Easy-to-implement Bayesian methods for dose-escalation studies in healthy volunteers.

Area of Science:

• Pharmacology
• Clinical Trials
• Biostatistics

Background:

• Phase I clinical trials are crucial for assessing drug safety and understanding dose-response relationships in healthy volunteers.
• Current methods for dose selection may not fully optimize learning or adequately mitigate risks associated with escalating doses.

Purpose of the Study:

• To develop a Bayesian decision procedure for selecting optimal doses in Phase I clinical trials.
• To balance the need for learning about the dose-concentration relationship with the imperative to minimize the risk of high-dose toxicity.

Main Methods:

• A Bayesian decision procedure was developed, incorporating prior information and observed volunteer responses.
• Pharmacokinetic data, after logarithmic transformation, were modeled using linear mixed models.
• Prior information was integrated as 'pseudo-data', and posterior distributions were maximized for dose selection.

Main Results:

• The proposed procedure effectively identifies optimal doses for Phase I trials, balancing learning and safety.
• The method is implementable using standard linear mixed model software.
• Illustrative examples using real and simulated data demonstrate the procedure's utility and advantages over existing methods.

Conclusions:

• The developed Bayesian decision procedure offers an improved approach to dose selection in Phase I clinical trials.
• This method enhances the efficiency of learning about drug pharmacokinetics while maintaining a strong focus on participant safety.
• The procedure's integration with standard statistical software facilitates its adoption in clinical research.