Multivariate patient simulation for clinical trial optimization in COPD
D. Teutonico1, F. Musuamba1, H.J. Maas2, A. Facius3, S. Yang2, M. Danhof1, O.E. Della Pasqua1,2
(1) Division of Pharmacology, Leiden/Amsterdam Center for Drug Research, Leiden, The Netherlands; (2) Clinical Pharmacology Modelling & Simulation, GlaxoSmithKline, United Kingdom; (3) Dept. of Pharmacometrics, Nycomed GmbH, Konstanz, Germany.
Objectives: Clinical Trial Simulation (CTS) can be a valuable tool to improve drug development [1]. However, in order to obtain realistic simulation scenarios, the patients included in the CTS process must be representative of the target population. This is particularly important when covariate effects exist that may affect the outcome of a trial. The objective of this exercise is to evaluate the performance of different methods to simulate demographic covariates of patients for a Chronic Obstructive Pulmonary Disease (COPD) trial.
Methods: Virtual patients with varying demographic characteristics were simulated by re-sampling with replacement, sampling from a univariate distribution and sampling from a multivariate distribution. Simulations of continuous and categorical covariates were performed in R according to the method described by Tannenbaum et al. [2]. A KPD model was used to generate FEV1 responses in the COPD trials and results compared with the data from a real patient population.
Results: Covariate simulation using a multivariate distribution allows covariate correlations to be characterised using an empirical distribution. Moreover using the multivariate distribution is also possible to simulate new populations stratifying for specific covariates of interest.
Conclusions: Multivariate distribution methods may be applied to continuous and categorical covariates. This procedure is valuable for the optimisation of the design of clinical studies in which covariate effects are known to influence treatment outcome (pharmacokinetics or pharmacodynamics).
References:
[1] Girard P. (2005) Clinical Trial Simulation: A Tool for Understanding Study Failures and Preventing Them. Basic & Clinical Pharmacology & Toxicology. 96:228-234.
[2] Tannenbaum et al. (2006) Simulation of Correlated Continuous and Categorical Variables using a Single Multivariate Distribution. Journal of Pharmacokinetics and Pharmacodynamics. 33(6):773-794
