Magazine Article | October 2, 2014

Combination Cancer Immunotherapy — A Virtual Roundtable: Part II

Source: Life Science Leader
wayne koberstein

By Wayne Koberstein, Executive Editor, Life Science Leader magazine
Follow Me On Twitter @WayneKoberstein

Llew Keltner, M.D., Ph.D., Roundtable Moderator

A series on the challenges and opportunities of using new agents to rally the immune system against cancer.

PART TWO: Key Opinion Leaders Benchmark the Science

Drugs that target the immune system, not the tumor, may have entered the long war on cancer at last. Although some still dispute the validity of cancer immunotherapy, others are charging ahead. Results of large Phase 2 trials showing durable responses from a new class of drugs called checkpoint inhibitors, along with progress on the vaccine/immunostimulator front, have fired up supporters and attracted new interest from former doubters.

Our virtual roundtable, a compilation of responses to questions from key experts and players in the field, considers the ramifications of a growing consensus that using cancer immunotherapies in combination, rather than as single agents, will be essential to have maximal effect for patients. In a matter of years, many cancer immunotherapies will be available, many combinations will be possible, and the choice of combinations will be quite challenging from a clinical, regulatory, and reimbursement perspective. Biomarkers and companion diagnostics may also play a big role in guiding the way, as will a deepening understanding of immunotherapy mechanisms and cancer response. But who should decide how combinations are tailored and delivered to individual patients, and on what grounds should they base those decisions?

Whether by written or in-person response, our panelists all tackle the same set of issues in combination cancer immunotherapy — in this second part, the deeper issues involved in therapeutic choices — by physicians, payors, and regulators. Their answers to our questions, along with further queries and interjections by the moderator, follow.

QUESTION: What are the essential constituents of any cancer immunotherapy combination?

KELTNER (Moderator): Even as this discussion goes on, new targets for immunotherapies continue to proliferate — from new checkpoint inhibitors, to immunostimulators, to entirely new classes. Thus, the discussion of the emerging alternatives and their respective roles is bound to show both agreement and divergence among the panelists.

SZNOL: We don’t yet have the answers to this question. In animal models, you see strong activity when you combine two checkpoint inhibitors, such as PD-1 and TIM3, PD-1 and LAG3, or PD-1 and CTLA-4, and very impressive data with blockade of one checkpoint, such as PD-1, together with one co-stimulatory agent, such as CD-137 or OX-40. Even if we developed biomarkers, we might still have to make guesses in the clinic. A given patient, based on a biopsy, might need two checkpoint inhibitors, and if the patient does not respond, we might add another agent.

Inhibiting checkpoints is not always necessary. Interleuken-2 works without inhibiting any checkpoints in a subset of patients. Why does it work at all? We don’t know, but it does not require PD-1 or CTLA-4 blockade or anything else. How does CTLA-4 work without blocking PD-1 when the T cells that infiltrate the tumor upregulate PD-1 or are subject to PD-1/PD-L1 suppression? Why does cell therapy work? When you give the engineered cells back, they upregulate PD-1. It is an analog, not a digital, system, and if you drive your system enough, you may overcome checkpoint immunosuppression.

URBA: Combinations one day will be developed based on a personalized approach addressing tumor and host characteristics. Generically, I would include an antigen-priming or boosting step. This would most likely be a vaccine, although it could be an intratumoral injection of a lytic virus, cytokine, bacteria, chemotherapy, or radiotherapy administration. One would want to ensure the proper local milieu, so some form of adjuvant tickling the appropriate receptors might be included. Future combinations are likely to include a checkpoint inhibitor and perhaps also an immunostimulatory component. I could see inclusion of specific strategies to alter potential inhibitory processes such as elimination of T regs, neutralizing negative factors (IDO, arginase, TGFbeta), and modification of macrophages.

