Read the Full Video Transcript

Oliver Sartor: Hi, I'm Oliver Sartor. I'm with UroToday and ready to welcome Andy Armstrong from Duke. Andy has a variety of titles, a professor of many departments. Congratulations, Andy. And also, Director of the Urologic Research at the Duke Cancer Institute, so welcome, Andy.

Andrew Armstrong: Thanks, Oliver. It's great to be with you today.

Oliver Sartor: You had a publication recently that really did catch my eye and it was a beautiful integration between the genomics and the clinical factors that feed into prognosis. And I wonder, first of all, if you might tell us a little bit about the background of the study, how did you get involved? And then number two, we'll begin to discuss what the implications of the study might be.

Andrew Armstrong: Sure, thanks for that question. Just to back up, in terms of long history of prostate cancer, as a medical oncologist when you and I see patients with metastatic hormone resistant or castration resistant prostate cancer, there's a wide range of patient outcomes that we see. We all know that patients with very high PSAs, liver metastases, pain, rapid progression, poor performance status, have different prognoses as compared to those that lack that. And that really comes from Susan Halabi's seminal work now over 20 years old, where she developed prognostic models using clinical phenotypes or clinical features.

But we also know that tumors also come in many shapes and sizes based on their genetics. And to date, there hasn't been a combination of clinical features and genetic features and kind of a holistic prognostic model that's better able to capture the patient experience and survival, so we leveraged a large phase three clinical trial. It's called the Alliance, which is a national cooperative group trial, AO31201 trial. This tested Enzalutamide plus or minus Abiraterone. It was a negative study. It showed basically that two RPs was not better than one so it did not change practice.

But that wasn't a problem for the purposes of a prognostic model. When you're trying to predict the natural history of patients, we're able to combine both arms and look at over 1,000 patients using plasma sequencing. And so this work was really a big team project led by Scott Dehm doing all the sequencing work, in which we published earlier this year in Nature Communications, the development of this assay, which we called the ARCT Detect. I'll go into that in a minute. And then Susan Halabi, our statistician, who really combined the clinical data and the genetic data to be able to predict survival.

Oliver Sartor: Well, that's pretty important. And one of the things that I've complained about over the years is that our typical ctDNA, whether or not be foundation or a variety of other assays like Guardant don't really focus on prostate and they don't really give credence to the importance of AR and some of the parameters like the upstream elements that help to drive the energy receptor, the enhancer. So I'm really pleased to see this new assay, which I think adds a little depth and breadth to what we ordinarily might be able to see. But don't let me comment, I want you to comment, so tell us about this assay in a little more detail, please.

Andrew Armstrong: Yeah, I'm glad you pointed that out. That is one of the motivations for developing our own CLIA approved assay by taking a deeper dive around the AR with Scott Dehm, who actually was one of the key discoverers of these AR genomic structural rearrangements and AR splice variants. We've also validated splice variants like ARV7 as being an important mediator of ARPI resistance. And the question would be, can a deeper dive into AR add to the prognostic significance? Can it predict ARPI success and survival? And in the clinical genetic model, we show that knowledge of all these genetic events adds to the prognostic significance of predicting survival by about 6%, which is in statistical language, a humongous improvement.

Susan Halabi has been trying to work on improving her model for about 20 years and this is the first time that she has topped that herself. I would say that the AR itself was one of the most important drivers. The AR enhancer, which is upstream of the AR promoter and gene body was one of the most important. When the AR enhancer is gained and the gene body is gained, that's one of the more important drivers of poor outcomes. We see genomic structural rearrangements being important, particularly at progression.

And we have a second paper that we've submitted to look at that question post-Enzalutamide and Abiraterone. A number of non-AR factors are also important in driving primary resistance. As you know, P53 and MEK, RB, PTEN a constellation of other genetic events are in this panel and they're all prostate specific. So they combine to give you a sense of how your patient's going to do in terms of survival. With this kind of paper, we would love for commercial vendors to read it and to think about integrating a deep AR sequencing much like Scott's lab has done to improve upon their assays, make them a little more prostate specific.

Oliver Sartor: Well, it certainly seems to add value. We're going to be coming out of that alliance trial, metastatic CRPC with Enzalutamide, plus/minus Abi, not much distinction between the arms. You're going to be doing the new assay, so I wonder if you might be able to give a brief summary of what the key findings in the prognostic paper actually were?

Andrew Armstrong: Sure. When you're judging a prognostic model, there's a statistical variable called the area under the curve, and that tells you how good you are at predicting survival. 50% is like flipping a coin, 100% means you're perfect. Our best clinical models are performing at about 72% and we're now able to push it up to 77% with the addition of all the information we can get from cell-free plasma DNA. The plasma DNA fraction is very important. Much like circulating tumor cell enumeration, it's a surrogate of tumor burden. A number of our patients have zero cell-free DNA, for example, even though they have metastatic disease, and that's really important prognostically. But for those patients where ctDNA is detectable, we can infer prognosis by mashing up all these different things into a model that's now readily available to estimate survival.

