GSK plc ($GSK)

It's my pleasure to introduce GSK. We've got Hesham, he's the Head of Oncology R&D at GSK. Exciting times just coming out of ASCO. I guess, Hesham, thank you for joining.

Thank you, Michael.

I guess to start, if you sort of think about GSK's Oncology R&D, what are the 2 or 3 decisions you're most proud of that shaped current portfolio? And is there anything you deliberately stopped doing to focus resources?

Yes. It's interesting because I think one of the first things that I think about just in terms of the journey that GSK has had coming back into oncology is really how the field continues to evolve. During one period of time, probably early on, there was a lot of interest in immuno-oncology and checkpoints as well too. And I think in many ways, if you look across the industry, at times, we may have actually over-indexed on checkpoints. And I think it's one of the areas that probably as I took on my role, I tried to really strike the right balance of thinking about precision oncology, a focus on what are probably much more well-validated platforms and molecules and a move more towards precision oncology and a bit away from checkpoint exploration. And I think probably, if anything, that was probably the first decision that I think is really important and helping shape at least the oncology portfolio at GSK. The second really was at a time when I think, maybe early on, there were certainly some mixed feelings about BLENREP, especially after the results of the DREAMM-3 study early on which was the first Phase III study that didn't meet its primary endpoint in a late-line multiple myeloma population. Again, mixed reviews, I think not everyone believed in the asset at the time. And we kept moving forward with DREAMM-7, DREAMM-8, really optimizing the design of those studies. And ultimately, they really led to successful outcomes, including DREAMM-7, which demonstrated statistically significant PFS and overall survival benefit and DREAMM-8 with a statistically significant improvement in PFS. The third really is around dostarlimab. I think at the time, if you look at certainly the landscape and opportunities to help differentiate its development program, given how competitive the PD-1, PD-L1 landscape was and identifying opportunities like first-line endometrial cancer, a space where the standard of care hadn't necessarily evolved, and for quite a lengthy period of time, about maybe 15 years or so with platinum-based chemotherapy. A space like, for example, dMMR/MSI-high, locally advanced rectal cancer, where we've seen, at least through the Memorial Sloan experience, 100% clinical complete response rate and initiating a pivotal development program in that space in GI malignancies. I think those were certainly also decisions that I feel like we're really the right ones to help us start to build the pipeline and then expand. And then probably, I would say like the last one is really business development. A lot of collaboration with Chris Sheldon's team on the BD side, that's really helped enable us access, not only the right technology platforms, but the right assets. And you look at, for example, acquisitions like Sierra Oncology with momelotinib or IDRx with velzatinib as well, too, as another means to complement the pipeline. So overall, I would say the decisions that I'm really proud of are the ones that, one, really helped us address key unmet needs for patients. Two, helped optimize the probability of success for GSK and its pipeline and rebalanced its risk profile. And then three, this focus on moving towards precision oncology and higher positive outcomes for our assets.

Excellent. And if we stick with BD for a minute and sort of try to link that to internal. So coming out of ASCO, we said it in our note, I think GSK maybe a little bit surprisingly came out as somewhat of a winner out of ASCO for us because we didn't quite expect the IDRx data. And I think what we've seen with other ADCs and PROC, your B7-H4 looks pretty good. Their licensed asset or externally sourced assets. So when you think about that internal versus external mix, what's the ideal mix?

