Keywords: dose-escalation, single-patient acceleration, early-phase oncology trials
The ever-changing landscape of oncologic drug development poses new challenges and opportunities in early-phase dose escalation (DE), where there is a high demand of more efficient plans. Three categories of methods exist: rule-based, CRM-based, and TPI-based, and each can be implemented on its own or in combination with a Single-Patient Acceleration (SPA) “lead-in” that only one single patient (pt) is tested. Their performances differ in terms of four distinct but correlated criteria that are of major considerations in an FIH study: (1) patient safety; (2) accuracy in identifying RP2D; (3) sample size; and (4) trial duration. We performed simulations to evaluate the operating characteristics (OCs) for those novel “hybrid” methods (e.g., R-TPI and i3+3) that claim to incorporate advantages of established methods from different categories, with or without SPA, and in case of SPA, with or without a pre-specified “cap” that forces to switch to other methods after certain number of dose levels are tested. We provided recommendations for determining the most appropriate method, based not only on the representative simulation results but also practical considerations commonly encountered while designing an FIH oncology study.