CHECKMATE649: Trial Design and Efficacy Data

Sam Klempner, MD: Do you want to describe the CheckMate649 study and the top-line results?

Daniel Catenacci, MD: This is a large study, one of the largest that’s been conducted for this disease. It was conducted globally, and it was a randomized 3-arm study. The 2 arms that have been reported to date are the chemotherapy backbone control arm of FOLFOX [folinic acid, fluorouracil, oxaliplatin], and the open-label FOLFOX plus Nivo [folinic acid, fluorouracil, oxaliplatin, nivolumab]. The third arm, which we’ve not had reports on yet, is Nivo, Ipi [nivolumab, ipilimumab]. What we know is the chemotherapy controls with or without nivolumab, and the primary end point of that study was changed a few times. But the final primary end point was overall survival in patients with a PD-L1 score of CPS [combined positive score] 5 or higher. There was a secondary analysis, should that have been met, to look at all-comers and look at CPS 1 or higher.

A few important points are that the way the PD-L1 testing was scored was with a different antibody than had been used up until that point, based on the pembrolizumab studies with the 22C3 antibody, and this was now with a different antibody, 2088. There’s potentially not much difference, but they’ve not been compared directly, at least in this tumor type, so there’s that variable. And this was the first time that nivolumab studies had looked at a CPS score, as opposed to TPS, or tumor positivity score. Based on that, we saw first that the incidence of CPS 5 or higher was about 60% of the patients, and the CPS 1 or higher was just over 80% of the patients. Then the remainder were CPS negative, which I have to say at least in my experience, I’d love to hear what you say, but in my all-comers, I feel like our incidences are much lower for the CPS high scores, but that’s what was seen in this study.

Then the final outcome was that survival was improved, as you mentioned, with FOLFOX plus Nivo [folinic acid, fluorouracil, oxaliplatin, nivolumab]. It was one of the few studies that have shown in a phase 3 setting a median overall survival above 12 months. The only other one that had that was the AVAGAST study, which was considered a negative study, but the investigational arm was 12.1 months or something like that. So this was an improvement from 11-ish months to 14 or so months, an improvement of around 3 months median in the CPS 5 or higher group. Then the patients in the all-comer assessment, when including all-comers, it was still improved, and the hazard ratio was diluted to some degree and was still statistically significant.

What we just learned recently from the actual publication and the updated report at ASCO [the American Society of Clinical Oncology annual meeting] this year was, as you’ve been pointing out, that those in the low level arms are not driving much benefit, with a hazard ratio of 0.94, and the median overall survival of 12.3 vs 12.4 months. It’s being driven by patients at the higher expression level of PD-L1. Based on that, what do you make of these results, considering the results of KEYNOTE-062, which was a study of chemotherapy with or without pembro [pembrolizumab] but was considered negative in the first line? How do we reconcile that?

Sam Klempner, MD: A lot of data to dive into for sure and all good questions. In the biomarker-enriched CPS greater than 5 and CPS greater than 10 [groups], the differences are a little bit less between CheckMate649 and KEYNOTE-062, although 062 was negative, including in that subset. Some of that is perhaps unbalanced post-trial therapies. There was a greater post-trial use of second line and later therapies in 062 compared to 649. There was certainly an enrichment for the population that’s driving the greatest benefit as you just alluded to, the CPS 5 or higher made up 60% of the trial. There’s also the impact of the large sample size and ability to granularly assess some of these subgroups in more detail. If you put it in the context of all of the other trials, ATTRACTION-4 and KEYNOTE-590, the overall trends are consistent that we can enrich for people who are going to derive a greater magnitude of benefit. Some of the differences in outcomes may come down to the use of later line therapies and the difference in the populations comprising the trials. But for the practicing oncologist, the take-home message is that this is a biomarker that matters even though the approval is independent of the biomarker, which I was surprised by it, but it is still an important one to assess.

Daniel Catenacci, MD: I would agree with you, and surprise is a good word. My phrase these days has been, “Just because you can, doesn’t mean you should.”

TRANSCRIPT EDITED FOR CLARITY