Ask the Expert: Predicting Antidepressant Response

This Month’s Expert Predicting Antidepressant ResponseTCPR: Dr. Leuchter, is there a science to choosing an initial antidepressant for a depressed patient? Is it still predominantly personal preference or is there a better way to make that choice?

Dr. Leuchter: At this point, it is still very much clinician and patient preference. I don’t think that there is any effective, proven way to predict medication response. I have been involved in research on biomarkers—we call them “response endophenotypes”—such as EEG, where we look at the physiologic response of the brain to a new drug and try to determine whether the drug will work for that patient. There have been several replication studies of our EEG work, but we would not yet advocate introducing this test into clinical practice.

TCPR: Can you describe very briefly some of the physiological changes that you might see on EEG?

Dr. Leuchter: These are functional changes in the prefrontal brain regions. We discovered that there are signals that emerge from the prefrontal regions that indicate whether patients are responding or are likely to respond to the medications they are receiving.

TCPR: How exactly does this work?

Dr. Leuchter: My research team and I record a baseline EEG and then record a second EEG after the patient is on a drug for one week. We look to see shifts in energy in certain frequency bands, specifically alpha and theta, that appear to be characteristic of early response to the drug.

TCPR: So you don’t get specific readouts or specific patterns with individual drugs? Instead you see either a yes or a no response to whatever drug you are applying at that time?

This article originally appeared in The Carlat Psychiatry Report -- an unbiased monthly covering all things psychiatry.
Want more, plus easy CME credit?
Subscribe today!

Dr. Leuchter: We actually see both. When we put somebody on an initial medication—usually an SSRI—we see a signal of whether they are going to respond to that drug or not. Interestingly, in our studies we’ve found that if patients have a negative indicator in response to the SSRI, they are likely to have a positive indicator with regard to bupropion (Wellbutrin). Now, obviously, not everybody who fails to respond to an SSRI goes on to respond to bupropion, so it is not a one-to-one correspondence. But we do know that having a negative indicator on one drug seems to indicate a greater likelihood of having a positive indicator on a drug with a totally different mechanism of action. We have published these results (Leuchter AF et al, Psych Research 2009;169:132–138) but have not yet published independent replication of those findings. [Editor’s note: for more on this, see also TCPR, November 2009.]

TCPR: That is something that we generally observe in practice, too. When patients fail one or multiple SSRI trials, they may have success with a trial of a different class of medication. So it sort of corroborates clinical experience.

Dr. Leuchter: Yes. What’s interesting is that it corroborates clinical experience but not clinical trials. Reading the literature, we know that if we switch to a drug with a totally different mechanism of action that it does not statistically increase the chances of response. However, we have all had the clinical experience where many people who are SSRI-nonresponders turn out to be responders to an SNRI or a drug with an entirely different mechanism. It may be that without this type of biomarker, we cannot reliably identify those subjects who will benefit from a medication with a different mechanism of action.

TCPR: What is a “response endophenotype”?

Dr. Leuchter: It’s a term we use for a behavioral or neurobiological feature that we can measure in a patient early in the course of treatment that may predict a patient’s ultimate outcome.

TCPR: Another predictor of response would be pharmacogenetic testing. Although clinicians do not seem to use that regularly, where does it stand right now?

Dr. Leuchter: I think the reason that people don’t use it regularly is because it hasn’t been shown to work (Garriock HA et al, Biol Psychiatry 2010;67(2):133–138). There are multiple genes that confer risk for depression and that are involved in mediating drug response. So since we don’t really have a good model for how people get depressed, it is very hard to have a good model for what genes we ought to look at when we attempt to predict response.

TCPR: Patients are often concerned about side effects. Is there any way to use either your endophenotype approach or another biomarker measure to predict whether a patient will have side effects from a given medication?

Dr. Leuchter: There are some promising leads that may have clinical applicability, particularly when it comes to drug metabolism, such as cytochrome P450 2D6 and 3A4 activity. Researchers have established that the rate at which a patient metabolizes some of these compounds is probably related to the likelihood of side effects or drug interactions (D’Empaire I et al, J Psychiatr Pract 2011;17(5):330–339). So I think some of this testing is ready for prime time. Of course there is the question of cost and whether you get more from the testing than you would from close clinical monitoring. The jury is still out on that.

TCPR: Another area of your research is placebo response. Could you summarize your work?

Dr. Leuchter: A decade ago, our group was the first to demonstrate that there were physiologic changes in the brains of placebo responders that were distinct from those of patients who didn’t respond to placebo or who responded to active drug (Leuchter AF et al, Am J Psychiatry 2002;159(1):122–129). So “placebo response” is not just a psychological phenomenon, but a physiologic phenomenon. We also have promising findings to suggest that brain function even before treatment might indicate the likelihood that one will respond to placebo, and that there are certain genetic polymorphisms, specifically in MAO-A and COMT genes, that make one more likely to be a placebo responder—at least in the context of depression (Leuchter AF et al, J Clin Psychopharmacol 2009;29(4):372–377). We continue to pursue this work and have been looking more recently at the role of expectations and different personality variables as mediators of placebo response. We hope to report these results within the next few months.

