Hi, everyone. Thank you so much for being with us here today. My name is Courtney Breen . I am the U.S. Biopharma Analyst here at Bernstein. I am thrilled to be sharing the stage today with Kaivan Khavandi. He is the SVP and Global Head of Translational and Development Sciences at GSK. I know you have a focus and are responsible for respiratory, for hepatology, for immunology, inflammation. Can you give us a little bit of context about the scope of your role and the things that you focus on in the company? Because it sounds quite vast.
Yeah. Great. First of all, Courtney, thanks for having me. The role, I guess has two aspects. One is the end-to-end accountability for R&D for respiratory, immunology, and inflammation. So that's a fully integrated unit from discovery, target choice, through to translation, clinical development, asset leadership, through to product approval. The full life cycle of assets. As you described, it currently represents a lot of activity through to the late stage in respiratory medicine and hepatology.
But also, I think some very promising programs in what we're increasingly thinking of just as comorbid disease that's the consequence of chronic inflammatory pathology. The other hat is the Translational Development Sciences, which is a relatively recent organization, which integrated three key components of the organization cross-category, so oncology, specialty, and vaccines ID. That was Asia R&D for reasons that probably are obvious, and the opportunity that we see represented by China, in particular, perhaps the unrealized opportunity to partner with China for translational medicine.
The second is genetics, genomics, and single-cell technologies, which we see as one of the most high-confidence areas to be able to build an understanding of target trait pairings. The third being real-world evidence or development sciences, which integrates epidemiology and causal inference from large-scale routine healthcare data.
Fantastic. It's a broad scope, and I think the seat that you have covers and means that you must have to think about some of the topics that we've been discussing throughout the day today in terms of the role of China and innovation that's happening there, as well as AI and drug discovery and drug development, and these are critically important themes that I think keep popping up in so many of our conversations. I might squeeze in a couple of questions on those as well. But perhaps to start with your therapeutic area orientation. Respiratory has been a long-standing strength for GSK, and you have had leadership in the COPD space and among others. How do you see the franchise evolving from here, and where do you see the biggest opportunities for growth over the coming years?
Yeah. You're right to point out that GSK's got what I would really say is a unique heritage in respiratory medicine. But I think to pioneer is to be forward-looking and to continue to think about what's coming next rather than reflecting backwards. So, whilst that heritage and in health therapeutics and I guess the first wave of targeted biologics with the IL-5 franchise continues to be really important to the portfolio. Trelegy, Nucala are incredibly important products. I think what we're seeing is the maturity of how we think about asthma is not necessarily true of a disease like COPD. So in asthma, we know that there are very effective therapeutics, and part of the reason is that we understand that the majority of that disease process is driven by Th2-driven inflammation.
A number of now monoclonal antibodies have been developed that are effective. One of the key challenges is biopenetration and persistence. There's an unusual level of poor persistence to biologics in asthma. And in fact, the majority of patients come off short-acting biologics within a year, despite the fact that they are safe, well-tolerated, and the efficacy is unequivocal. That's led us, of course, to think about what could enable penetration and persistence.
The two things that we've identified is, one, a deliberate approach to the comorbid profile and the indication mix for any mechanism, and we're seeing that different biologics have a slightly different flavor of where they're effective across different disease processes, a number of which overlap in patients with asthma. The second is how do you reduce the burden of the management and administration? Obviously, that's led us to have confidence that an ultra-long-acting biologic, which almost completely relieves the patients of any burden of engaging in their management with just two injections a year, is going to be one of the key enablers.
In contrast, in COPD, it's more a case of understanding the underlying drivers of risk. It's a much more scientific mechanistic proposition that we see. The scale of the problem is vast in COPD, so it's 300 million patients affected globally, a life-threatening disease. As of today, only two approved advanced therapies with Dupixent and Nucala. Both of those are reserved for one segment of the disease. We really have invested time to understand what's driving different segments of COPD and then what's the rational approach to try and interrupt those disease processes with the right mechanism.
That led us to diversify alongside IL-5, which remains a very important mechanism in obstructive lung disease, to acquiring an ultra long-acting TSLP and also progressing our program for a long-acting IL-33, as well as a number of other approaches, including a PDE3/4 inhaled approach and also what could end up being the first oligonucleotide in respiratory medicine. We have a program that's currently in phase I-B, which is an siRNA.
