Good afternoon, everyone. Thanks for joining us on day two of our Global Healthcare Conference from sunny Miami. I'm Andy Behrens, Senior Biotech Analyst at Leerink. I'm very happy to have Tango Therapeutics', Barbara Weber, President and CEO of the company. Thank you, Barbara, for joining us.
Thank you for having me.
Why don't we start with a brief overview of the company for those in the audience that may not be familiar with Tango?
Tango Therapeutics is a company that we started back in 2017 with the idea that the ability to build a really comprehensive genomics, functional genomics platform could yield a whole new generation of cancer targets in synthetic lethality and related fields, and that's really what we've done. We now have four molecules in the clinic representing three different programs and approaching data on some of them this year.
Great. Why don't we start with the PRMT5 inhibitors? There's two of them. What is the difference between the two molecules?
TNG908 is the first PRMT5 inhibitor of ours that went into the clinic. That is a molecule that has the unique feature of being blood-brain barrier penetrant, which is important in this space because glioblastoma is a common disease for which there really is almost no effective treatment, particularly once patients have relapsed. Frequency of MTAP deletions, which is required for activity of 908, is about 50% of GBM patients. The other molecule, TNG462, is a later molecule, came into the clinic about a year later and represents some enhancements to the profile of our PRMT5 inhibitor program, being more selective against MTAP-deleted cells, more potent, and having enhanced PK profile that gives it a very long half-life, which is nice because the flat PK profile inhibits the target very effectively. However, it's not blood-brain barrier penetrant.
So that's really the difference between the two.
Okay, and we've seen some data already for 908. Can you just give a summary? And I know, and kudos to you, you've stuck to the timelines and milestones very, very rigorously. We did see some data, and I think going into that data set, you told us we'd be at dose level 5 and not to expect responses. And can you just summarize what you showed us at that update? I think you were the first company to show clinical data for any of the PRMT5s.
That's right, and what we did was essentially proof of mechanism back in May of 2023, showing for the first time that you could differentially inhibit PRMT5 in normal cells versus MTAP-deleted cancer cells, and that was a very striking difference to be able to do that, and it's really the basis of why any of these molecules should, in fact, work. And then, as you know, later that year, in August and September, October, Amgen and Mirati showed their first clinical activity data with the proof of concept that it does work.
Okay, and what do we know about the differences between those three molecules?
These, the mechanism for the three molecules or four, all four molecules, the mechanism is the same. We don't know a lot about the Mirati molecule because they didn't disclose... Is that what you're talking about?
Yep.
They didn't disclose any PK data, but they clearly showed that it was efficacious over a number of dose ranges and reasonably well-tolerated. Amgen showed a very, to my mind, a very similar level of activity, probably less well-tolerated than the Mirati molecule, but holding up, I think, the thesis that we and others have had, that the activity of these PRMT5 inhibitors is more related to the MTAP deletion than to histology, because there are responses across a whole range of histologies for both of those molecules.
Right. Okay, and your data-
It's pretty unique, right? It doesn't always happen that way.
Right. And so your data set established that the whole MTA, requiring MTA to be present, could widen the therapeutic window.
Yes
... for the molecule. And then you also showed some downstream suppression of the pathway. What did you show there?
So SDMA is a marker of the immediate downstream methylation effect of PRMT5, and you can show that you can fully suppress the presence of, or the detection of SDMA in tumor cells that have been treated with these PRMT5 inhibitors. In fact, we haven't presented all the data yet, but we've shown now that it's very dose-dependent, at least with our molecules. That is a marker of saying that we have good target engagement, and then the PD effect is dose-dependent, and that we can get good, almost complete, if not complete, suppression of the enzyme shown by that PD marker.
Okay, and I know we've talked about this before. There are a number of first-generation PRMT5 inhibitors that had therapeutic window issues, but some investors point to the fact that they got pretty good suppression of that pathway, but yet it didn't translate into clinical activity. So what, you know, what do you think the reason for that is?
So the difference between the types of molecules is that the original PRMT5 inhibitors, of which there were four, don't have a mechanism that allows them to differentiate inhibition of PRMT5 between normal and cancer cells, whether they're MTA deleted or not. That's because they're what's called SAM cooperative. The new generation uses the generation of a MTA, which accumulates in cells in the presence of an MTAP deletion in order for them to work effectively, and it's really that difference that allowed us to say, "This should work." The concern that got raised is sort of a technical point, which is that we use only tumor SDMA staining with an IHC assay.
