Hello? Yeah. Good. Hi, I'm David Berman, and I'd like to welcome you to the Moderna ASCO Oncology Event. Here are our forward-looking statements, and you can read them at your leisure. Here's the agenda for today. I'm going to give you an overview of intismeran. Michelle Brown from Moderna is going to highlight the neoantigen selection and the mechanistic foundations of intismeran. We're very honored to be joined here today by Dr. Matt Carlino, who will give a reprise of the phase II KEYNOTE-942 study that he presented today. Dr. Ryan Sullivan will join us. He presented translational data on intismeran yesterday at ASCO, and so he'll be sharing that data. I'll conclude, and we'll have some questions and answers. For those of you I haven't met, I'm the Chief Development Officer, and I joined Moderna about three months ago.
About 20 years ago, I had the honor to participate in the world's first checkpoint antibody for cancer. That was ipilimumab. Over the past seven years, I've had the honor to participate in the world's first TCR therapy for cancer. I joined Moderna three months ago because I believe intismeran has the potential to be the world's first RNA immunotherapy for cancer. You'll see why I am so enthusiastic today. intismeran is the most advanced individualized neoantigen therapy in development. In brief, a tumor and blood samples are taken. They are sent to a lab for sequencing to identify neoantigen mutations. These neoantigen mutations are then designed into an individualized mRNA. We manufacture one lot per patient, individualized. This is then shipped back to the investigator. The investigator then administers this. You'll hear more about this from Michelle in a few minutes.
We've begun to develop a stepwise experimental framework for understanding how intismeran works. First, of course, is the production of intismeran, and as mentioned, Michelle will walk you through this in detail. Once intismeran is produced and injected into the muscle, the lipid nanoparticle and intismeran migrates to the regional lymph node, where it's picked up by dendritic cells, converted into peptides, which are presented on the surface of the dendritic cells, and then the dendritic cells activate neoantigen reactive T cells, so T cells that are specific for the neoantigens encoded by intismeran. These neoantigen T cells then circulate in the blood, and hence, there is an increased clonality in the blood. And the T cells then migrate to where the tumor is and identify those neoantigens and kill the tumor. This is the theoretical framework.
Personalized or individualized vaccines have been studied for several decades, predominantly as peptide, adjuvanted peptides, and as DNA. It has been shown that novel T cell clones in the blood can be produced or induced as a result of a personalized vaccine. However, until intismeran, no one has ever demonstrated efficacy, and especially not in a randomized trial. Today, as mentioned, Dr. Carlino will present the five-year updated efficacy in the form of RFS, DMFS, and overall survival, from our randomized phase II trial. He will also take a step to the left and show you that we do see an increased level of novel clones in the blood, and he will show that on the intismeran randomized arm, that patients who are relapse-free, who don't relapse, have higher levels of T cell clones in the blood than patients who do relapse.
This provides a mechanistic link between intismeran and the efficacy. The outstanding question to date has been, what are these novel T cell clones that you'll see in Dr. Carlino's presentation? What exactly are they? This is where Dr. Sullivan's presentation will come in, and he will complete this loop by showing that these novel clones that appear in the blood are, in fact, T cells that react to specific neoantigens encoded by the intismeran product. Therefore, we have this complete circle, and to me, someone who has worked in immunotherapy for several decades, this is the strongest proof of confidence for any immunotherapy, of course, prior to a randomized phase III trial. Why have prior therapies not worked? Why have prior individualized vaccines not produced the results, and why do I think intismeran works? First, of course, is the intismeran drug product.
It's a modified mRNA in a lipid nanoparticle. We know that this type of product is highly immunogenic to induce T cells. We've known this for COVID, for flu, and for RSV. We now know it as well for intismeran. The second is where intismeran was studied. If you think about it, prior studies were in heavily pretreated patients with metastases, usually multiple lesions, and one lesion of many is usually biopsied to make the individualized product. However, in this late stage, there's probably intralesional variability. Producing a neoantigen vaccine from one tumor may not represent other tumors in the patient. Second, of course, the T cell fitness in heavily pretreated patients is not good. Finally, patients with metastatic disease usually have somewhat rapid progression, and this limits the time for the vaccine to take effect. All of these are eliminated in the adjuvant setting.
There is one single dominant lesion that is sequenced for the vaccine that's removed, allowing maximal T-cell fitness as good as you're going to get, and there's enough of a runway to produce the intismeran drug product. Despite the successful development of anti PD-1s in the high-risk adjuvant setting, there is an unmet need. For example, bladder cancer, 64% of patients treated with anti PD-1 will still recur, and up to half of patients with adjuvant high-risk melanoma will also recur. You can see even in other tumors, there is a significant high risk of recurrence. Unmet need still remains. Pancreatic cancer is here in gray because, of course, anti PD-1s are not approved here, but this remains, in the adjuvant setting, an area of high unmet need.
This is why intismeran was initially focused in the adjuvant setting and in tumors that are known to be IO sensitive, because IO sensitive means that checkpoints work here and checkpoints unleash T cells to target neoantigens. Therefore, it makes sense to study intismeran in adjuvant settings that are IO sensitive, melanoma, lung, and genitourinary tumors. We also have trials recently initiated, for example, the stage I phase III non-small cell lung cancer, which is the first time that we're actually studying an intismeran arm that's monotherapy. Then we have our phase II non-muscle invasive bladder cancer. We decided to also study this in the metastatic setting. Now, this is a higher risk for the reasons I articulated, but it is certainly worth studying in a phase II trial.
We have also decided to explore intismeran in tumors that are generally less sensitive to immune therapy, gastric cancer, and pancreatic cancer as well. These are single-arm trials that we hope to share results for soon. I'm now going to call Michelle up to talk about neoantigen selection and the development, how we make intismeran.
Thank you, David. Hi, everyone. It is nice to see you. It is nice to see you at the end of ASCO. We're very excited to have this as one of our third type of conference events around an intismeran data set, and so we're very excited to see Dr. Carlino and Dr. Sullivan showcase the data that was presented over the weekend. As David alluded to, he highlighted the unmet need that still remains in adjuvant melanoma and the adjuvant setting and tumors overall, and also the basic principles behind why we think intismeran can be extended into the type of clinical portfolio we have today. One of the things that we think about, and I think it's important to highlight with intismeran, is that it is not just a traditional clinical product.
It really is a process, and it's a process that sits at the culmination of mRNA LNP advancements on genomic medicine with next-generation sequencing, digital bioinformatics, small-scale manufacturing and really, it is because of all of that that we're able to generate a personalized approach that we think has broad applicability to a number of tumor types. My intent here is to double-click on some of the technologies that funnel in to intismeran to be able to make it as applicable and where we're advancing and making learnings. As you see here, one of the starting point, as David alluded to, is that with intismeran, it starts and ends with the patient.
This is very true with the leverage of the patient's tumor tissue and their blood sample, and then the use of next-generation sequencing to understand the wide variety of mutations that are actually present in a patient's tumor. Using that litany of information, we can understand what mutations the tumor is using to generate these neoantigens. What neoantigens are those aberrant tumor mutations that are recognized by the immune system. It's not enough for a mutation just to be present. It also has to be expressed, and so we use RNA-Seq to understand or predict the expression profile of these mutations, and then also HLA typing to understand how the immune system is going to recognize these aberrant proteins.
