June 2, 2014. In the Tuesday, April 8 plenary session at the Schizophrenia International Research Society meeting in Florence, Italy, entitled "Update on Therapeutics: Improving the Clinical Yield," speakers from academia and industry discussed lessons learned from past failed clinical trials and ways to improve those of the future. One common theme was the need to think beyond a single definition of "schizophrenia" and to select subgroups of patients that are relevant for the specific compound being tested.
In support of this argument, Bruce Kinon, Eli Lilly, Indianapolis, Indiana, provided several possible reasons why Lilly's mGluR2/3 agonist pomaglumetad methionil failed to meet its endpoints (see SRF related news report; SRF news report). The antipsychotic-responsive patients (who presumably have a dopamine-mediated illness) who were used in trials may not have responded to a glutamate-based treatment, he suggested. The glutamate drug could also require a patient who is hyperglutamatergic, which studies have suggested is a state associated with early-stage patients, not the more chronically ill tested in Lilly's trials. A third reason for the drug's failure, said Kinon, could be mGluR2 downregulation resulting from prior exposure to atypical antipsychotics that block 5HT2A receptors. In fact, a post-hoc analysis found that patients treated only with D2-predominant drugs did respond to pomaglumetad (in contrast to those exposed to serotonin-blocking agents).
Richard Keefe, Duke University, Durham, North Carolina, also emphasized a careful consideration of the characteristics of trial participants. Although it may be easier to recruit stable, chronic patients with schizophrenia, he said, this population may not be the mostly likely to show cognitive improvement. Instead, he suggested that younger subjects earlier in their course of illness may be a better choice, citing data showing that younger subjects benefit more from cognitive remediation than older ones.
Keefe also emphasized the need to pay attention to study parameters such as the time of day that cognitive tests are performed. He described two clinical trials (encenicline and GABA agonist/dopamine antagonist BL-1020) in which larger effect sizes of cognitive improvement were achieved in patients who received baseline and endpoint cognitive batteries at the same time of day compared to subjects with inconsistent timing of testing. Ensuring consistency across raters is also important, he added.
The challenges of assessing negative symptoms in schizophrenia were addressed by Daniel Umbricht, F. Hoffman-La Roche, Basel, Switzerland. He discussed whether caregiver information should be incorporated into ratings of negative symptoms, which often deal with inner mental states that can be difficult to observe, and presented data from Roche's Phase 2 trial of now abandoned glycine transporter inhibitor bitopertin demonstrating that placebo response was significantly smaller in patients for whom caregiver information was incorporated into clinical assessments.
Umbricht also discussed whether negative symptoms should be lumped together or split into two factors (avolition/asociality and expressive deficits), arguing for separate evaluations of key dimensions. In order to minimize the expectation bias that can contribute to placebo response, he underscored the need for fully blinded raters. Finally, he also emphasized the importance of detailed characterizations of patients in order to identify subgroups of responsive patients.
Chris Schmidt, Pfizer, Groton, Connecticut, discussed lessons learned from the PDE10 inhibitor drug development program. He suggested that, in addition to looking for efficacy, researchers need to pay attention to the value of clinical trials for furthering the understanding of the underlying neurobiology of schizophrenia. "We need to walk away from these programs actually knowing that we tested the mechanism," he said.
Jonathan Rabinowitz, Bar-Ilan University, Israel, reviewed the advantages of data sharing across clinical trials and showed how real-time data sharing can improve patient selection by identifying and eliminating duplicate enrollments. He also presented data suggesting that studies can be made shorter and smaller, with the benefits of reducing placebo exposure and conserving resources.
Finally, session chair John Kane, Zucker Hillside Hospital, Glen Oaks, New York, touched on the struggle of the ever increasing placebo response across time. He also called for better assessments of inter-rater reliability, which he reminded the audience can greatly affect statistical power. Like many of the other speakers, Kane emphasized the importance of selecting the specific population of patients that is most likely to benefit from a particular treatment. "We need to think a lot about subgroups," he said.—Allison A. Curley.