Jean Lavigne (1), Mary Lor (1) and Silvia Pace (2)
(1) Celerion, Montreal, Canada, (2) Sigma-Tau, Rome, Italy
Objectives: Develop a population pharmacokinetic (PK) model for DHA by pooling data from 5 studies and apply it to predict DHA PK in pediatric patients (6 – 12 months) infected with Plasmodium falciparum malaria after administration of a new dispersible formulation.
Methods: Subjects/patients with at least one measurable DHA concentration were included in the analysis for a total of 201 DHA profiles, 3460 samples (2340 were measurable). The MLEM algorithm in ADAPT5[1] was used to estimate the population parameters. Concentrations below the limit of quantification were treated as censored. The M3 method from Beal[2] was used. The covariates age, body weight (WGT), body surface area, sex, race, fasted/fed (FED), health status healthy/patient (PAT), formulation old/new (FORM) and crushed/not crushed were explored. The general additive model in R[3] version 3.0.1 was used for covariate selection. The Bayesian Information Criteria (BIC) was used for model discrimination and covariate inclusion/exclusion.
Results: A one-compartment model with lag time and zero-order absorption was the structural model that best fitted the DHA data. Body weight corrected dose improved the BIC. PAT was a significant covariate on Lag, zero-order duration (Tk0), and relative bioavailability (Frel) (on healthy). FED was a significant covariate on Lag and Tk0. FORM was a significant covariate on Frel (on the old formulation). For the simulations, WGT was simulated according to the WHO training[4]. Two thousand infants were simulated (gender balanced) receiving 10, 20, or 40 mg of DHA depending of their WGT once a day for 3 consecutive days. AUC, Cmax and Tmax were calculated. For the new dispersible formulation, the simulated results suggest that the geometric mean of DHA AUC (Dose/Clearance) and Day 3 Cmax should be 1160 ng/mL*h and 407 ng/mL, respectively under fasting condition and 1180 ng/mL*h and 237 ng/mL*h, respectively under fed condition. The median Day 3 Tmax would be 2.5 h and 5.1 h under fasting and fed condition, respectively.
Conclusions: A one-compartment structural model with lag time and zero-order absorption best described the PK of DHA. Body weight, health status, food and formulation were the 4 covariates which improved the model. It is expected that DHA will have similar exposure under fasting and fed conditions. Cmax under fed condition would be about half of that under fasting condition and Tmax should be delayed about 2.6 h under fed relative to fasting condition.
References:
[1] D’Argenio, D.Z., A. Schumitzky and X. Wang. ADAPT 5 User’s Guide: Pharmacokinetic/Pharmacodynamic Systems Analysis Software. Biomedical Simulations Resource, Los Angeles, 2009.
[2] Beal SL. Ways to fit a PK model with some data below the quantification limit. Journal of Pharmacokinetics & Pharmacodynamics. 2001;28(5):481-504.
[3] The R Project for Statistical Computing, R Manuals (http://www.r-project.org/)
[4] Training Course on Child Growth Assessment – WHO – Module C: Interpreting Growth Indicators. http://www.who.int/childgrowth/training/module_c_interpreting_indicators.pdf
Reference: PAGE 23 () Abstr 3122 [www.page-meeting.org/?abstract=3122]
Poster: Drug/Disease modeling - Paediatrics