Skip to Content

EEG vigilance and response to oral prolonged-release ketamine in treatment resistant depression – A double-blind randomized validation study

Psychiatry Research: Neuroimaging

Abstract


Treatment-resistant depression (TRD) is associated with reduced quality of life and elevated mortality, posing a major challenge to psychiatric care. After non-response to conventional treatments, next-level interventions such as (es)ketamine are recommended, though remission rates remain variable. Identifying reliable markers of treatment response is therefore critical. Recent evidence suggests that a higher percentage of electroencephalography (EEG) vigilance stage A1 is associated with response to intravenous ketamine in major depression. We aimed to corroborate this finding in TRD patients from a recent phase-2 randomized controlled trial of oral prolonged-release ketamine. An algorithm classified vigilance stages in 21 10-minute resting-state EEG recordings. While no significant response × treatment interaction emerged for percentage of vigilance stage A1, a small-scale meta-analysis showed a significant pooled mean difference between ketamine responders and non- responders. Applying a previously proposed A1 cutoff (43 %) yielded chance-level prediction accuracy in the combined ketamine group, but 75 % accuracy in the 240 mg subgroup. Moreover, responders to 240 mg ketamine exhibited a significantly more stable vigilance over time compared to non-responders. Although further validation in a larger sample is warranted, these findings support the clinical value of EEG vigilance as a predictive biomarker for treatment outcomes in depression.

Psychiatry Research: Neuroimaging Vol. 350 2025


Authors

Monn, A., Eicher, C., Rüesch, A., Kronenberg, G., Offenhammer, B., Adank, A., de Bardeci, M., Ip, C. H., Scherer, H., Schaekel, L., Colla, M., Brühl, A. B., Seifritz, E., & Olbrich, S.

  https://doi.org/10.1016/j.pscychresns.2025.112001

Share this post
Assessing the robustness of deep learning based brain age prediction models across multiple EEG datasets
bioRxiv