Read the Full Video Transcript

Neeraj Agarwal: Welcome to The UroToday. Today we have two esteemed guests, Dr. Emmanuel Antonarakis and Dr. Justin Hwang. Dr. Antonarakis doesn't need any introduction. He is a professor of medicine in the Division of Hematology Oncology, and Dr. Justin Hwang is the assistant professor in the same Division of Hematology Oncology at the University of Minnesota.You both published a quite impactful paper in my view in the Journal Clinical Cancer Research titled "Structurally-Oriented Classification of FOXA1 Alterations—Identifying Prostate Cancer with Opposing Clinical Outcomes with Distinct Molecular and Immunologic Subtypes." So I'd like to discuss the findings of this paper with you.

So, Emmanuel, please give us a background for this study. How did it start, and how did you become interested in this topic?

Emmanuel Antonarakis: Thank you for the invitation, Neeraj. It's a pleasure to be here. And thank you for noticing the paper. I didn't know that anyone would read it.

But FOXA1 is a gene that I've been interested in for a long time. And one of the challenges with FOXA1 is it's a transcription factor. It enables the androgen receptor itself to do its job. So it is a pioneer factor that helps the androgen receptor.

And many of the next-generation sequencing companies do not include this gene on their panel. For example, Foundation Medicine does not have it. Guardant 360 does not have it. But when I came to Minnesota, we started working with Caris Life Sciences, and Caris includes this gene on their panel. So all of a sudden, I started to see mutations in FOXA1 in my patients that I was managing in the clinic, and I must admit, initially I didn't know much about it. I had not heard about it.

But then they kept popping up over and over again. And there were some very high-profile papers published a few years ago about this gene as a pioneer factor for androgen receptor, but no one had really shown the clinical significance of this. So we decided to partner with Justin Hwang to study this using the Caris Life Sciences database, which currently contains about 5,000 tissue sequencing results of DNA and RNA from localized and advanced prostate cancer patients.

You would be surprised, Neeraj, to hear that this gene is mutated in 15% of prostate cancers. So this is not a rare genetic alteration. The other thing, just as an aside, it's also mutated in about 5% of breast cancers, 5% of bladder cancers, 5% of salivary gland cancers. All of those are cancers that are driven partially by the androgen receptor.

So there are AR-positive breast cancers, AR-positive bladder cancers, and salivary gland cancers. So this is fascinating because in cancers where the androgen receptor can be a driver, then FOXA1 mutations also increase. So that was the background, and we decided to look at a large database that captures this gene to understand the biological implications but, more importantly, the clinical significance.

Neeraj Agarwal: So now, Justin, could you describe the Minnesota classification of FOXA1 variants? What are the different classes?

Justin Hwang: Again, I echo Emmanuel. Thank you for having me. I really do appreciate this opportunity to share our work. And I say, the Minnesota classification is honestly partially inspired by Emmanuel's thoughts in that we need to make some real, able clinical decisions when we see these alterations in patients, really. But there was so much literature out there—and we can't take all the credit for this—a lot of it is done by our predecessors that really have studied FOXA1 in so many ways.

So we really considered a lot of their knowledge up front. For example, we are very aware that there is a certain domain in the protein called the forkhead box domain that's extremely important for function, and obviously with today's technology, knowing that, for example, protein structure can deeply affect its function, we really oriented probably our classification scheme around changes both based on the type of mutations that are constantly observed in patients with prostate cancer but also thinking about the forkhead box domain. And then we use both these factors to orient ourselves toward what are the classifications.

So really for a clinician, I would argue that you can think about the type of mutation and then where it is with respect to the forkhead box domain. That's a very simple distillation of it, and I think the details you can definitely catch in our manuscript. I hope you guys read it as well.

Neeraj Agarwal: Thank you, Justin. So, Emmanuel, what are the key clinical observations of your study? Do FOXA1 alterations vary by race?

