The September 2005 publication of the first results from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) created a stir among psychiatrists and patients. An ambitious multicenter, randomized, double-blinded study of antipsychotic medications funded by the National Institute of Mental Health, CATIE is the first study directly comparing a first-generation antipsychotic with several second-generation antipsychotics in a large number of patients with chronic schizophrenia. On behalf of Medscape, Randall F. White, MD, FRCPC, interviewed Jeffrey A. Lieberman, MD, the study's director, to explore the surprising results of this landmark study. Dr. Lieberman is Lawrence E. Kolb Chairman of Psychiatry of the Department of Psychiatry and Lieber Professor of Schizophrenia Research at Columbia University College of Physicians and Surgeons in New York City. He is also Director of the New York State Psychiatric Institute.
Medscape: Can you describe the rationale for the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE)?
Jeffrey A. Lieberman, MD: The reason for doing the CATIE study was that, since the introduction of clozapine in 1990, a series of newer, second-generation antipsychotic drugs have been approved by the US Food and Drug Administration and increasingly used by clinicians. The treatments have been embraced on the assumption that they are better and safer. As a result of their increasing use, total treatment costs have risen substantially to about $10 billion in the United States and over $14 billion worldwide in 2005. Although this price may have been required for progress, it wasn't clear whether the preferential use and cost of the new medications were actually justified. At the time the study was conceived, circa 1998, it had not been clearly established that the second-generation drugs were truly superior to the first-generation drugs and that they were cost-effective. So the CATIE initiative was undertaken to answer 3 simple questions:
Medscape: The CATIE results published in The New England Journal of Medicine in September 2005 are from phase 1 of the study. Which aspects of the trial does that cover?
Dr. Lieberman: The initial report described the primary outcomes of efficacy and safety from the first phase of the study. The design of the CATIE study had 3 novel aspects. The first was that it included all patients with schizophrenia. Instead of having numerous, rigorous exclusion and inclusion criteria, it took all comers, except people in their first episode of schizophrenia or people with a history of treatment resistance. Everybody else was included (across all adult age groups), even if they had comorbidity or required concurrent medications.
The second novel design aspect was its long-term structure -- 18 months -- which allowed us to see the effects of treatment over time and to measure effects on functioning. The third novel aspect of the CATIE study was its hybrid construction - ie, it was both a standard clinical trial and a pragmatic clinical trial. The intention was to study interventions as they are administered in real-world treatment settings.
Although our study started out as a randomized, controlled trial of 5 different medications in people with schizophrenia followed for 18 months, it involved the potential for patients to be re-randomized a second and even a third time in the second and third phases of the study, if the effects of the initial treatment did not meet patients' and their clinicians' expectations.
The September 2005 article reported the primary outcomes of the first phase of the study, in which 1500 people with schizophrenia were randomly assigned to a second-generation medication marketed at the time of the study (ie, olanzapine, quetiapine, risperidone, and ziprasidone, or a first-generation medication).
Because we couldn't study all the first-generation agents, we selected a proxy for these, which was perphenazine, a non-high-potency antipsychotic drug. We wanted one that was less notorious so people wouldn't avoid enrolling in the trial. We also wanted to give the first-generation drugs a fair chance at competing with the second-generation agents.
Medscape: The phase 1 results were widely reported in both the medical and general media, sometimes with simplistic conclusions, such as that most antipsychotic medications are not effective. In your opinion, in what way should these findings actually change clinical practice?
Dr. Lieberman: A number of key findings emerge from this report. First, although all these treatments are effective, they have clear-cut limitations in their ability to control symptoms or in their potential to cause side effects. That's reflected by the fact that only a quarter of people remained on their [initial] treatment for 18 months; three quarters of patients switched to something else in the hope it would work better.
Second, of the treatments used, olanzapine appeared to work best in terms of controlling symptoms; however, this medication also produced the highest rate of side effects. So the most effective treatment also fostered the greatest safety concerns.
Third, the most surprising finding was that when you use a non-high-potency drug in a moderate dosage, the first-generation medications do well compared with the new medications. The difference between them is not substantial.
Fourth, even though all of the medications worked more or less the same, meaningful differences in efficacy and side effects were observed. Although the magnitude of the distinctions between the drugs is not huge when you look at group comparisons, they are meaningful for an individual patient. Therefore, these differences can be used to individualize treatment.
In the past, clinicians have sought to make these comparisons by a process of extrapolation and clinical inference. The CATIE study, by virtue of its being the first to compare this many medications simultaneously in the same patient group, provides the most comprehensive "map," if you will, for these relative differences in efficacy and safety between them.
Medscape: Can you explain why olanzapine was the medication that patients stayed on the longest in the trial?
Dr. Lieberman: I think the most likely explanation is that it has the highest degree of efficacy. Like clozapine, its pharmacology is such that it alleviates symptoms of the illness better than other agents. In other words its more clozapine-like.
Medscape: How do patients subjectively experience the adverse effects that olanzapine causes vs the adverse effects caused by classic antipsychotics?
Dr. Lieberman: They may subjectively tolerate metabolic side effects and weight gain better than the neurologic side effects of classic antipsychotics. The rate of discontinuation or switching from olanzapine was lower despite the fact that it had the highest side-effect rate. Even though 30% of the patients gained 7% or more of their baseline body weight, that didn't make them or their doctor want to switch. They felt that the benefit from symptom control was sufficiently valuable to tolerate the weight gain.
