NONMEM Population Models of Cytosine Arabinoside and Fludarabine Phosphate in Pediatric Patients With Leukemia

Abstract

Computer software NONMEM for nonlinear mixed-effects modeling was used to determine the population pharmacokinetics (PPK) and pharmacodynamics (PPD) of intravenously administered nucleoside analogs cytosine arabinoside (ara-C) and fludarabine phosphate (F-ara-A) in pediatric patients with leukemias. A description of the major NONMEM tasks as well as the statistical models has been given for these drugs. Two pharmacokineticpharmacodynamic (PK-PD) models have been developed for ara-C and one for F-ara-A. These NONMEM population analyses present a novel approach for determining the PK parameters in many patients; these parameters are superimposable to the ones derived from the classical two-stage PK approach. However, by minimizing errors, population analyses explained the significant variabilities in the elimination of these drugs from the central compartment and the variable high peak plasma concentrations in younger leukemia patients. Moreover, the influence of the covariate product (TR AGE × SA) demonstrated an increase in the PK parameter values with increasing patient age (AGE) and surface area (SA). The evaluation of the intercompartmental clearance of ara-C demonstrated statistically significant linear relationships with the covariate product (AGE × SA). These results explained the lower drug plasma concentration in the older children as compared to the infant and younger patients with leukemias. In addition to the classical population pharmacokinetic and population pharmacodynamic parameters of half-lives of elimination, clearance, and volumes of distribution, these analyses in pediatric patients determined the effects of deamination by cytidine deaminase on the clearance of ara-C from the circulation. Moreover, the K _m of activation of these drugs to their respective phosporylated anabolites in patients, and under treatment conditions, was determined by the Michaelis-Menten equation, which was fused into the INPUT PK-PD subroutine. Overall, NONMEM population PK-PD analyses are complicated and time consuming and require a rich database from many patients. However, they produce the best possible PK-PD analyses for a drug and its metabolites, the parameters of which give better understanding of the biochemical pathways in plasma and in leukemic blasts simultaneously, which can then be applied to many other patients in future clinical trials.