Phase I/II Design for Selecting Subgroup-Specific Optimal Biological Doses for Prespecified Subgroups.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Pharmacology

Background:

• Determining the optimal biological dose (OBD) is crucial for drug development.
• Patient populations can exhibit heterogeneity, leading to subgroup-specific dose-response relationships.
• Existing dose-finding designs may not optimally handle uncertainty in subgroup effects.

Purpose of the Study:

• To propose a novel phase I/II trial design for optimal biological dose (OBD) selection.
• To accommodate scenarios where the OBD may differ across two prespecified patient subgroups.
• To provide a flexible design that adapts to emerging trial data regarding subgroup effects.

Main Methods:

• A utility function was employed to balance efficacy and toxicity trade-offs.
• A Bayesian model with spike and slab priors was utilized to estimate subgroup effects on toxicity and efficacy.
• Simulation studies were conducted to evaluate the proposed design's performance.

Main Results:

• The proposed design demonstrated strong performance, comparable to specialized designs under different assumptions.
• It performed nearly as well as ignoring subgroups when relationships were similar across groups.
• It performed nearly as well as independent subgroup dose-finding when relationships differed.

Conclusions:

• The adaptive design effectively identifies optimal biological doses (OBDs) even with uncertainty about subgroup differences.
• This approach offers a robust strategy for dose-finding in heterogeneous patient populations.
• The design provides a unified framework for selecting common or subgroup-specific OBDs.