White Matter Integrity and Non-verbal Intelligence in Autism

Background: A growing number of neuroimaging studies report differences in white matter integrity between children with autism spectrum disorder (ASD) and typically developing (TD) controls. Findings from several of these studies show differences in white matter integrity across widespread areas of the brain, suggesting that ASD is a disorder related to connectivity. However, far less is understood about how aberrant connectivity relates to neuropsychological assessments of intelligence, which often differ between ASD and TD subjects. 

Objectives: We sought quantify how a standardized measure of non-verbal intelligence – the computerized test of non-verbal intelligence (CTONI) - relates to differences in white matter integrity between children with high-functioning ASD and TD controls. 

Methods: Ten children with high-functioning ASD and 10 TD controls matched on the basis of age (mean 10.7 vs. 11.4) and gender (9 males) were scanned at 3T (Philips Intera). Neuropsychological assessments were obtained in each subject, which included the CTONI. A high-resolution T1-weighted anatomical MRI and a 32-direction diffusion-weighted sequence were acquired. A standard tract-based spatial statistics (TBSS) analysis was performed with contrasts testing for voxel-wise differences in fractional anisotropy (FA) between ASD and TD controls, and differences in the correlation of the CTONI intelligence quotient between ASD and TD groups in areas showing significant (p<0.01, Threshold-Free Cluster Enhancement) FA differences.

Results: CTONI IQ differed significantly between groups (85.3 vs. 108.4, p=0.003). Using TBSS, we found two areas where FA was significantly higher in TD vs. ASD, which includes white matter terminations near the medial aspect of the superior frontal gyrus in the right hemisphere and the superior temporal gyrus of the left hemisphere. We also found areas where FA was significantly higher in ASD vs. TD, including white matter terminating near medial and lateral middle frontal gyrus and interior frontal gyrus (p. orbitalis) of the right hemisphere, and white matter terminating near the middle occipital gyrus, cuneus, thalamus, and inferior frontal gyrus (p. orbitalis) of the left hemisphere. In the network of areas that showed FA differences between the groups, we tested for a significant difference between the groups in the correlations between FA and CTONI scores. We found significant differences between correlations of FA and CTONI sources near terminations of white matter that included the left hemisphere of the middle occipital gyrus, thalamus, and inferior frontal gyrus. In these regions, children with ASD showed a greater positive relationship between FA and CTONI scores compared with TD controls. 

Conclusions: Although preliminary, these results implicate a thalamo-cortical white matter network, possibly including the inferior fronto-occipital fasciculus of the left hemisphere, as a possible neural substrate underlying differences in non-verbal intelligence in children with high-functioning ASD. This anatomical network may mediate the transfer of visual information from occipital areas directly to frontal cortex for further processing. Children with high-functioning ASD may rely more on this network, to the detriment of performance, than TD children for carrying out the cognitive operations necessary to complete some neuropsychological tests. Ongoing studies will test this hypothesis in the context of more specific behavioral measures.