Sex Differences in Brain Structure of Preschool-Aged Children with Autism Spectrum Disorder

Background:  Autism spectrum disorder (ASD) affects 1 in 88 children in the United States, but the disorder is much more common in boys than in girls. Although this disparate sex ratio is among the most highly replicated findings in studies of ASD, sex differences in the neuropathology of ASD remain poorly understood, particularly in young children. Normative brain development is sexually dimorphic, but we do not yet know how biological sex affects the developing brain in ASD.  

Objectives:  We evaluated within- and between-sex differences in cortical gray matter volumes in a large cohort of preschool-aged children with ASD and age-matched typically developing controls. 

Methods:   We acquired structural T1-weighted MRIs in 174 children with ASD (140 male, 34 female) and 90 age-matched typically developing (TD) controls (59 male, 31 female). Mean age at time of MRI acquisition was 37 months. Cortical gray matter volumes, parcellated into 34 gyral regions in each hemisphere, were obtained through Freesurfer v5.1.0. The automated processing stream includes motion correction of MRI image, removal of non-brain tissue, transformation to Talairach space and intensity normalization. We utilized multivariate pattern analysis to identify patterns of brain regions that discriminate between sex and diagnosis. We performed cross-validated linear support vector machine (SVM) analyses controlling for total gray matter volume and age for the following comparisons: ASDf v TDf, ASDm v TDm, ASDf v ASDm, TDf v TDm, and ASD v TD (with sex as additional covariate). We also recursively selected features that contributed the greatest weights for the classifications by performing t-tests (alpha = 0.1).  All steps were performed iteratively using training sets (not including test set) to remain unbiased and avoid overgeneralization. We report: (1) classification accuracy of test sets, (2) feature weights that contributed to the classification, and (3) the overlap and non-overlap in brain regions across different comparisons.

Results:  Our preliminary results from a subset of the full sample (156 ASD, 77 TD) suggest that classification accuracy was greatest for the ASDf v ASDm comparison (86%) suggesting that male and females with ASD do indeed have distinct patterns of neuropathology. Accuracy for all other comparisons ranged from 65-70%. For the within sex comparisons (ASDf v TDf and ASDm v TDm), the only overlapping region is left superior temporal gyrus, a region that has been widely implicated in ASD and has many functions, including semantic language processing and social communication. Regions specific to the ASDf v TDf comparison included regions in neural systems related to understanding mental states of others (right temporal pole) expressive language and reception of facial communication (bilateral pars opercularis, cingulate gyrus) as well as right postcentral gyrus, right superior frontal gyrus, and left transverse temporal gyrus. Interestingly, two of these regions, right temporal pole and right pars opercularis were also sexually dimorphic in typically developing controls (TDf v TDm) comparisons.

Conclusions:  These data suggest that females and males with ASD have patterns of neural abnormalities that are more dissimilar than similar.