Early Prenatal Phthalate Exposure, Sex Steroid Hormones, and Birth Outcomes

Abstract

Context:

Adequate sex steroid hormone concentrations are essential for normal fetal genital development in early pregnancy. Our previous study demonstrated an inverse relationship between third-trimester di-2-ethyl hexyl phthalate exposure and total testosterone (TT) concentrations. Here, we examine early-pregnancy phthalates, sex steroid hormone concentrations, and newborn reproductive outcomes.

Design:

We examined associations between urinary phthalate metabolite concentrations in early pregnancy and serum free testosterone (FT), TT, estrone (E1), and estradiol (E2) in 591 woman/infant dyads in The Infant Development and Environment Study; we also examined relationships between hormones and newborn genital outcomes using multiple regression models with covariate adjustment.

Results:

E1 and E2 concentrations were 15% to 30% higher in relation to 1-unit increases in log monoisobutyl phthalate (MiBP), mono-2-ethyl hexyl phthalate, and mono-2-ethyl-5-oxy-hexyl phthalate concentrations, and E2 was 15% higher in relation to increased log monobenzyl phthalate (MBzP). FT concentrations were 12% lower in relation to 1-unit increases in log mono(carboxynonyl) phthalate (MCNP) and mono-2-ethyl-5-carboxypentyl phthalate concentrations. Higher maternal FT was associated with a 25% lower prevalence of having a male genital abnormality at birth.

Conclusions:

The positive relationships between MiBP, MBzP, and DEHP metabolites and E1/E2 are unique and suggest a positive estrogenic effect in early pregnancy. The inverse relationship between MCNP and DEHP metabolites and serum FT supports previous work examining phthalate/testosterone relationships later in pregnancy. Higher FT in relation to a 25% lower prevalence of male genital abnormalities confirms the importance of testosterone in early fetal development.

We found significant relationships between prenatal phthalate exposure and sex steroid hormone concentrations. Free testosterone was associated with 25% lower prevalence of male genital abnormalities.

The primary programming period for human genital development is the first 5 to 18 weeks of pregnancy when genitourinary structures differentiate under precise genetic and hormonal signaling controls (1–3). Testosterone produced by the fetal testes drives normal male genital tract development, and its excess in female fetuses can lead to virilization (3). In the absence of testicular development and androgen production, female external genital development occurs, with ovarian development beginning early in gestation (2–4). Sex steroid hormone production occurs in the maternal, placental, and fetal compartments, which work in concert to achieve hormonal homeostasis, and the influence of each compartment changes in different gestational periods (1, 2, 5). The placental compartment does not become fully functional in producing steroid hormones until approximately 12 weeks, and even thereafter, it relies on maternal and fetal precursors for production (1, 2). Therefore, sex steroid concentrations in early pregnancy reflects a combined fetoplacental-maternal unit, and as such, maternal circulating sex steroids can reflect fetal circulation and vice versa (1). As pregnancy with a male fetus progresses, fetal testosterone production primarily occurs via testicular production, with the maximum concentrations of androgen secreting Leydig cells at 15 to 18 weeks of gestation (2, 6). Estrogen production, meanwhile, is maternal in origin in early pregnancy and then is under direct control of the fetal placenta, with concentrations starting to rise in week 8 and increasing steadily throughout gestation through aromatization of androgens derived from the fetal adrenal gland (1, 2, 7, 8). When typical fetoplacental-maternal endocrine production is disrupted, reproductive development may be altered. For instance, genetic conditions such as congenital adrenal hyperplasia lead to increased androgen concentrations and subsequent female virilization (9). Genital abnormalities have been reported among babies born to women with increased phthalate exposures (10), a type of endocrine-disrupting chemical (EDC), as well as those with increased estrogenic activity thought to be due to EDCs (11).

