Nondriver NSCLC: Immunotherapy and PD-L1 Expression

Benjamin P. Levy: I think immunotherapy has altered the treatment paradigm for patients with advanced non—small cell lung cancer. Without a driver, you now have the option to use immunotherapy up front with chemotherapy, which I don’t do yet. This would be carboplatin/pemetrexed with pembrolizumab. In an unselected patient population, you don’t need PD-L1 to make that decision. You can do it. And then, if you’re not going to use it up front, it is one of the options for second-line—again, independent of PD-L1. Atezolizumab, pembrolizumab, and nivolumab can all be used second-line. Atezolizumab and nivolumab can be used independent of a PD-L1 score second-line. Pembrolizumab, if you’re going to use it second-line, has to have a tumor proportion score greater than 1%.

It has completely altered the way that we view lung cancer, the way that we think lung cancer behaves, and the way that the immune system interacts with the cancer. But we have a lot of options, and I don’t know if we know what the best option is yet. Should we use immunotherapy up front with chemotherapy? Should we save it for second-line? Are there other combinations that may work better? That’s really the role for immunotherapy. Immunotherapy is great, and I’m not saying that I’m not excited about immunotherapy. It doesn’t work for everybody, and I think that’s really important. Again, the advertising campaigns would make you think it works for everyone, but it doesn’t. If you look across the board, even in the second-line, the response rate to a single-agent checkpoint inhibitor is 20%, so 80% of patients don’t respond. We’ve got to do better, and we will, but I want to temper the enthusiasm. I’m excited, and I want to move this field forward, but we have to be mindful about what the actual data show.

Immunotherapy is probably not the best drug for patients with driver mutations, specifically EGFR and ALK. If you look at the subset analysis of patients from these trials, in the subset analysis of patients who were EGFR mutated, in every trial they did better with docetaxel than they did with immunotherapy. So, these specific genotypes are generally not ones that we are really enthusiastic about using immunotherapy with. In a retrospective analysis of a very small data set, the response rate of patients with EGFR mutations treated with a checkpoint inhibitor was around 5% to 10%. Response rates to ALK rearranged lung cancers to immunotherapy were 0%—again, very small numbers. But I think we’re learning that these specific genotypes are not the ones that are the best candidates for immunotherapy.

I think that the scenarios where I see immunotherapy as probably not my first option are, first, in a patient who’s either EGFR or ALK rearranged, who has received targeted therapy and then gotten chemotherapy, or who has even received targeted therapy in whom you’re entertaining immunotherapy second-line rather than chemotherapy. I’m not sure we know how to sequence immunotherapy for these patients who are EGFR or ALK mutated. Again, the data would suggest that docetaxel is better than immunotherapy. So, that’s one patient group. The second patient is one we’ve talked about, who is the rapidly progressing first-line patient. So, a patient who’s getting 4 cycles of chemotherapy, they are rapidly progressing and their PD-L1 score is low. These are patients in whom you need to elicit a response. If you look even at the subset analysis from CheckMate-057, which had around 108 patients who had low PD-L1 scores—less than 1%—the response rate was only 9% to immunotherapy, half of what it is for the intention-to-treat analysis. So, that would be the second scenario where I would—not saying I wouldn’t use immunotherapy—think very critically about what I’m going to do second-line.

And then, obviously, the third scenario is a patient who’s just not going to be able to tolerate it: a patient who’s on steroids in whom you’re blunting the effect of the immunotherapy that you’re giving or a patient who has an autoimmune disease. These are 3 broad scenarios, and I’m sure there are others, where I would critically think about how I’m going to do things.

PD-L1 expression as a predictive marker for these 3 agents—nivolumab, pembrolizumab, atezolizumab—really depends on what line you’re thinking about. I’ll go with first-line first. We know that PD-L1 expression greater than 50% is the predictive biomarker for pembrolizumab up front. That’s clear now, and every patient who has a TPS PD-L1 expression score greater than 50% should get pembrolizumab. If we look at pembrolizumab in combination with chemotherapy up front, it didn’t matter what the PD-L1 was. In fact, in the patients with low PD-L1, there was a response rate that was quite high by adding pembrolizumab. So, the jury is still out on that. In the second-line, I’ll make a very general broad statement: If you look across the board in the second-line, for the most part, the higher the PD-L1 expression for these patients, the more likely they are to respond to pembrolizumab, atezolizumab, or nivolumab.