Defining Utility as a Measure of Preference Under Uncertainty in Phase I-II Oncology Dose Finding Trials

Area of Science:

• Clinical Trials
• Biostatistics
• Decision Theory

Background:

• Dose finding trials are crucial in oncology for identifying optimal drug doses.
• Current methods may not fully capture complex clinical decision-making regarding efficacy and toxicity.
• Bayesian approaches offer a flexible framework for adaptive trial designs.

Purpose of the Study:

• To propose a novel Bayesian decision-theoretic framework for dose finding trials.
• To develop a utility function that incorporates both efficacy and toxicity, reflecting clinical judgment.
• To introduce the Reference Dependent Decision Theoretic dose finding (R2DT) method.

Main Methods:

• Application of a Bayesian decision-theoretic approach.
• Development of a new utility function grounded in von Neumann-Morgenstern (VNM) utility theory.
• Incorporation of reference points to model varying risk attitudes for gains and losses.

Main Results:

• The R2DT framework demonstrates good operating characteristics in simulations.
• R2DT shows improved performance in identifying optimal doses near efficacy and toxicity thresholds.
• A flexible utility function enhances the probability of finding the optimal dose.

Conclusions:

• The R2DT method provides a proof-of-concept for a more clinically relevant dose finding approach.
• This framework can lead to improved trial outcomes by better aligning with clinician beliefs.
• Further evaluation of R2DT in diverse clinical settings is warranted.