Antidepressants are among the top five most commonly prescribed drugs in both the U.S. and Canada. And yet, it can be difficult to predict the effectiveness of these drugs when first prescribed, leading to a lot of trial and error.
Electroencephalography (EEG), it turns out, may finally offer some clues to help fast track treatment.
More than 265 million people globally suffer from depression, and it is the world’s leading cause of disability, according to the World Health Organization. Fortunately, there are many effective treatments including several classes of medication.
It can, however, take a long time – trying one drug and then the next – to find the best treatment for any given individual. Sometimes this means months or years of unnecessary suffering and death.
For this reason, researchers have long tried to find biomarkers that can help predict which treatments might work best. A biomarker is a measurable substance (like an antibody) or activity (like a certain brain wave emission) in the body that can be associated with a disease. EEG is often used to find biomarkers associated with epilepsy, for example.
For the last 40 years, researchers have also been using EEG to look for biomarkers linked to depression. According to an invited commentary in JAMA: Psychiatry, published in June 2020, most of the evidence for finding these kinds of biomarkers has so far been weak. A new paper, however, offers a breakthrough.
The paper, also published in JAMA, shows that using machine learning on large EEG data sets may provide answers. The study highlights the importance of focusing on specific symptoms associated with the condition rather than on one single depression score. Researchers measured symptoms such as loss of insight, appetite changes, and energy loss.
They looked at these and other symptoms in 518 patients diagnosed with major depression and given antidepressants. The study participants were evenly divided into three groups depending on which antidepressant they took: escitalopram oxalate, sertraline hydrochloride, or extended-release venlafaxine hydrochloride. No placebo was included. The patients were assessed using EEG at baseline and after 8 weeks on their respective medication.
According to the researchers, data on the power of the EEG signals in each frequency range at each electrode site were extracted using the Welch method for spectral density estimation. The data was then run through a machine-learning algorithm to compare baseline data with the data gathered after eight weeks of treatment.
They found several baseline EEG biomarkers associated with improvements in symptom scores. The most intriguing of these results? Improvement in insight was associated with baseline (pre-treatment) EEG occipital δ band power, and improvements in energy and psychomotor retardation (another depressive symptom) were associated with baseline EEG power in the δ and α bands.
The authors acknowledge this is just the beginning of our understanding into how biomarkers like these and EEG studies may be used clinically to help expedite treatment for patients with depression. However, they add, “our models provide a first proof of principle that noninvasive neurobiological markers and pre-treatment symptom assessments may be used to determine whether specific symptom domains are likely to persist with standard antidepressant treatment.”