From Trial to Treatment: Exploring the Lasting Impact of KEYNOTE-426 in Kidney Cancer - Brian Rini

June 20, 2023

Pedro Barata and Brian Rini unpack the findings and implications of the KEYNOTE-426 trial. The trial assessed the efficacy of a pembrolizumab and axitinib regimen in advanced clear cell kidney cancer, contrasting it with the then-standard sunitinib monotherapy. The KEYNOTE-426 trial's success contributed to changing the standard of care in this field. It demonstrated superior response rates and progression-free survival, regardless of the risk group. Dr. Rini presents five-year follow-up data, reflecting the sustained benefits of the tested regimen, and emphasizes the importance of giving patients access to all active mechanisms upfront. Dr. Rini also suggests that limiting TKI use and focusing on immune response in future trials may yield valuable insights, highlighting the potential role of HIF alpha inhibitors like belzutifan.


Brian I. Rini, MD, FASCO Vanderbilt-Ingram Cancer Center, Nashville, TN

Pedro C. Barata, MD, MSc, Leader of the Clinical GU Medical Oncology Research Program, University Hospitals Seidman Cancer Center, Associate Professor of Medicine, Case Western University, Cleveland, OH

Read the Full Video Transcript

Pedro Barata: Hi, I'm Pedro Barata. I'm a GU oncologist at University Hospital Seidman Cancer Center, Case Western University of Cleveland, Ohio. I'm very, very happy to be joined today by my mentor, friend, colleague, kidney cancer expert, Dr. Brian Rini. He's the Chief of Clinical Trials at Vanderbilt-Ingram Cancer Center. Dr. Rini, such a pleasure to have you here today.

Brian Rini: Pedro, thanks. Appreciate it.

Pedro Barata: Absolutely. So first of all, great job with, as usual, presenting the data on KEYNOTE-426. Once again, another podium presentation at ASCO, which is a larger stage, very important data, and so, amazing session. So congrats for that. And I'm thinking maybe we can check a little bit about it. Some of the folks are familiar with KEYNOTE-426. Others might not be as familiar, at least with that name. Can you just briefly summarize the design of the KEYNOTE-426, which you conducted a few years ago? And why you presenting the data with such long follow up at ASCO? What's the opportunity for that?

Brian Rini: Sure, thanks Pedro. So KEYNOTE-426 was one of the several IO based doublet trials that has set the standard of care in front-line advanced clear cell kidney cancer. So it was a large trial that randomized previously untreated advanced clear cell RCC, to either the old standard of care, sunitinib monotherapy, b542, which was the control arm for all the trials. Or, an experimental regimen of pembrolizumab plus axitinib, so-called IO/TKI. At the time, there was a clinical reason to do that. TKI monotherapy was a standard of care. Immune therapy clearly had activity. Nivolumab mono approved in the refractory setting, probably right around that time. So it was, in some ways, just a clinical A plus B combination. And then there's also some preclinical rationale about VEGF being immunosuppressive, and that may be by inhibiting VEGF slash receptor that you could make immunotherapy work better. I don't know that the latter has worked out, or is proven or disproven. I think there's clearly an additive effect. But anyway, that was the rationale.

Trial was conducted as you said, five plus years ago now, and this is actually the fourth analysis. So there was the initial analysis and publication, because it had hit a pre-specified number of events for PFS, but OS happened to be very positive at that first interim analysis as well. Hazard ratio of 0.53.

There was a second analysis with a minimum follow-up of about two years. And then, a protocol specified final analysis with a minimum protocol follow-up of three years. All that's been presented slash published. And the data were pretty consistent, in terms of response rate, PFS, survival advantage, et cetera, as we've seen across regimens.

At ASCO this year, we presented the five year, so five-year minimum follow-up for all patients, and median of 67 month follow-up. So it was actually beyond the protocol specified final analysis. And I do give Mark credit for doing this, because it didn't really need to be done, it wasn't in the protocol. But as I think we'll dig into, these long-term follow-up data sets are really important, and have stimulated a lot of, I think, really good discussion, and that's always good for the field, I think.

Pedro Barata: Right. No, those are fantastic summary, and I agree with you. It's so important. So let's talk on that, since we're talking about follow-up. You presented it for follow up five years, right? And there's a couple of things that were striking, I think, and people look at that, and kind of try to put that into context. One is actually quite interesting to see, that number one, the median OS has reached for all subgroups, including good risk, right? And we're talking in the 50% range or so. It's also interesting to see that actually, over 70% of patients treated sunitinib end up getting subsequent therapies. And I think, over 60% on axi/pembro end up receiving subsequent therapies. I find those numbers actually be higher than I would particularly anticipate. Would you agree with that?

Brian Rini: Yeah. I think that's a good thing, right, is this study was the earliest of the IO/TKIs, but there was still approved subsequent therapies around the world, not in every country, and in every therapy, of course. Certainly, in the US there was. So that's what we want to see.

