Checkpoint Inhibitors in NSCLC

Transcript:Mark A. Socinski, MD: Now, let’s discuss the availability of the newest class of drugs. We have the checkpoint inhibitors. In the past year or so, we’ve had two approvals for this. Tom, walk us through the FDA approvals, both for nivolumab as well as pembrolizumab.

Thomas E. Stinchcombe, MD: The first, nivolumab, was approved for the squamous histology patients based on phase III trials that compared it to docetaxel and showed a tremendous survival benefit. The hazard ratio was 0.59, about a 3-month improvement in median overall survival. Then the second indication was in a non-squamous patient, another phase III trial that showed a significant survival benefit at this point. Neither of those studies tested for PD-L1 testing in order to enroll on the trial. There was some retrospective testing done, and so nivolumab is now indicated without any regard to PD-L1 testing. Pembrolizumab is also approved. It was initially approved based on a phase I trial—in a large expansion cohort phase I trial—and it was approved for patients who had PD-L1 expression greater than 50%.

Mark A. Socinski, MD: I get confused. I always thought the PS was performance status, but it’s proportion score in this case.

Thomas E. Stinchcombe, MD: And that showed a very dramatic response rate of about 45%, and that’s what necessitates the PD-L1 testing with that agent. I think it’s important to put it in context of the recent KEYNOTE-10 data—that was presented in Singapore—that showed survival benefit to pembrolizumab compared to docetaxel in all patients with a PD-L1 proportion score of greater than 1%. So, the package label may be changing based on the new data. Mark A. Socinski, MD: Right. So, Ben, in practice, do you check PD-L1 status, or how do you use it in your practice?

Benjamin P. Levy, MD: Sparingly. I think that, as Tom mentioned, at least in the squamous population—a genetically complex disease—PD-L1 did not drive any of the effect that we saw with nivolumab compared to docetaxel. In the non-squamous study, it seemed to do this. Patients that seemed to have a better outcome if they were PD-L1 expressers if they got nivolumab versus docetaxel, and that wasn’t seen in the patients who did not have high expression of PD-L1. It really begs the question of what to do, not necessarily in the squamous cell population, but in the non-squamous, in which we saw this effect. I must tell you that I’m not confident that PD-L1 is a reliable predictive biomarker for these class of drugs. We’ve had very different data suggesting that these drugs may work in the absence of PD-L1 from other studies. We know that, at least in the CheckMate study, that if you were PD-L1—negative, you did at least as well with nivolumab versus docetaxel, a better tolerated drug. So, based on all of that, I’m not routinely testing for PD-L1. I have issues sometimes at my institution with tissue procurement and having sufficient tissue for all the testing that needs to be done, not only for genetic interrogation that’s done routinely but also for clinical trials. So, no, we don’t generally test for PD-L1.

Mark A. Socinski, MD: Do you think the cheapest biomarker we have, which would be smoking history, would do as well?

Benjamin P. Levy, MD: Perhaps. I think it’s rare for me to see a smoker not garner a benefit from these drugs.

Mark A. Socinski, MD: We would have to take a history.

Benjamin P. Levy, MD: That’s right, which may take more time than PD-1 testing for me. But, overall, I think there may be a future role for this biomarker. I think the data is just so challenging and difficult to distill out in terms of whether this is truly a reliable biomarker. And this is based on tumor heterogeneity. This is based on different competing platforms having different cutoff values. I think it’s a challenge, and we’re not using it routinely yet.

Mark A. Socinski, MD: And Tom, any difference between nivolumab, pembrolizumab? Is it Coke, Pepsi? How do we use them differently?

Thomas E. Stinchcombe, MD: I think the toxicity profiles appear to be very similar, and I would argue the efficacy profiles are pretty similar. The only real difference is the every-2-week schedule with nivolumab and every 3 weeks with pembrolizumab, and the tissue testing—PD-L1 testing—associated with pembrolizumab at this time.

Mark A. Socinski, MD: Greg, with nivolumab being approved without a biomarker, I always think back to the initial BR21 data with erlotinib versus placebo. There’s no one that might not get a benefit from erlotinib because everyone got it. This was 10 years ago. Are we in the same spot with nivolumab at this point?

Gregory J. Riely, MD, PhD: We may very well be, but I think what we currently have with biomarkers is not what we need. So, with today’s technology, we’re where we were with EGFR IHC and EGFR FISH. It’s a test that ended up being not useful at all. And I think there’s no doubt in my mind that 5 or 10 years down the road, we’re going to have a better biomarker for these agents. I think it will be exciting when we get there. On the way there, I think the current biomarker just doesn’t identify people who won’t benefit, so as a consequence I don’t find it useful.

Transcript Edited for Clarity