GSK plc (GSK) Earnings Call Transcript & Summary

Thank you very much, and good afternoon, and good evening, and welcome to GSK's Investor Science Event Getting Ahead of Anemia from Chronic Kidney Disease for investors and analysts from the American Society of Nephrology Kidney Week 2021. I'm Mick Readey, your moderator for today's call and a member of the GSK IR Team. We're here today for an update on daprodustat, a potential best-in-class new medicines for the treatment of anemia of chronic kidney disease. As usual, the presentation materials are available on gsk.com and were sent to our distribution list earlier today. Please also note that the data presented on today's call are available on the ASN website or in the accompanying publication on the New England Journal of Medicine. Before we get started, I would like to thank Frannie DeFranco from the IR team for her contribution to today's call. Please turn to Slide 2. Now a quick reminder of the usual safe harbor statement and forward-looking statements disclaimer. Please turn to Slide 3. We are joined on today's call by Dr. Hal Barron, Chief Scientific Officer and President of R&D; and Luke Miels, Chief Commercial Officer at GSK. We're also pleased to welcome Dr. Ajay Singh, Senior Associate Dean for Postgraduate Medical Education from Harvard Medical School, and Principal Investigator for the daprodustat's ASCEND Phase III program, including the ASCEND-D and ASCEND-ND Phase III data that were presented as late-breaking abstracts at ASN on Friday. Please turn to Slide 4. This is today's agenda. I anticipate the call will last around 1 hour with half the presentation and half of the remainder for Q&A. Naturally, we will try and ensure we take as many questions as possible. For the Q&A point -- from the Q&A portion of the call, we will be joined by John Lepore, SVP Research; and Chris Corsico, SVP of Development. [Operator Instructions] Before I hand over to today's presenters, I'd like to remind you that the call is being recorded and that a replay will be available after the event. And with that, I will now hand over to Hal. Please turn to Slide 5.

Thank you, Mick, and a warm welcome to everyone on today's call. Please turn to Slide 6. Let me start by saying how pleased I am with the positive Phase III results from the daprodustat ASCEND program that were presented on Friday. Daprodustat is a drug that acts as a HIF-prolyl hydroxylase inhibitor, a target we chose to pursue because the genetics strongly suggested a role in stimulating erythropoiesis. As well as compelling world-class biology that actually led to a Nobel prize. Daprodustat's success in Phase III lends further credibility to our R&D strategy, which focuses on human genetics and functional genomics to identify target with a higher probability of success. Our robust clinical development program recruited more than 8,000 patients in well-designed studies with active controls, and resulted in very consistent clinical findings in both efficacy and safety. This positions daprodustat as a potentially best-in-class oral agent for anemia due to chronic kidney disease. We're also quite proud that the ASCEND program reflected real-world management of patients with chronic kidney disease, and specifically included both dialysis and non-dialysis-dependent patients in separate, well-powered studies aligned with requests from regulators. Lastly, we are inspired by what this potentially transformative medicine could mean for patients. There are more than 700 million people worldwide living with chronic kidney disease and 1 in 7 of them suffers from anemia. The current standard of care for these patients is an erythropoiesis-stimulating agent or an ESA, administered via subcutaneous injection or as part of dialysis. Although these can provide benefits to patients, only around 30% of non-dialysis patients today receive ESA treatment due to the administration challenges. With daprodustat, we believe we have a convenient and flexible oral treatment option for both dialysis and non-dialysis patients, which has shown improvements in quality of life and predictable increases in hemoglobin. With that, it's my pleasure to hand over to Dr. Singh to discuss the results of the studies presented at ASN on Friday. Dr. Singh?

