October 2014, Vol 3, No 7

← Back to Issue

Persistence and the Expedition to Mine the Immune System for Cancer Cures:

An Interview with Dr James Allison of MD Anderson Cancer Center

James Allison, PhD

Interview with the Innovators

The recent triumphant emergence of immuno-oncology as a powerful cancer treatment venue with rich potential has the oncology community’s full attention. While the spell lasts, we do well to learn not only the empirical considerations of this stunning “overnight success after a 30-year quest,” but the underlying principles that sustained it along the way. Persistence, indeed tenacity, is essential to the pursuit of cures for anything as deeply complex as cancer. It is the fundamental pillar on which research goals and grand visions depend for the time they need to incubate and mature into substance. Persistence, even more than genius, is the cardinal virtue of medicine, and all stakeholders must practice it: providers, payers, corporate shareholders, research teams, pharmaceutical marketers, accountable care organizations, and, of course, patients, their families, and caregivers.

A talk on the power of persistence was delivered in an arresting, offbeat keynote address delivered at the 2005 Annual Meeting of America’s Health Insurance Plans (AHIP). There a packed hall of 3500 healthcare professionals received the very last analogy they might have expected on medical progress. Celebrity author Malcolm Gladwell, perhaps best known for his bestseller The Tipping Point, made his case for persistence in drug research using the unlikely example of the road to success of…Fleetwood Mac.

Yes, Fleetwood Mac. This particular British hippie rock group formed in 1967 and, unlike so many instant apparitions of the British rock invasion that often died out as quickly as they hit the Top 40, Mac underwent years of trial and error before finding its stride, its sound, and its public. The broad applicability of this band’s experience is not fleeting for Gladwell either: he delivered another keynote address adapting its principles to educational innovation at an annual meeting of the National Educational Computing Conference a few years later. In his AHIP address he showed the length of time for the band to find its identity, leaving a trail of failed albums along the way until they achieved a trademark sound that defined them as an iconic band selling “…more than 100 million albums worldwide, making them one of the best-selling bands of all time.” Of paramount significance, he remarked, is the contrast between the patience of record corporate leaders then and since. The music corporate culture stayed with Fleetwood Mac as it changed its lineup and its sound, unlike the current recording companies that demand a quick return on investment. This stifles experimentation, and Gladwell observed how this can hollow out medical innovation just as thoroughly as it does musical innovation. The maxim is that those who expect a venture or a great performer to succeed overnight ignore the realities of enterprises of great depth.

Gladwell holds to the premise “…that effort is more important than talent. ‘When we look at people who come to master something…we have a tendency to telescope how long that learning took place – to think that the learning happened overnight.’…In fact, almost every successful individual or organization puts in at least 10,000 hours of practice first, which averages out to about 4 hours a day for 10 years. Successful learning begins not with talent, but with an approach to the task, an approach that says, ‘I believe that my effort is crucial for getting somewhere.’”

Whether or not healthcare researchers pursuing the immune system instead of tumors were listening to Gladwell’s concept, they adopted it. Immuno-oncology has begun to hit its stride, with the journal Science naming it Breakthrough of the Year for 2013. An article this year in Biotech Now encapsulated the process with its title: “Immuno Oncology – A 30-Year Overnight Success Story.”1 The pattern it chronicled of the pit bull determination in the immuno-oncology epic might have been lifted from Gladwell’s description of Fleetwood Mac’s series of album failures after an initial success, its changing lineup of musicians led by a visionary who needed time to perfect his craft. The Biotech Now article vindicates Gladwell’s hypothesis: “For such a seemingly simple yet powerful idea of using one’s own army of immune cells against tumorous ones, the field has endured countless setbacks….” It progressed, regressed, and emerged triumphant “… from a promising hypothesis to a new modality of cancer treatment of today.”1

Recently, Personalized Medicine in Oncology had the privilege of discussing the journey of immuno-oncology with one of its pioneers: Dr James Allison of The University of Texas MD Anderson Cancer Center. Here he shares his experience and insights into how this paradigm shift in cancer management is moving forward, what caused its “overnight success” after 30 years of failure, and how it relates to the premises of personalized medicine. The following highlights just begin to shed light on the results of this persistent quest to crack the code of the immune system on behalf of cancer treatment, helping practicing oncologists and indeed the entire spectrum of stakeholders to appreciate the profound changes in expectations for cancer care that this portends.

