|
Abstract:
|
In immuno-oncology, many recent survival trials are planned in clinical settings where non-proportional hazards (nPH) are expected. Traditional designs with a survival endpoint were mostly developed under the proportional hazards (PH) assumption. Those methods may not be appropriate in the nPH setting, likely leading to suboptimal designs, particularly, with the use of group sequential procedures (GSP). Currently, such trial designs and the evaluation of their operating characteristics (OCs) are conducted using simulations due to the lack of closed-form solutions. In practice, it is didactic to obtain a good analytical understanding of the design performance through the explicit functions of design parameters, particularly, in the nPH setting. There are several design parameters that influence trial performance more in the nPH setting than in the PH setting. Closed-form solutions for the nPH design facilitates the evaluation needs and informs optimal design selection. We develop and provide such an analytical solution for trial designs in the nPH setting when the GSP is used. The evaluation of trial design and its OCs are illustrated using both analytical and simulation approaches.
|
