Maternal Exposure to Environmental Disruptors and Sexually Dimorphic Changes in Maternal and Neonatal Oxidative Stress

Abstract

Context

Early pregnancy exposure to endocrine disrupting chemicals (EDCs) may contribute to poor birth outcomes through oxidative stress (OS)-mediated disruption of the maternal and fetal milieu. Most studies have investigated the effect of single EDC exposures on OS.

Objective

Assess the association of uniquely weighted mixtures of early pregnancy exposures with the maternal and neonatal OS markers.

Design

Prospective analysis of mother–infant dyads

Setting

University hospital.

Participants

56 mother–infant dyads.

Main Outcome Measures

The association of OS markers (nitrotyrosine, dityrosine, chlorotyrosine) in maternal first trimester and term, and cord blood plasma with maternal first trimester exposure levels of each of 41 toxicants (trace elements, metals, phenols, and phthalates) from 56 subjects was analyzed using Spearman correlations and linear regression. The association of OS markers with inflammatory cytokines and birth outcomes were analyzed by Spearman correlation and linear regression analysis, respectively. Weighted mixtures of early pregnancy exposures were created by principal component analysis and offspring sex-dependent and independent associations with oxidative stress markers were assessed.

Results

(1) An inverse relationship between levels of maternal/cord OS markers and individual EDCs was evident. In contrast, when assessed as EDC mixtures, both direct and inverse associations were evident in a sex-specific manner; (2) the maternal term OS marker, nitrotyrosine, was inversely associated with gestational age, and (3) both direct and inverse associations were evident between the 3 OS markers and individual cytokines.

Conclusions

Provides proof of concept that effects of exposures on OS varies when assessed as EDC mixtures versus individually.

Over the past 30–40 years, there has been a considerable increase in noncommunicable diseases (NCDs) such as metabolic, cardiovascular, immune, autoimmune, and neurological disorders, and some types of cancers (1). The increased incidence of NCDs over a short duration of time cannot be explained by genetics alone (2). Environmental factors (3, 4) and gene–environment interactions (5) are being considered as major contributors to this sudden increase in NCDs. In addition to poor nutrition and lifestyle choices, environmental endocrine disrupting chemicals (EDCs) are gaining wide recognition for their role in development of NCDs (6). The National Health and Nutrition Examination Survey has found a wide range of environmental chemicals, such as phenols, phthalates, parabens, perfluoroalkyl substances, and polychlorinated biphenyls, among others, detected in samples from US adults and children (7, 8). Many of these environmental chemicals have been associated with increased incidences of various NCDs (7, 9–11). The importance of exposure to environmental endocrine disruptors and their contribution to development of adult-onset disorders, both individually and as mixtures, is also recognized by the publication of the position paper by the Endocrine Society (12, 13).

Several EDCs act as hormone receptor agonists or antagonists and can disrupt metabolic processes, cell proliferation, and differentiation (14). Owing to the high rate of metabolism, cell proliferation, and differentiation, a developing fetus is particularly vulnerable to EDC exposure (2, 15). The detection of a multitude of EDCs in pregnant women from National Health and Nutrition Examination Survey, the Canadian Maternal-Infant Research on Environment Chemicals, and our group’s Michigan Mother Infant Pair (MMIP) studies (16–18) emphasizes the potential for increased risk from fetal exposures. Gestational exposures to bisphenol (BP) A, BPS, phthalates, ethyl paraben, cadmium, and copper have been linked to poor birth outcomes such as preterm delivery, need for obstetric interventions (eg, management of pre-eclampsia and gestational diabetes or requiring cesarean delivery), intrauterine growth restriction, and small or large for gestational age babies (19–25). For instance, first trimester maternal levels of BPA (20), BPS (18), and lead (26) were found to be negatively associated with birthweight. Likewise, polycyclic aromatic hydrocarbons in maternal circulation were negatively correlated with offspring birthweight (27). Other studies found gestational dibutyl phthalate exposure was linked to reduced birthweight (28), and mono (3-carboxypropyl) phthalate (MCPP) to increased birthweight (18). Considering that poor birth outcome is an established risk factor for development of adult pathologies as proposed in the Developmental Origins of Health and Disease hypothesis (29), and environmental EDCs have been shown to have an impact on birth outcomes, it is imperative to obtain a clear understanding of the effect of such EDC exposures on the mother and fetus.

Pregnancy is a precisely coordinated dynamic process that involves systemic and local changes in the mother to support nutrient and oxygen supply to the baby for growth in utero. Disruptions in this process can result in pregnancy complications, alterations in fetal growth trajectory, and preterm birth (30). The homeostasis of maternal milieu is maintained by various mediators including hormones, cytokines, oxidant status, and nutrient supply. Factors that disrupt this homeostasis, such as nutritional deficiency or excess, inflammation, oxidative stress, and lipotoxicity, may compromise fetal growth and development. The effect of EDCs in disrupting the maternal and fetal milieu is therefore a fertile area for investigation. We have previously reported that early pregnancy EDC exposures in the MMIP cohort are associated with changes in the maternal and neonatal inflammasome (31) and are correlated with negative birth outcomes (18, 20, 28).

