Study cohort and study design

Our subjects were from the TMICS, as previously described in detail [7]. In brief, pregnant women in their third trimester at nine hospitals located in northern, central, southern, and eastern Taiwan between October 25, 2012 and May 21, 2015 were invited to participate in the TMICS. Participant inclusion and exclusion criteria are shown in Fig 1. Self-administered questionnaires, anthropometric measures, and bio-samples (i.e., urine and peripheral blood) were obtained from participants during the baseline interview. Umbilical cord-blood samples were immediately collected after babies born and stored at –20°C until analysis. The study protocol was approved by the Institutional Review Board of the National Health Research Institute (NHRI) and our nine collaborating hospitals, the full names of the nine IRBs are listed in S1 Table. A written informed consent was obtained from each participant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart of participants enrolled in this study.

https://doi.org/10.1371/journal.pone.0297631.g001

Phthalate exposure

One-spot urine samples were collected during pregnancy at outpatient departments, delivered to the central laboratory at the NHRI, and certificated by an international laboratory comparison program (G-EQUAS 59) [7]. The concentrations (μg/L) of eleven urinary phthalate metabolites were determined using a solid-phase extraction method coupled with liquid chromatography-electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS, Agilent 1200/API 4000). The phthalate metabolites we measured included MEP, monomethyl phthalate (MMP), monoisobutyl phthalate (MiBP), monoisononyl phthalate (MiNP), mono-(2-ethyl-5-hydroxylhexyl) phthalate (MEHHP), mono-(2-ethyl-5-oxohexyl) phthalate (MEOHP), mono(2-ethyl-5-carboxypentyl) phthalate (MECPP), and mono(2-carboxymethylhexyl) phthalate (MCMHP) (Hsieh et al. 2019). In brief, we prepared 100 μL urine sample aliquots containing 20 μL ammonium acetate (1 M, pH 6.5), 10 μL β-glucuronidase, and a mixture of isotopic phthalate metabolite standards. The samples were incubated at 37°C for 1.5 h and treated with 270 μL solvent (5% acetonitrile and 0.1% formic acid) in glass screw-cap vials and mixed for quantitative LC-ESI-MS/MS after hydrolysis. For quality control, a blank sample’s measured concentration below two-fold the limit of detection (LOD) value (LOD for MMP, MEP, MnBP, MiBP, MBzP, MiNP, MEHP, MEHHP, and MEOHP was: 0.12, 0.06, 0.06, 0.12, 0.06, 0.6, 0.12, 0.06, 0.12 μg/L, respectively) for urinary metabolites was required. Each phthalate metabolite’s recovery rate ranged from 80% to 119%. Phthalate metabolite concentrations below the LOD were assigned a value of 1/2 LOD.

Considering the urinary dilution effect, urinary phthalate metabolite levels were divided by urinary creatinine levels measured using the ADVIA 1800 Clinical Chemistry System (Siemens, Erlangen, Germany) at the Union Clinical Laboratory (UCL; Taipei, Taiwan). Because MECPP and MCMHP tests were only available for limited study participants, we calculated the micromolar sum of MEHP, MEOHP, and MEHHP levels to estimate DEHP exposure (ΣDEHP).

Anogenital distances

Among the 564 babies, 263 were assessed for their AGDs by trained neonatologists who had been trained by watching an instructional video. AGD was measured as the scrotum-to-anus distance in male newborns and the distance from the center of the anus to the posterior convergence of the fourchette in female newborns. Babies were kept in a supine position, and their legs were held back in a frog-leg posture to identify the central position of the anus. The basic AGD-measurement procedures followed the methods developed by the Infant Development and Environment Study [19]. Each AGD was measured twice per baby using a dial caliper and the averaged distance was used for data analysis.

Sex steroid hormone concentrations

Clinical-biochemistry concentration were determined in radioimmunoassay (Diagnostic Products Corporation, Los Angeles, CA, USA) which was performed at UCL. The measures included five cord-blood sex steroid hormones, i.e., progesterone (ng/mL), E2 (pg/mL), frTT (ng/dL), sex hormone-binding globulin (SHBG, nmol/L), and follicle stimulating hormone (FSH, mIU/mL), as well as three maternal sex steroid hormones, i.e. progesterone (ng/mL), E2 (pg/mL), total testosterone (TT, ng/dL). The detection limit values of progesterone, FSH, E2, free TT, and SHBG were 30 ng/mL, 0.3 mIU/mL, 11.8 pg/mL, 0.04 ng/dL, and 0.35 nmol/L, respectively.

