Read the Full Video Transcript

Oliver Sartor: Hi, I am Dr. Oliver Sartor. I'm here with you today, and we have three special guests today: Russ Pachynski from Washington University in St. Louis, Chris Maher from Washington University in St. Louis, and Dr. Aadel Chaudhuri from Mayo Clinic, but ex-Washington University in St. Louis. And we're going to be talking today about a new potential assay that could serve as a biomarker for those being treated with advanced prostate cancer. And I'm not exactly sure who's going to do the presentation, but let's take it away with the presentation. Then we'll come back and ask some questions.

Aadel Chaudhuri: I'm going to take you through our study. We started with 129 mCRPC patients enrolled from three different institutions, and we began by validating the findings from our previous paper that we published in JCO Precision Oncology in 2020, showing again that alterations within the AR gene, including the upstream enhancer, are highly prognostic for both progression-free survival and overall survival in these patients who are being treated with AR-directed therapy. And we extended the findings from the previous study to the pre-treatment setting in 63 patients and then validated them in the on-treatment setting, where now we're seeing a more significant result with a higher hazard ratio, again showing that alterations within AR, including the upstream enhancer, are highly prognostic in this patient population.

We then wondered if we could learn more about the underlying biology of high-risk lethal mCRPC by studying the epigenomics in these patients with resistance to AR-directed therapy. And so we extended the analysis, going beyond the targeted sequencing of the AR region, including the upstream enhancer, and extending that to the entire genome, performing whole-genome sequencing as well as methylation sequencing in order to look for epigenomic features that are associated with increased lethality.

And we started by doing nucleosome profiling, so we can infer essentially almost a cell-free version of ATAC-seq. We can infer which regions of the genome are active versus not active by looking at where the nucleosomes open up and allow for transcription factors to bind their transcription factor binding sites. And I'm showing an example here of HOXB13, which has decreased coverage for patients with AR/enhancer-altered lethal disease but doesn't have that decreased coverage in the patients who don't have alterations within the AR locus. And indeed, I'm showing this here in another way, where you're seeing decreased normalized coverage in the patients that are AR/enhancer-altered lethal compared to those that are AR/enhancer wild type.

When we do this form of analysis, we did it across 10,000 transcription factor binding sites across the genome. We found that the top 20 transcription factors with binding sites that were most accessible in high-risk mCRPC, when we performed gene set enrichment analysis, we strikingly found signatures that were related to development and stemness. We saw IPS cells. We saw gastrulation. We saw stem cell pathways. We saw embryo. We saw all of these things that we didn't necessarily expect to see, but it just dropped out of the analysis, these signatures related to development and stemness.

We wanted to query this further, so what we did is we acquired single-cell RNA sequencing data from mCRPC that had been published earlier by He and colleagues in Nature Medicine, and we applied CytoTRACE, which is a nice algorithm that was published by Gunsagar Gulati and colleagues in Science in 2020, and we identified a stemness signature from the single-cell RNA sequencing data. And then we developed novel methodology to apply that stemness signature to plasma cell-free DNA methylation sequencing data, and we identified a stemness signature, and we found that this stemness signature was strikingly associated with survival outcomes.

So looking here within our cell-free DNA cohort, patients that had an elevated stemness, which we were able to infer from the plasma methylation sequencing data, these patients had worse progression-free survival outcomes and trending worse overall survival outcomes. And importantly, when we tested the same signature in a completely held-out cohort Abida and colleagues published in PNAS in 2019, these patients were profiled by tumor tissue RNA sequencing. Again, though, despite the differences in the modalities, the difference in cell-free versus tissue, the patients with elevated stemness had worse survival outcomes. So that's pretty much it.

Oliver Sartor: Gosh, well, thank you so much for hearing us through that recent publication. And there are a couple of interesting points that I'd like to amplify on. And first of all, I might go to the clinician. Russ Pachynski is with us today. Now, Russ, when I was actually looking at that AR/enhancer for on-study, I was looking at hazard ratios of 15, which is pretty phenomenal. Do you believe this could have clinical utility, first of all?

Russ Pachynski: Yeah, yeah, absolutely. In our first study that we published in JCO Precision Oncology a few years ago, we actually took a subset of the cohort, and we sent off the CTC AR-V7. At that time it was Genomic Health, now it's Epic Sciences, on oftentimes the same plasma sample, but very frequently within a very short timeframe of the sample, so it was temporally contemporaneous, I should say. And we did a subset analysis and just showed that looking at the enhancer in our assay, EnhanceAR-Seq, was much more sensitive and specific.

