|
Abstract:
|
With the recent advancement in oncology immunotherapy, the next stage of early oncology development is to focus on identifying the best combinations of established immunotherapies with new agents. Although the combination therapy is the primary interest, safety profile of the new agent alone must also be explored. As a result, many trials were designed to have both monotherapy and combination therapy arms. Finding the MTD for the new agent in both arms is critical. Current oncology dose-finding methods, e.g., 3+3, mTPI, CRM, and BOIN, as well as the MTD estimation algorithm, pool-adjacent-violators algorithm (PAVA), do not handle the correlation and interplay between the two arms, resulting in disregarding useful information for dose-finding decisions and MTD estimation. Also, due to relatively small sizes of phase 1 trials, the selected MTDs may contradict with each other when the dose-finding was conducted independently for each arm. To overcome these issues, we applied a two-dimensional PAVA (2D-PAVA) to MTD estimation, and modified the standard BOIN to allow for information flow between arms during dose-finding. These new approaches were assessed with simulation.
|
