All right, well, thank you everybody for joining us. We're gonna, we're gonna jump right into it. We've got Gritstone management here, with Andrew Allen, CEO here. Thank you so much for joining us in Miami, and— Thank you, Jon. Yeah, look forward to talking about Gritstone and your upcoming programs. Let's start with just very briefly, heading into 2024, what do you think people should be focused on? And seems like a pretty obvious answer to that question, I think. But, Is that a softball question, don't we call that? That's a softball lead-in, and then we'll get into the nitty-gritty in a sec. Yeah, so, we have data, preliminary data off of our, randomized phase II study with our personalized cancer vaccine in frontline metastatic colorectal cancer, expected in Q1. So that's obviously our near-term catalyst. And then we have this collaboration with BARDA on a, on Project NextGen for a second-generation vaccine against COVID, which obviously has interest in its own right, since that remains a, you know, north of $10 billion global market. But of course, really acts as an exemplar of potential superiority of our self-amplifying mRNA platform over other platforms, and though that trial will be enrolling through 2024, hopefully starting in Q1. So progress against that goal is the other key objective for us next year. Excellent. Well, let's focus on that neoantigen vaccine that you mentioned. Granted, this is obviously, as you say, the near-term focus, but it is the key issue for investors on Gritstone at this point, especially given the partnership with BARDA on infectious disease. There's robust investor debate on the neoantigen vaccine field right now, how to interpret Merck and Moderna's data in adjuvant melanoma. How appropriate do you think it is to read across from one neoantigen program to another? How differentiated do you think these various personalized cancer vaccine programs are from each other, both yours from theirs, but also the field more broadly? I think there's some read-through. Obviously, I think it would be foolish to say there's none, but it's pretty limited. The product is different. You know, these are sophisticated products that have several elements. The two key elements are prediction of neoantigens and getting that right, and secondly, using vectors to deliver your predicted neoantigens that prime strong CD8 T cell responses in humans. And those are both complex areas, and I believe our product is highly differentiated from Merck-Moderna's. Although, when I say that, I'm basing that on my semi-informed speculation because they don't really publish any data as to how their product actually works. All we see is some clinical phenomenology in a randomized trial, which is very encouraging, but I can't really see behind it. So, it's really, it's really hard to know just how correlated that is with what we expect to see. But nonetheless, what they showed was very convincing. It was a hazard ratio of 0.56 against recurrence-free survival in the adjuvant melanoma setting. Now, obviously, we're in a cold tumor, and that's a little different from melanoma. Of course, the product is designed to work in a cold tumor. At some level, actually, the risks of the melanoma study are not trivial for a personalized cancer vaccine, because pembrolizumab, the control, works really well. So, given that what we know about checkpoints is that they work in patients who've got a preexisting neoantigen-specific CD8 T cell response, the question for a cancer vaccine in that setting is, can I add to that, presumably by either making T cells against new neoantigens that are hitherto unrecognized by the patient's immune system, or in some way I'm augmenting the immune response? That's kind of a different question from the question we're asking in colorectal cancer, which is: Can I make an immune response that doesn't exist? And at some level, that's actually an easier question to contemplate because it's kinda like running a placebo-controlled study. Yeah. Right? The bar is lower. So, it's fundamentally different. I think we can all agree on that. We obviously have precedent data in colorectal cancer in third line, which is an even tougher context, and we saw a very clear signal of efficacy. So we're not being hopeful about the current study. We are simply asking the product to do what it's done before. And if you actually look at the Moderna study and ask the question: Do they have any information there at all on cold tumors? There is perhaps a little bit of information, r ight? Because there is a PD-L1 negative subset of melanoma, and it's a small subset, so let's acknowledge that up front. Nonetheless, in that PD-L1 negative subset, the effects of the vaccine were very striking. The hazard ratio was much stronger than that observed in the overall population. It was 0.17. So for what it's worth, which may not be much, given that that's, like, 20% of the patients, the effects in the cold subset of melanoma were really striking, which the thesis would suggest, actually, when you look at those curves, the reason that that hazard ratio was perhaps so dramatic is that pembro