Transcript:Benjamin Levy, MD: Let’s stick on this theme of biomarker here, Ed. And quite a contentious issue, [just] maybe not in squamous. There’s a lot of work that needs to be done. I agree with molecular enrichment strategies for squamous cell cancers and their ability to predict efficacy of these drugs—the story is a little murky in adenocarcinoma—but let’s just talk about PD-L1 as a biomarker and maybe dovetail that into the KEYNOTE data because I think this is where people, community physicians and even academic physicians, started to understand. Maybe there is an enrichment strategy with PD-L1. We saw higher response rates, and there was a trend based on the cutoff of PD-L1 in this study. Maybe can you distill that down and make it a little easy for me to understand the PD-L1 biomarker?

Edward S. Kim, MD, FACP: You understand it just fine. Paul could give this lecture in his sleep. The excitement and the confusion has come from two simultaneous drugs approved in similar settings, but one with a biomarker and one without a biomarker. Before, if you took away the biomarker, you would look at dosing, you would look at schedules, three-week versus two-week and efficacy, which was similar between the two. You throw in the biomarker contingent with one of the drugs, pembrolizumab, and that was based on the study that Eddie Garon reported. That changes things, and on a superficial basis, the biomarker data is impressive. The cutoff point they used was greater than 50% and had a 41% response rate.

Sound familiar to what we have in first-line with nab-paclitaxel? The same thing is happening here. You’re going to ignore that because it’s not about testing the biomarker. Every center across the country, when this drug was approved, said “Let’s just reflex test on our biopsy frontline.” Problem is, you can’t use biopsies that are greater than six months old if this is a second-line drug. Which test would you use and how should you do it? That forced you in the second-line setting to make a decision. Will you just write the order for nivolumab and not worry about the biomarker? If your patient rapidly progressed on first-line treatment with squamous therapy, you could use that archived biopsy if there was enough left to send for staining, and if there was a greater than 50% [chance], then it would direct you to use pembrolizumab. You could get turnaround times that are about 7 to 10 days on those types of tests or you just default to nivolumab because you don’t want to have to re-biopsy, risk that patient—they’re progressing.

The part that’s been a bit of a shame—because I like that biomarker-directed data, and both drugs are wonderful drugs—is that it’s clear that we cannot find a biomarker-enriched subset as great as we can with pembrolizumab and PD-L1 testing. This will all change, this is caveated. In about a year, if the first-line data comes out, there will be reflex testing and people will have to do that. Which kit, I don’t even want to go in to, and what score. But right now, we have two great options, especially in patients with squamous cell cancer. We have discussed this extensively within our lung section group, and we have put both of them as preferred regimens and one that just requires PD-L1 testing. What not everybody knows is you can’t use that archived tissue if it’s greater than 6 months old, preferably the newer the better. You don’t want it sitting on the shelf for five-and-a-half months.

Benjamin Levy, MD: You mentioned some challenges with the PD-L1 biomarker. Tumor specimens older than 6 months probably are not optimal for tumor heterogeneity. the different platforms, the different cutoffs. There is a lot of confusion there about whether this should be routinely incorporated or whether we should just de facto give these patients checkpoint inhibitors. In the squamous cell population, at least in my center, we’re not testing for PD-L1, given the data from CheckMate. Nevertheless, compelling data out of the KEYNOTE study showing that different cutoff, depending on the PD-L1 cutoff, led to higher response rates with this drug. Paul, thoughts on the PD-L1 biomarker? Should it be used in the squamous cell population? Should this be something that’s just reserved for the non-squamous tumors? Where do we sit with this in your mind?

Paul K. Paik, MD: For now, it’s pretty clear we should not use it in squamous lung cancer, in part because—and this is the main thing—it’s not predictive for nivolumab and because the biomarker PD-L1 testing results have not been stratified by histology, biomarker positive/negative for adenocarcinoma, for squamous in the KEYNOTE study. So we don’t know. But if you put those two together, then you shouldn’t test. Then it becomes a pretty simple answer: if you’re not going to test, then you give nivolumab because you’ve got to test if you’re going to give pembrolizumab.

Edward S. Kim, MD, FACP: All these Merck people are crying now.

Paul K. Paik, MD: I know. Practically speaking, adenocarcinoma is a lot more difficult, as both of you had mentioned, because pembrolizumab is approved with a biomarker. But there’s no guidance to what degree or percent positivity for the tumor that you’re going to use. We know, taking a look at least by response rate, which was gradated by more than, just less than, or greater than 50%, that there’s variation of response. It’s very high, just south of 50%—if you’re 75% to100%. If you’re less than 1%, it’s 8%, and then it sort of goes up by grades from there. This is important because if you’re going to give pembrolizumab, and let’s say your score comes back as like 8% tumor positivity, what do you do with that?

You could do that exercise of going to the appendix—because it’s in the appendix of the study to see what the response rate is—and then try to figure out if that makes sense relative to docetaxel/ramucirumab. I’ve done that before but it’s a lot of extra work, and I don’t think community oncologists have time to do that for every single patient they’re going to see. We have a situation where it’s biomarker-approved. We don’t have guidance, but we do know it makes a difference in terms of what percent tumor positivity is present. And so you’re left with sort of a mess. In addition to that, I think as Ed was talking about, we want a biomarker that’s really good. We want something that’s discriminating. If it’s present, you’ll respond. If it’s absent, you won’t respond.

Benjamin Levy, MD: Right, something binary.

Paul K. Paik, MD: Right, and PD-L1 testing does not do that. If it’s absent, you still may respond to the drug and clinical characteristics may be better predictors. If you’re a never-smoker, you probably shouldn’t receive it. You should probably get docetaxel/ramucirumab instead. So there are these subtleties that are involved that make it difficult to figure out what we’re going to do in addition to all the things that you guys mentioned.

Edward S. Kim, MD, FACP: This reminds me of the NCI Director's Challenge Consortium that nobody wants to talk about anymore; the adenocarcinoma was going to solve the rule and get us away from a histology-based program and all the molecular trees. And we learned that it’s equally important. The clinical characteristics are equally as important as the molecular characteristics.Benjamin Levy, MD: Especially with this drug. So a lot of outstanding questions with biomarker, clinical versus molecular enrichment, with the institution of these drugs.

Transcript Edited for Clarity