Chemoimmunotherapy in Advanced NSCLC Treatment

John V. Heymach, MD, PhD: We’ve spoken about immunotherapy or combination immunotherapy but without chemotherapy. We did get additional data about chemotherapy-immunotherapy combinations with a few studies as well.

Sandip, tell us about the KEYNOTE-407 study update in squamous patients.

Sandip P. Patel, MD: The KEYNOTE-407 trial was a study in the frontline space, metastatic squamous histology for KEYNOTE-407 that looked at carboplatin-taxane, which could be paclitaxel or nab-paclitaxel in this study for 2 years. We saw similar results and durability of responses. One of the paradoxes with immunotherapy is that squamous is almost universally associated with smoking. Also, we associate immunologic responses with smoking-related neoantigens. But if you look at any study—not just KEYNOTE-407, but any of the squamous studies, any of the subgroups—the squamous arm tends not to do as well as nonsquamous. Initially, I thought, “Maybe it’s because they have comorbidities.” But even if you look at PFS [progression-free survival], it’s not quite the same.

That said, overall survival [OS] benefits broadly in all comers, regardless of PD-L1 status. That’s 1 of the key aspects for the chemotherapy–I/O [immuno-oncology] studies: we’re seeing benefit. With KEYNOTE-407 though, we didn’t see the same level of benefit in PD-L1–negative patients as we saw in KEYNOTE-189, which had a phenomenal hazard ratio, I think 0.56, 0.58 for PD-L1 negative. We didn’t see that with KEYNOTE-407. This is 1 scenario in which you can absolutely use chemotherapy-pembrolizumab. That’s 1 of the most commonly used regimens in frontline metastatic squamous non–small cell lung cancer, but there are others as well.

John V. Heymach, MD, PhD: You’re right. It’s surprising. I wonder if it has to do with tumor bulk, why squamous patients aren’t getting as much benefit there.

Stephen, we brought up cemiplimab before as a single agent in the high PD-L1s, but they’ve also conducted a study with chemotherapy. Do you want to tell us about that?

Stephen Liu, MD: This was the EMPOWER-Lung 3 study, very similar to the IMpower study, the atezolizumab studies. EMPOWER-Lung 3 was a randomized phase 3 trial where patients were randomized 2:1 to cemiplimab plus chemotherapy vs chemotherapy alone. This merged both histologies. What they showed, no surprise, was that there was a survival benefit. Overall the hazard ratio was 0.71, and the 1-year OS rate improved from 56% to 66%. It was a better PFS, better response rate, better duration of response.

When we look at some of the subsets there, we see that there was quite a bit of benefit for squamous with a hazard ratio of 0.56. On the other hand, when we look at the PD-L1 negative, maybe the area where we need this the most, the benefit wasn’t as large. In fact, the hazard ratio was right at 1. The challenge with a study like this is when you merge the histologic subtypes, you can do your study quickly but you’re a little less well powered to look at each subtype. That’s the problem with the stats here. Overall, this reinforces the paradigm. When you add an active PD-1 or PD-L1 inhibitor to platinum doublet chemotherapy, you improve outcomes across the board. Thus, there are no real surprises here and maybe another option in the future.

John V. Heymach, MD, PhD: That’s another possible option. You mentioned the squamous subset had a little better hazard ratio there than we saw in the KEYNOTE-407. What are your thoughts? Is that a criterion that may push you, or do you think that small subgroups are not powered appropriately?

Stephen Liu, MD: Ideally, when we look at a study like KEYNOTE-407, it’s just in squamous. We break it down by a PD-L1 subtype. Now it’s a 2x2 matrix histology and PD-L1 expression in some of the other cohorts. My subsets get smaller and smaller. Just as I won’t read too much into the squamous, in a positive sense, I won’t read too much into the PD-L1 negative in a negative sense in overall. When we add these together, my expectation is that we improve outcomes. I don’t expect this to be any different—no worse, maybe no better. As you mentioned, I’m hoping we can build on this backbone to improve outcomes.

John V. Heymach, MD, PhD: Yes, that’s right.

Transcript Edited for Clarity