doesn't work very well in cold tumors. At some level, there is actually some analog there for what we're trying to accomplish in colorectal. One other interesting thing that came out of the ESMO update of Merck Moderna's data is the suggestion that the bulk of the benefit, regardless of PD-1 or positivity or negativity, is actually being driven by BRAF mutant patients, which struck us as particularly interesting because that's not something we typically associate with being a particular driver of response to pembro here in the melanoma setting anyway. But does that make sense to you? Do you consider those patients to be relatively colder than the general melanoma setting? No, like you, I that was surprising to me, and so that could just be noise, right? Obviously, we're into post-hoc subset analysis. You know, it goes back to my med school training, when you looked at the old ISIS trials of thrombolysis, that if you were Sagittarius, you did really badly. I think these are the hazards of retrospective subgroup analyses. But, so I was surprised by that, and I think obviously you can wave your hands and say, "Are there differences in the immunogenicity?" Again, we have no data to inform that speculation. Were the T cell responses as good, not as good? I don't know. What they alluded to in that presentation was that there were more—there was a mismatch, an imbalance of PD, of ctDNA positivity in that subgroup- and that there were more ctDNA-positive patients in the BRAF wild type, in the active arm. And that group, prognostically, will do less well, they'll progress faster. And they inferred, I think, that that might be explaining at least some of that observation. But again, we're into subgroups of subgroups here, so I think repeat the study and we'll see. So moving on to your GRANITE readout in the first quarter of 2024. So I believe we're expecting 50 patients worth of data, greater than four months of therapy. Can you just walk us through the expectations of the trial? Which of the efficacy endpoints, ctDNA, PFS, iPFS, like, what do you expect to be most robust? And can you walk us through, like, the various scenarios and how you're thinking about the bar of success internally? Like, what if one endpoint shows a signal, but the other endpoint doesn't have— Sure. Anything about that? So the question really for us as we design this phase II study is, you know, what is the right phase II endpoint? So let's be clear, this is not intended for registration, and people get a bit confused about that. So we are not using ctDNA for registration. Let's be clear, right? That's a separate question. We're looking for signal in a phase II study that will catalyze initiation of a phase III, and likely for us, I think, partnership with pharma. So the question is, what is the best surrogate for overall survival? Because it's likely that the phase III study is gonna have a survival endpoint of some form, either as a primary endpoint or at least as a confirmatory endpoint. So OS is what matters, we all know that. So the question is, what's the best surrogate for OS that I can power a phase II study to deliver against? Now, historically, objective response rate has been useful, but what's become clear with immunotherapy is that's not a good surrogate for overall survival. And so, and neither is PFS. And, and you know this because lots of, lots of, studies have shown this dissociation with immunotherapy between PFS and OS, and everybody's basically a little bit lost right now in terms of what's the best surrogate. The good news is, circulating tumor DNA is emerging, with lots of manuscripts now showing this, as a much better surrogate for overall survival in patients treated with immunotherapy. So now we have something that I think is a good surrogate. So that's how we designed this study, using ctDNA response as the primary efficacy endpoint. There are different ways to measure ctDNA. We don't have time to go into details, but you can either use the standard panels, where you're looking for known drivers, like the Guardant panel, looking for KRAS and P53 and so on. Or you do what's called tumor-informed, where you actually sequence the tumor, you pick a set of mutations that you're gonna follow longitudinally over time. The latter is the approach we're taking. It's what Natera does, if you know that company. It's a much more sensitive assay, and that's obviously for good reason, I think, the right one to be using in this setting. We've set a threshold of 30% reduction in ctDNA as a response. That is based on our data. When we looked at our phase I/II data in advanced disease, a 30% delta was the best predictor of overall survival benefit. Others have used a similar cutoff in this, in this field, but there's no very well-established benchmark at this point. But at least with our data, 30% reduction looks to be the best choice. And what is clear is that down is good, up is bad, further down is better, right? So that's pretty unambiguous. So that's the primary endpoint. Now, of course, we're gonna be measuring PFS, and we'll get those data. Now, patients, we start their baseline ctDNA assessment