Oxidative stress as a potential mechanism linking gestational phthalates exposure to cognitive development in infancy

Abstract

Background:

Gestational exposure to phthalates, endocrine disrupting chemicals widely used in consumer products, has been associated with poor recognition memory in infancy. Oxidative stress may represent one pathway linking this association. Hence, we examined whether exposure to phthalates was associated with elevated oxidative stress during pregnancy, and whether oxidative stress mediates the relationship between phthalate exposure and recognition memory.

Methods:

Our analysis included a subset of mother-child pairs enrolled in the Illinois Kids Development Study (IKIDS, N = 225, recruitment years 2013–2018). Concentrations of 12 phthalate metabolites were quantified in 2^nd trimester urine samples. Four oxidative stress biomarkers (8-isoprostane-PGF_2α, 2,3-dinor-5,6-dihydro-8-iso-PGF_2α, 2,3-dinor-8-iso-PGF_2α, and prostaglandin-F_2α) were measured in 2^nd and 3^rd trimester urine. Recognition memory was evaluated at 7.5 months, with looking times to familiar and novel stimuli recorded via infrared eye-tracking. Novelty preference (proportion of time looking at a novel stimulus when paired with a familiar one) was considered a measure of recognition memory. Linear mixed effect models were used to estimate associations between monoethyl phthalate (MEP), sum of di(2-ethylhexyl) phthalate metabolites ($\mathrm{\Sigma }\mathrm{DEHP}$), sum of di(isononyl) phthalate metabolites ($\mathrm{\Sigma }\mathrm{DINP}$), and sum of anti-androgenic phthalate metabolites ($\mathrm{\Sigma }\mathrm{AA}$) and oxidative stress biomarkers. Mediation analysis was performed to assess whether oxidative stress biomarkers mediated the effect of gestational phthalate exposure on novelty preference.

Results:

The average maternal age at delivery was 31 years and approximately 50% of participants had a graduate degree. A natural log unit increase in $\mathrm{\Sigma }\mathrm{AA}$, $\mathrm{\Sigma }\mathrm{DINP}$, and $\mathrm{\Sigma }\mathrm{DEHP}$ was associated with a statistically significant increase in 8-iso-PGF_2α, 2,3-dinor-5,6-dihydro-8-iso-PGF_2α, and 2,3-dinor-8-iso-PGF_2α. The association was greatest in magnitude for $\mathrm{\Sigma }\mathrm{AA}$ and 2,3-dinor-5,6-dihydro-8-iso-PGF_2α (β= 0.45, 95% confidence interval= 0.14, 0.76). The relationship between $\mathrm{\Sigma }\mathrm{AA}$, $\mathrm{\Sigma }\mathrm{DINP}$, $\mathrm{\Sigma }\mathrm{DEHP}$, and novelty preference was partially mediated by 2,3-dinor-8-iso-PGF_2α.

Conclusions:

Gestational exposure to some phthalates is positively associated with oxidative stress biomarkers, highlighting one mechanistic pathway through which these chemicals may impair early cognitive development.

Introduction

Phthalates are a group of synthetic chemicals widely used to improve qualities in plastics.¹ Phthalates are found in 80–90% of global consumer products, ranging from household appliances to personal care products, and can easily leach into their surrounding environment, leading to chronic human exposure which is inexorable and universal. National representative studies of the U.S. population find that nearly all individuals have detectable concentrations in urine of at least a few phthalate metabolites.² Individuals who use cosmetics, soaps, shampoos, and other personal care products more frequently have elevated concentrations of phthalate metabolites relative to those who report less frequent use of these products.^3,4 This is particularly concerning as phthalates are known endocrine disruptors, thereby altering metabolism and interfering with hormone regulation, affecting growth, fertility, and reproduction.¹