GRETEN: Vaccines are a bit more complicated because it’s very difficult to develop one to be used off the shelf, rather than an individualized vaccine for every single person, as in prostate cancer with the Provenge vaccine, which requires a very costly, time-intensive, and difficult procedure. My basic idea is, why not just use treatments that you do anyhow, such as chemotherapy or ablation, optimize them, and use them as a kind of vaccine to induce a response, then combine them with immunotherapy to achieve better responses?

WOLCHOK: There will be a very broad list of different combinations — combinations of immunotherapies with other immunotherapies; with regional therapies such as oncolytic viruses; locally destructive techniques such as radio frequency ablation, cryosurgery, and radiation therapy; and oncogenicpathway targeted therapies. It will still require a lot of investigation, but we have already established a firm foothold for immunotherapy in producing a durable response for patients with multiple kinds of malignancies.

SZNOL: This is not a war between immune therapy and targeted therapy or chemotherapy. We can figure out a way to use all those approaches to treat patients more effectively. In melanoma, we treated patients who failed to respond to immune therapy with dabrafenib or trametinib and got a substantial clinical benefit.

QUESTION: Do you believe PD-1/PD-L1 therapy, or some other approach, will be the “backbone” of cancer immunotherapy combinations? Or will consensus on a hierarchy of therapies evolve along with the growth of scientific understanding and use of biomarkers?

KELTNER: The exploration of overall safety profiles in combination immunotherapies may be just as important as the overall efficacy of combinations, favoring immunotherapy mechanisms with inherently low toxicity — such as newer-generation, off-the-shelf cancer vaccines — in selection of combinations.

FONG: With other immunotherapies coming down the pipe, it will be important to see their clinical activity, but just as important, their toxicities. If you have a relatively nontoxic treatment, it really makes it that much easier to layer things on top of it, whereas if something already has a lot of side effects, we must try to navigate around them in any combination. Even if the immunotherapy only had the same efficacy as chemotherapy regimens, but had no toxicity, it could become the backbone of cancer treatment.

Other approaches are very complementary — for example, a cancer vaccine such as Provenge or another prostate cancer vaccine called PROSTVAC-VF, which is a recombinant viral vaccine. These could stimulate immune responses in a patient.

WOLCHOK: Safety profiles definitely have to be considered, on an agent-by-agent basis. There are some situations where we’re going to need to use therapies that target the tumor to release antigens and alter the microenvironment, and in fact some of them may even have immunomodulatory effects as well.

SZNOL: At least, with immunotherapies, you may have a different return for the adverse effects — a better chance of being able to live the rest of your life without having to take a drug chronically and going back to a fairly normal quality of life. But these drugs can produce permanent side effects: Some people lose their pituitary and have to take a prednisone or thyroid pill for the rest of their life; and if they get sick without a pituitary, they need to get replacement hormone therapy. So patients don’t go back completely to normal in some cases, but it’s pretty close to normal.

ALLISON: When CTLA-4 was new, there were a lot of lessons learned; it was a whole new concept, a new way of treating cancer, and it was terra incognita from the moment the field opened. The adverse events were not expected. There had been no toxicity in mice, no toxicity in monkeys, so they first popped up in humans. It took awhile to get a handle on it, but now there is a dosing algorithm, so when PD-1 came along, everybody was ready. Immunotherapy probably is inherently less toxic than chemo, but it can have severe toxicities. People died before we had the algorithm. There is also an indication that some people who only receive anti-PD-1 recur more often than people who just get anti-CTLA-4, so the backbone will probably be those two in combination along with a third one — not anti-PD-1 by itself.

SHARMA: Anti-PD-1 is also administered for a much longer period of time as compared to anti-CTLA-4. Anti-PD-1 is given for a period of two years of therapy in the current clinical trials, while anti- CTLA-4 was FDA-approved to be given over a period of about three months as four doses of the antibody, with each dose being given three weeks apart, and then you’re finished with the treatment. Anti-PD-1 may become a backbone where other treatments can be given during the two-year period of anti-PD-1 therapy.

