Welcome. Thanks for joining us today.
Yudai Ishiyama: Thanks, Pedro, for a kind introduction. It's an absolute honor to be talking with you here today and discuss the study and showcase our work.
Pedro Barata: No, absolutely. Thank you for joining us, and congratulations. You did a great job around these, I'll call it, fancy methodologic technique that you used looking at the contemporaneous management of advanced renal cell carcinoma. You basically look at different IO combos, at least those who have shown survival advantage for patients present with advanced clear cell RCC, and you embark on for what many of us, it's not as clear, or at least we don't dominate statistics at that level. Things like restricted mean survival time, area under the curves, scores are things that I'm not sure they're very digestible for many of us. That's the reason why we have you to explain to us, break it down to me, at least, what that means and the value.
Let me start first. I congratulate you. This is really good job that you did and presented as a AUA. Tell me a little bit, why did you decide to embark on this project?
Yudai Ishiyama: Absolutely. The concept started off with the understanding as a clinician that it's sometimes very hard to digest and interpret and also to explain to the patients what exactly the benefits shown in the clinical trials mean. Because standard or conventional survival metrics, such as overall survival or the ratios, do not always convey the absolute benefit. They show us the relative number but not the absolute benefit.
It is also said that it's sometimes and many times dilated in contemporary mRCC trials. We know that this restricted mean survival time, or RMST, exists out there, and it does provide absolute time-scale measures of benefit. And it's regardless of proportional hazard ratio assumption.
Our main goal for this study was, by using RMST, can we normalize or generalize the clinical benefit also in terms of side effects and cost so that, if it's normalized by the time we gain from this novel therapies, it will be a little bit more easier when we counsel the patients.
Pedro Barata: Gotcha. Let me do a deep dive into the statistical analysis that you did. I mean, I'm under the impression that you basically got the Kaplan-Meier curves from published phase three trials, ipi-nivo, cabo-nivo, len-pem, axi-pembro, and you kind of do this reverse engineering where you're able to get to extract the data and to basically run it statistically.
Is that correct? Can you correct what I just described? And can you add to it? Because, on top of that, what you did further, I think, was to, as you said, this OS cost per month, which is a concept that might not be very familiar to many of us. Help me to understand those two concepts that are embedded in your project.
Yudai Ishiyama: Absolutely. I think most of what you just mentioned stated is correct. When we have a Kaplan-Meier curve, it usually goes down, but you can calculate the actual area under the curve, and you can compare this to the area that are under this curve A and curve B. Basically, what you're looking at is this difference right here. And we call that a difference in RMST, which will come out as an absolute time measure, and we call that absolute time we gain by using this novel therapy. That's basically an easy way to interpret how we calculate RMST.
The second question is pretty straightforward. We calculated the cost from Medicare. We also included using some of the things that are presented previously, the costs that needed to manage the adverse events and also the personnel cost that needs to be put in if we're doing the IV or infusion therapies. And including all that, all that net cost was divided by the RMST. The difference in cost between sunitinib and the experimental drugs were divided by the difference in RMST.
To be stated in a different way, the difference in cost was divided by the difference in the time gain, meaning, by using this drug, we are spending this much money for this much amount of time that we have to buy.
Pedro Barata: I understand. And I believe you did this that I believe it was around the 42-month range time point, which I'm assuming it derives from where you have enough follow-up for all the four regimens. Those was in common at the time that you did the analysis, but I don't know if that's correct or incorrect because we've seen longer follow-up data for all of these trials. Is the reason why the 42-month truncation point that you all chose?
Yudai Ishiyama: That's a very good question. And that was actually the question that was raised when I was presenting.
Just as you mentioned, it's the longest follow-up that we could use across four trials. And I know there might be some data that were presented in the congress or meeting, but we did ignore those, what we call, immature data. We just stick to the ones that are in a full published style.
As of when we did the analysis that was last year, last summer, that was the longest follow-up that were available for use.
Pedro Barata: Understood. You basically have median follow-up of three-and-a-half years, roughly, and you're actually looking at that survival-cost-per-month gain. Tell me, what were the findings of the study? How do you summarize them?
Yudai Ishiyama: Absolutely.
In terms of the RMST analysis, there were a clear difference in overall survival and progression-free survival. When look at overall survival, the difference between across four trials were very minimal. The least that we could find, that was for KEYNOTE-426, axitinib and pembrolizumab, that's 2.7 month. Now, the longest one that we could find was per nivo-cabo that was still for one month. One-month difference in over survival, if we can call that clinically significant or not. I think that can be argued that can be discussed with the patient.
But, in terms of progression-free survival, there was a lot of bivariation. Ipi-nivo, the DRMST was only 1.4 month, which is sometimes for some people is ignorable. Whereas for lenvatinib-pembrolizumab measurement, the DRMST was over 10 month, which could be clinically significant difference. That was a key finding.