when they enter the maintenance phase of treatment. So just to quickly recap, we randomize patients at the beginning. Everybody on the two arms of the study, and it's a 1:1 randomization, gets induction chemo, mostly FOLFOX/bev. Some young patients will get FOLFOXIRI bev. Typically, you're on that for about five months, and then neuropathy kicks in, and you stop the potent chemotherapy, and you then continue on maintenance 5-FU bev. That's the standard. So that's what our control arm gets. The test arm or active arm, at that point of moving to the maintenance phase, we initiate immunotherapy on top of 5-FU bev, okay? We start the ctDNA clock, as it were, at that initiation of maintenance point. What we observed in our single-arm study in advanced disease was a 50% molecular response rate. So I think this, these are healthier patients, and we've got an optimized regimen. So I think 50% is probably a conservative number, but let's go with that. We're expecting to see something like a 50% molecular response rate in the active arm. In the test arm. Sorry, in the control arm, we're expecting to see basically zero because those patients have been on their chemotherapy for five months on average by the time they enter the maintenance. They're at a nadir of ctDNA at that point anyway. The data suggests that chemo does drop ctDNA, but it's quick. It happens within two-three months. So you'll hit a nadir, and then you're gonna start to rise. And the Natera data suggests your ctDNA goes up about 6 months before radiologic recurrence. Median PFS in this disease is 11 months. 11 minus six is five months. Guess what? That's exactly when they're starting maintenance. So the control arm is expected to have hit their nadir at month two, three, and then be flat or rising as they begin the maintenance. Therefore, the molecular response rate in the control arm should be close to zero. Now, never say zero. Let's call it 10. So we're expecting 50 versus 10, something like that. So a pretty significant difference. That's our expectation based on everything we've seen hitherto. In terms of PFS, the worry, and this is an mostly an academic worry, is about pseudoprogression. Are people gonna get T-cell proliferation in lesions such that the lesions expand by 20% and they hit progressive disease criteria per RECIST, and they get labeled as PD, even though actually they might be doing very well? That's just a worry we have. I don't know if that's gonna happen or not. We will have those data. We will share those data when we present, the ctDNA data. So it's possible it's all gonna be easy and great, and everything will be fine, and the PFS curves separate nicely. If there is pseudoprogression, we'll know that because we're monitoring ctDNA. It'll be a more complicated presentation, so we have to worry about that. Do we have enough time to put it all together into a 10-minute oral or something? But I think the truth will kind of shine through the data. You are using iRECIST, iPFS. That's also being used, yeah, as pre-specified, yeah. Which presumably helps with that pseudoprogression issue a little bit. It should, based on checkpoint data, but no one's done what we're doing. So we're just in the unknown here. This is the price of innovation. Makes sense. Makes sense. Besides the concern of pseudoprogression, like, how should we think about the scenario in which ctDNA reduction endpoint hits in a stat sig, but PFS and even iPFS is not stat sig? Like, how do you— Well, then you should say, which is the better surrogate for OS? And you know the answer to that. It's ctDNA response. So rationally, you will say, "I believe in this product on this trial", but the emotional part of you will say, "But I love PFS. I'm so used to it." So you're gonna have to wrestle with that. Well, let's hope that the market behaves rationally and not emotionally. That's right, 'cause we're well-known to, obviously. Will the FDA agree with that as well? Yeah, I mean, the FDA is pretty unemotional. I think one would generally align behind that statement. So, most important to us actually is pharma, because obviously for us, you know, partnership to really expand the program, assuming we see positive data, which is what we anticipate, we need pharma to believe in ctDNA, and they do. I mean, pharma's publishing on this stuff, as you're probably well aware. So when I say that ctDNA is a better predictor of OS than PFS, that's based on data from AZ and Roche and Regeneron and many companies that have been working with checkpoints and have themselves kind of seen the limitations of radiology as applied to immunotherapy. Before we wrap up with some of the infectious disease stuff— I want to make sure to ask about your SLATE program, which is not— Yeah. A personalized neoantigen vaccine. It's a more of a fixed SLATE of, of neoantigens. It's been developing a little bit slower. You've been obviously prioritizing the near-term readout, GRANITE, but how are you positioning that fixed, neoantigen vaccine program at this time? But beyond bandwidth limitations, what's driving your focus on the personalized program? Yes, good question, Jon. So