Good afternoon, everyone. My name is Akash Tewari. I cover pharma and biotech companies here at Jefferies. This is day two of our healthcare conference in New York. I have the pleasure of hosting the Terns management team, including Amy, who recently joined the company, and Emil. Amy, why don't I hand it off to you for some introductory remarks, and then we'll get started on the one-on-ones?
Great. Thank you so much for having us, Akash and the Jefferies team. I will just start off by saying we'll be making forward-looking statements, and you can refer to our website for relevant disclosures. We're excited to be here. 2024 is an exciting year for Terns. As we've guided too, we have two exciting data readouts in the second half of this year. You know, we're a company that focuses on small molecules with validated mechanisms of action, and we're in both oncology and obesity. And so it makes for an interesting year and an interesting company.
That is right. So, you know, I just came back from ASCO and dealt with that ride from O'Hare, which is painful, but Scemblix kind of stole the show. I mean, if you're talking about practice-changing datasets, DB-06 was up there, but so was Scemblix, and the data that they were you know, Novartis was able to show in first-line CML. And I think, you know, a lot of people ask you how to develop your program, and I actually kind of respect your answer 'cause like, I don't know. Like, depending on what Novartis does, will determine what we do from a clinical development perspective, and I think that's right. I think the question I get from investors right now is: Okay, is the Scemblix data good or bad for Terns?
On one end, higher, more selective inhibition with, you know, these next-gen TKIs is showing a benefit on MMR in a first-line setting versus Gleevec. At the same time, they were able to show efficacy with QD, which was, you know, a talking point I think your team had. And there's this question of, well, is the bar too high now for Terns? I'd love to get your take, how you view that data, and looking at what they showed in first line, where would your compound differentiate versus Scemblix?
Yeah. So that's an easy answer for us, Akash. We think it's great news for Terns, and most importantly, it's great news for patients with CML to see, you know, a compound after almost 25 years that has better safety and better efficacy. And that's a lot of the reason why we've been working in this space and developing another allosteric, because we knew that that asciminib is a great drug, and we're not surprised to see this data. At the same time, we see opportunities to potentially improve. We've already shown clinical data showing that we do not have the food effect that asciminib has. And we see asciminib taking quite a significant share of first line. We don't yet know how much.
You know, Vas, the CEO of Novartis, made a statement that they project $3 billion in revenues, which I'd imagine they have good data to support that. And so, you know, but regardless, we think that generic first and second gen TKIs will still be used, and so there will be the opportunity for Terns to do the same, a similar trial, the ASC4FIRST, if we choose to. We also see that because first and second gen TKIs will continue to be used in the front line setting, that there'll be a real opportunity in second line, where asciminib does not have a label, and Terns is currently enrolling patients in our dose escalation trial.
Understood. I mean, maybe to that point, I actually think Novartis is probably being conservative here. I'm guessing the street's probably more like 5+. But even if it is 5+ and it's just first line, it still only implies probably 40% of patients are going to get on Scemblix first line. So there are gonna be plenty of patients in the second line setting who haven't been dosed with an allosteric TKI agent. Is that the way to kinda think about it? 'Cause like, again, base case, what do we know now? Is the way to think about the Terns compound, if you were to make a decision today, whether you go head-to-head against Scemblix, or you attempt the first line opportunity, or you're going second line. It does seem like second line is the most logical step, but you tell me.
Yeah, I mean, I'd say we're not gonna go head-to-head with asciminib to be in front line. So I think we have the optionality to do either or both, and that to be in the front-line setting, that we can do a similar trial because there will still be patients being put on first and second gen. So that trial is repeatable.
Would there be an appetite on your end to maybe not go against Gleevec, but a second-gen TKI in first line? 'Cause again, like we're seeing the second-gen TKIs are going generic. I think if you look at ASCO, one of the pushbacks in the discussion panel, 'cause CML is actually one of the cancers where oncology cost seems to be relevant. What about not going against Scemblix, but going against a second-gen TKI that's about to go off patent? Do you think that could be an option for you in your front-line setting?
