Johanna Bendell, MD: Absolutely. Moving from esophagus, we’re going to move through the gastroesophageal [GE] junction. What do you think, Ian? What’s your interpretation?

Ian Chau, MD: For gastric, the data are going to be mainly driven by the data from CheckMate649 because that really is, at this moment, the only positive study of using chemotherapy plus nivolumab. Certainly, for a CPS [combined positive score] of 5 or more, I’m very convinced about the data. For a patient with gastric adenocarcinoma, what you call type 2, type 3 junction of adenocarcinoma, I think that would be my preference.

Between CPS 1 and 5 is going to be the harder part. Certainly, CPS 1 or more, you can still see benefit. In fact, Yelena already mentioned that for all-comers, it was statistically significant. That was with a big sample size and a hazard ratio of 0.8, so you may not think that benefit is enough for all-comers.

To me, when I look at CPS 1 or more—in CheckMate649 that is 80% of the patient population, which is a large proportion of patients using that particular immunohistochemistry antibody to test for PD-L1 expression—the hazard ratio is still quite impressive. If you look at the survival curves, they do clearly separate as well. I would have that discussion with the patient, to say that there is some benefit and that it has become more difficult, whether that is going to benefit to them because clearly there is a proportion of patients who don’t benefit in that kind of lower PD-L1 expression.

My thought at the moment is that a CPS 5 or more is clearly something that I would discuss and recommend to patients. Anything lower than 5 needs to be quite a careful discussion. Especially in many countries outside the Untied States, maybe southern hemisphere countries, reimbursement is a very important factor. Taking these different subgroups to the reimbursement agency and having that discussion will be the challenge in the next 12 to 24 months about who would be reimbursed.

Johanna Bendell, MD: Dan?

Daniel Catenacci, MD: I would bring up what Ian brought up. I was talking about the CheckMate649, that CPS 5 or higher was the amended primary end point. Midtrial, it was amended to make the primary end point, and the sample size was increased. That was the primary end point. It was clearly positive, and we will be using FOLFOX [folinic acid, fluorouracil, oxaliplatin] and nivolumab. The question of whether less than 5 is very important.

What we’ve seen in the public data set here at the European Society for Medical Oncology Congress that was presented is the 1-or-higher group lumped together with the 5-or-higher group. That’s about 80% of all patients, as Ian mentioned, and then we have all-comers. We see the outcomes are “better and statistically significant,” but they look like a crescendo plot. I dubbed the yin-yang plot of monotherapy, where you see some detriment and then crossing of the curves. What we see here with all-comers especially is this crescendo curve, meaning there’s a bunch of patients at the beginning. They’re not deriving much benefit, and as you go on, you see this increased benefit over time.

What I need to see before I’m going to be giving my patients with less than 1 CPS or the 1 to 5 CPS are the actual data of those patients, not with the 5-or-higher pulling the curve over. I want to see that. That’s not been presented. Hopefully it’s in the publication, and if not, then I will still be asking that question. Then again, Ian’s point is about reimbursement. It will be very interesting to see how the different regulatory bodies—FDA, EMA [European Medicines Agency], and the Asian countries—approach these studies. For example, KEYNOTE-181, which again looks identical to KEYNOTE-590, they looked through and approved only CPS 10 or higher in squams. How are they going to look at KEYNOTE-590, and how are they going to look at CheckMate649? Are they going to expand to the secondary end points of 1 or higher or all-comers? Are they going to hold it at 5 or higher? Are they going to demand to see those data? That’s what I’m demanding. I hope we see those data shortly.

Transcript Edited for Clarity