Associations Between White Matter Diffusion Properties and ASD Impairments: A Tract-Based Spatial Statistics Study

Background:

Autism spectrum disorder (ASD) has often been characterized as a disorder of connectivity. Studies have revealed white matter anomalies across a wide range of networks including those commonly associated with ASD such as motor and social networks. Diffusion tensor imaging (DTI) has been used to investigate structural connectivity across major white matter pathways. In an effort to shed light on the inconsistent and widespread findings from prior studies, we opted to utilize an unbiased approach for localizing anomalous regions of white matter connectivity in children with ASD. This study employed tract-based spatial statistics (TBSS), a non-hypothesis driven approach of the whole brain rather than localized regions and/or networks.

Objectives:

To isolate differences within FA and MD and symptom severity between children with ASD and typically developing (TD) children within major white matter connections of the whole brain.

Methods:

Diffusion weighted imaging was acquired on 50 children with ASD and 50 TD children aged 8-12 years. Groups were balanced for sex, age and cognitive ability. All children were right handed. TBSS was used to create a skeleton of major white matter tracts. A whole brain voxel wise analysis was performed to identify areas of significant difference within the white matter skeleton between groups for fractional anisotropy (FA) and mean diffusivity (MD). Threshold-free cluster enhancement was used to control for multiple comparisons. Atypical white matter regions in ASD were localized using a standardized whole brain atlas. Pearson’s correlations were used to examine associations between MD and symptom severity.

Results:

Preliminary analysis from TBSS revealed widespread increased MD in children with ASD as seen in Figure 1. When these significant networks are overlaid with the standardized whole brain atlas, findings were localized to 39 of the 90 regions; all but 10 of these 39 regions lie in the left hemisphere. Pearson’s correlation further revealed a positive correlation in children with ASD between the MD and ADOS communication score in 10 regions with only two within the right hemisphere. These regions included the left body of the corpus callosum (p=0.045, r=0.285), left cuneus (p=0.018, r=0.334), left lingual gyrus (p=0.017, r=0.335), left precentral gyrus (p=0.049, r=0.279), left inferior occipital gyrus (p=0.041, r=0.290, left middle occipital gyrus (p=0.004, r=0.404), left posterior thalamic and optic radiation (p=0.020, r=0.327), left superior corona radiate (p=0.038, r=0.294), right angular gyrus (p=0.020, r=0.327), and right middle occipital gyrus (p=0.010, r=0.361). No differences in FA were observed.

Conclusions:

Consistent with prior findings, we found that children with ASD show increased MD in major white matter regions including but not limited to motor and social networks predominately within the left hemisphere. This left lateralization is supported by several studies have found networks including motor and social networks to have decreased FA and increased MD in children with ASD. Increased MD in children with ASD is associated with increased impairment in communication including functional, social and gestural communication as measured by the ADOS communication score.