Johanna C. Bendell, MD: You know what everybody is hot on now—aside from being able to use antiangiogenic agents for gastric cancer—is immunotherapy, immunotherapy, immunotherapy, and I think you in Japan were the first to present randomized data on the use of nivolumab for gastric cancer. Can you tell us about ONO?
Kohei Shitara, MD: Yes, this is an Asian trial including not only Japanese but also Korean and Taiwanese patients who were randomized patients to nivolumab and a placebo arm. The subjects are refractory with at least 2 or more previous chemotherapy treatments. Actually, overall survival was improved with use of nivolumab. Median survival was 1.2 months, but the hazard ratio is 0.63—clear benefit. One-year survival was more than doubled, and nivolumab was very well tolerated in this heavily pretreated patient population. So, nivolumab was approved last September and included as a standard third-line or later-line treatment in Japan.
Johanna C. Bendell, MD: One difference between Japanese guidelines and the United States guidelines is PD-L1 status. Can you tell us a little bit about why there’s that difference there?
Kohei Shitara, MD: Actually, in this ATTRACTION-2 Asian trial, PD-L1 status was evaluated, but the tumor tissue was just available from 40% of patients, and their PD-L1 status was just evaluated in cancer cells, not in immune cells. So, as a result, PD-L1 expression on tumor cells did not show any difference between positive patients and the negative state of patients, in terms of hazard ratio and overall survival, which were also included in the paper. As a result, nivolumab can be used regardless of any status of PD-L1 in Japan. But I don’t think this is universal, because this was based on a very small sample size by molecular analysis. Maybe we need more data from good tumor specimens.
Johanna C. Bendell, MD: Yelena, what do you think? Do you think it’s a PD-L1 story?
Yelena Y. Janjigian, MD: Well, looking at the pembrolizumab data and the positivity score, it’s a moving target. Looking at the stroma and the surrounding tissue is very important, so that’s the distinction. While looking at CheckMate-032 or the ATTRACTION-2 studies, the positivity rate was just 30%. If you look at PD-L1 expression by the combined positivity score of 1% or more, that’s 70% of the patients, so you’re really not excluding that many patients. Maybe it’s logistically a little tricky to get the tissue and get the testing done before you can get your patients to pembrolizumab. But the majority of the institutions are quite comfortable with PD-L1 testing because of the lung cancer data. So, you just need to educate your pathologists, saying that the bar for PD-L1 positivity in gastric cancer is quite low; even if it’s 1%, that’s what we need to report. Whether or not PD-L1 positivity is the biomarker, that’s to be determined. Ideally, we hope to find an even better biomarker, but it is important to narrow down the scope of patients who will receive this therapy to improve the benefit to the overall population.
Johanna C. Bendell, MD: We’ve seen this data from KEYNOTE, which is showing that in the earlier lines of therapy when you treat with the pembrolizumab, potentially, you do better. What are your thoughts on whether or not we might be moving the checkpoint inhibitors up to an earlier line? Is that the right thing to do?
Yelena Y. Janjigian, MD: For MSI patients, absolutely; for EBV patients, probably; for unselected population, even patients with PD-L1 expression, may not be. The truth is, gastric cancer is an aggressive disease, and even in second-line therapy, patients present with peritoneal carcinomatosis, a lot of symptoms, and if you don’t induce response quickly, you may lose some patients. What the randomized study has shown—which Kohei will be presenting at this meeting—is that in most patients, the chemotherapy is still going to be an important part of this puzzle, probably because, first of all, we’re not enriching well enough, and secondly, because the time to respond for immune checkpoint inhibitors is not as brisk as with chemotherapy. So, if you do respond, you tend to respond longer, and the survival curve has a tail on it. But if you don’t respond quickly, you’re in trouble. Kohei, maybe you can give us a preview of what you’re going to discuss.
Kohei Shitara, MD: This is a very important study based on the KEYNOTE-059 trial, which showed lines of treatment with PD-L1 positive. So, the KEYNOTE-061 trial was conducted to compare second-line paclitaxel and pembrolizumab for PD-L1—positive tumors as a primary endpoint. The population with CPS of 1 is the same as the FDA indication. Now, we did not use paclitaxel and ramucirumab, because this trial was performed in 2014. At the time, paclitaxel and ramucirumab were not available in many countries. So, as a result, this trial could not show a statistically significant prolongation of overall survival. Hazard ratio was 0.82, with a P value of 0.042; very borderline results. Because we did the interim analysis, and PFS was also included as a primary endpoint, the data are not sufficient to show a positive result. But a very important point is the molecular analysis. Plus, we also included the PD-L1—negative patient population, so we can show the negative PD-L1 patient population outcome, and our abstract has shown that a higher PD-L1 expresser showed a higher benefit, with a hazard ratio of 0.64 compared with paclitaxel. This is a very remarkable benefit, so it should be grounds for additional studies.
Transcript Edited for Clarity