When we take all of the genomic information, the RNA-Seq information, and the data analysis, we leverage all of that data inputs into integration and interpretation with our bioinformatics algorithm to actually manufacture intismeran. It's a different way of looking at the little cycle that you guys are used to seeing. As you see here, one of the foundations building on all of the inputs for the next-generation sequencing is the algorithm itself. The algorithm, you have to think about it as a node for this technology, because what it does is it takes all of that input, reviews the litany of mutations that are present, and basically says which out of these are going to be the ones that are most able to mobilize the immune system.
It then rank orders the amount of neoantigens and puts them in and selects the top 34 that we think will activate the immune system to generate de novo T cell clones and increase endogenous T cell responses. This type of mechanistic hypothesis is the foundation of what Dr. Ryan Sullivan will be presenting today. If you believe this, what it means is that right now the current algorithm is well-controlled and is essentially applying the same basic rules and principles for every intismeran patient that we have. What we know is that science advances and technology advances, and we have a litany of clinical studies, as David alluded to, that are going to drive learnings. While the algorithm itself right now is fixed, it doesn't mean that we don't have an approach where we could adapt to those learnings and have it iterate on itself.
In fact, we're in conversations with the agencies to understand how we could adapt the algorithm based off of learnings today. A lot of people ask, this seems like a relatively complicated process. There's a lot of steps, there's a lot of arrows. Do we really have to do this? The answer was presented back in SITC in 2024, where we essentially looked at the patients in KEYNOTE-942, so the phase II study that Dr. Carlino will be discussing, and looked at their intismeran antigen cassette. What we found in that was that the majority of intismeran plus pembrolizumab patients had a full 34 cassette. 91% of the patients that were treated had 34 neoantigens. You can see that there's a span. It goes from nine up to the 34.
That's not surprising because melanoma tends to have a lot of tumor mutations, we have a lot to choose from and select from. What was interesting was that 99.1%, over 99% of the neoantigens that were selected were unique. They did not overlap between patients. For that less than 1% of patients that did have overlapping neoantigens, these were BRAF mutations and NRAS mutations. What we know is that the algorithm is actually accounting for your traditional biological relevant mutations and then selecting hotspots that are relevant for those specific patients. It's important to note that even though some of these were overlapping neoantigens, the actual neoantigen itself, like the cassette frame, was different because it's based off the patient's biology, it's based off their HLA type, and it's based off the specific sequence for themselves.
Even though they're targeting, let's say, BRAF, it's different parts of that BRAF that are tailored to that patient. Really what this tells you is that we do have to go through all of the steps to generate an individualized neoantigen therapy because no two patients and no two tumors are the same. That funnels into the mechanism which I think you've seen before. The idea here as David alluded to, is that we use all of the digital infrastructure in the next-generation sequencing to program and select these 34 neoantigens, which then integrate seamlessly to our manufacturing site to generate an mRNA cassette and encapsulate that into an LNP, which is then administered IM.
Once the intismeran enters into the body, it tends to go to the lymph node where it uses natural cellular processing to generate peptides that are expressed on MHC-1 and MHC-2 to activate CD8, CD4 cells, which then generate effector immune memory phenotypes. Essentially this was our hypothesis for how intismeran should work, and it's essentially saying that we're training and activating the immune system in a targeted way. The thought was that if we combine this with a PD-1 inhibitor, you're essentially taking the brakes off the immune system and mobilizing and maintaining activity of these targeted T cells. One of the things we've been so excited about is not only the clinical data that's being presented, but the translational story that's following this mechanism.
I think it's important to double-click on that translational story because what we talk about a lot of the time is the clinical data set, the clinical portfolio. One of the foundations with this type of technology is, again, the amount of learning that we are going to be able to have. It is important to note that because of the litany of clinical studies we have in collaboration with Merck, we do have a lot of translational learnings to guide our future states. We do have, across the studies, multiple sample collections basically from every patient that comes on. We have a tumor sample for sequencing. We have a blood sample and this gives us baselines, but it also can give us baseline about their biology, their biological state in the blood.
Within each study we have a number of samples that are being collected to generate additional learnings. That can teach us on the mechanisms of recurrence or progression. It can also teach us about the use of ctDNA, or really how the neoantigen and the immune system are interacting and interfacing. All of these learnings can then be put back in to the platform to help us with either future clinical trial designs or, again, future intismeran design. Right now, if you take a look at the totality of the clinical studies that we have, what it would mean is that we would have a repository of over 5,000 patients that we would have samples for to help us learn, which represents the largest portfolio for a neoantigen approach to really drive those learnings.
I think it's important to know, as David alluded to, that we do have a very large portfolio of studies, and each one is addressing different questions. Where we started in this immune sensitive foundational tumor, as Dr. Carlino will talk about, with adjuvant melanoma for the highest probability of success, we then expanded to tumor types that had biological adjacencies that would make sense with the mechanism that we talked about. After that, we started exploring the bookends, because really this type of approach is tumor specific. We wanted to test how early can we go, especially with the safety profile we see, and that's one of the reasons we're so excited for the stage 1 study, but also for the NMIBC study.
How late can we go, because maybe we really do have the ability to activate that immune system in a metastatic way. Lastly, we're able to do this with multiple combinations, so we aren't just combining with the PD-1, we are combining with standard of care where it makes sense, because again, the mechanism is distinct and the safety profile is distinct. All of this really started with the phase II P201 study that we presented back in 2023 at AACR with the primary analysis, and what Dr. Carlino is going to talk about today with the five-year analysis, which represents the longest duration we have for follow-up for this type of technology. Thank you.
Thanks, Michelle. Hi. My name's Matt Carlino. It was exciting then to think about the concept, but the last few weeks have sort of closed the loop, both presenting this data, but also visiting the production facility at Moderna and seeing the change from a trial that I think we put 15 or 20 patients on at our site and 107 patients collectively, to now an infrastructure that's ready to make thousands, hopefully for commercial product very, very soon. This is the slide deck I presented this morning. I'm happy for people to interrupt. I'll also present it slightly differently given that some of the information overlaps with David and Michelle, and maybe have some commentary as I go. ASCO makes you do your key findings first.
The key finding of the study was intismeran pembrolizumab really works. It reduces the risk of recurrence by 49% or 59% for distant metastasis-free survival. To put those numbers in perspective, I think the benefit of pembrolizumab over placebo is smaller than that. One of the things I was taught when I was reading papers as a trainee was the easiest study to do was when your study is of an active drug versus nothing, or an active drug versus an inactive drug. A study of placebo versus an active drug, placebo versus pembro, or an active drug versus an inactive drug, dacarbazine versus nivolumab. It's actually harder to get a big benefit when you're comparing yourself against something that's effective.
I think those numbers are more effective when the comparator is actually pembro, not placebo or we weren't silly enough to give adjuvant dacarbazine ever. The second take home point is the safety profile I'll touch on when I get to the safety slides, but I think when it comes to adjuvant treatments, there's really important safety factors that this does that other adjuvant trials hasn't done. The last point I'll come to is that David and Michelle touched on is the translational data that really links how intismeran is designed to what we're seeing. This is clearly on target benefit in these patients. I think the translational data that's, for want of a better word, a little bit superficial from this that I'll show you to the really deep dive that Ryan will show you, closes the loop of the story.
