Abstract
BACKGROUND: Although autism spectrum disorders (ASD) are among the most heritable of all neuropsychiatric syndromes, most affected children are born to unaffected parents. Recently, we reported an average increase of 3-5% over general population risk of ASD among offspring of adults who have first-degree relatives with ASD in a large epidemiologic family sample. A next essential step is to investigate whether there are measurable characteristics of individual parents placing them at higher or lower recurrence risk, as this information could allow more personalized genetic counseling. METHODS: We assembled what is to our knowledge the largest collection of data on the ability of four measurable characteristics of unaffected prospective parents to specify risk for autism among their offspring: (1) sub clinical autistic trait burden, (2) parental history of a sibling with ASD, (3) transmitted autosomal molecular genetic abnormalities, and (4) parental age. Leveraging phenotypic and genetic data in curated family cohorts, we evaluate the respective associations between these factors and child outcome when autism is present in the family in the parental generation. RESULTS: All four characteristics were associated with elevation in offspring risk; however, the magnitude of their predictive power-with the exception of isolated rare inherited pathogenic variants -does not yet reach a threshold that would typically be considered actionable for reproductive decision-making. CONCLUSIONS: Individual specification of risk to offspring of adults in ASD-affected families is not straightforwardly improved by ascertainment of parental phenotype, and it is not yet clear whether genomic screening of prospective parents in families affected by idiopathic ASD is warranted as a clinical standard. Systematic screening of affected family members for heritable pathogenic variants, including rare sex-linked mutations, will identify a subset of families with substantially elevated transmission risk. Polygenic risk scores are only weakly predictive at this time but steadily improving and ultimately may enable more robust prediction either singly or when combined with the risk variables examined in this study.