Cortical folding abnormalities in autism revealed by surface-based morphometry

Abstract

We tested for cortical shape abnormalities using surface-based morphometry across a range of autism spectrum disorders (7.5-18 years of age). We generated sulcal depth maps from structural magnetic resonance imaging data and compared typically developing controls to three autism spectrum disorder subgroups: low-functioning autism, high-functioning autism, and Asperger's syndrome. The low-functioning autism group had a prominent shape abnormality centered on the pars opercularis of the inferior frontal gyrus that was associated with a sulcal depth difference in the anterior insula and frontal operculum. The high-functioning autism group had bilateral shape abnormalities similar to the low-functioning group, but smaller in size and centered more posteriorly, in and near the parietal operculum and ventral postcentral gyrus. Individuals with Asperger's syndrome had bilateral abnormalities in the intraparietal sulcus that correlated with age, intelligence quotient, and Autism Diagnostic Interview-Revised social and repetitive behavior scores. Because of evidence suggesting age-related differences in the developmental time course of neural alterations in autism, separate analyses on children (7.5-12.5 years of age) and adolescents (12.75-18 years of age) were also carried out. All of the cortical shape abnormalities identified across all ages were more pronounced in the children. These findings are consistent with evidence of an altered trajectory of early brain development in autism, and they identify several regions that may have abnormal patterns of connectivity in individuals with autism.

Figure 1.

Method for generating sulcal depth maps. A, Coronal section through a high-resolution T1-weighted MRI for an individual subject. Hemispheres are cropped and analyzed separately. B, A segmentation boundary running midway through cortical gray matter (shown overlaid on the MRI in pink) is used to generate a fiducial surface (C). D, A cerebral hull boundary is created by dilating and expanding the segmentation boundary and a cerebral hull surface is generated (E). F, Sulcal depth maps are generated by calculating the distance from each node on the fiducial surface to the nearest node on the cerebral hull surface. Darker areas represent deeper sulcal depth measurements.

Figure 2.

T-maps for cluster size analyses (A) and interhemispheric correlation analyses (B) for each subgroup comparison across all ages overlaid on the PALS-B12 inflated and very inflated surfaces. A, T-statistic maps are thresholded to p = 0.01. Clusters that surpass the size threshold for significance are identified by arrows (left parietal operculum in HFA group and left frontal operculum in LFA group). B, T-correlation maps are thresholded to twice the t-value (p = 0.01) used in corresponding single hemisphere analyses. Bilateral clusters that surpass the size threshold are identified by arrows (intraparietal sulcus in the ASP group and parietal operculum in the HFA group).

Figure 3.

A, Schematic of a fiducial surface contour and the overlying cerebral hull. The sulcal depth measurement, denoted by the dotted line, is the distance from a single node on the fiducial surface to the nearest node on the cerebral hull. B, C, Schematics of two different anatomical configurations that could give rise to differences in sulcal depth measurements between groups. The gray lines represent the control scenario. B, Deeper sulcus: the relative position of the gyri and sulcus are the same, but the red fiducial surface has a deeper fundus than the control. C, More prominent gyrus: the difference in height of one gyrus causes the cerebral hulls to diverge. The green fiducial surface has a deeper sulcal depth measurement than the control, but the region associated with the depth difference is in the neighboring gyrus, not in the fundus of the sulcus. No depth difference appears on the gyrus itself in C because sulcal depth is always zero at the crown of a gyrus.

Figure 4.

Visualization of the folding abnormality in the LFA group in children. A, The LFA cluster in the frontal operculum identified by the cluster size analysis is shown on the PALS-B12 very inflated surface. The black line denotes the location of the coronal slice depicted in C. B, Plots of individual mean depths within the significant cluster for LFA and TD groups. C, Outlines of group-averaged fiducial surfaces (LFA in red; TD in blue) are overlaid onto the MNI-152 MRI template. The black arrow indicates the location of the cluster in the frontal operculum. Note that the red and blue surfaces overlap almost completely at this location. In contrast, the yellow arrow depicts the gyral region in the inferior frontal gyrus that is different between groups. The LFA group extends more ventrally than the TD group. This difference occurs only in the left hemisphere. D, The green outlines on the group-averaged LFA and TD fiducial surfaces depict the gyral region most strongly associated with the sulcal depth abnormality. The yellow outlines depict a more liberal estimation of the location of the gyral shape abnormality for the LFA group (pars opercularis of the inferior frontal gyrus and ventral portions of the precentral and postcentral gyri). E, The affected area for the LFA group is overlaid onto the PALS-B12 inflated surface with Brodmann areas identified.

Figure 5.

Visualization of the folding abnormality in the HFA group in children. A, The HFA cluster in the parietal operculum identified by the cluster size analysis is shown on the left hemisphere PALS-B12 very inflated surface. The black line denotes the location of the coronal slice depicted in C. B, Plots of individual mean depths within the left and right significant clusters for HFA and TD groups. C, Outlines of group-averaged fiducial surfaces (HFA in red; TD in blue) are overlaid onto the MNI-152 MRI template. The black arrow indicates the location of the cluster in the parietal operculum. The HFA surface is deeper than the TD surface by ∼1 mm. D, The yellow outline overlaid onto group-averaged HFA and TD fiducial surfaces depicts the location of the putative gyral shape abnormality in ventral postcentral gyrus. Note that although the right hemisphere is not depicted in A or D, the abnormality is bilateral, and the corresponding region in the right hemisphere is also affected.

Figure 6.

Visualization of a folding abnormality in the intraparietal sulcus of children in ASP group. A, B, Bilateral clusters identified from the intrahemispheric correlation analyses are projected on to left and right group-averaged fiducial surfaces for the ASP group (A) and TD control group (B). The intraparietal sulcus is deeper in the ASP group than the TD control group. C, Outlines of group-averaged fiducial surfaces are overlaid on an MRI template. Arrows depict the intraparietal sulcus. The ASP fiducial surface (red) is significantly deeper in the intraparietal sulcus than the TD fiducial surface (blue). D, E, Left hemisphere clusters are projected onto the PALS-B12 inflated surfaces with visuotopic areas (D) and Brodmann areas (E) identified (Van Essen, 2005).

Figure 7.

Correlations between sulcal depth and age (A) and full-scale IQ (B) for the left intraparietal sulcus cluster identified in the ASP group across all ages. Depth is correlated with age in the ASP group but not the TD group. Depth is correlated with full-scale IQ in both ASP and TD groups.