I'm Marc Frahm from the biotech team. So, we're really welcome, happy to have with us for this next panel on novel IO targets and approaches, the CEOs from Adaptimmune, Adrian Rawcliffe, Gritstone bio, Andrew Allen, Shattuck Labs, Taylor Schreiber, and Xilio Therapeutics, René Russo. We have a whole list of questions, and I'm also joined by my colleague, Ernie Rodriguez Dumont, who helps me on several of these names. And we have our whole list of questions, but feel free to raise your hand, ask follow-ups if you have them. We'll try to pepper those in as well. But maybe to start off with, we'll start with some general questions, and then we'll go into some company-specific questions. So just starting off, obviously, IO went through an explosion in the 2010s of development programs, lots of different targets. Unfortunately, many of them have fallen by the wayside. You know, maybe... And that's led to some apathy among a lot of investors about novel IO. Maybe we'll start with Taylor. Just why is now a good time to kind of come back to it, to come back to, in some cases, the exact targets being approached in a different way, and other times, it's just novel targets. Just why, why is it now a good time to think about IO? I think because we can learn from a very large clinical experiment that's now been done over the last 15 years, and, you know, there's some key takeaways. If you look at what's worked and what hasn't, a very small number of checkpoints have shown any evidence of any clinical activity, and those are the checkpoints that sit at the top of a hierarchy that includes a whole bunch of other checkpoints that have since fallen by the wayside or, at most, provide incremental benefit over what anti-PD-1 or anti-CTLA-4 provide on their own. And at the same time that clinical experiment has been going on, we've also seen the clinical experiment with T cell engagers, with CAR T, and with different co-stimulatory inhibitory antibodies in autoimmune disease. What that set of parallel experiments tells us is that an appropriately activated immune cell is very, very difficult to suppress. So I think it, it points to mechanisms of agonism, perhaps being something that might build on the cornerstones of checkpoint inhibition more than incremental checkpoint additions going after TIM-3 or TIGIT or LAG-3 or IDO or kynurenine or anything else that's failed along the way. Adrian, you want maybe add on to that? You know, a little bit different perspective within cell therapy, where it's not so much the modifying—altering the immune response that's happening in the cell, but introducing kind of a completely new immune response. Yeah, I mean, I think the first point I'd make is, I don't think it's particularly surprising that the IO field, like every other space that is subject to irrational exuberance, goes through the normal cycle of, "This is gonna be the greatest thing since sliced bread. Everything's gonna work. Let's de-risk, everything's de-risked because of our barely mediocre or minor experience with the first few things that work," followed by a period where the laws of drug development and discovery tend to weigh in, and not all targets are great, not all targets are clinically relevant, not all products are without adverse events that stop their development. And you learn from that, as you say, and the next cycle, I think, is better informed in that regard, and I think that's true. That's, you know, my little microcosm of engineered T-cell therapies. That was particularly true. I mean, if you think about the CAR T experience with CD19. You know, CD19, CD19 was definitive proof that CAR T was gonna be the solution for all cancers, and that's why we have 147 CD19 programs. Many of them now have been repurposed into autoimmune disease. It's not that easy, but I think we do learn from the mistakes. And I think I also think the investment community learns from mistakes. So to your question about why would this be a good place to get back in, well, the valuations are completely different. They no longer assume success, and I think the investment community has learned an awful lot about what makes a good program a good target, and what questions to ask about early-stage IO programs that I think the next phase should be better experienced than the last. René, maybe you want to add in, just contrast here, where you, you know, you're using a mechanism that, unlike some of the others that are to be proven in the clinic, we know IL-2 fundamentally can work. Just, maybe you would add your perspective there. Yeah. So I think we're taking a completely different approach at Xilio, which is taking historical targets, be it IL-2, IL-12, ultra-potent CTLA-4 therapy enhanced, that we know from history, shrink tumors, but we know have a therapeutic index problem, and we know even as we use them today, we're leaving efficacy on the table, and it's been proven clinically, that if we can get to higher concentrations, we get a better outcome. And of course, we can't combine any of these agents because of the tox. So our thinking is, let's not try to figure out novel targets, because that's really tricky. Let's take these ones we know work, and let's bring this sort of new concept of geographic precision in drug delivery. Let's not expose the entire body systemically to these potent immune stimulatory agents, but let's find a way to preferentially push their activity into the tumor. G ets really high exposures there, and then start to do combinations. And what we really like is multifunctional molecules, where you build them in the same molecule, but have that tumor selectivity. And we think a little bit, we're sort of taking the risk of the target off the table, but now we're learning about how to, how to leverage the tumor biology against itself. Andrew, feel free to comment on that exact topic, but maybe also layering in, just this idea of showing single agent activity, which may be possible in some cases like IL-2, or CTLA-4, but in others, you know, it hasn't always been there, but has, has even with an IO led to approvals now with, LAG-3, right? Where there wasn't much single agent activity. Yeah, I think as my panelists have said, I think very eloquently, you know, we're learning, and we learn as we go. And as always in biotech, you should follow the science, and we've learned a ton about the science. So if you cast your mind back, just, you know, 15 years ago, right when checkpoint started, all we knew is that they activated T cells. There was some magic that they turned on T cells. We didn't particularly know which T cells, we didn't know what the targets were. We had very little knowledge. And we've learned a ton about the targets. Neoantigens clearly are one of the key targets, or the target class that matters the most. It all makes sense, because, of course, you're not tolerized to those, because they're not around when you're a fetus or a neonate undergoing a process of central tolerance. So you retain this T cell repertoire that recognize neo-antigens, they recognize targets in tumors. Sometimes they successfully eliminate tumors before they even develop. That's what we've learned, and then occasionally, unfortunately, too often, tumors outsmart the immune system and start to grow in the face of either no immune response, because they've hidden themselves, or in the face of an established immune response, because they have developed some adaptive resistance mechanism. So we've learned all of this, and we started to learn about the nature of T cells that we like, and the expression profiles, the transcription factors that are turned on. So we're following all of this, and the clinical question is then, from an investor point of view, single agent therapy activity. It's desirable, of course, because it makes everyone's life easier, but the playbook of targeted therapy is not the right playbook for immunotherapy. I think not everybody has really kind of tuned into that yet. If you just step back, what are we trying to do? We're trying to prolong survival of people with cancer. That's our goal. Historically, RECIST response was a good predictor of that, with targeted therapeutics and with chemotherapy. It's no longer true. I was reading a paper on the plane yesterday from Memorial that made