Milligan P.1, Karlsson M.2, Nichols D.
. Early Clinical Research Group, Pfizer Central Research, Sandwich, Kent. CT13 9NJ United Kingdom 2. Division of Biopharmaceutics and Pharmacokinetics, University of Uppsala. S-75123 Uppsala. Sweden
One of the principal aims of performing mixed effects modelling has been to provide a means of identifying and quantifying the factors that may produce pharmacokinetic and/or pharmacodynamic variabilities (interindividual and intraindividual). Following the elucidation of an appropriate structural pharmacokinetic model covariates can be incorporated into the model to account for interindividual variability in plasma concentrations. For drugs that are hepatically metabolised frequently pathophysiological factors are tested as potential indicators of metabolism. These demographically derived covariates can be considered as either a readily quantifiable measure of “hepatic size” e.g. height, weight, gender, age or “hepatic efficiency” e.g. age, and they are simply acting an extrinsic surrogate descriptors of the “true” covariate of interest – an individuals’ intrinsic hepatic function. It may be argued that armed with information on how an individual, or group of individuals, intrinsically handle a drug, it would be possible to construct a pharmacokinetic model with more of the interindividual variability explained. This approach would be of use where a large observed interindividual variability bore little or no relationship to demography e.g. in healthy (demographically homogenous) volunteers studies. Where such intrinsic hepatic function is genetically determined it may be possible to obtain a clearer understanding of the “true” situation across a population either by administering a probe compound that is considered to exhibit a high correlation in vivo to the route and rate of metabolism of the drug under study i.e. phenotype assessments, or by a more direct quantification of the level of genetic deficiencies in the metabolic enzyme(s) responsible for the elimination of the drug under study i.e. genotype. Information on the former has been available for a number of years e.g. debrisoquine and dextromethorphan for CYP2D6, and information on the later has recently become more readily available (and inexpensive) e.g. PCR based assays. The challenge is to discern how these data should be incorporated into the eventual model as a number of issues exist:
a) which is the most informative, genotype or phenotype for the compound of interest
b) how best should genotype or phenotype information be defined, as categorical or continuous variables
c) what degree of accuracy in the genotype assessment is appropriate
d) can genotype or phenotype be applied to both PK and PD
e) can any resultant model be applied to individuals where genotype or phenotype information is lacking i.e. can a descriptive model be used in a predictive sense
This paper will describe experiences following incorporation of both phenotype and genotype information into a pharmacokinetic model constructed to describe a compound currently under development at Pfizer which exhibits a moderate degree of interindividual variability, some of which may be apportioned to CYP2D6 polymorphism. Reference will be made to some of the issues outlined above.
Reference: PAGE 6 (1997) Abstr 600 [www.page-meeting.org/?abstract=600]
Poster: oral presentation