15907
Altered Amygdala Nuclei Projections in Young Adults with Autism Spectrum Disorder

Thursday, May 15, 2014: 10:54 AM
Imperial A (Marriott Marquis Atlanta)
C. R. Gibbard1, J. Ren2, D. H. Skuse2, J. D. Clayden1 and C. A. Clark1, (1)Imaging and Biophysics Unit, UCL Institute of Child Health, London, United Kingdom, (2)Behavioural and Brain Sciences Unit, UCL Institute of Child Health, London, United Kingdom
Background:

Structural magnetic resonance imaging (MRI) studies of autism spectrum disorder (ASD) report both amygdala enlargement and reduction in ASD, whilst functional MRI studies have shown reduced amygdala activation in response to social cues.  The amygdala comprises several nuclei, each with specific white matter (WM) connections.  Recent diffusion tensor imaging studies segmented the healthy adult amygdala into subregions in vivo using WM connectivity-based parcellation schemes.

Objectives:

To investigate amygdala sub-structure in ASD using in vivo WM connectivity information.

Methods:  

25 high-functioning ASD (mean age: 24.7yr) and 26 neurotypical (mean age:23.2yr) subjects underwent whole-brain T1-weighted (1mm3) and diffusion-weighted (2.5mm3; 60 directions b=1000s/mm2; 3 b=0) MRI on a 1.5T Siemens Avanto scanner.  Diffusion data were pre-processed using FSL.  Amygdala and whole brain regions of interest were delineated using the FSL tools FIRST and SIENAX.  Cortical regions were generated using FreeSurfer.  Cortical targets were grouped into frontal, parietal, occipital, and temporal lobes, and the insula using FSL utilities.  All regions of interest were registered to diffusion space using the FSL tools FLIRT and FNIRT.  TractoR was used to seed probabilistic tractography from each amygdala voxel to the five cortical targets.  An iterative 'winner takes all' process was used to assign a winning target to each voxel.  Voxels maximally connected to the same target were considered a cluster.  WM tract fractional anisotropy (FA) and mean diffusivity (MD) were measured.  Group comparisons were made using linear regression.  Correlations with the self-reported autism quotient (AQ), a measure of ASD severity, were made within the ASD group using partial Spearman correlation.  Age, gender and full-scale IQ were covariates.

Results:  

Right amygdala volume, normalized as a percentage of whole brain volume, was significantly elevated in ASD compared to neurotypical controls (t=2.01; p=0.05).  There was no significant group difference in left amygdala volume.  MD was significantly elevated in the ASD group in WM tracts connecting the left amygdala with the left cortex (t=2.53; p=0.02) and the right amygdala with the right cortex (t=2.95; p=0.005).  The 'winner takes all' algorithm resulted in clusters of amygdala voxels connecting to all cortical targets, aside from the occipital lobe.  MD of WM tracts connecting the left amygdala-left temporal lobe cluster was significantly higher in ASD (t=2.11; p=0.04).  Within the ASD group, negative correlations were observed between AQ score and FA of WM tracts connecting the left (rho=-0.51; p=0.009) and right (rho=-0.51; p=0.009) amygdala-temporal lobe clusters (both survive FDR-correction).

Conclusions:

We report alterations in amygdala volume and structural connectivity in young adults with ASD.  These findings were particularly apparent for the right amygdala, indicating that it is more involved in, or affected by, ASD pathology than its counterpart in the left hemisphere.  Our structural parcellation results demonstrate that local connections between the amygdala and the temporal lobe are particularly affected in ASD, with greater aberrations in these local connections strongly related to increased ASD severity.  This is the first study to parcellate the ASD amygdala based on its WM connections; our findings indicate that amygdala sub-regions have distinct involvement in ASD impairment.