All Times EDT
The genomic revolution has changed the world of medicine. In many ways. Modern biology is discovering potential therapeutic targets and associated targeted therapies. And these discoveries are happening daily. What was once a single disease is now many diseases. Soon every patient will have an ultra-orphan disease. Large two-armed clinical trials that focus on a narrowly defined disease will be impossible. The traditional statistical approach to clinical trials must adapt to the biological revolution. Relating measurements of individual patients over the course of their disease will be an essential aspect of making progress. For example, correlating imaging, circulating tumor DNA, and patient-reported outcomes over time will be fundamental in clinical cancer research. Clinical trial statisticians will have to become steeped in longitudinal disease modeling. Of special importance will be the increasing use of early biomarkers that reflect therapeutic effect and that predict longer term clinical endpoints. In this presentation I will give examples of actual clinical trials that have used or are using such early biomarkers. In cancer and other diseases. Observations on such biomarkers can be useful in adaptive clinical trials for getting tentative answers early and for modifying the trial accordingly. This is so whether or not the early biomarkers are or have the potential to become surrogate endpoints in the regulatory sense. I will also give my perspective on, and examples of, what is required to demonstrate that a particular biomarker is a “reasonably likely” surrogate endpoint in the sense that it can be used to support accelerated marketing approval. I will also address methods and data necessary to demonstrate that a biomarker is a “validated” surrogate.