A Generalized Phase I/II Dose Optimization Trial Design With Multi-Categorical and Multi-Graded Outcomes

Area of Science:

• Clinical Trial Design
• Biostatistics
• Pharmacometrics

Background:

• Advances in clinical trial design are driven by accurate observations and rational assumptions.
• Recent trials increasingly collect multi-graded outcomes for enhanced analytical insights.
• Non-monotonic dose-efficacy relationships are more realistic than traditional assumptions.

Purpose of the Study:

• To propose a novel phase I/II clinical trial design.
• To simultaneously handle multi-categorical toxicity and efficacy with multi-graded outcomes.
• To accurately identify the optimal biologic dose (OBD) under non-monotonic dose-efficacy relationships.

Main Methods:

• The design integrates multi-categorical toxicity and efficacy data using quasi-continuous probabilities.
• It employs weight matrices for clinical significance and toxicity probability intervals to screen doses.
• Adaptive dose escalation/de-escalation decisions are made based on posterior distributions of dose desirability (utility).

Main Results:

• Numerical simulations demonstrate robust safety, accuracy, and reliability across multiple scenarios.
• Benchmarking against existing designs shows superior overall performance.
• The proposed design effectively identifies the OBD in non-monotonic dose-efficacy settings.

Conclusions:

• The proposed phase I/II design offers a significant advancement in clinical trial methodology.
• It provides a more realistic and accurate approach to dose selection by accommodating complex dose-response relationships.
• The design demonstrates superior performance and robustness compared to current methods.