutrition 2009; 90: 1061-9. [21] Lyngsø J, Ramlau-Hansen CH, Høyer BB, Støvring H, Bonde JP, Jönsson BA, et al. Menstrual cycle characteristics in fertile women from Greenland, Poland and Ukraine exposed to perfluorinated chemicals: a cross-sectional study. Human Reproduction 2014; 29: 359-67. [22] Marshall JC. Regulation of gonadotropin synthesis and secretion, in: L.J. DeGroot, J.L. Jameson (Eds.), Endocrinology, fifth ed., Elsevier, Philadelphia, PA, 2006, pp. 2687-91. [23] Waller K, Swan SH, Windham GC, Fenster L, Elkin EP, Lasley BL. Use of urine biomarkers to evaluate menstrual function in healthy premenopausal women. American Journal of Epidemiology 1998; 147: 1071-80. [24] Windham GC, Elkin E, Fenster L, Waller K, Anderson M, Mitchell PR, et al. Ovarian hormones in premenopausal women: variation by demographic, reproductive and menstrual cycle characteristics. Epidemiology 2002; 13: 675-84. [25] Wise L, Mikkelsen E, Rothman K, Riis A, Sørensen H, Huybrechts K, et al. A prospective cohort study of menstrual characteristics and time to pregnancy. American Journal of Epidemiology 2011; 174: 701-9. [26]
Barsom
SH, Mansfield
PK, Koch
PB, Gierach
G, West
SG.
Association between psychological stress and menstrual cycle characteristics in perimen opausal women. Womens Health Issues 2004; 14: 235-41.
17
Figure captions
Figure 1. The mean urinary concentration of estrogen-equivalent total paraben (ETP) according to 3 groups of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary ETP concentration. The mean adjusted for age, BMI, age at menarche and meat consumption. ETP was the sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6]. “Shorter” menstrual cycle length group includes female students with < 30 days (25 percentile) of menstrual cycle length and “longer” group includes > 36 days (75 percentile) of menstrual cycle length.
18
Figure 2. The mean urinary concentration of butyl paraben (BP) according to 3 groups of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary BP concentration. The mean adjusted for age, BMI, age at menarche and meat consumption. “Shorter” menstrual cycle length group includes female students with < 30 days (25 percentile) of menstrual cycle length and “longer” group includes > 36 days (75 percentile) of menstrual cycle length.
19
Figure 3. The mean urinary concentration of propyl paraben (PP) according to 2 groups of intra-individual variability of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary PP concentration. The mean adjusted for age, BMI, age at menarche and menstrual pain. “Small” intra-individual variability group includes female students with ≤ 6.1 days (median) of menstrual cycle length and “large” group includes > 6.1 days of menstrual cycle length.
20
Figure 4. The mean urinary concentration of estrogen-equivalent total paraben (ETP) according to 2 groups of intra-individual variability of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary ETP concentration. The mean adjusted for age, BMI, age at menarche and menstrual pain. ETP was the sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6]. “Small” intra-individual variability group includes female students with ≤ 6.1 days (median) of menstrual cycle length and “large” group includes > 6.1 days (median) of menstrual cycle length.
21
Figure 5. The positive association between menstrual cycle length and intra-individual variability of cycle length (Spearman coefficient correlation, r=0.54, p< 0.001, n=128). X-axis represents mean value of mean menstrual cycle lengths of subjects during 5 months survey period.
22
Table 1. Biological attributes and lifestyle of subjects in this study. N
Mean ± SD
Median (min-max)
Age (years)
127
20.2±0.9
20 (19-22)
BMI (kg/m2)
111
20.8±2.0
20.6 (16.6-27.2)
Age at menarche (years)
120
12.2±1.4
12 (10-16)
35±12
32 (21-98)
10±13
6.1 (0.7-70)
4±1
4 (1-7)
Menstrual cycle length (days)a < 25 percentile
31
25 – 75 percentile (30-36 d)
66
75 percentile <
31
Intra-individual variability of menstrual cycle (days)b ≤ 50% 50% < Menstrual cycles per women (cycles)a
60 65 128
Menstrual pain No
23
Yes
101
Smoking Non-smoker Smoker
124 2
Alcohol consumption (g/week)
128
29.6±52.4
8.13 (0.0-29.4)
Caffeine consumption (mg/week)
128
556±494
490 (175-740)
Physical activity
a
No
104
Yes
23
Menstrual cycle length was mean values of mean menstrual cycle lengths of each female
students for 5 months. b Intra-individual variability is expressed as standard deviation of menstrual cycle lengths within an individual.
23
Table 2. Concentrations of parabens in urine (n=128).
MP
EP
PP
BP
ETPb
Detection rate (%)
Unit
100
ng/mL
97
98
70
NAc
ng/mL
ng/mL
ng/mL
μM
Arithmetic mean ± SDa
Geometric mean (SD)a
Median (min-max)a
Unadjusted
369±532
206 (3.28)
273 (2.97-4107)
SG-adjusted
340±366
205 (3.28)
285 (2.48-2738)
Unadjusted
16.1±25.5
4.80 (5.47)
4.04 (< LOD-166)
SG-adjusted
18.2±36.4
4.77 (5.54)
3.45 (< LOD-288)
Unadjusted
31.2±55.9
7.73 (6.91)
8.39 (< LOD-441)
SG-adjusted
29.7±45.8
7.69 (6.89)
7.38 (< LOD-245)
Unadjusted
5.00±12.0
0.565 (10.7)
0.634 (< LOD-73.7)
SG-adjusted
4.58±10.6
0.562 (10.6)
0.690 (< LOD-67.0)
Unadjusted
24.9±35.7
10.4 (4.33)
12.0 (0.234-224)
SG-adjusted
23.7±30.1
10.4 (4.28)
12.3 (0.226-168)
Abbreviations: MP, methyl paraben, EP, ethyl paraben, PP, propyl paraben, BP, butyl paraben, ETP, estrogen-equivalent total paraben, SG, specific gravity a
Means and median were calculated by substituting 1/2 of detection limit for subjects with urinary concentration below detection limit. Detection 24
limits for MP, EP, PP and BP were 0.54, 0.25, 0.15 and 0.064 ng/mL, respectively. b ETP was the sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6]. c Not Applicable.