Pregnant women and the developing fetus are uniquely vulnerable to the effects of phthalates given the rapid hormonal changes occurring during pregnancy, and immaturity of fetal metabolic pathways.⁵ Phthalates can cross the placental barrier and disrupt hormonal homeostasis in utero, a period critical in proper regulation of the maternal-fetal unit, through modulating the expression of key hormonal activities.^6–10 Even minor alterations during gestation, particularly to the rapidly developing brain, can induce abnormalities in biochemical parameters that increase offspring susceptibility to disease.^11,12 In recent years, numerous prospective studies have reported adverse associations between gestational exposure to specific phthalates, including di(2-ethylhexyl) phthalate (DEHP) and diisononyl phthalate (DINP), and neurodevelopmental disorders in childhood relating to deficits in externalizing and internalizing problem behaviors,^13,14 executive functions,¹⁵ psychomotor function, social interactions, attention, and memory.^16–22 Within the Illinois Kids Development Study (IKIDS), we have shown that higher urinary concentrations of monoethyl phthalate (MEP), and of biomarkers of DEHP, DINP, and anti-androgenic phthalates (AA) are associated with a reduction in novelty preference in early infancy, which reflects worsening recognition memory.²³

Oxidative stress has long been implicated in the pathogenesis of disease, and is of particular importance during pregnancy.²⁴ Oxidative stress is a naturally occurring phenomenon reflecting an imbalance between the production of reactive oxygen species (ROS) in the body and its counteracting antioxidant mechanisms.²⁵ Isoprostanes are considered some of the best biomarkers of oxidative stress and are of particular importance during pregnancy as they activate various stress-sensitive signaling systems that influence fetal programming.²⁶ A robust body of research has shown that exposure to phthalates can induce oxidative stress. Specifically, levels of urinary 8-isoprostane-prosaglandin-F_2α (8-iso-PGF_2α) are elevated among pregnant persons following exposure to DEHP. ^27–31 Separately, studies have linked elevated levels of isoprostanes to adverse pregnancy outcomes, including preterm birth and preeclampsia, as well as adverse cognition outcomes in offspring.^32–38 We hypothesize that oxidative stress may be one mechanism through which gestational phthalate exposure may influence neurodevelopment, as our prior work in the IKIDS cohort has shown that elevated levels of F₂-isoprostanes during pregnancy is associated with poor recognition memory in early infancy.

Here, we leveraged the prospective IKIDS cohort to examine whether gestational exposure to select individual phthalates, assessed from the following urinary biomarkers, namely the sum of di(2-ethylhexyl) phthalate metabolites [$\mathrm{\Sigma }\mathrm{DEHP}$], sum of di(isononyl) phthalate metabolites [$\mathrm{\Sigma }\mathrm{DINP}$], sum of anti-androgenic [$\mathrm{\Sigma }\mathrm{AA}$], and MEP, was associated with elevated levels of F2-isoprostanes. Additionally, we evaluated to what extent our previously observed effects of phthalate exposure on novelty preference operated through oxidative stress (Figure 1).²³

Figure 1:

Analytical framework for mediation analysis of urinary phthalate exposures, oxidative stress biomarkers, and visual recognition memory outcomes.

Methods

Study Participants

Mother-child pairs included in the present analysis were a subset of those enrolled in the Illinois Kids Development Study (IKIDS), an ongoing, longitudinal birth cohort designed to investigate prenatal exposure to environmental chemicals and association with offspring physical, behavioral, and cognitive development. Details of the study design and recruitment protocols have been previously described.³⁹ Briefly, participants were recruited from prenatal clinics in Urbana and Champaign, IL beginning in 2013. Pregnant women were eligible for inclusion if they were pregnant with a singleton at less than 15 weeks gestation, were between 18 and 40 years of age, and had a low-risk pregnancy. The study was approved by the Institutional Review Board at the University of Illinois Urbana-Champaign, and all participants provided full, written informed consent prior to enrollment. For this analysis, we included two subsets of participants: 1) those for whom information on urinary phthalate metabolites and oxidative stress biomarkers were available (N = 225, analytic sample 1), and 2) the subset of the 225 for whom infant cognitive outcomes were assessed at 7.5 months of age (N =143).

