Good afternoon, everyone. I'm Tara Bancroft, a senior analyst at TD Cowen. Thank you for joining us at TD Cowen's 44th Annual Healthcare Conference. Today we have Brian Sullivan here from Celcuity. It's a pleasure to have you. Thanks for joining us.
My pleasure.
So I guess to start, can you just begin with a general overview of Celcuity and a general update to kick things off?
Sure. I founded the company about 10 years ago, initially focusing on development of a platform that could analyze signaling activity in a patient's live tumor cells. And one thing led to another, and we eventually evolved to development of therapies that are targeting cancers that involve the PI3K/AKT/mTOR pathway, or PAM pathway as we call it. Our lead candidate is gedatolisib, PAM/PI3K/mTOR inhibitor, highly differentiated. I'm sure we can go into that in the rest of the questions, but highly differentiated from other drugs in this class, and essentially solves the riddle of how to effectively blockade this pathway, very complex pathway, without inducing excessive toxicity. Two programs that are active currently. We have a phase III program in advanced breast cancer, second-line setting.
Then we've just started enrolling patients in a prostate cancer study, men who've progressed on their prior androgen receptor therapy and are castration-resistant.
Okay, great. So let's get right into the lead program, geda. And so can you start by just providing some current and historical context for the PI3K/AKT/mTOR pathway inhibitors and how geda is differentiated?
Sure. So the pathway was discovered roughly 20 years ago as an oncogenic driver. And the structure of it was characterized, and it became clear that this pathway involves six interactive components, four Class I isoforms, mTORC1 and mTORC2, two subcomplexes of mTOR, and that the biological imperative to address disease that involved this pathway would require comprehensive shutdown. And so almost immediately after the pathway's role was discovered, nearly every major pharmaceutical company developed PI3K/AKT/mTOR inhibitors because that was the biological imperative. But they weren't great inhibitors because they weren't able to be very potent against these different targets. And they were also too toxic. So none of them made it out of phase I. And so it was like they touched the third rail. And so then the development strategy evolved to narrowly focusing on single components, essentially PI3K alpha, mTORC1, AKT.
But the challenge is that while they may have resulted in more tolerable drugs, they gave up, in our opinion, a lot of efficacy. Because to control this pathway, if you're only inhibiting a single node, you'll induce compensatory resistance along the other components. Inhibiting alpha will potentially activate mTORC1. And so geda is differentiated because essentially we're addressing the biological imperative. We're inhibiting all four Class 1 isoforms equipotently, low nanomolar potency, subnanomolar. And we are also addressing mTORC1 and mTORC2. But because of the activity of the drug, we are able to achieve very high potency, low concentrations of drug. And the drug is IV administered. One of the challenges of drugging this pathway is that it involves critical metabolic processes, including your glycolytic system. That activity takes place in the liver.
So if you have an orally administered drug that's hitting PI3K alpha, last place you want it to be, you're going to induce a lot of hyperglycemia. So we avoid it on first pass because it's IV administered. But because of the PK profile of the drug, it has very balanced volume of distribution, essentially balanced concentration of plasma and the liver or tissue in general, we essentially don't get overly concentrated in the liver. So we report just a fraction of hyperglycemia relative to the PAM inhibitors that have preceded us, as an example.
Okay. So you have some pretty robust PK data. So can you explain how that really makes geda special and how that translates to PI3K inhibition, the target inhibition?
Sure. Right. So the two factors in, I think, the activity of geda, one is that by controlling the pathway comprehensively, you are not fighting or experiencing the negative effect of cross-activation of these other pathways. And so we have, I think, a pretty robust set of non-clinical data that shows the effects in live cells when you comprehensively blockade this pathway compared to what occurs when you are only addressing a single target, i.e., alpha, mTORC1, or AKT. And we found, for instance, that geda is 300 times more potent than these other drugs in live cells, and that it's the only one that's cytotoxic. It's the only one that can induce cell death at high rates. And so we think that's a function of the mechanism. You can isolate the mechanism with that type of study.