In modern populations, exposure to some EDCs is nearly ubiquitous. For example, phthalates, a class of synthetic chemicals in widespread commercial use, are found in the majority of the human population (12). In rodent models, some phthalate esters, including di-2-ethyl hexyl phthalate (DEHP), dibutyl phthalate (DBP), and benzyl butyl phthalate (BBzP), downregulate fetal testosterone production (13), resulting in alterations in reproductive tract development (14–18). This has been described as the “phthalate syndrome” in rats and demonstrates that in utero exposure to antiandrogenic phthalates during the male programming window exert a direct testicular toxic effect leading to reduced testosterone production and conditions such as hypospadias and cryptorchidism (18–20). Diisononyl phthalate is a known replacement for DEHP and has also been found to have antiandrogenic properties in utero in rodent models (18, 21). In humans, phthalate exposure during pregnancy is associated with a number of infant and child developmental endpoints that are androgen mediated, including reduced anogenital distance (AGD) (22, 23), a measure of fetal androgen exposure, and changes in sex-specific behavior (24, 25). Few human studies have examined phthalate exposure and androgen concentrations during gestation; prenatal exposure to DEHP has been negatively associated with free testosterone (FT) and total testosterone (TT) concentrations in pregnancy and in umbilical cord blood (26, 27). Unanswered questions remain, such as timing of exposure in relation to effects and the degree to which maternal sex steroid hormone concentration reflects that of the fetus.

Similarly, the relationship between phthalate exposures and estrogens during fetal development has not been studied extensively. The few studies published to date suggest conflicting results and focus on estrogen production during adult life. DEHP metabolites have been negatively associated with estradiol concentrations in adult female rodents and in ovarian granulosa cells via decreased aromatase transcription (28–32), but this relationship has not been observed in human studies (27, 33, 34). In contrast, DBP and BBzP exposures have been associated with increased estrogenicity in breast cancer cell lines and in vivo in offspring exposure in utero or in adult female rodent models (35–37). None of these studies examine early-pregnancy exposures and placental estrogen production during gestation, an important period for fetal genital development. In animal models, lack of estrogen during fetal development results in a normal external phenotype but with abnormal internal ovarian/uterine development resulting in impaired fertility (38, 39).

In an earlier study, we examined phthalate exposures and serum sex steroid hormones, measured in late pregnancy. In that study, DEHP and DBP metabolites were inversely associated with FT and TT concentrations but were not associated with estradiol (E2) concentrations (27). The current study examines the relationship between phthalates and hormone concentrations during the genital programming window in early pregnancy when sex steroid concentrations are most important for normal genital development. We also examine the relationships between early-pregnancy hormone concentrations and newborn AGD and genital birth abnormalities.

Methods

Study participants

Details of participant recruitment in The Infant Development and the Environment Study (TIDES) cohort have been published elsewhere (40) and are summarized here briefly. TIDES recruited mothers in their first trimester (T1) at four clinical centers (University of California, San Francisco, University of Minnesota, University of Rochester Medical Center, and Seattle Children’s Hospital/University of Washington) from 2010 to 2012. Eligibility criteria included: <13 weeks pregnant, singleton pregnancy, English speaking, age ≥18, no serious threat to the pregnancy, and plans to deliver at a study hospital. Interested women who met eligibility criteria signed an informed consent for themselves and their infant and were then enrolled in the study. All study centers received human subjects’ approval prior to the start of recruitment. All participants completed a questionnaire and gave blood and urine samples on the same day, and the majority of samples were collected in the late first trimester or early second trimester. Here, we report on 591 women who completed the first trimester questionnaire and had urinary phthalate measurements and serum hormones measured in early pregnancy samples.

