First Trimester Urinary Bisphenol and Phthalate Concentrations and Time to Pregnancy: A Population-Based Cohort Analysis

Abstract

Background

Increasing evidence suggests that exposure to synthetic chemicals such as bisphenols and phthalates can influence fecundability. The current study describes associations of first trimester urinary concentrations of bisphenol A (BPA), BPA analogs, and phthalate metabolites with time to pregnancy (TTP).

Methods

Among 877 participants in the population-based Generation R pregnancy cohort, we measured first trimester urinary concentrations of bisphenols and phthalates [median gestational age, 12.9 weeks (interquartile range, 12.1, 14.4)]. We used fitted covariate-adjusted Cox proportional hazard models to examine associations of bisphenol and phthalate concentrations with TTP. Participants who conceived using infertility treatment were censored at 12 months. Biologically plausible effect measure modification by folic acid supplement use was tested.

Results

In the main models, bisphenol and phthalate compounds were not associated with fecundability. In stratified models, total bisphenols and phthalic acid were associated with longer TTP among women who did not use folic acid supplements preconceptionally [respective fecundability ratios per each natural log increase were 0.90 (95% CI, 0.81 to 1.00) and 0.88 (95% CI, 0.79 to 0.99)]. Using an interaction term for the exposure and folic acid supplement use showed additional effect measure modification by folic acid supplement use for high-molecular-weight phthalate metabolites.

Conclusions

We found no associations of bisphenols and phthalates with fecundability. Preconception folic acid supplementation seems to modify effects of bisphenols and phthalates on fecundability. Folic acid supplements may protect against reduced fecundability among women exposed to these chemicals. Further studies are needed to replicate these findings and investigate potential mechanisms.

Increasing evidence suggests that synthetic chemicals can influence both male and female health and fecundability, which is the biological ability to conceive a pregnancy. Persistent organic pollutants have been linked to a host of reproductive disorders in both men and women (1, 2), including prolonged time to pregnancy (TTP), a marker of couple fecundability. The literature on nonpersistent chemicals is more limited. Prior studies of female urinary bisphenols and TTP have not found an association, but have relied exclusively on linear models (3–5). Results of studies investigating female phthalate metabolites and TTP have been inconsistent (3–8). Differences among these studies’ conclusions may stem from variations in study size and population, as well as design, because they did not all measure the same metabolites and some measured TTP prospectively, whereas others relied on retrospective recall of TTP from the first trimester of pregnancy.

DNA methylation is a potential mechanism through which bisphenols and phthalates may be related to female reproductive disorders (9). Bisphenol A (BPA), in particular, has been shown to target reproductive tissues and affect reproductive outcomes in numerous animal studies (10). In a now-classic experiment, mice exposed to BPA during pregnancy were more likely to give birth to offspring with yellow coats as a result of decreased methylation upstream of the Agouti gene. The effect was negated when BPA-exposed dams were supplemented with folic acid, a methyl donor (11). Folate depletion has been associated with global hypomethylation, but also with targeted hypermethylation [reviewed in Crider et al. (12)]. A recent study among women undergoing infertility treatment reported that high urinary BPA concentrations were associated with lower probabilities of implantation, clinical pregnancy, and live birth, but only among women who consumed <400 μg/d of dietary folate; there were no associations among women who consumed ≥400 μg/d (13).

Phthalates also affect reproductive outcomes in animals and humans (10) and, depending on the metabolite, have been shown to be associated with either DNA hypermethylation or hypomethylation. Among 336 Mexican-American newborns, phthalate metabolites [in particular, metabolites of di-2-ethylhexylphthalate (DEHP)] in maternal urine samples collected during pregnancy were associated with increased DNA methylation in cord blood (14). By contrast, among 181 mother-newborn pairs in China, maternal third trimester urinary mono(2-ethyl-5-hydroxyhexyl)phthalate and mono(2-ethyl-5-oxohexyl)phthalate were associated with decreased DNA methylation in placental tissue (15).

