SLIDES & TRANSCRIPTS
Tuesday, March 6

Clinical Trial Designs for Target-Directed Therapies
Richard L. Shilsky, MD

targets have been identified based on understanding the molecular pathogenesis of gastric cancer; target-directed inhibitors have been produced and shown to bind to the target with adequate affinity and specificity; and that these inhibitors produce regression or growth inhibition in preclinical models. I think the preceding session demonstrated that these assumptions are pretty valid, and that we have targets and we have, and will have, inhibitors.

There are, however, a number of important issues in the development of these target-directed therapies that we hope to be able to address during this session. Early on, of course, is the question of dose optimization for some of these novel agents. What dose reproducibly inhibits the target in tumor tissue? It is an important question and one that is not always easy to answer in the clinical setting. Now, we have heard a lot of talk about surrogate markers. I actually think it is important to distinguish two kinds of surrogate markers. There are what I call surrogate effect markers. To assess these, we need to know what biochemical or molecular changes in perhaps normal tissue reliably correlate with target inhibition in tumor tissue. In my mind, a surrogate marker for effect might be a test done in a normal tissue or perhaps an imaging strategy that predicts, with a high degree of reliability, whether or not the target-directed agent is actually hitting the target.

Then there are what might be called surrogate benefit markers. The issue there is whether the target inhibition, whether assessed in tissue, normal or tumor, or assessed non-invasively, actually predicts clinical benefit in the patient. It is fine to know that you can inhibit the target, but the real question for the clinic is, if you inhibit the target, does that actually produce a meaningful clinical benefit for the patient. I think when we think about surrogate markers, it is useful to try to distinguish those things that are surrogates for an effect and those things that may be a surrogate for a clinical benefit.

TOP

There are a lot of issues, as I said, regarding the development of some of these target-directed agents that we will try to address in this session. For example, how do we manage tissue heterogeneity in interpreting the results of clinical studies? We have heard something about that already. How do we manage assay variability in interpreting results? How do we, in fact, assess clinical benefit for some of these newer types of agents? How do those assessments impact on the selection of the appropriate patient population and on the end points for clinical trials? Our commonly-used end points are things like disease-free survival or progression-free survival or overall survival. Are those, in fact, still going to be valid clinical end points and is one preferred over another for a particular type of agent or clinical setting?

TOP

As a basis for discussion, and not to suggest that this is an optimal design, but for purposes of discussion, we are proposing the following trial paradigm. For patients who have potentially resectable gastric cancer, there might initially be an endoscopic biopsy with evaluation of the tissue, a period of treatment of the patient with a new agent, some clinical assessment, followed by primary resection of the tumor. Then, following that, if appropriate, adjuvant therapy, which might be done in the context of a randomized trial of standard adjuvant therapy with or without the new agent that was evaluated early on in the course of treatment. Obviously, this is one of many possible clinical trial paradigms, but it gives us an opportunity to have a baseline tissue assessment, a period of treatment, some clinical assessment, perhaps with a novel imaging modality, get tissue again to assess the biological effects of the intervention, and then potentially do a randomized clinical trial to determine definitively the benefit of the new agent.

TOP

There are, of course, a number of statistical issues with respect to this and other clinical trial designs that will be discussed. They include, for example, the impact of the target prevalence on the sample size that is required to do the study. Again, how does one manage heterogeneity and target expression as well as assay variability. How do all of these things impact on the magnitude of the difference that is actually clinically detectable with appropriate confidence intervals. How do we incorporate issues of correlating biological markers and clinical outcomes, and what are the strategies we can use to develop appropriate information about how markers correlate with each other, as well as how markers correlate with what may be a variety of clinical end points. Finally, how do we actually demonstrate clinical benefit in the context of clinical trials of novel agents.

TOP

So, to discuss these issues, we have three speakers this afternoon, Emily Bergsland, who is a medical oncologist at the University of California, San Francisco, Gary Rosner, a statistician from M.D. Anderson Cancer Center, and Rich Wahl, who is a radiologist at Johns Hopkins. We will ask each of them to give their presentation in this order. We will take time for a few pertinent questions after each presentation. Then we will have a discussion with group participation at the end.

TOP