Emmanuel Antonarakis: So we'll get to the race question in a second because it's very interesting, but let's start with the overall conclusions. If one looks at FOXA1 alterations and does not separate into the different classes, all of the Kaplan-Meier curves look identical. It neither improves the prognosis nor makes the prognosis worse. So if you just did a very preliminary analysis, you might say this is not interesting at all, has no clinical relevance, but the fascinating part comes when you begin to take a deeper dive into these different classifications.

And what we found is that there were some types of FOXA1 mutations that increased androgen receptor addiction and hence increased androgen receptor sensitivity. The most important is what we call the class I B, which are in the forkhead domain that Justin mentioned, and mutations in this forkhead domain, the wing 2 region—which is what we call the class I B—those are associated with improved, better outcomes, particularly with hormone therapies and ARPIs, but not with taxane chemotherapies.

Now, on the opposite end of the spectrum, the class II alterations—there is a hotspot residue; it's the arginine-219 (R219). That residue is fascinating. First of all, it's a hotspot. So there's recurrent mutations there. But the most fascinating thing is this mutation, which we call class II, causes androgen insensitivity. So these patients have poor outcomes to ADT and ARPI, and they have a much higher chance of lineage plasticity and a neuroendocrine transformation.

So one gene, different types of mutations: on the one end of the spectrum, they increase androgen sensitivity; on the other end of the spectrum, the same gene, different mutation, causes lineage plasticity and small cell differentiation. So it's absolutely fascinating to me.

Now, there are major racial differences. We talked about the 15% prevalence of FOXA1 alterations overall. If you look at Asian Americans or Pacific Islanders, the prevalence is 25%—one quarter of Asian patients. One quarter is going to have an alteration in FOXA1, and by the way, most of them are the class I B, the one that enhances hormone sensitivity. So this might be one of the reasons that Asian Americans or Asian prostate cancer patients may actually do better with ADT and ARPI.

Now, on the other hand, African Americans—the prevalence was also increased, 20%. But it was increased by the wrong type, the bad type, class I C. So the African Americans, yes, they have a higher prevalence, but most of those are the poor prognosis classification, class I C.

Neeraj Agarwal: It's fascinating. So, Justin, one of the advantages of Caris Life Sciences' platform is the RNA sequence in addition to the DNA sequencing. So what did you discover from the transcriptome of FOXA1-mutated patients?

Justin Hwang: So as Emmanuel has alluded to, these single classes have very divergent outcomes. What the Caris platform allows us to do, because it has a DNA/RNA matched platform, is to really indicate, for example, are there differences in overall behavior of the tumors using the surrogates of RNA sequencing. And I do think RNA sequencing these days tells us a lot. The Caris platform—not that I would like to advertise them—they really capture almost every gene that we would have interest in, in terms of thinking about oncogenesis. We're allowed to study all of these hallmark pathways that might be associated with specific outcomes, and two of the ones that we may be interested in are, for example, epithelial-to-mesenchymal transition, which is a pathway that really drives poor prognosis.

On the other hand, we obviously know that androgen receptor activity is a big component of this. So based on the Caris platform, we were simply able to look at holistic differences between all the seven subclasses, which were indeed very different by class. That tells us by itself the subclasses that we hadn't annotated were indeed very different even amongst themselves in terms of their impact on the transcriptome.

And then further on, like Emmanuel said, class I B, which indicated more sensitivity to, for example, androgen therapies, there was an enrichment of, for example, androgen receptor signaling. On the other hand, the class II, which we see with very poor outcomes, was certainly more enriched—compared to every single one of the classes—in something like epithelial-to-mesenchymal transition. So this allowed us to really bridge the research of our thoughts and concepts that we have perpetually found in the laboratory setting and cell models, to really see this in actual patients. And I think in also connecting this with outcomes, it really gave us a more comprehensive understanding of what is going on in patients and how that might be associated with outcomes. So to me, it was a very exciting process that we got to really think about this altogether.

