Magazine Article | September 1, 2016

Cancer Immunotherapy - Simpler Or More Complex?

Source: Life Science Leader

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

There comes a time when the present can no longer be described in terms of the past. A great leap forward redefines our normal experience and resets our expectations. We must invent and adopt a new lexicon for the future. And yet the transformation can happen silently, invisibly, and unfelt right in front of us, leaving the past behind like a vaporous dream.

Such a change of scene is happening right now, in the form of cancer immunotherapy. You won’t see much about it in the general press or on the nightly news, and even the central players in this dramatic shift seem to be doing their best to keep a lid on things in one way or another. But slowly and surely, the wheel is turning, and when we look back from the future, we will see that cancer treatment has made a giant leap forward with immuno-oncology (IO).

Two years ago (September 2014 issue), we initiated a five-part series that captured the thoughts of many key IO players and anticipated most of the issues the field faces even today. A “virtual roundtable” of a dozen key opinion leaders, including some of the main pioneers in IO drug research and development, along with executives in 16 companies conducting clinical trials in the space, contributed their views and expertise to the series. We then followed with an update a year later (September 2015 issue) on the current clinical results and insights.

The series and update shed light on how science can drive business in the life sciences industry. Companies have scrambled to keep up with the emerging science of cancer immunotherapy, and the articles in this series supply an essential layperson’s understanding of the scientific progress in IO and its course-changing effects on the IO business space.

This year, with most of the major cancer meetings concluded, yet so much in the IO space still in full flux, we are turning to someone who first proposed, then served as moderator of, the virtual roundtable in the original “Combination Cancer Immunotherapy” series: Llew Keltner, M.D., Ph.D. Keltner is an oncologist long active in drug development, company startups, and partnerships in the cancer space, particularly in immuno-oncology. He continues to be an everpresent witness and participant in the IO arena, constantly interacting with scientific and business leaders working on the remarkable new generation of cancer immunotherapy approaches.

"Anti-PD-1/anti-PD-L1 is the backbone of cancer immunotherapy, except in some very unusual situations."

Llew Keltner, M.D., Ph.D.


In the following exchange, Keltner sums up the key issues and findings that fired up the IO field during the past year. He reports on new clinical trial results and evolving theories about immunotherapeutic mechanisms and agents in cancer, as presented at the leading oncology meetings and discussed in the IO community. In many cases, the new findings bear out predictions made in our original series; in other cases, they challenge previous expectations. Safe to say — in most cases, controversy among the various players in the field continues despite, or because of, the new insights, and Keltner expresses his own views as well.


The bottom line, which became clear at the AACR (American Association for Cancer Research) annual meeting, and was even more clear at ASCO (American Society of Clinical Oncology), is that anti- PD-1/anti-PD-L1 is the backbone of cancer immunotherapy, except in some very unusual situations. More money has been spent on development of anti-PD-1 drugs, by about a factor of three, than any other type of drug in history. We now have more than 100 anti-PD-1s, and more than 30 anti-PD-L1s, in development worldwide. So the reality is anti-PD-1/anti-PD-L1 therapy is the biggest focus of drug development that has ever existed.


In specific subsets of patients with nonsmall cell lung cancer (NSCLC) treated with anti-PD-1 as a sole agent, we can hit about a 40-percent response, compared to around 10 percent with other drug treatments — and this isn’t a typical progression-free survival response. When patients respond, when their tumors shrink with immunotherapy, and when they live longer than a year, they tend to respond dramatically, beyond all precedent. That is the magic of immunotherapy.

In some of Dr. Jedd Wolchock’s 10-plus-year studies with anti-CTLA-4, a very high percentage of the melanoma patients who survived for a year are still alive. So response in cancer immunotherapy is completely different than what we’ve all thought of as a response in solid-tumor therapy with radiation, chemotherapy, antineovascular agents, and targeted agents. All of those earlier therapies can destroy tumor tissue, but they don’t treat the disease very well at all, which is why they show so little improvement in overall survival, except in small trials with highly selective patient populations.

Oncologists have become accustomed to image-based response measurement and the idea that if they’re killing tumor tissue it must be good. That’s a nice idea, but, unfortunately, it doesn’t correlate well with the patient’s survival, and all of the conventional drugs — and of course radiation — have very bad side effects. With cancer immunotherapy, we are now seeing what we hoped and expected to see — when patients respond to an IO drug, they have a very good potential for long-term survival.

