Modeling and Simulation to Integrate Efficacy and Safety Data Following Full Development: a Case Study in Schizophrenia and Bipolar Disorder
Rik de Greef (1), Lena E. Friberg (2), Sunny Chapel (3), Matthew M. Hutmacher (3), Marita Prohn (1), Alan Maloney (4), Thomas Kerbusch (1), Mats O. Karlsson (2), Carl C. Peck (5)
(1) Clinical Pharmacology and Kinetics, Organon, a part of Schering-Plough Corporation, Oss, The Netherlands; (2) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden; (3) Ann Arbor Pharmacometrics Group, Ann Arbor, MI, USA; (4) Exprimo, Mechelen, Belgium; (5) NDA partners LLC, Washington, DC, USA.
Objectives: While modeling and simulation is an excellent tool to quantitatively inform decision making in early clinical development phases, there have been few reports of its use to summarize all relevant data at the end of a compound's full development. We present an example of such application following Phase 3 development of asenapine, a novel psychopharmacologic agent for the treatment of schizophrenia and bipolar disorder.
Methods: A population pharmacokinetic model was developed based on intensively sampled data from 11 clinical trials with asenapine in healthy volunteers or patients. The model was applied to obtain individual pharmacokinetic parameter estimates for use in PK-PD, based on sparse sampled PK data from the efficacy trials. Population PK-PD models were developed to characterize the time course of efficacy of asenapine on PANSS total, including drop-out, (schizophrenia indication) and Y-MRS (bipolar indication). Safety parameters included in PK-PD evaluations were QTc prolongation and extrapyramidal symptoms (EPS), as rated through the Simpson-Angus rating scale (SARS) and based on reported adverse events.
Results: The model describing the time course of PANSS total showed a significant exposure response relationship for asenapine . Combined with a model for drop-out, simulations were used to demonstrate consistency between the apparently mixed clinical trial results. Despite limited data on doses below 10 mg BID, the PK-PD model for Y-MRS time course quantified an exposure response relationship, enabling prediction of relative efficacy at lower doses. The limited effect of asenapine on QTc prolongation, indicated by the PK-PD analysis of the thorough QTc trial, was confirmed by a predictive check on ECG data from Phase 3. Exploratory analyses of the time course of SARS could not identify an exposure response relationship. Also, a model describing the probability of EPS-related adverse events did not detect a clear dose response trend for asenapine; individual plasma exposure was no better predictor for EPS than dose. Simulations from the four models quantified the exposure response of asenapine and associated uncertainty on the different efficacy and safety parameters, enabling an integrated assessment of benefit-risk.
Conclusions: Modeling and simulation provided an integrated quantification of the benefit-risk for asenapine, indicating a good balance for the proposed 5 - 10 mg BID dose range.
 LE Friberg et al., Modeling and Simulation of the Asenapine Exposure-Response and Drop-Out in Acute Schizophrenia. PAGE 17 (2008) Abstr 1283 [www.page-meeting.org/?abstract=1283]