An R package for simulation experiments evaluating clinical trial designs

Area of Science:

• Biostatistics and Clinical Trial Design
• Computational Biology and Bioinformatics

Background:

• Evaluating clinical trial (CT) designs is crucial for drug development.
• Existing simulation tools may lack flexibility or user-friendliness for biostatisticians.

Purpose of the Study:

• To present an open-source, extensible R application for simulating and evaluating competing clinical trial designs.
• To demonstrate the application's utility by assessing the impact of pharmacokinetic heterogeneity on a Phase I "3+3" design.

Main Methods:

• Developed an object-oriented application using R, featuring four core classes (population, CT design, outcome, evaluation) and five key methods.
• The application utilizes clear interfaces for extensibility, allowing integration of new subclasses and instances.
• Simulated patient data and trial outcomes to evaluate design performance.

Main Results:

• The application successfully simulated clinical trial scenarios, demonstrating flexibility and extensibility.
• Evaluated the performance of the "3+3" Phase I design under varying degrees of patient pharmacokinetic heterogeneity.
• Quantified the impact of heterogeneity on the "3+3" design's ability to identify safe and effective doses.

Conclusions:

• The open-source R application provides a powerful and adaptable tool for clinical trial design evaluation.
• Biostatisticians can leverage this tool to optimize trial designs by considering patient-specific factors like pharmacokinetic variability.
• Simulation-based evaluation enhances the efficiency and reliability of clinical trial design.