A Simulated Exploration of Alternatives to Standard Expansion Cohorts in Phase 1 Trials
Jacobs, Justin Leon
0000-0002-5621-5962
:
2022-11-18
Abstract
In phase 1 clinical trials, dose escalation is an iterative process whereby a maximum tolerated dose (MTD) is identified for subsequent testing. Patients are treated with escalating dose levels to determine which may result in too many dose-limiting toxicities (DLT). This is sometimes followed by dose expansion, both in order to confirm accuracy of the MTD estimation and to assess other safety and efficacy concerns before moving to phase 2.
While Continual Reassessment Method (CRM), modified Toxicity Probability Interval (mTPI), and Bayesian Optimal Interval Design (BOIN), have been utilized for dose escalation, the traditional 3+3 method remains popular due to its ease of use and conservative nature.
Expansion typically involves enrolling a prespecified number of patients at estimated MTD to assess responses, but Iasonos et al. suggested modified expansion cohorts, enrolling patients at the MTD and adjacent dose levels to assess accuracy of MTD estimation, safety, and efficacy. Accuracy of MTD estimation for these alternative expansion methods is explored herein through a simulation study, across multiple dose toxicity curves and escalation methods.
Simulations using CRM reveal a modified expansion cohort method using the estimated MTD and adjacent dose levels can improve estimation accuracy compared to standard expansion or escalation only, while not substantially increasing DLTs. In mTPI and BOIN, none of the modified expansion methods significantly improve accuracy compared to standard expansion. With 3+3, expansion of any kind does not significantly improve accuracy.