Xencor, Inc. (XNCR) Earnings Call Transcript & Summary

Good afternoon. Thanks for joining us for another session at the 42nd JPMorgan Healthcare Conference. I'm Brian Cheng. I'm one of the senior biotech analysts here. I'm joined by my associate, [ Sean Kim ], on stage. We have Xencor's CEO here with us. I'll pass the mic to Bassil for a really quick presentation, and then we'll jump right into a fireside chat. And if you are joining us live, you can submit the questions to our portal. Bassil, the stage is yours.

Thank you so much, Brian. It's a pleasure to be here at what I think is my 22nd JPMorgan Healthcare Conference.

Nice.

About half of them. Majority of my adult life. So this is our forward-looking statement slide. I won't belabor the slides. I think it's better to have a more of a conversation, but I will just state who we are. We're a protein engineering company. We focused on monoclonal antibodies and used our expertise in engineering the entire antibody structure, in particular, antibody Fc Domains for that bottom part of the antibody in the constant region to build a toolkit of really stable modular units that we can use to both build our own internal pipeline, where we've got a range of molecules we're advancing as well as build molecules and create intellectual property, we can license to have essentially another arm to the business through partnering. There's currently 3 marketed products with our technology embedded in it or that we've entirely made. And we have partnerships that we think complement but don't interfere or conflict with our own internal development. Our goal is to become a fully integrated drug company, and we know that developing our own drugs is the way to really unleash the value of this platform that we've created over the years. So we have, again, this tiered model where the core and the bedrock of the company is always staying at the cutting edge or beyond of protein engineering tools for therapeutics discovery. Again, we parlayed that into a very successful partnering model that generates capital. We haven't raised -- we haven't sold stock to raise money in 6 years. And we started this year -- it's pro forma actually closer to $680 million because of a royalty monetization we did in Q4 of last year. So we are guiding runway into 2027. Now the strategy when you have this kind of broad toolkit is to use our technology to build a diverse pipeline and to keep advancing the technology. So 15 years ago, we created our Xtend technology for enhancing antibody half-life in the circulation. That's led to the drug's Ultomiris and sotrovimab for COVID. And we created our bispecific platform, which now has a rich pipeline of molecules. Between partners and Xencor, I think currently in active development is 8 or 9. I always lose track. And we built our cytokine platforms and the next wave of bispecifics. That's led to a set of molecules within early clinical testing at Xencor where we try to do clear clinical experiments to decide whether a molecule is worth advancing internally, hopefully, to late-stage registrational studies that one day can lead to drug approvals to partnering if we feel like a partner is needed to create the synergies we need or to find the resources we need to really unleash the value in a molecule or we'll stop programs if it's warranted, like we've done on a number of occasions because our goal is to get to that orange bucket on the far right side of the slide. It's all based on this modular approach where I won't get into the -- delve into the molecular engineering of it, but we can mix and match all these modules, constant domains, variable domains, create brand-new bespoke variable domains of various ranges of activity and put them together in a way that tries to address the unique biology of each target and therapeutic opportunity. But at the bottom of the slide, it tells you really the core requirement we always impose on our molecules is you have to create molecules that mimic the best of what nature has given us. The antibody scaffold is highly stable. It's highly soluble. It's produced in massive amounts in your body and it can be produced in massive amounts in industrial manufacturing places and stored for a long time. These are wonderful properties that make them great drugs. We have to preserve those and not create in our engineering solutions problems that didn't exist before. That's always been a challenge in protein engineering, keeping what's great and just modifying what you need. We think we've done a great job and our scientists has done a great job at preserving those properties, so the molecules we make are robust and long-acting and easy to make and always plug into standard manufacturing and development processes, which is what allows our modular partnering approach that doesn't take a lot of Xencor resource. So I'll stop with this slide and just highlight a few points of strategy and some brief updates that we've had in the new year. So at the top of our pipeline, our most advanced internal program is vudalimab, that's a PD-1 x CTLA-4 bispecific antibody that I'm sure we'll talk about with Brian. It was engineered to be highly specific to target T cells for derepression and hopefully antitumor activity that express both antigens. So blockade of CTLA-4 is very potent, but also can be quite toxic. We've made that conditional upon binding PD-1 as well. And so that molecule is now in 2 Phase II studies. Most recently, we started dosing patients last quarter in a study in frontline non-small cell lung cancer, where we're in a dose selection run-in phase prior to a randomized portion against pembrolizumab. So that trial just started. And we expect to provide data updates in the first half on our metastatic prostate cancer studies that have been ongoing. The focus of the company and what we've really been applying the -- putting the accelerator down and pushing our resources has been on our T cell engagers. There's been a step-function advance in validation clinically of solid tumor-targeting T cell engagers and solid tumors have really been resistant to targeted T cell therapies like bispecific antibodies and CAR-T therapies because of the unique microenvironment -- immune microenvironment around them, a molecule we engineered for one of our partners, Amgen, AMG 509, really showed a crack in that solid tumor armor in prostate cancer at the ESMO conference last October that I think shows our design tools can now start addressing these hard-to-reach tumors. And we think there's an enormous potential there. Our lead internal program in Phase I is XmAb819, targeting essentially a novel target in renal cell carcinoma that we think fits the bill for what it takes, and we'll talk about why that's a great target for CD3s. We just filed the IND and expect to have dosed this first half. At the bottom of that interior trio are XmAb541 program, CLDN6 targeting CD3 for ovarian and other cancers. Tough target to engineer against. I think our team did a great job pulling one together. And in the middle, been in the clinic just for about a year is our CD28-targeting bispecific XmAb808, a pan-tumor target B7-H3 that's going to bring, we hope, CD28 co-stimulatory activity in a tumor-specific way. We have a unique low-potency design for that CD28 domain that we hope overcomes some of the toxicities that might have emerged. And then finally, our work in engineered cytokines right now is really focusing on putting a bow on it and getting our molecules to demonstrate the favorable pharmacokinetic/pharmacodynamic and safety profiles we've seen with our IL-15 program, partnered with Amgen, in these molecules. And then looking at external data to apply a stringent filter for further internal development of those programs as we really concentrate our resources on our T cell engagers and vudalimab. So the cytokines are going to be something we really kind of put a bow on we hope, this year. So with that, I'll kind of -- as a set up to the strategy, we've got a focusing of the pipeline on T cell engagers that we think we can really pursue now that the science has shown us which of our platforms to really invest in heavily right now. And I'll leave it at that to maybe dive into the individual programs as you see might be most interesting, Brian.

