Beam Therapeutics Inc. (BEAM) Earnings Call Transcript & Summary

Good morning, everyone. My name is Gena Wang. I'm SMid-Cap Biotech Analyst at Barclays. Welcome to Barclays Global Healthcare Conference. It is my great pleasure to introduce our next presenting company, Beam Therapeutics. With us, we have Giuseppe Pino Ciaramella, Chief Scientific Officer and President. Pino, do you want to give a brief overview before we dive into specific questions?

Yes, of course. And thank you very much for the opportunity to come and share some of our updates. Yes, so Beam Therapeutics is a next-generation gene-editing company that uses primarily this technology known as base editing. It is still a Cas9-based technology. But we use Cas9 as the landing pad and the targeting domain. But we have changed in a very important way so it no longer makes the double-stranded break, which we believe leads to some unwanted outcomes in editing. Instead, fuse to the protease, is an additional protein, which is a human enzyme called deaminase, which is capable of catalyzing the direct conversion on the gene of a nuclear base to another. And depending on which deaminase we use, we can convert a C to a T or an A to a G. And obviously, as you know, an A or a C are present in any base pair in the genome. And so this gives us the opportunity to generate a very versatile editing technology. They not only correct point mutation, which is -- has proven to be very challenging for the first generation of editors based on nucleases, but also gives us the opportunity by doing that to do all sorts of different edits. In fact, we have products where we can knock out proteins, particularly in our BEAM-201 oncology cell-based therapies. We can actually activate genes by making changes in promoter regions of genes. This is the case for our BEAM-101 sickle cell disease program, as well as doing this in a multiplex fashion without running into the consequence of -- because of double-stranded breaks of chromosomal rearrangements, which obviously can lead to some unwanted effect. And so that's -- it's proven a very robust versatile technology that works well, both in replicating and non-replicating cells because it's based on a direct enzymatic conversion of one nuclear base to another.

Thank you, Pino. So maybe I wanted to ask you your technology in terms of -- because you do -- based on the CRISPR-Cas but you use deaminase. And then you did have already IND clear as well and that's the ex vivo part. So for the in vivo systemic delivery, do you think [indiscernible] the, say, the IND package or safety package what will be additional -- anything differences if we think about the R&D package submission compared to, say, class of CRISPR/Cas9 technology?

Yes. So for the in vivo submission for classic CRISPR/Cas9 versus base editing, I don't see any difference between the two. The only sort of minor thing is that, there is a one-off target biology assay that is more relevant to base editor that it's not. But it's essentially a version of the same assay that is used in the context of nucleases. In that context called GUIDE-seq, we have ONE-seq and essentially, what it is, is an in vitro assay that is capable of generating essentially to recognize wherever in the genome a base editor has been introduced. And this is, thanks to the fact that a C is usually converted to a U or uracil. Before -- and that's read as a P. And the adenosine is converted to an I, inosine, which is read as a G. But both U and I in the context of DNA can be very specifically recognized by particular enzymes. And so those when they recognize they make a cut. And then at that point, it becomes like GUIDE-seq you essentially introduce probes wherever that cut has actually occurred. So that's the only, asset is different, but the package for the rest is essentially the same. And the question about is the package the same between ex vivo and in vivo, there is really one additional considerations that selling regulators and our interactions with the FDA in pre-IND discussion for 301 suddenly confirm that, that is all to do about the potential for transferring the edit to progeny. And so obviously, in ex vivo setting that's not a concern. But in vivo, that could be a concern. And so part of your package needs to ensure that both through a biodistribution analysis but also maybe through progeny studies, you demonstrate that essentially you are not targeting reproductive cells. And that's really the main difference. The rest is fundamentally actually pretty similar to in vivo LNP packages that I've seen in the past in my previous life from Moderna, where we also have been seen P&L. So there is no real unexpected concerns.

Okay. Very good. I have several questions. Regarding , I think the first step, you were saying the ONE-seq you are using. How do you tell the, like A-to-I or A-to-G, C-to-U from the [ polymorphism ]?

