Dose Selection Using Nonlinear Mixed-Effect Model Averaging Approaches in Conjunction with Adaptive Optimal Design Techniques

• Activity Number: 17 - Dose Selection and PDUFA VI: Advancing the Model Based Drug Development in Regulatory Environment
• Type: Topic Contributed
• Date/Time: Sunday, July 29, 2018 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Andrew Hooker*
• Companies: Uppsala Universitet
• Keywords: Pharmacometrics; optimal design; model averaging; dose-exposure-response; adaptive designs; clinical trials

Abstract

A recent EMA/EFPIA workshop on dose finding found consensus that selection of dose for phase III clinical trials is an estimation problem and should not be addressed via hypothesis testing (Musuamba, CPT:PSP, 2017). Nonlinear mixed-effect model (pharmacometric) approaches have been widely used for the analyses of phase IIb clinical trial data to increase the accuracy of dose selection for phase III clinical trials and to characterize the dose-exposure-response (DER) relationship, useful for dose selection, as well as understanding and characterizing other aspects of a pharmaceutical compound, for example, drug-drug interactions. However, one problem with pharmacometric analyses is the potential bias introduced through model building. Here methods are presented that assume a number of predefined model structure candidates and then combine or select those candidate models; their benefit over pairwise comparisons in dose selection are shown. Further, using these candidate models one can create adaptive designs for phase IIb studies using robust multiple-model optimality criterion which could improve the characterization of DER relationships and/or lower sample size in these studies.