Exposure to Phthalates and Phenols during Pregnancy and Offspring Size at Birth

1Diesters of phthalic acid (phthalates) and phenols are found in many consumer products. Low molecular weight (MW) phthalates (MW < 250 g/mol) are used in personal care products (perfumes, cosmetics) or as coating for pharmaceutical products. High MW phthalates (MW > 250 g/mol) tend to be used in polyvinylchloride floor and wall covering, food packaging, and medical devices. Phenols are used in food packaging [bisphenol A, (BPA)], polycarbonates (BPA), cosmetics (parabens), soap [triclosan (TCS)], and sunscreen [benzophenone-3 (BP3)]. Precursors of dichlorophenols (DCPs) are used as intermediates in the production of several herbicides or insecticides.

2Widespread exposure to phthalates and phenols has been documented for pregnant women in several industrialized countries. Some of these compounds can cross the placenta in humans, and phthalates have been detected in cord blood, amniotic fluid, and meconium.

3Little is known about the consequences of prenatal exposures to phthalates and phenols on fetal growth. An American cohort study of 404 mother–infant pairs reported an inverse association for 2,5-DCP maternal urinary concentrations and a positive association for BP3 urinary concentrations with birth weight in male but not in female newborns.

4Our aim was to study the relationships between prenatal exposures to phthalates and phenols and weight, length, and head circumference at birth among male newborns.

5Study population. We conducted a case–control study of male malformations of the genitalia nested in the EDEN (Etude des Déterminants pré et post natals du développement et de la santé de l’Enfant) and PELAGIE mother–child cohorts. These cohorts are described elsewhere. Briefly, the EDEN cohort consists of 2,002 pregnant women recruited before the end of the 28th gestational week from April 2003 through March 2006 in the obstetrical departments of the University Hospitals of Nancy and Poitiers, France. The PELAGIE cohort consists of 3,421 pregnant women enrolled before 19 weeks of gestation from April 2002 through February 2006 in three districts of Brittany: Ille et Vilaine, Finistère, and Côtes d’Armor, France. The present study includes all of the male newborns with undescended testis or hypospadias (identified at birth by pediatricians; n = 48 in EDEN and 24 in PELAGIE). In addition, three male newborns without congenital malformation of the genitalia (controls) were matched to each case by recruitment center, date of recruitment (± 6 months), day of week (weekend yes/no), and gestational week when the maternal urine sample was collected, for a total of 288 mother–newborn pairs (72 cases and 216 controls). Participants provided informed consent for data and biological sample collection for themselves and their offspring. These cohorts received the approvals of the appropriate ethical committees. The involvement of the Centers for Disease Control and Prevention (CDC) laboratory was limited and determined not to constitute engagement in human subjects research.

6Outcomes. We extracted from hospital maternity records weight, length, and head circumference assessed at birth.

7Exposure assessment. Urine was collected between 6 and 19 gestational weeks in the PELAGIE cohort and between 24 and 30 gestational weeks in the EDEN cohort. Assessment of biomarker and creatinine concentrations in maternal urine was done by the National Center for Environmental Health laboratory at the CDC in Atlanta, Georgia, USA. We measured the urinary concentrations of 11 phthalate metabolites in samples from all mothers (n = 287; 72 cases and 215 controls because the vial of one control broke during transport to the CDC laboratory). Nine phenols were measured in urine samples collected from mothers in the EDEN cohort only (n = 191; 48 cases and 143 controls) (phenols could not be measured in the PELAGIE cohort because a preservative added to the samples interferes with the assay used).

8Molar concentrations of four metabolites of di(2-ethylhexyl) phthalate (DEHP) [mono(2-ethylhexyl) phthalate (MEHP); mono(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP); mono(2-ethyl-5-oxohexyl) phthalate (MEOHP); and mono(2-ethyl-5-carboxypentyl) phthalate (MECPP)] were summed as total DEHP. Concentrations of phthalate metabolites of MW > 250 g/mol and of MW < 250 g/mol were summed as total of high MW phthalates (ΣHMW) and of low MW phthalates (ΣLMW), respectively. Total paraben concentration (ΣPB) was calculated by summing methyl paraben (MP), ethyl paraben (EP), propyl paraben (PP), and butyl paraben (BP) molar concentrations. Monoethyl phthalate (MEP) and monobenzyl phthalate (MBzP) concentrations have been corrected [multiplied by 0.66 (MEP) and 0.72 (MBzP)] because the analytical standards used were of inadequate purity (Calafat AM, personal communication).

