Grantee Research Project Results

1) Determine whether prenatal exposures to the endocrine disrupters, polycyclic aromatic hydrocarbon (PAH) and bisphenol A (BPA), are associated with adverse neurobehavioral outcomes in peri-pubertal children, as measured by diagnostic assessment of child psychopathology and cognitive functioning.

 

2) Determine whether prenatal exposure to PAH or BPA is associated with epigenetic changes in umbilical cord white blood cells (DNA methylation validated by gene expression) in candidate genes/ pathways is associated with endocrine disruption and immune dysregulation, and whether altered methylation and gene expression in these candidates is associated with the neurobehavioral outcomes.

 

3) Using GIS, determine the extent to which neighborhood-level conditions contribute to neurobehavioral outcomes and/or moderate the individual-level associations between exposure to PAH or BPA and child neurodevelopment.

Aim 1: PAH exposure

 

During this grant period we have continued our analysis on the impact of prenatal PAH exposure on the neurodevelopment of our cohort as they grow and develop. At age 5 years, using the Wechsler Preschool and Primary Scale of Intelligence (WPPSI), after adjustment for potential confounders, we found that children highly exposed to PAH(> the median of 2.26 ng/m³) had Full-scale and Verbal IQ scores that were 4.31 and 4.67 points lower, respectively, compared to less exposed children (Perera, et al., 2010). These reductions are similar in magnitude to the effects of low-level lead exposure and show that prenatal exposure to PAH at levels encountered in NYC air adversely affected child IQ scores at 5 years of age

 

Using a more informative scale for older children, the Weschler Intelligence Scale for Children (WISC), the same children were tested for IQ at ages 7 and 9. After adjusting for potential confounders, high PAH exposure was inversely associated with Full-scale IQ and Working Memory at ages 7 and 9 and with Verbal IQ and Processing Speed at age 9. It is known that toxic exposures are higher in communities of color, and frequently co-occur with chronic economic deprivation. With this knowledge we examined the interaction between prenatal PAH exposure, measured by presence or absence of adducts in cord blood, and material hardship, based on questions regarding housing, food, electricity, and clothing given prenatally, and child age 6 months, and 1,2, 3, and 5 years, on child IQ at 7 years. After adjustment for potential confounders we found significant (p < 0.05) inverse effects of PAH exposure on Full-scale IQ , Perceptual Reasoning, and Working Memory, in children of mothers with material hardship during pregnancy or recurring material hardship during the child’s early years, and not in children whose mothers did not experience maternal hardship. Additionally, there was a significant interaction (p < 0.05) between high cord adducts and prenatal hardship on Working Memory and between high cord adducts and recurrent material hardship (Vishnevetsky, et al., 2015). These results indicate a need for multifaceted approach to intervention.

 

Continuing our work on PAH exposure and neurodevelopmental outcomes we found that high PAH exposure, measured as number of DNA adducts specific to benzo[a]pyrene (BaP), a representative PAH, in in cord blood, was significantly (p < 0.05) positively associated with symptoms of anxiety/depression at age 4.8 and with attentional problems at ages 4.8 and 7 years on the Child Behavior Checklist (CBCL) after adjusting for potential confounders. Additionally, higher PAH exposure was associated with 8.30 times the odds of having a borderline or clinical classification on the Diagnostic and Statistical Manual of Mental Disorders (DSM)-oriented Anxiety Problem Scale at age 4.8 years (p < 0.037) (Perera, et al., 2011).

 

Further examining the relationship between prenatal PAH exposure and symptoms of Anxiety/Depression and Attentional Problems on the CBCL, we used both prenatal air monitoring data and measurement of DNA adducts specific to BaP a representative PAH, in both maternal and cord blood to determine if the relationship persisted using additional markers of PAH exposure. In our model, after adjusting for potential confounders, we found that high prenatal air monitoring, maternal BaP adducts and cord BaP adducts were significantly (p < 0.05) positively associated with symptoms of anxiety/depression and attentional problems on the CBCL measured at ages 6-7. High PAH measured via prenatal air monitoring was associated with 4.59 times the odds of having borderline or clinical classification on the DSM (p < 0.009) (Perera, et al., 2012). These consistent results provide evidence that environmental PAH at levels encountered in NYC air can adversely affect child behavior, especially internalizing problems, including both anxiety/ depression, and attention problems that could affect school performance.

 

These results are consistent with those in the Center's parallel experimental studies that have observed effects of prenatal PAH on anxiety-like behaviors in mice (Project 3).

