Adjustment for Inconsistency in Adaptive Phase 2/3 Designs With Dose Optimization

Area of Science:

• Clinical Trials Methodology
• Oncology Drug Development
• Biostatistics

Background:

• Adaptive Phase 2/3 designs are crucial for oncology drug development when Phase 1 data is limited.
• Identifying optimal doses is challenging due to insufficient early-phase data.
• Inconsistencies between Phase 2 and Phase 3 results pose regulatory and practical hurdles.

Purpose of the Study:

• To develop statistical methods for adaptive Phase 2/3 designs that explicitly address dose selection inconsistencies.
• To balance regulatory caution, sponsor interests, and practical considerations in adaptive trial designs.
• To account for the probability of "picking-the-winner" in dose optimization.

Main Methods:

• Incorporation of inconsistency concerns into statistical analysis.
• Specification of an inconsistency cutoff.
• Utilization of three hypothesis testing strategies: conservative, aggressive, and neutral.
• Accounting for the probability of "picking-the-winner".

Main Results:

• The proposed methods provide a framework for managing dose selection imperfections in adaptive Phase 2/3 designs.
• Demonstrated a way to balance competing interests in adaptive trial design.
• Highlighted the importance of addressing potential inconsistencies proactively.

Conclusions:

• The statistical strategies enhance the reliability of adaptive Phase 2/3 designs in oncology.
• This approach offers a more robust method for dose optimization in drug development.
• Further research is warranted in this underexplored area of adaptive trial design.