Mark A. Socinski, MD: We know that a very similar trial design, CheckMate-026, which enrolled patients greater than 1%, compared single-agent nivolumab to platinum-based chemotherapy in that population. It was a negative trial. Here, we have KEYNOTE-042 for the greater than 1% population. This is a positive trial. Dr. Langer, what are your thoughts?

Corey J. Langer, MD, FACP: I predicted that this trial would be negative, largely because of CheckMate-026. Our track record there, in a less selected population, is completely negative. It wasn’t worse than chemotherapy, but nivolumab showed no advantage whatsoever. In stark contrast, we saw a striking response in the progression-free survival and overall survival advantage. So, I was highly skeptical about this trial. I figured by including anyone with any degree of PD-L1 expression that we would dilute down those results. I figured that we’d see a trend because we were still including a fair number of folks with 50% or higher expression. But, in fact, the trial was positive. It’s positive at each of the 3 cut points—at 50%, at 20%, and at 1%. We see an 8-month advantage in median survival for the highest expressers—12 months going up to 20. We see a less striking advantage for the 20% or higher group—16.7 versus 13, and then 15-something versus 12 for all-comers. I think the big question becomes, is this advantage driven solely by the 50% or higher group, or are we also seeing it in the 1% to 49% group?

Mark A. Socinski, MD: It’s important to note that the 50% or higher group represented 50% of the patients in the trial.

Corey J. Langer, MD, FACP: That’s not a surprise. When you look at KEYNOTE-189, and even at KEYNOTE-021 cohort G, it’s roughly a third in each. So, if all-comers had to have some PD-L1 expression, you’re splitting that 67% or so. And 50% would be 50% or higher.

Benjamin P. Levy, MD: I just have a question to ask the group. How do we reconcile CheckMate-026 with KEYNOTE-042? In CheckMate-026, greater than 1% patients were allowed, but they looked at greater than 5%. But that was the majority of patients.

Karen Kelly, MD: Even the 50% did not do well.

Benjamin P. Levy, MD: How do we reconcile the fact that there are some limitations that we need to think about with KEYNOTE-042? We’ll need to look at the subset from 1% to 49%. Are these drugs different? This has come up a lot recently.

Mark A. Socinski, MD: It was but, again, there was nowhere near the greater than 50% PD-L1 expression in CheckMate-026. I think that’s a factor. I think there were some bits of unluckiness in CheckMate-026 that led to some imbalance of known prognostic factors. Again, these were not quite equal trials because of these things. It used to be based on the second-line data. I thought that nivolumab, pembrolizumab, and atezolizumab were Coke, Pepsi, and Dr. Pepper. As the first-line data have emerged, I’m not sure that this is the case, right?

Benjamin P. Levy, MD: Right.

Mark A. Socinski, MD: It will be interesting, moving forward, in terms of how we look at these agents. We had the debate on the greater than 50% population. Would you use monotherapy or chemotherapy? We talked about that. What about for the 1% to 49% group?

Benjamin P. Levy, MD: It will depend on KEYNOTE-042. If that shows a benefit, the question becomes, do you use single-agent immunotherapy if it’s positive? Or, do you use the triplet regimen?

Mark A. Socinski, MD: Do we have the same argument about the bulk of disease, symptomatic patients, and that sort of thing? And maybe for those low-volume patients who are either not symptomatic or are minimally symptomatic, you would do immunotherapy as monotherapy, even in that population.

Corey J. Langer, MD, FACP: Certainly in older, frailer patients who are chemotherapy-averse. But I think I’d have a lower threshold in the 1% to 49% group to continue with chemotherapy and immunotherapy in combination.

Karen Kelly, MD: We’re going to see the pie slicing start to happen here, as we’ve already seen. We cut it in half with the 50%, but we’re going to start seeing the slices become even more refined. So, I think that we have to wait to see the data to do that, but it’s a challenge.

Corey J. Langer, MD, FACP: Do we need to do PD-L1 testing? I think this is an argument very much in favor of doing it. As time goes on, we will be looking at different populations. And then, if TMB hits on survival, we’re going to have to figure out a way to integrate that. I’m just hypothesizing. TMB and the survival advantage is realized in CheckMate-227. It’s not part of our clinical armamentarium, but should that happen, I would personally have equipoise, in a PD-L1 0% population with high TMB, for comparing chemotherapy and immunotherapy, chemotherapy/pembrolizumab, with the ipilimumab/nivolumab combination.

Benjamin P. Levy, MD: We may need that.

Transcript Edited for Clarity