It is really hard to conduct a study with crossover, and we've tried that in kidney cancer unsuccessfully. And that's always the criticism. Well, sunitinib patients didn't cross over, blah, blah, blah. But that's just not how real life works. And patients don't get subsequent therapy, either because they unfortunately pass away, right? So it's an argument for giving your best therapy upfront. Because, as you well know, not everybody makes it to second-line, even on a very selected trial setting, or access issues. But despite those issues, the vast majority got subsequent therapy. In fact, the majority got multiple lines of subsequent therapy. 80% of sunitinib patients who got therapy got a PD-1. So I think this, I hate to use the term real world, because people throw that around, but it was a reflection of what was going on at the time, and still, is that patients get multiple lines of therapy. That's how it should be.

Pedro Barata: Right. So speaking of that, let me point the data on the good risk, right? Because there's been a lot of conversations around that.

Brian Rini: Yeah.

Pedro Barata: So for those who are listening to this, might be a little bit less familiar with that. The median OSs were actually very similar, hazard ratios over one. No difference, basically, no statistically. The study was not powered to find a difference in that subgroup, I think that's relevant. KEYNOTE, they enroll about 30 or so percent of patients with good risk. But it's interesting to see, is the curves do get closer towards the end. And so, I guess, some folks look at that and might potentially think, "Oh you know what? There's no difference in OS for patients getting sunitinib or IO/TKI. Do you think that can be a dangerous statement out of the data the way we're looking at it? What is your take on it, and what do you think is going to be the impact for the community providers out there, when they're looking at these data?

Brian Rini: Yeah. It's a good question, and a complicated one. As you say, it's always dangerous to draw sweeping conclusions off subset analyses. And of course, comparing subsets across trials is even more difficult.

Good risk patients are good risk for two reasons. One is, they likely have more indolent disease. So showing an advantage to that initial therapy, which is given for, call it 15, 16 months, when these patients live five plus years, is going to be challenging. And also, we know from biologic studies, that they are more angiogenic driven. Not exclusively, but call it 60% from some of the IMmotion151 data, predominantly, let's say. Well, the IO/TKI versus TKI, both arms are getting anti-angiogenic drugs. So you're really counting on that immune subset of favorable risks to drive the survival benefit. And that's going to be smaller. I don't know what the number is, but let's say it's 20 or 30%. So I don't think it's surprising that there's not a survival difference.

Importantly, there is a response rate CR and PFS advantage. Which in some patients, is more important, and in some patients, less so. So that's an individual patient decision. I'm not a believer in giving TKI monotherapy, largely because nobody is cured with that therapy. Not a single patient is cured with TKI monotherapy, without resection of mets, or some other modality. So as we just said, not everybody makes it to second-line therapy. So 20% of the sunitinib patients didn't make it. Now maybe some of that was access, but as mentioned, some of that's because they get sick and die. So I'll take my chances with a doublet, even if I can't stand up and say there's a survival benefit, because there's other clinical benefits. Because I know I'm giving the patient all active mechanisms upfront, and frankly, we don't want to lose patients, right?

Pedro Barata: Right.

Brian Rini: And I think the final point is, there's not a lot of added toxicity. We didn't represent toxic data honestly, because there was nothing new, and we only had 12 minutes, so we focused on other things. But nothing, absolutely nothing, was new from a toxicity standpoint. And yeah, there's some immune mediated this and that, of course, but there's really... And you've given plenty of these regimens. There's not a whole lot of difference between giving sunitinib and giving axi/pembro. And I know those aren't the choices these days, but in terms of a tox to the patient quality of life, we don't see much difference. I think there's little downside and some to gain.

Pedro Barata: Yeah, I agree. Great way to summarize that. Another perspective, or a lot of pieces of data that you present, is kind of how patients who completed two years of IO did afterwards, right?

Brian Rini: Right.

Pedro Barata: And one way, I guess, to summarize it can be, patients who get there, the patients who do better on-

Brian Rini: Do better. Yeah.

Pedro Barata: ... tend to do better afterwards, right? And you've seen this across different studies. We've seen it with cabo/nivo, so with lev/pem. Focusing on axi/pembro, when you're thinking of that, knowing everything we know now, there's a big discussion about, should have we stopped at two years, the IO? Can we continue TKI? Should have we done the other way around? When we look at the feel of the curve, maybe 30% or so, what is your thoughts? If you were to design this study again, would had you thought of a different design, where actually you would continue the IO and hold the TKI? Because to your point, TKIs, we probably not, they don't hear anybody probably in that setting beyond two years, right?

Brian Rini: Yeah.

Pedro Barata: So what do you take on what happens to patients who do great after two years?

Brian Rini: So the last few slides were, what we call completer analysis. So about 28% of patients randomized to axi/pembro, I think it was a 120 patients, finished two years, 35 doses. The characteristics of those patients are exactly as you'd expect. A little more favorable, a little less bulky disease, enriched for responders, 85% response rate. I think almost 20% CR. So exactly as you'd expect. And as you say, not surprisingly, patients who do well, do well. So they did well over time. 70% alive at five years, tail of the PFS curve above 30%.

