Microstructural alterations of association tracts in autism localized with along-tract mapping

379 

Abstract Submission 

Gaon Kim¹, Sebastian Benavidez¹, Bramsh Chandio¹, Katherine Lawrence¹, Paul Thompson¹

¹Imaging Genetics Center, Keck School of Medicine of University of Southern California, Los Angeles, CA

Gaon Kim  
Imaging Genetics Center, Keck School of Medicine of University of Southern California
Los Angeles, CA

Sebastian Benavidez  
Imaging Genetics Center, Keck School of Medicine of University of Southern California
Los Angeles, CA

Bramsh Chandio  
Imaging Genetics Center, Keck School of Medicine of University of Southern California
Los Angeles, CA

Katherine Lawrence, PhD  
Imaging Genetics Center, Keck School of Medicine of University of Southern California
Los Angeles, CA

Paul Thompson, PhD  
Imaging Genetics Center, Keck School of Medicine of University of Southern California
Los Angeles, CA

Autism is a heterogeneous neurodevelopmental condition characterized by subtle and widespread changes in brain morphometry and white matter microstructure of both gray and white matter as revealed by previous neuroimaging studies (1-7). While diffusion-weighted magnetic resonance imaging (dMRI) research has uncovered microstructural alterations associated with autism, region of interest based approaches or tract-based spatial statistics lack the precision to offer a fine-scale local mapping of microstructure (4-7). In this pilot study, we used the advanced tractography-based approach, BUndle ANalytics (BUAN) to investigate white matter microstructure of association tracts in autism at a more refined anatomical scale.

We analyzed 3D dMRI brain scans from 172 participants (mean age: 24.3+15.2 years, 99.4% male) - 107 individuals with autism and 65 neurotypical subjects. Data was sourced from one NIMH Data Archive site and two Autism Brain Imaging Data Exchange sites and all three sites used 3T scanners (b=1000-2500s/mm²; voxel size=2-3mm). Preprocessing steps included denoising and correction for eddy currents, head motion, bias field, and gradient distortions. Standard DTI metrics - fractional anisotropy (FA), and mean, axial, and radial diffusivity (MD, AD, RD) - were computed at each voxel. Whole-brain tractograms were generated using a constrained spherical deconvolution model and local deterministic tractography. The tracking algorithm was set to start from voxels where FA>0.3, seed count=10, step size=0.5 and stopped tracking if FA <0.2 (8). Each white matter tract was extracted using the auto-calibrated RecoBundles (9) and a standard atlas of major white matter tracts (10). We focused on 6 separate bilateral association tracts: arcuate fasciculus (AF), extreme capsule (EMC), inferior fronto-occipital fasciculus (IFOF), inferior longitudinal fasciculus (ILF), middle longitudinal fasciculus (MdLF), uncinate fasciculus (UF). Using BUAN, each of the association tracts was divided into 100 segments per subject. Linear mixed models were used to compare microstructural metrics between the autism and neurotypical groups, adjusting for age and sex while accounting for subject and site variability. The false discovery rate was applied for multiple comparisons correction.

We found significant tract segment differences for mean FA, MD, AD, and RD in association tracts (Table 1). Mean FA differences were detected in the left AF, left IFOF, left ILF, bilateral MdLF, and bilateral UF (Fig 1A-D). For mean MD, the following tracts showed significant differences between the autism and neurotypical groups: right EMC, bilateral IFOF, right ILF, left MdLF, and left UF (Fig 1E-F). Significant mean AD differences were observed in right EMC, left IFOF, left MdLF, and right UF. Finally, for mean RD the left AF, right EMC, left IFOF, right ILF, left MdLF and left UF tracts showed significant differences between groups. In sum, we found localized differences in association tracts in autism using BUAN.

In this pilot study, we used BUAN - an advanced along-tract analysis method - to investigate the fine-grained microstructure of association tracts in autism. We found localized microstructural changes in the AF, EMC, IFOF, ILF, MdLF, and UF in autism compared to neurotypical controls. Microstructure was altered in localized regions, at a scale that may not be resolved with standard ROI analyses. Future work will include larger and more diverse samples, as well as associations with clinical and behavioral assessments. Whole-brain tractometry may also help to identify subgroups within autism cohorts, providing insights for interventional studies.

Neurodevelopmental/ Early Life (eg. ADHD, autism) ¹

Diffusion MRI Modeling and Analysis ²

Autism

White Matter

WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC

Other - association tracts