Yes. Michael, it's a really interesting question. I would probably say like all big pharma companies are struggling and grappling with this. I don't know if there is a specific split that I would say is most optimal, right? But what I can say is, I don't think any company could sit here and say, hey, by the way, we are going to source innovation purely from an internal standpoint. Like it's not realistic. It's not possible. Each company has unique strength, unique capabilities. Like for us, for example, I think our antibody engineering, our small molecule engineering are focused on, for example, functional genomics, immunology, like those are all areas of expertise for GSK. But for us to say, hey, we're going to like -- have like the broadest expertise and the deepest expertise across all areas of platforms, it's not realistic. And I think we have to really think broadly, like so one, we have to think about where to source the innovation from, right? So historically, it's been much more like from the U.S., from Europe. Now, of course, we're seeing it emerge in Asia and certainly in China as well, too. And the bottom line is making sure that there is an openness towards what can be done externally and what can be accessed externally. Once you've prioritized like and clearly delineated what you're interested in, and I think to me like that's a like fundamental fact of like what's defining -- what's your strategy? What are your prioritized disease areas? What are your prioritized technologies or platforms? And what are the best out there? Do they exist internally? Do they exist externally? And can you benchmark them in their performance as well, too? And there's -- to be honest, there's an element of like what are you looking for? Are you looking for first-in-class? Are you looking for best-in-class, right? And then you can think about, is it developed internally and in-house? Or do you try to kind of access an iteration or a best-in-class asset externally as well, too. But at the end of the day, even now and as we think about discovery research organizations, I think the biopharma space has to really look at it and say, what is the right-sized research organization for this day and age, given where we are and given the open source innovation that exists as well, too, right? Do you have to have an internal presence for a research organization? Of course, you do, right? But at the same point in time, like, does it need to be like tremendously large. I think that's a separate question and very debatable as well, too.

Perfect. And could we maybe briefly try to layer into that sort of the technology angle, everybody talks about AI. There's digital pathology. There's tumor modeling. Like how does that feature into that mix?

Yes. A really important question, Michael. I mean for us, it really starts everything from discovery, right? And I think data and technology go hand in hand. So it's everything from accessing proprietary data sets from either external partners or collaborators or academic institutes to help us with antigen discovery to certainly medicinal chemistry design and how AI/ML can certainly help enable that and is embedded within GSK. Two, basically, and I think to me, this is fundamentally the area that where I try to spend a lot of time is prosecuting and better characterizing disease biology, right? I mean if you think about the challenge that we have in oncology, and I would argue probably across different disease areas is oftentimes, we're only looking at one singular piece of the pharmacology in the context of the biology of the disease without better understanding what's happening around it. And for us, we focus really right now on building what our foundational models, right? And these foundational models help us get a better understanding of the disease biology in terms of progression, resistance, relevance of different targets and combinations. And at the same point in time, they have to be powered by data. How are they powered by data? They're powered by data that we access externally. So multiomic data, so everything from, of course, clinical outcomes annotated data. to genomic, proteomic, transcriptomic deep immune profiling data, at a tumor microenvironment level, at a spatial level, right? And at the end of the day, the ability to be able to, again, use AI/ML to generate these biological models is really important. And you also have to keep them up to date because there's new data that's continually emerging as well. And then basically link them to, not only discovery, but also your translational strategy and hypotheses. For example, we acquired sometime in early 2025, a company called CELLphenomics, and this company focuses on developing novel in vitro model systems. What do they do? They're called organoids. And basically, they try to replicate what happens within a patient's tumor, but outside of the human body. And you're able to, not only do that through the lens of better, for example, replicating what happens in the patient's tumor through unique knowledge, insight and culturing but also be able to visualize it in 3D fashion and collect all of the data that's being developed in this living model system, right? And then see if you can replicate what happens in the organoid from the patient. And then follow the 2 in parallel to see is the drug predicting, for example, in the organoid what's happening in the patients. And can you predict the resistance mechanisms that are likely to occur, response, lack thereof, et cetera. So to me, I think fundamentally, like that's equally important. And then finally, maybe the third piece is we're talking about like predictive biomarkers and also diagnostics. Computational pathology is really critical. The ability to be able to once again start looking at biomarkers through a different digital lens and start to move the diagnostic discussions away from just simple testing to more probably slide and AI/ML assessment is also something that we're evaluating and assessing as well too.

