Pietro Scilipoti: Thank you very much.
Leslie Ballas: Tell me a little bit about this abstract.
Pietro Scilipoti: So yes, it's a study that we have been contacting together with my senior author, Thomas Sen from actually a period where I was doing my research fellowship and my surgical fellowship in Paris. And we investigated actually which intermediate clinical endpoints can mediate overall survival. So we have had a lot of trials lately because it's one of the field that is the most exciting in bladder cancer right now, especially muscle invasive bladder cancer.
And what we wanted to assess was whether EFS was a good surrogate for OS and whether PCR was a good surrogate for OS. It's quite complex when we discuss about surrogacy. It must not be confused with prognostic because it's a totally different thing. When we talk about surrogacy, we need to satisfy basically certain criteria. And if these criteria are satisfied, then the FDA we can use. We know that in terms of approval, we can use these endpoints to substitute OS, which makes the trials less expensive, and we know that we are spending less and in the end we have this intermediate endpoint palliated for OS and we can interrupt the trial before and approve the drug faster.
So we assessed both endpoints and we found that when we look at first and second step of surrogacy, EFS was a good surrogate and PCR was not a good surrogate. If I were to go a little bit in the details of the numbers, we had a total of 23 trials that we selected up to the last trial, which was the trials on EVP just published in cisplatin-ineligible patients in New England Journal of Medicine and among these trials 12 had also data on PCR.
And what we found was that the R-squared of the correlation between the treatment effect on OS and treatment effect on EFS was quite high. It was 0.87, which means there is quite well above the 0.7 market that is accepted to say this is a good surrogate in second step analysis. And the ST was 079, which means that to mediate OS in a trial, you should predict that you have at least 80%, we can say so 20% reduction in the risk of EFS in a clinical trial that we call positive.
Instead, when we look at the PCR, what we found was the R-squared was 0.42, which means that there is not a good surrogate. And we did some also sensitivity analysis finding always the same results. So in general, what we concluded is that when we plan a trial, PCR is not a good surrogate. EFS, it's a good surrogate. So PCR should not be used as a standalone when we are planning this trial to have approval from the FDA.
Leslie Ballas: And EFS is defined ... What are the events that define it?
Pietro Scilipoti: There is precise definitions given by the FDA. Basically it includes, it's used when we have perioperative trials because the event can occur even before the radical cystectomy. For example, an event can be, as we know, death or progression or recurrence following the surgery, but even the progression after the neoadjuvant therapy before radical cystectomy is an event. So the impossibility to undergo radical cystectomy is an event.
And of course, this is a limitation of this study because trials that we included, they begin in 2003 when we have the GROSSMAN trial, which is the biggest one back then. And then the outcome was a little bit different and so you didn't have these events. Now it's like quite homogeneous in the last three years, we have the trials using the same event. So of course it's quite heterogeneous and we have to use an umbrella definition at the end of the day. There is EFS based on the definition of the trial because IPD was unfortunately not available.
Leslie Ballas: And this surrogate endpoint of EFS is for perioperative trials as opposed to TMT trials in muscle invasive bladder cancer.
Pietro Scilipoti: Yeah, exactly. We did not include any TMT trials or bladder-sparing trials first because of course we didn't want to include too much heterogeneity and it might be done when we have more trials, but of course to fetal regression to perform this analysis, just for example, in TMT trials, you would need at least four or five trials to make some conclusions.
So of course this is only for trials where cystectomy is planned and this is exactly another point because of course in all these trials, you have also to account for attrition bias. There is trials such as the NEO-BLADE trial or we had the SWOG trial in which the attrition was well above 20%, 30% and which means that of course when we look at the PCR, there we had to use ETT analysis, of course, and the proportion of CR is calculated on the overall population. So there is some sort of bias as well for this.
Leslie Ballas: Yes. Well, congratulations on this abstract.
Pietro Scilipoti: Thank you very much.
Leslie Ballas: We look forward to reading the paper.
Pietro Scilipoti: Yeah, hopefully soon.