Yeah. So to Amy's point, number one, we don't need to go head-to-head against Scemblix, and we know that the regulators are not gonna make us do that. I think a control arm should reflect real-world practice. which is why I commend Novartis for designing their control arm the way they did. That 50% stratification to Gleevec and the second-gen TKI is exactly reflective of the current practice in front line, and they showed that they beat that, that control arm. We would probably do the same thing.
Got it.
I think to your point around second-gen TKI, yes, the cost differential was there, but now you're talking about a drug that has shown superiority on both efficacy and safety. And even though, and they clearly did show that they won on both fronts very significantly versus Gleevec, they won against 2G TKIs numerically looked equivalent to slightly better, but the key there is the safety was definitely substantially better than 2G TKIs.
So in a disease where the control arm and the current standard of care have already changed overall survival to match that of age-matched controls, an equivalent or slightly better efficacy with substantially better safety is really valuable for both clinicians and patients. So that's why we think that the feasibility of doing a study very similar to ASC4FIRST, with a very similar blended control arm, has a very high PTS and is very, very feasible based on the fact that we still expect a good proportion of patients to get generic drugs in the front line.
Interesting. Okay, maybe to that point, now I'm gonna ask you, 'cause I've always told you guys, "Publish more of your data." I still. I'm gonna say that on a transcript right now, but, look, this is the question that I think a lot of investors have is there's not been a lot disclosed on from an ADME perspective on your drug. So we have the Hansoh trial has been going on for two years. So I'm gonna ask you this, Nikhil. Were you surprised to see MMR rates that high, given that you know, you've actually probably explored very high dose levels in the Hansoh trial, that's ongoing in China? So were you surprised with asciminib data-
Not at all.
Knowing what you know with Hansoh?
Not at all, because you're talking about apples and oranges when you're talking about a frontline population versus a relapsed refractory population. So in a frontline setting t he same requirement on target coverage is not there as it is for a third, fourth line patient population, where the disease is now more advanced, more clonally evolved, has more resistance mutations. So you need more target coverage, therefore, more dose to get the equivalent efficacy. Now, keeping in mind that, you know, the asciminib dose range that was tested in the phase I, study was a wide dose range. It went from 40 all the way up to a cumulative dose of 400.
Right.
And by the way, they also have a 400 milligram dose approved in, in relapsed refractory patients, specifically for T315 mutant patients. So there's nothing unusual about the wide dose range that we're testing with 701 in Hansoh and our study, because the mutational spectrum of the disease, as you advance to third, fourth line, is much wider than it is in front line. So you've, you've heard about data where, for example, with even second gen active site TKIs, the dose of the drug doesn't have to be as high. We've seen some data with dasatinib 50 milligrams versus 100 milligrams in the front line, arguably being as good. So the, the ability... The disease is much more sensitive in the front line, and that's why I'm, I'm not surprised that they hit the, the MMR rates they did-
At 80?
At 80.
Okay, this is a really provocative statement because I think a lot of people, even me, I was looking at the data. I'm if I look at the doses you're testing, you go up to 500, and, you know, there's not the. There's many a biotech who says, "Hey, I have really good potency in vitro." Then you don't take into account plasma protein binding, bioavailability, et cetera, et cetera, and then the human dose ends up being higher than expected. You could say there's some analogies in oral GLP-1 space, which we'll hit into on that very topic. But what you're really telling me is, we should not take the fact that you're dosing in a late line population as high as 500 to say that on a human adjusted exposure basis, that you are meaningfully different than asciminib. Is that the correct way to think about this?
No. So, yeah, kind of. I think what I'm saying is that this, the metric that you use to set your target dose, target efficacious exposure, usually at the starting point of your development, which is the relapsed refractory p opulation, is based on the preclinical data, where you look at what the IC50, IC90 values are in, you know, specific cell lines engineered to have either native or mutant BCR-ABL, which in CML historically has translated into, you know, relatively well into the clinic. But that being said, what you see in a cell line is no.t a linear.
Right.