We collected medical birth record data and performed an extensive physical exam shortly after birth for which all examiners completed a 2-day training before beginning the study (41). A pediatric urologist trained examiners to inspect and identify any minor or major genital anatomic abnormalities, including hypospadias, cryptorchidism, hydrocele, fused labia, clitoromegaly, and intersex genitalia. If there was a question about an anatomic finding during an infant exam, a study physician was consulted. Exams included visual inspection, palpation, and measurement of the genital area. Two measures of AGD [anoscrotal distance (AGD_AS) and anopenile distance (AGD_AP)] were taken for each male infant and two measures [anofourchette distance (AGD_AF) and anoclitoral distance (AGD_AC)] were taken for each female infant. Further details on AGD measurement methods and the rigorous quality control program used for all measurements and exams are described elsewhere (41). Data on AGD was originally reported in a previous publication and is also given in this study because of relevance to the study questions (42).

Serum hormone measurements

All serum hormone assays were performed at the Endocrine and Metabolic Research Laboratory at Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center using assays that have been validated and reported. Serum TT was measured by liquid chromatography (LC) tandem mass spectrometry (MS/MS) as previously described, with minor modification to shorten the runtime and system parameters (43). LC-MS/MS runs were performed with a Shimadzu high-performance liquid chromatography (HPLC) system (Columbia, MD) attached to an Applied Biosystems API5500 LC-MS/MS (Foster City, CA) equipped with a turbo-ion-spray source using positive mode. The calibration standards showed a linear response from 2.0 ng/dL (0.069 nmol/L) to 2000 ng/dL (69.3 nmol/L) for testosterone. The within-run and between-run precision was less than 5%, and the accuracy of samples spiked with the steroids was between 100% and 113% for testosterone. The lower limit of quantification for testosterone was 2 ng/dL (0.069 nmol/L). FT represents the unbound, biologically available testosterone and was measured by equilibrium dialysis using labeled testosterone as described previously (44).

Serum E1 and E2 concentrations for all mothers were measured with a validated LC-MS/MS method. The estrogens were separated on a column with a gradient profile started at 63% methanol using a Shimadzu HPLC system and a triple quadrupole mass spectrometer (Applied Biosystems API5000 LC-MS/MS) operated in the negative mode using multiple-reaction monitoring. The calibration curve was linear over a concentration range of 2 to 2000 pg/mL for both E1 and E2. The lower limit of quantification was 2.0 pg/mL for both E1 and E2. The within-run and between-run precision (% coefficient of variation) are from 2.6% to 5.6% and from 3.9% to 4.6% for E1, from 4.3% to 5.0% and from 4.6% to 5.2% for E2, and from 4.1% to 5.7%.

Phthalate metabolite measurements

First trimester urine was collected in sterile and phthalate-free specimen cups during initial recruitment visits, transferred to cryovials, and stored in freezers at <–80°C. We measured specific gravity using a handheld refractometer at the time of urine collection, which was calibrated with deionized water before each measurement. Phthalate metabolite concentrations were analyzed at two different sites. Samples from the mothers of girls were analyzed at the Environmental Health Laboratory at UW per a modified version of the Centers for Disease Control and Prevention (CDC) method 6306.03. Glucuronidated phthalate monoesters underwent enzymatic deconjugation, followed by online solid-phase extraction coupled with reversed HPLC-electrospray ionization-MS/MS to quantify the simple monoesters in urine (45). Samples from the mothers of boys were analyzed by the Division of Laboratory Sciences, National Center for Environmental Health, CDC. At the CDC, urine samples were analyzed using a modified method described in Silva et al. (46) that involved the enzymatic deconjugation of the phthalate metabolites from their glucuronidated form, automated online solid-phase extraction, separation with HPLC and detection by isotope dilution MS/MS. Process and instrument blanks as well as field blanks were run in each laboratory for quality assurance of analytical and sampling procedures. For the field blank collection, deionized water was purchased, poured into phthalate-free urine cups, and transferred with disposable pipettes to 5-mL cryovials. These blanks were then interspersed with subject samples to be shipped to laboratories.