At this year’s ASCO meeting the talk was, now what do we do with survivorship? We’re seeing cancer patients who, thanks to immunotherapy, are living for very long periods of time. What does that mean for us as a field? How do we treat these patients? How do we monitor these patients? What kind of follow-up is necessary? We haven’t really had this problem before. It’s very similar to what happened with AIDS after the protease-inhibitor cocktails came into use. Among the up to 40 percent of NSCLC patients who respond to anti-PD-1, any one of them may have a good shot at a normal lifespan, and certainly a much longer extension of life than with other drugs.


The evidence suggests that a very important difference in the nonresponder populations is that they have low- TIL tumors; their tumors do not contain activated CD8 T cells, called tumor-infiltrating lymphocytes (TIL), so they’re TIL-negative. In patients who are TIL-positive, the correlation with response to anti-PD-1/anti-PD-L1 is extremely high. In the current jargon, TIL-positive tumors are hot and TIL-negatives are cold. That idea was first discussed intensely at the SITC (Society for Immunotherapy of Cancer) meeting in the fall last year, at several IO conferences last winter, and then at AACR and ASCO this spring. Now hot and cold tumors seem to be the “hottest” thing in the field these days.


At ASCO, there were probably at least 100 presentations or abstracts related to ways to convert cold tumors into hot tumors. In general, the idea is gaining a huge amount of credibility. Much of the data seems to show that it is quite reproducible with all sorts of different mechanisms. One reason for the widespread belief in the idea is the use of biopsies in cancer-drug trials, which has just gone through the roof. Ten years ago, biopsy trials were not rare, yet not that common, but these days the number of trials that involve biopsies is just enormous.

Yet researchers are using biopsies to determine whether patients are PD-L1 positive, HER2 (human epidermal growth factor receptor 2) positive, or positive for some other marker to predict their response to a particular molecule. They typically look at gene signatures in tumors, which is unfortunately misleading because tumors are so massively heterogeneous. A needle inserted into a tumor at a particular location, at a particular moment, may sample a bunch of cells that show XYZ. Another needle in the same tumor at the same moment won’t show the same thing. Tumor heterogeneity is why cancer immunotherapy works where the other forms of cancer therapy fail. A biopsy showing high or low TIL may also be more indicative, since TILs are living cells, and when appropriately activated, they can proliferate and destroy cancer cells throughout the tumor.


Yes. It all starts with T cells. CD8 T cells, or effector T cells, are just T cells that express the receptor protein CD8 on their surface. There are progenerative T cells — immune stem cells that have the ability to go in a zillion different directions. They can become T regulatory (Treg) cells when they express the Foxp3 protein on their surface, or CD4-T cells when they express the CD4 protein on their surface.

We had always thought Tregs are bad in cancer because the tumor uses them to suppress CD8s. It turns out Tregs can stop expressing Foxp3 and start expressing CD8 and begin killing tumor cells. All T cells are capable of becoming tumor killers, or TILs. It’s just a matter of expressing DNA and changing pathways and functions within the T cell, probably in response to the triggering of multiple pathways inside the cell. In general, any T cells expressing the CD8 marker are very likely TILs, and they will kill tumor tissue as long as they’ve been activated to one or more of the tumor antigens.

Despite all of the proposals and research on how to convert TIL-negative to TIL-positive patients, however, only a small number of commercial candidates exist. One avenue companies have explored is apoptotic ablation, which makes tumors hot by creating neo-antigen containing apoptotic bodies in the tumor that traffic to the tumor-draining lymph nodes and get exposed to T cells. CD8-T cells get activated, wander around the body, and locate those tumor antigens on any remaining tumor or metastasized tumors. In a certain percentage of patients, you get abscopal effects — shrinking or disappearance of tumors outside the local treatment area. If the patient can then be treated with an anti-PD-1, anti-PD-L1, and/or an anti-CTLA-4 or TNFRSF (tumor necrosis factor receptor superfamily) co-stimulator, the abscopal response rate may approach 80 or 90 percent or perhaps more.

There are also a small but increasing number of cancer vaccines with data demonstrating they can convert tumors from cold to hot, but only one cancer vaccine platform, the allogeneic transfected whole tumor cell technology from Heat Biologics, has produced data from human clinical trials demonstrating the vaccine can convert 100 percent of tumors from cold to hot in multiple indications. There are lots of other agents that can help — old targeted drugs and even chemotherapies will trigger some apoptosis and immune activation in a tumor, causing conversion of a low percentage of patients from cold to hot.