this point very eloquently. They just looked at just over 1,000 subjects with non-small cell lung cancer, treated with checkpoints at Memorial. It's not a randomized trial, it's a single center experience of everybody with lung cancer treated with a checkpoint. If you were treated with a checkpoint, and you did not respond, you immediately progressed. You died about 4-5 months later. Right? That's lung cancer. That's static lung cancer. And if you responded, great, but only 20% of people responded. But then they progressed, and as the paper was all about acquired resistance to checkpoints. So at the point of progression, you would then think, "Okay, they're now in that I'm doing badly mode. I'm gonna die after 4-5 months." No, median survival after progression was about 20 months. So the immune therapy is doing something that is completely under or, in fact, unappreciated by this myopic focus on lesion shrinkage. The immune system is much more nuanced, and we're establishing degrees of control over tumors using T cell and adaptive immunity that are important, they're durable, and they are not captured by this focus on the things shrink quickly. So I think we have to get away from this relentless focus on single agent shrinkage activity as monotherapy. It's just not capturing the benefit that we actually care about. So you do randomized trials, and they're small. You can argue about power, but I think that's the only way you can start to really peel the onion here and start to identify regimens that are gonna drive OS benefit. Because everyone's getting a checkpoint, and we all want to treat early, and everyone's getting checkpoints early. So let's just agree that everyone's gonna get a PD-1 antibody at a minimum, plus my new therapy. That's the basic concept that makes sense, and that's what we're all gonna be doing. Taylor, with your mechanism, SIRPα, you know, again, not designed to have tremendous from a RECIST perspective, single agent activity, but just, you know, how do you prove out—go about proving out efficacies and in early trials in late stage disease, where you—unfortunately, that's where the trials have to start? Yeah. Well, thanks, first of all, for the unprompted plug of our paper. I think you just have to set a high bar. I mean, we had some single agent activity. So, for those who aren't familiar, our lead molecule is a fusion protein that simultaneously inhibits CD47 and activates CD40. It's designed with a null Fc domain to avoid hemolytic anemia that killed magrolimab. And because it's designed with a null Fc, then the only patients that are gonna respond to monotherapy are those tumors that spontaneously, for whatever reason, decorate themselves with an abundant source of a prophagocytic protein. So we saw some single agent activity, and but that's, you know, because we were fortunate to treat some of those patients whose tumors, for whatever reason, had that phenotype. But it's not - it wasn't an expectation going into the study, and I think I agree with many of the comments you made, Andrew. But I think what we do have to have is very stringent bars for what is medically meaningful in a combination, in a small sample size. So, at our personal view, this is not everybody's, is that if you're not triple what the single agent response rate is for whatever you're combining with, in our case, azacitidine, in the first 20 or 25 patients you treat, you're probably better off using your resources elsewhere. René, how do you, how do you approach this, this issue of, you know, kind of establishing proof of concept early on? Yeah, I mean, it's obviously our biggest challenge, right? Because especially when you're working on mechanisms that rely on some active immune system there, right? I mean, for an IL-2, for example, right, you need, you need some CD8s in the environment. You need something to stimulate, right? Because IL-2 is a part, is a leg of a stool, but it, in and of itself, isn't the whole stool. So, and when we look historically at the IL-2 activity that we saw with aldesleukin, first of all, the excitement was around the 10-year durability, right? That was the excitement. Maybe it was around a 15% response rate, but it was first line. It was first line everything, right? So, now what do we do? I think what we're all struggling with this because there is this bar for the single agent activity, and we have seen it in our phase 1s from time to time, which is great, because then you can sort of... it does make things easier. But we're also looking at, and this is also difficult, matched tumor biopsies, looking at the PD in the tumor. Are you seeing consistent, you know, for, in our case, CD8s and NKs going tumor across the board? Are you seeing some level of dose response? Are you putting these pieces together? But I think it's a real challenge. The other way we're looking at it, and I know others here are looking at MSS colorectal as an example, is where you're studying in combination where the PD-1 has a less than 5% response rate. That's what we're doing with our CTLA-4, is we're going in with a PD-L1, and, you know, we know historically the contribution of components is very low from the PD-1. We know this is a high unmet need area. It's a high bar, but then you clearly know that the contribution of those components is valid. So I think you have to look for those examples where, you know, the PD-1 alone is really negligible. Andrew and Adrian, your response to this. We'll start with Adrian, just on kind of that contribution of the components. With your... You know, you've done some testing of your first generation MAGE-A4 product in some tumor types, but it's not exhaustive because also with a personalized cell therapy, like, you have a pretty high efficacy bar to justify the complications of therapy. You know, now you have your second generation going. How do you think about kind of proving that out, and how does the FDA approach it when, you know, you don't have hundreds of patients on without the second modification, right? To say exhaustively exactly what outcomes are for a first generation. Just how do you approach proving that it is dramatically better, that it really is, that you need all those components? Yeah. So, I mean, I think the only way it's approached at the moment is just, okay, this is a different product. Okay, and you test it, and you see what effect size you get, and you see, does that justify continued development? Now, in the case of our next generation therapy targeting MAGE-A4, we had some level of benchmark with the first generation. We had 16 patients in non-sarcoma indications where we saw two responses. So that was showing that the T-cell receptor is probably active, but that it was probably insufficiently active to drive an accelerated development program based on the response rate we examined. So the next generation sort of had to beat that, and the high watermark for us was that, in the sarcoma indications, we saw a very good response rate in that first generation. So we saw, like, a 40% response rate. So when we repeated that in a basket of, a basket trial with the second generation approach, across all tumor indications, we saw a 35%-40% response rate. That seemed good enough. I think it's, it—you, you are getting into, you're gonna get into real difficulties if you are, if you get to, be asked, "Okay, dissect that and prove to me that each additional component makes a difference." I think you've got to think about the safety of the individual components before you go into a trial. But ultimately, what you're proposing isn't, isn't, an incremental approach with, incremental registration. It's, at the moment, it's a brand-new registration strategy. Now, there may come a time where we think about things as a platform, and we can insert new TCRs, and we can insert new next generation modifications. That's not the reality of development and approvals at this point, though. How do you approach that with, as you've moved into the maintenance setting, and, you know, there are several drugs being put in there that have not been, you know, fully studied in that setting, right? Yeah, we're different from cell therapy. I think the challenges to cell therapy are very distinct, where there is a focus on single arm studies. For