The limit of detection (LOD) of metabolites was between 0.2 and 2.0 ng/mL for the UW samples and 0.2 and 0.6 ng/mL for the CDC samples, and all blanks were below the LOD. For participant concentrations below the LOD, a value equal to each sample’s specific LOD divided by the square root of 2 was used (47). All urinary phthalate metabolite levels were adjusted for dilution using specific gravity measurements and logarithmically transformed to normalize distributions. To calculate the molar sum of the DEHP metabolites, mono-2-ethylhexyl phthalate (MEHP), mono-2-ethyl-5-hydroxy-hexyl phthalate (MEHHP), mono-2-ethyl-5-oxy-hexyl phthalate (MEOHP), and mono-2-ethyl-5-carboxypentyl phthalate (MECPP) were divided by their molecular weights and added: ∑DEHP metabolites = ([(MEHP/278) + (MEHHP/294) + (MEOHP/292) + (MECPP/308)] × 1000). Other metabolites measured included monoethyl phthalate (MEP), monoisobutyl phthalate (MiBP), monobenzyl phthalate (MBzP), monobutyl phthalate (MBP), mono(carboxynonyl) phthalate (MCNP), and mono(carboxy-isooctyl) phthalate (MCOP).

Statistical analysis

We examined the range of all exposure and outcome variables under consideration, including concentration and distribution of maternal serum hormone concentrations and urinary phthalate concentrations. The distribution of both urinary phthalate concentrations and maternal hormone concentrations was skewed and therefore log transformed. We excluded one woman with an outlier testosterone concentration of 464 ng/dL because this was deemed clinically not feasible by our reproductive endocrine and infertility clinical specialist and 34 others because they reported having polycystic ovarian syndrome, a hyperandrogenic condition that could confound results.

We examined covariates chosen a priori (maternal age, race/ethnicity, education, income, study center, prepregnancy body mass index, parity, smoking during pregnancy, and gestational age at blood draw) for estimation of association with log hormone concentrations. We chose these factors based on past studies examining prenatal hormone concentrations as well as studies examining phthalate exposures. We explored whether maternal hormone levels differed by infant sex and whether sex modified the relationship between phthalate exposure and hormone concentrations using an interaction term for infant sex, and results from both of these analyses were insignificant. Therefore, we include infant sex as a covariate only. We also explored time of day of urine collection as a precision variable given that it can be related to phthalate variability within the day (48) but did not observe a relationship and therefore did not include it in final models. We used multiple linear regression to explore the associations between log-transformed prenatal urinary phthalate metabolite concentrations and concurrent log-transformed maternal serum hormone concentrations in individual models. We used multiple linear regression to explore the associations between log-transformed prenatal hormone concentrations and reproductive outcomes, including frank genital anomalies and AGD_AP, AGD_AS, AGD_AC, and AGD_AF, in each sex and adjusted for weight for length z score, infant age at exam, study center, maternal age, maternal race/ethnicity, and gestational age at serum draw. We present final regression results as percentage change in hormone concentration through antilog back transformation.

Results

Pregnant women were primarily white (69%) and between ages 20 and 40 (95%) and had a college degree (76%). More than 99% of women were within the first 20 weeks of gestation (the genital programming window is 5 to 18 weeks) at the time of blood and urine collection (Table 1). We observed abnormalities in 43 newborn males and no frank abnormalities in females; cryptorchidism, hypospadias, and hydrocele comprised the majority of the 43 cases (Table 1). Mean male AGD_AP was 49.7 mm [standard deviation (SD) 5.9] and female AGD_AC was 36.7 (SD 3.9). Further explanation of the AGD measurements and genital anomalies is reported in Swan et al. (42) and Sathyanarayana et al. (10).

Table 1.

