Read the Full Video Transcript

Tom Keane: Hello everybody, this is Tom Keane, coming to you from Charleston and USC and UroToday. Today I'm delighted to say that we have a presentation from Dr. Eugene Cone from Harvard who is going to give us a talk. The title of this talk is "Lower Odds of Cardiac Events for Gonadotropin-Releasing Hormone Antagonists versus Agonists." Now, as you know, this has been an area that I've done a lot of work in. I am a firm believer that there are differences, and I think you will find the following presentation extremely enlightening. Dr. Cone, welcome to UroToday and thank you very much for taking the time to do this.

Eugene Cone: Absolutely. Thank you so much for having me on. So I am a urologic oncology fellow at the Harvard program between Mass General Hospital and the Brigham and Women's Hospital. As part of my research, one of the things that we were looking at is the medical effects of androgen deprivation therapy and if there were differences by the modality of androgen deprivation that was administered.

So I think as most of the listeners are aware, ADT is really the standard of care for advanced prostate cancer. Studies have generally found that there is increased cardiac risks associated with ADT, but the risk of newer agents and the risk of even the more traditional agents, especially in comparison to each other was somewhat under-studied. So we looked at more traditional gonadotropic releasing hormone, GnRH agonists, such as leuprolide, lupron, goserelin, triptorelin, and then compared them to GnRH antagonists, which until more recently, the only one on the market really was degarelix. Then, obviously, there's been some more interest recently in relugolix, which was published in the New England Journal of Medicine, but we didn't have data on that. So we limited our analysis primarily to degarelix. The database that we were using to analyze this is called VigiBase; it is a pharmacovigilance database administered by the World Health Organization, and it collects drug incident reports or safety reports from more than 130 countries worldwide.
You have representatives from every continent, and it categorizes all of the reports into various subtypes based off of organ system affected, severity, it does collect some basic clinico-demographic information, but unfortunately, there's not a ton of granularity to it. But it's very good from a pharmacovigilance standpoint to try to pick up if there's a disproportional signal, that there are more events than would be expected attributed to a given drug or medication.

So, we had to study up a little bit on pharmacovigilance research. It's widely used in the pharmacology world. Again, the FDA and the World Health Organization use these methodologies basically to detect if there are unusual signals or things that would require post-marketing warnings to be publicized, but it's not commonly used in urology, so we did have to do some due diligence on that. The main methodology that is used, that we wound up using as well, is called disproportionality analysis. So for a traditional study, looking at an adverse drug reaction, an ADR, you would look at people who received the medication and you would subcategorize them into people who did or did not experience a given side effect. Then you would look at people who did not receive the medication and look at the number who did or did not experience the given side effect in order to calculate a baseline rate. You would compare the two, and if the rate of the event is higher in people who took the medication than it is in the baseline, non-medication takers, then that would be a sign that the medication may be causing or increasing the risk of that event.

The problem with pharmacovigilance is that you don't have numbers or reliable data on the number of people who took a given drug and did not experience an event. The reason for that is the only way to get recorded into a pharmacovigilance database is if you had some kind of adverse drug reaction. So everyone in the database by definition had some kind of side effect. So to get around that and figure out what the baseline rate would be so that you know what you're comparing it to, you basically look at ... If we're looking in this case at, call it leuprolide, and we're saying, "Okay, how many patients with leuprolide had a cardiac event, and is that disproportionally high?" We look at the number of cardiac events recorded for leuprolide, we then look at the number of any other event recorded for leuprolide, so everything from fatigue, myalgias, the common headache complaint with any medication.

That that gives you the leuprolide rate and then you compare it to, for every other drug in the database, so everything from aspirin to chemo, how many of those patients had a cardiac event compared to any other event? So that serves as sort of the comparison for your baseline rate. You compare two and if the proportion reporting an event while taking a drug is higher than the proportion taking any other drug with the same event, that's the disproportionality signal and that's what the FDA or the World Health Organization or researchers will use as a signal that the given drug may be associated with a higher risk of that event.