Phthalate exposure assessment

As part of IKIDS, participants provided a first morning void urine sample at 16–18 weeks’ gestation using polypropylene collection cups. Urine samples were stored at −80 °C prior to quantification of 12 phthalate metabolites at the Centers for Disease Control and Prevention (CDC) Division of Laboratory Sciences by solid phase extraction-high performance liquid chromatography-isotope dilution tandem mass spectrometry, as described before.^23,40 The involvement of the CDC laboratory did not constitute engagement in human subjects research. See Table S1 for a complete list of all phthalate metabolites measured. To account for urinary dilution, concentrations of phthalate metabolites were specific gravity corrected. Phthalate metabolite concentrations below the limit of detection (LOD) were imputed using the instrumental reading if available; if not, the imputed value was the LOD divided by the square root of 2. The detection frequencies for individual phthalate metabolites are provided in Table S1. For consistency with our prior work,²³ our analysis focused on $\mathrm{M}\mathrm{E}\mathrm{P}$ (main metabolite of diethyl phthalate), and the following phthalate biomarkers: $\mathrm{\Sigma }\mathrm{D}\mathrm{E}\mathrm{H}\mathrm{P},\mathrm{\Sigma }\mathrm{D}\mathrm{I}\mathrm{N}\mathrm{P}$, and $\mathrm{\Sigma }\mathrm{A}\mathrm{A}.\mathrm{\Sigma }\mathrm{D}\mathrm{E}\mathrm{H}\mathrm{P},\mathrm{\Sigma }\mathrm{D}\mathrm{I}\mathrm{N}\mathrm{P}$, and $\mathrm{\Sigma }\mathrm{A}\mathrm{A}$ were calculated using the following equations:

$$\mathrm{\Sigma }DEHP=\left(\frac{MEHP}{278}\right)+\left(\frac{MEHHP}{294}\right)+\left(\frac{MEOHP}{292}\right)+\left(\frac{MECPP}{308}\right)$$

$$\mathrm{\Sigma }DINP=\left(\frac{MiNP}{292}\right)+\left(\frac{MCOP}{322}\right)$$

$$\begin{gathered} \mathrm{\Sigma }AA=\left(\frac{MiBP}{222}\right)+\left(\frac{MBzP}{256}\right)+\left(\frac{MEHP}{278}\right)+\left(\frac{MEHHP}{294}\right)+\left(\frac{MEOHP}{292}\right)+\left(\frac{MECPP}{308}\right) \\ +\left(\frac{MCOP}{322}\times 0.43\right)+\left(\frac{MiNP}{292}\times 0.43\right)+\left(\frac{MBP}{222}\right)+\left(\frac{MHiNP}{238}\right)+\left(\frac{MHBP}{238}\right) \end{gathered}$$

All phthalate biomarker concentrations were right skewed and natural log transformed for analysis.

Urinary stress biomarker measurement

Urine samples collected at up to two timepoints during pregnancy (16–18 weeks gestation [N=225] and 22–24 weeks gestation [N=117]) were frozen at −80 °C prior to oxidative stress biomarker measurement. The Eicosanoid Core Laboratory at Vanderbilt University Medical Center quantified levels of four oxidative stress biomarkers using liquid chromatography–mass spectrometry.^41,42 8-iso-PGF_2α, and its two major metabolites 2,3-dinor-5,6-dihydro-8-iso-PGF_2α, and 2,3-dinor-8-iso-PGF_2α, and the cyclooxygenase-derived prostaglandin-_2α (PGF_2α). Oxidative stress biomarker concentrations were normalized to specific gravity to account for urinary dilution. Oxidative stress biomarker concentrations below the LOD were substituted by the LOD divided by the square root of 2.⁴³ All biomarker concentrations were right skewed, and natural log transformed for analysis.

Cognitive Assessment

Infant cognition was assessed at 7.5 months using a visual recognition memory (VRM) paradigm that employed infrared eye-tracking technology to assess looking behavior data. The VRM task was completed using SR Research EyeLink 100 to record looking behavior. To complete the task, infants sat on their caregiver’s lap, and stimuli were shown on a television screen. Infant looking behavior was first calibrated using animated audiovisual clips. For the VRM task, each assessment consisted of three trials, with black and white photographs of faces as the stimuli. In the familiarization trial, infants were shown two identical faces side by side. In two subsequent test trials, infants were simultaneously shown two faces: one familiar and one novel. In test trials, novelty preference was calculated as the proportion of time spent looking at the novel face, relative to the familiar face, where a larger proportion of time looking at the novel stimulus was an indicator of better recognition memory. Data were processed using the SR Research DataViewer software (SR Research Ltd., Mississauga, Ontario, Canada) and were averaged across each trial.⁴⁴

Statistical Analysis

Distributions of demographic characteristics and cognitive outcomes (i.e., novelty preference) within the two analytic samples were tabulated with counts, proportions, means and standard deviations (SDs). Distributions of urinary phthalate exposure biomarkers and oxidative stress biomarkers were assessed using geometric means, geometric SDs, and selected percentiles. Pearson correlations coefficients were calculated to assess correlations between phthalate exposure biomarkers, oxidative stress biomarkers, and novelty preference.

