Distinct Sources of Genetic Risk for Autism from Social and Non-Social Domains

Background: Common genetic variation contributes to a substantial proportion of the variation in autism liability. There is considerable phenotypic and genetic heterogeneity within the autism spectrum. It is unclear how common genetic variation in social and non-social domains in the autism spectrum contribute to the liability in the condition.

Objectives: To investigate the genetic architectures of social traits (self-reported empathy, cognitive empathy, social and communication difficulties, and social relationship satisfaction) and non-social traits (systemizing) related to autism in large cohorts and map their contribution to autism liability.

Methods: We investigated the contribution of social and non-social traits for autism using GWAS for multiple traits (5000 > N > 140000) in three cohorts: ALSPAC, 23andMe, and the UK Biobank. Self-reported empathy was measured using the Empathy Quotient (EQ), cognitive empathy was measured using the Reading the Mind in the Eyes Test, Social and Communication Difficulties was measured using the Social and Communication Disorders Checklist, Social Relationship Satisfaction using two brief questionnaires: family relationship satisfaction and friendship satisfaction, and Systemizing was measured using the Systemizing Quotient-Revised (SQ-R).

Results: Using GWAS autism data from the iPSYCH consortium and the PGC, we identified a positive genetic correlation between autism and higher systemizing, lower social relationship satisfaction, and lower self-reported empathy. However, there is no genetic correlation between systemizing and the social traits i.e. social relationship satisfaction, self-reported empathy, and social and communication disorders. We further demonstrate that social autistic traits are positively genetically correlated with each other. Both social and non-social traits are genetically correlated with measures of intelligence, with significant negative genetic correlations between social traits and measures of intelligence and significant positive genetic correlations between systemizing and measures of intelligence. Finally, we demonstrate that the positive genetic correlation between systemizing and autism is independent of the genetic correlation with measures of intelligence.

Conclusions: Our results suggest that there are at least two independent sources of genetic liability for autism stemming from social and non-social traits. This is the first empirical evidence of distinct pathways to autism liability from common genetic variants.