19196
Anomalous Anatomical Connectivity Networks in Children with High Functioning Autism Spectrum Disorder

Saturday, May 16, 2015: 11:30 AM-1:30 PM
Imperial Ballroom (Grand America Hotel)
D. J. Peterson1, R. A. Vasa2,3 and S. H. Mostofsky1,3, (1)Kennedy Krieger Institute, Baltimore, MD, (2)Center for Autism and Related Disorders, Kennedy Krieger Institute, Baltimore, MD, (3)Johns Hopkins School of Medicine, Baltimore, MD
Background:  Altered cortico-cortico connectivity has been implicated in the neurobiological basis of autism spectrum disorder (ASD). Several lines of evidence indicate that ASD may be associated with underconnectivity between distant cortical regions, and a corresponding over-abundance of local connections. Graph Theory (GT) methods applied to structural connectivity networks derived from diffusion tensor imaging (DTI) may help directly address this hypothesis. GT is a branch of mathematics and computer science that has recently been adapted to neuroimaging studies of whole-brain connectivity. The two principal measures that can be derived from GT are the clustering coefficient (CC) and the characteristic path length (CPL). The CPL is generally taken to be a measure of long-range connectivity, while CC is a measure of local inter-connectivity, averaged over the whole brain.  Clinical group differences as well as correlations with behavior have been observed in these network measures for a variety of psychiatric disorders [Bullmore 2009].

Objectives:  1) To examine whether the anatomical connectivity networks of children with ASD differ from Typically Developing (TD) children in their overall topological properties, using DTI tractography. 2) To assess the hypothesis that ASD is associated with long-range underconnectivity and an overabundance of short-range connections. 3) To examine whether graph theory metrics correlate with core ASD symptoms.

Methods:  DTI was acquired in 35 Children with High-Functioning ASD (HF-ASD), and 35 TD controls, aged 9-14 years. The groups were matched on age, sex, IQ and handedness. After preprocessing with CATNAP and eddy-current correction, cortical regions were parcellated using a semi-automated atlas-based procedure. Each subject's FA and B0 images were transformed to the JHU-ICBM template using multi-channel Large  Deformation Diffeometric Morphic Mapping (LDDMM). The ROI labels from the JHU-ICBM atlas [Mori 2008] were then back-projected onto each subjects' standard-space images. Fiber-tracking was then initiated in each cortical label, using FACT as implemented in CAMINO. The number of streamlines connecting each cortical region to each other cortical region was taken as a measure of connectivity between cortical regions. GT measures were computed using the Brain Connectivity Toolbox [Rubinov 2010].

Results:  The mean CC over all nodes was significantly higher in children with HF-ASD (p=0.0007, Wilcoxon rank-sum) than in TD children, indicating a greater degree of local interconnectivity in children with HF-ASD. There were no significant group differences in the mean CPL. In the HF-ASD group, CC was found to be positively correlated with the stereotyped and restricted interests subscore of the ADOS (p=0.0332 r=0.361), and CPL was negatively correlated with the stereotyped and restricted interests score (p=0.0478 r=-0.336), and the ADOS total score (p=0.0341 r=-0.130).

Conclusions:  The anatomical connectivity networks of children with HF-ASD show increased short-range connectivity compared to TD children. The findings also demonstrate that graph-theoretic properties of full-brain cortico-cortico anatomical networks correlate to measures of symptom severity in children with HF-ASD.