One size does not fit all: Single-subject analyses reveal substantial individual variation in electroencephalography (EEG) characteristics of antidepressant treatment response

Electroencephalography (EEG) characteristics associated with treatment response show potential for informing treatment choices for major depressive disorder, but to date, no robust markers have been identified. Variable findings might be due to the use of group analyses on a relatively heterogeneous population, which neglect individual variation. However, the correspondence between group level findings and individual brain characteristics has not been extensively investigated. Using single-subject analyses, we explored the extent to which group-based EEG connectivity and complexity characteristics associated with treatment response could be identified in individual patients. Resting-state EEG data and Montgomery-[A]sberg Depression Rating Scale symptom scores were collected from 43 patients with depression (23 females) before, at 1 and 12 weeks of treatment with escitalopram, bupropion or both. The multivariate statistical technique partial least squares was used to: 1) identify differences in EEG connectivity (weighted phase lag index) and complexity (multiscale entropy) between responders and non-responders to treatment ([&ge;]50% and <50% reduction in symptoms, respectively, by week 12), and 2) determine whether group patterns could be identified in individual patients. The group analyses distinguished groups. Responders showed decreased alpha and increased beta connectivity and early, widespread decreases in coarse scale entropy over treatment. Non-responders showed an opposite connectivity pattern, and later, spatially confined decreases in coarse scale entropy. These EEG characteristics were identified in [~]40-60% of individual patients. Substantial individual variation highlighted by the single-subject analyses might explain why robust EEG markers of antidepressant treatment response have not been identified. As up to 60% of patients in our sample was not well represented by the group results, individual variation needs to be considered when investigating clinically useful characteristics of antidepressant treatment response. Author summaryMajor depression affects over 300 million people worldwide, placing great personal and financial burden on individuals and society. Although multiple forms of treatment exist, we are not able to predict which treatment will work for which patients, so finding the right treatment can take months to years. Neuroimaging biomarker research aims to find characteristics of brain function that can predict treatment outcomes, allowing us to identify the most effective treatment for each patient faster. While promising findings have been reported, most studies look at group-average differences at intake between patients who do and do not recover with treatment. We do not yet know if such group-level characteristics can be identified in individual patients, however, and therefore if they can indeed be used to personalize treatment. In our study, we conducted individual patient analyses, and compared the individual patterns identified to group-average brain characteristics. We found that only [~]40-60% of individual patients showed the same brain characteristics as their group-average. These results indicate that commonly conducted group-average studies miss potentially important individual variation in the brain characteristics associated with antidepressant treatment outcome. This variation should be considered in future research so that individualized prediction of treatment outcomes can become a reality. Trial registrationclinicaltrials.gov; https://clinicaltrials.gov; NCT00519428