Table 3. Spearman correlation coefficientsa between the concentrations of parabens (n=128). MP
a
EP
PP
MP
-
EP
0.36
-
PP
0.48
0.37
-
BP
0.31
0.26
0.46
BP
-
Correlation coefficient between log-transformed SG-adjusted concentrations. All of the coefficients were significant (p<0.01).
25
Table 4. The result of ordinal logistic regression analysis for menstrual cycle length (n=128). Adjusteda
Crude Odds
a
95% confidence interval of Odds lower limit
upper limit
p
Odds
95% confidence interval of Odds lower limit
upper limit
p
Urinary MP concentrationb
0.98
0.74
1.3
0.86
0.88
0.62
1.3
0.49
Urinary EP concentrationb
1.0
0.82
1.2
0.98
1.0
0.82
1.3
0.86
Urinary PP concentrationb
0.88
0.74
1.1
0.16
0.85
0.70
1.0
0.10
Urinary BP concentrationb
0.89
0.77
1.0
0.11
0.83
0.70
0.99
0.037
Urinary ETPbc
0.83
0.66
1.0
0.10
0.73
0.56
0.96
0.027
Categorized menstrual cycle length were adjusted with covariates; age, BMI, age at menarche, and meat consumption. b SG-adjusted and log-
transformed value was used. c Estrogen-equivalent total concentration (ETP): The sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6].
26
Table 5. The result of binary logistic regression analysis for intra-individual variability (n=125). Adjusteda
Crude Odds
a
95% confidence interval of Odds lower limit
upper limit
p
Odds
95% confidence interval of Odds lower limit
upper limit
p
Urinary MP concentrationb
0.95
0.70
1.3
0.72
0.79
0.53
1.2
0.24
Urinary EP concentrationb
0.91
0.74
1.1
0.39
0.83
0.64
1.1
0.17
Urinary PP concentrationb
0.86
0.72
1.0
0.12
0.83
0.67
1.0
0.087
Urinary BP concentrationb
0.98
0.84
1.1
0.77
0.92
0.77
1.1
0.38
Urinary ETPbc
0.87
0.68
1.1
0.26
0.77
0.57
1.0
0.093
Categorized intra-individual variability was adjusted with covariates; age, BMI, age at menarche, and menstrual pain. b SG-adjusted and log-
transformed concentrations were used.
c
Estrogen-equivalent total concentration (ETP): the sum of the urinary concentrations of 4 parabens
weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6].
27
S0890-6238(16)30116-2 http://dx.doi.org/doi:10.1016/j.reprotox.2016.05.010 RTX 7301
To appear in:
Reproductive Toxicology
Received date: Revised date: Accepted date:
19-10-2015 2-5-2016 13-5-2016
Please cite this article as: Nishihama Yukiko, Yoshinaga Jun, Iida Ayaka, Konishi Shoko, Imai Hideki, Yoneyama Miyuki, Nakajima Daisuke, Shiraishi Hiroaki.Association between paraben exposure and menstrual cycle in female university students in Japan.Reproductive Toxicology http://dx.doi.org/10.1016/j.reprotox.2016.05.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Title Association between Paraben Exposure and Menstrual Cycle in Female University Students in Japan
Author names and affiliations Yukiko Nishihamaa, Jun Yoshinagab, Ayaka Iidac, Shoko Konishid,e, Hideki Imaif, Miyuki Yoneyamag, Daisuke Nakajimag, and Hiroaki Shiraishig
a
Department of Environmental Studies, University of Tokyo, Kashiwanoha 5-1-5, Kashiwa,
Chiba 277-8563, Japan. b Faculty of Life Sciences, Toyo University, 1-1-1 Izumino, Itakura, Ora, Gunma 374-0193, Japan.
c
Department of Environmental, Occupational Health and
Toxicology, School of Allied Health Sciences, Kitasato University, Kitasato 1-15-1, Minami, Sagamihara, Kanagawa 252-0373, Japan. d Department of Human Ecology, Graduate School of Medicine, University of Tokyo, Hongou 7-3-1, Bunkyou, Tokyo 113-0033, Japan.
e
Department of Anthropology, University of Washington, Seattle WA 98195, USA. f Department of Nursing, Tokyo Healthcare University, Higashigaoka 2-5-1, Meguro, Tokyo 152-8558, Japan. g National Institute for Environmental Studies, Onogawa 16-2, Tsukuba, Ibaraki 3058506, Japan.
Corresponding author: Jun Yoshinaga, PhD Faculty of Life Sciences, Toyo University, 1-1-1 Izumino, Itakura, Ora, Gunma 374-0193, Japan. Tel/Fax: +81-276-82-9031. E-mail address: [email protected]
1
Highlights
Possible effect of paraben exposure on menstrual cycle was explored.
Urinary concentrations of parabens were used as biomarker of exposure.
Shorter menstrual cycle was associated with higher urinary paraben concentrations.