Dr Allison is professor and chair of The University of Texas MD Anderson Cancer Center Department of Immunology in the Division of Basic Science Research. He directs MD Anderson’s Immunology Platform and is Deputy Director of the David H. Koch Center for Applied Research in Genitourinary Cancers, Department of Genitourinary Medical Oncology Research.

PMO Thank you for talking with us about the phenomenon of immuno-oncology: a 30-year “overnight success story.” Can you describe its plight and how it finally turned the corner?

Dr Allison There are basically 3 reasons why people for decades have been interested in manipulating the immune system as a means of treating cancer. The first one is specificity. T cells, the part of the immune system we’re talking about today, largely recognize peptides, short bits of proteins that are presented on the surface of cells in the context of structures called MAC antigens. The immune system is trained to recognize self, so if it’s self-peptides, nothing happens, but if a peptide is recognized as foreign, then the T cell will kill it.

What’s really apparent now due to a lot of recent work is that not only do T cells recognize what used to be called cancer differentiation antigens, but recent evidence suggests that what T cells are recognizing are important molecules called neoantigens that are generated by mutations inherent to the cancer process itself. Cancer is really a disease of genomic instability and results in accumulation of mutations that eventually result in drivers or oncogenes that cause a cell to become a cancer cell. Along the way there are a lot of mutations that really don’t have much functionality in the tumor cell.

The tumor biologists pretty much ignored them for a long time, but those are what the immune system can recognize. Bert Vogelstein and I proposed it about 8 years ago. But now there’s emerging data from a number of labs all over the world that report that’s what T cells really see in cancer cells: these mutant peptides.

So that being the case, the specificity of T cells is for the process that causes cancer itself and not for any particular type of cancer. The second thing about the immune system is that it has memory. Once you generate a family of T cells that can recognize an individual peptide, their descendants will most likely stay with you for the rest of your life. If you get a flu vaccine or something, you’re protected for a time until the flu changes, but for a lot of childhood diseases the immunity lasts a long time. No drug can do that. You give the drug and it’s gone after a while, and if it hasn’t taken everything out, it hasn’t worked.

The third thing is adaptability. The immune system is even more adaptable than the cancer cells. The problem with the genomically targeted therapies is the adaptability of the cancer cell. You target one molecule that’s causing the cancer and you can kill all the cells that express that molecule either by mutating that molecule, or by selecting for variants in other molecules that result in activation of a different pathway. The immune system is not fazed by such mutations because it is potentially capable of mounting responses to the products of “passenger” mutations not related to the ones that drive the tumor cells, but also to new antigens generated by the new driver mutations that allow the tumor to escape the original drug and result in resistance.

But as the tumor cell changes, the immune system is designed to protect you against virtually anything nature can throw at it, including tumor antigens, mutant peptides, and so there’s really no resistance to immunotherapy. There are ways theoretically that tumor cells can protect themselves, but if a patient responds once, they tend to respond again if they need to.

For these reasons people have been trying to employ immunotherapy for decades, and we’ve known what tumor antigens are for 15 to 20 years now. We know a lot about how the peptides are processed and present on the surface of cells. We know about the features of the innate immune system that are required to initiate an immune response. But yet attempts to really mobilize the immune system to protect against cancer have really not been very successful.

I think that’s because we didn’t really recognize how complicated the process is. In the late ’80s when we started our work in this area, it was thought that recognition of antigens was by itself enough. It was thought that recognition of antigens by the T-cell antigen receptor was by itself enough to activate a T cell. But then by the late ’80s it became evident that wasn’t the case, that there was another molecule called a costimulatory molecule that had to be engaged. This costimulatory receptor, called CD28, had to be engaged by structures that are really found only on antigen-presenting cells, so you really need 2 signals: an antigen-specific one and then this costimulatory signal that’s not specific to an antigen. So it’s sort of like the ignition switch and the gas pedal.