Neeraj Agarwal: Thank you. So, Emmanuel, what are your future research directions?

Emmanuel Antonarakis: So, of course, when you have a study, you want to validate that. We have the challenge that many of the other NGS companies don't include FOXA1, but Tempus includes it. So one initial step is to use the Tempus database to study this, and also in the Caris database, these were all tissue biopsies. As you may know, Tempus has a very large liquid biopsy ctDNA database as well, and we would like to use that database to validate the clinical findings and maybe some of the molecular findings as well.

The second thing is if there are investigators such as yourself or others who have their own in-house clinical genomic databases with a clinical annotation, we would love to collaborate. So one of the purposes of this interview, hopefully, is to attract collaborations with anyone who might be listening. And, of course, we will not take away any of your credit. You will get the credit for your own work, but we would love to help you. If you have internal databases that capture FOXA1 gene in prostate cancer or, by the way, if you might be a breast cancer researcher, this gene might be relevant in other cancer types.

And then the last thing is I'm extremely fascinated by this arginine-219, the lineage plasticity mutation. It's quite rare. It's only about 1% prevalence. But when it occurs, the prognosis is so terrible, and these patients do not always have histological signs of small cell or NEPC. Sometimes they just look like adenocarcinomas, but they are hiding this underlying genetic variant that can lead to transdifferentiation.

So I would love to form a consortium of sites, maybe put together 30 patients or 50 patients that have this lineage plasticity FOXA1 mutation, and deeply characterize their clinical outcomes. So those are the three avenues that I would like to take in the next 12 months.

Neeraj Agarwal: Those are the clinical directions you are going to take. And by the way, I looked into our database. We already have found 50 patients with very deep clinical annotation, and I'm really hoping to be able to collaborate with you and Justin in this regard. And I'm really hoping many more people will come up and share their databases. So very good point, which will actually impact patient care down the line.

So, Justin, before we wrap up, what are your future laboratory directions in this regard?

Justin Hwang: Yeah, that's a wonderful question because we have thought about this in so many ways. And honestly, some of my initial entry into prostate cancer has been related to understanding FOXA1 function. At the time, we didn't know exactly the differences in outcomes, but all of this data gives me confidence to actually really focus on specific variants that may impact, for example, plasticity.

And nowadays we have—even at the University of Minnesota—we have all this technology where we can do, for example, proteomics, phosphoproteomics, and we can study this with respect to each variant, in addition to whether it has the ability, for example, to cause worse outcomes for more tumors that are neuroendocrine-like. So I'd like to almost reopen the dialogue in the laboratory setting as well, to think about all these impacts that specific variants may have in the same setting, especially using some of the exciting CRISPR knock-in technology that we have today. I feel like we can get a much better answer. And obviously, I think this hopefully mirrors Emmanuel's results that'll help us all understand this phenomenon altogether so that we have a joint sort of laboratory finding as well as a sort of clinical impact as well.

Neeraj Agarwal: Thank you very much. So just to summarize for our viewers, Dr. Emmanuel Antonarakis and Dr. Justin Hwang from the University of Minnesota just reported the presence of FOXA1 mutations in patients with advanced prostate cancer. And it's fascinating to see that it is present in about 15% of patients. In some races, it is even more—25% in Asian patients, 20% in Black patients. And a more fascinating aspect, in my view, is that depending upon the type of mutations, these patients may have different prognosis. And down the line, I'm expecting you to tell us how they are associated with response to therapy, such as ARPIs.

Congratulations again for this outstanding work and for publishing this work in the Clinical Cancer Research. We look forward to seeing more data in this regard from you both so that we can improve the practice and lives of our patients. Thank you very much.

Emmanuel Antonarakis: Neeraj, thanks very much for the very kind invitation. It was a pleasure to do it with you.

Justin Hwang: Thank you for giving me the opportunity to share as well. This is great.