Electrophysiological neural network reorganisation in patients with depression undergoing electroconvulsive therapy is linked to treatment response

While electroconvulsive therapy (ECT) is highly efficacious in severe mental disorders, such as schizophrenia, catatonia, mania, and depression, its underlying mechanisms of action remain unclear. One possible candidate discussed in recent literature is a potential reorganization of functional brain circuits following ECT. Functional brain network analyses have often been limited by comparability and thresholding problems. The minimum spanning tree (MST), as a state-of-the-art method, offers a solution to overcome these difficulties. Using repeated resting-state electroencephalography (EEG) measurements in depressed patients undergoing ECT, we investigated changes in the functional brain network using MST parameters, as well as the association of these changes with the therapeutic effect of ECT.

Patients (N = 42, mean age = 42.5 years, 19 females) who underwent bifrontal ECT due to a depressive episode were included. High-density resting-state EEG (64 electrodes) was measured before and after the acute ECT series. Sensor-level minimum spanning trees in the alpha frequency band were computed based on the corrected amplitude envelope correlation (AECc) and the phase lag index (PLI). For both functional connectivity measures, the MST network was characterized using the tree hierarchy, the diameter and the leaf fraction as network metrics. The tree hierarchy quantifies an optimal balance between large-scale integration and hub overload, the diameter represents a measure of network efficiency by referring to the largest distance in the network between any two nodes, and the leaf fraction is a measure of network centrality. Paired t-tests were computed to compare the parameters from pre- to post-ECT, and an FDR p-value correction was applied in the main and the subgroup analyses. Treatment response was classified by symptom reduction of at least 47% in the Beck’s Depression Inventory (BDI), the Hamilton Depression Rating Scale (HAMD-21), or the Montgomery-Åsberg Depression Rating Scale (MADRS).

Pre- to post-ECT, the PLI-based tree hierarchy increased (p=0.004, p_fdr=0.040, d=0.39), but not for AECc. There was no association of the increase in tree hierarchy with the pre-to post-ECT changes in the MADRS score. However, subgroup analyses showed an increase in tree hierarchy for PLI and AECc in the responder group (p<0.05, p_fdr< 0.05, d=0.45-0.47), but not in the non-responder group. There were no changes in leaf fraction or diameter pre- to post-ECT for both AECc and PLI in the full sample. However, the non-responders showed an increase in diameter in the PLI-based MST, which did not survive p-value correction, though (p=0.044, p_fdr=0.22, d=0.64).

In general, the increase in PLI-based tree hierarchy suggests an improved balance between electrophysiological neural network integration and hub overload following an acute ECT series, while network efficiency and centrality remained unchanged. The observation that the tree hierarchy increased for both functional connectivity measures in responders only is indicative of a response-specific network optimization following ECT. These findings highlight the need to further investigate electrophysiological functional brain network reorganization as a potential mechanism of action of ECT. Future studies should also investigate the potential association between specific functional brain network changes and the cognitive side effects of ECT.

References

No references