Other paradigms: better treatments are identified by better trials: the value of randomized phase II studies.

The emergence of many newer, molecularly targeted anticancer drugs requires that we rethink the way that we conduct phase II trials in oncology. In particular, we can no longer afford to advance drugs (or combinations) to phase III trials with a high risk of failure to improve on outcomes. Drawing on phase II trials of therapeutics for other chronic and progressive diseases in medicine, we find that a randomized design is essential not only for selecting agents for further study but also for optimizing the design (dose, patient population, and endpoints) of the subsequent phase III trials. We use the example of advanced nonsmall cell lung cancer to demonstrate how randomized phase II trials have already made an impact in oncology, whereas single-arm phase II trials have led to negative phase III trials in the same disease. Finally, we make the case that randomized phase II trials are feasible, as long as reasonable statistical standards are applied.