Aripiprazole, risperidone, ziprasidone
(Abilify, Risperdal, and Geodon, respectively)
(updated 12/ 2008)

Are these really  "mood stabilizers"?  The term "mood stabilizer" is starting to get a little misleading itself, though no more misleading than "bipolar" or "hypomania", so I guess we shouldn't worry about it too much.  After all, when we have some decent biological markers for what these things actually do, we'll have to re-name everything anyway, so there's not much point in grinding out a proper definition for this term. (For the very curious, here's my approach to the definition.)  

All of these medications have been shown in randomized trials to have anti-manic effectiveness.  For example, even the most recent entry, aripiprazole, appears to be a pretty good anti-manicKeck, though a little slow compared to Zyprexa and Depakote-loading.  But does it treat depressive symptoms?  As of October 2007, new data have just arrived on that, which I have outlined below. Conclusion: there is suggestive evidence that aripiprazole may have antidepressant effects, perhaps particularly at low doses for the "softer" versions of bipolar disorder such as Bipolar II/Bipolar Spectrum patients.  However, so far we do not have randomized trial evidence for such benefit in patients with Bipolar I, although in such trials the evidence did trend in the direction of antidepressant effect (details below). 

That is why, for now (2007), I regard Zyprexa and quetiapine (Seroquel) as the best choices in this family of "atypical antipsychotics" for use in Bipolar II where depression is the dominant problem. Because Zyprexa has considerably more weight gain risk than Seroquel, for the moment this means that in the treatment of Bipolar II one atypical antipsychotic stands out above the rest. That is Seroquel, at least if you go by the current standard -- namely, as one of my professors once said, "why don't you start with the ones for which we have evidence that they work?" (This is not always a fair standard when some treatments have more research funding than others, but in the case of the atypical antipsychotics, they all have big pharmaceutical companies behind them). Finally, I should remind you that I take a lot of money from the company that makes Seroquel so you should judge this conclusion very carefully, even harshly.

Worse yet, all of the atypicals  have been implicated in causing mania (e.g. quetiapineLykouras and ziprasidoneNolan  ). For risperidone and olanzapine (Zyprexa), older versions of this "second generation" of antipsychotics:  there was an article a while back implying equivalent risk of inducing mania for each of these two medications, by Aubry and colleagues.  In my clinical experience -- which is of course no substitute for, someday, a rigorous experimental approach -- risperidone has far more potential to induce manic symptoms than olanzapine.  In fact, except in the elderly, where risperidone alone seems to work like a gem (very low doses such as 0.5 mg daily seem sufficient, for example), I do not trust risperidone by itself to be a comprehensive "mood stabilizer", based on this clinical experience. (Update 3/2005: Dr. Aubrey's colleague, Dr. Berstchy, again conducted a review of the available reports of inducing mania or hypomania in 2004.  Same results, now with additional data for the newer medications.Rachid

Risperidone Update 12/2008:  a new result just published surprised me, although it is very consistent with my concerns about risperidone not having the right properties to be relied upon as an overall "mood stablizer".  Here's the background:  experience has shown olanzapine (Zyprexa) to be very effective against agitation and anxiety in patients with bipolar disorder, although very risky because of extreme weight gain and glucose level increases. Likewise quetiapine (Seroquel) has also shown value in targeting anxiety in people with bipolar disorder, though it too poses the same metabolic risks (not as much, but still a lot). Not clear yet what aripiprazole (A'... I hate that trade name, let's boycott it; you know, the cutesy one that's in all the magazines, "Abilify"?) can do. But risperidone is generic now, that ought to count for a lot, keeping costs down. Thus we're fortunate that a research team studied risperidone for the treatment of anxiety symptoms in people with bipolar disorder. Unfortunately, they found it no better than a placebo.Sheehan

The result is consistent with my experience using risperidone in people with bipolar disorder: it just doesn't act like olanzapine or quetiapine (even though it's cheaper, and about the same metabolic risk as quetiapine). I was just suprised because sometimes it works great, particularly in folks over about 60 years of age.  And it ought to have at least been a little bit helpful, a little better than a placebo, I would have thought. Conclusion: this reaffirms my disinclination to rely on risperidone as a comprehensive mood stabilizer. Anti-manic, yes, but that's about it. 