Translation to the clinic. In the clinic, you're gonna have much more heterogeneous populations of patients with native, with mutated. So you always target to get to an exposure that not only gets to your target efficacious dose based on the basic KCL-22 cell line is what Novartis used for asciminib. We've already put out data from our healthy volunteer PK data, to your point, that we should be publishing more of our data, which we just did, right? We just put out our healthy volunteer data showing that at the 80 milligram dose of 701, you're already hitting target coverage at trough. That's the IC90 of KCL-22, protein binding adjusted.
Which is exactly what Novartis used to set that target of 80. What we're saying is our 160 milligram dose is folds above that. Why are we doing that? Because we think this therapeutic index of this class is so high that you should take advantage of it, that you're seeing a safety profile that allows you to achieve multiple fold target coverage over not just native BCR-ABL, but the full heterogeneity of mutations that you run into in a relapsed refractory population. So our strategy is we think that allosterics can be better dose optimized. Asciminib 80 milligram a day is a great dose.
Right.
But there are some confusion around what the dose of that drug is. There's 80 approved, there's 40 BID approved, there's 200 BID approved. We think that the therapeutic index of these drugs is such that you can get to a target dose at which you get full coverage of the mutational spectrum of the disease, that allows you to simplify dosing to a single dose once a day, hopefully for every patient, so that you don't have this confusion around which dose do I use, and that'll translate to even potentially better efficacy while maintaining this really great safety profile.
Okay, that's, that's really interesting. I don't think that's well understood by investors either. So bottom line, you feel confident that with your drug, you're going to be able to get target coverage that could be maybe orders of magnitude higher than the asciminib 80 milligram dose. Is that a fair characterization?
Our aspirational goal is to get there.
Understood. Let's put it this way, you've seen data from Hansoh so far that obviously isn't publicly disclosed, where you have, you feel comfortable from a safety perspective of getting those types of exposures?
So as we've, as we, you know, guided to, as we put out, a statement last year when we announced the design of our phase I, study our starting dose of 160 milligram QD was based on data that we were able to leverage from our partners' phase I, study in China, where they had covered doses below that. They started at 40 milligram, and they go up to 400 milligram in their study. But we were able to leverage that data to get to a starting dose, like I said, is already above the dose that I just said is 80 milligram, where we're achieving trough target, you know, for native BCR-ABL, that showed good safety and early evidence of meaningful clinical activity. So we're confident that hopefully we're on our way to get to that aspirational goal.
Understood. Maybe last on CML, and again, like, if I were to give you a piece of advice, disclose your baseline data before you publish. Because, you know, Emil, I've talked to you about this. I think you make an amazing point. Context matters here, right? Are patients getting off a second gen TKI 'cause they can't handle the safety, or is it because they need to get higher log drops? You could show what wouldn't be optically impressive log drops, but it might be entirely because, A, you haven't followed the patients long enough, but, B, they might be coming off other regimens. You know, take this opportunity, tell us how to read this data. What are the patients you're enrolling with your data set that's coming off in the back half of the year?
How long of a duration are they on your drug? Is it appropriate to make any takes on efficacy? And can you outline it maybe qualitatively-
Yeah, yeah.
What these patients are coming off of in your study?
So, you know, without going into obvious specifics and details about our patient population, I think I'll just say that interpreting efficacy data in CML takes a couple of extra steps of calibration than you would otherwise do for, you know, other oncology indications. So, for example, if you're looking at lung cancer, you know, the typical baseline characteristic shows, you know, age, number of prior lines of therapy.
Usually, the number of prior lines of therapy alone is sufficient to give you an idea of how high that bar is for the drug in terms of achieving efficacy. CML is a little different. Because it's a chronic disease, number one, it takes time to get to the deeper responses we're used to seeing, like MMR at six months or MMR at 12 months. So that's really hard to show meaningfully in an early data cut from a phase I, study.
How early is the data cut?
So again, we've been enrolled—we'd have been enrolling, we're guiding to initial data from our dose escalation cohorts in the second half of this year, which point we would have not been enrolling for more than a year for this study, right? About just about a year. So it's a very early data cut, but that doesn't mean it's an interpretable data set. What you have to calibrate for is not just prior lines of therapy, but what the starting point in terms of baseline BCR-ABL transcript level is in these patients. So, for example a two prior TKI patient whose baseline transcript level is 0.1 or 0.15, which is right above MMR, has a much lower disease burden than a two prior TKI patient with a baseline transcript level of greater than 10%.
