Precision Immunotherapy 'Still Years Away'

Stefan Zimmermann, MD; Ignacio Melero, MD, PhD; John B. Haanen, MD, PhD


January 04, 2019

Stefan Zimmermann, MD: Hello. I am Stefan Zimmermann. I am associate head of immune-oncology phase 1 research at the Centre Hospitalier Universitaire Vaudois, in Lausanne, Switzerland. Welcome to Medscape Oncology Insights, coming to you from the 2018 Congress of the European Society for Medical Oncology (ESMO).

Today we will discuss the concept of highly personalized, or precision, immunotherapy. Joining me in this discussion are Dr Ignacio Melero, professor of immunology at the Academic Hospital of Navarra and the Center of Applied Medical Research at the University of Navarra in Spain; and Dr John Haanen, chief scientific officer in immunotherapy and staff scientist in the division of immunology at the Netherlands Cancer Institute in Amsterdam.

At the ESMO 2018 annual meeting, presidential sessions were, once again, full of news about checkpoint inhibitors in previously uncharted disease settings. We're giving checkpoint inhibitors upfront to just about everyone, and the choice is not biomarker driven. It's not a selective population; rather, we are giving everything that works to everyone, which seems a bit counterintuitive to me. A lot of the focus in drug development presently is toward a more rational or more tailored approach in developing these drugs and new combinations based on how the environment looks. Are we doing this wrong?

Ignacio Melero, MD, PhD: In a way, we are. Obviously, you have to bring the clinical benefit to patients, and the sooner the better. We probably can identify patients who are more likely to respond, and certainly we can identify patients who are unlikely to benefit. [But] for each treatment combination, this becomes different; sometimes it's slightly different but sometimes it is highly different in regard to biomarkers. So it seems that the clinical development of drugs and combinations of agents is outpacing our ability to develop companion diagnostics and biomarkers. We will have to learn as we go, but there certainly is a lot of potential for biomarkers in that space.

John Haanen, MD, PhD: I agree. We are still lacking proper biomarkers. Of course, some of the biomarkers are already in the clinic; some people are actively using PDL-1 expression and also tumor mutational burden in lung cancer, and with that you already select a bit.

Much of the work being done at the moment is in developing this into an assay that can be used on a large scale. In the non–small cell lung cancer area, this may help with selecting patients who require combination immunotherapy. But there are still so many things we do not know. We are not able, in a validated manner, to know whether tumors are immune-desert or are highly infiltrated, because those tumors would be approached in different ways. So I completely agree: The trials are ongoing with all of the combinations, but we lack a lot of information on the tumor microenvironment. That is something that we in academia must push for because I'm not sure all of this work will come from pharma.

Zimmerman: So we are still years away from being able to identify whether a specific tumor behaves in this fashion and that phenotype is immune-excluded, for example. And then we need to combine those two compounds to aid in drug delivery. If it's a hot tumor, then we combine two elements that will [disengage] the inhibitory checkpoints or activate the checkpoints that are engaged.

Melero: John Haanen was a coauthor of Christian Blank's beautiful review in Science,[1] which described, in a very graphic way, multiparametric biomarkers and trying to calculate the probability of response based on several immunologic parameters and nonimmunologic parameters that could be involved in shaping the tumor microenvironment. And they explained how, by using these multiparameter tasks, we should [be able to] develop algorithms.

But again, the way biomarkers are usefully developed in oncology—and this goes back to the ages when we were using categorical types of biomarkers, not quantitative biomarkers—came from the fact that we could distinguish among people who were positive or negative and then make the comparisons. Now that we have a continuous spectrum, it makes things much more complicated because you have to set up cutoffs. In addition, it is more important that each combination treatment becomes, in a way, independent. So you really need to develop the companion diagnostics as a biomarker; because of the dichotomic type of quantitative biomarkers, you had to validate them prospectively. That is one of the reasons that clinical trials looking at efficacy are outpacing our capability to use immunologic biomarkers. That is my perception. What do you think, John?

Haanen: I totally agree with you. The "cancer immunogram" you were referring to is based on the parameters that we found, in melanoma, to be important drivers for a response and nonresponse to checkpoint inhibition. If you have seven parameters, for example (and I think that's a very low number of parameters; I believe there are many more), how can we combine that into one algorithm that you can use to decide what patients will need? This is extremely complicated and will also make it difficult to select because, as you said, what are the thresholds for it? And if someone is below a threshold, does this mean that he will never respond [to a given treatment]? Or is the chance so low that you can say that this is certainly not a drug that we should give to this patient? Or is it even more complicated than that? And this may be different from tumor type to tumor type and may be different for any combination that we want to give.

We are still far from having validated and good biomarkers that help us select patients. There is still a lot of work that needs to be done. At the moment, people are looking at this on a large scale, which may be the way forward. We need to look at many patients, to get enough data that will be powered enough to show whether a certain biomarker is present or not present and whether or not that is really important. Big data may help us in this respect to identify a good biomarker.

Melero: On top of that, I would add that once we give the first round of immunotherapy, either in the context of a clinical trial or in usual clinical practice, and the patient progresses after transient control or something else, then we have to monitor closely to find the reason for resistance to treatment. In that space, for example, it may help to find alternative checkpoints or other molecules that could be dominant in that setting. So, personalized medicine may have a role in addressing refractory patients. I believe that has been important in the development of LAG-3 inhibitors and could be important in other immunotherapies as well.