There was a molecule called CTLA-4 that had been identified in the early ’90s in a lab headed by Pierre Goldstein in France. All that was known about it was that it was expressed in activated T cells but not in resting T cells. But it was highly homologous, structurally very much like CD28, and it was shown that it bound to the same molecules and had the same counterreceptors as CD28. So it was suggested it was another costimulatory molecule. You have to stop immune responses or else they’ll just take over your body, because T cells divide very fast once they are activated. So at the time, it was thought that the T cells stopped by a process called activation-induced cell death – that after a time they just basically burned themselves out and died.

But when we started studying CTLA-4, as did a colleague named Jeffrey Bluestone who was at the University of Chicago at that time, we proposed in the mid ’90s that actually CTLA-4 was not another costimulatory molecule, it was an inhibitory molecule, and its expression was turned on when you tried to activate a T cell.

This was kind of controversial for a while, but finally the data were overwhelming and began to be accepted. But when we realized this, I had the notion that maybe what was going on with the therapeutic vaccine strategy was that when you try to vaccinate, you give the “on” signal, but also as a result of that, this initiates the expression of CTLA-4, which is the brakes, and that the harder you try to give the “on” signal you get an “off” signal, and after a while the cells can’t respond because they just accumulate the CTLA-4, which keeps them from dividing.

That was the idea. The idea was, well, if you just take the brakes off by blocking CTLA-4, then T cells could keep running for a while longer and maybe have enough time to mop up more tumor cells. We showed that in mice and published it. It was first presented in 1995 and then published again in 1996, showing that just injection of monoclonal antibodies specific for this one molecule, CTLA-4, was enough to unleash the immune system in mice to allow it to cause rejection of tumors and also the induction of lifelong immunity to the same tumor.

That was what led to production. We teamed up with a company called Medarex. They made the first fully human monoclonal antibody to CTLA-4 and took it into clinical trials about 2001 and saw responses in metastatic melanoma, even in the phase 1 trial. By 2011, after a number of phase 1/2 trials and finally a randomized placebo-controlled trial in metastatic melanoma, it was shown to increase survival by 4 months.

It was the first drug of any kind that had ever done that in a randomized clinical trial. It was 4 months, and that was enough to get it approved by the FDA. I was shocked because that was what a lot of people paid attention to.

But the really interesting thing about those results was that very often the tumors progressed for a time before they started regressing. The usual way of evaluating cancer drugs is progression-free survival. But tumors have to shrink. The definition of resistance is when the tumor grows in the face of the drug. That happened in many patients and in all of our mice. In every mouse experiment we did, the tumors grew before they went away. By those criteria that we used in the clinic, all of our mouse experiments were failures, even though in many experiments we cured 95% to 100% of the mice.

In any event, because of those complications, Meda­rex and Bristol-Myers Squibb changed the end point from progression-free survival to overall survival, so the trial took almost 5 years to complete because you just had to measure survival before it could be unblinded. But when the trial was finally unblinded, not only was that 4-month increase in median survival achieved, but there was a tail where at about 2.5 years to 3 years, the survival curve flattened out at 20% and stayed there for 3.5 years, 4 years, 4.5, 5 years. Recently, there was an almost 5000-patient retrospective study of people that had been treated, and what it showed is that after about 3 years patients really don’t succumb to cancer, they succumb to something else, and that’s after a single treatment.

Then another molecule called PD-1 was identified. It was another immune checkpoint or inhibitory molecule. A completely independent system, completely different molecule, ligands.

After the success of CTLA-4, there was a big rush to get PD-1 evaluated, and it turned out that it did the same thing. It’s different in some ways. Its ligands are completely different, and one of them can be expressed on tumors, whereas the ligands of CTLA-4 and CD28 are expressed only on antigen-presenting cells.

The PD-1 data were presented in 2012 at ASCO. The PD-1 results showed much the same as CTLA-4. There were a lot of responders: 20% to 30% of patients in many kinds of cancer.

It’s interesting to note that neither of these drugs target the cancer cell at all. They target the immune system, so it’s a completely different way of going about treating cancer, because you basically ignore the cancer.

PMO That’s just amazing, the entire reorientation of the target. What are you after?

