Abstract

Purpose: To evaluate the influence of covariates on the PPK of both parent drug and metabolite(s) from clinical trial data with neither separate metabolite administration nor excretion sampling.

Methods: A model drug (P) and its active/toxic metabolites (M1 and M2) with a metabolic pathway of P → M1 ↔ M2 plus intact excretion of all three species were used to simulate the evaluation of covariate effects on PK parameters of a 5-compartment model (2 for P, 1 for M1, and 2 for M2). A Phase II dataset was simulated, with 1690 P, M1, and M2 samples from 338 patients, collected around 0, 3, 7, 14, and 23 hr after an IV infusion (90 min) of 100 mg/m2 P. The typical excretion characteristics of the drug were assumed or available from prior studies. Fifteen covariates on five clearances (parent and metabolites) were evaluated via NONMEM® V.

Results. The problem of parameter identifiability in simultaneous PPK modeling for both parent drug and metabolite(s) was best solved with assumptions or application of prior information on the drug excretion characteristics. Eight significant covariates were identified based on alpha=0.001, independent of the accuracy of the excretion ratios which linearly determined the accuracy of the PK parameter estimates. When different excretion ratios were assumed, the relative position of the individual PK parameter values in the population distribution of the PK parameter was unchanged. The relative significance of each covariate (defined as the ratio of the coefficient of the covariate to the typical PK parameter estimate) was not significantly changed with the excretion ratios.

Conclusions: The method of fixing excretion ratios as a solution to the problem of parameter identifiability does not significantly influence the identification of significant covariates for simultaneous estimation of PPK for parent drug and metabolite(s).

American Association of Pharmaceutical Scientists (AAPS); Toronto, Ontario, Canada; November 2002

By A. Xiao and Jill Fiedler-Kelly