understandable reasons, randomization is impossible, or blinded randomization, of course, is completely impossible, and even randomization is very challenging and unnecessary early on. But I think the focus appropriately is gonna be swinging to durability. It's all about durability. You know, IL-2, back in the day, right, when I, my first job in the Bay Area at Chiron on high-dose IL-2, if you judged it by RECIST, it was about 10% responses, about 4% were durable, and they turned into cures. So we were curing 4% of people with metastatic melanoma and metastatic renal cell. And that obviously was just phenomenal. And we've just been learning since then how to try and drive that cure rate up. That, to me, that's the prize, right? Responses and all this malarkey, sort of, it's fine, but it actually, who cares? Survival is what matters, and we're trying to drive that tail up the curve, right? Right shifting a curve by two months is not interesting to patients. It's interesting to investors because you can make money out of that, but it's not interesting to patients, and I think it should not. We should not be overly focused on that. We're trying to lift the tail with immunotherapy. That's the whole point of immunotherapy. So combinations are the way to go. We can do randomized trials because we have a sort of easier construct, and therefore, you just have to get into randomized studies, and there's no way around it. You're looking for something that is the best possible surrogate for overall survival in a randomized setting. And whether that's PFS or some other marker, and we've obviously been exploring circulating tumor DNA because there are some challenges with PFS, potentially. But all we care about is surrogacy for OS. Show that in a randomized study. And then, as you say, we are using a checkpoint, and just as René noted, checkpoints have been tested in standard microsatellite stable colorectal cancer and have been found lacking. In whichever context you do that, whether it's frontline combo therapy with chemo, whether it's maintenance therapy, where we did a big study of atezolizumab as maintenance, none of it has been shown to be effective beyond the couple of responses. No, no shift in PFS or OS. The best data was from AstraZeneca, second-line, combination of PD-1 plus their CTLA-4 tremelimumab, and it shifted the curve a little bit to the right, but it was not enough to be interesting to anybody. So you don't need to repeat all those studies, and physicians won't do them because checkpoints are not without toxicity. So when we set up our randomized study, which is chemo plus or minus immunotherapy, we spoke to the agency, but actually first, we spoke to our physicians and said: Would you do a three-arm study, chemo, chemo plus atezo, chemo plus atezo plus vaccine? And they said, "No, because we've done the chemo plus atezo, and that was a big study we all participated in. No efficacy, but toxicity. So it is unethical to, for us, in our minds, to do that trial." So we go to the agency, and the agency happily agreed. So it's a two-arm study with the combination, now looking at these different efficacy endpoints, PFS and various criteria for molecular response. So I mean, that's the way you do it. There's no easy answer here, and then you, if that's good, you then move to a phase II study. I mean, it's not, it's not rocket science. It's kind of a little bit old school, I guess. We're back to the future where, you know, we used to do phase I, and phase II, and phase III, and that's what we're doing. And then each one of your companies and programs are looking at different settings in the adjuvant maintenance or different histologies. So what do you think makes the optimal pathology or settings? How do you evaluate the potential or opportunity to determine which are the best settings as for the IO? We will start with Jeremy. Sure. I think we're probably all each gonna have different answers to this. And the reason for that is because the right setting is dependent on what the underlying effector mechanism is of the IO therapy you're using. And in Shattuck's case, we're using a therapy that is designed to activate antigen-presenting cells, and overcome situations where the primary immune defect is at the point of antigen presentation, or myeloid-mediated immunosuppression. And so those, that includes, you know, a list of tumors like ovarian cancer, certain heme malignancies, there's a number of other ones that are out there, where the tumor itself at baseline is defined by a dearth of T cells. A dearth of cells that have shown prior evidence of antigen engagement, and an abundance of cells that are thought to act at the point of suppressing antigen priming in the first place, immunosuppressive myeloid cells, in our case. And then secondly, it depends upon, you know, what are the key ingredients that are required for your drug to drive maximal responses, in addition to whatever the drug is doing on its own. And as I mentioned a minute ago, in the case of our compound, there needs to be a pro-phagocytic signal present, in order for the CD47 effector mechanism to effectively bridge to CD40 agonism, which is what the second arm of our drug provides. And that can come from decorating your tumor cell targets with high concentrations of antigen-targeted antibodies that have an active Fc domain on the tail end, or it can come with combining with certain chemotherapies that lead to decoration of those tumor cells with endogenous pro-phagocytic signals. So it is very mechanistic, mechanism-driven, I'd say, in our case. So I think with an engineered T cell, I think the unique selling point of the engineered T cell is that you tether the effector mechanism directly to the targeting mechanism. And if you get that signaling right, you end up with a very, very, very powerful effector mechanism that you can then subsequently optimize as the cell itself, as an effector mechanism, you subsequently optimize. But I think the advantage of that is that what you're really doing is you are removing reliance on an endogenous T cell response to. So then you think about, well, where could that be applicable? And the answer is actually from a histology perspective, it's probably applicable everywhere, but there are some places where actually reliance on that endogenous T cell response works quite effectively. So where checkpoints work, they obviously have a neoepitope-driven and T cell response seems to be also where TILs are effective. And therefore, what's the additional benefit of a T cell, a highly engineered T cell? Well, it's probably still effective, but actually, there are spaces where you can't rely on that endogenous T cell response, and including right at the other extreme from melanoma would be something like synovial sarcoma. So it's not coincidence that in synovial sarcoma we get fantastic responses as a cold tumor that has not seen immune activity. It just hides to a point Andrew was making. And then when you show it to a T cell population that's engineered to kill it, you get profound responses, and they are very durable because it does not have the ability to evolve out very quickly. So that's a good starting point, but there are other places like ovarian cancer, where they're marked by a lack of T cell, effective endogenous T cell response. And so, and I think you sort of work backwards across there. Not to say that we won't go into melanoma, not to say that people who are going into melanoma, that's a bad idea, but it's more crowded, and there's other approaches to T cell engineering or to immune therapy that will be effective there, whereas they won't in these other spaces. Yeah, I agree with all that, and I think, well said. So cold tumors are where the unmet need is, and as far as we can judge, many patients do not have a T cell response to their tumor neoantigens. So you're starting where just giving some kind of nonspecific immune stimulant probably is not gonna do the job because you don't have that right substrate there. And that's obviously why we focused into that space, because our whole therapeutic hypothesis is that we will give you the T cell repertoire, or not the repertoire, but the response that you currently lack. Giving an exogenous cell population is one way to do it. We