Prenatal phthalates, gestational weight gain, and long-term weight changes among Mexican women

Abstract

Background:

Phthalates are endocrine disrupting chemicals that may influence weight status; however, few studies have considered weight gain during pregnancy and subsequent long-term weight changes in women.

Objective:

To determine associations of prenatal phthalate exposure with maternal weight during pregnancy and through up to seven years post-delivery.

Methods:

We analyzed 15 urinary phthalate biomarker concentrations during the 2^nd and 3^rd trimesters among 874 pregnant women enrolled in the Programming Research in Obesity, Growth Environment and Social Stress Study in Mexico City. We examined three time-specific maternal weight outcomes: gestational weight gain (between 2^nd and 3^rd trimesters), short-term weight (between 3^rd trimester and 12 months post-delivery), and long-term weight (between 18 months and 6 – 7 years post-delivery). We used Bayesian Kernel Machine Regression (BKMR) to estimate associations for the total phthalate mixture, as well as multivariable linear mixed models for individual phthalate biomarkers.

Results:

As a mixture, 2^nd trimester urinary phthalate biomarker concentrations were associated with somewhat lower gestational weight gain between the 2^nd and 3^rd trimesters (interquartile range, IQR, difference: −0.07 standard deviations, SD; 95% credible interval, CrI: −0.20, 0.06); multivariable regression and BKMR models indicated that this inverse association was primarily driven by mono-2-ethyl-5-carboxypentyl terephthalate (MECPTP). Prenatal (2^nd and 3^rd trimesters) urinary phthalate mixture concentrations were positively associated with maternal weight change through 12 months postpartum (IQR difference: 0.11 SD; 95% CrI: 0.00, 0.23); these associations persisted from 18 months to 6–7 years follow-up (IQR difference: 0.07 SD; 95% CrI: 0.04, 0.10). Postpartum weight changes were associated with mono-3-carboxypropyl phthalate (MCPP) and MECPTP.

Conclusions:

Prenatal phthalate exposure was inversely associated with gestational weight gain and positively associated with long-term changes in maternal weight. Further investigation is required to understand how phthalates may influence body composition and whether they contribute to the development of obesity and other cardiometabolic diseases in women.

1. Introduction

Pregnancy, and its associated biological and lifestyle changes, is a critical event in a woman’s life course that predisposes later weight gain, metabolic disturbances, and the development of chronic diseases (Gunderson 2009; Rasmussen et al. 2010). Traditionally pregnancy is considered as a developmental window for child health and development, but the same physiological changes that influence child development are experienced by the mother and may play a role in her long-term health trajectory. Women with high gestational weight gain are more likely to develop gestational diabetes mellitus, gestational hypertension, and postpartum weight retention (Champion and Harper 2020; Kominiarek et al. 2018), all of which are risk factors for poor cardiometabolic health outcomes, including type 2 diabetes, cardiovascular diseases, and obesity (Kim et al. 2002; Lind et al. 2014; McDowell et al. 2019). Although diet and exercise are strongly associated with weight gain both during and after pregnancy (Amorim Adegboye and Linne 2013), there is growing evidence that environmental exposures, such as endocrine disrupting chemicals, may have influential roles (Heindel et al. 2017; Muscogiuri et al. 2017).

Phthalates are environmentally ubiquitous endocrine disrupting chemicals used in food manufacturing (e.g., processing and packaging) and various consumer, personal care, and industrial products, which may enter the body through ingestion, inhalation, and dermal absorption (Katsikantami et al. 2016). Among pregnant women, major sources of phthalate exposures include personal care products (e.g., cosmetics, perfumes, and lotions), beverages and foods in plastic packaging, and use of microwave with plastic containers (Bustamante-Montes et al. 2021; Fisher et al. 2019; Just et al. 2010; Pacyga et al. 2019; Serrano et al. 2014). Phthalates are hypothesized to manipulate body composition by interfering with energy balance as well as glucose and lipid metabolism (Desvergne et al. 2009). Their actions may be particularly potent during pregnancy (Clewell et al. 2008). In female rodents, dietary exposure to di-2-ethylhexyl phthalate (DEHP) was associated with increased appetite, body weight, and percent visceral fat prior to pregnancy, which persisted through delivery and lactation (Schmidt et al. 2012). Among non-pregnant women, higher urinary concentrations of specific phthalate biomarkers were associated with greater body mass index (BMI) and waist circumference (Buser et al. 2014; Díaz Santana et al. 2019; Dong et al. 2017; Hatch et al. 2008; Lind et al. 2012; Yaghjyan et al. 2015), as well as modestly faster rates of weight gain during a 10-year follow-up period (Song et al. 2014).

