Practice Changing Trials in Advanced Non-Small Cell Lung Cancer - Episode 7
Benjamin P. Levy, MD: Does 189 (KEYNOTE-189), or even 021G (KEYNOTE-021 cohort G) for that matter, answer the clinically relevant question? These are patients, all PD-L1 expressers, being randomized to carboplatin/pemetrexed versus carboplatin/pemetrexed/pembrolizumab. So, the greater than 50%, which I assume will be about one-third in each arm; the control arm is carboplatin/pemetrexed, not pembrolizumab. And so, it really doesn’t answer a question for me. I assume the greater than 50% are going to do better with a regimen with pembrolizumab than a regimen without pembrolizumab. And then the less than 50% is the group that I’m really interested in seeing something, whether that will really be powered to show survival differences. I think they’re going to look at this. But does this answer the question, will 189 lead us down a road of conclusion and what to do?
Hossein Borghaei, DO, MS: Well, again, we don’t know at this point. I think those are really good questions that all of us are asking. And, again, as I said, is the overall survival in 189 driven entirely by one particular group based on PD-L1 level of expression? I don’t know. Is it possible that we’re going to get mutation or burden analysis as part of 189 and that would be a better-clarifying biomarker as to what to do with this patient population? But we simply don’t have the data. However, I would argue that if you have a young patient with good performance status and with a huge tumor burden sitting in front of you today in the clinic without an EGFR and without an ALK, and you’re looking at data that say with a triple combination you have a 55% response rate versus your usual 30%, it is hard to argue that you should not use the triple combination because you don’t have the subgroup analysis done, right?
Benjamin P. Levy, MD: Right.
Hossein Borghaei, DO, MS: So, I think when you’re making the day-to-day clinical decisions, based on what we have now as imperfect as it is, you have to consider the triple combination as a potential option for somebody who needs it. Now, I would not give the triple combination to somebody who I know has a PD-L1 expression of 50% or higher.
Benjamin P. Levy, MD: You would not.
Hossein Borghaei, DO, MS: No. I would say clearly in that patient, based on everything that we just discussed, 3-year median survival with follow-up, why do I need to bring the cytotoxics into it?
Benjamin P. Levy, MD: Even if they’re a never-smoker?
Hossein Borghaei, DO, MS: Well, again, that’s not the population that we necessarily have the data on. But sure, if they don’t have an EGFR, they have 50% or higher.
Benjamin P. Levy, MD: You’re giving them a single agent.
Hossein Borghaei, DO, MS: I would definitely give them single-agent pembrolizumab.
Roy S. Herbst, MD, PhD: But, Hoss, even in the 50% group, 55% did not respond, 45% responded. So, you treat someone with PD-1 alone, you give a few cycles, they’re starting to grow, do you add chemotherapy in there?
Hossein Borghaei, DO, MS: Well, that’s the subject of a study that we have proposed to do as part of the NCI Cooperative Groups. We don’t have the answer to that. But if I look at the cohort G data and now KEYNOTE-189, and maybe even IMpower, I can make an argument reasonably that maybe I try 2 cycles with the combination and see if I get a better response out of that patient. So, I don’t think it negates that if you started with the I-O (immune-oncology) alone and you don’t get the desired effect that you should not or cannot add the chemotherapy. What other options do you have in that patient except going back to your traditional platinum doublet where you know the response rate is 20% to 30%?
Roy S. Herbst, MD, PhD: I agree. The way I’m looking at these data, even before 189 came out, is I’m very clinical trial focused, as I think we all are. But in the real world, we don’t always have a clinical trial. It would be nice if we did. But if someone is PD-L1—high, I agree, I’m going to give them pembrolizumab, although if they’re symptomatic, you can make a case for giving them the pembrolizumab. But if someone is low, in the absence of a trial, I’m using the combination. Again, it’s only for the nonsquamous. I’ve been doing this a long time.
