I think this should be on. Good? Okay, perfect. Thank you. Thanks, everyone, for joining us for the next Fireside. My name is Brad Canino. I'm a senior analyst here at Guggenheim. Thanks for attending our Second Annual Healthcare Innovation Conference. Very excited to the next Fireside with Pyxis Oncology. I've got Lara Sullivan, CEO and Founder here of Pyxis. Thank you so much for joining us.
Thank you for having me.
Okay. Let's just kick it off with an introduction of the lead asset, MICVO, and talk a little bit about the development priorities that you have for the asset now.
MICVO is an ADC. It's a non-internalizing ADC, which we originally licensed in from Pfizer, who did the construct work of building the ADC. It attacks a target called EDB, which is extracellular domain B. It's a noncellular target present in the tumor microenvironment across a variety of solid tumors. As such, the extracellular mechanism allows for the ADC to attack that target, the noncellular target, cleave the ADC in the tumor microenvironment, and then it activates the bystander effect as well as through direct tumor killing for the payload-mediated approach to killing tumors. It has a third component of its mechanism, which is immunogenic cell death, where it activates the local immune environment to augment the tumor cell killing. As such, it has a variety of potential applications because the target is present in the extracellular matrix across a variety of solid tumor types.
We have chosen, based on our phase I, part one data that was released last year, to focus our development going forward on head and neck. We currently have two programs underway in head and neck: second, third line monotherapy and first and second line targeting in combo with pembro.
Okay. Could you talk a little bit more about the phase I in terms of key learnings that you think you pulled out of that in terms of the clinical activity and the safety profile of the drug?
Yeah. The phase I-I study was a basket dose escalation trial, and we had 10 tumor types that were eligible to participate in the part one. The selection of those tumor types was driven by the target expression information that we had across those tumor types, as well as early PDX models that showed tumor regression across multiple tumor types. As in any good dose escalation study, especially for something that's a novel target with a novel mechanism, we work very deliberately as we escalate across tumor dose levels. We got the dosing up to 3.6-5.4 mg/kg, which was identified as the efficacious dose range. We dosed a total of 80 patients in that study, of which 77 patients we reported safety data for in that part one, and about 65 or 66 were efficacy evaluable.
What that study showed was tumor regression observed across six different tumor types with confirmed responses and a particularly pronounced confirmed response rate in head and neck. From a safety perspective, we showed broad tolerability across a number of dimensions. In particular, with any ADC, we tend to keep an eye out for payload-mediated side effects such as ocular tox, neutropenia, neuropathy, cutaneous lesions, pneumonitis. Those are the things we all watch for. We showed minimal grade three events, in fact, none in ocular tox and neuropathy, treatment-related AEs grade three and above. We had modest cutaneous and neutropenia observed in grade three and above from a treatment-related AE perspective. These were extremely sick patients. They were end-of-life patients. They had a median of four prior lines of therapy, so we were median fifth line.
We had many end-of-life patients, which means their next step after treatment with our agent was if their disease was progressing, hospice, or passing away. We had some patients who passed of disease, not of drug-related effects. Overall, the part one showed a nice tolerability profile within the expectations of what you would expect for an ADC, as well as broad efficacy with rapid and deep responses observed across six different tumor types. In particular, in head and neck, the confirmed response rate was 50%, five, zero.
Okay. To double-click on the safety profile, can you just talk a little bit about what the dosing strategy and philosophy was for that part A of the phase I for MICVO to define that dose range? Then how much data do you have to support the go-forward expansion dose?
Yeah. Different companies have different philosophies around dosing strategies, especially for new agents. In the case of our agent, we felt it was important to really continue to escalate the patients and the dosing schema right up to the maximum tolerated dose, the MTD dose line. The MTD is typically the dose at which patients are unable to tolerate the therapy, and you stop dosing at that point. It is a bit of an art as well as a science to find how high you can escalate just before you get to that point. By doing that, by definition, you invite some AEs because by getting close to that line, you measure that by the amount of AEs that you have. We would expect, as we saw in our safety table, to have some AEs.