Dr Allison We’re not after killing the tumor cells, we’re after unleashing the immune system. My assumption when I started this is taken basically from the immune system’s viewpoint: cancer is cancer. The immune system doesn’t really know or care if it’s melanoma or kidney cancer or lung cancer. It just knows there’s stuff in the cell that ought not to be there.

When PD-1 came along, what became apparent is that some patients who didn’t respond to anti–CTLA-4 did respond to anti–PD-1 and vice versa. Again, PD-1 works differently. Instead of interfering with the costimulatory molecule, what it does is recruit a phosphatase. It interferes with antigen receptor signaling. In other words, rather than taking the brakes off, it interferes with the ignition switch, if you will. But that being the case, we reasoned that if you put them together, they might work…they’re not redundant. If you put them together, it might be additive. And that proved to be the case.

In data presented at ASCO in 2013, it was a test of the antibodies in combination in metastatic melanoma, and fully 50% of the patients showed tumor shrinkage. Actually 65% showed some tumor shrinkage; 50% shrunk enough to be objective responses.

There was a follow-up this year at ASCO, a survival study, because those data matured in the combination trial enough to really look at survival. It’s a small-numbers study, but the 1-year survival with the optimum doses of each antibody was 94%, and the 2-year survival was 88%, so almost 9 of 10 patients were alive 2 years after treatment.

Remember, the cutoff with anti–CTLA-4 is 3 years. Patients that make it 3 years are pretty much done. So that number is probably going to drop. It wasn’t a randomized study. It could have been selection, but even so, the number that’s falling out is going to be high.

Anyway, that’s the excitement, and although most of the studies are being done in metastatic melanoma, data are coming along also in lung cancer and in kidney cancer that are very similar, so now the effort is to bring that combination of drugs to many other types of cancer.
Bristol-Myers Squibb is sponsoring a single trial that we’re doing here that targets pancreatic cancer, triple-negative breast cancer, gastric cancer, and small cell lung cancer. It’s too early to say what’s going to happen, but there’s no reason why there shouldn’t be some success.

PMO That’s extremely promising. Thank you for a very complete answer to the chronology of its pathway to becoming 2013’s breakthrough of the year. With all that potent effect, what is its adverse event profile?

Dr Allison The hope was that the immune system is so precise and specific that there would be no collateral damage, if you will, and all the things would just affect the tumor cells. In fact, in the mouse models that we did, we never saw any adverse events except occasional depigmentation in the mice with melanoma, which was expected. And in the monkey toxicity studies at very high doses, there was also no toxicity observed.

But early on one of the concerns was that there are adverse events in people, particularly with anti–CTLA-4: really severe diarrhea. The thing about this is you’re taking the brakes off the immune system. It’s so powerful that it’s too much to hope that that can be done without any bad consequences at all.

That was pretty frightening. With time, algorithms were developed for dealing with this. They involve just systemic administration of steroids. Some version of that can almost always deal with the adverse events, and then you can taper the patients off of the steroids, as the inflammatory responses that are causing the problems don’t come back. So it’s not autoimmunity, and that’s very important, because a lot of people say autoimmunity accompanies this. That’s not really true. That’s very, very rare.

There are other problems to contend with: colitis, uveitis, hepatitis, but they’re almost all reversible. As a consequence of developing algorithms to deal with these adverse events, over time their frequency and severity have really decreased. But they aren’t gone. Patients have to be watched.

PMO What is the patients’ health-related quality of life as they’re taking this? How are they feeling?

Dr Allison Initially there’s tiredness. Once the diarrhea or the rash goes away, I think it’s good. I know a number of patients that were treated in the early days who were uncomfortable while they were being treated, but later on they were fine. I think it’s extremely better than the adverse events associated with conventional chemotherapies. It’s not even comparable.

PMO What do you think the long-term overall survival potential is with this approach to treatment?

Dr Allison As I said, in melanoma – and here I’m talking about intent to treat, just some patient that walks in the door that’s healthy enough to receive treatment – I think with metastatic melanoma it’s going to be perhaps as high as 70%. Right now, at 2 years it’s almost 90%. Again, with anti–CTLA-4, with ipilimumab by itself, the long-term survival was roughly 20%. And that’s a study of 5000 patients.