ALLISON: And 20 percent of those people treated with anti-CTLA-4 are still alive after a decade or more. With anti-PD-1, survival data is still early and patients are still under observation.

BERINSTEIN: As is the trend in cancer therapy, there is a push to make cancer immunotherapy more personalized. Anti-PD-1 or PD-L1 therapies only have activity in a subset of patients, which may be disease-specific. It is likely that other immune inhibitory pathways are more dominant in the nonresponders. In the future, it may be possible to identify more dominant pathways by interrogating each patient’s tumor individually. In this way, immune modulation can be prescribed for each patient in a more logical and patient-specific fashion. However, in the immediate/short term, it is likely that the hierarchies of appropriate combination treatments will develop based on welldesigned and controlled pivotal clinical trials, and these will inform the best combination therapies moving forward for a variety of cancer types. It is also possible that a complex combination therapy that will likely include the PD-1/ PD-L1 pathway will be able to provide a safe and effective approach to treating cancer without seeking a personalized approach. The field is not there yet.

KELTNER: Proof that “we’re not there yet” already exists in data from the ASCO presentation given by Pam Sharma entitled “PD-1/PD-L1 Inhibition: Identifying Relevant Biomarkers,” as she discussed ASCO abstract 5012 composed of an anti-PD-1 clinical trial and abstract 5011 composed of an anti-PD-L1 clinical trial. The trial targeted patients whose tumors expressed the PD-L1 antigen, and the responses in those patients were dramatic. But some PD-L1-negative patients also responded, casting doubt on the usefulness of PD-L1 as a biomarker for patient-treatment selection. The field of prostate cancer biopsies has clearly demonstrated the danger of making negative determinations based on single biopsies, just as the presence of a target on some cells in a tumor, at a fixed point in time, cannot guarantee all of the tumor cells will respond well to a targeted drug. Such challenges do not bode well for biomarkers to guide patient-by-patient selection of immunotherapies.

In the genomic era, we all got caught up in the idea that you look for the mutation, then you look for the drug to target the mutation. But that only addresses the static state — the mutated gene itself is static, but the immune system is dynamic. If a patient’s tumor biopsy is PD-L1-negative today, it could be PD-L1- positive tomorrow. You might think the PD-L1-negative patient will not respond to PD-L1 or PD-1 therapy, but that is not true. Some of them do respond, and the reason may be not that the target is absent, but that the target is dynamic and changes over time based on the presence of other immunologic factors such as IFNgamma, and you cannot label “positive” or “negative” on a single biopsy.

ALLISON: T cell infiltration produces gamma interferon, which upregulates PD-L1 in the tumor, and if you biopsy the wrong portion of the tumor where there is not an effective immune response, the tumor is not going to express PD-L1. As a matter of fact, if there really is no PD-L1, it means the immune system is quiescent in the patient, so that ought to be reflected in low T cell counts and all the immune factors. Instead, people just jump to one biomarker. When you give anti-CTLA-4, you are going to see a lot of PD-L1-negative patients get these ICOS-positive cells that make gamma interferon; they’re going to make the tumor cells and other cells in the microenvironment express PD-L1, which provides the perfect opportunity to give anti-PD-1 or anti-PD-L1 therapy. But you couldn’t have known that the tumor cells would express PD-L1 based on the initial biopsy.


"Immunotherapy has taught us we can’t predict which cancers are going to actually respond."

LAWRENCE FONG,M.D., Professor, Department of Medicine (Hematology/Oncology), UCSF

WOLCHOK: We would love to have a perfect predictive biomarker for these treatments. But the truth is, we don’t. The PD-L1 tumor expression has been considered very carefully over the past few years since the initial data began to emerge, and it did seem to identify patients who are more likely to respond to PD-1 blockade, but that is really the limit of the association. It is not a binary indicator. It is an inducible dynamic biomarker unlike a BRAF mutation, a static biomarker. PD-L1 is expressed on normal endothelial cells and antigenpresenting cells, and by interrupting that pathway, you may augment immunity even if you are not directly affecting the interaction between PD-L1 on the tumor cell and the T cell. This pathway has a role in immune regulation outside of the tumor microenvironment as well. In my opinion, as it stands now, we should not use PD-L1 as a biomarker to restrict access to these medicines, which can provide long-term disease control.

