|
Abstract:
|
In complex diseases, such as oncology and cystic fibrosis, it has become apparent that effective treatment requires a combination of agents. While dose selection via modeling approaches for a single agent is well-studied, a framework to achieve the same for a combination of drugs is yet to be established. The task can be further complicated in studies with rare diseases due to limited sample size and prior information. A well-understood response surface (RS) can facilitate making drug combination dosing decisions moving forward into confirmatory trials. In this research, we bridge the gap between RS modeling methodologies applied in early stage clinical trial settings and the practical aspects of dose selection in phase 2 trials. We design simulation studies under various factors including different choices for the true underlying RS, sample size, and adaptive features. We identify existing and propose novel RS models based on both pharmacological and statistical rationale. The performances of RS models are evaluated using metrics such as the proximity of estimated minimal effective doses to the truth.
|