Michelle, you stole my slide, or maybe I stole your slide. Just reminding you of the mechanism action. The goal is to expand existing and create new T-cell clones, targeting those up to 34 neoantigens that's in the intismeran, and we'll show you that that's actually what's happened. The theory is by giving the pembrolizumab, you're potentiating that response. You're, if you like, getting that response and then driving it forward with the pembro. As I said, background, nivo and pembro both work. In the adjuvant setting in stage II, stage III, and stage IV, they reduce the risk of recurrence by 40%-50%. That's how I, and I'm sure Ryan's the same, describe it to patients. If your risk of recurrence started at 60%, roughly we're making it 30 with pembro.
That means whatever we're doing, a significant proportion of patients still develop metastatic disease and still die. An improvement on this matters, because someone who gets adjuvant pembro or nivo and subsequently recurs has a very high chance of succumbing from their disease. I think the other thing to remember is no trial has improved on this. I've been involved in four studies that have tried to improve on pembro or nivo. One of them stopped early, the drug was so bad. Three of them were negative, including two agents that were positive in the metastatic setting. The addition of ipilimumab and the addition of avelumab don't improve nivo. I think the other thing about those two studies is both of those agents increase tox. In the adjuvant setting, the thing I don't like is permanent toxicity.
pembro, relatlimab, and ipilimumab all drive permanent toxicity, and the one that bothers me the most is patients I make diabetic or patients I give cortisol deficiency to. I guess the alternative trials that we filed and failed had the added negative of causing the toxicities we don't want in the adjuvant setting. The previous analysis were positive. The job today was to update the five-year data and give that beginnings of translation data. Here's the study design. There was a question today from the audience, or maybe a comment from the audience, which was a compliment to me, but really a compliment to Moderna and a slap in the face to everyone else.
Jeff Sosman is the most famous question asker in melanoma, and he made the point that this study was so important to be a randomized phase II, and the other two people I shared the stage with had single arm studies. I think that's right. If Moderna had done 200 patients with intismeran and pembro, and then tried to compare it to some historical control, rightly, we would be more skeptical, or you should be more skeptical, we all should be more skeptical than a randomized phase II. I think this is a perfect example of the power of randomized phase II trials. Not to get a drug approved, but to say it's worth and it's likely to win in phase III, and it's worth going forward with. It was quite a high-risk population, which I also think was another smart move.
We enrolled stage 3B patients, they had to have recurred first, and actually there's no stage 3B patients in the study. By going for a high-risk population, you can get away with 150-patient study, and you get an answer early. There's nothing biologically different between stage 2 melanoma and stage 3. It's just you can get an answer with a small study, and I guess from a Moderna point of view, a quick answer with a relatively small investment. Patients needed to have tissue available to make or design and make intismeran. That's something we were really worried about early on. The investigators thought, oh, gee, we're going to need large chunks of tumor to get to make this thing. We learned in this study, in the subsequent phase III study, that's not necessarily the case.
A third of the patients on this study actually had micrometastatic disease, they didn't have big lumps of tumor. It was microscopic. Certainly when we all see the results from the phase III study, probably two-thirds of patients had microscopic stage 3 disease. You don't need big lumps, you do need tissue. Patients have tissue. Even stage 2 patients will have blocks of tissue available. Am I going too slow?
No.
Okay. This was the primary endpoint, relapse-free survival. Couple of things, what I touched on before. A hazard ratio of 0.51 is really impressive against an active drug. I imagine, and you're not allowed to do this, so I'll do it. If you put the placebo curve from the 054 study down there, and you looked at the equivalent patients, not the whole 054 study, the 054 is pembro versus placebo. If you put in the placebo curve, I think I calculated it, that it probably sits at about 30. You're going from 30 to 50 to 72. That really, it's an additive benefit. Not many drugs do this. I was once going to talk when we developed Well, it was well before my time.
When people got hormone therapy for breast cancer, tamoxifen versus nothing was really impressive, Arimidex versus tamoxifen wasn't that much better. Same with Herceptin. Herceptin versus nothing is fantastic. Adding in Perjeta, small benefit. This tells you, unlike ipilimumab, unlike relatlimab, this is a new class of drug, and that's why you're getting that significant delta. The other thing to say is a delta of 23%, an absolute benefit of 23%, or if you put the placebo curve in, an absolute benefit above 40%, 40% absolute is kind of huge. This slide really just showed that the benefit is durable and maintained. First analysis, the hazard ratio actually gets better. Often with trials that have transient benefit, the first ASCO presentation, the hazard ratio is as good as it's going to get, and then it gets worse.
Typically what happens is ASCO presentation one, best data you'll get. Poster number two, because you don't get a second oral presentation, the hazard ratio gets worse, you don't even bother with the third presentation. This is the opposite. Improves maintain benefit in terms of relapse. Acknowledging this is a small study, I think this gives us confidence that it's not just one small group of patients we're helping. The combination was better if you had stage 3 disease, stage 4 disease, PD-L1 status, BRAF status, left-handed, right-handed, high tumor mutation burden, et cetera. I think it's really promising going into the phase III study that we're unlikely, I think, to have a population of melanoma patients that aren't going to benefit from the incremental benefit of intismeran pembro. Distant metastasis-free survival. This has a ratio of 0.41.
Got a lot of talk, I think probably at the first presentation. Another thing that you often see if you look at an adjuvant study, the relapse-free survival hazard ratio is typically better than the distant metastasis-free survival ratio. There aren't many studies where distant metastasis-free survival is better. In some ways, if as a patient, what do you want, right? If you said you have to have a recurrence, certainly I'd choose a local one. A drug that shows a greater benefit when it comes to the stuff that matters, this means liver mets, brain mets, lung mets, in some ways. I'm really interested to see if this pans out, and certainly biologically, there's an argument why this immune surveillance may be helping with distant metastasis.
Once again, an absolute benefit of, I think this is 18%, is huge in terms of distant metastasis-free survival. Incrementally, that's bigger than the benefit of pembro over nothing. Overall survival is a really high benchmark. To give you an idea, KEYNOTE-054, pembro versus nothing, there's still no overall survival available. It's at nine years, right? To see even a hint of an improvement in overall survival, hazard ratio of 0.47, the reason this crosses 1 is the lack of events. If this pans out, we may be getting an improvement of overall survival with intismeran versus pembro when pembro versus placebo has none, which will kind of be interesting for maybe we won't be allowed to use pembro unless it's given with intismeran. Okay. Side effects. A couple of things about side effects for me.
I'm not particularly concerned that any treatment related AE, in some ways, there's an incremental slightly higher number with the combination versus the single agent. That doesn't bother me in the slightest. These incremental increases relate to pretty predictable side effects, and these will be better manageable today than they were on the study. Remember, when we were enrolling these patients, it was either no patient had ever seen an mRNA vaccine because it was pre-COVID, or it was post-COVID, where everyone who got febrile made us panic, and we had to lock them up if they were febrile. I think having slightly higher treatment related AEs is of not a lot of consequence, particularly when we know what they are. The most important point to me is this, the red box, immune related AEs being equal or at least numerically potentially lower.
When I give adjuvant treatment, how I describe it to patients and the wording I use is, "If you get insert name of side effect here and it makes you unwell for a week, but your melanoma doesn't come back, you should still thank me. If I give you type 1 diabetes, if I give you cortisol deficiency, you still have to thank me, but you're allowed to be angry at me." There's no increase in those type of immune toxicities with this. This in some ways, for us it's a relief that we're not talking about giving adjuvant Ipi/Nivo and giving people a 10%-15% chance of cortisol deficiency because when this is approved, we'll be able to say your risk of cortisol deficiency remains 1%-2%. The intismeran side effects are pretty predictable. It's interesting now that we've got a placebo-controlled trial.
