Jack West, MD: What do you guys think about your use of a platinum/pemetrexed/pembrolizumab combination, before—based on the phase II data and approval—and then the subsequent phase III data from KEYNOTE-189 and also its applicability across the PD-L1 spectrum?

Charu Aggarwal, MD, MPH: So, this was a very well conducted trial, KEYNOTE-189, and as Greg pointed out, we are able to now really discern—because of the largeness of these 3 groups: PD-L1 expression being the negative, 1% to 49%, and greater than 50%—that no matter which group you were to use this in, the benefit still applied for the triplet. And personally, in using the triplet, I find that it’s actually very well tolerated without adding much cumulative toxicity above and beyond what I would have obtained or had as a side effect with just chemotherapy alone. So, I find that it’s really practice changing and reinforces the data from the phase II trial.

Jack West, MD: Jared, your thoughts?

Jared Weiss, MD: So, KEYNOTE-021g didn’t change practice for me at all. In the greater-than-50% population or the 50%-and-greater population, I already thought that single-agent immunotherapy fit the human goals of my care a little bit better. And the question of interest to me would have been immunotherapy versus immunotherapy plus chemotherapy, not chemotherapy plus or minus immunotherapy. In this trial, there was greater crossover than in the KEYNOTE-042 study. I think it was about 40% from the chemotherapy alone arm, maybe 41%.

Jack West, MD: A little higher.

Jared Weiss, MD: But still imperfect. Some of that will be because the patients haven’t progressed yet. But there was imperfect crossover here. And so, I think this is practice changing. It makes the triplet a reasonable option, perhaps a preferred option for the 0%-to-49% patients. The question that remains for me is that in the real world, I think that crossover is much better than that. And my question, is it mandatory to give it in frontline? The PFS curves, to me, speak to an additive benefit, not a synergistic benefit. Then you get a better overall survival difference, particularly in the high expressers, than you have on the PFS curves.

And so, I hearken back to some of our old data and ask, “Well, is this guaranteed exposure to the agent?” But then I look at my practice and say, “Well, I think crossover, quite a bit more than 40-something percent of people.” And so, I definitely think that this trial is practice changing in the sense of making that triplet eminently reasonable. My question that I have not fully resolved is for the 0% to 49%: Is it mandatory or sequential therapy guaranteeing exposure to immunotherapy in second-line still reasonable? And my personal answer to not totally cop out is, I think it’s still at least reasonable.

Jack West, MD: I agree. I think that there were still differences, where the vast results with the combination was in the high—PD-L1 group in whom monotherapy remains a very strong choice and you can avoid the issues and the side effects of chemotherapy, at least initially, and have that available for later. The 1% to 49% were still, I would say, compelling benefits, and I would even say that the 0%, the less than 1% had a benefit from the combination. But I agree with you, Jared, that if you have 50% or fewer of the patients crossing over, it does lead to the question of, could you still do comparably well? The progression-free survival, hazard ratio, in the less than 1% was 0.75; the curves are kind of crossing over and touching in a few points. So, I wouldn’t say we want to be too heavy handed that it is an absolute mandate to do this, particularly if you’re really confident that for the patient in front of you with more indolent disease or very good performance status, you’ll be able to deliver it sequentially, if not concurrently.

And I agree that in practice, I think it’s important to distinguish. People say, “Well, we only deliver a 50% rate of second-line therapy with docetaxel, looking back at our clinic records,” but that’s docetaxel. And I would say that there’s not that much enthusiasm for it. The toxicity burdens are real. Very few of our patients are prohibited from getting immunotherapy for these reasons. If they were not candidates for immunotherapy second line, they were probably not candidates first line either.

Jared Weiss, MD: Let me just also… You first.

Charu Aggarwal, MD, MPH: I would just add that our ability to predict as oncologists which patient will go on to second-line therapy is still limited. I think we still need more than just an intuition or gut to say, “OK, I can save immunotherapy for this patient second line.” I think there is value to giving the best and—whatever that best may be, as defined by data—up front because we can’t predict clinically who is or not going to make it to second line.

Jack West, MD: No, I think that’s fair. But at the same time, there are costs to doing it all up front and burning that option, particularly if there are some data out there, imperfect though, that by giving subsequent therapy, you can get better results with chemotherapy after immunotherapy, etc. I think that we still have a lot to learn about sequential, and I think it’s fair to say there is a difference between a 20% or a 40% crossover. And in my clinic, it exceeds 80% or 90%. I would say that I think it’s fair to argue the point of give your best up front but there is still more to learn about this, particularly if we could deliver everything sequentially versus concurrent.

Jared Weiss, MD: Yes. And just briefly, I think it’s delightful how quick our data are moving, all these advances for patients. But one thing that’s lost is the opportunity to ever learn important questions. And I think that there’s a sequence question here that our data have moved so fast that I fear we may never answer. Is it best to use these agents all up front? Is it best to do a consolidation or maintenance serial approach? I fear we may never really learn that.

Jack West, MD: One small point to make is that unlike the KEYNOTE-021g trial, KEYNOTE-189 included cisplatin or carboplatin at the discretion of the investigator. The benefits were comparable either way. There was a higher incidence of nephritis in KEYNOTE-189 that would lead to some, at least, vigilance if, not caution about that. And my own preference would be to favor carboplatin. I would say in the United States, that’s the clear prevailing approach, but either would be a reasonable one.

Transcript Edited for Clarity