Kerr, Tannock on the 'Triangle of Mediocrity' Holding Oncology Back

David J. Kerr, CBE, MD, DSc, FRCP, FMedSci; Ian F. Tannock, MD, PhD, DSc


November 16, 2018

David J. Kerr, CBE, MD, DSc, FRCP, FMedSci: Hello. I'm David Kerr, professor of cancer medicine at the University of Oxford. Welcome to Medscape Oncology Insights, coming to you from the 2018 Congress of the European Society for Medical Oncology (ESMO). Today we will be discussing two dilemmas which are facing the broad field of oncology: the limits and boundaries of precision medicine, and sustaining the development of innovative medicines. Joining this discussion is an old friend and colleague, Professor Ian Tannock, who is emeritus professor of medical oncology at Princess Margaret Cancer Centre and University of Toronto. Ian, good to see you.

Ian F. Tannock, MD, PhD, DSc: You too, David.

Limits and Boundaries of Precision Medicine

Kerr: How long have you been a cancer doctor?

Tannock: I think I finished my oncology residency training in 1978, so 40 years.

Kerr: It's 34 years for me, so between us we have almost three quarters of a century of accumulated knowledge. Are we allowed to say that?

Tannock: Accumulated knowledge and accumulated skepticism. But it's healthy skepticism.

Kerr: That is part of it. We have been around long enough to see cycles come and go, and we see our young colleagues and fellows fascinated by specific subjects and thinking that their approach just might be the one to break through.

I think we are overly in love with a P value of.05.

Tannock: We have to accept that enthusiasm is not always borne out by reality. I don't think that failure is the norm. Things are a lot better than when you and I started, but not because of big breakthroughs; those have been very rare.

Kerr: Remember all of those mouse experiments we did a thousand years ago? If we were mouse doctors, we would be king. There is no mouse cancer that we cannot cure. But the leap from mouse to man has proven harder than perhaps we had imagined when we were kids.

Tannock: There is another lesson from doing mouse or cell experiments, because when we are in the lab, we have to repeat things several times. We often find that we have a one-off result that looks fantastic, but when we repeat it, it does not work. And yet when there is one positive clinical trial, we accept that as the reality. But it may actually be nothing more than a one-off result. Repeating consistency in clinical trials is, for me, much more important than any arbitrary P value that is set for them.

Kerr: I think we are overly in love with a P value of .05. Richard Peto [University of Oxford epidemiologist] would agree with us, were he here. Let me challenge you a little. When we were youngsters, trials of a couple hundred patients needed to be done and repeated. You do not think modern trials, sometimes with thousands of patients, high statistical power, and all sorts of paraphernalia that come with them, can be sufficiently compelling as a one-off? That still we just need to dot i's and cross t's?

Tannock: It depends on the situation. You have to look at the summary of evidence. We have seen some very good trials where all of us are impressed. After 100 years or so of dabbling in immunotherapy, we've at last got something that really works. That is very positive. But we also see companies doing trials with me-too drugs. They put 1000-2000 patients into those trials because registration agencies like the European Medicines Agency and US Food and Drug Administration will generally register any drug that shows a difference with a P value < .05. If you have a very small effect, you need lots of patients to show it. I would argue that, in general, we are wasting resources by showing tiny effects, particularly with drugs that are not different in any big way from those that have gone before.

Kerr: I agree entirely. We somehow set the bar too low. Although we may meet the primary endpoint of the trial as defined statistically—tied up in P values and confidence intervals—there is a gap between statistical certainty and clinical validity.

Tannock: That is very true, and I do applaud the efforts of ESMO[1] and the American Society of Clinical Oncology[2] for developing value systems. We should be registering new drugs and new therapies on their value for patients. It's not easy to measure that, but efforts that have gone on by the ESMO team, led by Nathan Cherny and others, are very worthwhile. I would like to see registration agencies start to take note of that and use a value system. There was a recent paper in BMJ by Davis and colleagues,[3] showing that only half or less than half of the drugs that were registered over a decade actually improved survival or health quality for patients. A study we did showed that only about a third of the positive trials are sufficiently beneficial for patients to meet a criterion that is set by the ESMO benefits scale.[4] I think we can do better than that.

