White Matter Microstructure in Girls with Autism Spectrum Disorder: Comparison with Neurotypical Controls and Unaffected Siblings

Background:

Studies implementing tract-based spatial statistics (TBSS) in samples consisting mainly of boys have shown abnormalities in white matter (WM) microstructure in autism spectrum disorder (ASD).  One such study reported similar WM abnormalities in unaffected sibling (US) and ASD groups, suggesting that US share some neurological vulnerability with probands.  Sex-comparison studies have indicated that ASD is associated with sex-specific neurological effects.  However, ASD girls have been grossly understudied relative to boys and no study to date has implemented diffusion tensor imaging (DTI) in the exclusive examination of girls with ASD or US.

Objectives:

Investigate WM structure in ASD girls in comparison to neurotypical (NT) girls and US sisters.

Methods:

This study consisted of 31 girls: 13 ASD (mean age:8.9yr; IQ:101), 11 NT (mean age:9.3yr; IQ:102), and 7 US (mean age:12yr; IQ:102).  Participants underwent diffusion-weighted MRI (2.5mm³; 30 directions at b=1000s/mm²; 5 b=0; runs=3) on a 3T Siemens scanner.  Diffusion volumes with excessive motion were removed prior to averaging acquired runs.  Averaged scans were pre-processed using FSL, including eddy current correction and estimation of the diffusion tensor, which enables calculation of WM microstructure metrics: fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD).  Voxel-wise group comparisons in WM microstructure were performed using TBSS which co-registers all diffusion data and generates an average WM skeleton on which statistical comparisons are made.  The values for each WM metric were also averaged across the WM skeleton and correlated with parent-reported social responsiveness scale (SRSp) t scores.  Age and full-scale IQ were covariates.

Results:

There were widespread regions of the right hemisphere in which MD was significantly elevated in US compared to NT.  There were also significant clusters widespread in the left hemisphere where RD was significantly higher in ASD compared to NT (both p<0.05; FWE-corrected).  No other group comparisons reached significance.  SRSp score was strongly correlated with DTI measures averaged across the WM skeleton (FA: p=0.002; MD and RD: p<0.001; AD: p=0.009) in NT girls.  SRSp and WM correlations did not reach significance in ASD or US.  Voxel-wise correlations within TBSS (p<0.05; FWE-corrected) showed that the NT FA and RD relationships with SRSp were widespread bilaterally, whilst the MD-SRSp relationship was localised to the left hemisphere.  The AD correlation did not show a significant voxel-wise effect.

Conclusions:

This is the first study to show that both ASD girls and US sisters have WM abnormalities in comparison to NT girls, thus indicating that siblings and probands exhibit structural abnormalities in the brain.  In controls, SRSp score, a measure of ASD-like behaviours, was highly related to WM microstructure; this relationship was not observed in the other groups.  This suggests that both ASD girls and US sisters show an atypical relationship between WM microstructure and ASD traits.