There's a couple of ways to report this. You can report this using a Bayes estimator, which unless you're a statistician would take way too long to get into the weeds on, or you can use a reporting odds ratio, which is basically a relative of the odds ratio. So we used an empirical Bayes estimator as a sensitivity signal to say, you know, if a given drug had a significantly disproportional signal and the empirical Bayes estimator, then that was something we would take a closer look at and then we would calculate reporting odds ratios because they're easier to digest.

So when we looked at comparing GnRH antagonists — with degarelix — against agonists, such as leuprolide, goserelin, triptorelin, we were able to capture 877 cardiac events that were reported in VigiBase over the entirety of the collection period. We didn't limit it, it basically ran from the '90s up until present day. The most recent one that we accessed was in November of 2019. We had reports from all across the globe: the Americas, Europe, Australia, Africa, et cetera. Importantly, we did find that the majority of these events were reported outside of a clinical trial.

That's important in this space because cardiac events take a long time to occur, they're the culmination of a lifetime of abuse. So if you limit your observation period just to a clinical trial, most clinical trials aren't going to run for long enough to pick up the true increased risk of a cardiac event with a given drug. So about a third of our reports came from clinical trials, the rest did not. The majority of the events were limited to being suspected to be due specifically to the GnRH agonist or antagonist. In terms of the time to onset, the meantime from initiation of drug therapy until the cardiac event of interest was just under two years, it was about 550 days. The vast majority of these events were severe: 82% of them were severe and 20% of them actually wound up with death as the outcome.

Now, when we say cardiac events, we're talking about there's the general categorization of any kind of a cardiac event per VigiBase, but we also subcategorized into myocardial infarction, heart failure, any of the carditises, so endocarditis myocarditis, pericarditis. We looked at arrhythmias and we also looked at new onset valvular dysfunction.

So what we found when we were comparing antagonist therapy to agonist therapy, is we found that the reporting odds ratio for any cardiac event for GnRH agonist was 1.2 with a significant effect seen there. The confidence interval was pretty tight, 1.1 to 1.3, but there was no significant increase in cardiac events for patients taking GnRH antagonists. So there was a higher risk of a cardiac event if you were on leuprolide and no risk associated with degarelix.

When we looked a little bit closer, we saw that the risk of heart failure, the reporting odds ratio, was 2 for heart failure for GnRH agonists. It was 1.7 for MI and again, we didn't really see a significant effect when we looked at the GnRH antagonists. So really the take-home message here was that in our pharmacovigilance analysis, we saw a significantly higher incidence of cardiac events that were driven by myocardial infarction and heart failure that was associated with GnRH agonist therapy. That was in comparison to the baseline risk by all other drugs and all other reports in VigiBase, but we did not see that effect for GnRH antagonists. The meantime to the onset of these events was almost two years and the majority did not come in for clinical trials, so that may partially explain why this is not data that has been reported previously.

So, we think that a strength of this work is that again, it has a long timeframe. It looks outside of clinical trials and in addition to the fact that the timeframe is important, we also think it's important to be looking at non-clinical trial data because given relatively stringent enrollment criteria, clinical trial populations often do not really reflect the real-world use and population that may take medication. So for example, if you had increased cardiac risk factors, you had a prior MI, you're a smoker, et cetera, you may not have been eligible for a clinical trial examining the two, whereas, in the real world, we all know that those patients often receive these medications when they develop advanced prostate cancer.

So this is a real-world data source. It's absolutely huge. It's an absolutely huge database of these clinical safety reports and the fact that we were able to pick up this signal, we think is important. For me personally, what it's meant is that when I see a patient come into my clinic with advanced prostate cancer, and we're talking about referral to med-onc and we're talking about what to do next, and maybe they have a biochemical recurrence. If they're someone who is obese, smoker has either a personal history of MI or just generally has many cardiac risk factors, I think I'm much more likely to reach for a GnRH antagonist than an agonist.

