Bradley J. Monk, MD, FACOG, FACS: Let’s transition to immunotherapy. Immunotherapy is the newest and the greatest treatment of solid tumors, and even liquid tumors. Unfortunately, except in the MSI [microsatellite instability]-high endometrial cohort and in the PD-L1 [programmed death ligand 1]—positive cervical cancer cohort, it’s all off label. And even in cervical cancer, it’s only a response rate of 14.3%.
There was a paper presented last year at ASCO [the American Society of Clinical Oncology] on KEYNOTE-100. I just accepted it today in Annals of Oncology. I don’t know if that’s a secret, but I don’t think it is. Tell us about KEYNOTE-100.
Kathleen N. Moore, MD: So KEYNOTE-100 was a very large monotherapy study of nivolumab?
Bradley J. Monk, MD, FACOG, FACS: Pembrolizumab.
Kathleen N. Moore, MD: Pembrolizumab—gosh, I’m losing my mind as well. Pembrolizumab, in recurrent ovarian cancer. It was really a study looking at efficacy, but it was also trying to figure out what the predictive biomarker could be for response. It had 2 cohorts. One was patients who had 1 to 3 prior lines of therapy and the other smaller cohort was 4 to 6 lines of therapy.
In the first cohort, they used the first hundred patients or so to identify what the cut point for the combined positivity score should be, or the CPS score. And so, the CPS score is not PD-L1—positivity as has previously been described. It’s number of tumor cells, macrophages, and inflammatory cells that are positive for PD-L1 divided by the total number of tumor cells times 100.
So it’s a different score. You could be PD-L1 positive but CPS 0. So it’s a different way of looking at more of the tumor microenvironment. That’s how they scored it, and they came up with the score of less than 1 as a cut point. One, greater than 1, and then greater than 10, and that’s the score. It doesn’t have a percentage. So those are the cut points they identified in the first hundred-ish, 90-ish patients on the 1 to 3 cohort, and then did a validation with that group, and then they validated it again in the 4 to 6 group.
What they found was that the CPS score of 10 was predictive of overall response. You did see that response rate increase from 8% to 18%, roughly. But the confidence intervals, even with the pretty large number of patients on the study—over 400 patients—still overlapped. Every one of them overlapped, so it wasn’t a statistically significant difference; and the durations of the responses were relatively short. So it was sort of a mixed message study. Greater than 10 was probably a good cut point for picking those patients who would have the best likelihood of responding, but it still didn’t pick a group of patients who were going to do really well.
Bradley J. Monk, MD, FACOG, FACS: So let me go through the 3 conclusions. All-comers—what’s the activity of pembrolizumab in ovarian cancer?
Kathleen N. Moore, MD: All-comers—8%.
Bradley J. Monk, MD, FACOG, FACS: Ten percent, 8%, 9%—bad.
Kathleen N. Moore, MD: Bad, yes. And the KEYNOTE-100 was like 8% overall.
Bradley J. Monk, MD, FACOG, FACS: And is there T-cell exhaustion? In other words, regarding the number of lines of therapy, is that an important thing? Is earlier treatment better, or is it about the same?
Kathleen N. Moore, MD: We would think that early treatment would be better, but the cohort of 4 to 6, in smaller numbers, had higher response rates. So it’s a bit backwards to me.
Bradley J. Monk, MD, FACOG, FACS: There you go. So T-cell exhaustion, number of lines of therapy, an important baseline activity, 8% to 12%. And then third, is there an opportunity for this enrichment with the PD-L1 score, however you measure it? And the answer is, maybe.
Kathleen N. Moore, MD: Maybe.
Transcript Edited for Clarity