PMO Are we closing in on patients not dying of cancer with this treatment?

Dr Allison Well, I’m not unbiased, but I would say I think we will be able to deal with the majority of patients with metastatic melanoma. I see no inherent reason why that couldn’t be true of other tumor types as well. Lung is going to be the big target, and that’s in process right now. I think there’s reason for a lot of optimism.

PMO Can you summarize the basic distinctions between targeted conventional and genomically targeted therapies versus immune therapies?

Dr Allison When you give conventional therapies that kill the tumor cells, the tumors generally start shrinking right away. That’s what the cytotoxic drugs are designed to do, to kill the cell. The same is true for most of the genomically targeted therapies. They shut off the motor that’s driving the cell to be a cancer cell, and if anything keeps growing or if a new tumor forms in the face of it, that drug is not going to stop it.

On the other hand, with immune therapies you’re not killing the tumor cell directly. What you’re doing is unleashing the immune response. There are lots of reasons why the tumors may get larger. One is it may just take time for the immune system to mobilize in sufficient numbers of T cells to actually kill the tumor cells.

A second thing that has actually been observed is that the tumors swell because they’re filling up with T cells, because T cells go in there and it’s hand-to-hand combat, if you will. When the tumors get bigger, you stick a needle in them, and what you find are activated T cells and a bunch of carcasses of tumor cells. Very often there are no live tumor cells in the biopsy even though the tumor’s gotten bigger. And the same is true if you look at CT scans, and this explains, I think, why the objective response underestimates the survival response. Again, this isn’t like something that goes in and kills the tumor cells and they just go away. This is warfare.

The T cells, while they’re killing the tumor cells, are spitting out inflammatory things like gamma interferon and tumor necrosis factor-alpha (TNF-?), and the cytotoxic T cells are putting out granzyme. So it’s killing cells. It’s low-class warfare, producing a lot of scar tissue that hangs around that the CT scan cannot identify…it simply shows up as a spot on a CT scan.

Allow me to speculate on the other thing that may happen, though I don’t know if it’s been well documented. It’s possible that there is a small sphere of residual tumor cells being kept in check by the immune system, so there could even be some residual tumor cells being held back in a process called equilibrium that Bob Schreiber at Washington University has talked about. The question is whether the patient has no apparent disease, because many of those patients do have dots on CT scans. But we’re not treating the CT scan, we’re treating the patient. Being alive is more important than having a clean scan, I think.

PMO It’s eminently rational that you would have these carcasses, this debris. It’s the warfare aspect of this. This is a different kind of war that’s going on.

Dr Allison Again, we’re not directly killing the tumor cells. We tried to do that, but unless you have the immune system involved when you treat with a cytotoxic, whether targeted therapy or otherwise, unless you take out every tumor cell, it’s going to come back. It may be in some cases you don’t have to, because the immune system will catch on and keep it in check, but the whole method of engaging cancer is different.

PMO It certainly is. What is the status of concomitant treatment modalities with this approach? I’m wondering how clear this is going to be at the clinical level, at the practicing level. In other words, when to use it. There arises the matter of the range of other medications to use or not to use. What gets displaced, what doesn’t get displaced in the treatment?

Dr Allison I think with time a few things will become apparent. One is in addition to CTLA-4/PD-1, there are more of these molecules. A few of them are in clinical development. So it may be that it takes a combination of blocking, as the example of blocking CTLA-4 and blocking PD-1 together at the same time is much more effective than either one independently. So it may be that in that last 20% of melanoma or whatever, there may be a third molecule or a third and fourth molecule that’ll work. But maybe that also differs in different tumor types, particularly since many of these T-cell inhibitory molecules have things on them that can bind them, and so that could play a role. I don’t see why it shouldn’t. I mean, this really is personalized medicine, right? It’s your immune system. It’s educated to recognize your normal cells and protect your cells against something, so it’s personalized medicine of the highest order.

