18085
Hubs of Functional Brain Networks Are Atypically Organized in Children with Autism
Hubs are uniquely situated brain regions that play a critical role in integrating modality-independent information across distributed brain networks. Remarkably, very little is currently known about the anatomical location as well as the functional organization of these hubs in autism spectrum disorder (ASD). Such knowledge is critical for understanding the brain’s information processing architecture underlying cognitive dysfunction in ASD.
Objectives:
To characterize hubs of functional brain networks in children with ASD, and compare their features in terms of anatomical location as well as functional connections with the ones characterizing the typical developing brain.
Methods:
One cohort of 40 children (ASD = 20, age: 9.9±1.6; TD = 20, age: 9.9±1.6) was recruited at Stanford University, and a second cohort of 40 children (ASD = 20, age: 11.04±1.3; TD = 20, age: 10.83±1.6) was recruited at Georgetown University. Task-free fMRI were acquired from both cohorts.
Preprocessed fMRI datasets were parcellated into 264 cortical and subcortical regions using previously published functional templates. Correlation analysis of the extracted regional fMRI-time series was used to compute whole-brain inter-regional functional connectivity. Graph-analytical methods were used to identify putative hubs in the whole-brain functional connectivity networks. Specifically, participation coefficient - a network-measure of number of functional systems a region participates - was computed for each brain region. Using this metric, a hub was identified as a brain region with a high participation index thereby allowing the region to facilitate interactions between distributed functional systems. Critically, identification of hubs using participation coefficient overcomes the limitations of conventional degree-based hub identification techniques. Finally, to investigate how ASD pathology affects hub organization, we compare participation coefficient of each brain region between the ASD and TD group.
Results:
We examined the participation coefficient values obtained for the whole-brain functional connectivity network of each participant. In the Stanford cohort, the ASD group, compared to the TD group, showed smaller participation coefficient values for multiple brain regions including the right insula, right lateral occipital cortex/fusiform gyrus (LOC/FG), right superior temporal gyrus, and postcentral gyrus. Higher participant coefficient values in the ASD group, compared to the TD group, were observed for the posterior cingulate cortex. We repeated our entire analysis on the second group of children from the Georgetown cohort. Remarkably, in spite of differences in scanner (GE vs. Siemens), fMRI pulse sequence (spiral in-out vs. echo planar imaging) and other data acquisition protocols, results from the Georgetown cohort replicated the Stanford-cohort finding of atypical hub organization in children with ASD. Specifically, in the Georgetown cohort, the right insula and the right LOC/FG – key nodes of the salience and face processing systems - showed smaller participation coefficients in the ASD group, compared to the TD group.
Conclusions:
Our findings provide robust evidence for atypical organization of hubs in childhood autism, particularly those situated in brain systems critical for attention to salient stimuli and face processing. We propose that dysfunctional hubs may lead to impaired signaling between brain systems important for processing salient socially-relevant stimuli, potentially resulting in social deficits among children with ASD.