In the current study, we evaluate associations of urinary bisphenols and phthalate metabolites with TTP among 877 women participating in a large population-based pregnancy cohort. Because both bisphenols and phthalates affect DNA methylation, we assess potential effect measure modification by preconception folic acid supplementation, which prior studies have not explored. Our analysis expands upon earlier studies in three following additional ways. (1) Where prior studies focused exclusively on BPA, we also investigate eight BPA analogs, chemical replacements that are commonly found in “BPA-free” products and have yet to be reported on in human fertility research. (2) In addition to measuring associations between molecular weight groupings of urinary phthalate metabolites and TTP, we also examine TTP in relation to phthalic acid (PA). As the final common metabolic form of all phthalates, PA represents a proxy for total phthalate exposure and captures exposure to phthalates for which we do not have individual metabolite measures. (3) As contrasted to prior studies, which are restricted to women who have conceived naturally, we include women who conceived using some form of infertility treatment, permitting the investigation of potential associations of bisphenol and phthalate exposure with longer TTP.

Methods

Study design and population for analysis

The current study was embedded in the Generation R Study, a population-based prospective cohort study from early pregnancy onward that was approved by the Medical Ethical Committee of the Erasmus Medical Centre in Rotterdam (16). In total, 8879 women were enrolled, 76% before gestational age of 18 weeks. Written consent was obtained from all participants (17).

Bisphenol and phthalate concentrations were measured in first trimester urine samples among a subgroup of 1396 women with singleton pregnancies whose children also participated in postnatal studies when the children were age 5 years. Of these, 519 participants were excluded because of missing data on TTP and whether they had used infertility treatment, yielding an analytic sample of 877 women.

Urinary bisphenol and phthalate measurements

Bisphenol and phthalate concentrations were measured in a spot urine sample obtained from each participant during the first trimester visit (median gestational age, 12.9 weeks, interquartile range, 12.1 to 14.4 weeks). All urine samples were collected between February 2004 and July 2005. Details on collection, transportation, and analysis methodology are provided elsewhere (18).

For analysis, we grouped bisphenols together and grouped phthalate metabolites according to their molecular weight categories and parent compounds. Phthalate metabolites and bisphenols were only included in groupings if <80% of the sample concentrations were below the limit of detection (LOD). Additionally, individual bisphenols were analyzed separately if <50% of the sample concentrations were below the LOD. We calculated the weighted molar sums for groups representing total bisphenols, low-molecular-weight (LMW) phthalates, high-molecular-weight (HMW) phthalates, and two subgroups of HMW phthalates, total DEHP and di-n-octylphthalate (DNOP) metabolites, using the formula: [(concentration in ng/mL) × (1/molecular weight) × (1/10⁻³)] + [(concentration in ng/mL) × (1/molecular weight) × (1/10⁻³)] and so on. PA was analyzed separately as a proxy for total phthalate exposure. For bisphenol and phthalate concentrations <LOD, we substituted LOD/√2 (19). Table 1 shows the concentrations and detection rates of all of the bisphenols and phthalate metabolites we analyzed, both individually and in groups.

TTP and infertility treatment

Information on TTP (months) and whether pregnancy was the result of any kind of infertility treatment (ovulation induction, surgery, artificial insemination, or in vitro fertilization) was obtained from the first questionnaire at enrollment. When we modeled TTP as a continuous outcome, we assigned 12 months as the TTP for all participants who received infertility treatment, reflecting the clinical definition of infertility (Zegers-Hochschild 2009), regardless of when they commenced intervention, to avoid bias.

Covariates

Potential covariates were identified via causal diagram and a review of the literature. Information on maternal and paternal age (years), parity (nulliparous/multiparous), maternal educational level (low/high), ethnicity (Dutch or European/non-European), prepregnancy weight (kilograms), and preconception folic acid supplementation (Y/N) was obtained from the first questionnaire at enrollment. Maternal height (centimeters) was measured at enrollment without shoes or heavy clothing. Paternal weight and height were measured once during pregnancy. Maternal prepregnancy and paternal body mass index (BMI) were calculated according to the standard formula (kilograms/square meter). Information on maternal smoking and alcohol consumption (Y/N) was assessed by questionnaires in each trimester.

Statistical analysis

Descriptive statistics were performed to assess participant characteristics. Depending on the distribution of the variables, Pearson or Spearman correlations were used to test for collinearity among potential covariates. Missing data for covariates were imputed using multiple imputation. Five imputed datasets were created and pooled for analyses. The percentage of missing values for any given covariate within the analytic sample was ≤10% except for folic acid supplementation (13.3%). For the main regression analyses, all bisphenol and phthalate urinary metabolite concentrations were natural log-transformed to reduce variability, and all models were adjusted for urinary creatinine concentration.

