New Biomarkers for Immunotherapy: The Microbiome and Beyond

Jeffrey S. Weber, MD, PhD; April K.S. Salama, MD


June 28, 2018

Jeffrey S. Weber, MD, PhD: Hello. I am Dr Jeffrey Weber, deputy director of the Laura and Isaac Perlmutter Cancer Center at the NYU Langone Health in New York City and co-director of the melanoma research program there. Welcome to Medscape Oncology Insights. Joining me today is Dr April Salama, an associate professor of medicine and director of the melanoma program at the Duke Cancer Institute in Durham, North Carolina. Welcome, April.

A lot of attention has been paid to using biomarkers [to predict] outcomes in patients receiving our innovative immunotherapies. We have many new approvals, but tell me about the state of the field of biomarkers that we can use to select patients for specific treatments. How far along are we?

April K.S. Salama, MD: Certainly this is an interesting and ever-changing topic that is key as we begin to move forward and figure out how to better select patients for therapy. One of the biomarkers that probably receives the most attention as it relates to checkpoint inhibitor therapy is PD-L1 status, and this has been a mixed picture as we look across different disease types. Certainly PD-L1 could be used to select therapy in lung cancer, but in other tumor types the picture is more mixed. The challenge is that any single biomarker alone is probably not going to be enough to ultimately select folks [for a specific therapy]. [Biomarker status] is so dynamic; it changes over time, and that's really one of the key issues.

Weber: What about tumor mutational burden (TMB)? That has received a bit of attention. I was quite impressed that, in a recent BMS publication from one of their trials, [the investigators] stratified patients strictly by TMB.[1] That was quite interesting. Is there a future for TMB as a biomarker?

Salama: I believe that is a possibility. I don't know that it is ready for prime time today, but I certainly think it is interesting to explore. TMB is becoming more readily obtainable in commercially available assays, so clinicians have access to this information. I do believe that it has some of the same issues as PD-L1 status in that there is this dynamic; it may relate to therapy time points or heterogeneity within a tumor, so there are certainly issues. TMB also requires tissue and biopsies, which may limit its applicability to a wide range of patients.

Weber: David Rimm and Kurt Schalper from Yale published a very interesting article.[2] They showed these fantastic photos of where they would do a needle biopsy of the tumor. They had removed the tumor, and they showed a needle going in from the bottom, a needle going in from either side, and one from the top, and depending on where you put the needle, the PD-L1 staining would either be absolutely negative or it could be highly positive. This doesn't increase your confidence that PD-L1 staining in the tumor would be useful if you're going to get a core biopsy of a single lesion. If you do the numbers, if you look at the area under the curve for its value as predictive marker, it is about 0.7, which is not that impressive—at least as I was taught. Will a PD-L1 [-positive] patient be more likely to respond and do well than a PD-L1-negative patient? Yes, absolutely. My concern is that, as an individual marker, PD-L1 is not something I could routinely obtain on my patients and [reliably] tell them, "You're going to respond" or "You will not."

Salama: You could not use it to exclude someone from therapy, for example.

Weber: You raise a good point. To me, the value of a biomarker is that if it is negative, you will not begin a [specific] treatment. If a patient is BRAF wild-type, meaning negative, the patient definitely will not benefit from BRAF/MEK drugs. If the patient is PD-L1 negative, he or she might benefit from some of these immuno-oncology drugs, and if the TMB is low, they still could benefit. It depends [on whether the patient has] that one mutation that stimulates the right T cell. But I agree: More mutations is better than fewer, but this is a real problem.

Tell me about things coming up the road. What is in development that may do better than an area under the curve at 0.7? What will give you an area under the curve of 0.9? What about the microbiome?

Salama: That is a very interesting topic, as you know, and there have been a lot of recent high-impact publications by different groups suggesting that this could be a potential marker, or series of markers, to select patients. I would be interested to hear your thoughts, given that each of these groups has identified different biomarkers. How do we put that together and how do you see it moving forward?

