A stochastically curtailed single-arm phase II trial design for binary outcomes

Area of Science:

• Clinical Trials
• Biostatistics
• Oncology Drug Development

Background:

• Phase II clinical trials are crucial for drug development, but rising costs necessitate more efficient designs.
• Current phase II trials for cancer treatments often have binary outcomes, with decisions to continue or cease development based on hypothesis tests.
• Existing early stopping rules can reduce sample size but may miss opportunities for earlier intervention.

Purpose of the Study:

• To propose novel phase II clinical trial designs for binary outcomes using stochastic curtailment.
• To address limitations of existing early stopping methods by allowing cessation when results are highly likely, not just certain.
• To enhance the efficiency and reduce the sample size of phase II cancer trials.

Main Methods:

• Development of new design approaches for single-arm, phase II binary outcome trials.
• Utilizing exact distributions to avoid simulation-based estimations.
• Considering a broad spectrum of potential trial designs.
• Incorporating stochastic curtailment for early stopping of promising treatments.

Main Results:

• The proposed designs enable early stopping for promising treatments, not just definitive outcomes.
• The use of exact distributions ensures accuracy without simulation.
• A wider range of designs were evaluated, offering more flexibility.
• The new approaches result in considerably reduced average sample sizes.

Conclusions:

• Novel phase II trial designs incorporating stochastic curtailment offer significant improvements in efficiency.
• These designs allow for earlier decision-making, potentially accelerating cancer drug development.
• The methods provide reduced sample sizes on average, optimizing resource allocation.