Game of Exomes: Battle of the Rare Variants for Association with Autism Spectrum Disorder

Background:  Autism spectrum disorder (ASD) is a highly prevalent developmental disorder, affecting an estimated 1 in 88 individuals, and is associated with significant morbidity.  Despite demonstrating high heritability estimates, only a small proportion of the genetic risk for ASD is explained.  While previous research, including genome wide association studies, has focused on the effects of common variants, recent exome sequencing studies suggest that rare variants (RVs) contribute significantly to ASD risk.  

Objectives:  To identify RVs associated with ASD using two approaches; 1. An agnostic approach testing all genes on the Exome Array for association with ASD 2.  A candidate gene approach testing ~1,000 genes previously implicated in ASD for association with ASD, including 254 genes previously reported as associated with ASD in a gene-based analysis using whole exome sequencing data.

Methods:  Participants were drawn from a large, family-based study of ASD and included 995 unrelated cases and 650 controls that were genotyped using the Illumina HumanExome-12v1 Array.  To identify genes associated with ASD we performed gene-based analyses testing autosomal genes genotyped on the Exome Array for association with ASD.  The sequence kernel association optimal test (SKAT-O) was used for these analyses.  All analyses were performed within our genetically-defined White, Non-Hispanic sample.  A Bonferroni correction was made to adjust our alpha levels of significance.

Results:  We tested a total of 12,922 genes across the genome for association with ASD in our sample.  While no genes were statistically significant, 14 genes were nominally associated with ASD (p<0.001).  Of the ~1,000 ASD candidate genes tested, 19 were nominally associated with ASD (p<0.01), with the most significantly associated being adenylate cyclase 5 (ADCY5) (SKAT-O p=0.002) and dynein, axonemal, heavy chain 5 (DNAH5) (SKAT-O p=0.001).  

Conclusions:  While our study failed to identify RVs to be significantly associated with ASD, we did identify several nominally significant associations providing suggestive evidence for a role of RVs in ASD risk.  These suggestive associations include ADCY5 which is an excellent ASD candidate as it plays an important role in G-protein signaling and two separate reports have identified cases with de novo mutations in the gene.  Further investigation into the role of RVs in ASD with larger sample sizes is necessary.