Sex-based neurobiological heterogeneity in autism is poorly understood. Research is disproportionately biased to males, leading to an unwarranted presumption that autism neurobiology is the same across sexes. Previous neuroimaging studies using amalgamated multi-center datasets to increase autistic female samples are characterized by large statistical noise.
We used a better-powered dataset of 1,183 scans of 839 individuals-299 (467 scans) autistic males, 74 (102 scans) autistic females, 240 (334 scans) control males and 226 (280 scans) control females-to test two whole-brain models of overall/global sex-modulations on autism neuroanatomy, by summary measures computed across the brain: local magnitude model, where the same brain regions/circuitries are involved across sexes but effect sizes are larger in females, indicating quantitative sex-modulation; and spatial dissimilarity model, where the neuroanatomy differs spatially between sexes, indicating qualitative sex-modulation. The male and female autism groups were matched on age, IQ and autism symptoms. Autism brain features were defined by comparisons to same-sex controls.
Across five metrics (cortical thickness, surface area, volume, mean absolute curvature, and subcortical volume), we found no evidence supporting the local magnitude model. We found indicators supporting the spatial dissimilarity model on cortical mean absolute curvature and subcortical volume, but not other metrics.
The overall/global autism neuroanatomy in females and males does not simply differ quantitatively in the same brain regions/circuitries. They may differ qualitatively in spatial involvement in cortical curvature and subcortical volume. The neuroanatomy of autism may be partly sex-specific. Sex-stratification to inform autism preclinical/clinical research is needed to identify sex-informed neurodevelopmental targets.

Copyright © 2021. Published by Elsevier Inc.

Author