Increased Global Covariance of Cortical Volume in Children with ASD

Background: Structural covariance (SC) is the correlation between structural brain features and this metric may reflect long-term developmental influences, the effect of plasticity and mutual trophic influences [1, 2]. Regions with high SC often belong to brain systems such as salience or default mode networks [3] which serve specific functions [4]. By examining SC in ASD it may be possible to determine which regions are developmentally linked and further our understanding of brain development in this disorder

Objectives:  We aim to examine global patterns of cortical volume SC in ASD and typically developing controls (TDC) in both children (age 7–13 years) and adolescents (age 13–18 years). We aim to: 1) characterize global patterns of cortical SC in ASD and TDC 2) determine if differences in SC exist between ASD and TDC and 3) whether changes in SC occur with age.

Methods: Structural MRIs and demographic information were obtained from ABIDE [5]. We investigate SC in a cohort of right handed male subjects which consists of 64 ASD (average age ± std: 10.88 ± 1.45 years) and 75 TDC (10.67 ± 1.35 years) children and 63 ASD (15.34 ± 1.58 years) and 61 TDC (15.49 ± 1.43 years) adolescents. MRIs were preprocessed and cortical volumes were computed using FreeSurfer [6]. SC was calculated using Pearson’s correlation between the 34 cortical volumes listed in Table 1 and this resulted in 2,278 pairs of inter and intra hemispheric SCs. Significant SC differences were determined between ASD and TDC for children and adolescents separately. Permutation testing was performed (N = 10⁶) to verify results.

Results:  Results indicate inter and intra hemispheric SC differences in ASD children and that children with ASD have higher SC than TDC for most pairs (87%) of brain regions. In adolescents this difference was not present. The median SC were as follows: ASD children: 0.59, TDC children: 0.42, ASD adolescents: 0.39 and TDC adolescents: 0.39. Highly significant ASD vs. TDC SC differences were found in children (Figure 1A) but differences were not significant in adolescents (Figure 1B) after correction for multiple comparisons. Permutation testing further validated the above result. SC pairs containing left entorhinal cortex with left rostral anterior cingulate, left fusiform, right medial orbitofrontal, and right fusiform survived Bonferroni correction (P<2.2x10^-5) in children.

Conclusions:  Many of the SC pairs that survived multiple comparisons correction in children have been implicated previously in ASD. Regions such as the right fusiform and medial orbitofrontal gyri are part of the face processing network[7] and the anterior cingulate has been implicated in social impairment[8]. This study raises the possibility that SC between the left entorhinal cortex, which plays a role in memory[9], and the above regions may underlie social impairment in ASD. We report altered cortical topography (both inter and intra) in children with ASD that normalizes during development in adolescence. To our knowledge this study is the first to characterize global cortical SC differences in ASD.