Thanks for inviting us to join you today, Corinne, we appreciate it. So, Gritstone is now at a very interesting stage in its evolution. We're about eight years old, and we have, the GRANITE program, which is personalized cancer vaccine in metastatic colorectal cancer, which is a potentially game-changing, opportunity. Because we've seen immunotherapy, of course, be wildly effective in a variety of solid tumors, but really mostly limited to the so-called hot or inflamed tumor types, PD-L1 high, melanoma, MSI tumors, subsets of lung cancer, gastroesophageal, and so on. What we haven't seen yet is success penetrating into the colder tumor types, which unfortunately are the majority. Most cancer deaths relate, to so-called cold tumors, and don't respond very well to immunotherapy. So that's a big opportunity now in a randomized phase II trial to deliver data, opening up this really important and substantial opportunity for patients and commercially. And then off of the same platform, we have an infectious disease business, which obviously recently was validated by the U.S. government, as BARDA has awarded us a contract to run a 10,000-patient randomized phase II-B study. This is a COVID vaccine study with a clinical endpoint of PCR positive symptomatic disease. And BARDA is offering a contract of up to $433 million to fund that study, which will compare our product versus one of the approved vaccines. And that, if all goes well, will begin in the first quarter of next year. Behind that, of course, we have many other infectious disease pathogens that one can make vaccines against, and there's growing interest in the field about multiplexing, putting multiple pathogens into a single vaccine. The obvious example these days is respiratory viruses. So we have to have currently three different shots, one for influenza, one for SARS-CoV-2, and one for RSV. And, I think that one of the attractive concepts is to put all of those into a single product. And self-amplifying mRNA, which is our platform, lends itself remarkably well to this concept of multiplexing, all being done at relatively low doses, which is good from a cost of goods perspective. So we have two different programs, one in oncology, one in infectious disease, both in advanced randomized trials. Very exciting setup for us, and potentially, obviously, the commercial opportunities are not that remote now. Great. So I'd like to focus today on GRANITE, given there will be an important catalyst for that that program next year, early next year, I believe. So let's start with the approach. What is the proposed mechanism of a cancer vaccine, and GRANITE in particular? So what we've learned from all of the other data, from checkpoints and so on, is that key effectors, the cells that you want if you're a cancer patient, are killer T cells or cytotoxic CD8 T cells that recognize your tumor neoantigens. That's the cell type that really sits at the heart of the successful immunotherapy that we've seen. So CD8 T cells, obviously, most people know what those are. Those are the killer forms of T cells. They're hard to make. Let's be clear, most vaccines don't drive very good CD8 T cell responses. So when you think about vaccines, it's not enough to make just T cells. CD4s, the helper cells, are relatively easy to stimulate with a vaccine. You need CD8 T cells, which limits the numbers of vectors that can actually be used to really drive that CD8 response. And then secondly, you want them targeted to neoantigens. Neoantigens are the name given to mutations in the tumor, which create slightly altered proteins that look foreign to your immune system. And of course, your immune system is designed to recognize and kill foreign cells because the immune system assumes that they're virally infected cells. Now, they look foreign because they've actually mutated their DNA and created this abnormal surface protein, the so-called neoantigen. So that is the exquisitely, well-designed target for cancer immunotherapy because those targets are only found on tumor cells. So if you can generate strong CD8 T cell responses targeting those tumor-specific neoantigens, in principle, you will eradicate tumor cells everywhere, every cell, and drive long-term clinical benefit with very, very low toxicity. That's really the goal for the GRANITE product. Right. So you've selected colorectal cancer, which you mentioned is the initial indication. Maybe walk us through why you think that's an attractive first target for this program. Yeah, obviously, some people say to us, "Well, shouldn't you be focused on hot tumors?" And of course, hot tumors have responded well to just checkpoints, and the reason for that, we think, is that the T cells already exist. So if you have melanoma, if you're unlucky enough to have cancer at all, you wanna have something like melanoma because we've got really good immunotherapy for melanoma now. Because you, naturally, you will have T cells forming spontaneously that recognize tumor neoantigens, and they just need an extra kick from a PD-1 antibody, let's say, to really go back into action and kill tumor cells. But in the colder tumors, as I mentioned, colorectal, ovarian, prostate, most breast cancer, the big, solid tumors, generally patients do not have those naturally forming neoantigen-specific CD8 T cells. And so our entire thesis really is that we should be able to, to drive real benefit in those patients if we can