Sequential Interacting Multiple Model (IMM) Bayesian Analysis of Carbamazepine and Valproate Repeated Therapeutic Drug Monitoring (TDM) Data from Epileptic Patients
I. Bondareva, K. Bondareva
The Research Institute of Physical - Chemical Medicine, Moscow, Russia
Objectives: The objective of the study is to examine whether the new Interacting Multiple Model (IMM) sequential Bayesian method and the earlier developed nonparametric population models for carbamazepine (CBZ) and valproate (VPA) monotherapy can track and quantify the pharmacokinetic (PK) behaviour of these antiepileptic drugs (AEDs) in adult epileptic patients during their long-term therapy.
Methods: The population PK analysis was performed using the USC*PACK software based on the earlier developed linear one-compartment models for CBZ and VPA PK and repeated TDM data (peak – trough sampling strategy). This study included 42 adult epileptic patients for whom at least three pairs of serum levels measured on different occasions during different time periods of epileptic therapy were available. These data were not included in the developed population PK models of these AEDs.
Results: Epilepsy is a chronic disease, and its treatment in most patients involves long-term mono- or polytherapy. During long-term AED therapy individual PK parameter values appears to vary due to physiological reasons, concomitant medical conditions, etc. These changes can lead to significant interoccasional variability in serum AED concentrations. The traditional Bayesian algorithms are based on conventional assumption that the estimated parameter distributions are fixed and unchanged throughout the observational period, they can not describe possible PK parameter changes. When interoccasional variability is relatively high, information on serum levels measured on only one occasion is useless for future serum level predictions using conventional methods. In contrast, the IMM sequential Bayesian algorithm permits PK parameter values to jump from one support point to another, as new data are available. It was shown that the IMM can capture changes due to physiological reasons, pregnancy, adding another AED, switching to another dosage form, etc in individual PK parameter values of epileptic patients who received CBZ or VPA monotherapy chronically. Magnitude of these changes, when PK data of the individual patient measured on different occasions were analysed separately by the traditional methods, varied from 5 up to 58%.
Conclusions: The study demonstrated that the IMM algorithm described and tracked repeated TDM concentration data of AEDs well in epileptic patients with changing PK parameter values during their long-term epileptic therapy.