Matthew Cooperberg: Thanks for joining us.
Martin Gleave: Yes, yeah.
Matthew Cooperberg: So, maybe to get started just a little bit, even before we get into GUNS itself, a little bit of the history of neoadjuvant therapy for high-risk prostate cancer. And you had a lot of experience in the early Canadian trials with ADT. Where have we been and where are we today?
Martin Gleave: Sure. I mean, the goals of neoadjuvant therapy are really twofold. Firstly, to improve regional control, reduce positive-margin rates, but ultimately, to reduce metastasis-free survival. Secondly, a goal of neoadjuvant therapy is, as a window of opportunity trial, to study therapeutic response and to provide us with tissues to understand emergent resistance by comparing pre and post. So, it's an excellent model in that way. Over a period of decades, we've tried to improve upon surgical outcomes by combining initially ADT alone. And then, in fact, we looked at short-term versus long-term, zero versus three, three versus eight months. It showed in those large studies that surgical margins decreased, but that was insufficient for decreasing PSA recurrence rates. What we've learned at that time from those studies is, number one, we didn't enrich for aggressive disease. So, we enrolled too many patients who were not going to have an event down the road. And number two, that margin rates and PSA recurrence rates are not the right endpoints. As we developed new active agents in metastatic disease, naturally, they move forward into the localized space through Canada, through CUOG.
We did a phase two, and then that supported the phase three trial with CALGB or Alliance, the 90203 PUNCH trial. 750 men treated with ADT, plus or minus docetaxel. And while there were hints of benefit, probably not enough to lead to a change in practice. Again, that was ADT plus docetaxel. The new, more potent AR antagonists emerged, and they were evaluated a number of phase two trials led by Mary-Ellen and the Dana-Farber group. And again, suggested that we were getting deeper responses consistent with what we were seeing in advanced disease. And that provided a new foundation to be tested in the neoadjuvant space. And that's where GUNS came in, from a model perspective. So now, we had a much better ARPI doublet foundation to use. And we wanted to use that in a precision-oncology framework, to use genomics to then guide combination regimens in a biomarker-directed way, to determine whether or not pathologic responses could be deepened as a surrogate of contextual lethality to help guide subsequent clinical trials, not so much in the neoadjuvant space, but much more in the advanced space. So, if we could understand which genomic frameworks responded better to this combination, or, just as importantly, were less likely to respond to an ARPI doublet, then you can start to use that in guiding patients with metastatic disease. For example, who to use a chemo triplet M, who needs it, that's still driven by clinical correlates, not so much genomic correlates. And so, we wanted to use GUNS to help guide those questions, not answer those questions, but to guide those questions in more advanced disease.
Matthew Cooperberg: So, walk us through GUNS. What's the framework? What are the sites and which drugs are you using?
Martin Gleave: So, it's an investigator-initiated trial, and we're grateful to Janssen for being really the anchor tenant in providing apalutamide and funding to get this international IIT up and running. Not easy, cross-border IITs, because we hold the IND for Canada at University of British Columbia. And initially, Vancouver and Toronto were the two main sites. It took us a while to open up cross-border US sites just because not so much of regulatory FDA issues, but much more institutional contract and insurance issues that vary across jurisdictions. Over a period of several years, all sites are now active, but we've now enrolled across GUNS, about 280 patients, targeting close to 350. But GUNS is a neoadjuvant adaptive umbrella trial, which means, of course, it's given before surgery. It's adaptive because as protocols open, they are designed to fail early, or expand if they don't. And arms that can be closed allow us to then open up new arms over time as new drugs are added to our quiver, so to speak. Currently, men with... And again, the other aspect of it is, we're building upon PROTEUS, and this is an important space right now. PROTEUS is a large global phase three trial. Men with high-risk prostate cancer treated with ADT alone, plus or minus APA six months before, six months after, enrolled close to 2,000 patients. It should read out by ASCO this year.
And it uses as two co-primaries, MFS plus a complete response rate. So, I use complete response rate as a bit of a focal point right now. You treat a patient with HER2-positive breast cancer with a HER2 inhibitor. Your complete response rates with monotherapy alone is close to 40-50%. You treat a muscle-invasive urothelial cancer. Now, with various agents, you're getting complete response rates of 60%. So, when we try and treat a high-risk localized prostate cancer with the very best systemic therapies known to control the disease for five, seven, nine years, we have complete response rates that are in the low single digits. And so, there's something about prostate cancer that makes it challenging to have complete response rates, heterogeneity, tumor stromal interactions, immune, the surveillance issues, etc., and etc. So, that, though, allows us to use depth of pathologic response as a CR or MRD, which is a low event rate, to allow us design a trial, to see small increases above that low event rate where our sample sizes are quite feasible. So, that's the design for GUNS, men with high-volume, high-grade cancer. These are not single-biopsy four-plus-fours. These are multiple cores of grade group four and five, are enrolled to an eight-week master protocol. And they're treated with an ARPI doublet with APA. During that period of time, needle biopsies are acquired, sequenced, and based upon that sequencing information, they're then assigned biomolecular tumor board to one of now six subprotocols.
