An Integrative Epigenome-Wide Autism Association Study

Background: It is becoming clear that in order to understand the complex molecular architecture of autism it is important to examine and relate specific environmental exposures, genome-wide genetic, and epigenomic data. While the value of an integrated approach is now widely recognized, very few autism studies have compiled this information, especially from the same individuals. The Study to Explore Early Development (SEED) is one of the only case-control epidemiologic studies of autism with comprehensive phenotypic evaluation, broad prenatal environmental exposure information, genome-wide genotyping data, and whole blood available for epigenomic measurements, from the same individuals. Thus, we utilize samples from SEED to carry out the first large-scale effort to integrate genetic, environmental exposure, and epigenetic data and further our understanding of the molecular basis of autism.  

Objectives: The overall purpose of this study is to identify sites of altered DNA methylation (DNAm) associated with autism and relate them to genome-wide genetic and prenatal environmental exposure data from the same individuals. In addition, we plan to evaluate DNAm as a potential biological mechanism for gene-environment interactions identified in our parallel SEED gene-environment-wide interaction study, using measurements from the same individuals.  

Methods: Genotypes for 606 autism cases and 742 controls were measured using the Illumina 1M-Quad and Affymetrix Axiom arrays. Several quality control steps were implemented, at both the sample and SNP level, and SNP imputation was performed with IMPUTE2 using all individuals in the 1000 Genomes Project as a reference, resulting in genotyping data at over 13 million loci for 1,348 SEED children. We will focus on 5 specific in utero exposures (collected via maternal self-report using a structured interview) including maternal use of tobacco, alcohol, β-2 adrenergic receptor agonist or antidepressant medications, and maternal infection. For 600 SEED children, we will measure DNAm using the Illumina Infinium 450K methylation platform. A data quality control pipeline, already developed, will be implemented to remove potential methylation measurement errors at both the sample and locus level. Analyses to identify sites of altered DNA methylation associated with autism, environmental exposures, and genotypes will be performed using a previously described analytic framework and generalized regression models.  

Results: We are currently measuring DNA methylation for 600 SEED children (292 cases and 318 controls) with comprehensive phenotyping, genome-wide genotyping, and prenatal environmental exposure data already available. Analyses will be performed to: (1) identify genomic regions associated with altered methylation in autism; (2) find differentially methylated regions (DMRs) of the genome associated with specific in utero environmental exposures; (3) correlate autism and exposure associated DMRs with genotypes; and (4) assess DNAm as a potential biological mechanism for gene-environment interactions recently identified in our parallel SEED gene-environment interaction study, using data from the same individuals. Results from these analyses will be presented at the conference.  

Conclusions: We present the first autism study to examine and relate genome-scale methylation, genotyping, and in utero environmental exposure data from the same individuals. Our integrated approach will likely provide a more comprehensive understanding of the molecular underpinnings of autism.