EEG Vigilance Regulation Patterns and Their Discriminative Power to Separate Patients with Major Depression from Healthy Controls

Background/Aim: Recently, a framework has been presented that links vigilance regulation, i.e. tonic brain arousal, with clinical symptoms of affective disorders. Against this background, the aim of this study was to deepen the knowledge of vigilance regulation by (1) identifying different patterns of vigilance regulation at rest in healthy subjects (n = 141) and (2) comparing the frequency distribution of these patterns between unmedicated patients with major depression (MD; n = 30) and healthy controls (HCs; n = 30). Method: Each 1-second segment of 15-min resting EEGs from 141 healthy subjects was classified as 1 of 7 different vigilance stages using the Vigilance Algorithm Leipzig. K-means clustering was used to distinguish different patterns of EEG vigilance regulation. The frequency distribution of these patterns was analyzed in independent data of 30 unmedicated MD patients and 30 matched HCs using a χ2 test. Results: The 3-cluster solution with a stable, a slowly declining and an unstable vigilance regulation pattern yielded the highest mathematical quality and performed best for separation of MD patients and HCs (χ2 = 13.34; p < 0.001). Patterns with stable vigilance regulation were found significantly more often in patients with MD than in HCs. Conclusion: A stable vigilance regulation pattern, derived from a large sample of HCs, characterizes most patients with MD and separates them from matched HCs with a sensitivity between 67 and 73% and a specificity between 67 and 80%. The pattern of vigilance regulation might be a useful biomarker for delineating MD subgroups, e.g. for treatment prediction.