Demographic Characteristics of 591 Mother-Infant Dyads Within the TIDES Study With Maternal Urinary Phthalate and Serum Hormone Measurements

Characteristic	n (%)
Study center
UCSF	123 (21)
UMN	175 (30)
URMC	155 (27)
UW	122 (21)
Maternal age (years)
<20	15 (3)
20 to <30	211 (36)
30 to <40	347 (59)
≥40	17 (3)
First trimester body mass index (kg/m2)
<18.5 to <24.9	322 (55)
25 to <29.9	136 (23)
>30	128 (22)
Race/ethnicity
Non-Hispanic white	406 (69)
Non-Hispanic black	63 (11)
Non-Hispanic Asian	28 (5)
Hispanic	52 (9)
Other/mixed	36 (6)
Highest education attended
Up through high school	77 (13)
Some college	65 (11)
College/postgraduate	443 (76)
Any smoking during pregnancy
Yes	42 (7)
No	549 (93)
Parity
Nulliparous	224 (39)
Parous	346 (61)
Income
<$25,000	130 (23)
$25,000–$74,999	158 (28)
≥$75,000	283 (50)
Infant sex
Male	287 (49)
Female	304 (51)
Gestational age at blood draw (weeks)
≤12 weeks	352 (59.5)
>12 to 20 weeks	236 (39.9)
>20 weeks	3 (0.5)
Infant genital abnormality (male)a
Undescended testes	5 (1)
Hydrocele	30 (8)
Hypospadias	3 (1)
Other	4 (1)
Infant AGD measurement (mean [SD])
Male AGDAP	49.7 (5.9)
Male AGDAS	24.7 (4.6)
Female AGDAC	36.7 (3.9)
Female AGDAF	16.0 (3.2)

Abbreviations: UCSF, University of California, San Francisco; UMN, University of Minnesota; URMC, University of Rochester Medical Center; UW, University of Washington.

^aInfants may have more than one malformation. In addition, results for genital anomaly distribution are reported in Sathyanarayana et al. (10).

MEP was found in the highest concentrations followed by MCOP, a metabolite of a phthalate replacement chemical, diisononyl phthalate (Table 2). The mean concentrations for E1, E2, FT, and TT were 1017 pg/mL, 1768 pg/mL, 0.34 ng/dL, and 74 ng/dL, respectively (Table 3). Pearson correlations between E1 and E2 were >0.7 and were similarly high for FT and TT. Correlations between the estrogens and testosterone measures ranged from 0 to 0.25.

Table 2.

Specific Gravity–Adjusted Urinary Phthalate Concentrations During Early Pregnancy (µg/L, N = 591)

	% >LOD	25th Percentile	Median	75th Percentile	Mean (SD)
MBP	92	4.83	8.51	13.97	12.86 (25.09)
MBzP	88	2.06	4.09	8.56	8.13 (17.27)
MiBP	97	2.90	5.14	8.88	7.34 (7.41)
MEP	99	13.20	30.53	79.75	142.34 (640.3)
MCNP	96	1.45	2.18	4.35	6.05 (21.48)
MCOP	100	8.24	15.39	45.01	43.21 (70.55)
MEHP	66	1.38	2.56	4.35	5.14 (20.62)
MEHHP	97	4.35	7.66	12.66	16.76 (65.75)
MEOHP	97	3.22	5.44	8.46	10.92 (40.42)
MECPP	97	5.89	9.53	15.71	18.32 (48.41)
Sum DEHPa	N/A	15.73	25.67	39.70	51.14 (172.25)

^aUnits are in μm/L.

Table 3.

Sex Steroid Hormone Concentrations During Pregnancy (N = 591)

Hormone	% >LOD	25th Percentile	Median	75th Percentile	Mean (SD)
E1 (pg/mL)	100	354	725	1340	1016.58 (956.84)
E2 (pg/mL)	100	811	1440	2260	1767.89 (1355.32)
TT (ng/dL)	100	44.4	64.3	89.3	73.77 (43.6)
FT (ng/dL)	100	0.2	0.29	0.43	0.34 (0.22)

