Your browser doesn't support javascript.

Biblioteca Virtual em Saúde


Home > Pesquisa > ()
Imprimir Exportar

Formato de exportação:


Adicionar mais destinatários
| |

Estimation of unbound carboplatin clearance from total plasma concentrations as a means of facilitating therapeutic drug monitoring.

Moeung, Sotheara; Chevreau, Christine; Poinsignon, Vianney; Guitton, Jérôme; Lelièvre, Bénédicte; Ciccolini, Joseph; Gladieff, Laurence; Massart, Christophe; Fléchon, Aude; Delva, Rémy; Gravis, Gwenaëlle; Lotz, Jean-Pierre; Bay, Jacques-Olivier; Gross-Goupil, Marine; Delahousse, Julia; Filleron, Thomas; Lochon, Isabelle; Chatelut, Etienne; Thomas, Fabienne.
Ther Drug Monit; 2018 Oct 03.
Artigo em Inglês | MEDLINE | ID: mdl-30299429


Therapeutic drug monitoring of carboplatin is based on its unbound clearance (CLU) determined by Bayesian analysis on unbound (U) concentrations. However, the ultrafiltration of plasma samples presents technical and time constraints. Therefore, the present study aims to estimate CLU using total plasma (P) concentrations.


U and P concentration data of 407 patients were obtained from two clinical studies in which actual CLU had been determined for each patient. The patients were then split into development (277 patients) and prospective datasets (130 patients). Two approaches were evaluated. PK-model-only approach: a three-compartment pharmacokinetic (PK) model based on U and P concentrations and taking into account the protein binding process was developed. The model with patient covariates was also evaluated.Linear regression approach: an equation (CLU = a·CLP + b) was obtained by linear regression analysis between actual CLU and CLP which is the total plasma clearance obtained by analyzing P concentrations according to a two-compartment PK model. Predictive performance was then assessed within the prospective dataset by estimating CLU from P concentrations using each approach and computing the relative percentage error (PE) between estimated CLU and actual CLU.


The linear regression equation was CLU (L/h) = 1.15 CLP (L/h) + 0.13. The mean PE (MPE) between CLU (estimated using the equation) and the actual CLU was +1.2% (ranging from -31% to +33%) and the mean absolute PE (MAPE) was 9.7%. With the three-compartment PK model, the MPE was +2.3% (ranging from -41% to +31%), and the MAPE was 11.1%. Inclusion of covariates in the three-compartment model did not improve the estimation of CLU [MPE: +6.3% (from -33% to +37%); MAPE=11.4%].


The linear equation gives a relatively good estimation of CLU based on P concentrations, making PK-based carboplatin dose adaptation possible for centers without ultrafiltration facilities.
Selo DaSilva