TCR: Dr. Ghaemi, I thought we could discuss this rather controversial issue of whether antidepressants (ADs) should be prescribed in bipolar disorder, particularly in light of the recent article published in the American Journal of Psychiatry by Altshuler et. al. (1). The 2002 American Psychiatric Association (APA) guidelines for bipolar disorder discourage the use of ADs in bipolar disorder, and yet this recent article seems to cast doubt on this wisdom.
Dr. Ghaemi: It certainly is a controversial issue. The context is that until 5 or 10 years ago, ADs were used very extensively in the treatment of bipolar disorder, even by academics and researchers. Community practitioners still use them quite a bit: In our studies at the Cambridge Hospital we’ve found community usage rates on the order of 80% for ADs in bipolar disorder, compared to my own clinic, where I use ADs in only about 20% of my bipolar patients.
TCR: What do the new APA guidelines say with regard to AD use?
Dr. Ghaemi: They came out in 2002, and the previous ones were published in 1994. The older guidelines said that the preferred treatment for bipolar depression was an AD plus a mood stabilizer, which is the way most clinicians still practice. But the 2002 guidelines recommend that the first-line treatment should be lithium or Lamictal alone for mild to moderate depression, and only in severe depression should an AD potentially be a first-line agent. I’m even more conservative than that, in that I don’t even use ADs first-line for severe depression; instead, I optimize mood stabilizers.
TCR: How does the Althshuler study fit into this issue?
Dr. Ghaemi: This studies is a great case example of the utility of a good knowledge of statistics for studies in psychiatry. Her study design was observational, not randomized: she observed the natural course of patients who got ADs and mood stabilizers for bipolar depression and then compared those who came off their ADs vs. those who stayed on. Her results were that those who came off were more depressed in a one year follow-up. Based on these findings, she concludes that therefore patients who respond to ADs for bipolar depression should stay on them.
TCR: Ok, so what’s wrong with that conclusion?
Dr. Ghaemi: Exactly. It sounds reasonable enough if that’s all you know about the study. But the first question one should always ask about any clinical study is “Is there any confounding?” In other words, are there any other factors that might explain the results. Now, Altshuler’s study was an observational or naturalistic study, so the patients were not randomized. There’s nothing wrong with that, I publish studies like that as well, but when you do observational studies, you have to make some effort to control for confounding, since you know that’s going to be a risk. Altshuler et al did not take either of the usual measures,which are to stratify the two groups or to statistically analyze the results in a regression model that controls for confounders.
TCR: So how does that affect the validity of the study?
Dr. Ghaemi: There are many potential confounders, and the obvious one is the issue of rapid cycling, because we know that if there are more rapid cyclers in one group than another group, the rapid cycler group will relapse more quickly. In this study, clinicians were allowed to make their own decisions about which patients to withdraw from ADs. The study doesn’t tell us how they made these decision. It is possible, for example, that clinicians are more likely to stop ADs in rapid cycling patients.
TCR: And why is that?
Dr. Ghaemi: Because many clinicians are aware of the data indicating that many bipolar patients (about 20%) will be converted to rapid cyclers if maintained on ADs. And if, in the Altshuler study, rapid cyclers were overrepresented in the group that stopped ADs, the finding of a higher relapse rate would be completely explained by the characteristics of the patients, rather than by any differences in treatment.
TCR: So basically, it’s very hard to interpret the study, because the groups were not randomized, leading to the possibility of confounding.
Dr. Ghaemi: Right, and I really don’t think that you can make a judgement like they did at the end of their study that these data should impact treatment guidelines, and that maybe bipolar patients should be kept on ADs. In fact this could be a dangerous recommendation, because if clinicians routinely start leaving their bipolar patients on ADs long-term, about 20% of these patients could be converted into rapid cycling patients, a finding that is not at all refuted by Altshuler’s study.
TCR: Related to this is the very confusing issue of whether some ADs are better than others in bipolar disorder.
Dr. Ghaemi: We don’t have a lot of evidence, but the evidence that we do have supports the idea that Paxil and Wellbutrin may be better choices. By “better” what I mean is that they possess less of a risk of causing acute mania within the first 2 months. I don’t mean better efficacy, nor do I mean a lower long-term risk of inducing rapid cycling or a worsened course of illness. But in terms of acute mania, the only drugs that have been shown in double-blind randomized trials to have a lower risk of causing mania than tricyclics in bipolar type I are Paxil and Wellbutrin.
TCR: How convincing is this data?
Dr. Ghaemi: The data is limited, but it is more impressive for Paxil, for which there is double blind data on over 150 patients in three studies; with Wellbutrin you have 15 patients in one study. For bipolar type II depression there is some good double-blind data with moclobemide, which is not available in the U.S.
TCR: So can you give us some bottom-line recommendations for what we should do with our bipolar depressed patients?
Dr. Ghaemi: Bottom-line is to use mood stabilizers aggressively in these patients for their depressive symptoms. I think one reason clinicians have gotten so focussed on using ADs is that they tend to under-dose and under-use mood stabilizers. It’s a self-fulfilling prophecy-if you just give people a little bit of lithium here, and a little bit of Depakote there, and they’re still very depressed, it will seem that you have to use ADs. In some observational data that I’ve published we’ve shown that lithium and Depakote, either alone or in combination, led to a 50% improvement in Hamilton drepession rating scale scores in a 9 month follow-up.
TCR: So you recommend using lithium, Depakote, and Lamictal for bipolar depression.
Dr. Ghaemi: And Tegretol (carbamazepine), and potentially Trileptal (oxcarbazepine). So I would use these mood stabilizers aggressively and keep them in the picture, often combining two or more. And then I would go to atypical neuroleptics, since there’s some evidence they may have some benefit in depression (see TCR 1:2). Once I do all of that, I find that I don’t need to go to ADs in many of my patients. In those in whom I still need to use ADs, I certainly would, and I would use Paxil and Wellbutrin preferentially.
TCR: At what doses?
Dr. Ghaemi: Usually at low doses. I start at 10 mg of Paxil or 100-150 mg of Wellbutrin SR, and often I don’t need to go higher than that. Part of the reason I keep the dose low is that that there is probably a dose-related risk of switching to mania. Then, two to six months after they get better, I usually withdraw the AD, because while there are actually no randomized data showing that ADs prevent depression in bipolar disorder, there are some randomized data showing that it can make these patients worse over the long term. Now in some patients, as the Altshuler study suggests, you stop the AD and the patient relapses, and we have all had this experience. This happens in about 20% of patients in our clinic, and in these patients I will leave them on the AD longer. So for me, it’s not a 100%, yes or no issue, it’s just a matter of how much emphasis one puts on mood stabilizers versus ADs. I personally emphasize mood stabilizers more.
TCR: That’s saying a lot. Not only are you someone who does research and knows statistics, but you treat a heck of a lot of patients, so you really speak from clinical experience.
Dr. Ghaemi: And I’ll tell you, the most common clinical scenario in my practice, especially in consultation, are patients who have already been on many ADs, they usually either have not been on mood stabilizers or they’ve been on minimal trials, and they have chronic unremitting bipolar depression. You don’t want to have to say to these patients, “Well let’s try another antidepressant and hope the fifteenth one works,” because that almost never happens. Whereas to be able to say to them, “There may be a different strategy that will work” actually provides them a lot of hope, and I’ve seen many patients turn around with this approach.
1. Altshuler L, Suppes T, Black D, et. al. Impact of antidepressant discontinuation after actue bipolar depression remission on rates of depressive relapse at 1-year followup. Am J Psychiatry 2003 160 (7): 1252-1262.