Leora Horn, MD, MSc, FRCPC: You used the word “marker,” so I have to ask this question. In the PACIFIC trial, we saw some subset analyses with different markers. When the data first came out in the EGFR-positive patients, the data sort of crossed the confidence interval in terms of progression-free survival for the hazard ratio. We don’t have the overall survival data.

And then, at the World Lung Conference, we saw that there were differences in benefit based on PD-L1 [programmed cell death ligand 1]—positive versus PD-L1 expression of less than 1%, to the point where Europe has actually only approved durvalumab following concurrent chemoradiation in patients who have PD-L1 expression that is 1% or higher. What do you make of these subset analyses? They are retrospective, from large trials. Is that impacting who you’re giving durvalumab to right now?

Edward B. Garon, MD: At the World Lung Conference, before the data was actually presented, this was the talk that everyone would have over drinks. Europe had restricted use based on the 1% cut point for PD-L1, which was particularly unusual because the 1% cut point is not a defined cut point at all in the PACIFIC trial. The PACIFIC trial was designed to look at greater than 25% or less than 25% expression. The United States FDA [Food and Drug Administration] has been, of course, very reluctant to take an unplanned subset analysis of a trial and do that sort of approach.

I do think this is a meaningful approach. One of the things that one always has to remember is that this is one of the few times you were able to basically run a PD-1 [programmed cell death 1] or PD-L1 inhibitor versus placebo. That was what this study was. In this group of patients who had no PD-L1 expression, with all of the caveats of an unplanned subset analysis, the placebo wasn’t beaten. And, in fact, the survival, at least numerically, looked a little better in the placebo group.

Would this entirely affect my decision? Maybe not entirely, but it does give me pause that I already had on patients for whom I considered not particularly likely to respond to a PD-1 or a PD-L1 inhibitor in the first place—patients with EGFR mutations who, maybe, weren’t smokers and have low PD-L1 expression. Although there may be some patients who have no expression for whom I would still treat, it does make me reluctant.

Leora Horn, MD, MSc, FRCPC: At our institution, for patients with early stage disease, it’s kind of like “don’t ask, don’t tell.” We don’t know if they’re PD-L1—positive. We don’t know if they’re EGFR- or ALK-positive. Part of the rationale is, if they progress to metastatic disease, we’re going to get that information because we’re going to rebiopsy and test. Do you have that information available for your patients with early-stage disease at your institution?

Edward B. Garon, MD: We do. We have tried to do that. There are a host of reasons, including a very practical one of not cutting into the block to make slides a bunch of times. So we generally do have that information.

Leora Horn, MD, MSc, FRCPC: Are you testing everybody?

Thomas E. Stinchcombe, MD: We reflexively test for PD-L1 regardless of the stage. I think all of our patients have that done. Part of it’s like what Eddie said: The efficiency is there. How to apply that information—I don’t really know at this point. I think Eddie brought up a good point. We’ve always seen the PD-L1 in the context of other clinical factors; not always, but most of the time.

Leora Horn, MD, MSc, FRCPC: It’s hard to know. At every talk that you go to, every meeting ends with, “We need better biomarkers.” PD-L1 was not a great predictor against docetaxel in the second-line setting with nivolumab and atezolizumab. So we need the better biomarker to really figure out who that patient is.

Transcript Edited for Clarity