Vermeulen, An; Piotrovsky, Vladimir and De Ridder, Filip
Advanced PK/PD Modeling & Simulation, J&JPRD
Introduction: A new drug candidate in development was associated with an increased risk to develop orthostasis at the start-up of therapy. Orthostasis was defined as a drop in systolic blood pressure of more than 20 mmHg when changing posture from supine to standing. This side effect was especially predominant in healthy volunteers, and was less common in the target patient population. It was postulated that the rate of change in plasma concentrations could explain this finding. Therefore, a slow release formulation (SR) was developed such that the rate of change in plasma concentrations was substantially decreased, but Cmax remained similar, i.e. Tmax occurred at a much later time point than for an oral solution (~22h versus ~2h). Subsequently, two clinical trials were designed, a single dose trial in healthy volunteers, and a multiple dose trial in patients to proof the concept.
Objectives: The objectives of the current analysis were:
· to model the pharmacokinetics (PK) of the compound after administration of an oral solution and a SR formulation to healthy volunteers and patients;
· to model the incidence of orthostasis as a function of plasma exposure;
· to check the robustness of the model by performing a posterior predictive check.
Methods and results:
Two population PK models were developed using NONMEM software: one for the oral solution and the other one for the SR formulation. The two-compartment models only differed in their absorption parameters (lag times, duration of the zero order input and absorption rate constants), whereas apparent oral clearance, intercompartmental flows and apparent volumes of distribution were shared parameters in the two models.Model predicted individual empirical Bayes estimates of basic PK parameters were obtained, and used as input in the longitudinal logistic regression model. The probability to develop orthostasis at each time point was estimated. The population PK/PD model contained a baseline model, a placebo model and a drug effect model. The drug effect model was an Emax-type model, characterized by an EC50, a Hill coefficient, Emax, and an additional parameter describing the tolerance.
The final population PK/PD model was used consecutively to perform posterior predictive checks on the two trials.
Reference: PAGE 13 (2004) Abstr 498 [www.page-meeting.org/?abstract=498]
Poster: poster