TCR: Dr. Gitlin, to begin with, my readers would be interested in your impressions of the new kid on the block, Cymbalta (duloxetine).
Dr. Gitlin: I have had approximately two months of clinical experience with Cymbalta, so I am still getting a sense of both its tolerability and its efficacy. As we know from a theoretical standpoint, the difference between Effexor (venlafaxine) and Cymbalta is that Cymbalta has dual reuptake inhibition at every dose, even at low doses like 20 mg, whereas with Effexor, the norepinephrine reuptake inhibition kicks in somewhere between 100 mg and 150 mg for most people.

TCR: Have you noticed any differences between the two?
Dr. Gitlin: One clear difference in my clinical experience is that Cymbalta "feels" different from Effexor in its side effects. Effexor has the exact same side effect profile as any SSRI, the one exception being the dose-related hypertensive effect that really only starts at 150 mg and is more prevalent as the dose increases. Cymbalta, however, has other side effects that are more associated with what we tend to think of as "anticholinergic" effects, such as dry mouth and constipation, as well as showing high rates of nausea. In my experience, nausea has not been a big problem--I have had only one patient who found nausea to be more than minimally difficult, and for most people it just goes away within the first week.

TCR: How many patients have you had on Cymbalta so far?
Dr. Gitlin: Twelve to fifteen, and overall it has been pretty well tolerated. Of course there are a few people who didn't like it at all and a few people liked it a lot. Do I think it is anything extraordinary, that it is a great leap forward? The answer is no. I am always glad to have another tool in my shed, but do I think it has absolutely unique properties? Not so far.

TCR: What about the idea that it has something unique to offer people with physical pain?
Dr. Gitlin: That is a difficult one. Certainly, Lilly's data show pretty clearly that it is effective for physical symptoms associated with depression and it did get the FDA indication for use in neuropathic pain. However, we don't know what would happen if we compared Cymbalta's efficacy for somatic symptoms of depression to other dual action drugs, like tricyclics, Effexor, even Remeron (mirtazapine). Would it really be any different? Unclear.

TCR: There has been a buzz in the field now for several years that Effexor might be more effective than the SSRIs, at least for remission of depression. Now that we have had several years to think about all these metanalytic statistics, what is your current take on this issue?
Dr. Gitlin: Well, after several years, what we have is a much deeper appreciation of the muddle! I don't think we know the answer any more now than we did when Thase published that first meta-analysis in the British Journal. In that pooled analysis and then in the subsequent one done with a larger database published later, it does look as if Effexor has higher remission rates than do the SSRIs. There are, of course, difficulties with that database, one of which is that remission was not the endpoint in any of those studies. So these were all posthoc analyses and there is only one later study (comparing Lexapro to Effexor), that examined Effexor and an SSRI in which the stated purpose of the study was to compare remission rates. Now, frankly, if I were Wyeth, I wouldn't do any prospective studies in this area either. They have gotten hoards of mileage out of these posthoc analyses, and yes it would be nice if they did a real prospective study, but what if it didn't work? I think they would lose all of the great marketing they have gotten, so I suspect they have no intention of doing that.

TCR: In your clinical practice, have you noticed efficacy differences between Effexor and the SSRIs?
Dr. Gitlin: Well, I would have to say that, given the small differences reported in the pooled analysis (45% remission for Effexor, 35% for SSRIs, and 25% for placebo), it is almost impossible for an individual clinician to notice that level of difference in a patient population. You would have to have the most extraordinary memory and the ability to keep separate each individual patient over months and years. And no one is able to do that. If I can answer a related question, there are lots of patients I treat who have failed full trials of one or more SSRIs who I switched to venlafaxine and they really have a qualitatively better response. Now, the reason that is not an answer to your question is, if I hadn't switched them, they might have done better anyway. What is the effect of time? The problem we have as clinicians is the difficulty in generalizing from any individual case because there are no controls.

TCR: Let's move on to Lexapro, which has been billed by the Forest reps as the "perfect" SSRI.
Dr. Gitlin: My feeling is that Lexapro is a very good SSRI that is relatively well tolerated, as was Celexa, although I don't like the way Forest has marketed it--as the "perfect SSRI," as if stereo isomers were the next great revolution. The major clinical difference I see between Celexa and Lexapro is that Lexapro is slightly less calming, tranquilizing, and sedating, and that is neither a good nor a bad point: It depends on the patient. There are times I would want to use Celexa, if somebody is a little more agitated or insomniac and sometimes I will want to use Lexapro, so it is a neutral quality depending on the patient.

TCR: Getting into other treatment issues, when do you decide to switch out of the SSRI class when a patient is not responding?
Dr. Gitlin: If one of my patients has failed a full trial of an SSRI, I either add an adjunct or I switch outside the class. If you look at the studies of people who have failed a trial of an SSRI, and there are about five of them, all methodologically flawed, it looks as if 50% of people who fail one SSRI will respond to a second. Clinically, however, that is not my experience--I would say the response rate is more like 20-25%. Well, 20-25% is not trivial but it is also not dramatic, which is why I tend to move on after one trial.

TCR: What do you tend to move on to?
Dr. Gitlin: I will either switch to Effexor (and possibly now Cymbalta), or of course, Wellbutrin (bupropion), the idea being that you are just trying to do something that is biologically distinct from what has already not worked.

TCR: And you would tend to switch before you augment?
Dr. Gitlin: You know, I have thought long and deeply about this issue, and I cannot tell you that I have a rational scheme for why I do what I do. I clearly do both. There are numbers of patients who fail an SSRI and I will add Wellbutrin and there are probably an equal number of patients who I will switch to Effexor or Wellbutrin. Some patients are fine adding and others want to keep things simple; they are ambivalent about taking psychotropic medications to begin with, so for them I will do more switching. I think the data so far suggests that adding may be slightly more effective than switching, but there is not sufficient data to answer that question in any meaningful way.

TCR: And what do you turn to for adjunctive medications, usually?
Dr. Gitlin: The most common adjuncts I use are other antidepressants with a different mechanism action. The classic example is adding Wellbutrin to a strongly serotonergic drug, whether it is an SSRI, Effexor, or Cymbalta. Sometimes I add a stimulant to the original antidepressant . It is a shame that there are truly zero double-blind studies examining the efficacy of stimulants as adjunctive antidepressants. There actually are a handful of controlled studies from the '70s looking at the efficacy of stimulants as solo treatment for depression. The studies showed that they were sort of effective, less than tricyclics, but somewhat better than placebo. However, there are no studies looking at it in the way that I use it all the time, which is as an adjunct to another antidepressant. Nonetheless, my clinical experience and that of most of my colleagues is that stimulants are pretty effective adjuncts and so I use them pretty regularly. I also use Provigil (modafinil) as an adjunct--as you probably know, there are some double-blind studies out there that indicate something of an energy effect and not much of a mood effect. Nonetheless, I have had some very good responses to Provigil as an adjunctive antidepressant.

TCR: Looking beyond Wellbutrin, Effexor, and stimulants, do you have a next tier that you tend to go to if those don't work?
Dr. Gitlin: From a data point of view, the two adjuncts for which there is the most evidence are Cytomel (T3) and lithium. However, I tend to use them relatively late in my algorhithm simply because I have not been convinced that Cytomel works all that well, and because lithium is poorly accepted by patients. Furthermore, I have not found that it is as effective as the literature would suggest. So, if I really had a third choice for adding an agent, it would probably be low doses of an atypical antipsychotic--for example, 0.5 to 1 mg of Risperdal or 2.5 mg of Zyprexa or 2.5 to 5 mg of Abilify.

TCR: Thank you Dr. Gitlin.