Optimal Designs for Low-Dose Linear Extrapolation in Carcinogenesis Experiments

• Activity Number: 225
• Type: Topic Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Melvin Slaighter Munsaka*+
• Companies: Takeda Global Research & Development Center, Inc.
• Keywords: Optimal design ; Low dose extraplolation ; low dose slope

Abstract

This paper investigates the utility of optimal designs in improving inference associated with low-dose linear extrapolation. The investigation is exploratory with the objective of characterizing designs for which the magnitude of the upper confidence limit of the low-dose slope is small. Hence, the question of primary concern is: Can carcinogenesis experiments be designed optimally to make low-dose linear extrapolation less restrictive? A simulation study is used to identify designs that are optimal in the sense that the upper confidence limit of the low-dose slope estimated using a specific model-free extrapolation procedure is minimized. An empirical study is used to identify optimal designs in the sense that the variance of the estimate of the low-dose slope estimated using the one-hit model is minimized.