Synergistic Co-Activation Probabilities of Large-Scale Resting State Networks in Major Depressive Disorder

Abstract

Background: Major depressive disorder (MDD) involves subtle, distributed alterations across multiple large-scale resting-state brain networks (RSNs), highlighting the need for integrative approaches to uncover synergistic network patterns driving clinical symptoms.

Methods: In this study, we employed a dynamical systems approach to investigate patterns of simultaneous RSN activation - i.e. co-activation - in 867 participants, including 487 healthy controls (HC), 175 patients with current MDD (cMDD), and 205 with remitted MDD (rMDD) from the Marburg-Münster Affective Disorders Cohort Study. Using a pairwise Maximum Entropy Model, we estimated RSN co-activation probabilities based on resting state fMRI data of seven RSNs-default mode network (DMN), frontoparietal network (FPN), sensorimotor network (SMN), visual network (VIS), salience network, dorsal attention network (DAN), and language network (LAN)-capturing 128 possible states of co-activation.

Results: General linear models revealed elevated co-activation probabilities in cMDD, particularly for states involving DMN, FPN, and VIS, with the co-activation state involving DMN, VIS, DAN, FPN, and LAN showing the strongest association with MDD diagnosis, clinical status, and symptom severity. Canonical Correlation Analysis (CCA) on the full sample further identified two distinct network-symptom profiles: Canonical variate (CV) 1 linked high DMN and DAN co-activation probabilities to cognitive, insomnia, and mood/anhedonia symptoms, while CV2 tied SMN and VIS to cognitive and somatic symptom domains.

Conclusions: These results demonstrate that MDD, especially during acute episodes, is marked by a dominance of DMN, FPN, and VIS co-activation, pointing to altered dynamic network organization. They highlight how changes in brain state dynamics are linked to MDD symptoms.

Keywords: Canonical Correlation Analysis (CCA); Co-activation; Major Depressive Disorder; Maximum Entropy Model; Resting State fMRI; dynamical systems.