Goldman Sachs 47th Annual Global Healthcare Conference 2026

Okay, we'll continue with the next session, which is Foghorn Therapeutics. I'm Paul Choi, and I cover the SmidCat biotech sector. It's my pleasure to have Adrian Veltschuk to my immediate left here. Maybe what we'll do is what we've done in prior sessions, and maybe we'll let Adrian sort of describe what are Foghorn's strategic priorities for the remainder of 26 and going into 27, and then we'll go into Q&A. Sounds great, Paul. Thanks for having me.

Thanks to everyone for being here today. Yeah, look, I think next up and most importantly as we think about 2026 is our FHD909 program. This is the selective SMARCA2 inhibitor molecule that is partnered with our colleagues at Eli Lilly in a 50-50 strategic collaboration with them. So our guidance hasn't changed over the course of the last six or so months. I anticipate that sometime over the course of the summer, we'll be in a position where the collaboration will make a decision whether we proceed to expansion or not, and I'm sure we'll get into some of the details around that. So that's coming up over the next several months. We've got an exciting proprietary pipeline behind the Lilly collaboration, roughly this order. We have our CBP selective degrader. That's for ER-positive breast cancer. We've actually completed our non-GLP tox studies. We're actually doing a bunch of animal pharmacology work, both as a single agent, but also in combination with oral SIRS and CDK4-6. That's gating for us to then go into GLP tox, which would likely come towards the end of the year, setting up an IND and potential entry into the clinic in 2027. We have our selective EP300 degrader, which is targeted for a range of hemoalignancies, but certainly multiple myeloma, potentially diffuse large B-cell lymphoma. Very excited by that program. The gating or next critical step for us is formulating that as a once-a-week subcutaneous formulation. If we're able to do that, that's tracking to the non-GLPTOC studies sort of in the Q3, Q4 timeframe, frame, also setting up a potential IND next year. Again, excited by that just based on some competitor data in the clinic that has already validated that target or program. We have an INI asset that's not disclosed on our pipeline. We're still keeping that very much below the radar screen. I won't have too much to say about the target or anything related there, but that's a small molecule. It looks in several different animal models, looks as good as one of the major antibodies that's used in this particular autoimmune disorder. That's tracking to non-GLP tox studies towards the end of the year, also in IND potentially next year. And then last year, ARID-1B selective degrader. We've made really nice progress on that much earlier than the other three things I just mentioned. That's tracking to in vivo proof of concept this year. That's sort of the setup and lay of the land. So a lot of different things between the clinical decision for 909 with Lilly and then our proprietary pipeline.

Great. Investors may be somewhat familiar with Smarca, given that there was an asset that was being investigated for a little bit. I think it got put on hold. Can you maybe tell us what do we know about Smarca, I guess, at this point, and just sort of what is its prevalence across various or major tumor types here? Right. So maybe I'll just take a step back on this and

and just describe the biology at a high level, why we're interested in the target, and then talk about some of the relevant tumor types or where we're interested and excited. So SMARCA2 and SMARCA4 actually, ATPases, they're actually part of a broader chromatin remodeling complex that's responsible for opening and closing chromatin, vast oversimplification in the biology there. They're mutually exclusive, meaning that the machinery only has either SMARCA2 in it or SMARCA4. And the observation that has come out over the better part of a decade now is that individuals that lose that SMARCA4 engine, or ATPase, for their survival on SMARCA2. And so we had set out actually at the inception of the company roughly a decade ago to try and selectively drug SMARCA2. We ended up finding a selective allosteric inhibitor, which is 909. And that's what we've actually put into the clinic with Lilly. And the relevance here, specifically in non-small cell lung cancer, is about 10% of non-small cell lung cancer harbor these individuals, these patients, have a SMARCO4 mutation. It's mutated in about 5% of all different solid tumors, but we don't know exactly the relevance of that. We don't know exactly the synthetic lethal setup there. The most well-established one that we've spent time on within the collaboration is indeed in non-small cell lung cancer. We'll see whatever the clinical data shows. But that's sort of why it's relevant. And maybe one last point that we know from the various literature and studies that have been done is the patients that have, the non-small cell lung cancer patients that have a SMARCA4 mutation, their prognosis is far worse than those who are wild type. There's plenty of literature out there that shows that the patients who receive chemoimmunotherapy, if you have a SMARCO4 mutation, basically you cut their efficacy across all parameters in about half.

