Spontaneous Alpha Oscillations Stratify Children Across the Autism Spectrum Based on Cognitive Ability

Background:

Electroencephalography (EEG) measures of spontaneous brain oscillations hold particular promise for improving our understanding of the neurobiological bases of the clinical heterogeneity observed in autism spectrum disorders (ASD). Of particular relevance within the EEG signal are spontaneous alpha oscillations. The typical increase in peak alpha frequency seen during childhood is said to reflect the development of thalamo-cortical connections (Valdés-Hernández et al., 2010); disruptions to which are said to also manifest through atypical levels of alpha power (Doesburg et al., 2011). Therefore, measures of peak alpha frequency and alpha power may provide a sensitive marker of altered trajectories of neurodevelopment in ASD at an early age.

Previous studies have reported atypical alpha power (Wang et al., 2013) and peak frequency development in ASD (Edgar et al., 2015). However, there are no reports of spontaneous alpha oscillations in an ASD sample with variable cognitive abilities. It is currently unknown whether spontaneous alpha oscillations are altered consistently across a representative sample of children with ASD or whether they distinguish subgroups based on cognitive ability.

Objectives: We asked whether the power and frequency of spontaneous alpha oscillations distinguished children with ASD from age matched typical children and, moreover, whether alpha power and frequency differentiated children with ASD who had comorbid intellectual disability (ID).

Methods: Fifty-seven children with an ASD diagnosis (mean age=5.6 years, range= 2-12 years) were split into ‘Low-IQ’ (IQ = 11-69; N=30) and ‘High-IQ’ (IQ= 70-126; N=27) groups based on an IQ cut-off of 70. Forty age-matched typically developing (TD) children (IQ = 95-148) served as a control group. High-density EEG was recorded under task-free conditions while children watched bubbles on a computer screen. Relative alpha (8-12Hz) power and peak frequency were calculated.

Results: TD children showed the expected increase in peak alpha frequency with age (r=>.5, p=<.001), as did children in the high-IQ ASD group (r=0.35, p=0.05). However, this relationship was absent in the low-IQ ASD group (p=>.5). Low-IQ ASD participants showed lower alpha (p=<.01) power across the scalp and lower frontal peak alpha frequency (p=.02) compared to controls. Alpha power (p=>.52) and peak frequency (p=>.30) across the scalp did not differ between high-IQ ASD participants and controls.

Conclusions:

Our results suggest that while the typical development of peak alpha frequency with age is disrupted in ASD, this disruption seems to be driven by co-occurring ID. We also find reduced alpha band power differentiates children with comorbid ID and ASD. Together, these findings suggest that electrophysiological markers of atypical neural function may not be consistent across the spectrum in ASD. Whether these oscillations represent a biomarker of ID rather than ASD will require investigation in children with ID without ASD.

Furthermore, in the context of previous literature, the present study indicates that the level of disruption to the development of thalamo-cortical connections may contribute to the variability in EEG power measures previously reported in ASD. Our future research will utilize longitudinal data to determine how changes in peak alpha frequency are associated with the development of cognitive abilities across the autism spectrum.