Of course, there are issues with the CTC-based assay. First of all, you've got to get the CTC. You have to look at AR-V7, which is only one variant, and it has to be located in the nucleus. And so I think that was a first-generation liquid biopsy that we had for prostate cancer. We feel like this is now a second, maybe even third generation of a ctDNA-based assay that you can use. And these patients that are more advanced typically have higher tumor burden. I don't think we had any failures of capturing ctDNA at all, and so the assay hopefully will be more clinically useful than the CTC-based assay.

Oliver Sartor: Yeah, I would think so. Circulating tumor DNA is pretty accessible today. Chris, I wonder if you could maybe talk about the assay per se. Are you looking at an amplification? Are you looking at any methylation changes? Tell me a little bit about the assay per se. And then importantly, is this going to make its way into a commercially viable product? And if so, then I think it can broaden out and be accessible to many patients nationally. So first of all about the assay, and then about the commercial applicability.

Chris Maher: Yeah. So this assay is based on some of our early tissue-based studies, which focused on 100 metastatic castration-resistant prostate cancer patients. And we designed this targeted panel to essentially capture all of the most commonly altered genes in this highly aggressive setting. But most notably, one of the most recurrent events that we unexpectedly found associated with treatment response was not just genetic alterations and amplifications of the androgen receptor gene body, but there's a regulatory region about 600 kilobases upstream, an area where we normally don't monitor, that actually can drive the expression of androgen receptor. And so our panel is uniquely positioned to capture AR and this essentially underrepresented non-coding region. We've also established the informatics software to process all of this data, so it's highly reproducible, which is important for assay development and really deploying this.

Which brings me to your other question about the clinical and commercial applicability. The idea that we have essentially honed down and we have a highly sensitive way and reproducible method for monitoring these critical alterations throughout patient care is extremely important, and again, with the bioinformatics software too, highly reproducible, such that when clinicians are using this, you can get highly reliable and sensitive results in real time as you're monitoring patients.

Oliver Sartor: So one more follow-up question. So we've got the androgen receptor itself, and then we have the enhancer that's going to be upstream. In my imagination, I would see that many times these would be co-amplified. Are these co-amplified events? Are these independent events? Do you find one without the other? How's the relationship between AR amplification and enhancer amplification?

Chris Maher: Yeah, that's a great question. So in all of the patients that we've looked at, we've seen a variety of scenarios. We've seen some patients where they have them co-amplified. We see some patients where they have just the androgen receptor gene body. But noticeably, about 10% of patients actually have an amplification of this enhancer, which up until recently had been previously overlooked. So you really need to monitor the whole region between the gene body and the enhancer. But what's also interesting is there are some patients that have multiple events, where you might actually see both the enhancer and gene body amplified and then a subsequent amplification of even just the enhancer region. So there's a strong pressure for this region, both in combination or even individual pieces, to be amplified and drive overexpression.

Oliver Sartor: That's interesting. Now, I didn't see any actual protein data. In other studies, have you been able to relate the enhancer amplification to protein overexpression, like we've seen with amplification of the AR itself?

Chris Maher: So we didn't look at protein expression, but for some of our earlier work, we had matched RNA-seq expression, so we knew exactly what type of amplification they had in their androgen receptor gene body or the enhancer compared to patients that lacked any genetic event controlling androgen receptor. And then we looked at each individual patient and their mutation status and how that affected androgen receptor expression, and we saw a high level of concordance between the patients that had an amplification of the enhancer or the gene body with elevated expression of androgen receptor, and it was much higher than patients that lacked these alterations.

Oliver Sartor: Interesting. And that's exactly what I would have expected, but I was just trying to carry it through. Now, Aadel, you then did the confirmation on the AR front and then pivoted into this stemness, which, to my knowledge, is really quite novel. I may not be totally familiar with the literature here, but you're being able to look at basically the ability for a transcription factor to bind by the openness on your DNA through the methylomic signatures and then infer what genes might be activated. Now, that's pretty interesting, because the signature that you've come up with had not previously been described. At least, if it has, I'm not aware of it. Can you talk about the originality of this observation and how it might compare to others? What about other cancers? Stem cells are presumably present in many cancers, but we don't really know how to assay them very well.