It would be ridiculous to say for that reason that it’s going to replace targeted therapies, but there is one place where I think it will continue to be used. We’ve been talking up until now about just giving these immune checkpoint blockers without anything else. These incredible responses have been seen in many kinds of cancer. My own feeling, understanding the mechanism, is that they all work on T cells that have been primed, and so when you give these by themselves, there’s got to be tumor death coming from someplace. It may just be that every now and then some tumor lesion gets so big that it cuts off food supply or oxygen supply or something and starts dying, and that causes inflammation and alerts the immune system to come in and respond to bits of tumor that are being dumped out. But if you can make that happen, if you could kill tumor cells, I think you could start everything faster and get responses in a much higher fraction of patients.

I think if you’ve got non–small cell lung cancer (NSCLC) with an epidermal growth factor receptor (EGFR) mutation that’s driving it, which is very common, and there are drugs that can work through that mutant EGFR that cause amazing tumor shrinkage very quickly but only last a few months for the reasons I mentioned earlier, the tumor cells can adapt and become resistant. But you could really amplify the effectiveness of those drugs, I think, and we’ve done this a bit in mice, by initially coming in with that genomically targeted agent against NSCLC, for example – something that kills, that works through that molecule and kills tumor cells and then immediately follow that with the immune therapies.

Of course, in the case of metastatic melanoma, vemurafenib is a drug that targets the BRAF mutation that’s found in about 60% of people with melanoma. Again, it’s short-lived, but already the response rate with a combination of anti–CTLA-4 and anti–PD-1 is higher than the response rate to vemurafenib.

I think some of these targeted therapies are going to fade.

PMO When you treat patients for autoimmune adverse events of the new immuno-oncology drugs, does the treatment, like steroids, nullify the positive effects of the immunotherapy?

Dr Allison The treatment with steroids to alleviate the adverse events surprisingly enough does not interfere with the antitumor effects, at least if it occurs late. Let’s turn that around. There are still plenty of responses in patients that have received steroids. It may be because the cells that are responding to the tumor cells may already have been primed. Memory T cells are resistant to the effects of steroids, so it may just be that. Anyway, they don’t seem to interfere much.

PMO Why is retreatment with immunotherapy often effective, but retreatment with chemotherapy and nonimmunotherapy, targeted therapy, seldom effective?

Dr Allison The reason that tumor cells become resistant or don’t respond after they become resistant to chemotherapy and targeted therapies is the tumor cell has figured out how to evade.

PMO That memory you talked about before?

Dr Allison In the immune system you’ve got memory, but also if the tumor changes by a process involving generation of new mutations, that makes the initial drug ineffective. The immune system can respond to the mutation itself, whatever it’s in. The immune system is designed to protect you against anything nature throws at it.

The antigen receptors on T cells are generated by a fairly random process and then selected to not attack anything that’s in your normal cells. It’s pretty effective, because other immune diseases are pretty rare. It’s been estimated that as many as 1015 different antigen receptors can be generated by the immune system. That’s billions of times more than the total number of cells in anybody’s body, and it can change because it’s an ongoing process. You’ve got probably 100 million different T-cell receptors in your body. A set of them gets activated to a tumor and then the tumor changes and something new comes up, and some different T cells will just move in. The immune system is even more plastic, even more adaptable than the tumors.

PMO Why do certain cancer cells survive initial treatment?

Dr Allison It’s possible that there are certain molecules that are involved in expression of antigens on the cell surface. If there are mutations in those, then the tumor cells lose the ability to present the targets that T cells would recognize. That so far has not proven to be an issue in these sorts of treatments, as far as I know. You can select for mutations, for example, that would make a tumor cell resistant to killing by gamma interferon. Gamma interferon binds to receptors on cells and activates a chain of signals down to the cell that says, “Kill it.” And if you have a mutation anywhere in the tumor cell, anywhere in that pathway, then that mechanism of killing, for example, becomes ineffective. That has been observed, but that’s also rare.

But those are describing ways that individual tumor cells might escape. I think a bigger question is, why do some patients with cancer respond and some don’t? We don’t really have a good answer for that. It may be that some tumor cells express multiple checkpoint molecules, and you have to block them all. Or maybe the T cells just aren’t being effectively primed. If that’s the case, then you can take care of that by just adding a cytotoxic genomically targeted drug along with immunotherapy.