Would I design it differently? Yeah. I would love to see designs that stop a TKI, or give it intermittently. Or frankly, not give it at all, if we can get there. But those aren't data we have. There are some small studies ongoing looking to limit TKI, and I applaud investigators for doing that.

In practice, I don't think giving more immune therapy is the issue, because we didn't see all those patients fall off a cliff. We didn't see them progress and die after stopping immune therapy. And that's not what immune therapy is, right? Immune therapy is give it upfront, generate an immune response, and then stop it. I stop at practice in two years, and I prep patients a few infusions ahead of time. Hey we're coming to two years, we're going to stop, just so I don't spring it on them. Because they don't know per se.
In practice with a TKI, I'm very liberal. And I've had patients who say, "Hey doc, I don't want to take this pill the rest of my life." I say, "Great. Let's stop and see what happens if you're willing." Or, you know what? Take liberal breaks.

And also, if the patient's tolerating it wonderfully, then it's less of an issue, than if they're having all sorts of chronic nagging side effects, right?

Pedro Barata: Right.

Brian Rini: So I have no problem stopping, but I have to tell a patient we don't really have data to support that. Although, my sense is, it's more important upfront. It's disease control or shrinkage, and then let the immune component kick in. And I don't have great data to support that, but that's my sense.

Pedro Barata: Right. No, that's a great way. So final question before I let you go. We're seeing a little bit of novel MOAs, maybe HEF2 is perhaps, the winner among the ones that are probably closer to get available to the treating providers out there. And this moving earlier is being studied in the frontline, but I'm thinking, how can we go from this? We definitely improve outcomes of men who come to us with advanced RCC, particularly clear cell RCC. And it sounds like we were using the drugs we had available, then we are trying to combine them, potentially, for a synergic opportunity there, if not additive value. How can we go from here? That's maybe an opportunity to highlight your OPTIC study, and others, talking about a biomarker based approach.

Brian Rini: Yeah. I think the way we've cured solid tumors is combining multiple mechanisms of action. And that's what we're trying to do in RCC. COSMIC was a first attempt at that, that was limited by toxicity. Could a HIF alpha inhibitor, like belzutifan, replace a TKI, clearly more tolerable? Could it have the same effect but less toxicity, and i.e., be more combinable? That is being tested in one of Merck's triplet trials, or their triplet trial. So I think, we're going to see more attempts at triplet with the mechanisms we have, and maybe some novel mechanisms like belzutifan, like LAG-3 inhibition, et cetera.

And then, as you mentioned, we're doing a biomarker based study looking at RNA-seq and using the IMmotion151 clusters to assign treatment. And I view that study, number one, as an operational challenge. If we can get that done, it at least sets the platform for future study. I don't think that initial study is going to be definitive, to be all and end all as much as I'd like it to be. And then it gives us the opportunity to test other drugs, and other mechanisms, and other clusters. Then, it's a platform to test a whole lot of things, and perhaps get towards personalized medicine, which we really don't have in kidney cancer. Right?

Pedro Barata: Right.

Brian Rini: There's a lot of hand waving, but we don't have personalized medicine. We have my biases, and yours, and somebody else's. And good drugs. And we're making patients live longer, but it's very empiric. We give drugs, and we see what happens.

Pedro Barata: Right.

Brian Rini: I use that line with patients, and I say, "It's not terribly scientific, but that's what we do. So let's get you to the first scan. Let's see how you do, and we'll all be a lot smarter about how you're going to do." Right? So I'm all for biomarker based approaches. We put a ton of effort into the OPTIC study. And we just need to do more of it. And frankly, we need big pharma support to do it, right? Ours is a DOD funded relatively small effort. But if we're really going to validate biomarkers and use them clinically, it's, as you know, it's going to have to be pharma funded by and large.

Pedro Barata: Yeah. No, very well said. Of course, so bias, I have to disclose that we reach out, we're trying to participate and help. Your study is really relevant. To your point, I think we'll set up the stage for the next level of biomarker based approaches in developing this kind of platform. Whether it's by RNA-seq or something else. Or a score, combining that plus other information, that get us better than the clinical and laboratory characteristics we're now using to define risk groups, right? And maybe to improve IMDC.

Brian Rini: Yeah. I think IMDC was meant to be a prognostic tool, not predictive, as Danny Heng, himself, would admit. We've really painted ourselves into a corner with IMDC, right?

Pedro Barata: Right. Right.

Brian Rini: I love that. It's great. We participated. Danny was in Cleveland when he started it. But we've really put patients, we've boxed them into this clinical characteristic, when it's not predictive, really, at all.

Pedro Barata: Right.

Brian Rini: Yet, we're making treatment decisions based on it. I think it's a great prognostic tool. I think we need better predictive tools.

Pedro Barata: Right. That's well said. And a well way to summarize this. Dr. Rini, congrats. Great session. I welcome everybody who didn't get the chance to be there live in Chicago, to actually go and watch the kidney session around Keynote-426. Great presentation, great discussion. And always a pleasure to have you, Dr. Rini. Thank you.

Brian Rini: Pedro, thank you. Appreciate it.

Pedro Barata: Talk soon.