Late-life depression (LLD) has been associated with both generalized and focal neuroanatomical changes including gray matter atrophy and white matter abnormalities. However, previous literature has not been consistent and, in particular, its impact on the topology organization of brain networks remains to be established. In this multimodal study, we first examined cortical thickness, and applied graph theory to investigate structural covariance networks in LLD. Thirty-three subjects with LLD and 25 controls underwent T1-weighted, fluid-attenuated inversion recovery and clinical assessments. Freesurfer was used to perform vertex-wise comparisons of cortical thickness, whereas the Graph Analysis Toolbox (GAT) was implemented to construct and analyze the structural covariance networks. LLD showed a trend of lower thickness in the left insular region (p < 0.001 uncorrected). In addition, the structural network of LLD was characterized by greater segregation, particularly showing higher transitivity (i.e., measure of clustering) and modularity (i.e., tendency for a network to be organized into subnetworks). It was also less robust against random failure and targeted attacks. Despite relative cortical preservation, the topology of the LLD network showed significant changes particularly in segregation. These findings demonstrate the potential for graph theoretical approaches to complement conventional structural imaging analyses and provide novel insights into the heterogeneous etiology and pathogenesis of LLD.
Keywords: Brain network; Dementia; Depression; Graph theory; Imaging; Late-life depression; Psychiatry.
Copyright © 2016 The Authors. Published by Elsevier Inc. All rights reserved.