N. Bleyzac1,2 , F. Vial1 , B. Allard-Latour1 , J. Mouret1 , R W. Jelliffe2 , P. H. Maire1,2
1 ADCAPT, Hôpital A. Charial, 69340 Francheville, France - 2University of Southern California, Laboratory of Applied Pharmacokinetics, Los Angeles, USA
Drug disposition in man can particularly be affected by biological rhythms, such as diurnal variations observed in excretion rate by organs like the kidney. All these pharmacokinetic variations during the day should certainly be taken into account to plan drug dosage regimens, notably when toxicity-related drug accumulation may occur. However, chronopharmacokinetic studies are not easy to carry out, especially because a high number of blood samples is necessary for each subject.
In these cases, population pharmacokinetics can be particularly interesting to determine pharmacokinetic parameter values corresponding to different parts of the day or night, because it offers the possibility to use firstly, few plasma drug concentration measurements and secondly, data from retrospective studies not especially designed in a chronobiological way. In addition, clinical software collections like USCPACK are particularly useful in their way of managing patient data on the one hand, and estimating pharmacokinetic parameter complete distributions on the other hand. Indeed, the included patient data files program (PASTRX) allows to incorporate each time of drug administration and of blood sampling for a given day of treatment. Therefore, patient data treatment for chronopharmacokinetic analysis can be done more easily, especially when dealing retrospectively with clinical data from routine therapeutic drug monitoring. Because the nonparametric EM algorithm (NPEM) is able to estimate the entire pharmacokinetic distribution in a nongaussian way in addition to mean parameter values, small diurnal variations can be detected even if statistical moments remain similar. The general method allowed by the use of the USCPACK software collection (both PASTRX and NPEM) is : firstly, to divide each patient file in several subfiles according to the parts of the circadian rhythm to be studied ; secondly, pharmacokinetic parameter values are estimated using NPEM by fitting the model only to the serum levels drawn in a defined time period, keeping all the doses.
Thus, we have detected significant differences between daytime and nightime pharmacokinetic parameter estimates of amikacine and vancomycin for a same subject. In particular, elimination parameter estimates were higher in the first part of the day, according to the known diurnal rhythm of Glomerular Filtration Rate. These findings suggest to include chronopharmacokinetic data in Bayesian adaptive control of antibiotic dosage regimens in order to both increase efficacy and decrease toxicity.
Reference: PAGE 6 (1997) Abstr 653 [www.page-meeting.org/?abstract=653]
Poster: poster