Matthew Cooperberg: Hi, I'm Matt Cooperberg. Welcome to another installment of our live Localized Prostate Cancer Center of Excellence interviews for UroToday. I am pleased to be joined now by Chris Barbieri who's an associate professor at Cornell and has been one of the major drivers of our understanding of the genetics and genomics of prostate cancer and how we might dig deeper into the molecular biology of the cancer to really better individualize and personalize treatments for our patients. Chris, welcome.

Christopher Barbieri: Thanks. It's a pleasure to be here.

Matthew Cooperberg: Yeah. Thanks for coming. Tell us a little bit about your perspective on where we are and where we're going, maybe a little bit of where we've been, for prostate cancer, genetics and genomics. Both in terms of clinical practice today, but also where we're heading.

Christopher Barbieri: Yeah, of course. As you know, it's only fairly recently we've come to the conclusion that prostate cancer is not a single homogeneous disease, but it's can be considered a collection of molecular subtypes, basically. And that it's identifiable by underlying molecular and genomic features that separate these subtypes and they may act a little differently. And so I think I'm tremendously interested in continuing to define what makes those things different. How do we use that for patient benefit? And more than that, how do we drill down on the why of why they look different and how you can leverage that for how we can treat patients differently and really more of a precision medicine approach for different patients with different kinds of diseases.

Matthew Cooperberg: Tell us a little bit about the difference. Well, the markers we have today and the extent to which they're prognostic versus how are we doing in terms of subtyping and actually using those subtypes to develop actually predictive tests that can help us choose treatments and not just tell which cancers are going to do worse?

Christopher Barbieri: No, absolutely. It's a critical question. So like you said, prognostic markers of which cancers are more aggressive or not, are valuable, but only to a certain degree, because at some point you need to say, "Okay, what are we going to do about that?" In terms of the more aggressive cancer. What we'd ideally are the markers that tell us this cancer will do better with this therapy and worse with this therapy. And like you said, those are predictive markers. I think the fact is we're still learning the alphabet a little bit in terms of this field.

First, we have to have the catalog of what's there and what looks different. And then we've got to do the right experiments, both clinically and lab-based to understand what acts differently. And when we put those cancers under different stresses, when do they respond better to different things? We need to keep attacking this both from a lab-based perspective and saying, "Okay, can we from a top-down approach say what cancers, when we push them with these drugs seem to do better?" And the bottom-up approach, look at the clinical data, learn from our patients and say, "When these patients have these markers, which ones do well, which ones don't?"

Matthew Cooperberg: And where do you think we get the most mileage both today and in the future, looking at DNA sequencing, looking at RNA expression, RNA-seq some of the more advanced sequencing technologies out there.

Christopher Barbieri: Yeah.

Matthew Cooperberg: Attack methylation, these sorts of things.

Christopher Barbieri: Yeah. Yeah, no. It's really exciting in terms of the technology we can bring to bear on these cancers. I think most of the markers that we have deployable now and utilizable today are RNA based markers. And so when I of think about these markers, there's a balance between how robust the analyte is and how close to biology the marker is. DNA is an extremely stable, easy to test analyte. You can test it forever on crappy samples and it works well. So, it's a really robust analyte, but it's pretty far away from the biology, basically. It's just the blueprints for telling you what's going on. At the other end of the spectrum, proteins are extremely close to the biology, but miserable to try to test for. They're not stable. They go away in a second. They change.

RNA’s been the sweet spot so far, which is a nice balance between how stable and robust the analyte is and how close a good of a snapshot of the biology you get. There's also advantages to things like DNA in that the genomic mutations in the DNA are things that are oncogenic drivers. You can say that's something that does something. And we, as scientists, can take that and put that in a mouse and say, "Okay, what does that do? Is cancer different now?" It's harder to recapitulate that the closer you get to just signatures that are closer to biology, basically.

Matthew Cooperberg: And what's your sense of the current lay of the land? If you look at the current means we have the subtyping prostate cancers, whether we're talking about PAM50 or some of the decipher signatures.