give them the right CD8 T cell population. So it really is the right test of the hypothesis, where effectively, it's not quite placebo-controlled, but immunotherapy alone really just doesn't work at all in those patients. And so you've got a very clean canvas to paint on. It gives some benefits in terms of trial design, which we can talk about. It's a very big indication. You know, it's the second most common killer of humans in the United States, and has seen very, very few advances of note in the last 20 years. Immunotherapy has just passed it by. But we're going there not because of hope or driven by commercial opportunity. We're going there because it is where our therapy should work, and indeed, in our phase I/II first-in-human studies, we saw very clear and strong signal of efficacy in advanced colorectal cancer. We published this in Nature Medicine last year. And so for all of these three reasons, we've seen signal, it's a big opportunity, and it's where our thesis should play out. That's why we've chosen colorectal cancer. Great. So maybe you've kind of alluded to it, but how do you define the addressable patient population there in the frontline maintenance, colorectal cancer setting? How many patients are there? What are they currently getting, and what's the level of effect that the existing standard of care provides? Yeah, so there, there are about 55,000 people a year dying in the U.S. from advanced colorectal cancer. So we're at steady state, relatively speaking, so that's kind of roughly the number to be thinking about. Another way to think of it is about 20% of patients who are diagnosed with colorectal cancer are diagnosed with metastatic disease. So it is a common tumor. Sadly, it often presents already metastatic, and even for those who are diagnosed with a localized disease, who have surgery, about a quarter will progress to develop metastatic cancer. So it's a big problem, and five-year survival is less than 20%. Today, when you're diagnosed with metastatic disease, you are treated with chemotherapy, not chemoimmunotherapy. Immunotherapy's been tried, has not worked to date in the form of just PD-1, PD-L1 antibodies, et cetera. So you're given simple chemotherapy, and the standard treatment for most patients is to receive 5-FU and oxaliplatin. That's given in a regimen called FOLFOX, together with Avastin, bevacizumab. So that is the standard form of treatment, chemotherapy. You typically get about five or six months of the FOLFOX plus bevacizumab, and then you have to stop the oxaliplatin because of neuropathy. You drop the oxaliplatin and continue on 5-FU and bevacizumab maintenance, and that's for about another six months or so. So about five months of induction chemo, six months of maintenance chemo, then median progression-free survival is around 11 months for this indication. And that's this, the setting where we're running our randomized phase II trial now. Great. And I guess, kind of moving from the next question I have is, this is a modality that has obviously, you've alluded to it, it proves challenging historically. What are some of the challenges that they faced as cancer vaccine developers? And then what are some of the features of GRANITE that sort of try to overcome these historic challenges? If you could talk through those, that'd be really helpful. Yeah, it's a great question. We've obviously seen a lot of cancer vaccine failures, and I think as you, as we look back on it, it's not surprising that they all failed. There's two real dimensions that we think about that matter. First is the antigens. You know, what am I targeting? The second is the nature of the cells that my vaccine is driving. We've had problems historically on both of those dimensions. If we start with the antigens, people generally worked using what we think of today as tumor-associated antigens. These are normal proteins found in normal cells that are often upregulated on a tumor. But your immune system does not really care about the amount of the antigen, and so just because there's a bit more of a protein on a tumor cell, does not make it a good target from an immune response perspective, because we have this magnificent phenomenon called tolerance. By design, as a fetus, we become tolerant to our normal self-proteins. Of course, we have to, otherwise we'd end up just rampant autoimmune disease. And so tolerance is a huge problem. If I'm tolerant to HER2, why do I think that putting HER2 in a vaccine will be able to drive a strong HER2 T cell response that matters? If tolerance is real, you basically shouldn't be able to, and in fact, often those T cells won't exist. They were deleted because they were autoreactive. And so self- proteins are a problem, hence the excitement when neoantigens arrived. Neoantigens are not normal self, and we couldn't really understand them until we could do next-generation sequencing for each individual patient quickly and cheaply, and that arrived in around 2015. So prior to 2015, we couldn't really think about using neoantigens as targets 'cause we just didn't have a way of deriving them from sequencing. Now we do. That's a total game changer. And then secondly, as I said, you've got to drive CD8 T cells, we, we believe. CD4s historically are not the cell type that matters. And if you look in tumors responding to checkpoint, one of the strongest predictors of response is the presence of pre-existing neoantigen-specific