Population modeling of tumor growth inhibition in vivo: application to anticancer drug development
Simeoni M (1), Poggesi I (2), Germani M(2), De Nicolao G(1), Rocchetti M(2).
(1)University of Pavia, Italy; (2)Pharmacia Italia S.p.A., Nerviano, Italy
Introduction: The in vivo evaluation of the antitumor efficacy of compounds in animal models is a fundamental step in the development of anticancer drugs. In these experiments, efficacy is expressed as percentage of decrease of the tumor weight in treated animals compared to control animals. We developed a minimal pharmacokinetic-pharmacodynamic model linking the dosing regimen of an anticancer agent to the tumor growth in animal models. The growth of tumors in non-treated animals (unperturbed growth) is described by exponential growth followed by a linear growth phase. The rate of tumor growth in treated animals (perturbed growth) is considered decreased by a factor proportional to both plasma drug concentrations and number of proliferating tumor cells. A transit compartmental system is used to model the delayed process of cell death. The parameters of the pharmacodynamic model are related to the growth characteristics of the tumor, to the drug potency and to the kinetics of the tumor cell death. Since the unperturbed and perturbed growths are measured in different groups of animals and considering that in this model the perturbed growth collapses into the unperturbed one in the absence of treatment, the simultaneous fitting of the two average growth curves was adopted for estimating the model parameters.Aim: In this communication we report examples of the use of population approaches for modeling the outcome of these experiments. This would allow estimating the different sources of variability.
Methods: The examples refer to treatments with paclitaxel and an anticancer candidate synthesized as part of our discovery programs. Pharmacokinetics were described using 2-compartments open models with first order elimination from the central compartment. Tumor weights were modeled using the approach described above. The PK-PD model was implemented using NONMEM (v. V). For both PK and PD data inter-group and inter-animal variability were described using multiplicative models. Different models were used for describing the random error.
Results: The model fitted well the experimental data. The use of the population approaches allowed the identification of the PK-PD parameters and the description of the relevant sources of variability.
Conclusion: The use of the non-linear mixed effect model allowed the full exploitation of the capabilities of the PK-PD model of tumor growth. Since the model was proven effective also in predictive mode, based on the outcome of a preliminary experiment, stochastic simulations can be implemented for efficiently simulating the whole campaign of tumor growth inhibition studies for a novel anticancer agent.