come at it from the vaccine perspective, and of course, there is interesting evidence that maybe these two approaches actually synergize, and that's for another day. But we take the vaccine-based approach, and so we went into a cold tumor, and we did phase 1, as you do, and we saw very clear evidence that this was working. It It was eliciting T cells, entered into cold tumors, proliferated, and killed tumor cells. So then we scaled up and entered a randomized phase 2 study, as I described. So far, so good. Now, we looked at adjuvant because that's something I haven't spoken about yet, but obviously, the data from Moderna, in particular Moderna Merck, in the adjuvant setting is very encouraging with personalized cancer vaccines in melanoma. And of course, the checkpoints alone are showing really exciting activity even when used in the perioperative space, whether that's neoadjuvant, adjuvant, or both. Thinking of that KEYNOTE-671 study, everybody wants to be treating early, which I think is the right instinct because we all know the best chance at cure is your first treatment. So we need to just get everything lined up, so that first treatment phase, and it's probably gonna end up with different phases of different drugs and so on as we figure this out. That first is the one that the patient should be most focused on. So we looked at actually doing an adjuvant study, and we began two studies, one in adjuvant colorectal, one in metastatic, and we made a judgment call, I think correctly, that, like, the markets hate us because they hate cancer vaccines. They kinda hate the IO field in general, and we're never gonna be able to finish these two trials if we, if we press ahead, even though it makes total sense to do that. So we actually stopped the adjuvant one, knowing that the metastatic one would go faster, and I think that was the right call. But we're on the cusp of data, and we'll see what signal we're able to generate in the metastatic setting, and obviously optimistic for that outcome. And then, if we get a nice signal there, the question is, okay, do we keep going metastatic, or do we actually then move into an adjuvant space? And do we stay in colorectal and think about something else? 'Cause if it works in colorectal metastatic, we're probably gonna work in most places, as I'd say, right? It works in melanoma, but do we wanna go, continue there in that relatively small population? Do we wanna think about lung? So it does open up the space for you, but I think you've just got to keep following the science and the sort of basic therapeutic hypothesis that's been tested proactively, retrospectively, in all the sort of standard ways. You just follow the what signals you get forward to the next stage. Yeah, I don't think I have anything else really to add to everything that was said. I mean, for us, it's mechanistic focus or for IL-12, we are more interested in colder tumor types as ovarian head and neck, as someone mentioned. However, I think the added layer for us, right? Because we, we're using the protease biology of the tumor against itself, we're also continuously profiling. We have over now 1,000 human solid tumor samples, and currently, we're profiling the protease biology and the MMP. So we're matching, right, the indication and tumor type, also with the protease biology, and that can even be specific to looking at something like MSS CRC with liver metastases and looking at even in patients, are we seeing activation of our molecule in liver mets in this type of tumor? I think that's an added layer for us that's been very helpful. Yep. We will now segue into more just purely company-specific questions, and we'll kinda do this round robin style, just so that I'm not picking favorites alphabetically. So we'll start with you, Ad, with the Adaptimmune. Just, you, you have BLA now on file, with priority review. So one, congrats on getting that in. And then also, you know, what disclosures is Adaptimmune kinda planning to make along the way, especially in cell therapy, where manufacturing really is most of the review, right? You know, things like around inspections and, and just what, what's the approach on the disclosure there? Yeah, so just since that, we've got a PDUFA date to the fourth of August. That's a Sunday. So So I'm not anticipating we get it on the fourth of August. And we are very much now on the FDA clock priority review. We have scheduled all of the things that you would anticipate you schedule as part of a priority review, mid-cycle review meeting, PAIs, GMP, GCP, late-cycle review, et cetera. And so we're responding to information requests as we speak. We anticipate that we will update as we tick off those major items. And we have taken the approach of being reasonably granular, historically in our development program. So as we have the various meetings, as we have the inspections, we will update everybody. The first opportunity to do that is probably at our investor day, which we will be having on April 18. Invites will go out later on this week to everybody, and I encourage anybody who wants to understand where we are in the process to attend that. That'll happen at our Navy Yard facility down in North Carolina-down in Pennsylvania, sorry. Wrong place. Down in Pennsylvania. Probably Pennsylvania somewhere. In Philadelphia. The opportunity there, not only to understand where we are in the process, but also how we see the opportunity from both a patient and a physician perspective as well. It's not North Philadelphia either. It's not- Fresh Prince. Not north. I know, but it was a rolling submission, so some of those modules have actually been in for quite a while at this point. You know, so how should we think about overall timeline, you know, given that some of those modules have already been in for a while? Is there a real chance that, well, maybe not all the way to Sunday, August fourth, but that this doesn't have to go all the way to Friday, August second? So, yeah, in theory, we have a rolling submission. The obvious advantage of that is that the preclinical and clinical modules that have been submitted now for some time can be reviewed by the agency, and there's good evidence that they went through those. However, as you pointed out, almost all of the interesting discussions with the agency in the CSRs, the RNA, the CMC portion that was submitted as part of the final submission in December last year. So we anticipate that, and we have priority review, so it's not like there's a long time clock on this. So I anticipate that we will get approval on or around the predicted date. Okay. And maybe turning to Andrew Gritstone, we started touching a bit on kind of what the personalized cancer vaccine is, but maybe just kind of fill out some of the missing pieces there of kind of the overview of the approach. And you also hinted at the data that Moderna and Merck have shown, but maybe contrasting there, what's similar, what's a little bit different about how they've approached a personalized cancer vaccine? Sure. I'm first still in shock that I didn't win the alphabetical with the name Andrew. How am I normally- I went by company, by company. Yeah, okay. Fair enough. All right, putting that aside. So the notion of personalized cancer vaccine, I won't repeat it because I think we've touched on it, and you're probably all familiar with it, the basic therapeutic idea. There are two elements. First one is figuring out from a tumor sample or biopsy, of the many mutations, which ones create real neoantigens. So that's step one. Rosenberg's data suggests that 2% of non-synonymous exome mutations forms a real neoantigen. That's probably an underestimate, and so I think the real number is probably closer to 5%, thereabout, but no one actually knows the answer. There is no truth out there, so we have to figure that out. That's step one. Step two, having now done that, I now need to induce a strong, appropriate CD8 response against those neoantigens, and those CD8 T cells will traffic into the tumor and kill tumor cells. I've got 20 different targets. We've got 20 different neoantigens on our vaccine. And that obviously is really important, I think, because the thing we know about solid tumors is they're really good at developing acquired resistance. If you put a strong