Optimizing randomized phase II trials assessing tumor progression

The traditional development paradigm for phase II trials in oncology has been challenged in recent years by the introduction of cytostatic therapies. These agents slow the growth of tumors rather than cause high rates of shrinkage, this argues for the use of endpoints that measure growth inhibition such as progression free survival. We have previously argued the need for randomized trials in this setting. Here we discuss methodological solutions to enhance the development decision at the end of phase II in the context of progression endpoints employed in randomized trials. There are well recognized issues associated with progression endpoints relating to bias in the timing and interpretation of assessments. In this paper we present design and analysis solutions that will minimize bias by using methods that are either partially or completely time independent. We also discuss other design features to maximize the information yielded in a phase II setting. We advocate the creation of progression endpoints that utilize all available progression data rather than early fixed time-point analyses and show that little is to be gained by assessing progression status any more frequently than would be required in routine clinical practice. Such design and analysis measures will optimize the development decision made at the end of phase II clinical evaluation.