Optimal adaptive two-stage designs for phase II cancer clinical trials

Area of Science:

• Oncology
• Clinical Trial Design
• Biostatistics

Background:

• Single-arm, two-stage designs are standard for phase II oncology trials, particularly for cytotoxic therapies.
• Simon's optimal design is popular for minimizing sample size under the null hypothesis but uses computationally intensive search algorithms.
• Current search methods often restrict parameters, potentially compromising design optimality.

Purpose of the Study:

• To extend fixed sample size phase II designs by incorporating adaptive stage two sample sizes based on early results.
• To develop a more efficient numerical algorithm for an exhaustive search of optimal phase II designs.
• To compare the performance of proposed adaptive designs against existing literature designs.

Main Methods:

• Developed an adaptive phase II design where the second stage sample size is contingent on the number of responses in the first stage.
• Implemented a novel, efficient numerical algorithm to facilitate a comprehensive search for optimal designs.
• Conducted comparative analyses between the new adaptive designs and traditional fixed sample size designs.

Main Results:

• The proposed adaptive designs, identified through an exhaustive search, offer improvements over existing phase II designs.
• While overall sample size reductions are often moderate, significant decreases are achievable for specific parameter settings.
• The new search strategy enhances the efficiency of identifying optimal adaptive designs.

Conclusions:

• Adaptive phase II designs can lead to more efficient sample size utilization in oncology drug development.
• The developed numerical algorithm enables a more thorough exploration of the design space, potentially uncovering superior designs.
• This work provides a valuable alternative for optimizing phase II clinical trial designs in cancer research.