Title: A Bayesian dose-finding design in oncology using pharmacokinetic/pharmacodynamic modeling
Authors: Xiao Su - The University of Texas MD Anderson Cancer Center (United States)
Yisheng Li - The University of Texas MD Anderson Cancer Center (United States) [presenting]
Peter Mueller - UT Austin (United States)
Kim-Anh Do - University of Texas MD Anderson Cancer Center (United States)
Abstract: Most oncology phase I dose-finding trial designs are either algorithmic or empirical model based, with the goal of identifying the maximum tolerated dose (MTD). Efficiency may be lost in these designs due to failure to exploit the drug concentration-time profile that can be modeled based on available pharmacokinetic (PK) data. To extend the existing designs that make use of low-dimensional summaries of the PK profiles and/or PD endpoints, we propose a Bayesian framework to model the dose-concentration-pharmacologic effect-toxicity (D-C-E-C) relationship, by using PK/PD modeling and modeling of the relationship between a latent PD outcome (dynamic pharmacologic effect) and a binary toxicity endpoint. An important feature of the proposed framework is that it allows explicit modeling of the effects of treatment schedule and method of administration (e.g., drug formulation and route of administration), which are difficult to model under the existing design framework. We compare the performance of the proposed designs with common phase I designs and a nonparametric benchmark design via simulation studies. We illustrate the proposed design by applying it to a phase I trial of a $\gamma$-secretase inhibitor in metastatic or locally advanced solid tumors conducted at MD Anderson Cancer Center.