Genetic-Independence of Key Early Contributors to Autistic Trait Severity

Background: Previously, in separate lines of research, we have shown that social visual engagement (Constantino et al., Nature, 2017) and quantitative autistic traits (QAT, see Constantino and Charman, Lancet Neurology 2016) constitute highly-heritable autism endophenotypes. Here, we examine the relationship between these and other highly- heritable predictors of autism recurrence in infancy and early childhood.

Objectives: In a prospective longitudinal study of epidemiologically-ascertained infant twins, we examined the extent to which variation in social visual engagement at 18-24 months, familial QAT background, inattention at 36 months, and variation in motor coordination at 36 months predicted quantitative autistic trait burden at 48 months.

Methods: Visual social engagement was measured using eye tracking methods as described in Constantino et al. (2017), involving infant viewing of social scenes on video, and deriving for each subject a total proportion of visual orientation to eyes, mouths, bodies, and objects. We obtained ratings of variation in attention and motor coordination by parent report using respectively Achenbach’s Child Behavior Checklist and the Developmental Coordination Disorders Checklist (DCDQ). We derived a quantitative index of parental QAT using the biparental mean of the adult SRS-2 completed by spouse report.

Results: Remarkably, despite the established relationship of each predictive phenotype to autism, none were correlated with one another in early childhood (p>0.05 with false discovery correction applied), overwhelmingly indicating that they constitute non-overlapping traits. Moreover, Bayesian regression revealed that visual social engagement—despite its strong association with categorically-defined autism—did not predict variation in QAT among normal 48 month olds in this population sample. In contrast, biparental mean QATs was a distinct positive predictor of child QAT and explained 6% of the variance in child QATs in hierarchical regression. Furthermore, two neurocognitive phenotypes non-specific to autism, attention and motor coordination, explained the largest amount of the variance (18% and 10% respectively) in hierarchical regression, and both were significant predictors of QATs at 48 months. All three phenotypes appeared heritable in female twin pairs, while there was a large overlap in the 95% confidence interval for ICC in male monozygotic and dizygotic twin pairs in males on CBCL attention scores, suggesting sex-specific differences in heritability for this phenotype. Finally, we observed genetic overlap only between motor coordination and QATs, suggesting that genes affecting motor coordination may also directly affect autistic traits.

Conclusions: Social visual engagement, familial background for quantitative autistic traits, variation in attention, and variation in motor coordination represent genetically-independent contributors to autistic symptom burden. Linking genetic variants to these underlying traits rather than to a diagnosis of “autism” may be more productive in devising personalized approaches to developmental intervention, especially if autism represents an epiphenomenon of earlier-interacting susceptibilities.