Williams Syndrome-Specific Neuroarchitectural Profile and Its Associations with Cognitive Features

Abstract

Williams Syndrome (WS), a rare genetic disorders caused by hemizyous deletion of ~26 genes on the chromosome 7, has unique cognitive features and neuroanatomic abnormalities. Limited in statistical power due to its rareness had led to inconsistent in many direct comparisons using structural magnetic resonance imaging (MRI), and their associations with cognitive features of WS are not clear. Here, we used a novel approach to derive a WS specific neuroarchitectural profile and tested its association with cognitive features of WS. Using a WS adult cohort (n = 43), we trained a logistic elastic-net model to extract a sparse representation of WS specific neuroarchitectural profile. The predictive performances are robust within the training cohort (leave one out cross-validation AUC = 1.0) and generalized well in an independent teenager WS cohort (n = 60, AUC = 1.0). The WS specific neuroarchitectural profile includes multiple MRI measurements in the orbitofrontal cortex, superior parietal cortex, Sylvian fissures, and basal ganglia, whereas its variations reflect the underlying size of hemizygous deletion, and mediated the disease impact on the cognitive features of WS. In this study, we demonstrate the robustness of the derived WS specific neuroarchitectural profile, suggesting the joint developmental abnormalities in the cortical-subcortical circuitry cause the unique features of WS cognition.