SZNOL: It is going a bit too far to rule out PD-L1 as a biomarker because there is unquestionably an association between PD-L1 expression and response. It is not a perfect predictive biomarker, but it certainly does stratify groups into those with relatively higher response rates and relatively low response rates. If it were so dynamic, there would be no correlation, but in fact there is a fairly strong correlation in several different diseases. I believe PD-L1 to be very useful as a biomarker for PD-1-based therapies, and with combinations, for stratifying groups by probability of response to PD-1 alone. Phase 2 trials should target activity for the combination of different antigens in those two groups. If you treat 100 patients with PD-L1-low tumors, your response rate might be 15 percent. With a combination, you may increase the response rate from 15 to 40 percent. Patients with PD-L1- high tumors may have an expected response rate of 35 to 40 percent, so with a combination, you have to target a higher response rate.

What I’m hoping is we’ll have enough information over the next few years to combine patient characteristics and some relatively simple information from the tumor biopsy to make a good guess on the right treatment approach for any individual patient. Because immunotherapy has produced long, durable remissions in a subset of patients, it will change patients’ expectations. Now, when someone comes in with cancer, they won’t want their doctor to say, “You’re going to live four more months, you will have to take this pill, and you will have to deal with side effects during the whole time.” Patients want to say, “I’m willing to give you three or four months of my life. I’m willing to accept some side effects, but at the end of those four months, I want my cancer gone, I want my life back, and I don’t want to think about this anymore.” That’s what they want. If we develop therapies that give them that, those are the therapies patients will choose and the doctors will want to give.

KELTNER: Still, in the real world of cancer therapy — outside of clinical trials and academic institutions — reliable tumor biopsies are available in only a minority of cases. The situation suggests a major goal for industry should be developing set combinations with high response rates across most patients and tumor types.

QUESTION: By what criteria will physicians select specific immunotherapy combinations for individual patients or patient groups? Or will regulatory and reimbursement realities dictate the combinations?

KELTNER: The history of drug therapy in cancer is one of development, approval, and sale of single agents, with very recent ventures into approval and sale of agents in combination. For the foreseeable future, approvals of single or dual agents with single labels will continue to determine reimbursement for expensive cancer immunotherapies. If a combination does not get reimbursed, few clinicians will have the luxury of prescribing it on a “personalized” basis, even with good disease/biomarker data to support it.

FONG: Those realities, regulation and reimbursement, will dictate the combinations. Unfortunately, treatments such as anti-CTLA-4 and Provenge are already very expensive. We’re really constrained in using these drugs by what is reimbursed, and that unfortunate reality will dictate the usage of these different drugs, especially in combination. More often than not, drugs are approved as single agents and insurers may not be willing to have the doctor prescribe more than one at the same time. But that is where we will need trials that show the combinations actually work. Unless there’s a change in the cost and access to these drugs, it’s not going to be like chemotherapy where we oncologists can prescribe different drugs and be able to tailor a treatment “on the fly.”

I do not see all the different immunotherapies as being mutually exclusive. If anything, it is great that we’re hitting different elements of the immune system. Once we know where the fundamental blocks are and can identify those within a patient, we might figure out in advance, rather than empirically, which of the combinations would be good for them.

KELTNER: Paying for combination immunotherapies might be less of an issue than paying for drugs with a less substantial therapeutic payoff — that is, if the new therapies have the impact on safety and survival the current data suggests.

WOLCHOK: Whether a drug is approved in a given disease affects whether it is reimbursed. It seems to be less of a problem with the numerous PD-1 pathwayblocking medicines being explored in a wide variety of diseases. We are entering a new era in the tools we will have at our disposal to treat cancer. Immunotherapy is now a standard approach to cancer treatment, and we will see it integrated into treatment programs, not just for one or two diseases, but for a wide variety.

