Herbst Tackles Key Questions in Emerging PD-1/PD-L1 Immunotherapy for NSCLC

Beth Fand Incollingo
OncLive
Published Online: Saturday, August 2, 2014

 
Dr. Roy S. Herbst

Emerging agents and pressing research questions in immunotherapy were reviewed in depth during the 15th Annual International Lung Cancer Congress, with Roy S. Herbst, MD, PhD, of the Yale Cancer Center, leading the way.

Herbst was the recipient of the 2014 Addario Lectureship Award, which the Bonnie J. Addario Lung Cancer Foundation bestows annually at the Physicians’ Education Resource (PER) conference. 

At Yale, Herbst serves as the Ensign Professor of Medicine (Medical Oncology), professor of Pharmacology, chief of Medical Oncology, and associate director for Translational Research. He also is serving as a co-chair of the Lung-MAP trial, an innovative biomarker-driven study that is evaluating five novel therapies for the treatment of squamous cell non–small cell lung cancer (NSCLC).

As this year’s Addario honoree, Herbst delivered a keynote lecture that explored the biology of the immune system in lung cancer. In addition, he led a panel that discussed the future of immunotherapy.

He also specifically discussed four monoclonal antibodies in development that target either programmed death-1 (PD-1) or its ligand, PD-L1. The anti-PD-1 agents are nivolumab from Bristol-Myers Squibb and pembrolizumab (MK-3475) from Merck. The agents targeting PD-L1 are MPDL3280A from Genentech and MEDI4736 from MedImmune, AstraZeneca’s biologics research and development arm.

In an interview with OncLive during the conference, Herbst shared key takeaway messages about the immune system and immunotherapy in lung cancer.

OncLive: The proteins PD-1 and PD-L1 have emerged as the foremost targets for lung cancer immunotherapy. What is known about this treatment strategy so far, and what still needs to be determined?
Herbst: The exciting news is that we’re seeing response rates upward of 20% across the board with all the different antibodies in unselected patients with refractory lung cancer. We can do even better if we select the patients and enhance using biomarkers, which are still very early in development. With biomarker assays, you can bring that response rate up as high as 40% to 50% or more.

We still have to figure out what to do with patients who don’t benefit. Are there ways to enhance the immune response, combine therapies together, combine immunotherapy with targeted therapy, with standard chemotherapy? But clearly the development of this field for lung cancer, along with other tumor types, is something we should keep a close eye on. It’s very important for patients. The clinical trials are accruing quite rapidly, and many of these include tissue analysis, and old biopsies or repeat biopsies, to figure out mechanistically how this is working.

MEDI4736 is especially interesting to us because in the large lung cancer master protocol [Lung-MAP], which I co-lead, we are including an arm of immunotherapy [MEDI4736]. I’m excited about that because it gives the 40% or 50% of patients in this large NextGen-sequenced, selected protocol the option to receive immunotherapy versus chemotherapy, and also because we’ll have access to their sequencing information. Hopefully, we can begin to learn a little bit more, mechanistically, as to who responds best to some of the immune therapies.

Have PD-L1 expression levels been established as a biomarker for therapies targeting PD-1 or PD-L1? If so, is there a valid test for measuring the levels, and have thresholds been established?
It’s under development. Currently, these assays tend to be propriety to each of the groups that are developing the antibodies. But these assays are quantitative as best they can be. Most look at the PD-L1 staining on the tumor; some look at PD-L1 staining on immune infiltrate, which will include tumor but also immune cells—the entire microenvironment.

The cutoffs are picked using whatever data are available at the time you set a cutoff; then, you validate it as you treat more patients. Most groups now have locked into what they’re using as a cutoff, because you have to have your cutoff set before you start a big phase III trial. I think these cutoffs will enhance the number of patients who will benefit from immune therapy. Will they be the ultimate cutoffs that are used, the ultimate antibodies that are used, the ultimate tests? I don’t know.

I think there’s going to be a desperate need for some sort of harmonization of these tests. How can any one hospital run four different tests? Even now, as we do clinical trials, to have different tests for different drugs is quite confusing. But until something’s approved, we work with the best assay we have for any given study.