Few studies have investigated associations of prenatal phthalate exposures and maternal body weight during pregnancy (Bellavia et al. 2017; James-Todd et al. 2016; Perng et al. 2020; E. M. Philips et al. 2020) or the postpartum (Perng et al. 2020; E.M. Philips et al. 2020; Rodríguez-Carmona et al. 2019), with differing results. In studies of pregnant women, higher first trimester urinary monoethyl phthalate (MEP) was associated with greater early gestational weight gain (defined as weight gain between the first and second prenatal visits) (Bellavia et al. 2017) and greater odds of excessive total gestational weight gain (defined as weight gain between the first prenatal visit and delivery) (James-Todd et al. 2016), while DEHP was associated with lower early gestational weight gain (Bellavia et al. 2017) and lower weight at delivery (Perng et al. 2020). In studies of women during postpartum follow-up, DEHP was associated with slower weight loss during the first year postpartum (Perng et al. 2020); phthalic acid, low molecular weight phthalates, and dioctyl phthalate (DNOP) metabolites were associated with weight gain from delivery through six years follow-up (E.M. Philips et al. 2020); and mono-3-carboxypropyl phthalate (MCPP) was associated with higher annual rates of weight gain through nearly 11 years follow-up (Rodríguez-Carmona et al. 2019).

Previously, we reported a high burden of prenatal phthalate exposure, particularly dibutyl phthalate (DBP), diisobutyl phthalate (DiBP), and benzyl butyl phthalate (BBzP), as well as an alternative phthalate, di(2-ethylhexyl) terephthalate (DEHTP), among women enrolled in a birth cohort in Mexico City (Wu et al. 2020). Given the paucity of literature examining phthalate exposures and maternal health, we now utilize longitudinal data from this cohort to determine whether prenatal urinary concentrations of phthalate biomarkers are associated with changes in weight during pregnancy through seven years post-delivery.

2. Methods

2.1. Study design and population

Data were collected from the Programming Research in Obesity, Growth Environment and Social Stress (PROGRESS) Study, a longitudinal birth cohort of mother-infant pairs in Mexico City, enrolled 2007–2011. Details of the study are published elsewhere (Wu et al. 2020). Briefly, pregnant women were recruited from prenatal clinics of the Mexican Social Security System. Eligibility criteria included: age 18 years or older; gestational age less than 20 weeks at enrollment; planning to reside in Mexico City for the next three years; free of heart or kidney disease; not using steroids or anti-epilepsy drugs; not daily consumers of alcohol; and able to access a telephone. After enrollment, women had follow-up study visits during the 3^rd trimester (27 – 36 weeks’ gestation) and 1 month, 6 months, 12 months, 18 months, 2 – 3 years, 4 – 5 years, and 6 – 7 years post-delivery. Women provided written informed consent. Study protocols were approved by institutional review boards at the Brigham and Women’s Hospital, Icahn School of Medicine at Mount Sinai, and the Mexican National Institute of Public Health. There were 948 women enrolled in the study who delivered a live birth; 874 women had prenatal urine collection and at least two weight measurements, and were included in the current analyses.

2.2. Urinary Phthalate Biomarker Concentrations

Women provided random spot urine samples at the 2^nd and 3^rd trimester visits. Samples were divided into 2 milliliter (mL) aliquots in phthalate-free tubes, stored at −80°C, and shipped on dry ice to the CDC in December 2017. Phthalate metabolite analyses used isotope dilution high-performance liquid chromatography coupled with tandem mass spectrometry to quantify 15 urinary phthalate metabolites (Silva et al. 2007): mono-n-butyl phthalate (MBP), mono-isobutyl phthalate (MiBP ), mono-hydroxybutyl phthalate (MHBP), mono-hydroxyisobutyl phthalate (MHiBP), MCPP, MEP, mono-2-ethyl-5-carboxypentyl phthalate (MECPP), mono-2-ethylhexyl phthalate (MEHP), mono-2-ethyl-5-hydroxyhexyl phthalate (MEHHP), mono-2-ethyl-5-oxohexyl phthalate (MEOHP), monobenzyl phthalate (MBzP), mono(carboxy-isononyl) phthalate (MCNP) mono(carboxy-isooctyl) phthalate (MCOP), monooxononyl phthalate (MONP), mono-2-ethyl-5-carboxypentyl terephthalate (MECPTP). The linearity of measurement range, from lowest to highest calibrator, for the analytes was verified by R² of the standard curves between 0.95 and 0.99.