We're currently running the metastatic trial of pembro plus intismeran for stage 4 melanoma. Either I'm very lucky at getting my patients randomized to the combination, or a lot of my patients/myself have the placebo effect because everyone gets chills and a sore arm. It doesn't seem to be too bad. They are really manageable. Now that we know what we're doing with the current trial and the previous phase III trial, it was responsive really to NSAIDs, paracetamol, a little bit of fluids. This is the translational data that Michelle touched on, which I actually think was the best part of this presentation rather than the update. I'll just talk you through it because this took me a while, and I think Moderna's immunologist had to slow it down for me.
What this is a measure of clonality. Clonality means have you got a large amount of T cells attacking one clone, and up is more clonality, and down is more diversity. A spread of T cells targeting lots of different targets. Well, not targeting for that matter. Their sort of inverse relation, the more clonality, the less diversity, and the more diversity, the less clonality. The time point to baseline, this is before they get the drug, they've signed their consent, Moderna's busily making the product. They haven't yet got any pembro. We give everyone two doses of pembro, and they haven't yet got the intismeran, which is often given with the third dose.
This is after two, potentially with the third intismeran at this time point, and this for most patients is a year after starting, but importantly six months after the last intismeran dose for the vast majority. I'm going to talk this backwards to what I did today. I'm going to talk about the blue line first. pembro alone does not increase clonality. The way I think about that is pembro's activating your immune system, but it's not specific to any one thing really because otherwise, if it was specific to certain targets, you would expect clones to kind of come out. That's certainly how I describe pembrotox to patients. I say, "I'm trying to activate your immune system to attack melanoma.
Because I'm not smart enough to point it towards the melanoma, you're going to get a side effect that attacks your thyroid. intismeran's the opposite. We're getting an increase in clonality, right? We're getting our immune response directed. This does not tell us, this curve doesn't tell us that we're directing the immune response towards those 34 neoantigens. The question is, if it's not directing it to those 34 neoantigens, what are you directing it to? Ryan's going to show you that we are in fact directing it towards those 34 neoantigens. This also tells you why you don't get immune-related toxicities additively, because your immune response from intismeran is directed at those 34 neoantigens, and hopefully the algorithm is choosing neoantigens that are novel to the tumor, not host neoantigens. Similarly, this is new clones now.
What they've done here is forget about any T-cell clone that existed before the patient walked into the study. Same thing, we are getting new clones coming out, at the early time point, because the two arms of the study are identical, as many new clones for pembro as you get for pembro because both arms are getting pembro. You get a greater number of clones with the intismeran. Once again, the suggestion or the hypothesis confirmed by Ryan or whoever did the work Ryan presents, is this delta, this extra number of clones, is due to clones directed at those 34 neoantigens in the product. Okay. Closing the loop is this. Just for a moment, forget about that curve. This is now just the patients who got combinations. intismeran, pembrolizumab, everyone in this curve, it's the same time point.
Red patients did not recur. These are the patients who, if you like, cancer hasn't come back. Maroon patients unfortunately did recur. It's really linking the mechanism to outcome. The patients who didn't recur got more novel clones than the patients who did recur. Closing the loop, that association between novel clones and outcome wasn't seen with pembro alone. Once again, really strongly suggesting a mechanistic link between intismeran and the improved outcome. As I said, I think the support for this data that Ryan's data gives is huge in my mind. I think hazard ratios of 0.41 are hard to look aside, but when you see the data that proves how it happened, it really does close that loop, even based on 157 patients. This is essentially a return to the things.
It works, it reduces the risk of stuff coming back a lot. The safety and tolerability is fantastic. Really, the side effects you see are pembrolizumab side effects. I cannot think of a patient that I am willing to give adjuvant pembrolizumab to that I'm not willing to give adjuvant intismeran and pembrolizumab to. There are many patients I could consider giving intismeran to that perhaps wouldn't have pembrolizumab, it's that good a safety profile. The translational data, yes, the superficial translational data I showed to you really does, I guess, suggest a link mechanistically, Ryan's data is going to take that suggested link to a kind of a dead certainty in a moment. I think like always, I think finding so much information from a randomized phase II is a positive, not a negative, ASCO emailed me, said I had to have some study limitations.
I actually think the ability to make such a strong scientific argument based on a randomized phase II is a strength, not a weakness. The job of this study, and it did it in spades, was not only to lead to a registration phase III in melanoma, but to lead to multiple registration studies. I'm glad that the first registration positive study will be melanoma, and like always, all the other cancer types can follow. When we see the registration study soon, I think at least I'll remember, and that it kind of started with 157 patients five years ago. Okay. Thank you.
Okay, I get to take us home. One of the joys I've had with working with Moderna and with Matt and the other investigators is having the opportunity to present some of the translational data with this study over a few years. When we were presenting that data that Matt just showed with the expanded clones, people said, "Well, what are they against?" I said, "I don't know." There was initial collection of large numbers of patients with leukapheresis to do very sophisticated analysis and to type these T-cell clones and see if they were neoantigen specific. We've ultimately gotten to the point where we have some of that analysis. That's what I'm going to share. You've seen much of this.
I will say that KEYNOTE-603, which was the phase I study of intismeran, did show in the publication that came out last year that all of the patients that they had T cells collected and analyzed, they could show that there was neoantigen-specific T cell clonal generation. 30% of those neoantigens elicited immune responses. You could see both expansion of CD8 and CD4 positive T cells. In KEYNOTE-942, again, intismeran shows higher new clonality, and those were sustained over the course of the study. In fact, in those patients that didn't recur, it seemed to be going up, and the novel T cell clonal type expansion associated with better outcomes.
What I'm going to do over the next few minutes is talk to you about the efforts that have been done to link the de novo T cell clonotypes to intismeran encoded neoantigens and to characterize the resulting neoantigen specific T cell responses in patients who received the combination. Importantly, this study looked at seven patients' worth of samples. Samples were collected, I wish I had that little figure that had each of the time points, but essentially baseline after starting pembrolizumab, after starting intismeran, and then a long-term follow-up, which was usually a year after therapy started and about six months after intismeran ended. From these seven patients, three were actually from the phase I study and four were from the KEYNOTE-942 study.
What this figure shows is that using a really cool assay where you can essentially load the peptide and have a reporter assay that has a T cell receptor that can recognize that peptide, or you have the T cell receptor that are on these T cells, you can see whether or not there was a target engagement, which essentially means that it was a neoepitope specific TCR. This is a patient, and just showing the example that you can identify a number of different TCRs against the same neoantigen. Remember, up to 34 neoantigens are in the cassette. This was looking at neoantigen 14, these were all generated for neoantigen 14 and characterized. These happen to only be CD8. This patient generated CD8 T cell clones against neoantigen 14, against neoantigen 27 and 28, similarly just CD8, but also CD4.
Proving the point that when you create a neoantigen therapy, you may get both CD4 and CD8, and we think that's probably a good thing. This is taking that same patient and then looking at all the T cell clones that were generated against those specific neoepitopes. Again, 14, 27, 28. This is only CD8 positive T cells against 14, but as I mentioned, CD4 and CD8 against 27 and 28. The ones that are in color are the ones that have been characterized to date. It turns out it takes a lot of time to do this, and really the point of doing this was just to show, yes, we can do this, we can follow this, but there's a ton more that were generated.
