Time-to-event model-assisted designs for dose-finding trials with delayed toxicity.

Area of Science:

• Clinical Trials
• Drug Development
• Biostatistics

Background:

• Accelerating drug development requires increasing patient accrual and using adaptive designs.
• Challenges arise when outcome evaluation lags behind patient accrual or when toxicity is late-onset, hindering timely adaptive decisions.

Purpose of the Study:

• To propose a general methodology for model-assisted designs to manage pending toxicity data in phase I dose-finding trials.
• To facilitate seamless decision-making despite delays in observing trial outcomes.

Main Methods:

• Developed a novel formulation and approximation of the likelihood of observed data.
• Proposed time-to-event model-assisted designs that evaluate each dose separately.
• Dose-escalation/de-escalation rules are pre-tabulated for simplified trial conduct.

Main Results:

• The proposed designs effectively handle pending toxicity data due to fast accrual or late-onset toxicity.
• Demonstrated desirable finite and large-sample properties for the new designs.
• Achieved performance comparable to more complex model-based designs.

Conclusions:

• The proposed model-assisted designs offer a practical and efficient approach for phase I dose-finding trials.
• These designs simplify trial conduct and improve decision-making processes.
• User-friendly software is available for implementing these novel designs.