Application of the optimal design approach to improve therapeutic drug monitoring for cyclosporine
Stefanie Hennig1, Joakim Nyberg1, Samuel Fanta1,2, Andrew Hooker1, Mats O. Karlsson1
1. Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden. 2. Department of Clinical Pharmacology, Helsinki University Central Hospital, Helsinki, Finland.
Objectives: Previously an intensive sampling schedule spread over 2 days was used to identify the pre-transplantation oral (PO) and intravenous (IV) pharmacokinetics (PK) of cyclosporine in paediatric renal patients and to predict the optimal post-transplant dosing strategy.[1] The aim of this study was to develop a reduced optimal pre-transplant design, within clinical restrictions, for estimation of individual empirical Bayes estimates (EBEs).
Methods: The newly developed monitoring schedule was supposed to be reduced to 3 samples/dose/patient and both doses administered within one day. Further constrains on maximum doses and infusion rates applied. The following design variables were optimized within these constraints: sampling times, doses of cyclosporine (IV,PO), time of second dose, duration of the IV infusion, administration order.
The model was based on a previously determined population PK model[1]with the population parameters being used as prior information. To optimize on the individual level, the EBEs of 8 parameters were transformed to fixed effects. The design was optimized across a discrete distribution of individual parameter vectors obtained from 77 patients, who received IV and PO cyclosporine.[1] The main covariate relationship between weight (discrete distribution), clearance (CL) and volume of distribution was included into the model and the doses were optimized as mg/kg. The ED-optimization was performed in PopED v.2.0. (http://poped.sourceforge.net/).
Results: The above method for maximizing the precision of EBEs using optimal design was implemented for the first time, as was the application of continuous distributions to represent prior information and discrete distributions for ED-optimality.
Several optimization options were explored incorporating different constraints. The pre-transplant monitoring schedule could be reduced to a total of 6 samples within an 8 hour observation interval. Both doses could be administered within this time interval complying with the constrains above. Clinically suitable designs were found for both combinations PO dose first or second. The designs were optimized to give precise individual estimates for CL and bioavailability (F) or for all PK parameters. The expected coefficient of variation on the EBEs for CL and F could be reduced on average by 50% applying these designs compared to the prior information.
Conclusions: Optimal design techniques can be applied for therapeutic drug monitoring schedules to maximize the information about the individual EBEs.
References:
[1] Fanta S, Jonsson S, Backman JT, Karlsson MO, Hoppu K. Developmental pharmacokinetics of ciclosporin--a population pharmacokinetic study in paediatric renal transplant candidates. Br J Clin Pharmacol. Dec 2007;64(6):772-784.