Each of them designed from biologic rationale perspective to capitalize on their genomic vulnerability, so to speak. And so, along the way, we, of course, now have a genomic sequencing on over 200 patients, a genomic landscape that's dominated by what we call AR-associated alterations. The AR is a co-oncogene. It doesn't become genomically altered until we selectively target it, but the most common genomic alterations are those that associate with the AR. FOXA1, SPOP, obviously an E3 ligase that degrades the AR, or ETS fusions where, of course, AR drives ERG and other genes in that pathway. So, those are the three most common alterations. They all go into subprotocol one, where subprotocol one randomizes those patients to an ARPI doublet with APA or a triplet with APA plus abiraterone. The hypothesis being that by intensifying AR pathway inhibition in a group of patients that are AR-driven, but more importantly, lack other nasty genomic alterations, these may be the subtype that benefit from hitting the AR axis harder, AR pathway harder. And indeed, again, the statistics are Simon's two-stage. We enrolled 24 men in each arm and the MRD rate, which is a primary endpoint, was up, I can't remember now, in the high 40s versus 16%. So, a clear benefit to the ARPI triplet versus the doublet. So, that protocol is now expanded. Second stage is complete, over a hundred patients enrolled in that subprotocol, but we haven't got all the pathologies in to read that subprotocol out. The second subprotocol was the subject of the GU ASCO presentation here.
These accrue to the second most common alteration group, which are the aggressive tumor suppressor gene losses, p53, PTEN, RB. And these are aggressive genotypes. We know that they're more difficult to control with AR pathway inhibition. So, the hypothesis there was that in these aggressive genotypes, if we hit them with ARPI doublet plus docetaxel, do we see a better response? And if we just hit it with an ARPI doublet? And the bottom line is that yes, we do. The ARPI doublet MRD rate is zero, so zero out of 24, consistent with what we see clinically, that this is a group of patients who derive a shorter-term control with systemic therapy. When you add docetaxel, nine out of 23, or 39%, achieved that MRD rate. So, pretty big difference, surprisingly. And so, that justifies expansion of that subprotocol now to its next second stage. But there are a couple of other alterations when we looked at genomics of the people who achieved an MRD versus non-MRD in the docetaxel triplet group. Those who had actually a co-occurring ETS fusion plus either p53 or PTEN, they had a much higher rate of intraductal carcinoma, which is an aggressive phenotype, and a much greater likelihood of not having an MRD. So, whereas if you had an SPOP co-occurring with p53 or PTEN, they were enriched in the MRD group. So, we start to see the impact of co-occurring alterations and therapeutic sensitivity. So, that's the subprotocol two group. Subprotocol three is enrolling in classic HRR-altered into a PARP inhibitor arm that's enrolled over 30 men now. Subprotocol four, the high-TMB MMR-altered, they get atezolizumab as a PD-L1 inhibitor. And we've recently opened up two subprotocols.
Subprotocol five captures the, what we call plasticity-prone cancers, those that have RB or p53 or co-occurring alterations that we know when we suppress the AR axis, lineage plasticity is activated and you get a neuroendocrine kind of phenotype, even early on, detectable with transcriptional assays. So, that arm now uses the epigenetic modulator EZH2 in combination with the ARPI doublet. And then, the six subprotocol captures two groups. One is the subprotocol one genotype, which are what we call genomically favorable. They go into 6A where they get plus or minus capivasertib, the AKT inhibitor. Based upon the hypothesis that when we inhibit the AR pathway, AKT is activated. If you inhibit AKT, cross-talk activation of AR, worked by Brett Carver and others a long time ago. So, by hitting that genomically favorable subgroup with the capivasertib combination, we may get a deeper response by inhibiting survival pathways activation. The second is 6B, where we now put PTEN-altered and AKT-activated into a capivasertib arm as well. So, that's the evolving and multi-armed aspects of GUNS.
Matthew Cooperberg: And this is all mutationally driven. Are there other... First of all, is your default subprotocol one if you don't find mutations, or are you looking at transcription as well? Are there other?
