All right. Well, good morning, everybody. It's a real pleasure to be here. My name is Mike Nally. I'm the CEO of Generate Biomedicines, and excited to share a little bit about the work we've been doing over the past eight years to try and usher in an era of programmable biology. Here's our forward-looking statement. Generate's been at the forefront of what we think is a pretty material shift in how we explore biology.
Around nine years ago, the technology industry started to open up opportunities using the transformer architecture to understand complex domains with a data-driven approach. Over the past eight years, Generate has been pursuing this with a distinct focus on making proteins programmable. What we've been able to show is that through an integrated dry and wet lab, our ability to explore biology continues to progress very, very quickly.
In the last year, we've really pushed hard, given the tools that the large language models are enabling, to create an agentic native infrastructure. What we're finding now is that we have a suite of molecular design capabilities that allow us to pursue very difficult therapeutic tasks, whether that be through internalization, through selectivity, through pH-dependent binding.
A whole range of very difficult-to-drug domains are starting to become druggable using protein-based therapeutics. We've been fortunate to work with some extraordinary collaborators. We did our first partnership with Amgen back in 2021. We extended that work with Novartis in 2024.
We're working with now, most recently, Zealand Pharma in the obesity space, using these sort of technologies. One of the things that excites me about the opportunity is that the technology is translating very, very quickly. Five and a half years ago, we had no therapeutic programs.
We were purely a technology company. Our lead asset entered phase III trials in under five years. In the chronic care space, that's pretty quick. What we're seeing is an ability not only to accelerate the time to therapeutic innovation, but also a very clear ability to drug domains that were almost impossible to drug in the past.
Our second program is looking at this with a protein that binds the toxin of MMAE-based ADCs. This was a really difficult toxin to drug, but we've been able to create a molecule that is specific to that toxin without interrupting the intact ADC.
Where in the first instance we've shown an ability to come up with better answers to existing therapeutics, what we're seeing in subsequent rounds of innovation is the ability to drug more undruggable domains. The pipeline starts with GB-0895. This is an anti-TSLP molecule.
The molecule has some exquisite characteristics, a half-life of 98 days, the binding affinity of 106 fM. We moved it into phase III trials in asthma in December of last year. We'll talk more about this in a little bit. We're also investigating it in COPD. We shared some data at the European Respiratory Society in September that showed our ability to effectively modulate the key biomarkers of FeNO, IL-5, and IL-13 out to six months in COPD.
GB-4362 is the molecule I mentioned a moment ago. This is the ADC toxin neutralizer. That molecule is rapidly enrolling in its phase I dose-finding portion of the trial. Once we complete the dose finding early next year, we will likely push into an expansion cohort to see whether or not we're able to stop the progression from grade 1 to grade 2 of neuropathy.
GB-5267 is an armored CAR T program that we are working with our partners at Roswell Park on. The team at Roswell Park are some of the best innovators in the cell therapy space. They had asked us, could we use our technology to optimize the binder and framework for CAR T therapies? What we have been able to find is probably the most potent CAR T to be investigated in solid tumors, and that is enrolling in a phase I trial in Buffalo, New York.
As I mentioned a moment ago, we have a range of different partnerships. We do think this is a new era. We do think these technologies will allow us to not only help enable the ability to design molecules that are transformative and out of the reach of current protein engineering approaches, but also change the underlying process of drug discovery.
I will share a little bit more about both sides of that equation in subsequent slides. One of the benefits of this is what we are seeing is a much bigger and cheaper early exploration of biology. We are seeing a time advantage, but also, importantly, the quality advantages are starting to be realized, where we are able to go after things that were previously undruggable. Our partnerships have been a great asset in this regard.
Working with teams at Amgen, working with teams at Novartis, some of the best protein engineers on the planet, when they cannot solve problems and we start to use our technology to be able to solve those domains, it tells us these are getting after undruggable pockets. That is one of the things I think that excites us most about the future of this programmable biology space.
At the core of the company is this integrated dry and wet lab. I think oftentimes when you hear AI drug discovery, people think of models. Certainly the models are important and will continue to push the frontiers of computational models, but the amount of innovation experimentally is as profound as it is computationally.
You see one of our scientists on the lower right-hand side of this chart, Kirsten. She is holding a test tube. In that test tube, we are doing 10 billion purposefully designed variants to drive an experiment around pH-dependent binding.
This allows you to find the gradient in really complex domains of biology to really fully leverage these computational advantages. We will push the frontiers of computational modeling, but we will also push the frontiers of experimental innovation. On the other side, on the left-hand side here, you see our scientist, Claudio.
He is working on our cryo-EM studio. We bought four cryo-EM microscopes about four years ago, and what Claudio has been pushing on is making cryo-EM a high-throughput instrument. All of these protein engineering models have been trained on the Protein Data Bank.
