Population pharmacokinetics of AZT and its active metabolite AZT-TP in HIV patients: joint modelling and design optimisation
C. Bazzoli (1), S. Retout (1, 2), E. Rey (3), H. Benech (4), J.M. Tréluyer (3), D. Salmon (5), X. Duval (1, 6, 7), F. Mentré (1, 2) and the COPHAR2- ANRS 111 study group
(1) INSERM, U738, Paris, France; Université Paris 7, Paris, France; (2) AP-HP, Hôpital Bichat, Paris, France; (3) AP-HP, Hôpital Cochin, Département de pharmacologie clinique – EA3620, Paris, France; (4) CEA, Service de Pharmacologie et d'Immunologie, DSV/DRM, Gif sur Yvette, France; (5) Université Paris 5, AP-HP, Hôpital Cochin, Service de Médecine Interne, Paris, France; (6) AP-HP, Hôpital Bichat, Service des Maladies Infectieuses B, Paris, France; (7) CIC 007, Hôpital Bichat, Paris, France
Objectives: To determine a joint pharmacokinetic (PK) population model of azidothymidine (AZT) and its active metabolite AZT-TP in HIV infected patients and to optimise several designs for further joint population PK analysis of AZT/AZT-TP.
Methods: In the COPHAR2 - ANRS 111 trial, 75 naïve HIV patients received orally 300 mg twice daily of AZT, as part of their tritherapy treatment. Four blood samples per patient were taken after two weeks of treatment to measure the concentration at steady state at 1, 3, 6 and 12 (trough) hours. Concentrations of AZT, quantified by HPLC, were measured in 73 patients. AZT-TP concentrations were measured in 62 patients using a direct LC/MS/MS, a costly method performed in a specific laboratory in France. Using the SAEM algorithm implemented in the MONOLIX software version 2.4, which can handle data under the LOQ [1, 2], a population PK model was developed in order to, for the first time, simultaneously describe the PK of AZT and AZT-TP. Based upon this model, we first evaluate the design used in COPHAR 2 assuming 50 subjects, called the empirical design. We then explored D-optimal population designs for further joint population AZT/AZT-TP analysis using the Federov-Wynn algorithm implemented in PFIM 3.0 [3]. To keep the same constraints as for the empirical design, we first optimise population designs with only four sampling times common to both measures with a set of 12 admissible sampling times at 0.5, 1, 1.5, 2, 3, 4, 5, 6, 7, 8, 10, 12 hours. Due to the different PK profiles between the plasma and intracellular concentration, we also optimize other population designs with different interval of admissible times for AZT and AZT-TP and with different constraints regarding the number of samples per patient.
Results: A one compartment model with first order absorption and elimination best described AZT concentration [4], with an additional compartment describing the metabolism of the drug to AZT-TP with first order elimination. Optimal design, with quite similar constraints to the design used in the trial has a better efficiency. More general optimisation show that optimal designs allow as precise parameter estimates as the empirical design but with less samples per patient.
Conclusions: A joint model was found to describe adequately AZT and AZT-TP concentrations and was used to estimate population PK parameters of AZT-TP. We optimised population designs with lower number of AZT-TP samples involving thus a more reasonable cost.
References:
[1] Samson A, Lavielle M, Mentré F. Extension of the SAEM algorithm to left-censored data in nonlinear mixed-effects model: application to HIV dynamics model. Computational Statistics and Data Analysis 2006; 51:1562-1574.
[2] www.monolix.org
[3] www.pfim.biostat.fr
[4] Panhard X, Legrand M, Taburet A.M., Diquet B, Goujard C, Mentré F and the COPHAR 1-ANRS 102 study group. Population pharmacokinetics analysis of lamivudine, stavudine and zidovudine in controlled HIV-infected patients on HAART. European Journal of Clinical Pharmacology 2007; 63:1019-1029.