It feels like I have to pick one. But I want to...

No, we can start at the top.

I'll start with the top. I started covering you, what, 1.5 years ago, 2 years ago. One thing that I always notice is that your platform clearly is already validated, right? You have Ultomiris, sotrovimab, Monjuvi all based on your technology, right? So -- but when you look at plamo for instance, that's -- I think, to me, is a great example where it seems that you didn't move fast enough back then. And -- so now that you -- in the last couple of weeks, you had a major development where you finally -- you sold a portion of your royalty stream, right? Now that you have a cash pile, multiple programs, and it seems like you also kind of narrowed down, you're going to focus on T cell engagers as well. So does that change a way of how you're going to operate as a whole where you're going to be more aggressive about achieving clinical proof of concept and take on a little bit more risk?

Absolutely. The time is now. Again, the science has shown itself that there's an enormous opportunity to open up the solid tumor microenvironment to T cells using these bispecific antibodies that we can tune to balance the high activation you want with the danger of overactivating the immune system, in particular, in healthy tissues. So now that we have these CD3 bispecifics we can make more selective for targeting tumors, I think now is the time to really double down. It's an area where there's not a lot of options. And what we have, we think, a technology lead, then that those leads we all know last a very short time in our industry. And so now it's the time to put the pedal to the metal, be a little bolder and more aggressive in clinical development. And there's such a richness, even within our 3 molecules right there, there's a richness and a hedging of risk by having those. I think now is the time to step back from the sort of hedging of risk we had by looking at multiple modalities now that one has shown itself. So now it's time to really, I think, put the pedal to the metal, fully accelerate and we've changed around some of the team we have on our clinical side, bringing on people who have deep, deep experience in late-phase development so we can ready ourselves for that jump. So exactly, as you said, now is the time to not -- to fully commit.

