Great. Okay. Well, let's go ahead and get started with this next session, Genome Editing: Next- Wave Platforms Beyond CRISPR, what we call CRISPR 1.0. My name's Keay Nakae. I'm the Director of Research and one of the Senior Biotech Analysts at Chardan. I'm happy to introduce our participants, Dr. Amber Salzman, the CEO of Epicrispr, and Dr. Jian Irish, the CEO and President of Metagenomi. I think most of you are familiar with the Nobel Prize-winning gene editing science. Essentially, we're using a guide RNA to locate a section of DNA where the enzyme then makes a double-stranded break. That's fairly invasive. The repair process introduces its own set of complications. I think that we wanted to highlight two different next- wave editing approaches that can affect one-time treatments.
Epicrispr is leveraging the epigenome, the endogenous system that precisely and durably regulates gene expression via methyl markers without changing the sequence of the DNA. Metagenomi uses CRISPR-like programming editing logic but leverages its own proprietary library of nucleases to achieve targeted gene editing and gene insertion. With that, thank you both for joining us today. Amber, let me start with you. Fundamentally, your epigenetic therapeutics seek to remethylate or methylate a target chromosome to control gene expression. Can you briefly describe the advantages of your epigenetic approach and your design platform?
Sure. Just to add, we don't just methylate or remethylate. We do histone acetylation. We basically regulate genes up and down. As you noted, the work that was done by the Nobel laureates for CRISPR was amazing. I grew up in GSK, where developing a drug was really hard and took a really long time. To come across this technology where within months you could come up with a construct to bind to the location needing correcting was just so exciting and mind-boggling. However, we really needed to advance that.
What we've done is said, okay, now that we bind to the precise location, rather than cutting the DNA and really taking that brute force approach and potentially having genomic instability and not being able to really regulate it precisely, what we've done is we've deactivated the nuclease of the Cas molecules that we don't cut, and instead we bind to it one of our proprietary modulators so that we can then precisely control gene regulation. Where the power in that is that if you're trying to activate, you may want to sometimes only do twofold activation. For example, in a haploinsufficient disease, you've got one bad allele, one good allele. You only want to double the expression of the good allele. You sometimes really want to be careful not to go much higher because it can be toxic.
So that kind of precise activation is critical, so you want to have a modulator that could do that. In other situations where you have a compensatory gene that you want to express, like utrophin to make up for a lack of dystrophin, you want to be able to shout that one out. So you want to have modulators that do that. So that kind of precise regulation is critical, and really controlling the expression is kind of the next generation, is the way I would put it. In addition, just from a safety perspective, most of our DNA is non-coding. So if you even had off-target in the 80% non-coding and you methylated or remethylated or did something, it is not going to make a hell of a difference already there. But if you cut it could make a difference.
Regardless, you still have to make sure you are looking at the off-target. For example, for our lead indication in the clinic, we looked inside, upside down to make sure there was no off-target anyway. So it is really a way to precisely control that gene regulation. Hope that helps address it.
Yeah. So your initial target, EPI-321, is targeting FSHD. It is a muscular dystrophy. We have abnormal expression of the DUX4 gene. Why does this represent such an attractive target for your approach?
Yeah. So first of all, this is a really debilitating disease with nothing on the market for these patients. So it is something that really needs attention. While it is rare, it is not that rare. It is kind of medium rare, 40,000, 50,000 in the U.S. alone, so it really needs help. And where we can make such a significant difference is that we can go into the root cause of the disease, as you stated, is the expression of DUX4, because there is this region that is hypomethylated that allows DUX4 to leak out.
We can literally go in, bind upstream, and suppress that expression and stop that poison coming out and killing the muscle cells. Where we stand out and why we were very deliberate about that indication is that today, to deliver these kind of constructs, the only known way to I should not say known way. The only clinically validated way to get into the muscle was with AAV, and as we know, the AAV payload is really small, and Cas9 couldn't fit. We have our own proprietary Cas molecule, which is truly compact. We're in a unique situation where we could fit our Cas molecule in, our modulator in, our guide in, still with room to spare, and really use our superpower to use AAV and deliver it and stop that poisonous DUX4 from leaking out.
You've moved quite quickly into the clinic. You've actually presented human clinical data. Maybe highlight some of the key observations thus far.