The enzyme is -- these are really clever. There's essentially because in DNA, you don't see U or I presumably. So they can actually scan the DNA. And whatever there is a U they actually make a cut or whatever there is an I, they make a cut. And that's -- so the -- if you remember, when we talked last time, essentially, the strategy for off-target biology in general is essentially a funnel way. So what you do at the very beginning, you create the worst-case scenario. And typically, you do this by a series of in silico and in vitro assays, where essentially you expose naked DNA, which is the worst possible case to saturate the doses of your editor and you generate basically the possibility wherever it might happen of creating essentially a base edit in this case off target. That then allows you to generate essentially PCR probe at all of those places where then when you do the edit in the context of your cells of choice, let's say, CD34 or primary HSPC, you then use the PCR to very specifically determine whether the theoretical off-target has actually occurred or not. And on top of that, you can actually determine the percentage to which it occurs, if it does occur. So it's really a very, I would say, very sensitive actually funnel that characterizes that off-target biology in a very meaningful way.

Sorry, Pino, so like you still have like that bring back like biased versus unbiased full genome sequencing. So here will be biased because you have your certain sequence in mind?

The original assays would be what you would call biased because they are based on a degenerative sort of library, but it's very extensive. But you can couple that to a non-biased assay, which is called Digenome-seq. And basically, the two together will give you essentially the broadest possible opportunity to interrogate off-target biology.

Right. And with this combination, FDA is totally fine?

Yes, we're talking about this now 5x or so with the FDA and is absolutely acceptable to them.

Okay. Good. Very helpful. And the other question is the ex vivo versus in vivo, you progenic cells distribution in the tissue. So maybe any learning you can have from, say, Intellia's R&D clearance or your partner, Verve hold anything you can learn with your 301?

Yes. So we obviously, we have had extensive discussion with the pre-IND with the 301. So of course, I don't have access to the submission from Intellia. We know a little bit about the Verve, but the -- really, the pre-IND conversation we had with the FDA was very clear in terms of the guidance they provided. And so it was also clear to us that the fact that the Intellia has actually cleared the IND is not surprising to us because in the pre-IND feedback that we've received, there was -- it's obviously a very significant data package, but there was nothing that would prevent us almost like a systemically ever sort of creating this impossibility for an IND to be done. We didn't see that. It's just -- it's a very heavy data package, but nothing more than that. With Verve, our perspective is maybe they didn't quite meet all of the data that they wanted to see. But it wasn't anything that was of concern of base editing versus nuclease. There's no biological rationale for that to be the case.

Okay. And then are -- are you on track to submit by the end of this year ?

Yes. So for 301, we guided that we'll be either at the end of this year or early next year, and we're on track for that. We've also accelerated 302. This is the alpha-1 antitrypsin program to the extent that we can, and we are continuing to guide to be on track for. In this case, it's probably likely to be a CTA submission early in '24. And the reason why we are going for a CTA submission rather than IND list in the first place is because -- in many parts of Europe, the protein replacement therapy for alpha-1 has not been licensed. And therefore, it gives us the opportunity to generate a much clearer readout so that we don't have to worry about [indiscernible] apart the up regulation by the edit versus the protein replacement that's there. And so that's why we like it. But eventually, we will, of course, generate data in the U.S. as well.

Okay. Very helpful. So going back to 301, the R&D package like now we still have maybe less than a year. What are the steps you need to prepare and what data package?

Yes. So basically, there is a sort of IND tox studies, which is similar. Typically, it's rodents and non-human primates. And then the -- really the -- sort of the one thing that makes this guidance, to some extent, a little bit movable at the moment is also the formal biodistribution study needs to be done with the final drug products. So the final mRNA for the edited -- and the final LNP. And the biodistribution, although we have data already that we've done in non-GLP conditions, there may be the odd cell that might be different between the two. And so we don't know until the data packages we generated, how many off-target biology cell packages we need to do. So one of the additional things that FDA want to see in the in vivo setting is that the off-target biology should be done not only in the target cells, but also in all of the cell types where the drug product distributes to. And that's not just at the tissue level. So for instance, if it goes to the spleen, they want to see also which cell types within the spleen does the RNA actually go to. And then they want to see the same off-target package for that cell as well, regardless of whether you see editing or not. And so it may swing, if you will, the amount of data that we need to generate at that point. But other than that, the assays are the same. It's just a matter of generating the information.

Okay. That's super, super helpful. So tox study of rodent, non-human primates, usually how long they would need?

They usually -- so there is a period on target. Basically, there's going to be -- we're going to follow them for a month, but then we also have a recovery period that will be longer than that. And so overall, the study will take between 4 and 6 months.

Okay. And then the -- I think if I got you correct, the non-GLP, the biodistribution and then you are not sure if GLP biodistribution will be required by FDA?

We want to do the biodistribution with a formal final drug products. So this is really the material that essentially is going to go pretty much into the... And so that's really the -- we reserve a little bit of flex in that. But we're not expecting huge deviations from what we're seeing already.