In general, inflammatory states correlate with oxidative stress (32). Oxidative stress involves predominance of either or both oxidants and nitrosative species (reactive oxygen/nitrogen species [ROS/RNS]) over antioxidants (33). Indeed, studies in BPA-treated rodents and sheep, as well as preliminary observations in a small subset of MMIP women, found a positive association between gestational exposure to BPA and increased levels of oxidative stress markers in maternal circulation and neonatal cord blood (34). Similar links have been reported in both humans and animal models between individual phenols, parabens, phthalates, and benzenes with increased oxidative stress state during pregnancy (35–39). However, all these studies have examined the impact of each EDC individually. Human exposure to EDCs rarely occurs in isolation (40, 41) as humans could be exposed to many of the over 82 000 existing chemicals according to the current US Environmental Protection Agency inventory (42). Because the net biological effects of cumulative EDC exposures might vary from individual EDC exposure due to their additive, synergistic, or antagonistic properties (40), there is a need for examining the effect of EDC mixtures on the oxidative stress state. It is an important biomarker to assess, as oxidative stress can have a negative impact on both the mother and her offspring due to its association with several pregnancy-related complications such as gestational diabetes, pre-eclampsia, and preterm birth (43–46).

Our previous study examined the relationship between EDCs and inflammasome in the MMIP cohort and provided evidence that the effects on the inflammatory cytokines varied depending on whether individual or cumulative effects of EDCs were evaluated (31). To build on this proof of concept, this current study assessed the impact of early pregnancy phenol, phthalate, and metal exposures, both individually and in combination, on the maternal and neonatal markers of oxidative stress. In addition, as inflammation and oxidative stress are pathophysiologically linked (32), the association of maternal/neonatal oxidative stress state with the inflammasome was also investigated. Considering the bidirectional communication between the mother and fetus and potential for sexually dimorphic effects, sex-specific association of oxidative stress with exposures and, finally, the association between oxidative stress with birth outcomes, gestational age, and infant birthweight, were also investigated.

Methods

Subject recruitment

The subjects studied were recruited under the MMIP birth cohort study (2010 to present) at the University of Michigan (UM) as described previously (18). Briefly, pregnant women were recruited at their first prenatal appointment between 8 and 14 weeks of gestation with the following eligibility criteria for study participation: age between 18 and 42 years, had a spontaneously conceived singleton pregnancy, and intended to deliver at the UM hospital. A subset of MMIP participants from among these were selected for full exposure assessment and oxidative stress biomarker analysis, and selection criteria for these families were previously reported (18). Criteria for inclusion include the mother–infant pairs having complete demographic, survey, and health information at their initial study visit survey data and availability of all biospecimens at all time points from mother and child. Study procedures were approved by the UM Medical School Institutional Review Board, and all participants provided written informed consent.

Sample collection

Spot urine and venous blood samples were collected during the first prenatal appointment between 8 and 14 weeks of gestation (first trimester), and again upon arrival to the hospital for delivery (term), prior to intravenous placement. Cord blood was collected immediately after placental expulsion during the delivery. Maternal and cord blood samples were collected in ethylenediamine tetra-acetate containing vials and centrifuged to obtain plasma and aliquoted into glass vials. Urine samples were collected in polypropylene containers before being transferred to glass vials. All samples were stored at –80°C until further analysis.

Exposure assessments

Exposure measures carried out in maternal urine samples collected between 8 and 14 weeks’ of gestation have been previously described (18, 28). Briefly, exposure assessment was carried out following protocols from the Centers for Disease Control and Prevention (CDC) laboratory procedure manuals utilizing isotope dilution liquid chromatography/tandem mass spectrometry (MS) for phthalate metabolites (CDC method 6306.03) and environmental phenols (CDC method 6306.01), and isotope chromatography plasma/tandem MS for metals (CDC method 3018.3) with modifications to accommodate the expanded EDC panel and available instrumentations as reported before (18). In total, 41 exposure measures comprising a wide range of commonly encountered environmental EDCs, metals, and metalloids were quantified per subject. These measurements were also adjusted to specific gravity of maternal urine samples, an indicator of urine dilution, measured via a digital handheld device (ATAGO Company, Ltd, Tokyo, Japan).

The phthalates detected included metabolites of mono (2-ethyl-5-carboxylpentyl) phthalate (MECPP), mono (2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono (2-ethylhexyl) phthalate (MEHP), mono (2-ethyl-5-oxohexyl) phthalate (MEOHP), mono-isobutyl phthalate (MIBP), and mono n-butyl phthalate (MnBP). In addition, concentrations of mono-benzyl phthalate (MBzP), mono-carboxy isononyl phthalate (MCINP), mono (3-carboxypropyl) phthalate (MCPP), mono (6-COOH-2-methylheptyl) phthalate (MCOMHP), monoethyl phthalate (MEP), and mono-isononyl phthalate (MINP) were also assessed. Phenol analytes measured included parabens (butyl, ethyl, methyl, and propyl paraben [BuPB, EtPB, MePB, PrPB]), bisphenols (BPA, BPF, BPS), 2,4 and 2,5-dichlorophenol (DCP24, DCP25), benzophenone-3 (BP3), triclocarban (TCC), and triclosan (TCS). The heavy metals and metalloids were also quantified. These included arsenic, barium, beryllium, cadmium, chromium, copper, mercury, manganese, molybdenum, nickel, lead, selenium, tin, thallium, uranium, tungsten, and zinc.