Sure, yeah. We were really pleased that one of our PIs gave a talk at the World Muscle Society meeting a couple of weeks ago. What he highlighted was we finished dosing. We had six patients at the starting dose, six at higher. Our first six patients have made it to their six-month visit. He reported on the interim data, and what was really remarkable was that the six patients, when we did imaging to look at their lean muscle volume, all gained lean muscle volume where they were all projected to lose lean muscle volume because that's what happens with this disease. What was even more remarkable was when we looked at quantitative muscle testing to see where they were gaining strength or losing strength, they gained strength in the maneuvers that were tied to where they gained lean muscle volume.
That really gave a nice totality of data to say we're having a drug effect. To complement that, we showed with a circulating biomarker that we did in fact reduce ZBTB17, which is a direct transcriptional target of DUX4, and the fact that we could bring that down again complemented that we were having target engagement. That really was remarkable. As you know, the others in the field are also doing MRIs, but nobody's talking about their data, which you can speculate is good data if they're not talking about it. It really stood out in terms of other ways to go about this.
Maybe help investors better understand your data in comparison to both your digital twin cohorts as well as, and maybe equally or more important, the patients in the natural history study that is ongoing.
First, just a little bit about the digital twin. We are working with Springbok Analytics, who is a company that really are experts in doing muscle imaging. They look at whole-body images and measure lean muscle volume, fat fraction, and such in about 140 muscles across the body. What they have done is they have read in MRIs from hundreds of patients over time, and based on that, they put a machine learning algorithm in place so that if you give them a baseline MRI for a patient and some basic demographics, they can then tell you what that patient's MRI is projected to look like in six months.
What we have done is we have given them the baseline MRIs for our first six patients and some demographics, and then they said, "This is what each of those digital twins would look like in six months," so that we had a comparator. If you do not like AI and machine learning, we also have natural history data to compare it to. To your point, this external control, there is actually a lot of data. We are very fortunate that Sanofi has made, because they had a deal with Fulcrum, they have made the Fulcrum data available, and Fulcrum ran a phase II and phase III. They had placebo arms that measured lean muscle volume as well as functional measures. We can look across more than a couple hundred patients' worth of data and actually see what you would expect the MRIs to progress based on that data.
Even more compelling, besides looking overall that we are doing better than what would have been expected, which is loss of muscle, we are taking an approach that the FDA really is, they are saying is the gold standard, which is when you have a natural history study, especially with these rare diseases that have a lot of variability, they like you to do propensity score matching to say, of those hundreds of patients, which ones at baseline were really similar to the patients that you happen to enroll in your study?
We fed in the baseline demographics, how fast they could walk and timed up and go, et cetera, from our six patients, and then fed that in with propensity score matching to the natural history and said, "Pull out the data for the patients that most similarly match ours." It is a really good. I think it is even better than a placebo-controlled because we would not have had hundreds of patients worth of data. Then you can really match them and see how remarkable it is that our patients are gaining lean muscle versus those of the comparatives which were losing lean muscle.
You mentioned you have the ability to assess a circulating biomarker. I think most people are familiar that Novartis, which acquired Avidity, has their own program in FSHD, and they have their own circulating biomarker, KHDC1L, or they call it cDUX. Can you maybe compare and contrast—
Sure—
—your biomarker versus theirs?
KHDC1L is a protein, and unfortunately, Novartis is keeping their assay proprietary. We would love to compare head-to-head because I know we run circles around them, but they are not making it available. What we and others in the field are doing, along with the FSHD Society, is developing another assay to basically measure KHDC1L. We have saved samples so that we will be able to do that. It will never be the exact head-to-head because it will be different antibodies. However, just in terms of comparing the KHDC1L to our DUCZ. KHDC1L, as I said, is a protein. At least to date, we have not seen any publications showing that there is a correlation between that marker and disease severity. That leaves me a little bit nervous. However, the biomarker that we are using, which is developed by University of Colorado, is looking. Again, it is not a protein.
They are looking at cell-free DNA. They are looking at DNA fragments. Let me back up. The way they came up with this biomarker, again, cell-free DNA, as you know, it is not novel. People use it in fetal-maternal health every day. They use it for looking at tumors. What the team in Colorado did was they looked at samples from healthy individuals and patients and said, in an unbiased way, what looks different between patients and healthy? This gene came out, ZBTB17, which we call DUCZ, and it is a direct transcriptional target of DUX4, which is critical. What is even more compelling is they looked at the FSHD patients, and they lined them up based on disease severity, and the biomarker correlated very strongly with disease severity.
