|
Abstract:
|
Oncology dose-finding designs currently being deployed use dose limiting toxicity (DLT) events in the first cycle to make dose escalation decisions. This leads to huge loss of information on other relevant adverse event data, and the grades and types of the adverse events, thus resulting in Maximum Tolerated Dose (MTD) estimation with large bias. To avoid this, Yin et al.(2016) proposed the Bayesian Repeated Measures Design to model a semi-continuous endpoint that incorporates toxicity types and grades from multiple cycles. However, this design follows a decision rule of minimizing point-estimate based loss function, which can be less reliable due to small sample sizes. To address this concern, we proposed a design with interval based decision rule to select the dose with the highest posterior probability of a pre-specified target toxicity interval. Through simulation, we compared this new design with the design in Yin et al (2016), as well as popular designs such as Continual Reassessment Method. The results demonstrated that our design outperforms all other designs in terms of accurately identifying the target dose and assigning more patient to the therapeutic dose levels.
|
