Evan Yu: Good morning. I'm here at ASCO in 2025 in Chicago. And I'm here today talking to Dr. Matt Siskin from New York University. And we're going to be talking about a really interesting abstract that he's going to be presenting at ASCO today. It's about androgen receptor ligand binding domain mutants and a couple, in particular, that have biologic implications and potentially therapeutic implications.

So Matt, why don't you tell us a little bit about the mutations that you looked at?

Matthew Siskin: Great. Well, thank you for having me. Excited to be here. So our project focused on AR T878A and AR L702H, which are both missense mutations in the ligand binding domain. We chose those two, because they're the most prevalent among the missense mutations in the ligand binding domain. So we thought they'd be the easiest to get sufficient patients to do a formal analysis.

Evan Yu: So tell me a little bit about these mutations. What implications do they actually have? Do they occur frequently?

Matthew Siskin: Yeah. So we've known about these mutations in the field for quite some time. Their prevalence in prostate cancer patients definitely varies by disease state. They're present, but at pretty low frequencies in hormone-sensitive patients. And they become much more common in castrate-resistant patients, particularly patients who have been exposed to second-generation ARPIs.

In that patient population, prior data has shown prevalence rates of about on the order of 20% to 25% depending on the specific study used, but importantly, that number encompasses all AR ligand binding domain mutations, and there are several that have been described.

As I mentioned already, the most common seem to be L702H and T878A, which together probably account for 80% to 90% of the mutations. And then there are rarer ones. They have important implications for biology that we know. But we don't have a tremendous amount of large-scale clinical data to say how these really impact clinical management as of yet.

Evan Yu: I see. I've heard something about these mutations and T878A. My understanding is it generally occurs after someone's been exposed to abiraterone. Is that right?

Matthew Siskin: Definitely. And that's one of the first settings in which it was found. And conversely, L702H more often arises after enzalutamide.

Evan Yu: OK. Do we know how they work biologically? Like are they still active? Do they confer a worse prognosis, worse outcome?

Matthew Siskin: Yeah.

Evan Yu: How does it work?

Matthew Siskin: There are some cell-based data that give us some inclinations into that question. Broadly speaking, AR ligand binding domain missense mutations work by either making the androgen receptor more, quote unquote, "promiscuous," more responsive to non-testosterone steroid hormones in the body, or by changing the way that the androgen receptor responds to medications.

So some of the first-generation AR antagonists, in particular, have been shown when certain AR ligand binding domain mutations come about, they actually transform from antagonists into agonists.

In regards to the specific two mutations we were interested in this study, it's thought that L702H makes the AR partially responsive to glucocorticoids, in addition to testosterone, and T878A is thought to confer some AR responsiveness to estradiol and progesterone.

Evan Yu: Is it like AR splice variants, like AR-V7? Do the patients generally— is it felt that they have a worse outcome or worse prognosis?

Matthew Siskin: Generally, it's thought that they have a worse outcome. It's a different biology than the AR splice variants. But it is broadly thought that way.

Evan Yu: OK, great. Thanks for that background. It's very, very important. All right, so tell us what you did in this study and what you looked at?

Matthew Siskin: Great. So what we did in this study was we used the GuardantINFORM database. We worked with partners over at Guardant as well as some other oncologists around the country. We used their database, which essentially combines genomic data that they've collected with health care claims data to provide the clinical information.

And we extracted out a matched cohort of patients. And we looked at T878A versus a population of patients with L702H. And we also compared T878A to AR amplification and to AR wild-type.

We chose these three comparisons because we really wanted to get a clear sense of how the T878A patients are behaving clinically, specifically patients being treated with enzalutamide. As you already mentioned, the T878A mutation was originally found to be arising in patients treated with abiraterone. So we had the clinical question of whether those patients would respond better to AR antagonists.

Evan Yu: OK. And so was that the case?

Matthew Siskin: Yeah, so the comparison between T878A and L702H shows a prolonged time on treatment and time to next treatment, which are two different variables, both broadly looking at how long the patients are staying on enzalutamide.

And the T878A patients were specifically significantly longer than the L702H patients. The OS data trended toward favoring T878A versus L702H when treating with enzalutamide, but didn't reach statistical significance.

Evan Yu: Do you think this will help us in the future? I know that we know that most patients that have progressed on one androgen receptor pathway inhibitor generally don't do as well on another one.

Matthew Siskin: Yeah.

Evan Yu: Do you think that this might help us select out whether we should be doing that strategy or not, or select patients that maybe might benefit from that strategy?

Matthew Siskin: Yeah, I think that's a great question. And that's certainly the hope. But I would still say that this work is hypothesis-generating. The population that we looked at matched patients by the line of therapy in which they received enzalutamide.