URBA: Regulatory and reimbursement issues always affect how patients are treated. Except in clinical trials, physicians can only use agents that have been approved for clinical use. Current economics now dictates a value assessment for new therapies that was not required before. Value will certainly depend on price, and there will be ways to price oneself out of the market and make combinations financially impossible for individual patients or their insurance/health plans. The big plus for immunotherapy is the durable nature of responses compared to far shorter remissions for patients with solid tumors receiving chemotherapy or targeted therapy. This is the true value — the benefit to the patient — and this is where it should, and I believe will, eventually be judged.

I believe physicians will select on the basis of what is best for their patient and in the future this will be based on characteristics of the tumor and the patient’s immune status. In the future there will be an immunoscore for patients at diagnosis that will help their doctor know which immunotherapy or combination will be required. This will assess both the type and extent of immune response in the tumor and the general health of the cancer patient’s immune system.

SZNOL: We already pay a great deal of money for treatments that are really suboptimum. We are developing new treatments that may be expensive, but would be returning people to fulltime jobs in three or four months, without continuing their treatments indefinitely until they die. Maybe society will consider that a worthwhile cost.

SHARMA: Insurance companies tend to pay for treatments that are approved by the FDA, and most physicians tend to prescribe treatments along these same lines. But again, we need to make sure we’re not selecting for treatment on a single biomarker such as PD-L1. If a patient has a tumor that is deemed to be PD-L1-negative, this should not exclude patients from treatment and they should still be considered for treatment with anti-PD-1 or anti-PD-L1. I consider it an ethical issue. The response rate is lower if a patient’s tumor is PD-L1-negative, but it is not zero. For patients with no other treatment options or for whom there are limited treatment options that may only provide a few weeks to months of life, it’s worrisome that these patients would be denied the opportunity to have access to an immunotherapeutic agent with the potential to provide long-term survival.

QUESTION: Will the most effective immunotherapy combinations be specific to traditional cancer indications (NSCLC, HCC, etc.) or tend to have general effectiveness against all or a wide range of cancers?

KELTNER: Although panelists have already hinted at immunotherapy’s potential to treat a multitude of cancers, the effects of having highly safe and effective agents (in combination) that were also widely applicable would be unprecedented and profound. At the same time, panelists may have different reasons for, and place different limitations on, those effects.

SHARMA: It will depend on the number of patients who respond in all tumor types. We don’t know that number yet. Initially, all the protocols were focused on melanoma, then skin cancers, now lung cancer, but we are now conducting trials with patients in small-cell lung cancer, gastric cancer, pancreatic cancer, triple-negative breast cancer, bladder cancer — all those tumor types on one protocol — and already seeing a few responders in those patients. It will be interesting to see what happens as more and more data comes in from the studies of other tumor types with new clinical trial protocols, where I believe we will see responses as well.

ALLISON: Some tumor types will probably be more responsive than others. You can get the immune system primed, and when you know it’s primed, you know exactly when to treat. But I believe eventually we will see immunotherapy used for every kind of cancer.

FONG: Immunotherapy has taught us we can’t predict which cancers are going to actually respond. The big surprise was anti- PD1 and anti-PD-L1 antibodies working in lung cancer, because we have seen so many therapies fail in that disease. We need to try immunotherapies in many different cancers just because they may work.

GRETEN: There is no reason to believe that immunotherapy would work in only one disease and not in others. The whole field was surprised to see the data on using checkpoint inhibitors in nonsmall- cell lung cancer, and there is a lot of debate about why these patients actually show responses, but none of the hypotheses have proven correct. It is the same with targeted therapies; we have no biomarkers, and biomarkers would not only help us select patients that would respond, but also to understand why and how a therapy works or not. In general, even with the best treatment for solid tumors, we never get more than 50 percent of the patients to benefit. That means the majority of the patients do not benefit. One good approach is to understand why we have such good data in non-smallcell lung cancer but not in all the other diseases. What happens to those patients where the drug doesn’t show a response?

