All right. Hey, guys. We're going to get started here. For this fireside chat, we have Xilio Therapeutics. René, CEO, Yariv, CMO, and Uli, CSO, which is probably for the best that we have all three of you guys, because I do want to get a little bit more into some of the newer assets you guys are moving into the clinic soon and areas of differentiation. But maybe to start, if you guys can introduce us to yourselves. I know you're new, so it would be great to get a little bit of your background and what attracted you to this company, and just where you guys are as a company, where you are as far as clinical progress and milestones upcoming in 2027.
Great. So maybe I'll kick off, and then we can go down. So hi, everyone. I'm René Russo, president and CEO of Xilio Therapeutics. We're a protein engineering company focused on masked IO therapies. We have just put our fourth IND. Our fourth IND just got cleared in mid-year, and we put our fourth program into the clinic, which is a PD-1 masked IL-2 that we're very excited about. The quickest way to kind of give the update on the history of the company is we have clinically validated masking technology. We started out with simple molecules, monoclonal antibodies, and cytokines, demonstrated in the clinic activation in patient tumors with these molecules, and have multiple pharma partnerships, including two active ones with Gilead and AbbVie, based on that validation of our initial molecules.
Now, coming up in 2027, we're moving into these more complex multispecifics, again, starting with the PD-1/IL-2, and also masked T-cell engagers that incorporate multi-targeting and co-stimulatory signaling built into the molecule. So we'll talk a little bit more about what's coming up in 2027.
Hi, everyone. I'm Uli Bialucha, chief scientific officer at Xilio Therapeutics. I'm a cancer biologist by training. I'm a strong believer in immunotherapy. I think the field generally appreciates that immunotherapy can drive profound responses in cancer patients, but that oftentimes we're limited by how high we can dose. So the therapeutic index is a challenge that we've been geared up to solve at Xilio Therapeutics, and as you will learn today, we're doing so across a variety of therapeutics in development.
Hi, everyone. I'm Yariv Houvras. I'm a physician scientist and medical oncologist. I joined Xilio two months ago. My prior experience in drug development includes T-cell engagers and ADCs in solid and hematologic malignancies. I was really delighted to join Xilio because I think they have best-in-class protein engineering in IO, and I think we really have a tremendous opportunity to bring safe and effective medicines to patients with advanced solid tumors.
Thank you.
Great. Maybe to start with XTX501, your IL-2 PD1. Can you maybe as we start getting into the whole masking, talk about learnings that you guys have from your previous assets and what you think the importance and the differentiation of masking such an asset is here?
Yeah, absolutely. I'll start, and everyone can jump in. We've been working in the masking space for a while. We have a lot of experience masking cytokines, and we recognized pretty early on that putting a cytokine into a PD1 bispecific really was differentiated to really bring that cytokine to PD1-positive cells, in particular IL-2 with that known biology. But of course, IL-2 can be incredibly toxic, can have PK challenges, right, if it gets taken up in the peripheral circulation. So the masking is really key, and we've also now had the benefit of experience, I think, in the whole field that the right design for an IL-2 is not full beta-gamma, but is more of a wild-type IL-2 that has the right alpha tuning.
It's a combination of the potency of the IL-2, the alpha in the IL-2, and the masking that allows it to have long half-life and activation on the PD1-positive cells in the tumor. That's the key. Would you like to add anything, Uli?
Maybe just adding that the field obviously has had or has pursued IL-2 for a while, and it's never been a challenge around IL-2 not being relevant to T-cell biology. I'd argue it's a key pathway, and we know that if you deliver IL-2, this T-cell agonism can drive profound responses. The challenge has always been around how you engineer it so that it has the right receptor bias, that you can give it in a tolerated way, and then more recently, an appreciation that driving it to the right cell type is really critical.
That's what we're really aiming to do with 501, drive the IL-2 to the PD1-positive T cells that we know are enriched for antigen-specific T cells that are seeing and fighting the tumor, keeping it in check in the periphery using the mask, and then when it's in the tumor microenvironment, remove the mask efficiently and allow for the right receptor bias, right? So that you get, again, more preferential activity on those PD1-positive recently activated T cells that also have the alpha receptor.
