Maturation of Resting State Power Spectrum and Functional Connectivity in Autism Spectrum Disorder

Background: Understanding the developmental trajectory of neuronal mechanisms underlying autism spectrum disorder (ASD) is critical to developing effective treatments. Resting state EEG provides an index of cortical neuronal oscillations that reflects brain function. Previous studies have used resting state EEG to compare local and long-range synchronization of neuronal oscillations between individuals with ASD and neurotypical controls. Although ASD-related alterations have been described in a large number of studies, findings are very heterogeneous and inconsistent. Well-powered studies providing an unbiased evaluation of putative resting-state EEG abnormalities in ASD are urgently needed.

Objectives: The present study evaluates the developmental trajectories of resting state power spectrum and functional connectivity in high functioning ASD from childhood to adulthood in a comprehensive and large dataset and in comparison to TD.

Methods: EEG data were acquired as part of the EU­AIMS Longitudinal European Autism Project (LEAP) (www.eu-aims.com). Clinical diagnosis of ASD was established with DSM-IV / ICD-10 or DSM-5. In this analysis, 141 TD and 153 ASD participants between 6 and 30 years and IQ>70 were included. Subjects underwent a resting state EEG (2 min eyes open, 2 min eyes closed in alternating blocks of 30 sec, 5 centers, 60 or more EEG channels). Power spectra and functional connectivity were evaluated in sensor and source space. Source time series were extracted using beamforming and finite element forward models built from individual MRIs. Power spectra (1-32) Hz were computed with Morlet wavelet transform. Functional connectivity was estimated between sensors and between 50 regions of interest (source space analysis) using phase and amplitude synchronization measures (phase locking value, phase lag index and power correlations). Statistical modeling was performed with linear-mixed effects models with group, age, gender, IQ, and all interactions with group as fixed effects, recording site as a random effect, and allowing for different variances for ASD and TD groups (nlme package in R). The analysis was unbiased with respect to directionality of the effect, spatial location, and frequency; cluster based permutation techniques were employed to account for multiple comparisons.

Results: Highly significant age-related changes were found both for power spectrum and functional connectivity. Relative power decreased with age at lower frequencies (2-6Hz) and increased at higher frequencies (10-32Hz). Functional connectivity showed complex patterns of age modulations. Whole brain averaged functional connectivity values increased robustly with age in 4-20 Hz range for power correlations and phase locking value. These age-related modulations are in line with previous literature demonstrating the validity and sensitivity of our power spectral and functional connectivity estimates. No significant group effects (p>0.05, main effect or interactions with age, IQ or gender) in mean or variance were found in power spectrum or functional connectivity.

Conclusions: Both resting state power spectrum and functional connectivity showed strong age-related maturation from childhood to adulthood in high functioning ASD and TD cohorts. However, we did not find differences between ASD and TD in both measures. This indicates that, across brain rhythms, local and long-range synchronization in high functioning ASD may mature according to a typical developmental trajectory.