Naiyer Rizvi, MD: We now have pembrolizumab approved as a monotherapy for patients with greater than 50% PD-L1 [programmed death-ligand 1] expression. Should we be using monotherapy for everyone more than 50%, or are there situations where you would use chemotherapy and immunotherapy [I/O] in combination? Leora, do you want to comment on that?

Leora Horn, MD, MSc: A lot of people discuss that maybe we should use the chemotherapy-I/O combo, particularly in those patients over 50%. We saw from KEYNOTE-189 that the response rate was higher than what we see with pembrolizumab monotherapy, where the response rate is around 50%, and with the combo response rates are over 70%. So maybe if you have this patient who is rapidly progressing, who you want to get a quick response with. To be honest, the couple of times that I've done that, the patient rapidly deteriorated and ended up in the hospital with a complication from their therapy. So I am not a big fan of the pembrolizumab/chemotherapy combo when they're over 50% and symptomatic. Maybe it's a lung mass, and you can rather radiate and still use pembrolizumab monotherapy, but there are some folks who think that in those patients over 50% who have a high symptom burden, that maybe with monotherapy we'll get quicker symptom relief.

Naiyer Rizvi, MD: Tim, do you want to comment on whether some of the upfront radiation can be useful to manage acute issues?

Tim Kruser, MD: Certainly there's a growing body of evidence that using radiation in conjunction with early immunotherapy seems to augment the systemic response. So we tend to have a low threshold for finding a reason to palliate something. If there's a borderline lesion in the hilum that might be threatening a central airway, we will often radiate that in a prophylactic palliation framework in someone who is going to get single-agent monotherapy, in an effort to keep them locally controlled, but also potentially augment their systemic control.

Naiyer Rizvi, MD: I'm not sure I completely agree with Leora. I think for patients who have a big disease burden and a lot of osseous disease, very symptomatic, I get a little nervous about just pembrolizumab monotherapy in the more than 50% group. Josh, what are your thoughts, and maybe Jacob can comment as well.

Joshua Bauml, MD: I think the fact is that PD-L1 is not a great biomarker. So I don't have a lot of confidence. When I see PD-L1 greater than 50%, it's not like EGFR. When I see EGFR, I'm going to give a drug and I know that it's going to work, or I'm pretty darn confident. I would tend to agree with you, Naiyer, that if I have a patient with greater than 50% who presents in extremis, I feel like I only have 1 shot on goal.

Now, Leora, I think you're right as well. A lot of times they end up in the hospital because they're so sick. But I feel like if I have 1 shot on goal, I want to go with all guns blazing here and use chemotherapy-I/O. But that is a small percentage of my patients. Someone really has to be quite ill for me to say that that's the approach that I use. For the vast majority of my patients with greater than 50%, I give pembrolizumab monotherapy. If you look at the overall survival data, not comparing just ORR [overall response rate], I look at the overall survival data. The medians are pretty similar for the greater than 50% patients on KEYNOTE-189, KEYNOTE-407, and KEYNOTE-024. So I feel pretty comfortable with it. It’s that early phase that I get a little worried about.

Naiyer Rizvi, MD: Jacob?

Jacob Sands, MD: Yes, I agree. If I have a patient who is really symptomatic; recently someone who had a big pleural effusion and bulky disease and was very symptomatic from it, young, healthy, I ended up doing a combination of chemotherapy plus pembrolizumab to increase the likelihood of response. And fortunately, she has done well. You can't draw conclusions from 1 individual, and obviously, we do better by putting all of the data together and letting that drive the decision making. But otherwise, aside from those with bulky disease, trying to really get as much of a response as early as possible, I tend to use pembrolizumab monotherapy. But I'll also add, we talk about greater than 50% PD-L1 expression because that's really where the bulk of the data are, but there are data sets recently published looking at groups of 10% ranges—so 50% to 60%, 60% to 70%, 70% to 80%, and up. And that greater than 50% isn't all the same either.

When you have someone with greater than 90% expression, it looks like there is an even higher likelihood of response. So someone with a 55% expression versus someone with a 95% are likely to have significant differences and median outcomes as well. Now, how much that particularly drives decision making, well, that comes down to when you have that person where you're trying to get as much of a response in outcomes from, where if they've got a 95% PD-L1 expression versus a 55%, those might not actually be the same population either.

Essentially, what I'm saying is for the majority of patients with greater than 50% expression, I will treat with pembrolizumab alone. In someone who I really want to get as much of a response as I can, I'll do the combination, even if it's greater than 50%. But then also raising the topic of even that greater than 50% group isn't all the same either. And as you see increasing PD-L1 expression, we may actually have differences within those patients as well.

Joshua Bauml, MD: Jacob, it's interesting because there was a great paper that came out from your group. It was really interesting. But I've been finding that our pathologists really score it in a rather bimodal fashion. They know that 50% is the cutoff, so they'll say 50%. And they won't really break it down to 65%. I don't think they're doing as much cell counting. I think that as we go into AI [artificial intelligence]-based assessment of this, we may get more precision and more of a bell-shaped curve. But for now it ends up being a lot of peaks at 0% and a lot of peaks at 50%.

Jacob Sands, MD: You're absolutely right, and it is interesting, but that's where those 10% ranges end up being helpful in that analysis. Because I agree if you're looking at just 55% or 50% versus those who are 95% and those being different, that could also just be a user taking anyone who is in the 50%-ish range and calling them 50%, 55%. So that's a very good point, but also in those 10% ranges, it does seem like there is some legitimacy to that as well. And at the same time, also recognizing there's a lot for us to learn, and even biopsies from different areas will give you different percentages of PD-L1 expression. So that's not something that ends up being all of the tumor that they have has that percentage. It can range from sites you biopsy. Back to your earlier point that PD-L1 is certainly not a perfect test. There's still a lot for us to learn from this. We kind of categorize it as best we can.

Generally for greater than 50% expression I do single-agent pembrolizumab, but in those patients where I really want to get a response as quickly as possible and I think they can tolerate it, I'll do the combination.

Transcript Edited for Clarity