Thank you very much, Dr. Barron. Before I get onto the next slide, let me just say thank you, everyone, for joining. I know it's Sunday, and there are many better things you probably have to do, but I really appreciate the time you're giving. I think it's important to give you a little bit of background about myself before I talk through the slides. I've been practicing nephrology for the past 35 years, both in the U.K. and in the United States. And have been involved in anemia studies for the past 20 years. 15 years ago, I published the CHOIR study in the New England Journal of Medicine and testified to the House Ways and Means Committee in the U.S. Congress, and spoke about the limitations of current therapy. A few years later, we published the TREAT study, a placebo-controlled trial, again, reinforcing the limitations of therapy. So it's really exciting for me, at a personal level, to share with you this data because I think we are meeting an unmet need that currently exists in the treatment of anemia among kidney disease patients. Not only because current therapies have raised issues around safety, including cardiovascular safety and tumor-related issues, but also the fact that we haven't in any trials been able to effectively demonstrate that there's an improvement in quality of life. So this trial program, I think, is -- I'm particularly proud of because I think it could be an important step in the right direction for our patients with kidney disease, particularly those with chronic kidney disease who are not receiving therapy, as Dr. Barron spoke about. Let's turn to the next slide, Slide 8. In this slide, what you see is an overview of the ASCEND clinical trial program. It included 5 Phase III clinical trials in more than 8,000 patients that was targeted to investigate the efficacy and safety of daprodustat across a spectrum of patients with chronic kidney disease or CKD. The CVOT trials that you see at the top, the clinical trials labeled ASCEND-D in dialysis and ASCEND-ND in non-dialysis comprise of approximately 6,800 patients and approximately 14,200 patient years of experience. So a tremendous amount of experience that we're presenting today. My presentation will focus on the 2 cardiovascular outcome trials in the dialysis and non-dialysis population, although I will share with you program level cardiovascular safety at the end of the presentation. Can I have the next slide, please? This next slide shows you the overall structure of the ASCEND-D and ND trials. Both trials were event-driven, open-label, randomized, active-controlled, parallel-group, multicenter Phase III trials conducted in 41 countries worldwide. Both trials included, as you can see on this slide, a 4-week screening period and then a 4-week run-in period before randomization marked by the letter R occurred. And patients who were randomized 1:1, either to daprodustat, oral administered daily or to an ESA comparator. Now if you see this box, you see ESA (D) and darbepoetin SC (ND), and let me explain that to you. For ASCEND-D, epoetin alfa was administered to hemodialysis patients and darbepoetin alfa to peritoneal dialysis patients. That's because peritoneal dialysis patients are home patients and they needed a subcutaneous injection. And that's why darbepoetin was used instead of epoetin alfa. For the ND study, ASCEND-ND study, darbepoetin alfa was the active comparator. For both trials and for both arms, the drug was dosed to a hemoglobin target of 10 to 11 grams per deciliter. Now remember, in normal, healthy people who are not on dialysis or with kidney disease, the normal hemoglobin is around 13 to 15 grams per deciliter. But the target of 10 to 11 was something that was decided in conjunction with the regulators, both in the United States and in Europe. Now the same iron management protocol is applied to both arms, and you see that in the box right in the middle. Now with respect to the co-primary endpoints, that's shown on the right-hand side of the slide, these were 2. One was the mean change in hemoglobin from baseline to the average during the primary evaluation period, that represented week 28 through week 52. And the second co-primary endpoint was timed to Major Adverse Cardiovascular Event, or MACE. And as you know, MACE is defined as all-cause mortality, nonfatal myocardial infarction or nonfatal stroke. Again, these were events-driven trials where the trials ended when the target number of MACE events reached 664. The key entry criteria are highlighted on this slide as well. Details of these entry criteria included in the published manuscript that became available to everyone, I'm sure you've received it. It was -- they were published in the New England Journal of Medicine. And I'm not going to go into the details of them, but you can see the key criteria. And of course, you can look at them in more detail in the papers. If you go to the next slide, this slide discusses the patient disposition. The first most important point here is that we looked at the intention-to-treat -- we used an intention-to-treat analysis to look at this population. So that's the ITT population. And the key points here with respect to patient disposition were a few patients withdrew, and the withdrawal rates were similar across treatment groups. Second, the premature discontinuation of randomized treatment was balanced across treatment groups. And three, known vital status was high in both trials across both treatment groups. Why is all this important? Because it really speaks to the high internal validity of the trials. It's very important to know what happened to the patients and whether -- whatever happened to the patients with respect to withdrawal, was balanced between the 2 arms. And I think this is something I have a lot of pride around because I think