Benjamin P. Levy, MD: So, this whole concept of biomarkers linked to immunotherapy is a very confusing one. It’s one that we’re just beginning to tell the story of and trying to understand what a reliable biomarker is. And what that means is defining molecularly, or within the tumor microenvironment, those patients and those tumors that are most likely to respond to these agents. And that’s the way we’re heading for all of lung cancers, precision medicine—defining a target and developing a drug against that target.
This is a little different. We’ve got a great drug that works in a broad spectrum of patients. But now we’re trying to sort out, “Okay, who are those patients that are more likely to respond?” If we look at the data across the board, only 20% of all lung cancer patients respond to these agents. And the question is, why are 80% not responding? But more importantly for the 20%, who are the ones that really respond, and why do they respond?
There’s been a lot of work with PD-L1. PD-L1 is a biomarker. It’s an immunohistochemistry test that we can do on a patient’s biopsy. And it turns out, long story short, the higher your PD-L1 expression is on the tumor, the more likely you are to respond to these agents. That said, if you’re negative for PD-L1, there’s still a chance you can respond. So, the question is, is it a reliable biomarker to test for PD-L1? And it turns out that they all have different cutoffs. They all have different terms for what’s positive versus what’s negative. There’s no harmony.
There are four different platforms for testing for PD-L1, and what we found is that there’s a discordant rate between these platforms. And it really begs the question, is this a reliable test if we have different platforms that can be used, what we call the negative predictive value of the test? If it’s negative, there’s still a chance that patients could respond. Currently, there are two drugs approved for lung cancer: Opdivo (nivolumab) does not require a PD-L1 test and Keytruda (pembrolizumab) does require a PD-L1 test. That’s just the way the data shook out.
Outside of PD-L1, there are other biomarkers. In fact, at ASCO, there are a tremendous amount of data looking for better biomarkers than PD-L1. One is total mutational burden. This is looking at how many mutations are in the cancer. And you can actually calculate this doing something called whole exome sequencing. The concept is the more mutations you have, the more likely you are to upregulate antigens in the tumor. The more likely you are to upregulate antigens in the tumor, the more likely the immune system is to recognize the cancer’s foreign.
So, that’s certainly another biomarker. I’m not sure if it can be recapitulated in a commercial test, but it’s something that’s being looked at. There are blood tests that are also looked at. There’s some data we got from ASCO this year looking at cytokine levels in the blood and whether that would be something we could use. I can tell you right now the biomarker search for immunotherapy is still in its infancy. I don’t think we really know yet.
This is a very complex story, this interplay between the immune system and cancer. This is very different than the targeted therapy paradigm. Targeted therapies, mutations are fixed. You either have them or you don’t. If you have them, you give a targeted drug. Immunotherapy is a little different. It works in very complex ways that employs the immune system, and I think it’s going to be a bit longer time before we really have answers on this.