Sample size calculation for small sample single-arm trials for time-to-event data: Logrank test with normal...

Area of Science:

• Clinical Trials
• Biostatistics
• Oncology Research

Background:

• Sample size calculations are crucial for clinical trial design.
• Standard software for time-to-event endpoints in single-arm trials has limited options, often relying on asymptotic normality.
• Existing methods like the log-rank test and exponential distribution-based tests perform well for moderate-to-large sample sizes.

Purpose of the Study:

• To evaluate an exact parametric test for sample size calculations in single-arm trials.
• To address the challenges of small-to-moderate sample sizes common in oncology due to cost and slow accrual.
• To compare analytic results with Weibull distributed survival times using simulations.

Main Methods:

• Utilized an exact parametric test with a chi-square distribution for sample size calculations.
• Compared this method with standard approaches for time-to-event endpoints.
• Employed simulations to verify the analytic results for sample size calculations.

Main Results:

• Simulations indicate practical benefits of the exact test for small sample Phase II studies.
• The exact test can positively impact study feasibility, timeliness, financial support, and clinical novelty.
• It is a viable option for small-to-moderate sample trials with adequate accrual and follow-up.

Conclusions:

• The exact parametric test is a valuable tool for sample size calculations in small-to-moderate sample single-arm trials.
• Statisticians should assess the sensitivity of calculations across different methods before recommending a final sample size.
• This approach can enhance the practicality and efficiency of oncology clinical trials.