Fig. 1: Subnetwork of the structural connectome with decreased connectivity in depressed individuals without anxiety disorder (MDD) and increased connectivity in depressed individuals with comorbid anxiety disorder (MDD + ANX), as compared to healthy controls (HC).

Network of white matter fiber tracts capturing differences among healthy controls (HC) as well as depressed individuals without (MDD) and with comorbid anxiety disorder (MDD + ANX). A Axial view of the network identified via NBS. Node size represents the number of affected edges connected to each region, with larger nodes indicating hub regions within the network. Nodes with the highest number of affected edges are labeled with their anatomical region names. Edge thickness reflects the effect size (t statistic) for each connection, with thicker edges indicating stronger group differences. B Between-group differences in the identified network, with effect sizes shown as Cohen’s d. C Frequency of edges connecting frontal, temporal, parietal, occipital, and insular regions in the identified network, compared to the frequency of the respective edges in the full connectome. D Individual estimates of total fiber length across all edges in the identified network versus the first of 1000 size- and topology-matched random networks, with comparison of mean total fiber length (averaged across individuals) in the identified network to mean total fiber lengths in all 1000 randomly selected networks (upper right corner). E Pattern of between-group differences in robustness checks (see Supplementary 9 for effect sizes). Robustness checks included: (1) separate analyses for males and females; (2) separate analyses for participants from either study site; (3) separate analyses for participants with acute, partially remitted or remitted depressive episode; (4) correcting for non-linear age effects; (5) including head motion as a covariate (derived from FSL’s eddy output); (6) excluding connectome outliers; (7) using an alternative measure of structural connectivity (streamline volume density) corrected for gray matter volumes; and (8) using the original Desikan-Killiany atlas that includes subcortical regions. See online version for colored figures.