Measurement of oxidative stress markers

As markers of oxidative stress, protein-bound oxidized tyrosine moieties, 3-nitrotyrosine (NY), 3-chlorotyrosine (CY), and o, o′-dityrosine (DY), were quantified in maternal plasma samples at recruitment, at delivery (term), and in neonatal cord blood using isotope dilution high-performance liquid chromatography electrospray ionization tandem MS (HPLC-ESI-MS/MS) as described previously (34, 47). Briefly, the plasma proteins were precipitated with ice-cold trichloroacetic acid (10% vol/vol), delipidated with water/methanol/water-washed diethyl ether (1:3:7 vol/vol/vol) and known amounts of isotopically labeled internal standards 13C6-Y and 13C6-NY, 13C6-CY, and 13C12-o, o′-DY added. These preparations were further hydrolyzed at 110°C for 24 hours in 4 M methanesulfonic acid solution saturated with 1% benzoic acid and subject to solid-phase extraction. The oxidized amino acids were quantified by HPLC-ESI-MS/MS with multiple reaction monitoring by integrating peak areas of the labeled standards and the analytes. The levels of the oxidized amino acids were then normalized to the precursor amino acid tyrosine content. The levels of oxidized tyrosine products are expressed as the ratio of the oxidized product over the total tyrosine. Intra-assay coefficients of variation calculated using a pool of all samples were 1.98%, 3.67%, and 13.57% respectively for NY, DY, and CY. Interassay coefficients of variation for all 3 measures were calculated from performance of a commercial pooled plasma and were <15%.

Relationship of oxidative stress markers with inflammasome

Oxidative stress marker levels were correlated with previously reported inflammatory cytokines from the same 56 subjects as described before (31). Briefly, inflammatory cytokines were assessed using the Luminex xMap assay (Millipore, Burlington, MA) as per the manufacturer’s recommendation at the clinical chemistry laboratory at the Michigan Diabetes Research Center (Ann Arbor, MI). The inflammatory markers investigated include granulocyte macrophage colony-stimulating factor (GM-CSF), interferon gamma, monocyte chemotactic protein 1, monocyte chemotactic protein 3, macrophage inflammatory protein 1, tumor necrosis-factor ɑ, vascular endothelial growth factor, interleukin (IL)-1β, IL-6, IL-8, and IL-17ɑ.

Birth outcomes

The details on birth outcomes and potential confounding variables from each of the recruited families were collected from the medical record as described before (18). Recorded birth outcomes include estimated gestational age at delivery (determined by last menstrual period or ultrasound dating, at the discretion of the obstetric provider) and birthweight. Potential confounding variables assessed were maternal prepregnancy body mass index, mode of delivery (vaginal versus cesarean), infant sex, maternal age, and history of smoking.

Statistical analysis

Spearman correlation coefficients were calculated to study the pairwise association between the 41 first term exposure variables (metals, phenols, and phthalates) and the 3 oxidative stress markers (CY, NY, DY). The EDC variables were first adjusted for urine specific gravity and then natural log transformed, prior to the correlation analysis. These correlations were analyzed cross-sectionally at each of the 3 oxidative stress measurement points: maternal first trimester, maternal term, and neonatal cord blood. This analysis first performed the pairwise association of each of the 3 oxidative stress marker with each of the 41 individual exposures, and then utilized Benjamini–Hochberg (BH) false discovery rate (FDR) to account for the multiplicity in the hypothesis testing. A BH FDR 0.1 was adopted for a cutoff for the statistical significance in the analyses.

Multiple linear regression was used to analyze the covariate-adjusted association between the 3 oxidative stress markers and the 41 individual exposures, in order to mitigate potential confounding effects in the association analysis. Prior to analysis, the exposures were natural log-transformed, and the oxidative stress values were transformed using inverse-normal transformation to ensure the outcome in the regression analysis to be normally distributed. Confounding variables considered in the analysis included: specific gravity of the urine sample, history of smoking, prepregnancy body mass index, mode of delivery, maternal age, infant sex, and gestational age. A BH-adjusted P value was used to address the multiplicity in hypothesis testing, with FDR 0.1 considered for statistically significant associations. Additionally, linear regression was employed to test the association between oxidative stress markers and birth outcomes. Birth outcomes considered were gestational age and birthweight, measured in grams. The oxidative stress values were transformed using standard normal transformation prior to analysis, and birthweight Fenton z-score was calculated and used as the outcome variable.

Principal component (PC) regression was conducted to analyze the effect of EDC mixtures on maternal and neonatal oxidative stress (48). The 41 EDC variables were reduced to 11 orthogonal PCs using the means of PC analysis, each of which represented an orthogonal linear combination of the EDC measures. The make-up of the PC variables are detailed in Table 1 and Reference (49). These 11 PCs collectively explained 80% of the variability in the EDC data, and were utilized simultaneously in a multiple linear regression analysis to study the association between PC-type EDC mixtures and oxidative stress markers. The loading coefficients for the individual EDCs within a single PC could be positive or negative. The sign of the loading coefficients (negative or positive) affected the sign of the association between the components of the PCs and the outcome of interest, as detailed further with examples in the Results section. This analysis was conducted for the full cohort, as well as on stratification by female and male infants, to understand both the composite and sex-specific associations. For the PC analyses, the 11 PCs were modelled together with the same confounding variables that were considered in the linear regression analysis, and the oxidative stress variables were transformed using inverse-normal transformation. As we considered the associations of the PCs with 3 different oxidative stress variables at each of our measurement points, a BH-adjusted P value was used to address the multiplicity in hypothesis testing. As in the linear regression analysis, a BH-adjusted P value with FDR < 0.1 was considered for statistically significant associations.

Table 1.