"The T cell therapy, like CAR-T, may work, but it is mostly wishful thinking right now. Even if it turns out to be doable, it will be ridiculously expensive."

ALAN VENOOK, M.D., Professor, Department of Medicine (Hematology/Oncology), UCSF.

WOLCHOK: There’s a lot of talk about the importance of antigen expression, and a lot of interest focused on the so-called neoantigens formed from the mutations that arise during carcinogenesis. It is probably not a surprise that the diseases where immunotherapy has shown significant benefit have cancers with an abundance of mutations — melanoma, tobaccorelated cancers, now bladder cancer, head and neck cancer — which, because of their association with environmental exposure, have numerous mutations. There now are very important trials going on looking at the activity of immunotherapy in patients who have familial syndromes involving mismatched repair defects and “microsatellite instability” and are overrepresented with passenger mutations because of the defects that lead to the cancer.

QUESTION: Do you see limits on the practice model for cancer immunotherapies, i.e., will cell-based approaches remain restricted to a small number of patients in intensive care or salvage settings?

KELTNER: Most of our panelists work on the side of “off-the-shelf” immunotherapeutic agents, but they may acknowledge some plausible technological solutions and see a defined role for cell-based therapies.

ALLISON: If you had asked me about cell-based therapies two or three years ago, I would have said they tell us what the immune system can do, yet I doubt they are exploitable. But with the new techniques, such as the chimeric antigen receptor approach and the T cell receptor constructs, they now seem more practical. The new techniques allow physicians to take the patients’ own blood and redirect their T cells’ specificity, making them more useful. However, although cell-based therapy looks impressive in leukemia, because leukemia has antigens that are simply not found on other tumor cells, it is not the same with solid cancers. It is possible to do antigen spreading and combine checkpoint blockade with adoptive cell therapy, but I do not believe it will ever be front-line therapy.

SZNOL: I am on a scientific advisory board of one of the companies developing cell-based therapies, and I do believe they will work. There will be ways to do relatively rapid isolation of cells from tumors that are tumor-antigen specific, and relatively simple ways to do rapid expansion and reinfusion. So far, the data suggests if you progress on checkpoint inhibitors, you can still respond to the cell therapies, though they may not become front-line treatments. They could become commercially available as a process where you take a piece of tumor, put it in a bag, ship it overnight, and three weeks later get back a bag of cells to infuse — then someday become a blood-bank procedure.

URBA: Widely applicable, off-the-shelf combinations would be ideal, and there will likely be examples of them for specific disease types and perhaps even shared among a few different types of cancer. However, based on what we are learning of the genetic complexity of tumor cells and the plethora of immunological states within a tumor microenvironment, it looks like there will be significant personalization for some time. Cell-based therapy is one example. In some instances, it appears that once certain principles are understood, anti-CD19 CAR-T (Chimeric Antigen Receptor Therapy) could be more widely available. But picking individual T cell receptors, as is happening in some of the solid tumor work, is not likely to be broadly available any time soon.

FONG: Dendreon has shown you can apply cellular therapies to a broad population. Whether that is a viable or sustainable business model, I defer to the business folks. The model of moving cells to a centralized production facility from apheresis centers, which are widely available, is possible today. But any treatments, such as some adoptive T cell therapies that could cause patients to get significant side effects and end up in the intensive care unit, may not be appropriate in a typical community setting. It would be preferable to have an off-the-shelf therapy, and one of the hopes with the patient-specific therapies is we could learn how they work and potentially transform them into an off-the-shelf type of therapy.

GRETEN: I do not think that these highly individualized therapies, where you have to produce the drug for every single patient, will be something feasible long term. The incidence of leukemia in patients who actually require such treatments is so much lower than patients with solid tumors. It would be a significant challenge for the healthcare system to provide cell-based therapies for solid cancer.

