|
Abstract:
|
The primary objective of Phase I-II trials of multiple-agents in oncology is to locate an optimal dose combination based on toxicity and efficacy endpoints. The statistical methodology underlying the design of these trials is intended to search a drug combination matrix comprised of several discrete doses of each agent. The combination-outcome probabilities associated with the two-dimensional grid are characterized by a partial order. Vaccine-based combination immunotherapy may investigate treatment strategies that do not consist of several multi-agent doses, creating an unconventional partially ordered combination space. Motivated by an early-phase trial of a novel vaccination approach in melanoma, a model-based design is described for determining the optimal treatment regimen, based on bivariate binary measures of toxicity and biologic activity. A simulation study demonstrates the method's ability to effectively recommend optimal regimens in high percentage of trials with manageable sample sizes. A brief discussion of the design's flexibility in handling other complex dose-finding problems, such as different treatment schedules, is provided.
|