I think it's important to note this straight line is when patients are getting pembro, and then these steep upward sloping lines are after intismeran dosing. Dozens of different neoepitopes for specific TCRs were mapped to each neoantigen. The other point to make is that some of these are going down, but most of these are sticking around, and it's been six months since this person received intismeran and yet they still have at least the same, if not higher numbers of T cell clones against, for example, neoantigen 14. When we look at all seven patients, we were able to identify at least one T cell clone or neoantigen specific T cell. This was this patient. The highest of the seven was 18 out of the 34.
Importantly, and I think it was mentioned earlier, that there is rare shared neoantigens, but those that are shared are either BRAF or NRAS neoantigens. That intismeran, again, elicits CD4 positive and CD8 positive and combined CD4 and CD8 positive to different neoepitopes. When you begin to look and say, "Okay, that's cool. You have T cells that can recognize the neoepitope. What are those T cells doing?" In this, we're looking at staining of the T cells. What you can see when these T cells are engaging the antigen is that they're making granzyme B, which is a good thing if you want T cells to kill things. They're making interferon gamma. This is a different patient. No, still patient 3, and looking at the responsiveness against neoantigen 14 and neoantigen 27.
Not only are they there, and we can identify them, but they seem to be acting like the type of T cell we want to generate to kill cancer cells. In addition, you can then begin to look and say, "Okay, what type bucket would you put these T cells in?" Are these the cytotoxic T cells that are terminally differentiated, and killing and then dying? Are these central memory cells that may not be as close to the action, but can stick around a while? Or are they T effector memory cells? You can characterize that. What's happening here is you're doing a dextramer stain of the neoepitope, and then you can see which actually bind to the dextramer, and then you can see where they are. This is a control against viral antigens.
What you're seeing, that these T cells, when exposed to viral antigens, are terminally differentiated. These are ready to go. They're trying to get rid of those virally infected cells that they thought they got exposed to. When we look at the intismeran treated T cells, these are falling more into the T effector memory. They're functionally active, they're positioned to do rapid surveillance, and they can also potentially leave and go to other areas of the body. To conclude, we saw durable de novo neoantigen-specific T cell responses in patients who were treated with intismeran and pembrolizumab. The team at Moderna was able to functionally validate and directly map the intismeran encoded individualized neoantigens. They validated the neoantigen-specific TCR clonotypes and expanded during intismeran therapy, and most importantly, they persist over time. We could see both CD4 and CD8 T cell responses.
The nice thing about this is if I'm standing in front of somebody talking about intismeran and they say, "That was cool what you showed about those TCR clones, but do you even know that those things are going after neoepitopes?" I can say, "Yeah, I think we do know that." What's next, Dave?
Okay.
Okay.
Thank you very much. Wonderful presentations. We appreciate you both taking time out from your busy schedules. Today you heard about intismeran, and that's frankly why I came to the company. Since joining the company, I have also discovered quite an exciting oncology pipeline, including our 4359, which we had an oral presentation at AACR. 4106 and 4200, which are off-the-shelf cancer antigen therapies. A T cell engager, in vivo T cell engager program, which is really interesting, 2808. In vivo cell therapy enhancers. These three programs, as well as our emerging oncology research programs, which will enter the clinic, will be highlighted at our Moderna Science Day later this month, and I encourage you to attend that. Now I'd like to invite up our speakers, and we're happy to take question and answers. Guys, if, and Michelle?
Great. Thank you everyone for fantastic presentations. We're going to start the Q&A session. When you do take the mic, please introduce yourself and then ask your question.
Hi, Tenthoff from Piper Sandler. If I may, I have two for Michelle, and if you permit me, I have one for Ryan. It was very interesting to see in melanoma that that small percentage of the antigens, albeit unique to the patient, were kind of what I would call shared or known driver mutations, i.e. RAS, Raf/MEK. In other cancers, are you seeing more of these shared ones? The one that jumps to mind right away would maybe be the phase II for CRC. Are you seeing more KRAS or things like that? Are you seeing in different patients, depending on the tumor type, a different flavor and/or number of these shared antigens? The second part where the question's going is, how do we know these neoantigens are actually better than shared antigens?
I guess it's kind of impossible to prove that out until maybe we do an intismeran versus one of your shared antigen vaccine trials. I don't know. How do we actually know that these are better?
Yeah. Maybe I'll start and then phone a friend if someone else wants to answer.
Yeah. Happy to help, yeah.
I think the key is that in the bioinformatics algorithm, it weighs out all the mutations, and that includes hotspot mutations, irrespective of the tumor type we're going into, which is why you see the type of clinical trial portfolio we have. In melanoma, you would expect BRAF and RAS mutations to pop up and be within the ranking system. For non-small cell lung cancer, I would expect EGFR, ALK, ROS. The question becomes when we rank them based off HLA presentation, expression level, and a prediction to engage the immune system, do those shared mutations actually rank to the way the neoantigens do or the novel clones? In some patients, that's the case where you see it for the less than 1%, in some patients, you don't see it.
The answer is, in all tumor types that we have, there is the potential to have some of those shared incorporated into the neoantigen cassette. The question becomes, do we have data yet across the rest of the studies to start showing this? The answer is, we did see this in the P101 study back from some of the older papers where we were mapping the different neoantigens, but we haven't accessed the phase II studies yet that are ongoing or any of the new phase III studies in the rest of the tumor types. That is one of the things that we are really looking forward to is the data sets there to see exactly how those neoantigens map. Do you want to add?
I was going to add, it's highly likely the answer is going to be yes. There's a long literature on this, going back to Steve Rosenberg's interrogation that shared neoantigens are just not common. That probably has to do with the fact that there's evolutionary immune pressure. Even in hotspot mutations, they tend not to be presented or they're presented at very low levels. I think that's to be the case. Then the answer to your second question, which was, what was the
How do you compare the shared versus-
Oh, yes. Yeah. The shared antigens versus neoantigens, of course, we won't know, but scientifically, neoantigens have not gone through thymic selection, and so there are T cells that are going to recognize it with sufficient avidity, whereas shared antigens, by definition, have gone. Even shared antigens such as cancer testes that are not usually present in adult tissues, they're present in the testes or the placenta, so they have gone through some thymic selection. That's the scientific hypothesis, but of course, we won't know until we test.
That's a great point. I really do appreciate the comment that you're making about it's really about how the patient is going to respond to the antigens. My quick question for Ryan was, and maybe I misunderstood this really cool presentation, but on slide 39 where you showed the three antigens that were activated in expanding the T cells, does this mean that only three of the neoantigens were taken up and activated that way, or were these just the top three that you had come across and analyzed yet? Thank you.
When characterized, each of the seven patients were characterized, and they were tested against each of their 34 neoepitopes, and the range was one TCR expansion against the one neoepitope to 18 neoepitopes that generated a TCR expansion. I think it's why does one happen or 18 happen? Why do you get a dozen or two dozen or three dozen TCRs against one neoepitope? I think these are questions that are hard to answer from a small data set, but maybe areas of study, particularly if we can. All these patients that I showed, none of them had recurrence. It's not like I can say, "Oh, the one had one recur," but the one that had 18 didn't recur.