Parabens exposure may be one of environmental determinants of female fecundity.
Abstract
Parabens have been known to have estrogenic activity in many in vivo and in vitro studies and biomonitoring data indicated ubiquitous exposure in general populations. However, there are few human studies on reproductive effects of parabens. In this study, menstrual cycle length and its intra-individual variation were investigated by bleeding record over the period of 5 months and urinary concentrations of parabens were measured for 128 female Japanese university students. We found significantly negative relationships between menstrual cycle length and urinary estrogen-equivalent total paraben (odds=0.73, 95% CI 0.56-0.96) and butyl paraben concentrations (odds=0.83, 0.70-0.99), which indicated shortened cycle length in women with high urinary paraben concentrations. This study indicated that paraben exposure at not excessive levels is associated with menstrual cycle length or its variability among general female subjects. This results suggest a possibility that parabens exposure is one of the environmental causes of human reproductive problem.
Keywords paraben; urine; menstrual cycle; female fecundity.
2
1. Introduction
Alkyl esters of 4-hydroxy benzoic acid (parabens), used as a preservative in personal care products (PCPs) such as cosmetics and body creams, in medications, and in foods, are suspected endocrine disrupting chemicals (EDCs) [1, 2]. In Japan, the maximum permissible level of parabens in cosmetics is regulated by the Ministry of Health, Labor and Welfare (MHLW) [3] to be less than 1.0 g /100 g as a mixture of esters and/or sodium salt. Commonlyused parabens including methyl parabens (MP), ethyl paraben (EP), propyl paraben (PP) and butyl paraben (BP) are considered relatively safe because acute toxicities such as dermal and eye irritancy, and chronic toxicities are not seemed to be observed for these parabens [1]. However, many in vitro studies have reported estrogen receptor binding capacity of parabens [4-6]. Additionally, in vivo studies have demonstrated increased uterus weight in female rats administrated with parabens [7]. Meanwhile, a limited number of studies explored reproductive effects of exposure to parabens in humans. Buttke et al. [8] investigated the association of urinary paraben concentrations with the age of menarche for 12-16 years old female participants of the NHANES study (n=287) to find non-significant association. Smith et al. [9] examined the relationship between urinary paraben concentrations and marker of ovarian reserve (day-3 follicle-stimulating hormones (FSH), antral follicle count (AFC) and ovarian volume) in a prospective cohort study of 21.0-46.7 year-old women seeking fertility treatment in Boston, Massachusetts (n=192). They indicated possible reduction of ovarian reserve by paraben exposure. Apparently more epidemiologic studies are needed on a possible reproductive effects of parabens when we take into consideration their estrogenic activities observed in in vitro and in vivo studies. Ubiquitous exposure to this group of chemicals has been revealed by frequent detections in urine samples in a number of biomonitoring studies on general population [10-
3
12]. The purpose of this study is to investigate the relationship between paraben exposure and menstrual cycle characteristics of young Japanese female subjects. Menstrual cycle is considered to be regulated by endocrine actions and thus it could be affected by the exposure to EDCs such as parabens.
2. Material and methods
2.1. Subject
The subjects of this cross-sectional study were female university students. They were recruited at Tokyo Healthcare University located in Tokyo during 2012-2013 at the orientation of class for clinical laboratory practice (n=178). The purpose and procedure of the study was explained to students by survey staffs after distributing a brochure describing the same content. We approached 178 students who attended orientation for class of clinical laboratory practice, and obtained written consent from all of the students. The ethics committees of the Tokyo Healthcare University, the University of Tokyo and National Institute for Environmental Studies approved this study.
2.2. Sampling
A single spot urine sample was collected from subjects in the clinical laboratory practice typically 1-2 months after the orientation. Urine sample was taken by the subjects herself in a paper cup during the class time and was brought to class immediately. An aliquot was dispensed into polypropylene (PP) tubes by survey staffs. Travel blanks for urine samples were prepared.
4
All of the PP tubes for urine sampling were rinsed with ultrapure water and methanol (HPLC Grade, Kanto Chemical Co. Ltd., Japan) prior to use. The samples were stored at – 20°C until analysis.
2.3. Urinary parabens analysis
Analysis of MP, EP, PP, and BP in urine samples were carried out by liquid chromatography-tandem mass spectrometry (LC-MS/MS) after enzymatic deconjugation and solid phase extraction [13]. In brief, 1 mL of urine sample was added with d4-labelled internal standards, enzymatically deconjugated, and was subject to solid phase extraction (Sep-Pak ® Vac 6 cc (500 mg) tC18 Cartridges, Waters, Massachusetts, USA), which was followed by LC-MS/MS analysis (LC: Agilent 1200 series, MS/MS: Agilent 6460 Triple Quad LC/MS, Agilent Technologies, CA, USA). No parabens were detected in travel blanks. Recoveries of the internal standards were monitored for all of the samples and the result was 50±17% (MP), 44±15% (EP), 52±17% (PP) and 47±14% (BP) (n=128). The instrumental detection limit for the parabens was defined as the concentration corresponded to S/N=3 calculated from 7 repeated measurements of 2.0 ng/mL (MP) or 0.5 ng/mL (EP, PP, and BP) standard solutions. The recovery-corrected detection limits were 0.54, 0.25, 0.15 and 0.064 ng/mL in urine for MP, EP, PP, and BP, respectively. An in-house quality control urine sample was included in every batch (typically 20 samples/batch) of sample preparation and measurement. The repeatability and reproducibility of analysis were 1.0-2.4% and 1.4-3.3%, respectively, for the 4 parabens. The urinary concentrations of parabens (ng/mL) were adjusted with specific gravity (SG) of 1.020.