CD8 T cells. That's telling us something. So what vaccines drive CD8 responses? And we looked at the literature. We didn't have a particular platform to work with when we started the company, so we had the luxury of choice. The literature gave a very clear signal. Adenovirus is the strongest vector to drive CD8 T cell response in higher species in humans, and it can prime CD8s, meaning you can be immunologically naive to these antigens, and an adenovirus will prime a CD8 response. So we used an adenovirus as our priming vector, but we've also developed a self-amplifying mRNA as a boost. It is not yet as potent as adenovirus at priming, so let me be clear about that. That's why we continue to use the adenovirus, but it does boost very effectively, and we have reason to believe it may be more potent than, than mRNA, so when you combine self-amplifying mRNA as a boost on top of a potent adenovirus prime, that cocktail, referred to technically as a heterologous prime boost, potentially can drive very, very strong CD8 T cell responses against the encoded antigens. And for us, those are the neoantigens. So that's the platform that we've built. Very helpful. How does EDGE function, and how would you compare the product to the other algorithm-based neoantigen-targeting vaccines in development? EDGE is the tool that we built to predict which mutations create neoantigens. So it's important to recognize this point. Tumors have hundreds of mutations typically, but only a small number, maybe 5%, actually function as targets for the immune system, as antigens. And I wanna be clear about language, 'cause there's some fluffy language used here. So we have a whole set of mutations, then we predict, and we have candidate neoantigens. But something is not an antigen until you have a T cell response to it, or an immune response, let's say, 'cause antibodies for other antigens, but in our case, T cells. So three different categories: mutations, candidate or predicted neoantigens, and then true neoantigens, because you know that there's a T cell response to it. So we built a platform of EDGE to try and help us work our way through this complexity. So when we started the company, there was a way to try and do this, where you would plug in all the mutations, and the system, called netMHC, would try and predict which ones would function as antigens based purely upon whether the mutant peptide would bind to the HLA molecules, which are those tissue-type platform proteins displayed on every cell. The peptides, these short mutated peptides, are presented by HLA, so you have to think of them as a complex. And the original approach, this netMHC approach, had very, very low positive predictive value, and just wasn't fit for purpose. So we recognized this challenge, and we built our own system called EDGE, and the way we did it was to take real human tumor samples, to do a technique that isolates the peptides, these antigenic or neoantigenic peptides, off of the surface of those real human tumors. So this is truth, right? These are mutant peptides displayed on the cell surface, and then we sequence those same tumor samples, DNA and RNA, and we let deep learning mathematics figure out the genomic features that predicted probability of peptide presentation. So I guess we were early users of what's now loosely termed as artificial intelligence, and our model, EDGE, has continued to iterate. We now use large language models, because of this ability to actually bring different data points together in a complex fashion and understand relationships between different data features. We put all of that together, and we now have this prediction model, where all you have to do is sequence a tumor, plug the sequence data into the EDGE model, and it predicts which mutations will be neoantigens with over 80% positive predictive value. So it's working extremely well for us. Now, your question about comparison is interesting. We don't know how Moderna and Merck do their prediction, and they never talk about it, except to say it's proprietary. So we can say nothing about that comparison, and they don't publish T cell data, so we don't really know how effective they are at making T cells against their candidate neoantigens. We do have some data from BioNTech, our other major competitor in the space. There was a paper published online in Nature earlier this year from Memorial Sloan Kettering in the adjuvant pancreatic cancer setting, and they showed some data there that in 16 patients with localized pancreatic cancer, each of whom received a 20-candidate neoantigen vaccine, this is the BioNTech vaccine, given weekly intravenously times eight. They observed T cell responses to the administered neoantigens in half of the patients. So eight of the 16 had no measurable T cell response against the administered neoantigens, and eight of them did. And of the eight that did, half of them had a T cell response to one of the 20 administered candidate neoantigens. So that's the BioNTech data set. What we've published, this was in our Nature Med paper last year, is that when we vaccinate our subjects, we see strong T cell responses to neoantigens in everybody, and we see it against typically 10-15 of the 20 administered neoantigens. So by using that metric of measured T cell response, we clearly significantly outperform the BioNTech product in this cross-trial comparison. Now, of course, I can't tell you whether that