Abstract

EEG alpha rhythm has been used as a neurophysiological measure of activation, ranging from relaxation at rest to task-related local cortical synchronization or desynchronization. Although the topographic capacity of high density EEG is limited by volume conduction, CSD transformation provides reference-free measures that represent the neuronal generator pattern underlying alpha. CSD-based frequency principal components analysis (CSD-fPCA) of resting EEG has identified posterior, condition-dependent (eyes open/closed) alpha as a predictor of antidepressant treatment response (Tenke et al 2011). However, alpha also varies during the performance of behavioral tasks, and these changes may be quantified as event-related spectral perturbations (ERSP). The present study combined resting- and task-related (novelty oddball) CSD alpha measures obtained from a 67-channel montage in depressed patients prior to treatment with a serotonergic antidepressant (n = 77) or bupropion (n = 23). Oddball EEG was epoched (-200 to 800 ms) for each condition (target, nontarget, novel), CSD-transformed, and submitted to FFT-based TF analysis to yield ERSP and corresponding baseline spectra. The TF dataset was simplified and quantified by unrestricted PCA (CSD-tfPCA), yielding a prominent, unambiguous factor quantifying alpha desynchronization (ERD), characterized by loadings peaking at 490 ms (350-700 ms) and 10 Hz (7-12.5 Hz) and scores with a posterior topography for novels and targets, but not nontargets. Resting and task-related baseline EEG alpha were quantified by CSD-fPCA, and submitted to repeated measures ANOVAs with treatment (serotonergic, bupropion), response (responder, nonresponder), and gender as between-factors. The Beck Depression Inventory (BDI) measure of pretreatment depression was used as a covariate because BDI was correlated with condition-dependent alpha at rest (r = .253, p = .01) and post-stimulus ERD (r = .212, p = .03). For alpha at rest, we replicated the finding of greater condition-dependent alpha in responders than nonresponders for both treatments. In contrast, task-related alpha (prestimulus baseline or ERD) showed no association with treatment response. However, the overall amplitude of task-related alpha showed a significant treatment by response interaction, where bupropion responders showed less task-related alpha than nonresponders, but this difference was not seen for serotonergic treatment. In conclusion, alpha oscillations at rest and during a novelty oddball task are related to depression and antidepressant treatment response, suggesting distinct processes that modulate default mode activity.

Supported by NIMH grants MH36295 and MH094356.