Continuous endpoints in Bayesian two-stage designs.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Oncology

Background:

• Phase II clinical trials often use binary endpoints derived from continuous measures, leading to information loss.
• Traditional designs may not fully leverage the available data for treatment efficacy assessment.

Purpose of the Study:

• To propose an improved two-stage clinical trial design for phase II studies.
• To directly utilize continuous endpoints, avoiding information loss associated with dichotomization.
• To apply a Bayesian predictive approach for enhanced analysis.

Main Methods:

• Development of a single-arm, two-stage clinical trial design.
• Incorporation of a Bayesian predictive framework.
• Direct use of continuous efficacy endpoints, such as tumor shrinkage.

Main Results:

• The proposed Bayesian design effectively utilizes continuous data, preserving information.
• Numerical results demonstrate the design's applicability in phase II cancer trials.
• The approach offers a more sensitive assessment of treatment effects compared to binary endpoints.

Conclusions:

• The Bayesian predictive two-stage design offers a statistically robust alternative for phase II trials.
• This method enhances the evaluation of experimental treatments by preserving data integrity.
• Recommended for oncology trials assessing continuous efficacy measures like tumor response.