Aadel Chaudhuri: That's a great question, Oliver. We didn't intend to be going down this stem cell route in this project. We openly queried the genome using this emerging branch of -omics, which is cell-free DNA epigenomics by querying methylation, and also fragmentomics, looking at the nucleosome positions and trying to infer which areas of the genome are on versus off, essentially by looking with high granularity at coverage maps, which parts of the genome have lower coverage, which we can infer that the nucleosomes are opening up, letting a transcription factor sit down. And because they're opening up, that area of the genome gets degraded in cell-free DNA, and so we see a dip in the central coverage versus areas that have nice, deep coverage where the nucleosomes are really right next to one another and protecting the DNA from nucleases. So we really went into this with an open book, trying to query the mechanism underlying what Chris really described really nicely, stemming back to his earlier work that he had published in Cell that we had really gone down this whole pathway on for this project.

And what we found was... We performed this nucleosome positioning analysis in a completely unbiased fashion, looking at 10,000 transcription factor binding sites. And really what dropped out of the data were the stemness pathways, as I described. And we were able to then really work that up specifically by applying the CytoTRACE algorithm, translating that technology to the cell-free DNA space with some technology development within our labs, and then showing that, in addition to cell-free DNA space, this stemness signature can also be applied to standard tumor tissue transcriptomic data. So we can apply this to bulk RNA sequencing data from a completely separate cohort and see the same exact result, that these patients that are stemness-high, that we can infer from the -omics data, that these patients have worse survival outcomes.

Oliver Sartor: Yeah, it's really fascinating to be able to see this, to envision in real time following patients sequentially, being able to see the emergence of these patterns, which have a high degree of prognostic importance.

Now, as a clinician, the next thing that I want to do is to interfere with the process. I don't like stemness factors in that patient's cancer cell. I want to figure out how to kill it. I don't know what's on the cell surface. I don't know if I have to send in a little nuclear bomb and blow it up, or maybe I'd be able to find some sort of Wnt signaling pathway that I might interfere with. This is a very speculative question. Any concepts about how we might interfere with this process in a clinically relevant way, in addition to just monitoring as it occurs? That's a big question, and if you know the answer, it'd be fabulous, but I doubt if you know the answer.

Aadel Chaudhuri: I wish we knew the answer, but I think I speak for all of us that we unfortunately don't know the answer now, but the data we have here lets us start to ask the question. Indeed, we would love to find these targets that... What is the Achilles' heel or Achilles' heels for stemness that we could then target and prevent the cancer cell from going down this lethality pathway? Presumably these exist. Can we identify cell surface markers, or can we identify other potential markers that are druggable, whether by radionuclides or antibody-drug conjugates, in order to arrest the cell from going down this stemness pathway, letting them be in a less lethal state, and allowing a clinician to more effectively treat this patient and let this patient achieve a longer survival interval, more like some other metastatic prostate cancer patients, versus going down this highly lethal pathway where we're seeing such short survivals?

Russ Pachynski: And I think, Oliver, to your point, I think using this approach allows us to do serial monitoring easily over time, whereas before we're using old plasma markers that aren't great, or doing solid-tissue biopsies, which over time are very hard to do. So I think utilizing this tool and then bringing in the therapeutic slant, I think is... Sounds pretty promising.

Oliver Sartor: No, I agree completely. Guys, I'm going to have to wrap it up here in just a second, but just as a commentary, one of the things I really like here is, in my own mind, we've been stuck between this adeno and neuroendocrine cancers divide for a long time, and I think what you've done is introduce a whole new level of conversation around stemness that we've been titillated by over the decades but never really in a demonstrable way like this. Here you have an assay that can actually pick out that stemness. Once you can measure something, then you can begin to not only monitor, but potentially interfere. And then if once you can interfere, maybe something therapeutic can come out. So a lot of provocative work here.

I'm going to need to wrap it up, but I'm going to say thank you to Russ, to Chris, to Aadel for being here today. I think this is an important new working concept that you've presented, and I look forward to the next iteration and being able to monitor patients going forward, and most of all to maybe being able to fix this problem and fix some of the patients that we see dying of prostate cancer. So again, thank you for being here.

Russ Pachynski: Great. Thanks, Oliver.

Chris Maher: Thank you.

Aadel Chaudhuri: Thank you, Oliver. Appreciate it.