The Protein Data Bank has about 200,000 high-quality, high-resolution structures of proteins. What Claudio and the team are doing are actually solving structures where our protein folding models have low confidence. It is expanding the universe of effective exploration of proteins. Secondly, they are also creating new data types.
While most of the Protein Data Bank are static images of proteins, what Claudio and the team are doing by both experimental resolution, but then deconvoluting using machine learning practices, is actually moving it to a 3D image of the protein in its dynamic state.
A lmost solving the video of the protein interaction. That data, we think, will be very powerful in feeding our models into the future. We were fortunate around the turn of the year to really have some scientists that said, "A new way of discovery is upon us." We were fortunate to work with our partners at Anthropic around how we could leverage agentic science to explore biology in a different way.
This was largely as coding agents became more and more powerful. We asked ourselves, could we take on a very complex scientific domain using these sort of approaches? We put six scientists into a room. We gave them a budget of $250,000, asked them to explore and reason over a complex domain of biology.
They chose the intracellular signaling domain of in vivo CAR T as the domain that they wanted to reason over. That has been largely static over the past few years. We typically use the 4-1BB CD3 zeta signaling domain. Basically, could they distill it down to the essential elements and ultimately to create space to add additional cargo to the delivery vehicles in in vivo CAR T?
This team of six scientists over the past six months have now come up with answers that are 70% shorter and driving a logarithmic improvement in function. We're seeing with these sort of tools, with great scientists, with a rapid verification engine, the ability to pursue complex domains of biology in very, very short order. These are just the early results. So you see up on the left-hand side the traditional design.
You see these refactored leads over on the right-hand side, and what you're seeing is a left-hand shift in tumor cell killing. You're seeing an increase in cytokine release. You're seeing, not shown on this slide, a log improvement in surface expression.
You're seeing improvements in durability. Ultimately, what that's now leading to is a big shift down in B-cell depletion. We think this technology has application across five or six different very meaningful therapeutic domains, whether that be hematological malignancy, solid tumors, immune reset, and fibrosis.
The technology now is being pushed forward in those different domains. Over the next 12- 18 months, it's going to be a busy time at Generate. We think our lead program, GB-0895, continues to make great progress in asthma. We hope to fully enroll the study by the end of next year, the two studies.
The GB-4362, our MMAE neutralizer, we should have the phase I results from the dose-finding portion in the first part of next year. We will quickly go into an expansion cohort and should have that data by the end of next year, which should show proof of concept in our ability to reduce peripheral neuropathy. GB-5267, we should have early phase I results.
Then, based on the productivity of the platform, we'd expect to do one or two more strategic collaborations over the course of the next 12 months. So, a lot on the horizon for Generate. Maybe I'll just quickly dive into GB-0895, our lead asset. This is our best-in-class TSLP antibody. As I mentioned earlier, we started the phase III trial in December of last year.
We believe this gives us an advantage over all other long-acting anti-TSLPs, given that we started the phase III about nine months earlier than anyone else. The molecular properties, as I mentioned a moment ago, are best in class. The half-life at 98 days compares favorably. The binding affinity compares favorably. Ultimately, we believe the unmet medical need here is really, really profound.
The severe asthma market is only about 20% penetrated, and we believe a six-monthly alternative will be very well received by both patients and health systems around the world. This is the opportunity set.
We see 1.9 million biological eligible patients in asthma, an even bigger opportunity in COPD, and a vastly under-penetrated market in COPD. Beyond those first two indications, we see utility across a number of other domains. We obviously saw recent data from tezepelumab in EOE that looked very promising.
Chronic rhinosinusitis has shown proof of concept. We think this molecule offers a pipeline and a product-type potential. The early data on GB-0895 from the phase I study showed a 98-day half-life. The ability to modulate key biomarkers across FeNO, EOS, IL-5, and IL-13 were at the same or better levels than tezepelumab at all time points.
Our target saturation based on our PK/PD modeling showed 99.9% target engagement across that six-month period. The safety and ADA rates were unremarkable, just as we like to see those sort of things.
As I mentioned earlier, we were able to migrate directly from phase I in mild to moderate asthmatics into a phase III trial with regulatory support in severe asthma across a severe asthma population. These are the biomarker reductions, as I mentioned a moment ago. You can see how they compare to tezepelumab.
This is using one dose of GB-0895 at day one, measuring at six months, versus a monthly dose of tezepelumab measured at six months in a comparable population. The phase III studies are up and running. We had two replicate designs measuring a primary endpoint of asthma exacerbation rate at 52 weeks. The key secondary endpoint is the reduction in asthma exacerbation rate in the less than 300 EOS cohort.
