Challenges in Oncology Trials: Analyzing Endpoints

ACT: What are some of the unique challenges facing oncology trials today?

Buyse: Well, I think there are many challenges, but one of them is the fact that when we do a randomized clinical trial in oncology, we collect a lot of data, and then at the end of the trial we analyze a primary endpoint, a so called designated primary endpoint, and the fate of the trial completely depends on whether this primary endpoint shows a treatment benefit or doesn't. And sometimes there is a treatment benefit, but it just fails to be statistically significant, in which case the trial is considered to be negative, and the rest of the data serves no useful purpose. Now, in many cases, that is a problem, because some other endpoints actually show a benefit of the new treatment, but then these endpoints are not analyzed, and even if they are analyzed, it’s descriptive, and it's not considered to be statistically significant because the primary endpoint failed. So that is a big problem, I think, because first of all, we may miss good drugs by doing this. Secondly, patients are asked to go through a lot of examinations—a lot of lab tests, and a lot of exams—and then the data of all these exams eventually may not be used or useful. In that case, of course, it's been a huge waste of time. And so I think we have done a pretty bad job, if we look at the totality of clinical trials that have been conducted—not so much in missing really effective drugs—but in missing drugs that had some benefits, but also perhaps not as much as we thought, or perhaps some harms as well as benefits, so I think what we need is a much more powerful approach, or a more sensitive approach. To use the statistical word, for statistical sensitivity, means that if there is a true treatment effect, you're going to see it with high probability and I think the trials that we've done in the past, for many of them, lacked that sensitivity, and therefore perhaps resulted in not approving drugs that would have worked.

I think it's fair to say that when we collect a lot of data, we have to analyze the data, and we have to look at the totality of the evidence in favor or against the treatment before we make a decision. And so the method that we have implemented at One2Treat, and that we continue to develop, is a method that allows us to include more than one outcome or endpoint in our analysis. So instead of focusing entirely on a single endpoint, we try to now think of the endpoints that are relevant for patients, that are important, that are clinically the sorts of events that patients will worry about or will be concerned about, and then build the trial around these patient-centric clinical endpoints, rather than focusing on one traditional endpoint that has always been used. And so it's a complete change in paradigm. We want to move away from a sort of very focused, very narrow-sided view of the trial as having a single endpoint of interest, to a much more holistic view, much more patient relevant view of the trial, where in fact, all the clinically relevant endpoints or outcomes are taken into consideration simultaneously. That's the basic idea behind the methodology that we have adopted.

Coppe: Last part as Marc was mentioning, most of the of the clinical trials in oncology were focusing on proving that the efficacy was better or superior than any kind of other already commercialized products. And from time to time, you may just want to demonstrate that your new innovative drug that you want to bring to markets is as efficient as another drug, but maybe also related to less toxicities, less secondary effects. And of course, that's directly related to quality of life, and that's not always easy. So in the past, the way this has been managed was also to run what we call non-inferiority trials where these are typically long-lasting clinical trials where you need to recruit a lot of patients to demonstrate this kind of non-inferiority, and that's not always the nice way to, I would say, motivate both conditions and patients to take part in this kind of clinical research, right? So again, trying to make sure you can involve—as Marc was mentioning—multiple patient relevant outcomes in the clinical trial design may help you consider both efficacy and safety outcomes in your primary endpoint within one single metric, and this helps you avoid those challenging non-inferiority trials. And that's better for patients, that's better for clinicians, and for pharma sponsors, and ultimately from an overall standpoint, bringing efficient drugs to market in a faster way.