What is PAGE?

We represent a community with a shared interest in data analysis using the population approach.


2001
   Basel, Switzerland

Influence of LOQ on the pharmacokinetics parameters after an IV bolus dose

Vincent Duval and Mats O. Karlsson

Division of Pharmacokinetics and Drug Therapy, Uppsala University

This study aimed to evaluate the influence of the limit of quantification (LOQ) on the estimation of pharmacokinetics parameters after an IV bolus dose.

Method : Nine initial datasets (100 patients/400 observations) were simulated. The difference between those datasets is based on the ratio of the AUC of the distribution phase over the total AUC (0.1, 0.2 or 0.3) and on the ratio of the distribution half-life to the terminal half life (0.03, 0.1, 0.3). The terminal half life was arbitrary defined around 20 hours. Based on those hypothesis, the populations parameters were determined, intraindividual variability on each PK parameters (25% and 30%) and a 30% residual error were added. Six "sub-datasets" were build by taken away 5, 10, 20, 30, 40, 50% of the lowest concentrations values for each dataset. In order to correct the influence of LOQ, the first value below the LOQ for each subject is changed to the half of the value of LOQ (suggestion by Lewis Sheiner on nmusers). The parameters were estimated on NONMEM VIbeta with FO, FOCE and FOCE interaction methods.

Results/Discussion : The three methods result in almost the same parameters estimates. Influence of LOQ value on parameters depends of AUCs and half life ratios. The larger the half-lives ratio, the greater the influence of LOQ. The distribution parameters are the most influenced by LOQ, central volume is not as affected but it leads to a decrease of clearance. These modifications of primary parameters mainly results in an increase of terminal half life. The addition of the correction for LOQ values improves the estimation of the parameters except in the case of the greatest half-lives ratio and smaller AUC ratio.

Conclusion : After IV bolus administration, informative missingness due to LOQ value may induce bias in parameter estimates. Influence of LOQ value seems to be related to AUCs and half-life ratios. The use of a simple replacement rule considerably improved estimation.



Top