That gave us a lot of comfort, more than KHDC1L, where at least to date, nobody has been able to show that. Could be true, we just have not seen that data yet. The fact that it lined up with disease severity was huge. Then we sent Colorado our samples, and what was really interesting was that our baseline data, when you lined it up with disease severity, it fell in the severe range, and the patients that came onto our study were in fact severe. Then when they looked at those same patients six months later, their level of DUCZ lined up between no symptoms and very mild symptoms. It is a very different biomarker, but because it is a direct transcriptional target of DUX4, and disease severity correlates with it, I think it is a really exciting biomarker for this disease.
Okay. Well, great. We will look forward to your next set of data, I believe sometime in 2027 from the study.
Yeah. We will have 12 patients, 12 months data.
Yeah. So congrats so far on what we're seeing. Jian, let me switch to you. Your lead, MGX-001, targeting hemophilia A, the promise of a one-time curative treatment. You do so in the wake of BioMarin's Roctavian, which notably, an approved gene therapy. The promise was it would be one time. Commercially, not very successful. We knew there was going to be durability issues, and maybe some other issues as well as far as just the patient inconvenience with that particular drug. I think at the end of the day, a lot of important learnings that someone like you as a fast follower can benefit from. So maybe to start, talk about how your drug is different, and how that is leading to some really impressive NHP data that we've seen thus far.
Wonderful. Hemophilia A is probably a more familiar disease to us, so let's start with the patients. Despite the modern therapy, children with severe hemophilia today still grow up expecting joint damages from cumulative bleeding. There are 25,000 patients in the U.S. with hemophilia A and more than half a million worldwide. We do talk to the physicians and the patients, and we continue to hear strong desire for a one-time curative treatment. Let me quote, the community is saying to free the people with hemophilia from treatment burden and providing a functional cure with a single intervention, and that's what they are looking for. So you are right. One of the advantages we have as a next generation company into this space is our ability to learn from the early challenges. So how is our approach like?
We take a structurally different approach from Roctavian rather than deliver factor VIII gene in the episomal as in Roctavian, and we integrate factor VIII directly into the chromosome. What that gives us is expected durable expression as the liver cell divides. So how do we do it? We deliver a factor VIII DNA template in the promoterless AAV cassette first, and then separately deliver the LNP that provides mRNA encoding our CRISPR nuclease, along with a guide RNA to make a targeted cut in the albumin gene, and then the factor VIII integrates into the albumin gene, and then just lets the liver's machinery drive the expression of that. We have shown preclinical data and tested out early in the study. First is the durability.
Roctavian would struggle with the durable factor VIII expression, and in our preclinical studies with juvenile non-human primates, we have seen the factor VIII activity reach the peak level within two weeks post dose. It remains stable through the end of the study of approximately 19 months. Not only that, we see the factor VIII activity correlates with the integration as we measure it. That's exactly the mechanism does its job as designed for. The edit stays as a permanent part of the DNA. The second study we're also very excited about is a dose range finding study in non-human primates with exact same drug product components. What we have seen is most animals in the target dose cohorts reach a normal factor VIII range, which is 50%-150% in our targeted dose cohorts. That's exactly what we were looking for.
In addition to that, we did not see any animals exceeded upper limit of 150%, and that gives us confidence in the safety margin. Most importantly for clinical trial design, we saw clear dose-dependent response of both component of the AAV delivers the DNA template, and the LNP delivers the CRISPR gene editing cargo. We are excited about totality of the data. We really think that it supports a differentiated profile. We have been focusing on now initiating phase I/II study next year.
Okay, so let's move on to CMC. Obviously very critical for gene therapy. People would say, again, BioMarin's wasn't optimized for Roctavian. What are the key lessons you have been able to learn from that to optimize your CMC?
That's a really good question because structural approach is different for us. In fact, it works out very well for us. Let me make three points of the difference where we see. First of all, a B domain deleted factor VIII gene. It is still very large gene, takes up the most capacity of the AAV cassette. Being able to rely on the endogenous albumin promoter that eliminates the need to build in the promoter elements like in Roctavian. What does that do is help preserve the AAV genome integrity during manufacturing process, which is one of the critical quality attributes we measure in CMC. The second point is the activity levels we achieved in our preclinical study I just mentioned was based upon a lower AAV dose.
