Neeraj Agarwal: Hello, welcome to UroToday. My name is Dr. Neeraj Agarwal. I'm a GU medical oncologist at the Huntsman Cancer Institute, University of Utah in Salt Lake City. Today, I have the pleasure of having Dr. Benjamin Lowentritt join us to discuss his most recent study on real-world experience of apalutamide treatment in patients with metastatic hormone-sensitive prostate cancer where he looked at the efficacy of apalutamide versus abiraterone. Just to introduce you, Ben, you are the director of the prostate cancer program at Chesapeake Urology Associates. So welcome.

Benjamin Lowentritt: Yes, thank you so much for having me. I'm really excited to be here.

So this was a really interesting effort where we followed the best evidence-based FDA guidelines for real-world evidence to look into a large dataset of urology patients, patients in community urology practices using a couple of different datasets actually. So we used the PPS dataset, which is an EMR extract from about 90 urology practices around the country, and then paired the data from those with a well-known claims-based database called the Komodo dataset. And that allowed us to get a broad picture of what was happening to these patients. So these were patients that went through a number of different criteria to really try to nail down that these were mCSPC patients, and then we followed and saw how they did over time. And this was overall survival. We've previously published on PSA response and PSA90, PSA undetectable using similar methods. But by pairing with the Komodo dataset, we really were able to get to overall survival.

And we had a prespecified data point of two years that we were looking at to see if there was any difference in survival between patients being put on apalutamide versus those that were put on abiraterone. There were about 2,000 patients in each group, so a really robust number of patients. And we used a technique called inverse probability of treatment weighting to adjust for any slight differences in demographics or clinical characteristics. And with a dataset this large, there's not a lot of adjustment that was needed because they were fairly similar, but that's a common method used to control for differences in a real-world dataset like this.

And essentially what we were able to show is that at that two-year point, there was a 26% reduction in the overall survival risk in favor of the patients receiving apalutamide compared to those getting abiraterone and prednisone. So it was following what we had already seen in some of the PSA response and other things that we had looked into in the past, but it was a really significant finding when you're only at two years. So just to be clear, and you can see this in the graph, the overall difference is a 26% risk reduction, but still you don't have a lot of survival events. You still see over 80% of patients in both groups are alive at two years, which is what you would expect with these therapies. But we do start to see the lines separate in a time frame where we know that there are patients that progress rapidly. So I think it was a very interesting result.

Neeraj Agarwal: Very interesting data indeed, Ben. First of all, this is a really large number of patients. We are talking about 4,000 patients, beyond the scope of any phase III trial in prostate cancer. Wouldn't you agree?

Benjamin Lowentritt: Yes. And I think that's where we should see value in real-world evidence, right? For so long, we were talking about real-world evidence as a concept, but it really was just a couple of institutions and a retrospective chart review and things like that. And this is leveraging the power of big data, computer-based EMRs and the ability to extract that data and find large numbers of patients across a broad network of sites. So I think that's exciting. Yeah.

Neeraj Agarwal: And you said 90 urology practices. This is a massive amount of data.

Benjamin Lowentritt: Yes, it is.

Neeraj Agarwal: First of all—

Benjamin Lowentritt: Sorry—

Neeraj Agarwal: Sorry to interrupt you, but I just wanted to ask you about the risk reduction. 26% reduction in risk of death with apalutamide versus abiraterone.

Benjamin Lowentritt: Yes.

Neeraj Agarwal: How do you explain that within two years of follow-up?

Benjamin Lowentritt: Well, once again, it was still overall over 80% of patients who were surviving in both groups. So it's that risk reduction that we're talking about. But I need to say very clearly, this was not designed to determine the cause of the difference. So anything here would just be purely speculative. But I think that you can imagine a number of different reasons why people might do better on one therapy over another. Tolerability may be one. There's always the possibility of dose reductions because of side effects. And are you able to keep people on a higher dose and therefore get a better response? We tend to do a lot of dose adjustments in these patients. They're older men typically, and there are a lot of reasons that we might need to dose adjust. But that's all just speculating as to why this may be. But in the end—

Neeraj Agarwal: It's there.

