Driving brain state transitions in major depressive disorder through external stimulation

Abstract

Major depressive disorder (MDD) as a dysfunction of neural circuits and brain networks has been established in modern neuroimaging sciences. However, the brain state transitions between MDD and health through external stimulation remain unclear, which limits translation to clinical contexts and demonstrable clinical utility. We propose a framework of the large-scale whole-brain network model for MDD linking the underlying anatomical connectivity with functional dynamics obtained from diffusion tensor imaging (DTI) and functional magnetic resonance imaging (fMRI). Then, we further explored the optimal brain regions to promote the transition of brain states between MDD and health through external stimulation of the model. Based on the whole-brain model successfully fitting the brain state space in MDD and the health, we demonstrated that the transition from MDD to health is achieved by the excitatory activation of the limbic system and from health to MDD by the inhibitory stimulation of the reward circuit. Our finding provides novel biophysical evidence for the neural mechanism of MDD and its recovery and allows the discovery of new stimulation targets for MDD recovery.

Keywords: brain state transitions; excitatory stimulation; functional dynamics; major depressive disorder; whole-brain model.

FIGURE 1

Computing the probabilistic metastable substate (PMS) space for whole‐brain activity. (a) Leading eigenvector dynamics analysis (LEiDA) method for the time series in every brain region of each participant. (b) Blood‐oxygenation‐level‐dependent (BOLD) phase coherence matrix and its leading eigenvector at a given time point. (c) Clustering analysis for leading eigenvectors at all time points in all subjects. Each cluster is represented by a central vector (green, red, and blue), representing a recurrent pattern of phase coherence or substate. (d) Three PMS spaces for major depressive disorder (MDD) and health. Substates 1 and 2 are significantly different in terms of the probability of occurrence between MDD and health.

FIGURE 2

Schematic of the whole‐brain computational model. (a) the whole‐brain model with the local dynamic mean‐field model, in which then underlying anatomical connectivity was determined using diffusion MRI and tractography techniques. (b) Simulated time series and its corresponding simulated probabilistic metastable substate (PMS) space. (c) Empirical time series from the empirical functional magnetic resonance imaging (fMRI) and its corresponding empirical PMS space. Notably, the whole‐brain model was fitted to the empirical neuroimaging data described by the PMS space.

FIGURE 3

The optimal spatiotemporal fit of the whole‐brain model to the probabilistic metastable substate (PMS) space. (a) Fitting of the whole‐brain model to all subjects by optimizing the G parameter. (b) Fitting the whole‐brain model to two radically different brain states for major depressive disorder (MDD) and health

FIGURE 4

Forcing the transition of brain states between major depressive disorder (MDD) and heath. (a) The whole‐brain model with external excitatory or inhibitory stimulation. (b) Schematic of the strategy for forcing the transition between source and target brain states. (c) Forcing the transition from MDD to health by excitatory external stimulation (intensity = 0.02). (d) Forcing the transition from health to MDD by external inhibitory stimulation (intensity = 0.02)

FIGURE 5

Impact of different external stimulation intensities on the transition between major depressive disorder (MDD) and health and vice versa. (a) Forcing transitions from MDD to health by excitatory stimulation with different intensities. (b) Forcing transitions from health to MDD by inhibitory stimulation with different intensities