And because of the mechanism, you are able to induce cytotoxicity at very low concentrations of drug. And so we've shown that we can maintain coverage above the IC80 value throughout the seven-day treatment period. And we've got animal PK/PD data that shows that this concentration is directly related to antitumor control. So we think the drug is able to maintain its effective concentration. Three weeks on, one week off is a very manageable schedule, we think, for patients. And more importantly, it can induce the appropriate levels of pathway control.
What % of target inhibition would you say is the threshold to show clinical efficacy?
Well, I think our data has shown that at the current concentrations, at the end of the dosing period, we're inhibiting roughly at least 80% of downstream signaling pathway components. And so we think 80% is the threshold I think other people have used. And we think that the non-clinical data points us to that number. And then we think that's backed up by looking at the clinical data, where we've reported very robust antitumor activity when geda is added to an existing regimen.
Actually, can you go into that?
Sure. Okay. Sure.
Give us a brief recap.
The relevant data is we did phase I data that evaluated initially patients in an escalation arm to essentially confirm the dosing schedule, dosing concentration to use, and then evaluated four groups of patients. One was treatment naive. One had prior endocrine treatment without CDK4/6. Then two arms had prior treatment with CDK4/6. Treatment naive patients showed very, very robust response, 79% response rate, 48 months median PFS, which was to put that in context, current standard of care outcomes for patients who receive CDK4/6 plus letrozole was 25 months and 55% objective response rate. We compare very favorably to the existing regimens. Then in a second line setting, with the dosing that we're taking to our phase III, reported 63% objective response rate, 12.9 months median PFS.
You have to break out the kind of context between patients who have a PIK3CA mutation versus those that don't. In the patients that have a PIK3CA mutation, the standard of care or the reported results for standard of care drugs are between 5.5 and 7 months. In the patients who lack PIK3CA mutations, wild-type patients, current standard of care reports results from two to four months. Essentially, these patients today with current standard of care treatments are not getting substantial benefit from those therapies. If we were to compare the results we've reported, if we are able to replicate those results, that's always the big question. We think we would establish a new standard of care for the second line paradigm for this patient population.
Have others that have targeted this pathway before, have they ever been able to reach the wild-type patients?
They haven't. It's interesting. So there was a very relevant study that just was reported out last year and the drug was approved. It's called capivasertib/AKT inhibitor. And they found that they could induce about a three, three and a half-month improvement in PFS in patients who had a PIK3CA mutation or an AKT mutation, but no treatment effect in patients that had none of those mutations. And we think that's consistent with the biology of the disease. We think the primary role that pathway plays is intrinsic. It's not a function of the presence of an activating mutation. Patients that have a tumor with an activating mutation may benefit from those drugs. That's been consistent. But we think all things being equal, you will get a better benefit if you more comprehensively blockade, whether you do or don't have one of those mutations.
In that phase I beat, you had several arms. Did you do any analysis of chemo-free patients and PFS there?
Sure. So we had two arms that had patients who received prior CDK4/6. And that's the paradigm. 80%+ of patients today in the front line setting are receiving CDK4/6. So you really have to isolate that patient group, or at least in our opinion, you do, to report out actionable results. And in the two arms that we evaluated, or two arms we had, one arm had roughly 50% of patients had prior chemo. And the other arm had roughly 20%. And there was a difference in outcomes, which you would expect. Alpelisib did a study where they ran one arm on patients without prior chemo treatment, one arm with prior chemo treatment, and saw that the patients without prior chemo treatment reported roughly a 30% higher median PFS than those who had prior chemo treatment.
In our phase III study, we wanted to design one that provided essentially almost a pure second-line look at that population. That meant not including patients who had prior chemo. In our early phase studies, we had patients with prior chemo. We didn't break out because the numbers get too small and we just don't want to ascribe significance by reporting them out. We think it's fairly well established that with targeted therapies, patients who have had prior chemo will, all things being equal, be less responsive to a target therapy than those who haven't had prior chemo.
Okay. Before we get into that, since you mentioned the hyperglycemia and everything with the mutant-specific inhibitors, can you talk about the safety profile so far? Because I think it's important to then get into the metabolism and the excretion.
