Waveform Morphology in EEG Parses Heterogeneity in Autism and Schizophrenia

Background: Autism (ASD) and Schizophrenia (SCZ) share core and comorbid symptoms such as sensory sensitivities, reward processing impairments, and social difficulties. Electrophysiological (EEG) investigations of these symptoms have primarily focused on oscillatory activity; however, recent research suggests that important aspects of EEG are not strictly rhythmic. Instead, they include non-stationary bursts and waveform shape, biophysically meaningful features used often in clinical evaluation of the EEG. Nevertheless, these features are infrequently used to assess brain activity between and within diagnostic groups.

Objectives: To use innovative metrics of EEG shape to quantify resting brain activity and its association with diagnosis and clinical characteristics in ASD and SCZ. Specifically: (1) the slope of the resting power spectrum (PSD); (2) the presence of oscillatory “bursts;” (3) Nonsinusoidal features of EEG shape.

Methods: Resting EEG was collected in 105 individuals: 33 ASD, 28 SCZ, and 44 typically developing (TD) controls. EEG was recorded for 4 minutes with eyes open and 2 minutes with eyes closed. Slope of the PSD was calculated from 1-30Hz. Waveform shape was quantified as the symmetry between the peak and trough, or rise and decay, of a cycle, aspects of the EEG that reflect underlying physiology but are not captured in frequency analyses (e.g., the symmetry between the rise and decay influence the EEG’s tendency to exhibit a “sawtooth” shape). Bursts were defined as at least three cycles that exhibited the same symmetry characteristics. These EEG metrics were calculated cycle-by-cycle across frequency bands, and averaged, per-person. A regression tested EEG association with diagnosis and correlations assessed EEG association with symptomology.

Results: Compared to TD, SCZ had fewer bursts in alpha across the scalp, in beta across occipital regions, and in gamma frontally [parameters ranging b=-.14, p=.017 to b=-.20, p=.001]. Compared to ASD, SCZ exhibited fewer bursts in alpha [parameters ranging b=-.13, p=.03 to b=-.18, p=.006]. With respect to waveform shape, compared to TD, both SCZ and ASD exhibited increased peak-trough symmetry in frontal theta (ps<.005), indicating waveforms with sharper troughs. Individuals with SCZ exhibited faster oscillatory bursts (higher peak frequency) in the theta, alpha, and beta bands [parameters ranging b=2.9, p=.03 to b=10.1, p=.02]. Waveform shape correlated with symptomology, with faster rises in frontal theta (r=-.28, p=.004) and drops in occipital beta (r=.24, p=.015) predicting greater ADOS severity scores. With respect to PSD shape, SCZ differed from TD with more negative slope (b=-.07, p=.03) in the occipital regions, indicating more high-frequency power associated with increased excitatory activity.

Conclusions: The shape of EEG waveforms reflects underlying cortical activity. While assessment of EEG shape is common clinical practice, research has focused, primarily, on oscillatory approaches. We show, for the first time, that objective, non-oscillatory measures in clinical groups strongly associate with diagnosis and symptoms. This approach offers insight into low-level activity generated by cortical inhibition and excitation, supporting a better understanding of the biological substrates. Ongoing analyses incorporating sensory and ASD symptomology will further explicate these relationships and allow us to better differentiate subgroups of individuals exhibiting similar brain activity and symptom profiles.