General Composition and Weightings of the 11 Orthogonal Principal Components (PCs) Generated Using Principal Component Analysis From the EDC Variables Measured

PCs	General Composition
PC 1	Positive metals (As, Ba, Be, Cd, Cr, Cu, Hg, Mn, Mo, Ni, Pb, Se, Sn, Tl, U, W, and Zn)
	Positive phthalates (MECPP, MEHHP, MEHP, MEOHP, MIBP, MnBP, MBzP, MCINP, MCPP, MCOMHP, MEP, and MINP)
PC 2	Negative metals (As, Be, Cd, Cr, Cu, Hg, Mn, Ni, Pb, Se, Sn, Tl, U, and Zn)
PC 3	Negative phthalates (MECPP, MEHHP, MEHP, MEOHP, MIBP, MnBP, MBzP, MCINP, MCPP, and MCOMHP)
	Positive phenols (BuPB, EtPB, MePB, PrPB, BP3, BPA, BPS, and DCP24)
PC 4	Positive metals (Ba, Be, Cd, Cr, Cu, Mn, Ni, Pb, Se, Sn, and Tl)
	Negative phthalates (MECPP, MEHHP, MEHP, MEOHP, MnBP, MBzP, MCINP, MCPP, MCOMHP, and MEP)
PC 5	Negative phenols (BuPB, MePB, PrPB, BPS, DCP24, DCP25, BP3, and TCS)
PC 6	Positive phthalates (MECPP, MEHHP, MIBP, MnBP, MCINP, and MCPP)
	Negative phenols (BPF, BPS, DCP24, DCP25, BP3, TCC, and TCS)
PC 7	Negative phenols (BuPB, MePB, PrPB, BPA, BPF, BPS, DCP24, DCP25, and TCC)
	Negative phthalates (MECPP, MIBP, MnBP, MBzP, MCINP, MCPP, MCOMHP, and MEP)
PC 8	Mix of group weightings
	Positive weighting in certain metals (Cr, Tl), phthalates (MEP), phenols (BP3, DCP25)
	Negative weightings in certain metals (Hg, Pb, Sn), phthalates (MBzP, MCOMHP, MEHP), phenols (BPA, BPS, BuPb, PrPb)
PC 9	Mix of group weightings
	Positive weighting in certain metals (Cu, Mg, Ni, Sn), phthalates (MzBP, MCOMHP, MEP), phenols (BPF, BP3, BuPb, EtPb, TCC)
	Negative weightings in certain metals (Ba, Mn, U, W), phthalates (MCINP, MCPP, MECPP, MEHHP, MINP), Phenols (BPA, BPS, DCP24, MePb, PrPb, TCS)
PC 10	Mix of group weightings
	Positive weightings in certain metals (Ba, Cd, Pb, U, W), phthalates (MBzP, MCOMHP, MCINP, MINP), and phenols (BuPb, EtPb)
	Negative weightings in certain metals (As, Hg, Mn, Se), phthalates (MIBP), and phenols (BPA, BPS)
PC 11	Mix of group weightings
	Positive weightings with metals (Ba, Cr, U), phthalates (MBzP, MCOMHP), and phenols (BP3, EtPb, MePb, TCC, TCS)
	Negative weightings with metals (Mn, Ni, Pb, W, Zn), phthalates (MCINP, MnBP, MINP), and phenols (BPS, BPF, BuPb)

Abbreviations: BP, bisphenol; BP3, benzophenone-3; BuPB, butyl paraben; MBzP, mono-benzyl phthalate; MCINP, mono-carboxy isononyl phthalate; MCOMHP, mono (6-COOH-2-methylheptyl) phthalate; MCPP, mono (3-carboxypropyl) phthalate; MECPP, mono (2-ethyl-5-carboxylpentyl) phthalate; MEHP mono (2-ethylhexyl) phthalate; MEOHP, mono (2-ethyl-5-oxohexyl) phthalate; MEP, monoethyl phthalate; MePB , methyl paraben; MIBP, mono-isobutyl phthalate; MINP, mono-isononyl phthalate; MnBP, mono n-butyl phthalate; MzBP, mono-benzyl phthalate; PrPB, Propyl paraben; TCC, triclocarban; TCS, triclosan.

Additionally, a longitudinal analysis was conducted, using a linear mixed-effects model (LMM) with a random intercept to account for the within-subject correlation among repeated measurements. LMM enables researchers to study the relationship between the longitudinal concentrations of maternal oxidative stress in the first trimester and term and exposures to the PC-type EDC mixtures. This LMM model incorporated interaction terms between time and the PC-type EDC mixtures, and thus analyzed if the associations between the maternal oxidative stress measures and the exposures shifted from baseline to term. We conducted a test for each of the 3 oxidative stress variables, and we calculated BH FDR-adjusted P values to account for multiple hypothesis testing, with a FDR < 0.1 considered for statistical significance.

Finally, a descriptive analysis based on Spearman correlation analysis was performed to describe the pairwise association between the oxidative stress markers and the inflammatory cytokines previously studied by Kelley et al (31). Pairwise correlations were calculated cross-sectionally at each of the two maternal time points (first semester and term) as well as neonatal cord plasma. All statistical analyses were carried out using R Software, version 3.5.

Results

Subjects, first trimester EDC exposures, and maternal and neonatal inflammasome and oxidative stress markers

The demographics of the 56 mothers in the mother–infant dyads from the MMIP cohort and their infant birth outcomes have been reported previously (18, 31). First trimester urinary concentrations of metals, phenols, parabens, and metabolites of phthalates for these 56 women have been previously detailed (18) and women in general had an average of 30 detectable analytes (average 12.8 metals, 10.5 phthalate metabolites, and 7.7 phenols metabolites) during the first trimester. The levels of inflammatory cytokines in the maternal first trimester and term and neonatal cord blood have also been previously reported (31), and on average the 12 cytokines assessed were detectable in approximately 70%, 58%, and 74% of the samples, respectively. The mean and standard deviation for NY, CY, and DY, the 3 markers of oxidative stress measured in maternal first trimester and term, and neonatal cord blood plasma are shown in Table 2.

Table 2.