immune attack on a single HLA peptide complex, tumors are pretty good at throwing off lots of mutations, one of which will probably drop that HLA allele so that the tumor no longer presents that particular neoantigen. Steve Rosenberg described this, of course, as he always does before everybody else, about a decade ago in a beautiful paper in The New England Journal, of a patient getting cell therapy, this was TCR T, against a particular mutant form of KRAS, presented by HLA-C*08:02. Patient developed a resistance lesion in the liver, which they were fortunately able just to resect en bloc, and that lesion had dropped HLA-C*08:02, exactly as you would expect. So that's good news, bad news. Good news, the drug is doing precisely what it's intended to do. We're driving a really strong immune attack against that neoantigen. Bad news, tumors can mutate. So many of us, I think, feel that a real long-term success with cell therapy or vaccine therapy, and they are kind of related. You have to either kill all the tumor cells immediately or you need to have multi-targeted attack or both. Right, so that's why there's interest in this notion of synergy. So we do the multi-targeted attack piece, and I think that really matters. This notion of what is the right CD8 response, no one really knows the answer. I think it's CD8, right? There's lots of fluff and guff about, "Oh, maybe it's CD4s, and maybe..." But CD8, it's always CD8. When you look at the data, when you look at that beautiful paper you just published, it's CD8, right? There's no mystery here. We're putting class one neoantigens into our vaccine, and class one is recognized by CD8. Okay, so CD4s are important, but they're not the effectors that just do the job of killing tumor cells through recognition of a class one neoantigen. That's a CD8 cell. So we've got a vaccine system that generates really good CD8 responses. Are they the optimized phenotype? I don't know, because no one knows what the optimized phenotype is. We're getting into this notion of expression and this idea that exhausted, that stem-like T cells are the ones you want. So there's more work to be done there, but what we have seen is efficacy. So we're killing tumor cells, and so far, so good. So that's what we do, and we've built these two elements of the platform. Merck Moderna, no one knows how they do neoantigen prediction. They keep a veil over that and just mutter things about proprietariness, which is not very helpful. But that's, I can't tell you. I have no idea how they do it. BioNTech acquired a company called Neon that was one of our rivals back in the early days of this, in 2015. So Neon was a Third Rock-funded company doing similar work to us, and it was this idea that you're trying to predict neoantigens based on observed human tumor peptides displayed on the surface of tumor cells, and you correlate those peptides with the genome, so you sequence, and you then predict from sequence data which peptides will make it. And we started using machine learning, so it's, you know, we used to talk about AI. Obviously, these days, we use large language models, et cetera, et cetera, but it's all part of this prediction piece. And BioNTech bought them, so I think I know what BioNTech does for their program. I don't know what Merck Moderna does. In terms of then generating CD8s, we chose an adenovirus as our priming vector because it drives really strong CD8 responses. They were both, I think I'll use the word shackled, I think it's true. They're shackled to the mRNA platform because that's what they do, and it doesn't drive as good a CD8 response as an adenovirus. That's very clear. Does that matter? Don't know. When you look at the Merck Moderna data in melanoma, the thing everybody's pointed out is those curves don't separate till about 10 months. Now, is that because the T cells take that long to form? Maybe. We don't know because they've never published any T cell data, so we, that's a thesis, not, not a fact. What we do know is they give a lot of product, so for COVID, you received a 100-microgram injection intramuscularly times two. The cancer program, they give a 1-milligram injection, 10 times the dose, times nine, so it's on a three-weekly basis. So they give a lot of product, and I assume that eventually they get the T cells they need because that presumably is what's driving the efficacy that they've observed, but the kinetics are not yet clear. So that's how we're different. Is that better? I don't know. We'll find out. I do know that, you know, they've abandoned cold tumors. They had abandoned metastatic tumors because they just didn't see signal there. Although interestingly, about 10 days ago, Moderna started talking about maybe doing trials in metastatic disease. So that was the first time we've heard that for quite a while, so that was interesting. So it's hard to be super specific given how little information is published by our competitors, but that's the way I think in an investor's head, you should frame it in that way. Think about prediction and generation of T cells and then clinical data. And obviously, we were all cheered by the data that they demonstrated in adjuvant melanoma because it's the first for the cancer vaccine space. So that was wonderful for all of us, for sure, and particularly, of course, for those patients where, as I said earlier, they seem to be lifting the tail of the curve. This is adjuvant, right? So response is not relevant. There's no measurable disease when they treat the patients, but the key is, after 3 years, have I stopped recurrence in a greater fraction of patients, i.e., are more people out there living normal lives free of melanoma? The answer appears to be yes, by a pretty significant margin. It goes from, forget the exact numbers, but like 30% to 50%, so almost doubling. That's the goal. So kudos to them for kind of finding the path and, yeah, we hope to follow them, but in colder tumors and therefore potentially more curable. You were starting to touch on this on one of the earlier group questions also, just you had that adjuvant trial, you had the maintenance trial, you know, keep doing the maintenance trial, and maybe you'll come back to adjuvant later, but just why are those the right settings to go for this type of approach? Yeah, I mean, the basic idea should work in cold tumors. That's kind of the whole premise. We had data from our phase 1/2 study in third-line colorectal that suggested it was working, so we're following the data. And then there is the pragmatism piece about investors don't have infinite patience for novel platforms in cold tumors. There's a lot of people sort of just say, "Oh, well, you know, immunotherapy doesn't work in cold tumors," and they just write you off. And that's obviously decerebrate, but not atypically. So we, we have to kind of show this, and we have to show it quickly, and therefore, metastatic disease compresses the timelines. And as I said, we were able to do a two-arm study, not a three-arm study. So just a little bit of this is pragmatism. A lot of it is following real data, and an important driver is unmet need. So there's kind of three- But the main contrasting also the, you know, you had some early data in third line, but why, why move to the maintenance setting instead of doing large third-line trial, right? Large... Yeah. Oh, yeah, because the general vibe for immunotherapy is if everything works better earlier, when patients are healthier, better immune systems, just go earlier. And so obviously, we've been inspired by the checkpoint data as checkpoints have migrated earlier, and I think the exciting area for us is like, how do we add on to that? How do we get- Because again, it's all about lifting the tail, and the evidence is, the earlier you start, the more aggressively you hit, the broader you hit, the more we're gonna lift that tail up. And so that's always been our motivation for it, so just get as early as you can. Now, with personalized, doing it neoadjuvant, it's tricky because... you know, it's a 10-week minimum timeline for making, sequencing, making, and releasing product. Releasing hopefully will get easier, 'cause right now, every product has to undergo exhaustive testing as if it was a vaccine for 1 million healthy subjects. We're literally