A fireside chat with Boundless Bio. I'm really pleased to welcome Zach Hornby, CEO. Zach, welcome. Thanks for joining us. Thank you, Michael. Thanks for hosting. All right, I'm gonna, I'm gonna start out, just discuss with a really high-level question. It's probably as high as they come. Zach, can you start by discussing your view on some of the largest unmet need in cancer? Yeah, absolutely. Happy to. So I think one of the things we've seen over the last several decades is different new modalities or approaches to treating cancer. Since about the turn of the century, we've seen the precision medicine revolution with targeted therapies going after oncodrivers, and then more recently, immunotherapies, cell therapies have kind of had their day as well. And one of the things that I've personally observed over the course of my career working in oncology is that each of those different revolutions of new therapy classes work very well for certain types of patients, but of course, there's other patients who get left behind. A specific subset of tumors that have not been well addressed, either by targeted therapies or immunotherapies, are tumors with what's called oncogene amplification. And so there's some tumors where they have a genetic driver, which is like a gene fusion, and you're familiar with these types of drugs for ALK fusions, ROS1 fusions, RET fusions, TRK fusions. We've had really good drugs, really successful companies going after those. We also have patients whose tumors are driven by point mutations. You know, the one - the most high profile right now are things like KRAS G12C, G12D, and clearly, we've got good drugs in the pipeline that are working against those. In the immunotherapy space, yeah, there's things like MSI-high, others. But when it comes to patients with oncogene amplifications, instead of having a fusion of the gene sequence or a point mutation of the gene sequence, the sequence is normal, but there's just too many copies. And in that setting, where... And this is, by the way, this is 25% of all cancer, so it's a huge percentage of the population. Those same drugs that work and get approved for fusions or point mutations, they just don't work in the patients with amplification, and so these patients, they have no standard of care. They have objectively worse survival and outcomes. Their tumors are more aggressive. And so, you know, to your question of what I think is one of the largest unmet needs, I think it's patients with oncogene amplification. Right. And Boundless Bio, I think, was founded based on a premise of leveraging ecDNA as a diagnostic and a treatment approach. And so maybe first, just stepping back again, remind us what ecDNA is and what the role is in cancer. Yeah. So ecDNA stands for extrachromosomal DNA, and literally, it's kind of what the word sounds like. If you recall back to high school biology, we all learned that our genomes are encoded on 23 chromosomes, and we have a pair of each. And that's kind of what we learn. We learn about Mendelian genetics and how we inherit things from our parents. But it turns out that in the case of cancer, certain cancers, not all of the genome is encoded on the chromosomes. Instead, units of DNA can actually break off of the chromosomes. They literally fracture off of the chromosome, and then they circularize, almost like bacterial or viral plasmids, and then those little circles can independently replicate into multiple copies. They can transcribe. The DNA becomes RNA and ultimately protein. These ecDNA, these circular units of DNA, they're still in the nuclei, but they're physically separated from the chromosomes. These are a unique feature of cancer cells, and their purpose is for cancer cells to consistently drive genomic diversity, heterogeneity, and ultimately, different clonal populations that might have different fitness advantages that help the tumors survive and recur. These circles of ecDNA, this is the primary mechanism by which oncogene amplifications occur. It's kind of you take this unmet need, you take this biology that that drives the unmet need, and then you begin to ask the question: Okay, well, if these ecDNA are a unique feature to cancer cells- Right ... is there something we can do about it? Right. And how does your platform, or how do you approach that with your platform? How do you take advantage of that? Yeah. So our platform is called Spyglass, and what we set out to do at the beginning is basically understand, well, if these ecDNA are common in cancer. It turns out they are. They're found in 15%-17% of tumors at the time of diagnosis. So if we know they're common, then we need to generate some models that we can test things on, that represent these tumor settings. So our Spyglass platform, we set out to first, develop and characterize hundreds of in vitro models, thousands in silico, and then dozens in vivo, where we can ask types of questions like: Which targets, when pharmacologically inhibited or genetically knocked out, will kill the cancer models that have this ecDNA but will spare models that don't have ecDNA? So we generated this library of ecDNA-positive tumor models and then ecDNA-negative models, and we run differential analyses looking for targets whose knockout only kills the ecDNA-positive models and then spares the healthy