We had modest grade three treatment-related AEs, as I mentioned, but that's part of the profile of getting to a risk-benefit optimized component. The 3.6-5.4 as that optimal dose range was identified through a couple of manners. One was direct clinical observation around where we started to see meaningful tumor regression, of which 3.6 was the lower boundary of that. The other was leveraging some prior information for the ADC construct itself in the setting of the Pfizer HER2, which is a different target, a different program, but it was the same ADC construct. We knew from the Pfizer program that 3-4 mg/kg was about what the efficacious dose range or tolerability range ceiling was for HER2. We looked at that as a guide as we were going through the escalation.
It turned out because of the specificity of our target and the extracellular nature of our mechanism that we were able to dose escalate above what Pfizer was able to do with that same construct in a different target. I think that was a really important piece of information for us because it shows the power of this target and this mechanism by having the same ADC construct applied in two different settings.
Okay. We are waiting for some clinical data to come, but in the interim, you've presented a lot of translational biology work throughout the course of 2025. How does that work build your confidence around the mechanism of MICVO?
Yeah. As I mentioned, the mechanism is a three-pronged mechanism, right? We've got the payload-mediated component of direct tumor killing and then activating the bystander effect, which is when the tumor cell lyses and the payload is recycled, and then the third prong being the immunogenic cell death. Those three components of the mechanism, we had hypothesized that was how MICVO was working as we took it into first-in-human studies. To your point, Brad, we then began a journey of understanding the biology and building up the translational insights to support our hypothesis on that mechanism. In fact, we put out 10 posters just in the last four or five months alone that validate this mechanism and provide proof of mechanism. Those posters were presented, six of them at triple meeting, two at AACR, and two at ESMO.
Those posters demonstrated some key points that support this mechanism. For example, we put forward in these posters the identification and concentration of some of the key proteases that exist in the tumor microenvironment, which cleave the ADC and allow the payload to begin to do its thing. We also demonstrated some data that showed T cells being activated locally. We also put some data out that showed the nature of the tumor microenvironment from a mechanical or physical perspective. We used AI-enabled digital path tools that showed almost these tunnels of fibers within the head and neck tumors that allows the payload to traverse, whereas in pancreatic, where we saw less responsiveness, those tunnels were sort of blocked. The fibers were more tangled.
These are all components of the translational work and the biology work that point to the explanation of how our three-pronged mechanism works, as well as why it works in certain tumor types, maybe more than in other tumor types. That goes to the nature of the tumor microenvironment and how the cellular and noncellular elements are behaving in the tumor microenvironment.
Okay. On the clinical side, can you discuss the separate cohorts of head and neck patients that you've been enrolling and why you've been enrolling those particular cohorts?
The head and neck signal that we showed in part one with the confirmed ORR of 50% and also a confirmed complete response was in those heavily pretreated patients of four prior lines of therapy. We felt that was actually a very significant signal that clinically these patients were responding to our MICVO agent while being in the end-of-life setting. We said, "Let's see what that signal looks like when we take it into earlier lines of therapy." We designed a monotherapy program addressing second and third line head and neck patients. We have divided that population into two arms. Those arms are based on prior treatment history.
It's the same recruiting characteristics, the same second or third line population, but if you're assigned to arm one, it's because you have had prior platinum and PD-1 treatment, which is today's standard of care. If you're assigned to arm two, it's because you've had EGFR and PD-1 treatment, which we believe will be potentially the future standard of care. By designing the trial this way, what we're showing is that there's a place for our agent in today's standard of care and a place for our agent in potentially tomorrow's standard of care. It speaks to the agility by which this agent can compete in the head and neck marketplace.
To take it a step further, the arm one cohort, which is the platinum and PD-1 resistant patients, these are the patients that we have all seen data recently from Merus, Bicara, J&J, Corbus. Our arm one data set will be directly comparable to those competitor data sets. Our arm two data set will be essentially for a new line or a newer line of therapy that is emerging, which is the line that would come into place post-EGFR treatment. Today, it is a cats and dogs picture. There is no defined therapy that comes after cetuximab. People might do a platinum or a chemo. The response rates in that line today are pretty low. They are maybe 10%-15%. We believe there is a space for a clear alternative coming into that post-EGFR line. That is the monotherapy program.
