Group-sequential and adaptive designs in immuno-oncology

Abstract

Here I talk about designing a clinical trial for an immuno-oncology endpoint where we expect a delay in the seperation of the survival curves. Starting with the properties of a standard design under non-proportional hazards, I believed I had made a great improvement by adapting the sample size based on a “ratio of hazard ratios”. However, a well-chosen group-sequential design made an equal (slightly better) improvement to the operating characteristics, and is much simpler.