Sex-Modulated Structural Covariance Networks in Autism

Thursday, May 11, 2017: 12:00 PM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
R. A. Bethlehem1, M. V. Lombardo2,3, A. N. Ruigrok1, B. Auyeung4, J. Suckling5, E. Bullmore5, M. Consortium6, S. Baron-Cohen1, B. Chakrabarti7 and M. C. Lai8, (1)University of Cambridge, Cambridge, United Kingdom, (2)University of Cambridge, Sacramento, CA, (3)University of Cyprus, Nicosia, Cyprus, (4)University of Edinburgh, Edinburgh, United Kingdom, (5)Brain Mapping Unit, Department of Psychiatry, University of Cambridge, Cambridge, United Kingdom, (6)Institute of Psychiatry, Psychology and Neuroscience, London, United Kingdom, (7)School of Psychology and Clinical Language Sciences, University of Reading, Reading, United Kingdom, (8)Psychiatry, University of Toronto, Toronto, ON, CANADA

Atypical neural connectivity has been proposed as a biomarker for autism, entailing decreased fronto-posterior and enhanced parietal-occipital connectivity, reduced long-range and increased short-range connectivity, and temporal binding deficits. Empirical findings vary substantially depending on the aspects of connectivity examined, the developmental stage of the individual, the spatial and temporal scales, task versus no-task conditions, how motion artefacts are handled, and the specific neural systems of concern. Heterogeneity in connectivity findings is likely further due to key aspects of sample variation, such as sex/gender.


(1) To explore if biological sex moderates the characteristics of structural covariance in autism; (2) To explore the atypical connectivity hypothesis of autism using structural covariance network analysis.


Structural covariance of T1 weighted MRI data from 117 individuals in 4 groups (males with autism N=25, neurotypical males N=33, females with autism N=30, neurotypical females N=29), matched for age and IQ. IQ was in the average range or above. Data was analyzed using Freesurfer segmentation for cortical thickness and custom Matlab code to assess graph theoretical and network properties. We specifically investigated metrics describing the relation between cortical thickness and network topology in reference to long versus short-range connectivity. For example, we looked at the relation between anatomical distance and covariance correlation strength and properties of the degree distribution. Statistical significance was assessed using pair-wise Monte-Carlo permutation tests.


Inter-regional correlation strength as a function of Euclidean distance differed across all 4 groups. Specifically, the female autism group showed a much steeper decline in correlation strength with increased anatomical distance, indicating a balance that more strongly favours local over global connections. In addition, the overall cumulative distribution of the degree of network parcels showed a strong sex difference, with the male control group having a reduced incidence of high degree nodes. This difference was absent in the autism group.


We report evidence for atypical connectivity in adults with autism compared to neurotypical adults, but the pattern is heterogeneous, both moderated by sex and dependent on the metrics examined. More fine-grained descriptions on patterns of atypical connectivity are needed. These results challenge an over-simplified view that general hypo- or hyper-connectivity marks the atypical neurobiology of autism.

See more of: Genetics
See more of: Genetics