Oxidative Stress Biomarkers Measured in Maternal Plasma Collected During First Trimester and at Term, and in Neonatal Cord Plasma

	Maternal First Trimester	Maternal Term	Neonatal Cord
Oxidative Stress Marker	Mean (SD)	Mean (SD)	Mean (SD)
NY	0.82 (1.08)	0.86 (1.16)	0.62 (0.81)
CY	1.54 (2.35)	1.83 (2.30)	2.70 (3.14)
DY	0.28 (0.35)	0.36 (0.58)	0.24 (0.37)

Oxidative stress markers are expressed as the ratio of the oxidized product over total levels of tyrosine.

Abbreviations: NY, 3-nitrotyrosine; CY, 3-chlorotyrosine, DY, o, o′-dityrosine; SD, standard deviation.

Association between oxidative stress measures and first trimester exposomes

Among the first trimester measures of individual exposures, phthalate metabolites were independently associated with maternal first trimester CY using Spearman pairwise correlation coefficients (Fig. 1). Significant negative pairwise correlations were observed between phthalate metabolites MCINP, MCPP, and MECPP and maternal first trimester plasma levels of CY.

Figure 1.

Cross-sectional Spearman correlations between early pregnancy EDC exposures and maternal first trimester and term, and neonatal cord blood plasma levels of oxidative stress markers. The pairwise correlation between oxidative stress markers with metals, phthalates and phenols are depicted using the black and white color spectrum, with the shaded circles indicating positive correlation and unshaded (white) circles indicating a negative correlation. The size of the circle reflects the size of the correlation coefficient, with larger circles indicating a larger correlation coefficient. Significance by BH FDR corrected p values are noted as follows: ***P < 0.01, **P < 0.05, *P < 0.10. Abbreviations for oxidative stress markers: NY, 3-nitrotyrosine, CY, 3-chlorotyrosine, and DY, o, o′-dityrosine. A color version of this figure is available in Reference (49).

When adjusted for confounders using linear regression with individual exposure biomarkers, significant associations with oxidative stress markers in first trimester and neonatal cord blood plasma were evident and are summarized in Table 3. Using FDR-BH correction to control for false discoveries related to multiple comparisons with FDR of <0.10, significant negative associations between first trimester phthalate metabolites MCPP and MCINP with CY was evident. Likewise, in the neonatal cord blood plasma significant negative associations for phthalate metabolites MBzP and MCOMHP with NY, and metals lead and copper with CY and NY, respectively were observed (Table 3) (49). No significant associations were evident when compared with maternal term plasma levels of oxidative stress markers. When associations between exposures and the differences between the maternal term and first trimester plasma levels of oxidative stress markers NY, CY, and DY were examined, no significant associations were evident (data not shown).

Table 3.

Associations From Linear Regression Model Between First Trimester EDC Exposure Analytes and Maternal (first trimester or term) and Neonatal Oxidative Stress Biomarkers

			Oxidative Stress Markerb
Time Pointa	EDC	EDC Category	Positive Association	Negative Association
First trimester	MCPP	Phthalate		CY
First trimester	MCINP	Phthalate		CY
Neonatal cord	MBzP	Phthalate		NY
Neonatal cord	MCOMHP	Phthalate		NY
Neonatal cord	Pb	Metal		CY
Neonatal cord	Cu	Metal		NY

Only significant relationships under the BH FDR 0.1 control for multiple testing are shown.

Abbreviations: CY, 3-chlorotyrosine; DY, o, o’-dityrosine; EDC, endocrine disrupting chemical; MBzP, mono-benzyl phthalate; MCINP, mono-carboxy isononyl phthalate; MCOMHP, mono (6-COOH-2-methylheptyl) phthalate; MCPP, mono (3-carboxypropyl) phthalate; NY, 3-nitrotyrosine.

^aAll EDC exposures were measured in first trimester.

^bOxidative stress markers were measured at the timepoint listed.

As EDC exposures rarely occur individually, the cumulative effects of the different EDCs on oxidative stress markers were also assessed, utilizing PCs representing groups of correlated exposures. The individual PC components are detailed in Table 1 (49). Among the 11 PCs, 6 PCs showed statistically significant associations with the first trimester, term, and neonatal oxidative stress markers and results are summarized in Table 4 (49). A BH FDR-adjusted P value with FDR < 0.10 was again used to assess statistical significance to account for multiple hypothesis testing. For example, in the maternal first trimester plasma, levels of oxidative stress marker NY showed a negative association with PC 1, which has higher positive weighting for metals and phthalates. These data therefore suggest that higher metal and phthalate levels may be negatively associated with first trimester NY levels (Table 4). In contrast, PC 3, which is weighted positively for phenols and negatively for phthalates, has a positive association with maternal first trimester CY levels (Table 4). These data therefore suggest that phenols are positively associated while phthalates have an inverse association with maternal first trimester CY.

Table 4.

Associations From Linear Regression Model Between First Trimester EDC PC and Maternal (first trimester or term) and Neonatal Oxidative Stress Biomarkers

	Oxidative Stress Markerb
PC Variablea	Positive Association	Negative Association
First trimester oxidative stress levels
PC 1		NY
PC 3	CY
PC 4	CY
PC 10		CY
PC 11	NY, CY
Term oxidative stress levels
PC 9		CY
PC 11	DY
Neonatal cord oxidative stress levels
PC 1		NY
PC 10		CY
Differences in oxidative stress levels between first trimester and term
PC 9		CY

Only significant relationships under the BH FDR 0.1 control for multiple testing are shown.

Abbreviations: CY, 3-chlorotyrosine; DY, o, o′-dityrosine; EDC, endocrine disrupting chemical; NY, 3-nitrotyrosine; PC principal component.

^aPCs were based on EDC exposures measured in first trimester.

^bOxidative stress markers were measured at the timepoint listed.