Are there any other markers that might predict which patients would be the best candidates for immunotherapy?
Yes, some. We can look at gene signatures or RNA expression profiles. Those, of course, are hard to use as predictive markers, but they at least allow us to understand what’s going on. There are some markers of T regulatory cells, suppression T cells, that can be looked at and can help one gauge how turned on or off the immune microenvironment is. So there are a number of things under investigation. Right now, PD-L1 is the main focus.

Is there any means for new biomarkers to emerge out of the Lung-MAP trial?
There are ways that will happen. We’re sequencing many genes, but only using a few of them to sort the patients. So we will see if any of the other abnormalities happen to assort. One drawback of the platform is that we’re not doing “match normal,” so the chance of finding new variants is a little bit low, though we can still do it using different bioinformatics algorithms. If we see exceptional responders, we can go back and go deeper, and maybe do a whole genome analysis of those patients’ tumor samples.

Discovery is a big part of this. For example, the arm getting immunotherapy is unmatched, so we’re giving it to those patients regardless of their PD-L1 status, yet we’ve talked at this meeting about the fact that PD-L1 may be important. So now we can look at those patients and look at those who respond and don’t respond, and say, “What is it about their mutational status?” We’ll have that. We can go retrospectively and look at PD-L1 status, so we can do biomarker discovery. I truly believe that you must do the best you can for patients in real time, but learn as you go and get as much information as you can to help the next patient better.

Getting back to the Addario lecture, what else did you discuss?
We talked about the specificity, adaptability, and memory that are inherent to the immune system, and how that can be used to tackle lung cancer. One point I made was that the very thing that makes lung cancer so difficult to treat—the many mutations and heterogeneity—could be an advantage to using immune therapy against lung cancer, because having a large number of mutations per cell does give more opportunity for T cells to recognize and attack a tumor. This PD-L1/PD-1 interaction inhibits or cloaks the tumor so that it can’t be killed by the immune system in a large number of patients, and now we’ve learned how to identify it and block it and reinvigorate T cells.

And science needs to accompany this. In the oncology world, we tend to very quickly say, “Oh, here’s a drug, and here’s another drug that works—let’s combine them together.” That’s all well and good, but we really—more than ever before, now that we’re targeting multiple compartments, tumor cells, the microenvironment, immune cells, and endothelial cells—need to think carefully as to how we do these studies.

Some of them need to be preclinically modeled. Preclinical models are difficult because, by virtue of the way they are developed, many of the models we use are immune incompetent, so we can’t really test immune-modulating agents there. There are some models that can be used, and for the most part, clinical studies now can be started early—phase I trials—but I would make the case that those trials should include biomarkers, pre- and post- biopsies, ways we can get a sense of what’s happening in the immune microenvironment, so we can either enhance it or maybe not stimulate it as much.

We also have to watch for toxicities with these agents. Many of these agents are not completely without toxicity, and while they don’t produce the usual cytotoxic toxicities of a platinum—for example, low blood counts or neuropathy— they can produce endocrinopathies, issues with the thyroid, adrenal glands, [factors] that one has to watch for very carefully with these immune-altering therapies.

At the annual meeting of the American Society of Clinical Oncology this spring, you presented a study on tumor infiltrating lymphocytes (TILs) being associated with a better outcome in NSCLC. How can measuring TILs be used in immunotherapy?
It’s not enough just to have the immune checkpoint PD-L1, and to block that, if you don’t have TIL cells—they’re the effector arms of the whole process. We’ve seen that there are lung cancers that might have PD-L1 expressed on the tumor, but there are no immune cells; the tumor has never been inflamed. For whatever reason, there’s a zone where the T cells can’t get into the tumor, and we need to figure out how to break that barrier. Is it something about the endothelial cells? Do we need to target the angiogenesis, the epithelial cells? Do we need to, in some way, inflame the tumor?

Do we need cytokines, interleukins—for example, interferon? Other things to sort of stimulate the tumor to bring the T cells in are all going to be incredibly important questions. Then, there’s also the whole idea of CART cells [chimeric antigen receptor T cells], bringing in reprogrammed T cells. That’s something ongoing in lung cancer—not with the most momentum, just because finding unique antigens in lung cancer has been hard—but there are so many possibilities right now.

- See more at: http://www.onclive.com/conference-coverage/ilcc-2014/Herbst-Tackles-Key-Questions-in-Emerging-PD-1-PD-L1-Immunotherapy-for-NSCLC#sthash.GUQFO62p.dpuf