I think the answer to the question of why some patients don’t respond really lies in combination therapies. I’m an optimist and not unbiased about the power of the immune system, but I think our experience in mice has shown that we have rarely, if ever, found a tumor model that we couldn’t selectively attack by using the appropriate combination of agents.

PMO Introductory statements to an article in the November 16, 2012, issue of Nature Immunology state: “Immunology beats cancer: a blueprint for successful translation.” It goes on to say, “Immunology offers an unprecedented opportunity for the science-driven development of therapeutics. The successes of antibodies to the immunomodulatory receptor CTLA-4 and blockade of the immunoinhibitory receptor PD-1 in cancer immunotherapy, from gene discovery to patient benefit, have created a paradigm for driving such endeavors.” I’d like your thoughts on whether that is going to clear things up for everybody? Is this a good way to encapsulate the concept, or would you state it differently?

Dr Allison I think that’s right on. Let’s back up a bit – early on, people were just saying, let’s get the immune system, let’s do this, we’ll activate it this way, we’ll activate it that way, we’ll activate it the other way, let’s throw IL-2 at the tumors, whatever. None of those early attempts at immunotherapy were really based on understanding of mechanisms, whereas CTLA-4 is a case in point. Nobody really understood what it did until Jeff Bluestone’s lab and mine said this is a negative regulator, and once we saw that, we asked, are there really enhanced immune responses if you take the brakes off? So we said okay, here’s 1 molecule. Let’s take it off. Actually, I was astounded when we saw what happened in mice with just 1 molecule blocked, and it was enough in many, many mouse models to get tumors to shrink and get long-lived immunity. So it came out of mechanistic studies. Nobody saying “Let’s treat cancer” would ever have taken a look at CTLA-4. It just wouldn’t happen then because the knowledge wasn’t there.

There was such a rush. In the clinic people didn’t really pay all that much attention to mechanism, so mouse models in my lab and other ones were doing it, and that provided the rational basis. I’m sure PD-1 blockade and CTLA-4 blockade together would have come about anyway, but we provided a mechanistic reason for understanding why those two were so good together…and that was from mouse studies.

I think in the future what we’ve got to do is really study patients. So that’s what we’re doing here in the immunotherapy platform as part of the Moon Shots Program at MD Anderson. A colleague of mine at MD Anderson named Dr Padmanee Sharma is a clinical oncologist that specializes in urinary cancers. And she was interested not so much in clinical response, although that’s what you’re after, but in understanding the cellular molecular mechanism. So what she pioneered is something that we’re following now as we can, and that’s treating patients who are going to surgery. It’s hard to learn a lot from a biopsy specimen, although it’s better than nothing. But if you really want to understand what’s happening and get enough T cells out to understand not only their phenotype but their function, you must treat a patient and then get the whole organ with the normal tissue, the tumor tissue, etc, and see the impact of the drug. This is what she’s done now in bladder cancer and in prostate cancer.

When you do these things along with mouse studies, you can get data from the patients that allow you to generate hypotheses, and you can then test them using the mouse models where you have genetically modified mice in which you think a given molecule is important, so you can see how well the drug works when that molecule isn’t around.

Here’s a case in point. Dr Sharma treated bladder and prostate cancer patients with anti–CTLA-4, and these were patients with localized disease, and these are small trials, 10- to 12-patient trials. Every single patient showed a several-fold elevation of a kind of T cell called an ICOS-positive T cell – which stands for inducible costimulator. ICOS is another molecule in the CD28, CTLA-4 family, but it’s distinct and has its own ligands and everything. At any rate, you wouldn’t expect to find T cells expressing that in tumors, because they make the wrong kind of cytokines. You want a T cell in a tumor to make gamma interferon and TNF-?. This happened in all her patients, and so it serves as a good pharmacodynamic model for CTLA-4 therapy, because if you give the drug and you don’t see these cells, that means it’s probably not going to work and maybe something’s wrong with the patient’s immune system.

We did a collaborative study – her studies were done in patients with localized disease. But when I was still at Memorial Sloan Kettering with Dr Jedd Wolchok, Dr Sharma told us about this, so then we looked at melanoma patients. It turned out you could divide melanoma patients into patients in which ICOS increased and stayed increased for the whole 3 months of therapy.

