GUGGENHEIM 3RD ANNUAL HEALTHCARE INNOVATION CONFERENCE - EDITAS MEDICINE INC

Good afternoon, and thank you for joining Guggenheim at our third annual Healthcare Innovations Conference. I am Debjit, one of the therapeutic analysts, and my privilege to welcome Gilmour O'Neill, President and CEO of Editas, and Amy Parrison, Chief Financial Officer. Thank you both for making time for us. It's great to be here. Thanks for having us. Obviously, the Editas story has evolved over the last couple of years. Gilmore, why don't you just give us a rundown where you are currently and then we can dial in. Happy to. So we're really at the point where we have driven or really close to complete, we have completed sort of the evolution of Editas where we are a entirely focused in vivo CRISPR editing company. And our lead asset, which we are on track to have patients dosed this year, really epitomizes our strategy, which is one of using the unique capabilities of CRISPR to create non-coding edits that functionally upregulate disease-mitigating proteins. And in this case, EDIT401, our lead asset, replicates a naturally occurring human genetic variant by editing the three prime UTR of the LDR gene, resulting in a significant upregulation, a direct increase in upregulation of that, which is associated with, in our non-human primates, a 90% reduction in LDL cholesterol, along with significant and comparable reductions in Lp and ApoB. And just at AHA, the American Heart Association, two days ago, we presented an update on the data showing that those effects now are durable out to 10 months in our human primates. Got it, yep, there's plenty to unpack from the AHA. but we'll get to that later, but could you help us understand, at least given the non-human primate data that we have seen from the classical PCSK9 editing approaches versus the one that you're pursuing, why does increasing the LDL receptor give you such a deeper LDL-C knockdown? Well, in all cases, they're trying to increase the LDR receptor. We believe that the reason we're able to do this with this effect size in the non-human primate, and hopefully humans, is because we are directly increasing the synthesis of the LDL receptor, whereas the PCSK9 inhibition approach is really trying to inhibit the degradation or recycling. So while they are actually targeting recycling of the LDL-receptor, we are correctly going to the core of the issue, which is increasing synthesis of LDL receptor. Got it. Let's talk about the HORIZON study and implications for 401 development program. The Phase 1-2 study is currently underway, right? Given the learnings from the HORIZON study, is there a specific subset of patients that you would enroll, given you can lower Lp and LDL-C? Because there was a signal in that small subset of of patients who are very high LDL-C. Yes. So we've looked at the LP little a as what we call almost like a no-cost or zero-cost upside. You know, the core central part of our thesis is the LDL C reduction. So there are a couple of takeaways from the Horizon study. You quite rightly point out the importance of enrolling high-risk patients. And in many ways, it validates our current approach in our Phase 1-2 to target very high-risk patients. In this case right now, we're leading with HEFH, but we can expand to non-HEFH patients, but very high-risk patients with arteriosclerotic cardiovascular disease who cannot get to target LDL cholesterol with their current standard of care. And I have to say, one of the takeaways from Horizons was it absolutely validated the critical need of getting LDL cholesterol low in these patients. You will recall that the event rate in the Horizons study was actually very low and is probably driven by the LDL cholesterol. But to your point about the subset of patients, those high-risk patients, those are patients who are already enrolling. We are really focused on the LDL as a selection criteria. And I think at later points, we would look at the LP little a effects. Got it. And from the dose that you plan to take in the clinic or where you would like to see therapeutic activity, help us understand the window of opportunity you have, because in the end, this is all LNP-mediated? Yes. So from the point of view of the effect size, obviously we're going into a start part one of our phase one, two study is a serial cohort dose escalation study. We're starting at a non-therapeutic dose for obvious reasons, just to ensure and maximize subject safety. And then we will escalate to higher dose and get to what we think is our target efficacious dose in subsequent cohorts. Apropos the LNP delivery, I think and I hope that I'm getting the answer to the right question. In the context of HEFH, obviously patients don't have or have an abnormal, one copy of the LDLR gene is abnormal and therefore could impair naked LNp uptake. So in this context, we have dressed our lipid nanoparticle with the GalNac. So that leverages, kind of enters the hepatocyte, if you like, by the back door, essentially leveraging another receptor, the SPGR receptor family. And we have very good data to show that in the context of both wild type animals and in mice who have a single or heterozygous for a loss of function LDLR, we're getting the same efficacy. Got it. And when we see the data at some point next year, what's your target number of patients that you hope to get to? Yes. So we essentially have designed the study with at least three patients per cohort, at least four cohorts. And so that's kind of sort of the broad number. Obviously, in the context of a phase one trial, like many trials, it's a three plus three per cohort design in case we need to expand the patient population. But essentially, that's the overall design