Benjamin P. Levy, MD: For all patients, Roy? Any patient who walks in with less than 50%, you’re generally offering the triplet therapy?
Roy S. Herbst, MD, PhD: Well, first we look at one of the 10 different trials we have. Of course, doing the Cooperative Group trials first. However, if the patient isn’t qualified, the only patients I’ve seen with metastatic lung cancer absent a driver, actually even with a driver, who have lived long-term—2, 3, 4 years—have gotten a PD-1 or a PD-L1 inhibitor. So, my philosophy is get that into them as quickly as possible.
Benjamin P. Levy, MD: Like a one shot on goal, get all active drugs in.
Roy S. Herbst, MD, PhD: Exactly. You might not have a chance for a second-line treatment. So, I think the KEYNOTE-021, albeit small, gave us that hint. But now with 189, even without the data—but that data will be out there pretty soon, I’m betting—I think that just confirms they hit their endpoints, they saw a survival. So I would use it, and I do use it when I don’t have a trial.
Suresh S. Ramalingam, MD: And I think this is an area where consensus is still emerging. It’s clear that from these trials, we look at non—small cell lung cancer patients without a driver mutation in 3 different contexts: the high PD-L1 group, the low PD-L1, and the absent PD-L1 group. And each of these trials that are going to be done, we have to look at each specific subset and see how they did. And I think that’s where we’re going to learn the definitive answers. I agree with Roy that if you have a low–PD-L1 patient and the patient was having very aggressive disease, and we talked that the patient is only going to have one shot at treatment, then you want to give them the triple combination. In my clinic, if they don’t fit in that category, I still start with a platinum doublet and then go to second-line checkpoint inhibition.
Benjamin P. Levy, MD: And that has been my practice. And I don’t know if we have a right answer here, but I think if they’re less than 50%, if they’re symptomatic, or they have a heavy smoking history—not that that’s the albeit best clinical enrichment factor—I would probably start with a platinum doublet and then sequence them to immunotherapy. You could make the argument, however, that they may never see an immunotherapy agent if you do it that way. So, I think there’s a lot that we can learn over the next 6 to 12 months. I’ll be very interested to see the subset analysis from 189 for the less than 50%. I don’t know if that will be powered for overall survival or PFS, but I’ll be very interested to see what that shows.
Hossein Borghaei, DO, MS: But that’s going to be the discussion. If it’s not powered, are you still going to avoid using the triple combination in low expressers?
Suresh S. Ramalingam, MD: Even if you don’t have a strong P value, if you see the curves separate and stay separated, you will use them.
Benjamin P. Levy, MD: Tom, in your practice, what is your approach to using this in all patients, symptomatic patients?
Thomas E. Stinchcombe, MD: In the PD-L1—highs, I generally give single-agent pembrolizumab, although generally below 50%, I do carboplatin/pembrolizumab for the borderline performance status, symptomatic, the people who I think I have 1 opportunity. A place I use carboplatin/pembrolizumab alone is generally the people with no PD-L1 expression, never-smokers, and I think there’s a debate about how much incremental value there is in adding chemotherapy. Granted, that’s more my own personal bias than evidence-based medicine like we’ve talked about.
Benjamin P. Levy, MD: Yes, we all have our own evidence-based consensus.
Thomas E. Stinchcombe, MD: We just don’t have evidence.
Hossein Borghaei, DO, MS: We call it bias. We don’t call it evidence anymore. It’s personal.
Benjamin P. Levy, MD: It’s a lot of confusion there. I think that we’ll find out a lot more very soon, and I don’t know if we’ll still have a right answer or not. I think treatment decisions will still have to be individualized.
Hossein Borghaei, DO, MS: Right, but every study, no matter how well designed it is, leads to some answers and leaves you some questions you still have to answer for the subsequent studies. And I think that’s where we’re heading, and we have to figure out what to do for the other populations.
Transcript Edited for Clarity