CheckMate 227: TMB Versus PD-L1 in Advanced-Stage NSCLC

Transcript:Benjamin P. Levy, MD: We’re going to move off chemotherapy as a first-line backbone for advanced NSCLC with immunotherapy and talk about the role of dual checkpoint blockade with PD-1 and CTLA-4. We’ve had some very thought-provoking data published in the New England Journal of Medicine recently with CheckMate 227. Sanjay, do you want to walk us through those data briefly and tell us about your insights?

Sanjay Popat, PhD, FRCP: CheckMate 227 is a large trial. It’s quite a complex trial aiming toward several questions. It builds on the background that Bristol-Myers Squibb has that the potential for combining a PD-1 inhibitor and a CTLA-4 inhibitor, ipilimumab and nivolumab, may potentially have a superior efficacy than just using each as a monotherapy. In CheckMate 227, patients with advanced disease were randomized to either a combination of nivolumab and ipilimumab or histology-specific chemotherapy with PFS [progression-free survival] and OS [overall survival] as coprimary endpoints. Outcomes are being presented according to the PFS, PD-L1, and the tumor mutational burden [TMB]. Because at the same time all of this has been going on, we’ve had a number of academic data sets that have generated a nice signal that tumor mutational burden—defined as the number of nonsynonymous mutations per megabase—seems to be influencing the benefits of the combined immunotherapy.

What the data have shown is that there seems to be potential benefit for nivolumab/ipilimumab versus chemotherapy. But potentially, it’s restricted to those who had high tumor mutational burden, defined, according to the CheckMate 227 data set, as patients with 10 or more mutations per megabase. But at the moment, this is restricted to a PFS endpoint, and we do not have any overall survival data as of yet.

Benjamin P. Levy, MD: That was a very nice review. Where does that leave TMB as a biomarker? Is this something that we should be doing? Is it for research use only? Is there ever a time in the treatment continuum of a patient with non—small cell lung cancer when it makes sense to do it off a clinical trial? Does anyone want to tackle that? Sanjay?

Sanjay Popat, PhD, FRCP: I think we’re at the beginning of the TMB story. These are data that are exciting, new. One of the key aspects of the whole TMB story is that TMB-positive patients, whatever your definition is, probably represent a separate group from those with PD-L1 expression. Ram, you’ve been involved in a trial with Solange that has demonstrated that TMB-high patients are a separate group of patients from strongly PD-L1—expressing patients. We do have different groups of patients. Now the challenge is defining TMB, ensuring we have homogeneity in what the definition of TMB is and how we can implement TMB in routine practice. Because in CheckMate 227, the application of TMB was retrospective to outcomes. Patients weren’t randomized on the basis of TMB. It was a retrospective analysis of the TMB status.

Suresh S. Ramalingam, MD: I want to draw on something Sanjay said, which we found in CheckMate 227 and other studies as well. The high-TMB group and the high—PD-L1 group of patients were not one and the same. There’s a direct discordance in who is high PD-L1 versus high TMB. What I think TMB does is help you select a subgroup of patients who don’t have high PD-L1 but who could still benefit from immune checkpoint inhibition. Even along the low–PD-L1 group, we now have data that show that if your TMB is high, you will actually have better outcomes with immunotherapy combinations compared with chemotherapy alone. I think we will be studying this further. We all agree that PD-L1 expression is an imperfect biomarker, and maybe TMB will turn out to be another imperfect biomarker, but it will be helpful, regardless, to select treatment for certain patients.

Solange Peters, MD, PhD: I’m quite convinced about these TMB data myself, and the fact is that we’d like to identify for our patients some modalities of treatment that might delay exposure to platinum-based chemotherapy. We can still see that the magnitude of high-grade toxicities by using immunotherapy is always significantly less than that of platinum-based chemotherapy. The problem, as you said, is really how we can routinely assess TMB and have a result in a decent time frame. In most other places in the world, it’s still not feasible. But it will come in the course of probably the next year. There are many efforts of harmonization. When it comes, it will just complete the picture. We’ll have more data across other trials, but it will complete the picture. I’m sure there will be a nice algorithm allowing us to classify patients according to what would be the best frontline treatment option.

I’m sure ipilimumab/nivolumab will come. It will come for PD-L1—negative, high-TMB disease, and it might also come to some subgroups where platinum-based chemotherapy is the best thing we have, right? It will really help us in classifying patients, and we should not feel beaten by the complexity of a new biomarker. Remember how PD-L1 was complex? Now we can move with TMB and try to be rational about it.

Benjamin P. Levy, MD: At my institution, we have moved to an in-house NGS [next-generation sequencing panel] that calculates TMB. I’ll be very honest with you: I’m not sure what to do with those data right now. I think you’re right. I think after more refinement, we may have some clues and hints about how to use this. Is it in the high-TMB, low—PD-L1 setting, where patients are offered dual checkpoint blockade? As Sanjay discussed, I think we’re in the very early stages of understanding how to implement this biomarker in tissue and, dare I say, blood with some of the trials that were presented at ASCO [American Society of Clinical Oncology Annual Meeting], such as the B-F1RST trial, looking at TMB in the blood.

Transcript Edited for Clarity