relates to the prediction of neoantigens or the vector that you're using to try and drive the T cell response, 'cause you need both of those things to take you from nothing to a neoantigen-specific T cell response. They have to work together. So when we're just measuring the sort of output, I don't know the relative contribution of prediction and vector. But, you know, ultimately, the output is what matters. Right. Okay, that makes sense. You've previously demonstrated some data with GRANITE, so maybe you could walk us through the clinical data that you've presented to date for that platform in colorectal cancer in particular. Yeah, so what we showed is that we're sort of checking the key boxes for the entire concept. So in colorectal, we're starting with cold tumors, so these are PD-L1 negative, TMB low, interferon gamma signature low, no T cells infiltrating. When you look in the blood of the patients, you cannot find neoantigen reactive T cells. So that's the start point, clearly cold tumors. We vaccinate in the way I've described, and we do use atezolizumab in our randomized phase II study. And we do two little shots of subQ epi adjacent to the vaccination site. So we do combine with a checkpoint, particularly the continuous background of atezolizumab now. What we showed in the phase I/II is that, we're able to prime the CD8 T cell responses, as I described, in every patient against the majority of the administered candidate neoantigens, so at least 10 out of the 20 administered. Those T cells are in the blood, and then we showed that those T cells traffic into tumors, and that they proliferate in tumors. This is done using an on-study tumor biopsy. And so we were able to show that these T cells get to the right place. Then we're able to show that the T cells kill tumor cells. The way you can show that is to use a marker of tumor mass, which is circulating tumor DNA, and basically, with effective therapies, ctDNA goes down as you kill tumor cells, and with ineffective therapies, ctDNA goes up. So what we saw was what we term a molecular response, i.e., a ctDNA decline, in half of the patients that we treated with colorectal cancer. If the patient had an elevated CEA, one of those old-fashioned protein biomarkers, that typically paralleled the ctDNA exactly. So very clear evidence of direct tumor kill in half the patients. Parenthetically, why only half? We don't know. Obviously, we're looking hard at that. We'd like to obviously increase that number. But today we've looked at the obvious explanations, and none of them seem to be pertinent, so we're exploring that as key research effort for us. But half the patients responded, and then most gratifyingly, over time, as the data matured, what we could see is that the half of patients who had molecular responses had extended overall survival. The median overall survival in that group of responders was not reached, but was over 22 months, whereas the half who did not have molecular responses had a median survival of about 7.5 months, which is exactly what you expect in the third-line colorectal cancer. So it's not a randomized trial, obviously, and the numbers are, are small, so let's be clear. However, everything is lining up and suggesting that the product is doing exactly what it was designed to do in approximately half of those patients with third-line colorectal cancer. It was really taking that set of insights that we moved into a randomized phase II setting in frontline metastatic colorectal cancer. That's a perfect segue for me. So looking forward, we anticipate results from the phase II portion of that GRANITE study, the study in frontline maintenance colorectal cancer next year. So maybe just first, can you set the stage for us? What's some of the basics of trial design? How many patients would you expect to see at that data update? What's the primary endpoint that you're looking for? And what's the study kind of powered to show, if that's a function here? Yeah. So we designed this as a randomized phase II/III, and people have forgotten a little bit, I think, about phase IIs and randomized phase IIs. So obviously, these are not small phase IIIs, right? You generally use a different endpoint. If you just run a small phase III with a standard phase III endpoint, it's an underpowered phase III study. That's not interesting to anybody. So you need a surrogate endpoint that predicts overall survival, 'cause it's clear that the phase III primary efficacy endpoint will be overall survival. It's the one thing everybody cares about. And in colorectal cancer, from time of diagnosis of metastatic disease, median survival is just two years. It's depressingly short and has not changed much. Very different for MSI, that little subset with high mutational burden, they respond beautifully to checkpoint inhibitors. So there's nothing inherently resistant to check to immunotherapy in colorectal cancer. You just have to get the right T cells and the right antigens. So we're up against overall survival, and we're looking for a good surrogate. And as I said, with immunotherapy, radiology is clearly not a very good surrogate for survival. We've seen that many times now. Lots of checkpoint studies where PFS and OS are disconnected. That leads to lots of confusion and hanging approvals and so on, dangling approvals, to use the FDA's jargon. And Immunocore has given us the best example, where a product which had a dramatic