To examine associations of first trimester urinary bisphenol and phthalate concentrations with TTP modeled continuously, we used discrete Cox proportional hazard models. Participants who conceived by use of infertility treatment were censored at 12 months. Proportional hazard assumptions for covariates were checked using plots of Schoenfeld residuals for continuous variables and log-minus-log plots for categorical variables. The resulting fecundability ratios (FRs) represent the probability of becoming pregnant in a single menstrual cycle per log unit increase in bisphenol or phthalate metabolite/group (note: FR > 1 indicates shorter TTP, whereas FR < 1 indicates longer TTP). To select potential covariates for inclusion in the regression models, Cox proportional hazards models were manually fitted using log-likelihood ratio test–based backward selection. Nonlinearity of exposure variables in the Cox proportional hazards model was tested using quintiles.

Sensitivity analysis was performed excluding participants who used infertility treatment. Given the biological plausibility that folic acid may influence the mechanism by which bisphenol and phthalate compounds affect female reproductive function, we stratified on preconception folic acid supplement use and performed a test for interaction among participants with nonmissing values of folic acid supplementation. Interaction on the multiplicative scale was considered significant at P < 0.1.

For all models with statistically significant results, subanalyses of individual bisphenols or phthalate metabolites were performed to determine which compounds were driving the association.

Cox proportional hazard models were performed using R statistical software, version 3.3.2, for Windows (package survival and msm). All other analyses were performed using the Statistical Package of Social Sciences, version 21.0, for Windows (SPSS Inc., Chicago, IL).

Results

Participant characteristics

Table 1 shows first trimester urinary bisphenol and phthalate concentrations of participating women. Because of low detection rates, not all bisphenols and phthalates were included in subsequent analyses.

Maternal participants were mean age 31.2 years (standard deviation, 4.4) and had a median BMI of 22.7 kg/m² (interquartile range, 20.9, 25.2). The majority were Dutch or European (68.9%), nulliparous (62.7%), and nonsmokers (73.9%). The median TTP was 3 months (range, 1 to 85), and 52 pregnancies (5.9%) resulted from any form of infertility treatment (Table 2).

We noted strong correlations between maternal and paternal age (Pearson correlation coefficient 0.6) and weak correlations between maternal and paternal BMI (Spearman’s rho 0.2) (data not shown). To avoid multicollinearity, only maternal covariates were used in regression analyses.

Generation R participants who met the eligibility criteria for our study (enrollment <18 weeks’ gestation, information on TTP and infertility treatment use, singleton live born children, 5-year postpartum visit, n = 2835) had generally similar population characteristics to those who additionally had first trimester urine samples and were included in our analysis (n = 877) (Supplemental Table 1).

Associations with time to pregnancy

Schoenfeld residuals of all continuous covariates were evenly distributed using the Kaplan-Meier time scale, indicating that the proportional hazards assumption was met (Supplemental Table 2a). In log-minus-log plots, lines for parity crossed and had a P value of 0.055 for nonproportionality (Supplemental Table 2b). Therefore, all Cox proportional hazards models were stratified for parity. Although the P value for nonproportionality for folic acid supplement use was 0.058, proportionality was assumed because the lines did not cross.

None of the first trimester urinary bisphenol or phthalate groups or compounds was associated with fecundability (Table 3). Modeling quintiles of exposures did not reveal any nonlinear associations (data not shown). Sensitivity analysis, using only participants who conceived naturally, yielded similar fecundability ratios (Supplemental Table 3A).

Effect measure modification by preconception folic acid supplement use

Among women who did not use folic acid supplements preconceptionally, each log unit increase in total bisphenols and PA decreased fecundability [FR = 0.90 (95% CI, 0.81 to 1.00) and 0.88 (95% CI, 0.79 to 0.99), respectively]. Slightly weaker associations were detected for BPA and bisphenol S (BPS) (FR = 0.93; 95% CI, 0.86 to 1.01 and 0.94; 95% CI, 0.87 to 1.01, respectively). Correlations between first trimester BPA and BPS are weak (18). BPA and BPS were assessed in a simplified multipollutant model by adding both in the same model. This yielded exactly the same estimates. Among preconception folic acid supplement users, bisphenol and phthalate compounds were not associated with fecundability. Tests for interaction indicated effect measure modification by folic acid supplement use for associations of total bisphenols, BPA, BPS, PA, HMW phthalates, and DNOP metabolites with fecundability (P < 0.1). In all cases, fecundability was lower in women without preconception folic acid supplement use, indicating longer TTP (Table 4).

