PD-L1 Testing in Advanced Lung Cancer

Transcript:Mark A. Socinski, MD: We’ll talk a little bit about immunotherapy in a moment, but I wanted to engage Dr. Weiss and just introduce the concept, because probably the latest biomarker that we’re asking is the PD-L1 status. And getting back to your point about a lot of aspirations being done: almost all the PD-L1 work, if not all the PD-L1 work, is done on core biopsies, which is a different strategy and not always easy to do in patients. But give us your perspective on PD-L1 testing and the pitfalls.

Jared Weiss, MD: I’ll start with one thing that is not as much of a pitfall as people sometimes make it out to be, and that’s quantity of tissue. I think sometimes we talk about PD-L1 tissue as if it were the newest, most advanced next-generation sequencing with some kind of advanced bioinformatics. It’s a simple IHC, just like the TTF1 or p40 we were talking about before. And so, tissue quantity is not always a problem in that sense. On the other hand, the operating characteristics of this biomarker have been far more limited in real humans than they were in the preclinical studies. And that may have to do with the transition from complete surgical tissue in a sacrificed animal in the lab down to these core biopsies in the clinic. What we’re seeing is a very poor negative predictive value of the PD-L1 assay with variable positive predictive value, depending on which company’s test you’re speaking of. And when you consider the FDA-approved indications for these therapies at the current moment, which is second-line lung cancer, as a clinician, I feel that negative predictive value isn’t so helpful to me.

I’m a little bit skeptical of just how well these really can be reconciled, how much things can be improved. There are some impressive efforts underway to try to understand the interaction between the different companies’ tests. But, at the core of it, PD-L1 obtained by an FNA or a core biopsy may simply be a flawed biomarker.

Mark A. Socinski, MD: John, your perspective on PD-L1 testing from a pathologist’s point of view.

John W. Longshore, PhD: I think PD-L1, we hear a lot about it, but the expression level is dynamic. It’s very heterogeneous, depending on the tumor type you look at. The expression will vary quite a large amount between the diagnostic biopsy that was taken and the biopsy that was taken at progression from the patient. Further confounding it is, we have multiple therapeutics available in the market, and each therapeutic has a different biomarker test that’s IHC-based at this point that goes along with it. So, we may actually be to the point in the testing landscape where we actually have to do the specific IHC marker that is associated with a particular therapeutic, which would be unique in the companion diagnostic landscape for the past few years. I think the good news, since we’re discussing non—small cell lung cancer today, is that the correlation for non–small cell lung cancer samples with PD-L1 antibody seems to be very good. This may be the best case scenario that we have for working with PD-L1 antibodies. They tend to be much more variable with other tumor types.

Mark A. Socinski, MD: But right now we have a companion diagnostic with one agent in a...

John W. Longshore, PhD: Complementary diagnostic.

Mark A. Socinski, MD: Complementary, thank you. It’s complementary, which means it can provide information that is not necessary. Are people using both of these antibodies in practice to figure this out? I find, in my practice, that I don’t necessarily use the biomarker clinically because I know there’s activity if the biomarker is negative. And I can use a drug, and we’ll talk about this later, that’s not linked to a biomarker without having that information. At least we have one of them that is available. I’m going to come to the BluePrint project, which Jared alluded to, in a moment, but how should we be thinking about this with two antibodies out there?

John W. Longshore, PhD: Right. It’s interesting that you have a complementary diagnostic and a companion diagnostic, where one test is required for use of one therapeutic and is not required for the other. I think prescribing patterns may be altered because of this. It may be a competitive advantage for one therapeutic versus the other if you do not have to have a biomarker test to predict efficacy in patients.

Jared Weiss, MD: In the clinic, it becomes more of a practical question. The data on both nivolumab and pembrolizumab are very impressive. I think there’s no clear distinction of one being more effective than the other. The advantage of nivolumab, as you’re referencing, is you don’t have to test. But for the patient that travels a great distance to come to the doctor’s office, which for many of us at this table can be quite a far travel distance, the ability to use an every-three-week therapeutic, as opposed to every two weeks, can matter for some patients. And so from a very practical perspective, that drives some testing.