Fantastic. How do you think about positioning some of these assets relative to one another? There's obviously a lot of potential innovation. There's also a lot of competitive pressure in some of these places, so can you just help contextualize the relative positioning that you're contemplating, at least at this point in the development process?
The simple framework that has been extrapolated from the asthma field is one that stratifies based on eosinophils as a surrogate of Th2- driven disease. You have these thresholds of patients with eosinophils over 300. An intermediate level of Th2- driven risk, which is eosinophils over 150, where we acknowledge there is going to be other pools of risk. Then eosinophils under 150, which is considered to be Th2 low.
What we understand, of course, is that IL-4/13, as an example, is restricted to those with eosinophils over 300. Nucala was able to generate an evidence base that showed clinically meaningful effects over 150, enhanced effects over 300. Again, the TSLP mechanism appears to be, although obviously being tested in pivotal studies now, effective in those with eosinophils over 150.
Now we are starting to gain a more sophisticated view. There are certain mechanisms that are effective in improving FEV1, lung functional spirometry. There are certain mechanisms that make the patient feel better, and those may be discordant with approaches that improve exacerbations or even improve survival. We know this from other settings. The analogy I always draw is to heart failure, where diuretics make you feel great. They do not make you live longer. Beta blockers make you live longer and do not improve how you feel. ACE inhib[crosstalk].
Make it worse sometimes.
Exactly. ACE inhibitors are the most prognostic. I think in COPD, we are going to start seeing this more sophisticated understanding of what mechanisms deliver what, and that, again, will likely lead to combination approaches as well. Of course, IL-33 now, and we predicted this, and we initiated programs, for example, testing IL-33 in non-CF bronchiectasis which is a disease process that's largely not Th2- driven. It's neutrophilic inflammation that drives that disease process. At the ERS International Congress that we've just come from last month, we've now seen that the IL-33 mechanism is effective in patients without elevated eosinophils or Th2 drivers at risk.
Fantastic. Continuing on the IL-33 mechanism, we've seen some mixed results for some of your competitors in this space. I'm thinking here of Sanofi and Regeneron. I'd love to get your thoughts on what did you learn from some of those studies, and what gives you confidence in the GSK995 asset relative to what we've seen play out elsewhere?
Yeah. It's been a storied history for IL-33. Look, I think there's probably a clear molecular differentiation between ST2 receptor monoclonal antibodies, which might limit the mechanism to one potential pathway and then everything else. People might be aware of the hypothesis that there's a redox dependent set of activity that IL-33 signals through. It starts off in its reduced form that can signal through the ST2 receptor, but there is also an oxidized form, which again is oxidized from the reduced form that can signal independently of ST2, potentially through the RAGE pathway. If you are an ST2 receptor blocker only, you may miss a potentially relevant signaling pathway through RAGE.
If you are a potent neutralizer of the ligand, which is what our program is, then you are going to neutralize the entirety of that reduced form, and it does not have a chance to become oxidized. The sort of black and white differentiation of our program is that it is long-acting, so every three months.
Yes.
But what we really invested time in, as you say, was to first of all, understand from competitor pharmacology and clinical data reading out where we are seeing evidence of a differential effect based on patient characteristics. And we have also got a universe of data internally in respiratory medicine, genetics, genomics, clinical data, as well as a phase II study with IL-33 to also integrate into those data sets and understand what is the correct population that is going to be responsive to IL-33.
Based on what I have just described, the astegolimab data from Roche at 15% annualized exacerbation reduction is understandable if the ST2 receptor was limited in terms of what it could do. The confusing data was itepekimab, the AERIFY studies from Sanofi. But we do also recognize that those studies were challenging, and one of the studies there was effectively almost There was very little exacerbation.
I think you have to separate the operating characteristics of a study to be able to demonstrate a treatment group difference from biological enrichment. Tozorakimab then took forward a phase III population where they mandated that patients answered yes to a CAT score domain, where they said, "Do you have a particular problem with cough from your COPD, and do you have a particular problem with productive mucus from that cough?" That was because in the phase II study, they saw an enhanced effect in patients that had imaging evidence of a mucus plug. That is a surrogate for neutrophilic inflammation.