But it's important to recognize that this data set goes back several years, up to 10 years, if not slightly longer. So many of the patients were receiving enzalutamide after abiraterone, receiving multiple ARPIs, receiving single-agent GnRH in the hormone-sensitive setting.

So these patients weren't necessarily managed the way that standard of care today would manage patients. That being said, they were matched, and therefore, similar in terms of comparison. But I think that ideally we would be able to acquire that data in a cohort of patients that's being treated by modern standards.

Evan Yu: One thing I noticed, because I read through your abstract, is that you looked at a lot of patients, like over thousand-some patients.

Matthew Siskin: Yeah.

Evan Yu: But the mutation rates for T878A and L702H were maybe, like 50 plus patients each, which quick math shows it's around 1% or less.

Matthew Siskin: Right.

Evan Yu: I mean, you mentioned earlier traditional data showed maybe 15% to 20% in some of the ctDNA studies. Why do you think it was so much lower in your analysis?

Matthew Siskin: So I think that there's a couple of possibilities. One is the requirements that we put on our selection of patients. So we require patients to have multiple health care claims. And it's possible that some of the patients just fell off of the database.

Another important feature, I think, is that AR mutations often co-occur. You'll see T878A with other AR mutations. And in order to isolate specifically the significance of T878A, we restricted patients such that they had no other AR alterations. And in doing so, we may have eliminated a large swath of patients.

Evan Yu: I see. So it was the exclusion criteria that you had might have taken out some of those other patients.

Matthew Siskin: I think that's possible.

Evan Yu: OK. Also in the field, I'll say that there are a couple other categories of AR-targeting agents that are being developed at this point in time, namely androgen receptor degraders and CYP11A antagonists that do adrenal annihilation and take out all these other steroid hormones.

I mean, I think what you're saying and what I've read is that the principles of these AR ligand binding domain mutants is that you can get promiscuous stimulation from other steroid hormones, not just sex steroid hormones but any other steroid hormones.

So you think this has major implications for new drug development for prostate cancer patients in the future, specifically targeting patients with these mutations?

Matthew Siskin: Well, I think that the CYP11 data and the PROTAC data are very exciting. Still early phase, but definitely very promising.

One thing that I think will be important for the field as it moves forward is recognizing that all the AR ligand binding domain mutations may not be the same. So as we group these patients together for these analyses, it's possible, to use the example of CYP11 inhibitors, that not every AR binding mutation requires complete annihilation of the adrenal gland. Perhaps only certain steroid hormones are really responsible for that particular patient population's resistance.

So I think looking at subgroup analyses, as those agents move into larger and larger cohorts of the different types of mutations, will be important to understand which patients really benefit from these agents.

Evan Yu: Well, interesting. So I got another question—hopefully throw some curveballs at you. So L702H traditionally is felt to be stimulated agonistically by glucocorticoids.

Matthew Siskin: Yeah.

Evan Yu: A lot of our patients with metastatic castration-resistant prostate cancer, especially those that have had abiraterone, are on steroids, like prednisone. Of course, enzalutamide less so.

But I guess my question is, do you have a way to look at the data through these claims databases and see whether the patients were on steroids as well, and determine outcomes based on that? Because the whole idea is the patient is on steroids and has that mutation; maybe they're stimulating the prostate cancer that way.

Matthew Siskin: Right. It's an important question and not just for abiraterone patients, but also docetaxel and cabazitaxel often have significant steroids that are used in combination with those agents as well. This clinical data set, to my knowledge, doesn't contain that level of granular clinical data.

So actually, just within the last week, I got access to the raw data for the different lines of therapy that the patients in this cohort were given. But those data focus on the agents that were targeting the prostate cancer and don't include whether the patients did or didn't get steroids as part of their overall treatment regimen. So I think it will be an important question, but not one that we can answer with this particular data set.

Evan Yu: All right, so I'll just wrap up with what I think is always a very interesting question, because I think you've done some great work. I think what I want to know and I'm sure what our readers want to know, is what are you going to do next? Where are you going to take this work next? What's your interest here?

Matthew Siskin: Yeah, so we are trying to do a similar analysis in patients treated with abiraterone and enzalutamide to try to get a better sense of different AR-targeting agents. They work differently in these different populations, and I think we'll be able to do some interesting comparisons with that. I also think that getting answers to what specific lines of therapy all the patients are given in these cohorts is going to be really important for clinically interpreting the data.

So I'm hoping to expand this analysis. Ultimately, in the future, I am hoping this analysis can inform clinical drug development, as we've already talked about with novel agents. And I think there's exciting work to be done.

Evan Yu: That's great work that you've done. Really super interesting biology with these androgen receptor ligand binding domain mutants. And I'm really looking forward to seeing more work out of you and your group.

Matthew Siskin: Thank you. I appreciate the time.

Evan Yu: Thanks for coming today.