5
2.4. Menstrual cycle characteristics
Questionnaire survey on menstrual cycle in this study has been described in detail in our previous study [14]. The subjects were asked to keep a diary for recording their menstrual bleeding for 5 months to a calendar we prepared for this purpose from the day of their agreement of study participation (generally 1-2 months before urine sampling). Menstruation was defined as ≥ 3 consecutive days of bleeding by following the definition of the Japan Society of Obstetrics and Gynecology (JSOG) guideline [15]. Menstrual cycle length for each subject was defined as the mean of the intervals from the first day of an index menstruation to the day before the next menstruation during 5 months survey period. If the truncated days (days from the first day of survey to the first day of the first menstruation, or from the last day of an index menstruation to the last day of record) was longer than the mean menstrual cycle of the subject during 5 months survey period, we regarded the truncated days as one menstrual cycle and recalculated the mean menstrual cycles by adding this cycle. Intra-individual variability of menstrual cycle was defined as the standard deviation (SD) of menstrual cycle lengths recorded for each individual. The subjects were also asked to answer a brief questionnaire which asked about age at menarche, menstrual pain, oral contraceptive (OC) use in the past 6 months, and experience of pregnancy. The menstrual pain was asked by the following 4 categories: “almost none”; “feel pain, but no problem in daily activity”; “can go out but exercise is difficult”, and “too painful to get out of bed” [16].
6
2.5. Biological attributes and life style
The subjects were also asked to answer another questionnaire on biological attributes (age, height, and weight), lifestyle (active/passive smoking, alcohol and caffeine intake, physical activity), and food habit. Food habit of the subjects was asked by frequency of consumption of 6 foods, which were considered major sources of exogenous estrogen or estrogenic chemicals [17, 18]. For the consumption of meat, milk, milk products and egg, the frequency was asked by the following 7 scales: “every meal”, “daily”, “four and five times a week”, “two and three times a week”, “less than once a week”, “one and two times a month”, or “none”. For fish and soy products, the frequency was asked by 6 scales: “every meal”, “daily”, “four and five times a week”, “two and three times a week”, “less than once a week”, or “none”.
2.6. Statistical analyses
SG-adjusted urinary parabens concentrations were used for the following statistical analyses. The distribution of these concentrations skewed towards higher value; therefore, the concentrations were transformed to natural logarithm. When a paraben concentration was below the detection limit, 1/2 of detection limit value was substituted in statistical analysis. In addition to individual paraben concentrations, we used estrogen-equivalent total paraben (ETP) in our statistical analysis by summing the individual concentrations of the four parabens weighted by their relative estrogenic activity (eq. 1) [13]. The weighing factor was derived from in vitro yeast estrogen screen assays [6]. [ETP] = [1*MP + 16.7*EP + 83.3*PP + 250*BP] (μM) …… (eq. 1) The subjects were categorized into 3 groups by their menstrual cycle length in the quartiles: subjects whose mean menstrual cycle length was < 25 percentile, 25-75 percentile,
7
and > 75 percentile were categorized into “shorter cycle”, “moderate cycle” and “longer cycle”, respectively. With regard to intra-individual variability of menstrual cycles, the subjects were categorized into “small” and “large” by the median of SD of individuals’ menstrual cycle length. The menstrual pain was divided into two categories “no” (“almost none”) and “yes” (the rest) groups. With regard to frequencies of food consumption, subjects were re-categorized into 2 categories by one of original 6 or 7 scales which was mentioned in the 2.5. Biological attributes and life style so that number of subjects in the two was roughly equal. For the analyses of the relationship between the 3 categories of menstrual cycle length or 2 categories of intra-individual variability and their biological attributes and lifestyles, chisquared test, Kruskal-Wallis test, or Mann-Whitey’s U test were used. An ordinal logistic regression analysis was used to explore whether urinary paraben levels are associated with the 3 groups of menstrual cycle length (“shorter”, “moderate”, and “longer” cycle). Either one of the 4 urinary parabens concentration and ETP was used as an independent variable. A (binomial) logistic regression analysis was conducted to investigate if urinary parabens levels are associated with their intra-individual variability with 2 groups: classified intra-individual variability (“small” and “large”) as a dependent variable and each 4 urinary parabens and ETP concentration as an independent variable. In addition to biological attributes of the subjects (age, BMI, and age at menarche), other variables were included in the logistic regression analyses as independent variable when the variable was found to be statistically significantly associated with dependent variables in the preliminary bivariate analysis. We used forced entry method for logistic regression analyses. Statistical analyses were performed with SPSS for Windows (version 12.0; SPSS Japan Inc., Tokyo, Japan). Statistical significance was defined as probability of < 0.05.