Can you maybe walk us through, you know, what is the reason for targeting SMARCO2 instead of SMARCO4, you know, and why is selectivity important here? And, you know, I've always admired the medicinal chemistry, you know, prowess at Foghorn, and how were you able to overcome that technical challenge of targeting SMARCA2 with an inhibitor here?

So we know from our own experiment, we had a prior molecule, which was FH2286, that we stopped the clinical studies on. It was actually a dual inhibitor, equipotently targeting both SMARCA2 and SMARCA4. Healthy cells have both of these proteins, by and large, in it. And the hypothesis is that, you know, you'd have a much narrower therapeutic window. And that's actually what turned out to be true. We ran that experiment with 286. We ran it in uvile melanoma and AML. It just wasn't very well tolerated. We did see efficacy, but, you know, you just couldn't push the dose as high as you needed to, quite frankly. And so the setup here is to try and make sure you're only drugging or being more selective for SMARCA2 so you're not engaging SMARCA4. you're not going to get all of these tolerability side effects, potential safety issues that arise from dual inhibition. And again, I've commented publicly, I haven't commented on any advocacy, but by and large so far in our clinical study, the drug has been generally safe and well-tolerated. So again, we're still escalating. We're backfilling cohorts that we can talk about. So the how do you drug it selectively, that has always been the issue. And the two approaches that have sort of surfaced over the course of the last several years have been enzymatic inhibition, which we've obviously achieved, as well as targeted protein degradation, which Prelude and some other current competitors are trying to tackle. The challenge with drugging this target selectively is that SMARCA2 and SMARCA4 are about 90% similar if you look at the amino acid sequence of the structure. So it's really hard to find something that differentially engages one of the proteins, but not the other. And if you look at the ATPase binding pocket of the proteins, they're almost identical. And so we set out with our platform where we could make, I'll simplify, this full remodeling complex with the relevant SMARCA2 or SMARCA4 protein, and we could then screen chemical matter and see actually which ones were more selective for one over the other when it was actually studied in its native context. That's a pretty great simplification for how we were actually able to find selective chemical matter.

Great. You mentioned earlier that smarker mutant patients might constitute approximately 10% of the non-small cell population. So pretty sizable here. But can you maybe walk us through, you know, currently how are these patients identified? Do they fit into the sort of regular screens or do you need NGS here? And then more importantly, what is sort of the typical prognosis based on the available treatments that are out there for non-small cell lung cancer?

Right. So my understanding, at least sort of in the United States, is if you have lung cancer and you get a biopsy, this is just part of the standard NGS panel that's run. So as an example, the SMARCO4 mutations are in the foundation screening in Tempest, et cetera. So finding out whether you have a smarcoform mutation seems very readily accessible and available, especially in this day and age. The second part of your question, sorry, was...

Just what is the clinical outlook or prognosis for these patients? Yeah, what's the prognosis

for these patients? So unfortunately, if you harbor a smarcoform mutation, you have a far worse prognosis than not. Maybe some other important facts here, which is if you look at some data coming out of Memorial Sloan Kettering, and you look at and see when they sequence these tumors, there's by and large not a lot of overlap with other driver mutations like EGFR, ELK, RET, or MET. However, there's about a 33% or for every 10 patients that have SMARCA4, roughly three out of those 10 patients also will have a concomitant KRAS mutation. So let's just deal with just the the pure SMARCA for patients for a moment. So in some of the literature and studies that have been done, you know, in the retrospective analysis on the patients and how they performed, so in a frontline study, you know, first line non-small cell metastatic setting, patients receiving a PD-1 and chemo, response rate for wild type is about, call it 39-40 percent, median PFS of about six months, overall survival is about 15 months. In the SMARCA4 mutant population, you cut that in about half. So response rate's about 22%. Median PFS is about 2.7 months. Median OS is about eight months. So they progress rapidly through the various treatments. They have about half the efficacy in that frontline setting. They're relatively non-responsive to that frontline treatment. And again, because they don't overlap with any of the other major driver mutations, There's not a lot. I can tell you without getting into any of the clinical data per se, we're seeing patients in the backfill cohorts as we're enriching for lung cancer. They're predominantly fourth or fifth line patients. They have progressed very rapidly through the other lines of treatment. So these individuals, unfortunately, have a much worse prognosis. The unmet need here is pretty significant.

