Application of Sparse Sampling Approaches in Rodent Toxicokinetics: A Prospective View

Abstract

In toxicology experiments in rodents, separate animals in satellite groups are routinely incorporated to assess the pharmacokinetic (toxicokinetic) characteristics of the drug. This approach has two major drawbacks: 1. limited or no individual exposure estimates are obtained for the animals used in the toxicological evaluation, preventing a quantitative assessment of the concentration/(tox)effect relationship, and 2. the use of satellite animals increases the number of animals and workload involved in a toxicological study. The combined use of sparse sampling and mixed-effects modeling is proposed to overcome these disadvantages. Depending on the specific objectives of the study, a pharmacokinetic-model-independent or a pharmacokinetic-model-dependent approach can be used. The former is based on appropriate transformation of the concentration data. A general linear model type of methodology, relating the main study variables such as dose level, time, and gender directly to the transformed concentrations, is then applied. The model-dependent approach uses nonlinear mixed-effects modeling with the NONMEM software: a pharmacokinetic (PK)-model is used to describe the concentration data in which the study variables (covariates) influence the structural parameters of the model.

Both approaches were applied to a four-week and a 26-week oral toxicity study in rodents. No satellite animals were used and sampling was performed on the animals which were evaluated for toxicity as well. Both approaches were able to summarize average concentration versus time behavior and to quantify sources of intra- and interanimal variation. The PK-model-dependent approach offers the possibility for elucidation of pharmacokinetic processes and for individualization by means of Bayesian regression. In the four-week study, the calculation of the post hoc conditional estimates in NONMEM IV yielded the individual structural PK parameters, allowing the reconstruction of the individual concentration-time profiles and the calculation of exposure measures. This option offers the possibility of directly correlating pharmacokinetic characteristics with the observed toxicological findings. Future methodological development should focus on providing and optimizing the tools necessary for establishing and describing this relationship.