Abstract
In Phase I oncology trials, the primary goal is to assess dose limiting toxicities (DLT) and estimate the maximum tolerated dose (MTD). The classical 3+3 design is still used in the vast majority of studies. In this chapter, we review the 3+3 design and the new Bayesian Optimal Interval (BOIN) design. BOIN is easy to implement, similar to the 3+3, using a simple table to guide dose escalation/deescalation. As opposed to the 3+3 design, BOIN can target a DLT rate well above or below the usual 25-33% target. We explain how computer simulations can be used to evaluate phase I designs and present results comparing the designs under a large number of true dose-toxicity scenarios. We show that BOIN has better performance than 3+3. BOIN selects the true MTD at a much higher rate and treats a higher percentage of patients at the MTD. BOIN allocates fewer patients to low toxicity doses. Unlike older Bayesian designs (e.g., modified continual reassessment method), BOIN does not require a statistician to be available during the trial. Readily-available, free software makes BOIN simple to implement. We recommend the use of BOIN over the 3+3 design.
Original language | English (US) |
---|---|
Title of host publication | Phase I Oncology Drug Development |
Publisher | Springer International Publishing |
Pages | 95-107 |
Number of pages | 13 |
ISBN (Electronic) | 9783030476823 |
ISBN (Print) | 9783030476816 |
DOIs | |
State | Published - Sep 16 2020 |
Keywords
- 3+3
- Bayesian optimal interval
- Statistical properties and performance Novel phase I design
ASJC Scopus subject areas
- General Medicine