The quasi-CRM shift method for partially ordered groups.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Pharmacology

Background:

• Dose-finding studies are crucial for determining optimal and safe drug dosages.
• Existing methods for dose-finding in clinical trials often do not account for pre-defined group ordering or utilize ordinal toxicity data effectively.
• A gap exists in methods that integrate group-specific dose-finding with ordinal toxicity assessment.

Purpose of the Study:

• To propose a novel phase-I clinical trial design for identifying group-specific doses using ordinal toxicity.
• To address the limitations of existing dose-finding methods by incorporating group ordering and ordinal toxicity.
• To develop a method that avoids dose reversals and enhances patient allocation to optimal doses.

Main Methods:

• The proposed method employs ordinal toxicity data to guide dose escalation within distinct patient groups.
• It specifically handles scenarios where groups are partially or completely ordered based on prior knowledge.
• Simulations were conducted to compare the proposed design against methods that either ignore group information or ordinal toxicity.

Main Results:

• The proposed design successfully avoids dose reversals, a common issue in methods using only ordinal toxicity.
• It demonstrated a higher frequency of allocating patients to optimal doses throughout the trial.
• Simulations indicated superior performance in selecting optimal doses by the trial's conclusion.

Conclusions:

• The novel phase-I clinical trial design effectively integrates group-specific dose-finding with ordinal toxicity.
• This approach offers a significant improvement over existing methods by preventing dose reversals and optimizing patient allocation.
• The proposed method provides a robust framework for dose-finding in trials with ordered patient groups and toxicity data.