Functional coherence of insula networks is associated with externalizing behavior

Abstract

The externalizing spectrum encompasses a range of maladaptive behaviors, including substance-use problems, impulsivity, and aggression. Although previous literature has linked externalizing behaviors with prefrontal and amygdala abnormalities, recent studies suggest insula functionality is implicated. This study investigated the relation between insula functional coherence and externalizing in a large community sample (N = 244). Participants underwent a resting functional MRI scan. Three nonartifactual intrinsic connectivity networks (ICNs) substantially involving the insula were identified after completing independent components analysis. Three externalizing domains-general disinhibition, substance abuse, and callous aggression-were measured with the Externalizing Spectrum Inventory. Regression models tested whether within-network coherence for the 3 insula ICNs was related to each externalizing domain. Posterior insula coherence was positively associated with general disinhibition and substance abuse. Anterior insula/ventral striatum/anterior cingulate network coherence was negatively associated with general disinhibition. Insula coherence did not relate to the callous aggression domain. Follow-up analyses indicated specificity for insula ICNs in their relation to general disinhibition and substance abuse as compared with other frontal and limbic ICNs. This study found insula network coherence was significantly associated with externalizing behaviors in community participants. Frontal and limbic ICNs containing less insular cortex were not related to externalizing. Thus, the neural synchrony of insula networks may be central for understanding externalizing psychopathology.

Figure 1. ICA-Derived Insula Intrinsic Connectivity Networks (Primary Networks)

Note: Three bilateral intrinsic connectivity networks (ICNs) overlapping at least 750 voxels with atlas-defined insula (red mask), including: A) posterior insula and Heschl’s gyrus (purple), B) anterior insula, ventral striatum, and anterior cingulate cortex (orange), and C) anterior insula and orbitofrontal cortex (green). The images are in neurological convention (i.e., left is left).

Figure 2. Suppression Effects for Posterior and Anterior Insula Portions with General Disinhibition

Note: Illustration of cooperative suppression between two insula intrinsic connectivity networks (ICNs) with the general disinhibition domain. All variables are in z-score form. A) Small to negligible pairwise correlations between the posterior insula and anterior insula, ventral striatum, and anterior cingulate cortex (AI-VStr-ACC) networks with general disinhibition domain. B) Increased coefficients for both the posterior insula and AI-VStr-ACC when included in the regression model together. Turquoise points represent the general disinhibition scores along the gray, fitted regression plane. The ICN coefficients in the regression model displayed were computed using only two predictors (which does not include the covariates used in the main analyses).