Arvinas, Inc. (ARVN) Earnings Call Transcript & Summary

I think we've got everyone assembled. Hi, everyone. Welcome to the fireside chat with Arvinas. I'm Yigal Nochomovitz, one of the biotech analysts at Citi. I think everyone knows me. If you do have a question, just you can e-mail me directly at yigal.nochomovitz@citi.com, and I'll try to check that during the session and ask your question. You can also just use the chat feature on the showcase platform and ask your question that way, it will also get sent to my e-mail. So it's my pleasure to welcome the Chief Scientific Officer of Arvinas, Ian Taylor. Ian, nice to see you. Welcome, and thanks for doing the session. Maybe as a starting point, just to kick things off, I think it will be valuable if you could give a 3- to 5-minute overview of the Arvinas protein degradation platform. What are protein degraders? How do they work? What are the key assets that you're developing, 110 and 471? And then we can take it from there.

Great. Thank you, Yigal. Yes. Thanks very much for invitation and the opportunity to chat with you today. Yes. So we're Arvinas. We are the PROTAC company. We are the leaders in the very hot field now of targeted protein degradation. We're located in New Haven, Connecticut. We were founded in 2013 by Craig Crews, a professor at Yale. He's one of the inventors of PROTACs. He's been working in this space for nearly 20 years now. So what is a PROTAC? PROTAC, it stands for proteolysis targeting chimera. It's a small molecule that induces the degradation of disease-causing proteins by using the natural way that the cell turns over proteins, whether they be mutated or just to outlet their usefulness, which is called the ubiquitin-proteasome system. So how does a PROTAC work? Again, it's a small molecule, one end of the molecule binds to the target of interest that we want to degrade. And the other end of the molecule binds to the workhorse enzyme of the ubiquitin-proteasome system, that's called an E3 ligase. And what the PROTAC does, therefore, is bring together the target protein and the E3 ligase. There is actually 600 E3 ligases in the human genome, more than there are kinases. And -- but the PROTAC just picks out one specific E3 ligase to do its job. And so PROTAC is facilitating a protein-protein interaction between the target protein and the E3 ligase in what we call as a trimer complex. And once it does that, the E3 ligase's job is to add a ubiquitin molecule to the target protein. It actually does that 4 times. And there's a chain of 4 ubiquitins attached to the target protein we want to degrade. And that's the signal to the cell, this protein is ready to go off to a structure in the cell called the proteasome, which then degrades that protein into its component peptides and, ultimately, amino acids. So that's why it's called the ubiquitin-proteasome system. The target protein gets tagged with ubiquitin, goes off to the proteasome and gets degraded. So it's really quite elegant. The PROTAC is basically facilitating the complete elimination of this protein from the cell. And it does it very potently and very, very specifically. You can think of PROTACs as having the power of the gene-based medicines like CRISPR, antisense oligonucleotides in terms of eliminating proteins but with the drug-like features and the benefits of small molecules, because it is a small molecule. So I know you'll be asking later about our 2 programs that are in the clinic. We have 2 PROTACs in Phase I clinical trials. Our first one, the very first PROTAC in clinical trials is called ARV-110. It is a degrader of the androgen receptor, which, as you know, is a driver of prostate cancer. So our Phase I trial is in men with metastatic castrate-resistant prostate cancer. And in May of this year, at virtual ASCO, we presented the very first efficacy data for a PROTAC and showed that we were seeing, in a subset of patients, 2 PSA50 responses and a RECIST response. We also showed for the first time of a PROTAC, degradation of a protein in human tumors. So we show decreased levels of androgen receptor in a human tumor. We also updated the PK and the safety, showing that these molecules are very drug-like. The PK has been dose proportional, the safety profile has been very benign to allow us to dose escalate very quickly. So we're really excited by the data that we showed at ASCO because it showed the proof-of-concept for our PROTAC platform and really demonstrated that PROTACs can be drugs that can benefit patients that have been highly pretreated. The patients in our trial, the median lines of prior therapy is 6. So these are patients in which standard of care inhibitors have failed, and we were able to show clinical benefit and, in fact, efficacy in those patients. And our second PROTAC that's in the clinic is called ARV-471. It's an orally bioavailable, much like ARV-110 is orally bioavailable, estrogen receptor degrader that's being developed for patients with ER-positive, HER2-negative advanced or metastatic breast cancer. All these patients have been pretreated with CDK4/6 inhibitor. Back in October of last year, we showed initial safety and PK data from the first cohort of patients, which we were very nice, it was a nice complement to what we showed for ARV-110. But then in our update in May, we further disclosed an interim update that plasma exposures, as we've continued to dose escalate, have been dose proportional. We've not seen a DLT, and again, have early evidence of ER degradation of tumors. So again, a very nice complement to the whole platform that PROTACs are behaving like traditional drugs, but with degradation properties. So our plan for the next update for 471 as well as 110 is in the fourth quarter of this year. Just to round off the company. We have several neuroscience programs headed by brain-penetrant PROTACs that are degraders for tau and alpha-synuclein. That was one additional breakthrough. So one of our breakthrough several years ago was making PROTACs orally bioavailable. There's always skepticism that you can do that, same for brain penetrance. That was another breakthrough that we had to allow us to get into the neuroscience space. All of the programs that I talked about and including the rest of our undisclosed pipeline, which we'll be talking about later this year, some of the programs, both in oncology and neuro, are wholly owned by Arvinas. But we also have 3 strategic partnerships, one with Genentech, with Pfizer and with Bayer, which -- those are discovery-based collaborations in which they give us their targets they want us to degrade. We work on them collaboratively to develop PROTAC degraders, which ultimately we hand back to those partners at certain stages. And lastly, with Bayer, we set up a joint venture called Oerth Bio, which has brought PROTACs into the agricultural arena going after weeds and fungi and pests, which is really novel. To be honest, when I joined the company a little over 4 years ago, I've never envisioned, and I don't -- I'm not sure even Craig Crews envisioned PROTACs in the agricultural space, but then we're doing that with Oerth Bio. So watch that space, it's pretty cool.