of the of the protocol. And given that you have GalNac, do you think, at least with LNP mediated delivery, there is a lot more heterogeneity? Do you think GalNAC takes that out? We think that actually does take it out. In our experience, you know, in In the non-clinical setting, it has reduced heterogeneity. In fact, if you look at VIRV-102, they've also used a GalNac. And that actually is certainly, I say, restricted. And they have a very narrow variance in heterogeneities there, too, in humans. Let's say you get to, I don't know, to that comment about at AHA, we don't have a therapy which goes above 50%. Let's see you get that, not the 50%. what would be your target population for a registration study? So if we get to, well, frankly, we have more ambition, you know, to get more than 50%, if you will. So our target population will definitely be HEF-H. And, you Know, we are actually considering and looking at the possibility of expanding beyond that to include also another very highly motivated population with very high risk. And those are patients with established cardiovascular disease who sit in that very high-risk category where they've already had events and that actually population together United States numbers about 10 million patients got it and in terms of cadence of data that we should be expecting clearly the first dose is not gonna be therapeutic but do you wait for your full dose escalation to complete or if you get to I'm making up a number 60% you put that data out first yeah so the way we're thinking about what we've guided to is sharing announcing safety and and tolerability in Q1. So that's what we're tracking to, which will be from the first cohort. And then we anticipate sharing and have guided to sharing top level data from subsequent cohorts, which will include LDL cholesterol reductions and efficacy as long as, along with safety and tolerabilty in the second half of 2027. Got it. And when you talk to physicians, what's the enthusiasm for a one and done approach? Because in the end, it's lower for longer. Yes. So I think I would say there are three categories of physicians. There are those who kind of say, well, you know, I'd like to see how gene editing goes. There's a second category, which are, you would actually say, if you have equivalent efficacy and a gene editing approach that helps patients get down and get that area under the curve of exposure to LDL cholesterol, that's great. I think where people are really excited is with the potential offering that we have. which if we validate the non-human primate data with that 90% reduction, we can actually help them address the need that they are wrestling to deal with, which are those patients who, despite using chronic combination aggressive standard of care, cannot get to target. And as I say, that number can represent as many as 30% of patients with cardiovascular disease if you look at the clinical trial setting. Outside the clinical trials setting, that percentage or proportion of patients who can't get to targets is even higher. Got it. And any theoretical risks of overexpressing LDL receptor? Yeah, there are hypothetical risks, which we've actually looked at very closely. So, you know, I think the two key ones relate to, one, you know, a question people said, is there an LDL level that's too low? And, you know, the late Gene Brownwald described LDL as a toxin. And I think that position, which is shared by many of the experts we've talk to is really validated by a number of converging data points. One, the Icelandic kindred that pointed us to the LDL-R variant that we are creating, where they have very low LDL cholesterol. You know, some of them as low as, you know, 13 milligrams per deciliter. Then there are the PCSK9 homozygous knockouts, some of whom actually have single digits, which is below the level of quantification. And then when we look at Fourier and other long-term cardiovascular outcome studies when you look at patients who have achieved the lowest targets in those studies we have patients again in single digits so that's the first point so we believe that that can actually be done very safely I can go into more detail if you want and then from a second risk which is a hypothetical risk has arisen is this idea that there are some LDL our variants that if overexpressed could actually, or driven up an expression, could create an endoplasmic reticulum stress. However, we look very closely at that and I think one of the reasons we feel very comfortable about dismissing that hypothetical is that LDL-R is actually a low abundance protein. And where we have seen examples of high abundance proteins, which can actually cause that you really have to be very abundant you know classic example is the alpha 100 trypsin where these liver cells are producing grams of the protein every day whereas the ldlr is in sort of the nanograms so if there is a very high fidelity translation from the non-human primary data to humans what does it mean for lpa because isn't there a theoretical risk of new answer diabetes if you go down to like single digit lpa levels so i think it's very interesting that this diabetes story keeps coming around a lot. And, you know, this is a signal that was highlighted sort of in some of the earlier cardiovascular outcome studies for the statins. What's interesting is if you look at sort of various meta-analyses, that risk is actually extraordinarily low and, in fact, may be close to, if not at background rate. Obviously, we'll keep a very close eye and monitor that, but I think in many ways that the diabetes risk has been carried forward very robustly, notwithstanding, as I say, substantial meta-analyses. If you look at the Lancet, I don't think the