Corey J. Langer, MD, FACP: I have problems with the way the data were presented. Certainly, at the AACR meeting, the implication was that it was no more toxic then chemotherapy.

Karen Kelly, MD: Yes, right.

Corey J. Langer, MD, FACP: Granted, I was not involved in the study, but I certainly, like all of us, have given patients ipilimumab/nivolumab. Some patients sail through, but others really crash and burn. It is a potentially very toxic regimen.

Mark A. Socinski, MD: In this population, I think you raise the ante, in terms of risk of toxicity.

Corey J. Langer, MD, FACP: Intriguingly, as Karen has pointed out, the relevant toxicity table is actually buried in the supplement, in the New England Journal of Medicine. The folks who had high TMB, who were exposed to ipilimumab/nivolumab, were on it longer. Therefore, they had the opportunity for more toxicity and, clearly, did have more toxicity. The second issue is that not all grade 3 and 4 toxicity is alike. Myelosuppression, which is going to be much more common with chemotherapy, is frequently totally asymptomatic. So, we have to distinguish between clinical toxicity and paper toxicity. Most of the ipilimumab/nivolumab toxicity was clinical.

Mark A. Socinski, MD: It was not paper toxicity.

Corey J. Langer, MD, FACP: Not at all.

Mark A. Socinski, MD: Yes, exactly.

Corey J. Langer, MD, FACP: The other point that I want to make is that this trial underwent a number of events as it went along. For the less than 1% group, they actually included another amendment. They added nivolumab to chemotherapy and compared it to chemotherapy, alone. So, it’s a separate subset. Hoss Borghaei is reporting on the data at this meeting, which does show a higher response rate. There was a higher progression-free survival in the 0% group [NOTE: CheckMate-227 study stratified patients by PD-L1 expression ≥1% or <1%]. The analogous trial would potentially be KEYNOTE-189, although this is for both the squamous and nonsquamous patient populations. We haven’t seen the survival data yet.

Karen Kelly, MD: The other point that is important, that has been consistently seen, is that PD-L1 and TMB are independent of each other. They saw benefit in the lows and in the highs. There was a little more benefit in the nonsquamous patient population.

Corey J. Langer, MD, FACP: But it was independent of histology.

Karen Kelly, MD: Yes.

Mark A. Socinski, MD: Ben?

Benjamin P. Levy, MD: One other point that was mentioned not in the appendix but very deep in the discussion, or in the results was that they did a secondary analysis looking at single-agent nivolumab versus chemotherapy and used a TMB cutoff of 13. There was no difference. So, I’m not sure of the biological rationale for dual checkpoint blockade outperforming chemotherapy for high TMB, whereas single-agent doesn’t. I don’t think we have ironed that out.

Mark A. Socinski, MD: Yes, I think there are a lot of issues with TMB that we haven’t resolved at this point, and that’s one of them. CheckMate-026 suggested that the TMB level could be helpful in distinguishing this. I am just concerned that the message that we thought CheckMate-227 was going to bring us is going to be diluted with all of these subset analyses in different populations.

Corey J. Langer, MD, FACP: The overall trial had something on the order of 1700 or 1800 patients. This is only 300 patients or so from that much larger trial. As far as I’m concerned, it’s tantalizing. It’s certainly hypothesis-generating. It’s not yet a game-changer. I need to see the survival data.

Mark A. Socinski, MD: Walking away from the CheckMate-227 presentation, what advice would you give to your community colleagues? Is TMB a standard that we should be looking at?

Corey J. Langer, MD, FACP: I think we’re on the cusp.

Mark A. Socinski, MD: It’s not ready for prime-time.

Benjamin P. Levy, MD: I think it needs refinement. It’s certainly something that you can obtain when you send off a FoundationOne report. You’re going to get the TMB on these patients. Is it helpful in treatment-naive patients in the context of KEYNOTE-189? I don’t think so. Every patient population in KEYNOTE-189 derived a benefit, even those never-smokers. There may be certain instances when I would consider it—when treatment decisions become individualized. Certainly, we’re not talking about small cell disease, but we see it in the refractory setting. TMB may perhaps drive an effect with dual checkpoint blockade in small cell disease. But for non—small cell disease, it’s not something that I would use in everyday practice, yet.

Karen Kelly, MD: That hazard ratio in KEYNOTE-189 is so good, and this does not compare with that hazard ratio.

Mark A. Socinski, MD: And, as you pointed out, the overall survival is not positive.

Karen Kelly, MD: It’s immature.

Corey J. Langer, MD, FACP: This brings up another point. Sixteen months in the control arm is better than we had seen in IMpower150 and certainly in KEYNOTE-189. So, this is not our average patient. They had to wait until TMB was actually available. There’s a time element here that’s not…

Karen Kelly, MD: Well, they didn’t wait.

Corey J. Langer, MD, FACP: They were randomized but, still, TMB was a feature.

Karen Kelly, MD: Right.

Corey J. Langer, MD, FACP: Certainly, in clinical practice, we’d have to wait if we were going to select people on that basis.

Karen Kelly, MD: Remember that if you’re a squamous cell patient, you’re not getting next-generation sequencing. You would have to be getting next-generation sequencing for patients, just for that piece. The other issue is that various assays and platforms are out there. There is an ongoing harmonization study for TMB so that we can answer the questions just like we had with PD-L1. I think that will be important. It appears that having a TMB greater than 10 is going to be where the cutoff will be. But we’ll see. Others have higher cutoffs. I think that as long as you’re above 10, you’re probably going to be in the general ball park. Can we refine it? TMB, remember, is crude.

Corey J. Langer, MD, FACP: It’s more complicated than getting a PD-L1.

Benjamin P. Levy, MD: It takes more time, too.

Karen Kelly, MD: It takes 3 or 4 weeks. These are, again, some of the issues that are going to be challenging to our community physicians. You really would have to send TMB out.

Transcript Edited for Clarity