everybody loves off-the-shelf because obviously it's easy to make, and could be used instantly in a neoadjuvant setting without waiting to make product. The challenge with an off-the-shelf is that in the ideal world, you want to deliver multiple relevant tumor-specific antigens to each patient, and you cannot get there if you rely on classical driver mutations from the exome. In other words, KRAS and TP53 are rarely shared, and when they are shared, the chances that you have the right HLA molecules to present both of those is even lower. And so the amount of sharing, if you rely on those classical driver mutations, is incredibly low. And therefore, the current version of SLATE that we used really delivered a single KRAS neoantigen to patients. We know that an immune response to a single neoantigen can be productive, but can also be mutated around by a tumor. Steven Rosenberg published this in a beautiful paper in The New England Journal a few years ago. Cell therapy against KRAS G12D, patient developed acquired resistance because they dropped the HLA allele, happened to be HLA-C*08:02, that presents G12D. They had a liver lesion. Happily, it was an isolated recurrence. They resected it, and that liver lesion had done exactly what you'd expected. It dropped the HLA-C*08:02 selectively. The other HLA were intact, but that one was gone. So that's the worry, has to be the worry. With GRANITE, we don't have this problem because we deliver multiple neoantigens. So if you want SLATE to work as well as GRANITE, then you need to deliver multiple tumor antigens. Now, to get there, you're probably gonna have to get into the dark matter of the genome, and that's a whole rich topic we could talk about. Suffice to say that there are a lot of companies forming now, starting to look for tumor antigens from within what's called dark matter. And what that means is the non-exome part of the genome, things we don't often talk about, but things like, alternative splicing, where introns suddenly appear in proteins, or introns and exons are spliced out, and you get new exon, exon boundaries that create novel antigenic substrate. Endogenous retroviruses can be reactivated. Strange fusions can happen. There are these non-canonical, coding sites that appear in tumor genomes. So tumors just mess up their genomes in all manner of ways, and there's a lot of stuff on the surface of a tumor cell that is not found on a normal cell. That's where we—everyone's now looking for shared antigenic material to start creating off-the-shelf products that deliver multiple relevant tumor antigens. So my expectation is that over the next decade, and I think it's gonna take a bit of time, we will end up with off-the-shelf products that are, that are as potentially as good as the personalized. It's just gonna take us a bit of time. Makes sense. In the last couple of minutes, let's return to the infectious disease program. You mentioned the partnership with BARDA, that study getting started next year. You mentioned your self-amplifying RNA program there. We've seen others look at similar approaches with self-amplifying RNA, but haven't seemed to take off. Notably, BioNTech looked at that for their first-generation COVID vaccine, but eventually dropped it. And then Pfizer, well, BioNTech, Pfizer, yes. Pfizer would be upset if I didn't mention them. What's driving your enthusiasm for samRNA here? And what part of the data that BARDA saw was most compelling to them? Data is what drives our enthusiasm. Self-amplifying mRNA, as we have seen it, and we have seen no published data from Pfizer-BioNTech, so again, it's a hard question to answer. But what we have seen is that we induce antibodies that persist for at least 12 months, clearly differentiated from first-gen products. We include T-cell epitopes in our vaccine that potentially drive T-cell responses to conserved viral antigens, such that your immune response is not gonna be lost just because spike keeps mutating. That notion of variant-proof immunity is very important to Project NextGen, the BARDA campaign. And then thirdly, and very importantly, we're able to give low doses and potentially combine low doses of different products to multiplex. So this growing excitement around the idea of putting multiple different pathogens into a single vaccine is something that you can do, or we can do with self-amplifying mRNA, because we can put several different products together. We have big payload capacity. We can do two different 5-microgram doses of two different products into a single syringe, 10-microgram total dose, but 5 micrograms is still very well immunogenic, and still, that 10-microgram dose is well tolerated. So this flexibility of the platform, its ability to generate strong T-cell responses, to drive variant-proof immunity, and the persistence of neutralizing antibodies, these are three distinctive elements, all of which I think BARDA found attractive. Interesting. Well, we are out of time, unfortunately, but we're looking forward very much, obviously, to seeing the GRANITE data first quarter next year. As do we. Thank you, Jon. Thanks, Jessica. Thank you.