models. And that's how we begin to isolate targets that are unique to the ecDNA biology. Okay. And can you talk a bit about how your latest product candidate, BBI-940, potentially could take advantage of that? You recently had an IND accepted, and I think phase 1 study going up very soon. ... In fact, today, the clinical trials listing went live on ct.gov, so the trial is official as of today. This program is called BBI-940. We are targeting a novel kinesin. We discovered this kinesin through the platform I was just describing. As we were learning more and more about the biology of these ecDNA, we started to appreciate different facets of the life cycle. How did these things form, and how do they function? And as we better understood that, one of the features we learned was that their location matters. So how they segregate, how they localize when cells divide or undergo mitosis, there is some order into how they go from a parent cell into the daughter cells. And it turns out that there is a kinesin that is not essential in healthy cells. So in normal tissue, this is not absolutely necessary protein. It's redundant. In fact, there are knockout models that are viable. But it turned out that this kinesin was essential when it came to ecDNA. It appears to play a critical role in how these ecDNA move during mitosis. And so when we discovered this target, we then validated it in multiple different tumor models. We came up with compounds to inhibit it, and then ultimately, we developed an oral degrader. That's our compound, BBI-940. And then, as mentioned, we just cleared an IND in January. The trial just went live on ClinicalTrials.gov today, which means it's now open for screening and enrollment, and we hope to dose the first patient in the next couple weeks. Yeah. So and kinesins are an interesting class of targets. I think there have been some efforts historically targeting different kinesins that didn't make it to approval, obviously. There's a couple other new companies on the radar now targeting KIF18A, and so, yeah. How interesting is this class? Maybe stepping back again, you know, this class of targets today, and maybe just talk a bit more about how or what differentiates BBI-940 from some of the other programs, including the KIF18A inhibitors. Everything you just said is accurate. The kinesin class has been known. I I believe there's more than 40 different kinesins within the category, and historically, there has been some pharma interest in the class because they are involved with mitosis, and cancer cells are replicating quickly. Some of the previous pharma efforts went after a specific kinesin called KIF11. The drawback with that particular target is that it is an essential target, so healthy cells rely on that target for survival. In DepMap, it would be considered highly essential, and so I think the prior efforts to drug that kinesin were stymied by the therapeutic index. They were too toxic in the clinic. More recently, there are a couple, at least private companies, pursuing a new kinesin called KIF18A. I think you were alluding to that. That one is less essential, less toxic than KIF11. Sure enough, these companies are apparently generating some really encouraging data. While they remain private, you know, I've heard from investors or bankers that they're starting to see some nice signals in ovarian cancer, among other tumor types, and they appear to be reasonably well-tolerated. What's unique about our kinesin, it is absolutely non-essential, and so therefore, would be predicted to have quite a clean target therapeutic index in the clinic. Certainly, preclinically, that has been borne out by our studies, that our GLP tox studies were extremely well-tolerated. So we are looking forward to bringing this program into humans and seeing whether indeed we validate in the clinic that it's well-tolerated. But if so, that could make for a very interesting profile. Right. Yeah, definitely interesting to see the emerging POC for in this class of targets in general, I think, for industry. And, I think... So your molecule is a degrader, not an inhibitor. Maybe talk about, you know, why you pursued a degrader approach. One of the interesting things and challenges about this target is, it, it was completely novel. There was no evidence that anyone in this industry globally had ever worked on this target or tried to drug it, which on the one hand is very exciting, because it's first in class, and it's a great opportunity. On the other hand, it's very hard because there was no roadmap of how to go about drugging this. So we had to do everything from learning how to express the protein, create the first crystallization structure, crystal structure, had to run different high-throughput screens of compounds to try to generate binders and then hits that were inhibitors. So this was a real labor of love. It took us multiple years. We screened over 1 million compounds. We ultimately only got a single hit from multiple screens of over 1 million compounds. Single hit that was, like, 50 micromolar, not very potent. Our team slaved away at medicinal chemistry. SAR drove better and better binders, inhibitors, and eventually got down to single-digit nanomolar inhibitors. Yet they were still not having the same cellular pharmacology in cancer cells that the