Benjamin Lowentritt: It's there, right?

Neeraj Agarwal: It's there.

Benjamin Lowentritt: The data are there, the result is there. So whatever the reason may be, this is something that we were able to find.

Neeraj Agarwal: So again, as we discussed, as you mentioned, these are real-world patients. Electronic medical record-extracted database. So the data out there are not really driven by the recollection bias of physicians. They're not reporting. The data are being extracted directly from the electronic medical record. And these are really large numbers of patients on both sides. 2,000 patients treated with apalutamide, 2,000 patients treated with abiraterone. And I thought it was quite remarkable that even within two years of follow-up, there was a 26% reduction in risk of death. So how do you match these patients to make sure that the underlying characteristics are similar in these two cohorts?

Benjamin Lowentritt: So that was a combination of both the claims and the EMR dataset. I think the EMR largely was to try to identify what had happened up to the point of starting therapy and then seeing that the patients were given therapy. And you can see this in the paper itself, but there was a lot of narrowing down of the patients to get to the cohort we actually were able to evaluate. And that was making sure that if there was any question whether they were previously treated and they were actually in the castration-resistant stage or something like that, any patient that may have had previous therapies were excluded. So we tried to exclude those with previous hormonal therapies other than in the setting of radiation. So we really tried to get to those patients that were newly diagnosed as mCSPC, not necessarily all synchronous patients, but patients that were truly newly diagnosed.

And then we were able to identify a lot of the clinical information: What's the site of metastasis? And that came both from extraction from records of imaging when available, but also from codes, the codes necessary to start a patient on therapy, whether it was lymph node, visceral or bone disease or some combination. And that came from both claims and from the EMR. But the main thing that the claims data were able to show us that we can never get from EMR is a date of death. And that's pretty reliable there, where it's very unreliable within the EMR itself. So the information on medications, especially the start date, is all there. The information on ADT is all there within both the electronic medical record and effectively the practice management or the billing data on the EMR side as well. So it's a very robust dataset. And then being able to get all of the demographic data from the EMR and match it to the claims data from Komodo was really powerful in this dataset.

Neeraj Agarwal: And as you said, there is no ambiguity about the endpoint of overall survival; a patient is either dead or alive. And that takes away so many biases, which can happen with determining who progressed earlier. Progression can be based on PSA, can be based on clinical or radiographic criteria. And overall survival is such an absolute yes or no. And I think that's why I like the overall survival endpoints in any analysis.

Benjamin Lowentritt: I agree completely, and I think it really opened eyes to be able to provide that data. And I think that's so powerful about this claims-based dataset, in addition to the EMR-based dataset, being able to have that firm endpoint. And as you said, in doing good FDA-directed real-world evidence, you have to find a prespecified endpoint. So that's why, even though the visual that I provided shows the lines continuing to separate beyond two years, that's not the endpoint that we're talking about. We had to prespecify a specific endpoint in order to compare the two populations, and that was at two years, which I think is a reasonable time point. In order to get those kinds of numbers and get to four or five years, which would be a wonderful study to be able to have as well, we just have to have more time because the treatments weren't really approved in the mCSPC space until 2019 or so. So in order to get adequate numbers of patients in there, we have to give it more time.

Neeraj Agarwal: So next, I was going to ask you about the next steps. So next steps will be longer follow-up of these patients who are being treated with apalutamide versus abiraterone, and looking at the directionality of the curves, it looks like they continue to remain separate and maybe we'll see wider separation down the line. But we look forward to seeing those data from you, and hopefully we'll be right here on UroToday discussing those data two years from now.

Benjamin Lowentritt: I hope so too. Thank you so much for having me.

Neeraj Agarwal: Thank you for taking the time.