In the longitudinal LMM analysis, there were statistically significant negative interactions between time of maternal CY oxidative stress measurement and 2 of the PCs: PC 7 and PC 9 (Table 5). These negative values indicate a negative shift, or decrease, in the association between the PC and maternal oxidative stress measurements at term versus first trimester, a change that is not captured by the single time-point analyses conducted by linear regression. This longitudinal analysis also indicated statistically significant associations between the PCs and the maternal first trimester oxidative stress levels, which were similar to the cross-sectional linear regression analysis results (49).

Table 5.

Association from Linear Mixed Model of EDC PC with Change in Oxidative Stress Measures From Maternal Baseline to Term by Longitudinal Analysis

	Oxidative Stress Marker
PC Variable	Positive Association	Negative Association
PC 7		CY
PC 9		CY

Only significant relationships under the BH FDR 0.1 control for multiple testing are shown.

Abbreviations: CY, 3-chlorotyrosine; EDC, endocrine disrupting chemical; PC, principal component

Association between oxidative stress measures and exposures by offspring sex

The linear regression analysis of sex-dependent association between the levels of oxidative stress markers at 3 different time points: maternal first trimester and term plasma and the neonatal cord blood plasma and are summarized in Table 6 (49). Sex-dependent associations were evident at all time points (Table 6). For example, maternal first trimester plasma levels of CY were negatively associated with PC 1 for mothers of female offspring and positively associated with PC 2 (Table 6). PC 1 is positively weighted for metals and phthalates while PC 2 is negatively weighted for metals (Table 1), and these data suggest that there is an overall inverse association between maternal first trimester levels of metals and CY.

Table 6.

Gender Specific Associations From Linear Regression Model of EDC PC From Measures in First Trimester With Oxidative Stress Levels in Maternal (first trimester and term) and Neonatal Cord Blood

	Offspring Gender			Oxidative Stress Markerb
PC Variablea	Composite	Males	Females	Positive Association	Negative Association
First trimester oxidative stress levels
PC 1	X	X	X		NY
			X		CY
PC 2			X	CY
PC 3	X		X	CY
PC 4			X	NY
	X			CY
PC 5			X		NY
			X		CY
PC 6		X		NY
		X		DY
PC 7			X		NY
PC 10	X				CY
PC 11	X		X	NY
	X			CY
Term oxidative stress levels
PC 4			X	CY
PC 7			X		CY
PC 9	X				CY
PC 11	X			DY
Neonatal cord oxidative stress levels
PC 1	X		X		NY
PC 4			X		DY
PC 8			X		DY
PC 10			X	DY
	X				CY
Differences in oxidative stress levels between first trimester and term
PC 4		X			CY
PC 8		X			CY
PC 9	X				CY

Only significant relationships under the BH FDR 0.1 control for multiple testing are shown.

Abbreviations: CY, 3-chlorotyrosine; DY, o, o′-dityrosine; EDC, endocrine disrupting chemical; NY, 3-nitrotyrosine; PC, principal component;

PCs were based on EDC exposures measured in first trimester.

Oxidative stress markers were measured at the timepoint listed.

Association between oxidative stress measures and birth outcomes

Linear regression analysis demonstrated one significant negative association between maternal term plasma levels of NY with gestational age at birth (estimate [–2.970], SE 1.077 and P = .009). No significant associations were evident between the first trimester maternal or neonatal oxidative stress markers and offspring birthweight.

Association between oxidative stress and inflammasome measures

As oxidative stress and inflammation go hand in hand, we explored whether the previously reported levels of cytokines in the same sample (31) set were associated with markers of oxidative stress assessed in the current study. Spearman rank-based pairwise correlation analysis showed positive and negative cross-sectional associations with maternal (first trimester and term) and neonatal cord plasma cytokines and oxidative stress markers. For example, the inflammatory cytokine GM-CSF was negatively correlated with NY in the maternal first trimester time point (Fig. 2).

Figure 2.

Cross-sectional descriptive Spearman correlations between inflammasomes and oxidative stress markers from respective maternal first trimester and term, and neonatal cord blood plasma. Using the black and white color spectrum, the black shaded color indicates a positive pairwise correlation, while the white un-shaded color indicates a negative pairwise correlation for oxidative stress markers NY, CY, and DY. The size of the circle reflects the size of the correlation coefficient, with larger circles indicating a larger correlation coefficient. Abbreviations for oxidative stress markers: NY, 3-nitrotyrosine, CY, 3-chlorotyrosine, DY, o, o′-dityrosine. Color version of this figure is available in Reference (49).

Discussion

This study revealed the proof of concept that maternal and neonatal oxidative/nitrosative stress status is associated with maternal first trimester exposure burden and that these associations are evident when considering exposures individually or as mixtures. Even more importantly, the findings reveal that many associations of exposure PCs with oxidative stress markers are offspring sex specific. The mixture analysis of exposures with maternal and neonatal oxidative/nitrosative state and birth outcomes not only highlights the risk posed by environmental exposures but also the need to consider exposome as a whole rather than at the level of individual exposures in analyses. Efforts are ongoing at the Endocrine Society by creation of “EDC Clinical Strategy Task Force” to educate clinicians about the risks posed by EDCs. As epidemiologists and clinicians delve more and more into such investigations and relate to clinical outcomes, it is important to recognize possible interactions among the various EDCs that humans are exposed to in parallel and likely to differ from individual to individual. The implications of these findings are discussed below.