effect on overall survival, so immunotherapy T cell engager, had really no meaningful impact on RECIST radiology compared to an effectively inactive control arm. And so we needed a better surrogate for survival, and circulating tumor DNA is proving out to be a much better surrogate for OS. We chose ctDNA response as our primary efficacy endpoint for this randomized phase II part of the phase II/III GRANITE study. The good thing about ctDNA response is it is what's called a categorical variable, meaning you're either a responder or a non-responder. Because it's just binary, you have much greater statistical power. That was the endpoint that we set. It's a frontline metastatic colorectal study, so the way it works is that we randomize subjects at time of diagnosis. We randomized about 100 subjects. They receive induction chemotherapy with FOLFOX bev primarily, as I said, and that's symmetrical across the two arms. We're making the product for the active arm during that time period. As they reach the maintenance phase at around five months, then the treatments diverge. The control arm just gets maintenance 5-FU bev, which they've been on from the beginning. But now the active arm receives 5-FU bev plus our vaccine, plus monthly atezolizumab, and those two little doses of [IP] just to sort of try and light the fire initially. So that's the regimen. Median PFS in the control arm, we anticipate 11 months. So that's the time period we have. And we're measuring ctDNA monthly from the start of maintenance. We've set that first ctDNA measurement at the beginning of maintenance as the baseline, and a response is defined as at least a 30% reduction in ctDNA. What we expect across the two groups, well, in the active arm, what we saw in third line was a 50% molecular response rate. So conservatively, I think, let's say 50% is expected in the active arm. In the control arm, because their therapy is de-intensifying, the ctDNA, which may have come down a little bit during induction chemo, is expected, based on everything we know, to have hit a plateau and be flat or actually even rising as they begin the maintenance stage. Therefore, the ctDNA response rate in that control arm should be low, let's say 10%. So 50 versus 10 is a delta of 40%, and you need less than 50 subjects to detect that kind of delta with 90% power and high probability of a statistically significant result. So this is the context we're in. You actually with roughly 100 subjects, you can pick up a 20% delta, much smaller than we anticipate, with good power. So I think given the nature of the endpoint we've chosen, we're well-powered for a positive result here. Okay, so then to confirm, when we get these results next year, will we have all 100 patients that were enrolled in the phase II portion? And then how should we think about the maturity of progression-free survival at the time of that data cutoff? Yeah, so in Q1 of next year, so very soon, we anticipate showing data from approximately 50 subjects. So that is likely to be meaningful, given everything I just described. And then as we reach the middle of the year, we should have complete data from just about everybody. We randomized our last patient in July this year. So obviously, do your 11-month math, that takes you to a median PFS of around the middle of next year. So by mid-next year, we should have pretty good data on the entire population. PFS, we're a little bit leery of PFS because there's clear risk of pseudo progression. With immunotherapy, you're driving T cells into tumors. We've shown we actually very clearly do this. We also saw in phase I too, that sometimes tumors get bigger in the short term, presumably because of T-cell infiltration. Using RECIST response rules, which were, of course, developed for cytotoxic chemotherapy, any expansion of a tumor or even a lymph node is regarded as progressive disease. There's a clear risk of mislabeling patients who are actually doing very well and have got florid immune responses with T-cell proliferation, could be labeled as progressive disease, an erroneous label, and that's referred to as pseudo progression. We worry about that. I don't know how big a problem it's gonna be because no one's ever done what we're doing before, but we did see evidence of pseudo progression in the phase I/II study, so it has to be something that you treat with respect. Also, with immunotherapy, PFS and OS are often not very well correlated, so I'm not quite sure what to expect on PFS. We will show the data, but I don't know how useful it's gonna be because I don't know how much of this pseudo progression problem will be pervasive. That's why we chose ctDNA, 'cause the data are very clear from lots of different settings. If you're treated with an immunotherapy and your ctDNA goes down, you will do much better than if your ctDNA goes up, and using our own data, we set that 30% threshold. The other truism is that the deeper it goes, the better you will do. So of course, we're gonna be looking at that, and we'll be presenting all of those data, starting in Q1 of next year. And if we're successful, as I say, this is a really important study to the field, because nobody can touch metastatic colorectal cancer with simple immunotherapy at this point. Lots of people have tried. Nothing has moved the needle, you know, in terms of overall survival. And so, if we're successful with this therapy in metastatic colorectal cancer, in