M. Higashimori (1), A. Viberg (2) and M. Kågedal (2)
(1) Pharmacokinetics group, Clinical Science Division, AstraZeneca, Osaka, Japan; (2) Clinical Pharmacometrics, Clinical Science, AstraZeneca, Södertälje, Sweden
Objectives: In clinical trials investigating the treatment effect on neuropatic pain, the treatment effect is often compared at end of treatment as the difference in response between the active and the placebo group. A problem with these designs is that a pronounced and variable placebo response often is seen. The studies therefore need to be very large to provide the information needed. The objectives of this study was to see if a placebo run-in could increase the information acquired in a phase IIb study by reducing the variability in placebo response.
Methods: In neuropathic pain studies, pain is measured using an 11 graded numerical rating scale (NRS) where 0 is no pain and 10 is the worst imaginable pain. This is recorded by the patient daily before treatment until end of study. Placebo data from previous studies was used to develop a placebo model. The model predicts that the placebo response mainly occurs during the first 10 days, and thereafter the effect is stable. A design including placebo treatment for all patients during the first 10 days was suggested. Studies of the new and the previous design were simulated in NONMEM assuming a drug effect which follows a linear exposure response with delta NRS of 1.2 at a dose of 150 mg. The results from the two different designs were thereafter compared both in terms of statistical significance levels to detect an exposure response but also in precision of the estimated dose to achieve delta NRS of 1.2.
Results: With the new design the number of individuals could be reduced to 20% compared with the previous design and still achieve sufficient statistical significance to detect an exposure response. Even more important, much fewer individuals were required to be able to correctly determine a dose for phase III.
Conclusions: By using a study design where all patients in a pain study were treated with placebo initially and thereafter randomized to treatment it could be shown, using modeling and simulation, that the dose required for effective treatment in phase III could be better predicted. This will increase the probability of a successful phase III study or alternatively reduce cost and time for the phase IIb study
Reference: PAGE 21 (2012) Abstr 2437 [www.page-meeting.org/?abstract=2437]
Poster: Study Design