Role of oxidative stress in pregnancy

Oxidative stress reflects the imbalance between ROS/RNS generated through metabolic processes and the antioxidant defense mechanisms. ROS/RNS have been shown to participate in placental development during early pregnancy (50). When the metabolic activity of the growing fetus increases, the metabolic activity of placenta increases (51), resulting in increased production of ROS/RNS (52). In normal pregnancies, the buildup of ROS/RNS is offset by corresponding increases in antioxidants (53). Uncontrolled accumulation of ROS/RNS have been reported in conditions with pregnancy complications and poor birth outcomes such as miscarriage, pre-eclampsia, fetal growth restriction, and preterm labor (43). In spite of the samples used from the MMIP cohort consisting only of full-term pregnancies, a negative correlation was found between term oxidative stress marker NY and gestational age, consistent with the contribution of oxidative stress to gestational age (45). The association of shorter gestational age only with oxidative stress marker NY but not DY or CY stresses the importance of the specificity of the biomarkers being investigated in addressing such relationships. In another study, the oxidative stress marker 8-isoprostane but not 8-hydroxydeoxyguanosine was found to be associated with preterm delivery (54). Although multiple reports exist that oxidative stress state is associated with low birthweight (43, 55), a similar association was not evident in the present study—possibly a function of the full-term cohort studied with no neonates having low birthweight (<2500 g).

Maternal EDC exposures and oxidative stress

Imbalance in ROS/RNS and associated oxidative stress is a common finding across many different classes of environmental pollutants including xenoestrogens, pesticides, and heavy metals (39, 56). Several studies suggest that oxidative stress during pregnancy could be predictive of adverse birth outcomes (57). Considering that maternal EDC exposures are implicated in offspring outcomes (12), oxidative stress provides a plausible biological pathway for EDCs to bring about its effects. Although oxidative stress and exposure to EDCs have been associated with many of the same health effects, until recently they have been studied rarely in parallel and often as individual chemicals, not mixtures. Our studies with humans and several animal models also found an association between high first trimester BPA levels and increased NY (34), a product of tyrosine nitration mediated by RNS and a marker of cell damage and inflammation. The fact that this association of BPA with NY was not found in the present study may be a function of conscious selection of subset of samples with very high maternal BPA levels in the previous study (34). A significant association between BPA and parabens with markers of oxidative stress (8-hydroxydeoxyguanosine and isoprostane) was found among Puerto Rican pregnant women (36). While an association of BPA with oxidative stress was not evident in the present study, a significant inverse association between individual phthalate metabolites and oxidative stress marker CY during first trimester, and individual phthalate metabolites and metal copper with neonatal cord blood NY and metal lead with CY was evident (Table 2). Association of 9 different phthalate metabolites with urinary isoprostane levels has been found in pregnant women studied in Boston and in Puerto Rico (37, 58, 59). Copper exposure in the third trimester has been previously associated with preterm birth (25), and prenatal lead is known to influence multiple maternal and child outcomes (60, 61). These studies and our data suggest that oxidative stress may be a mechanism underlying some of these effects.

Interestingly, most of the associations between oxidative stress and individual exposures in our study were inverse, the opposite of what was expected based on other studies (62, 63). However, in studies reporting positive correlations between oxidative stress and phthalates, markers of lipid or DNA oxidative damage were used whereas our study measured markers of amino acid (tyrosine) oxidative damage. The products of oxidative stress assessed in previous studies namely markers of lipid peroxidation such as 8-isoprostance or DNA oxidation such as 8-hydroxydeoxyguanosine are relatively unstable, nonspecific, and are not reflective of the underlying mechanisms activating oxidative stress. In contrast, because of their covalent nature, products of oxidized amino acids are stable and plasma levels serve as an accurate index of systemic oxidative stress over longer duration (64). Moreover, these oxidized amino acids provides indications of underlying pathways of oxidation that are activated (eg, CY formation is catalyzed by myeloperoxidase while NY are products of RNS) (65). These stable products of protein oxidation have been found to be associated with many pathological conditions (59–62).

The uniqueness of the present study is that the effects of several EDCs were investigated both individually as well as in a mixture context with 3 different stable oxidative stress markers. Additionally, the study was not restricted to identifying association between EDCs and oxidative stress but extended the equation to relating oxidative stress with pregnancy outcomes. Although the findings from this study found associations between individual phthalate metabolites/metals with maternal and/or neonatal oxidative stress, the identification of associations between different combinations of EDC mixtures and maternal/neonatal oxidative stress measures that differed from that identified by individual EDCs emphasizes the need for future studies to consider exposome as a whole. Interestingly, while individual EDCs showed negative associations with oxidative stress biomarkers, when analyzed as mixtures many positive associations became evident indicating the EDCs in mixtures may have a more profound effect. In addition, the data are also suggestive that similar class of chemicals may have similar effects as evident by the positive association of oxidative stress marker CY with PC 3 and 4 groupings both of which have negative weightings for phthalate metabolites. In contrast, negative association with CY in groupings with both positive and negative (mixed) weightings for example combination of phthalate metabolites, phenols, and metals, such as that seen in PC 10 (Table 1) could reflect the sum effect of additive, synergistic, and antagonistic actions of the various components, depending on their chemical properties and biological pathways that they affect. It is important to note that many of these associations also varied depending on stage of pregnancy (first trimester vs term) or compartment (maternal vs fetal) examined. These data extend our earlier proof of concept study, which related EDC mixtures with inflammasome in the same MMIP families (31) and draws similar conclusions (66).