the conduct of trials is a very important part of doing trials. The next thing you see, the number of randomized patients. We had 2,964 patients randomized in the dialysis study, and 3,872 patients randomized in the non-dialysis study. And you can see that when you looked at this with intention-to-treat, the numbers in both arms. If you go to the next slide, you see the co-primary endpoints. Before I just speak to that, I just want to add one little piece of information. In the papers, you will see baseline data on the 2 patient populations, the samples that were recruited in the study, the D study and the ND study. And baseline data or Table 1s, as we traditionally call them, are really important to look at because the Table 1s in both studies showed that there was balance, again, on baseline characteristics between the 2 trials. That's really important for 2 reasons. One, it shows that the results reflect randomization having worked. Because if you randomize properly, you'd expect it to be balanced between the 2 arms, again, speaking to internal validity. And the second reason it's important to know about the baseline data is you want to know whether these patients reflect what we generally see in clinical practice. Are these results generalizable to clinical practice. And in this -- both these cases, the answer was they were. And again, I refer you to the papers and the journals for more information about them. If you now look at the next slide, this next slide looks at co-primary efficacy endpoint for hemoglobin. And remember, this is defined as the mean hemoglobin change from baseline to the valuation period from week 28 to 52. And again, we're looking at, in a prespecified way, the intention-to-treat population. And we're first looking at the ASCEND-D data. And in blue panel, you see the major -- the message from this. Daprodustat was noninferior to ESA for the mean change in hemoglobin from baseline to the evaluation period. That's from week 28 to 52 showed in the shaded area for the graph just below it. Now it is important to know that in the ASCEND-D trial for the co-primary hemoglobin endpoint, this -- the adjusted mean treatment difference was 0.18 grams per deciliter, with a 95% confidence interval of 0.12 to 0.24 grams per deciliter. And the noninferiority margin, which is, again, prespecified and agreed upon with the regulators, the noninferiority conclusion was supported here because the confidence interval for the treatment difference is entirely above the prespecified minus 0.75 grams per deciliter that represents that noninferiority margin. Now what you also see on the slide, the mean hemoglobin profile during the trial for dapro in purple and for ESA control in orange. And you can see that the mean hemoglobin for both treatment groups remain in the analysis hemoglobin range of 10 to 11.5 gram per deciliter. And this was the case for the ESA hyporesponder subgroup as well, supporting the notion that, a, dapro works and works effectively. And that our protocol worked, and the protocol worked well in terms of determining the dosing of dapro. But the second is that this -- that dapro worked well in patients who are ESA or ESA hypo-responders. If we now turn to the next slide, that's Slide 10. We see the co-primary efficacy end point for hemoglobin for the ASCEND-ND trials. And here, similar to the previous data, dapro was also noninferior to darbepoetin alfa for that same co-primary hemoglobin endpoint. The mean adjusted treatment difference was 0.08. You can see that in the box on the right-hand side, with a confidence interval of 0.03 to 0.13 grams per deciliter, which is entirely above the prespecified minus 0.75 gram per deciliter for the noninferiority margin. So it met the criteria for noninferiority. Now the corresponding graph on the slide here, that's on the left-hand side, displays the mean hemoglobin profile during the trial for dapro, which is marked in purple and darbepoetin in orange, and you can hardly see any difference and we superimposed them over other especially during that shaded evaluation period. So in this trial as well, the mean hemoglobin for both treatment groups remained in the analysis hemoglobin range of 10 to 11.5 gram per deciliter. And again, in both -- for both the D and the ND trial with respect to noninferiority, we met the noninferiority criteria for this. So the trials were successful with this co-primary endpoint. Now if you go to the next slide, that's Slide 13. We are looking here at the first occurrence of adjudicated MACE. So let me just spend a second on this. Adjudicated MACE, adjudicated meaning that these were events that were adjudicated independently by Duke Clinical Research Institute that were the endpoints committee, the clinical endpoints committee for this trial. And MACE, as you recall, is mortality, nonfatal myocardial infarction and nonfatal stroke. And these analyses are all prespecified in our statistical analytical plan. And our analysis that are based on the ITT population based on, obviously, agreement with the regulators. Now the -- let's now look at the results here. On the left are the results for the ASCEND-D trial where 25.2% of dapro patients and 26.7% of ESA-controlled patients experienced a MACE event. Dapro was noninferior to ESA with a hazard ratio of 0.93 and a 95% confidence interval of 0.81 to 1.07. And we'll discuss why in a second, but let me just hold that thought with you. So it was -- you know that -- and I shared with you the hazard and the confidence intervals. Now on the right are the results for the ASCEND-ND trial where 19.5% of dapro patients and 19.2% of darbepoetin patients experienced a MACE event in the trial. And here again, dapro was also non-inferior to darbe with a hazard ratio of 1.03 and a 95% confidence interval