Right.
That's about a 10,000-fold higher disease burden, leukemic burden, in patient number two versus patient number one. So you're gonna set a very different. In the clinic, you set a very different goal for that second patient versus the first patient, where you're not gonna get to an MMR signal. It'll be, you know, most TKIs won't get you at an MMR signal within 3 months with that, the second patient with more than 10%.
Right.
So we intend to, to your point, baseline characteristics are super important in CML. We intend to do sort of an investor day event, mid-year this year, where we wanna sort of walk people through how you calibrate appropriately to interpret an early CML data set that takes into account these and other factors.
Awesome. I'm very excited for that. I will be.
Me too.
Okay, let's go to the GLP-1. And, you know, we published a deep dive telling people not to own your stock going to this event, and the idea was, dose matters. When we look at, you know, there's companies like Regor or the old Pfizer danuglipron team, they've been able to get, danuglipron modified, two modifications. You get to about 8-12 hours of half-life, so you don't get 44 hours like we see with oraglipron. You're gonna have PK variability, and then on top of that, plasma protein binding adjusted, you're getting to 120. Isn't that exactly what Structure's data was?
Outside of it, maybe, let's say, being an oraglipron-like scaffold, but they get what? 8-15 hours half-life. They have 60% vomiting. I don't think that improves that much despite titration. But you see the stock reaction, you see what investors think. So I ask you this: are you comfortable with the safety and efficacy bar that Structure showed, and what is your confidence you might be able to deliver that or improve on it with 601?
Akash, I think we see it a little bit differently. I think we see the bar really as other, other GLP-1s with 28-day data. That's the first point. I think that the Structure data with 12 weeks is different than a 28-day data set.
Fair.
You don't have as long to titrate. You can only lose so much weight in 28 days. And so we really set the bar at 3%-5% body weight loss over those 28 days, with the thought that the AEs are more difficult to manage in that type. So we've seen a pretty wide range of AEs in that.
Right
Timeframe a nd really, what the 28-day data is telling you is, do you have a drug here that's worth taking into a 12-week trial to be able to evaluate potential differentiation? So I think that's one, one sort of point of view that that question brings up. I think the second really is about you, you talk solely about half-life. I think, you know, as you mentioned in your note and you've talked to us about before, I mean, we see it more as AUC and target coverage during waking hours as being really important.
Right.
I mean, we've seen, we've seen effective GLP-1s with a wide range of half-life, and that AUC is important, and I think the data is really evolving about peak to trough and where that's important in terms of AEs. And so we're really just gonna have to see how the data plays out for TERN-601. We're in that trial. We're gonna be putting that data out in the second half of this year, and that there is, you know, other data evolving at the same time that's gonna help to evolve our understanding, which is still in a bit of the nascent stages.
What other data are you referring to there that's coming out?
I think Pfizer data-
Oh, okay, QD Dan-
Yeah, I would say other data's coming out. There seems to be lots of folks in this space, so I don't have a memorized list in my head.
No, okay. No, no, but I just wanted to make sure, QD danuglipron.
Yeah, yeah.
Understood. Okay, very interesting. So one of the things that I think is, you know, I feel like Pfizer is probably kicking themselves with the titration scheme they picked, right? Like, in the end of the day, you can't down titrate patients discontinue the drug. They had very high vomiting rates, too, very much like Structure, but Structure did not have a lot of discontinuations, right? Instead, they decided, let's have, you know, 40% down dosing. You have learned a lot, I'm sure, from, you know, how to design a good titration profile with your oral GLP-1. How have you designed that profile? Do you allow the ability for patients to down titrate, and what triggers that threshold to down dose in your protocol?
So, yeah, I think this relates to the earlier point you made around sort of the bar in terms of the Structure data and the reaction to that. I think some of my key takeaways were very aligned with what you put in your report is, number one, there's still a very strong appetite for oral GLP-1 molecules to get into this space. Number two, I think people are essentially realizing that GI tolerability, GIEs are part and parcel of this mechanism.
