Flexible Bayesian methods for cancer phase I clinical trials. Dose escalation with overdose control

Area of Science:

• Clinical Trials
• Biostatistics
• Pharmacology

Background:

• Phase I clinical trials are crucial for determining safe cancer drug dosages.
• Modeling the dose-response relationship, specifically the maximum tolerated dose (MTD), is essential for patient safety.
• Prior distributions significantly influence the estimation of MTD and the overall safety of dose-escalation studies.

Purpose of the Study:

• To investigate the impact of different prior distributions on the safety and efficiency of dose-toxicity modeling in cancer phase I trials.
• To evaluate a joint prior distribution with a negative correlation structure for key dose-toxicity parameters.
• To compare the performance of this novel prior against independent priors in simulations.

Main Methods:

• Utilized a dose-toxicity model parameterized by the maximum tolerated dose (MTD) and the probability of dose-limiting toxicity (DLT) at the initial dose.
• Employed the Escalation with Overdose Control (EWOC) method for MTD estimation.
• Conducted simulation studies to assess trial safety and MTD estimation efficiency under various prior specifications.

Main Results:

• A joint prior distribution with a negative a priori correlation between the MTD and initial dose toxicity probability demonstrated improved trial safety.
• The use of independent priors for these parameters resulted in less safe trial designs.
• The efficiency of MTD estimation remained largely unchanged when using the negatively correlated joint prior.

Conclusions:

• A negatively correlated joint prior distribution offers a superior strategy for modeling dose-toxicity relationships in early-phase cancer trials.
• This approach enhances patient safety during dose escalation without sacrificing the precision of MTD estimation.
• The findings advocate for the adoption of informed prior structures in clinical trial design to optimize safety and efficacy.