We saw activities reach normal level at 5e12 vg/kg , which is more than 10 x below Roctavian dose level. What does that mean? It is less pressure on the manufacturing scale, the advantages for cost of goods. It means a lot for CMC development to supply, in addition to the safety factor. The third I want to mention is our proprietary CRISPR nuclease also helps us. We use MG29-1, that is type V nuclease enzyme. It appears with a substantially short guide RNA. As you know, synthetic RNA yield drops as the construct gets longer. With a much shorter guide RNA, we have a better yield and better purity profile. Finally, we leverage the platform strengths in-house as well as CDMO.
In fact, we've completed the manufacturing and released the batches for clinical supply that goes into the phase I/II study next year.
Okay. I think an underappreciated but critical function for any company, especially as they move into the clinic for the first time, is how you've set up your clin ops team. Maybe talk about how you've done that to minimize problems and ensure success as you move into the clinic.
That's very important to have a team that have experience. Hemophilia is a familiar disease, so we build our team around the people who have done it. For instance, our Medical Director to lead this clinical study is Dr. Kapil Saxena, who is himself a pediatric hematologist, used to treat the hemophilia young patients. Before he came to the industry, he was Director of Hemophilia Treatment Centers at Boston Children's Hospital and in Oklahoma. After he joined the industry, he led the pivotal study of [inaudible] that is the extended half-life factor VIII replacement therapy at Bayer. He knows the patients, and he's taken product through BLA approval. We build our clinical development function around that.
In addition to that, in designing our clinical studies, we're also taking inputs from global KOLs who have had hands-on experience in developing hemophilia gene therapy, as well as former FDA leaders in cell and gene therapy and hematology to help us with the study design and the regulatory strategy.
Okay. Then, again, one of the drawbacks for Roctavian was the adjunctive corticosteroid use. Given the fact that you can dose lower, how does that impact that aspect, if at all, for your drug and your patients in the study?
Wonderful. Of course, our clinical steroid regimen will be defined based upon the observation in the clinic. But we know Roctavian had a prolonged corticosteroid use, and that was one of the major safety-related concerns by the clinicians in the practice. That was significant in phase III, Roctavian had 79% of patients were on corticosteroids for the length of media eight months to reactively treat the elevated enzyme levels. For us, what we have seen in the preclinical studies was that only mild transient elevation of a liver enzyme with the study. We did one prophylactic steroid treatment before the AAV and LNP dose. Then just it resolved without additional intervention. Given that we anticipate a lower AAV dose going into the clinic, we have the reason to believe that we will have a lower immune burden.
That is what we expect to closely monitor during our phase I/II study and then define what additional treatment we need.
Okay. Well, let me ask you one final question. I know you have not guided to a lot of the timelines on this study, but clearly, your preclinical data shows that the drug can affect factor VIII expression quickly, and we get to steady state levels relatively quickly. Would we be wrong to assume that as you get into the clinic, and that translates the same way, that you would be in a position to report that fairly early on after starting enrollment?
Yes, certainly, we hope the preclinical data plays exact the same in the human trial. In fact, many gene editing therapies have translated very well in the clinic. We definitely have the reason to expect that. I would like to probably address the clinical study design first. We have had a discussion with the FDA in the preclinical pre-IND meetings, mostly about how the study design should look like. It's going to be a phase I/II study, single arm, open label. We'll conduct the study at multi centers. Of course, include the U.S. centers as well as international hemophilia treatment centers that have had hands-on experience and infrastructure ready for our study. We have really strong interest from the investigators in participating in our study. In fact, we have more interest than what we can accommodate in the phase I/II study.
We expect, once we start study next year, we will have a data readout, as in preclinical. The activity rises, reach the peak, right, within two weeks after dose. Of course, we expect to see safety signals very quick in the study. We believe the early data readout from our phase I/II study next year will be meaningful as we guide how we steer this towards the later stage studies.
Okay. Well, we've reached the end of our time here. I think it should be clear to all of you that 2027 is setting up to be a great year for both of you. Thank you so much for joining us.