But there are new therapeutic candidates out there that will do a fantastic job of it, and if the pharma companies will pick them up, push them forward, and really drive hard, we may see some dramatic developments. But we also have to be cautious; some companies that were making cytotoxic agents two years ago now say they are doing cancer immunotherapy, converting tumors from cold to hot. Maybe so — in rats or mice — but they are a long way from human proof of concept.

With one approach, blocking PD-1, now dominating the IO field, work on new immunotherapy agents continues but new approaches are being almost desperately pursued. Keltner discusses some important developments with other immunotherapies, vaccines, and co-stimulators.


First, I don’t like calling anti-PD-1 a checkpoint inhibitor. There’s new data emerging that suggests the most important mechanism of action of anti-PD-1 may not be its checkpoint inhibition, which interferes with the binding of PD-1 and PD-L1 on tumor cells. PD-1 is expressed by the CD8 T cells, and when they bind to the PD-1 ligand PD-L1 on the tumor cell, the CD8 T cells become inactive.

The theory was that anti-PD-1 merely blocked the process of inactivating T cells. But studies are now showing that when anti-PD-1 binds to the PD-1 on a CD8 T cell, likely all of the TNF (tumor necrosis factor) family of receptors — OX40, 4-1BB, GITR, ICOS, and especially TNFRSF25 — gets dramatically upregulated on that T cell, which causes the population of activated T cells to proliferate. That is why the TNF receptor agonists are called co-stimulators, now recognized as a new class of immunotherapeutics.

This year at ASCO, the level of excitement about the TNF family of co-stimulators did not go way up, but it didn’t go down, either. Most of the community was looking at the latest clinical data for the co-stimulators, mainly anti-OX40 and anti-4-1BB. Other companies are also developing the co-stimulators, but Pfizer and Roche were the only ones to present clinical data at ASCO. AstraZeneca did not present data on their three ongoing combination studies involving anti-OX40 or the OX40 ligand fusion protein.

Some analysts and others at ASCO said the costimulator data was disappointing because the reported clinical responses were not overwhelming. However, these were first-in-man dose escalation studies in small numbers of patients, monitored for relatively short periods of time. Just as occurred in early trials of anti-CTLA-4, the patient responses from new immunomodulatory agents can be surprising — and in some cases can occur very late. All data suggesting the combination of the co-stimulators with anti-PD-1/anti-PD-L1 in humans is theoretical, based primarily on in-vitro and mouse tumor model research. We are in very early days, and there is an astonishing amount to learn about these agents — which are, again theoretically, not likely to have much effect as monotherapies.

The safety data for 4-1BB and OX40 was great. Many people have been worried about safety with the costimulators because they can have some seemingly opposing effects. OX40 and TNFRSF25 cause a large and immediate up-regulation in Tregs. That concerned those who still think Tregs are evil, but we know the greater the Treg stimulation, the more CD8 memory activity results. According to research published more than a year ago, the Tregs that accelerate immediately are required for the maturation of activated or “memory” CD8s. The research found that cytokines secreted by the Tregs are absorbed by the CD8s, pushing them to become memory CD8s, another term for TILs, which go after the tumor.

As research continues to elucidate immunotherapeutic mechanisms and test new targets, corporate constraints, payer pushback, clinical challenges, and regulatory conundrums are shaping the IO business and market, for better or worse.


Anti-CTLA-4 really is one of the few actual “pure” checkpoint inhibitors out there, and the only one so far to go into the clinic, as well as the only one to be approved. Some say the IDO (Indoleamine 2,3-dioxygenase) inhibitors are checkpoint inhibitors, but they’re not. They’re just affecting mechanisms in tumor cells that alter the ability of T cells to survive as affected by the tryptophan breakdown pathway (tryptophan [TRP] to kynurenine [KYN] metabolic pathway). But they don’t target immune checkpoints; they are probably only involved in facilitating T-cell proliferation.

Any human cell can get around an IDO inhibitor, and any cell can get around a TDO (TRP-2,3-dioxygenase 2) inhibitor. Tryptophan breakdown goes through multiple pathways in the cell, each one ending with kynurenine, which is the bad actor with T cells, sending them into apoptosis. Kynurenine is produced from tryptophan even if you block IDO completely, or if you block TDO completely. And if you block those two things completely, some researchers believe that there will be severe toxicity. One company, Kyn Therapeutics, is working on a way to cleave the kynurenine, but it’s at a very early stage.

