2026 Volume 31 Pages 28-41
Objectives Designs of phase I oncology trials are generally classified into 3 classes: rule-based, model-assisted, and model-based designs. 3+3 design, Bayesian Optimal Interval design, and Bayesian Logistic Regression Model design are representative designs from each class. In Japan, Pharmaceuticals and Medical Devices Agency has released a “Check List for 30-day-Clinical Trial Notification Review on an Initial Clinical Trial Notification (Oncology Drugs)” for 30-day-CTN Review of oncology drugs and an “Early Consideration” specifying points to consider particularly with a focus on safety when evaluating operating characteristics of a dose-escalation trial based on statistical considerations. The performance of the adopted design should be evaluated by simulation according to these notifications. Corresponding R packages and software such as Trial Design are useful, but study-specific changes, such as application of single patient acceleration (SPA) at low doses, may also be considered. In such cases, tools to apply such modifications more efficiently are desirable considering development timelines and technical complexity. Therefore, we develop an application tool using Shiny to evaluate the operating characteristics of each class of designs incorporating SPA.
Conclusion The developed tool is capable of calculating evaluation metrics required in the Early Consideration, enabling efficient and effective consideration of designs and negotiations with regulatory authorities. The R script download function is employed, ensuring reproducibility of results. The tool is not intended to have all of the expected functionality. Users are advised to download the R script and make additional modifications (e.g., data generation, discontinuation criteria, and so on) as needed when study-specific changes are to be incorporated.