Subanalysis of bisphenol F did not reveal further associations (Supplemental Table 4). Subanalysis of individual HMW phthalate metabolites showed effect measure modification by folic acid supplement use for mono(2-ethyl-5-carboxypentyl)phthalate, mono[(2-carboxymethyl)hexyl]phthalate, mono(3-carboxypropyl)phthalate, and monobenzylphthalate in models that included interaction terms (P < 0.1). Modeling quintiles of bisphenol and phthalate compounds did not show any nonlinear associations (data not shown).

Discussion

In this population-based birth cohort study with recalled information on TTP, we found total bisphenols and PA in the first trimester of pregnancy to be associated with decreased fecundability among women without preconception folic acid supplementation. We found no associations of first trimester urinary concentrations of bisphenols and phthalates with fecundability in unstratified models. The addition of an interaction term suggested effect measure modification by folic acid use of the association of total bisphenols, BPA, BPS, PA, HMW phthalate metabolites, and DNOP metabolites with TTP. Among those with preconception folic acid supplementation, bisphenols and phthalates were not associated with TTP.

We did not find maternal BPA to be associated with TTP, in line with previous studies (3–5). Both animal and human studies, however, have suggested a role for BPA in the pathogenesis of several known causes of female infecundability, including endometriosis and polycystic ovarian syndrome [reviewed in Ziv-Gal and Flaws (20) and Huo et al. (21)].

Results from previous studies examining associations between phthalates and TTP have been inconsistent (3–5, 7, 8). Among prospective cohorts of couples discontinuing contraception, the North Carolina Early Pregnancy Study reported no associations between female urinary phthalates and TTP (4), the Danish First Pregnancy Planner Study reported female urinary monoethylphthalate to be associated with a longer TTP (7), whereas the Longitudinal Investigation of Fertility and the Environment study observed an association between female urinary mono(3-carboxypropyl)phthalate and shorter TTP (3). Among pregnancy studies that used recalled TTP, as we did, the Canadian Maternal-Infant Research on Environmental Chemicals study reported no associations between first trimester urinary phthalate concentrations and TTP (5), whereas the multisite European INUENDO (Biopersistent organochlorines in diet and human fertility) study reported a shorter TTP per log unit increase in maternal serum DEHP concentrations during pregnancy (8).

Our findings of effect measure modification of the associations of total bisphenols, BPA, BPS, PA, HMW phthalates and DNOP metabolites with TTP by folic acid supplementation, and of reduced FRs for those without preconception folic acid supplementation support our hypothesis that bisphenols and phthalates may influence fecundability by inducing changes in DNA methylation. This study reports this for phthalates and is in line with a recent study among women undergoing infertility treatment that reported that high urinary BPA concentrations were associated with lower probabilities of implantation, clinical pregnancy, and live birth among women who consumed <400 μg/d of dietary folate (13). Diet interactions might explain the null findings reported in previous studies. DNA methylation has been identified as a potential mechanism in the association of persistent organic pollutants and reduced fecundability. A recent study in European adult males observed a weak association between exposure to persistent organic pollutants and global DNA methylation in sperm (22). In rodents, polychlorinated biphenyls have been identified to impair endometrial receptivity and cause failure of embryo implantation by disturbing the methylation level of the implantation-associated gene Homeobox A10 (23). However, no studies have investigated the role of preconception folic acid supplementation in this association. Further studies, ideally with maternal preconception serum folate concentrations rather than retrospective self-reported supplement use, are needed to confirm these results.

Strengths and limitations

Strengths of this study include the relatively large sample size of 877 participants with information on the use of infertility treatment and TTP. Participants were recruited in early pregnancy and asked to recall TTP in the first questionnaire at enrollment. Although prospectively measured TTP would have been preferable, the only published validation study comparing prospectively measured TTP with TTP recalled during the first trimester of pregnancy reported perfect agreement between measures among 53% of their sample. Recall error was, on average, small (only 12% had a discrepancy of ≥2 months), with a median of 0 and a mean of −0.11 months, indicating reasonable validity (24).

Detailed information on a large number of potential confounding factors was available, although we lacked information on several potential covariates predicting TTP such as timing/frequency of sexual intercourse and medical history, which may have introduced residual confounding. Most covariates used for this analysis were self-reported, which may have led to misclassification and consequently underestimation of effects. In the stratified and interaction models, we adjusted for maternal age, education, and parity, which are all associated with self-reported preconception folic acid supplementation in this cohort (25), but the possibility of residual confounding by other sociodemographic factors remains.

