The Role of Mutation Burden as a Biomarker in NSCLC

Transcript:

Everett Vokes, MD: Is mutation burden something you consider of value in second-line therapy?

Roy Herbst, MD, PhD: Not really, because we’re going to give the drug to the patient anyhow. That said, I think mutation burden is certainly a surrogate for patients who are in that higher group to respond, and I think it is going to play a role—probably not alone, but probably with PD-L1 or a surrogate for PD-L1, some sort of gene signature, and perhaps biomarkers to be named later. I would, again, say that’s probably going to be something that looks at the immune microenvironment in some way—looking at the T cells and the quality of the T cells. But, yes, it’s certainly very helpful to have that. Not in the second-line, but it might help more in the frontline setting, or maybe even as we start to think about adjuvant therapy or earlier disease.

Everett Vokes, MD: Yes. Fred?

Fred Hirsch, MD, PhD: I know we have Naiyer here, in the panel, who has been pioneering the mutation burden, and I think the mutation burden is a very interesting potential biomarker. I’m a little bit more skeptical than others on this stage for the following reasons. I think we will learn that all mutations are not equal. I think we will learn that there are differences in the mutation burden depending on what kind of mutations we are talking about. Some mutations are more immunogenic than others, but we don’t have that knowledge, at this stage.

My question is, and Naiyer might correct me, do we have a harmonized and standardized way to define mutation burden? Is it based on 16 genes? Is it based on 80 genes? Is it based on 300 genes? Does it matter? I think there needs to be some standardization, also, with this biomarker (as we have talked about earlier).

Those are 2 reasons why I’m a little bit skeptical on this issue at the current stage. I do agree with Roy, based on the CheckMate-026 data, where, of course, a tumor burden by itself made a difference in favor of nivolumab. But if you are looking on the curves, the curve, which flattened (and that is a curve we would like to see in the future), is based on high mutation burden and high PD-L1 expression. So, that curve, even if there are a small number of patients in this stage, and the combination approach, also for biomarkers, need to be studied further. I think we are focusing too much on one single assay rather than what a combination of assays can perform in the future.

Roy Herbst, MD, PhD: Rather, a harmonization project for tumor mutation burden.

Fred Hirsch, MD, PhD: We haven’t talked about that yet.

Roy Herbst, MD, PhD: I think it may be a good thing.

Everett Vokes, MD: Naiyer, you’ve thought a lot about this?

Naiyer Rizvi, MD: To do a mutational burden analysis correctly, you have to do whole exome sequencing. These extrapolated mutational burdens do tend to break down for lower mutation counts. I don’t know that it’s ever going to be a perfect tool because it’s just extrapolated data. And if you look at CheckMate-026, the upper tertile of mutations (where patients did the best), had around 250 mutations or greater. That’s per exome, so that’s about 8 mutations per megabase. But to actually achieve a threshold with some of the foundation panels, you have to get 15 or 16 mutations per megabase. So, you’re really just picking the top high-level mutation. I don’t think it’s every going to be very, very good to just do it with a targeted panel.

Everett Vokes, MD: It’s not just that it’s not standardized, the methodology is not available yet to be precise enough?

Naiyer Rizvi, MD: I think you’re never going to be able to perfectly align it with exome data, because there’s not enough genes covered.

Suresh Ramalingam, MD: And the key is, you’re really trying to find out which mutations accord to a neoantigen that’s recognized by the immune system. The mutation burden is, rather, a broad basket to catch, as with many of these mutations. There isn’t an ability to sort out one from the other, and until we have the ability to do that, this is going to just be a broad estimated enrichment method, rather than a true biomarker.

Everett Vokes, MD: Just like PD-L1, which is an enrichment rather than a true biomarker?

Naiyer Rizvi, MD: Right.

Transcript Edited for Clarity