Prenatal Exposure to PM2.5 Species and DNA Methylation in Newborns: a Novel Statistical Framework

• Activity Number: 260 - SPEED: Environmental Statistics Methods and Applications, Part 2
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 3:05 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Jenny Lee* and Tamar Sofer and Andres Cardenas and Brent A. Coull
• Companies: Harvard School of Public Health and Brigham and Women’s Hospital, Harvard Medical School and University of California, Berkeley - School of Public Health and Harvard T. H. Chan School of Public Health
• Keywords: A-clustering; Sparse Canonical Correlation Analysis; Epigenetics; Global Test; Particulate Matter

Abstract

Exposure to ambient particulate matter during gametogenesis can result in epigenetic perturbation of the developing fetus, which in turn can negatively impact health throughout the lifecourse. Our goal is to assess associations between exposure to PM2.5 species and DNA methylation in newborns. To achieve the goal of quantifying the association between high-dimensional epigenomic outcomes and multiple correlated exposures, we propose a two-stage analysis pipeline that identifies a subset of PM2.5 species and a subset of DNA methylation probes that are associated with each other as follows: 1) apply an Adjacent Site Clustering (A-clustering) algorithm that clusters correlated neighboring CpG sites, and 2) analyze the association between each cluster of regions and PM2.5 species concentrations using Sparse Canonical Correlation Analysis (sCCA). We evaluate the performance of this two-stage analysis when using a variety of penalty functions in the sCCA. Through simulation studies, we show that the proposed method yields greater efficiency relative to existing methods.