Sex-specific relationship of EDCs with oxidative stress outcomes

Because EDCs impact endocrine functions including hormones involved in sexual differentiation, it is not surprising that they can have sex-specific effects. Studies in animal models have pointed to sex-specific programming effects of individual EDC exposures on organ differentiation and adult outcomes (67, 68). The impact of prenatal exposure to EDCs on sex-specific effects on maternal and neonatal hormonal/metabolic milieu are limited. Few studies in animal models and humans that have addressed the impact of gestational exposure to individual EDCs on oxidative stress state have not dissected out sex-specific responses (34, 37, 54). Considering the bidirectional communication between the mother and developing fetus, the sex of the developing fetus has the potential to lead to sexually dimorphic effects both at the maternal and at the offspring level. Sex-specific effects of individual EDCs reported for BPA include low birthweight and longer gestational age among female offspring from mothers with higher BPA levels at first trimester and term, respectively (20). In contrast, male offspring with larger size at birth were born to mothers with higher phthalate metabolites during the first trimester (28). The present study is unique in that it addresses the impact of gestational EDC mixtures during a critical period of fetal differentiation and points to EDC mixtures having an impact on the maternal and neonatal oxidative stress in a sex-specific manner, an observation that needs to be validated in larger cohorts.

Yin–Yang of oxidative stress and inflammation

A tight interrelationship between oxidative stress and inflammation has been shown in many systems including placenta during both physiological and pathological states (69). Although in most conditions, oxidative stress and inflammation seem to occur simultaneously and can promote each other to further enhance the pathological state (70), oxidative stress can itself stimulate expression of many pro- and anti-inflammatory cytokines (71–74). Consistent with this finding, markers of maternal and neonatal oxidative stress in our study correlated with many of the 12 cytokines assessed (Fig. 2). The association between cytokine GM-CSF and NY, for example, is similar to that observed in chronic inflammatory conditions (75) and endothelial cells (76). Although maternal and neonatal inflammatory markers were also associated with both infant birthweight and gestational age (31), associations for oxidative stress markers were only found with gestational age in the present study. This suggests that offspring gestational age may be influenced by oxidative stress and inflammation state of the maternal and/or neonatal milieu.

Limitations and strengths

Some limitations to be considered in interpreting these findings include that these studies were carried out in a small sample set from a nondiverse cohort with exposome measures carried out only during the first trimester. However, the lack of diversity among study subjects is also a strength as it may have contributed to circumvent the effect of confounding factors arising from race, socioeconomic status, and educational background. Although exposure to EDCs can vary throughout the pregnancy and thereby impact the oxidative and inflammatory status in a stage-specific manner, the uniqueness of this study is that exposure biomarkers were assessed during a critical period of fetal development encompassing the sexually dimorphic window and vulnerability.

Another limitation to consider is the oxidative stress biomarker assessed. Oxidative stress state is associated with oxidation of cellular lipid, protein, and nucleic acids (71). In this study, 3 major markers of oxidative/nitrosative stress that mainly measured the oxidation and nitrosylation products of protein tyrosine moieties were assessed. However, oxidized and/or nitrosylated tyrosine moieties are stable products that can be easily assessed and are known to be associated with various disease and inflammatory states (77, 78), thus serving as a useful biomarker to assess the association between EDCs and oxidative stress. In addition, considering that oxidative stress state is also affected by changes in the antioxidants (33), the relative levels of antioxidants and their associations with EDC exposure also needs to be considered in a future study. Other possible confounders not addressed include hemodilution or hemoconcentration of plasma samples, maternal prepregnancy comorbidities, maternal weight gain during pregnancy, parity, and medication use. The potential also exists for diet and dietary supplements such as vitamins and exercise to influence inflammatory and oxidative state (79, 80). For example, vitamins E and C can help boost antioxidant and anti-inflammatory capabilities (80). Studies with animal models (81–83) and epidemiological data (84) point to EDC–diet interactions, although such investigations are limited to individual EDCs and/or dietary type or supplementation. Future studies need to take into account these confounders when impacts of EDCs are being addressed. In spite of the limitation of not including such variables, the present proof of concept longitudinal study stresses the need for future studies to consider exposome as a whole instead of studying individual EDCs. This is particularly important as health outcomes are ultimately affected by the sum of the additive, synergistic, or antagonistic properties of all the EDCs acting via several common signaling pathways.

While the results of this study need to be confirmed in larger and more diverse cohorts and causality of the relationship between exposure burden and oxidative stress established, this study provides the proof of concept that exposures, as mixtures, in a sex-specific manner have the potential to impact the oxidative state of the maternal and neonatal milieu.

Glossary

Abbreviations

BH

Benjamini–Hochberg

BP

bisphenol

BP3

benzophenone-3

BuPB

butyl paraben

CDC

Centers for Disease Control

CY

3-chlorotyrosine

DY

o, o′-dityrosine

EDC

endocrine disrupting chemical

EtPB

ethyl paraben

FDR

false discovery rate

HPLC-ESI

high-performance liquid chromatography electrospray ionization

IL

interleukin

LMM

linear mixed-effects model

MBzP

mono-benzyl phthalate

MCINP

mono-carboxy isononyl phthalate

MCOMHP

mono (6-COOH-2-methylheptyl) phthalate

MCPP

mono (3-carboxypropyl) phthalate

MECPP

mono (2-ethyl-5-carboxylpentyl) phthalate

MEHP

mono (2-ethylhexyl) phthalate

MEOHP

mono (2-ethyl-5-oxohexyl) phthalate

MEP

monoethyl phthalate

MePB

methyl paraben

MIBP

mono-isobutyl phthalate

MINP

mono-isononyl phthalate

MnBP

mono n-butyl phthalate

MS

mass spectrometry

NCD

noncommunicable disease

NY

3-nitrotyrosine

OS

oxidative stress

PC

principal component

RNS

reactive nitrogen species

PrPB

propyl paraben

ROS

reactive oxygen species

TCC

triclocarban

TCS

triclosan

VEGF

vascular endothelial growth factor

Additional Information

Disclosure statement: The authors have no conflicts of interest to disclose.

Data Availability: All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.