Okay. Well, that was a very good intro. Obviously, you mentioned the ASCO data, which is front and center for a lot of investors. And you mentioned the 2 PSA50 responses, quite deep responses, I think, 74% and 97%. And those were in the ones that had the AR mutations. And then you had 5 patients that didn't show response, and I just want to get a sense as to, those are patients that could have the potential for response? And is it just a question of perhaps more time on therapy? Maybe you could comment a bit as to why we haven't seen a response yet for them. I know for 4 of them, they were really only on therapy for a very short period of time and wouldn't necessarily expect to see PSA50 response after 4 weeks on at 280?

Right. Exactly. Yes. So limited follow-up for several of the 280s at the time of the ASCO presentation back in May, you're 100% right. Some of the other ones we were able to show the data. Yes, exactly. So as part of our protocol with ARV-110, we're doing FoundationOne Liquid circulating tumor DNA analysis. Looking at AR mutation status, so that's how we knew those patients that had a response had 2 particular AR mutations, which we know that we can degrade. But we also see whatever else is in the about 100 gene panel that's in their liquid test. And so we've observed mutations in other genes, which have been seen in the literature as potentially non-AR-dependent resistance mechanisms. And so of course, if the tumor is being driven by something other than AR, even if it's wild-type AR, and we degrade, we're not going to see a response there. So those are hypotheses that we've generated as to why we're not seeing responses in some of those patients. It's really too early to detail what we've seen because it's still small patient numbers. But again, I will say that the mutations we're seeing are not novel, these have been reported in the literature. So as we enroll more patients at higher doses as well as backfill into some of the lower doses that the protocol allows and we generate more data, we can test that hypothesis to see, okay, maybe the percentage of tumors that are driven by AR is less than we thought at the start of the trial because, again, going back to the number of prior therapies, the median is 6. Our trial has -- an AR-directed agent has never been tested in such a late-line population. And so I think a lot of the data that was published before may not be 100% relevant to our particular study, and these are the things we're learning as we go forward.

And so is it still an open question as to those 5 wild-type ARs that didn't respond? I mean, do you know yet if they had other driver mutations at this point or...?

We do.

Okay.

We do. We do. Yes, exactly. And those are the hypotheses that we're generating. So we've noticed mutations in at least some of those patients that potentially could be non-AR-dependent drivers. Again, such a small patient number that I don't want to get into what they are because we'll generate that data as we get more patients, but that's one of the things that we're investigating as to why those patients were not responding relative to the other patients who did, right, they had 2 specific AR mutations. Maybe that's a signature that those tumors truly were dependent on AR and degrade and then gave those deep responses, PSA as well as one RECIST response. So it's, again, small patient numbers there, too, but that's the type of data that we're going to be generating going forward as well.