Our primary analysis investigated the relationship between the summed phthalate biomarkers ($\mathrm{\Sigma }\mathrm{DEHP}$, $\mathrm{\Sigma }\mathrm{DINP}$, and $\mathrm{\Sigma }\mathrm{AA}$), MEP, and oxidative stress biomarkers using two approaches. First, we used linear mixed effect regression models to account for the longitudinal study design and repeated measures of oxidative stress biomarkers taken to account for differing levels of oxidative stress across pregnancy.³⁸ Linear mixed effect models included a random intercept for participant ID. Second, we used linear regression models to examine associations between individual phthalate exposure biomarkers and averaged oxidative stress biomarkers, allowing us to create a more robust measure of oxidative stress across pregnancy. To calculate the average, we took the geometric mean of oxidative stress biomarker concentrations across the two urine collection time points, if one sample was available per participant, we used that value. Linear mixed-effect models were minimally adjusted for gestational age at sample collection. Maternal age, maternal education, and pre-pregnancy body mass index (BMI in kg/m²) were retained as covariates in fully adjusted models (both linear mixed effect and linear regression) and were identified using a Directed Acyclic Graph (DAG; Figure S1). We checked our standard regression assumptions by examining qq-plots for each model. As a secondary analysis, we additionally examined the association between the 10 individual urinary phthalate metabolites (restricted to those with >60% concentrations above the LOD) and oxidative stress biomarkers.

To assess whether our previously observed associations between phthalate exposure biomarkers and novelty preference operate through oxidative stress, we estimated natural direct and indirect effects using the mediation R package.^45,46 The goal of this analysis is to quantify the extent to which the association of phthalate exposure with novelty preference is mediated by oxidative stress. As shown in Figure 1, the direct effect describes the effect of an individual exposure on novelty preference, or c’. The indirect effect is comprised of two parts: the effect of an individual phthalate exposure biomarker on an individual oxidative stress biomarker (i.e., the mediator), or a, estimated using the model: $oxidativestress\sim phthalate+covariates$, and the effect of the mediator on novelty preference, or b, estimated using the model: $cognition\sim phthalate+oxidativestress+covariates$. The two coefficients were multiplied together to obtain the indirect effect (i.e., the effect of the exposure on outcome that occurs through the mediator variable, or ab). The total effect is the sum of the direct and indirect effects. During complete mediation, the entire total effect is dependent on the indirect effect (i.e., the direct effect is close to 0). Partial mediation occurs when both direct and indirect effects contribute to the total effect. In mediation models, averaged levels of 8-iso-PGF_2α, 2,3-dinor-5,6-dihydro-8-iso-PGF_2α, and 2,3-dinor-8-iso-PGF_2α were treated as separate, individual mediators, while $\mathrm{\Sigma }\mathrm{DEHP}$, $\mathrm{\Sigma }\mathrm{DINP}$, $\mathrm{\Sigma }\mathrm{AA}$, and MEP were treated as separate, individual phthalate exposure biomarkers. Although PGF₂ was associated with phthalates exposure in our primary analysis, it was not associated with oxidative stress in our prior work in the IKIDS cohort.³⁴ Thus, we did not include PGF_2α as a possible mediator. All models were run on complete cases and p <0.05 was used to indicate statistical significance. Analysis was performed in R version 4.3.1.

Results

Across both analytic samples (analytic sample 1 included those with measured phthalate exposure biomarkers and oxidative stress, and analytic sample 2 included those additionally with information on novelty preference), the average maternal age was 30 years. The majority of participants held a graduate degree (49% for analytic sample 1, 54% for analytic sample 2), and self-identified as non-Hispanic, white (>80% in both analytic samples). The distribution of infant sex was equal across analytic samples. The average novelty preference in analytic sample 2 was 56.8% (Table 1).

Table 1:

Distribution of demographic characteristics and visual recognition memory outcomes in the Illinois Kids Development Study.