Fantastic. So we start talking a little bit more about the ADCs. I guess that -- so to me, the SGO data -- at SGO, you show data for Mo-Rez that look really good and look very clean. So it seems to me that ADCs are still poorly understood, and they're complicated. There's the payload, there's the target, there's the chemistry. It looks like with your B7-H4 that could be quite differentiated. So if I take a step back, what do you think ADCs are still misunderstood? And if I could link the next question into that, how do you decide whether to go single agent or combo with ADCs?

Yes. It's interesting because I know like -- we think back to maybe like 10, 15 years ago. And the only thing that, like, with the first iteration of ADCs that we thought about was, hey, like the therapeutic index, that's the issue that -- like they seem to be really difficult. I think we've moved like beyond that. And you're absolutely right. Like, I think now, we're looking at ADCs through the lens of like there's 3 different components to them, and they all matter and they matter in different ways. So is it the antigen and not only its expression but also selectivity, of course? Because that matters in terms of on off-target effects. Is it the linker and its stability? And again, because of the toxicity, but also the efficacy as well, too. And also as we think about like things like bystander effect, right? And is it the payload? Where is it on the spectrum of like potency, toxicity, right? And its mechanism of action because you're also thinking about combinations. What types of synergies could there exist, right? Could it be combinations? I'm just saying like, with like, let's say, DNA damage response agents with, of course, checkpoints and others as well, too. Like does it -- does the payload causing, for example, let's say, DNA damage, elicit an immune response through new antigen or neoantigen generation. So all 3 components matter. I think we're still following a space where we're trying to optimize dosing, if I may say, because I think it becomes much more important to patients. At the end of the day, we're trying to basically deliver what our targeted chemotherapy to tumor cells. I think the biomarker space is still one that is, I would say, important to consider. I think we've moved beyond target expression. It is no longer a target expression game. It is actually one where we have to think about multivariate biomarkers that incorporate the targets, the linker and the payload and sensitivity to all 3. And I think there is an element of, we have to think about what's to come in the next 3 to 5 years as these patients start to progress on more commonly used payloads and we identify what the resistance mechanisms are, so novel payloads are going to be equally important as well, too.

Fantastic. And maybe this is unfair, but if I get you to rank B7-H4 versus B7-H3, I think my impression in the past has been B7-H3 was maybe just slightly more interesting asset to you with the data that we've seen, it seems like it's flipped a little bit. Is that a fair interpretation or not?

So I look at the entire portfolio, and I get these questions often like even internally, by the way, like what's your favorite asset? What are you most excited? I'm going to say both of them to me are equally important and for the following reasons. I think you look at B7-H4 and because of its expression profile, it is highly directed towards gynecologic malignancies, broad expression in ovarian, endometrial cancer, a lesser extent, maybe in cervical, a little bit in triple-negative breast and biliary tract cancers. But ovarian and endometrial, very clear, right? And the way that the ADC is designed as well too. It's a [ dara 6 ], it's cleavable linker, it's a Topo-1 payload and the activity that we've observed, the data that was presented at SGO you alluded to, 62% confirmed response rate in platinum-resistant ovarian cancer, 67% in second-line endometrial cancer, 5 Phase III pivotal studies launched. The drug is active irrespective of H4 expression. And the safety profile, I think, certainly, it's very consistent with that of the Topo-1 payload, just the heme tox, which is quite manageable because of how familiar medical oncologists are with it as well, too, the neutropenia and so forth. And so I think about it through that lens and the drug has got a really interesting efficacy profile. On the other hand, B7-H3 as well, too. I would say what's unique about it is this broad expression profile of the target. And not only is it expressed across thoracic malignancies, lung, head and neck, GI malignancy, CRC, GU, prostate, bladder and also sarcomas. But also, I think maybe this is something that's common to both of them. Not only our -- like not only is the target differentially expressed on tumors versus normal tissue. Both B7-H3 and B7-H4 are checkpoints. So there is this like dual role that they have, right, which I think is an area that we're focusing on better characterizing because I think it plays a really important role in how we think about combinations and combining with checkpoint inhibitors as well, too. And so we're learning a little bit more about that in small-cell lung cancer, where we have a breakthrough therapy designation, but also in osteosarcoma. And we've already initiated a Phase III study in second-line small-cell, but stay tuned. I'm going to say, Michael, for 2 things. One, several more Phase III studies that will be starting with B7-H3 over the next few months. Two, data, multiple data sets from our own development program and from the Hansoh development program that will actually be presented at a key scientific congress in the second half of this year.

