Toxicity-evaluation designs for phase I/II cancer immunotherapy trials.

Objectives: To facilitate more efficient Phase I/II cancer immunotherapy trials by incorporating statistically rigorous safety analysis.

Setting: The standard Phase I oncology trial is designed to find the maximum tolerated dose (MTD) in a setting where serious drug-related toxicity is expected. However, many newer agents hope to show the efficacy without increasing the background rate of adverse events. Formal statistical designs in this setting are needed.

Results: The Phase I/II toxicity-evaluation design is suitable when the therapeutic dose is expected to be well below the MTD. In Phase I, the design enrolls multiple cohorts at the target dose, possibly after an initial dose titration stage, and tests a formal safety hypothesis using a standard 3 + 3 enrollment scheme. Phase I serves as an interim safety analysis before proceeding to Phase II efficacy testing. We give an exact upper confidence limit on the toxicity rate at the therapeutic dose using the combined Phase I/II toxicity data, as well as the maximum likelihood estimate of the toxicity rate. We describe an example where the design has been used for a Phase I/II trial of immunotherapy in leukemia.

Conclusions: Phase I/II toxicity-evaluation designs are simple to execute and may be suitable for some cancer immunotherapy trials. We show how to compute power, expected sample size, and expected number of dose-limiting toxicities, as well as the maximum likelihood estimator and exact small sample confidence intervals for the toxicity rate at the therapeutic dose. More flexible designs are briefly discussed.