Mark A. Socinski, MD: We have a little bit of conflicting data with IMpower131, which is squamous, which was kind of positive but not fully positive.

Paul K. Paik, MD: Right. Yes, it was not fully positive. It was a very interesting ASCO [American Society of Clinical Oncology meeting] last year, where they presented on this in 2 different sessions. I think in KEYNOTE-407, in particular, the secondary look at the OS [overall survival] came out later. That was in the Clinical Science Symposium, so it made it very difficult to catch both.

But KEYNOTE-407 really is similar to KEYNOTE-189. It is practice changing. This was a randomized phase III trial of a backbone of carboplatin and taxane. It was investigator’s choice, as to whether patients received paclitaxel or nab-paclitaxel. And the randomization was to pembrolizumab with this followed by maintenance pembrolizumab or placebo. It was positive by PFS [progression-free survival] across all PD-L1 [programmed death-ligand 1] cut points. It was very similar to what we saw with KEYNOTE-189. The OS benefit really has been quite powerful. And this is where, in part, IMpower131 has failed. There is no OS advantage. There is PFS, but no OS.

And then, at least in the initial look at the data, there was this sort of squirrely middle section for PD-L1—positive patients, intermediate, where they seemed to do worse when you added atezolizumab. It was very difficult to make heads or tails. It also makes it difficult to interpret.

Mark A. Socinski, MD: Yes. So one of the things that bothered me about those 2 trials is the performance of the control arm. The control arm on KEYNOTE-407, where 60% of the patients got carboplatin/taxane and 40% got carboplatin/nab-paclitaxel, is 11 months. And then on IMpower131, where the control arm was all nab-paclitaxel, the median survival was 14 months. Obviously, it’s harder to show a difference if your control arm does better. So I haven’t really understood why. There is some suggestive data that nab-paclitaxel may have more activity in squamous disease, so maybe that’s the explanation, although I don’t necessarily believe that. Do you have the same thoughts, concerns?

Paul K. Paik, MD: Yes, I have the same thoughts. It’s very difficult to know for sure. These are cross-trial comparisons that are there. So each study has to be interpreted really within the context of what happened in that study. And that’s sort of where we ended. But again, I think your point is a good one. The performance of the control arm does matter. There’s that signal, potentially, of increased efficacy with nab-paclitaxel. Is that part of it? I don’t necessarily think so, because of other things that we’ve seen with the performance of pembrolizumab-based regimens versus other immunotherapy-based regimens.

Mark A. Socinski, MD: Corey?

Corey J. Langer, MD: If you go to the KEYNOTE-407 trial, and I believe this was in the forest plots, when they actually combined, they looked at the hazard ratios with nab-paclitaxel as opposed to solvent-based, it was actually a bit lower. So that may actually be the preferential regimen to build upon. And again, it may be a superior squamous drug. So the fact that one trial exclusively used nab-paclitaxel, or at least in that cohort that was compared, and this one with a taxane, it may be less of a difference. I want to emphasize one thing that we have not mentioned. All of these trials featured a crossover in the control arm at the time of progression. Patients on the control were eligible, and it was built into the trial to get single-agent pembrolizumab. And despite that, and including this population, we still sought a consistent overall survival advantage.

Heather A. Wakelee, MD: Yes, that’s an important point.

Mark A. Socinski, MD: Remind me how many people actually have crossed over?

Corey J. Langer, MD: The majority. At least in KEYNOTE-189, it’s over 50%. When you actually look in the intent-to-treat population, it’s about 50%. When you look at those who are eligible for crossover in these trials, it’s on the order of 60% to 70%. It may not be pembrolizumab on trial, necessarily, but it’s usually some sort of immunotherapy.

Mark A. Socinski, MD: But Dr Julie Brahmer presented some of that crossover data. You would think if you were greater than 50%, using the oncogenic driver paradigm, does it really matter if you get a TKI [tyrosine kinase inhibitor] in the EGFR setting in the first or second line, as long as you get it at some point? The suggestion from KEYNOTE-024 was that even if you were highly positive and got pembrolizumab in the second line, the survival didn’t catch up, right? So I don’t know what that tells us but….

Corey J. Langer, MD: I think it’s the nature of the population. Some of these people just develop fulminant progression. They can’t get to the crossover. Or perhaps you give your best drug first.

Transcript Edited for Clarity