Shuhua Hu, Vitalii Nazarov, Kevin Feng, Michael Dunlavey, Robert H. Leary, Michael Tomashevskiy
Certara/Pharsight, Cary, NC, USA
Objectives: Clinical trial simulation (CTS) using Monte Carlo (MC) technique has been increasingly used in pharmaceutical industry to make drug development more efficient, robust and informative. However, it is time-consuming due to the large number of virtual individuals simulated. In the case where MC simulations are used to compare two treatment arms, we propose to use interim analysis (e.g., see [1, 2]) to avoid simulating unnecessary large number of subjects and hence improve the efficiency. The goal here is to demonstrate this approach through an example.
Methods: The example used to demonstrate this is based on the study conducted in [4], where MC simulation was used to confirm that the proposed dosing regimen and the approved one have equivalent clinical outcomes. This was done in [4] through comparing the steady-state time-concentration profiles between these two dosing regimens, where 100 replicates were simulated with each replicate consisting of 100 virtual individuals for each dosing regimen. Instead of visually comparing the steady-state time-concentration profiles of these two treatment arms, we borrowed the FDA standard for the bioequivalence study (e.g., see [3]), 90% confidence interval of ratios of the area under the concentration time curve (AUC) of the two dosing regimens contained in the range of 80%-125% (same rule applies to the peak plasma concentration, Cmax), to ascertain whether they are equivalent. The repeated confidence interval approach [2] was used to determine whether these two treatment arms are equivalent and when to stop the simulation. Specifically, an innovative simulation engine using Pharsight modeling language was used to simulate one subject at a time, and the repeated confidence interval approach was used to ascertain whether the simulation can be stopped.
Results: Numerical results show that after simulating 30 replicates, these two dosing regimens were found to be equivalent. This eliminates the need for simulating another 70 replicates as done in [4] to obtain the same conclusion, and hence reduces the simulation time.
Conclusions: This example demonstrates that interim analysis combined with simulation engine using Pharsight modeling language can be used to improve the efficiency of clinical trial simulations in treatment comparison studies and create statistical results for decision-making.
References:
[1] C. Jennison and B.W. Turnbull, Group sequential methods with applications to clinical trials, Chapman and Hall/CRC, Boca Raton, FL, 1999.
[2] C. Jennison and B.W. Turnbull, Interim analysis: the repeated confidence interval approach, J. R. Statist. Soc. B, 51 (1989), 305-361.
[3] S. Rani and A. Pargal, Bioequivalence: an overview of statistical concepts, Indian J. Pharmacol, 36 (2004), 209-216.
[4] Yim, etc., Population pharmacokinetic analysis and simulation of the time-concentration profile of etanercept in pediatric patients with juvenile rheumatoid arthritis, J Clin Pharmacol., 45 (2005), 246-256.
Reference: PAGE 25 (2016) Abstr 5731 [www.page-meeting.org/?abstract=5731]
Poster: Methodology - Study Design