And when you said putting a bow on the cytokine programs, how aggressive are you in pursuing potential partnership? And given the recent interest in the I&I space, is there a timing that you have to kind of execute on those potential transaction?

I think there's always windows of time when there's strong partnering interest in a particular asset class. And those windows can -- when they wane, they necessarily go away forever, interest ebbs and flows over the years. But I think there's a window of time, you're right, there's a strong interest in novel assets in I&I. Having done many, many partnerships with various kinds of things we've licensed, whether it's just patents or, in some cases, clinical programs, they're unpredictable. So I don't ever want to promise or guide as part of our annual strategy this molecule partnered that way. That said, given that we believe we can rapidly demonstrate the PK/PD and safety profiles that our reduced potency cytokine approach seems to offer and has, in other cases that we've partnered with Genentech, we're certainly open to with that kind of data in hand pursuing partnerships that could deliver value for these and have allow us to really focus our cash pile where we said on the T cell engagers.

Great. So maybe we'll switch into vudalimab. For those of you who are in the audience, you can raise your hands. There are runners on the floor. I think it would be great if you can just kind of go through it by order. Vudalimab, your Phase I -- Phase Ib/II -- Phase I/II in frontline on small cell with chemo, you just started dosing relatively recent. So when you think about the Part 1 success, how do you think about Part 1 success? Because there are 2 doses that you're testing. 2 flat doses. And so what are you looking for in Part 1 so that you have confidence moving into Part 2? I mean there's also another piece to it is that there's also competition. So what specifically are you looking for so that you can confidently go into Part 2 and put more resources into Part 2?

Right, right. So I think the number one goal of this dose -- 2-dose initial run-in on top of backbone chemo for frontline lung cancer, the initial goal is to make sure we have a regimen that's tolerable. We think both doses are well within the active dose levels for vudalimab. But we know from a competitive program at a large pharma that had tried a similar study with a similarly designed bispecific antibody, that a too-high dose of the antibody can lead to an unacceptably high discontinuation rate. The goal of this therapy is to give patients a drug that can deliver CTLA-4 blockade over time to hopefully enhance efficacy but do it in a safer way than you can with your monovalence -- monospecific, I should say, antibodies against CTLA-4 like vudalimab. So an unacceptably high discontinuation rate where a majority of patients drop out is not going to cut the mustard. We're doing 2 doses, so we essentially hedge our bets and have a quick answer is our sort of standard acceptable dose as a monotherapy for vudalimab good or do we need to knock it down. So we hope to know that fairly quickly. It will be safety-driven. Of course, we're going to characterize efficacy in these patients. But with the small numbers, roughly a dozen per, it's not about quantitatively accessing Fc. So discontinuation rate, tolerability, that's key. Make sure there's activity and then move as expeditiously as we can to the randomized portion of that study. So once we pick the dose, we would jump right into a portion of the study, where we're going to 2:1 randomize, 2 vudalimab, 1 pembrolizumab on the standard chemo backbone that's used in frontline lung. We're looking at the lower PD-L1 expressers, so 50% -- 49% and down. And we will run that out. And I think the key endpoint there is looking at the durability of response in PFS.

Any questions from the audience?

Yes, vudalimab. You've got AstraZeneca data, you've got Akeso data, they're ahead of you. What do you view as the benchmark in terms of data? And also how do you position yourself as a chaser? Are there unique attributes to the way you've constructed your bispecific compared to theirs? Or is there some kind of little loophole on the data you think you can catch and close the ground?

Yes, I would say that the benchmarks are a discontinuation rate better than ipi/nivo combo, which I think it's quite possible the AstraZeneca molecule is going to do at their lower dose. I would say that we think our design is very similar to the AstraZeneca molecule, and those are the only 2 molecules that have that design of a truly selective for CTLA-4 design. So that is, let me reframe that, a truly PD-1 dependent selectivity for CTLA-4. So if you don't have both antigens, you're not just going to bind every cell that expresses CTLA-4. We've characterized that thoroughly preclinically, our biomarker data in human [ seems ] supports that. We know the AZ molecule is designed similarly. Other molecules don't seem to have that same design. They have higher monovalent -- monospecific binding. Now with what's the opportunity, our molecule is a little bit lower potency than the AstraZeneca, that the goal here is a safe regimen. You can keep the CTLA-4 blockade going on for as long as possible. It's a slight difference. I think for us, the benchmarks on efficacy are really going to be read out in the randomized portion, not in the safety run-in, where the early data was about a 4- or 5-month improvement in PFS that AstraZeneca saw, which is very encouraging. And we designed our study to read out to give us the ability to discern that. I think that would be very meaningful as an improvement over PD-1 therapy alone. So those are the benchmarks that I think we need to hit that we designed against and now it's about executing against our time lines.