BERINSTEIN: There are pros and cons for both approaches. Clearly, using patient-tumor-specific therapies should be able to provide a level of specificity and memory which may not be possible in all patients using the off-theshelf therapies. Using patient-derived immune cells that are engineered and reinfused can also be effective (CAR therapy). However, the antigens being targeted must be selected carefully to avoid serious toxicities. Unfortunately, these personalized treatments require higher levels of infrastructure and cost. But the considerable redundancy in immunosuppressive pathways requires off-the-shelf immunomodulators be combined to achieve suitable efficacy, and the most effective combinations will likely need to also incorporate strategies to enrich tumor-specific T cells. We need to identify the appropriate combination of therapies, effective and safe, that will provide a clinical benefit in a majority of patients and eliminate the need to resort to patient-specific approaches.

QUESTION: Is it premature for companies to jump in and start developing immunotherapies?

KELTNER: In point of fact, and unlike other fields where the KOLs are exclusively academic, scientists/opinion leaders are leading much of the commercial development in the cancer-immunotherapy sector. On the other hand, skepticism remains in the KOL community.

VENOOK: It’s fine for companies to do, but most agents are not ready for Phase 3. Again, I just think they’re building expectations excessively. The T cell therapy, like CAR-T, may work, but it is mostly wishful thinking right now. Even if it turns out to be doable, it will be ridiculously expensive. Look at Provenge and ipilimumab — is either one enough of an advance for the cost? I believe they are on the margin.

KELTNER: Is it still too early to do late-stage trials with immunotherapies, single or combination, on the scientific and regulatory fronts? Or do immunotherapies, which may be more effective in earlier- stage disease, defy the old paradigm of conducting new-drug clinical trials only in late-stage patients?

SHARMA: We have always had the paradigm in oncology of starting out with Stage 4 metastatic-disease patients for whom nothing else has worked, and then moving it earlier into a clinical trial setting where patients with earlier-stage disease receive treatment before surgery. I believe we’ll see the same paradigm with immunotherapy.

"Because immunotherapy has produced long, durable remissions in a subset of patients, it will change patients’ expectations."

MARIO SZNOL, M.D., Professor of Medicine (Medical Oncology); Clinical Research Program Leader, Melanoma Program, Yale Cancer Center

FONG: This is a really important question. As an example, if the FDA continues to mandate survival as an endpoint, it will get increasingly difficult for us to move immunotherapy into early-stage disease — especially in prostate cancer, one of the areas we focus on. If the FDA wants to support faster development of drugs, it will need to change the approvable endpoints in these cases.

ALLISON: Earlier may be better, but it can be too early. All this works by getting the T cells primed, and you have to have tumor-cell death to get the information to the immune cells, so if you have just a tiny tumor mass, that’s too early — unless you have a genomically targeted drug that just kills all the tumor cells, or at least enough of them to cause inflammation.

GRETEN: The biggest fear I personally have is that, instead of doing small, very careful studies, companies face too much pressure to immediately start large randomized Phase 3 studies, which in the end may fail. That is always a problem with a new movement in any disease. Obviously, there is huge competition among the companies, small and large, but it may lead to making mistakes that in the end will harm the field.

A lot of the rules that we have in oncology all of a sudden don’t apply any more, especially with early drug development. We like to look at progression free survival [PFS], or response rates, but now we may not even see dramatic changes in PFS. We have to come up with new, immune-related response criteria, but at the same time, some patients show really extended overall survival that is not mirrored by an early signal.

Let’s say checkpoint inhibitor XY does not work in colon cancer — does that mean it will work if you combine it with something else or give it earlier in the disease progression? We should slow down development and initiation of huge Phase 3 studies until we really understand the mechanisms. If not, what are we going to do with all the negative studies?

Ending this part of this series with a provocative question primes the discussion for the highly interested industry participants in Part Three, “Combination Cancer Immunotherapy — Companies at Stake,” coming next month. Meanwhile, please join the discussion on Twitter at #CCIRLSL.