Just so I'm clear, I mean you're not ruling out the response, not seeing a response with the 280. I mean, it's only been 4 weeks, at least to have a large data cut.

Correct. Oh, yes. No, no, we're definitely not ruling it out, and we're not ruling out that we can work in -- that ARV-110 will work in patients that don't have those driver mutations, absolutely not. But seeing that signal, of course, you have to follow it because it's really such a strong signal and could be a path to accelerated approval, right, if it holds up as we enroll more of those patients.

Right. I mean another question we were getting a lot after ASCO is just time to response. And I don't know if you shared that in a lot of detail. I know there's a -- there can be a tail on the PSA50 response that can start after a matter of weeks, but it can also take several months. To what extent can you comment on -- for those 2 deep PSA50 responders, when did you start to see those responses develop?

Yes, right in that range. So one of them was basically within a month hit PSA50. And then the other was, it took closer to 3 months. And then, of course, you have to have the confirmatory reading as well. So that's where the Prostate Cancer Working Group 3 kind of does the cutoff for PSA response at 3 months. And so both of those patients were in that range, and which is why the other patients were too early to really evaluate relative to that working group criteria.

Okay. And then for the next data update, you mentioned ARV-110 data in the fourth quarter. Could you just help us set expectations there in terms of how many patients we're going to see? What the degree of follow-up will be at 280 milligrams? Will patients have reached 12 weeks on 280? And will we see anything from the 420-milligram cohort?

Yes, absolutely. So those patients will have reached 12 weeks by the time we do the next disclosure. We'll be disclosing the data on 420 and any doses that we might have beyond that. So it will be the same sort of the disclosure that we had at ASCO in terms of showing PSA50 responses, any RECIST responses, any AR degradation data that we have, more PK, more safety, et cetera. So basically, whatever data we have, whatever cutoff we set for that disclosure, we'll be putting into that fourth quarter release.

And is that going to be in conjunction with the 471 update? Or are you going to have separate updates for those 2 programs?

We haven't fully decided. To be honest, I mean, it's going to be -- the 110 disclosure is going to be at a venue of our choosing. Basically, it's something that we host. And we haven't really talked about the 471 release. It could be a medical conference, it could be at the same time at 110 or different. We want to give ourselves flexibility to have as much of a fleshed out story in terms of the amount of data that we have before we do those.

Okay. And what -- and in terms of the longer-term progression of the dose escalation, I mean, is the goal here to go all the way to an MTD? I know that according to the animal work, that you could go up to 900 milligrams in humans safely. So are you going to take this to an MTD? Or do you feel that that's not necessary, and you might stop sooner?

Right. Yes. So the goal -- the original goal was to try to reach an MTD as our dosing paradigm. So far, of course, we've not reached that as to the -- at least at the time that we disclose the data. If we don't reach an MTD, of course, we'll have a recommended Phase II dose that's not MTD. And that will be based on the compilation of the data. So for example, will the PK have plateaued with higher doses? If we're not seeing any more exposure, then there's probably no reason to go any higher. Whether we have any additional biomarker data that tells us something about the level of AR degradation that we're achieving. If we're getting really, really high, and it's not getting any higher whatever biopsies we have, then, again, we may decide that there's no reason to go higher still if we don't reach an MTD. And of course, there's always the option, as many oncology trials compounds have done, to have 2 doses that we bring forward into expansion, right? So there's no real need to wait to get to an MTD if we feel like -- as we're seeing the exposures and we know where we are relative to that preclinical data that you cited, maybe that's good enough to bring forward as a first expansion dose and then have a second one later. So that's another option that we will certainly explore. We're trying to be as aggressive as we can with this molecule because we've seen the efficacy that we've talked about, and we really want to push as fast and as hard as we can. If that means multiple doses, then we'll do that, too.

And in terms of the new patients that you're enrolling, I mean, I know you can't get into specifics, but in general, are you making sure that there are enough patients in the dose escalation that have the verifiable AR mutation such that they would be a degradable mutation so that you just -- you don't run into a situation where everybody is AR-wild-type and might have another driver?

