Abstract
The primary aim of this dissertation was to investigate the value of neurophysiological biomarkers measured by the EEG, in the prognosis of treatment outcome in depression. Wishing to improve treatment outcome, we described our first steps towards the implementation of an EEG biomarker informed protocol.
A meta-analysis of sixteen studies on the diagnostic value of the biomarker frontal alpha asymmetry (FAA) in showed a non-significant, negligible ES, demonstrating limited diagnostic value of FAA in MDD. We demonstrate that FAA is a stable trait, robust to time, state and pharmacological status. Response rates in patients with a normalized EEG taking sertraline were 5.2 times (significantly) larger than in subjects treated with escitalopram/venlafaxine. For the improvement of assessment of abnormal EEGs in the depressed population, the computed features CNN probability, the dominant frequency, and the tBSI all showed good performance in identifying the specific and notably “light” abnormalities. A random forest model containing the combined features did not reliably predict treatment outcome. We developed a protocol in which all knowledge on biomarker informed AD prescription was implemented. In this feasibility trial, seventy patients stratified to AD based on their EEG biomarkers, had better symptom improvement than 52 control patients.
The complexity of a heterogeneous depressed population makes it impossible to find one or few treatments that fit all. However, this heterogeneity can be embraced by employing biomarkers that are capable of identifying homogenic subgroups, and potentially predicting treatment outcome. Established quantification methods already seem to allow us to predict response to treatment. Combined with qualitative assessment of subclinical EEG abnormalities in our newly developed protocol, the first results for EEG-informed prescription of antidepressants show sufficient feasibility in a clinical setting. To our knowledge, this is the first attempt to elevate the treatment of depression through these biomarkers, which not only shows the stratification protocol is non-inferior: patients actually show a modest increase in symptom improvement. The proposed protocol therefore not only makes our methods easily translatable to clinical practice, it bears the promise of a small but much needed achievement of higher treatment standards, in a new form of neuropsychiatric health care for depression.
A meta-analysis of sixteen studies on the diagnostic value of the biomarker frontal alpha asymmetry (FAA) in showed a non-significant, negligible ES, demonstrating limited diagnostic value of FAA in MDD. We demonstrate that FAA is a stable trait, robust to time, state and pharmacological status. Response rates in patients with a normalized EEG taking sertraline were 5.2 times (significantly) larger than in subjects treated with escitalopram/venlafaxine. For the improvement of assessment of abnormal EEGs in the depressed population, the computed features CNN probability, the dominant frequency, and the tBSI all showed good performance in identifying the specific and notably “light” abnormalities. A random forest model containing the combined features did not reliably predict treatment outcome. We developed a protocol in which all knowledge on biomarker informed AD prescription was implemented. In this feasibility trial, seventy patients stratified to AD based on their EEG biomarkers, had better symptom improvement than 52 control patients.
The complexity of a heterogeneous depressed population makes it impossible to find one or few treatments that fit all. However, this heterogeneity can be embraced by employing biomarkers that are capable of identifying homogenic subgroups, and potentially predicting treatment outcome. Established quantification methods already seem to allow us to predict response to treatment. Combined with qualitative assessment of subclinical EEG abnormalities in our newly developed protocol, the first results for EEG-informed prescription of antidepressants show sufficient feasibility in a clinical setting. To our knowledge, this is the first attempt to elevate the treatment of depression through these biomarkers, which not only shows the stratification protocol is non-inferior: patients actually show a modest increase in symptom improvement. The proposed protocol therefore not only makes our methods easily translatable to clinical practice, it bears the promise of a small but much needed achievement of higher treatment standards, in a new form of neuropsychiatric health care for depression.
| Original language | English |
|---|---|
| Qualification | Doctor of Philosophy |
| Awarding Institution |
|
| Supervisors/Advisors |
|
| Award date | 1 Jul 2020 |
| Place of Publication | Enschede |
| Publisher | |
| Print ISBNs | 978-90-830-0133-3 |
| DOIs | |
| Publication status | Published - 1 Jul 2020 |