We take those insights, astegolimab just published in The Lancet, pulled data from across their pivotal program. You can see, for example, in patients with more severe lung function decline, the mechanism appears less able to provide a benefit. That might not be surprising when you consider this is not a circulating cytokine. It is expressed on the lung epithelium. In patients with advanced lung dysfunction, emphysema, they have lost that cell type in the lung. Internally, the data set that really is proprietary to us was based on the observation that IL-33 is largely expressed in the lung epithelium, but the vascular endothelium.
IL-33 has clear evidence of causing endothelial dysfunction. We signaled a couple of months ago that our intent is to take forward a more conventional pivotal program to demonstrate the benefits on exacerbation reduction. For patients with COPD, they have an incredibly high comorbid burden, and what you really want is to keep them alive and out of hospital. You do not really want to be separating whether the reason they have decompensated and arrived in hospital was because of a degree of heart failure or their COPD or their metabolic profile.
We have been waiting for the right mechanism to be able to evaluate a study that says, "Actually, we are going to evaluate the benefit on CV hospitalizations, respiratory hospitalizations and mortality." That is the cardiopulmonary outcome study that we have shared. We plan to start next year as the third pivotal program there.
Fantastic. I know you have the TSLP, the long-acting TSLP, also in development. This is already a pretty clinically validated mechanism. What do you believe your asset needs to show to demonstrate differentiation going forward, particularly recognizing pipelines are of long-actings are coming in a few different places?
Yeah. Increasingly, we want to bag the benefits of the modality being ultra long-acting, and that is not trivial, as I described, because most patients come off short-acting biologics. I think that the long-acting components, if it replicated a Tezspire like data set, would be a highly successful product. We know that Tezspire is effective in asthma irrespective of eosinophil levels. It was demonstrated what I consider to probably be best in disease activity in chronic rhinosinusitis with nasal polyps with the WAYPOINT data.
COPD, not confirmed efficacy yet, but being studied, but the COURSE data suggested that it is efficacious in patients with eosinophils over 150 and 300. If you were able to replicate that and provide a product that patients could persist on, then I think that is going to be highly successful. However, the modality plus component comes in with, what do we understand from the mechanism? How can we bring our innovation to trial design to the benefit of the product profile? For example, for that program, the PERSIST ASTHMA study is recognizing that if you want to have long-term benefits of treatments and achieve clinical remission-type benefits, you really need to try and intervene earlier in the disease process.
Now, that makes sense, but from a trial point of view, if you do not have enough substrate of events to demonstrate a treatment group difference, that is very difficult. We have one pivotal study that is more classical for asthma with TSLP, which is those patients have exacerbated twice in the prior year. We have another study where the patients are only required to exacerbate once.
We've got some other criteria to make sure that we are able to simulate an event rate that's going to give us confidence that we can show a treatment group difference. That's a great example where we've borrowed from competitor data. We've got an understanding ourselves of how the mechanism links to the disease process, and then we've got this wealth of understanding around clinical trial design to be able to ultimately achieve a product profile that's going to allow for eligibility of patients at a less advanced stage of their disease. We're doing exactly the same thing with Extencha for COPD.
Absolutely. You're across those studies that you referenced, I think it's six in total that you're advancing 283 and two in phase III. The conviction required to initiate those all in parallel in some respects rather than sequentially, does that come purely from this very well-understood biology in some of these other pieces you're speaking to, or have you seen something internally that kind of makes you say, "All right. This is the right time to go big on this asset?
Yeah. All of the above, and we did have a phase II study in asthma, the NAZARE study. We're going to share those data next year, but they were very clear that we achieved the pharmacology necessary for twice-yearly dosing. That was obviously something that we weren't going to take a risk on, and we wanted to make sure we had empirical data in our hands. We did, of course, benefit again from our partners in China and Hungary, who also evaluated the same molecule in patients with nasal polyps.
Actually, we've got a wealth of data. We've got a GSK phase II confirming the pharmacology necessary. The data was very strong indeed, actually. We've got competitor data, we've got our China partner data, nasal polyps, and so I would describe that as an extremely high confidence set of pivotal study starts.
Great. Yeah, that makes a huge amount of sense to move forward with initiating all these studies. I want to continue in this space, but go beyond to pulmonary hypertension. You've got HS235, I believe, which is an activin trap in development. Can you talk a little bit about what gives you conviction in this mechanism? Why do you believe that this has the right balance of safety and efficacy to advance at this time point?