And then on the other 2 fronts, recent KOL call from someone who's been around a long, long time, as long as you, Bassil, basically put a stake in that. Just said, look, we've seen what we've seen out of these modified IL-2s, that's just not a very attractive mechanism of action in I&I. What's your thought to that? What are you trying to show? I know you've got an indication that may not be a commercial indication, more of a proof of principle. But how do you think you're going to be able to disprove that thought process in the IL-2 mutein field?

Well, our design was always based on the idea of disproving the idea that cytokine can't be used safely and tolerably. We reduced the potency of our cytokines 100 to 1,000 fold. They have a long-acting Fc on them. That said, what we haven't done -- and we demonstrated with our IL-15 using that design, very long duration, tolerable amplification of NK cells and cytotoxic T cells in cancer with our Genentech collaboration. Our single-dose data with our IL-2 targeting T regs XmAb564 showed for a single dose, similar kinds of properties. We're doing the multi-dose now. I think what that KOL is probably referring to is there isn't -- there isn't certain data that shows that doing a good job of amplifying T regs with a modified IL-2, like 564, there's no data that shows that doing that is actually going to be superior to other modalities for treating autoimmune disease. And what we're going to do is demonstrate the PK/PD. We hope best-in-class properties that we've seen with our IL-15 for this IL-2 and then await how external validation or disproving of the therapeutic hypothesis is to support our own decision to go forward beyond this initial characterization. And I think that's kind of where Brian's point was, partnering can play a role in that. So I don't -- I think it's premature to say put a stake in it because there have been many resurrections in biotech of mechanisms long dead, right? But I think we do want to allow others to invest ahead of us. We feel with the best-in-class profile we hope we can get with our platform, that's fine. We're happy to be best-in-class for a proven mechanism rather than be the ones disproving the mechanism.

So maybe just on the prostate indication for vudalimab. Data sometime this year, give us a little bit more understanding on how it performs there. But given where you are in non-small cell, what will trigger the next step in prostate?

Yes. So I see the 2 as really quite distinct experiments that don't really read through too much to each other. The non-small cell has -- it's a frontline setting. We've not been in frontline before. These are checkpoint inhibitor naive patients. So you've not pre-filtered those checkpoint-sensitive people out. And it's in a checkpoint-sensitive indication. So that's an experiment that's going to run separately. Do we have the tolerability and duration of treatment? And do we have the efficacy? Boom. On the colorectal -- on the metastatic castrate-resistant prostate cancer, that's a realm where it seems that adding CTLA-4 blockade to PD-1 does open up activity you don't get with PD-1s alone. We saw it in our Phase 1, competitors have seen it with PD-1, CTLA-4s. The question is, is it enough? Is it tolerable? And how does it fit into this rapidly changing landscape? I think the biggest change in the late-line prostate cancer landscape was actually the solid tumor CD3 AMG 509 molecule that we designed with Amgen that showed really powerful data in -- at ESMO. Late-line visceral met patients, the people with the worst disease, with the poorest outcome, 40% OR, very promising early DOR of above 9 months, so that will have to firm up. So I think within the context of that change in the landscape with the spread of PLUVICTO, having an active checkpoint inhibitor has to be viewed through the lens of how might you combine it with these other agents. It's not going to be a monotherapy. Or how are you going to combine it with chemo? And so the lens we're going to look at it with is how is our chemo combo data? What kind of monotherapy activity do we see suggesting how much of a spread do we have from PD-1 therapy alone? I'm hopeful that it will be significant, but we don't know. And if we have that significant spread from PD-1 alone and good activity in combo with chemo, we have an agent that we could then convince ourselves could be the combo checkpoint of choice with these new modalities because that's where I think it's going to go in late-line prostate. And then we have to sort out what's the right trial for us to invest in. So that's an unknown still. So we should have all those decisions this year, however.