held to that same bar. So it takes time, and it, and it costs some money. Now, that's likely to change over time, I think, so that timeline should compress, but for some indications, you just don't have a lot of neoadjuvant window, like lung cancer, you generally, you can get 2 doses of checkpoint inhibitor before the patient has surgery. These people are just they're not yet comfortable enough sitting on a tumor in situ for a long period of time, right? It varies a bit, tumor by tumor. Breast cancer, people will, will give you months of neoadjuvant space, but you don't have that in other solid tumors, so time matters. Adjuvant is probably much more straightforward. Taylor, the CD47 surface state has seen a fair amount of up and down with it, mostly. What do you think it has been the biggest takeaway, as has that informed the development of your agent, and what do you think is still to be understood? Yeah. Well, people these days seem to think magrolimab was the first CD47 inhibitor ever developed, and it wasn't. It wasn't even the first CD47 inhibitor the company Forty Seven developed. They first developed an IgG1 backbone CD47 inhibitor that was similar to CC90002 and a series of other first-generation CD47 inhibitors that all had active Fc domains, and all were discontinued very early in clinical development due to very severe hemolytic anemia and other cytopenias. So Forty Seven changed gears, and they switched the Fc domain over to an IgG4. It is true that IgG4s have less Fc gamma receptor binding than IgG1s, and they were able to develop a priming regimen in their phase 1 studies that lessened the amount of cytopenias, and in particular, hemolytic anemia that magrolimab caused, but they did not get rid of them. You know, in their own publications, you see that on average, patients are transfused with 7 units of packed RBCs following their first dose of magro, even with that priming regimen. You know, people read the manuscripts on the Phase 1b studies with magro, and they see that over a third of patients had more than a 2-gram-per-deciliter drop. I think it was a 2.2 g/dL drop within the first week of their first dose of magro, and they misinterpret the conclusion of that statement to mean that a third of patients dropping 2.2 g/dL is the nadir of what magro will do. What that actually means is that a third of patients have a 2.2 g/dL drop in the setting of aggressive transfusion management. 'Cause the second you see that drop, you start giving those 7 units of packed RBCs to a patient. You know, now we're seeing the results of what happens when you move from a drug that is clearly a narrow therapeutic window in a phase 1b trial that can be managed in phase 1 centers, to what a drug like that does when it's being tested in a more real-world setting in a phase 3 randomized global clinical study, where not every patient is being kept overnight, not every patient is getting a CBC within a few hours of infusion, and you start to see more patients that have those 6-gram-per-deciliter drops, which can be lethal. And so I think, rumor has it that Gilead's gonna be on the podium at EHA, and I truly hope that's the case. I think it will be valuable for the thousands of patients that have been treated with, with magrolimab, for the field to understand that this is a drug, not a target problem. And, I can tell you that, you know, we are one of the companies developing CD47 inhibitors that were designed not to engage Fc gamma receptors to avoid those cytopenias. ALX Oncology is another one that made that same design choice. Trillium, now Pfizer, has actually an active Fc domain, but it's a much lower affinity binding event than, than what's seen with magro, and that seems to skirt some of the hemtox issues. Now, I can tell you, all of us, when magro was put on partial clinical hold, last August, received FDA information requests to submit all of the hemtox data we had, all of the cardiotox data we had, because they were clearly looking at something that the FDA was trying to make certain wasn't a class effect. And, you know, fortunately, all of us are continuing. So I think that by the middle of this year, you're gonna have an interesting mixture of data where ALX is gonna have what I hope to be a good update from their gastric cancer trial at ASCO. We're gonna give an update from our frontline TP53 mutant AML and higher risk MDS cohorts at EHA. You know, if those updates come, concurrent with Gilead talking a bit more, specifically about what led to the futility of magrolimab in the enhanced trials, I think we'll have another target on our hands, that will be more clearly understood. We'll go to René next. Just, even the approach of masking, there have been a number of masks tried in different settings historically, that you're not the only company there, and obviously w e've also had a pretty big update from another company in the last week with masking. Just what do you think has been proven, overall, in this approach? What are kind of still the outstanding questions that you're hoping to address? Yeah, I mean, masking's been being tested for quite a while now, and I guess, I would say CytomX is probably, like, the earliest, as far as I know, in this space. And like anything else, we learn as we go, so we learned from many who have gone before us. I think at this point in time, we at Xilio have learned a few things, having been at this now around six years. We know that the target and the mechanism is the most important thing, right? Masking an agent that doesn't have a therapeutic index issue, where you're not leaving efficacy on the table, is not useful, like a PD-1. Masking ADC doesn't really work for a number of reasons, but I think we've learned that what is the best approach is you're masking an agonist immune stimulant that has a clear therapeutic index problem, and you know if you can get higher concentration, you improve efficacy. And that, I think, is the key. So first, it's what are you masking? And of course, this is like, you know, looking in the rearview, right? It's always easy to say. Two, I think we understand now that you have to build redundancy so that multiple MMPs can activate your molecule in different settings. We know that the 3D conformation really matters, so this idea you could take the same mask and linker and stick it on different molecules over and over and get the same result is absolutely not true, because the way it's presented in the 3D conformation really matters. And we know that bringing those masked molecules, you're also working with half-life extension domains. Do they come off with the mask? Do they stay with the molecule? If you empirically make those decisions without putting them all the way through non-human primates, you will make the wrong one. I think we, we've learned that as well. So you can't shortcut, and you really got to do the work. And I think testing the actual activation in human tumors, ideally in patient setting, which is hard to do, but we've done that. In human solid tumor samples, ex vivo, before you even select your final candidate, is really critical. But I think we know it works, right? We're seeing it works, and we're seeing it can make a big difference. And you are gonna have some of your own data shortly. You wanna kind of frame up, you know, what investors should expect in terms of patient numbers, follow-up, and maybe also just lay out, you know, the changes in the cohort design as you got to this cohort, versus the prior parts of the phase one. Yeah. So we have, we have three data reads this year, two phase 2 and a phase 1. The most near-term data readout will be our IL-2. This is gonna be a relatively small phase 2 cohort, but at a very high dose. So we, we have been able to get IL-2 up to 4 milligrams per kilogram, long-term use in the outpatient setting. So I think that's really key. And we've now shown back in November at SITC that we have PD in the tumors, we have PK in the tumors, we're getting activation. So now we'll look at the first cohort of efficacy in renal and melanoma at 4 mg per kg. Be about 20 patients in that readout, and that'll come by the end of this quarter, which is very soon. The next readout we expect will be in the second half of this year, will be our safety PK/PD on our