Christopher Barbieri: Yeah.

Matthew Cooperberg: Are any of these getting us close to a clinically actionable predictive test, or do you think we still have a long way to go?

Christopher Barbieri: Yeah, I think we're getting close. I think we're actually getting really close. Things like PAM50, the Decipher signatures, the ability to look at more luminal versus less luminal. And there's lots of different gradations in that. And how you want to get into that is actually telling us about cancer behavior. It's telling us about cancer origins. It's telling us about cancer behavior and it's got real potential for predictive biomarker ability because the fact is luminal phenotypes are really associated with the lineage of the prostate itself. And the main drug we use systemically attacks the lineage of the prostate itself in terms of the androgen receptor targeting. It looks good as a predicted biomarker in that way, because again, those things are really tied together. So I think that's a big step in the right direction at the very least.

Matthew Cooperberg: Are we close to driving trials though? Based on 1050 and similar. AR intensification.

Christopher Barbieri: Yeah, no, no, no. I think it's the right question. I think it's one of those things that you don't know until you do the experiment, to some extent, if it was the right experiment to do. You have to design the trial well, ask the right question and see if the resolution of the markers we have now is good enough to get us information. Or if you get the answer of it's the right question, but that's just not the right test right now. You know what I mean? And unfortunately, it can sometimes be expensive and time-consuming to get a negative answer for things like that too.

Matthew Cooperberg: Well, as one example, do you think we are far enough along in terms of identifying the DNA repair path mutation subset for predictive response to PARP inhibitors? There's been a lot of discussion about this, of course, with the PARP trials.

Christopher Barbieri: Yeah, yeah. Of course. Obviously with two trials that were designed very similarly getting different results.

Matthew Cooperberg: Right.

Christopher Barbieri: One of the obvious questions there is of who they called biomarker positive or not, is that the main difference? And so it's the elephant in the room of how do you say this is the patient we think has a DNA repair deficiency that is targetable by PARP inhibitors. I don't think we have an answer yet. I think we know that true germline carriers of BRCA genes and other things are clearly some of those patients, but they're not all of them clearly. There's not going to capture all the patients. And it's tough because you could argue that the true metric of a functional DNA repair deficiency is response to the PARP inhibitors, which is what you're trying to find surrogate market test for basically.

Matthew Cooperberg: Right.

Christopher Barbieri: Yeah.

Matthew Cooperberg: Right, right. I guess going back to the luminal question for a second, do you think we are close enough to predicting androgen response with the existing luminal, the PAM50-ish signature that is out there now? As opposed to the AR-V7 story has more or less fizzled. Do you think we're there or do you think ... Because this is of course borrowed from breast cancer.

Christopher Barbieri: Right.

Matthew Cooperberg: The luminal phenotype. Do you think it's going to be good enough or do we still need to get a little bit more specific in terms of AR predictiveness?

Christopher Barbieri: Yeah.

Matthew Cooperberg: That's a hard question.

Christopher Barbieri: No, no, no. It's a very hard question. Not just for me, for the field.

Matthew Cooperberg: No. I know.

Christopher Barbieri: The fact is ...

Matthew Cooperberg: You can predict a few.

Christopher Barbieri: The value of things like PAM50 is that it's there.

Matthew Cooperberg: Yeah.

Christopher Barbieri: It's there.

Matthew Cooperberg: Right.

Christopher Barbieri: It's deployable now.

Matthew Cooperberg: Exactly.

Christopher Barbieri: We can try it today. The short answer is, do I think it will be good? Probably not. Do I think it will be good enough? Maybe actually.

Matthew Cooperberg: Fair enough.

Christopher Barbieri: I think it's one of those things that you're always going to do better if you can define your marker for your population in your context that you care about. PAM50 is not that. It's being poured over from something else, but it might be good enough.

Matthew Cooperberg: Yeah. Yeah. And finally, tell us a bit about your own work. What's your lab up to these days?