9Statistical analyses. We replaced concentrations below the limit of detection (LOD) by LOD/√–2. We corrected the overrepresentation of congenital abnormalities, induced by the case–control design, using a reweighting approach. We used center-specific weights corresponding to the inverse of the inclusion probability of controls, so as to give cases and controls the same relative weight as in the original cohorts. Sensitivity analyses restricted to controls were also conducted.

10To limit the impact of between-subject variations in urine sampling conditions, we standardized biomarker concentrations using a two-step standardization method based on regression residuals. First, we estimated associations between ln-transformed biomarker concentrations and sampling conditions [hour of sampling (for EDEN only), gestational age at collection, duration of storage at room temperature before freezing, and season and day of sampling] using separate linear regression models for each biomarker adjusted for maternal age, body mass index (BMI) before pregnancy, parity, year of sampling, education, current occupation, active smoking, and center. Next, we used these data to estimate biomarker concentrations that would have been observed if all samples had been collected under the same conditions (e.g., 0730 hours for hour of sampling). Unless otherwise specified, all concentrations are the standardized values; we also report associations with biomarker concentrations not standardized for sampling conditions.

11Associations between each standardized maternal urinary biomarker concentration and birth weight, birth length, or head circumference were estimated using separate weighted linear regressions. Biomarker concentrations (micrograms per liter) were ln-transformed or coded in tertiles. We performed tests of heterogeneity in outcome value across exposure tertiles; p-values of trend tests were estimated using categorical variables whose values corresponded to the tertile-specific median biomarker levels. Situations in which biomarkers exhibited heterogeneity in outcome across tertiles (low p-value, heterogeneity test) but with little support for a trend (high p-value, trend test) were considered as suggestive of a nonmonotonic association. [A monotonic trend does not reverse direction but may have flat segment.] We also estimated adjusted relationships between standardized biomarker concentrations and birth weight using restricted cubic splines. Adjustment factors were all variables possibly related to birth outcomes (based on a priori knowledge), including maternal prepregnancy weight (broken stick model with a knot at 60 kg), maternal height (continuous), maternal smoking (never, 1–5, ≥ 6 cigarettes per day), parity (0, 1, ≥ 2), education level (high school or less, up to 2 years after high school, ≥ 3 years after high school), gestational duration (linear and quadratic terms), and recruitment center. In addition, we adjusted for urinary creatinine concentration (continuous, nontransformed) as a marker of urine dilution. Gestational duration was estimated using the date of last menstrual period (LMP) or gestational duration assessed by the obstetrician if it differed from the LMP-based estimate by > 2 weeks. Adjustment for the obstetrician-estimated gestational age instead of the LMP-based gestational age did not modify results (not shown). Models for head circumference at birth were also adjusted for the mode of delivery (because passage through the birth canal may influence head circumference at birth).

12We performed sensitivity analyses excluding women with pregnancy-induced hypertension (n = 17) or gestational diabetes (n = 15). All analyses were performed using STATA/SE, version 11 (StataCorp, College Station, TX, USA).

13To draw our conclusion, we gave more weight to results in agreement with our a priori hypotheses, namely, an effect of BP3 and 2,5-DCP on male birth weight. Other associations (highlighted on the basis of their p-values, without relying on threshold p-value to define statistical significance) were considered as hypothesis-generating.

14Study population. The 287 women were 29 years old, on average; 17% smoked during the first trimester of pregnancy (Table 1). Average gestational age at delivery was 39.8 weeks, average birth weight was 3,393 g (5th–95th percentiles, 2,640–4,130 g) and seven newborns (2%) weighed < 2,500 g. Average birth length was 50 cm (5th–95th percentiles, 47–54 cm), and average head circumference at birth was 35 cm (5th–95th percentiles, 32–37 cm) (Table 1).

15We detected 8 of the 11 phthalate metabolites and 5 of the 9 phenols in at least 95% of the samples (Table 2).

16Phenols and birth outcomes. Birth weight decreased by 49 g [95% confidence interval (CI): –86, –13] in association with a 1-unit increase in ln-transformed 2,5-DCP concentration. After categorizing exposures in tertiles, boys in the highest exposure tertile were 152 g lighter, on average, compared with boys in the lowest tertile (95% CI: –299, –5) (Table 3). We observed a similar association between 2,4-DCP and birth weight [Table 3; see also Supplemental Material, Figure 1 (http://dx.doi.org/10.1289/ehp.1103634)], consistent with the high correlation between 2,4-DCP and 2,5-DCP (r = 0.95). 2,5-DCP ln-transformed concentration was also inversely associated with head circumference at birth (–0.1 cm, 95% CI: –0.2, 0.0).