principle, it should work in other solid tumors, other cold solid tumors, both in metastatic and of course, in the adjuvant setting, which is a place we're very much looking forward to going. Helpful. I think you walked through a lot of my questions on the trial, but I'll finish with this before we move on to next steps. As you think about kind of risks, you've got this readout coming, what keeps you awake at night in terms of risks to the trial's success, and what are some of the things that you're keeping your eye on? Yeah, well, because we saw the product doing everything we wanted it to do in advanced disease, generally, immunotherapy is gonna work better in less advanced disease. So that is obviously encouraging for us. We have more time, and the patients are healthier, and they have more time to develop the necessary immune response. We did do a switch from nivolumab to atezolizumab, and we did that because atezolizumab was actually tested in frontline MSS colorectal as maintenance by Roche in a big randomized phase II study that convincingly showed it did absolutely nothing, PFS or OS. It's called the MODUL study. And that meant that when we went to the FDA a couple of years ago to discuss this trial, we talked about whether we needed a third arm, right? You've got the chemo arm, the control arm. You've got vaccine plus checkpoint. Do you need a checkpoint-only arm? Logically, obviously, in principle, you do, and we knew that. The good news for us was that Roche had done that trial, and they'd shown that atezo alone, in exactly the same maintenance setting, really didn't work. Physicians we were talking to said they felt it inappropriate to repeat that because the question had been answered, and happily, the FDA agreed. So that's why this is a two-arm study. But we did make that switch from nivo to atezo. So as I think about reasons why the trial might not replicate what we saw before, it's pretty hard to come up with good reasons. The only one I can come up with is that, and it's not expected at all. Obviously, nivo and atezo perform very similarly in many different settings. So, basically, I sleep well. That's good to hear. Okay, so as you mentioned, this study is part of a larger phase II/III design. So walk us through kind of the next steps pending you see what you'd like to see, with the results next year. So if we're positive on the randomized phase II, we go to talk to the agency in mid-2024, and then if all goes well, we move into an overall survival endpoint phase III study. And obviously, it's an unambiguous endpoint. No one's gonna discuss that at all, and we just get started, right? We randomize patients 1:1, and off we go. The interesting part is whether there's an opportunity for accelerated approval. So I think getting the study going, you're not gonna be thinking about this too much. But as we enter 2025, our phase II survival data will be maturing, and what we may show, and frankly, I expect this, is that ctDNA response predicts extended overall survival. If we can show that in a randomized setting with our therapy in metastatic colorectal cancer, now you're starting to potentially substantiate ctDNA response as a regulatorily acceptable surrogate endpoint sufficient for accelerated approval. So remember, the language of the statute is that the surrogate endpoint has to be, quote, "reasonably likely to predict clinical benefit." Now, it's vaguely worded to give, obviously, the agency a lot of discretion, but there's a lot of data out there now in many different settings showing exactly that, that ctDNA response, reduction in ctDNA using an analytically valid assay, that's an important caveat, is reasonably likely to predict clinical benefit, i.e., extended overall survival. If we're able to build on that existing dataset and show that that is true using our randomized trial data, then I think you can have an interesting conversation about accelerated approval. With the FDA's sort of revised position on how to run those trials for accelerated approval, obviously we're pretty well-tailored to their expectations, which is you begin a study, it's randomized with an overall survival endpoint, and the study should be meaningfully, if not fully, enrolled before accelerated approval is granted, ideally with no crossover, because crossover is what often obscures potential survival benefit. Obviously, we would be worried about crossover 'cause we seem to have learned that our therapy, even if administered late, drives a significant survival benefit, so crossover might be toxic to a clean trial outcome. Therefore, we would approach the agency with an ongoing randomized phase III with an OS endpoint, with the potential discussion around accelerated approval for that very study. We'd then continue the study to get confirmation of the overall survival data, no crossover. So that's kind of a good setup. You know, that's a conversation for 2025, so we're a little ways off still. But that potentially is a way to, if everything's working, to bring this forward and help patients. Because, you know, when you spend time talking to colorectal cancer patients, the frustration for them is that treatment just hasn't changed, and once you have metastatic disease with a five-year survival of less than 20%, you are desperate for something better. And as you're probably aware, there's been an increase in the incidence of colorectal cancer in young people, so