Weber: You raise a good point. If you look at those three articles, one from Laurence Zitvogel, MD, PhD, at the Institut Gustave Roussy in France,[3] one from Thomas Gajewski, MD, PhD, in Chicago,[4] and the third from Jennifer Wargo, MD, MMSc, at MD Anderson in Houston,[5] they each enrolled about 100 patients, either with lung cancer or melanoma. They collected feces pre-treatment and looked at outcomes with PD-1/PD-L1 antibodies. They all had slightly different [findings]: Bifidobacterium for Tom Gajewski's group, and Bacteroides and Clostridium for the others. I believe that we will see a pattern and that Bacteroides and Clostridium are going to be the good guys and the bad guys.

I like the idea but it needs to be validated in a large trial. Someone needs to get 1000 or 2000 fecal samples from some adjuvant trial, and then you have some numbers. Then you need to sequence the ribosomal DNA which is how you characterize the bacteria, and then do what they call shotgun sequencing, which gives you information about the function, and make some real decisions. This is not cheap; this is expensive. Once that is done, we will have an answer. It will either be important or it will not. The other good news is that if you have the wrong bacteria in your gut, you can change it, [whereas] if your TMB is low, you can't raise it.

I guess you could alter the microbiome with antibiotics. And by the way, there was this thing about antibiotic use being a negative factor. I am not convinced. But you can cure C difficile with a fecal transfer, which, by the way, requires 200 cc's of liquid fecal material being inserted into your colon via colonoscopy. I am not so sure I would volunteer for that if I were a patient, but you could lyophilize bacteria and put them in a pill and swallow them. You could go the other way, which wouldn't be popular. The nice thing about the microbiome is that you can actually make an impact. I like that.

What about other, more systemic peripheral blood markers? You are familiar with this.

Salama: This is an area with a lot of excitement because, as you pointed out with the analogy to BRAF, we have had more and more patients ask things like, "Well, do I have a PD-L1-positive tumor or not?" What is so interesting about some of the current markers is that they are dynamic. If your tumor is PD-L1 positive today but you progress on therapy, can we really say what the status of that tissue is 6 months from now? Invasive sampling through tissue carries risk, and we believe that peripheral blood markers may be a way to look at that. The technology is advancing. At Duke, some of my colleagues are working on being able to isolate melanoma cells based on pigment, based on different photo wavelengths and photo acoustics; and being able to conduct detailed genetic analyses of these tumors in addition to being able to look at different immune cell subpopulations. There certainly have been data out around the ratio of T-regulatory cells to CD8-positive T cells.

Weber: But photo acoustics. Tell me about that.

Salama: This is a complicated technology. It is very preliminary work at this point, but [researchers are] looking at being able to isolate melanoma cells based on the different wavelengths that the melanin absorbs [relative to] other cells.

Weber: This was developed from a CIA technology used to detect Russian submarines, which I thought was very interesting. I heard about this a few years ago, but I haven't heard anything since. And the other [developments you mentioned] are interesting. It would be so much nicer to use a blood sample [than a biopsy]. You can get a blood sample any time as opposed to getting a biopsy. The problem with the biopsy is that a biopsy is invasive, potentially morbid. It could differ from tumor to tumor; your lung met [metastasis] could be different from your liver met; and as Schalper and Rimm showed, even within the same met, you might find a difference in your marker. It would sure be nice to get a blood biomarker to measure these cells.

Salama: And [to be able to get them] at different time points on therapy with an ultimate goal of helping to uncover other mechanisms of resistance and potential ideal combinations.

Weber: Everyone focuses on the tumor, but I have been involved with some work with collaborators looking at the proteomics in the serum of patients receiving checkpoint blockade. Of interest, one of the usual suspects always pops up, and it's C-reactive protein and other acute-phase reactants. It turns out—no big surprise—that acute-phase reactants are bad for you. So, chronic inflammation, bad; acute inflammation, good. And that may help us develop some new biomarkers in the future.

April, thank you very much for joining me. This has been a fantastic discussion. This is Dr Jeffrey Weber for Medscape.


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.