Irene-Ariadne Kechagia and Aris Dokoumetzidis
School of Pharmacy, University of Athens, Greece
Objectives: To develop and assess the performance of a meta-analysis method to estimate the population pharmacokinetic parameters with the respective inter-individual variabilities (IIV) from reported non-compartmental parameter values (NCP), taking into account the sampling schedule that was used to calculate the latter.
Methods: A structural model was constructed to calculate the NCPs, namely Cmax, AUC, tmax and t1/2. Unlike other methods [1, 2] in which the NCPs are calculated according to theoretical formulas, here they are calculated exactly as reported. Using a one-compartment model with first order absorption and lag time, the concentrations at specific sampling time points are simulated. Cmax and tmax are the maximum of these concentrations and the respective time, t1/2 is calculated from the slope of the log transformed three last points and AUC by the trapezoidal rule and extrapolation to infinity. A two stage approach is applied to estimate the compartmental parameters (CP), CL, V, ka and tlag. Initially, geometric means of the NCPs, namely Cmax, Tmax, AUC and t1/2, are used to estimate population means of the CP. Inter-study variability is also included as random effects to account for random variability of the different reported NCPs from which covariates can be determined. At a second stage, coefficients of variation (CV) of the NCPs are calculated by Monte Carlo sampling and used to estimate the IIV of the CP, except for tlag. Estimation was performed using FOCE method in NONMEM V7.2.0 and a custom written Fortran subroutine for the PRED. To evaluate the performance of the method simulations were carried out. The method was applied to develop a PopPK model for valaciclovir administered to paediatric population [3].
Results: The per cent relative bias in the population parameters (IIV), CL, V, ka and tlag, was -1 (-13), -3 (-20), 14 (57) and -0.2 respectively. The respective per cent RMSE was 4 (22), 6 (28), 17 (62) and 13. The results show that the method is capable of estimating the population means with satisfactory bias and precision, while the variances exhibit higher but acceptable imprecision, except for ka for which IIV cannot be estimated precisely. A paediatric PopPK model for valaciclovir was developed having age as a covariate on CL and V.
Conclusions: The method can cancel out the bias of the NCP values introduced by the relatively sparse sampling, by taking into account the exact way these were calculated in the first place and thus develop a population pharmacokinetic model from the NCPs, particularly applicable to paediatric sparse data.
References:
[1] Wang Z, Kim S, Quinney SK, Zhou J, Li L. Non-compartment model to compartment model pharmacokinetics transformation meta-analysis–a multivariate nonlinear mixed model. BMC Syst Biol. 2010 May 28;4 Suppl 1:S8.
[2] Dansirikul C, Choi M, Duffull SB. Estimation of pharmacokinetic parameters from non-compartmental variables using Microsoft Excel. Comput Biol Med. 2005 Jun;35(5):389-403.
[3] EMA, Public Assessment Report for paediatric studies submitted in accordance with Article 45 of Regulation (EC) No 1901/2006, as amended for Valtrex, http://www.hma.eu/fileadmin/dateien/Human_Medicines/CMD_h_/Paediatric_Regulation/Assessment_Reports/Article_45_work-sharing/Valaciclovir_2012_02_45_PdAR.pdf
Reference: PAGE 23 () Abstr 3134 [www.page-meeting.org/?abstract=3134]
Poster: Methodology - Other topics