Spatial Two-stage Designs for Phase II Clinical Trials.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Oncology Research

Background:

• Single-arm phase II studies commonly use tumor response as a binary endpoint.
• Simon's two-stage minimax and optimal designs are standard but present trade-offs in sample size.
• The minimax design minimizes the maximum sample size, while the optimal design minimizes the expected sample size under the null hypothesis, often leading to large differences in total sample size.

Purpose of the Study:

• To develop novel phase II clinical trial designs that balance sample size criteria.
• To create designs that compromise between minimizing maximal and expected sample sizes.
• To offer more appealing alternatives to existing Simon's designs.

Main Methods:

• Development of novel spatial designs utilizing information on first-stage and total required sample sizes.
• Analysis of the properties of these proposed spatial designs.
• Comparison of spatial designs against Simon's minimax and optimal designs, and an extension by Lin and Shih.

Main Results:

• The proposed spatial designs offer a compromise between the optimality criteria of existing designs.
• These novel designs avoid the large sample size discrepancies seen in Simon's designs.
• Demonstrated advantages of the spatial designs over Simon's designs and its extension.

Conclusions:

• Novel spatial designs provide a balanced approach to sample size determination in phase II clinical trials.
• These designs are more appealing to investigators by mitigating the trade-offs inherent in traditional methods.
• Applications in Hodgkin disease and head and neck cancer studies demonstrate the practical utility of spatial designs.