A Bayesian adaptive phase I-II trial design for optimizing the schedule of therapeutic cancer vaccines.

Phase I-II clinical trials refer to the class of designs that evaluate both the safety and efficacy of a novel therapeutic agent in a single trial. Typically, Phase I-II oncology trials take the form of dose-escalation studies, where initial subjects are treated at the lowest dose level and subsequent subjects are treated at progressively higher doses until the optimal dose is identified. While dose-escalation designs are well-motivated in the case of traditional chemotherapeutic agents, an alternate approach may be considered for therapeutic cancer vaccines, where an investigator's main objective is to evaluate the safety and efficacy of a set of dosing schedules or adjuvant combinations rather than to compare the safety and efficacy of progressively higher dose levels. We present a two-stage, Bayesian adaptive Phase I-II trial design to evaluate the safety and efficacy of therapeutic cancer vaccines. In the first stage, we determine whether a vaccination schedule achieves a minimum level of performance by comparing the toxicity and immune response rates to historical benchmarks. Vaccination schedules that achieve a minimum level of performance are compared using their magnitudes of immune response. If the superiority of a single schedule cannot be established after the first stage, Bayesian posterior predictive probabilities are used to determine the additional sample size required to identify the optimal vaccination schedule in a second stage.