Of course, because it's an adjuvant study, you can't look at response or PFS or other endpoints that just make you feel more certain about what's clinically happening from an anti-cancer standpoint.
The other thing with the one, my understanding is that one, that patient had one neoantigen with a response. It wasn't one TCR.
Correct. It wasn't one TCR.
Even for a given HLA, you may have more than one TCR, and some neoantigens will be presented in different HLA.
Correct.
I think there was that illustration where one of them had multiple T cell clones against one antigen. That was the norm rather than the exception.
That was the norm.
The norm was multiple different T cell clones against one antigen.
What we didn't show here, but showed on the poster, there was an example of a patient who had across HLA-A and HLA-B that they would generate against the same neoantigen.
Which gives this innate biological redundancy. It's not HLA-A*02:01 limited.
Correct.
I think I was surprised by the number of different T cell clones against a given neoantigen.
Yeah. No, that was clearly the most interesting-
Yeah
what I would not have guessed before seeing the data.
Mm-hmm. Particularly when you go, "Oh, we've chosen that clone because it binds well to HLA X.
Right. Yep.
In your head, the cartoon you drew as a student was one T cell finds that HLA with that neoantigen. Clearly, there is so much redundancy in the immune response.
Thank you.
Hey guys, Alec Stranahan from Bank of America. Thanks for holding this event. Really great discussion. Just two questions from me. Maybe first, I believe that the phase III is enrolling Stage 2 disease as well.
I guess, what percent does this represent within the phase III data set? I think we saw, I guess, directionally higher hazard ratio in the Stage 3C versus the Stage 4 in the 5-year follow-up. Small n, but how does that maybe bode for even earlier patients? Then maybe one quick one for Dr. Sullivan. In the poster yesterday, it looked like there was maybe some variability in terms of the number of immunogenic neoantigens sort of within that 34 total delivered between patients. Could you maybe speak to the variability and if there's maybe a minimum threshold of truly immunogenic neoantigens you'd expect to drive efficacy?
I'll take the first one, which is, we wouldn't disclose the number right now on an ongoing trial, but I would say it's likely to be sufficient to enable labeling negotiations. In terms of efficacy, I think you're right. I think the lesson we're learning is that the earlier you go, the more likely you're going to have a treatment effect. Of course, that's balanced. The earlier you go, the less events you're going to have. It'll be a balance between that.
My gut feeling of that is that the benefit will be the same, but I predict there's some chance the benefit will be greater. When you develop a hazard ratio off a curve, the hazard ratio is a reference to the whole curve compared to the whole curve, which sounds intuitive. The higher risk your population, the earlier you get recurrences. In adjuvant studies, you get a group of patients who recur within weeks, and they're the absolute highest-risk patients. If a treatment starts at week six, I predict the hazard ratio in the slightly lower risk patients, not low risk, because stage 2C is still a high-risk population, but they're less likely to recur three weeks after you start the pembro.
If the hazard ratio, instead of being 0.51, is 0.45, my hypothesis is it's that by having a more, if you like, typical adjuvant population, paradoxically, we get a better hazard ratio because when the curves run together, that feeds into the hazard ratio of the whole curve. I actually think it's going to be a positive that we've balanced the population because really this has got an artificially early recurrence population, not necessarily higher risk, but macroscopic disease has those early recurrences, whereas the 3Bs, the 3As and the 2Cs and 2Bs don't get that. They recur at first scan as often. I predict we might get a better number with a typical population.
Your question about the number of neoantigens that you get T-cell responsiveness to. We certainly, again, we can't answer that in this population because it's, one, we don't have enough patients, and we don't have a splay between patients who recurred and patients who didn't recur. There is two pieces of information that speak to this. Both is with a competitor called autogene cevumeran. One is in the pancreatic cancer data set, the initial data set that came out of MSKCC, where they treated patients in the adjuvant setting with their neoepitope therapy. The patients that generated responses against the neoepitopes didn't recur, and the patients who didn't generate neoepitope responses did recur, and that's very clean data. In another trial with that drug, we did a phase II randomized study in front-line melanoma, previously untreated.
This actually was launched around the same time, so there weren't a lot of patients who got adjuvant therapy on that study. It was mostly just first time at being exposed to checkpoint inhibitors. Very similarly to that, we could see that patients who generated more neoepitope and more responses to more neoepitopes had better outcomes than the patients who didn't. It's not a specific answer to intismeran, but presumably these things work together and that biology is biology, if you're having an immune responsiveness to a therapy like this.
The one other piece I'd add onto what Dr. Sullivan's saying is, at least in the poster, right? You had that splay between one and 18. We know one can be sufficient, at least for disease control in that one patient. This is one of the reasons we're so excited for the 5,000+ samples that I highlighted because these are the types of sizes I think we're going to need to understand how many neoantigens do you need? What makes the good features of a neoantigen? How do we start parsing that out? I don't think we can do it with just the seven.
Yeah, you basically need all the samples from all the patients or nearly all the samples from all the patients that you can really get a good look at how it correlates with outcomes.
Thanks.
Cory Kasimov with Evercore ISI. Thanks for taking the question. It's probably for Dr. Carlino, but anybody can chime in. Curious as to your thoughts about how the upstream migration of immunotherapy to the neoadjuvant setting could eventually impact intismeran. Once available, how would you choose which patients get neoadjuvant therapy versus an intismeran-based adjuvant therapy?
Yes, that's a good question. The one question I asked, expecting to get that question this morning but not getting it, was on this study, even though they're 3C, remember a lot of 3C patients are microscopic 3C. Even on this study, a third of patients have micrometastatic disease. By definition, those people can't get neoadjuvant. When you look at the registration trial, the prediction or the estimation is 2/3 of those patients have micrometastatic disease. The micrometastatic population is a significant proportion of the Stage 3s, and by definition, the Stage 2, despite the presentation this morning, Stage 2 don't really have an option of neoadjuvant because the definitive, the biopsy removes all the tumor. You've got all the Stage 2s and most of the Stage 3s, neoadjuvant isn't even on the cards for. I think that's the first thing.
This is the dominant market, if you will. I think clearly, I guess the provocative question is, at the moment, if someone walks in with macroscopic nodal disease, they're getting neoadjuvant, not adjuvant. I think that will remain the standard unless the hazard ratio on the registration trial is so good where you'd actually say to someone, "Look, I could give you neoadjuvant pembro, same toxicity." I think intismeran outdoes the neoadjuvant benefit, hypothetically. When you're saying, "I'm going to give you neoadjuvant ipi/nivo," different question. I think the short answer is there's a huge market or a huge group of patients that need adjuvant and can't have neoadjuvant. That's the easy population. If the hazard ratio is spectacular, it becomes quite a complex discussion because the lack of increased tox compared to NADINA, because NADINA adds permanent tox.
I think the most interesting question, the next study is for the, can intismeran add to the neoadjuvant therapy? If you give intismeran pre-surgery, can you improve on the NADINA or the SWOG result? That's an important study to do, so we have to do that study where you start SWOG or start NADINA, and you start intismeran before the operation. The easier group of patients to study is the patients who get neoadjuvant, whichever one, and then post neoadjuvant, if they don't have a complete response, they get their adjuvant therapy, typically nivo or pembro, plus intismeran. That's a very easy study to run.