We also have a combo therapy program in combination with PD-1. We have a supply agreement with Merck. We are combining MICVO with Keytruda, and that is in dose escalation phase as well.
Okay. And the scope of the data update, from which cohorts will you pull patients to disclose?
We have guided to an upcoming data disclosure in the fourth quarter of which we are in. We have decided to include preliminary data from both arm one and arm two of monotherapy, as well as the preliminary emerging data for the head and neck participants in the combo setting. The upcoming data update will include each of these three components of data. We are doing that because we believe that given the evolving competitive landscape where we saw two recent data sets come out of ESMO, from Corbus and from J&J, it is important that we participate and help shape the understanding of the emerging head and neck landscape.
The data sets will allow investors to see holistically how MICVO behaves in the second and third line setting in the case of the monotherapy arms, and in the first and second line setting in the case of the combo study, as well as how MICVO behaves depending on the prior treatment lines. That is the difference between the monotherapy cohort one and the monotherapy cohort two.
Yep. You mentioned some of the data sets in the space, but what do you think is the efficacy required in second, third line in particular to be a credible competitor in the space?
Yeah. I think that expectation has been well set and is very entrenched at this stage. Merus kind of got us out of the gate on that in around a 35% ORR, which we've seen recapitulated and potentially improved upon with the J&J data set recently. I think at a minimum that 35% is the floor to be a credible competitor for those platinum and PD-1 resistant patients that are second and third line. We could argue, is it 35? Is it 40? Is it 42? But they're all within the same band. That's sort of the minimum. I think in terms of our arm two, where it's the post-EGFR line, there is no established expectation right now because that line has not really been well characterized.
The dialogue in head and neck over the last two years has really been around Merus or Bicara, which one, right? We've had our bulls on either side. Now we've got J&J in the mix, and now we have another mechanism in the mix with ADCs being available. The nuance of how the market is going to shape up is starting to emerge. We feel that we can actually participate in a variety of these different segments because not just around the line of therapy and the prior treatment, but our data set that we put out last year was HPV agnostic. We saw responses in both HPV positive and HPV negative. The EGFR competitors are not able to say that. They've seen differential response rates in those two different populations. There's a number of ways that we can participate.
I think with our data disclosure, we will also include some guidance and commentary around how we see the competitive landscape shaping up and where we can compete in there.
Is there any mechanistic reason for why MICVO might have different clinical activity in an EGFR naive or EGFR exposed population?
Yeah. There really is no reason to think that arm one, the platinum and PD-1 experienced population data will differ from the EGFR and PD-1 resistant population. Mechanistically and biologically, it's the same patient pool. In fact, we showed in our part one data set already last year that we had tumor regression in EGFR treated patients, taxane treated patients, prior other patients that were on other ADCs with other payloads, other IO. We have no reason to believe either from the clinical data that we've already accumulated, which showed tumor responsiveness to any of those agents, as well as the mechanistic and biological insights that we've generated recently, as well as the fact that the patient pool is essentially the same other than the prior treatment regimen that's participating in our trial.
All those three things combined lead us to believe there'll be no difference in responsiveness from arm one to arm two. I mean, a basis point or two or the denominator, right? But you're not going to see 30 basis points difference.
Okay. Since this will be an interim cut, so you'll still be enrolling patients into the cohorts, how do you think about the interpretation of unconfirmed responses for this update?
Yeah. This is always the dilemma that every early stage company or early in development company faces, right? Because we all have the desire to get our data out there while it's fresh, and yet there's the competing desire that is out there to see data with durability and lots of validity behind it. Every individual company kind of makes a choice on this and a philosophy on this around whether to go out with a mixture of confirmed and unconfirmed data or to just hold and wait until you have everything fully confirmed. We have not taken a position yet on our intended approach on that. We do know from our part one data set where we did have a mixture.
