2012 - Venice - Italy

PAGE 2012: Study Design
Daniel Tatosian

Strategic comparator modeling and simulation to optimize dose selection for a dose-range finding study

D. A. Tatosian (1), J. Z. Peng (2), A. Zandvliet (1), J. Elassaiss – Schaap (1), Ashley Martucci (1), Rajesh Krishna (3), Anya Kauh (4), Gendrano III (5), B. Musser (4)

(1) Clinical PK/PD, Merck & Co., Inc., USA; (2) Modeling & Simulation, Merck & Co., Inc., USA; (3) Product Value Extension, Merck & Co., Inc., USA; (4) Clinical Pharmacology, Merck & Co., Inc., USA; (5) Clinical Research, Merck & Co., Inc., USA

Objectives: Drug X is under development using a rapid straight-to-IIB development paradigm. To support dose selection for this dose-range-finding study, Pharmacokinetic/Pharmacodynamic (PK/PD) models were developed for drug X and a reference drug Y to conduct clinical trial simulations.

Methods: PK/PD models describing the PK and PD of an intermediate biomarker for Drug X and an internal reference Drug Y were developed. Additional PD-PD models relating the observed trough and individual-predicted average levels of intermediate biomarker A to the primary efficacy endpoint were developed. Using the linkage through the intermediate biomarker, the combined PK/PD models were used to conduct clinical trial simulations to evaluate the probabilities that given a selected set of Phase IIB doses, a dose-response relationship could be accurately determined. 

Results: The plasma pharmacokinetic and intermediate biomarker data for Drug X could be described by nonlinear distribution of the drug to target biomarker present in both the central (sampling) and peripheral compartments. A linear 2 compartment model described the PK of Drug Y. The relationship of drug concentration to the intermediate biomarker was well characterized by an empirical Emax model. The relationship between the trough levels of the intermediate biomarker and primary endpoint was captured by sigmoidal Emax models. As a sensitivity analysis around the assumption of trough relation to efficacy, a second model relating the individual-predicted weighted average intermediate biomarker levels to primary endpoint was also developed. Trial simulations from both models were conducted for various IIB study sizes and dose selections, with parameter distributions obtained from bootstrapping the datasets. From the clinical trial simulations, probabilities to accurately capture the predicted dose-response model were calculated by refitting each simulated trial to a dose-response model and comparing the accuracy to the predicted true dose-response model.

Conclusions: PK/PD models were developed for a new drug under development and a reference drug, with a shared intermediate biomarker. By using the shared intermediate biomarker to bridge, trial simulations were conducted to enhance the decision-making and guide the development team to optimize the doses selected for the dose-range finding study.




Reference: PAGE 21 (2012) Abstr 2588 [www.page-meeting.org/?abstract=2588]
Poster: Study Design
Top