Implication: are "comprehensive mood stabilizers" supposed to have antianxiety effects? I think so. Some tools are available that do, particularly valproate/divalproex (formerly Depakote) and quetiapine.  Olanzapine too, but the risk side is too great to justify this medication unless symptoms are terrible or a lot of other things have been tried. 

New Data on Aripiprazole

The company has funded two studies showing an antidepressant effect in people with Major Depression that did not respond to a standard antidepressant.  Notice, that is Major Depression -- which means "unipolar", not bipolar depression. This is interesting, because they also funded two studies in bipolar depression which did not show positive results by the standard way we judge outcomes. Here are the details on these studies, obtained by request from the manufacturer.  From these data I conclude, for now:  aripiprazole does probably have an antidepressant effect for some people, so it is still worth thinking about as a candidate treatment for bipolar depression, even though the data did not officially support this approach by the usual standard. But for the present it does not meet the usual standards of evidence in this respect.

Bipolar Depression Studies  

In these two unpublished trials, aripiprazole did not prove to be better than a placebo, at least not to a standard level of statistical significance. In the following graphs, improvement is shown by decreasing scores on a standard depression scale.

CN 138-096    CN 138-146

As is nearly always seen in studies like this, placebo performed relatively well, leading to the obvious improvement you can see in these graphs.  You can also see that aripiprazole, the yellow line, is leading to greater improvement.  However, as indicated by the asterisks, which signify "statistically significant differences", the difference was not maintained at all points in the study, particularly at the end, when the two pills begin to look quite similar in terms of the level of improvement produced.

Patients in these studies had Bipolar I, not Bipolar II or other variations from the bipolar spectrum.  They were taking no other mood medications.  Compare the results in patients with Major Depression when aripiprazole was added to their antidepressant, shown next.

Major (unipolar) Depression Studies  

CN138-139 (Berman)


(The study on the left has been published, the one on the right is in preparation)

These graphs are not exactly comparable with one another; notice that the scale changes slightly, on the left (they do this to make the differences look as impressive as possible). But overall, you can see that the drug-placebo differences in these unipolar depression studies are more consistent than in the bipolar depression studies. The company is trying to get FDA approval for marketing aripiprazole as an add-on medication in unipolar depression, based on these data. They were surely intending to try to get a similar FDA approval for use in bipolar depression, but because the results did not maintain statistically significant difference (as shown above in yellow), they are not likely to be able to get that approval.

These FDA approvals matter more for marketing purposes for the company than they do for those of us who are interested in what treatments to use. Indeed, I think the fact that the yellow lines above trend toward differences is still useful information. Why would the medication work in unipolar depression, but not in bipolar depression? This is an important question. It may actually turn out that these studies reflect a real difference.

Meanwhile, in a big 100-week maintenance studyKeck, aripiprazole did very well against preventing relapse if patients had gotten well on this medication. but all of the preventive benefit manifest in keeping people from having manic episodes, not in keeping people from having depressed episodes (where it was no better than a placebo replacing the aripiprazole).

So, we might eventually learn something like "well, it just depends on what kind of depression you have -- some are more responsive to aripiprazole than others"; and that "bipolar versus unipolar" is not the way to identify who will respond to this medication. Earlier "open trials", without a placebo comparison group, seemed to suggest that aripiprazole has a role to play in bipolar depression. My colleague Dr. Tam Kelly in Colorado has used this strategy a lot and thinks that very low doses are useful in Bipolar II, as low as 2.5 mg every other day or even every third or fourth day.  Obviously that is a very low dose compared to the 15 mg which is the standard dose for bipolar mania.  However, he is a smart guy, and he may be on to something.  In the Major Depression studies above, the researchers used a starting dose of 5 mg and a dose as low as 2 mg was allowed.  In the bipolar studies, the starting dose was 10 mg.  Maybe they did not see what they hoped to see because the starting dose was too high?  Or perhaps Bipolar II patients respond better than the Bipolar I patients they studied in a trial shown above. We will have to wait and see, if the company chooses to conduct a study in patients with Bipolar II, which they generally do not. Maybe in this case it would be worth it...