Standard anti–CTLA-4 treatment is 4 doses at 3-week intervals. The frequency of T cells expressing the ICOS molecule went up at least 2-fold and stayed there for the 3 months of treatment, and then there were others that didn’t. So if you divided those and asked how long they lived, it turns out the median survival of the patients in whom that frequency did not increase, or only went up a little bit and then came back down, was about 8 months. In the patients in whom it went up and stayed up for 12 weeks, the median survival was 20 months. That’s not going to tell you who’s going to respond and who’s not going to respond prior to treatment, because this only happens after treatment. It may tell you who’s not going to respond, though, because if it doesn’t increase, the chances of anything happening are not good in that study. There need to be more numbers.

That leads to the hypothesis that ICOS might be important in the therapeutic effect of anti–CTLA-4. So you can make that hypothesis, except in clinical data you could also say well, maybe the patients in whom it didn’t go up, they’ve just been treated so much with chemotherapies and whatever that their immune system was beaten up and was nonresponsive.

So you can’t answer that question, you can’t distinguish between those 2 possibilities really in patients. What Dr Sharma did with our help was to study the efficacy of anti–CTLA-4 in mice that genetically lacked ICOS. What she found is that the anti–CTLA-4 treatment only worked about half as well or less in mice that lacked ICOS. So that now gives us not only a pharmacodynamic marker but tells us that this is important in the function. Now ICOS is a costimulatory molecule, so a signal through it will help an immune response.

So this experience raises the possibility that if you block CTLA-4, then you see these ICOS-positive cells appear and you didn’t give a signal through the ICOS molecule, the anti–CTLA-4 might work a lot better; and indeed that’s the case. In mice we showed that if you give an ICOS signal while you’re blocking CTLA-4, it increases the efficacy by at least 4-fold. We recently published this in the Journal of Experimental Medicine.

In any event, this can lead to a whole new way of treating cancer where you combine blocking negative signals with giving positive signals, all targeting the immune cells, not the cancer cell. That’s where we’re going with this research: to really understand the mechanism in the patients. Confirm it in mouse models and then take you to the next level. That sort of paradigm is what I think that sentence that you read is talking about, about developing a drug that targets ICOS.

PMO You are bringing this new strategy into focus, not only for the researchers, but the oncologists in the trenches too.

Dr Allison What’s apparent now, and I think is probably the reason for this discussion, is that if you go to a major medical center and you have metastatic melanoma, you’re most certainly going to get anti–CTLA-4, or ipilimumab, and increasingly 1 of the 6 anti–PD-1 targeting agents that are out there. It’s really slow to make it into the community.

PMO Yes, the diffusion of this knowledge is critical.

Dr Allison Right, and I think in part that’s because of lack of understanding of the adverse events and others.

PMO Yes, addressing the full array of facets of this thing called immuno-oncology will require extensive drilling down into its components and its value. We could not have asked for a better introduction into the topic than what you have provided us here today. Thank you for this outstanding contribution to our knowledge of this paradigm shift in the making.

Dr Allison It was my pleasure.


  1. Avery M. Immuno Oncology: A 30-Year Overnight Success Story. BIOtechNOW. February 20, 2014. www.biotech-now.org/health/2014/02/immuno-oncology-a-30-year-overnight-success-story#. Accessed August 24, 2014.
Uncategorized - October 17, 2014

Combo of BRAF and MEK Inhibitors Improves Survival in Advanced Melanoma

Combination therapy with a BRAF and a MEK inhibitor improves outcomes in advanced BRAF-positive melanoma, according to 2 phase 3 studies presented at the 2014 ESMO Congress. These studies support the hypothesis that inhibition of both BRAF and MEK will improve survival in melanoma by overcoming the mechanism of acquired [ Read More ]

Uncategorized - October 17, 2014

HER2-Derived Vaccine Cuts Recurrences in High-Risk Breast Cancer

A vaccine derived from the GP2 peptide was safe and reduced the rate of breast cancer recurrence in women with high-risk breast cancer in a phase 2 clinical trial. The vaccine was particularly effective in reducing recurrence in women with HER2 overexpression, reported Elizabeth Mittendorf, MD, PhD, at the 2014 [ Read More ]