And Susan created three or four risk groups, she has both in the paper, that allow you to kind of communicate to your patient how they might do. And the prognosis differs from basically under a year for a poor risk group to more than five years for a good risk group and that's a big difference, I mean in all in the first line MCRPC setting. Now, there's limitations because you and I, Oliver, we treat our patients with ARPIs now in the hormone sensitive setting as do most academic and GU medical oncologists. But I think this model is still relevant for patients who have not had an ARPI in that earlier setting. We're now thinking of expanding it to earlier settings or developing a post-ARPI prognostic model. So those are different contexts where this model wouldn't be directly applicable.

Oliver Sartor: Yeah, some of the things you picked up, we've known about, RB loss, MEK amplification, AR amplification, the enhancer. We're going to come back to the enhancer just for a second, Pten loss, et cetera. Let's actually come to this enhancer. Some people know about it, some people don't.

Andrew Armstrong: That's right.

Oliver Sartor: Explain what it is and how it sort of ranks in the prognostic importance scheme, if you will.

Andrew Armstrong: Enhancers are parts of a gene that are usually way upstream, sometimes even on different chromosomes that can interact and regulate the gene expression of the gene they're paired with. The AR enhancer has now been very well described, including by yourself, Felix Feng, and many colleagues from the Stand Up to Cancer team as being independently amplified sometimes without the AR itself being amplified. And it basically regulates the transcriptional rate of AR. So when the enhancer is amplified, AR transcription rate is higher, then it drives over-expression of AR, which makes it harder for our AR ligand binding domain inhibitors to be successful.

So it kind of makes sense that if you have way too much AR, you're going to have relative AR therapy resistance. Scott has also shown that when you have structural rearrangements along with AR amplification in our nature communications paper, that infers that the AR may have left the nucleus and is on these extra chromosomal DNA locations. So a weird thing about prostate cancer along with other cancers is that the oncogene can be excluded from nuclear DNA, normal chromosomal DNA and be on these extra chromosomal circular loops like double minutes.

And that allows for super high levels of AR and allows the cancer to survive load levels of testosterone. So our model really shows the importance of those features, which would not be captured by standard assays like Foundation or Tempus or Keras or Guardant. So our hope is that by extending the sequencing to include the enhancer, but also to cover the introns where a lot of these genomic structural rearrangements encompass, you'll be able to really capture not just the enhancer, the gene body, but also all the genomic rearrangements in between.

Oliver Sartor: Those genomic rearrangements were unknown to me until, and I think it was Scott Dehm who pointed them out, I don't know if he was the first to observe them, but he was certainly the first that made me aware of not only their presence, but their importance.

Andrew Armstrong: Absolutely. And others have now shown that when you have this extra chromosomal DNA AR gain, it really makes the tumors exquisitely resistant to our conventional hormonal therapy. I think for future drug development, this could be helpful. You have a biomarker of ARP resistance, maybe you have a new drug that is selectively active when you have these AR amplifications or extra chromosomal DNA, or we have drugs that can overcome some of the other poor prognostic features like BRCA2 mutations, for example.

Oliver Sartor: I'm a little curious, Andy, and perhaps this is off-topic so forgive me if it is, but the use as the Hopkins people help to originate the bipolar androgen therapy and you get a significant down regulation, will that have a potential role in these individuals with a high AR expression? And I just ask that because it comes to mind, not because it's that clinically relevant in most practices.

Andrew Armstrong: I've been impressed by the bipolar androgen therapy working in selected patients, but I have not been able to identify who's going to benefit from it ahead of time. And having an assay like this could enable that. So far what I've seen is that perhaps bipolar therapy works better in patients with DNA repair defects, but AR splice variants has not been predictive of bipolar therapy. It's possible that having knowledge of the AR enhancer, the AR gain, the structural rearrangements will give a little more information to identify who could benefit from that. It could be more complicated than that.

The nice thing about this assay, which we call ARCT Detect, is it includes all of the genes that we know about that are relevant to prostate cancer. I'm sure we're missing some. We had this as a grant funded by an NIH award that was started about three years ago. And so when we developed the panel, in the past three years, there have been a couple more genes added that have been important. There's a lot of missing information in the panel like epigenetic signatures or acetalomics or metabolomics, proteomics, a number of other important phenotypes that we don't capture just with DNA. But this is a good first step to show that the information encoded in the tumor DNA does add to our clinical features.

Oliver Sartor: It certainly does and you have helped to demonstrate that. And you've made many contributions, this is your latest contribution. I'm sure you'll have many more to come. We're about to wrap up but before we do, I wonder if you might have any final words or thoughts on the topic for our listeners?

Andrew Armstrong: Yeah, I think our next work that we've submitted for publication is asking the question, how does this assay change? What are the genetic events that happen, not just at baseline to predict the future, but when a patient progresses? And we think that will be much more relevant where these genomic structural rearrangements, AR enhancers, eCDNA emerge. It's going to be under that evolutionary selection pressure that the ARPIs confer, regardless of which ARPI it is. And stay tuned for more information on that progression model.

Oliver Sartor: I look forward to learning more, Andy, and thank you for teaching myself and the field about what's important in prostate cancer. Thank you for being here.

Andrew Armstrong: Thank you, Oliver.