It was primarily confirmed, but we had some unconfirmed that were in there that you do not always, you cannot necessarily use the mixture or the proportion of unconfirmed or confirmed to be a predictor of what the final maturing data set will be. We have not sort of clarified our position on that yet, but we are empathetic to the dilemma that comes with how do you decide what to include and what not to include?
How much might you be able to say about durability, and how might durability even be attempted to be described in an update like this?
Again, that's one of those components that's got a little bit of a natural tension in the desire to get data out and the desire to have the ability to say how long that data has incubated for. The way we tend to handle that is we rely on vignettes, right? Some of the earlier enrolled patients can provide a really rich source of insights, even if we're not able to give statistically significant durability stats. Vignettes is one of the ways. I think another is just pattern recognition, reverse engineering, right? We did put out swimmer's plots with our part one. We had them posted for the entirety of the patients that were in the study, as well as a double click on the head and neck population in particular.
I know in your coverage initiation, you took a look at that swimmer's plot in head and neck and did some analysis on it. I think that's also very helpful because when we go back and we look at some of those earlier data points from part one, they can be potentially very informative in triangulating, right, to what we may see in the future. We would sort of counsel investors to take a look at the analysis you've done in the coverage initiation report and to pay attention to some of the vignettes that we tell because in the absence of letting the data cook for a year, that's usually the best we can do.
Okay. How do you understand where MICVO might be able to go in front line from this data update? What are you particularly going to be focused on?
Yeah. In terms of the front line opportunity, we've seen some great data sets again from Merus and Bicara in combo with Pembro, setting the standard at 65%-ish, right, plus or minus a few data points, which is phenomenal for patients. I mean, if we even think for these head and neck patients, what's just been becoming available to them over the last couple of years compared to no real evolution for a long time. It's fabulous. We look at that as great for patients. We look at that as setting a new sort of expectation, and we look at our mechanism and point to the three prongs, in particular the immunogenic cell death component and the prior history that we've seen with ADCs plus IO combos.
Padcev plus Keytruda is a great example of the power of those two things together in bladder, where they saw a phenomenal response rate. We think that if we're able to compete with the expectation that Merus and Bicara have set around that level of responsiveness, there could be a place for us in that segment as well. Obviously, the data will be the main driver on that.
Okay. Coming out of this update, how do you plan to further develop MICVO, and what might be the right time to start to define some of these pivotal populations just at the start for second, third line?
Yeah. When we provide our data update, we will provide commentary on how we see this competitive landscape shaping up, where we see the scope of competition, where things are more open. Again, HPV status is clearly one of those areas where we can potentially see a differential competitive set. There are also ways to think about this with respect to other cuts, right? We talked about prior treatment status. There is also tumor size, anatomical location, CPS score. There are a number of ways that I think this market is actually more sophisticated and nuanced than maybe the dialogue has been to date. We will include commentary about that, and that will naturally inform the commentary we will provide around the next steps of our development path.
You'll get data, market segmentation, and observations on how we see it evolving, and then how we plan to compete by developing the agent in the clinical path ahead.
Okay. Do you consider MICVO to be a head and neck drug only, or where else do you see opportunities?
Yeah. I think the answer to this question goes all the way back to kind of the origins of MICVO and what's unique about it. The target being expressed across a variety of solid tumor types and the mechanism of extracellular cleavage being feasible across a variety of solid tumor types lends itself to a broad potential set of uses in a variety of tumor settings over time. I think as a small biotech, thinly resourced, whether it's people or capital, right? We're living in these times of constrained capital. We want to make sure first and foremost, we are highly focused on head and neck. We deliver the data sets there. We prove the value of this target, this agent, and the biology that we've been doing will also provide the clues for future opportunities for value.
Head and neck to start and potentially more exciting things to come over time.
Okay. Unfortunately, we're out of time, and we'll have to leave it there, but very much looking forward to the clinical update later this year. Thank you so much for joining us, Lara.
Thanks, Brad. Thank you.
Thank you for listening.