A need for clarity in clinical trial end points

Share :
Published: 13 Dec 2012
Views: 2838
Dr Clifford Hudis – Memorial Sloan Kettering Cancer Center, New York, USA

Dr Clifford Hudis talks to ecancer.tv about the variation of end points which are used to gauge the success of clinical trials. Clinical study success is measured by a range of outcomes such as survival benefits or patient quality of life, and there is a need for greater patient input to determine what the priorities for future trials should be.


Although there are clear definitions of terminology used within early stage cancer trials, this is not the case for later stage studies. Dr Hudis explains why he believes clarification is necessary and outlines how this could be achieved.

SABCS 2012


A need for clarity in clinical trial end points


Dr Clifford Hudis – Memorial Sloan Kettering Cancer Center, New York, USA



We only get drugs and treatments into patients if we do prospective research and identify where to use them and where they’re not effective and so forth. This is especially true when we’re talking about regulatory issues, meaning you can’t sell a drug if it’s not known to be effective. I know everybody knows this but it’s worth making the point because, in the US at least, only 2-3% of adults with solid tumours go on clinical trials, so about 97% or 98% of all the people who are exposed to all the therapies we try in advanced cancer aren’t contributing to future knowledge.


The second problem, of course, is that the cost of drug development has, on the commercial side, really sky-rocketed and I’ve heard estimates, and I can’t be certain which of these are right, but a range of anywhere from $1-4 billion now, all in, to get from a novel potential agent to an approved drug for patients.


The issue that I addressed was in that mix we have a problem in advanced breast cancer, and in other solid tumours, agreeing on what we think is important. So the way this has played out in recent years has been very controversial, arguments, decisions, discussions around things like bevacizumab for advanced breast cancer. Whether or not it’s effective is subservient to the question what is effectiveness? What defines a better drug for advanced cancer? And that’s what my talk was about – the clinical trials endpoints. Because ultimately we have to agree on what matters before we can argue about what really works and we haven’t really done that yet.


So this brings us into the meat of the matter. One could be hard-nosed about this and simply say, “When I’m treating advanced cancer I want to cure it. Since we don’t cure it I will settle for improving overall survival.” That immediately raises questions of cost trade-offs for modest or small benefits in overall survival. But put that aside for a minute. If you can’t prove that overall survival is better, does that mean that no drug is ever better than another? Well no, there are intermediate end-points, what we might call surrogates for overall survival, or surrogates for other things that matter like quality of life. That brings us to response rate, cancers getting smaller, and it brings us to progression free survival, which is one time-dependent measure of how long the patient is able to go without getting the bad news that her cancer is worse. There is again the issue, because which endpoint we select for a clinical trial has something to do with whether we call the trial successful or a failure and indeed how we present it to ourselves and our world. So I focussed on response rate, progression free survival and their sometimes loose relationship with overall survival. I want to be clear, I’m not arguing that you can only approve a new drug or use a new regimen because it improves overall survival, I would argue quite the opposite actually, but our community really does have to get together and come to some sort of understanding and agreement about which of these things we value and then agree that when we make a threshold we will call it better and when we fail to make it we will agree that it isn’t better.


Is there a need for more regulation at endpoints?


I don’t want to say more regulation of endpoints, in fact I might say the opposite. I would say more clarity and openness about what it is we’re looking for. So my own experience with this really began years ago when the aromatase inhibitor trials came out. The five original trials that showed that aromatase inhibitors were superior to Tamoxifen in early stage breast cancer; they all, four of the five, defined the improvement in terms of disease free survival. Fair enough, for early stage cancer if your cancer doesn’t come back that’s a good outcome, right? One of the trials used a different term, event free survival, and we started digging into the differences. It turned out that all five studies used different definitions of what was better; four of them called it disease free survival and I used to always say this is sort of frustrating because if you tell me that you’re 1.6m tall, I shouldn’t ask whose metres are you using. A metre is a metre is a metre. So disease free survival should be disease free survival period. That lead to a paper in the JCO and at the end of the list I gave the defined endpoints for early stage trials. I want to be clear, we’ve never suggested being prescriptive about this, I don’t actually care very much about what people use. What I care about is that we all agree on what the words mean, what endpoints are bundled into groups and what we call success. So that’s a segue into the advanced disease setting where I think we’re grappling with this right now.


How is this officiated?


The way we did it in early stage cancer is that we had a series of meetings that the NCI helped to organise and supported by some funders and we worked it out in a manuscript. The thing is that it wasn’t especially charged, it’s very wonkish, it’s very technical. All we wanted to be able to say is that when a paper says disease free survival it means exactly and only one thing. If you want to modify that, if you have a reason that you want to include other endpoints, you put a little tag on the DFS to explain the variation. That way you don’t get into arguments about A is better than B, C is better than B but C is more better than B and I want to choose C indirectly. It’s not a productive way for us to do science; we wouldn’t tolerate it in a quantitative laboratory experiment so we shouldn’t here.


Now the point of all this is that ultimately the regulators are our partners and they’re looking to us for guidance and help. In fact, quite frankly, at yesterday’s oral session here there was a long discussion from several people about the use of pre-operative in breast response to novel agents as a potential pathway for drug approval. What you really heard from the Food and Drug Administration was a description of what they know and, in a sense, a request for help in sorting out this data so that they can arrive at a definition of success that we would all agree on. So I think it’s a real chance for partnership and to make an advance in the science of how we do our clinical trials.


Do you think patients should be involved in the design of trials?


I think very much so and that actually is a key point here. So if in the advanced disease setting we are not improving overall survival, a fair question that a savvy patient might ask is, “So why should I put out my arm and get intravenous chemotherapy? You just told me if it has an impact on my likelihood of survival it’s pretty modest, so what are the benefits to me?” I might say, “You would be able to go eight months right now without your cancer getting worse whereas if we didn’t give you this treatment you might only go three or two or four, I don’t know, whatever it is.” And a patient might say, “Well if I’m still likely to pass from this Earth at the same date in the future, I’d rather not spend my time getting this treatment, even though you have a metric of improvement.” Another patient might say, “Well, what does that correlate with? Does that mean I’ll be able to go windsurfing and skiing and to parties and out or will my quality of life be ruined by the treatment and I’ll be housebound and unable to enjoy life?” I think weighing the value of those changes in quality of life, mixed in with the things we measure, is important, it does require input from patients.


I do have to add, though, one of the unfortunate misunderstandings sometimes is the notion that there’s this kind of a firewall between the physician scientists doing clinical research and the patients. Let me just say this: some of the physicians are patients and some of them, all of them, most of them, will be patients at one point. So I think it’s important for active patients and those who have been through treatment to have an input into all of this but I think we should be a little easier on the physicians sometimes and not assume they know nothing about it.