Tom Keane: Excellent. I mean, that gets to the core of the issue. I mean, for a number of years, there's been a group of us talking about antagonists versus agonists and it takes large numbers to see significant changes. I mean, I remember the first time that I was at Emory, when [Inaudible] Sanders published an article on fractures using ADT, and we were almost laughed out of the place. "What are you talking about? There's no fractures and ADT." If you ask the urologists, they had 40 or 50 patients, well, if it's a 1% to 2%, or sometimes 3% risk, they're not necessarily going to see those fractures. Now today, everybody fully realizes the fracture risk and the risk of bone health when you're on any kind of ADT. I think this is a very similar situation, where we have known, we have seen cardiac events occurring with patients on ADT. We don't see it a whole lot in the trials and that's always brought up as well. "Why didn't we see it in the trial then?" Your explanation of real-world experience is exactly what I felt for a long time, that when you deal with the mass population, we're not pre-selecting patients to go into trials. This is real world. I think that is one of the main reasons that we look at this and that we are finding these changes.

I was also when we look at it, O'Farrell had a great observational database that was published, I think from Sweden at one stage of 80,000 patients, and that showed the same thing. So I do believe that there are very substantial differences between these two agents, between agonists versus antagonists. Now, I think the arrival of relugolix may make things somewhat easier for the antagonist's arm because, at the present time, it's degarelix has been the main agent and it is a 28-day injection, and it is a subcutaneous injection. There is a rate of injection site reactions, which I've heard a lot of people put off by that. Now, the fact that there's an oral agent that's hopefully going to be approved in the very near future, that's going to change the field completely in my opinion.

I did have one question for you, that when you said that most of these events occurred after one year, there was, I think it was Albertson's paper where he put together six different trials, which had large numbers of patients and looked for cardiovascular events. They showed that it occurred within one year and there was one other paper, but I think that's probably because of the design of these studies and going back, and they were taking patients from various trials and putting them in but I would understand perfectly when you say that we need to be looking for it after one year, because there was some information saying, "If it occurs in a year, then you're okay, after that ... " I don't believe that. I think your data shows very clearly that we have to be extremely vigilant and it's not just for a year, it needs to be over the years. Have you any data at further out as to what's the five-year risk? Did you get any information on that?

Eugene Cone: So, yeah, we don't have data specifically on what the five-year risk is, but we do have just a time series showing when these events occur. We certainly saw a very long tail trickling off and we saw events at four years, five years in one case, 10 years after administration of therapy. So, to your point, I think it really is about the real-world usage and the fact that, like Albertson's paper, as you said, it synthesized data from existing trials, many of them with their own inclusion criteria that are going to limit the time period that they're looking at.

Tom Keane: Yeah, yeah. Then it is also interesting, there was a paper from Grace Lu-Yao which I think was published in 2019 in European Urology, they actually looked for events occurring with the androgen receptor inhibitors. In other words, the newer agents, enzalutamide and abiraterone, et cetera ... well, I don't know if darolutamide was included, but apalutamide and other things. The main gist of her paper was, I think it was mainly enzalutamide and abiraterone, but we also see significant cardiac events in these patients also, which was very interesting to me.

So it basically shows us that as we treat ADT, I think hormonal therapy, in general, is a risk that needs to be assessed in terms of cardiovascular risk. But I do think that we need to be very careful with what agents we're giving when we're not alone in terms of the main driver of ADT, which will be an agonist versus an antagonist. I believe it should be an antagonist, but even as we move further along and start giving further treatments, as the disease progresses into castration-resistant status, then we need to be equally as vigilant in assessing cardiovascular risk in these patients. What's your opinion on that?

Eugene Cone: I think it's really interesting that you brought that up and we actually have some other work that's examining the cardiac risk associated with abiraterone and enzalutamide that preliminarily, I would say, tracks reasonably well with the data that you've quoted. And I would be delighted to come back and discuss that further in a future episode.