Age and Cognitive Ability Predict Peak Alpha Frequency During Resting-State Electroencephalography in Autistic Children

• In Event: Socio-cognitive abilities in children with Autism Spectrum Disorder

Abstract

Background: Peak alpha frequency (PAF) is an electrophysiological marker sensitive to developmental changes in neural networks and related to age in typical children (Cragg et al., 2011). A recent study found PAF to associate with cognitive ability but not age in autistic children aged 2 to 10 years (Dickinson et al., 2018). We sought to extend this finding to an older sample of autistic children without intellectual disability and to assess relationships with symptom severity. Method: Participants were 76 autistic and 58 typical children aged 8 to 17 years, matched on age and full-scale IQ (FSIQ). Social responsiveness (parent-report) and adaptive behavior (interview with clinician) were assessed. At least thirty seconds of artifact-free electroencephalogram (EEG) during eyes-closed resting state was available for each participant. PAF between 6 and 13 Hz was extracted from occipital electrodes using open-source code from Corcoran et al. (2018). Results: Mean PAF did not significantly differ by group, t (132) = -0.81, p > .05. Regressions were run with PAF as the dependent variable and age entered in the first block and FSIQ entered in the second block. In autistic children, age significantly and positively predicted PAF, F (1, 74) = 16.61, p < .05, and accounted for 18.3% of the variation in PAF (Figure 1); adding FSIQ significantly (p < .05) explained an additional 4.4% of variation in PAF, F (2, 73) = 10.72, p < .05. In typical children, age also significantly and positively predicted PAF, F (1, 56) = 9.97, p < .05, and accounted for 15.1% of the variation in PAF (Figure 1); adding FSIQ explained an additional 0.1% of variation in PAF, which was not significant (p > .05), although the overall model remained significant, F (2, 55) = 4.91, p < .05. Regressions were also run with PAF as the dependent variable and social responsiveness or adaptive behavior as the independent variable. Social responsiveness did not significantly predict PAF in either autistic (F (1, 73) = 0.83, p > .05) or typical (F (1, 54) = 0.15, p > .05) children. Similarly, adaptive behavior did not significantly predict PAF in either autistic (F (1, 69) = 1.25, p = > .05) or typical (F (1, 53) = 0.88, p > .05) children. Conclusion: Contrary to Dickinson et al. (2018), autistic children show the typical increase in PAF with age. Because this sample was older than the prior study, this suggests the relationship between PAF and age may normalize with age in autistic children. Similar to Dickinson et al. (2018), cognitive ability explained variance in the prediction of PAF in autistic children only. Thus, PAF may serve as a marker of cognitive ability above and beyond age and independent of symptom severity in autistic children.

Authors

• Armen Bagdasarov, Duke Univeristy
  Presenting Author

• Adam Naples, Yale University
  Non-Presenting Author

• James Charles McPartland, Yale University
  Non-Presenting Author