Purpose: To characterize the population pharmacokinetics of oral busulfan in 48 children including pooled data from three transplantation centres with the aim of estimating the variability in the kinetics of busulfan and to identify covariates that could be used for dose calculation.
Methods: A total of 508 plasma samples from 250 administrations (mean 9 samples per patient over 4 days of treatment) were collected from 48 children receiving busulfan orally every 6 h. The dosing varied between 13 and 20 mg/kg with seven patients receiving a dose of 600 mg/m(2). The busulfan formulations administered varied considerably. They included 2-mg tablets (Myleran), gelatine capsules, crushed tablets suspended in water and suspension for administration via nasogastric tube. Samples were analysed for busulfan either by HPLC using postcolumn photolysis or by LC-MS. Plasma concentration-time data were analysed by population pharmacokinetic modelling using NONMEM.
Results: Busulfan kinetics were best described by a one-compartment model (subroutine ADVAN 2 TRANS 2). Residual variability was modelled using a combined additive and proportional error model. The influence of different covariates on the pharmacokinetic parameters was tested. The best results were obtained by inclusion of body surface area (BSA) as a covariate for clearance (Cl/F) and volume of distribution (V/F). The final population estimates were: Cl/F 4.13 l/h per m(2) +/-26%, V/F 21.3 l/m(2) +/-31% and ka 1.31 h(-1) +/-110% (population mean +/- interindividual variability, IIV). Variability in one patient during the 4 days of treatment (interoccasion variability, IOV) for Cl/F (10%) and V/F (19%) were calculated to be less than interindividual variability, fulfilling the condition for individualization of busulfan dosage regimens.
Conclusions: In our paediatric population, BSA, not body weight, is the best predictor of Cl/F and V/F. Our final estimations reflect the wide interpatient variability after oral administration of busulfan with an IIV for ka of 110%.