Limits of detection (LOD) ranged from 0.2 to 1.2 nanograms (ng)/mL, depending on the metabolite. We replaced concentrations that were below the LOD (<0.6% of all samples) with the lowest instrument reported concentration for that metabolite (Schisterman et al. 2006). Digital handheld refractometer (AR200, Reichert Technologies, Buffalo, NY) measured specific gravity (SG). We used the median value of 1.016 for samples with missing SG measures (n = 101, from the 2^nd trimester visit only), the robustness of this approach within this cohort has been discussed previously by Wu and colleagues (Wu et al. 2020). The formula for dilution normalization of phthalate measurement by SG is: Pc = P{(SGm-1)/(SG-1)}, where Pc is the SG-corrected metabolite concentration (ng/mL), P is the measured phthalate metabolite concentration, SGm is the median SG for all samples, and SG is the specific gravity for that individual urine sample. In addition to CDC laboratory standard analytic quality control protocols, we included a pool of anonymous adult human urine (Bioreclamation IVT, New York, USA) as blinded replicates, randomly inserted 92 times throughout the study samples. The coefficient of variation for all metabolites ranged from 2.1% to 16.5%, with a median of 8.1% across metabolites. We calculated molar sums of DEHP (ΣDEHP = MEHP + MEHHP + MEOHP +MECPP), DINP (ΣDINP = MONP + MCOP), DIBP (ΣDIBP = MHiBP + MiBP), and DBP (ΣDBP = MBP + MHBP). We computed the geometric mean across 2^nd and 3^rd trimester measurements for statistical modeling to address variability in phthalate biomarker measurements. There were 156 women who did not have phthalate biomarker measurements at the 3^rd trimester visit. Exclusion of these women from statistical analyses did not substantively change our results; therefore, we presented our results based on the full cohort.

2.3. Maternal Anthropometric Measurements and Other Characteristics

Maternal height and weight were measured by trained staff using a stadiometer and scale, respectively, while women wore light clothing and no shoes. Height was measured at enrollment and weight was measured at all study visits except delivery. These measurements were used to calculate maternal BMI (kilograms (kg)/meters (m)²). We examined three distinct, time-specific maternal weight outcomes in relation to pregnancy: 1) gestational weight gain, defined as the difference between measured weights at the 2^nd and 3^rd trimester visits; 2) short-term weight, defined as the difference in weight measurements collected at the 3^rd trimester visit and at the 1-month, 6-month, and 12-month post-delivery visits; and 3) long-term weight, defined as the difference in weight measurements at the four post-delivery visits, 18 months, 2 – 3 years, 4 – 5 years, and 6 – 7 years, after the 12-month post-delivery visit. Short-term weight is reflective of women’s weight changes within the first year postpartum, while long-term weight is reflective of women’s weight changes throughout several years post-delivery.

Self-reported sociodemographic information was collected at enrollment and included: maternal age (years); education (<high school, high school graduate, >high school); parity (nulliparous or parous); environmental tobacco smoke in the home (collected at enrollment – 6-month post-delivery visit, yes or no); and alcohol use (collected at enrollment – 6-month post-delivery visit, yes or no). Socioeconomic status was calculated based on an index created by the Mexican Association of Market and Public Opinion Research Agencies (Spanish acronym: AMAI) (Carrasco 2002). There were 13 variables derived from a questionnaire regarding head of household’s education level, number of various items in household, and number of different types of rooms in the household. It was then collapsed into six SES levels ranging from lowest to highest SES status within the cohort. For better interpretability, we further collapsed the index into three categories; low, medium, and high. We also included seasonality of urine collection (November – February, March – April, May – October) as a covariate to represent long-term and seasonal trends. Gestational age was calculated based on two methods: subtraction of the self-reported date of last menstrual period from delivery date and the Capurro method (Capurro et al. 1978; Rodosthenous et al. 2017). If the gestational age estimated by the two methods differed by more than three weeks (n=40), we used the gestational age from the Capurro method. All postpartum models excluded women who were diagnosed with preeclampsia (n=47) or who became pregnant during follow-up (n=30).