	Analytic Sample #1 (N=225)	Analytic Sample #2 (N = 143)
Maternal Age at Delivery (years)
Mean (SD)	30.4 (3.9)	30.8 (3.9)
Pre-pregnancy Body Mass Index (kg/m2)
Mean (SD)	26.1 (6.3)	26 (6.4)
Missing	4 (1.7 %)	1 (0.7 %)
Maternal Education
Less than College Degree	31 (13.8_%)	15 (10.5 %)
College Degree	84 (37.3_%)	51 (35.7%)
Graduate Degree	110 (48.9_%)	77 (53.8 %)
Maternal Race/Ethnicity
Asian/Pacific Islander	13 (5.8_%)	7 (4.9%)
Black	11 (4.9)	<5
Hispanic	<5	<5
Other/Multi-racial	14 (6.2_%)	10 (7.0%)
White	183 (81.3%)	119 (83.2 %)
Infant Sex
Male	103 (45.8%)	64 (44.8 %)
Female	122 (54.2%)	79 (55.2 %)
Parity
1+Births	140 (62.2%)	85 (59.4 %)
No Prior Births	85 (37.8%)	58 (40.6 %)
Novelty preference (%)		56.8 (6.73)

Note: Analytic sample #1 corresponds to the phthalates and oxidative stress analysis. Analytic sample #2 corresponds to the phthalates, oxidative stress, and VRM analysis.

When examining the distributions of individual phthalate biomarkers, we observed that the geometric mean of MEP was slightly larger in analytic sample 1 (geometric mean= 19.41 ng/mL, geometric SD = 8.79) than analytic sample 2 (geometric mean= 16.01 ng/mL, geometric SD = 13.37). For the oxidative stress biomarkers, the geometric mean for 2,3-dinor-5,6-dihydro-8-iso-PGF_2α was lower in analytic sample 1 (geometric mean= 0.47 ng/mL, geometric SD = 5.30) relative to analytic sample 2 (geometric mean= 0.57 ng/mL, geometric SD = 4.89) (Table 2). The distributions of $\mathrm{\Sigma }\mathrm{DEHP}$, $\mathrm{\Sigma }\mathrm{DINP}$, $\mathrm{\Sigma }\mathrm{AA}$, 8-iso-PGF_2α, 2,3-dinor-8-iso-PGF_2α, and PGF_2α were similar across analytic samples. Individual phthalate biomarker distributions are shown in Table S1. Pearson correlations coefficients showed phthalate biomarkers were moderately to strongly correlated with each other and weakly to moderately correlated with oxidative stress biomarkers (Figure 2).

Table 2:

Distribution of urinary, specific gravity corrected, phthalate exposure biomarkers at 16–18 weeks gestation, and oxidative stress biomarkers, averaged across 16–18 and 22–24 weeks gestation, in the Illinois Kids Development Study (2014–2016).

		Percentile
Phthalate exposure biomarkers	Geometric Mean (Standard Deviation)	5%	25%	50%	75%	95%
MEP (ng/mL)
Analytic Sample #1	19.41 (8.79)	5.82	11.45	21.76	40.83	138.74
Analytic Sample #2	16.01 (13.37)	5.29	11.50	21.46	39.39	109.48
ΣDEHP
Analytic Sample #1	0.07 (2.48)	0.02	0.04	0.07	0.11	0.31
Analytic Sample #2	0.08 (2.69)	0.02	0.05	0.07	0.12	0.38
ΣDINP
Analytic Sample #1	0.05 (3.58)	0.01	0.02	0.04	0.11	0.52
Analytic Sample #2	0.05 (3.44)	0.01	0.02	0.04	0.11	0.50
ΣAA
Analytic Sample #1	0.31 (2.21)	0.08	0.19	0.30	0.48	1.14
Analytic Sample #2	0.31 (2.20)	0.08	0.20	0.30	0.44	1.00
Oxidative stress biomarkers
8-iso-PGF2α (ng/mL)
Analytic Sample #1	1.14 (1.79)	0.43	0.85	1.22	1.60	2.66
Analytic Sample #2	1.12 (1.90)	0.37	0.84	1.2	1.61	2.64
2,3-dinor-5,6-dihydro-8-iso-PGF2α (ng/mL)
Analytic Sample #1	0.47 (5.30)	0.03	0.16	0.73	1.53	4.80
Analytic Sample #2	0.57 (4.89)	0.03	0.18	0.79	1.96	6.08
2,3-dinor-8-iso-PGF2α (ng/mL)
Analytic Sample #1	4.24 (1.87)	1.95	3.27	4.52	6.12	9.19
Analytic Sample #2	4.42 (1.77)	2.24	3.39	4.73	6.18	9.62
PGF2α (ng/mL)
Analytic Sample #1	2.34 (2.28)	0.45	1.52	2.82	3.98	6.95
Analytic Sample #2	2.00 (2.49)	0.41	1.19	2.34	3.73	6.41

