I came armed basically with one question: Which drug is best for whom? Seated around the table were Maurizio Fava, M.D., director of the depression clinical and research program at MGH, and associate directors Jonathan E. Alpert, Ph.D., M.D., and Andrew A. Nierenberg, M.D., and myself. (Hara Estroff Marano)
I didn't get the answer I wanted. What I got was more interesting.
AAN: Matching patients to antidepressants is opinion-based not data-based.
MF: There are many different positions within our field. Some believe you can predict who is most likely to respond to a drug or who is most likely to have difficulties in tolerating a drug. But the data do not support that anyone has figured this out yet in any systematic way. Then there are the nihilists, who say you can't predict anything so you might as well go by cost, something you can measure. Some people feel that regardless of the prediction factor, some antidepressants are more efficacious, at least for certain subtypes of depression. Their impressions are typically based on their own limited experience and are totally self-fulfilling. If I have a lot of faith in a drug I'll use it a lot, and use it first-line—and have a lot of success. If I don't believe in a drug I'll use it in a limited fashion in more resistant patients and the outcome will be less good.
The most rational approach is to present the patient with a menu of reasonable choices and discuss them. If you talked about 30 antidepressants, the patient would be overwhelmed. Even then the patient will ask, "but doctor, which one would you take if you were depressed?" That's a completely subjective answer.
AAN: There's another aspect. Everybody weighs the risk of side effects differently. We published data showing there's more weight gain with one SSRI [Paxil] than with others and that it causes more sexual dysfunction. There's also a problem with discontinuation reactions—feeling terrible when the drug is stopped. We might weight these more than other groups might. That's one reason why there's great variation in what people prescribe.
JEA: One handle is co-morbid illnesses. Many who present for treatment with depression have other disorders, especially anxiety disorders. The serotonergic drugs such as the SSRIs have been very well studied for the major anxiety disorders: panic, social phobia, generalized anxiety disorder, obsessive-compulsive disorder. For someone who has depression and social phobia, it's a reasonable starting point to use a medication whose effectiveness had been well-documented for both disorders.
Some of the SSRIs have been well studied for eating disorders, so they would be reasonable first choices in someone with an eating disorder too. There's not such good evidence that SSRIs are helpful for ADD, but there's a little evidence that buproprion (Wellbutrin) and the tricyclic desipramine are helpful.
Co-morbid illness provides the illusion of choice—because we have only a little positive data showing that a drug may work and no data comparing one drug versus another. Yes, some studies clearly show that Paxil works better than Wellbutrin for social phobia, but in general we don't know that something is not effective—because it might not have been studied.
MF: Some people take the view that a drug's indication [approved usage] is secondary to the investment of the pharmaceutical company, that if other drugs were studied they would probably have that indication as well. Not all SSRIs have an indication for a particular condition, but we tend to say, if one SSRI has it probably all have it, that it is a class effect.
However, the tricyclics used to be standard treatment for panic disorder, and the tricyclics mostly work on norepinephrine, so it's not necessarily true that you can only work on anxiety disorders by affecting serotonin. The more you know the literature, the more you are not so certain that there is, or should be, a first choice.
HEM: Do the atypical antidepressants give you more leverage with norepinephrine as well as serotonin systems?
AAN: If you're not that sophisticated, you'll think it works that way. But the more one looks into this, the more closely the systems are linked together, and it's hard to change one without changing the others.
MF: No one has linked neurotransmitter systems with the pathophysiology of depression yet in a convincing way. In fact, some people argue that antidepressants work by changing gene expression to promote nerve regeneration, by stimulating brain-derived neurotrophic factor. Wellbutrin is not an uptake inhibitor and it affects primarily norepinephrine, but it works as an antidepressant. Whatever theory you have, something disproves it.
AAN: That said, we could probably still agree that the drugs of the newer generation are more tolerable overall, and patients are more likely to take them in the short and long run. By and large we tend not to choose the old generation—the tricyclics and monamine oxidase inhibitors—as first-line drugs. Over time we've seen people have trouble tolerating them. It is also accurate to say that antidepressants don't work if people don't take them. If they're not willing to put up with the side effects, they're not going to take them.
MF: Once you move beyond that point into the newer antidepressants, it's difficult to know which drug to pick. Most comparative studies fail to show a difference in efficacy among the newer drugs.
And with some exceptions, comparative studies are inadequate for using tolerability as a [selection] factor. Regarding sexual dysfunction, studies favor the atypical agents. With weight gain, evidence also favors the atypicals, particularly buproprion and nefazodone (Serzone). But nefazodone bears a warning from the FDA on liver failure. So a drug promoted for sexual function and weight stability could cause you to die of liver failure.
AAN: Even that presents a problem in interpretation. The risk of liver failure with Serzone is one in 350,000 patient years of exposure. Is that a big or little risk? It's a matter of clinical judgment.
MF: The argument regarding tolerability of risk is the same one that makes people hate flying but feel comfortable driving. The risk of a car accident is much higher but is perceived to be very low. The risk of liver failure is very low, but it's fatal, you can't predict it and it's not dose-dependent.
