Statistical tests for two-stage adaptive seamless design using short- and long-term binary outcomes

Area of Science:

• Clinical Trials
• Biostatistics
• Pharmaceutical Development

Background:

• Adaptive seamless designs integrate Phase II and III trials for efficient drug development.
• These designs often involve two stages with distinct objectives: treatment selection and efficacy comparison.
• Current methods may lack optimal power or conservativeness in specific scenarios.

Purpose of the Study:

• To propose an exact conditional test for two-stage adaptive seamless designs.
• To evaluate the performance of this test using simulation studies.
• To compare the proposed method with existing combination test approaches.

Main Methods:

• Focus on a two-stage design with short-term binary endpoint for selection and long-term binary endpoint for comparison.
• Utilize a bivariate binomial distribution for an exact conditional test based on interim analysis.
• Incorporate the mid-p approach for enhanced test conservativeness.

Main Results:

• The proposed exact method effectively controls the type I error rate at the nominal level.
• Performance is consistent across varying numbers of initial treatments and endpoint correlations.
• The proposed methods demonstrate superior treatment comparison power compared to combination tests when one treatment is effective.

Conclusions:

• The proposed exact conditional test is a robust and powerful tool for two-stage adaptive seamless trials.
• This method offers improved statistical efficiency in drug development.
• It provides a reliable approach for selecting and comparing treatments based on distinct endpoints.