Diffusion Tensor Imaging in Unaffected Siblings of Individuals with Autism: A Pilot Study Using Tract-Based Spatial Statistics

Background:  Numerous studies have been published using diffusion tensor imaging (DTI) to demonstrate white matter abnormalities in autism spectrum disorder (ASD). However, very few studies assess whether white matter abnormalities exist in unaffected siblings (UAS) of those with ASD. Among those published, including a recent report from the authors of this abstract, samples of modest size included both boys and girls in the same analysis. Consequently, findings may be influenced by the structural variability associated with gender. To address this limitation, the authors compare newly-recruited male UAS with stringently-matched control participants.

Objectives:  To evaluate whether previously reported intermediate neuroendophenotypes in mixed-gender samples of UAS can be replicated in a new male-only group of participants, who were individually matched to typically developing males on age and IQ.

Methods:  Participants included 12 pediatric male UAS (age in months: M = 129.99, SD = 41.68, Range = 64.9-203.8; Full-scale IQ: M = 112.92, SD = 11.21, Range = 96-135) and 12 pediatric male control participants (age in months: M = 132.47, SD = 45.7, Range = 62.9-197.57; Full-scale IQ: M = 109.83, SD = 14.62, Range = 89-141). T1-weighted and diffusion-weighted MRI (directions = 30 and b0 = 5) were acquired using a 3-Tesla scanner. FMRIB Software Library (FSL) was used to process/analyze diffusion-weighted data and compute the following WM microstructure metrics: fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD). Voxel-wise analysis of multi-subject diffusion data was conducted using FSL’s Tract Based Spatial Statistics (TBSS). Areas of significant difference were computed using Threshold-Free Cluster Enhancement (TFCE) and displayed as p-value images, where p < 0.05 corrected for multiple comparisons across space.

Results:  There were no statistically significant differences in age [t(22) = 0.14, p = 0.89] or IQ [t(22) = -0.58, p = 0.57]. There were no statistically significant group differences in FA, MD, AD, or RD when UAS were compared to control participants. Trends toward statistical significance were also not observed in comparison of the aforementioned WM microstructure metrics.

Conclusions:  Unlike previous mixed-gender comparisons between UAS and control participants, this preliminary male-only comparison does not support the presence of an intermediate neuroendophenotype in male UAS. Lack of statistically significant differences in the UAS and control comparison does not preclude that aberrant WM microstructure exists in UAS males, especially if the effects are small. Future studies should investigate whether current findings replicate in larger studies with closely-matched comparison groups.