A lot of learnings have gone into the design of this molecule, and I think based on our preclinical profiling, even in relative to Innovent's IBI363, we have a high conviction that this has a best-in-class potential.
Compared to Innovent's data to date, when you guys look at that data, in what areas do you think you could clinically differentiate, and I guess, what will you be looking for in your early dataset next year?
Maybe I'll just kind of start out that I think we're all very excited about what we've seen with the IBI363 data, in particular, really meaningful efficacy in the post-PD1 setting, which is very exciting, right? And we think these type of molecules have the potential, right, to really move up earlier line and ideally, right, move to become the next pembro, right? Become the foundational backbone. What you need though in order to do that is you need excellent tolerability, so it can be as early line. You need to be able to combine it, and you need the efficacy, right, to really be driven as far as possible, even in the post-PD1 setting. So, Uli, do you want to talk about what we
Sure
are hoping to see here?
Yeah. No, I think, just to build off your comments, I believe that XTX501 may be a next-generation PD-1/IL-2, because of multiple design features and the rigorous preclinical evidence that was generated. XTX501 is bivalent for PD1, whereas IBI363 is monovalent. So we know we're a stronger binder at PD1. We have an alpha-optimized IL-2, with significantly stronger affinity and activation of the IL-2 receptor, and we're masked. We're fully masked, and the preclinical evidence demonstrates really nicely that the masked molecule has minimal activity and unmasked is very potent at inducing tumor regressions in vivo. I think all of this will translate into safety and efficacy in our first clinical trial, which has started. We enrolled our first patients several weeks ago.
Our focus there is on the treatment of lung cancer patients in a dose escalation BOIN design. I think it's these features of the molecule which will hopefully translate into wide therapeutic index, tolerability, safety, and efficacy.
Got it. You've said you've dosed your first patient. Can you speak to what dose you're starting at relative to what dose level you expect to start seeing clinical activity?
It's a great question, Laura. We have disclosed that we started at a dose of one mg per kg. This, we believe, is very close to what will be an efficacious dose in humans.
What I can say is that we have a very standard and relatively aggressive dose escalation regimen, which is typical of solid tumor.
Late 2027 is when we should expect data from you guys first, right? Should this include efficacy, or will this be more safety PK focused?
I think we are hoping to bring efficacy data forward in Q4 of 2027.
Are you guys doing any type of biopsy work or anything to really show what's happening in the tumor microenvironment?
Well, that's a great question. We have a number of exploratory endpoints built into the study. We're very, very interested in understanding the pharmacodynamics of treatment with this molecule.
Okay. Why start in the post-PD-L1 lung cancer? This is historically such a hard indication. I guess you do have the biological de-risking from the competitor here. But why start there, and what could be the next steps beyond that for this asset?
Well, as you know, lung cancer has a tremendous unmet clinical need in the U.S. We're really excited by the Innovent data, but I would note that their patient population is mostly current and former smokers. Their efficacy is significantly higher in squamous than adenocarcinoma. So, there is an opportunity with the next-generation PD-1/IL-2 to examine safety and efficacy in the U.S. in lung cancer patients. I agree that I think it is de-risked.
To use your phrase. We're hoping that we can really try to understand the efficacy across all histologic subtypes and a population that's reflective of lung cancer in the U.S.
Can you speak to what kinds of combinations might be on deck here? Should you show monotherapy activity and good safety?
Well, I think ultimately we'd love to see this as a frontline therapy to replace pembrolizumab in combination with chemotherapy.
Mm-hmm. Yeah.
That'll depend on whether or not the safety profile permits that combination.
At what point do you think you guys will be at that stage where you could start making the decision to move into expansion dosing or into combination cohorts?
Yeah. I think our initial the focus initially on non-small cell lung in the post-PD1 setting, right, it's a very challenging approach, but we think we have a very clear efficacy sort of base now.
with the Innovent data. Also, when we look at the deal that they did with Takeda, right, that is, in our mind, kind of the floor, right, for us. I think what we'll be looking for is a balance of how much data do we need to really drive value.
for shareholders with the existing dataset. This dataset in Q4 will be early, but we hope that it includes efficacy. That's our plan.
Yeah.
But we'll let that data mature somewhat. But there may be opportunity, depending on that data, to work with potential partners to move into combos.
