Taright N., Mentré F. ,Jouvent R.*, Ammar S.*, Mallet A.
INSERM U 194 et Service d'Informatique Médicale, *CNRS EP 53 et Service de Psychiatrie, CHU Pitié-Salpêtrière, 91 bd de 1'Hôpital, 75634 Paris Cedex 13
A population analysis of the kinetics of lithium is performed on a set of experimental and observational data, composed of 37 healthy volunteers and 76 patients with bipolar disorder. A bicompartment model is used to describe the kinetics of lithium. Age, body weight, height and serum creatinine are included as covariates. The population characteristics are provided by the nonparametric maximum likelihood method (NPML) which estimates the whole distribution of the kinetic parameters and of the covariates. The results of the estimation are consistent with those of earlier studies and confirm a large interindividual variability. Relationships between kinetic parameters and covariates are studied.
Data from a separate group of 35 patients are used to assess the obtained results. All of them were assigned to the same dosing scheme and had a concentration measurement 24 hours after : (1) the estimated distribution of the concentrations obtained with the NPML analysis has been found to be similar to the observed one; (2) the relevance of the covariates has been assessed using the estimated distribution of the concentrations conditional on covariates. The mean standardized difference between expected concentration and observed one is lowered when using the covariates. Furthermore, 83% (n = 29) of the observed concentrations lie within the estimated 80% confidence interval.
The satisfactory results obtained in the validation step allow to propose a Bayesian approach for individualizing the dosing scheme. The method chosen is based on the minimization of a quadratic risk function with respect to the residual concentration at steady state. A prospective evaluation of the method is in progress.
Reference: PAGE 2 (1993) Abstr 912 [www.page-meeting.org/?abstract=912]
Poster: oral presentation