DNA Methylation As a Biomarker for Prenatal Exposures Implicated in Autism Spectrum Disorders

Background: Environmental exposures have recently been shown to play a larger role in autism spectrum disorders (ASD) than previously thought. For example, reports have shown prenatal exposure to select serotonin reuptake inhibitors (SSRIs) is associated with an increased risk for ASD (Croen et al. 2011; Rai et al. 2013). However, most population-based autism studies lack appropriate and reliable information on prenatal exposure status, making it difficult to assess and definitively determine the relationship between ASD and exposure status. A molecular signature of prenatal exposures, present during childhood, which could serve as a proxy for prenatal exposure when investigating exposure-disease relationships in lieu of having actual prenatal exposure data would be of great utility. Animal models have shown DNA methylation (DNAm), a type of epigenetic mark, is a useful biomarker of environmental exposure. Here we plan to assess the feasibility of DNAm as a biomarker of 6 specific prenatal exposures that have been implicated in ASD. 

Objectives: The purpose of this study is to define DNAm signatures, present in blood during childhood, that are associated with prenatal exposure to smoking, alcohol, maternal infection, folic acid, SSRIs, and B2AR medications.

Methods: Among 611 children (both ASD cases and population controls), aged 3-5 years, enrolled in the Study to Explore Early Development (SEED) we used the Infinium450 BeadChip to measure DNAm at over 485,000 loci. Rigorous quality control measures were implemented, leaving 609 children available for downstream methylation analyses. Prenatal exposure data were collected via telephone interview with the mother. For each exposure examined in this study, we created a dichotomous ‘exposed’ versus ‘unexposed’ indicator for each trimester and for pregnancy, as a whole, for our primary analyses. For each exposure, we plan to perform regression analyses, adjusting for blood cell composition estimates, race, and age, to identify novel associations with DNAm. Additionally, for smoking, we have analyzed 26 loci shown to have significant correlation between maternal cotinine levels during pregnancy and DNAm levels at birth (Joubert et al. 2012); thus, we examined the persistence of these DNAm changes into childhood. 

Results: Our preliminary analysis indicates the approximate frequencies of maternal active smoking, alcohol, folic acid, B2AR, and SSRI use, as well as infection at any point during pregnancy, are 8%, 12%, 90%, 8%, 8%, and 29%, respectively. Our examination of the 26 smoking associated loci, previously identified in birth samples, revealed striking concordance between the current and the previous study in DNAm changes associated with exposure status. These changes appear to be tobacco specific since the changes in DNAm at these 26 sites were not replicated when we examined prenatal exposure to infection, alcohol, or B2AR or SSRI medications. We are currently performing analyses to define DNAm signatures for all exposures and plan to include any additional findings in our presentation. 

Conclusions: We show for the first time that a DNAm signature identified at birth associated with prenatal tobacco exposure persists in 3-5 year old children; thus, demonstrating blood-derived DNAm measurements obtained during childhood can retain signatures of in utero exposures.