So before we get into your own wholly owned T cell engagers, maybe one quick question on the Amgen collaboration. Amgen recently set rapidly advancing. To that note, what's your expectation in terms of how -- what does this mean by rapidly advancing? How quickly can they move in RCC?

You mean in CRPC?

In CRPC. And also, what is the near-term milestone that still remain? Can you just remind us of that?

Sure. So the deal we structured with them is a royalty milestone deal. It is a mid- to high single-digit royalties, and there is about $250 million odd remaining in milestones. Those are -- those are clinical trial milestones, regulatory filing milestones and then commercial sales milestones. The clinical milestones are upon initiation of Phase II, Phase III, et cetera. And so those would be the next milestone payments we get. And we don't break out individual milestone payments. I think we break them out by groups and it's in the multiple tens of millions of dollars for the remaining clinical milestones, and it's in our filings for the details. At the moment, they're doing a Phase I, that's your classic modern oncology Phase I. It's going to end up being multiple hundreds of patients with many, many different lines. And in the old days, it would have been multiple Phase IIs, but here we are. Rapidly, I don't know. I know what we would do with an asset like that in hand. And I know my colleagues at Amgen who have been tremendously -- tremendous collaborators, fantastic collaborators and they're working incredibly hard on this program, so I don't know what they're going to do. But what I would do is I would say, well, look, you have this really excellent activity monotherapy in patients that have exhausted the last line of chemo, what can you do with the monotherapy there? And can you get an active enough OR and DOR signal perhaps convince the FDA for the first time to allow an accelerated approval in prostate cancer. That would be worth a shot or at least a discussion with data in hand. I think you would immediately also want to do a chemo combo and right after androgen deprivation failure of that therapy. So you'd want to do a chemo combo with the taxane current backbone that's used to see how the orthogonal toxicities of a CD3 in a taxane lineup, I think that's a very promising avenue. And I do know from their Phase I work that they've talked about, they're also even looking to combine it with androgen deprivation. So I would develop it broadly, and I'm hopeful that they do the same.

Any questions from the audience? So we have about 15 minutes left, which I think is a good segue into asking the next question. Out of the 3 T cell engagers in front of you, which one is your favorite?

This whole thing about, oh, how do you pick your favorite child, but I've been doing this 26 years. I totally pick favorite children. I even tell my children sometimes, depending on who washed the dishes, "You're my favorite tonight." I would say right now, and it's always going to change based on data, that XmAb819 is the one that I'm most keen on, that I'm following most closely in our pipeline because it is designed very similarly to the AMG 509 molecule. And it's molecular details and in the decisions we made about how to tune selectivity by picking affinities. In fact, we learned from what we did on XmAb -- sorry, on AMG 509 for 819, added some extra tweaks on how we engage CD3 to make it lower affinity but the same potency to hopefully spread that therapeutic window even wider. And we've been 18 months in. Hopefully, as we keep progressing dose escalation, we can start getting to active doses and start characterizing more patients at a given dose. It's just the nature of the target we feel is very promising in terms of its antigen distribution and matches up beautifully with this CD3 platform we created that the AMG 509 molecule suggests has got real great legs. So I think the setup is good for that. As a scientist, that's the hypothesis I'm most eager to test. So it's also no coincidence as the furthest along in our early clinical pipeline, it's the one I'm going to get the data on first. So that's the caveat with that.

Have you started the subcu formulation, incorporating it into the Phase I? And can you tell us how far you are in picking the recommended Phase II data?