There is a lot of combos we are excited about, even potentially within our own pipeline.
Yeah.
Of course, there is so much opportunity at other solid tumors, and kind of following behind Innovent, we can really be selective and have a focus so that we can have kind of meaningful data readouts and have to do less exploration.
Makes sense.
We are not going to wait too long, but we want to make sure we get that initial value creation of this first dataset.
Definitely looking forward to seeing that data next year. So maybe pivoting into the T-cell engagers, before we start diving into the individual assets, if you guys could maybe provide a little context of how you guys were able to seemingly very quickly pivot into this platform.
Mm-hmm. Yeah, maybe I'll start and then Uli can pick up. Yeah, we have a lot of kind of relevant, I would say full circle, discovery translation into the clinic now cycles, and that has given us the confidence and the ability to move much faster into more complex molecules and know that we are getting the masking exactly right. I think this is a bit hard to follow, right? You can't just jump in and do these types of molecules from scratch. You really have to build that capability. What Uli and his team have built really has this modular architecture, so we can look at a variety of constructs with our TCEs, right?
We have a co-stim built into all of the molecules, which is, I think, very unique, and we have the ability to mask really any of the components, as well as target multiple TAAs simultaneously. That flexibility has been really unique for us and has allowed us to move fast and pivot. But Uli can talk a little bit more about what we have coming up with regard to TCEs.
Yeah. For us, it's always been the desire to complement what we have with our CTLA-4, with the cytokines that are really about stimulating an existing immune response and making that more effective with synthetic immunity, right? Which is you're creating a novel immune response from naive cells, and that is T-cell engagers. Both of these mechanisms, they rely on essentially activating T cells. We know that T-cell engagers have been very limited in terms of their tolerability as they've attempted to move into solid tumors, because the reality is that most of the targets are expressed not just on cancer, but also on normal tissue. And that's a prime place for masking. We were relatively well-positioned, I would say, with everything that we've learned about masking to pivot into that area, given also the protein engineering expertise we have.
But I think where we, over the last couple of years, have really invested time and effort is not just to generate the next masked TCE, but really think about what would the best possible molecule for a given tumor type look like. And so when we looked at prostate cancer, the reality is that many of the commonly known antigens, like PSMA and STEAP1, they are quite heterogeneous in their expression, which means that about 30% of patients have tumor where not every tumor cell is PSMA positive. And that is a challenge as we think about resistance, lack of durability. So we said, well, really what would be a best molecule here would be one that can target either/or, either PSMA or STEAP1. And so that is how we built it.
Additionally, we felt that ideally a T-cell engager should not just trigger signal one, which is the T-cell receptor signaling, but bring along co-stimulation built into the same molecule. And so that is what we have now built in the context of this modular platform that René talked about. And just today, we announced that our second TCE program targeting CLDN18.2 now also brings along co-stimulation.
Can you speak to the chemistry of the masking and how it varies across the different modalities in your pipeline?
Yeah, I would be happy to. So we do not just have one fixed way of masking molecules. In many ways, our philosophy there has been what would be, again, like the best masking solution for this molecule? You have one shot of putting a molecule in the clinic. It better have what you consider the best possible shot and the best possible mask. So what does that mean? For an antibody, we know cyclic peptides can work quite well, bind in the cleft, and then essentially block that interaction. But there are other targets that you may want to mask, like a cytokine or a single domain antibody, where the surface is suboptimal for masking with the cyclic peptide.
What we decided to invest in is capability that allows us to essentially identify a mask independent of the structure that you're trying to go after, which means we have custom masking solutions. In the case of our IL-2, it's a single domain-based mask that is very efficient at shutting that IL-2 down, and that's part of XTX501. Then the T-cell engagers, we knew we needed to lock down that anti-CD3 domain. Again, we went at it with a broad discovery campaign, identifying multiple masking solutions that could work, then selecting the one that had the overall best property. If you look at how effectively the CD3 domain is masked with our molecule relative to what's in various patents or relative to what Janux have shown, we feel confident that we might have the best masking efficiency on that anti-CD3 domain.
For the T-cell engagers, can you speak to like half-life and affinity and making sure that this is getting into the tumor microenvironment and hitting the targets it needs to be?