8
3. Results
Of 178 subjects agreed to participate in this study, 145 (81%) completed bleeding diary and questionnaires. Of the 145, 15 OC users in the past 6 months and 2 subjects whose age was regarded as outlier by Smirnov-Grubbs test (25 and 36 years) were excluded, thus leaving 128 subjects for further analyses. No subject reported experience of pregnancy. Table 1 shows the characteristics of the 128 subjects. Mean ± SD (min-max) of age and BMI was 20.2±0.9 (19-22) years and 20.8±2.0 (16.6-27.2) kg/m2, respectively. Mean ± SD of menstrual cycle length was 35±12 days with the median (min-max) being 32 (20.9-98.0) days. The mean of intra-individual variability (SD of cycle lengths within individuals) was 10±13 days and the median (min-max) was 6.1 (0.66-70) days. Mean ± SD (min-max) of menstrual cycles per women were 4±1 (1-7) cycles. Preliminary bivariate analyses between length and intra-individual variability of menstrual cycles and biological, lifestyle factors and dietary habit found a suggestive association between menstrual cycle length and meat consumption (chi-squared test, p=0.090), and between intraindividual variability and menstrual pain (chi-squared test, p=0.060). These variables were used as covariates in the following logistic regression analysis. Table 2 shows unadjusted and SG- adjusted mean, geometric mean and median urinary concentrations of MP, EP, PP, and BP of the present subjects. MP, EP, PP and BP were detected in 100, 97, 98 and 70% of the subjects, respectively. One half of detection limit value, 0.13, 0.075, and 0.032 ng/mL for EP, PP, and BP, respectively, was substituted for undetectable subjects for the calculation of mean value (Table 2) and in the following statistical analyses. SG-adjusted mean ± SD concentrations were 340±366, 18.2±36.4, 29.7±45.8 and 4.58±10.6 ng/mL for MP, EP, PP and BP, respectively and 23.7±30.1 μM for ETP. The relative contribution of PP, MP, BP and EP to ETP was 46.1%, 21.7%, 19.9% and 12.3%, respectively.
9
Table 3 shows Spearman correlation coefficients between the SG-adjusted concentrations of 4 parabens in the urine samples. We found that all of the paraben pairs had a significantly positive correlation (r=0.26 to 0.48, p< 0.01). The results of logistic regression analyses using menstrual cycle length and intraindividual variability as dependent variable were shown in Tables 4 and 5. Only the results of models with a significant independent variable were shown in these tables. We found a significant relationship between menstrual cycle length and urinary ETP concentrations (adjusted odds= 0.73 (95% CI: 0.56, 0.96), p=0.027, Table 4) and BP concentrations (adjusted odds=0.83 (0.70, 0.99), p=0.037, Table 4). The urinary concentrations of ETP and BP were higher in subjects with shorter cycle lengths than those with moderate and longer cycle lengths (Figure 1 and Figure 2). Urinary MP, EP and PP concentration was not significantly associated with menstrual cycle length of the subjects. Variables used as covariates, i.e., age, BMI, age at menarche and meat consumption were not significant in all of the models. Urinary PP (adjusted odds=0.83 (0.67, 1.0), p=0.087, Table 5) and ETP concentrations (adjusted odds 0.77 (0.57, 1.0), p=0.093, Table 5) were selected as independent variables with suggestive association for intra-individual variability in the binary logistic regression analysis. Urinary concentrations of PP and ETP were higher in subjects with smaller intra-individual variability than in those with larger variability (Figure 3 and Figure 4).
4. Discussion
4.1. Menstrual cycle characteristics
The median cycle length of the present subjects (32 days, Table 1) was consistent with,
10
but intra-individual variability (median 6.1 days, Table 1) was larger than, that of the previous study on Japanese women of similar age (31 days and 3 days) [19]. However, the median cycle length of the present subjects seemed greater than those reported for women in US and Europe [20, 21] without known reasons.
4.2. Urinary parabens level
Median urinary concentrations of parabens of the present subjects (Table 2) were similar to those obtained for pregnant women (32.6 years, 108, 7.26 and 0.80 ng/mL for MP, EP and BP) in a previous study in Japan [13] except for PP for which higher median concentration than ours was reported (33.3 ng/mL). Median urinary concentrations were also similar to those for the U.S. female (> 6 years old), i.e., 137, 1.3, 29.1 and 0.50 ng/mL (unadjusted) for MP, EP, PP and BP, respectively [11]. Thus exposure levels of parabens of the present subjects were considered not apart from levels of general populations in Japan and the US. We found significant correlations between the concentrations of parabens (Spearman’s rank correlation coefficient: 0.26-0.48) (p< 0.01, Table 3). This result was probably due to the fact that more than one paraben are often added to a PCP in many cases to enhance preservative effect [1]. This means that co-exposure to multiple parabens takes place by the use of PCPs. Since the 4 parabens measured in this study are known to have estrogenic activity in common, but with varying relative potencies [6], additive effect is expected by the co-exposure. This was the reason of our usage of ETP in this study. The contribution of each of the 4 parabens to the median ETP concentration was 21.7%, 12.3%, 46.1% and 19.9% for MP, EP, PP and BP respectively. The greatest contribution of PP to ETP in the present subjects suggests that PP has the greatest risk potential for the present subjects among the 4 parabens. BP is almost same contribution of MP though the average concentration was the lowest (median: 0.690 ng/mL,
11
Table 2).
