Biomarkers of Social Perception in Children with ASD and Loss of Function Gene Mutations

Background: Recurrent disruptive mutations in genes such as CHD8 and DYRK1A have been implicated as contributing to approximately 1% of autism spectrum disorders, (ASD; O’Roak et al., 2012; Devlin & Scherer, 2012). Work has begun to identify ASD subtypes associated with specific loss-of-function genetic mutations (e.g., Bernier et al., 2014). Despite the hallmark social impairments of ASD, to date, no one has characterized the impact of genetic mutations associated with ASD on social cognition. Social perception biomarkers are ideal for assessing social cognition in children with widespread behavioral capacities, given the robust nature of passive viewing tasks (e.g., Kaiser et al., 2010). Reduction of mu attenuation in response to viewing social versus nonsocial motion is one such biomarker of social perception. However, discrepant results of mu rhythm in ASD are reported (e.g., Bernier et al., 2007; 2013; Oberman et al., 2005; 2013), potentially highlighting existing heterogeneity that may be driven by etiological differences.  

Objectives: We aimed to compare the biomarkers of social perception in children with ASD with recurrent, ASD associated, loss of function mutations (e.g., CHD8, DYRK1A) to children with idiopathic ASD and typically developing children.

Methods: Children completed an EEG session during which they watched videos of social and nonsocial motion. Mu attenuation was computed across electrodes surrounding C3 and C4 as Fast Fourier transformed power (8-13 Hz), relative to resting. We targeted children with ASD and loss-of-function gene mutations falling into an interactive protein network (ASD-LOF). Comparison groups included age- and gender-matched children with idiopathic ASD (ASD-NON) and typical development (TYP). See participant characterization in Table 1. Single-trial analysis conducted in SAS 9.3 used REML multilevel models and Bonferroni correction for pair-wise comparisons.

Results: Observed mu attenuation for each group is presented in Figure 1 across trial and on average for social (green) and nonsocial stimuli (orange). First, a model comparing both ASD groups to TYP indicated no group differences of condition, F(1,7023)=.94, p=.33. Second, a model comparing all three groups indicated a group effect of condition, F(1,7031)=30.93, p<.0001, such that ASD-NON group did not exhibit mu attenuation condition differences, t=2.49, p=.19. The ASD-LOF and TYP groups exhibited more mu attenuation for social than nonsocial stimuli, respectively, t=5.39 and t=6.09, p’s<.0001. Post-hoc ANOVAs indicated that the children with CHD8 and the DSCAM genetic events exhibited equivalent mu attenuation to both social and nonsocial stimuli, F(1,58)’s<3.4, p’s>.068. The other ASD-LOF children exhibited greater mu attenuation to social compared to nonsocial stimuli, similar to the TYP group. There were no significant effects of IQ, age, or gender.

Conclusions: Traditional analysis of TYP versus ASD suggested no difference in mu attenuation (e.g., Bernier et al., 2013). Implementation of a “genetics-first” approach specified that ASD-LOF children exhibit patterns of mu attenuation similar to TYP. However, children with specific genetic mutations (CHD8, DSCAM) exhibit mu attenuation more similar to ASD-NON. These results indicate the necessity of careful consideration for genetic mutations contributing to phenotypic social cognition profiles of ASD.