A Simulation Study to Compare Modeling Approaches for Dose Proportionality Data

• Activity Number: 670
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Andrea Maes*+ and Wen-Lin Luo and Yu Ding and Deborah Panebianco
• Companies: Novartis Pharmaceuticals Corporation and Merck & Co., Inc. and Merck Research Laboratories and Merck & Co., Inc.
• Keywords: Pharmacokinetics ; Power model ; Contrasts

Abstract

A dose proportionality study is conducted during drug development to support a claim of linear pharmacokinetics in the label. Dose dependent pharmacokinetic (PK) parameters from the study are often analyzed using a power model in which log(PK) data are regressed onto log(dose). An estimate of the slope and its confidence interval (CI) are calculated from the model with 1 reflecting proportionality. Through simulation, we compare the power and type I error of asserting proportionality after modeling log(dose) as a continuous versus categorical factor in a linear mixed model. The slope is estimated through contrasts in the latter approach. Variance heterogeneity and slight deviations from log-scale linearity are also considered. Results show that the contrast method has greater power and closer to nominal CI coverage under slight non-linearity, but may be more sensitive to heterogeneity.