Some may argue you don't need the study. Certainly once you've got the positive phase III study, I've just given you an idea of two other studies, and the post-neoadjuvant one, I'm confident we could design a study with 200 patients. Those patients have such a high risk of recurrence, they're technically already on label. We could do another quick study with 200 patients and get that answer relatively quickly, expanding the population. I think my main answer in summary is that the microscopic population is still the bulk of the patients slash stage 2, and this will be tested in neoadjuvant. Those two studies have to happen. The lung cancer guys have already got those studies going on. We saw on that first slide the post-neoadjuvant study already happening in the equivalent lung cancer.
Lung cancer's not allowed to beat melanoma, so we'll.
Okay. Thank you
get there quickly.
Myles Minter from William Blair. Congrats on running a controlled study here.
Thank you.
I think you deserve that comment earlier today. My question is actually very much coming from investors, Dr. Carlino, I was interested, you were kind of inferring where the placebo curve would be on that RFS for the five-year data. Investors bring up where the pembro curve from KEYNOTE-054 would've fitted on that. Do you think that pembro performed as expected in this trial? Did it underperform?
Exactly. If you go to the 054 papers, you can't use the whole 054 population. The best way to capture the prediction is just look at the 3C group. That one of the 054 papers, we sort of update 054.
Four year.
two years. One of the 054s has the 3A, 3Bs, and 3Cs, and that 3C group is the best approximation for this, and I think it's smack on 50 at four year. It's very close. That's what I was using in my head. I was remembering that. I think it's actually, there's one figure because there's, I remember there's three, it's A, B, C in the middle. I can't remember which paper it is, but I'm pretty confident it's smack on.
Hi, this is Greg Torres, representing Tyler from TD Cowen. Thanks for taking my question. How confident are you that the high rate of INT manufacturing you've achieved in clinical trials will be replicable in the real world? Are there any considerations related to feasibility of this in the Stage 2 and 3 versus Stage 4 patients? Thanks.
Yeah, maybe I'll take that one. I'll start with the Stage 2 and 3. Obviously we've done it for the 001 study. On top of it, one of the things that we're so excited about with some of the other studies is that, especially for like NMIBC, the amount of tumor is actually even smaller than what you could get into the Stage 2. Not only are we expanding the hypotheses within the clinical studies, but we're also testing the bounds for the tissue requirements as well. For that, I think we've been able to showcase that based off the enrollment that we've had. As far as the global expansion, you have to remember that this study for KEYNOTE-942 ran from 2017 to 2019, and in that timeframe, we manufactured 107 intismeran doses over a three-year span.
The 001 study started in 2023. Since then we have the litany of the pipeline. To date, we've dosed over 2,000 intismeran patients. You're talking a 10-fold range in the past three years. Again, that ability to scale that we saw with the COVID era is what we are able to scale with intismeran. It's actually a miniaturization of the process and then a scale out across the globe. What we saw with KEYNOTE-942 was that we had a U.S. and Australia footprints, and now our global studies with Merck are in over 46 countries. Again, that turnaround time has really maintained itself. Even with that scale up of the 10-fold, the scale out globally, we're still able to actually maintain the distribution time in the manufacturing chain.
That gives us confidence that if we enter into the commercial sphere, that we'll be able to deliver the way we expect.
When we visited, well, I visited the Moderna plant, and I think you really got the feel having visited that it was set up to do 50,000 and it almost felt like if it wasn't doing tens of thousands, you could see the exponentialness of the processes where it felt very impressive to see how this process had turned down at one thing and then it would just repeat it and if you wanted to double it, you times it by four. I think probably of the ASCO week, the visit to that plant was up there with how much thought had been put into the scalability, I think as someone who doesn't make things. It seemed to be ready and raring, where it almost felt like it was built and sitting there ready to go.
Hi, Lili Nsongo from Leerink Partners. Very quick question going back to the clonotyping and then linking that to the epitope selections. Up to 34 neoepitope selection, I think it was down to nine, between nine and 34, and then one to 18 that are immunogenic. Can you maybe give us a sense of how much do we know in terms of whether there is any correlation between any baseline characteristic of those patients, namely PD-L1 status, stage, and what you're able to do in terms of neoepitope selection and then what you see in the clonotyping?
Yeah, I think it's an excellent question and an important one. I think with seven patients across two trials, it doesn't matter what they were in terms of this PD-L1 status or TMB or things like that. I think we need to look at more patients and to be able to answer that question with any certainty.
I guess intuitively it's unlikely to be a simple thing like stage in my mind.
I would agree with that.
It's the biology of both the immune response and the tumor. If we look at TMB, we don't have a huge difference in TMB between stage IIIC and stage 2 melanoma. Certainly actually in Australia you probably get a higher TMB in stage 2 because they're slightly older and we get these group of patients who have stage 2 melanoma in older people, which goes with chronic sun damage and a TMB. I don't think it's got to be something like stage or PD-L1 or something simple.
I guess the only thing, I think this is where this could only get better, is hypothetically you could imagine an analysis in 5,000 patients where you run the algorithm and say, "Actually, this type of neoantigen isn't as good, so you no longer make the cut of the 34." I think the exciting thing for me is you're choosing the 34 best. I think it's impressive that they chose the 34 best not having ever treated a patient and that's worked out pretty bloody good. If you can feed the positive and negative feedback into the algorithm, could you improve it? I guess the question is do you need to improve it? Until the curve's flat you can always try.
I actually think that if there was an opportunity to evolve the algorithm, and I know that's complicated and we'll need discussions, but certainly the algorithm's pretty impressive. Theoretically, I think that was where the difference might come, not something about the stage of the patients.
True, who knows? Like somebody with a big bulky tumor may be different than somebody with microscopic disease. I think it's valid to look into that, but we also have to just need to get data from many more patients.
Thanks, guys. Matt Hagood for Michael Yee from UBS. One for Dr. Carlino. I just wanted to ask if you could expand a bit on the safety profile here compared to IO. I'm curious your thoughts on whether this kind of expands the population of patients who would get treated in the adjuvant setting with an agent. Then if I could just squeeze one more in, too, on the earlier stage pipeline, I wanted to ask when we might start seeing some phase I updates from those programs, too. Thanks.
Look, a couple of things about the safety. I think the decision to use intismeran pembrolizumab, from a safety point of view, is completely driven by the pembrolizumab toxicity. Right. Am I using intismeran pembrolizumab in someone with a kidney transplant adjuvantly? No. Nothing to do with intismeran, it's because of pembrolizumab. The only area, I guess, it may expand the population theoretically is I gave a second talk at ASCO this afternoon about tox, and my first slide was about risk benefit. Right. Every patient has a risk profile which can alter based on their comorbidities, and every patient has a risk appetite. Right.
As a patient, a stage 2 patient may say, "Look, for an absolute benefit of 10%, I'm unwilling to take the risk of a 1% risk of cortisol deficiency." Whether you would make that decision or I would make that decision, that individual patient may decide not to take a given risk. intismeran doubles the benefit and does nothing to the tox, the risk-benefit discussion in some ways becomes easier. Certainly that's how I discuss adjuvant, risk-benefit. X benefit, X risk, and you as an individual have to decide. We're not touching the risk side of that equation. If we're doubling the benefit side of the equation, you may expand the population.