Right. No. And in some ways, it's as important as the ability to effectively target the pathway because that's been the big barrier. And so in our early phase study, we reported only less than a 9% discontinuation rate due to an adverse event profile. And that's in 138 patients. So a meaningful enough population, you could draw some conclusions, I think. Just to get a context of that number, the CDK4/6, palbociclib, fulvestrant discontinuation rate was 6%. So essentially, you'd see a 3% difference in discontinuation rate, which suggests, we have to confirm, that geda doesn't add a meaningful level of adverse events that induces discontinuation. And so the second, then getting into the more detailed review of the AEs, hyperglycemia, we had no discontinuations due to hyperglycemia, 7% Grade 3, Grade 4 rate of hyperglycemia.
That compares to, let's say, Alpelisib, which has reported 26% or actually 37% in their phase III study. So clearly different levels, doesn't require meaningful patient management, at least the feedback we've received. The one adverse event that we do have to manage is stomatitis. In the early phase study, no prophylaxis to treat stomatitis was used, which was unfortunate because Novartis did a study with their drug everolimus, a randomized study called SWISH, which compared patients who received for two cycles of treatment prophylactic mouth rinse, dexamethasone, versus those that didn't. They found that Grade 2 or greater stomatitis was reduced by 90%. And there was no Grade 3 or Grade 4 stomatitis. So it can really impact favorably the incidence of stomatitis. We're including that in our phase III study.
So we would expect that we will report much, much lower stomatitis in the phase III study than was reported in the early phase study. And then the other class of AEs that have been a concern in this setting have been GI related. For instance, there were delta inhibitors, PI3K delta inhibitors in the heme setting. These high levels of toxicity along the GI tract, colitis or diarrhea, vomiting, we don't induce any such. We haven't reported any colitis. The vomiting, diarrhea are at levels comparable to very, very well tolerated drugs and more favorable than when we compare single agent data to these other inhibitors. And again, that's a function of its route of administration. You're avoiding these target organs on first pass and then this balanced volume of distribution.
Just for reference, those mutant-specific ones, like for hyperglycemia, GI, stomatitis, what are the rates that they show relative to the numbers that you've given?
So if you look at the lab values, which are really the most objective way of assessing impact on glucose of these drugs, capivasertib was around 60% all grade. Everolimus was about 70%. Alpelisib was about 80% all grade. Everolimus, capivasertib were about 10% Grade 3/4. And alpelisib as a single agent was 26%. But then if you look at gedatolisib, it's about 25% all grade. So essentially less than half or a third of these other drugs. And the Grade 3 was, as a single agent, was 1%. So as a single agent, it again does not significantly disrupt your glycemic system, regulatory system.
All right. Thanks for all that context. Because I think it's really important to understand very deeply that historical context because the phase III now, I think, is something that everyone should be caring about right now. You mentioned a couple of things like inclusion criteria. But can you walk us through the design of that trial?
Sure. So it's an all-comer trial where we're enrolling patients. The primary relevant criteria is that they must have had prior CDK4/6 treatment. And then we do an assessment of their PIK3CA status. Patients who lack a detectable mutation are assigned to arms A, B, C, right? Triplet, doublet, singlet, and fulvestrant. And that study is powered independently, a separate independent statistical analysis plan. So for all intents and purposes, it's a separate study. And then for patients that have a PIK3CA mutation, they're randomized to arms D, E, F. Again, where you have a triplet, geda, palbo, fulvestrant. And then the comparator arm in that setting is alpelisib fulvestrant. And we're also evaluating geda fulvestrant. I should have mentioned the comparator arm in the wild type is fulvestrant. And again, they have context of what the comparators are.
Fulvestrant's reported in several recent clinical trials between two to three months median PFS. So we think it's unlikely that we'll get surprised by that. And then for alpelisib, the data has ranged from, I think, reasonable to assume seven months as the bar that we would expect to need to beat.
Okay. Is there anything you can discuss about powering at this point, alpha allocation?