The trial is powered for actually that key secondary endpoint at 94% power of a 40% reduction in that key secondary. That means we're 99% powered in the primary endpoint. The studies have now been approved in all regions. We're enrolling across 400 different sites. The early progress is suggestive that we'll continue to complete the trial by the end of next year.
A big question we oftentimes get is, what gave us the confidence to migrate from phase I to phase III? There were four primary reasons. First, the molecule binds the same epitope as tezepelumab. This was important actually in the early days of Generate. The real question we were getting consistently was, would computationally designed proteins be immunogenic?
The only way to answer that question was actually to bind a biologically de-risked site and observe the technology. What we saw now is that they do not induce a higher rate of ADAs than native proteins. In this study, what we were also able to show was that we saw those modulation of those key biomarkers in a comparable manner. Secondly, the dose selection is based on the target engagement data.
We modeled GB-0895 versus tezepelumab across a range of disease severity, a range of EOS levels. What we saw was that the target saturation was extraordinarily high across those different domains.
Importantly, in the TESI phase II study, tezepelumab ultimately made a pragmatic choice on dose. As you see here in the third column, what we saw was at 70 mg, 210 mg every month, and 280 mg every two weeks, so an eightfold difference in dose, you saw no statistically significant difference in asthma exacerbation rates.
Ultimately, a pragmatic choice on dose was made. The question we were asking was, if we were to run a similar phase II study and ultimately make a pragmatic choice on dose, it would cost us $50 million, take two years. We said we're better off making a pragmatic choice on dose based on those first two columns.
Lastly, but importantly, the regulatory precedent was established with GSK and depemokimab. Depemokimab is now approved in both the U.S. and Europe. They followed exactly the same strategy, going into mild to moderate asthmatics in phase I and then severe asthma in phase III. COPD is the next frontier.
We think TSLP as a mechanism has huge promise in COPD. If you look at the current COPD market, the two approved medicines, DUPIXENT and NUCALA, are only working on a very small sub-segment of COPD patients. These are the approved biologics, I should say.
Obviously, we saw recent data at ERS that's suggestive that IL-33 could be approved across all comers in COPD. If you look at the relevant market for GB-0895 based on the phase II TESI data, we think it has particular relevance in that greater than 150 EOS cohort.
If we think about how does it compare relative to the currently marketed alternatives, as well as TEZSPIRE in those cohorts, importantly, TEZSPIRE would be the only potential medicine approved in the less than 150 cohort with a 23% exacerbation rate reduction. In the 150 to 300s, the TEZSPIRE data is actually pretty compelling relative to what we saw in the phase III protocols for TEZSPIRE. We saw about a 34% reduction versus a 25% reduction with TEZSPIRE.
The greater than 300 EOS cohort, the TSLP numerically has shown kind of top end of the range efficacy as well. We think given the under-penetrated market in COPD, given the unmet medical need and the fact that COPD is one of the five greatest burdens of disease globally, this has a very meaningful role to play for patients with COPD.
We have fully enrolled in our phase Ib trial in COPD. We looked at two different doses. The results in this trial basically showed that we were able to modulate those key biomarkers in a robust fashion. We saw no difference between the 300 and 600 mg doses, which gives us confidence that the same dose that we have selected for the phase III trial in asthma is likely the right dose in COPD.
Maybe now I'll switch quickly to our next-generation oncology asset in GB-4362. GB-4362 is a truly unique molecule. The observation that led to this program was we saw a transcendent impact on overall survival with enfortumab vedotin coupled with KEYTRUDA in first-line urothelial cancer patients.
The challenge with the molecule has been roughly two-thirds of subjects experience neuropathy, and neuropathy is the dose-limiting side effect, which leads to a significant rate of discontinuations, a significant rate of down-dosing, or dose holidays.
The question we were asking ourselves was, could we find a molecule that would only bind the circulating toxin, remove that from the body without interrupting the intact ADC? GB-4362 achieved all of those sort of criteria. We think there is a huge unmet medical need. We think it has an opportunity to not only reduce neuropathy and other dose-limiting toxicities, but ultimately,
If that is shown to be true, has a potential to improve overall survival in the long term. The other point that is worth noting is that while we are studying it initially in combination with PADCEV plus KEYTRUDA, the utility will be germane across all MMAE-based ADCs.
The initial study will be in this single setting. The utility should work across all MMAE ADCs. The study that we are working on right now is a dose escalation study. The first cohort enrolled faster than we thought. The second cohort is well underway. We believe by the end of the year, we will be able to select the dose, then we will trigger this dose expansion cohort that is noted here,
Where we will look at our ability in the phase Ib cohort to stop the progression from grade 1 to grade 2 neuropathy, which we believe will be proof of concept. That is the extent of what we were hoping to share today. We have a couple minutes. I do not know if there are any questions from the audience. We are happy to take them now if so.
Cool. Well, thank you guys very much for your time and attention. Appreciate it.