Mark A. Socinski, MD: Let’s switch gears and go back a little bit. Ticiana brought this up before, and this gets back to the biomarker thing. When I want the truth about something, I always ask Heather. We’re going to talk about tumor mutation burden [TMB]. We had a little flash in the pan a year ago, I guess, at AACR [the American Association for Cancer Research], and there was a lot of optimism. That’s kind of been quelled, and I just want to get your thoughts and others’ thoughts on whether we should use it.
Heather A. Wakelee, MD: I think they’re ongoing flashes, right? And I think that’s where it’s so confusing, because again, when we’re talking about biomarkers there’s the really clear-cut, no question biomarkers like having an EGFR mutation or not. And then there are the somewhat fuzzy ones, like PD-L1 [programmed death-ligand 1], and then there’s the ones that really could be a lot of different things, like tumor mutation burden. Because to establish a tumor mutation burden cutoff, it’s going to be dependent on how many mutations you are looking for. And in most of the definitions, it’s excluding all of the known drivers. So you’re just looking at the other things that we don’t yet understand. And different assays have different numbers of mutations they’re looking at, and they’ve established different cut points.
And so, whenever I see a biomarker where there’s gradation and people pick their cut points, I get a little skeptical. I think this is one of those settings. There absolutely have been studies where they picked a good cut point and were able to show that this mattered, and others where it hasn’t. And so, it’s hard to know what the actual truth of it is. That being said, I think there is some truth to it.
When I’m trying to describe immune therapy to patients, I talk about the 2 aspects of it. You’ve got to have something about your tumor that’s different than all of your normal tissue—and that is tumor mutation burden but it’s not exactly tumor mutation burden—and then you have something you’re doing that helps the immune system recognize it. With the PD-1 [programmed cell death protein 1], PD-L1, you’re turning off one of the stop signals. And if that’s the stop signal that’s important, great. That’s really critical. But if there are other stop signals, then that’s irrelevant. And if you take away your stop signals but there’s nothing that helps the immune system recognize the tumor, then you’re also stuck. So I think both sides are probably relevant, but I don’t think that the tumor mutation burden as it’s defined today is the full truth.
Mark A. Socinski, MD: And often, depending on where you may send your samples, for those people who are sending samples outside, you often may get that in your report. We won’t name names, but, Paul, your thoughts? Because I get asked by community oncologists, “What do you do with TMB,” and I say, “I ignore it right now.”
Paul K. Paik, MD: Again, I think there are 2 ways to look at this. One is from the perspective of TMB as a binary biomarker, as you can get where you have essentially very little tumor mutation burden, or thousands. The colorectal cancer experience tells us that this is functionally important in terms of mismatch repair-deficient colorectal cancers, Lynch syndrome, right? So, in practice, using that theory, it does matter. The question becomes, in the context of other malignancies like lung cancer, small cell, non—small cell, in the spectrum of tumor mutation burden that we see, is this functionally operant as a biomarker? I think the jury is still out on that. There was that initial promise with CheckMate 227 last year. It was not borne out on overall survival. MYSTIC was not borne out.
Overall, in terms of tumor mutation burden for the tumors themselves, that wasn’t borne out. There’s this blood-based mutation burden at a different cut point that seems positive. Again, it is very difficult to make heads or tails of, in terms of what exactly is going on. But I think for all of the reasons that Heather had mentioned and the fact that it’s just very difficult to get anything consistent, it makes it a much less reliable biomarker. I’m similar to you. When I do get it, which is also by and large much after I have been treating the patient, it’s sort of, “Well, this is interesting.” Maybe it’s like a fifth-line or sixth-line therapy. I’ll give ipilimumab/nivolumab, or something like that. And so, for me, that’s the utility.
Mark A. Socinski, MD: Corey?
Corey J. Langer, MD: I divide the world into 3 groups. There are the TMB cultists, and they know who they are. They are a small minority. TMB agnostics. I think I’d probably put myself in that group. I’m not sure what to believe. I agree with Paul that the jury is out. And then there are the TMB skeptics. Skeptics can also be agnostics. I’d probably say that the majority of us are in 1 of those 2 categories. But that being said, as you both pointed out, there are tantalizing data. There was a blood-based TMB analysis. Granted, it was post hoc. It was hypothesis-generating. But in the 20% or so who had higher than 20 mutations per megabase in this blood-based assay in the MYSTIC trial, there was a survival advantage. Is that enough to carry forth a separate phase III trial comparing a CTLA-4 [cytotoxic T-lymphocyte—associated protein 4]/PD-1 inhibitor in combination versus chemotherapy? Remember the control was just chemotherapy, and that’s not what we’re using standardly. We’re using chemotherapy plus a PD-1 inhibitor.
The other group is actually one from the subanalysis from Dr Hoss Borghaei’s presentation last year at ASCO [the American Society of Clinical Oncology], and that’s what I call the immunotherapy desert—low TMB, low PD-L1. In that group, the combination was certainly no better, and frankly, inferior. There was that early dip in survival. But chemotherapy did just as well as chemotherapy plus nivolumab in that population. It was small numbers, but there is, without question, a population that we have not fully identified, where checkpoint inhibitors, at least in their current status, are probably doing the patients absolutely no favors. And that is a group where I would have equipoise. I don’t think any company would do this, but maybe the cooperative groups would. Looking at chemotherapy, perhaps chemotherapy/bevacizumab versus checkpoint inhibitor in combination with chemotherapy, they would probably do just as well.
Paul K. Paik, MD: Mark, as you had pointed out before, in CheckMate 227 we also saw, in the ipilimumab/nivolumab patients, that early degradation of overall survival and PFS [progression-free survival] before the curves cross. And that is an area of concern, still, for a subset where clearly it’s not working, and they’re just not doing as well.
Transcript Edited for Clarity