17Each 1-unit increase in ln-transformed BP3 concentration was associated with an increase of 26 g in birth weight (95% CI: –2, 54) and of 0.1 cm in head circumference at birth (95% CI: 0.0, 0.2) (Table 3).

18For BPA, estimates suggested an inverse U-shape association: birth weight increased by 169 g (95% CI: 14, 324) in the second BPA concentration tertile and by 85 g (95% CI: –62, 233) in the third concentration tertile, compared with the first [Table 3; see also Supplemental Material, Figure 2 (http://dx.doi.org/10.1289/ehp.1103634) for associations based on a restricted cubic spline model]. BPA concentrations were positively associated with head circumference, which increased by 0.8 cm in the highest BPA concentration tertile compared with the lowest tertile (95% CI: 0.2, 1.3) (Table 3).

19For all other phenols, there was no evidence of associations with offspring measures at birth [all p for heterogeneity were > 0.24 (Table 3), and curves obtained using restricted cubic splines did not clearly support an association. See Supplemental Material, Figure 2 (http://dx.doi.org/10.1289/ehp.1103634)].

20Phthalates and birth outcomes. For phthalate metabolite concentrations and birth weight, all p for trend were > 0.14 (Table 4). There was some evidence of heterogeneity in mean birth weight across concentration tertiles for some phthalate metabolites: the lowest p-values for heterogeneity were observed for mono(3-carboxypropyl) phthalate (MCPP) and MECPP; given the high values of p for trend, results were suggestive of nonmonotonic associations (Table 4). For MECPP, but less so for MCPP, this nonmonotonic association was also supported by the restricted cubic spline analysis to some extent [see Supplemental Material, Figure 3 (http://dx.doi.org/10.1289/ehp.1103634)].

21Regarding other birth outcomes, the lowest p-values for heterogeneity were observed with MCPP, monocarboxy-isooctyl phthalate (MCOP), monocarboxy-isononyl phthalate (MCNP), MEHP, and ΣHMW phthalates for birth length, and with ΣLMW phthalates for head circumference at birth (Table 4).

22Sensitivity analyses. For phenols, associations with birth outcomes remained similar after exclusion of 48 cases of male malformations of the genitalia [See Supplemental Material, Table 1 (http://dx.doi.org/10.1289/ehp.1103634)]. Using biomarker concentrations not standardized for sampling conditions instead of concentrations standardized for sampling conditions did not markedly change associations between birth weight and, variously, 2,4-DCP, 2,5-DCP, BP3, or BPA. Similarly, associations between head circumference at birth and either BP3 or BPA remained similar (see Supplemental Material, Table 1).

23In analyses restricted to controls only or using concentrations not standardized for sampling conditions, findings concerning phthalates were consistent with those obtained in the main analysis, except for MECPP: The birth weight change observed in the second MECPP concentration tertile was –141 g (95% CI: –277, –5) in the whole study population, –79 g (95% CI: –220, 61) after exclusion of cases, and –51 g (95% CI: –187, 84) using nonstandardized biomarker concentrations [see Supplemental Material, Table 2]. These sensitivity analyses are difficult to interpret; restriction to controls decreased population size, and standardization sometimes induced strong variations in the distribution of biomarker concentrations and hence in the cutoff values of tertiles.

24Excluding women with pregnancy-induced hypertension or gestational diabetes did not modify our main results (data not shown).

25Within our study population of male newborns, maternal urinary concentrations of 2,4-DCP and 2,5-DCP were associated with a birth weight decrease, whereas urinary concentrations of BP3 were positively associated with weight and head circumference at birth. BPA urinary concentrations were positively associated with head circumference. There was no evidence of monotonic associations between phthalate metabolite concentrations and birth weight.

26Phenols and birth outcomes. In our study population, birth weight decreased by 152 g in the highest 2,5-DCP concentration tertile compared with the lowest tertile (95% CI: –299, –5). The only other human study addressing this issue reported a decrease by 210 g in the third 2,5-DCP concentration tertile compared with the first (95% CI: –348, –71). Concentrations of 2,5-DCP (and hence tertiles) were much higher in the study by Wolff et al. (median, 53 µg/L) than in ours (median standardized concentration, 6.4 µg/L). Wolff et al. also reported that boys were 0.3 cm shorter at birth per increase by 1 in ln(2,5-DCP) (95% CI: –0.6, –0.4). Our results did not clearly support such an association (0.0 cm per increase by 1 in ln(2,5-DCP); 95% CI: –0.2, 0.2), but birth length is not very accurately assessed, implying potentially strong measurement error. 2,5-DCP is a metabolite of 1,4-dichlorobenzene, which is used as chemical intermediate in the production of dyes and organic chemicals and found in mothballs and toilet-deodorizer blocks. Dichlorophenols may also be released from water treatments.