Note: Analytic sample #1 corresponds to the phthalates and oxidative stress analysis. Analytic sample #2 corresponds to the phthalates, oxidative stress, and VRM analysis. Values below the LOD were imputed with LOD divided by the square root of 2.

Figure 2:

Pearson correlation coefficients for natural log-transformed urinary phthalate biomarkers and oxidative stress biomarkers corrected with specific gravity, and visual recognition memory outcomes in the Illinois Kids Development Study (N=143).

In fully adjusted linear mixed effect models, we observed that an increase in natural log transformed $\mathrm{\Sigma }\mathrm{DINP}$ was associated with a significant increase in 8-iso-PGF_2α (β= 0.08, 95% confidence interval [CI]= 0.01, 0.15), 2,3-dinor-5,6-dihydro-8-iso-PGF_2α (β= 0.40, 95% CI = 0.23, 0.58), and 2,3-dinor-8-iso-PGF_2α (β= 0.10, 95% CI= 0.03, 0.17) (Figure 3; Table S2). Similarly, $\mathrm{\Sigma }\mathrm{AA}$ was significantly associated with an increase in 2,3-dinor-5,6-dihydro-8-iso-PGF_2α (β= 0.45, 95% CI= 0.14, 0.76), and 2,3-dinor-8-iso-PGF_2α (β= 0.14, 95% CI = 0.01, 0.27). Increasing $\mathrm{\Sigma }\mathrm{DEHP}$ was also associated with an increase in 2,3-dinor-8-iso-PGF_2α and 2,3-dinor-5,6-dihydro-8-iso-PGF_2α but confidence intervals included the null. MEP was associated with a non-significant reduction in 2,3-dinor-8-iso-PGF_2α and 2,3-dinor-5,6-dihydro-8-iso-PGF_2α, and a non-significant increase in 8-iso-PGF_2α (Figure 3; Table S2). Linear regression models, which included averaged oxidative stress biomarkers as the outcome, produced similar results (Table S3).

Figure 3:

Associations between natural log transformed, urinary, specific gravity corrected phthalate biomarkers at 16–18 weeks gestation, and oxidative stress biomarkers at 16–18 weeks and 22–24 weeks gestation, estimated using linear mixed effect models in the Illinois Kids Development Study.

Note: Minimally adjusted for gestational age at study visit and include a random intercept for participant ID and fully adjusted models additionally include maternal education, maternal age, and pre-pregnancy body mass index. Sample sizes for each model are provided in Table S2.

Mediation analysis yielded varying results, with evidence of partial mediation by 2,3-dinor-8-iso-PGF_2α. We observed that $\mathrm{\Sigma }\mathrm{DINP}$ was indirectly associated with novelty preference through 2,3-dinor-8-iso-PGF_2α (indirect effect= −0.25, 95% CI= −0.60, −0.01, proportion mediation = 0.34, 95 % CI = −4.76, 7.92) and 8-iso-PGF_2α (indirect effect= −0.09, 95% CI= −0.34, 0.06, proportion mediation = 0.08, 95% CI = −2.86, 2.62) (Figure 4; Table S4). $\mathrm{\Sigma }\mathrm{AA}$ and $\mathrm{\Sigma }\mathrm{DEHP}$ were also indirectly associated with novelty preference through effects on 2,3-dinor-8-iso-PGF_2α ($\mathrm{\Sigma }\mathrm{AA}$ indirect effect= −0.21, 95% CI= −0.66, 0.07, proportion mediated = 0.16, 95% CI = −1.36, 2.38; $\mathrm{\Sigma }\mathrm{DEHP}$ indirect effect= −0.11, 95% CI= −0.43 0.10, proportion mediated = 0.09, 95% CI = −0.30, 0.93). We did not observe any evidence of mediation by 2,3-dinor-5,6-dihydro-8-iso-PGF_2α (Figure 4; Table S4).

