S.C. van Dijkman (1), W.M. Rauwé (1), O.E. Della Pasqua (1,2,3)
(1) Division of Pharmacology, Leiden Academic Centre for Drug Research, Leiden University, The Netherlands; (2) Clinical Pharmacology Modelling & Simulation, GlaxoSmithKline, Uxbridge, UK (3) Clinical Pharmacology and Therapeutics. University College London, 29-39 Brunswick Square London WC1N 1AX United Kingdom
Objectives: Due to high variability in pharmacokinetics, pharmacodynamics, disease progression and maturation, clear dosing recommendations regarding anti-epileptic drugs are often not available. A hidden Markov model is proposed, that incorporates the pharmacokinetic maturation over all age ranges and takes into account the disease progression. Dosing recommendations are made to improve the clinical management of this disease when treating paediatric patients with lamotrigine.
Methods: Clinical trial data in patients using lamotrigine alone or in combination with other drugs were used. Datasets included both sparse and rich steady-state PK samples across a wide range of ages (from 01 month to 18 years) and types of epileptic seizure. Drug concentrations were first used to fit a PK model that included the effects of weight and maturation. The resulting PK model was subsequently used to simulate systemic exposure, as determined by AUCs. These results were used as input for a Hidden Markov model implemented in NONMEM [2], in which changes in transition rates and probabilities of seizure occurrence were evaluated.
Results: The population PK model showed good performance with adequate accuracy and precision. The effect of weight was added allometrically and the inclusion of maturation on clearance significantly improved model fit. The hidden Markov model provided a good description of the cumulative seizures over time and the trend in seizures, with limited misspecification of active and inactive states. Transition probabilities between states were found to be significantly altered by both placebo and treatment effects, with epilepsy type and age as covariates.
Conclusions: Markov processes appear to better describe seizure frequency in epilepsy [1,3], as compared to standard time to event analysis. In addition, the use of a hidden Markov model seems to provide a more robust description of the process, when compared to Markov. It can be anticipated that dosing recommendations may be derived for children which take into account the dynamics of disease.
References:
[1] Delattre M, Savic RM, Miller R, Karlsson MO, Lavielle M. Analysis of exposure-response of CI-945 in patients with epilepsy: application of novel mixed hidden Markov modeling methodology. J Pharmacokinet Pharmacodyn. 2012 Jun;39(3):263-71.
[2] Diack C, Ackaert O, Ploeger BA, van der Graaf PH, Gurrell R, Ivarsson M, Fairman D. A hidden Markov model to assess drug-induced sleep fragmentation in the telemetered rat. J Pharmacokinet Pharmacodyn. 2011 Dec;38(6):697-711.
[3] Trocóniz IF, Plan EL, Miller R, Karlsson MO. Modelling overdispersion and Markovian features in count data. J Pharmacokinet Pharmacodyn. 2009 Oct;36(5):461-77.
Reference: PAGE 23 () Abstr 3191 [www.page-meeting.org/?abstract=3191]
Poster: Drug/Disease modeling - CNS