The Rich-Club Organization of the Brain in Autism Spectrum Disorder
Understanding of the organization of structural and functional neural networks on a large scale has been advanced in recent years through a set of general techniques under the umbrella of graph analysis. Through applications of these and related techniques, comparisons of regional and global network parameters can be estimated and compared across groups, leading to the discovery of several previously hidden brain networks. Recently, the so-called “rich club” organization of the human brain has been documented (e.g. van den Heuvel & Sporns, 2011), revealing a surprisingly small number of subcortical and neocortical hubs that have a tendency to be more highly connected to each other than to other brain regions. This suggests that focal structural or functional abnormalities may have a greater than expected impact when localized to brain regions that are part of the ‘rich-club’. Simulation of network failure in these hubs is particularly useful for investigating potential connectivity difficulties in ASD.
In the current study, we examined this notion in autism spectrum disorders (ASD), to test whether alterations in the morphological properties of 82 brain regions are 1) systematically different in ASD vs. Controls, and 2) whether any observed differences map onto the 12-16 regions thought to be part of the brain’s rich-club.
Resting state data from 184 subjects (79 with ASD) from the ABIDE database were analyzed. Subjects were matched on age (ASD=14.52; TD=15.81) and IQ (ASD=107.91, TD=113.15). Additionally, all subjects were collected at the same site. Regions of interest (82 regions indicated in the rich club of the brain) were extracted for each subject and averaged by group. Minimum and maximum density measures of connectivity were computed for each group to conduct regional network analysis. Finally, network attacks were simulated for each group.
Preliminary analyses demonstrate the ability of utilizing the rich club connectome to generate adjacency matrices. These matrices are then applied to a brain connectivity toolbox capable of examining the impact of connectivity failures in ASD.
Results demonstrate a potentially compromised network of connections in ASD. When network failure occurs in a rich club node, a greater detrimental impact on other connections in ASD. Thus, results indicate alterations in overall neural functioning in these individuals. Further, this analysis provides a data-driven approach to parsing well-documented heterogeneity within ASD. The relationship of real-world functioning to indices of altered connections in ASD provides a potential for targeting treatment.