A method for therapeutic dose selection in a phase II clinical trial using contrast statistics

Area of Science:

• Biostatistics
• Clinical Pharmacology
• Drug Development

Background:

• Phase II clinical trials are crucial for dose-finding.
• Statistical methods are needed to reliably determine therapeutic doses.
• Prior dose-response data can inform confirmatory trial dose selection.

Purpose of the Study:

• To propose a novel statistical method for therapeutic dose determination.
• To utilize phase II clinical trial data (3-4 doses) for dose selection.
• To establish a robust method for selecting dose-response patterns and therapeutic doses.

Main Methods:

• Assumes normal distribution of primary variable with common variance.
• Employs contrast statistics (linear, plateau) to identify response patterns.
• Uses posterior probability and a cut-off value to validate selected patterns.
• Simulation study with a specific loss function to determine optimal cut-off values (0.75 for 3-group, 0.70 for 4-group trials).

Main Results:

• The proposed method successfully identified appropriate response patterns and therapeutic doses in a leukotriene receptor antagonist trial.
• Simulation results provided specific cut-off values for 3- and 4-group trials to minimize expected loss.
• The method demonstrated effectiveness in selecting therapeutic doses based on estimated dose-response patterns.

Conclusions:

• The developed statistical method offers a reasonable approach for therapeutic dose determination in clinical trials.
• The method integrates phase II data to enhance the precision of dose selection in phase III trials.
• Application to a real-world case study validates the method's practical utility and effectiveness.