Although the weight gain is unhealthy and could cause distress for cosmetic reasons, it isn't necessarily a physically uncomfortable side effect, whereas Parkinsonism, akathisia and dystonia are. And as for the metabolic side effects, if you have subclinical hyperglycemia or hyperlipidemia, you don't know unless someone tells you, and you're not uncomfortable because of it. Of course, there's a delay in the consequences of these effects, so they don't have an immediate impact on compliance or motivation to switch.
Medscape: Given the differing rates among medications of extrapyramidal side effects (EPS) and metabolic effects, can clinicians determine who might do better on a first- or a second-generation agent at the outset of treatment?
Dr. Lieberman: We haven't yet done an analysis of risk factors to know whether people of a certain gender or race, or those with a particular clinical history, are more prone to develop weight gain or EPS; that's something we do intend to do. But at this point one could look at treatment history and determine the patient's sensitivities. Let's say somebody has been prone to gain weight, and their symptoms are not particularly resistant to treatment. One might be disinclined to use olanzapine and more inclined to use ziprasidone, which in CATIE was associated with the least weight gain, or risperidone, or even perphenazine.
If, on the other hand, somebody has a history of EPS, one would want to stay away from perphenazine. Perphenazine didn't have an overwhelmingly higher rate of EPS than second-generation drugs, but it had a slight incremental risk of EPS reflected not by direct but indirect measures of this side effect. So, for people who are sensitive to that, perphenazine could be an undesirable choice, as could risperidone, which had the next highest risk for EPS.
Medscape: In the study published as "Comparative Efficacy and Safety of Atypical and Conventional Antipsychotic Drugs in First-Episode Psychosis," you reported that "olanzapine and haloperidol did not substantially differ in their effects on psychosocial functional and health status improvement in the first year of treatment following the onset of a schizophrenic psychosis."  But 50% of patients discontinued medication, more often haloperidol than olanzapine. In your opinion, are the considerations different when choosing an antipsychotic for a first-episode patient vs for a chronic patient?
Dr. Lieberman: Yes. First-episode patients are more responsive to treatment than at any later stage of illness. They're more sensitive to the therapeutic effects and to the side effects of medication. They also require doses about 50% lower, on average, than patients with chronic schizophrenia. So the considerations are markedly different for first-episode patients.
Medscape: And what about the finding that haloperidol and olanzapine did not differ substantially, unlike in CATIE?
Dr. Lieberman: What that means is that haloperidol works as well as olanzapine at controlling symptoms and facilitating recovery. Only a small proportion of people are treatment-resistant at that stage (10-15%), so most first episode patients are going to get better on just about any antipsychotic, but they're going to be very susceptible to side effects. Haloperidol was associated with a much higher incidence of EPS, and a lot of people dropped out because of that. So the side-effect profile becomes the determining factor in adherence to treatment in first episode patients.
Medscape: In the Veterans Administration (VA) cooperative study of the cost effectiveness of olanzapine vs haloperidol, the outcomes and treatment retention for the 2 medications did not significantly differ, but the costs to the VA were greater for olanzapine. For example, there were no savings because of less inpatient treatment among patients on olanzapine. Do CATIE results suggest that longer studies on cost would show different results because of better treatment adherence with olanzapine?
Dr. Lieberman: First of all, the complete results of CATIE have not yet been reported. The phase 2 results for people who switched medicines and were re-randomized will be published in the April issue of the American Journal of Psychiatry. The cognition data, the cost effectiveness data, and a variety of other analyses will be reported over the course of the next year.
One can make certain inferences or extrapolations based on the first report, but I would caution against doing so. For example, we don't know whether people who were taking olanzapine required more in the way of healthcare services, or whether the reduced amount of hospitalization required because of better efficacy was enough to offset the differential in drug cost. And we don't know how effects in other areas, such as cognition, may enter into the equation to determine cost-effectiveness.
In the VA cost-effectiveness study, one needs to be mindful of the patient population and the specific aspects of that study design. Although it's a perfectly valid design and an important population of interest, it may be that the responsiveness of this group was not sufficient to reflect differences between the treatments. To put it another way, the VA patients may have been at a fairly advanced stage of illness and was therefore somewhat more resistant to treatment.
Medscape: What psychosocial interventions might make a difference in treatment adherence and long-term outcomes in a patient population similar to that of the CATIE trial?
Dr. Lieberman: I think there's good evidence that a number of psychosocial interventions would have a positive effect on treatment adherence and outcome. First, any type of compliance enhancement therapy by case management, assertive community treatment, and psychoeducation would be useful. In addition, having some type of supportive employment would be highly desirable. These were not tested in CATIE, and they're all too infrequently provided in mental healthcare treatment settings, but one can reasonably assume that were they to be applied, they would enhance adherence to treatment and improve outcomes of patients.
Medscape: Would you like to mention anything else about CATIE?
Dr. Lieberman: Studies like CATIE, which was sponsored by the National Institute of Mental Health to evaluate the comparative effectiveness of marketed treatments, are far too infrequent. Yet, they are increasingly important given the proliferation of new medications, the costs they entail, and the dilemma they present to clinicians who are trying to select the best medication and the most effective way to sequence treatments. Studies like CATIE are long overdue, and we should be demanding that more of them be done so that clinical practice is guided not by intuition and marketing but by empirical data.