An adaptive dose-finding method using a change-point model for molecularly targeted agents in phase I trials

Area of Science:

• Oncology
• Clinical Pharmacology
• Biostatistics

Background:

• Oncology drug development is shifting towards molecularly targeted agents (MTAs).
• Unlike cytotoxic agents, MTAs can exhibit non-monotonic dose-efficacy relationships.
• Existing adaptive dose-finding methods often require additional parameters to model this non-monotonicity.

Purpose of the Study:

• To propose a novel adaptive dose-finding approach for MTAs in phase I trials.
• To address non-monotonic dose-efficacy and dose-toxicity relationships.
• To utilize binary efficacy and toxicity outcomes for monotherapy.

Main Methods:

• Developed a dose-efficacy model with parameters that adapt near a change point.
• Identified the change point by maximizing the log-likelihood of dose-efficacy and dose-toxicity models.
• Employed a weighted Mahalanobis distance algorithm based on posterior probabilities of outcomes.

Main Results:

• The proposed method effectively models non-monotonic dose-efficacy patterns.
• Simulation studies demonstrated competitive operating characteristics compared to existing methods.
• The approach showed sensitivity across various simulated scenarios.

Conclusions:

• The novel adaptive dose-finding approach accurately captures non-monotonic relationships for MTAs.
• This method offers a robust alternative for optimizing dose selection in early-phase oncology trials.
• It enhances the development of targeted cancer therapies by better understanding dose-response.