Longitudinal, Voxel-Based Analysis of White Matter Contributions to Processing Speed in Individuals with Autism Spectrum Disorder

Background: Slowed processing speed has been commonly reported in autism spectrum disorder (ASD) and has major implications for social and cognitive functioning. A recent longitudinal investigation found that processing speed was substantially slower across the life span in individuals with ASD compared to individuals with typical development, and  processing speed was correlated with whole-brain fractional anisotropy (a diffusion tensor imaging measure of circuit integrity) (Travers et al., 2014). However, until recently, it has not been possible to examine which white matter tracts within the whole brain are longitudinally related to processing speed in ASD and in typical development. Given that processing speed is a difficulty for many people with ASD, understanding the specific white matter tracts that underlie processing speed difficulties is essential.

Objectives: The objective of this study is to use a voxel-based, longitudinal approach to examine the white matter tracts in the brain that are related to processing speed in individuals with autism and individuals with typical development.

Methods: Fifty-one males with typical development and 70 males with ASD (ages 6.4 to 40.6 years old) underwent longitudinal processing speed assessments and longitudinal DTI scanning on Siemens 3T scanner at four points across a 10-year period of time (DW, single-shot, spin-echo EPI, b=1000, 12 non-collinear directions, 4 averages). Groups were matched on age (p = .48). Voxel-based, linear mixed-effects analyses examined fractional anisotropy as a function of group and processing speed, controlling for age, head motion, and a head coil replacement (between Times 1 and 2).

Results: In both the ASD and typically developing groups, the results suggested that processing speed was significantly related with a number of white matter areas, including the bilateral posterior medial parietal cortex, the tip of the anterior genu/cingulate, left temporal parietal junction, and the right prefrontal white matter (p< .05, FDR corrected). However, a group x processing speed interaction demonstrated that the group with ASD had weaker correlations between these areas and processing speed than the group with typical development (See Figure 1, p< .05, FDR corrected).

Conclusions: We found that processing speed in both groups was related to the white matter microstructure of areas previously reported to be implicated in processing speed (i.e., the bilateral posterior medial parietal cortices, the tip of the genu of the corpus callosum/anterior cingulate, the temporal white matter, and prefrontal white matter) (Turken et al., 2008). However, there were significantly weaker correlations in the ASD group between these white matter areas and processing speed, suggesting that these areas may not contribute as much to processing speed in the ASD group as in the typically developing group.  At the group level, these data imply that white matter contributions to processing speed are similar in both groups, albeit these relations were weaker in ASD, suggesting compensatory mechanisms more inter-individual variability in ASD.