IL-12, our tumor-activated IL-12. So we're now into the third dose level. This is about 100 times the MTD of human recombinant IL-12 from historical studies, so it's a significant dose. And we've had no DLTs reported through that last update that we provided into dose level three. And then our next readout, I think, which is, you know, I think a really meaningful one, as we talked about, is in a colder tumor type. So this will be by Q4 in a combination study with our CTLA-4 and atezolizumab in MSS colorectal. We expect to have about 20 patients, and that'll be a beta initial phase 2 readout. And that will include patients with liver mets, which I think is very different than what's been done historically with this combination. And that's about 70% of these patients. In terms of the one that's coming this quarter, right in melanoma, RCC, just, what are your kind of success criteria as we see that data? So it's funny, but I think we hit on this quite a lot. I would say from the perspective of what we talked about, in a lot of ways, we've had success with our IL-2 program because we've treated now close to 100 patients. We have clear tolerability, and back at SITC, we showed a patient, a stage 4 colorectal patient on our IL-2 for 15 months with stable disease. They resolved 3 of 4 non-target lesions, and the only adverse event they had was very minimal. I think it was looking at Uli, like he remembers, like grade 1 fatigue. So that is what we're talking about because that would be unexpected course for that patient's natural history. So I think in a lot of ways, we feel like this molecule is ideal for a combination now because of the tolerability, because of the PD we see consistently in the tumors. It's really driving NK and CD8 consistently. So whether this is a T-cell engager, whether it's a vaccine, whether it's a TIL, you know, we think this is ready for combination. What we were looking at, though, also, as we talked about, is there monotherapy activity even in these sort of later line, previously IO-treated melanoma and renal patients? And is that, is there any reason to bring forward a monotherapy? I think that's really what we want to see with this data set. And that would be, you know, somewhere in a range of a 10%-15% response rate. But again, small numbers, it's about 20 patients that will be read out. The day after the PDUFA, when hopefully there's an approval, and we've got the first solid tumor TCR T-cell approval. Just how should we think about the launch here? I think we've seen as the CAR T cells have proceeded with, you know, the ability to sign up centers and work out contracting has gotten much faster. You know, just how should we think of kind of that trajectory in the first few quarters in terms of warehousing patients? Is there any warehousing of patients in sarcoma that we can think of? Often not in oncology, but sometimes there is. So, I think you hit on the key thing, one of the key things to get right with a cell therapy launch, which is the scale-up of the treatment centers. We are targeting getting to about 30 treatment centers in the first 2 years. We'll start with 6 or 10. That is still a 6-9-month process to stand up a treatment center. And so, you can imagine that we're already on that, but the first wave that we anticipate will be open around about the launch time. So I think those, for us, are... It's a very focused set of treatment centers, and they are majority of them, the vast majority of them are where we did the clinical trial for and or for afami-cel. So we know them, and they know our product well. So that's a good thing. Second piece is the supply side. So that's where you're gonna generate your initial demand. Supply side is also where a lot of existing cell therapy launches have struggled. We have the advantage, if you like, of this being a relatively modest indication. You know, we think there's about 400 patients in the United States who are the right HLA, the right target expression, for afami-cel, and they will go through those 30 centers. And so our ability to manufacture is pretty strong. We did all of the manufacturing for the pivotal trial for afami-cel out of a facility in Navy Yard, Philadelphia. and, we did all the pre-BLA work on process qualification, et cetera. We will launch out of that facility. We can scale up to maximum capacity or to supply pretty much the entire market out of that facility. So we feel pretty good about that. In terms of the timing of patients, the thing you have to recognize, don't anticipate a vast bolus of patients and warehousing in this space. I'll say there won't be any, but we don't anticipate it vast. And to the extent that there is some, we anticipate it actually being spread out and mediated through the treatment sites, because those patients are still going to need to get to the treatment sites, not being out. So it's if this isn't a bit of an oxymoron, maybe a 6-9-month bolus would be appropriate. And then we think this ramps up, but I think importantly, the best way to think about this from a modeling perspective is that the whole launch is offset by the time it takes to screen the patients, manufacture the patients, and then reinfuse, which is 2-3 months, probably, in launch. So the whole thing is offset by about a quarter. That's why we anticipate first revenues on first treatments of patients in this year, subject to the approval in August. And then also, you are working on the second generation where, you know, there are trials going on. Just kind of lay out the next updates there, that we should expect on the second gen. Certainly. So the second generation version of afami-cel, which is called uza-cel then, that product was previously called ADP-A2M4CD8 because it had a next generation CD8 alpha homodimer to give it more potency. And we saw a 35% response rate in that phase 1 trial that I talked about earlier. We're moving forward in ovarian cancer in a phase 2, potentially registrational trial called Surpass 3. That trial will recruit throughout 2024 and into 2025. And so data updates from that are probably in 2025 at the earliest. And those will be the interim data updates once we've enrolled the full trial, because I don't want to bias patient selection as we go through that trial. Behind those, in, behind ovarian is bladder cancer and head and neck cancer, and as we've spent plenty of time discussing, our objective there is to get into earlier lines of therapy in combination with checkpoint inhibitors, where checkpoint is our standard of care in various earlier lines of settings in those indications. We're recruiting patients in a phase 1 study in those earlier lines in combination with Keytruda, and we anticipate that at the tail end of this year, we'll have sufficient patients and sufficient data to be able to make a decision on which patient populations to move forward in those indications. Great. Andrew, you're, you have data coming as well, similar timeline to René. Yeah, this quarter, which is now how many days we got left? Yeah, we got a couple of days. Nine days or something. Yeah. So you want to bring up that... We've touched on the trial design already, but just, you know, like, what is good data here? The primary endpoint, ctDNA, where I think investors don't have a lot of experience looking at that as a, as the kind of main endpoint. But you'll also be reading out, it won't be as mature, but PFS. So just how should we think of what, what's a good outcome from that trial? Yeah, so it's 100 subjects, roughly randomized 1 to 1, to receive standard of care chemo, which is induction with FOLFOX-bev, and then usually after about 5 months or so of oxaliplatin, the neuropathy kicks in, down to 5-FU bev maintenance. So that's the control arm. The active arm is same induction, and then it's 5-FU bev plus immunotherapy on maintenance. As you say, we spend a lot of time thinking about endpoints, and PFS is obviously the accepted surrogate. Again, we're thinking about what's the surrogate for OS, right? Obviously, we will follow these patients to OS. Median OS is about 2 years for newly diagnosed metastatic colorectal cancer. So that's a ways off still. We enrolled the last patient last summer. So PFS, median PFS is about 1 year, 11 months. So those data, you know, the last patient