Dose-finding based on bivariate efficacy-toxicity outcome using Archimedean Copula.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacometrics

Background:

• Clinical studies frequently involve multiple primary endpoints, such as efficacy and toxicity.
• Simultaneous evaluation of efficacy and toxicity is crucial for determining optimal drug dosage.
• Existing methods may not adequately handle correlated efficacy-toxicity outcomes, especially with mixed data types.

Purpose of the Study:

• To propose a joint statistical model for correlated efficacy and toxicity outcomes using Archimedean Copula.
• To extend the continual reassessment method (CRM) for bivariate trial designs.
• To develop a robust method for mixed continuous and discrete correlated outcomes in dose-finding studies.

Main Methods:

• Development of a joint statistical model incorporating Archimedean Copula to capture dependencies between efficacy and toxicity.
• Extension of the continual reassessment method (CRM) to a bivariate framework for simultaneous endpoint assessment.
• Adaptation of the joint model to accommodate mixed continuous and discrete outcome data.

Main Results:

• The proposed Archimedean Copula-based joint model effectively handles correlated efficacy-toxicity data.
• The extended bivariate CRM demonstrates excellent operating characteristics in simulations.
• The methodology provides a robust approach for mixed outcome types in dose-finding trials.

Conclusions:

• The novel bivariate trial design using Archimedean Copula offers an effective approach for joint efficacy-toxicity analysis.
• This method optimizes dose selection for Phase III by considering both efficacy and toxicity simultaneously.
• The proposed statistical framework is suitable for complex clinical trial scenarios with mixed outcome data.