Okay. Good. Very helpful. And then actually, biodistribution, would they have some kind of requirement, how much excess of doses if you try to translate that to say -- therapeutic window in human, will FDA have certain requirement, what dose you need to begin. Depending on the dose, maybe the biodistribution will be slightly different?

Yes, they don't. But typically, what we do, due to the highest dose that you can actually go in biodistribution. And so it's just -- again, you create the worst case scenario so that it offsets the risk.

So basically, animal like a maximum tolerated dose.

Yes, it would be a little thereabout. You don't want to go to the -- yes, it will be the maximum tolerated dose. You don't want to push it where signs of intolerability exists because then that could change the biodistribution as well.

Okay. Very helpful. And then CTA and R&D package, any differences there?

Yes, there is some -- typically, the FDA is a little bit more conservative in the package that they want. So for instance, the progenic study, it's unlikely to be fully required by regulators, although they certainly prefer to see that data prior to dosing of some cohorts. And there may be that the requirement of the different cell types for off-target biology is less stringent in other. But we don't yet have formal regulatory interactions, frankly, with other agencies to be able to completely guide you on that. But that's from prior experience and also what we know from other...

Since you are going back to U.S. anyway in the future, like will you just collect -- or the same level of the details as if you prepare for the 301?

With the 301, we're still considering both U.S. and ex U.S. and in part that's driven also by where the patient population actually is. And so we are definitely aiming to prepare a package that is consistent with U.S. and then obviously, some of the package can be redeployed in undersetting as well. We will be doing the same with 302. As I said, we plan to have IND filed for the two programs anyway, and we're just going to plan ahead with doing that.

Okay. That's good. Very helpful. And I do learned a lot with this. And now going back to your clinical program, sickle cell, the BEACON study. So first patient already enrolled in November 2022? Have you dosed that patient? Any status regarding the patient enrollment?

We have not dosed yet. We guided that typically, this is -- remember it's a very complex process that they need to but we expect on average that to take about 6 months, and we're on track for that kind of time frame. Remember, these patients need to, first of all, go into transfusion to count the bone marrow. They then need to go through mobilization, typically 2 to 3 cycles of mobilization that's to essentially generate as many CD34 cells that we possibly can. Then those cells are sent to the manufacturing sites, whether they're electroporated and the editor is introduced. Then there will be frozen and undergoes all sorts of tests for release, then back to the clinic where there will be essentially conditioning of patients initially in this case with busulfan, but hopefully in the future with our improved conditioning regimens, and then finally dose. And so you can understand why technically it takes the 6 months. But we're on track. And in fact, we recently -- we announced a protocol amendment that actually is accelerating the way in which the sentinel cohort can actually be enrolled into the study. So there is much more compression between the first patient and patient#2 and 3. And such that we have guided now that we believe that we will complete the sentinel dose and we will initiate enrollment of the expansion cohort by the end of this year. And that expansion cohort is really what is on the critical path to the BLA. The design of the study is intending to be pivotal and in a design which is very similar to what you've seen for CRISPR, Vertex. And so the quicker we start with the expansion the quicker obviously, we can go to the end finish, and that's our aim. And part of the acceleration on 101 is to reduce the time between Vertex launch and us coming to the market.

Okay. Very helpful. And so when do you think it will be realistically shown -- like the -- show initial data?

We announced that we're going to be waiting until 2024 really to share some more meaningful clinical data. We have chosen not to do what has been done in [ Adicet ] where single patient data was really -- we just don't think that it really clinically meaningful to do. Having said that, we will probably make some -- we haven't decided yet, but it's likely that we could let -- obviously community know that we have formally initiated the enrollment of the expansion cohort and you can infer from that, that we must have had a successful engraftment for the first patient and first sentinel. Otherwise, we wouldn't be able to get there. So there is an element of being able to be reassured that the progress of the study is on track, whilst not disclosing clinical data, they may not necessarily be meaningful to interpret following just a single patient for instance.

But then with all your data, would that be a little bit difficult to enroll patients to let them know?

No, we're actually seeing incredible desire actually to come on to our trial, that there are several sites that are waiting for this sentinel to progress in order to actually -- patients have already been guided for that. So there is a real desire actually for coming on to the trial that we are set up regardful of the data. And I think that frankly, our preclinical data with models that are very translatable on what we've seen is certainly the best that I'm aware of that has been published. And so if that's what to be reproduced in the clinic, there is a lot of interest to be that patient.

