Comparison of Three Methods to Analyse the Population Pharmacokinetics of High-Dose Carboplatin in Patients with Testicular Carcinoma
Georg Hempel1,2, Christiane Eickhoff3, Charlotte Kloft3, Ulrich Jaehde4
1 Universitäts-Kinderklinik, Abt.Hämatologie/Onkologie, Münster, Germany; 2 Institut für Pharmazeutische Chemie, Universität Münster, Germany; 3 Institut für Pharmazie, -Klinische Pharmazie-, Freie Universität Berlin, Germany; 4 Pharmazeutisches Institut, Universität Bonn, Germany
Introduction: Carboplatin is a widely used cytotoxic agent showing activity against several solid tumours. To reduce the variability in the individual exposure, dose adjustment based on renal function (e.g. creatinine clearance CrCl) has been suggested. To investigate the influence of several covariables for the kinetics of ultrafilterable carboplatin, we analysed a data set using three different approaches.
Methods: Blood samples from 29 patients receiving 400 to 500 mg/m2 carboplatin as a 1h infusion on three consecutive days were collected and analysed for ultrafilterable platinum. On average, 14 samples per patient were available. In addition, the following covariables were assessed: age, weight, height, body surface area (BSA) and CrCl (estimated by the Cockroft-Gault formula). An open two-compartment model was used. Three different methods were applied to analyse the data: standard two stage (MicropharmÒ ), the EM-algorithm using PPharm and NONMEM.
Results: The population pharmacokinetic parameters found are as follows:
|
STS |
NONMEM |
PPharm |
|||||
|
Mean |
SD |
Mean |
SD |
Mean |
SD | ||
|
CL [l/h] |
6.78 |
1.8 |
7.62 |
1.8 |
6.63 |
3.2 | |
|
Vcentral [l] |
21.3 |
3.6 |
21.4 |
2.9 |
19.9 |
1.6 | |
|
t1/2 alpha [h] |
2.00 |
0.50 |
1.93 |
0.48 |
1.99 |
0.42 | |
|
t1/2 beta [h] |
35.4 |
6.5 |
34.3 |
8.5 |
35.2 |
6.4 |
SD: standard deviation
For carboplatin clearance, CrCl and height were identified as cofactors by the PPharm analysis. In contrast, in the NONMEM analysis besides CrCl BSA showed the highest correlation to carboplatin clearance.
Conclusion: The three methods used provide similar population parameters whereas different results were obtained in the covariate analysis. As reported earlier, by implementing CrCl in the model the prediction of the individual clearance can be significantly improved. Currently, the applicability of the NONMEM model is further investigated in a data set of 69 patients receiving carboplatin as a 1, 24 or 96 h infusions.