We observed 15% to 30% higher E2 concentrations in relation to a 1-unit increase in first-trimester log specific gravity–adjusted MBzP (14.7% [95% confidence interval (CI): 1.3%, 29.8%]), MiBP [30.0% (95% CI: 12.2%, 50.6%)], MEHP [25.0% (95% CI: 10.0%, 42.0%)], and MEOHP [14.7% (95% CI: 0.3%, 31.2%)] concentrations (Table 4). Similar results were observed for MEHP, MEOHP, and MiBP in relation to E1 (Table 4). We observed 12% lower FT concentrations in relation to a 1-unit increase in log specific gravity–adjusted MCNP [–12.4% (95% CI –22.5%, –1.03%)] and MECPP [–12.3% (95% CI –21.2%, –2.4%)] concentrations (Table 4). Of note, all DEHP metabolites were inversely related to FT and TT, but only MECPP reached statistical significance.

Table 4.

Percent Difference in Maternal Hormone Concentration in Relation to Log Specific Gravity–Adjusted Phthalate Exposure (N = 591)

Phthalate	E1	E2	TT	FT
MBP	8.84 (–8.55, 29.57)	7.87 (–6.57, 24.57)	1.62 (–8.5, 12.85)	1.81 (–8.61, 13.45)
MBzP	13.40 (–2.41, 31.76)	14.66a (1.30, 29.75)	3.25 (–5.68, 13.01)	0.05 (–8.88, 9.82)
MEP	3.42 (–7.17, 15.21)	−0.41 (–8.90, 8.87)	4.3 (–2.25, 11.33)	4.57 (–2.21, 11.81)
MiBP	30.38a (9.02, 55.96)	29.99a (12.20, 50.59)	7.40 (–3.66, 19.73)	8.62 (–2.88, 21.48)
MCNP	−14.18 (–29.45, 4.40)	−13.32 (–26.24, 1.86)	−6.76 (–17.07, 4.81)	−12.40a (–22.48, –1.03)
MCOP	−2.84 (–17.38, 14.24)	–3.95 (–15.96, 9.77)	−0.48 (–9.64, 9.57)	1.30 (–8.48, 12.10)
MEHP	28.23a (9.85, 49.69)	24.97a (10.00, 41.97)	0.00 (–9.01, 9.90)	−3.46 (–12.38, 6.37)
MEHHP	12.62 (–3.88, 31.95)	10.28 (–3.22, 25.69)	−3.57 (–12.36, 6.10)	−5.07 (–13.96, 4.74)
MEOHP	19.34a (1.39, 40.51)	14.71a (0.25, 31.22)	−4.79 (–13.72, 5.10)	−6.59 (–15.61, 3.37)
MECPP	4.95 (–11.71, 24.74)	3.18 (–10.55, 18.99)	−7.87 (–16.98, 2.21)	−12.30a (–21.17, –2.43)
Sum DEHP	14.10 (–4.32, 36.08)	11.02 (–4.02, 28.38)	−5.57 (–15.08, 5.03)	−9.18 (–18.57, –1.30)

Values adjusted for study center, maternal age, maternal race/ethnicity, gestational age at serum draw, first-trimester body mass index, and infant sex.

^aP < 0.05.

None of the early-pregnancy hormone concentrations were associated with AGD_AP, AGD_AS, AGD_AC, and AGD_AF in male or female infants (Fig. 1). We observed that higher maternal FT in early pregnancy was associated with a 25% lower prevalence of having a genital abnormality at birth in males [odds ratio, 0.10 (95% CI 0.01, 0.94)] (Table 5).

Figure 1.

Difference in male (n = 266) and female (n = 269) newborn AGD in relation to a 1-unit increase in serum estrone (E1), estradiol (E2), TT, and FT concentration in early pregnancy.

Table 5.

Odds of Having a Newborn Male Genital Anomaly in Relation to Early-Pregnancy Log Hormone Concentration (n = 266)

Hormone	Odds Ratio (95% CI)
E1	0.82 (0.23–2.96)
E2	1.68 (0.34–8.27)
TT	0.11 (0.01–1.05)
FT	0.10a (0.01–0.94)

Values adjusted for weight-for-length z score, infant age at exam, study center, maternal age, maternal race/ethnicity, and gestational age at serum draw.