One of the important considerations in that area is that if the biomarker happens to be a target for treatment, it adds double value. We already know that people who have a high LDH, or a high interleukin-8 or interleukin-6, are doing poorly, and this may be telling us that something about the metabolomics of the tumor are important for the T-cell response, or that these inhibitory cytokines and chemokines can be interfered with. So if the biomarker happens to also be a therapeutic target, that is "double useful."

Zimmerman: We see trials being designed now for early drug development. I get the feeling that many of these trials are designed with a minimum number of patients, just enough to demonstrate efficacy. And then they add new arms, these very highly adaptive trials, to channel the maximum number of patients to a maximum number of drugs. But we'll never get enough data to learn in what disease category the drug actually works.

Haanen: I believe that's one way to go wrong. This is certainly true; that is how trials are being designed at the moment. The problem is that the biological data that are behind these combinations oftentimes are not very solid, and many of these combinations will fail.

One way to design a lot of interesting research is to use new combinations at an earlier stage—in the neoadjuvant situation, for example. This allows you, very early on, to get signals of what your drug is actually doing. Is it really reaching the target? Is it driving the response or not? By looking at different combinations in that setting, you may find biomarkers that will help at that stage and also in later-stage disease. You can conduct a lot of in-depth translational research and hopefully that will help us to understand the biology. I'm in favor of doing these kinds of trials because it enables us to get more out of the research.

Melero: I agree. Rationally addressing these points is very important. We have so much research on the table. Human advances have benefited from serendipity and finding the unexpected. So I agree that the current activity is likely to produce breakthroughs. The best talent in the world is now in the field of immunotherapy. When we started immunotherapy, it wasn't the brightest of us who went into the field. But now we see all this expertise and this huge amount of activity. There are many reasons to expect big breakthroughs that we would not anticipate by rational approaches.

Haanen: More than 1000 combination-therapy trials are now ongoing. And of course, you can say that this glass is half empty, because many of these trials will fail. But as you said, some of them will make a breakthrough. There is also a lot of money going through the pipeline. But I agree with you that, for sure, some of the trials will make us think completely differently than we do right now.

Melero: I usually joke that we are going to win the lottery by buying all of the tickets.

Zimmerman: A lot of work still needs to be done before we can actually move away from one-size-fits-all to a more tailored approach. But with the current technology, it's also possible to be quite personalized and going in the direction of, for example, cancer vaccines, with the personal neoantigens of the patient. The pharmaceutical industry is also pushing this; some biotech companies are developing RNA vaccines and lipid nanoparticles or peptide vaccines. We have seen at this meeting an interesting presentation regarding this, based on data that were already published but recently expanded. Do you believe that this is the future of immunotherapy? Do you think it will crack the glass ceiling that we seem to be limited by, with these nontailored approaches?

Haanen: Certainly, I hope so. We have been using vaccines for a long time. In the past, vaccines usually failed because we were not targeting the right antigens and perhaps we didn't know exactly how to vaccinate, and I believe that's still an issue. But going from a specific vaccine (and perhaps the wrong antigens) to the neoantigens is again a step forward. We have a lot of data on neoantigens to make us believe that those are the ones that are being targeted by the T cells.

But I also believe that the jury is still out. You can induce responses against neoantigens—that's clear. The question is, do these responses translate into very good, lasting clinical effects? We still need to see more data and longer follow-up to be sure that this is the case. I am sure we can induce responses against the neoantigens. But the question is, are these the neoantigens that are important for the tumor? Are they being expressed by the tumor against the T cells that also see them? That question still needs to be completely addressed.

Melero: It is not a piece of cake, logistically. We have to improve, and improve a lot, in that regard. To begin with, the time to deliver the vaccines is long. To treat a patient with later-stage cancer is really compromised. Then consider the algorithms for prediction out of the whole exome sequencing of normal cells and the tumor, and the algorithms, which are based on the known epitopes of pathogens. The systems keep learning, but for example we are quite poor at predicting class 2 epitopes, the scenario where John Haanen was a pioneer. They could be very important as well, particularly for vaccination where you need T-helper cells to cooperate with cytotoxic T cells.

Vaccine development is kind of in its infancy. The RNA approach that you mentioned is especially attractive because it can incorporate more antigens and, in a way, has what we in immunology call self-adjuvanticity, which means that the RNA itself is engineered in such a way that it will stimulate the antigen-presenting cells, making them believe that they have been infected by an RNA virus. So they start behaving as if fighting a virus. This is extremely exciting and these formulations could be quite useful. But I believe that the field is still in its infancy. I completely agree with Dr Haanen that we have to practice medicine based on evidence, not on eminence. We could be wrong, and until we see a demonstration of clinical benefit, we must remain skeptical.

Zimmerman: If I hear you correctly, you are saying that, for the time being, the logistics are restricting us more or less to an adjuvant setting or to the patient with non–rapidly progressing cancer; and even then, we don't know the key ingredients—the right antigen, the right adjuvant, and the right vector.

Melero: It will take time before we really can deliver these in a fashion feasible for a large number of patients. But I have a lot of faith now in technologic advancement. Things are miniaturized and engineered in such a way that they are feasible. Everything becomes automatic through robots, and eventually we will be surprised. These days, I get more computer power in my cell phone than they had in the NASA computer when they landed on the moon. That is something that makes me believe this can happen.

Zimmerman: Ignacio and John, thank you very much for this fascinating discussion. This is Stefan Zimmerman, speaking from ESMO 2018 in Munich, Germany.


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