Scientists and companies propose all sorts of mechanisms for how they can affect T-cell behavior, with many related agents such as PEGylated IL-10, TIGIT (T-cell immunoreceptor with Ig [immunoglobulin], anti-TGF beta, and ITIM [immunoreceptor tyrosinebased inhibition motif] domains). But those are also not really checkpoint inhibitors. Compugen has a slew of real checkpoint inhibitors with novel targets. Bayer has licensed two of them and is carrying the development forward, but they’re not yet in the clinic. Compugen is developing its own targets and is no less than a year and a half from the clinic.


I don’t believe the patent challenges will succeed, especially if they are based on use of a natural DNA sequence. Out of the more than 100 anti-PD-1s in development, some of them will make it to market. One in particular will surely be the BeiGene product. BeiGene very likely has a “gold stamp” from the Chinese government, and patients are not in general being treated in China with Keytruda (pembrolizumab) or Opdivo (nivolumab) in a clinical trial or otherwise. The Chinese government has made it very difficult for biologics not manufactured in China to enter development for approval in China. BeiGene has developed and manufactures its drug in China as a Chinese company. But it is a public Nasdaq-listed company, now well into anti-PD-1 trials and smart as a whip, having just hired Merck’s cancer immunotherapy business development head, Ji Li.

Following the recent approval of Genentech’s atezolizumab, anti-PD-L1 drugs will add to the flow of new products into IO. There will be enormous commercial competition, which means more clinical trials, trying to generate more data, trying to widen indications, which the big players are doing. IO drugs are already winning approval for more and more indications, and the FDA is still granting breakthrough designations for the new indications. But anti-PD-1 and anti-PD-L1, whether you like it or not, is the backbone, and small biotechs in general today have little choice about using their drugs in combination with that backbone.


Not a lot, because the payers are being typically restrictive about it. As always in oncology, there is a fair amount of it for rich and influential people. It is a horribly unethical mess, the typical result of our strange nonhealthcare system, that poor people are clearly being discriminated against in the most powerful cancer therapy area. Poor people who have cancers that are not in approved buckets for Keytruda or Opdivo are not in general getting treated, unless the therapy is being subsidized at large public teaching hospitals or if the patient can get into a clinical trial for an unapproved indication.

And because of what’s happened to co-pays in the last two years, even poor people who do have on-label cancers are often not getting treated with Keytruda or Opdivo. The co-pays for a course of therapy with Opdivo in, say, the approved indication of renal cancer can amount to $18,000. A lot of the people I walked by on the street today don’t have $18,000; they never will have $18,000. Are Merck and BMS trying to help? Of course. There are very patient-oriented, ethical, and concerned folks at all of the oncology companies, and they don’t like this inequity either. But the very high price of new, truly innovative drugs is creating a serious problem that the industry must pull together to solve.


Yes, for example, Merck has almost 700 cooperative trials with other companies now underway or in planning in combinations with Keytruda. BMS is beginning to work with companies now. Pfizer is in early stages of creating collaborations and AZ is focusing on a small number of existing collaborations, though their anti-PD-L1 programs likely have less potential for use in combinations than anti-PD-1. Roche is very focused on using its IO platform, including anti-PD-L1, to shore up their aging anti-neovascular and targeted-drug approved therapies.

But the biggest nightmare we will have with combinations is data. Many combination trials are beginning to read out data, and they are all different from one another in patient groups, trial designs, endpoints, error estimators, biomarkers, translations from preclinical data, and dosages. The pharma companies will try to make useful comparisons, but they will almost all be wrong, because the trials are fundamentally statistically different, and comparisons will thus be by definition mathematically flawed. The number of reports we will see in the next year in the financial press, trade press, and scientific press making illegitimate but highly touted comparisons will be truly amazing.


A number of our original experts have, over time, expressed concerns in meetings and articles about the induction of various T-cell receptors in CAR T cells. Some have felt that the b7 family of receptor ligands might have unknown or untoward toxicity, while the TNFR family would be less likely to have the same risks. Given the recent disaster with the Juno construct, the concern may have been warranted. Although a great deal of fairly complex work will be required to sort out the potential differences, the relative lack of toxicities both in preclinical and clinical use of the anti-TNFR drugs (4-1BB, OX40, GITR, ICOS, TNFRSF25) seems to support the potential difference between the two receptor families. Until the work is done, the ethical issues of using any CAR-T cell preparations in humans should always be questioned, on a patient-by-patient basis.