^aP < 0.05.

Discussion

In our analysis of early-pregnancy phthalate exposure and hormone concentrations, our strongest findings showed a positive relationship between MiBP, MBzP, MEOHP, and MEHHP metabolites and E1 and E2 concentrations. These findings are unique and similar to animal and in vitro studies reporting phthalate estrogenicity; they contrast with those reporting suppression of ovarian granulosa cells by phthalate compounds (29, 36) and suggest that phthalates may have impacts on estrogen biosynthesis and/or metabolism. Our finding of lower FT in relation to MCNP and MECPP is consistent with published studies of antiandrogenic impacts of DEHP phthalate compounds, but relationships were not as strong as in our previous analysis nor were they sex specific (27). This discrepancy may be explained by the fact that, in the current study, the measured hormone concentrations were in early pregnancy, while, in the previous study, they were measured later in pregnancy when circulating hormone concentrations are higher and would more likely reflect the fetal milieu (27). We observed an association between higher FT in early pregnancy and 25% lower prevalence of male reproductive anomalies at birth. This finding aligns with literature documenting that adequate testosterone concentrations are needed for normal male reproductive genital development (1, 2).

The impact of phthalate exposure on estrogen steroidogenesis is not well understood, and in vivo/in vitro studies do not examine exposure/outcome relationships during gestation and/or do not focus on the placenta as a primary source of estrogen production. In human gestation, E1 and E2 are maternal in origin until about week 8, when placental production rises through aromatization of fetal adrenal androgens (1). BBzP and DBP exposure leads to increased estrogenic activity in vitro in ovarian granulosa cells, while DEHP has been associated with reduced estrogenic activity in breast cancer cell lines (31, 32, 36, 37). Although the mechanism underlying these observations is not known, these studies hypothesize that DEHP may stimulate peroxisome proliferator-activated receptors that suppress aromatization in the granulosa cell. Given that estrogens are produced by the mother (granulosa cells) in very early pregnancy and then the placenta starting in week 8 (2), effects may differ depending on tissue and physiologic state.

In vivo results differ with respect to effect of phthalates on estrogen production as well. DBP exposure during rat gestation was associated with significantly increased postnatal estradiol levels in offspring (35), but this study did not measure hormone levels during gestation. In two studies of adult female rodents, circulating estrogen concentrations were reduced following high-dose DEHP exposure, but these studies were not conducted in pregnant dams where hormone physiology changes significantly (28, 49). We observed higher estradiol concentrations in relation to increased MBzP, MEHP, MEOHP, and MiBP concentrations in early pregnancy. Future studies should examine the effects of phthalates on human gestational placental production of estrogens via aromatization. Our previous analysis did not find a relationship between late-pregnancy phthalate exposure and estrogen concentrations (27).

FT is unbound and bioavailable to peripheral target tissues, while TT reflects both bound [to sex hormone–binding globulin (SHBG)] and FT concentrations (1). The origin of prenatal circulating testosterone concentrations in humans is unclear but is thought to be maternal in origin for women carrying female fetuses and for male fetuses through fetal testicular Leydig cell production as pregnancy progresses (6, 50). In animal studies, DEHP, BBzP, and DBP exposure in pregnant dams led to lower intratesticular testosterone concentrations in male offspring via a direct testicular toxic effect on Leydig cells (14, 16, 51). One theory for this is that phthalates may affect the binding affinity for SHBG to testosterone, thereby leading to displacement and increased vulnerability to degradation (52). We observed lower FT in relation to all DEHP metabolites, but only the finding for MECPP was statistically significant (Table 4). Therefore, our study suggests that DEHP could be leading to changes in the ratio of free to bound testosterone as related to SHBG or potentially to reductions in early-pregnancy circulating maternal FT concentrations, but relationships for FT might have been much stronger in later pregnancy, when testosterone concentrations are higher with a higher contribution from the male fetus. DEHP exposure might also affect ovarian and/or adrenal production of testosterone, as there are no studies of this mechanism in the literature. Few studies have examined the antiandrogenicity of diisodecyl phthalate, the parent compound for MCNP, which we found to also be associated with lower testosterone concentrations. We did not observe a relationship between other known antiandrogenic phthalates, MBzP (metabolite of BBzP) or MBP (DBP), and testosterone concentrations in our primary regression analysis. However, our previous study found negative associations between DEHP and DBP metabolites and testosterone in the third trimester, suggesting that phthalates may affect fetal hormone production (27). This discrepancy may reflect differential effects based on timing of exposure/outcome assessment during gestation.