Martin Gleave: Well, 15%, when you think about needle biopsies, high-grade, high-volume. So, you'd think we'd have sufficient tumor cellularity, such that we'd have sufficient tumor-content DNA, or RNA, to get both a DNA- and RNA-seq pattern. We get both, but 15% of patients are non-evaluable because of low tumor content, even in... And that's just in part, I mean, because we're using a CLIA-certified assay from TEMPUS. They require 160 nanograms of DNA, tumor-content DNA, and you don't always get that. If we run it in a bespoke lab, like when Alex Wyatt takes those samples, runs it in our group, the vast majority of them have an interpretable assay. But that's in a bespoke lab, non-CLIA-certified, and you can't use that to guide therapy in a therapeutic sense. Genomics is being used to make therapeutic assignments, but a big part of GUNS, really, the reason for designing it is to gain insights into therapeutic resistance that emerge downstream of precision-oncology-framed combination regimens. So, we know that as PARP inhibitors are added to HR-altered patients, as we treat with IOs, as we treat with AKT inhibitors, so on and so forth, then emergent resistance is going to be different than if we just treat with an ARPI doublet, because the selective pressures are going to be different.
Matthew Cooperberg: Are you looking at germline as well?
Martin Gleave: Can't look at somatic without germline. And so, germline is obviously detected. And that leads to, of course, familial counseling where necessary. But the main areas there are things like, obviously, BRCA, CHEK2, and Lynch syndromes. We even have a few, a couple of "mesh in maya", I think it is a p53 alternative. The advantage of an umbrella trial is that you're able to detect the 1-2% rare birds that fall through the filter, and that's why an umbrella trial is so important for this type of analysis.
Matthew Cooperberg: What's your most exciting finding so far? Or any surprises, I guess, is the other question?
Martin Gleave: It wasn't so much... We expected subprotocol 2B, which again used chemotherapy as just sort of a holding pattern until we had new drugs to target that aggressive genotype. The absence of MRD in the ARPI doublet group wasn't surprising. What was more surprising was the high MRD rate when you add chemotherapy. And I think this provides some guidance to the question of who to use an ARPI doublet in versus a triplet a la PEACE, ARASENS, that if... The way I'd be guiding this going forward is that if you've got a high-burden metastatic prostate cancer patient, you know it's PTEN- or p53-altered, I'd be more inclined to add chemo earlier in that sense. And so, I think that's where a lot of people think of GUNS, "Oh, you're developing this to develop neoadjuvant regimens in the future." That's quite a ways downstream and contingent upon PROTEUS being positive. But right now, it helps provide questions that can be addressed in the systemic state with systemic therapies as we get better and better at segmenting heterogeneity in the metastatic state.
Matthew Cooperberg: But it's also an important philosophical stance on the cancer from the standpoint of identifying mutations in the trunk rather than the distal branches, right?
Martin Gleave: Yes.
Matthew Cooperberg: And even going around this meeting, I'm always struck as the armamentarium expands how frequently the decisions about advanced disease get made in the absence of updated somatic testing. This is by patient preference, or by insurance coverage, or all these non-biologic factors. So, as you do this, I mean, the assumption is that there is a predestiny written in the primary tumor that will predict response to these different regimens and will predict how the cancer's going to behave as it goes through therapy, right? So, I mean, first of all, do you think that's true? Do you think you can read the future from the primary tumor, which-
Martin Gleave: I think I like Yogi Berra said, "It's hard to make predictions, especially about the future." I think that there's many different ways that heterogeneity can shuffle the deck over time, and the addition of a therapy obviously adds great variety. We are doing, and again, there's a paper coming out with one of our computational people, Josh Scurll, where we're looking at our first 150, soon to be 200, baseline biopsies, looking at things like cancer cell fractions to better understand truncal versus subclonal alterations. And things like SPOP, things like T2Es, things like p53, these are truncal. PTEN is subclonal, and why there's great heterogeneity of PTEN loss. And it leads you into, especially as we start introducing AKT inhibitors into practice, "What is the best way to ascertain PTEN deficiency in cancers?" And what we are seeing is that PTEN genomics, copy number loss or mutations, detects only 50% of PTEN-deficient cancers, because loss of protein can occur for a variety of reasons. Most assays don't detect copy number loss completely. So, you've got to use a composite. And then, using IHC, it's got to be probably less than 10%, less than 5%. So, the more deficient you are, the more it isn't subclonal anymore. And we see some cases where there is a PTEN genomic alteration, in cases where PTEN protein is 50% positive, right?