of 0.89 to 1.19. Now for both these MACE endpoints, for both the trials, the noninferiority conclusions are supported because the confidence interval for the hazard ratio is entirely below the prespecified noninferiority margin of 1.25. Now also on this slide, you're all familiar with Kaplan-Meier curves. And so you see 2 Kaplan-Meier plots. One on the left for the D trial and on the right for the ND trial. And these show the percent of patients with a MACE event over time. Dapro is shown in purple and ESA control is shown in orange. And as you can see, the lines are not only very close together, they're essentially superimposable on each other. So it's not at all surprising that the noninferiority criteria were met. Now in the boxes in the middle, you see the components of the first occurrence of adjudicated MACE. And again, the details of all of this are much more -- are laid out in much more detail in the papers that were published on Friday, but let me just walk you through them. On the left, you see first occurrence of adjudicated MACE with dapro and ESA for the D trial. And they're really similar when you look at the components that are the all-cause mortality, myocardial infarction or stroke. And then on the right-hand side, you see the ND data, which again, depicted in a similar fashion, show a great similarity between both dapro and darbe. So again, the message here for this important co-primary endpoint was that noninferiority was achieved because the upper boundary of the 95% confidence interval of HR was lower and, therefore -- than the prespecified NI margin of 1.25. We go to the next slide. Here, we're looking at the MACE supplementary analyses. And of course, as you know, this is always done in much greater detail and discussed in the papers, but I'm going to share with you some key messages. What you see here are the analyses conducted in the co-primary MACE endpoint. And as you know, the results of the primary analysis was ITT, which was the prespecified analysis. So this included both on- and off-treatment-based events. And this is shown on the first row for both. So you see on the left the ASCEND-D and on the right, the ASCEND-ND. And the first row is the first occurrence of MACE, the primary analysis. And as you can see here that it's almost close to 1 in this case with respect to daprodustat versus ESA for both. So the noninferiority criteria were met here. Next, an on-treatment analysis was run, and this excluded MACE events that occurred more than 28 days after a patient's last dose of treatment. And you can see that on the left, very similar to the primary analysis. On the right, it's an outlier for first occurrence of MACE on-treatment. And I think that it's important to emphasize the reason for this is that we're looking at this as an on-treatment analysis. And as you know and I know through all these years, essentially an on-treatment analysis is not the same as the prespecified ITT analysis. And the prespecified ITT analysis is essentially what you do as a clinical trialist. Whereas the on-treatment analysis is essentially opening up the data set after the trial has essentially been done. And it's open -- the analysis is open to a variety of confounding factors as well as treatment biases. In this case, post hoc analysis demonstrated that the differential dosing of dapro compared to darbepoetin explained in large part this outlier effect. And the effect attenuated once you took the differential dosing frequency into account. The next 2 rows show further ITT analysis, which included both on- and off-treatment MACE that were run. On the left, you see the analysis that excluded MACE related to COVID-19. And also MACE that occurred after the target, 664 first events were reached in the dialysis trial. And again, very similar to the primary analysis. On the right, in the non-dialysis trial and additional prespecified ITT analysis was run that included additional covariance in the analysis model. And you see here, and again, very similar to the primary ITT analysis. And so I think that it's important to know that this is discussed in much more detail in the papers that were published. Now we go to the next slide, that's Slide 15. Here, we're looking at the principal secondary endpoints. And the principal secondary endpoints, the message here is that they did not meet multiplicity adjusted statistical significance for superiority using the Holm-Bonferroni method. On the left, you see the data for the ASCEND-D trial. On the right, you see the data for the ASCEND-ND trial for these principal secondary endpoints. The first 3 principal secondary endpoints were consistent between trials and included a superiority analysis for time to first adjudicated MACE, which had been assessed for noninferiority in the primary analysis. As well as superiority assessments for the adjudicated MACE plus thromboembolic events and adjudicated MACE plus hospitalization events for heart failure endpoints. Now it is important for one thing which I think stands out is the MACE plus thromboembolism events is 0.88 with a 95% hazard ratio, 0.78 to 1. So MACE plus thromboembolism favors dapro in terms of fewer MACE plus thromboembolism events with dapro as compared to darbe. And when you look at the Kaplan-Meier curves, and you can look at that in the paper, the D paper which is published, you see that Kaplan-Meier curves diverge in favor of dapro pretty early on after randomization occurs and they continue to gradually diverge. When you look at the thromboembolic events separate from MACE, there are a higher number of thromboembolic events among patients randomized to darbe compared to those randomized to dapro. So with respect to thromboembolism, lower number of events for dapro patients. And what seems to be driving this are vascular