2.4. Statistical Analysis

We examined associations for all 9 biomarkers jointly as a mixture using Bayesian Kernel Machine Regression (BKMR) with each of the three maternal weight outcomes. BKMR is adaptable to repeated measurements and can flexibly model multiple mixture components and exposure windows simultaneously (Bobb et al. 2015). It estimates individual mixture components (i.e., single biomarker) importance, total mixture effect, as well as explores non-linearity and interactions between different mixture components. Since BKMR is flexible to different hierarchical groupings of mixture components, we built the following BKMR models where the mixture components were: 1) ungrouped (no hierarchical structure); 2) grouped by trimester; and 3) grouped by parent phthalate compound. Collectively, these models allowed for assessment of model robustness and determination of whether a specific exposure time period (2^nd or 3^rd trimester) or parent phthalate compound was more relevant to the maternal weight outcomes. The results from all of the examined BKMR models were very similar; therefore, we present the ungrouped BKMR models of mean phthalate biomarker concentrations. For all BKMR models, we scaled all model inputs, including exposures, outcomes, and covariates, and specified Gaussian kernel and 50,000 iterations.

In addition to mixture models, we estimated associations of individual 2^nd trimester urinary phthalate biomarker concentrations and gestational weight gain using linear regression models. We used linear mixed models with random intercepts to estimate associations of 2^nd and 3^rd trimester mean urinary phthalate biomarker concentrations with short- and long-term weight to account for the repeated measurement design. We first modeled 2^nd and 3^rd trimester phthalate biomarker concentrations separately, but found no meaningful trimester-specific differences; therefore, we present models using the mean of the 2^nd and 3^rd trimester phthalate biomarker concentrations.

We included age, height, baseline weight, gestational age (weeks), socioeconomic status (collapsed to three-level index of low, medium, and high), education, and parity in all multivariable models. The post-delivery study visits were tightly scheduled and there was little variation in the timing of when these visits occurred relative to delivery date, particularly within the first 24 months post-delivery, therefore, models were not adjusted for maternal age at study visits. Alcohol and tobacco use during pregnancy, secondhand smoke exposure, and seasonality were not included in the multivariable model, but were assessed in sensitivity analyses. All analyses were conducted using R 3.5.3 (Team 2018). BKMR models and linear mixed models were fit using the bkmr and lme4 packages, respectively (Bates et al. 2015; Bobb et al. 2015).

3. Results

At enrollment, women had a mean age of 27.7 years (standard deviation, SD = 5.5) and second trimester BMI of 26.9 kg/m² (SD = 4.2). The majority of women were of low socioeconomic (a level 3 or lower among the six-level SES index, 74%) and education status (high school graduate or lower, 76%) and nearly one third of women reported environmental smoke exposure (Table 1). All of the measured phthalate biomarkers were detected in more than 85% of the samples, and most of the metabolites were detected in more than 99% of the samples. The highest prenatal urinary phthalate biomarker concentrations were observed for MEP, MBP, and DEHP (Table 2).

Table 1.

Baseline distributions of sociodemographic characteristics of women enrolled in the Programming Research in Obesity, Growth Environment and Social Stress Study (n=874)

Maternal Characteristic	Mean (SD)
Age (years)	27.7 (5.5)
Body Mass Index (kilograms/meters2)
2nd Trimester (n=874)	26.9 (4.2)
3rd Trimester (n=794)	29.4 (4.1)
6 Months Post-delivery (n=627)	26.9 (4.6)
6–8 Years Post-delivery (n=515)	27.8 (5.0)
	N (%)
Socioeconomic statusa
1 (lowest)	80 (9%)
2	375 (43%)
3	194 (22%)
4	129 (15%)
5	81 (9%)
6 (highest)	15 (2%)
Education
< High School	353 (40%)
High School	312 (36%)
>High School	209 (24%)
Nulliparous
No	471 (54%)
Yes	402 (46%)
Environmental Smoking at Home
No	603 (69%)
Yes	264 (30%)
Alcohol During Pregnancy
No	834 (95%)
Yes	40 (5%)

^aCalculated based on characteristics of the household according to AMAI criteria and collapsed into low (1 – 2), medium (3 – 4), and high (5 – 6) for statistical analyses.

Table 2.