You talked a little bit about backfilling. So maybe, you know, help fill in the picture on, you know, what is the phase one design that you and Lily are currently pursuing? How does this work in terms of the study you're conducting? And then you talked a little bit earlier about, you know, potentially talking about updates during the summer here. Can you maybe remind us what are some of the endpoints and metrics, and, you know, how do you assess this in this sort of later line population?

Right, right. So just to set context here as well. So Louie is responsible for all the day-to-day operations in the clinical study. We started dosing patients as a collaboration back in October of 2024. Initially, the sites we had open were about five sites in Japan, 15 or 16 sites in the United States. We started, and the escalation was sort of any solid tumor as long as you had any Smarco4 mutation or Smarco4 deletion. we've roughly cleared a cohort every couple of months or so we've been added 18 plus months you can do the math plus or minus in that range in terms of the number of cohorts that we've gone through in the latter part of q4 of last year per protocol we started backfilling some of the cohorts and and and the backfilling is relevant because that's where we're trying to enrich for non-small cell lung cancer patients, and we're also enriching for this loss of function or class one type mutation. That was triggered by either achieving one of the two things, which was hitting the predicted IC90 exposure, right? So we're doing blood draws, so we know what the exposure level, but based on the allometric scanning from the animal models, we believed we had either hit the IC90 or were close to it, and or we had seen some clinical activity, tumor shrinkage, some sort of sign that this drug was actually in therapeutic range. We've not commented on whether it was one or both of those things, but needless to say, we started backfilling the cohorts, you know, roughly in the late part of Q4 of last year. That's the group that I'm most interested in assessing for efficacy because that's where the lung cancer patients are. So based on how things have unfolded, and this is my guidance, I'm in no way speaking for our colleagues at Lilly, I believe that sometime over the course of the summer, I don't know if it will be, you know, July timeframe or August timeframe, I believe we'll have enough non-small cell lung cancer patients somewhere between 10 to 20, maybe a little more than that. We'll see where we'll have enough radiographic scans to understand is the drug active as a single agent and is it meeting whatever bar that Lilly has in their mind? I can speak to the bar that I have. Again, won't speak for Lily. So that's what I think what's going to unfold over the summer, fixated again on this non-small cell lung cancer patient population. Great. Since we don't have a historical

analog in terms of SMARCA targeting drugs in lung cancer, I guess maybe, Adrian, can you elaborate on that? I guess based on KOL feedback or physician feedback, have you heard, would constitute a good response or a good result, I guess, rather, in the fourth or sort of fifth line lung population here? Are we primarily focused on survival metrics, ORR, disease control rate? Maybe just

help fill in the picture. Right, yes. So obviously, it's a safety study first and foremost. So we want to make sure whatever doses we get are well-tolerated, generally safe, because the end game here will be to ultimately move towards frontline and certainly in combination, likely with pembrolizumab, but maybe with a KRAS inhibitor at the relevant point in time. So I'll give you my view, and I'll tell you sort of anecdotally what we've heard from some of the KOLs. I'll start with the KOLs. Depending on who you talk to, someone will tell you 10%, 15%, 20% with some duration of response, you know, a few months. Keep in mind, you know, I mentioned a few moments ago how these patients fare in the first-line setting when they're given a checkpoint inhibitor plus chemo. Median PFS in a large population is about 2.7 months with a response rate of about 22%. percent. Given that this is fourth, fifth line patients, there are no studies. There's not really, there's just anecdotal conversations. Patients who are coming into our study are incredibly sick individuals, usually life expectancy measured in number of weeks, not many, many months. And so my personal view on this is with a denominator of somewhere between 10 to 20 patients, I'd like to see three or so partial responses. I'd like those to have some durability. So I'd like to see in the three to four plus month range to make a decision from a development perspective. We're not gearing this towards some overall survival, but we'll have a sense for how long people have been able to stay. I don't think we have to follow patients for the entirety of the lifespan per se in order to make a decision. We need to know that this is active as a monotherapy such that whatever we go forward with, we have a reasonable chance of actually helping patients when we move into the earlier lines of treatment. So that's how I think about it. We'll see, obviously, how things go over the next few months. Great. Earlier, you mentioned

about maybe 3 in 10 patients here overlaps with a KRAS mutation. And so, I guess, given that subpopulation here, how do you think about potential either sequencing of either your drug or KRAS therapy in this population? Do combinations make rational sense here? And just sort of how do you think about the potential and developmental path, given that this is a substantial chunk of the addressable population?