Essentially, all of the SDMA work, biomarker work, and PD biomarker work that was done initially with those first-generation PRMT5 inhibitors is done with serum SDMA, and what's in your serum is a mixture of what's coming from normal and tumor cells. So if you look across the data sets, you really can't get more than 80% reduction in plasma SDMA. That probably doesn't even mean anything because you really are sort of mushing together what's happening in normal and tumor cells. The really important piece of this is, though, to say that once you inhibit PRMT5 in a normal cell to a point where you're starting to actually affect the function of the enzyme, you get significant toxicity. It's an essential gene, so you really start to damage normal cells, and it's that therapeutic index that you lose, regardless of what you see with the detectability of SDMA.
Okay. And then, just to dive a little deeper into the MTA SAM relationship, they both bind to the same place, in the receptor pocket, and so when SAM is bound, these drugs don't work, and conversely, when MTA is bound, they work. Is that correct?
That's exactly right. And the way the biochemists, kind of their hair stands on end when I explain it this way, but it's how I think about it, that SAM is basically an activator for PRMT5, and MTA is basically an inhibitor for PRMT5. And in normal cells, let's say there's this much SAM, and there's tiny little bits of MTA. When a cell has an MTAP deletion, it can no longer metabolize MTA, and the levels go way up. So now, instead of having a lot of SAM and a little bit of MTA, you have the same amount of SAM and a lot of MTA. And so if you do exactly what you just said, make an inhibitor that only functions in the presence of MTA, you're strongly biasing towards inhibiting PRMT5 in those cells with high MTA.
Okay. Now, there's another company that's also involved in this pathway, IDEAYA, and they are making their MAT2A with the PRMT5, with the hypothesis that you're blocking that pathway in two locations through the messenger RNA, but also they believe that, you know, that it's better to block SAM production, and avoid the resistance. So what, you know, what have you found about the combination approach? Because obviously, as far as we know, you're not pursuing that.
Yeah, so that's a multi-part answer. Hopefully, I can get it all in the right order to make sense. But what MAT2A inhibitors do is reduce SAM levels, and they reduce it equally across cells, whether they have MTAP deletion or not. So again, I'll use my hands to sort of illustrate this. If you've got a tumor with an MTAP deletion and your SAM level's here, your MTA level's here, and presumably, the bigger the difference between the two, the better an MTA cooperative PRMT5 inhibitor will work. What MAT2A inhibitors do is slightly lower the SAM level, making that difference a little bit bigger. However, if you have a selective enough PRMT5 inhibitor that you're getting already full inhibition without altering SAM levels, it doesn't really add any benefit.
It's clear, true biologic synergy, but maybe the way we think of it is more as a dose-sparing effect. If you need to reduce the level of PRMT5 inhibitor you're giving, or if you can't get quite where you want to be because of tolerability issues, that might tip you over to having better efficacy. In our case, we can easily achieve clinical exposures that are well-tolerated, well past where we would see any effect, at least preclinically, with a MAT2A inhibitor. So from our perspective right now, we don't see that we'll need it, but we'll obviously keep an eye on the data.
Okay. And what about potential durability? I mean, obviously, we're talking about responding, but do you think that there could be a benefit to giving it in terms of durability? Because that's what IDEAYA has been pointing to.
I mean, it's possible that they have data that I'm not aware of, but in our hands, we do see upregulation of MAT2A enzyme in response to treating cells with PRMT5 inhibitors. That is common, but it happens quite quickly after exposure, I think probably within a week or so, and I think that is not something that we see that alters the long-term responsiveness in preclinical models. So again, we'll keep an eye on the data, but it's not an observation that we've seen. The one thing I would say is that MAT2A inhibitors themselves don't have enough intrinsic single-agent activity or big, big enough therapeutic index to be given by themselves. So if you have a MAT2A inhibitor, you definitely need a PRMT5 inhibitor for maximal activity. If you have a good PRMT5 inhibitor, you may not benefit from adding MAT2A.
In fact, you may add both complexity to your development plan and toxicity.
Fair enough. Okay. Why don't we talk about the much-anticipated updates that are coming this year? For... I think both, right? You're gonna give both, 462 and 908.
Our official guidance is on 908, and that probably won't change. But what we've said is that, because of a couple of different factors about the 462 trial, it's gone very fast, and that it has essentially caught up with 908 in terms of getting into the effective, predicted, preclinically predicted efficacious dose ranges. So what we'd like to be able to do, and I think there's some likelihood of being able to do that, is when we do the clinical update this year on 908, to also provide some data on 462 that will give everybody directionality on what we're doing with the two molecules. Because, you know, it's a very reasonable question. You've got two of them, what are you gonna do with them?
Right. Okay. And I think you've said, like, initially, you said you'll be at dose level 5, don't expect efficacy. The last time we spoke, your guidance has been, we should expect efficacy.