Distribution of specific gravity-corrected urinary phthalate biomarker concentrations, stratified by trimester of pregnancy, in the Programming Research in Obesity, Growth Environment and Social Stress Study

		2nd Trimester (n=874)	3rd Trimester (n=792)
Parent Phthalate	Metabolite	GM (95% CI)	GM (95% CI)
DEHP	MEHP	5.67 (5.31–6.05)	5.79 (5.37–6.24)
	MEOHP	18.37 (17.23–19.59)	23.11 (21.63–24.69)
	MEHHP	19.92 (18.65–21.27)	23.62 (22.05–25.29)
	MECPP	42.29 (39.90–44.83)	50.28 (47.21–53.55)
DEHTP	MECPTP	1.83 (1.71–1.96)	2.33 (2.16–2.52)
DiNP	MONP	1.31 (1.23–1.41)	1.68 (1.57–1.80)
	MCOP	4.57 (4.30–4.86)	4.72 (4.44–5.03)
DiDP	MCNP	0.97 (0.93–1.03)	1.05 (0.99–1.1)
DOP	MCPPa	1.46 (1.37–1.54)	1.56 (1.46–1.66)
BBzP	MBzP	5.39 (4.98–5.84)	5.85 (5.39–6.35)
DiBP	MHiBP	3.46 (3.25–3.68)	3.83 (3.57–4.11)
	MiBP	8.96 (8.42–9.53)	10.61 (9.90–11.37)
DBP	MBPb	81.05 (75.72–86.76)	89.44 (83.26–95.06)
	MHBP	7.19 (6.68–7.73)	7.92 (7.35–8.54)
DEP	MEP	140.52 (128.73–153.39)	156.51 (141.74–172.81)

GM, geometric mean; CI, confidence interval

^aAlso a minor metabolite of several high molecular weight phthalates

^bAlso a minor metabolite of BBzP

3.1. Prenatal Phthalate Biomarkers and Gestational Weight Gain

As a joint mixture, 2^nd trimester phthalate biomarker concentrations were negatively associated with gestational weight gain between the 2^nd and 3^rd trimesters (IQR difference: −0.07 SD, 95% credible interval, CrI: −0.20, 0.06) in fully adjusted BKMR models (Figure 1A and Supplemental Table 1). In multivariable linear mixed models, only MECPTP was statistically significantly associated with gestational weight gain; each doubling of 2^nd trimester MECPTP concentrations was associated with a 0.19 kg (95% CI: −0.34, −0.03) lower weight gain between the 2^nd and 3^rd trimesters (Figure 2A, Supplemental Table 2). This was consistent with the BKMR model, which showed that MECPTP was negatively associated with gestational weight gain holding all phthalate biomarkers at their median concentration values (Figure 2B). Additional adjustment for time elapsed between prenatal visits did not produce meaningfully different estimates (Supplemental Table 2, model 3).

Figure 1 -.

Overall associations of the phthalates mixture on A) maternal gestational weight gain, B) short-term weight, and C) long-term weight. This figure plots the estimated weight differences and 95% credible intervals (*expressed as standard deviations) when all mixture components are at the indicated percentiles relative to when mixture components are all at the 25^th percentile (reference).

Figure 2 –

Associations of 2^nd trimester urinary phthalate biomarker concentrations and gestational weight gain (kilograms, kg) in models of individual phthalate biomarkers using A) linear regression or B) Bayesian Kernel Machine Regression (BKMR) as a mixture. The linear regression models are interpreted as estimated change in weight (kg) per doubling of phthalate biomarker concentrations and the corresponding 95% confidence intervals. The BKMR models are presented as univariate dose response curves with a scaled and centered exposure (x-axis) and outcome (y-axis). The shaded areas represent the 95% credible intervals. All models were adjusted for maternal age, socioeconomic status, education, primiparity, height, baseline weight, and gestation age at urine collection.