this problem is becoming ever more acute, and there's a desperate hunger for something better, and obviously, we profoundly hope that we're able to offer that something better here. Helpful. Since you brought it up, I'd be curious if you could just remind us all what the FDA's current kind of stance is on ctDNA. I know they've issued some guidance around that, so could you just remind us what that is? Yeah, so FDA obviously is paying a lot of attention to ctDNA. They, they think of it as very useful for patient selection, so as a stratification tool. As you're well aware, there's a lot of data that, particularly in the adjuvant setting, this is where there's a huge amount of data now, actually a lot of it in colorectal cancer. As I said, so colorectal is a very common tumor, and a lot of patients are present with Stage 2 or Stage 1, Stage 2, or Stage 3A disease. In other words, fairly localized disease that, in principle, can be surgically removed with curative intent. Now, if you've got Stage 1 disease, most surgery is curative happily, so that's great. But Stage 2 and Stage 3A, many patients, unfortunately, even though you, you resect the tumor in the local lymph node bed, those patients do recur. And so adjuvant chemo has had a complicated history in colorectal cancer, where, you know, it's, it's long-term chemotherapy, where many patients just don't need it. And so we're, we're looking for tools to try and focus the adjuvant chemo on people that really need it and spare those that just don't. And ctDNA is seeming to be a very helpful tool here, where if you have your surgery for localized colorectal cancer and you are- you remain ctDNA positive post-op, that group of patients will recur relatively quickly and clearly are likely to benefit from adjuvant chemotherapy. And so there have been some big studies presented, including at ASCO this year, suggesting that if you're ctDNA negative post-op, you do not need to receive adjuvant chemotherapy, and you will have the same outcome as, you know, patients treated where everybody gets adjuvant chemo. So in other words, you're using ctDNA as a tool to guide treatment with adjuvant chemotherapy. So that is well established now in the regulatory paradigm, and lots of trials are being run using ctDNA as a stratification tool. No one yet has used ctDNA response as a surrogate endpoint for accelerated approval, and the agency, I think, is waiting to see more data, as they should. So, you know, it's on people like us to generate those data, take the data to them, follow the paradigm, right? You have to have an analytically valid assay, so in other words, you're measuring exactly what you think you're measuring. You have to show that it has clinical utility, which I think is largely been shown in colorectal cancer now. We know that, you know, up is bad and down is good. And then the clinical validity piece for us will be showing that a change in ctDNA correlates with a hard clinical endpoint of overall survival. So that's why today it's not something we're thinking about, but go fast-forward a couple of years, absolutely, I think we should be thinking about this as a potential surrogate endpoint. And the FDA's been clear about the roadmap of how to get there, so we're on that journey, and, you know, hopefully it'll bear out. I think it will, because frankly, we need it with immunotherapy, because radiology is proving to be very problematic. Not surprisingly, again, because radiology is all based on the simple idea that good therapy kills cells quickly, kills tumor cells quickly, and shrinkage of lesions is all I need to think about. And with immunotherapy, that just isn't true. Obviously, we're driving cells in, they're proliferating, and these are the good guys. So lesion size is not a direct proxy of benefit, which doesn't mean that over time those lesions may not fade, but in the short term, there's a real disconnect, lesion size and true clinical benefit. And so radiology is having all kinds of challenges. Maybe, and I think likely, ctDNA is gonna solve those problems. Great. You made reference to adjuvant colorectal cancer. I'd be curious if you could expand on that point and also talk through any other indications you think would be kind of interesting to continue pursuing personalized cancer vaccine development in. Yeah, so the data with checkpoint suggests that as you move earlier, the magnitude of effect increases, and perhaps most importantly, cure becomes more likely, 'cause that's obviously what we're all looking for. So adjuvant's very exciting. The data obviously in lung cancer adjuvant, where we've got a lot of data and neoadjuvant, obviously, as well as melanoma, is suggesting that there's real benefits to moving the immunotherapy early. And I think this general truism is you want, you know, your best option for cure is your first treatment, whatever that is, whether that's in the localized setting or the metastatic setting. You want the best possible first treatment. And from what we know about cancer, chemotherapy, and immunotherapy, the treatment that has the best chance of driving long-term, let's use the word cure, is immunotherapy. That's really kind of why we got into this space in the first place. So adjuvant immunotherapy is absolutely where we should be focused. Now, there are big studies running in lung cancer and melanoma with just checkpoints. Melanoma, actually, most of them have been done