We're in the range where you should expect efficacy, yes.
Okay, so what-
We have very good tolerability. I would say that we certainly disclose that information, that for both 908 and 462, well into the active dose ranges, we have very good tolerability.
Okay. In GBM, which I guess is going to be the main, the patient—the way you enrolled patients, where you started enrolling in the 406 trial after the last cohort. So most of the patients at the higher doses with 908 are going to be GBM? Is it going to be all of them?
The way the— Yeah, and the way the trial is designed is that the actual dose escalation cohorts for 908 don't include GBM, but we backfill multiple cohorts, and we backfill with GBM patients. So there's, there will be a number of GBM patients at multiple dose levels.
Okay, will there be other tumor types at higher than dose level 5?
Yes, absolutely, because the GBM patients are not going into the dose escalation cohorts. The dose escalation cohorts themselves that have between 3 and 5 patients each, those are non-GBM.
Okay.
Then the backfill. Well, we can put up to 10, but usually there's not that many.
Okay. So in terms of the bar for success in GBM, what, what are you looking for in terms of competitive activity?
The bar for success in GBM is sort of tragically low. The response rates, even for drugs that are used, is pretty much zero, and their success is measured by PFS extensions of something like 2-4 months. So I think any, you know, even a small signal could potentially be an indication we're moving forward in development in that indication.
Okay. And just to be clear, this is going to be more of the slower-growing GBM or the aggressive GBM?
These are the aggressive GBMs. There's evidence that the low-grade gliomas also have MTAP deletions, and there's some evidence that the PRMT5 inhibitors that are brain-penetrant could work there as well. But our patients, the patients on our trial are all aggressive GBMs.
Okay. You did see, with 908 in the preclinical models of GBM, you saw apoptosis and?
Absolutely. In fact, you know, some of our most sensitive xenograft models are from GBM patients, from xenografts. I think the tricky thing, as I've come to understand after really digging into the historical data, is that the clinical course of GBM, once the relapses occur, is really short and difficult to manage because you have a rapidly growing tumor in an enclosed space, and you don't have a lot of room or time for that to regress. So what we're hearing very definitively from investigators, the leaders in the field, is that any signal in relapse refractory should really encourage people to think about moving more quickly to the post-surgical, post-temozolomide adjuvant regimen, where by definition, then you have to do a time-delimited study because those patients have had their tumors resected.
That's where they expect to be able to start to show benefit of a newly active drug.
Okay. Should we expect a reasonable amount of durability with this update?
I think we will certainly provide what we have on patients. It obviously just depends on when in time those patients have gone on study.
Okay.
But we intend to provide, you know, full swimmer plots on all the patients and the PK. So the data should all be available.
Okay. And, what are some of the challenges measuring responses in GBM? I know that there can be false responses or it's-
Yeah, I think part of the problem has to do, as I said, with measuring what's happening to a tumor in an enclosed space. It also has a lot of edema and often has fluid around it anyway. And so sometimes, if you look, for example, the bevacizumab data, people were very encouraged initially by thinking the response rates were 20%-50%. But in fact, what was really happening was that the edema was decreasing for short periods of time, so it looked like the tumors were getting smaller when they weren't. Sometimes you get a little bit of response, and you get swelling in the tumors, and it makes it look like they're growing when they're not.
So I think the investigators warned us in the beginning about how difficult it was to measure responses, actual objective responses in GBM, and I think that was a good warning.
Okay.
-based on historical data.
Right. Okay. I guess that makes sense that Avastin would decrease the edema, but do you think that the PRMT5 would also decrease the edema? Not well, not, I guess, decrease the edema, but not really cause tumor shrinkage.
There's nothing mechanistically to suggest that that would be true. Beyond that, I can't really comment.
Right. Okay. For 462, if you do give an update, it sounds like it'll be a small number of patients. What tumor types are likely to be in the cohort?
Yeah, it's actually interesting. I think with TNG908, we saw a much bigger range of histologies coming onto the trial. I think when we gave the update last May, we had 14 patients in 12 different histologies. We're seeing with TNG462, a lot more of what you might expect of the more common tumors, so pancreas, lung, glioblastoma, I mean, cholangiocarcinoma, things that are more common tumors.
Okay. And then, so it sounds like you, you guys are not combining with, with MAT2A, at least not at this point. Are there any other combination strategies you think might make sense in the PRMT5 and in certain tumor types?