Kerr: Is the treatment of cancer—literally our lifetime work—a Sisyphean effort where we are pushing a large boulder up a hill slowly, slowly, slowly rather than there being stepped breakthroughs? I guess you and I have contributed along the way with a nudge, a push, a gentle shove.

Tannock: We have contributed to the gentle differences. They add up and they are important. Occasionally we have something that is much bigger. When imatinib was first used for chronic myelogenous leukemia (CML), this was a knock-out. Others were cisplatin for testicular cancer and perhaps treatments for melanoma. BRAF inhibitors and immunotherapy have changed that disease where virtually nothing worked and now [we can] give realistic hope to patients. So, there have been a few steps, but most increases in outcome in the common diseases have been small.

Kerr: Sometimes I feel that there is a folie à deux or folie à trois between academia, the industry, and regulatory bodies, and that somehow we support a sort of triangle of mediocrity—small, expensive, complicated drugs with marginal benefits, huge amounts of paperwork, and huge amounts of oversight. Of course, we're focused on safety. But trials are much harder to do and much more expensive than when we were kids.

Tannock: That is very true. We obviously need to protect patients, but to some extent we have created a system that is not as beneficial for patients. There is good evidence that patients on trials do better than their colleagues not on trials; this may be due in part to selection of patients with better characteristics. Patients are well served by being in trials, and yet we have made them more expensive and more difficult to do. This has been partly stimulated by one or two bad events in trials that have led to deaths or severe toxicity. While this is extremely rare, it's changed the whole system. There needs to be a better balance.

Value in Oncology

Kerr: We touched a little upon the concept of value. This is a complicated word, because when my wife goes to the supermarket, the value counter is where the cheaper cuts of meat are, the slightly dodgy cheese, and the milk that has just turned. A "good value" is attributed to something of low quality. Whereas when we talk about value [in oncology], we are saying that for this investment of dollars, pain, and hassle of coming in for this treatment, we want you to live longer and better. But given the current costs of drugs and the rather meager returns, value seems somehow to be squeezed out.

Tannock: We are more conscious of value than we were a few years ago because of the ESMO effort. You are right. There are only two aims that any form of doctor can really have: to make patients either live longer or better. I think the problem in interpreting trials is that many of the trials are using surrogate endpoints—without showing surrogacy—that do not necessarily lead to those improvements in quality of life or in the duration of survival. Having the value system brings us back to concentrate on that, and that is valuable. I mean, we're not looking for the "cheaper cuts"; we're looking for things that really do make a real difference. The cost is just ridiculous. It seems to me absolutely crazy that there is no limit or control of costs. A lot of drug development is funded by governments and charities, and yet in the end, companies can essentially charge what they want. There is absolutely no relationship between cost and value as measured by improvements in survival or its quality. And that is actually different from any other commodity.

When you go out and buy a car, you can go and buy a Ford and pay a moderate amount of money, or you can go buy a Porsche and pay a lot of money. But at least you get something that is different and more valuable. That is absolutely lacking in cancer medicines.

Kerr: "Financial toxicity": Is that an expression you have ever used? Bob Carlson, who is the "grand poobah" of the National Comprehensive Cancer Network, introduces the concept of financial toxicity when he meets patients. Not neutropenia, diarrhea, and mucositis, etc, but financial toxicity. He bravely discusses this up front, telling them, "There's this pattern of treatment and there's this pattern of treatment. If you go down treatment route X, it costs a very large sum of dollars. Route Y costs a smaller amount of dollars but is perhaps marginally less good." Bob, God bless him, is having that conversation. I'm protected, working within a socialized healthcare system we really love, called the National Health Service, as are you. But isn't that extraordinary?

Tannock: It is extraordinary but I think it's necessary. In the United States, the commonest cause of bankruptcy is health costs. A lot of people there are foregoing treatment, even if they have insurance, because of the copay. If a drug costs $10,000/month and the copay is 20%, a lot of people can't afford $2000/month. And we need to look at the rest of the world. As people are living longer in middle-income and lower-income countries, cancer is becoming much more of a problem. They simply cannot afford these types of drugs. It's very sad that we are developing a lot of treatments for the rich but not for a lot of people who really need them.

Limitations of Precision Medicine

Kerr: Let's talk about precision medicine and actionable mutations. It sounds trendy and exciting. It's sounds like cutting-edge medicine: "We have discovered an actionable mutation, and regardless of tumor type or somatic mutational landscape of your tumor, this drug will work." This is overstretched.