4.3. The relationship between menstrual cycle and urinary parabens concentration
As far as we know, this is the first epidemiologic study that explored a possible relationship between menstrual cycle and parabens exposure. In this study, significant relationship between menstrual cycle lengths and urinary ETP and BP concentrations were found (Table 4), suggesting that exposure to parabens shortens menstrual cycle length. A mechanism for the observed relationship between menstrual cycle length and exposure to parabens is not clear. When estrogen level rises, a negative-feedback system works to decrease FSH secretion leading to LH surge and subsequent ovulation [22]. It is well known that shorter menstrual cycle length is accompanied with shorter follicular phase length [23]. Higher estrogen-related metabolite excretion is reported for women with shorter menstrual cycle length [24]. Thus, elevated level of circulating estrogen results in earlier transition from follicular phase to luteal phase, and exogenous estrogen-like substances, such as parabens, may similarly works. Suggestive relationship between intra-individual variability and urinary PP and ETP concentrations was found in the present study (Table 5). This result indicated that the subjects with smaller intra-individual variability were exposed to higher levels of parabens. The underlying mechanism of the relationship is not clear either but we noticed that subjects with longer cycle length tended to have greater intra-individual variability (Spearman coefficient correlation: r=0.54, p< 0.001, Figure 5). Taking this positive relationship between menstrual cycle length and intra-individual variability into consideration, if parabens have an effect to shorten menstrual cycle length, then reduced intra-individual variability among higher urinary paraben levels, as observed in this study, would be apparently expected. Another explanation
12
is that parabens exposure reduced intra-individual variability, which resulted in apparently higher exposure levels among the subjects with shorter menstrual cycle length. The former explanation seems likely because the former relationship is significant but the later was not significant. However, it is not possible to definitely distinguish the two explanations from this study result alone. Epidemiologic studies reported that women who had shorter menstrual cycle [25] were prone to have prolonged time-to-pregnancy. This indicated reduced fecundity among women with shorter menstrual cycle length. If exposure to parabens has an effect to shorten human menstrual cycle length as indicated by the present study, then reduced fecundity would be expected for the exposed women. This may pose a serious concern as one of potential environmental causes of subfertility/infertility problem prevalent in many countries. Therefore, further studies are warranted to confirm the present result. Reproductive effect of paraben exposure in humans has been explored in two previous studies as far as we are aware; Buttke et al. [8] examined if there is a relationship between age of menarche and exposure to phenols including parabens among 12-16 year-old girls. In this study, the sum of urinary concentration of MP and PP was used as a biomarker of exposure to parabens. They found no significant association between the two. Smith et al. [9] explored the association between measures of ovarian reserve (day-3 follicle-stimulating hormone, antral follicle count (AFC), and ovarian volume) and urinary MP, PP and BP concentrations for 21.046.7 year-old women who underwent fertility treatment (n=192). They found a suggestive, but not a significant, negative trend between AFC and urinary PP levels (p=0.07). Thus studies to date did not find significant reproductive effects of paraben exposure in humans. The strength of our study included 1) menstrual cycles of subjects were assessed by a recording method, not by a recall method, for 5 months. 2) paraben exposure was assessed by biomarker of exposure. Urinary paraben levels in spot urine samples were reported to represent
13
exposure level of subject over longer period of time [12]. Moreover, 3) the fact that we limited our subjects to university student resulted in small inter-individual variabilities in age, one of the most influential factors of menstrual cycle length, and other attributes [14]; this reduced the possibility of potential confounding, though we included potential covariates of menstrual cycle length in the regression models. However, it is still possible that we failed to include some other potential covariates, e.g., psychological stress [26], and this might be a limitation of this study, in addition to a moderate sample size.
5. Conclusions
We found significant association between menstrual cycle length and urinary ETP and BP concentrations (p< 0.05) among female Japanese university students indicating that exposure to parabens shortens menstrual cycle length. This study result suggests that paraben exposure plays a role as environmental causes of infertility/subfertility, which is one of a public health concern across the world. Further epidemiologic studies using a larger sample size are needed.
Conflict of interest None declared.
Acknowledgement
The authors sincerely appreciate the cooperation of the subjects. They also appreciate Dr. Kouda, Department of Nursing, Tokyo Healthcare University, for her help in the sampling.
14
References
[1] Soni MG, Carabin IG, Burdock GA. Safety assessment of esters of p-hydroxybenzoic acid (parabens). Food and Chemical Toxicology 2005; 43: 985-1015. [2] Błędzka D, Gromadzińska J, Wąsowicz W. 2014. Parabens. From environmental studies to human health. Environment International 2014; 67: 27-42. [3] MHLW (Ministry of Health, Labor and Welfare). Standards for Cosmetics (Ministry of Health and Welfare Notification No.331 of 2000). Tokyo, Japan: MHLW 2000. [4] Nishihara T, Nishikawa J, Kanayama T, Dakeyama F, Saito K, Imagawa M, et al. Estrogenic Activities of 517 Chemicals by Yeast Two-Hybrid Assay. Journal of Health Science 2000; 46: 282-98. [5] Golden R, Gandy J, Vollmer G. A review of the endocrine activity of parabens and implications for potential risks to human health. Critical Reviews in Toxicology 2005; 35: 435-58. [6] Routledge EJ, Parker J, Odum J, Ashby J, Sumpter JP. Some alkyl hydroxy benzoate preservatives (parabens) are estrogenic. Toxicology and Applied Pharmacology 1998; 153: 12-9. [7] Lemini C, Jaimez R, Avila ME, Franco Y, Larrea F, Lemus AE. In vivo and in vitro estrogen bioactivities of alkyl parabens. Toxicol. Ind. Health 2003; 19: 69-79. [8] Buttke DE, Sircar K, Martin C. Exposures to endocrine-disrupting chemicals and age of menarche in adolescent girls in NHANES (2003-2008). Environmental Health Perspectives 2012; 120: 1613-18. [9] Smith KW, Souter I, Dimitriadis I, Ehrlich S, Williams PL, Calafat AM, et al. Urinary paraben concentrations and ovarian aging among women from a fertility center. Environmental Health Perspectives 2013; 121: 1299-305.
15
[10] Ye X, Bishop AM, Reidy JA, Needham LL, Calafat AM. Parabens as urinary biomarkers of exposure in humans. Environmental Health Perspectives 2006; 114: 1843-46. [11] Calafat AM, Ye X, Wong LY, Bishop AM, Needham LL. Urinary concentrations of four parabens in the U.S. population: NHANES 2005-2006. Environmental Health Perspectives 2010; 118: 679-85. [12] Smith KW, Braun JM, Williams PL, Ehrlich S, Correia KF, Calafat AM, et al. Predictors and variability of urinary paraben concentrations in men and women, including before and during pregnancy. Environmental Health Perspectives 2012; 120: 1538-43. [13] Shirai S, Suzuki Y, Yoshinaga J, Shiraishi H, Mizumoto Y. Urinary excretion of parabens in pregnant Japanese women. Reproductive Toxicology 2013; 35: 96-101. [14] Nishihama Y, Yoshinaga J, Iida A, Konishi S, Imai H. Menstrual cycle length and source of its variation in female university students majoring in nursing sciences. Nihon Eiseigaku Zasshi 2015; 70: 139-48 (in Japanese). [15] Mikuni M, Fujimoto S, Toda T, Kohno Y, Terakawa N, Ishikawa M, et al. Diagnosis, Treatment and Management of Gynecologic Diseases, Endocrine Diseases. Acta obstetrica et gynaecologica Japonica 2002; 54 (12): N-557 - 8. [16] Konishi S, Nishihama Y, Iida A, Yoshinaga J, Imai H. Association of antimüllerian hormone levels with menstrual-cycle type and dysmenorrhea in young asymptomatic women. Fertility and Sterility 2014; 102: 1439-43. [17] Mendola P, Buck GM, Sever LE, Zielezny M, Vena JE. 1997. Consumption of PCBcontaminated freshwater fish and shortened menstrual cycle length. American Journal of Epidemiology 1997; 146: 955-60. [18] Hartmann S, Lacorn M, Steinhart H. Natural occurrence of steroid hormones in food. Food Chemistry 1998; 62: 7-20. [19] Yano C, Tomiyasu T, Anai T. A general survey on what consitutes a "normal" menstrual
16
cycle. Bosei Eisei 2005; 45: 496-502 (in Japanese). [20] Gaskins AJ, Mumford SL, Zhang C, Wactawski-Wende J, Hovey KM, Whitcomb BW, et al. Effect of daily fiber intake on reproductive function: the BioCycle Study. The American Journal of Clinical Nutrition 2009; 90: 1061-9. [21] Lyngsø J, Ramlau-Hansen CH, Høyer BB, Støvring H, Bonde JP, Jönsson BA, et al. Menstrual cycle characteristics in fertile women from Greenland, Poland and Ukraine exposed to perfluorinated chemicals: a cross-sectional study. Human Reproduction 2014; 29: 359-67. [22] Marshall JC. Regulation of gonadotropin synthesis and secretion, in: L.J. DeGroot, J.L. Jameson (Eds.), Endocrinology, fifth ed., Elsevier, Philadelphia, PA, 2006, pp. 2687-91. [23] Waller K, Swan SH, Windham GC, Fenster L, Elkin EP, Lasley BL. Use of urine biomarkers to evaluate menstrual function in healthy premenopausal women. American Journal of Epidemiology 1998; 147: 1071-80. [24] Windham GC, Elkin E, Fenster L, Waller K, Anderson M, Mitchell PR, et al. Ovarian hormones in premenopausal women: variation by demographic, reproductive and menstrual cycle characteristics. Epidemiology 2002; 13: 675-84. [25] Wise L, Mikkelsen E, Rothman K, Riis A, Sørensen H, Huybrechts K, et al. A prospective cohort study of menstrual characteristics and time to pregnancy. American Journal of Epidemiology 2011; 174: 701-9. [26]
Barsom
SH, Mansfield
PK, Koch
PB, Gierach
G, West
SG.
Association between psychological stress and menstrual cycle characteristics in perimen opausal women. Womens Health Issues 2004; 14: 235-41.
17
Figure captions
Figure 1. The mean urinary concentration of estrogen-equivalent total paraben (ETP) according to 3 groups of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary ETP concentration. The mean adjusted for age, BMI, age at menarche and meat consumption. ETP was the sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6]. “Shorter” menstrual cycle length group includes female students with < 30 days (25 percentile) of menstrual cycle length and “longer” group includes > 36 days (75 percentile) of menstrual cycle length.
18
Figure 2. The mean urinary concentration of butyl paraben (BP) according to 3 groups of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary BP concentration. The mean adjusted for age, BMI, age at menarche and meat consumption. “Shorter” menstrual cycle length group includes female students with < 30 days (25 percentile) of menstrual cycle length and “longer” group includes > 36 days (75 percentile) of menstrual cycle length.
19
Figure 3. The mean urinary concentration of propyl paraben (PP) according to 2 groups of intra-individual variability of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary PP concentration. The mean adjusted for age, BMI, age at menarche and menstrual pain. “Small” intra-individual variability group includes female students with ≤ 6.1 days (median) of menstrual cycle length and “large” group includes > 6.1 days of menstrual cycle length.
20
Figure 4. The mean urinary concentration of estrogen-equivalent total paraben (ETP) according to 2 groups of intra-individual variability of menstrual cycle length in Japanese female students (n=128). Error bars indicate 95% confidence interval of urinary ETP concentration. The mean adjusted for age, BMI, age at menarche and menstrual pain. ETP was the sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6]. “Small” intra-individual variability group includes female students with ≤ 6.1 days (median) of menstrual cycle length and “large” group includes > 6.1 days (median) of menstrual cycle length.
21
Figure 5. The positive association between menstrual cycle length and intra-individual variability of cycle length (Spearman coefficient correlation, r=0.54, p< 0.001, n=128). X-axis represents mean value of mean menstrual cycle lengths of subjects during 5 months survey period.