Statistical methods

We used geometric mean with geometric standard deviation and median with interquartile range to describe right-skewed variables and the Kruskal–Wallis H test to compare differences between groups. We categorized phthalate metabolite levels into low, median, and high based on tertile cut-off points for all phthalate metabolites but MBzP. Because of lower detection rate of MBzP, we assigned undetectable MBzP levels as low and dichotomized detectable values as median and high. Linear regression model was performed to estimate the changes of AGD corresponding to exposure levels of phthalate metabolites from low-to-middle and low-to-high levels. According to literature review, basic sociodemographic characteristics (household income, maternal education status), maternal physical activity, birth outcomes, gestational diabetes in pregnancy, and delivery records were considered potential confounders. Finally, we used the directed acyclic graph (DAG) analysis to determine the covariates which should be correctly adjusted in multivariable regression analyses (S1 Fig). Because body size and age largely explain variation in AGD, the newborn’s percentile of weight-for-length was included as a covariate [19] in predicting AGD. The norm of weight-for-length in the World Health Organization Growth Standard for babies 0–2 years of age [20] was applied. To explore the potential mediation, we additionally conducted general linear regression models to explore the associations of cord-blood and maternal sex hormone levels with newborn’s AGD in S5 Table, the sex hormone levels were in natural logarithm scale in these analyses.

All data analyses were performed using SAS software (version 9.4; SAS Inc., Cary, NC, USA).

Alterations of AGD associated with prenatal phthalate exposures

Decreased AGD is a sensitive marker detecting anti-androgen effect [14]. Our study reported that median levels of maternal MMP, MBzP, MEOHP, and ΣDEHP were associated with shortened AGD in male newborns, which was aligned with findings in past literatures. Previous studies have reported that a shorter AGD was related to prenatal DEHP and DMP exposure [10, 11, 17] and anopenile distance was negatively associated with in-utero MEP, MiBP, and MBP [21] in male newborns. The underlying mechanism has been related to phthalates’ anti-androgen effect such as testosterone inhibition in the animal studies. Male litters of rat dams gavaged with high doses of DEHP or butyl benzyl phthalate were more likely to have shorter AGD and reduced testis weights [22, 23], suppressed fetal testosterone [23], and malformation of reproductive systems [15, 22]. However, evidence gap exists in the female newborns due to limited amounts of studies reported in females [16]. Our study findings were aligned to the only available epidemiologic studies in the past, neither of which found a significant effect of maternal exposure of phthalates on the females’ AGD at birth [11, 24–26]. However, we observed borderline significant increased AGD and borderline significant decreased AGD in female babies with median maternal MiBP and MEHHP, respectively. There was also a meta-analysis showing an increased female’s AGD in female babies with higher maternal MBzP levels [21]. Although the effect of maternal phthalate exposure on female’s AGD at birth seems not to be so pronounced as that in males, more studies with sufficient sample size for baby girls are expected to further confirm the effect in females.

Associations between maternal urinary phthalate metabolites and cord-blood sex steroid hormones

Evidences have widely suggested that the reproductive toxicity of phthalates involves dysfunction of hypothalamic-pituitary-gonadal (HPG) axis [27]. Although the actual mechanism is still unclear, the reproductive toxicity of phthalates may be associated with its anti-androgenic property and seems to be more frequently reported in males than in females [17, 28].The upstream changes in HPG are associated with the abnormal expressions of peroxisome proliferator-activated receptors, aryl hydrocarbon receptors, and insulin receptors, which may alter the release of gonadal-releasing hormones and gonadotropins and induce cell damages [29], subsequently disrupt synthesis of sex steroid hormones and functions or morphologic changes of reproductive organ. The reproductive disrupting effect of phthalates can be even transgenerational to the unexposed offspring in animal studies, but it still needs further confirmed in humans [27, 30].

To the best of our knowledge, there were only two previous studies investigating the effect of maternal phthalate exposures on cord-blood sex steroid hormones [12, 13] and a cross-sectional study which measured phthalate metabolites and fetal hormone levels in amniotic fluid [31]. Among the two studies measured hormone levels in cord-blood, one study measured phthalate metabolites in maternal serum [12], another study was the only one measured phthalate metabolites in maternal urine which is now considered as the most appropriate matrix for phthalate exposure assessment [13]. Lin and colleagues analyzed the data of a birth cohort study conducted in 2001 in Taiwan, it was about ten years before the 2011 phthalate-tainted foodstuff episode uncovered in Taiwan. Our study enrolled study participants in 2012, at the time the control measures had been intervened. After that time, phthalate exposures in pregnant women have been reduced. The median levels of MEHP and DEHP in our study cohort (MEHP 5.0 μg/g creatinine, DEHP 0.13 μmol/g creatinine) was lower than that in Lin’s study (19.1 μg/g creatinine, DEHP 0.14 μmol/g creatinine, measured in 2000–2001) [13] and Kuo’s study (11.9 μg/g creatinine, DEHP 0.19 μmol/g creatinine, measured in 2009) in Taiwan [32] (S3 Fig).