Christopher Barbieri: Yeah. My lab is interested really in again, the “why” of why these subtypes look differently. What is the underlying molecular features that drive those differences? There are specific genomic alterations that cluster differently in different subtypes of prostate cancer. Some of them never happen together. Some of them happen together all the time. What are those molecular circuits that make that happen? How can we learn about them? How can we learn to break them for patient benefit? Because in general, I think if we're seeing different biology and we can model that different biology, biology's king, biology drives everything, so you can imagine that different biology might mean different therapeutic sensitivities definitely. You can imagine different biology might mean different patterns of spread, local versus distant quicker. You can imagine lymph node versus bone tropism might be slightly different among different biologies. You can imagine the way we use advanced molecular imaging, there might be different biologies that are better for certain. So once you of say, "We've got different biology, we can build models of it," there are tremendous number of questions that sort can fire off from that. And that's what my lab is doing.

Matthew Cooperberg: Terrific. Based mostly in primary prostate or in metastasis biopsies?

Christopher Barbieri: As a urologist, urologic surgeon, most of what I do is based on, focused on, untreated primary prostate cancer and interested in how that works, which is not the easiest to study.

Matthew Cooperberg: Well, no, but actually, the follow-up question, I think is one that I've been really interested for a long time. This is obviously specialty driven. We get the prostate, the oncologist get met biopsies, but how much of the story is written from the beginning? How much of the stories in the trunk versus how much evolves under selective pressure by the time you're biopsying the liver, after three rounds of therapy, is that the same cancer that you started with? Could you have predicted that in the prostate?

Christopher Barbieri: Great, great critical question. And actually, it's something we're really interested in is that idea of tumor evolution and how much of the end game is dictated by how you started. Basically, it does appear that certain types of tumors have their evolutionary paths toward treatment resistance are relatively limited. It's not like all the pathways are open to them. They choose certain paths more frequently. And again, that's a way you can leverage that information. You can imagine almost precision prevention of saying, "Okay, this is the man that we can string along and treat with androgen deprivation therapy intermittently for the rest of his life. No problem. This is a guy that's not going to work for and we need to get upfront different strategies basically."

Matthew Cooperberg: Terrific.

Christopher Barbieri: Yeah.

Matthew Cooperberg: Where are we going next? What's the next year, five years hold for us?

Christopher Barbieri: I think the next few years, what I'm most interested in is the ability of neoadjuvant treatments to really tell us not just who can do well with those therapies, but essentially, treat a patient before surgery with a therapy, take out their prostate and then say, "Okay, who did that work for?" Can that deliver the predictive biomarkers for which tumors did you hit really hard with that drug? Which tumors didn't it touch and why? And again, those are your predictive biomarkers for everybody that gets that drug potentially down the road.

Matthew Cooperberg: Yeah.

Christopher Barbieri: So, I'm super excited about the potential for those things.

Matthew Cooperberg: Fabulous. And it's got to be a much faster, cheaper path to drug discovery development than the phase three in the advanced population.

Christopher Barbieri: Exactly. And the other thing is the fact that in general, the way drugs are usually developed, the field asks a new molecular agent to run the gauntlet of the worst nastiest CRPC that has survived everything.

Matthew Cooperberg: Exactly.

Christopher Barbieri: And if it doesn't do that, you never get a chance to look at what could it do in an earlier cancer.

Matthew Cooperberg: Exactly.

Christopher Barbieri: The idea, the fact is there might be drugs that could cure hormone naive prostate cancer that you would never know about if you only test them in CRPC. These neoadjuvant pipelines, I think, are really exciting to test some of those ideas.

Matthew Cooperberg: Excellent. Lots of work to follow.

Christopher Barbieri: Lot of work to do.

Matthew Cooperberg: Science Based. Yeah. Keeps us all busy.

Christopher Barbieri: Yep.

Matthew Cooperberg: Thanks for joining us. Good talking to you.

Christopher Barbieri: My pleasure.