Distinctive modularity and resilience of structural covariance network in first-episode antipsychotic-naive psychoses

Abstract

AbstractObjectiveDivergent findings in structural brain alterations in psychoses suggest that often-observed regions may exist within a network of susceptible regions. We built structural covariance networks (SCN) of volumes, cortical thickness, and surface area using the Human Connectome Project atlas-based parcellation of 358 regions on structural MRI data from 79 first-episode antipsychotic-naive psychosis patients (FEAP) and 68 controls.MethodsUsing graph theoretic methods, we obtained representative graph metrics of integration, segregation, resilience, centrality, importance, community structure, and hub distribution for all 3-morphometric features. We compared networks for resilience by simulated removal (“attacks”) of nodes and hubs, and simulated replacement of correlations among nodes in FEAP with that of controls using DeltaCon metric.ResultsVolume and thickness SCNs of FEAP showed differences in many graph metrics in opposite directions compared to controls but not surface area SCN. Network resilience did not show differences in the replacement simulation and hub “attacks, but betweenness centrality-based node “attacks” showed FEAP network disintegrating with relatively fewer node removals with preserved global efficiency compared to controls. In FEAP, communities consisted of regions from different lobes and hubs were more distributed than in controls.ConclusionsOur findings suggest decreased heterogeneity and less differentiated community structure of FEAP network that may not be attributed to illness chronicity and medications. Contribution of cortical thickness but not surface area suggests that disease/developmental processes may preferentially affect cortical thickness. Near-similar resilience of FEAP and control networks may shield controls from developing illness but may prevent fuller long-term recovery in FEAP.