On Tuesday, Bristol-Myers Squibb received approval by the U.S. Food and Drug Administration (FDA) for expedited approval to review its antipsychotic medication, Abilify — already approved in the treatment of bipolar and schizophrenia — for treatment of depression in adults too. Typically such expedited approval is requested when a drug has been on the market for awhile and there is additional research data to support its use and effectiveness in another disorder. Keep in mind, “expedited approval” means little, a difference of only 4 months.
CL Psych decided to look at one of the two placebo-controlled studies the revised supplemental New Drug Application for aripiprazole (Abilify) and depression is based upon.
His first point, that a 3-point difference on the depression measure used, is clinically insignificant is important — there is no difference here from a clinical point of view. It should also be noted that the Montgomery-Asberg Depression Rating Scale (MADRS) is an older, rarely used clinical scale that has no widely-accepted cutoff for a definition of clinical depression (although 30 seems to be what is reasonable, Benazzi (1999)). These researchers found a mean score of 26 on this scale, meaning that, according to some researchers, the people studied wouldn’t even meet the accepted definition of clinical depression.
The MADRS is also a clinician-based scale, meaning the professional rates the patient on the severity of depression. Because this isn’t based upon the patient’s subjective depressive feelings, it is intended to be more objective but is also more susceptible to rater bias.
However, we don’t agree with the concerns about the study design. Of course the researchers already established the group treated with antidepressants aren’t likely to respond to that treatment when continued on them. Nor are they likely to show any greater response on placebos. So if Abilify has no effect on depression, we would likely expect there to be no response by the subjects who are taking Abilify. Objective data (in this case, the MADRS scores) don’t care about “a huge disadvantage,” because that’s a human concept, not a data- or logic-based one. From a standard research design perspective, there’s nothing wrong with the design the researchers employed (although a bit novel).
Keep in mind, the Associated Press story CL Psych quoted simply isn’t very factually accurate. BMS isn’t looking to have Abilify approved as a stand-alone treatment medication for depression, but rather as an adjunctive treatment to antidepressant therapy. This is an important difference and explains why this study may have been designed in the manner it was:
Priority Review status for an application or supplement for a drug product is assigned if a product, if approved, would be a significant improvement, compared to marketed products, including non-drug products/therapies in the treatment, diagnosis or prevention of a disease. The FDA goal for reviewing a drug with Priority Review is six months [Ed. – as opposed to the standard 10 month review].
This sNDA is based on data from two six-week, double-blind, randomized, placebo-controlled, multi-center trials (n=743) evaluating the use of adjunctive Abilify in adult patients with a primary diagnosis of major depressive disorder who had an inadequate response to monotherapy with one or more ADTs.
We’re all for picking apart studies (pharma or no), because there are so many badly designed studies out there to choose from! (Seriously, there could easily be an entire site devoted to picking apart the bad science of so much of what is published today as “research.”) But while this particular study’s measures leave a lot to be desired (and I hope the FDA sees through the weak outcome measures purposely chosen), the design is okay (acceptable, but not ideal).
Not sure where the other study on Medline is, it may have simply not made it into the database yet.
Reference: Benazzi, F. (1999). Severity gradation of the Montgomery Asberg Depression Rating Scale (MADRAS) in outpatients. J Psychiatry Neuroscience, 24(1): 51 — 52.
5 comments
John,
Thanks for discussing this issue. I think it would have been a better designed study had they taken a treatment resistant population (let’s suppose they failed to respond to an 8-week antidepressant regimen), then randomly assigned half to antidepressant + placebo and half to antidepressant + Abilify.
My guess is that the study was designed as it was (which was to only include those who had already shown inadequate treatment response on antidepressant + placebo) in order to give the best chance to make Abilify appear efficacious.
We can disagree on this point, but it seems we agree that the advantage shown on the MADRS was pretty small — i.e., Abilify didn’t look like a real breakthrough in the trial. I agree that a patient-rated measure (like the Beck Depression Inventory) would have made a nice addition — you’d think the opinion of the patient might be important in determining if a drug is an antidepressant!
Thanks for the reply. A study’s design actually does include the outcome measures chosen, and in this case, the poorly chosen outcome measures (a single, older, clinician-based rating) says a lot about the quality of this study. Which, I think, was your point (and which I’m in agreement).
Seriously, I would expect better for a study intended to gain FDA approval for a new use. (None of this, of course, stops docs from already prescribing Abilify for whatever they darned well please, though.)
Even at best, we are basing results based on clinician’s subjective reports. To mistake subjective observations that have been quantified as objective (scientific) data is the consistent error in supposed “empirical” research in the field of psychology.
So, normal: yes. Valid: not in my opinion. Subjectivity does not become objectivity simply by counting them up and showing numbers. Numbers look like scientific data, but disguise the way the data was collected. Subjects have bias, clinicians have bias.
So the studies for the drug were not the best designed, but then the question is why did the FDA approve its review? Doesn’t that say something about the nature of the FDA as well?
why does everything I read about abilify scare me away from taking it. I have read maybe 15 blogs and 2 were not negative. My dr. gave it to me and I’m scared to death to take it Annie