Yes. So in the escalation portion, that's really hard to do because you have to screen prospectively. And these mutations are at the 5% to 10% level of all castrate-resistant prostate cancer patients. So we don't want to hinder getting to the recommended Phase II dose, whether it be an MTD or otherwise. But certainly, when we get to the expansion, that's certainly one thing that we're exploring putting into the protocol, that would be part of an amendment, is to have more prospective screening so that we can have subcohorts that have those specific mutations. Again, not giving up on the other mutations, even V7 because V7 is only -- it's not clear if it's a driver of resistance, and it's always a subpopulation of the total AR, so we could still get a response there. But yes, in the expansion piece, we will be more focused on prospectively screening. But in escalation, that wouldn't really slow it down too much.

Okay. Fair enough. And in terms of patient biopsies, to look at the AR degradation histologically, are you doing that in certain patients or all patients? And just maybe recap what you've seen there from the immunohistochemistry?

Yes. So great question. So again, it's optional on the protocol to have paired biopsies. So a pretreatment and a biopsy at some point between cycle 1 and cycle 3, cycle being a month's worth of treatment. Again, it's hard to mandate those biopsies in a Phase I protocol because it's really -- it would really slow down your recruitment and your enrollment. So -- but we were able to get 2 pairs, 3 pairs really of biopsies in the data that we disclosed back in May at virtual ASCO. One of those pairs was a 35 mg patient. So again, not -- we didn't see any AR degradation there nor do we expect to based on the plasma exposure that, that patient achieved. We also had another paired biopsy. I think it's 70 mg, where one of the biopsies had no detectable tumor in it. So it was impossible to compare pre and post. And then, of course, the data that we showed at ASCO was a 280 mg patient. And so those -- those then were the only 3 paired biopsies that we have. We knew all along that getting paired biopsies in this particular indication would be difficult and that proved to be so. As a hedge for that, we included circulating tumor cells to be able to look at AR levels as well. And as I've said publicly, I think probably the only disappointment I have in the study so far is that the number of patients with usable CTCs or a number of CTCs and certainly those expressing AR has been much less than we expected based on literature reports. So the CTCs have just not been able to be interpreted in terms of whether we're seeing AR degradation. And that's a disappointment. I don't really have an explanation for why our patient population seems to have fewer CTCs than the literature. We're using the Epic platform. So it's a very validated platform. So we just don't have a lot of paired biopsies, and we don't have a lot of CTC data to complement that, unfortunately.

I know this is sort of -- it's early days in your studies, obviously. But in terms of the target product profile, what is the draft version of that look like today? I mean, is it -- are you looking at the patients that have, obviously, failed chemo, failed the AR mechanisms, that's where you would start? And do you believe that this protein degradation strategy could potentially be promoted to something that could either be used in conjunction with an enzalutamide or an abiraterone or maybe even ahead of it?

Yes. No, absolutely. So yes, our current population is, obviously, heavily pretreated, as I've mentioned. And as part of our expansion, we want to -- that we were talking about earlier, we want to get less-pretreated patients in the study. But long range, as those other agents like apalutamide, darolutamide move into earlier lines of therapy, non-metastatic castrate-resistant prostate cancer, castrate-sensitive prostate cancer, resistance mechanisms will arise there that we will be able to address. And so we can move into earlier lines as well. But ultimately, we're not really afraid of doing a head-to-head study to get into those settings as the preferred anti-androgen agent. Sure, based on our preclinical profile, doing head-to-head studies, we think the degrading is better than inhibiting. The other aspect that you mentioned is potentially combinations. You probably won't do one with enzalutamide because there's a competitive nature there. They're both binding to the ligand binding domain. But abiraterone is an interesting combination. It's one that we have modeled preclinically. It does make sense in terms of total pathway blockade. I know that enzalutamide and abiraterone combination didn't look so good, but I think apalutamide just had a publication recently where there was some benefit to the combination. So that's another potential area that we would look to test clinically as well as what we did preclinically.

And can you just spend a few minutes just talking about the -- you have a backup compound, the ARV-766, I believe. What is different about 766 versus 110? Or is it just a backup?