3.2. Prenatal Phthalate Biomarkers and Maternal Weight through 6 – 7 years Follow-up

As a joint mixture, 2^nd and 3^rd trimester urinary phthalate biomarker concentrations were positively associated with short-term weight (IQR difference: 0.11 SD, 95% CrI: 0.00, 0.23) in fully adjusted BKMR models (Figure 1B and Supplemental Table 1). In multivariable linear mixed models, all of the phthalate biomarkers were positively associated with short-term weight change (Figure 3A). Each doubling of concentrations of ∑DEHP, MECPTP, ∑DINP, MCPP, and ∑DIBP was associated with 0.11 – 0.21 kg greater short-term weight (only results for MECPTP, ∑DINP, and MCPP reached statistical significance, p <0.05). Further adjustment of models for gestational weight measurements did not meaningfully change the observed associations (Supplemental Table 3). In BKMR models, holding all phthalate biomarkers at their median values, MECPTP, MCPP, and DIBP were positively associated with short-term weight change (Figure 3B). Because weight retention and weight loss rate may differ throughout the postpartum, we conducted a sensitivity analysis to assess each of the three periods (third trimester – one month postpartum, 1 – 6 months postpartum, and 6 – 12 months postpartum) independently. We found positive associations across all three periods, but there was a difference in magnitude of effect when comparing the first two periods (third trimester – 1 month and 1 – 6 months) compared to the last (6 – 12 months) (Supplemental Figure 1).

Figure 3 –

Associations of mean urinary phthalate biomarker concentrations and short-term weight (measured weights from 3^rd trimester through 12 months post-delivery) in models of individual biomarkers using A) linear mixed models or B) Bayesian Kernel Machine Regression (BKMR) as a mixture. The linear mixed models are interpreted as estimated change in weight (kg) per doubling of biomarker concentrations and the corresponding 95% confidence intervals. The BKMR models are presented as univariate dose response curves with a scaled and centered exposure (x-axis) and outcome in standard deviations (y-axis). The shaded areas represent the 95% credible intervals. All models adjusted for maternal age, socioeconomic status, education, primiparity, height, baseline weight, and gestational age at urine collection.

For long-term weight, 2^nd and 3^rd trimester urinary phthalate biomarker concentrations were positively associated with long-term weight (IQR difference: 0.07 SD, 95% CrI: 0.04, 0.10) in fully adjusted BKMR models (Figure 1C and Supplemental Table 1). Prenatal concentrations of MCPP were positively associated with long-term weight change (B = 0.43 kg, 95% CI: 0.09, 0.77) in multivariable linear mixed models (Figure 4A). This association was attenuated after additional adjustment for maternal weight measured at 12 months (Supplemental Table 4). These findings were consistent with BKMR models, with MECPTP showing positive trends with maternal weight (Figure 4B).

Figure 4 –

Associations of mean urinary phthalate biomarker concentrations and maternal long-term weight (through 6 – 7 years post-delivery) in models of individual biomarkers using A) linear mixed models or B) Bayesian Kernel Machine Regression (BKMR). The linear mixed models are interpreted as estimated change in weight (kg) per doubling of biomarker concentrations and the corresponding 95% confidence intervals. The BKMR models are presented as univariate dose response curves with a scaled and centered exposure (x-axis) and outcome in standard deviations (y-axis). The shaded areas represent the 95% credible intervals. All models adjusted for maternal age, socioeconomic status, education, primiparity, height, baseline weight, and gestational age at urine collection.

We examined group (all phthalate biomarkers) and conditional (individual phthalate biomarkers) posterior inclusion probabilities (PIP) to determine whether there were differences in the relative importance of 2^nd trimester versus 3^rd trimester phthalate biomarker concentrations with maternal short-term and long-term weight (Supplemental Table 5). For both short-term and long-term weight, group PIP for the 2^nd and 3^rd trimesters were reasonably high but suggested that 3^rd trimester biomarker concentrations may be somewhat more influential than 2^nd trimester biomarker concentrations.

4. Discussion

In this prospective pregnancy cohort of Mexican women, we found that prenatal phthalate exposure was associated with changes in maternal weight during pregnancy and through approximately seven years post-delivery but in opposite directions. During pregnancy, 2^nd trimester joint urinary phthalate biomarker mixture concentrations were associated with lower gestational weight gain between the 2^nd and 3^rd trimesters; multivariable regression and BKMR models indicated that this negative association was primarily driven by MECPTP. During the post-delivery follow-up, prenatal joint urinary phthalate biomarker mixture concentrations were positively associated with changes in maternal weight through one year postpartum; these associations persisted from 18 months to 6 – 7 years follow-up and were associated with MCPP and MECPTP.