However, perhaps the elephant in the room here is the very large nondrug costs and complexity of administration and patient management with autologous CAR-T. It may be very telling that the large autologous CAR-T players (Novartis, Juno, and Kite) are all looking very hard at allogeneic CAR-T methods that may overcome much of the nondrug cost and complexity questions. Kite’s very recent license of the UCLA allogeneic CAR-T patent portfolio is an indicator of this nervousness. Many claim that some of the allogeneic approaches may also address some of the severe toxicity issues, but until human studies are engaged, this is unknown.


Nothing actually reported that really says anything. If there is a target — such as EGFR (epidermal growth factor receptor) or HER2 — then the CRISPR folks will claim all they have to do is change some genetic material and cancer will be cured. But they are even further from proving that claim than gene-therapy proponents are from proving theirs. CRISPR may be very useful for rare metabolic diseases where there is a deficiency in a protein. Fix the genes in some cells, and maybe they will produce enough of the required protein to reverse the symptoms. Pretty elegant actually. Of course, no one has ever been able to efficiently deliver gene constructs into cells. But, if there is some not-yet-invented way to modify some embryonic cells, and if it is allowed ethically and politically, then it might work in rare diseases.

Dr. Keltner anticipates continuing tumult and change in the IO space in the coming year, and we will continue to cover new developments, as well as prepare for another annual update about this time next year.

Neil Bernstein, M.D.
Professor of Medicine
University of Toronto
Odette-Sunnybrook Cancer Center



A member of the original panel of science experts and key opinion leaders in our Combination Cancer Immunotherapy virtual roundtable series (September 2014 to January 2015) responds to our query: What was the most important advance in immuno-oncology during the past year?

Advancement in translational science is the most important advance in immuno-oncology.

Now, finally, there have been successes in using the immune system to fight cancer. This follows a decade or more of failed clinical trials that told us more what not to do rather than what to do. With the clinical success of checkpoint inhibitors and some cell therapies, the situation now is different — translational science is providing significant understanding of how the new cancer immunotherapies (CIs) are working, and the field is suggesting rational science-driven strategies to further enhance these therapies.

The new insights are possible because of significant advances in technologies such as whole exome sequencing, multiplex immunohistochemistry, and gene expression profiling of clinical samples. For example, the evidence that the mutational load, lymphocytic infiltrate, and level of PD-L1 expression are predictors of response to some checkpoint inhibitors has fueled a number of hypotheses around how CIs are working and how to make them work better. From this data it is hypothesized CIs are allowing lymphocytes that recognize processed and presented neoantigens in the tumor to expand, infiltrate the tumor, and attack. Recent evidence for this includes relationships between mutational load and clinical response to CIs in melanoma and lung cancer. Moreover, cell therapy recognizing mutated neoantigens has been associated with clinical response in some patients.

More importantly, this translational data predicts that various strategies to increase tumor-specific lymphocyte infiltration into tumors should synergize with CIs. To accomplish this, strategies and trials combining CIs and vaccines, cell therapies, and local immunotherapies or cancer therapies are being initiated. Additionally, we are approaching the era where CIs or other immunomodulators may be combined with vaccines or cells generated to patient-specific neoantigens that are thought to be highly immunogenic. Given the strong translational science that has provided a rationale for these trials, there is a high likelihood of further significant advances and successes in immuno-oncology.


As noted in our Cancer Immunotherapy Update, industry companies are conducting a very aggressive search for new immunomodulatory targets. But new targets are becoming quite scarce — at least in the more obvious checkpoint inhibitor, PD-1, and co-stimulator realms. As a result, the industry is focusing much more intensely on bispecific technologies where a single molecule or construct can hit more than one target. Many companies are developing a wide range of platforms, from Heat Biologics’ whole cancer cells transfected with GP-96 fusion proteins and OX40 ligand fusion protein, to constructed bispecific combination molecule platforms such as BiTe (bispecific T-cell engager) and DART (dual-affinity retargeting), to single antibody bispecifics, including those from Xencor, Shattuck, and Crescendo. Large companies such as Amgen, Novartis, and others are betting heavily on the use of bispecifics in oncology. Preclinical data from some of these technologies shows quite interesting activity when targets such as PD-1, OX40, CTLA-4, and other immunomodulators are combined. Bispecifics are at an early stage, with many unanswered questions, but they are likely to be quite important in immuno-oncology (IO). One of the great advantages of bispecifics, if effective and safe, will be avoidance of “stacking” of IO drug pricing with too many single-target drugs on the market confusing physician choices and reimbursement.