Regional Brain Volume Differences Between Males with and without Autism Are Highly Age-Dependent
Meta-analyses suggest that the atypical neurobiology of autism may vary substantially with age. This also implies that individuals with and without autism may have different trajectories of brain growth.
To test (1) if brain volumetric differences between individuals with and without autism vary across chronological ages of 7-30 years old, and (2) if the volume of age-dependent atypical brain regions in autism is associated with current social deficits.
Voxel-based morphometry (using DARTEL) was applied on structural MRI (3T) images of 104 males with autism (mean=14.6 years, standard deviation, SD=4.4) and 90 typically developing (TD) males (mean=15.0, SD=5.9). A general linear model was used to test if there are significant diagnosis-by-age interaction effects in gray matter (GM) and white matter (WM) volume. If confirmed, main effects of diagnosis among different age-bands would be contrasted in three age-stratified subsamples (TD, autism): children [7-12 years old,, n=44, 37], adolescents [13-17 years old; n=17, 44], adults [18-30 years old; n=29, 23], with age (linear term) covaried in the models. Statistical outcomes were corrected at the cluster-level with topological FDR q<0.05 (corrected for non-stationarity) and cluster-forming voxel-level p<0.001. The Taiwanese Social Responsiveness Scale (SRS) was used to quantify current autistic-like social deficits.
For absolute total volumes, no significant diagnosis-by-age interaction was detected in GM (p=0.258) or WM (p=0.179). Across ages, males with and without autism did not significantly differ in total GM (p=0.069) or WM (p=0.096) volumes. For both groups, total GM volume decreased (Spearman’s Rs=-0.298, p=0.002 in autism, Rs=-0.557, p<0.001 in TD) and WM volume increased (Rs=0.219, p=0.03 in autism, Rs=0.274, p=0.009 in TD) with age. For relative (i.e., corrected for individual total) local GM volume, first in a simple model without considering age and diagnosis-by-age interaction effects, there were only trend-level effects of diagnosis (autism>TD) at bilateral frontal pole and medial prefrontal cortex (mPFC); no significant effects were found in WM. However, in a second model testing for main effects of diagnosis, age, and diagnosis-by-age interaction, significant interaction was identified at again bilateral mPFC, as well as cuneus (mPFC: q<0.001, 6546 voxels, peak-voxel T=5.2; cuneus: q=0.036, 1627 voxels, peak-voxel T=5.0). Here, between 7-30 years old, relative GM volume decreased with age in TD males, but increased with age in males with autism. Age-stratified analyses showed that compared to TD groups, children with autism were smaller in cuneus GM, adolescents with autism were larger in WM at the premotor portion of the corticospinal tract, and adults with autism were larger in mPFC GM. In all individuals with autism, larger relative GM volumes in mPFC and cuneus were both associated with increased ‘social awareness’ difficulty (derived from an independent factor analysis on the Taiwanese SRS) (mPFC, Rs=0.298, p=0.002; cuneus, Rs=0.241, p=0.015). Other SRS subscores did not show significant correlations.
Local brain volume differences between males with and without autism vary substantially with age between 7-30 years of age. Males with and without autism may have different brain growth trajectories beyond childhood, which needs to be investigated by future longitudinal studies.