Although maternal phthalate metabolites levels in our study were lower than that in previous studies in Taiwan, most phthalate metabolites were among the levels reported from Japan and US (S3 Fig). Under the exposure levels reported in our study, we found that maternal DEHP metabolites were in an inverse relationship with cord-blood progesterone and FSH in male and female newborns. FSH stimulate Sertoli cells proliferation and sperms production with testosterone (TT) in prepuberty [33]. Our findings were consistent with experimental study that showed reduced Sertoli cell number and inhibited cell proliferations in rodents gavaged with DEHP [34]. It was also partially aligned with Araki’ study findings in HCS. Araki et al. found an inverse association of serum MEHP with cord-blood progesterone and inhibin B (an FSH-regulated functional marker secreted by Sertoli cells in testes), but not with FSH levels, in male newborns [12]. Their findings may suggest a functional reduction of Sertoli cells associated with maternal DEHP levels in serum. Jensen et al. only examined effects of MECPP in male newborns since it was the most abundant metabolite of DEHP detected in amniotic fluid. In contrary to our findings, their study observed an increased progesterone associated with higher MECPP [31]. However, the study determined the sex hormones levels in amniotic fluid samples; their observations may not be directly compared with our study results. The anti-androgen effect of DEHP has been widely observed in cross-sectional study in adults [35, 36]. However, our study also observed that elevated maternal DEHP metabolites was associated with higher cord-blood frTT and lower SHBG in male newborns as well as lower frTT and higher SHBG in female newborns. The tendency toward decreased frTT of maternal phthalate exposures in female newborns has also been observed in Lin’s study, which showed an inverse association between maternal DEHP exposure and cord-blood frTT and frTT/E2 in female newborns, but they did not find that association in male newborns [13]. Jensen’ study findings in male fetuses were similar to ours, showing that amniotic fluid MECPP levels was positively associated with TT levels [31]. Araki’s study did not find any significant effect of maternal serum MEHP on cord-blood TT, SHBG and TT/SHBG levels both in males and females [12]. Similar to DEHP, maternal MMP was also related to higher cord-blood frTT and lower SHBG in male newborns but not in female newborns in our study. Araki’s study and Jensen’s study did not report the effect of MMP. The limited previous studies showed that maternal MMP was not associated with frTT, TT, or frTT/E2 levels in cord-blood [13] and in 2-, 5-, 8-, and 11-year-old children’s serum [37]. In terms of cord-blood frTT and SHBG, our study seemed to suggest an androgenic effect in males and anti-androgenic effect in females of maternal DEHP and MMP, in contrary to the effect in adults or cross-sectional study. The underlying mechanism of our study results is unclear and needs further study to confirm.

Our study revealed that cord-blood FSH levels was the most affected sex hormone by maternal phthalate exposures. We particularly noted that both male and female’s cord-blood FSH levels decreased as maternal MBzP increased. Similar results have been found in a study assessing sex hormones levels in minipuberty [38]. FSH stimulates Sertoli cells proliferation and promotes sperm production with TT in prepuberty [33]. In male adults, MBzP was reported to be negatively associated with lower sperm concentration [39] and lower sperm motility [17]. Our study findings suggest that prenatal MBzP exposure may inhibit early reproductive tract development in male and female newborns.

Pregnant women were suggested to avoid consume the packaged food with unsafe plastic wrap or containers [2, 40], and reduce the use of PCPs, i.e., body lotion, perfume, lip stick… etc., particularly the leave-on items [41, 42] and the products with long lasting fragrance. The careful consumption of packaged food and selective use of PCPs may protect themselves and their babies to minimize exposure levels of phthalates.

Strengths and limitations

Our study has several strengths. First, this study was one of the few cohort studies investigating the hormone-disrupting effect of prenatal phthalate exposure on cord-blood sex hormones and the associations between cord-blood sex hormones and AGD in newborns. Second, our study provided an opportunity to continually monitor the impact of relatively low prenatal phthalate exposure on fetal development after the 2011 phthalate-tainted-foodstuff episode in Taiwan. Third, we examined the babies’ cord-blood sex hormone concentrations, avoiding confounding effect of postnatal exposures of hormone-disrupting chemicals in environment.

Our study has certain limitations. First, we only collected one-spot urine samples in the third trimester, which may not fully represent the average phthalate-exposure level during pregnancy or the effective exposure levels on alteration of AGD at birth. A subset of participants (n = 142) in our study who provided urine samples in each trimester of pregnancy showed a weak-to-moderate consistency of DEHP metabolites between the third and first trimesters (Spearman’s correlation coefficient [ρ] 0.23 [MEHP]– 0.28 [MEMHP]) and between the third and second trimesters (ρ 0.17 [MEHP]– 0.42 [MEHHP]), indicating the likelihood of phthalate-exposure misclassifications in our study; nevertheless, the misclassifications should be non-differential and biased observed associations toward the null. Second, our study sample size may not be sufficient and thus reduced the statistical power in some analyses. Third, as we replaced the categorical form of phthalate metabolite levels with continuous concentrations, the significant associations between prenatal phthalate metabolite levels and male newborn’s AGD became statistically insignificant. We believed that might be partly resulted from limited sample size in our study. Another possible reason might be that the potential non-linear non-monotonic effects of phthalates on AGD were obscured in linear models. However, whether the relationships between prenatal phthalate exposures in such relative low levels were non-linearly associated with newborn’s AGD should be further validated.