Figure 4.

Stacked bar charts reflecting the direct, indirect, and total effects from mediation models assessing oxidative stress biomarkers as mediators of the relationship between urinary phthalate biomarkers and novelty preference (N=142).

Note: Beta estimate and 95% confidence intervals for the direct, indirect, and total effects are provided in Table S4. Models adjusted for maternal education, maternal age, and pre-pregnancy body mass index.

Discussion

Among mother child pairs enrolled in the IKIDS prospective cohort, we observed that gestational exposure to phthalates, particularly to DEHP and DINP, was associated with a strong, statistically significant increase in the F₂-isoprostane urinary metabolites 2,3-dinor-5,6-dihydro-8-iso-PGF_2α and 2,3-dinor-8-iso-PGF_2α. Furthermore, we demonstrate that 2,3-dinor-8-iso-PGF_2α acts as a partial mediator of the relationship between exposure to some phthalates and recognition memory in infancy. Findings from our study contribute to the growing body of literature recognizing oxidative stress – and specifically lipid peroxidation – as a biological mechanism on the causal pathway between environmental exposures and cognitive outcomes in childhood.

Our finding, which shows gestational phthalate exposure is associated with oxidative stress during the gestational period, is consistent with findings from other birth cohorts. For example, within the Puerto Rico Test site for Exploring Contamination Threats (PROTECT) and LIFECODES longitudinal cohorts, increasing urinary concentrations of individual metabolites of DEHP, DINP, and AA were associated with an increase in 8-iso-PGF_2α.^27,47 These results largely support what we observed with the 8-iso-PGF_2α parent compound and its metabolites in IKIDS. Similarly, there was a positive relationship between $\mathrm{\Sigma }\mathrm{DEHP}$ and 2,3-dinor-5,6-dihydro-8-iso-PGF_2α within The Infant Development and the Environment Study (TIDES).²⁹ However, the TIDES study also observed that increasing urinary concentrations of MEP were associated with 2,3-dinor-5,6-dihydro-8-iso-PGF_2α, which contrasts with what we observed in the present analysis. It is possible that these differences relate to underlying differences in phthalate exposures, as concentrations of MEP were much lower in our IKIDS cohort relative to TIDES. Notably, we observed that PGF_2α levels had a negative association with all phthalate exposure biomarkers evaluated. Because PGF_2α differs from other F₂-isoprostnaes in that it is a product of the “enzymatic” oxidation of arachidonic acid via cyclooxygenase (COX-1 and COX-2), it may thereby be more reflective of inflammation as opposed to oxidative stress and explain differences in patterns of effect across oxidative stress biomarkers.⁴⁸

Our study is unique in that we examined urinary oxidative stress biomarkers as mediators of the relationship between exposure to phthalates and novelty preference, representing an advancement over prior studies which have focused solely on the relationship between exposure to outcome, exposure to mediator, or mediator to outcome. In our mediation models, we observed that the effect of $\mathrm{\Sigma }\mathrm{DINP}$, $\mathrm{\Sigma }\mathrm{AA}$, and $\mathrm{\Sigma }\mathrm{DEHP}$ on novelty preference operated, in part, via 2,3-dinor-8-iso-PGF_2α. These findings are of critical importance, as DEHP is one of the most widely used phthalates and is the dominant plasticizer.⁴⁹ Notably, $\mathrm{\Sigma }\mathrm{DINP}$ and $\mathrm{\Sigma }\mathrm{AA}$ were more strongly associated with oxidative stress, which acted as a partial mediation of the relationship between these phthalate exposures and novelty preference, relative to $\mathrm{\Sigma }\mathrm{DEHP}$ where effects were not as strong. This supports the notion that DINP is potentially more toxic than other major phthalates, a conclusion of particular concern, as DINP has only been regulated in children’s toys, and it is one of the possible alternatives to DEHP following several regulatory guidelines limiting the presence of DEHP in consumer products. A closer look at the metabolites included in $\mathrm{\Sigma }\mathrm{AA}$ and $\mathrm{\Sigma }\mathrm{DINP}$ phthalate exposure shows that the metabolite MCOP appears to drive the association with oxidative stress. Additionally, many DINP, DEHP, and other phthalates have anti-androgenic properties with shared mechanisms of action, which supports our findings that $\mathrm{\Sigma }\mathrm{AA}$ impacts novelty preference through 2,3-dinor-8-iso-PGF_2α. However, we did not observe strong evidence of mediation by other oxidative stress biomarkers. Phthalate metabolites interact with a myriad of systems in addition to androgen, that interfere with endogenous hormones critical to female reproduction and fetal development. Prior work shows that gestational exposure to phthalates can alter both the maternal and fetal hypothalamic-pituitary-adrenal axis, inducing the dysregulation of glucocorticoids.^47,50,51 Furthermore, studies have shown that some phthalates, namely DEHP, affect metabolism and promote systemic inflammation.^52–55 Altering the natural balance of these systems can induce oxidative stress, specifically lipid peroxidation, that may influence fetal brain development through mechanisms of malnutrition, reduction of placental perfusion, and impaired neurogenesis and synaptic plasticity. ^33,56,57 Taken together, a complex interplay of pathways underlies our previously observed associations between phthalate exposure and recognition memory in infancy and can explain the differences of mediation between oxidative stress biomarkers.²³