TCPR: What types of physiological changes have you observed in placebo responders?

Dr. Leuchter: Like active-drug responders, they showed changes in neurophysiologic activity in the prefrontal region. However, these changes took longer to come on. In drug responders, we saw changes in many cases within 48 hours and certainly saw very significant changes by one week. The placebo responders didn’t tend to show any change until about two weeks, and even then, the changes that we saw frequently were in the opposite direction of those that we saw in drug responders. Based on the particular symptoms that they showed, there was no way that you could distinguish the placebo responder from the drug responder. But physiologically they looked different. We saw decreases in prefrontal cordance [a measure of regional brain activity] in the drug responders and increases in prefrontal cordance in placebo responders.

TCPR: Does that change normalize over time after the placebo effect wears off?

Dr. Leuchter: Well, actually, it normalizes in both groups; they tend to come back to their pretreatment baseline over time. That makes sense with the drug effect because if you block reuptake, for instance, the brain does eventually re-equilibrate. In placebo responders, why they get back to their baseline is an interesting question.

TCPR: Based on your research, are there patient-centered factors that predict a better response to a treatment, whether it is a placebo or an antidepressant?

Dr. Leuchter: There certainly appear to be some. Some of these are seemingly common sense factors, such as: the greater the expectation people have for a positive outcome of treatment, the more likely they are to respond to treatment. The interesting thing is that expectations seem to play a different role in medication responders than they do in placebo responders.

TCPR: Would it help outcomes if we simply assess a person’s expectations at baseline?

Dr. Leuchter: I think it would. There was an interesting study last year on placebo response where the conclusion was that a placebo works even if you know you’re getting it. But when you look at the study, people were told it was a study of placebo response, and they would either get a placebo or be put into a control condition. Those who did not receive placebo were disappointed, and therefore may have responded more poorly in the study (Kaptchuk TJ et al, PLoS One 2010;5(12):e15591).

TCPR: It’s hard to do a placebo-controlled placebo trial.

Dr. Leuchter: Exactly.

TCPR: What are your suggestions for enhancing patients’ expectations in an appropriate and ethical way?

Dr. Leuchter: We live in an age where people have to be fully informed of what they are getting and the consequences of those choices. But we should harness expectations whenever we can. When I prescribe a new medication for a patient, I always tell the patient, “You know, I think that this medicine that we are going to prescribe for you now is really going to help you.” If I didn’t think it was going to help, I wouldn’t be prescribing it. So there is actually nothing misleading about it. To some extent raising expectations can become a self-fulfilling prophecy. In clinical trials, the placebo response may be an annoyance. In clinical practice, placebo response is our “friend.”

TCPR: Do patients benefit from an explanation of a drug’s mechanism?

Dr. Leuchter: Yes, and there are two main reasons. One is that some patients won’t go for the treatment unless you “medicalize” it. So if you make it sound like a medical treatment and explain the mechanism through which drugs work, I think that increases acceptability and maybe adherence. Secondly, it’s a way to increase confidence and positive expectation because it is not just that you are saying, “I think this is going to help you.” You are also saying, “I think I know why this is going to help you.”

TCPR: When it comes to treatment-resistant depression, is the problem that we just don’t have the right tool to treat it, or is it largely a patient’s negative expectation: “Well, the last medication didn’t do anything for me so the chances are that nothing else will”?

Dr. Leuchter: Treatment resistance is a very complex phenomenon, and to some extent when individuals are treatment-resistant it is because we don’t have the right molecules—the right treatment—to correct whatever biological abnormality exists. At the same time, there probably is an element of pharmaco-conditioning going on. We have an article coming out soon that shows that when you expose individuals to one unsuccessful treatment after another you probably are conditioning them in some ways not to respond to the next treatment. When they have the experience again and again of taking a pill and not getting better, at some point negative conditioning may start to play a role. Jerome Frank, when he wrote Persuasion and Healing: A Comparative Study of Psychotherapy, said that one of the goals of therapy was the restoration of hope, and I think there is a lot of truth to that. We are in the business not just of prescribing treatment, but also encouraging patients to believe that the treatment is likely to benefit them. We are not really doing the patient full service unless we address their hopefulness or their hopelessness.

TCPR: Thank you, Dr. Leuchter.

Ask the Expert: Predicting Antidepressant Response

This article originally appeared in:

The Carlat Psychiatry Report
Click on the image to learn more or subscribe today!

This article was published in print 5/2012 in Volume:Issue 10:5.


APA Reference
Leuchter,, A. (2014). Ask the Expert: Predicting Antidepressant Response. Psych Central. Retrieved on May 28, 2020, from


Scientifically Reviewed
Last updated: 22 May 2014
Last reviewed: By John M. Grohol, Psy.D. on 22 May 2014
Published on All rights reserved.