The thing that we haven't touched on in melanoma but the lung cancer guys get to is single agent intismeran completely different conversation because it's probably safe to give a kidney transplant patient a single agent intismeran or a person with rheumatoid arthritis or ulcerative colitis, et cetera. I think the summary answer to me is the combination you probably have a slightly bigger population because of the risk benefit. If the population was going to expand hugely, it would be single agent. I guess a question for Ryan, when the drug's approved, if you had a patient who was contraindicated to pembro, theoretically there would be a temptation to use single agent intismeran.
Yeah. Probably wouldn't do it, but.
Yeah
that's generally because I like some data before I make a decision. I don't need all the data, but a little. I do want to come back to your point, though, Matt. I think not only is the risk benefit, ultimately the decision is the patient's, but the way that the physician talks to the patient really can influence the decision. Right now, I don't know about you, Matt, but if somebody has stage 3A melanoma, they have a 1.1 millimeter thick melanoma on their leg and they have microscopic deposits in their node, I'm not going to recommend pembrolizumab for that patient. I'll talk to them about it, and if they absolutely say, "I have to have this otherwise I'm not going to be able to live," then I'll entertain it. It's on label.
I think to Matt's point about what a patient's threshold might be, our threshold just kind of in our biases about how we think about risks and benefits, we might actually be more supportive of making such a recommendation.
Just to answer your second question, the 4359 we just updated at AACR. We now are expanding the data set to confirm the signal. That's ongoing and I would estimate probably next year would be an update on 4359. For the other three or four clinical programs, they're still mostly in dose escalation. We will share when we plan to update, I think, as we get closer. I'll just put a plug in for the Science Day we have later this month.
Yeah. The only one other part since we're talking ASCO is we did have the 2808 TiP actually as a poster session here. It does showcase that we're beginning to actually, one, have the trials in progress, and two, there's some preclinical papers that are out for that one as well as 4106 and 4200.
Okay.
Ellie Merle, Barclays. Thanks for taking the question. Maybe first just for the physicians, obviously really impressive hazard ratio in phase II. What would be the minimum effect size that you would want to see in phase III to adopt this and say widely add it on to all your PD-1 adjuvant patients given the risk-benefit comment that you make around the safety? Then just a question for the company. Obviously, really impressive hazard ratio in phase II. As we think about the interim analysis, any color on sort of what the minimum hazard ratio would be needed to be successful on the first interim? Then just based on the curves that we've seen in the IO trials, how we should think about the likelihood that if it misses the first interim it could be successful at the final analysis.
Just quickly on the statistics, we've said nothing on the interim analysis on the statistics, and with that I'll hand it over to one of the physicians, please, to speak to what they consider a good hazard ratio in the phase III.
Yeah. I guess the first thing is that there, in my mind as an Australian, I think there are two thresholds. As a patient, as in a clinician, when I think of tox, all I care about is serious tox.
Permanent tox is the big one for me. There's permanent tox or tox that bothers a patient's quality of life is what we don't want. Because the cost from the patient perspective, not financial, the cost from the patient's perspective is negligible, the benefit to consider doing it is lower in my mind. Hypothetically, if I had two trials that had the same hazard ratio, intismeran pembro versus pembro, or ipi pembro versus pembro in the adjuvant setting, and they were identical hazard ratios, ipi doesn't meet the bar intismeran does. From a patient and a clinician, I think the bar actually is quite low. Acknowledging that there's another part to cost, right? The hazard ratio that is required to make it financially worth it is hard as a clinician. We're not great making financial decisions.
Me personally, I'm also Australian, so therefore I come from a health system that has a central payer, and has those complexities. I don't know what number. I think certainly, and the other way to look at it is actually not hazard ratio, it's absolute benefits and numbers needed to treat. We've got an absolute benefit in that study of 22%, so a number needed to treat of five. Would I do this for a number needed to treat of 10? Yes. The payer bit is something I don't have the knowledge to answer on. From the patient point of view, the benefit could be really tiny because the patient's cost is not measured in dollars and cents. The patient cost is measured in tox. With the calculation being tox versus benefit, the benefit could be smaller than
Yeah. I think the answer to your question about what hazard ratio, whatever the hazard ratio where the P value is significant and the FDA can approve it, I'll use it.
Okay. We'll take our last question.
Hi, this is Morgan Lamberti, representing Salveen Richter at Goldman Sachs. For the physicians, how does the translational clonality data impact how you think about success in non-small cell lung cancer and indications beyond this?
So-
Oh, you go.
I'll just start and then I don't know. If you didn't have the translational data, but you had an amazingly positive phase III trial, it doesn't matter that you didn't have the translational data. It kind of matters in from a sort of scientific interest standpoint, like did it actually do what we think it can do? I think from a proof of concept and as you're also rolling out potential ways that you can monitor patients, it's way easier and cheaper to do TCR sequencing than it is to do all of that really sophisticated experiments to get that reporter assay and characterize all of the TCRs. On some level, I think seeing that there is a likely explanation of that rise in TCR sequence or TCR clonality, that correlates with actual expansion of neoepitope-specific clones.
It'll make us feel better about any of the trials if we just followed TCR clonality, through TCR sequencing, instead of doing these deep characterizations of all these. As a scientist, it also makes me feel good that we are saying we designed something to do something, we did it, and then we've actually shown that we were able to do the thing that we were trying to do. There's a lot of dos in there. Ultimately, I think it feels good when you can say, "This works, and we actually are doing what we think the drug's supposed to do, and we might even have a surrogate biomarker that can be reflective of the more sophisticated TCR characterization that can be rolled out into all of the trials.
Yeah. Look, I agree, Ryan. I think when you've got a randomized phase III trial that's positive, it almost doesn't matter what the drug does, right? By the same token, if I give you the coolest drug ever and the best preclinical data, if a properly designed and run phase III trial is negative, doesn't matter how cool the NECTAR, the science is, it's negative. The phase III trials thing, I think the big thing this does for me is, one, it gives you huge confidence that the phase II data is real. When you've got less than phase III evidence, predicting your drug works because of X and then showing that it actually happened gives you confidence. I think with the reference to non-small cell lung cancer, the other positive is this mechanism of immune response is not melanoma specific.
It actually gives you the confidence that this is likely not just to be a melanoma thing, right. I think at the point now where we're going, what's the likelihood of the registration melanoma study being positive? The translational data increases that off the scale. To be honest, once you've got this phase III data and you just get phase III positive trial after phase III positive trial, the translational data is really important about working out if there is a subgroup of patients you're not helping, what do we have to tweak to help them, and what's next. It's more about the confidence today about what's coming next year, really. That's the most important thing. Yes, it's cool, but it makes us way more confident than we otherwise would have been.
Yeah. Maybe I'll just add in the final minutes is, the most validated immunotherapy, of course, are anti-PD-1s. We know that anti-PD-1s work by releasing the brake or by reversing exhaustion on tumor-reactive T cells, which then go on to kill the cancer. That's the most validated clinically, immunotherapy. We know now that our therapy roughly does the same thing. We give the intismeran, it activates neoantigen specific T cells. From the randomized data, we see evidence in this phase II that the T cells are doing their job. I think to me, that's the reason it gives me reason to believe that this is not just a melanoma specific phenomenon. Anti-PD-1s are not just melanoma drugs. Of course, we'll have to wait for the randomized trial to find out.
With that, let me thank our investigators who joined us, and thank you all for joining us tonight. Thank you, guys.
Thanks, Matt.
Appreciate it. Thank you so much.
You're very welcome.
For spending the evening here. My pleasure. Guys