Nguyen L, Bastian G, Doz F, Urien S, Peng B, Chatelut E, Boddy AV, Newell DR, Rubie H, Houin G, Canal P
Hopital de la Pitie-Salpetrire et Institut Curie Paris, France, Cancer Research Unit Newcastle UK, Centre Claudius Regaud and CHR Purpan Toulouse , France
In paediatric patients, relationships between carboplatin haematological toxicity and plasma ultrafilterable carboplatin AUC have been observed. They were more significant than those between toxicity and dose. A first population pharmacokinetic study using Nonmem has been performed with the data from 57 children (2 months to 18 years old) according to a bi-compartment model and proportional error model for the residual and inter-patient variabilities.
The best fit for ultrafilterable carboplatin clearance corresponded to the formula: CL (ml/min)=3D 2.85. weight. (1-0.00357. Scr).(1-0.372. Np) + 8.7 (with Scr: serum creatinine in M, weight in kg, and Np =3D 1 or 0 for unilateral nephrectomy or not, respectively). [Clin Pharmacol Ther 1996;59:april]
The predictive performance of the formula was evaluated prospectively in 28 patients (9 months to 18 years old). The median percent of error in CL was -17% (range from -48 to 29%). The Nonmen analysis of the data from the whole population (85 patients) was then performed. The best fit was found with a similar formula: CL 3D 2.9.weight .(1-0.00368.Scr).(1-0.270.Np) + 9.9. Independent deletion of each covariate significantly increased the objective function. Since body surface area (BSA) is the usual co-variate for carboplatin dosage in children, predictive performances of the last formula and bsa were compared. The mean value of CL was 74 ml/min/m2. The median values of percent error in CL calculated according to the formula and bsa were -6 and -1%, respectively. The ranges and quartiles were -42 – +115% and -18 – +19% for the formula, and -40 – +337 and -16 – +20% for bsa. The interindividual variability in CL decreased from 69% (no covariate) to 44% or 28% by taking in account bsa or the three co-variates of the formula, respectively. More detailed results showed that formula improved the prediction of CL mainly for patients whith altered renal function.
The ability to obtain accurate value of CL from 3 samples per patient was investigated prospectively in ten patients (randomly selected among the 85 patients) by using Bayesian estimation under Nonmem (with the formula and a data base of the remaining patients). Absolute percent of error between calculated and actual value of CL never exceded 20%. Since carboplatin is administered by daily short-term infusions repeated up to fives times in many paediatric protocols, it would be possible from limited drug monitoring on Day-1 combined with nonmem analysis to adjust carboplatin dosage of the remaining injections in order to achieve a desired overall target AUC.
Reference: PAGE 5 (1996) Abstr 566 [www.page-meeting.org/?abstract=566]
Poster: poster