Sex differences in trajectories of cortical development in autistic children from 2-13 years of age

Abstract

Previous studies have reported alterations in cortical thickness in autism. However, few have included enough autistic females to determine if there are sex specific differences in cortical structure in autism. This longitudinal study aimed to investigate autistic sex differences in cortical thickness and trajectory of cortical thinning across childhood. Participants included 290 autistic (88 females) and 139 nonautistic (60 females) individuals assessed at up to 4 timepoints spanning ~2-13 years of age (918 total MRI timepoints). Estimates of cortical thickness in early and late childhood as well as the trajectory of cortical thinning were modeled using spatiotemporal linear mixed effects models of age-by-sex-by-diagnosis. Additionally, the spatial correspondence between cortical maps of sex-by-diagnosis differences and neurotypical sex differences were evaluated. Relative to their nonautistic peers, autistic females had more extensive cortical differences than autistic males. These differences involved multiple functional networks, and were mainly characterized by thicker cortex at ~3 years of age and faster cortical thinning in autistic females. Cortical regions in which autistic alterations were different between the sexes significantly overlapped with regions that differed by sex in neurotypical development. Autistic females and males demonstrated some shared differences in cortical thickness and rate of cortical thinning across childhood relative to their nonautistic peers, however these areas were relatively small compared to the widespread differences observed across the sexes. These results support evidence of sex-specific neurobiology in autism and suggest that processes that regulate sex differentiation in the neurotypical brain contribute to sex differences in the etiology of autism.

Fig. 1. Autistic sex differences in rates of cortical thinning and cortical thickness across childhood.

Cortical regions with significant sex-by-diagnosis interactions in trajectories of cortical thinning (A) and in cortical thickness at Times One, Three and Four (B). Line plots indicate trajectories for highlighted clusters, vertical dotted lines indicate mean ages at Times One, Three and Four. Bar plots indicate proportion of functional networks affected (total total cortical surface, DMN default mode, DA dorsal attention, FP frontal parietal, L imbic, SM somatomotor, VA ventral attention, V visual). A Significant age-by-sex-by-diagnosis interactions in cortical thinning trajectories were predominately identified as regions with faster thinning in autistic (ASD) females and slower thinning in ASD males compared to sex matched typically developing nonautistic (TD) individuals. These clusters involved 4.74% of the total cortex and were distributed across all seven functional networks, but most proportionately affected the SM (9.46%), FP (6.96%), VA (6.38%), and DA (5.33%). In contrast, regions where autistic females had slower and autistic males had faster rates of cortical thinning incorporated 0.95% of the total cortical surface, most proportionately affecting the L (4.08%) and SM (2.15%). B Significant sex-by-diagnosis interactions in cortical thickness at Time One were predominately characterized by regions of thicker cortex in autistic females and thinner cortex in autistic males compared to sex matched TD individuals. These clusters incorporated 5.92% of the total cortical surface, most proportionately affecting the V (7.02%), DA (6.95%), DMN (6.94%), SM (6.88%), and VA (6.68%). Regions where autistic females had thinner cortex and autistic males thicker cortex were isolated to select regions comprising 0.25% of the total cortical surface. Across childhood faster cortical thinning across several cortical regions for autistic females resulted in shifts in the spatial extent of significant sex-by-diagnosis interactions at Time Three and Four. By Time Four, regions in which autistic females had thicker cortex and autistic males thinner cortex reduced to 2.17% of the total cortical surface, most proportionately affecting the L (5.75%), SM (3.08%), DMN (2.26%) and VA (2.38%). Additionally, at Time Four regions in which autistic females had thinner cortex and autistic males thicker cortex increased to 0.80% of the total cortical surface and most proportionately affected the FP (1.67%) and SM (1.01%).

Fig. 2. Within sex effects of autism on rates of cortical thinning and cortical thickness across childhood.

Cortical regions with significant effects of autism diagnosis between autistic and nonautistic females, and between autistic and nonautistic males in rates of cortical thinning (A) and in cortical thickness at Times One (B), Three (C), and Four (D) are highlighted. Bar plots indicate proportion of functional networks affected within each sex (total total cortical surface, DMN default mode, DA dorsal attention, FP frontal parietal, L limbic, SM somatomotor, VA ventral attention, V visual). A Autistic females had significantly faster rates of cortical thinning from Time One - Four in several regions compared to autistic males (8.32% of cortex in females compared to 2.21% in males), in contrast autistic males and females had a comparable number of regions with less thinning than nonautistic participants of the same sex (1.05% of cortex in males compared to 1.18% in females). B At Time One autistic females predominately had significant increases in cortical thickness across several cortical regions. This was in contrast to autistic males whom at Time One had less regions of thicker cortex (8.83% of cortex in females compared to 1.67% in males) and more regions with thinner cortex (4.12% of cortex in males compared to 0.60% in females). C, D Faster thinning from Time One to Four in autistic females resulted in a reduction in cortical regions with thicker cortex at Times Three (4.81%) and Four (3.29%) compared to Time One, and more regions with thinner cortex (Time Three = 1.01%, Time Four = 2.26%). Autistic males continued to have regions with both significant increases and decreases in cortical thickness in comparative proportions across the cortex at both Times Three (0.99% thicker, 4.54% thinner) and Four (0.66% thicker, 3.65% thinner).

Fig. 3. Spatial Correspondence of typical sex differences and sex-by-diagnosis effects in rates of cortical thinning and cortical thickness across childhood.

Spatiotemporal FDR corrected and uncorrected F maps of typically developing (TD) sex differences in cortical thinning trajectories (A1), and cortical thickness at Times One (B1), Three (C1) and Four (D1). Spin tests were used to test the spatial correspondence of these uncorrected F maps with maps of diagnosis-by-sex interactions in cortical thinning trajectory (A2), and cortical thickness at Times One (B2), Three (C2), and Four (D2). Spin tests (n = 1000 permutations) were found to be statistically significant for all three measures (p < 0.001), indicating significant overlap between regions which display typical sex differences and sex-by-diagnosis interactions in cortical thinning trajectories and cortical thickness measures across childhood.

Fig. 4. Shared Differences in Cortical Thickness and Thinning Between Autistic Males and Females.

A Autistic males and females both had clusters of increased thinning centered within regions including left inferior parietal, middle temporal, lateral occipital, regions, the left supramarginal gyrus and rostral anterior cingulate, as well at regions of the right superior parietal, middle frontal, and postcentral gyri. B At Time One autistic males and females both had significantly thinner left parahippocampal and superior frontal gyri, as well as increased thickness in right inferior parietal, superior parietal, lateral occipital, medial orbitofrontal, praracentral, and fusiform gyri. C At Time Three regions in which autistic males and females both had thinner cortex included bilateral parahippocampal and paracentral gyri as well as the left rostral middle frontal gyrus. Thicker cortex at Time Three was observed in autistic males and females within bilateral lateral occipital cortex and the right fusiform and inferior temporal gyri. D At Time Four autistic males and females both had decreased cortical thickness with clusters centered bilaterally in the parahippocampal gyrus, as well as the left supramarginal, paracentral and middle frontal gyrus and right precentral and inferior parietal regions. Additionally, at Time Four both autistic males and females had increased cortical thickness within clusters centered in the right fusiform and superior temporal gyri.