Yeah. Acknowledging that this program is relatively earlier than everything else we've spoken about, I think that that mechanism is extremely exciting. Again, it builds on the evidence base that's been generated by WINREVAIR, sotatercept, which I think everyone in the field acknowledges as being transformative to pulmonary arterial hypertension, Group 1 pulmonary hypertension. The first mechanism of many, but the first mechanism that's demonstrated remodeling of the pulmonary vasculature that provides evidence of disease modification. That product has been very successful post-launch, but has some very real liabilities.
Even actually as recently as yesterday, their label was updated to reflect the HYPERION study from an efficacy perspective. Marketing data that's led now to new language around GI bleeding, so pericardial effusions, telangiectasia, potentially GI bleeds. Because of the unmet need and because of the efficacy, the product is still successful and penetrating well in PAH. If you can overcome that, then you have a product which the benefit risk profile becomes pretty remarkable, really.
The proposition was, if you can be more selective and not inhibit the ligands that are signaling through to the bleeding and the adverse events, but preserve the efficacy, then that would be a very clear proposition. 35Pharma came with a real heritage in understanding medicine design that modulates the TGF beta superfamily of signaling. In that instance, recognized that activin A and B are key to inhibit, to modulate the efficacy. But BMP9 inhibition is probably leading to the bleeding risks. And one of the reasons that we recognize that is that if you have a genetic loss of function of BMP9, you literally develop the phenotype of the adverse events.
You develop a hereditary hemorrhagic telangiectasia. So this medicine was designed to spare BMP9, i nhibit activin A and B. But the really exciting bit on top of all of that is that GDF8 myostatin is a protein that we know leads to skeletal muscle dysfunction, insulin resistance, and inflammation. So if you're not dose limited like sotatercept is, you're able to unlock additional metabolic pharmacology and benefits and anti-inflammatory benefits. Which then, if you think about Group 2 pulmonary hypertension, which again, sotatercept has a signal of efficacy from pulmonary vascular resistance in the CADENCE study. But those patients have an average BMI of over 30.
A lot of them have metabolic perturbations, diabetes, comorbid risk. Imagine if you're able to treat the pulmonary vascular remodeling, offload the pressures, and treat the patient more holistically in terms of their metabolic risk. Then that's quite tantalizing. Of course, we had clinical data. The point of the deal, we had multiple ascending dose data up to 20 weeks that did empirically observe actually zero of the adverse events of concern.
Wow. That's really exciting because it is a patient population with so much unmet need and with kind of adverse events that they do have to navigate today. We began pivoting towards kind of the metabolic angle, and I think that you've got a MASH product in development. Can you share, sorry, some thoughts on the potential differentiation of your asset versus kind of some of the experimental other FGF2s in development and Rezdiffra? How do you think about the MASH subgroups relative to the opportunity for your asset?
Yeah. I will start off with what attracted us to acquire Boston Pharma and access that program was the observations of what was achieved across the class with FGF21 analogs. That was unequivocally transformative observations in cirrhotic F4 MASH, where something that was thought to be impossible just a few years ago was demonstrated that you could actually reverse the patients and shift them from cirrhotic disease to non-cirrhotic disease. That is enabled based on the clear benefits on fibrosis and reversing fibrosis. That is in a setting where none of the other mechanisms you mentioned have had any degree of efficacy.
For example, semaglutide has been tested in F4 cirrhotic MASH, and what was observed was an effect that favored the placebo arm over. GLP-1. I think FGF21 has a very clear, compelling, and important proposition in cirrhotic MASH, and we know that those patients have an incredibly poor outcome. In F2, F3 MASH, again, fibrosis improvement is the single most important factor that predicts improvements in liver-related outcomes. Actually, the mechanism is also a pretty potent metabolic modulator. So it actually was initially developed for severe hypertriglyceridemia. It does improve atherogenic lipids alongside best-in-class fibrosis improvement.
Collectively, what that means is you have highly competitive data for MASH resolution. You have best in disease data for fibrosis improvements. Importantly, a mechanism that is probably entirely additive to GLP-1 agonists or, as you say, resmetirom. Actually, there was a clinical data set that was really important for us when we did the deal, which was a study undertaken by Stephen Harrison in Oxford, I think, in collaboration with Phil Newsome, where they had patients that were on stable GLP-1s, then they were given an FGF21 analog. What you saw there was that the benefits of the FGF21 were completely additive on top of the GLP-1.
