Association Studies of Environmental Exposures, DNA Methylation and Children's Cognitive, Behavioral, and Mental Health Problems.

Prenatal environmental exposures have been associated with children's cognitive, behavioral, and mental health problems, and alterations in DNA methylation have been hypothesized as an underlying biological mechanism. However, when testing this hypothesis, it is often difficult to overcome the problem of multiple comparisons in statistical testing when evaluating a large number of developmental outcomes and DNA methylation sites as potential mediators. The objective of this study is to implement a 'meet-in-the-middle' approach with a sequential roadmap to address this concern. In the Columbia Center for Children's Environmental Health birth cohort study, we implemented a 5-step sequential process for identifying CpG sites that mediate associations between prenatal environmental exposures and cognitive, behavioral, and mental health problems as measured by the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) and the Child Behavior Checklist (CBCL). These steps include 1) the identification of biological pathways that are relevant to each outcome of interest; 2) selection of a set of genes and CpGs on genes that are significantly associated with the outcomes; 3) identification of exposures that are significantly associated with selected CpGs; 4) examination of exposure-outcome relationships among those where significant CpGs were identified; and 5) mediation analysis of the selected exposures and corresponding outcomes. In this study, we considered a spectrum of environmental exposure classes including environmental phenols, pesticides, phthalates, flame retardants and air pollutants. Among all considered exposures and outcomes, we found one CpG site (cg27510182) on gene (DAB1) that potentially mediates the effect of exposure to PAH on CBCL social problems at children aged 7. This 'meet-in-the-middle' approach attenuates concerns regarding multiple comparisons by focusing on genes and pathways that are biologically relevant for the hypothesis.