 

High prenatal PAH exposure, represented by maternal adducts, was also significantly related to attention deficit hyperactivity problems in children at age 9. After adjusting for relevant confounders, high PAH exposure was significantly (p < 0.05) associated with increases on the DSM-IV total, DSM-IV Hyperactive-Impulsive and DSM-IV Inattentive subscales of the Conners’ Parent Rating Scale (CPRS) in addition to increased scores on the Conners' ADHD Global Index (Perera, et al., 2014). These analyses adjusted for important potential variables, including characteristics of the home environment, maternal ADHD and comorbid anxiety/depression in the children.

 

In an analysis of MRI brain scans in 40 cohort children, 20 with high PAH exposure and 20 with low, we assessed the effects of prenatal PAH exposure on brain structure and the cognitive and behavioral correlates of those abnormalities at ages 7-9. There was a dose-response relationship between increased prenatal PAH exposure, air monitoring measured in the third trimester, and reductions of white matter surface in later childhood. These reductions were significantly (p < 0.05) associated with slower information processing speed, on the WISC- IV and numerous CBCL measures, including externalizing problems and DSM ADHD symptoms (Peterson, et al., 2015).

 

Building on previous results from the cohort showing associations between PAH exposure and Anxiety/ Depression symptoms on the CBCL, we evaluated the relationship between PAH, measured via adducts in maternal blood, and deficient emotional self-regulation (DESR) defined as moderate elevations on 3 specific scales of the CBCL (Anxious/Depressed, Aggressive Behavior, and Attention Problems) and social competence (measured by the Social Responsiveness Scale) at multiple time points between ages 3 and 11. Additionally we measured the association of impaired self-regulation with social competence and whether impairment in self-regulation mediates the association of prenatal exposure to PAH with social competence. Using Generalized Estimating Equations to asses PAH on DESR there was a significant interaction between exposure and time (p < 0.05) showed that the developmental trajectory of self-regulatory capacity was delayed in the exposed children. Multiple linear regression revealed a positive association between presence of PAH adducts and problems with social competence (p < 0.04), level of dysregulation and problems with social competence (p < 0.0001), and evidence that self-regulation mediates the association of prenatal exposure to PAH with SRS (p < 0.0007) (Margolis, et al., Submitted to JCPP).

 

 

Aim 1: BPA Exposure

 

We analyzed the association between maternal prenatal bisphenol A (BPA) exposure (dichotomized into high/low groups at the upper quartile) and child behavior at ages 3-4 (assessed via the CBCL), using generalized linear models, adjusting for postnatal BPA exposure and other potential confounders. We observed significant interactions (p < 0.05) between prenatal BPA urinary concentrations and child sex, on Emotionally Reactive, Aggressive Behavior, and Internalizing Problems scales of the CBCL. After stratifying on sex, the BPA effects were positive and significant among boys on Emotionally Reactive and Aggressive Behavior indicating that boys with prenatal BPA exposure in the highest concentration quartile had, on average, more reported symptoms of problems in these areas. In contrast, among girls, high BPA exposure was associated with lower scores for Anxious/Depressed and Aggressive Behavior (p < 0.05), and Internalizing Problems (p < 0.1) indicating that girls in the high prenatal BPA exposure group had, on average, fewer reported problems in these areas than girls in the low exposure group. Postnatal BPA urinary concentration alone only had a significant negative effect only on Emotionally Reactive within the entire sample. Comparison of results before and after adjusting for postnatal BPA exposure found the effect estimates to be similar, suggesting that the prenatal period may be a more sensitive window for BPA exposure.

 

We built on these findings by analyzing the effect of maternal prenatal BPA (dichotomized into high/low groups at the upper tertile) and child behavior at ages 7-9 years (assessed by the CBCL) to determine if the association of sex specific behavior differences persisted. Using a stratified model, adjusted for relevant confounders, including postnatal BPA, boys in the highest BPA tertile had significantly (p < 0.05) higher Internalizing and Externalizing composite scores, and a higher score on their corresponding syndrome scales. Girls in the highest tertile showed a significant decrease (p < 0.05) in the Internalizing composite score (Roen, et al., 2015). Together these results suggest BPA exposure may affect childhood behavioral outcomes in a sex specific manner, and differently depending on timing of exposure.

 

Our recent results at ages 10-12, using outcome measures specific to anxiety and depression, the Revised Children’s Manifest Anxiety Scale (RCMAS) and Children’s Depression Rating Scale (CDRS), respectively, found that maternal prenatal BPA exposure (dichotomized into high/low groups at the upper tertile) is significantly associated (p < 0.05) with increased RCMAS total score, RCMAS social concern scale, and CDRS total scale in boys, but there are no significant associations in girls. These results indicate the persistence of sex specific effects of prenatal BPA exposure on the symptoms of anxiety and depression (Perera, et al., manuscript in prep.).

