Aberrant Interregional Correlations of Cortical Thickness In Autism Spectrum Disorders

Background: Prior neuroimaging studies assessing either white matter integrity or blood oxygenation suggest that aberrant neural connectivity is characteristic of Autism Spectrum Disorders (ASD).  Recent work indicates that hubs or networks can be derived using interregional correlations of cortical thickness among typically developing individuals.  No study to date has investigated whether these interregional correlations of cortical thickness are atypical in ASD. 

Objectives: Examine patterns of interregional cortical thickness correlations in 41 high functioning males with ASD as compared to a group of 40 typically developing (TD) males matched on age, IQ, and handedness.

Methods: T1-weighted 3T MPRAGE MRI scans were acquired from 41 males with an ASD (diagnoses based on DSM-IV criteria and scores from the Autism Diagnostic Inventory and/or Autism Diagnostic Observation Schedule) and from 40 TD males (screened for the presence of learning, developmental, psychiatric, neurological disorders) matched group-wise on age, IQ, and handedness.  Cortical reconstruction and volumetric segmentation were performed with the FreeSurfer image analysis suite.  Once cortical models were completed, the cerebral cortex was parcellated into 33 regions per hemisphere based on gyral and sulcal structure.  Average cortical thickness was computed for these 66 individual gyral regions.  Patterns of covariation in thickness among the gyral regions were obtained using Multi-dimensional scaling and K-Means clustering.  K-Means clusters were determined solely on the basis of the TD ROI-ROI correlation matrices.

Results: Four clusters were derived from the analyses:  Cluster 1 (C1)=cingulate cortex, Cluster 2 (C2)=lateral and inferior temporal cortex, Cluster 3 (C3)=medial temporal cortex, Cluster 4 (C4)=the remainder of the cortex.  Though global patterns of cortical thickness correlation were largely similar between groups, several significantly discrepant (p<.05 after Bonferroni correction) interregional correlations were found.  The ASD group had significantly greater correlations between C1 & C1, C1 & C2, and C1 & C3 than the TD group.  In contrast, the TD group had significantly greater correlations between C1 & C4, C2 & C4, C3 & C4, and C4 & C4 than the ASD group.

Conclusions: Consistent with prior findings of atypical connectivity in ASD, the present study demonstrates that ASD is associated with aberrant interregional correlations of cortical thickness.  The pattern of findings is striking in that (1) less differentiation of interregional relationships characterized the ASD group, (2) cingulate cortex is more strongly associated with the rest of the cortex in ASD than in TD individuals, and (3) group differences in correlations were not localized to regions previously associated with accelerated cortical thinning in this identical sample of ASD individuals (Wallace et al., 2010).  Of note, a large longitudinal study of typical development shows that the thickness of cingulate cortex follows an independent trajectory than the rest of the cortex during adolescence/young adulthood (Shaw et al., 2008).  The failure of cingulate cortex to decouple from the rest of the cortex in ASD during this age range may be the result of synaptogenesis gone awry.  Examining these relationships during different developmental periods and longitudinally will help to pinpoint the ontogeny of group differences.