Then our product, we were the first major pharma to transact on a deal for an FGF21 agonist. There was a number of factors that related to that, the most simple being that it is a monthly dosing proposition in the portfolio that, as you have highlighted in this chat, has significant conviction around the benefits of long-acting injectables. There was other attributes. We saw the ability to scale manufacturing better with mammalian-like glycosylation, very low anti-drug antibodies observed. Really, it is the class that we found incredibly exciting.
Then finally, I described this completely unique proposition in cirrhotic MASH. That pathology is quite adjacent to alcohol-related liver disease, which is actually the main driver of liver transplants in developed countries. If it is effective in F4 cirrhosis, we think that de-risks a completely open area for a number of reasons, stigma and otherwise, a prevalent life-threatening disease in alcohol-related liver disease. The proposition then is, if you are a clinician, do you really want to be trying to weigh up the impossible task of, is this patient's disease driven mostly because of metabolic risk, mostly because of alcohol intake, which of course fluctuates as well?
Or do you want a product that is effective in steatotic liver disease, irrespective of etiology or stage? I think that proposition is why we think it could be an incredibly important product.
Fantastic. Super, super helpful in understanding that scope and ability to apply that asset across the range of liver issues there. As we've been discussing some of these assets, many of them have come in through external acquisitions, external innovation. I think, what, 35 Pharma, RAPT, Nuvalent even recently. How do you think about the role of M&A or licensing in building the pipeline from here? Perhaps if you can make some comments, given the breadth of your role and the contemplation of early research in China and how you're assessing that external versus internal trade-offs.
Yeah. I think that there's now an appreciation that the conventional model of you do some target finding, you operate in a discovery setting, five years later, you look to test whether that's relevant in humans. You then have a sort of transition to development product mindset. You might have not generated evidence that actually would be really important for trial design. You might have generated evidence and delayed the product for things that aren't material to how you're going to develop the product. That world has changed now. For us, it's really a case of how do you build the translational confidence that circumvents that?
So can you build an understanding of a mechanism in a way that allows you to apply it directly to the patient population study? In doing so, therefore, short-circuit the long runway of "discovery." A good example of that is what we've just discussed. IL-33 has been around for a while. The first-in-class players weren't the ones that were successful. It was trying to deconvolute the target-to-trait pairing, and sometimes just tweaking the population and eligibility criteria in a trial is enough to overcome the heterogeneity of that disease.
When you think about that in the context of BD, we don't want to be testing for the relevance of a mechanism in biology with a potential zero-value proposition. Our CEO, Luke Miels, has been very clear on that he wants to take forward de-risked biology and immediately test for differentiation and the product proposition rather than whether this matters at all in the disease process.
That doesn't necessarily mean the biology is being de-risked for everyone. It means, can GSK bring the best of our translational understanding and apply it to datasets, ideally to have a proprietary interpretation of that, and interpret it as de-risked and take it forward into late-stage development? HS235, which is the pulmonary hypertension program, is a good example of that. It's an early-stage program, relatively small data set in phase I, but enough to understand the relevance of the biology and the mechanism to the disease process, the potential differentiation proposition, and then a relatively tractable small study to be able to confirm that and advance into product development.
China's the same enabler, right? You don't have to undertake 10 years of discovery and biology activity. You can circumvent that. We take our translational data sets and understanding around human causal confidence, and then we're able to test that at a pace that is unmatched today and probably will remain so. We've established something that we've labeled the Global China Translational Hub. Where in partnership with companies like Hengrui, we're able to, again, tell our biology teams, "You've got the privilege of taking that understanding of human disease and applying it in a way where you can truncate the distance from insight into trial design into product application.
Yeah, absolutely. One of the other things that came to mind as you were speaking through those elements is where AI might be helping you and your teams in the most material ways. We've had some various players on the stage here today, some that focus on applying AI in the discovery space, some that focus on applying AI in the development space. Perhaps with a more operational orientation, but can be equally as important in some respects in terms of outcomes. You've got obviously a pretty long purview in the company from that discovery all the way through in some of these indications.
Can you just speak to what you're seeing at the moment in terms of the potential for application, the potential for real impact today versus the hopes for the future?
Yeah. So, what's established and now business as usual is the application of AI to efficiencies, and those efficiencies range from real-time intel on trial enrollment across the global footprint, integrating that with competitive intel and insights, automated development of regulatory documents, so some pretty high-stakes activities. We've been able to reduce cycle times from completing a study to developing first draft of modules for a submission to regulators based on automated development of those documents. Medicine design, again, I think every company has to be using AI for both small and large molecule design.
