Precision generalized phase I-II designs.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Pharmacometrics

Background:

• Traditional clinical trial designs often overlook patient heterogeneity.
• Optimizing drug dosage requires balancing efficacy, toxicity, and long-term patient outcomes.

Purpose of the Study:

• To introduce a novel precision Bayesian dose optimization design, PGen I-II.
• To refine existing dose-finding methodologies by incorporating patient subgroup analysis.

Main Methods:

• The PGen I-II design utilizes early efficacy and toxicity data for adaptive dose adjustments.
• It models long-term treatment failure using a piecewise exponential distribution with subgroup-specific effects.
• Latent variables are employed for adaptive clustering of subgroups with similar dose-outcome profiles.

Main Results:

• The PGen I-II design demonstrated superiority over designs assuming patient homogeneity or separate subgroup trials in simulations.
• It enables subgroup-specific decisions for dose escalation, de-escalation, or randomization.
• The model allows for borrowing strength between similar subgroups to enhance statistical power.

Conclusions:

• The PGen I-II design offers a more precise and adaptive approach to dose optimization in clinical trials.
• Accounting for patient heterogeneity through subgroup analysis leads to improved treatment success.
• User-friendly software and guidelines facilitate the implementation of this advanced design.