Eugène Cox, Eric Snoeck & Jaap Mandema
Pharsight Corporation, 800 West El Camino, Suite 200, Mountain View, CA 94040, USA
The investigational compound PH2000*, active in the neurohormonal-gut axis, is claimed to improve the number of satisfying events (SEs) in symptomatic patients. An interim blinded data surveillance at 50 % of the total patient rec ruitment in three phase III trials indicated that the blinded average efficacy response rate was lower than expected on basis of the phase I I data. This raised the concern whether the ongoing phase III trials would have sufficient power to allow the demonstration of significant r esults in the Phase III program. Therefore, a population pharmacodynamic model was developed for the number of SEs on basis of available pha se II, dose-range finding trial data. This model was then used to perform clinical trial simulations of the ongoing phase III program and th e expected range of outcome was evaluated.
The weekly number of satisfying events were collected from 402 patients over a 4 week treatment period. Using NONMEM vers ion V, these count data were analyzed on basis of a probabilistic population pharmacodynamic model, describing the relationship between the mean rate of SEs (l ), dose and time. Between- and within-patient variance were modeled using a log-normal distribution and a Poisson distribution, respectively .
Simulation of the blinded data surveillance outcome on basis of this pharmacodynamic model showed that the simulated resp onse rate was similar to the actual observed response rate. Simulations also showed that the two active doses evaluated in the phase III pro gram, would have a high likelihood of showing a significant improvement in response rate compared to placebo. Subsequent unblinding of the d ata from the completed phase III trials indeed yielded significant response rates for the two active doses within the simulated confidence i ntervals.
In conclusion, clinical trial simulations on basis of suitable pharmacodynamic models are a powerful tool to evaluate the expected range of outcome and evaluate the impact of key assumptions and uncertainties on these outcomes.
*
The name of the drug has been sanitized for confidentiality reasonsReference: PAGE 9 (2000) Abstr 94 [www.page-meeting.org/?abstract=94]
Poster: poster