272,4-DCP was also associated with a birth weight decrease. It is a metabolite of 1,3-dicholorobenzene, a minor contaminant of 1,4-dichlorobenzene, which may explain the high correlation reported between concentrations of both DCPs. 2,4-DCP is also an environmental transformation intermediate of the antiseptic agent triclosan and of herbicides such as 2,4-dichlorophenoxyacetic acid and 2-(2,4-dichlorophenoxy) propionic acid. Our results concerning 2,4-DCP are difficult to compare with those of, who studied male and female newborns altogether for this compound.

28Taken together, these studies suggest an effect of dichlorophenols, or one of their precursors, on birth weight.

29BP3 urinary concentrations were positively associated with weight and head circumference at birth. Boys were 105 g heavier (95% CI: –40, 250) in the highest BP3 concentration tertile compared with the lowest. Similarly, noted a birth weight increase in male infants in the highest BP3 concentration tertile compared with the lowest (betas not reported); they did not report effect estimates for BP3 and head circumference in males. Exposure to BP3 likely results from use of consumer products as sunscreens or cosmetics.

30BPA urinary concentrations were positively associated with head circumference at birth. After categorizing exposures in tertiles, we observed an inverse U-shape association between BPA concentrations and birth weight. Such nonmonotonic dose–response curves between perinatal exposures to BPA and weight in early life have been reported in rodents. However, urinary BPA concentrations were relatively low in our study population, enhancing the analytical uncertainties and hence the potential for exposure misclassification, which may limit our ability to distinguish monotonic from nonmonotonic associations.

31Phthalates and birth outcomes. There was no strong evidence of monotonic association between phthalate metabolite concentrations and birth outcomes, except for the possible positive association between MCNP and birth length. Our analyses suggested nonmonotonic associations with birth weight and birth length for some phthalate metabolites. To our knowledge, such associations between phthalate metabolites and birth weight have not been reported previously in rodents or in humans and therefore should be considered cautiously.

32Study population. To correct the overrepresentation of cases induced by our case–control design, we weighted the observations in regression models. We also performed nonweighted analyses, restricted to controls, and our main results regarding birth weight remained similar to those based on the whole weighted study population.

33Exposure assessment. Sampling conditions, such as hour of urine sampling or storage duration before freezing, may influence the concentrations of several biomarkers. We used a two-step standardization method based on regression residuals to reduce undesirable variability in biomarker urinary concentrations due to sampling conditions. To our knowledge, it is the first time that such an approach has been applied to study associations between phthalate or phenol prenatal exposures and birth outcomes. We repeated our analyses using concentrations not standardized for sampling conditions, and associations between DCPs or BP3 and birth weight or between BP3 and BPA and head circumference were similar.

34Limitations. The assayed phenols and phthalates and their metabolites have relatively short half-lives in humans (typically ≤ 1 day), but accurate information is not available on half-lives in pregnant women, in whom metabolism may differ compared with nonpregnant women. We assessed exposures from the urinary concentrations at a single point during pregnancy; increasing the number of urine samples collected would have provided a more accurate estimate of the average exposure during the whole gestation. reported that for phthalates, the probability of correctly classifying a woman into a low-exposure group based on a single urine sample, if she truly had low exposure based on multiple measurements, was between 0.43 for monoethyl phthalate (MEP) and 0.95 for mono-isobutyl phthalate (MiBP). Concerning phenols, BPA concentrations also vary during pregnancy. Therefore, clearly there is exposure misclassification, whose amplitude differs according to the biologically relevant exposure window (if any) and the compound considered.

35We adjusted for many potential confounders, but residual confounding cannot be discarded. For example, specific metabolic disorders associated with both fetal growth and xenobiotic metabolism would constitute potential confounders. In our study, excluding women with pregnancy-induced hypertension or gestational diabetes did not alter associations with BPA and BP3 (not shown). Multiple comparisons are generally an issue in studies relying on several biomarkers. Although we did not formally correct for multiple comparisons, our choice to focus our conclusions on previously reported associations limits the risk of chance being a likely explanation for our main findings.

36In our study, there was no strong evidence of monotonic association between phthalate metabolite concentrations and birth outcomes. Urinary concentrations of 2,4-DCP and 2,5-DCP were negatively associated with birth weight, whereas BP3 concentrations were associated with a birth weight increase. Results concerning 2,5-DCP and BP3 are in agreement with another publication concerning male newborns from New York.