access thromboembolic events. And you can see that in the paper itself. And I just would encourage you to look at that data. When you look at the ND, there was no difference. They're very similar between both the dapro and the darbe. Now when you look at the ASCEND-D trial for -- with respect to the on-treatment average monthly IV iron dose, which is shown on the bottom left, that's the on-treatment average monthly iron dose in milligrams from baseline to week 52. We saw a benefit in iron parameters consistent with the Phase II finding. So there was a benefit in iron parameters. This included a lower level of hepcidin. And as you know, hepcidin is a very powerful regulator of iron within the body, and regulates iron egress out of the enterocyte, out of the small intestine into the body. And lower levels is better in terms of allowing for more absorption of iron. So the level of hepcidin was lower in patients randomized to dapro. There was also a higher level of TIDC in patients randomized to dapro. However, we did not see a significant treatment difference in iron utilization. Although IV iron use was reduced in both treatment groups. So it was a little bit lower, minus 9.1, wide confidence intervals, probably not clinically meaningful. We don't have to look to this in much more detail because there were a number of things that were going on. There was an iron management protocol. Most patients on dialysis receive iron intravenously, they don't receive it orally. Most of our patients did too. And so there may have been an opportunity to actually enhance oral iron absorption because these patients weren't receiving oral iron. And once it's speculated, that might explain why you see also a benefit in iron parameters but no difference in iron utilization. On the right-hand side, you see the other principal secondary endpoint, and that's for CKD progression in the ND trial. And CKD progression, as you can see on the slide, is defined as a 40% decline in eGFR or progression to ESRD. And these definitions now are widely accepted and widely accepted by the regulators. And the hazard was 0.98 with confidence interval 0.84 to 1.13. So no difference with respect to CKD progression in the ND trial. The next slide, that's Slide 16. We're now looking at adverse events. Now -- and we're looking at adverse events in the safety population. The safety population are defined as randomized patients who received at least 1 dose of drug. And we're reporting here treatment-emergent events. Now additional information about adverse events is contained in the primary management for both trials that have been published. So I'm only going to go over this in a summary fashion. In general, rates of treatment-emergent adverse events and serious adverse events were similar between treatment groups in both trials. Rates are predefined treatment-emergent adverse events of special interest or also generally similar between the treatment groups in both trials. Now 2 of the adverse events of special interest, esophageal and gastric erosions and cancer events actually displayed conflicting results between the trials, and this is undergoing further investigation. Now post hoc analyses that are described in the primary management for the primary ND trial, which explore the impact of the differential dosing frequency on the treatment-emergent results. And these analyses demonstrated an attenuation of the imbalance for cancer events but not for gastric/esophageal erosions. The other point is that these are, again, on-treatment analysis. And as you know, what's important to patients is really what happens in the ITT approach. We randomized this drug and you see events versus on the right to another -- the comparator and you see what the results are. When you compare it based on ITT, in fact, there is no difference, there is balance between the 2. So I think that for cancer, these results, the post-hoc analysis plus looking at this with ITT is very reassuring. Now let's go to Slide #18. And now we're sort of pulling the lens back and looking at the ASCEND program-level cardiovascular data. And we're looking at the data for cardiovascular events that were adjudicated in all of the 5 Phase III trials. Although it's important to note that the 3 smaller trials were not designed for a formal MACE evaluation. So the 2, the ND and the D, the results of which I've already described, were designed empowered to look at MACE, whereas ID in instant dialysis PD, which is 3x a week dialysis and NHQ were not designed for that. Now what we're showing -- what I'm showing you here are the 4 trials at the top. And just summarizing the MACE results rates for the ASCEND-NHQ, that's the trial that focused on quality of life because that trial was even more different than the others in that it was very short duration, only 28 weeks. Even so, and just focusing on NHQ for a second, the MACE events, 4.9% in dapro and 6.2% in placebo. The NHQ trial was a randomized controlled placebo -- double-blind, placebo-controlled trial, that's why placebo was used. The results of the NHQ trial were presented as an oral communication on Friday afternoon at the American Society of Nephrology Meeting. So anyway, that's the NHQ. Let's go back to the data above that. And what you see here on this slide is the first forest plot. And you'll see here the MACE results by treatment group expressed as a rate per 100 patient years for the 4 active control trials, and you can see that there. The second forest plot on the right-hand side contains the absolute rate difference for 100 patient years between the treatment groups. And you can see that the MACE profiles were generally consistent across treatment groups in all of the trials. So this is really important, right? Now we're looking at even more patients and even