In non-pregnant women, positive associations of specific phthalate biomarkers with anthropometric measurements, weight gain, and obesity are often reported (Buser et al. 2014; Díaz Santana et al. 2019; Dong et al. 2017; Hatch et al. 2008; Lind et al. 2012; Song et al. 2014; Yaghjyan et al. 2015). Evidence linking prenatal phthalate exposures with weight changes during pregnancy (Bellavia et al. 2017; James-Todd et al. 2016; Perng et al. 2020; E. M. Philips et al. 2020) or the postpartum (Perng et al. 2020; E.M. Philips et al. 2020; Rodríguez-Carmona et al. 2019) is limited to studies from three pregnancy cohorts. These studies observed generally negative associations between prenatal urinary phthalate biomarker concentrations and gestational weight gain. Among Mexican women in the Early Life Exposure in Mexico to Environmental Toxicants (ELEMENT) cohort, MBzP, MCPP, MEP, ∑DEHP, and ∑DBP were associated with lower maternal weight at delivery (Perng et al. 2020). In the Generation R cohort (Netherlands), phthalic acid and phthalate groupings (i.e., low molecular weight phthalates, high molecular weight phthalates, ∑DEHP, and ∑DNOP) were associated with overall lower weight gains during mid- and late pregnancy (>=18 weeks’ gestation), but not with weight gain during early pregnancy (E. M. Philips et al. 2020). Similarly, in the LIFECODES cohort (Boston, United States), ∑DEHP was non-linearly associated with lower early gestational weight gain between the 1^st and 2^nd trimesters (Bellavia et al. 2017); however, MEP was associated with greater early gestational weight gain (Bellavia et al. 2017) and excessive total gestational weight gain (James-Todd et al. 2016). Following pregnancy, studies observed generally positive associations of prenatal urinary phthalate biomarker concentrations and longitudinal changes in women’s weight. In ELEMENT, MBzP, MCPP, ∑DEHP, and ∑DBP were positively associated with weight change through one year postpartum (Perng et al. 2020). Only MCPP was associated with greater annual weight gain from one year through eight years post-delivery follow-up (range, 5.2 – 10.7 years), while MBzP was associated with lower annual weight gain during this time period (Rodríguez-Carmona et al. 2019). In Generation R, phthalic acid and phthalate grouping were associated with weight gain between pre-pregnancy and six years post-delivery (E.M. Philips et al. 2020).

Our results are consistent with those from previous studies, suggesting that prenatal phthalate exposures may be negatively associated with weight gain during pregnancy and positively associated with weight changes after delivery, and that specific phthalates, such as MCPP and MECPTP, may have influential roles. It is difficult to make comparisons across studies due to important methodological differences between them, such as number and timing (i.e., trimester of pregnancy) of prenatal urine collections and maternal weight measurements. In contrast to the majority of previous studies, which examined single exposure models (Bellavia et al. 2017; James-Todd et al. 2016; Perng et al. 2020; E.M. Philips et al. 2020; E. M. Philips et al. 2020; Rodríguez-Carmona et al. 2019), we used a mixture approach to assess the nine prenatal phthalate biomarkers during the 2^nd and 3^rd trimesters and account for confounding and interactions between them. This may explain some of the discrepancies in findings for individual phthalate biomarkers between studies. For example, DEHP and DBP metabolites were identified as contributors to postpartum weight previously (Perng et al. 2020; E.M. Philips et al. 2020) and in our multivariable linear models, but not when we considered them within the mixture. However, we did find that MECPTP, a biomarker of DEHTP, was associated with maternal weight outcomes, particularly lower gestational weight gain. DEHTP is a replacement compound for DEHP and other ortho-plasticizers that is used in a variety of products, including food packaging, medical equipment, and flooring (Silva et al. 2019). Considering the previous evidence for DEHP with maternal weight outcomes (Bellavia et al. 2017; Perng et al. 2020), our results suggest that DEHTP may have similar, if not stronger, metabolic health consequences to DEHP and warrants further investigation.

There are several biologically plausible mechanisms by which phthalates may adversely influence weight gain and other metabolic health outcomes. Phthalates can alter the expression of transcription factors and genes necessary for glucose and lipid metabolism and energy balance by activating peroxisome proliferator-activated receptors (PPAR) (Desvergne et al. 2009). For example, DEHP and its metabolites can directly bind to PPARγ (Kambia et al. 2008; Taxvig et al. 2012), a hormone receptor that is highly expressed in adipose tissue and involved in adipogenesis, adipocyte differentiation, and fat storage (Desvergne et al. 2009). PPARγ is also involved in insulin sensitivity and inflammatory responses (Casals-Casas and Desvergne 2011) and implicated in the development of diabetes, atherosclerosis, obesity, and hyperlipidemia (Desvergne et al. 2009). Contrary to our findings for post-delivery weight gain, we observed negative associations between prenatal phthalate exposures and gestational weight gain. During pregnancy, specifically, phthalates may interfere with thyroid hormones, progesterone, and estrogen (Huang et al. 2016; Johns et al. 2016; Sathyanarayana et al. 2014), which could influence gestational weight gain by reducing fetal, placental, and maternal tissue growth. There is some evidence that prenatal phthalate exposures are negatively associated with placenta and birth weight (Philippat et al. 2019; Zhao et al. 2015; Zong et al. 2015), but less is known about associations with maternal fat stores and other maternal tissues.

