Claude Magnard (1), Pauline Traynard (1), Monika Twarogowska (1), Géraldine Celliere (1)
(1) Simulations Plus, Lixoft Division, Antony, France
Objectives: Clinical trial simulations can guide the design of phase 3 trials, e.g. by suggesting the minimum sample size required to reach a given probability of success. Here we show an example of model-informed design of a bridging study to target approval of a new treatment against asthma in China, after its approval by FDA in the US. Our goal is to leverage the data already available from the phase 3 study for the new treatment in White patients, and combine it to the phase 2 data for the reference treatment in Chinese patients, to minimize the sample size and/or duration of the bridging study.
Methods: First a population model was developed using Monolix to capture the forced expiratory volume data across all studies (phase 2 and phase 3), including the reference and a new treatment for asthma, and White and Chinese populations. A simple exponential was found to provide the best agreement with the data. Effects of covariates were also investigated. The final model includes the reference versus new treatment effect, the baseline forced expiratory volume and weight which have different distributions in populations from the different regions.
Secondly, we used Simulx to perform simulations of a phase 3 trial with two arms of Chinese patients receiving the new treatment or the reference treatment. The simulation accounts for inter-individual variability, as well as the uncertainty of the population parameters. Many replicates of the clinical trial are simulated, with different individuals. Each simulation is then post-processed to assess if it demonstrates superiority of the new treatment compared to the standard of care. The percentage of successful simulated trials gives the expected power of the study. The sample size and duration were varied to observe superiority of the new treatment with at least 85% probability. The simulations can easily be defined either in the Simulx graphical user interface or via an R script calling Simulx via the lixoftConnectors R package.
Results: The simulations of the phase 3 clinical trial in Chinese patients predict that 80 individuals per arm are sufficient to conduct a successful bridging study if the duration is 6 months, and 90 individuals are sufficient for a 3-month study. This is much less than traditional phase 3 trials for asthma which usually involve at least 300 participants to show the efficacy of a candidate drug.
Conclusions: This example shows that with enough data available from different regions and treatments, a population model incorporating all the data predicts success of a phase 3 trial with a lower sample size and duration compared to the standard. Computing the explicit trade-off between cost and risk given prior information allows to efficiently enlighten decisions in phase 3 trial design.
The Simulx application of the MonolixSuite provides a straightforward workflow, where only a few clicks are necessary to assess different designs and their expected power. Models can be imported from Monolix and results are directly displayed in the intuitive graphical user interface.
Reference: PAGE 30 (2022) Abstr 10165 [www.page-meeting.org/?abstract=10165]
Poster: Methodology - Study Design