Sure. So initially, we started with our IV formulation and have been advancing an IV now for 18 months. Just last quarter, we started dosing patients with our subcutaneous formulation. We did that on purpose. We started at a very low dose with our Phase I in IV because the FDA for immunostimulatory molecules like CD3 bispecifics has very stringent dose selection criteria they want you to start low and prove that you're tolerable as you go up. That's gone very well. At a certain point, when we're high enough, we started doing the subcu, so we didn't have to waste effort and frankly expose patients at subefficacious doses more than necessary. So we're at a higher dose, started the subcu, that's going. It's very early, though. So our goal is to keep advancing the IV, while we move the subcu behind it, get to active doses with the IV that we think are promising to expand and start expanding on the IV even before the subcus gotten further, and we'll be able to judge as we go which one to pursue further. I don't have specific timing guidance because Phase Is, the timing of them is much less predictable than larger sort of studies where you can just keep enrolling patients every day. Phase Is, you have waiting periods and safety windows and whatnot. So for that, I would say that, given the dose levels you have to start out for CD3s historically, what we've seen is 2 to 2.5 years is usually what it takes to get to doses where you start able to characterize the antitumor activity of CD3s. Hopefully, that's lining up soon. We've been 18 months in with the ENPP3 program, and it's been moving forward. So we'll guide more specifically as we keep advancing.

Any questions from the audience? So when you think about the -- sorry, I just lost my train of thought.

No worries.

So ENPP3, so was there any historical precedence from your competitors to suggest that ENPP3 is indeed a good target to tackle for RCC?

Yes. So we initially became aware of this target by doing genetic screens of expression levels of antigens, healthy versus tumor tissue using public databases that we came up with novel search methods to interrogate specifically looking for ones that had bigger than threefold spreads between healthy and tumor tissue that we thought would be suitable for CD3 bispecifics. ENPP3 popped up in our filters and lo and behold, in about 7, 8 years ago, there had been an antibody drug conjugate made against ENPP3 that showed activity in renal cell carcinoma. It showed responses in very late-line patients. However, it suffered from earlier generation ADC technology payload toxicity of ocular toxicity keratitis that scuttled the program. And the innovator there didn't pick up on it because they had some corporate changes and moved away from -- from antibodies. So we saw that data as additional validation of that target suggesting, okay, this is one to give a shot. And so we -- that's part of the reason why we like the setup for the ENPP3 targeting XmAb819.

So you're also working on a companion diagnostic for ENPP3 as well. So do you think that, let's say -- you said 2 to 2.5 years is kind of where the range is, 6 years down -- 6 months down the line, do you think that you may have to enrich the patient population with a expression cutoff for ENPP3?

We don't believe that's going to be the case for the predominant population that expresses this antigen, which is the clear cell renal cell carcinoma subtype. That's the vastly preponderant subtype. I think 80% to 85% of RCC patients are clear cell type. The data from histology experiments that we've done as well as publications suggest that greater than 95% of patients have high expression of ENPP3 in their tumor cells. So we didn't need to screen. We just enrolled. There are multiple other tumor types that express ENPP3. There's reproductive organ tumors, there's lung cancer that express ENPP3 in a much smaller percentage of patients, anywhere from 15% to 30%. That's where the companion diagnostic comes along. We want to expand the population. We've been working on it. We've developed the basic tools, and we've started characterizing it now in our Phase I. And we hopefully, as we get to expansion cohorts, can add different histologies where those histologies are screened with the companion diagnostic, whereas clear cell renal cell we don't.

I think you have one slide that talked about just the range of ENPP3.

Sure. Do you want to take a look?

I saw it earlier.

You did. It is posted on our website as well. There we go.

So how do you think about just the range here? You said that the implementation of diagnostic is -- companion diagnostic is more just for you to tap into other indications. Given how excited you are about this program, how many indications can you take for right at the get-go?

Yes. I think in our initial expansion cohorts, it would be the top, maybe 2 or 3 other indications that we would start looking at. I think that can be done pretty efficiently within the context of an ongoing Phase I. You find the physician -- we're using predominantly big academic centers as our sites, highly sophisticated in doing clinical trials. I think we could -- we would probably recruit physicians specializing in those other histologies and see if we can bring them on board. So I think 2 or 3 is a manageable number while we also look at probably combining our ENPP3 with some of the agents that are known to be -- that are used in renal cell carcinoma, like tyrosine kinase inhibitors, perhaps ultimately looking at how we're going to position the molecule in different lines of therapy.

Any questions from the audience? So just going back to your last comment about combination with TKI and RCC. What do you look for in a monotherapy response? And how much response do you need for you to start doing combination work with TKI and RCC?