But the grail that I think everyone is focused on is can you use AI to reason over novel, I'm not going to call it discovery, and I'm not even going to call it target discovery, but target trait pairings. That's really our job is how do you understand whether your target is interrupting a disease process in a precise patient phenotype.
And that's where we do have examples, that I'm hoping will be substantiated shortly, where if you're able to integrate multimodal data sets, and you hear that phrase, but just to highlight what does that really mean. A multimodal data set could be, for example, how do you relate an observation from a cell phenotype to an imaging trait, to a clinical functional outcome, in a manner in which it informs how you match that target, as I say, to a patient population.
That's quite difficult to test through conventional a priori statistical methods. And we're seeing this more and more that, for example, you can learn something about, for example, what's happening in your lung alveolar macrophages based on what's happening in your bone marrow. Or you can understand a fibrotic pathology through a non-fibrotic cell type. And so, I label that as these latent spaces of information that have been untapped based on conventional methods. If you're able to host the data in the correct environment, and that brings lots of really important but boring things around the ontology of the data sets, and you can layer on top of that the appropriate tools. The tools are largely commoditized now.
And then query it with the right clinical and biological insights so that you're not unbiased because the universe of biology is too vast. But you're able to skew those queries and frankly curate the right data so that it has a higher yield and likelihood of surfacing something meaningful, then I think that gets us to that grail. And we've got an example of that. I shared with someone earlier today that t wo years ago, we had a meet the management event, and one such environment where we were able to integrate genetics, genomics, imaging data, clinical trial data, was in COPD.
Whilst I think some of our competitors were empirically testing bispecifics through trial and error. We systematically looked at every single combination in that multimodal data environment, AI enabled, and simulated which of the combinations is likely to be redundant, additive, or synergistic.
Okay.
We disclosed at the time, and we were the first to describe this combination, that TSLP/33 was the combination that would give you additivity and potentially synergy. At the time, that wasn't very well understood external to us why that could be the case, because the jury was out on IL-33. Now, after last month's data readout at the ERS, it becomes very clear and logical. I think now we're going to see a wave of activity around that combination approach. We applied that, of course, to a pipeline that's developing approach with that combination. We actually just started a phase II study where we're co-administering our two mono-specific antibodies.
Actually, we're using our presence in China to expedite that through to demonstration of additive efficacy.
Fantastic. One other kind of practical question about this, how much is this changing the way that your teams work or the way that you're thinking about your organization over the coming years? Are you needing to build things internally, or are you able to get solutions off the shelf from vendors that are enabling you to make some of these advances? I'm thinking all the way through development.
Yeah.
Not just[crosstalk].
Well, I think the magic happens, as I described, in the integration of those different enablers and domains. The good thing is GSK recognized very early the influence that AI was going to have, so we do have 150 AI engineers internally in a dedicated group in R&D who have built proprietary tools. We're not dependent necessarily on external frontier models. Again, we invested time to make sure that we were accessing data sets that were likely to have yield when applied to these technologies, and hosting them in the correct environments. We, again, were a company that made an early recognition that those had to be human data sets rather than animal models that don't translate.
Now, having the correct end-to-end mindset and increasingly physician scientists are leading discovery and translation through this kind of reverse translational approach where you take a clinical observation that might come from a large routine healthcare data. You qualify the causal association you might observed with high-confidence human causal instruments like genetics. You design the right pharmacological perturbation and clinical experiment to be able to test that. Then it seamlessly transitions to product development.
What does that mean for an organization design? It means this kind of siloed chronology of sequential, as I said earlier, discovery biology, going through the motions of IND enabling, start to think about trial design, start to think about a medicine profile. It's collapsed into basically a single effort that happens, frankly, at single time points. What that requires from a people point of view is the right experts that can think across that life cycle of an asset. The right data sets to be able to access and query, and the right tools to be able to enable those queries to inform actionable insights that inform product development.
Fantastic. It sounds like there's a lot of opportunity ahead, both with the pipeline you already have in hand and the work that you're doing to continue to build that over time. It sounds like there is some exciting innovation, not just on the kind of biology or pipeline, but innovation in the way that you're doing things as well. Thank you so much for your time today. It's been an absolute pleasure to have this conversation with you.
Great. Thank you, Courtney.
Thank you.