Yeah. This is, again, about finding the right components and ensuring that you're not just picking a component that might be available to you, but selecting the one that is really the best component for that molecule. To give you just a bit of a picture, for our PSMA-STEAP1 program, we screened over 1,800 molecules, right? Which means that we had a broad survey of what's the best STEAP1 arm, what's the best PSMA arm, what is the right affinity for those, what is the right CD3, and what's the right affinity there, what's the right geometry of the molecule? A lot of really empirical screening goes into the selection, then having the right readouts, right? We generally prefer function-forward screening, and that's now really baked into how we arrive ultimately at our development candidates.
What's the timeline for when these two T-cell engager assets could enter the clinic?
Yeah, so I think just this morning with earnings, we've kind of narrowed that guidance to both INDs in Q4 of next year.
So that will be fun for the team to do two together. Q4 is also when we'll have data on the PD-1/IL-2, so Q4 will be-
It's going to be a busy-
a busy-
busy quarter for you guys.
Yes.
On the PSMA-STEAP, to start, we do have some proof of concept data that we've at least seen for PSMA. Can you speak to what you will be looking to see as far as efficacy? We previously saw that deep PSA50 or PSA responses didn't exactly translate into the differentiated efficacy or durability everyone was hoping for. How do you think this methodology or this mechanism will differentiate, and what are you really looking for as far as efficacy to get excited? I know we're thinking far out here, but I think that's the direction people are going to start thinking once you get this into the clinic.
Yeah, definitely.
Yeah, we're a little far, as you know.
Yeah.
from being in the clinic, but I think in general, what we've learned in the field is that PSA reductions are necessary but not sufficient.
That we need to also see radiographic reductions and PCWG working group-
criteria for objective response.
Yeah.
Prostate cancer is also characterized by different clinical subtypes. We have patients with visceral disease, we have patients with bone-only disease, we have patients with neuroendocrine differentiation. We'll be looking to understand the efficacy across critical clinical subtypes of the disease.
Okay. Then the Claudin 18.2 asset, can you speak to how this approach differentiates versus other modalities on the same target?
Yeah, I can speak to that. For Claudin 18.2, it is a validated target in multiple GI cancers, including gastric, esophageal, and pancreatic cancer. There is an approved antibody, right, for high expression settings only, and then there are multiple ADCs, some T-cell engagers in development. Again, those are more geared towards high expression settings. We specifically went after a molecule with XTX602 that has activity in lower expression settings. So we went through some lengths to further affinity mature our Claudin 18.2 binding arm so that even in low density settings, it should drive activity, knowing that with the masking, we should allow for that activity to become focused on the tumor as opposed to being a challenge in the normal GI, right? Additionally, now we have the co-stimulation built into the molecule, right? We went at this in parallel, multiple different candidates, one with co-stim, without co-stim.
We wanted to understand what would one mask as opposed to two masks do, and drove multiple of these candidates forward. Again, ultimately, XTX602 had the overall best profile, allowed for co-stimulation in excellent tolerability, potency, and no evidence of target-mediated drug disposition in primates, which is generally something that Claudin 18.2 agents can experience, along with really a favorable GI toxicity profile, which has really been one of the main limitations of going after Claudin 18.2 to date.
Mm-hmm. As far as indications of focus, where do you think the clearest initial opportunity lies?
Well, I think, as Uli noted, it's probably gastric, esophageal
and panc. I think we're particularly excited about the opportunity in pancreatic adenocarcinoma, where there's a strong stromal component. We believe the proteases will be active, there's obviously an unmet need despite some recent very important successes for RAS inhibitors.
Yeah.
Right? Those patients, after they progress, are going to need additional therapies.
Absolutely.
I think I could see this molecule meeting an important unmet clinical need there.
Yeah, absolutely. That makes a lot of sense. As far as dosing, how often per week, or are we looking at twice every other week or every three weeks. What are you guys predicting might be the window here?
Yeah, based on the PK that we know to date, it should be Q3W dosing as kind of the floor across all of these molecules. In some cases, we've been able to extend that even further with our first series of molecules. Yeah, we're looking at likely Q3W or longer intervals between doses.
Awesome. Then, as far as cash runway, you guys have a lot going on end of next year, a lot to do to get there as well. What is baked in, and what could maybe some of the milestones from the partnerships that you guys have contribute to?