It's, it's an on-target effect, and so that gives you a certain sort of range of places you can play in terms of how far you can take that tolerability threshold. So I think that threshold is discontinuations, to your point. I think the positive reaction, I think if you look at the Pfizer phase II data that was, you know, recently put out last year versus this phase II data set, of all the things, numbers in terms of tolerability, the incidences were not different, but the key difference was discontinuation rate.
Right.
There was a 50% DC rate there with Danu versus 5% here. So what we've learned from Pfizer and these other programs is that you need to have flexibility in titration. And I think we also learned that the starting point of titration, especially in a 28-day data set, doesn't need to be the lowest dose that you're evaluating in SAD. Which is why I think we, in our study, put in a 14-day non-titrated MAD to be able to give us the flexibility to look at higher doses in non-titrated patients to see where do you start to see those on-target GIEs, 'cause those are where you know you're hitting the target meaningfully.
Yeah.
That allows us to start titration at a reasonably good point, to be able to then not be too far off your target max dose, and then be able to get to that target max dose as fast as possible, provided that you can get there and stay there without discontinuation. So our protocol also allows down titration, just like Structure, but we built in flexibility to make sure that down titration isn't 50%, but, you know, maybe 20%-30%, so that you're still maintaining dose intensity.
Interesting, and the Structure down titration is 50%?
I didn't know the details there, but we wanted to make sure that we didn't have to down titrate too much. So we have the flexibility as we go from cohort to cohort to put in intermediate dose levels to allow us to maintain dose intensity. So that's in, that's important to get to your maximal weight loss possible while minimizing discontinuation rates.
That's super interesting. Tell me what you learned from the Regor molecule, right? I think yours is, like, maybe one modification to danuglipron, maybe Regor's two.
Yeah.
There has to be some similarities. You're gonna have similar half-life, day coverage. When you look at the Regor data, I think at 28 days, they were showing about 3.5% placebo-adjusted weight loss and about 42% discontinuations.
Yep.
You know, but we went over this in our report, there's a lot about your compound we don't know. For example, how quickly you're hitting Tmax, your actual peak to trough in humans. Like, there's stuff here that we just didn't know.
Yeah.
Number one, why or why not is Regor maybe a comp to how 601 will behave? And what could be some differences, whether it's oral bioavailability, Tmax, that you could differentiate, despite also being a danuglipron derivative?
I mean, the main similarity is it's based off the danuglipron scaffold, just like we are.
Sure.
But there's still a lot of degrees of freedom and variables that you can tweak to, to get to that profile, which Amy alluded to, which is, you know, get to at least. If you can get to a Cmax. What we've learned from all these molecules is that efficacy you don't think is a function of how much fold above you are above agonism. There's a threshold of agonism above which whether you're tenfold above that-
Yeah
Twofold above that, you're gonna land at whatever weight loss. So that's. We've seen that, right? But we also know that if you're, what is linearly related or maybe non-linearly related to that threshold is safety. So if you are tenfold above that, you probably will get worse safety. So we think the ideal profile is a fast Tmax that gets you above that threshold, but then flat curve that stays above that threshold at least as long as waking hours, optimally during the full 24-hour period, and that could be tweaked based on the absorption profile of the molecule, not just the half-life.
That could be tweaked based on tissue half-life versus plasma half-life. There's plenty of other variables that you can use to get to that profile without necessarily just focusing on half-life, for example. So again, 601 was intended to solve for gaps in the pharmacology of danuglipron that we think can get us that profile. To your point, we haven't disclosed a lot of that, but, you know, I think what's gonna... The proof of this in the pudding, it's gonna be the clinical data.
Any chance you publish data going into that readout? I mean, it's like, it's coming out in a month.
Guided to second half clinical data, so I don't, I mean-
Okay, we're running out of time. That's fair. That's fair. Okay, I think we're out of time.
Okay.
I wanna put on the record, now we're saying go long into this readout 'cause after Structure... Thanks so much for everyone for joining us, sticking around at five.