Sure. So before we designed the study, or rather while we were designing the study, we interacted with the agency quite a bit. And so we had two type C meetings. We presented our initial design, got feedback, and then wanted to measure twice, cut once. So we wanted to confirm that our interpretation of their feedback was reflected in our protocol. And it was. So the statistical analysis plan was one of the areas that they reviewed. And we're using conventional powering levels. Whenever you have a control arm that has low median PFS, you can detect a statistically significant effect size difference with a relatively small population. I mean, you can detect a one and a half-month effect size difference with a small population and three months with a very tiny population.
So you will, in that case, have to overpower your study so that you have an appropriate number of patients for a phase three study.
Are you able to see? You probably are. The patient backgrounds, regardless of which arm that they're enrolled in? How do you know that you're enrolling the right patients?
Sure. No, a big part of managing the study is being very, very deliberate about assessing the patient screening documents, their screening assessments. And so that's what our medical directors do. We think so far, so good. We think we've done a very good job. IDMCs review how effectively we're screening these patients and enrolling these patients. And so I think our team has done a great job. And there's no evidence that there's any challenge in being able to make sure we're enrolling the right patients.
How about duration of response to prior therapy or even time since prior therapy? Because those things can also.
We have to control for that. These patients can have had more than a 30-day gap from their prior therapy. You want to manage that, or at least the 30-day period for washout, 14 days actually for targeted therapies. But one thing we control for in the study we have as a stratification variable is their time on prior treatment. There's groups of patients that are more or less sensitive to endocrine treatment. And so you want to make sure your arms are balanced. Obviously, randomization helps. But because that variable can be quite significant on the responsiveness to the drug or regimen, we want to include that as a stratification variable. We also include geographic location. FDA almost always requires you to have that as a stratification variable.
Okay. Then I know that there are several arms. So there I ask, what should we expect to see for efficacy? I wish we had a whiteboard so you could write down everything.
Right. Oh, exactly. Well, I think it's what we expect versus what we need. Yeah. Okay. I think if we were to talk to KOLs, what they generally consider to be clinically meaningful is a three-month improvement relative to your comparator. And so if we're able to show three months between geda fulvestrant versus fulvestrant, and three months on top of that with the triplet, I think the clinical community would consider that a home run. I mean, that's the words that have been used if we could achieve that. And our data suggests that's not a ridiculous bar to beat. And one other factor to consider when assessing probabilities, because ultimately with where we are, you have to have some assessment of the probability. So we reported very promising data, 13 months median PFS comparable between wild-type mutated.
But we're also in the phase III study enrolling, we think, a patient population that has more favorable underlying baseline characteristics. So for instance, we're not enrolling patients who've had prior chemo. We talked about that. We think by not including those patients, you are less likely to include patients who will be less responsive, which is the case we had in our early phase study. We're also allowing patients who have bone-only disease to enroll in the phase III study. Whereas in our early phase study, we weren't allowing those patients. You had to have visceral metastases. For instance, 80% of our patients had liver lung metastases, which is significantly higher than what you would typically see in a phase III study. So I think those tweaks to the characteristics of patients' eligibility criteria should be considered as to the goods, at least we do.
And then one final factor, which we won't control, but which we think will be related to those two components, those two variables, is the duration on prior treatment. So we reported about 13 months median PFS, which was about equivalent to the median duration of treatment on their prior therapies, which is generally considered to be very good. For patients who are coming off of a frontline therapy, their immediate prior treatment duration median would be about 20 months. So we think if we were to project, it would be more likely that the patients enrolled in our phase III would be closer to that 20-month number than to the 13-month number. So all in, there's always the unknown unknowns, which can bite you in the rear. But at least we have some favorable known variables that are to the good.
Net-net, we think the study is designed to really provide a clean read on how effective the drugs are in a true second-line setting.
I got to say, to hear a double-digit month, let alone 20, I mean, for the context in the breast cancer space, I mean, think about the SERD trials over the last few years where they were talking about, well, we would really like to see five months, six months. And now for you guys, we're talking 20 months. So that's.
Well, I'm not sure.
I know they're different. Yeah.
