Whole-brain functional connectivity during script-driven aggression in borderline personality disorder

Kai Ueltzhöffer,Sabine C Herpertz,Marlene Krauch,Christian Schmahl,Katja Bertsch

DOI: https://doi.org/10.1016/j.pnpbp.2019.03.004

2019-07-13

Abstract:Objective: Intense anger and anger-related aggression are frequently reported by patients with borderline personality disorders (BPD). Recent results suggest that anger-related aggression and its control is associated with a complex interplay of different neural systems in BPD. To further investigate this, we complement standard activation and seed-based connectivity analyses by examining whole-brain changes in functional connectivity during anger and reactive aggression in BPD. Methods: We reanalyzed functional MRI data from 33 women with BPD, all of them fulfilling BPD criterion 8, "anger proneness", according to DSM-IV, and 30 healthy women. Subjects performed a script-driven imagery task consisting of four phases: baseline, anger-induction by a narrative of interpersonal rejection, a narrative of directing physical aggression towards others, and relaxation. We used a data-driven, spatially constrained spectral clustering approach to parcellate the brain into 200 regions. For each script-phase and subject, we computed the full connectivity matrix using wavelet coefficient correlations in the 0.05-0.10 Hz range. We calculated the individual increase in connectivity from baseline to the anger-induction and physical aggression phases by subtracting the corresponding connectivity matrices per subject, as well as the increase and decrease from the anger-induction to the aggression phase. We then applied permutation-based sampling to determine a combined threshold on the strength of individual connections and the size of the discovered networks for these difference matrices. Results: We discovered a single, large network showing a significantly stronger increase in connectivity from baseline to the aggression phase in female patients with BPD compared to healthy women. This network consisted of regions in the anterior and posterior cingulate cortex, precuneus, dorsomedial prefrontal cortex, superior and middle temporal gyrus, hippocampus, insula, ventrolateral and dorsolateral prefrontal cortex, superior parietal lobe, thalamus, precentral and postcentral gyrus, caudate, pallidum, cerebellum, middle occipital lobe, lingual gyrus, calcarine sulcus, and fusiform gyrus. Hub regions with highest node centrality were found in the right caudate and left thalamus. We found no significant differences for the increase of connectivity from baseline to anger-induction, as well as for the increase or decrease from the anger-induction to the aggression phase. Conclusions: We identified a large network showing a significantly stronger increase in connectivity from baseline to the aggression phase in female patients with BPD compared to healthy women. The regions constituting this network belong to four previously described functional networks: The frontoparietal cognitive control network, the extended default mode network, the visual system, and the motor system. This stronger increase in connectivity between regions of different functional brain systems associated with cognitive control of behavior, socio-affective and self-referential thinking, as well as salience processing and emotion regulation, visual perception, and action is mediated via hubs in the thalamus and caudate, i.e., core components of the thalamocorticostriatal motor loop essential for action selection and initiation. These findings suggest increased interaction of prefrontal cognitive control processes with thalamocorticostriatal action-selection processes in female patients with BPD during the processing of aggressive action impulses, which are facilitated by states of high emotional salience and associated processes of self-referential and social processing, and ineffective emotion regulation.