Suggestion is very powerful—for patients and for clinicians. Strong beliefs about your choices can translate into a tremendous placebo effect. If I look directly into your eyes and say "I know exactly what's going to work for you and this is it," as opposed to "well, you have these options and what do you think, we're not sure which one is going to work for you," I'm already downsizing the placebo effect.
AAN: This is interesting for the fact that the drugs work fairly well. The effect size, which shows how something works versus placebo, is as good as many things in medicine. It's not that we don't know anything. It's just that it's much more complicated.
JEA: Residents often come to us for supervision saying "I think I made a mistake, I put someone on Zoloft yesterday and now I think Serzone would have been better." You can make a case that a person has a fighting chance to respond to either one.
When people ask what's the best antidepressant for me, we say the best antidepressant for you is the one you go on to respond to. There's a good chance they'll respond to the first or second one we prescribe, or they'll go on to respond to something else. Even there we don't know whether switching from one SSRI to another or from an SSRI to an atypical antidepressant is better. In the patient who is treatment-naïve, there's a 60 to 70% chance that whatever antidepressant we use, the person will respond.
MF: For that reason, the most important question about drug choice is what to do next, rather than what to do first, if whatever I prescribe first doesn't work.
AAN: And more important than which one first is the relationship you have with the patient. It's not just ok to say "you're taking medicine" and "goodbye" in two minutes. It's necessary to develop a caring, compassionate, genuine relationship, and part of that is also contained within the negotiation. You say, "here are several choices you could take. Here are some of the side effects. What's acceptable to you, what's not?" You treat people with respect. It's almost more the psychotherapeutic quality that happens in the first encounter.
MF: Any good psychopharmacologist keeps an open mind to the needs of the patients and is willing to switch treatment halfway if necessary, to adjust or augment and combine. There's no recipe that can be determined a priori to be best for a patient. It's an interactive process. You have to hear what the patient has to say and adjust the treatment. You don't say, "yes, you can't eat, you can't sleep, and you've lost 13 pounds but this is a great drug so just stick with it."
AAN: We take great pride here in being responsive to feedback. We don't tell the patient who is substantially worse after two weeks of treatment, "just keep doing it." We have to do something different if they can't tolerate it. But we don't spend five minutes with patients; we spend a half hour.
HEM: So is the secret that the drugs are irrelevant?
AAN: As the famous quote says, the secret of patient care is caring for the patient, making sure, once that initial treatment is chosen, and it doesn't matter that much which one is chosen, that it is an interactive tailored process over time.
MF: In order to establish a dynamic relationship you have to be willing to take responsibility for faulty decisions. And for having perhaps caused unwanted untoward side effects. What patients will say once you do that is, "I feel validated. I didn't know this was a side effect." Outcome is related to duration of treatment. If you stop the drug prematurely, you're not going to respond. You get people to stay on an antidepressant by establishing communication. In order to establish that communication, you have to be comfortable taking responsibility.
HEM: Why don't you just have a wonderful relationship with someone, prescribe chocolates and say, "these might put on a little weight but they'll make you happy."
MF: Because chocolate will have a response rate comparable to placebo, which is 30-40%, which will not match what you get with antidepressants. Which antidepressant is not as crucial as establishing a communication that allows you to change the course of treatment.
JEA: In infectious disease, getting it right on the first shot is the main issue. You get a blood or sputum culture, it tells you what the microbe is and what drug it's sensitive to, you prescribe a short course of antimicrobials and that's it. We, by contrast, are in it for the long term. The process of selecting a first-line antidepressant sets the stage for long-term management. A person will need to take an antidepressant for at least six months; side effects and other concerns will arise that we'll try to address.
AAN: Reasonable data suggest that after someone begins to respond to an antidepressant they have to be on it for another four months minimum. That usually translates into a year. Some people need to take it longer: those who've had three or more episodes in their lifetime, those above age 50 who have had two episodes, those who have had double depression, which is dysthymia, a smoldering more mild depression, on top of a major depression.
HEM: Once someone starts responding, what might make you change course?
MF: There are side effects that arise in the short term, side effects that arise in the short term and continue, and others that kick in three or four months into treatment. The one comparative study we did showed that Paxil caused more weight gain than other SSRIs after six months.
HEM: Do subtypes of depression guide you at all?
MF: MAO inhibitors have repeatedly been shown to be superior to tricyclics in the treatment of depression with atypical features. However, practically speaking, nobody uses MAOIs first-line because diet and medication restrictions are really onerous.
In general I use subtypes for what to do next. In melancholic depression, dual-action agents—Anafranil, Effexor, Remeron—may be better than single-action ones. In atypical depression, drugs that affect dopamine, like MAOIs, may be efficacious.
At best, the difference between drugs in effectiveness in a subtype is maybe 20%. What is a 5% difference? There are compounds on the horizon that are of theoretical interest for subtypes. People with pain or melancholic features may be particularly responsive to duloxetine, a balanced dual-action agent (equally active on both serotonin and norephinephrine systems). Gepirone may be particularly efficacious in atypical depression. The NK-1 [substance P] antagonists may work in diffferent subtypes. The CRF antagonists may be good drugs for people with PTSD. We need empirical evidence. We don't have it yet.