Isn't it just awful that evidence gets in the way, yet again, of a good idea?

Tannock: You have to look at evidence. The idea is good. It seems logical and sort of sexy. Patients like getting their tumor sequenced. But if you look at trials, which are largely single-arm series, the evidence is like a funnel. You have 1000 patients who sequence their tumors. They don't realize that if you take a biopsy from one part, it's going to be different from another part. We have learned over the past decade that tumors are continually mutating. There is a lot of heterogeneity; what may be driving the tumor at one point is not really driving it somewhere else or shortly in time. That is a limitation and it happens in about 60% of those who have the disease sequence. And of those patients, for perhaps half of them there might be some drug that seems to act against that mutation. But when you use that drug, only a proportion of patients tolerate it. Even if they do tolerate it, in all of the series that I'm aware of, you end up with a response of < 10%, which is actually not better than simply pulling a drug off the shelf. While we should be doing research to ask why that is and how we can make the situation better, unfortunately a lot of the research that is going on—including the current ESMO project of community testing and various projects around the world—is all doing the same thing. It's become much cheaper and easier to work out genetic sequence of a biopsy of a tumor, but that isn't taking us anywhere. We don't have the right drugs. The drugs are too toxic to be used in combination. And those are the things that need to be done intelligently in a more limited number of well-designed research projects. There is no evidence at all that precision medicine in the sense of wholesale genome testing has improved the outcome of patients. Of course, I'm not talking about knowing whether a woman with breast cancer is HER2+ or ER+, or if the melanoma is BRAF mutated. Those sorts of things can be done much more simply, and those biomarkers are important. But whole genome sequencing has not, as yet, shown any sign of improving the outcome of patients.

Kerr: I agree that we could have a double handful of biomarkers that we would use—companion diagnostics—to move treatment. There are two things to pick on. One is heterogeneity—I could not agree more. There is evidence of this in a whole range of tumor types. People say that we can overcome that by using liquid biopsies. This is another myth that has yet to be busted. Also, I question the idea that we can take a driver mutation in breast, bowel, and lung cancer and use it independently of what else is going on within the tumor, in terms of microenvironment or other mutations that are associated with those tumor types. It makes no sense to me at all that you would actually take such a reductionist approach. Isn't it just awful that evidence gets in the way, yet again, of a good idea?

Tannock: You're right. It's not only the thought that that can happen; there is very good evidence. People have looked at the BRAF mutation, which is very useful for deciding treatment of melanoma; but you find it in colon cancer and it does not make an atom of difference. Unfortunately, it's not that simple. We are also lacking in the drugs we use. We do have some useful molecular-targeted agents, such as trastuzumab [for breast cancer] and imatinib, which is probably the best one of all for CML. But most of the targeted agents are not hugely effective. They don't inhibit their targets very completely, and if they do, they are too toxic because these targets are also used in normal cells for them to function. Yes, there are lots of problems. I'm very supportive of good research projects that look for common mutations that might be targeted in different ways, perhaps with immunotherapy, but you have to do it intelligently. Unfortunately, almost all of the North American cancer centers advertise personalized, precision medicine and yet they are essentially doing much the same thing. They are not learning from other people's experience.

Kerr: It feels like a blunderbuss. It feels like a rather un-thought-through approach at the moment. I agree. There is enough accumulated evidence for us to pause for thought, exactly as you suggest, and think, There is something to this, but we need to step back a little and think about what it might be. Our bioinformaticians are doing a lot of work trying to do nodal connections, so rather than take an isolated approach, can we look at pathways, etc? Can't we just take a wee step back rather than use an ultra-reductionist approach of "one mutant + one gene = answer to cancer"?

Tannock: That is important. Other people are looking at epigenetic changes to the genome. It's not the genes that function; it's the proteins that they make. Good research is going on in some places, but I think too many cancer centers and organizations are going too quickly. Obviously, patients are desperate to have new things to help them, but unfortunately there is political pressure to do things that aren't evidence based.

Kerr: It's just been delightful catching up after all of these years. Fantastic fun. To our friends who have been listening, thank you very much indeed. This is goodbye—not from two grumpy old men but perhaps two thoughtful old men. We will take leave of you now from the ESMO 2018 meeting here in Munich. For the time being, Medscapers, over and out. Thank you.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: