BeOne Medicines AG ($ONC)

Hello, everyone. Welcome to our 2026 investor event at ASCO. My name is Liza Heapes, and I'm the Senior Director of Investor Relations at BeOne Medicines. We are very excited to host this event, both in person in Chicago and online for our global attendees. Thank you all very much for joining us. This is truly an exciting time at BeOne, and we are thrilled to walk you through our program. Before we begin, a bit of housekeeping. I would like to remind all participants that during this presentation, we may make forward-looking statements regarding, among other things, the company's future prospects and business strategy. Actual results may differ materially from those indicated in the forward-looking statements. You can find more details in our filings with the SEC, Hong Kong Stock Exchange and Shanghai Stock Exchange. I'm delighted that joining us to present today are John Oyler, our Co-Founder, Chairman and CEO; Dr. Lai Wang, President and Global Head of R&D; Dr. Mark Lanasa, Chief Medical Officer for Solid Tumors, who will emcee the section with our esteemed KOL guests, give program overviews and development opportunities; Dr. Shom Goel, who came in from Peter Mac in Australia to share the latest on our CDK4i clinical development data that were presented earlier today. Mark will then run us through highlights from tomorrow's Mini-Oral on our B7-H4 program; and Dr. Hong Jae Chon, joining us from CHA University in South Korea, will share data from tomorrow's Mini-Oral on our GPC3 x 4-1BB before turning it back to Mark. At the conclusion of the slide presentation, our speakers will come up for Q&A. Now please let's welcome John Oyler, Co-Founder, Chairman and CEO.

Thanks so much, Liza. It's great to be here. I get to turn the page. So welcome. I really want to thank everyone for coming to Chicago and those of you that are joining us online. We certainly know how busy ASCO is in this time of year. And we appreciate you spending time with us and the continued partnership that you have in helping us work as one together to fight against cancer. I think that ASCO is always a defining moment of the year in oncology. And for this year, it's really an opportunity for us to step back and to show you something important about our company. It's the next phase of BeOne Medicines growth. Today is deliberately focused on work beyond hematology. If you haven't noticed, I'm very passionate and excited about the incredible hematology franchise that we've built. And sometimes, people think I'm so excited about it that I don't understand and appreciate everything that's happening in our solid tumor portfolio. And it's true, it's hard because there's so much going on there. And really, we now have multiple programs that are accelerating this year towards late-stage development, and we have a pipeline that's increasingly positioned to drive meaningful growth. And I do want to say to the BeOne Medicines' team and all of the investigators that are working with us that we are very committed to this space and really excited and proud, so proud of everything that's been accomplished in that area. With that said, I think we're a company that is very focused on trying to do great science. You'll see that in the pipeline today. We're a very science-oriented organization. We believe in scientific integrity and getting the best medicines to patients. I mean you have to develop the best medicines and we're committed to doing that. And I think also, as a company that's 16 years old, we've always believed that we need to find the best ways to do things. And we believe that's by working closely with the community, being one with the community to fight cancer. As part of that process, we've built this BeOne Superhighway that you hear about, and it's taken time for us to build that, it's not easy. It's like raising a child. But at 16, BeOne Medicines is now old enough to go drive full speed on the BeOne Superhighway. And you'll hear and see evidence of how Lai and his R&D team are doing that all the time today. So with that, I'll turn you over to my good friend and our wonderful head of R&D and business development, Lai Wang. Thank you.

Thank you, John. Thanks, everyone, for joining us tonight. BeOne is turning 16, as John pointed out. When I look back, we have gone through distinct chapters. The first chapter was our early years, about a decade of building from the ground up, that's when we designed and developed BRUKINSA and TEVIMBRA and laid the foundation for everything that follows. Then we enter the next chapter with a simple question in mind. Was that success repeatable? Or was that success a onetime achievement? Over the past 6 years, we have been working to answer that. On one hand, we built a highly productive discovery engine focused on oncology, powered by more than 1,200 research scientists and supported by a range of technology platforms. In parallel, we built something very unique on the development side, a clinical development superhighway, a global engine designed to move programs forward with speed, scale and efficiency. These are very -- 2 very distinct capabilities. By putting them together, they change what we can do. For the first time, it feels like we are not just relying on individual assets. We are building a system that can consistently generate and deliver innovation. This is why I see 2026 as the beginning of a new chapter, one where this model comes together and starts to deliver repeatedly for patients around the globe. I think everyone -- John said he's not going to touch on the heme, but I'm going to have one slide on heme. I think everyone is very familiar with our CLL franchise. The key point I would like to really highlight it takes 3 foundational assets to build on. It started with an anchor BRUKINSA, which set the standard care for the frontline CLL patients and comes with a combination product, BEQALZI remember the name, BEQALZI, also known as sonrotoclax, strengthening that backbone. Then finally, you need something in the relapsed/refractory setting, that's why when we designed the BTK degrader, now called tacabrutideg. Putting together, this gives us a complete framework across lines of therapy. And this is the key point. We now believe this is a model we can take beyond CLL into other tumor types. With that strategy in mind, we have been very deliberate in how we build our disease franchise. Our strategy starts with a focus selecting a small number of disease areas where we believe we can lead and then building the depth, not just the single assets. What enabled this approach is our in-house technology stack, spanning CDAC, novel payload ADCs, you're going to see tons of that from BeOne as well as cell-therapy and emerging platforms like T cell engagers. We're not bound to a specific target or a specific platform alone. We systematically match the right biology with the right modality to build a pipeline that is deep, durable and sustainable. The engine has clearly accelerated. From 2011 to 2020, the initial 10 years, we delivered 11 new molecular entities. Between 2021 and 2023, that's 3 years, we added another 10 NMEs, demonstrating consistent productivity and execution. The momentum stepped up again in the last 2 years with 18 NMEs across small molecules, CDAC, ADCs and bi and tri-specific antibodies, reflecting the maturation of our in-house platforms. Looking ahead, we expect to sustain this cadence of roughly 8 to 10 NMEs a year from 2026 and beyond. Innovation at BeOne is accelerating, systematic and built to scale. Historically, BeOne solid tumor pipeline was heavily weighted towards immuno-oncology. The CDK4i chapter marks the start of the new wave with more diversified mechanisms and a clear tumor type focus. This molecule entered the clinic just 2.5 years ago, 2.5 years ago. And the Phase III now is already on the way. CDK4 is not an exception. We initiated a potential pivotal Phase II study for our GPC3 x 4-1BB, you're also going to hear about today within 2 years entering the clinic. In addition, 3 more molecules have achieved clinical proof of concept and are now moving towards registration. All are on track to start their first pivotal trial within 2.5 years of entering the clinic, some of the less than 2 years. This is a clear inflection -- reflection of the quality of our science as well as our unmatched clinical execution capability. Today, we will share the clinical update for the first 3 assets shown on this slide with data on the remaining 2 to be shared with everyone in the second half of this year. I hope you will find the data as exciting as much as I do. With that, I will turn it over to Mark.

Thank you, Lai. Thank you all very much for joining us here tonight. Your time at ASCO is precious. We're very grateful. I had mentioned to a couple of you in conversations before starting that I'm authentically really happy, really excited to share the data. I feel like with many of you, I've been saying for a year or 2 years, we're really excited about what we're seeing with these programs. So it's great to finally be able to share the data, so you can understand why we've been and why I've been so excited. As you just heard from Lai, our strategy in solid tumors is straightforward and ambitious to create the next CLL-like franchises from 3 solid tumor areas, breast and gynecologic cancers, GI malignancies and lung cancers. Importantly, this is not just pipeline expansion, it's a deliberate transformation with each great -- with greater breadth across targets and modalities and meaningful depth within each disease area. This slide highlights the scale we've built with our industry-leading research and discovery organization. We now have more than 20 solid tumor assets targeting over 20 different oncogenic drivers and spanning multiple modalities. Importantly, this is not a narrow opportunity set. The diseases we're targeting represent roughly 2/3 of newly diagnosed solid tumor cases, both here in the U.S. and around the world. And by intent, each of our 3 priority areas now has multiple shots on goal across these mechanisms and disease areas of focus. We are not dependent on any single asset or outcome. As you move across this pipeline slide, what I would like to highlight is how we're matching biology with the right modality. The depth within each disease area is by design to enable in-portfolio combinations, which we believe are the key to maximizing patient benefit. Today, I'll focus on the programs where we have the most meaningful data here at ASCO and the clearest path toward registration. These include our selective CDK4 inhibitor, a B7-H4-targeting ADC, and a first-in-class GPC3 x 4-1BB targeting bispecific. In addition to the molecules we are highlighting today, in gastrointestinal cancers, we're advancing assets like CEA ADC and our PRMT5 inhibitor. In breast and gynecologic cancers, we have several next-generation combination opportunities to build upon CDK4 data that you will see today. And in lung, we're building a portfolio that reflects the complexity of that disease with multiple approaches targeted across small molecules, biologics and ADCs. Let's start with BGB-43395, our selective CDK4 inhibitor. We view this as a foundational asset within our breast cancer franchise with potential to redefine the standard of care in hormone receptor-positive disease. Even though CDK4/6 inhibitors are widely used in impactful medicines and are standard of care in first-line metastatic breast cancer, the class has well-recognized limitations. As you can see on this slide, neutropenia remains a major issue. It often forces dose reductions and can ultimately impact how much drug patients actually receive over time. There's also a meaningful management burden, frequent monitoring, clinic visits, which is not trivial for patients. And beyond tolerability, both primary and acquired resistance remain a significant challenge. So while the class has clearly transformed the treatment landscape, there is still a real opportunity to improve, particularly around safety, durability and overall treatment experience. Against that backdrop, this is a very large opportunity with global sales exceeding $15 billion. What we're aiming to do is improve on all 3 of those dimensions, efficacy, tolerability and combinability. The data with BGB-43395 is very encouraging and is consistent with our therapeutic hypothesis. We're observing response rates above 70% at our Phase III doses, which is consistent with the best reported outcomes within the CDK4 selective class. At the same time, we're seeing a differentiated safety profile, particularly with respect to hematologic toxicity. Importantly, our initial registration trial in first-line metastatic breast cancer is already underway. So taken together, we believe this positions the program very well, both as a potential next-generation CDK4 and as a backbone for combination approaches. The differentiation with 43395 is due to both potency and selectivity. As this slide illustrates, the goal is to potently inhibit CDK4 while minimizing CDK6 inhibition, which is what drives much of the hematologic toxicity with current agents. In a preclinical cellular assay, our molecule shows the highest potency against CDK4 coupled with the greatest selectivity relative to CDK6. And that's what underpins both the safety profile we're seeing and the flexibility for unique combinations. Today, we are honored to have Dr. Shom Goel, a key opinion leader in breast cancer presenting the CDK4 data update. Dr. Goel is an associate professor and clinician scientist at the University of Melbourne and the Peter MacCallum Cancer Centre. He serves as the global PI and translational PI for 4 randomized clinical trials in breast cancer and has over 60 publications, including recent works in high-impact journals such as Nature, Cancer Cell and Nature Cancer. He is the first author for the data you are about to hear, and we are excited to have Dr. Goel's insights on our selective CDK4 inhibitor and the potential impact on breast cancer treatment.

Good evening, everyone. Thank you so much, Mark, for the kind introduction. It's my pleasure to be here with you. And to share with you the data that we presented earlier today in a poster at the ASCO Annual Meeting. But let me start by orienting you to the clinical program in its full and where today's data set comes from within that program. As you can see on this slide, the study, BGB-43395-101 is the global first-in-human Phase I study that has progressed in this stepwise fashion. On the left is when the study began back in December 2023, as you heard, with global dose escalation, and that evaluated BGB-43395 both as a monotherapy and in combination. From there, the program has moved into dose optimization shown in the middle. And the data set being presented today is highlighted in red. First, the safety expansion cohort in first-line HR-positive HER2-negative CDK4/6 inhibitor-naive breast cancer using BGB-43395 with letrozole. And as shown here, this cohort includes roughly 20 patients per dose level across 3 dose levels. And then on the right, the program is continuing to generate data in this Phase Ib dose expansion phase. There are 40 patients enrolled in the first-line combination expansion which is the same first-line HR-positive HER2-negative CDK4/6 inhibitor-naive population with BGB-43395 plus letrozole. And importantly, we'll provide our first clinical evaluation of co-administration of this agent with food. The key point really from this slide is that this is a substantial global data set now. And so with that context, let me move to safety first, particularly heme tox, because hematologic tolerability is a key point and differentiator for the CDK4 selective approach. So as you can see on this slide, we're looking at adverse events across 3 dose arms. We've got 240 milligrams, 400 milligrams and 600 milligrams, all given twice daily with letrozole, and what stands out immediately is the overall low frequency of hematologic toxicity. If you focus first on neutropenia, the top row in the table which is usually often a dose-limiting toxicity for CDK4/6 inhibitors, you can see that grade 3 or higher events are quite rare, just one event at the 240-milligram and none at 400 milligrams. And interestingly and importantly, even the low-grade hematologic toxicity events are uncommon. You see a similar pattern with thrombocytopenia and with anemia where all grade events are infrequent and no grade 3 events are reported at the Phase III dose, which is the 400 milligrams twice daily dose. So I think what this table really demonstrates is a remarkably clean hematologic toxicity profile, which is consistent with what Mark showed you, being the high selectivity of this agent for CDK4 over CDK6. And clinically, that is highly meaningful because it suggests the potential to maintain dose intensity over time and to avoid the frequent heme tox, which led to dose -- leads to dose modifications that we often see with CDK4/6 inhibitors. So let me now turn to gastrointestinal tolerability, which is the other key aspect of the safety profile here. As you can see on this slide, we are comparing outcomes in patients treated with food versus those treated without food, focusing on diarrhea, nausea and vomiting. Starting with diarrhea, without food, the overall rate is relatively high at 94% though the majority of events are low-grade and manageable. Now while the number of patients evaluated with food co-administration is currently limited at 8, that rate of diarrhea drops meaningfully to about 50%. And importantly, what we're seeing there is now limited to Grade 1 events. And a similar pattern has been seen with nausea and with vomiting. For nausea, the rate decreases from about 61% without food to 38% with food, again, with a shift towards these lower-grade events. And for vomiting, you see a reduction from about 42% down to 13% with food. So I think the key takeaway from this slide is that the emerging data suggests that gastrointestinal toxicity from this agent is substantially mitigated through co-administration with food, reduces both the frequency of these events and also the severity of these events. Also, as noted on the left side here, there are additional mitigation strategies, which are currently being investigated, including the use of low-dose prophylactic antidiarrheals and also step-up dosing. And from a clinical perspective, the ability to actively manage these events is critical, particularly in a first-line setting where long-term tolerability is essential. So let me now turn to efficacy. On this slide, you'll see the waterfall plots of best tumor response for 2 different dose levels, 240 milligrams and 400 milligrams, again, both in combination with letrozole. And what stands out visually is the consistent tumor shrinkage across patients at both dose levels. In the 240-milligram cohort, you can see that the majority of patients are achieving meaningful reductions in tumor burden, many beyond the threshold for partial response. And in the 400-milligram cohort, that pattern is also quite consistent with most patients achieving deep responses. And if you look at the numbers on the right, in the 240-milligram group, the confirmed overall response rate is 68.4% with an unconfirmed response rate of 73.7%. The patient with an unconfirmed response is still on treatment. And in the 400-milligram cohort, the confirmed objective response rate is 63.2% with a similar unconfirmed response rate of 73.7%, again, with one patient with an unconfirmed response remaining on treatment. So I think if we put all that together, my impression is that this is a strong efficacy signal in the first-line for HR-positive HER2-negative disease, particularly when we consider it alongside the safety profile, which I've just reviewed with you. So with that information in hand, let me now walk you briefly through the pivotal study design. And I'm pleased to announce that the first patient for this study was randomized at Peter MacCallum Cancer Centre in Melbourne earlier today. So as you can see on this slide, this is a randomized Phase III study in first-line HR-positive/HER2-negative advanced or metastatic breast cancer and patients are being randomized 1:1 between BGB-43395 at 400 milligrams twice a day with letrozole or a standard CDK4/6 inhibitor with letrozole. The key eligibility criteria includes patients who received no prior systemic therapy for advanced or metastatic disease, ECOG performance status 0 or 1. And the study is powered to enroll approximately 1,056 patients. The primary endpoint is progression-free survival assessed by blinded independent central review and the secondary endpoints will include overall survival, objective response rate, duration of response, safety and patient-reported outcomes. So this design really is set to directly compare this selective CDK4 inhibitor against the current standard of care. And it's well positioned to test whether both the efficacy and the tolerability advantages that I've shown you will translate into a clinically meaningful benefit. So to summarize, first of all, as a selective CDK4 inhibitor, what stands out to me here is the promising antitumor activity that we're observing in first-line metastatic breast cancer. As I said, we are seeing an objective response rate of around 70%, which I believe provides a strong clinical rationale to continue advancing this program, particularly in the frontline setting. Second, on safety. I think what's encouraging is the overall favorable profile. We're seeing very infrequent low-grade hematologic toxicities and GI events that are generally manageable, especially when the drug is co-administered with food, which might help to mitigate those effects. And then third, looking ahead, we're actively planning to generate evidence in early-stage breast cancer. So that's an important next step as we think about expanding the impact of this molecule beyond the metastatic setting. And overall, taken together, we think this profile supports both continued development and a potentially meaningful role for this drug in breast cancer treatment. Thank you for your attention. With that, I'd now like to turn it back to Mark.

Thank you, Dr. Goel. We're so excited that the first patient has now been randomized at your site in our global Phase III clinical trial. I'd like to circle back and say one of the most important implications of the safety profile is what you see on this slide, which ultimately leads to combinability. Because we're not seeing the same level of hematologic toxicity, not only high-grade events, but also low events, low-grade events, we have the unique ability to combine CDK4 with other complementary mechanisms. That includes assets within our own pipeline, such as our KAT6A/B inhibitor, our CDK2 degrader and our BCL-2 inhibitor. So this is not just about a single opportunity. It becomes a platform for in-portfolio combinations and further growth of our breast cancer portfolio. With that, let's now turn to our B7-H4-targeting ADC. Our goal is to establish a new benchmark in ovarian cancer and other B7-H4-expressing tumors. This program has advanced very quickly, moving from first-in-human to Phase III readiness in about 2 years. We have observed a potentially best-in-class safety profile which is a critical attribute for patients with advanced cancers, but particularly in an early-line or maintenance setting. We're seeing very encouraging activity across ovarian, endometrial and triple-negative breast cancer and our emerging efficacy data enables development regardless of B7-H4 expression in ovarian cancer. Importantly, we are planning to initiate a Phase III trial in ovarian cancer before the end of the year. And we believe this combination of product attributes and timing positions us as the leading B7-H4-targeting ADC. Let me take a minute to walk you through the slide because this is where you can see how the differentiation of our B7-H4-targeting ADC comes together. As you can see here, there are a few components that drive the overall profile of the molecule, and each of these has been very intentionally designed. First, starting with the target itself, B7-H4 is highly expressed in a number of tumors, particularly ovarian, endometrial and breast cancers, but has very limited expression in normal tissue. That gives us a strong starting point from a targeting perspective that provides a safety profile that has no apparent target-mediated toxicities. Next is the linker, which is engineered to be stable in circulation, which helps reduce off-target toxicity but then cleavable once inside the tumor cell, allowing for payload release and bystander effect. The payload is a potent topoisomerase inhibitor with a DAR of 6. Finally, while I will not present our PK data this evening, we have observed a half-life of approximately 7 days, which supports every-3-week dosing. We presented a poster at this congress sharing detailed PK data and also explaining how adjusted ideal body weight dosing rather than total body weight dosing provides consistent exposure profile. So when you put all of these pieces together, the target selection, the antibody design, the linker stability, payload potency and consistent PK, what you get is a molecule that is designed to maximize tumor delivery while minimizing systemic exposure. Now I'll put the Phase I trial into context. As you can see on this slide, this is a first-in-human multi-arm study evaluating the B7-H4 ADC across a range of solid tumors with a particular focus on ovarian, breast and endometrial cancers. At this point, we've completed dose escalation and have moved into dose expansion, which will also serve as our dose optimization phase. What you'll see tonight and at the Mini-Oral presentation tomorrow are the data from patients treated at dose levels under consideration for future development with sufficient follow-up to begin assessing both response rates and durability. As you'll see, the data set is still maturing, but it's already providing a clear signal, both in terms of the level of activity and the overall tolerability profile. And importantly, this is the data set that's supporting our confidence to move in the Phase III with initiation plan before the end of this year. Let me spend a moment on safety because this is a key part of the story. Here, we are presenting data from 2 dose levels under consideration as our Phase III dose. The overall safety profile is very manageable and importantly, differentiated relative to what's been reported with other topo I conjugated ADCs with ongoing or plan development in early-line ovarian cancer. If you look across the adverse events, what stands out is that most events are low-grade, and the rates of Grade 3 and higher toxicity remain relatively limited during the period of follow-up. You can see the dose reductions in treatment interruptions are infrequent, which is important. It suggests that patients are generally able to stay on therapy and receive consistent exposure. And from a tolerability standpoint, we're not seeing a single dominant toxicity that is difficult to manage. Instead, it's a profile that appears predictable and clinically manageable. So when taken together, what this slide shows is a safety profile that we believe supports earlier-line use where tolerability becomes even more important. Turning to efficacy. This is where you can see the real strength of the signal. As you can see on this slide, in advanced or heavily pretreated ovarian cancer, we're observing an unconfirmed response rate above 50%, with the majority of patients experiencing a reduction in tumor size. Although I am not sharing the data regarding response rate by expression, and I encourage everyone to attend our Mini-Oral presentation tomorrow morning where those data will be shared, we are observing responses across the range of B7-H4 expression, which supports a broad development approach in ovarian cancer without the need for patient selection. Finally, let me detail how this data helps us to arrive at our planned Phase III development. As you can see, we're focusing on the first-line maintenance setting in ovarian cancer. This design is relatively straightforward, evaluating the B7-H4 ADC in combination with bevacizumab compared to bevacizumab maintenance alone in patients who have completed standard platinum chemotherapy-based first-line treatment. The maintenance paradigm is well established in ovarian cancer. There is an unmet need among patients who are not candidates for PARP inhibitor. The study primary endpoint will assess progression-free survival, along with key secondary endpoints that will help to characterize the overall benefit. And importantly, this setting allows us to introduce the drug earlier in the treatment course where we believe both the efficacy and tolerability profile can have the greatest impact. So as this slide outlines, the strategy is very focused, setting a clear, registrationally relevant setting and positioning the program for potential near-term path to approval. To conclude, I'll turn to our exciting GPC3 x 4-1BB-targeting bispecific. This is one of the programs that best reflects how we do science, not by following the crowd, but by being deliberate about what should be possible. The field has learned the hard way that by broadly activating 4-1BB, this can create unacceptable systemic effects. So the question for us was, can we bring 4-1BB's potency forward and control where and when it is engaged? This wasn't something that we just stumbled upon. It started with a clear hypothesis. If we could focus the immune activation only where the tumor is, we could aim for real efficacy without the price of systemic toxicity. Think of it less like turning up the volume on the entire immune system and more like designing a key that only turns the right lock. The goal is precision immune activation exactly where it's needed. And the reason we advance this program is simple. Patients with HCC run out of options very quickly. When you see how outcomes fall off after immunotherapy and tyrosine kinase inhibitors, it's not an academic problem, it's a human one, something we take profoundly personally on behalf of the people who need us. So what you'll see in the next few slides is the result of being willing to be bold to believe we could design something better and then letting the data tell the story and that brings me to where we are tonight. Let me step back and give you a quick overview of our GPC3 x 4-1BB program and how we're thinking about the overall strategy. At a high level, BGB-B2033 is a first-in-class bispecific and our goal is to establish a new standard of care for patients with HCC and other GPC3-expressing tumors. Starting with the clinical profile, what's really notable is the monotherapy activity. As you can see, we're observing response rates over 30%, which is roughly 3x what you'd expect with current standard of care in the second-line setting. Second, safety is a key and central part of the story here. We're seeing a highly favorable and differentiated safety profile, which opens the door to combination approaches and importantly allows us to move into earlier lines of therapy. This combination of favorable efficacy and safety is strongly differentiated from GPC3-targeting CAR Ts or ADCs and in turn enables a broader development plan with applicability to more patients. And then third, in terms of development strategy, there are really 3 parallel paths. First is the potential pivotal expansion in late-line HCC that is already initiated. The second is that we're planning a global randomized study in the post-IO setting. And third, we're continuing enrollment in a front-line HCC triplet, which will be important in supporting development in earlier lines over time. So overall, between the level of activity, the safety profile and very deliberate development strategy, we see a clear path to building a potentially meaningful franchise. Before getting into the data, let me take a minute to walk you through the slide because it frames the opportunity. Starting on the left, you can see the 5-year survival rates across a range of tumor types. What stands out is that liver cancer, hepatocellular carcinoma, or HCC, remains among the lowest. 5-year survival remains around 20% with only incremental improvement over time and is meaningfully below most other solid tumors. So despite some advances in therapy, outcomes remain quite poor. This is due to a combination of unique biology and patient comorbidities that are common in HCC. Now if you move to the right of the slide, you can see the incidence and stage distribution at diagnosis. In the United States, this is not a rare disease with over 30,000 new diagnoses per year. A significant proportion of patients are diagnosed at intermediate or advanced stages. And as a result, most will ultimately require systemic therapy. Globally, as shown in the pie chart, this is also a very large and growing disease with the highest incidence in China but meaningful patient populations across Europe and Japan in addition to the United States. So in summary, this slide highlights the combination of poor survival outcomes and substantial global disease burden that characterize HCC. And that's particularly important in the context of later-line disease where there is a desperate need for therapies that can drive meaningful responses. Now, allow me to describe the mechanism because this is critical to understanding the clinical data. As illustrated on this slide, one arm of the B2033 bispecific binds to GPC3 on tumor cells and serves as a targeting mechanism. Once the molecule is anchored to the tumor, the 4-1BB arm engages T cells, providing a co-stimulatory signal that enhances activation, proliferation and cytotoxic function. So the key here is that the 4-1BB activation is conditional. It's happening in the presence of the GPC3-positive tumor tissue rather than systemically. You'll also see on this slide a number of design features including elements to extend the half-life and reduce off-target toxicity. The most critical design here is 4-1BB binding at a unique epitope. Those design choices are important because they help create a balance between potency and safety, which has been a major challenge for this class historically. So the hypothesis is very straightforward. Targeted 4-1BB within the tumor should drive antitumor activity while maintaining a manageable safety profile. Now I'm very pleased to introduce Dr. Hong Jae Chon. Professor Chon is a leading clinician and investigator in hepatocellular carcinoma and currently serves as Professor of Medical Oncology at CHA University School of Medicine and Director of the Cancer Center at CHA Bundang Medical Center in Seoul, Korea. His work spans clinical trials and translational research, focusing on GI cancers, especially HCC. Professor Chon has been the global lead enroller in this Phase I study of B2033, and he will also present at the Mini-Oral session tomorrow morning. So we're very fortunate to have him with us here tonight. With that, let me hand it over to Professor Chon.

Thank you for the kind introduction. I'm Hong Jae Chon from CHA Bundang Medical Center. Thank you, Mark. And I would like to start by building on a point that Mark made earlier. As you saw, hepatocellular carcinoma remains a disease with both substantial global incidence and few long-term survivors, particularly when we compare with other solid tumors. Even with some improvement over time, 5-year survival remains relatively low and the majority of patients are diagnosed at intermediate or advanced stages. The treatment landscape for advanced HCC has evolved with the introduction of combination immunotherapy in the first-line setting. Today, most patients roughly 75% to 90% of patients are treated with IO-based therapy upfront. However, once patients progress, the options become quite limited. In second-line, we typically use TKIs such as sorafenib or lenvatinib. In selected patients, we can apply ramucirumab. But beyond that, as highlighted at the bottom, there is no clear standard of care after both IO and TKI therapy. So if you look at the right side of the slide, this is very well reflected in outcomes, response rate dropped significantly as patients move through lines of therapy from around 30% in the first-line setting and just less than 10% in second-line and very, very low level of -- in the third-line and beyond. So overall, this slide highlights a very clear gap, particularly in patients who progress after both immunotherapy and TKIs where effective treatment options remain limited. Let me walk you through the 101 study design. As you can see, this is the first-in-human Phase I trial evaluating BGB-B2033. Part 1 focuses on monotherapy and begins with dose escalation, doses ranging from 1 milligram to 1,000 milligrams. And after that, it expanded into safety expansion cohort in patients with second-line or later HCC. Then as shown in the middle of the slide, we moved into dose optimization with around 50 patients enrolled evaluating active dose level in the same second-line plus population. And on the right, the program is progressing into dose expansion with around 40 patients enrolled today and a planned cutoff of 120 patients, especially in the post-IO and post-TKI setting. In addition, as shown at the bottom, there is ongoing combination program in Part B evaluating BGB-B2033 with tislelizumab with and without bevacizumab. So overall, this study provides a structured and progressive approach moving from dose finding into dose expansion and towards a potential registrational pathway. Now turn to efficacy. On this slide, you can see the response rate across the different dose levels: 300, 600 and 1,000 milligrams. At 300 milligrams, the confirmed objective response rate was 29%. At 600 milligrams, the confirmed objective response rate increased to 36%. And at 1,000 milligrams, the confirmed response rate is 20%. Although I would note that this cohort was relatively small, and with short follow-up. So at that time, the unconfirmed response rate was around 30% already. So if you look across all of these dose levels, what you can see is the consistent signal of activity with response rate around or above 30% at the active dose levels. The 6-month duration of response is almost 37%, again, higher than both available therapies and other GPC3-targeting investigational molecules. Importantly, as highlighted on the right side, the patients are heavily pretreated with a median of 2 prior lines of therapy and the majority having received both immunotherapy and TKI. So we are seeing responses across a range of patient subgroups including both viral and nonviral etiology and both intrahepatic and extrahepatic disease. So overall, this slide demonstrates a meaningful and consistent efficacy signal in a population where historically, response rate has been quite low. Let me turn to safety. As you can see on the left side of the slide, overall safety profile appears very, very favorable. Treatment-related adverse events occurred in 48% of patients with only 8% being Grade 3. No Grade 4 or 5 were reported. Serious adverse events are also limited at under 5%, and treatment discontinuation due to adverse event low at just over 3%. There have been only one dose-limiting toxicity. If you look at the right side of the slide, which shows adverse events by type, what stands out is that most events are very low grade. Most common events include AST, ALT increase, but importantly, this did not require high-dose steroids. Other events were generally manageable and occurred at a relatively lower rate. So overall, this safety profile is favorable, particularly in the context of 4-1BB-based mechanism, where historically, toxicity has been a concern. I'd like to show 1 of my 2 -- 2 of my patient cases because I think this can help illustrate what we are seeing clinically. Let me start with first patient, and he is a 53-year-old male and diagnosed with HCC involving liver and lung and also positive for hepatitis B virus, and he had treated multiple prior treatments, including TACE and 3 prior lines of therapy, including immunotherapy and TKI. At baseline CT scan, you can see intrahepatic lesions and multiple lung metastases. But after 16 cycles of treatment, there was a substantial reduction in tumor burden and multiple lung metastases almost disappeared. In parallel with this, there was a dramatic tumor marker decrease from AFP level from over 2,500 down to the normal range. So this patient is heavily pretreated patient but achieved radiologic response and meaningful biomarker response. A second patient is 52-year-old male and had HCC and lung metastasis, so also had hepatitis B virus and he had underwent liver resection and had received 3 prior lines of systemic therapy, including immunotherapy and TKIs. At baseline CT scan, you can clearly see the lung metastasis. However, after 2 cycles of treatment, there was substantial tumor shrinkage and almost disappeared. And again, this was accompanied by the tumor marker decrease from over 8,000 down to the normal range. This patient also was heavily pretreated patient but showed a very rapid and deep response. So -- and both of them are still ongoing and around 1 year. And I also have around 10 patients who achieved a PR and all of them are still ongoing. So I expect this compound can have very, very good duration of response. So BGB-B2033-induced response rates are commonly observed in this treatment setting further supporting the activity signal seen in the broad data set. So what's most important is that in second-line or later HCC, we observed a response rate of above 30% with BGB-B2033 as a monotherapy. And when you look at that in context, the level of activity is comparable to what's typically seen in first-line combination immunotherapy and substantially higher than what we can see in later-line setting today. So you can see the comparison here in third-line the response rate is around just 2%, but in second-line, they are generally below 10%. Today, actually, in [indiscernible] trial in second-line, lenvatinib data showed just 5% of response rate. So this is important. So this is a setting where historically, activity has been quite limited. And what we are seeing here is a clear step-up in efficacy and this is important because this is a monotherapy. So the ability to achieve that level of response without combination provides flexibility we can think about development going forward. So now if you move to the right side of the slide, you can see the safety profile compared to traditional TKIs. Across key adverse events, hepatotoxicity, hand-foot syndrome, hypertension and diarrhea, what stands out is the low rate of high-grade toxicity with BGB-B2033. You will notice that Grade 3 and higher events are consistently lower than what's typically seen with lenvatinib or sorafenib. And even at the overall level, the rate of this toxicity appears more manageable. And it's important because TKIs, while active, sometimes can be difficult to tolerate and often require dose modifications. So when you look at this holistically, what this slide is showing is a combination of meaningful efficacy in a difficult setting along with a favorable and manageable safety profile and that combination is what gives us confidence, not just in the late-line opportunity, but also in the potential to move earlier and in combination setting over time. So -- and now, I will turn it back to Mark for company perspective.

Thank you so much, Professor Chon. From a development perspective, we're extremely excited by the data you presented here tonight and for tomorrow's oral presentation. This is a priority program, and we'll have the full power of our development superhighway behind it. And we recognize the patient need, the differentiation of our asset and the opportunity ahead as a first-in-class bispecific demonstrating an unprecedented combination of efficacy and safety. Now let me walk you through how we're thinking about development. Starting on the left, 2025 to '26, this is all about establishing the foundation. We're focused on late-line HCC where we can most efficiently demonstrate proof of concept and generate registrational data. As you can see here, we've already received U.S. Fast Track and Orphan designation, which we think underscores the potential of the program. At the same time, we've expanded our 101 study to support a potential post-IO post-TKI approval pathway. And importantly, planning is already underway to initiate a global randomized study in the second-line post-IO setting. This phase is about building the core clinical and regulatory footing. Then, as we move into '27 and '28, we will advance into earlier lines, deepening the opportunity. Here, we're starting to layer in combinations, including the tislelizumab plus bevacizumab cohort, which we expect will inform a frontline pivotal study planned for the second half of 2027. In parallel, we're broadening the scope even further looking into intermediate-stage HCC in even perioperative settings. This is where the asset begins to move earlier and into larger patient populations. And then looking further out into 2029 and beyond, this is where we expand. At this point, we're really thinking about portfolio-level combinations and how B2033 integrates across our broader pipeline. That includes next-generation IO approaches, additional in-portfolio combinations and also novel second-line strategies aimed at patients who progress on the frontline emerging standard of care. So taken together, this is not just a single development path. It's a long-term plan to systematically create patient impact and to build value in HCC, starting in late-line, moving earlier and ultimately expanding this foundational medicine into a multi-combination franchise built to address the needs of this substantial population of people around the world with HCC. So to conclude, allow me to put into context what you've seen today. We see this ASCO as an inflection point for our solid tumor pipeline. Across these 3 programs, we now have clear proof-of-concept data. And importantly, each of these programs supports the potential to initiate registrational studies by the end of this year in 3 different but common tumor types. Starting with CDK4, this is in the first-line HR-positive breast cancer, where we're seeing compelling response rates along with an improved safety profile, particularly when administered with food. Second, our B7-H4-targeting ADC in frontline ovarian cancer maintenance, here, we're seeing strong efficacy coupled with what we believe is a best-in-class safety profile. And third, the GPC3 x 4-1BB program in HCC, where we're seeing unprecedented single-agent activity with a safety profile that enables development in both later and earlier lines of therapy. Beyond those 3, we expect proof-of-concept data for 2 additional internally discovered molecules in the second half of this year, CEA ADC and PRMT5. So overall, what you're seeing is not just progress in a few isolated programs, but a broader portfolio that is systematically moving from early promise into proof of concept. And importantly, this is just the beginning with a number of additional early-stage assets advancing quickly toward that same inflection point. This wraps up our presentation for today. I'd again like to thank Professor Goel and Professor Chon for their time and their insights. And I'll hand it back to Liza.

Thanks, Mark. As our panelists come up to the stage, please, to participate in the Q&A, let me go through some general housekeeping items. [Operator Instructions] Now let's open up.

Thank you, Liza. We have multiple questions.

Kalpit Patel, Wolfe Research. I had a question for Dr. Chon for the GPC3 x 4-1BB. Very good data there. How are you thinking about the combinability with other TKIs given the LFT increases that we've seen across various different drugs there? So thoughts there. And then there were some other data with ADC and then the CAR T for the same target at ASCO this year. So I'm just curious as to what your thoughts are there.

Thank you for -- it's a good question. And First one is the GPC3 compound has a highly favorable safety profile. So I think there are a lot of chances to have combination strategy like atezolizumab and first-line setting combination and add-on or second-line TKI and add-on. Everything is possible, I think. And second one -- second question is also very important. There are so many GPC-targeting ADCs currently. But hepatocellular carcinoma patient is a little bit different from other solid tumors. Most of them have underlying liver disease. They are very vulnerable to liver toxicity. So the safety profile is very, very import. In this sense, I think CAR T can be a very easy way to access some targets. But if we have better alternative option, I don't think we have to insist on CAR T. Actually CAR T, even though in early-stage clinical trial, they showed very good response. But they also have high toxicity and have to go through some complicated processes like lymphodepletion and have to wait for a long time, which is not available for HCC patients, especially later line. So I think CAR-T treatment cannot be generally applied to general HCC patients. But in that sense, this bispecific antibody shows a highly favorable safety. So it is the best -- it is the most easily applicable approach for HCC patients. And ADC, actually, they also have some signal but still have high toxicity. As I mentioned, toxicity is very important in HCC patients.

And just quickly add on the sponsor's perspective about the later-line combination as you heard from Professor Chon, what we think is so exciting about this molecule is magnitude of efficacy with an extremely clean safety profile. While we believe we likely could combine TKI, we would also be giving up that really clean safety profile with that combination. So for now, we're prioritizing monotherapy development in later line, but I wouldn't want to [indiscernible] closing the door to that later-line combination.

Yaron Werber, TD Cowen. I have 2 questions and a really nice presentation. The first one on GPC3 is you're aiming to combine with the checkpoint inhibitor. Do you have any sense do you need to lower the dose potentially? Or is it safe enough you're going to get too much immune stimulation? It sounds like you're already doing that combo. And then second question on the CDK4. What's the powering of the KANDELA study for PFS? It sounds like you're starting with about 15%, 20% higher response rate. How are you thinking about durability of the PFS superiority?

Dr. Goel, maybe so Yaron, we generally -- thank you very much for the question. We generally would not disclose directly the overall powering of the study, but maybe you could talk about what would be viewed as clinically meaningful in the frontline setting in a comparative study with the existing...

Yes. Sure. Thanks for the question. So as you know, we sort of have to have a benchmark in mind for what our standard first-line therapies can currently achieve and based on first-line trials that led to approval of the existing drugs, we have figures in our head that sit around about 2 years, right? That's -- and we also look towards more contemporary data sets to help refine and inform those estimates. As you heard, the primary outcome for this study is progression-free survival. So the question is what is a meaningful improvement in progression-free survival. Yaron, my honest opinion, and this is not the way the things play out, and I understand this, but even if theoretically, one had a drug that was equivalently active, but better experience for patients, that alone, I think, would be a good thing. But here, of course, we're looking for that improved tolerability and improve potency to actually translate also into an improvement in PFS. I would say, in my personal opinion, an improvement in PFS in the order of months would be clinically very meaningful for patients. I can't really specify this many months or that many months, but I would say something in the 4 to 6-month range would be really meaningful in my professional opinion.

To your first question about combinability with PD-1, it's a very fair question, one that we thought about in terms of there is a risk of mechanistic [indiscernible] toxicity. We mentioned at earnings that, that frontline cohort is well underway. It is early days. Professor Chon, I don't know if you have any clinical reflections on how that's going [indiscernible] in that cohort.

Actually, we have already 10 patients, but there is no toxicity issue. But I think the first-line combination of nowadays first-line setting most wisely used first-rate regimens, atezolizumab and bevacizumab combination. But nowadays, we are evaluating tislelizumab, bevacizumab and GPC3 as first line. But I think that is a very smart strategy because we have data, patient who doesn't respond to atezolizumab, they usually show the high GPC3 expression. But maybe this triple combo can compensate this kind of weakness of Atezolizumab with triple combo, I think.

Maybe I will provide qualitative statements [indiscernible] last 2 years and so we're off to a good start [indiscernible]

Maybe I'll just add a comment [indiscernible] mechanistically. This molecule was really specifically designed to be target immunotherapy with the GP end of it, but as well as on the 4-1BB side of it. We're not disclosing exactly how we design molecule in 4-1BB binding, but I can tell you it's very unique. And in a way, we're leading to tumor-local activation. So with that, also now we're having more than 50 patients with the triplet, we feel very comfortable with the combinations.

Ren Benjamin from Citizens. Congrats on all the data. Can you talk -- just sticking with the CDK4 inhibitor, can you talk a little bit about the impact of food on efficacy? Have you seen any sort of betterment in efficacy given the tolerability profile. Why you chose 400 milligrams as the dose versus 240? I guess, for Lai, given the superhighway and the speed at which you do clinical trials, can you just extrapolate based on the enrollment rates for the Phase I portion, how long a Phase III must take to read out?