This study assesses associations of BPA analogs such as BPS, which recent studies have shown to have adverse effects on the reproductive neuroendocrine system (26–29), and of PA, a proxy for total phthalate exposure, with TTP. Unfortunately, our exposure measures were based on a single spot urine sample in the first trimester of pregnancy and may not accurately reflect preconception levels. Both bisphenols and phthalates are reported to have short biological half-lives (<24 hours) (30, 31), although it has been suggested that a single urine sample for phthalate concentrations reasonably reflects exposure for up to 3 months (32). Studies investigating variability of bisphenols and phthalates before and during pregnancy are scarce. Variability has been reported to be biomarker specific, with reasonable correlations for BPA and DEHP metabolites and stronger correlations for LWM phthalate metabolites and monobenzylphthalate (33, 34). Cross-sectional studies are at risk for reversed causation. In the current study, the observed interaction with folic acid supplementation use negates potential reversed causation induced by preconception behavioral changes in women taking longer to conceive. The absence of paternal exposure measurements may have introduced residual confounding because several recent epidemiologic studies suggest that BPA and phthalates may diminish semen quality (35–39).

Our main analyses violated the assumption of noninformative censoring for Cox proportional hazards models, as those who were censored because they used infertility treatment likely had a lower probability of natural conception (40). To overcome this limitation, we performed a sensitivity analysis including only participants who conceived naturally and found comparable results (Supplemental Table 3) despite differences among those with TTP <12 months vs TTP ≥12 months in maternal age, daily dietary caloric intake, ethnicity, education, and folic acid supplementation (Supplemental Table 5).

Finally, although there was reasonable similarity between the subgroup of the Generation R cohort in which we performed these analyses and the entire sample (Supplemental Table 1), the response rate at baseline for the Generation R study was 61% (16). We therefore cannot rule out selection toward a relatively healthy population that may limit the generalizability of our results. As a population-based pregnancy cohort, all women in this study successfully conceived, with or without infertility treatment. Therefore, this study is less generalizable for the general population of couples discontinuing contraception, in which 15% to 30% of couples are estimated to suffer from unexplained subfertility (41).

Conclusion

Although we found no associations of first trimester bisphenol and phthalate urinary concentrations with fecundability in full-sample analyses, we detected evidence of effect measure modification by folic acid supplement use. Among women without preconception folic acid supplementation, increased bisphenol and phthalate concentrations were associated with reduced fecundability, suggesting that the mechanism by which these chemicals impair reproductive potential may involve changes in DNA methylation. Our findings add to the already substantial evidence of the benefits of preconception folic acid supplementation to reproductive health. Further studies are needed to replicate these findings and investigate potential mechanisms.

Abbreviations:

Abbreviations:
 
• BMI

  body mass index

 
• BPA

  bisphenol A

 
• BPS

  bisphenol S

 
• DEHP

  di-2-ethylhexylphthalate

 
• DNOP

  di-n-octylphthalate

 
• FR

  fecundability ratio

 
• HMW

  high molecular weight

 
• LOD

  limit of detection

 
• LMW

  low molecular weight

 
• PA

  phthalic acid

 
• TTP

  time to pregnancy

Acknowledgments

We gratefully acknowledge the contribution of children and parents, general practitioners, hospitals, midwives and pharmacies in Rotterdam. The Generation R Study is conducted by the Erasmus Medical Center in close collaborations with the School of Law and Faculty of Social Sciences of the Erasmus University Rotterdam, the Municipal Health Service Rotterdam area, Rotterdam, the Rotterdam Homecare Foundation, Rotterdam and the Stichting Trombosedienst and Artsenlaboratorium Rijnmond (STAR-MDC), Rotterdam.

Financial Support: This work was supported by National Institutes of Health Grant R01 ES022972 to L.T., Netherlands Organization for Health Research and Development Grant VIDI 016.136.361 to V.W.V.J., and the European Research Council Grant number ERC-2014-CoG-64916 to V.W.V.J. The general design of the Generation R Study is made possible by financial support from the Erasmus MC, University Medical Center, Rotterdam, Netherlands; the Organization for Health Research and Development (ZonMw); and the Ministry of Health, Welfare and Sport.

Disclosure Summary: The authors have nothing to disclose.

References