We observed a significantly lower prevalence of male infants with reproductive anomalies in mothers with higher FT concentrations (Table 5). Normal male virilization requires adequate testosterone concentrations, and reduced testosterone is known to be associated with abnormal male reproductive tract development (1, 2). Therefore, this finding confirms other findings in the scientific literature. We did not, however, find a relationship between prenatal sex steroid hormones and sex-specific AGDs. Although AGD is a known hormonally responsive outcome, we may not have observed a relationship because of variability in measurement of AGD outcome or because of residual confounding that we could not account for. We also did not observe a relationship between third-trimester hormone concentrations and sex-specific AGD outcomes in our previous analysis (27).

This study has several strengths. First, the timing of serum and urine collection occurred during the genital reproductive programming window early in pregnancy, which allowed for examination of hormone concentrations in relation to reproductive outcomes with appropriate chronology of exposure and outcome. Second, we were able to examine FT by equilibrium dialysis, which is the gold standard methodology and allows us to more precisely assess hormone concentrations in relation to exposures and outcomes. Third, birth outcome assessment was conducted by a physician-led standardized training with significant attention to intrarater and interrater measurement variability (41).

One limitation of this study is its reliance on a single spot urine sample in early pregnancy, given that phthalate concentrations can change substantially with time of day and depending on what the subject has eaten recently (53). One way to address this limitation in multiple regression models is by examining time of day of urine collection to decrease the likelihood of exposure misclassification. In our analysis, time of day did not change point estimates or CIs appreciably, likely because most women were sampled primarily in the middle of the day, when urinary concentrations have not reached their peak. Recent studies suggest that a spot sample can be representative of a 3-month period of exposure in adult women (54, 55). Several phthalates are highly correlated with one another. For example, MEHP, MEOHP, MEHHP, and MECPP are all metabolites of DEHP and have >0.7 correlation. Therefore, the issue of multiple comparisons correction does not apply here, but future analyses should consider looking at mixtures of phthalates in the same model. We do not have multiple hormone measurements within the same women, which would allow for longitudinal analysis to examine change in concentrations. We also did not have SHBG measurements, which would have been helpful in deciphering potential mechanistic pathways. We were unable to directly assess fetal hormone concentrations in this study, and existing methodology to do this (amniotic fluid collection) would not be ethically feasible in the general population for research purposes.

Conclusion

Understanding mechanistic relationships between EDCs and reproductive outcomes is necessary to elucidate how these chemicals may impact endocrine processes. The positive relationships between MiBP, MBzP, and DEHP metabolites and E1/E2 are unique and suggest these phthalates may have a positive estrogenic effect in early pregnancy. We observed that DEHP metabolites and MCNP are inversely related to serum FT concentrations, and these results were similar but less robust than our previous analysis of third-trimester phthalate/hormone relationships. Higher FT in relation to a 25% lower prevalence of male genital abnormalities confirms the importance of testosterone in early fetal development. Because fetal production of sex steroid hormones increases throughout pregnancy, our measured serum hormone concentrations might reflect the maternal compartment to a greater degree than the fetal compartment. Longitudinal studies of exposures in relation to serum hormone concentrations and birth outcomes are needed to confirm these findings and explore impacts on subsequent health outcomes.