 

 

Aim 2

 

We have successfully completed pyrosequencing assays to assess the methylation of COX-2, ERα, CEBPα, PDE4D, and TH. We are currently analyzing the association between prenatal PAH and COX-2 and between prenatal BPA and ERα, CEBPα, PDE4D, and TH, respectively. We are also analyzing associations between the methylation of CpG sites on these genes and neurodevelopmental endpoints including mental and psychomotor developmental indices (from the Bayley Scales at ages 12, 24, and 36 months), IQ (from the WISC at age 7) and child behavior (from the CBCL at ages 3, 5, 7, and 11). We are currently working to optimize the analysis of additional CpG sites on the genes listed above and to develop pyrosequencing assays for the remaining gene candidates (NMDAR2b, CCL17, AhR, INF-γ, and THRβ).

 

We subsequently added brain-derived neurotrophic factor (BDNF) to our list of candidate genes based on experimental studies in Project 3 (see progress report for Project 3) and results of our parallel studies in a Chinese cohort showing associations of BDNF with prenatal PAH exposure and cognitive development. Our new approach synchronizes our human and animal data to select additional CpG sites of interest, taking advantage of our parallel experimental animal studies that allow us to assess the concordance between CpG-specific methylation related to either PAH or BPA and gene expression in target tissue, which is not accessible in the human cohort. After checking the homology between mice and humans, we map the sequence of interest and identify ‘indicator’ CpG sites located within or around this region that have been assayed using two genome-wide techniques (Infinium 450K and HELP-tagging) in a subset of our cord blood samples from the cohort. Using these indicator CpGs, we can assess the association between methylation and PAH or BPA exposure, as well as neurodevelopmental outcomes. If we identify a signal between either exposure or neurodevelopment at these indicator CpGs and demonstrate that there is sufficient variation in cord blood at these locations for follow-up analysis, we then develop pyrosequencing assays covering the CpG sites of interest to measure the methylation status of these CpGs in the cord blood samples of the full cohort. This analysis is done on a gene-by-gene basis; we have thus far completed BDNF and are following a similar approach for ERα. We plan to complete remaining genes and data analysis in the coming year.

 

It is widely recognized that because DNA methylation is cell-specific, failure to account for cell distribution when examining methylation in a mixed cell population can confound hypothesized exposure-to-methylation associations, leading to spurious results or failure to detect true relationships. Methods have been developed using regression calibration approach to adjust for cellular mixture requiring reference methylation data. Currently, two reference datasets are now available, both derived from adult blood. Cells represented in these datasets may not reflect the most prevalent cellular components of cord blood, leading some to question whether they are optimal when applied to cord blood. We have generated a 450K reference dataset using cord blood samples and are currently evaluating whether this reference dataset works better at adjusting for cell mixture in a cord blood matrix than the reference datasets from adult blood. This work was presented at the PPTOX meeting and a manuscript is currently in preparation.

 

 

Aim 3

 

The proportion of households below the federal poverty line in a participant's neighborhood (1-km street network buffer) is significantly (p <0.05) associated with reduced scores on the psychomotor development index (PDI) and mental development index (MDI) on the Bayley Scales of Infant Development at age 3 (Lovasi, et al., 2010).

 

We found that the neighborhood context variables, (given by self-report and geocoded into a 1-km network buffer around the home) building dilapidation and linguistic isolation were significantly (p < 0.05) associated with lower performance IQ scores (measured by the Wechsler Pre-school and Primary Scale of Intelligence-Revised) at age 5 (Lovasi, et al., 2014). These data demonstrate that early life PAH and neighborhood-level linguistic isolation exposures independently predict lower cognitive abilities at age.

 

We geocoded 841 prenatal addresses from the study area in Northern Manhattan and The Bronx and are continuing the process for later ages (currently 4,693 addresses geocoded). In preliminary analyses of our GIS work, we found that neighborhood percent poverty confounds the effect of cumulative PAH exposure on hyperactive-impulsive behaviors at age 9. The effect of cumulative PAH exposure is underestimated without adjusting for neighborhood percent poverty, as neighborhood % poverty is positively associated with hyperactive-impulsive behaviors at age 9 (p < 0.01) and negatively associated with cumulative PAH exposure (p = 0.19). The effect of cumulative PAH exposure on hyperactive-impulsive behaviors at age 9 is modified by neighborhood-level educational attainment (p = 0.67) in the low educational attainment stratum, (p = 0.02) in the high educational attainment stratum). However, a formal test of interaction does not reach statistical significance (p = 0.12).

 

We are using the results from the GIS work to inform our future analyses of the associations between endocrine disruptors and neurobehavioral outcomes. These neighborhood-level factors are potential effect modifiers and confounders; and our previous results allow us to build the most accurate models possible.