And also, we also found that the cost is pretty different across all trials. And as we imagined, CLEAR trial, lenvatinib and pembrolizumab, was the most expensive. But, because the overall survival was similar those trials, that net-cost-per-month gain were pretty much in the order of the actual cost. Whereas progression-free survival, as I mentioned, pembrolizumab-lenvatinib does provide a lot of survival gain in terms of absolute month. If you normalize it, there was not so much difference in how cost-effective these drugs can be.
Pedro Barata: Understood. That's super, super, super helpful. As I was walking through the results of the study, one thing that comes to mind, just keeping in mind the discussion still ongoing around should we start with a dual immunotherapy combination with ipi-nivo, which you included in this study, versus an IO-TKI, one of the arguments would be the durability of the responses. I'm assuming you did this for intermediate and high-risk, by the way, because there was the intention to treat population at least for the primary endpoints on ipi-nivo. I'm assuming we did that for all, right? We didn't mention that before.
Yudai Ishiyama: Right. Right.
Pedro Barata: Right? But the question that I have for you is, if we were to redo this analysis with a follow-up of six, seven, eight years, when you got more than half of the events for sure, not only in terms of progress-free survival but also median overall survival, which is measured in the four-year range or so ... The question is, are we expecting similar results? Or do you think, as we see those tail of the curves, it would be an argument for those favoring dual IO as a preferred regimen upfront? A longer follow-up would favor ipi-nivo compared to the IO-TKIs?
Would that be a different conclusion? Or it's still aligned with what you found with the three-and-a-half median follow-up cut that you use for these analysis?
Yudai Ishiyama: Thanks. I think that's a great question, and I think it's very clinically relevant as well. The answer will be that, from how this RMST is calculated, we all know that how the Kaplan-Meier curves for overall survival for CheckMate 214 and other trials do look. We do have that data. And, as I said, just because how RMST is calculated, the further that the follow-up goes, it will probably favor, or most likely favor, ipi-nivo regimen when you look at overall survival.
And we also know that, for most of the IO-TKI regimen, the survivals after 48 month tends to cross or comes very similar. And because the experiment or curves does drop significantly at a certain point, that is another thing that will probably favor ipi-nivo regimen, and it will support its use when we look at overall survival which is a primary endpoint for all these trials.
That has already been discussed and been shown in our current analysis. There's not so much difference in overall survival. It may. And it's the most cost-effective, according to our results, so why not use the most cost-effective one when we have full regimens that doesn't show so much of a difference in overall survival. And if we do it for longer follow-up, probably it will even more support that statement.
Whilst, at the same time, we do have to think a little bit different about progression-free survival, which was not a primary endpoint. There should be an argument if we should be looking at this in the first place. But if we look at progression-free survival, we knew this, but it is also redemonstrated by our study that ipi-nivo is not going to help or not likely to help at all.
Pedro Barata: Understood. Thank you. That's super helpful. Before I let you go, is anything else you'd like to raise awareness or to provide some insights as the listeners go through your research?
Any point that I didn't mention and you'd like to bring up?
Yudai Ishiyama: Thanks. I think there's a couple points. But most importantly, this has been stated multiple times in the past, but we are looking at the trials. It is a regeneration and recalculation of the published results, meaning our CTs are not meant to be compared against each other, so we do have to pay a lot of caution whenever we do these kind of analysis. Although I'm an author, and we believe as a team in our own results, I also want to share that this was not meant to compare directly, but to provide more of a content that can be utilized as a segue for shared decision-making.
And I also want to emphasize and bring up one more time that we show that there is a difference in overall survival, not just the absolute benefit, but how cost-effective and how much risk of adverse events that we have to take or patients have to take in order to gain during the survival benefit in terms of month.
Pedro Barata: Gotcha. That's wonderful. And I would just add this is important work. It continues to raise awareness around the importance of offering a combination strategy with a backbone immunotherapy for patients who are facing mortality from advanced kidney cancer.
We've come a long way from that perspective, but, at a global level, the use of combinations is not homogeneous still. And there's a significant number of patients who, for many reasons, are not being offered a combination that's important. And we're trying to figure out what should be the combination we should start with. And of course, there's many factors that we cannot control here like access to subsequent therapies, sarcomatoid features, as you said, different patient populations, more symptomatic visual crisis, et cetera, that would justify as using a combination versus the other.
I agree with your comments completely around the possible drawbacks that you're using data that's available right there. And the goal of the game is not to compare what cannot be directly compared, if you will. Absolutely fair points. And there's actually very extremely valuable insights to the listeners.
Dr. Ishiyama, this is fantastic. Thank you so much. Great congratulations. I'm sure I'll be seeing you soon to discuss more of interesting data that comes out of your team as well. Thank you so much, and congrats.
Yudai Ishiyama: Thank you, Pedro. Thank you for having me here. And I hope our insights will be beneficial to the listeners. And we hope to work with you in the future.
Pedro Barata: Thanks.