Transcript:Benjamin P. Levy, MD: Let’s move, in the spirit of disagreement, to PD-L1 testing and where that fits in in terms of a predictive biomarker. There’s been some conflicting data about the PD-L1’s reliability. Marina, can you talk a little bit about PD-L1? We have, in the United States at least, a complementary and a companion indication with two drugs.

Mark G. Kris, MD: Would you translate that for me?

Benjamin P. Levy, MD: Yes, companion means you gotta have it. Complementary, it’s flexible.

Mark G. Kris, MD: What does that mean?

Benjamin P. Levy, MD: I don’t know. I think ‘complementary’ is the first time I’ve ever heard this being used. It’s tough for me to understand what this means. It’s interesting language that I think we’re all trying to get used to. Can you give us some ideas about where we stand with PD-L1 testing for the use of these drugs? Is it helpful in the second-line setting?

Marina Garassino, MD: So, I think that it is really a contentious story because nivolumab started without doing the study. It’s a preplanned subgroup analysis and they created the cutoff of 1%, 5%, and 10%. But, at present, we don’t know what will happen with 50%, for example. The study with pembrolizumab was the opposite because they used another antibody. At the beginning, they published in the New England Journal of Medicine that the cutoff was 50% of the proportional score, and now we know that that is 1% again.

The POPLAR study with atezolizumab, and also the BIRCH study, they said that it was important not to see only the presence of PD-L1 expression on tumor cells, but also on immune cells. If you remember, 1 year ago at ASCO they presented the data of the POPLAR study, and it seemed that the good patients to give the drug to were the patients with strong positivity of PD-L1, both in immune cells and in tumor cells. But, if you look at the data, the CR of POPLAR, they started with more events and with more follow-up. It seems that the PD-L1 is important. Because when you have PD-L1, you have more chances to benefit from the drug, but you can benefit also if you have low expression of PD-L1. This is the same for durvalumab. So, we have at least the four drugs, four antibodies, but a different cutoff. There is a tentative harmonization of the antibodies. We know that at least the three antibodies seem to go in the same direction, but we have the VENTANA PD-L1 assay, which is the antibody of atezolizumab, that seemed to go in another direction. We don’t know what the PD-L1-positive is at present.

Benjamin P. Levy, MD: You mentioned the data from the Blueprint Project.

Marina Garassino, MD: Yes, the Blueprint Project, which is small, but there are some data that we can draw from this small number of patients. So, I think that, at present, it’s quite difficult to give a drug according to PD-L1 expression.

Benjamin P. Levy, MD: So, you’re not routinely testing your patients for PD-L1.

Marina Garassino, MD: No, we don’t routinely test the PD-L1, because we don’t know which is the antibody, which is the cutoff. The eyes of the pathologists are also different. And then, if they have to see the PD-L1 expression on the tumor cells, on the immune cells, the proportion score, I think we will need to educate the pathologists more. Nowadays in Europe, nivolumab is approved independently from the mutation, from the PD-L1 status, so we give the drug.

Benjamin P. Levy, MD: So, nivolumab is your de facto regimen.

Marina Garassino, MD: Yes.

Benjamin P. Levy, MD: Outside, you don’t routinely test for PD-L1.

Marina Garassino, MD: Exactly.

Benjamin P. Levy, MD: Chandra, are you testing for PD-L1 routinely in your patient population?

Chandra P. Belani, MD: Yes, we are testing, but we are testing in a small group of patients. If we have the biopsy specimen, we’re able to do it, we do test it. So, if the PD-L1 is positive, we give pembrolizumab. If not, we give nivolumab.

Transcript Edited for Clarity