Strengths of this study include the repeated maternal weight measurements from pregnancy through 7 years post-delivery, measurement of MECPTP, and the use of statistical methods that accommodate phthalate biomarker mixtures. The use of BKMR allowed for a flexible mixture model that can accommodate correlated exposures, repeated measurements, and opposite directions of effects, as well as examining the importance of prenatal phthalate exposure windows (i.e., 2^nd versus 3^rd trimester). There are also several potential limitations that should be considered. Our study population was Mexican women and may not be generalizable to other pregnant populations. Urinary concentrations of phthalate biomarkers were collected during the 2^nd and 3^rd trimesters of pregnancy, but not at any other time point during pregnancy or the postpartum, so we could not assess potential impacts of 1^st trimester or postnatal phthalate exposures. Given the short biological half-lives of phthalates and known variability of phthalate metabolites throughout pregnancy (Wu et al. 2020), exposure misclassification was possible. Breastfeeding duration did not meet criteria as a confounder and was not included in statistical models; however, it is associated with weight loss during the postpartum period (Jarlenski et al. 2014) and therefore may have impacted the precision of our results. There may also have been residual confounding due to unmeasured factors, such as gestational glycemia (and related complications) and maternal dietary intakes. Regarding diet specifically, DEHP and DEHTP are found in plastic food packages, which includes unhealthy foods (e.g., fast foods and snack foods) that may be associated with body weight. Although we can’t discount the possibility of confounding, mixture modeling provides some advantage by partially removing bias from uncontrolled confounders (when they are common sources of the exposure components) and total mixture effects are unaffected by co-exposure amplification bias (Webster and Weisskopf 2020; Weisskopf et al. 2018). Lastly, we lacked repeated measurements of women’s body composition during pregnancy and the postpartum and were unable to determine whether weight changes reflected fat or fat-free mass during these time periods. Gestational weight gain is not solely attributed to fat deposition, but also fluid expansion and maternal and fetal tissue growth and development. Similarly, postpartum body weight varies by changes in subcutaneous versus visceral fat, which may have important lasting implications for maternal metabolic health outcomes (Janumala et al. 2020; Sohlström and Forsum 1995).

5. Conclusions

We found that prenatal phthalate exposure was negatively associated with gestational weight gain and positively associated with changes in maternal weight through approximately seven years post-delivery. These associations were primarily driven by MECPTP and MCPP. Given the paucity of research on this topic, further investigation is required to understand how phthalates may influence changes in maternal body composition and whether they contribute to the development of cardiometabolic diseases.

Abbreviations:

DEHP

di-2-ethylhexyl phthalate

BMI

body mass index

MEP

monoethyl phthalate

DNOP

dioctyl phthalate

MBP

mono-n-butyl phthalate

MiBP

mono-isobutyl phthalate

MCPP

mono-3-carboxypropyl phthalate

MHBP

mono-hydroxybutyl phthalate

MHiBP

mono-hydroxyisobutyl phthalate

MECPP

mono-2-ethyl-5-carboxypentyl phthalate

MEHP

mono-2-ethylhexyl phthalate

MEHHP

mono-2-ethyl-5-hydroxyhexyl phthalate

MEOHP

mono-2-ethyl-5-oxohexyl phthalate

MBzP

monobenzyl phthalate

MCNP

mono(carboxy-isononyl) phthalate

MCOP

mono(carboxy-isooctyl) phthalate

MONP

monooxononyl phthalate

MECPTP

mono-2-ethyl-5-carboxypentyl terephthalate

DBP

dibutyl phthalate

DINP

diisononyl phthalate

DIBP

diisobutyl phthalate

CDC

Centers for Disease Control and Prevention

LOD

limit of detection

ng

nanogram

mL

milliliter

SG

specific gravity

BKMR

Bayesian Kernel Machine Regression

kg

kilograms

m

meter

PIP

posterior inclusion probabilities