Dose and sample time optimization of drug candidate screening experiments
Joakim Nyberg, Erik Sjögren, Hans Lennernäs, Andrew Hooker
Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: The estimation of the metabolic stability, i.e. the metabolic clearance, is of importance to decide if a new molecular entity will be suitable as a new drug. The current "standard" method assumes a mono-exponential decay model of the clearance. This can be a good approximation for most drug candidates but some drugs have a non-linear elimination and therefore a Michaelis-Menten model is more suitable.
To increase the efficiency in various stages in drug development, optimal experimental design has been used [1]. This approach has mostly been used to optimize sample times but it is also possible to optimize other design variables [2]. Further, when optimizing more than one continuous design variable, the simultaneous optimization approach should be considered [3].
The aim of this exercise is to determine a more optimal dose and sampling scheme that could be used for estimating the metabolic clearance for drug candidates with linear or non-linear elimination.
Methods: An analytic solution to a one compartment PK model with nonlinear elimination was used [4]. For the design to be optimal over numerous drug candidates, a modified ED-optimal design with penalty was performed in PopED v2 [5]. Briefly, the ED parameters' prior was a multivariate nonparametric distribution of 76 Vmax and Km values collected from SIMCYP [6]. The penalty function was formulated to normalize the influence of each set of parameter values in the prior on the optimal design. The design for a new drug candidate comprised 15 elementary designs (groups) with one sample and one dose for each elementary design. The samples were limited to 0-40 min and the doses were limited to 0-100uM. An upper and a lower LOQ for the concentrations where set to 0.1uM and 100uM respectively. A proportional residual variability was assumed fixed to 7.5%.
Results: Expected CV's for the modified ED-design with penalty gave, in all cases, at least a 60% improvement in expected model parameter CVs. If a standard ED-design was used the most informative parameter values tended to over influence the design, resulting in design deficiencies compared to the standard design for some types of drug candidates.
Conclusions: A method for improving the estimation of metabolic clearance for new drug candidates has been implemented. Further this method assumes a more accurate model. However, the choice of the penalty function is important to make the design robust for new candidates.
References:
[1] Mentré, F., A. Mallet, and D. Baccar, Optimal design in random-effects regression models. Biometrika, 1997. 84(2): p. 429-442.
[2] Foracchia, M., et al., POPED, a software for optimal experiment design in population kinetics. Comput Methods Programs Biomed, 2004. 74(1): p. 29-46.
[3] Nyberg, J., M.O. Karlsson, and A. Hooker, Sequential versus simultaneous optimal experimental design on dose and sample times. PAGE 16 (2007) Abstr 1160 [http://www.page-meeting.org/?abstract=1160].
[4] Beal, S.L., Computation of the explicit solution to the Michaelis-Menten equation. J Pharmacokinet Biopharm, 1983. 11(6): p. 641-57.
[5] PopED, version 2.07 (2008) http://poped.sf.net/.
[6] Proctor, N.J., G.T. Tucker, and A. Rostami-Hodjegan, Predicting drug clearance from recombinantly expressed CYPs: intersystem extrapolation factors. Xenobiotica, 2004. 34(2): p. 151-78.
