David H. Ilson, MD, PhD: Now, the advent of immunotherapy is complicating the discussion of first-line treatment in gastric cancer. We have approval in the West for pembrolizumab, the anti–PD-1 [programmed cell death protein 1] drug. In Japan, there is approval for nivolumab, an anti–PD-1 drug. Both of these approvals are in the refractory disease setting. We’ll talk about later-line therapy, but in the first-line setting, we have results from the KEYNOTE-062 trial, which was a trial in patients who were PD-L1 [programmed death-ligand 1] positive at 1% or higher. It was a head-to-head comparison of chemotherapy alone, fluorinated pyrimidine platinum chemotherapy plus pembrolizumab, or pembrolizumab alone. What we saw from this study comparing chemotherapy with or without pembrolizumab is that there was no meaningful difference in combining an immunotherapy drug up front with chemotherapy compared to chemotherapy alone. There was a slightly higher response rate, but no meaningful differences in progression-free and overall survival [OS], really arguing against the combination of an immunotherapy drug with first-line chemotherapy.
Interestingly, between chemotherapy alone versus chemotherapy plus immunotherapy, there was no difference in terms of CPS [combined positive score] or PD-L1 score. If you look at the 1%, or the 10% or higher patients, there was still no benefit in adding immunotherapy to chemotherapy up front compared to chemotherapy alone. The exception would be patients who are MSI [microsatellite instability]-high. That was a very small percentage of patients on KEYNOTE-062: it was only about 6% or 7% of patients. But in those patients, the checkpoint inhibitor may have behaved better than chemotherapy as first-line treatment. There were higher response rates, more durable response, and median survival wasn’t even reached.
The exception in first-line treatment would be the possibility of using drugs like pembrolizumab for MSI-high patients. These patients have exceptionally high response rates and a durable response. On that trial, there was also evaluation in MSI-high patients of immunotherapy plus chemotherapy, and it didn’t seem that chemotherapy added benefit to immunotherapy. I think the exception would be MSI-high patients.
Now, we have to remember that sometimes when our patients with advanced metastatic disease walk in the door, we don’t have results back from mismatch-repair studies or MSI status or genomic profiling. My suspicion would be that these patients would start with chemotherapy, and then if we detected that they’re mismatch-repair deficient or MSI-high, we consider transitioning them to immunotherapy. It’s not clear whether combining chemotherapy with immunotherapy is any better than immunotherapy alone. In the first-line setting, immunotherapy can be considered for the MSI-high patients. Knowing that when they walk in the door we won’t have results of that testing, we would often start them on chemotherapy and then consider earlier-line use of immunotherapy drugs.
Kei Muro, MD: As you know, the KEYNOTE-062 trial demonstrated noninferiority of overall survival in pembrolizumab monotherapy compared to standard chemotherapy for the CPS 1 or higher cohort. In addition, the forest plots in subgroup analysis for OS show clear benefit in Asia in contrast to the West. The trend is more pronounced in patients with a CPS of 10 or higher. Furthermore, the MSI-high population, which occurs only 5% of the time in advanced recurrent gastric cancer, has a clear benefit from IO [immunotherapy]. When IO is used as a first-line therapy, MSI-high is the most promising biomarker in the case of a single-agent IO. CPS 1 or higher—especially CPS 10 and above—are also promising biomarkers.
Daniel V. Catenacci, MD: In the second-line setting, specifically with respect to MSI-high patients, immunotherapy, of course, is the standard option if they’ve not had exposure to immunotherapy earlier. Another subgroup of patients may benefit based on the KEYNOTE-061 study, which was a randomized study of paclitaxel or pembrolizumab, open label. In a subgroup analysis, it appeared that in addition to MSI-high tumors, patients with higher levels of PD-L1 expression—a CPS score of 10 or higher, specifically—seemed to derive a benefit. In other subgroup analyses of other studies, such as KEYNOTE-062, this was not the same. There was a yin-yang effect, even in the CPS 10 or higher group, in the KEYNOTE-062 study comparing pembrolizumab monotherapy to a chemotherapy arm.
Whether CPS 10 is the optimal cutoff is yet to be determined. These have all been retrospective unplanned analyses in small subgroups that are subject to a lot of random variation. In one study, we see benefit; in another study, we don’t. But I think that the general trend that we consistently see is that, after excluding MSI-high tumors—so among the MSS [microsatellite stable] patients—the higher the level of CPS score, the higher the probability that a patient will benefit. It’s not guaranteed that they will benefit at a very high level of CPS, CPS 9 for example, but they have a really good chance. The lower the level, the lower the chance of benefit, to the point where if you’re less than CPS 1, you have a very low chance, if any, of deriving benefit.
It’s more of a linear parameter. We like to think in a binary fashion of yes/no or positive/negative, where we have to make cutoffs; but ultimately, that’s going to be subject to false negatives and false positives at any cutoff level. Whether CPS 10 is the optimal cutoff, or an even higher CPS score, is yet to be determined. But I think that one day we will, in addition to MSI-high tumors, be offering monotherapy checkpoint blockade to this optimal subgroup that we can define: if not as second-line or first-line therapy, at some point early in their disease course. I think that will be best for those subgroups of patients.
Transcript edited for clairty.