Implications for animal-human scaling of the parallel elimination profile PK model
Matthew M Hutmacher (1), Roberta Bursi (2), Sunny Chapel (1), Thomas Kerbusch (2)
(1) Ann Arbor Pharmacometrics Group, Ann Arbor, MI, USA; (2) Clinical Pharmacology and Kinetics, Organon, a part of Schering-Plough Corporation, Oss, The Netherlands.
Objectives: Translation and scaling of pharmacokinetic-pharmacodynamic (PK-PD) models is performed routinely throughout early drug development to aide in dose selection in first-in-human and proof of mechanism/concept studies. The intent is to predict concentrations at the effect site and their effect in humans from PK-PD animal information. Effect site concentrations are often immeasurable in humans, e.g. the brain for CNS compounds. When human effect site data are unobtainable, assumptions are necessary to predict these concentrations. The animal PK model can inform these assumptions. Equilibration between the effect site and central compartments might suggest scaling based on the central compartment. If the central and effect site compartments have dissimilar profiles, then the effect site concentration might be considered. If the concentration profiles are not in equilibration, but have parallel elimination rates (the central-effect site rate constants are inestimable - likely due to the study design), a parallel elimination profile (PEP) PK model can be utilized. This work introduces the issues associated with scaling the PEP model for prediction.
Methods: The PEP model is derived analytically for IV dosing. The model is also described graphically through simulations, and is motivated through fits to plasma and effect site concentrations from an animal study.
Results: Fits indicate the PEP model to be a parsimonious, adequate description of the animal data. However, the PEP analytical result indicates that the VES estimate is related to the (immeasurable) central-effect site rate constant, and is thus a biased (inaccurate) volume estimate - essentially VES is an apparent volume.
Conclusion: Experimental designs can fail to support PK rate constant estimation. The PEP model can reduce the number of parameters facilitating model convergence, but yields biased (apparent) effect site volume estimates. Predicted human effect site concentrations from scaling the PEP to inform dose selection for human proof of concept/mechanism studies are biased as a result. Other information or pharmacological considerations might be necessary to improve the accuracy of the prediction to ensure suitable decision making.