Improving Drug Trials: Lessons Learned From Aripiprazole

Jeffrey A. Lieberman, MD


May 27, 2016

This feature requires the newest version of Flash. You can download it here.

Hello. This is Dr Jeffrey Lieberman from Columbia University, speaking to you today for Medscape. I'd like to offer a few observations on a recent communication[1] from the US Food and Drug Administration (FDA) about the antipsychotic medication aripiprazole, which is marketed as Abilify®, Abilify Maintena®, and Aristada™, among others.

This communication advised clinicians about a series of side effects that have emerged largely through the spontaneous Adverse Event Reporting System. It stated that from the time the compound was marketed in 2002, there have been 184 cases of what are described as "impulse-control problems." They were predominantly in the form of compulsive or pathologic gambling but also extended into other kinds of compulsive behaviors such as eating, shopping, and sexual activity. It made me appreciate several things that I think are noteworthy.

First, when a new medication is approved by the FDA for clinical use and marketing, we—the regulatory agency, healthcare providers, and the public at large—do not know the whole story about the drug's clinical profile. That is because the drug development process is such that, to gain FDA approval, only a limited number of people need to be exposed to it and for limited periods of time. Thus, for side effects that are not very common, it may be that they will only emerge when large enough numbers of people have been exposed to the medication for adequate periods of time or when the population that is prescribed the medication is expanded beyond the restrictive inclusion criteria for studies done in phase 2 and phase 3 of development. That is what I think has happened in the case of aripiprazole.

Aripiprazole, as you know, is an antipsychotic. It acts by mitigating the stimulation of endogenous dopamine receptors by dopamine. However, because it's pharmacodynamically unique among the antipsychotics by having partial agonist effects, meaning that it stimulates the D2 receptor at some percentage (approximately 30% of what dopamine would do), it can act as a dopamine agonist and is most likely to do so when administered to individuals who are not afflicted with the illness for which the medication is indicated primarily, meaning schizophrenia-related psychotic disorders. Aripiprazole is prescribed for a variety of things, such as mood disorders and, possibly, anxiety. It is in the off-label prescriptions that I would suspect that most of these reactions are occurring, also because the doses that are being administered are possibly higher than might be necessary for the target symptoms that the medication is being used for.

So, who is it being used for? It may be for individuals who have anxiety, individuals who have some type of a behavioral-control problem, mild forms of depression that aren't part and parcel of major depression, or depression with residual symptoms that have not responded to antidepressant medications.

This is an interesting phenomenon because it involves impulse-control or perseverative behaviors, which is consistent with the drug's pharmacology. As we know, dopamine is the main neurotransmitter that is involved in mediating reward and the reward pathways, hence the desire to repetitively pursue reward-providing behaviors. Second, dopamine is the most densely concentrated in the basal ganglia, which are the neural structures that are involved in perseverative or routinized motor behaviors. So, by increasing the activity of dopamine through this partial-agonist effect in that system, you could stimulate the desire to seek reward through various behaviors as well as enhance the impetus for repetitive behaviors, which are the features that characterize these adverse effects.

This report also brings to mind the situation that evolved with fluoxetine, or Prozac, in the early 1990s. Prozac was really the prototype serotonin reuptake inhibitor and was viewed as a breakthrough drug. But in the wake of its very popular use, a series of reports emerged about people who were engaging in very uncharacteristic and extreme behaviors such as sexual promiscuity, gambling, violent behaviors, attacking individuals, and even self-harmful behaviors and suicide.

I was a member of an FDA advisory committee that was sitting at the time and was asked to review the evidence for drug-induced adverse behavioral reactions with Prozac. We were faced with a real dilemma. There were empirical data that existed from the phase 1, 2, and 3 data studies done as part of the new drug application for fluoxetine, but the postmarketing experience was a small number of phase 4 studies with much larger spontaneous adverse events data. It was the latter that really reflected these unanticipated and self-harmful behaviors. I remember that the public testimonies of those who had experienced these things were chilling. The committee, however, found that the empirical data showed no adverse effect of the sort that was being suggested, and that the problem was that the spontaneous adverse event reporting and public testimonies were reflecting types of behaviors that could have been part and parcel of the illness that was being treated, meaning depression.

As everyone knows, depression has the potential complication of self-harmful behavior and suicide, so how do we disentangle the drug effect from the illness effect? In the end, it seemed that the majority of these instances were part of the illness. At the same time, knowing that Prozac was being prescribed liberally even beyond the boundaries of what it was indicated for, the possibility for serotonin agonism stimulating extreme behaviors in individuals who did not have the illness for which it was indicated remained. Although the situation with aripiprazole is different—it's a different illness and a different pharmacology—there are similarities in that aripiprazole is producing compulsive, impulse-driven, excessive, and potentially harmful behaviors, which are most likely occurring in individuals receiving the medication outside of the formal indicated illnesses.

There are two lessons to be learned. First, clinicians are now aware of this possibility and need to be vigilant for it. They also need to be judicious in who they are prescribing the medication for, because these risks are less likely to occur or they won't occur in the primarily indicated illnesses for the medication.

The second is how we can avoid this from happening in the future. What is the FDA to do? Well, the FDA is in a tough spot, because on one hand, it has to act as the safeguard of society and the American public against medications being marketed and used that aren't effective or safe. At the same time, everybody wants to see medications being developed more efficiently and made available more rapidly and less expensively. If more time is taken or required to provide the studies and the data to ensure absolute safety, it's going to take a longer time and more money in the development process. On the other hand, if it's made too quickly and easily, then it's possible for risky treatments to slip through and not be detected until the postmarketing experience, when some bad things could possibly happen.

One suggestion is that the FDA institute a formalized postmarketing phase 4 requirement on companies with newly approved pharmaceuticals. This would involve the requirement that phase 4 studies evaluate not just the effectiveness of the drug but, specifically, side effects that weren't picked up in the phase 2 and phase 3 experiences with the drug, and that these be recorded systematically as the population that receives the medication expands. The difference between that and what exists now is that adverse events are only reported spontaneously and voluntarily by physicians who use the medications and observe them in their patients, and that's not necessarily completely representative of the true clinical experience.

I believe that with these two things we can better manage the uncertainty of a drug's true potential for clinical effects when it's available to a large and not necessarily diagnostically homogeneous population.

Thank you for listening. This is Dr Jeffrey Lieberman at Columbia University, speaking to you today for Medscape.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.