Yeah, so we've done a lot of preclinical work in this space. We've actually had multiple models with pembrolizumab, one genetically engineered model, one syngenetic model where we're using, we haven't altered the animal's immune system. So those data are pretty striking. Those were presented at AACR just a couple months ago. On the KRAS side, our colleagues at Lilly obviously have their own portfolio of KRAS molecules. The preclinical data that's been generated has been generated with their KRAS therapies. And again, the combination of the 909 molecule plus whether was a pan-KRAS or a KRAS G12C or a G12D, looked pretty striking. I can't comment on when Lilly would want to potentially run that study or what the sequencing would be, but I think there's a lot of rationale to actually go out and look at 909 with a KRAS inhibitor. At present, as you probably know, the KRAS drugs, at least the ones that are currently approved, tend to be reserved for a second-line treatment because they haven't yet demonstrated that they can outperform from the IO chemo regimen in the frontline setting. That may change with some of the newer therapies that are coming with different side effect profiles or improved side effect profiles and potentially better efficacy. So I think it's fair to say that that is of interest to the collaboration in the fullness of time. We'll just have to see sort of how that field plays out. Obviously, there's a lot of exciting new data with KRAS these days.

I want to touch briefly on your other approach to SMARCO, your degrader program, but just sort of, you know, again, given the prelude analog in the category, you know, what are sort of the learnings there, I guess, and, you know, how do you think about potentially improving with a next-generation degrader like yours?

So one of the things that we know, having studied this target for as long as we have, is that you have to hit it really hard. If you're an enzymatic inhibitor, that means hitting at an IC90, not just in a brief little CMAX and then pop down, but LILI sustained target coverage. The same is true, if not even more important, when you're degrading this. And so I can't comment about the specific details of the degrader program with LILI or status per se, but we know that you need to be north of 90, even maybe 95% on a sustained basis for degradation. And there's actually a very nice linear correlation that we've been able to generate with ranges of degradation percentage and correlation. Again, it's actually a very nice linear relationship with the TGI or regression you achieve. And the only times that you can get regression is when you're at 90, 95 plus percent. And when you compare that actually to our inhibitor 909, we've not yet had a degrader outperform what we've been able to achieve with our inhibitor. Preludes, molecules, they had both a VHL, which was intravenously delivered. They had a cereblon, which was orally delivered. In our hands, those molecules never really got past about 80%. Again, we believe we made their clinical candidates based on their patents. We replicated the data. And I think that's part of the challenge is you need a very rapid degrader. Selectivity is important at some level, But I think you have to have a rapid, deep degradation of that target. So I think the fate of the degraded program is obviously tied to 909. We'll see what happens with 909. I think then we'll be able to comment further.

We've spent a lot of time talking about lung, but you have tested 909, and Lily has tested 909 in other tumor types. And so maybe just broadly, you know, what constitutes good results in those particular subtumor types from your perspective and what, I guess, you know, beyond lung would potentially make the most sense to go after should the results be positive?

So, you know, the gene or the proteins mutated in, you know, 5% of a range of different solid We have a relatively, I think I can say, eclectic mix of a lot of different solid tumor histologies in the escalation. I can't comment on which one would be the right one at this point to potentially prioritize, but I think the clinical data will sort of point us in that direction when all is said and done. I think it's all going to depend on the depth of the synthetic lethal dependency that exists, the type of mutations. Again, it's going to be a smaller end relative to lung cancer, which has always been the priority. So we'll just have to wait and see what tumor

types could be a potential interest for us. Great. I want to turn to some of your non-partnered assets and what else is in-house, and specifically CBP and EP300. Can you maybe walk us through what is the meaningfulness of selectivity between these two targets? Why is that sort of clinically important, and how do you achieve that from a drug development perspective?