So then maybe going to the competitive landscape. And so we saw quite a few others, all showing like around maybe 40% range, the fetal globin. So what will be like clinical profile, you think it will be competitive?

Yes. We would like to be able to see the reduction of hemoglobin S to about 40% or less. And in preclinical data, that's where we're at. We see a 60% to 40% ratio between the [ AFAP ] regulation that we make. And then the reason why that 60% is important, we feel is because it's only when you get to those levels, then actually you start to reduce in the hemoglobin S. Because, as you know, there is a -- is called the locus control region, but basically there's a finite amount of hemoglobin protein that any given cell can make, and you need to push the system to making a lot of F to reduce the amount of S that it's making. Vertex and CRISPR at the moment, they still have about 55% hemoglobin S last time I've seen it. And so that's certainly something that we will look at that. We expect that resolving VOCs will be similar to what Vertex and Bluebird have seen. So it's difficult to really compete above 90-plus percent. And so we expect our product to have to do exactly the same thing. But what we will be looking for is even deeper biomarkers of resolution of all of that inflammatory [indiscernible] that sickle cell disease makes over and above VOCs that actually contribute to the disease. And that's where we think our profile is very competitive. And it is still intriguing to me that at the moment from the Vertex, we've only seen the VOC data. To my knowledge, I haven't seen any additional biomarkers like resolution of hemolysis [indiscernible] with the blood, half-life of cells that would provide a much more comprehensive clinical package of resolution of sickle cell disease. And that's what we aspire to generate.

That's very helpful. So the -- you do have also Wave 2 and Wave 3 conditioning regimen. And I think that some feedback from sickle cell community, the conditioning regimen is one very major hurdle there, the important consideration for patient or willing to take the new -- next generation of genomic medicine. So maybe a little bit of update on the Wave 2 and Wave 3.

Absolutely. Yes. So clearly, busulfan, as you mentioned, is a major limitation, particularly due to the sterility that it causes. Having said that, though, I really wanted to stress the point that we do see that even in that setting, there is a very meaningful market. And if you look at that is every year, even with busulfan, there is about 1,000 patients that want this particular transplant with busulfan and yet only 20% are eligible because of the donor issue. In this case, we don't have the donor issues because it's autologous. And so even if you envisage a 1,000 patients a year, with a price point per treatment of $2 million to $3 million, that's a several billion dollar market already. And so I don't want investors to discard that like that. But having said that, we think we can expand that significantly if you eliminate the busulfan. And so one of the approaches that we're making is what we call ESCAPE, which is we are benefiting from the precision and the ability of base editing not to do double-stranded break. So that now we have a multiplex edit where not only do we have the edit that corrects sickle cell disease. But also now we have an edit that makes those edited cells basically stealthy to an antibody that can instead eliminate basically the non-edited resident cells. And through that approach, we feel that this would create a huge opportunity not only for sickle cell disease because obviously, we've eliminated the toxicity associated with busulfan but could potentially become a very transformational approach for transplants in general. That kind of approach we'll be able to be used in other transplant setting for oncology and other diseases.

That's super helpful. So the -- I think that we have -- only have one minute. I wanted to ask you about the allogeneic CAR-T program [indiscernible] and rationale for selecting CD7?

Yes. So still on track for dosing our first patient media. This is initially for CD7 positive T-ALL patients, but also expanding to CD7 positive AML which is, by the way, is about 11% to 15% of the patient population. And the reason for basically picking CD7 is that CD7 is well characterized and expressed target in all of the T cells. In fact, one of the reasons for going after that is that originally, this TLL were left a little bit behind because when you target CD7essentially, you are killing both your CAR-T cells as well as the cancer cells. This is a phenomenon called fratricide. And you need to eliminate CD7 from the cells in order to prevent that. And that then pushes you to do many several edits. In fact, BEAM-201 is the first quadruple edited CAR-T cells that I'm aware of, and we were able to do that without chromosomal rearrangements. And that's in part what we've done. For next generation of CAR-Ts, we really believe that we need to get to an allogeneic series of edits that are much more persistent than what has been possible to do so far. We don't believe the B2M knockout alone is going to be sufficient. And I think the clinical data is supporting that view. We're working very hard on making additional edit. But ultimately, we believe that you need 4 to 6 edits to be able to get there. And that's why base editing is ideally placed to be able to do that.

Okay. Well, thank you very much, Pino. We look forward to most of the updates. Thank you.

All right. Thank you.