genetic knockout had had. So we'd originally validated the target with genetic knockdown. And so we're getting these, like, really potent inhibitors, but they're not doing the same thing as the genetic knockdown. We're like: Why is that? And so we started to hypothesize, "Well, maybe inhibition is insufficient," even though this kinesin, it's a molecular motor, it's got a motor domain, but maybe inhibiting the motor domain wasn't enough. Maybe it's playing a scaffolding role or structural role. So that's when my scientific team had the breakthrough, which I think was wise: "Let's try degrading it and see whether that achieves something different." And almost instantly, that was a breakthrough for our efforts. That as soon as we went to degraders, even before they were optimized, we immediately started seeing cellular pharmacology. So it would seem to reinforce the hypothesis that this protein is playing more than an active role. It might be playing a structural role vis-à-vis the ecDNA, and therefore, its physical removal is necessary for the pharmacology. Right. Maybe just a technical question, is it a heterobifunctional degrader? It is. Yeah. Yes. Makes sense. And then, yeah, now that the phase 1 study is open for enrollment, yeah, could you just comment on the study design? You know, what types of patients do you plan on enrolling, and do you expect an efficacy signal as a monotherapy? The study design is a classical first-in-human phase I study that's got initial dose escalation and then some dose expansion. For the initial Part 1A dose escalation, it's a BOIN design, so 3-4 patients per level, and that's gonna be as a single agent. The patients who are eligible for that part, it's two subtypes of breast cancer. So one is ER-positive, HER2-negative breast cancer, and the second is a subset of triple-negative breast cancer, which is specifically called TNBC LAR, and the LAR stands for luminal androgen receptor. How these patients are selected is their tumors actually express androgen receptor. So these are ultimately two hormone-driven tumor types. One is ER hormone, the other is the AR hormone. Those are the patients who we're gonna enroll. The reason we're enrolling those patients is it's empirical. It's based on where we saw a signal preclinically. We will enroll those patients initially as a single agent, where we're looking for PK, we're looking for PD, which is degradation of our kinesin and target. We're, of course, looking for safety and tolerability as we escalate. And then, once we get into what we believe to be a bioactive range, we can expand our escalation cohorts through backfill, but we can also roll into an actual expansion phase of the study, where there will be two features to the expansion phase. One is, where we will combine our agent with fulvestrant, and those will be, in ESR1- negative patients. And then a second is that we will be, looking at a biomarker cohort, and this is as a single agent. We will be looking at patients who have FGFR1 amplification, and the reason for that is twofold. That biomarker cohort is empirically... In our preclinical studies, we saw a really nice association between high sensitivity to our compound with the presence of FGFR1 amplifications. So it appears to be mechanistically, that's a, a space we're gonna be active, but also in terms of unmet need, patients with breast cancer with FGFR1 amplification, that's one of the highest unmet needs within breast cancer. The presence of FGFR1 amplification is associated with, rapid resistance to CDK4/6 inhibitors and endocrine therapy. So this is a known population that doesn't do well with existing standard of care, and there's a potential that we can make a real impact there. Okay, what percentage of patients have FGFR1 amplification? It can be multiple different types, but you can find it within ER-positive breast cancer, and particularly those who are rapidly resistant to frontline therapy, and you can also find these FGFR1 amplifications within triple-negative breast. Okay, so you said it'll be a fulvestrant combo within the ER-positive setting and then the biomarker cohort. What about the TNBC opportunity? Sorry, what, what's the question about? The TNBC- Yeah. I mean, is there any other selection- Oh! - happening in TNBC, and how big is- Right ... the LAR, subset? Yeah, the LAR subset, it's a minority subset, maybe, like, 10%-15%. No, there's no other selection within the TNBC LAR. So it's the AR expression, is what characterizes that, and then in the expansion. Right, the FGFR1 will be really focused on the ER positive. And we find that the FGFR1 amplification can be in about 15% of ER positive at baseline, and then upon resistance, increases to as high as 30% of patients. Yeah. So it seems to enrich within resistance. Makes sense. And then, yeah, would you look at FGFR1 amplification via ecDNA? We will not look for that prospectively. So fortunately, FGFR1 is measured on most panels, whether those are universal panels like Foundation Medicine or Tempus or Guardant, or also even the local panels like MSK-IMPACT or Dana-Farber's OncoPanel. So as long as they detect an FGFR1 amplification, those patients will be eligible to be immediately enrolled into our study. We will, of course, collect tissue, both tumor tissue and liquid biopsy, from