40. EXPLORING THE CUMULATIVE IMPACT OF RARE VARIANTS IN AUTISM

Background While de novo variants have a disproportionate contribution to autism risk, a substantial proportion of autistic individuals do not carry such variants. Recent large-scale studies supported the contribution of rare inherited variants to autism. We are interested in evaluating the cumulative impact of rare variants in autism. Methods We explored this question using 7,663 exome sequenced quartet families, each consisting of a father, mother, autistic child, and undiagnosed sibling in the Simons Foundation Autism Research Initiative (SFARI) datasets. Among these families, 2,398 siblings (subgroup 1) shared a rare (< 1%) exonic copy number variant (CNV) or high-confidence protein-truncating variant (PTV) in constrained genes (top 30% of loss-of-function observed/expected upper bound fraction (LOEUF) scores), and 248 siblings (subgroup 2) shared such variants in genes implicated in autism and developmental disorder (DD). We used sHet (selection against heterozygous PTVs) to create an exome-wide burden score (sHet burden score) that quantified the cumulative impact of rare variants in different variant classes (rare exonic deletions, rare exonic duplications, PTVs, high-impact missense variants (Missense deleteriousness Prediction by Constraint (MPC) ≥ 2 and AlphaMissense pathogenicity ≥ 0.97), and synonymous variants). We used conditional logistic regression to evaluate the association between an autism diagnosis and these sHet burden scores in sibling pairs while controlling for inferred genetic ancestry, age, and sex. Results We found that an autism diagnosis was associated with higher sHet burden scores made using rare exonic deletions (odds ratio (OR)=9.24, FDR-adjusted p-value (p)=6.90e-12), rare exonic duplications (OR=4.37, p=2.70e-12), PTVs (OR=6.45, p=8.97e-14), high-impact missense variants (OR=2.89, p=5.23e-5), and the combination of all four variant classes (combined sHet burden score, OR=6.11, p=1.83e-36). Compared to the count of variants in each of the above variant classes, exome-wide or in constrained genes only, sHet burden scores yield higher ORs, indicating autistic children tended to carry not only more, but also stronger impact, variants. Excluding de novo variants from the combined sHet burden score, the OR reduced to 1.88 but remained statistically significant (p=0.0002). Among subgroup 1 sibling pairs, an autism diagnosis was associated with higher overall (OR=7.67, p=2.69e-11) and inherited (OR=2.71, p=0.007) combined sHet burden scores. Among subgroup 2 siblings, with a smaller sample size, we found suggestive evidence that an autism diagnosis was associated with a higher overall (OR=7.01, p=0.039) combined sHet burden score. Discussion Even when sharing a rare exonic CNV or high confidence PTV in genes under strong evolutionary constraint or implicated in autism or DD, the cumulative impact of rare variants measured by sHet burden scores was higher among children with an autism diagnosis, highlighting the importance of studying the joint impact of multiple rare variants in autism. In addition to the analyses presented here, we are expanding our work to incorporate common variants and richer phenotype information.