Benjamin P. Levy, MD: I just want to piggyback on this quickly. A practical consideration, also a logistical consideration in terms of how much tissue we may have left over when we’re doing molecular interrogation in tissue that’s needed for clinical trials, is, do we have enough left over for PD-L1 testing? And we’ve had some challenges with that in our institution, and I think that’s institution-dependent, but certainly these are challenges.

Mark A. Socinski, MD: I’m getting Jared’s point that currently these drugs are approved second-line, so often you get there and all your tissue may be exhausted by that time. Greg, did you have a point?

Gregory J. Riely, MD, PhD: I just wanted to connect the two biomarkers we’ve talked about. We talked a little bit about EGFR, and we talked a little about PD-L1. It would be easy to lump them together, that these are both biomarkers in the same way. But to piggyback on what Jared was saying, the negative predictive value of PD-L1 testing is pretty poor. The response rate for the PD-L1—negative crowd is on the order of 10%. We don’t have a lot of data looking at EGFR mutation–negative patients treated with EGFR TKIs looking at response rate. But the data we do have—the IPASS data—looking at patients with EGFR mutations and without, show that those patients who did not have EGFR mutations had a response rate of 1%. So gefitinib had a response rate of 1% if you didn’t have an EGFR mutation, 70% if you did. That’s a huge spread. That’s a biomarker. That’s what we need in this space.

Mark A. Socinski, MD: One of the other issues I also get a lot of questions from community oncologists on is the struggle with getting these tests done. At our institution, they’re done reflexively. If a patient has adenocarcinoma, the pathologist on their initial signing out of the surgical path will say, “Here’s the diagnosis, adenocarcinoma.” There’s plenty of tissue, it’s submitted for molecular testing, and an addendum will be issued when the molecular testing is done. It’s all done reflexively. I think a lot of community oncologists struggle with that, getting their pathologist to do it. It may be done at a different hospital. You may not know the pathologist, the pathologist doesn’t know you, you can’t figure out if there’s enough tissue to do this or where to send it. Should we be doing it like ER/PR? We don’t have to ask for ER/PR HER2, do we?

John W. Longshore, PhD: I think the ordering patterns, particularly in the community, really vary. And patients tend to fall between the cracks wondering if the oncologist is going to order the biomarker testing or if the pathologist is going to order them. So, any type of reflexive setting you have really helps the patient in the end, and that’s really why we’re all here, because patients do respond quite well to targeted therapies. And we want to make sure that we do not miss testing opportunities to predict if patients will be eligible for these therapies.

Mark A. Socinski, MD: And I think we’re getting better at getting the message out to our tissue procurers. The thoracic surgeons, the pulmonologists, the interventional radiologists understand that in lung cancer, we need more. So, they’re actually cognizant of that fact. I don’t know how pervasive that is at the community level.

Gregory J. Riely, MD, PhD: I think that it’s been years that we’ve wanted to know EGFR mutation status up front. It’s really been in the community at least for 5-6 years that we want this information. And there’s a lot of talk about how much time molecular assays take, and if we had reflex testing, nobody would care about time because it would be happening before that patient even walked into the oncologist’s office.

Mark A. Socinski, MD: Yeah, the clock would start 10 days earlier, at the time of the biopsy, versus starting just now. The other point that I wanted to make is, getting back to the PD-L1 testing. Some of the issues with that test have been, is it okay to use archived material? Do you need a fresh biopsy? Any perspective on that, John?

John W. Longshore, PhD: The concept of biopsy, re-biopsy at progression, really is emerging, particularly now that we have an approved therapeutic for T790M mutation—a primary resistance mutation for EGFR. We’re going to be doing a lot more re-biopsies to determine if patients are eligible for some of these targeted therapies. The PD-L1 expression level is quite dynamic, and you would probably have a more accurate representation of the expression level with a current biopsy rather than going back to a diagnostic biopsy, which could be a year or even two years old at the time you want to do PD-L1 testing. So, I think it certainly would be indicated.

Mark A. Socinski, MD: You would think that. The thing that was a bit comforting from that point of view is the data from KEYNOTE-010, where the benefit of, in this case, pembrolizumab was the same whether they were archived biopsies or new fresh biopsies. That suggests that maybe PD-L1 status at the time of diagnosis may be adequate at that point. So, maybe we don’t need to go back and do it.

Transcript Edited for Clarity