Population Pharmacokinetics of High-dose Carboplatin
Andreas Lindauer (1), Christiane Eickhoff (2), Charlotte Kloft (2,3), Ulrich Jaehde (1)
(1) Dept. of Clinical Pharmacy, University of Bonn, Germany; (2) Dept. of Clinical Pharmacy, Free University of Berlin, Germany; (3) Dept. of Clinical Pharmacy, University Halle-Wittenberg, Germany
Objectives: Carboplatin is widely used in the treatment of several malignancies. While for conventional dosing (approx. 400 mg/m≤) several strategies have been reported to individualise carboplatin dose based on renal function measurements (e.g. creatinine clearance), such approaches in high-dose regimens are rare.[1-4] On a heterogeneous dataset with patients from five different studies, including 13 paediatric patients (age < 11 years), we performed a population pharmacokinetic analysis to investigate the influence of patient-specific factors on the pharmacokinetics of carboplatin.
Methods: Carboplatin was administered by intravenous infusions of varying duration (1 h, 24 h and 96 h). Daily doses ranged between 300 mg/m≤ to 2000 mg/m≤, median: 500 mg/m≤. A total of 1109 concentrations of ultrafilterable platinum from 69 patients were measured (5 to 81 observations/patient). Of 9 patients more than one chemotherapy cycle was included in the analysis. The following patient characteristics were available: age, body weight (BW), height (HGT), body surface area, serum creatinine, creatinine clearance (CLCR), sex, and co-medication with amifostine. A two-compartment model was fit to the data using NONMEM VI. Covariate selection was completed in two steps. First, generalized additive modeling and tree based modeling were applied to the base model as well as to 200 bootstrap replicates of the base model. Covariates that were considered important in the first step were tested within NONMEM by the backward deletion strategy.
Results: Intercycle variability for carboplatin clearance (CL) and central volume of distribution (V1) were estimated to be 19 and 14%, respectively. The following covariates were included in the final model: creatinine clearance on CL, infusion duration (DUR) on CL, HGT on CL and intercompartment clearance (Q), age on Q, BW on V1. Population parameter variability for CL was reduced from 50% in the base model to 21% in the final model. Carboplatin clearance in this population can therefore be calculated as follows:
CL(L/h)=6.59 x (CLCR/103.1)0.57 x (HGT/176)1.43 x (1+DUR) , with DUR = 0 for a 1 h infusion; 0.224 for a 24 h infusion; 0.319 for a 96 h infusion.
Conclusions: Creatinine clearance, infusion duration and height were found to be important predictors of carboplatin clearance in our dataset. For dosing strategies aiming at a certain target AUC, precise estimation of patient's clearance is vital. The formula we provide could improve individual dosing for patients receiving high-dose carboplatin.
 Calvert AH, Newell DR, Gumbrell LA, O'Reilly S, Burnell M, Boxall FE et al. J Clin Oncol 1989; 7:1748-1756.
 Chatelut E, Canal P, Brunner V, Chevreau C, Pujol A, Boneu A et al. J Natl Cancer Inst 1995; 87:573-580.
 Shen M, Schilder RJ, Obasaju C et al. Cancer Chemother Pharmacol 2002; 50:243-250.
 Kloft C, Siegert W, Jaehde U. Br J Cancer 2003; 89:787-794.