Phase II multi-step planning methods in oncology: comparison, recommendations and potential applications

Area of Science:

• Clinical Trials Methodology
• Oncology Research
• Biostatistics

Background:

• Phase II clinical trials in oncology assess treatment response rates for phase III comparison.
• Early trial termination methods like Fleming's procedure and triangular test aim to optimize resource allocation.
• Statistical properties of these early termination methods require careful evaluation.

Purpose of the Study:

• To compare the statistical properties of Fleming's procedure and the triangular test for early phase II oncology trial termination.
• To evaluate performance based on type II error, sample size requirements, and conclusion probabilities at different trial steps.
• To assess the reliability of these methods across various minimum response rates.

Main Methods:

• Comparative analysis of Fleming's procedure and triangular test using a fixed number of steps (4 or 5).
• Evaluation of statistical properties including overall and observed type II error rates.
• Assessment of average and maximum number of subjects required to reach a conclusion.

Main Results:

• Fleming's procedure demonstrated consistent statistical power across low to high response rates.
• The triangular test showed significant deviations in actual power compared to theoretical power at very low (<20%) and very high (>80%) minimum response rates.
• For intermediate response rates, the triangular test required fewer average subjects but potentially more maximum subjects, with a higher probability of early conclusion.

Conclusions:

• Fleming's procedure is recommended for its reliable performance across diverse response rates in phase II oncology trials.
• The triangular test is not recommended in situations with extreme minimum response rates due to unreliable power.
• Multi-step methods are valuable for cytotoxic and cytostatic drug development, particularly when response rate or surrogate endpoints are used.