Sample size planning for phase II trials based on success probabilities for phase III

Area of Science:

• Clinical trial design
• Pharmaceutical development
• Biostatistics

Background:

• High phase III clinical trial failure rates are a significant concern.
• Overly optimistic assumptions in phase III planning, stemming from limited phase II data, contribute to these failures.
• Realistic assessment of success probabilities is crucial for effective drug development.

Purpose of the Study:

• To develop an approach for planning phase II trials in time-to-event settings that integrates the entire phase II/III development program.
• To derive optimal phase II stopping boundaries that minimize events while ensuring conditional probabilities for go/no-go decisions and phase III success.
• To provide recommendations for phase II sample size selection.

Main Methods:

• The study proposes a method for deriving phase II stopping boundaries based on conditional probabilities of correct go/no-go decisions and phase III success.
• Simulations were conducted to evaluate the impact of the number of events in phase II on unconditional probabilities.
• General recommendations for phase II sample size were developed based on simulation outcomes.

Main Results:

• The number of events observed in phase II influences unconditional go/no-go decision probabilities and phase III success probabilities.
• Simulations indicate that exceeding 150 events in phase II has a diminishing impact on these probabilities.
• The study highlights the importance of considering factors like the number of compounds and available resources for phase III.

Conclusions:

• Optimizing phase II planning with realistic success probabilities can mitigate phase III failures.
• A phase II sample size of around 150 events appears sufficient to achieve significant improvements.
• Phase II investment should be scaled based on the number of compounds and phase III resource constraints.