So we are targeting, shall we say, aggressive enrollment time lines, given the overall positioning within the space and the fact that [indiscernible] has largely completed, if not fully completed their enrollment. That said, this is a global enrollment. We have had a team that's had a large investigator meetings here at ASCO. There's high enthusiasm for the study. So if you're looking to enroll these 1,000 patients on a globally allocated way that will meet global health authorities as quickly as possible. A qualitative statement. Shom, I don't know if you have any comments?

Yes. So in respect to your question about whether concomitant administration with food affects efficacy, I think it's fair to say that with the numbers being what they are, one can't really make a meaningful comment there. What we can say is that in the limited number of patients that we've looked at, there's no significant difference in PK as to whether the drug is co-administered with food or not. And I think that's a very important point to note.

We completed the food effect study, and there's no impact on the PK. So with no impact on PK, we do anticipate this should not have impact on the efficacy side of it.

Strong efficacy, we see not only at the selected dose of 400 milligrams, but also at the lower dose of 240 milligrams that gives us confidence there's not going to be any impact to efficacy.

Maybe just one more comment on the superhighway. Thanks for highlighting that. Certainly, we are very proud of what we have achieved with the superhighway. And hopefully, today, showing you is not just one case. Actually the all 3 programs we are delivering with the same type of discipline and the speed and the efficiency. The CDK4 program is only in the clinic in 2.5 years. The Phase III, the first patient has been randomized today. Actually, we -- not just for these events, we were really pushing and trying to get going. And in terms of how long, we need to do better than Pfizer. So in terms of the GPC3 x 4-1BB programs, that's a program now already enrolled over 200 patients, and that's only in the clinic less than 2 years. It's remarkable for HCC patients. And then in terms of B7-H4 ADC programs, similarly, in 2 years' time period, we're already ready for Phase III. We probably are starting the Phase III within 2.5 years of the molecule entering clinic.

Yigal Nochomovitz from Citi. On the B7-H4 program, I noticed that it seemed like on the waterfall plot, everyone was an ovarian cancer patient, but I believe the study also enrolled breast and endometrial. I was just curious if you didn't show that data or those weren't -- those subtypes weren't enrolled. And then just going back to the question Ren was asking on the dose for CDK4, I think you had a slightly higher confirmed ORR in the 240, and I believe I thought I saw 1 CR, but nonetheless, you're going with the 400. So if you could just expand on your thinking as to why the 400...

Yes. So for B7-H4, tonight's presentation really focused on ovarian cancer because that's where we're going to have our initial Phase III study start. But at tomorrow's Mini-Oral, we will disclose data in the additional tumor types. The overall design of that study, dose escalation and expansion and now these tumor type expansions, what we're sharing is the dose escalation and safety expansion or backfill cohorts, there's some, shall we say, enrichment of ovarian from the investigators. It was looking good in ovarian, so they enrolled a lot of patients with ovarian. There are fewer patients with triple-negative breast cancer and endometrial cancer, but we'll share those data at the Mini-Oral tomorrow. We're excited about those data, and we leave that there's a development path in those additional indications that's open to us. Maybe if I can start for the CDK4 dose rationale, I appreciate the point that you're making that with roughly 20 patients enrolled across those 2 cohorts, the response rates are essentially the same. We, of course, on the back end, do detailed population PK analyses, exposure response analyses. And we felt confident that 400 milligrams provided the best balance of efficacy and safety the highest probability of success most likely for patient benefit. We shared all that data with the FDA, and that data selection was endorsed by FDA during our Phase III start-up.

Jess Fye, JPMorgan. For the GPC3 x 4-1BB bispecific how many patients of single-arm data do you think you would need to file for accelerated approval post-IO, post-TKI and how many of those would need to be Western patients? And then for B7-H4, what's the proportion of ovarian patients who meet that criteria for the maintenance trial you described like there's still disease or PR after frontline bevacizumab plus chemo and PARP ineligible?

Thank you, Jess. So thankfully, for ovarian cancer, the majority of patients will achieve PR, CR stable disease after the frontline therapy, roughly 90%. So thankfully, there are a few patients who are refractory from the beginning. The distribution of patients who are eligible for PARP inhibitor versus not the proportion of patients who are HRD versus HRP is roughly 40-60, something like that. So we do think that it's a substantial proportion of the frontline ovarian cancer population who would be eligible for the study. How many patients? Oh, yes. So this is speculative. As we talked about at our most recent earnings call, we have expanded our 101 study. We've added a cohort to increase the sample size of late-line monotherapy and that double pretreated patient population to 120 patients. However, the global regulatory consultations related to that patient number are ongoing. That said, in the context of having a response rate of greater than 30% with a highly favorable safety profile, standard of care of only 5%. We're optimistic going into those conversations, but that's our rationale. But we haven't had those conversations yet or we're having them right now.

Etzer Darout, Barclays. A couple of maybe strategy questions. One on B7-H4 ADC that conversation with physicians that have talked about the need or importance for a biomarker strategy in various cancers. And you've noted a couple of times around sort of maybe an all-comers approach. So I wanted to have your thinking around there as you sort of progress these different tumor types into pivotal studies? And then secondly, on the 4-1BB program, and we've seen other bispecific pairings where there's just been failures. But given sort of the level of activity you're seeing with your GPC3 program, I'm curious about what is it about GPC3 where you're seeing this level of activity? And are there other bispecific [indiscernible] that you view as more compelling given what you're seeing with the GPC3 x 4-1BB?

Yes. So Lai, would you like to talk about additional iterations on the 4-1BB?

Sure. In terms of the 4-1BB again, this is uniquely designed 4-1BB binder. The reason we started with GPC3 and part of people will think that's very crazy because initially, the 4-1BB monoclonal antibody certainly had a liver toxicity. We believe that GPC3 is highly expressed in the cancer cell, but not in the normal liver cell. That gave us the comfort to move forward with GPC3 X 4-1BB. We actually have a [indiscernible] program in our pipeline now to targeting various different type of tumors. We believe this MOA should not only help in HCC, with the platform we have set up, we are looking forward to bringing the other molecule into the clinic in the upcoming -- next few months to a year. And that will be -- this is actually -- I'm very excited about this platform. I think this potentially can be applicable to many different type of tumors. And also just to add to that, a lot of people work on the CD3 T-cell engagers for solid tumor. However, unfortunately, for solid tumor, not that many clean TAA available really for CD3 MOA but I think the 4-1BB is a little bit different. The MOA there and the internal selectivity requirement is a little bit different compared to CD3 and also 4-1BB is preventing the T-cell exhaustion. So in a way, in this microsupressed environment within a cold solid tumor, 4-1BB might come up to be a better way to activating T cells. Of course, we need more data to really test the hypothesis, but it's a really interesting area. We're going to really explore and put a lot of effort behind.