all of our patients, and then we can retrospectively measure ecDNA status. Right. And then, you know, have you done calculations? Have you done evaluation of, you know, what level of target degradation would be necessary in humans to achieve clinical activity? And, you know, how fast do you think you can reach active dose levels? In our preclinical studies, we looked at pharmacokinetic exposure, target degradation, I'll call that PD, and then efficacy. We found that the sweet spot was targeting 70% target degradation, that once we got to that level, that was really associated with robust efficacy. Then, even when we achieved higher degradation, like 80% or 90% max degradation, there wasn't really much more efficacy to achieve. Right. That's what we're gonna be trying to strive for, is 70% in the clinic. Right. And then I think you said, so any knockout animals are viable, but are there any on-target toxicities that you're watching out for? We did not see anything that we would deem on target preclinically. The compound was very well tolerated. We ran our GLP in rats and dogs, and there was no mortality. So there, there's really not much that we're looking for specifically in the clinic, but of course, we'll take the data as they come. Okay. Then maybe just comment on your plans of potentially disclosing data from the phase 1 study, or maybe stepping back- Yeah ... how fast, you know, do you expect the study to enroll and then disclosing- Right this data? ... So I think what helps is that breast cancers are a large market. Of course, this is a tissue-focused tumor type, not pan-tumor. We're optimistic it will enroll pretty well. We're thinking we might get through 3-4 dose escalation cohorts this year. I'll tell you that in the early days, I'll first be looking for PK exposure. Because this is an oral degrader, we wanna make sure we're getting reasonable bioavailability, so that'll be the first thing we're looking for. Assuming we are getting good exposure, then we will be looking for the target degradation, the PD marker, so looking to get up into that 70% range. Once we're there in the appropriate range, we're gonna wanna make sure it's well-tolerated, that patients can tolerate that level and stay on drug. So I'd say over the course of this year, I'm hoping our team can check the PK, PD tolerability box, and then once you know you're at the right dose level with the right target degradation, and patients can stay on drug, that's where you become a candidate for efficacy. And that's where these expansions matter, that we're now treating the right patients, whether that's biomarker positive or the combo, and we're keeping them on drug. And so I think therefore, you could imagine efficacy beginning to read out first half of next year. This is all subject to enrollment and how things go, kind of best laid plans, but I could see de-risking data over the course of this year, and then the really critical data first half of next year. Right. And then outside breast cancer, are there any other opportunities for this target, in your opinion? There are. So, we ran a cell line screen of about 800 cancer cell lines. We saw very high sensitivity in about 15% of all the cell line models. That was across multiple tumor types. We chose breast as the lead indication because it was one of the most encouraging, both in vitro and in vivo, but there were some other tumor types that also look encouraging. We're currently profiling those preclinically, so I don't wanna get out over my skis and say, like, the next one is gonna be this tumor type, 'cause if, if they- if ultimately they don't work in vivo, we won't pursue them. But suffice to say, there are other tissue-specific cancers that we're actively profiling in vivo right now, and it would be our hope then that we could add additional cohorts in the future. Okay. And then last question, and you obviously have that ecDNA Spyglass platform. How do you think about longer term adding additional product candidates to the pipeline? Over the course of the company, we've identified and validated approximately a dozen targets that we think are quite intriguing for the ecDNA biology. Some of these targets are completely novel to the field, in fact, the majority of them. So there's a handful of other ones that we've validated in multiple models and we think would be quite interesting for drug discovery efforts. In the near term, we're gonna allocate our resources to this clinical study, and that's gonna be the preponderance of our effort, because we're mindful of our cash runway. We've told the Street that we project our balance sheet into the second half of 2028, and so when you back out what I just described about the clinical study, if we have cash into second half of 2028, if we can generate interesting clinical data by mid-next year and report that out, that still is a year of runway that could put us in a strong position for our next capital-raising event. But should that be successful, then we absolutely have additional targets and, and nascent programs that would be interesting to drive forward. Okay, awesome. Well, that, I think, we'll wrap up. But, yeah, looking forward to hearing more about 940 and the phase 1 study that's up and running now. So thank you, Zach. Thanks again for inviting me, Michael. Awesome.