Fantastic. One that I didn't have on the list, but I hope you don't mind, just thinking about the combinability. There's data that suggests that Jemperli better than Keytruda in trials. What you just said about the combinability is that ADC approach and ability to bring that out more that you actually have a checkpoint inhibitor that is actually very competitive, but it's just been difficult to do the trials in a conventional way now that the ADCs are there, you can -- that's more enabling to get that through?

Yes. I think it's an interesting question. And I think you're referring to the PERLA data, of course, which is looking at -- it was a Phase II study designed to really just benchmark the clinical activity of dostarlimab to pembro. So dostarlimab plus chemo versus pembro plus chemo in first-line non-small cell lung cancer and what we saw was, at least, again, the study wasn't formally powered for superiority or noninferiority for that matter, but it was just a relative comparison. And what we saw was the response rate were relatively similar and some positive trends for PFS and OS that favored dostarlimab. And I think for us, the key was basically benchmarking the activity of dostarlimab to the market leader in pembro. And I think that was really helpful context for us. How we think about the B7-H3 and B7-H4 programs? We're thinking about it through the B7-H4, B7-H3 development hat. And so we think where it makes most sense to combine with our already approved PD-1 inhibitor than -- or PD-L1, then we'll do that because, of course, there are clinical, regulatory and certainly commercial considerations. And the clinical and the regulatory ones involve, of course, investigator use, physician use. They involve having maybe data or not in a given tumor type. They involve what we call contribution of components from a regulatory standpoint and the need to generate that. And then there's the commercial kind of element as well, too, like I said, around uptake and so forth. And where there's opportunities for us to combine with dostarlimab and it makes sense to do that, then we're going to try to do that to help the dostarlimab program out as well, too, from its development strategy. Now bear in mind, there's 3 pivotal Phase III studies ongoing for dostarlimab. So not only did the Ruby Part I and Part II studies read out positive, but also we have ongoing Phase II pivotal study in locally advanced dMMR/MSI-high rectal cancer. We have a Phase III study in dMMR/MSI-high colon cancer, and we have a Phase III study in locally advanced head and neck post chemoradiotherapy as well. So those components remain ongoing, and I think we're continuing to identify ways to maximize the value of dostarlimab, which has become a blockbuster medicine as well, too, recently.

And then maybe IDRx, which now is a different name, which I always forget. I still call it IDRx.

Velzatinib.

So the data at ASCO looked really interesting certainly in the first-line setting. You've gone into Phase III head-to-head versus Gleevec. That's a punchy trial. It's going to take a little while for that to come through. Should I think about that as the main path to first-line? Or is there a fast forward that could get you there quicker?

Yes. No, I think it's an interesting question. And you're right, you saw the data at ASCO, very encouraging. In second-line confirmed response rate, 38%, unconfirmed 40%, median PFS about approximately 14 months. Well tolerated, I think especially as you look at the second-line data with other agents that have reported data recently as well, too. Like I think probably the Grade 3 event rate is about maybe 30%, 33%. Other agents, it's like 2 drugs, combination regimens, 72% Grade 3 event rate. It's difficult for this patient population that is typically, I would say, more used to well-tolerated, gentler at least treatments like imatinib. And I think for us, second-line is a starting point. So we've had a Phase III study start in late '25 head-to-head against Sutent in second-line patients. And then we're starting this first-line Phase III study head-to-head against imatinib and again, the ASCO data set, 61% confirmed response rate, 65% unconfirmed response rate. We're very encouraged by it, and we're going to continue to identify ways to accelerate that trial. Let me leave it at that.