Your question on B7-H4 is a really important question. I think what it comes back to ultimately is clinicians are thinking about differentiation and what they're selecting when they have a group of different molecules to select from. So perhaps the treatment algorithm for ovarian cancer can be summarized that patients receive their frontline chemotherapy with platinum and taxol and then patients are essentially dichotomized as part eligible or not, and is biomarker testing needed for PARP selection. So I do think that then having a second biomarker to triage to which EDC is an additional level of complexity and it would be attractive to not have to run that second biomarker in the circumstance where you're not giving something meaningfully for efficacy. So again, we'll share our biomarker data tomorrow. I think that we are not asking patients to give something up whether their B7-H4 is high or B7-H4 is low. And then the other critical part of differentiation, particularly maintenance is the safety. We have a very clean safety profile, unlike HER-2 targeted ADCs, which have stomatitis, which can be quite problematic, we have low rates of hematologic toxicity. HER2 is in biomarker selected subgroup. So we think that there are a number of factors that would align towards having biomarker selected approach being favorable in this patient set.

It's Leo from RBC Capital Markets. I wanted to ask on the CDK4 maybe for Dr. Goel. I just wanted to ask on your clinical perspective on safety here. I guess, to what extent is the lower heme toxicity a must have given the current real-world management rather than a nice to have? And I guess, would it make you change your use given you may have more experience with prior CDK4/6 agents? And then maybe if I can squeeze one. And related to that, if the GI toxicity doesn't improve meaningfully, but the heme profile stays favorable, I guess, how does that change your view of how the drug would fit in?

Okay. So there's a lot there. Let's start with the heme. So the low rates of heme tox, that's been seen with this drug, to me, this is very important. It's not just something that is a low number on paper. And there's a few reasons for that. One is that what that indicates to me really is the selectivity that the drug has for CDK4 over CDK6. That's what I see that as a readout of and you saw the preclinical numbers from Mark to support that. And what that in turn suggests is that the thesis -- one of the theses behind these drugs is that by sparing CDK6 and reducing neutropenia, it allows us to hit CDK4 harder, right? So in that sense, I think it's more than just a low number. It potentially has deep biologic implications for the durability of cell cycle control. That's one point. Second point to remember, and it is not a trivial one, is that with our existing CDK4/6 inhibitors, there are a fraction of patients who actually come off therapy on account of hematologic toxicities. If you look at the first-line studies with the existing drugs, I think depending on the agent, between 5% and 10% of patients came off therapy for toxicity and a large number of those were heme tox. So I think again, by mitigating heme tox, I feel that you have less patients that go down that road. And then I guess, thirdly, just slightly as a tangential point and one that Mark mentioned, more forward-looking, I think that the lack of significant heme tox also opens up opportunities when we're thinking about combinations in the future. As many of you know, the field is moving towards triplet regimens for ER-positive disease. And I think less heme tox you have with your foundation, with your hormone therapy and CDK4, the greater your options for triplets. With respect to the GI tox, first of all, I would say the numbers that we've seen today in terms of concomitant administration with food are relatively small. But my sense, and I've looked after a number of patients on this agent is that the concomitant administration with food. That effect, I mean time will tell with larger numbers, but my sense, my clinical sense is that there is something real there. Your question was what if it's not the case? So I can also talk to the experience of the patients I look after who took the drug in the fasted state and also point to some data that was on our poster today, but which wasn't in my presentation, looking at the prevalence of Grade 2-3 diarrhea over time. If you look at that data, what you'll see is that by the time you get to cycle 3, there's a real drop-off in the rate of Grade 2-3 diarrhea and that, that just continues to tail off to even lower numbers as patients continue on treatment. And that is not because patients were coming off therapy for diarrhea in the first cycle, no one came off therapy because of diarrhea. To sort of turn that into a more simple sort of anecdotal sort of nature, my experience with the agent has generally been that people have very manageable GI toxicity. If needed, low-dose intermittent loperamide, patients and doctors learn very quickly during that first 1 to 2 cycles, how to manage it and get on top of it. And after that, our experience has been that it is a very manageable thing. So I don't think even if the concomitant administration with food, that data doesn't play out. I don't think that, that would necessarily limit my view of the long-term potential for the drug.

Thank you, Dr. Goel. I think we have time for one more question.

Yanan Zhu, Wells Fargo. I have 2 questions on CDK4 inhibitor and one on HCC. On the CDK4, I was wondering if you could frame the relative efficacy here compared with Pfizer's asset in a similar stage -- in a similar setting. And then just note there are significant dose interruption and dose reduction in the study due to the diarrhea effect. With food, that's in the data, but with patients taking it with food and being able to better tolerate the regimen, could we expect even better efficacy just given the level of dose reduction seen in the current data? On the HCC, it's a quick question. Given the high prevalence of HCC in China, what proportion of patients in your pivotal study could come from China and be okay with the FDA?

Sure. So I can start with the CDK4 questions. So I think doing sort of meaningful comparisons across different studies with different agents, honestly, is difficult with the number of patients that we've seen the data from. And so anything to sort of say, oh, this or that is very speculative. I think what I can say is from the data that I showed you today, both at 240 and 400-milligram doses of BGB-43395, the objective response rate compares very favorably to what we have seen in the global randomized trials with the currently approved CDK4/6 inhibitors. So that's very encouraging. But showing data from a limited number of patients with either agent, I can't really speak to comparative efficacy. And I'm just trying to recall what your second question was.

[indiscernible].

Yes, yes, the food. So I think that this is again speculative my answer here, but if the GI toxicity is significantly mitigated through concomitant taking food and yes, you would expect let's see how this plays out over time that, that would lead to fewer dose reductions. But I think to then take another step, and we want to maximize relative dose intensity, right? That is a good thing and fewer dose reductions maximizes relative dose intensity. But again, it would only be speculation to say and therefore, that would translate to deltas in efficacy.

If I can address your second question regarding regulatory expectation. To be very clear for everybody, we have not had that conversation with FDA yet. So this is purely my opinion based on prior experience and speculation. My prior experience is that FDA will not view that question from the perspective of there are a lot of patients in Asia, but rather is this a common disease or a rare disease in the United States and there are patients eligible for enrollment. And as I mentioned, this is actually a common disease in the United States with an incident rate between 30,000 and 40,000 patients. So I suspect that FDA's position will be that they expect that we will have a typical proportion of U.S. representation in the Phase III study. And the overall allocation may be weighted a little bit towards Asia potentially. But as the study -- as with all of our studies, we will ensure that the regional allocation meets the need of all major regulatory agencies. Okay. Thank you all very much for participating in our panel tonight. And now would you -- Lai, would you bring us to a close?

Yes. Thank you. I'm very happy to share a few closing remarks. First, this is an inflection point in our solid tumor program history. In just 2.5 years, our first wave of new assets has already delivered 5 POCs, including 3 we highlighted today. Second, momentum is building. The next wave is coming with our next-generation IOs beyond more than what you have seen from the public domain about our trispecific. We have other things which we will bring to the clinic very soon. Also the 4-1BB bispecifics, thanks for the question, we are very excited about this MOA. And in addition to that, we'll also have novel payload ADCs following right behind. And finally, this is about more than just individual assets. We have talked about this many times tonight. This is about innovation, execution at scale, powered by our franchise model, world-class discovery and our global development superhighway. Thank you for spending your precious time with us today, and we truly appreciate your time and interest and engagement. We hope you enjoy the rest of ASCO and a safe trip home. Thank you.