22
Table 1. Biological attributes and lifestyle of subjects in this study. N
Mean ± SD
Median (min-max)
Age (years)
127
20.2±0.9
20 (19-22)
BMI (kg/m2)
111
20.8±2.0
20.6 (16.6-27.2)
Age at menarche (years)
120
12.2±1.4
12 (10-16)
35±12
32 (21-98)
10±13
6.1 (0.7-70)
4±1
4 (1-7)
Menstrual cycle length (days)a < 25 percentile
31
25 – 75 percentile (30-36 d)
66
75 percentile <
31
Intra-individual variability of menstrual cycle (days)b ≤ 50% 50% < Menstrual cycles per women (cycles)a
60 65 128
Menstrual pain No
23
Yes
101
Smoking Non-smoker Smoker
124 2
Alcohol consumption (g/week)
128
29.6±52.4
8.13 (0.0-29.4)
Caffeine consumption (mg/week)
128
556±494
490 (175-740)
Physical activity
a
No
104
Yes
23
Menstrual cycle length was mean values of mean menstrual cycle lengths of each female
students for 5 months. b Intra-individual variability is expressed as standard deviation of menstrual cycle lengths within an individual.
23
Table 2. Concentrations of parabens in urine (n=128).
MP
EP
PP
BP
ETPb
Detection rate (%)
Unit
100
ng/mL
97
98
70
NAc
ng/mL
ng/mL
ng/mL
μM
Arithmetic mean ± SDa
Geometric mean (SD)a
Median (min-max)a
Unadjusted
369±532
206 (3.28)
273 (2.97-4107)
SG-adjusted
340±366
205 (3.28)
285 (2.48-2738)
Unadjusted
16.1±25.5
4.80 (5.47)
4.04 (< LOD-166)
SG-adjusted
18.2±36.4
4.77 (5.54)
3.45 (< LOD-288)
Unadjusted
31.2±55.9
7.73 (6.91)
8.39 (< LOD-441)
SG-adjusted
29.7±45.8
7.69 (6.89)
7.38 (< LOD-245)
Unadjusted
5.00±12.0
0.565 (10.7)
0.634 (< LOD-73.7)
SG-adjusted
4.58±10.6
0.562 (10.6)
0.690 (< LOD-67.0)
Unadjusted
24.9±35.7
10.4 (4.33)
12.0 (0.234-224)
SG-adjusted
23.7±30.1
10.4 (4.28)
12.3 (0.226-168)
Abbreviations: MP, methyl paraben, EP, ethyl paraben, PP, propyl paraben, BP, butyl paraben, ETP, estrogen-equivalent total paraben, SG, specific gravity a
Means and median were calculated by substituting 1/2 of detection limit for subjects with urinary concentration below detection limit. Detection 24
limits for MP, EP, PP and BP were 0.54, 0.25, 0.15 and 0.064 ng/mL, respectively. b ETP was the sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6]. c Not Applicable.
Table 3. Spearman correlation coefficientsa between the concentrations of parabens (n=128). MP
a
EP
PP
MP
-
EP
0.36
-
PP
0.48
0.37
-
BP
0.31
0.26
0.46
BP
-
Correlation coefficient between log-transformed SG-adjusted concentrations. All of the coefficients were significant (p<0.01).
25
Table 4. The result of ordinal logistic regression analysis for menstrual cycle length (n=128). Adjusteda
Crude Odds
a
95% confidence interval of Odds lower limit
upper limit
p
Odds
95% confidence interval of Odds lower limit
upper limit
p
Urinary MP concentrationb
0.98
0.74
1.3
0.86
0.88
0.62
1.3
0.49
Urinary EP concentrationb
1.0
0.82
1.2
0.98
1.0
0.82
1.3
0.86
Urinary PP concentrationb
0.88
0.74
1.1
0.16
0.85
0.70
1.0
0.10
Urinary BP concentrationb
0.89
0.77
1.0
0.11
0.83
0.70
0.99
0.037
Urinary ETPbc
0.83
0.66
1.0
0.10
0.73
0.56
0.96
0.027
Categorized menstrual cycle length were adjusted with covariates; age, BMI, age at menarche, and meat consumption. b SG-adjusted and log-
transformed value was used. c Estrogen-equivalent total concentration (ETP): The sum of the urinary concentrations of 4 parabens weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6].
26
Table 5. The result of binary logistic regression analysis for intra-individual variability (n=125). Adjusteda
Crude Odds
a
95% confidence interval of Odds lower limit
upper limit
p
Odds
95% confidence interval of Odds lower limit
upper limit
p
Urinary MP concentrationb
0.95
0.70
1.3
0.72
0.79
0.53
1.2
0.24
Urinary EP concentrationb
0.91
0.74
1.1
0.39
0.83
0.64
1.1
0.17
Urinary PP concentrationb
0.86
0.72
1.0
0.12
0.83
0.67
1.0
0.087
Urinary BP concentrationb
0.98
0.84
1.1
0.77
0.92
0.77
1.1
0.38
Urinary ETPbc
0.87
0.68
1.1
0.26
0.77
0.57
1.0
0.093
Categorized intra-individual variability was adjusted with covariates; age, BMI, age at menarche, and menstrual pain. b SG-adjusted and log-
transformed concentrations were used.
c
Estrogen-equivalent total concentration (ETP): the sum of the urinary concentrations of 4 parabens
weighted by relative estrogenic activity obtained in a yeast estrogen screen assay by Routledge et al. [6].
27