Fair enough. I tried. Maybe in the last minute, 2 questions. One, a little while ago, I was sitting somewhere, I think it was in France, and you said look at the ADCs and luckily I did. If I were to ask you, in the next 12 to 18 months, what's the data set that I should be focusing or investors should be focusing at that sort of show that continuity of look at what we've done with the ADCs, look at what we've done with IDRx?

Yes. I'm going to say probably like for B7-H3, multiple data sets coming out over the next 6 to 12 months from our own development program from the Hansoh program, starting with the second half of this year at a key scientific congress. So stay tuned. So I think that's maybe the first one. I think the second is really more of like continued follow-up from the first-line cohort for velzatinib, which was presented at ASCO, which I think is going to be very helpful. And then the third, continued data maturity for the B7-H4 data across ovarian and endometrial as well, too. And then finally, maybe just one more point to put out there for BLENREP. Interesting data that was presented from DREAMM-9 at ASCO, looking quadruplets. But also interesting data that was presented a few months ago by Evangelos Terpos from Greece, there was a Phase I study that he was conducting called the TERPOS trial, as we call it, looking at the triplet of BLENREP in combination with RD. And the data from that actually was -- or has been presented across ASH, EHA, a few times. What was interesting is at a recent meeting called EMN which is a European myeloma network meeting. He presented some interesting data. It was actually some modeling work that he did. And the modeling work looks at the data from his study, the first-line study newly diagnosed the triplet regimen, which is in DREAMM-10, head-to-head against the MAIA regimen, daratumumab. And it looks at across different dose levels, 2.5, 1.9, 1.4, all 3 - just all 3 regimens. It projects what the median PFS could look like based on the extent to follow-up he has right now. And the modeling projects that across all 3 doses, the median PFS would be about 115 months across the 1.9 milligram dose specifically, which is the dose that is being used in DREAMM-10. It's about 100 months. So imagine, we know that quadruplet CD38-based quadruplets are delivering about maybe 90 months PFS. Well, imagine if you could deliver 100 months with a triplet with a better tolerability profile in these newly diagnosed transplant ineligible patient segments. So I'll leave you with that and say that we're going to continue, of course, as I alluded to earlier, to think about business development opportunities. We've done a lot on the research, the discovery side. And I think at the end of the day, for us, as we think about things that complement the portfolio, they have to fit into this framework of: one, address a key unmet need just based on the pharmacology and the mechanism of action. Two, align to key prioritized tumor types where we have strong capabilities. Three, there's preliminary clinical data that's indicative of a high probability of technical success. And then four, of course, bring the right value proposition for the organization and complementarity with existing assets.

Fantastic. And then final question, and maybe the answer to this is, hey, I'm the R&D guy. But if I look at guidance or ambition for the company, 2031 in revenues is GBP 40 billion. The GAAP to consensus is GBP 5 billion. And so far, the answer has been the biggest gap between GSK's expectations and the market is like 50%, that's BLENREP. If I look at what you've got with 7-H4 with IDRx, Jemperli data, these data coming through that there's a lot going on. It's -- I don't need that much out of BLENREP to get to a more ambitious outlook.

Yes. I mean, Michael, I love your view like that. So I think what we need is for everyone to embrace that, that's really the key. And you're right. We just talked about all this data across velzatinib, H3, H4, dostarlimab and let alone, of course, momelotinib, which is growing quite well. Yes, I think there's a lot of potential. I mean, there's a slide that Emma presented probably sometime in early 2025, that looks at, for example, the contributions that specialty medicines would be making to the long-term growth of the organization. And I think when you look at, for example, oncology, it's expected to be significantly contributing by 2031 as well, too, and 2034. So you're right, I think there is a reason to believe, I guess, maybe is the best way to put it.

Fantastic. Hesham Abdullah, thank you very much.

Thank you very much.