Yeah. So, cash runway into Q1 of next year. That obviously Oh, sorry, Q1 of 2028. I already think we're in 2027.
Yeah, it is hard to keep up.
Yeah. Q1 of 2028, right? That allows us to get to this initial 501 data, which is really important. If you look at our existing partnerships, as well as a tranche of Series C warrants that become available December of this year, there is an opportunity to bring in up to $140 million in additional capital with existing collaborations and that Series C tranche. So, that could dramatically change runway and extend even into 2029. We have multiple levers that are, again, within existing collaborations and these Series C warrants that we are paying attention to. For now, right, just based on the cash runway, we can get through the INDs and initial 501 data, which is really key.
Got it. From the IL-12 partnership with Gilead, you guys have an option package that you will be bringing to them in the near future. What goes into that package, and what is ultimately the next steps should Gilead plan to pursue further here?
Yes. The IL-12, this is a masked IL-12. We did a phase I/II in collaboration with Gilead. They have an option on the molecule. We will be delivering a phase II data package to them the first half of next year. It is around 40 patients. It is not a huge data package. We are confident that that is going to happen. We have demonstrated, and what is in the public domain is monotherapy activity with this molecule, dosing at about 100x the MTD of human recombinant IL-12 safely. We think this is readily combinable. We have seen some exciting data, the PD in the tumors demonstrating that it is activating. So it is hitting kind of all of the key kind of check marks, right, for the target product profile we set out to design with this molecule.
They will have some period of time to review that data.
We would expect that we will know when the decision is made whether they want to bring it in or not, next year, within the year.
Sounds good. You guys also have a masked CTLA-4 that I believe you guys are looking for partnerships for.
Yes.
What's the progress on that, or the status of that asset and partnership discussions?
Yeah. I think there's some interesting opportunities to potentially find a path for this asset. Pretty recently, we demonstrated in late-line CRC in combination with atezolizumab, a unique biomarker that is high plasma TMB, which was not really well understood in CRC. We were able to identify that over 50% of patients with CRC actually have high plasma TMB. This was not really recognized. In that population, we had a 40% response rate with atezo, which is dramatically different than the 20%-30% we see in the all-comer population with most of these molecules. I think that biomarker strategy is exciting from a regulatory path, could be really differentiated to get to an approval path that kind of no one else has that opportunity with their dataset at this moment.
We're also looking at opportunities in neoadjuvant that we think are unique to a CTLA-4, especially one that can be given multi-dose safely.
Yeah.
I think one of the factors with our CTLA-4 is patients stay on it. We have patients on out to two years. We've demonstrated it activates in the tumor, and it's very well tolerated. I think there's some pretty interesting paths, and there's some more excitement around, especially neoadjuvant, with these type of molecules.
We're working on this. We see a couple potential paths that could make sense, and really, ideally, want to get this to patients. I think the need is very high.
Got it. As far as other potential partnerships, your other T cell engagers, for example, are they something you're looking to keep in-house or something you might be open to partnering?
Yeah, so I think the good news about this platform and the different components and the flexibility modular components is that we have many different directions and many different targets we can go into, and we have others progressing sort of behind what is in the disclosed pipeline. That really does allow us flexibility to partner some of these earlier stage molecules, de-risk, right, some of those, and allow us to potentially increase runway to get to, ideally, to get to that TCE data. So I think the flexibility is there. We will always be looking for the right partner in that disease area, like what is their commitment, what is their capability in that disease area?
Yeah.
But I would say, for the TCEs, it is something that we are actively thinking about across the pipeline.
For XTX501, on the other hand, I think we really want to drive. We think we can drive
Yeah
some significant value with this early dataset based on the de-risking that we have seen with the Innovent PD-1/IL-2.
Nice.
That one we want to progress on our own.
Yeah.
At least for a while.
Happy to hear that.
Yeah.
Well, we are just about out of time. Actually, right on time. But I really appreciate you guys coming out to our conference and-
Thank you.
Congrats on all the progress you've made in such a quick period of time as well.
Yeah, it's gone really well, and we're glad to have you, Yariv, on board now as well.
Yeah.
Thank you. Thank you for having us.
Of course.
Thanks for having us.
Thanks, everyone, for being here.