now. And of course, we should talk about Moderna here, that earlier this year showed a strong signal in a randomized phase II in adjuvant melanoma. And their signals in advanced disease were much softer, but they got a nice, clear signal in adjuvant melanoma. With their partner, Merck, they've now launched a phase III in adjuvant melanoma and a phase III in adjuvant lung. And I think that's a rational thing to do. It's a smart thing to do because there you have time. Patients do need time to mount the immune response and for the immune response to then clear those remaining tumor cells. So adjuvant is an exciting place to go. In colorectal, for example, one of the challenges with adjuvant trials has been that when you enroll a colorectal adjuvant study, many of the patients are actually cured by the surgery. And so they're in your trial, they're receiving therapy that always carries some toxicity, and they're never going to contribute an endpoint of recurrent disease. And so they make trials big, and slow, and inefficient. ctDNA is changing all of that because now we can run an initial adjuvant trial in exactly the population you want, i.e., those who are ctDNA positive, who are going to recur, who are gonna recur quickly, and who need help. So it's a very attractive idea, and Roche/Genentech, BioNTech, they have a collaboration on their personalized cancer vaccine, and they're running exactly that trial, an adjuvant colorectal gated on ctDNA positivity. So this is, I think, the natural direction of travel. You know, we're a smaller company. We didn't have the time, to be honest, to do an adjuvant study properly. We actually did start one at the same time we began our metastatic trial, but it became clear early on that we had to focus our resources, and we had to focus on the one that would give us data faster. So that's why we stopped the adjuvant study and focused on our maintenance study. Also, of course, we got the data out of our phase I/II, and it looked so strong and clear that we are confident about the metastatic setting. So that's where we've gone for POC initially, but I think the world really is going to be about adjuvant and perhaps frontline metastatic as well. Because, of course, many patients present with metastatic disease, and you don't want to leave those patients behind. So I think adjuvant and frontline metastatic, that's gonna be the future, using the new tools like circulating tumor DNA to drive efficient trials, where you get results relatively quickly that have huge impact on the patients that really need that benefit. Great. I'll maybe close out with a question from the audience that you sort of alluded to here in that response, which is: How should we think about cash runway and your ability to kind of finance next steps for these programs? Yeah, so we've got data coming in Q1. We currently have cash in through Q4 2024. So we're set to obviously get the data, and then obviously, ideally, I think we'll partner. You know, to run a series of phase III trials in parallel is our objective. We won't be able to do that on our own. We're gonna need a partner. If our data look good, I think partners will find this very interesting. Yeah, immunotherapy's been a big, big driver of pharma interest. It's perhaps stalled a little bit because many of the second-gen checkpoint inhibitors have not really opened up new space in the oncology treatment landscape. You know, we're seeing TIGIT as a clear example, and LAG-3 as well, where those antibodies do have activity, but it seems to be basically limited to PD-L1 high patients who are gonna do well on checkpoints anyway, and they can do a bit better if I add in an additional checkpoint. But they're not opening up the space in the colder tumors. So in lung cancer, the huge indication that everybody focuses on, if you're not PD-L1 high, if you're one of the 2/3 who are, you know, PD-L1 less than 50%, you don't do that well on pembro chemo. You don't benefit, as far as we can tell, from things like the addition of an antibody to TIGIT or LAG-3. And so you're kind of stuck. You're gonna progress on pembro chemo, and then it's second-line docetaxel, and then, you know, most people, unfortunately, still will pass away. So that's the huge opportunity, the unmet need that we really want to service. But to do those kinds of trials, we clearly need a partner. But I think if we show proof of concept in this setting, in metastatic colorectal, I think that's gonna be a really strong signal that we've done something really important. We've kind of cracked the nut on colder tumors. We've done it in a rational way, and none of this is particularly fancy, right? It's simply the idea that I need neoantigen-specific CD8 T cells. People who have hot tumors already get those naturally, but people with cold tumors don't have them. So I've got to give them neoantigen-reactive CD8. How do I do that? Oh, with a vaccine. None of this is that complex at some simple level. That's what we're about, and I think pharma is likely to find it very exciting if we're able to show positive data in Q1. Great. Well, I think with that, that brings us to time. I really appreciate the time here, Andrew and the whole team at Gritstone, for all the work that you guys are doing. Thanks to everyone who joined us via the Zoom. As a reminder, we're here all week, so continue dialing in for these Catalyst Clinic events. Really appreciate it. Bye.