No, I'm simply pointing to the profile of patients that we're enrolling. I think there tends to be a correlation between duration on prior treatment, depending on the class and what the prior experience is. At least we've seen some correlation there.
Okay. How about on safety? Are you getting any whispers of any kind of events like the hyperglycemia?
Right. So yeah, our CMO and I and other senior management are blinded to the data. But on the safety side, it's been reported that we are seeing far lower levels of stomatitis. And that's what we would expect with the use of something like this dexamethasone mouthwash.
Do all patients use the mouthwash?
Yes. That's another to the good, right? Phase I didn't have that. If you were to say what's the factor or the AE that leads to most dose reduction with stomatitis? If we were able to address stomatitis with the use of this prophylaxis mouth rinse, I think we'd get low dose reduction as a result.
I see. Okay. I know one thing we didn't talk about was the metabolism of it. Because I think one of the questions that investors wonder is, how are you avoiding hyperglycemia? Is it because of how it's metabolized and excreted? Or something else?
Well, it's very stable. So if you do an assessment, you find less than 1% of what's excreted as a metabolite. So it's a stable drug, mostly excreted through feces and urine. But I think the biggest variable to look at when it comes to hyperglycemia in particular is this volume of distribution and the fact that it's essentially balanced between plasma and tissue. And that's why it's able to sustain its concentration throughout the dosing window.
Okay. All right. In the last few minutes, I really want to get to the addressable market. So based on the patient population that you're targeting, what is the size of the addressable market? And what do you think is a reasonable estimate for peak cells?
Yeah, I think most people would say there's roughly 28,000-30,000 eligible patients in the second-line setting, post CDK4/6. And if our data comes through the way we hope, I mean, so this is all academic at this point, but it would be reasonable to model and say, okay, if you had 12 months treatment period, $12,000 a month, a drug cost, I'm not projecting it, but that's kind of consistent with targeted therapies in this space. That's a $5 billion addressable market. Now, I'm not estimating or we're not projecting anything like that. But I think it provides a sense that says, okay, you can provide a lot of variables to haircut from that market potential and still end up at $1.5 billion-$2 billion peak revenue. So I think in the second-line setting, there's a very, very large opportunity.
One way to kind of establish credibility for that is to look at alpelisib's data. Alpelisib, even though it's a drug that a lot of doctors are critical of because it's rough on patients, despite that, and despite that the median duration of treatment is only five and a half months, or at least that's been what's reported, they generated $500 million of sales in roughly 35% of the market. If you were to normalize that, say that essentially represents for the market as a whole, roughly $1 billion run rate. I guess our perspective is, okay, in that segment, they're able to generate that level of penetration and revenue. With our drug addressing a bigger population, more tolerable, longer duration of treatment, it certainly feels like $1 billion is a very, very, very conservative estimate of what we might be able to achieve.
Okay, great. We have one minute left. So I was wondering if I could ask one final question is what do you feel is the most underappreciated aspect of Celcuity by investors?
I think if people understood the importance of the pathway and appreciated that to address it completely can have a profound impact on the efficacy of a drug hitting this pathway. And if they appreciated how low the penetration is of PAM inhibitors in the breast cancer market or the prostate cancer market, another area where we're focused, they would better appreciate how significant this company can become. Because it is one of the, I think most people would say it's one of the most important oncodrivers in cancer. It's the least drugged if you just look at it on a revenue basis. HER2 drugs, which are targeted just for the most part, patients with HER2 mutations, $10 billion category, only 8% of solid tumor patients have HER2 mutations.
If you look at the drug category or the tumor types we're in, CDK4/6 inhibitors generate $8 billion of annual revenue. AR inhibitors in prostate generate $10 billion of annual revenue. Those are the key pathways to address. We think PI3K is one of the key pathways to address in those two pathways. We have a long way to go from where we are to accomplishing anything like that. But it just says we're fishing in a very big pond with a lot of fish. And that if we can follow through, we could build a very, very important company as a result.
Great. Okay. With that, we are out of time. I want to thank you, Brian, for coming and everyone for attending.