I mean, I think as an oncologist, my bias has always been that we have to move to combinations to really expect maximum benefit for patients. Every tumor we've ever cured or made big dents in have come from combination therapies, regardless of what the single-agent activity of the components are. So for us, I think the two that we're most interested in are CDK4/6 inhibitors and RAS inhibitors. The CDK4/6 inhibitors, because basically every tumor that has an MTAP deletion also has a CDKN2A deletion, so there may be some synergy there, and there certainly is preclinically. And then the RAS inhibitors, because particularly in lung and pancreatic cancer, the percentage of patients with the tumors with MTAP deletion, the RAS mutation frequency is the same as in the general population of lung or pancreatic cancer patients, so it's a big number.
It's an orthogonal pathway, and it could also lead to significant benefit. So those are our two favorites at the moment.
Okay, what about potentially, combinations with a checkpoint inhibitor?
I mean, I think that could potentially make sense, specifically where moving to an earlier line of therapy makes that an important combination. Somebody asked us today something I hadn't really thought about. We only have a 28-day washout of any previous therapy, so any patient who'd been on a checkpoint inhibitor before coming on to one of our studies would have had several months of active PD-1 around, and we haven't seen any evidence of difficulty tox, you know, toxicity in that setting.
Mm-hmm.
But, you know, it's something that has to be addressed.
Right. Is there any reason to think that there'd be overlapping amount of toxicity that's plagued some of the combinations?
We don't have any evidence of that so far. As I said, our both molecules are very well tolerated.
Okay. Great, well, we definitely look forward to updates. A lot of interest in the whole area. Why don't we move to 260? Can you give us a description of that molecule and the mechanism of action?
Yeah, I think 260 is the, maybe the most novel program I've ever worked on. It came from the hypothesis that thinking about synthetic lethality a bit differently to say: Could tumor suppressor gene loss not just alter these cell-autonomous pathways that we're used to thinking about in cancer cells, but could it actually affect immune evasion and the ability to change how tumors respond to checkpoint inhibitors or other immune therapies? And as we got to doing some of those in vivo experiments to ask those questions, we identified STK11 as probably the top tumor suppressor gene that did, in fact, result in checkpoint inhibitor resistance in a very striking way in preclinical models. At the same time, Skoulidis and others published data on patients with lung cancer who had STK11 mutations in their tumor, showing that they didn't respond to checkpoint inhibitors.
So that was really exciting to us because almost simultaneously with our preclinical data, we had clinical validation. We then went on to look for drug targets that would reverse that effect in STK11 mutant tumors and showed that CoREST, the CoREST complex, which is really a subcomplex of HDAC1, had a very striking effect preclinically. And that's the molecule that we took into the clinic last summer, looking to see, in fact, if that could be replicated in patients.
Okay, the plan there is to go into patients that become resistant to checkpoint inhibitors or eventually combine in early lines?
So the way that molecule was discovered, in fact, the whole target was discovered, was to reverse checkpoint inhibitor resistance, and there's very little evidence that it should work on its own. It works to actually change the immune milieu, milieu of tumors and make them look like they're STK11 wild type. So the trial design is that patients get three weeks of TNG260 single agent. That was mandated by the FDA, despite our argument that it wasn't going to do anything. So they get three weeks of TNG260, and then they start a checkpoint inhibitor. So every patient on the trial already is getting the combination with pembrolizumab. When we backfill patients, they aren't required to have that run-in, so they get the combination from day one.
Okay, and it is going to be enriched for STK11?
Selected only for STK11, right. All patients have to have a tumor with an STK11 functional mutation, yes.
Okay, and that is... Is that mutation a primary mutation, or is that an acquired mutation?
Interesting question. It's usually a primary mutation, but now there's evidence, I think Mark Awad published this, that STK11 mutations can actually occur as secondary resistance mutations in patients who become resistant to checkpoint inhibitors.
Okay.
Might not be right about the author, so don't quote me on that, but there are data out there that they can occur as secondary resistance mechanisms.
Okay, and what percentage of patients are we talking about in lung cancer?
That have primary mutations, about 15% of lung cancer patients, non-small cell lung cancer patients, have primary STK11 mutations. And then in the universe of STK11 mutant tumors, that's about two-thirds of them. The other third is made up of breast, lung, ovarian, and carcinoma of unknown primary.
Okay, and is that being tested for currently?
It is being tested for. Many patients who are being screened for actionable mutations in lung cancer will have an STK11 mutation test at the same time, so we are picking up a lot of that, and surprisingly, also happening in some of the other tumor types as well.
Okay. And, does it coexist with other mutations?
