Objectives: To compare volumes of clustered white matter parcellation units in low-IQ children with autistic disorder (LAD) and non-autistic low IQ children (NALIQ) with those from typically developing (TD) and high-functioning autistic disorder (HAD) groups.
Methods: Participants (aged 4.5-11 years) were assessed during preschool years using the Wing Autistic Disorder Interview Checklist and by expert clinicians. HAD children (9M,) had non-verbal IQ ≥ 80 while LAD (13M, 2F) and NALIQ (4M, 6F) had non-verbal IQ < 80; there were 29 TD children (14M/15F). MRI scans were collected during school-age years beginning in 1989 on either a GE 1.5 Signa or Siemens 1.5T scanner. Grey-white segmentation, cortical parcellation, and white matter parcellation of T1-weighted images were performed according to extensively validated methods developed by the Center for Morphometric Analysis, MGH (Meyer, 1999; Makris, 1999). Superficial (radiate) white matter parcellation units were determined by contiguity to gyral-based cortical parcellation units; deep white matter was divided using an algorithm keyed to topographical landmarks pertinent to tract architecture. To perform group comparisons, effect sizes were calculated (mean volume for diagnostic group minus mean volume of control group divided by pooled standard deviation of both groups). Because the volumes were found to be normally distributed, we used univariate regression models to compare each group to controls; sex, age, and diagnostic status were included as covariates. Proportionalized volumes (volume of interest divided by total brain volume) were also analyzed.
Results: We found significant differences in volumes of the superficial (radiate) white matter underlying the frontal (p=.001), temporal (p=.018), and occipital (p=.018) lobes (HAD > LAD > TD). When LAD were compared to TD using proportionalized volumes, only frontal lobe white matter was significantly larger in the mixed gender analysis (p=.011) but less so for boys only (p=.041). In contrast, we found no significant differences between NALIQ and TD for any white matter regions.
Conclusions: These results suggest that LAD shows a lesser degree of superficial white matter enlargement than HAD. This finding is distinct what is seen in the IQ-matched NALIQ group, which showed no significant differences from TD, suggesting that the effects of autism are distinct from the effects of low IQ. While the number of females in the current study was relatively small, our preliminary results suggest that girls may exhibit disproportionate volumetric changes. This finding underscores the importance of studying females with autistic disorder.