Dose Individualization for Phase I Cancer Trials With Broadened Eligibility.

Area of Science:

• Clinical Trials
• Oncology
• Biostatistics

Background:

• Broadening eligibility criteria in cancer trials improves generalizability and recruitment.
• However, expanded criteria necessitate careful design for patient safety and dosing efficiency.
• Existing dose-finding methods often assume known patient heterogeneity, which is insufficient for expanded eligibility.

Purpose of the Study:

• To propose a novel dose-finding design for phase I cancer trials with broadened eligibility and unknown patient heterogeneity.
• To simultaneously identify differentiating patient criteria and recommend subpopulation-specific maximum tolerated doses (MTDs).
• To address the limitations of assuming patient homogeneity in dose-finding.

Main Methods:

• Developed a two-in-one dose-finding design integrating criterion selection and MTD recommendation.
• Employed stochastic search variable selection to identify patient subgroups with distinct toxicity profiles.
• Conducted simulation studies comparing the proposed design against a naive homogeneity assumption approach.

Main Results:

• The proposed design demonstrated favorable operating characteristics across various scenarios.
• It allocated patients more effectively to their true MTDs during trials.
• The design successfully identified differentiating criteria and recommended multiple MTDs when necessary.

Conclusions:

• Assuming patient homogeneity in dose-finding can lead to unsafe or sub-therapeutic dose recommendations.
• The proposed design effectively handles unknown patient heterogeneity in expanded eligibility cancer trials.
• This approach enhances early-stage trial variability, leading to more precise and safer MTD recommendations.