Right. So maybe, again, just a context setting on CBP and EP300. These are histone acetyltransferases. Fancy where I'm just saying they're depositing acetyl marks on various proteins as they're regulating gene expression. um this is uh actually these targets or these proteins have actually been uh drugged or attempted to be drugged by others before uh genentech way back when had a dual uh cbp ep300 broma domain inhibitor format therapeutics which i believe was was acquired or assets required by novo similar thing um our colleagues at cell centric most recently are the ones that are actually being they've demonstrated i think some very compelling data in multiple myeloma the challenge with hitting both of these targets without selectivity is it ends up having a myelosuppressive effect so you end up impacting or you end up with the range of cytopenias from impacting platelets so you get thrombocytopenia neutrophils with neutropenia a bunch of other white blood cells and leukopenia and i think that's been clinically now demonstrated by our colleagues at cell centric and they're dosing regimen four on three off so we've set out to try and selectively drug either CBP in the case of ER positive breast cancer or selectively drug EP300 in the case of a range of heme malignancies, multiple myeloma, diffuse large B cell range of and potentially prostate cancer as well. And we've done enough of the experience where we've done selective drugging. We have a dual degrader, we have all the dual inhibitors, and we can see that this is very much with dual inhibition or dual degradation. It's very much an on-target pharmacodynamic effect of impacting these other cell lines. And we think ultimately that that's going to be rate limiting or efficacy limiting is probably the better way to describe it. Once one is in the clinic, you're either going to have to down-dose, as we're seeing with our colleagues at CellCentric, or you're going to have to down-dose and dose intermittently as they're doing four-on-three-offs. So I think that's been the challenge, again, sort of in our sweet spot of drugging really similar proteins and finding a unique way to

get at it. The world of multiple myeloma has become increasingly complicated in terms of treatments, whether it's CD38, BCMA, plus the classic Revlimid or Imids and things like that, plus POM and DEX. And so I guess as you think about the developmental path in multiple myeloma and other hematologies, you know, first, you know, how are you addressing this and identifying sort of, you know, what is a relevant, you know, population from a preclinical perspective, and then as you think about sort of down the road, you know, what is the, you know, sort of natural starting point from a clinical development perspective? You know, the majestic study that

you just referenced from J&J, the PCMA CD38, I mean, those are impressive data, and I think it's amazing for patients. I think over time that's going to maybe faster than we all think. That will certainly work its way into that frontline setting. But we know patients become refractory to treatment. And so I think there's still a lot of room and opportunity. Everything in multimyeloma at this point, as you were just mentioning, is combination. And so I think the goal for us is to enhance the selectivity and where I think we'll have a leg up over an over is to be able to combine with a range of those different therapeutics and not have an additional myelostepressive or myeloblative effect. I think the starting point for us clinically on this will be probably a straightforward dose escalation where we could take a range of different heme malignancies. Maybe we'll focus in one or two. And then as we get to relevant dose, ostensibly efficacy, I think then you'll see some expansion cohorts where we'd probably focus 20 to 30 or 40 patients in multiple myeloma, as an example, potentially, you know, a monotherapy sleeve, but likely combination minimally with an IMID or DEX or maybe combination with anti-BCMA. I think you'd see sort of similar sleeves, if you will, you know, again, obviously capital dependent, looking at diffuse large B-cell, maybe, you know, some of the T and B-cell other lymphomas that you believe we're relevant in. At this point, every multiple myeloma cell line that we've tested, I can't think of any exceptions right now, seem to be responsive to the agent. Diffuse large B cell, on the other hand, we believe we may have a marker, and it's about 60% of those cell lines. So there you may be able to focus on. But those are both very competitive heme spaces. So I think we could certainly develop this through sort of a phase 1A, phase 1B dose expansion. I think longer term in the multiple myeloma market to compete with the likes of a J&J BMS as they're continuing to iterate with their IMID pipeline, we would need a strategic partner down the road for this is sort of my belief. But I think there's still room to play in these therapeutic areas. Obviously, it's going to evolve over the next few years, but until we cure every patient with cancer, I think there's still room to play.

Yeah, I think, as you've sort of mentioned a couple of times here, the safety profile and the lack of issues in the heme compartment, right, probably suggests, you know, high combinability more than anything else. Turning back to the solid tumor space, you talked about potential combinations with CDK4-6 in the breast cancer side, SIRDS, and so forth. And so can you maybe tell us, you know, what preclinical work you've done there, both as a monotherapy and what early data suggests in terms of potential here?