It does. It sometimes coexist with a KEAP1 mutation, and the effect of that remains to be evaluated. We aren't excluding or requiring KEAP1 mutations in the study, so there will be a mutation, sort of representation of that that's consistent with what you'd see in the general cancer population. The other is RAS mutations, and there's a lot of conflicting data on whether the presence of a RAS mutation alters the effect of an STK11 mutation in lung cancer or not. So it turns out that about half of STK11-mutant tumors have RAS mutations, half don't. So again, we'll have a population of patients that will allow us to start to sort that out.
Okay, and I assume these patients are getting checkpoint inhibition and chemo?
They are. We don't require it because it's standard of care, though most of them have gotten it and already failed a checkpoint inhibitor.
Right. And how quickly do they fail?
... they usually fail within the first couple of months. So the response rate in STK11-mutant lung cancer patients, the response rate to checkpoint inhibitors is less than 10%, and the PFS is less than four months.
Okay, that's-
These are not patients who respond and then fail.
Right.
These are patients who are primarily resistant.
Right. Okay. Is there any correlation to PD-L1 status?
Yes, almost all STK11 mutant tumors are PD-L1 negative, and at least preclinically, treatment with TNG260 markedly upregulates PD-L1 expression on tumors.
Hmm. That's interesting. Okay, why don't we move to 348, USP1, and describe that program. We only have a few minutes left, but.
USP1 is a synthetic lethal partner to BRCA1, BRCA2, and other HRD-positive tumors. It's a similar concept to PARP, but it's a different DNA damage pathway repair. So if you think of BRCA mutations as double-strand break repair defects, PARP simplistically as single-strand repair defects, what USP1 does is inhibit translesion synthesis. So it's kind of like if you've got a four-legged stool, take out one with BRCA1, take out a second one with PARP, you take out a third one with USP1. And the preclinical data we have suggests that it works as a single agent in preclinical models, particularly in PARP-resistant models, and it also has strong synergy with PARP.
The PARP-resistant mutations, does it work with, like, reversion mutations?
I wouldn't expect it to work with a reversion mutation 'cause that is like turning a BRCA mutant tumor back into a BRCA wild-type tumor. So if you think of... I think there's not a great characterization of this out there, but if you make the assumption that PARP inhibitor resistance mutations are about a third BRCA reversions, a third PARP enzyme mutations that prevent binding of a PARP inhibitor, and a third, you know, mixture, who knows what? The two-thirds that are PARP enzymatic mutations or other will should work with USP1, or at least have the potential to respond to a USP1 inhibitor, whereas the BRCA reversions probably don't.
Okay. And what's the plans? I mean, that's a very complicated, obviously, area. It's changed, you know, what patients get, especially within regards to PARP inhibitors. So how do you see developing that asset?
Well, in the short term, I think we want to move quickly, both in the single agent and the combination. So we're doing the single agent dose escalation, but fairly, fairly quickly, we'll add in, in parallel, a dose escalation in combination with olaparib. In the longer run, I think you're right, it's a complex development situation, but we've mapped out pathways in essentially all of the big four PARP markets, which is breast, ovary, prostate, and pancreatic cancer.
Okay.
So there's still ways to thread that needle.
Okay. When might we see the first clinical data? Have you given guidance?
We haven't, and we just dosed the first patient in December of this, just a month or two ago, or three months ago now. So I think it won't be this year that we would expect to have USP1 inhibitor data.
Okay, now Roche also has USP1. What do we know about that drug, and what have they shown, and what do you think their strategy is gonna be?
Well, to answer this last question first, I don't know what they're gonna do. But the background on that molecule is that it actually came from KSQ. What we know about the KSQ molecule is that the dose escalation was done without patient selection for BRCA mutations or other HRD defects, but there was reportedly at least one PR in that dose escalation in a patient with a BRCA mutation in the tumor. And that in January of last year, they announced that they were moving into a PARP combination without having reached MTD because the drugs are very well tolerated. Shortly after that, we started hearing noise that there was exciting activity, and shortly after that, Roche licensed the drug, and everything went quiet.
Ah, okay. And what-
The only other thing we know is that in June of last year, the trial was stopped for about six months. It's now restarted. We think that was probably drug supply issue, but don't know for sure.
Okay, and do we know anything about the makeup of that drug?
Yes, we do, because KSQ's released their structure, we're about to release ours. And that all of the USP1 inhibitors are out that are out there, and you'll see actually there was a surprising number at the ACR abstracts that came out last week. There's actually now seven of them. They all are in some way related because they all bind to the same allosteric pocket in USP1.
Okay, great. Well, we look forward to learning about that one. There's obviously a lot of patients that are failing PARP inhibitors and big, big opportunity in a lot of diseases.
Absolutely.
Thank you, Barbara. Appreciate it.
Thank you.
Thank you, everyone.