Patient compliance estimated from pharmacokinetic sample: application to Imatinib
Emilie Hénin (1), Véronique Trillet-Lenoir (1,2), Brigitte Tranchand (1,3), Michel Tod (1), Pascal Girard (1)
(1) Université de Lyon, Lyon, F-69003, France ; Université Lyon 1, EA3738, CTO, Faculté de Médecine Lyon-Sud, Oullins, F-69600, France ; (2) Service d’Oncologie Médicale, Centre Hospitalier Lyon-Sud, Hospices Civils de Lyon, F-69310, Pierre-Bénite, France ; (3) Centre Léon Bérard, Lyon, F-69008, France.
Objectives: In a previous study [1], we developed a methodology allowing the estimation of patient compliance through the immediate sequence of doses preceding the dosing interval where a blood PK sample has been taken. We assumed that the correct amount of drug was taken at nominal times. The method has been evaluated in silico with the Capecitabine/FBAL example (t½ = 3 hours) and indicated that compliance to dose given up to 5 t½ in the past could be evaluated. We considered the ratio of the plasma half-life (t½) and the interdose interval (τ) as a characteristic for our method performance. In the case of Capecitabine/FBAL, this ratio equals to 0.25 and we expect, for a greater ratio, a good estimation for more than 2 takings. The objective of present work is to evaluate further this methodology on another anticancer oral drug, Imatinib (ratio t½ / τ = 0.625), which has a greater ratio than capecitabine's metabolite.
Methods: The idea is to estimate compliance according to a single PK concentration value measured on one dosing interval at steady state and sparse samples taken after first dose. A population PK model, including estimation of all fixed and random effects estimated on a prior dataset is also needed. The sparse samples taken after first dose are combined with the population model to provide the individual POSTHOC PK parameter estimates. From those Bayesian estimates, individual PK profiles are simulated according to various dose taking scenarios and compared to the steady state nominal PK concentration. In order to be able to estimate compliance using those limited data, several assumptions were to be made, that could be released later on: (i) times of all drug intakes are supposed to be known exactly; (ii) prescribed doses are assumed to be taken or not ("all-or-nothing" approach); (iii) only the previous n doses to a PK observation can be assessed, n being dependant on the half-life of the drug and our method sensitivity; (iv) there is no inter-occasion variability; (v) individual PK profiles can be derived from POSTHOC parameters estimated using sparse data sampled after 1st dose.
In order to decide which dose among the n previous doses taken before the SS PK sample, 2n different compliance scenarios of dose taken/not taken have to be considered. The observed concentration value assumed to be taken at SS is compared to the concentration distribution predicted from each compliance profile. The compliance profile was chosen as the one minimising the Euclidean distance between the observed PK and the predicted ones simulated without residual errors.
In silico evaluation - the Imatinib example: Widmer et al.[2] proposed a population PK model of Imatinib (t½ = 15 hours). Imatinib is prescribed with an administration every 24 hours. One thousand PK parameter sets were randomly drown according to their population distribution and Imatinib concentrations following several compliance patterns were simulated considering the last 4 doses as taken or not leading to 16 compliance patterns. We have also considered different sampling times and compared the performance.
The best estimation of compliance was obtained with SS PK sample taken 5 hours after last dosing time. In this case, compliance of 91.8% patients was well estimated for immediate last dose taken, 69.9% on the last 2 takings, 44.4% on the last 3 takings and 24.4% on the last 4 takings. In addition, non-compliance appeared to be better estimated than good compliance profile.
Conclusion: In the Capecitabine/FBAL application (ratio t½ / τ = 0.25), we were able to estimate correctly adherence of the past 2 last takings. In the Imatinib application, the ratio t½ / τ is not large enough to characterise correctly the adherence of the third taking in the past. However our results on Imatinib seem more reliable since patients are prescribed 1 pill per time (while capecitabine patients are asked to take up to 7 pills per administration); the "all-or-nothing" assumption is in this case reasonable. Nevertheless the PK method is not informative enough and should be associated to electronic monitoring devices which record the actual time of system openings.
References:
[1] Henin E, Trillet-Lenoir V, Colomban O, Tod M, Girard P. Estimation of patient compliance from pharmacokinetic samples. Abstracts of the Annual Meeting of the Population Approach Group in Europe : PAGE 16-Abstr 1193, 2007
[2] Widmer N, Decosterd LA, Csajka C, Leyvraz S, Duchosal MA, Rosselet A, Rochat B, Eap CB, Henry H, Biollaz J, Buclin T.Population pharmacokinetics of imatinib and the role of alpha-acid glycoprotein. Br J Clin Pharmacol 62: 97-112, 2006
