All Times EDT
Keywords: Bayesian, Two-stage design, dose expansion, oncology, early phase
Early clinical trials are relatively small and short duration and may not be suited for typical hypothesis tests to detect a clinically meaningful treatment effect. An alternative approach is to use decision rule-based designs to assess early signals of activity in a clinical candidate in the earlier phase of the development In this presentation, Bayesian rules will be considered, and objective response rate will be used for assessment of effectiveness. In a typical dose expansion design of early oncology studies routinely two-stage designs (Fleming’s or Simon’s) are utilized. These designs may have some limitations for early small studies due to their fixed nature and inflexibility in decision rules. In this talk, tool will be proposed that will convert a standard two-stage design (Fleming’s or Simon’s) into an equivalent Bayesian go/no-go based design. The parameters (such as response rates of the current drug and the control, power, alpha) of the two-stage designs will be used as inputs and the corresponding Bayesian Go/No-Go decision rules can be calculated. Additionally, a visual method to describe the rules including priors, posteriors and other quantities will be provided. Visual depictions and interpretations of the rules will be discussed along with the expectations and risks tolerances. Implications of sample size and the choices of various Bayesian decision rules will also be considered associated with early oncology dose expansion designs.