Rare variants in the outcome of social skills group training for autism

Background Exome sequencing has been proposed as the first-tier genetic testing in autism spectrum disorder (ASD). Here, we performed exome sequencing in autistic individuals with average to high intellectual abilities (N = 207) to identify a molecular diagnosis of ASD and genetic modulators of intervention outcomes following social skills group training (SSGT) or standard care. Methods Within a randomized controlled trial of SSGT, we performed exome sequencing to prioritize variants of clinical significance (VCSs), variants of uncertain significance (VUSs) and generated a pilot scheme to calculate genetic scores representing the genetic load of rare and common variants in the synaptic transmission pathway (GSSyTr and GSSyTc). The association with the primary outcomes (parent-reported autistic traits, Social Responsiveness Scale) was computed using a mixed linear model. Behavioral and genetic features were combined in a machine learning (ML) model to predict the individual response within the cohort. Results In total, 4.4% (n = 9) and 11.3% (n = 23) of the cohort carried VCSs and VUSs, respectively. Compared to non-carriers, individuals with VCS or VUS together tended to have limited improvements of the interventions (β = 9.22; CI (-0.25, 18.70); P = 0.057) and improved significantly less from standard care (β = 9.35; CI (0.70, 18.00); P = 0.036), but not from SSGT (β = -2.50; CI (-13.34, 8.35); P = 0.65). In addition, both GSSyTr and GSSyTc were associated with differential outcomes in standard care and SSGT groups. Our ML model showed the importance of rare variants for outcome prediction. Conclusions Autistic individuals with likely pathogenic rare variants identified by exome sequencing might benefit less from the standard care. SSGT could therefore be of heightened importance for this subgroup. Further studies are needed to understand genetic predisposition to intervention outcomes.