Impact of Environmental Chemical Mixtures on Maternal and Child Mental Health

Abstract

Background: Internalizing disorders, such as anxiety and depression, are of public health concern. The prevalence of these psychiatric illnesses has been increasing in recent decades, with vulnerable groups including children and young adults, as well as women; and this rise in diagnoses has individual and societal implications, as internalizing symptoms are associated with adverse social, emotional, and cognitive development; enormous clinical and economic burden; and higher risk of lifelong physical and mental morbidity. Although there is evidence that exposure to neurotoxicant chemicals (e.g., organochlorines, metals) may heighten sensitivity to internalizing symptoms throughout the life course, research specifically on the impact of prenatal exposures has been limited by 1) the duration of prospective follow up of these symptoms and lack of consideration of 2) co-exposure to several chemicals and 3) effect modification by non-chemical stressors. For studies of children’s health, few have assessed associations between multiple prenatal chemical exposures and internalizing symptoms evaluated beyond mid-childhood, even though the core risk period for anxiety and depression has not yet occurred. With respect to adult mental health, most studies are cross-sectional and consider a single chemical exposure. Furthermore, there is scarce literature on associations between exposures to chemical mixtures during pregnancy and internalizing symptoms in women, even though the pre- and postnatal period may be particularly sensitive time for the development of mood disorders. The purpose of this dissertation was to identify modifiable prenatal environmental risk factors for internalizing disorders among women and children; and in the studies of outcomes from mid-childhood to early adulthood, to examine whether non-chemical stressors modify the effects of prenatal chemical exposures on internalizing symptomology. We did this by 1) evaluating the overall associations of prenatal exposure to a chemical mixture of neurotoxic organochlorines and metals with internalizing symptoms in mid-childhood, adolescence, and early adulthood, and exploring whether these associations were modified by non-chemical factors hypothesized to alter sensitivity to chemical exposures; and 2) assessing the overall associations between early pregnancy exposure to a metal mixture and maternal prenatal and postpartum depressive symptoms. Methods: Participants were mother-child pairs enrolled in two separate Massachusetts’ birth cohort studies. Among participants whose mothers resided near the New Bedford Harbor Superfund site at birth (1993-1998), we evaluated associations of prenatal organochlorine [hexachlorobenzene (HCB), p,p'-dichlorodiphenyl dichloroethylene (p,p’-DDE), polychlorinated biphenyls] and metal [lead (Pb), manganese (Mn)] exposures with anxiety-, depressive-, and/or somatic-related symptoms measured in mid-childhood via the Conners’ Parent Rating Scale, in adolescence via the Behavior Assessment System for Children, Second Edition, Self-Report of Personality, and in early adulthood via the Generalized Anxiety Disorder Scale and Patient Health Questionnaire. We utilized co-exposure- and covariate-adjusted regression approaches to estimate associations between chemicals and internalizing outcomes, and either Bayesian kernel machine regression (BKMR) or quantile g-computation to explore the overall effect of the five-chemical mixture. In parametric and mixture modeling, we explored whether associations were modified by non-chemical stressors. Among women living in Eastern Massachusetts during pregnancy (1999-2002), we investigated associations of early pregnancy metal (arsenic, barium, cadmium, cesium, copper, mercury, magnesium, Mn, Pb, selenium, zinc) exposures with prenatal and postpartum depressive symptoms, measured at three timepoints via the Edinburgh Postnatal Depression Scale. We used co-exposure- and covariate-adjusted regression methods to estimate associations between metals and depressive symptoms and symptom trajectories, and quantile g-computation to characterize the overall effect of the mixture of essential and non-essential metals. Results: The presence and strength of associations between prenatal chemical exposures and internalizing symptoms differed when assessed in mid-childhood, adolescence, and early adulthood. There were no consistent overall or sex-specific associations between prenatal organochlorines and internalizing symptoms in mid-childhood and adolescence. However, when considering effect modification by non-chemical stressors, strata-specific associations between prenatal HCB and p,p’-DDE and risk of elevated anxiety and depressive symptoms in early adulthood were apparent. Although no associations were observed with mid-childhood internalizing symptoms, prenatal Pb exposure was adversely associated with anxiety symptoms in adolescence and early adulthood. Prenatal Mn was adversely associated with anxiety and depressive symptoms for girls, but not boys, in mid-childhood and adolescence, whereas no associations were present with elevated symptoms in early adulthood. While confidence intervals mostly included the null for BKMR and quantile g-computation estimates, the overall organochlorine and metal mixture was positively associated with anxiety and depressive symptoms in mid-childhood, adolescence, and early adulthood, with associations varying by non-chemical stressor groups; and negatively associated with somatic symptoms from mid-childhood through adolescence. In both parametric regression and quantile g-computation models, there was little evidence that the early pregnancy metal mixture was associated with elevated maternal depressive symptoms at mid-pregnancy and 6 and 12 months postpartum. Conclusions: This dissertation supports evidence for adverse associations of prenatal exposures to organochlorines and metals with subsequent anxiety and/or depressive symptoms in child participants, but not with maternal prenatal and postpartum depressive symptoms. It demonstrates the importance of the 1) informant and timing with respect to outcome assessment of internalizing symptoms, since these outcomes are inward emotions that are oftentimes not apparent to outsiders and tend to clinically manifest in adolescence or beyond; 2) evaluation of multiple co-exposures, as humans do not experience these chemicals in isolation; and 3) consideration of effect modification by non-chemical stressors, which may reveal adverse associations otherwise not apparent in models only including main effects.