No. No. Well, so we always planned to have a backup to 110. As I mentioned, ARV-110 was the very first PROTAC in the clinic. And so I think it was a very prudent decision to make sure we had a backup molecule because, obviously, it's an unmet medical need space and androgen receptor is the driver. So there'd always be a place for it. But of course, as I mentioned before, ARV-110, we knew did not degrade the L702H mutation in androgen receptor. We knew why there's a steric clash between the warhead that binds the AR and that histamine that comes up as a mutation. And so that was one thing that we wanted to make sure that we picked up. Even though 702H is, it's 3% to 9%, depending on the publications you look at of the CRPC population. So we wanted to make sure that in our backup that we picked up those patients. And so we were able to rejigger the compound and make sure it can degrade 702H. So that's the major difference to it. That molecule is now in IND-enabling studies currently. And once we get that data and then we have the 110 data from the clinic, we're going to be strategizing quite strongly to have a holistic approach to what we'll do with 766 relative to 110. But I wouldn't say it's strictly a backup because it definitely has a different profile than 110.

And you mentioned AR-V7, which is apparently detectable in 1/3 or so of metastatic castrate-resistent prostate cancer. What is -- what are your plans in terms of developing a degrader for AR-V7. How far away from the clinic are you and when could we see that one?

Yes. Yes, I know exactly. So on our public pipeline, we have V7 as a program because we have an AR franchise. The role of V7 as a resistance mechanism is, as you know, continued to be under debate in the field. Some people feel it's a driver, other people have data where it's a -- just a passenger as a consequence of AR amplification. So there's data on both sides. And we want to make sure, because we wanted to be in the AR franchise area, to have a molecule against V7. I think it's too early to project when that will be in the clinic, but we do feel it's important. And of course, we'll see with 110 and potentially 766 that goes forward because both of those molecules are ahead of V7, of course, whether we have activity in V7 patients. And if we do, if 110 and 766 take care of V7 patients, then we would drop our V7 program. But we're going to let the clinical data speak to, at some level, the strategy with that program as we continue to bring it forward.

Is there any specific rationale scientifically why 110 or 766 would work with AR-V7?

yes. So as I mentioned earlier, V7, even though it's expressed in tumors, it's always at the minority of the total AR species relative to full length, for example. And there is data that V7 functions as a heterodimer with the full length. It can homodimerize as well, but forms a heterodimered signal, so that if you degrade it to full length with 110 or 766, that would take away that heterodimer and that V7 function would be abrogated. So -- and there is data there, but there's also data that V7 can homodimerize. And that's -- so I'm sort of not an absolutist when it comes to V7. People always ask me, do you think it's a true driver or not. I think I've learned over my career that every tumor is different. So I suspect in some V7-expressing tumors, V7 is the driver. I suspect in other V7-expressing tumors, it is not. And so there's no universal answer there. So again, we'll see how the clinical data plays out, but that would be -- the heterodimer hypothesis would be why 110 and 766 could work in V7-expressing tumors.

Right. But if you had a tumor where everything was homodimered and everything was AR-V7 homodimers, then you would need to have a separate drug?

Yes, but that almost never happens. I mean there's never just tumors that express only V7 and not the full length. V7 is always the minority species. So there's always some full length floating around.

Got it. Okay. So for 471, you're obviously going to have data coming up in the fourth quarter as well. Maybe tell us what to expect there in terms of how many patients' worth of data, how much -- what's the duration of therapy going to look like? What dose level you might be at, things of that nature, just to help everyone?

Yes. Yes. No. So we haven't really guided to any of that, quite frankly. So again, it will be the same that we do with 110 that whatever data we have, whatever dose levels we have, we will show it, again, the RECIST responses that we may have, the ER degradation that we will have, the PK and the safety. So it's -- the dose escalation is going well. As I mentioned before, in May, we talked about seeing dose proportionality in terms of PK, no DLT. So we're still escalating and seeing ER degradation. So we will flesh that out in the fourth quarter with additional safety, PK and efficacy data. What we're looking for is a safety profile that, again, will allow us to move into first-line because I think you've seen with the other SERDs, because -- especially when we're post CDK4/6, all the non -- similar to what we talked about with AR, all the non-ER-dependent resistance mechanisms. Certainly, post CDK4/6, that's been very well documented recently, that it's a high percentage of patients that have non-ER-dependent resistance mechanisms of post CDK4/6. That's why you really need to get to first line. So we want to make sure we have a safety profile that will support that, pharmacokinetics that can support that. So that's kind of what we'll be shooting for in terms of what is could look like for that compound.

And in terms of sort of the lessons learned from 110 in terms of applications to 471, is there anything specific in terms of the study progress with 110 that's helped you in guiding the design or progress of the 471 study? Or are they just separate studies?