Lancets published a meta-analysis of millions of patient years of exposure just about six months ago, and I actually found that that risk had been probably over-ascertained in prior trials, and really they toned that down quite significantly. Got it. And in terms of off-target characterizations, how much work has gone into that? So we've done a substantial amount of work. In fact, dare I say that many of the tools that EDItask created in its early days are used across multiple companies as they characterize off target. So we have done substantial amounts of in silico, which is where you start, as well as in vitro and biochemical experiments to look at this. And we actually have found a very clean profile, as others have reported in this context. Got it. Even if you were to sort of limit 401 to the HEFH market, it is a very large market. It is. So can Editas in its current form serve that market, or would you need to partner the asset? Well, we actually believe that we could certainly do that. Indeed, one can look at one's ambition even more greatly than that when you actually look across a broader population. but certainly the HFH patient population is one that we could certainly service. Our intention is to actually ensure that we develop the best and optimal development package for approval to deliver value to patients and our investors, and we can do it through the HVH, but actually we believe that we can actually do it even with the larger, more prevalent population. Got it. Given this is a slightly different target than, say, PCSK9, et cetera, but the end result is still LDL-C lowering, do you think the traditional approval pathways of using LDL C as a surrogate to reasonably predict clinical outcomes still serves as the path to approval? Yeah, I think that's a great question. I actually think in many ways, and one has to be careful about that. Obviously, this is subject to approval with agencies and so on. But the first thing is that the LDL-R pathway is the final common pathway for LDL C reduction. So you've got LDLC reduction, the same final common path way. We're directly ultra-impacting it. PCSK9 inhibitors impact the LDLR expression through inhibition of degradation, and then statins impact it through other mechanisms. So overall, it's the same finally common pathway. and so I think that the LDL precedent and path to approval actually does have a very good probability of success for us and indeed you may have seen more recently just this year a number of programs both recently approved like elicitide and two others who had revised labels had an indication statement that included MACE, based on other people's cardiovascular outcome studies or other sponsors, COVOTs. So I actually think that, obviously, contingent on regulatory agreement, that the path to approval would be with LDL cholesterol, and there's ample precedent for that. Got it. And even though it's just an HEFH population, what kind of safety database do you think the FDA is going to ask for, as opposed to more of a standard model? So I think the precedent for HEFH approvals has centered around a couple or a few hundred patients. Obviously, that's contingent on the FDA and how they view editing. I think we'll have more clarity in the next couple of years, particularly as Verve will be progressing with Lilly with their program, and we'll actually have other precedents from other in vivo editors. So we actually think that's a few hundreds. And after that, if we go to a larger population, that will be something we'll have to discuss with the agency. Got it. And earlier this year, Amy, you had a very successful round of financing. Where does that leave you? And depending upon what the feedback is from the FDA, it's going to be very different cost structures. Yes. Yes. Thank you. So, yeah, we were able to raise one hundred and twenty five million dollars earlier this year. And that extended our cash one way into the second half of 2028. So just to kind of piggyback on what Gilmore was saying, that will allow us to complete part one of our phase one, two trial and really progress through most enrollment in the part two. And so the financing really left us in a great position, as Gilmore were saying, to continue to generate value and really show the potential of edit 401. And I should say that, and thank you, Amy, that I probably didn't do a good job of saying part one and part two. Part one is a dose escalation serial cohort, and part two is an expansion population when we select the dose from part one. Got it. And, you know, depending on what would be your best case outcome for the part one dose escalATION before you even worry about the dose selection part because well I think a best-case scenario is that we see a very robust reduction in LDL cholesterol with a good tolerability and safety profile which we anticipate based on our non-clinical talks work and as you quite highlighted the really rather remarkable translational prediction of the non-human primate for in vivo editing And in that best case scenario, we believe that we could then execute, design and execute a pivotal program to deliver at the very minimum HEFH. And frankly, we're interested in looking beyond that. Is there a sweet spot that you would love to land up at? Because for most of the other editing programs, it's like getting to mid-40s would be a good outcome. but your preclinical data are so far above everything else that you've seen. Would a number above 60% is something you would aim for? Well, we would like very much to have a number that is above 60%, yes. Got it. Well, looking forward to the data next year, and congrats on all the progress. Thank you very much. Appreciate the time. Thank you. Great. Thank you, sir.