Findings from our study should be interpreted in light of strengths and limitations. First, the analytic sample was comprised of predominantly non-Hispanic white participants who had a college or graduate degree. This limits our external generalizability to other study populations, and it is possible that our findings would be greater in magnitude among those populations who experience the highest phthalate exposures as a result of socioeconomic deprivation.^58–61 Additionally, our study was limited by the cross-sectional study design and relatively small sample size. Within our analysis, phthalate metabolites were measured at 16–18 weeks, while oxidative stress biomarkers were measured at 16–18 weeks on all participants and at 22–24 weeks for a smaller subset. In mediation analysis, we used averaged oxidative stress biomarker levels as the mediator (for those participants who had only one oxidative stress biomarker measurement, we used that measure). Future studies employing a true longitudinal design are needed to confirm our findings. Additionally, the half-life of phthalate metabolites is short and a single measure during early pregnancy may not reflect exposure across gestation. However, we note that exposure to at least certain phthalates used in personal care products is likely consistent over time.

Our study also has numerous important strengths. We used an automated VRM task to assess recognition memory via infant looking behavior measured by an infrared eye tracker. This provides highly accurate assessment of looking behavior vs older methods of videotaping the infant and hand scoring the length of time they are looking at each stimulus. Prior studies demonstrated that eye-tracking methods and novelty preference are predictive of memory and cognitive function later in childhood, underscoring the importance of using these methods in early infancy, when it may be possible to intervene and prevent downstream adverse effects or provide early treatment to disorders that impact learning.^62–65 For instance, studies demonstrate that VRM methods have potential in diagnosis of ADHD and autism.^66,67 Finally, we included multiple biomarkers of F₂-isoprostanes, which are considered to be the ‘gold standard’ biomarkers of oxidative stress as they are unaffected by dietary lipid content and are stable throughout the day.⁶⁸

Conclusions

Within a subset of participants enrolled in the IKIDS cohort, we show that gestational exposure to some phthalates was associated with an increase in oxidative stress during early to mid-pregnancy. Our findings revealed that the relationship between $\mathrm{\Sigma }\mathrm{DINP}$ urinary concentrations on novelty preference operated, in part, through effects on 2,3-dinor-8-iso-PGF_2α. Given that oxidative stress is a common intermediary in chemical toxicity, we highlight the need for potential mediation analysis of other endocrine disrupting chemicals that can also affect health and development. Early life exposures to phthalates are preventable; understanding the effects of these exposures elevates our capacity to address underlying factors that increase susceptibility to neurodevelopmental disorders, guide risk assessment strategies, and implement targeted interventions. Future studies conducted in racially, ethnically, and socioeconomically diverse study populations are needed to confirm our findings. Additionally, future work would benefit from considering statistical approaches for evaluating exposure mixtures and multiple mediators, as our study was limited to single pollutant effects.

Highlights.

• We examined gestational phthalate exposure in relation to oxidative stress

• Oxidative stress partially mediated association between phthalates and cognition

• Increasing DINP was associated with increasing oxidative stress biomarkers