KEYNOTE-811: IO-Based Therapy for HER2+ Gastric Cancer

EP: 1.Historical Perspective on Treatment of Gastric and GEJ Cancers 

EP: 2.Recent FDA Approvals in Gastric and GEJ Cancers

EP: 3.CHECKMATE649: Trial Design and Efficacy Data

EP: 4.CHECKMATE649: Safety Profile and Toxicity Management

Now Viewing

EP: 5.KEYNOTE-811: IO-Based Therapy for HER2+ Gastric Cancer

EP: 6.Biomarker Testing for Gastric and GEJ Cancers 

EP: 7.Frontline Therapy for Metastatic PD-L1–, HER2+ Gastric Cancer

EP: 8.Sequencing After Progression on IO-Based Therapy for mGC

EP: 9.Emerging Agents and Biomarkers in Gastric and GEJ Cancers 

Daniel Catenacci, MD: Let’s move to the HER2-positive. Do you want to tell us the background and the study design of KEYNOTE-811 and how the recent presentation was all about an FDA [Food and Drug Administration] approval?

Sam Klempner, MD: Yes. We’ve known since the ToGA trial that HER2 is an actionable subset of our patients, maybe 20% roughly across the board. Certainly, there’s some mixture within that population in terms of tumors that are very driven by HER2 and very high staining and those that are lower, but ToGA, relatively convincingly, showed in a biomarker selected population, giving trastuzumab [Herceptin] on top of platinum chemotherapy improves survival. That was again, one of the first trials to show a survival over a year. Then people have been trying to build on that for a while, unfortunately, unsuccessfully, and built on some of the data for synergistic combinations that you discussed combining the innate and adaptive arms. There was a single institution phase 2 [trial] looking at 5FU [5-fluorouracil] and platinum with trastuzumab and pembrolizumab [Keytruda], but essentially the KEYNOTE-811 regimen, single-institution frontline study that showed a very high overall response rate, very encouraging biologic activity, and an encouraging 12-month survival rates, that provided some of the foundational groundwork for 811. In order to validate those findings, they designed the phase 3 study, which was a 1:1 randomization of essentially 5FU platinum standard chemotherapy, in this case, CAPOX [capecitabine and oxaliplatin] plus trastuzumab which would be our current standard for HER2, I would argue. Then the experimental arm took the same regimen and added pembrolizumab. We haven’t seen anything beyond the first interim analysis but what was shown just recently at ASCO [American Society of Clinical Oncology] and it was nice to see some data since the data came after the FDA approval in May. The first interim analysis was built around response rate. What they showed is that their response rate was impressive. It was around 75% for the experimental arm versus around 52% for the standard CAPOX plus trastuzumab. Notably, if you look at the proportion of patients who remained on trial, which maybe give us a little bit of hint at an upcoming PFS [progression-free survival]. We see that there was like 12% or 15% greater proportion in the experimental arm remained on study with limited follow-up. The follow-up here was around more than 8 months. A couple of interesting things to note is that the population that was enrolled and turned out to be relatively selected…it’s 80% HER2/HER3-positive, which are the highest scoring patients. Basically, everybody, were almost 90% of patients were PD-L1 [programmed death-ligand 1] positive. I’ll let you comment on the differential activity within HER2 subsets. I know you’ve done some work with some other HER2 antibodies suggesting that you can super select patients. It was an interesting breakdown that they wound up [with]. In theory, it should benefit both arms, but it’ll be interesting to see what the outcomes are.

Daniel Catenacci, MD: I agree with that. It’s impressive results for an interim analysis to show the response rate, 22% higher in the pembro [pembrolizumab] arm. It is an interim analysis, and we should point out it’s a phase 2 part of the phase 2/3, which I believe is still ongoing with accrual, it’s not fully accrued and close to accrual as far as I’m aware. That said that this finding has led to an FDA accelerated approval as we all know in this eligibility for HER2 positive. It’s conditional based on the phase 3 part to read out as positive for its primary end point of survival. Your question within HER2-positive criteria, are there other predictive factors? We’ve known for some time that as you pointed out with the IHC3-positive versus IHC2-positive probably the crudest way to look at it, the 3 positive tumors have a better outcome, better predictive benefit than 2 positive tumors. Certainly, more so than say the 0 in 1+ FISH [fluorescence in situ hybridization] positives that used to be considered HER2-positive in the original ToGA study. Even in the tightened second-line study, which made up 25% and a third of the patients, respectively, for those 2 studies. We know that the low-level HER2 expressors, even if there’s gene amplification, don’t do as well as the high-level patients with respect to predictive benefit of trastuzumab. You point out that if this study is somehow imbalanced to having more IHC3-positive patients, we may be seeing something there. You also pointed out that the CPS [combined positive score] scoring is 90% of patients were PD-L1 positive by the CPS1 or higher, where in the KEYNOTE-062 study, which was done in HER2-negative tumors, it was more on the order of like 65% or something like that incident. That’s a big discrepancy and whether there’s some sort of selection bias that’s going on or if it’s an inherent property of HER2-positive disease, we need to see. Either way that leads me to my interpretation of this is that I would like to see the outcomes based on the PD-L1 score, not only above and below 1, which again, will have very few patients. It’s only 10% that are below 1 or 0, but what about the other cutoffs which we know have been even more predictive 5 or 10? What are the incidents first of all in the study, which we don’t know, has not been reported, and what are the differences in their benefit now? Because a lingering question for me, for all these immunotherapy studies up to date, whether it was single-agent immunotherapy or in combination with chemotherapy, or now this in combination with chemotherapy and trastuzumab, is that every time we’ve looked at lower level and negatives, they don’t drive benefit. I’d like to see these data here reported in such manner and hopefully, it’s different here, and that would be great. But if it’s not, we would also want to know. That would be my interpretation and position.

Sam Klempner, MD: I would agree. I’ll just ask the question that probably many people are wondering. With the approval, how do you imagine using this in clinical practice for the next HER2-positive patient that comes through your door?

Daniel Catenacci, MD: We asked that question basically for 649 or 592. With such early data and putting this into context, I would like to see if the patients with a true CPS0 are driving much benefit because we have to also remember there were still patients who weren’t responding to this on the left side of the waterfall plot. Are those the patients that are PD-L1 negative? Is it something that’s easy to predict or not? I guess the other way I would answer it is that being at an academic center where we have many studies, for me, studies take priority, and we would accrue a study instead. Many of which are immunotherapy-based studies, but for the general oncology practitioner who has an approval, as I just mentioned earlier, just because you can doesn’t mean you should. In this instance, it’s HER2-positive. These patients do well when they’re given chemotherapy and trastuzumab anyway in most cases and their median survival for IHC3-plus tumors is almost 18 months in the ToGA study and probably longer now with better later line therapy. We just must keep all that in mind. At my practice, I can only say what I would do, and I still use CPS to help triage patients to those who certainly should get immunotherapy to those who should find another way potentially if they’re on the low end to 0. That would be the way I would answer that question.

TRANSCRIPT EDITED FOR CLARITY