Right. So we're looking at this post-checkpoint post-TKI population, call it, third line, second line-plus population because a lot of patients get checkpoint and TKI right out of the gate. I think we have a recent precedent in Belzutifan, where it had roughly a 25%-ish response rate that I think sets a bar for what's worth pursuing and how you -- what you'd want to characterize from there. So that's kind of the general metric we're looking at.

Okay. And then just going to 541, CLDN6. What's the latest progress in the Phase I in solid tumors? And when could we see updates here?

Yes. So this one, we just filed the IND last quarter as we had guided, and we expect to start the Phase I study, be in patients this first half of the year. CLDN6 is a very, very cleanly expressed target in about half of ovarian cancer patients and pretty much all germ cell tumor patients and then maybe in 20% to 30% of a handful of other tumors, including some large indications like in lung. That said, it has close cousins, CLDN9, for example, that are expressed in all sorts of healthy tissues you don't want to be in, but you have to be very careful on how you engineer your antibody to avoid. So we've done one where we've got about 4 orders of magnitude binding difference, and we put it into our format that highly biases towards high expressing cells. So I think we did a good engineering job there. We're obviously going to have to prove this out in the clinic. We're not going to guide on timing there. It is one where we're going to certainly apply the lessons we've seen from our partner, Amgen for AMG 509 as well as our ENPP3 program to accelerate our way through dose escalation. That's in terms of selecting doses for each escalation cohort, how we use anti-inflammatory pretreatment to avoid infusion reactions and CRS, et cetera.

Last couple of minutes. I also want to touch on B7-H3. Mechanistically, how does this synergize with PD-1?

Oh, sure. So I think I've got a slide on that. So B7-H3 CD28 bispecific targets the bottom arm in this slide or sort of the other half of how T cells drive their active activation, proliferation and persistence in particular. CD28 is an activating marker, a co-stimulatory marker called Signal 2 that's needed for long-term persistence and heightened activity of T cells. It really -- the way it does that is it causes an IL-2 dump that then causes the T cells to enter into a more persistent state. That's Signal 2. You need to have something from what's called so-called Signal 1 or the T cell receptor in the natural situation, recognizing some foreign antigen presented to itself on a tumor cell or on a macrophage or on a B-cell, right? So what we do is we have a lot of ways to get Signal 1 going. PD-1 inhibitors derepress that Signal 1 because that Signal 1 on -- through the T cell receptor is blocked by PD-1. So PD-1 inhibitors take PD-1 out of the picture and a TCR can keep going. CD28 would amplify that and make it more persistent because right now, anywhere from 80% to 40% of patients in any given tumor type don't respond to PD-1 inhibitors, right? Even a sort of called warm tumor can often have 60% of patients that respond to it to PD-1. So if you can increase that percentage and by creating a -- by getting T cells to be active where they often wouldn't in some patients, that's great. Note you can also have Signal 1 driven by a CD3 bispecific because CD3 is part of the T cell receptor complex and a CD3 bispecific basically short-circuits that by turning on that Signal 1 any time you have the tumor antigen present. So you can actually combine a CD28 bispecific with a CD3 bispecific. And in fact, our partner, Janssen, just started 2 clinical trials with their CD28. So they're about a year behind us from B7-H3, started 2 clinical trials, one in prostate cancer, one in B-cell malignancy, combining the CD28s we made for them with CD3. So there's a lot of potential across the whole gamut of I/O to use CD28 bispecifics. So we're super excited to characterize them.

My last question, last minute that we have here, is, let's say we come back to the table next year. What do you think we will spend the most time on?

Gosh, I think it's going to be -- so vudalimab will be chugging along in its lung cancer study, and we'll have had some clear answers, go/no-gos and what the go would be in prostate cancer, but that would have been sort of determined already. I think we're probably going to be talking about what have we seen, when are we going to see more, when are we going to see data from XmAb819. I think that's going to be on the tip of the people's tongues. And probably if we're lucky, XmAb808 or CD28, that's the kind of efforts we're putting in. And hopefully, those programs can move quickly. Again, we're not formally guiding. But I think it's going to be all about T cell engagers.

Great. Thank you so much for your time today, and we look forward to having you again. Talk to you all soon.

Thanks so much, Brian.

Thank you.