Early childhood network alterations in severe autism

Abstract

Objective: To examine the differences in whole brain topology and connectivity in 17 children of the ages 3-8 years across severity of ASD, we performed resting state fMRI using a 3T MRI scanner and graph theoretical analysis of networks.

Method: Patients were partitioned into two cohorts based on the severity of ASD, determined using the Childhood Autism Rating Scale (CARS) scores (Mild, 30-36; Severe, 37+). Standard preprocessing pipeline was used, followed by independent component analysis (ICA) to identify regions of interest (ROIs) to construct subject-specific Z-correlation matrices representing the whole brain network. Following which, graph theory measures were calculated in the range of sparsity 6%-35% and statistically analyzed, and corrected for significance (FDR corrected, p < 0.05). Regional clustering coefficient that revealed significant between-group (mild vs. severe) differences were correlated against clinical scores (CARS).

Results: Children with severe ASD revealed significantly increased clustering coefficient and small-worldness compared to those with mild or moderate ASD. Region of interest analysis revealed altered clustering in the Heschl's gyrus that significantly correlated with CARS scores.

Conclusion: The findings from the current study provide early stage evidence of aberrant brain connectivity appearing in severe ASD, prior to the effect of environmental bias and pruning mechanisms. The clustering of the Heschl's gyrus correlated to the severity of ASD symptoms and agrees with current literature on ASD-associated cortical changes, reflecting early changes to language processing regions.

Keywords: Graph theory; Language processing; Resting state fMRI; Severe autism.