will be at their median roughly in the middle of this year. So that's, by then, we should actually have a reasonably mature data set. We're a quarter short of that, or maybe a little bit more, 4 or 5 months short of that. So PFS data will not be super mature, but I think we'll be instructive. And so obviously, as you say, most investors are focused on that because they understand it. And the only question is, is pseudoprogression an issue? And that's why we've always been a little bit leery here, 'cause we did see some evidence of pseudoprogression in late-line patients. And obviously, what I'm describing is where they get initially bigger and then cavitate and shrink away. And that's been described, obviously, with immunotherapy. What we think is going on is that T cells enter the tumors, meet antigen, and proliferate. That's their job. Now I've got lesions getting bigger, which, of course, by using RECIST criteria, can only mean disease progression. Well, no, it can actually mean the T cells are doing their job, and now what follows is that the T cells kill tumors and cavitation and shrinkage ensues, and we've seen that. Quantifying it, however, is impossible. No one's done what we're doing in this setting before, so we just don't know. We were always, like, cautious about PFS. Of course, we're measuring it, but we're cautious. We looked at other surrogates, and circulating tumor DNA can now be quantified in blood, and there's abundant evidence that it is a good surrogate for survival with immunotherapy. That's why we've kind of chosen it. Now, the rules of defining response have not been established at all. This is not a regulatory endpoint, and put that issue to bed. This is a phase II decision-making endpoint. Phase II is, do you see signals that mean we should move into phase II? Phase III is a separate discussion and probably will be either an OS or, if it works, a PFS endpoint. And that's fine. We all know how to run that. The question in phase II is, am I seeing signals? So we set molecular response, meaning ctDNA reduction, as the primary efficacy endpoint, because of our concerns about pseudoprogression. And we'll see how that performs. We've defined it in a way that's not been used before in this setting, so we're in sort of uncharted territory. There will be a group of patients probably whose ctDNA actually is negative at baseline, our baseline, which is the maintenance phase. Because what happens is when they present with metastatic colorectal cancer, everybody basically has measurable circulating tumor DNA, particularly with the high-sensitivity assays that we're using. But chemotherapy is pretty good, and it comes down in just about everybody, and sometimes it's gonna come down to below the limit of detection. And that therefore, there is no baseline to measure. So those patients will not be valuable for quote, response. However, time to becoming positive again, likely to be meaningful. And Natera has published a lot of data in the adjuvant colorectal setting, showing exactly that. That when you treat patients with localized disease, with surgery and adjuvant chemotherapy, if you are ctDNA negative, you're generally gonna do well, and if you start off positive, you're gonna do less well. If you're positive after adjuvant chemo, you do worse than all. So it's all actually very straightforward. It's like measuring minimal residual disease effectively. So time to positivity, I think, is gonna be meaningful. So we've got these different ways of analyzing these patients, and the meta question is: Is there a signal there? So that's what we'll see at the end of this month, and then we'll have more mature PFS data, more complete data on these other endpoints by the middle of the year. So it's all gonna read out, and then we'll know. We'll know whether this thing works or not. So, let's assume you don't see a significant pseudoprogression signal. When we get that full PFS data, what would you view as meaningful separation of that curve? As I said to you, and I will be consistent here, the separation, of course, is likely and important, but the lifting of the tail is what you really care about. Because I think if we separate the curves by 3, if we're a small molecule that you can make for a small amount of money, that's a product that everybody can get not very excited about, but it's something that might get approved and become standard of care. You know, if you look at colorectal cancer, we have drugs like that, right? So third-line treatment, regorafenib is approved. It's a multi-targeted kinase inhibitor. It's shifted PFS by about 2 months. It's not widely used because survival is basically no different, and it's marginally better, but it's really marginal. It's a fairly toxic drug. Even the agency, we know drugs are not that great when the agency says, "You don't have to make patients progress on that drug, even though we've approved it. You can actually set up your clinical trial as an alternative to using that drug." That is a pretty clear signal of just how underwhelming the product is. So, I think really for us, shifting the curves is likely if you've got an active product, but our goal is lifting the curve. Okay. Taylor, same theme of setting up data presentations. You did start to touch on the ovarian update and the AML MDS that are coming in the, you know, next quarter to two quarters. Just wanna bring up what's the data there? Yeah. So we submitted abstracts to EHA for both, the TP53 mutant AML and the high-risk MDS studies. The update we gave in December on TP53 mutant AML, we had 14 patients on study. We had 11 patients worth of follow-up bone marrow biopsies at that point in time. And TP53 mutant AML patient treated with azacitidine has a CR rate less than 10%, has a true CR less than 10% of the time. And a TP53 mutant AML patient treated with azacitidine plus or minus venetoclax lives for a median of 6 months. So we wanna see, as I said earlier... at least a tripling of the true CR rate, and we need to see OS trending well beyond, you know, 6 months, and something 9 months or greater would be interesting. And about 10%-15% of these patients get effectively bridged to transplant on aza plus or minus ven. So if you can significantly improve the proportion of patients that are getting bridged to transplant, then that portends a good OS benefit. We expanded, we amended and expanded that study in December, doubling the size of the patients that we had on study. We've since completed enrollment of that expansion. Clearly went quickly. So we'll add all those patients in the EHA update there as well. In high-risk MDS, that's also included in the EHA update. We had 24 patients on study. Fourteen of those... In the December update, we had CRs and marrow CRs in 9 out of those 14 patients, which is well above the 22% CR rate you expect for azacitidine alone, especially in a predominantly TP53 mutant high-risk MDS population. So we want to maintain that. We want to be, you know, double or triple, the CR rate that you expect for aza alone of 22%. And, that's interesting but medically irrelevant if it doesn't come with, a duration of response that's trending beyond 9 months, and more than half of our patients will be beyond that 9-month benchmark, by EHA. On ovarian as well? Ovarian. So, yeah, we had in the Doxil combo, we have now 3. Well, the patients that we reported on in November, we had 1 confirmed and 2 unconfirmed PRs out of the first 14 patients. Those unconfirmed PRs subsequently confirmed. We finished enrollment of the cohort, 21 patients last quarter. We need to maintain greater than a 25% response rate, and for those, the duration of response in those patients to be 5 months or greater, for that to be interesting in the PROC setting. On the mirvetuximab arm, which is the other combo in ovarian, we're enrolling patients across the spectrum of folate receptor alpha expression, and we're looking to improve upon the composite response rate that mirv has in folate receptor alpha low and mid patients, which is about 22%, as well as in the high subgroup, where it's 42%. And we'll be updating on all of those mid-year as well. Okay. Thanks. Unfortunately, it's all the time we have, so we're gonna have to cut off the conversation. But thanks, everyone, for joining.