So, there is a disconnect just because of the clonality aspects of it. And so, it comes down to understanding what that biomarker means as you try to select out those patients who you shouldn't give the drug for. So, I think that's important. And again, I think the other thing that I see that is concerning as a therapist is heterogeneity. There's a daunting heterogeneity at, obviously, in the pretreatment biopsies, but this is even increased more greatly post-treatment. And so, I think this daunting heterogeneity, which we can now see increasingly using spatial profiling, just speaks to the challenge as to why we don't see complete responses. The surviving population is able to, I think, have subpopulations that are either present, but more importantly, I think, emergent downstream of changes under selective pressures of therapy.
Matthew Cooperberg: Have you learned anything about these surviving cells yet? Is it a heterogeneous cluster? Are they stem cell-like? Are they... Or is it too soon to say?
Martin Gleave: A big part of GUNS, and this is part of the Pacific Northwest Sport, which we're part of, a Terry Fox program that we're funded for in Canada, is to use various omics to compare needle biopsy versus post-treatment biopsies. We're just starting to profile the first 45, actually the first 48, in subprotocol one. And that's not easy because post-treatment, you're dealing with small foci of cancers that are scattered throughout predominant stromal regions. So, you can't bulk sequence those and expect to get anything other than a smoothie out of it. So, I think this is the challenge. So, you have to go to digital spatial profiling approaches, and that is expensive and challenging as well, as you try to scale it up. So, we're working on that, and that's the ultimate goal. In some cases, and we've done some already outlier cases to help us get grant funding, because it's important to show outlier observations. Patient one in the trial was a patient who was actually MSH2 altered, a TMB of 60, went on to subprotocol four, had an amazingly deep response, a positive two-centimeter positive node disappeared to a complete response, a large-volume, high-volume, high-grade cancer, regressed down to an eight-millimeter focus.
So, not an MRD, but when you looked at that eight-millimeter focus, first of all, bulk sequencing, incredible upregulation of plasticity, ESC, neuroendocrine pathways that are emergent within six months of therapy. And when you look at it from a DSP perspective, multiple subpopulations that were all plastic, but actually expressed different components of neuroendocrine markers. So, I think that's an example of a cancer that is going to be challenging to control systemically. Obviously, there's changes in immune infiltrate based upon the PD-L1 therapy as well, but that, we only have about seven patients in subprotocol four just because those are quite rare outliers. But we see it across other genomic subtypes as well, that it's quite common to see plasticity pathways activation and neuroendocrine markers emerge, not neuroendocrine prostate cancer, just neuroendocrine markers that are unmasked, derepressed, so to speak, by AR pathway inhibition, which is the rationale for why we're introducing EZH2 inhibitors as a way to potentially suppress that pathway's activation.
Matthew Cooperberg: Last question, you've talked a lot about this, about GUNS as a platform for discovery and for validation with an eye toward late-stage disease, but what about the concept of neoadjuvant therapy and the role of surgery in early advanced disease? And we have the RAMP trial finally came out. This has obviously been controversial for a lot of years. Hard to get surgical trials done. But do you think the role of surgery, as part of a multimodal approach, including systemic therapy, has a future, or has a brighter future in the next years than it's had in the last few years?
Martin Gleave: Oh, definitely. I think that what's important, with regards to management of high-volume, high-grade prostate cancer, is recognizing that local control is important. Surgery and radiation therapy provide, I think, two modalities that address that need. The advantage of surgery is that you're able to obtain tissues out that help provide genomic biomarkers into the future, helps to secure the pelvis, helps to provide, I think, more accurate BCR status earlier, helps to integrate PSMA PET imaging earlier to then help guide subsequent metastasis-directed therapies, combined with systemic therapies into the future. So, I think that surgery will continue to provide an option in those patients. And this is where I think the concepts of multimodal therapy is improving control of patients, who in the past were deemed non-curable and therefore, not surgical candidates. It's a concept of control. And the convergent advances that surgery is a better-tolerated anatomic operation now, so is radiation therapy. The use of imaging is disruptive and enabling to us to detect early recurrences. If you're not detecting early recurrences, you lose a potential oligometastatic state that is then more controllable.
SBRT, metastasis-directed therapy is key to that. And I think a lot of our advances in the systemic state applied to that may be the sort of other aspect of conversion advances, in the same way that testes cancer is controlled by multimodalities and enabled by better systemic therapies. So, as we get better systemic therapies on top of an AR pathway inhibitor doublet, figure out who to add a PARP inhibitor to, who to add a PSMA conjugate of some sort into the future surgery, will continue to remain an important part of that advance.
Matthew Cooperberg: Thanks so much for your time. It's beautiful work. Very, very exciting space to keep our eyes on and we'll look forward to the next updates, hopefully, very soon.
Martin Gleave: Thank you.