Quantitative Analysis of the Shape of the Corpus Callosum in Autism

Background: Multiple neuroimaging studies of the corpus callosum have suggested the presence of morphometric abnormalities in autistic patients. There is at present a convergence of findings from structural studies indicating that any significant abnormality is always manifested as a smaller corpus callosum. This applies to both areal (i.e., single midsaggital measures) and volumetric assessments. The fact that the corpus callosum is smaller in autism is all the more striking when considering that total brain size appears, on average, to be increased.

Objectives: The study aimed to elucidate the nature of the corpus callosum abnormality in autism. It complemented previous morphometric studies by quantitating the shape of the corpus callosum boundary.

Methods: Seventeen high-functioning, autistic individuals were recruited (fourteen male, three female, all between 16 and 51 years of age). An equal number of non-autistic control subjects were enrolled, matched pairwise by age and sex with the autistic participants. T1-weighted images were acquired with Siemens MAGNETOM Vision 1.5 T scanners using an MPRAGE acquisition sequence. The corpus callosum was segmented from each image using a probabilistic model for corpus callosum shape. Boundary surfaces of the segmented corpora callosa were mapped into a standardized pseudocylindrical coordinate system (z, φ, ρ) to facilitate their comparison with each other. For each point z along a curvilinear axis from the genu to the splenium, and for each angle φ measured clockwise about this axis from anatomical right, ρ(z, φ) is the perpendicular distance from the axis to the boundary surface. The transformed surfaces were aggregated pointwise into a statistical parametric map of pairwise t statistics with 16 degrees of freedom. Regions of statistically significance were derived from the associated P-values using the method of Benjamini and Hochberg with a false discovery rate q^* = 0.05.

Results: White matter in autism was reduced bilaterally along the body of the corpus callosum, apart from an increase in the region proximal to the right cingulate gyrus. There was also a reduction at the extreme posterior end of the structure. There was otherwise no significant difference around the mid-sagittal plane.

Conclusions: Results from our study indicate a generalized reduction of the corpus callosum in autism. The findings acquire relevance from previous neuroimaging studies that varied among themselves in regards to a preferred anatomical subdivision (e.g., genu, body, splenium). Generalized findings, involving the rostro-caudal extent of the corpus callosum, is expected from a mechanism involving corticalization, i.e. a mitotic event affecting periventricular germinal cells and the total number of cortical minicolumns.