Amit Etkin, MD, PhD, on EEG as a Tool in Depression Treatment
A secondary analysis recently published in JAMA Psychiatry explored how electroencephalographic (EEG) connectivity features may moderate response to antidepressant treatment. The study analyzed data from the Establishing Moderators and Biosignatures of Antidepressant Response in Clinic Care (EMBARC) clinical trial.
In this podcast, researcher Amit Etkin, MD, PhD, of Stanford University and Alto Neuroscience, explains the reasons for the research, what the study found, and possible clinical applications of the findings.
Amit Etkin, MD, PhD is the founder, CEO and chairman of Alto Neuroscience, as well as a professor in the Department of Psychiatry and Behavioral Sciences at Stanford University and a member of the Wu Tsai Neuroscience Institute at Stanford. He has received multiple awards, most notably the NIH Director’s Pioneer Award in 2017, for groundbreaking work in clinical psychiatry and neuroscience. Dr. Etkin is trained as both as a neuroscientist and psychiatrist, having received his MD/PhD under the mentorship of Nobel Laureate Eric Kandel. The overarching aim of the Dr. Etkin’s work is to understand the neural basis of emotional disorders and their treatment, and to leverage this knowledge to better understand how the brain works and to develop novel treatment interventions. In support of this goal, Dr. Etkin also collaborates with neuroscientists, engineers, psychologists, physicians and others to establish a new intellectual, scientific and clinical paradigm for understanding and manipulating human brain circuits in healthy individuals and for treating psychiatric diseases including depression, anxiety and post-traumatic stress disorder.
My name is Amit Etkin. I'm a psychiatrist and neuroscientist. I'm a professor at Stanford [University] in the psychiatry department and in the Neurosciences Institute, as well as now CEO and founder of a startup called Alto Neuroscience.
For psychiatry and for neuroscience, the question is why go after trying to understand depression and who responds to an antidepressant in biological terms, and more specifically, why go after EEG as a way to get there.
Across all of our history in psychiatry, we've used every one of our interventions, whether it's pharmacological, psychotherapeutic, or brain stimulation, more or less blind information about the person. That has led us over the years to come to the conclusion that interventions seem to work less well than we had thought or hoped, and certainly leads to a lot of frustration on the part of patients and clinicians.
At the same time, our diagnoses have been very imprecise, based on very broad, often nonspecific clinical criteria.
What we wanted to do in this study was ask a much more basic question, which is, is the issue the imprecision of our diagnoses, and that we give a drug, in this case sertraline, to people in an unselected manner, some which have a brain signature that's associated with response versus a brain signature that's associated with nonresponse to the drug, or is it that the diagnosis is fine, and the drugs are simply not particularly more effective than placebos, which has been the feeling in the field.
The best way to get a viewpoint into that is to use a tool that can tell you something about the brain, how the brain works. In this case, EEG is a direct measure of brain function.
What we quantified was how connected different parts of the brain are and how they communicate with each other based on EEG, and then give people an antidepressant or give them a placebo, randomize them to one of the two, and ask the question of whether knowing something about their brain helps explain whether it's in fact that the drug just simply doesn't work all that well and that you can't predict how it works for some people different from others based on the EEG.
Or, is the diagnosis imprecise and can be better understood in terms of some additional test of brain function, in this case EEG, such that we can find those people for whom an antidepressant really works much better than placebo, and differentiate them from the people for whom an antidepressant doesn't work any differently from placebo.
Separate of that, we wanted to understand just the placebo response itself. Are there things in the brain that can help us understand who responds to placebo just on its own?
This study is one analysis in a study called EMBARC. EMBARC is a study that began now nearly 10 years ago, funded by NIH as a multi‑site clinical trial.
Its specific goal though is not to be a clinical trial to determine whether the drug works differently from placebo—this is already an FDA‑approved drug that has been studied extensively—but rather a study to add on imaging, behavioral, and other measures in the context of a clinical trial in order to answer the questions that I mentioned earlier.
In fact, this is the largest such trial, the largest trial with imaging in a randomized placebo-controlled context for depression.
What we did as part of that is take EEG data at baseline prior to the randomization and prescription of treatment for each patient and evaluated whether information in that EEG could then predict clinical outcome. That's the subject of this particular analysis to answer the questions I had posed earlier.
The main findings in this case were as follows. First, it does seem while the drug is only slightly better than placebo on average, that there are really important differences between people based on their EEG conductivity that we can assess and determine that for some people, their brain signature predicts a much better response to drug versus placebo or in some cases even placebo relative to drug, whereas in other people, their brain signatures suggest an equal response to the antidepressant and placebo.
It does seem like there's two major conclusions here. One is that people with depression, when diagnosed clinically, don't have exactly the same picture in terms of their brains, and it matters a lot with respect to understanding their treatment outcome.
The second, actually similarly exciting, is that EEG as a tool seems to carry a lot of that important signal. The implication there is that EEG, which is something that can be done relatively cheaply and implemented in clinics at scale, may be the kind of thing we want to scale on in order to be able to drive new diagnostics and, down the line, new therapeutics into psychiatry in a personalized, targeted, rational, neuroscience‑driven way.
One of the things we'd done very deliberately in how we did this analysis is we looked across the entire brain, all the connections across the cortex using this particular EEG measure. The reason for doing that is that this way, we decrease the chance for bias in terms of where we're looking in the brain.
I think most people at this point understand that when you constrain your viewpoint, you actually bias yourself to get the kind of answers you were hoping to get, but they may not be the most representative answers.
In fact, that's what we saw here, is some of the strongest signal in the brain in terms of which regions had connectivity that best differentiated between drug and placebo were actually in the back of the brain in an area called the parietal cortex, which is not a region we often think about when we talk about depression. Most of that we talk about with respect to the front of the brain, the frontal cortex.
However, in most cases also, people never even look at the parietal cortex, because they're so focused on the frontal cortex. That's again an opportunity for bias to creep in, just in terms of what we're looking at.
That was a surprising and interesting result. It came out of a, if you will, unbiased, data‑driven analysis and should lead people to start thinking some more about what does that really mean to see that there, how do we alter our hypotheses or ideas of what's going on with depression with respect to who responds to an antidepressant now knowing this. That's really what we'd hoped to achieve from the study.
The question of near‑term application for clinical practice is a very interesting one. We have a number of other papers currently either in press or in process in various ways that address that systematically and do have very specific clinical implications.
This is one paper out of that set. I think we'll see in the next 6 to 12 months, as these papers come to print, that EEG is really something that is maybe not today ready for prime time, but very soon will be, has all of the right attributes for a test that could be used clinically. Like I mentioned, it's cheap, it's scalable, it's something that already happens in neurology clinics.
It does seem to be sensitive to the individual differences between patients that map onto treatment outcome. We could call that a diagnosis. We can say that EEG is telling us something about that person that could be thought of as a diagnostic. I prefer to think of it as that's an objective test that can help the clinician make decisions.
I'm not trying in these studies to redefine depression purely based on these biological criteria. Rather, I think people have a sense of what depression is from just clinical experience, as well as patients from their own experience and, in fact, the lay public.
What we're trying to do is add the objective test to guide what a physician does in terms of what treatments they try for the patient and how we talk about depression with respect to the biology of the brain.
Rolle CE, Fonzo GA, Wu W, et al. Cortical connectivity moderators of antidepressant vs placebo treatment response in major depressive disorder: secondary analysis of a randomized clinical trial. JAMA Psychiatry. 2020 January 2;[Epub ahead of print].