Yeah, so this is our simulator program. We generated, we initially generated a bunch of in vitro data where we combined, we titrated in abemacyclib, the CDK4-6. we titrated in fulvestrant, and we saw, interestingly, we saw monotherapy when we didn't have either one of those two agents on board, again, in petri dishes, and then we titrated them, and we saw pretty striking results above and beyond just CDK4-6 and the fulvestrant. So that was the in vitro stuff. We then set out, and we ran over the course of the December a time frame from last year, a range of PDX models and saw both in ESR1 mutant as well as wild type, you know, some striking TGIs for PDXs. I would say when we looked and compared, eyeballed it against some of the CAT6A stuff that others have done, we found those monotherapy data to be impressive. It was a single dose. So obviously, you know, we need to explore the dose. We need to understand in vivo additional monotherapy as well as combination. So we're actually right now in the process of running a range of PDX models with some dose ranging as a monotherapy. We're also doing combination with oral CERD and a CDK4-6. That's what's gating for us to move CBP forward to the GLP tox studies. We have a once-a-week subcutaneous formulation. The hypothesis here, which we're still evaluating, is there's amplification of CBP. Interestingly, when you go read the literature, as often as the case, CBP is a CAT3A. So it's part of this broader acetyltransferases that work. And we obviously know from what Pfizer is doing, the combination data they have with the CAT6A. CAT6A is amplified in a bunch of ER-positive breast cancers as well. So we're pretty excited by this, but obviously we need to validate all of this through the animal pharmacology.

In our remaining time, I want to maybe switch subjects a little bit, which is we've spent a lot of time talking about oncology, obviously, but maybe expansion into non-oncology areas. You mentioned INI a little bit earlier, but just sort of as you think about, you know, broader chromatin biology and where the applications could be from a therapeutic development perspective, you know, what sort of lines up first and second for Foghorn here?

Yeah, I mean, we know, you know, from the literature, And again, when we're regulating gene expression, it sort of makes sense. There's a range of different disease areas. We've seen stuff in neurodegenerative diseases, autoimmunity, as you're altering sort of various T-cell lineage populations, even in metabolic endocrine. We've had some interesting data actually there with CBP. So, you know, I think for me, the most obvious next space is in the autoimmune setting, right? Neuro is hard. I worked at Biogen for many years. That is a very hard space for a number of reasons we don't have time to get into here, but maybe that are obvious. So I'm pretty keen on stuff in the inflammation and autoimmune area. And I think there's some potentially interesting novel targets to go after. I won't say more than that at this point, given what we're working on, other than for our INI program that we have. We've got multiple animal models that the asset that we have looks as good as one of the major antibody classes. In this case, it would be oral, and we're pretty excited by that. So I think there's a lot of room to continue to expand.

Great. Maybe in the minute we have left here, can you maybe update us on, you know, what is Foghorn's cash position and, you know, how are you thinking about your cash runway over the near to intermediate term?

Yeah. So as of the end of Q1, we had roughly $184 million on the balance sheet. Very clean. There's, you know, in terms of capital structure, there's no debt. The current cash runway guidance that we've been giving is into the first half of 2028, but there's some very important provisors or caveats to that. we assume as part of that cash runway guidance, this is not because any decision is made this is just how we model it out, is that we proceed into further dose expansion with Lilly for 909 we've earmarked somewhere between 40 to 50 million dollars to ensure that we can pay our part of the 50-50 but also set aside some money of that for CMC works for drug supply, scale up in the event that we go into a registrational study once that is completed. So there's about 50 million of that 180 that are earmarked for 909. What that allows us to do is to move everything that I just mentioned in the pipeline, including the INI asset, to an IND. But we would need to either raise more capital, look for strategic partners or some mix of that to actually advance the rest of the proprietary pipeline in the scenario where 909 is moving forward. We'd need to raise additional capital to actually go and complete and run phase 1A, phase 1B type studies. So that's sort of how we think about guiding on the cache basis.

My thanks to Adrian and Foghorn for joining us today, and we'll end it on that note. Great. Thanks, Paul. Thanks for having me. Appreciate it.