more patient years of experience And really, the results are very consistent with respect to MACE, not only with respect to ND and D, but also across the program as a whole. Now if you go to Slide 18, we're looking at, in Slide 18, as the ASCEND program-level cardiovascular data with respect to MACE components. Remember the MACE is mortality, nonfatal MI, nonfatal stroke. And what you see here is that the component events that makeup MACE will show consistency across the board with respect to all the trials and across the treatment groups, dapro versus comparator ESA. And this, again, reinforces that not only in the individual 2 cardiovascular outcome trials, but also across the program, the MACE profiles are very much consistent. And this, I think, is really important. The last slide is the summary and conclusions. And here, the important points are that dapro was as effective as conventional ESA therapy in treating anemia of CKD. Dapro was noninferior to ESA with respect to cardiovascular safety, and no new safety signals were observed. And so looking forward, I believe, and I think my nephrology community, speaking for them having done this for many, many years, can confidently say that dapro could represent an oral alternative to ESA for treating anemia CKD in both dialysis and non-dialysis patients. And I think it presents an opportunity, particularly so in patients, for example, as Dr. Barron pointed out, non-dialysis patients who hitherto have essentially been under-treated. I think that's a great opportunity. I think it also represents an opportunity in dialysis patients and non-dialysis patients because it might target 2 arms of the anemia treatment paradigm, not only ESA deficiency because kidneys become diseased and so you need erythropoietin, but also the iron. Now we haven't got -- we have to still do a lot more studies in iron, but I think the benefit on iron parameters is really very, very intriguing, and I think will need to be looked at in more detail. So I just want to say thank you again for your attention, and I'll turn it over to Dr. Barron gain.

Thank you, Dr. Singh. So if I could ask everyone to please turn to Slide 21. So firstly, on this slide, you can see that CKD is a significant global burden. And as Hal mentioned, there are more than 700 million people worldwide living with CKD. Patients in the more advanced stages of 3 to 5 are the target group for daprodustat. And we know that this patient group is underdiagnosed, undertreated or not treated to target levels. It's also important to note that this is a progressive disease, which requires many years of treatment and increasing levels of specialty care. The increasing burden for patients underlies the need for an effective, safe and convenient treatment. And these patients usually have other comorbid conditions like diabetes and TB disease, further complicating therapy choice and management. Currently, patients are treated with iron supplements and regular injections of an ESA, which brings administration challenges. Please turn to Slide 22. With the positive Phase III data that you've just heard about today from Dr. Singh, we believe there is a significant opportunity to provide a differentiated medicine in an area of high unmet need. In the non-dialysis population, there are more than 1.3 million people under treatment in the U.S. and the EU, and only about 30% of this population are treated with an ESA. Furthermore, around half of these patients discontinue their treatment within 1 year. Consequently, there's a major opportunity to impact the lives of patients in the non-dialysis setting, and we believe that daprodustat could potentially offer a new convenient oral treatment option. The dialysis-dependent population, I want to specifically highlight 2 populations that we think daprodustat could immediately help. Those already receiving dialysis at home, and those who are on a high dose of ESA and require muscle injections known as hypo-responders. Together, these 2 subpopulations make up around 1/4 and growing of the dialysis population. Patients receiving dialysis at home account for around 12% of total dialysis patients under treatment, and this proportion is expected to grow to around 25% by 2025. These patients need cold storage and to inject their medicine, which brings administration challenges. And an effective and convenient oral option has the potential to simplify the treatment regimen and improve their quality of life. For the hypo-responders, these patients also represent around 12% of the dialysis population, but they bear a significant proportion of the cost as they can receive up to 6x the standard ESA dose. In this setting, daprodustat has the potential to be a transformational medicine with simple oral dosing replacing the high and frequent injection burden and the costs. Please turn to Slide 23. In collaboration with our partner, KKC, Duvroq was launched in Japan in August of 2020, 9 months after the first HIF. The medicine was also -- has already surpassed the main competitor and is now the market leader with 47% class share. One of the main drivers is our favorable label, which enables patients to switch directly from an ESA, a competitive advantage we intend to seek to replicate in our label in other markets. In Japan, we're now seeing 75% of patients coming from switches and the other 25% from new patient starts. We're also encouraged by the feedback we've received from nephrologists, which is driving the strong momentum and uptake in both the non-dialysis and dialysis-dependent populations. The successful launch in Japan has provided us with lessons that we'll apply globally, including the importance of early treatment with external experts and health care providers. We've also seen positive trends in our target populations. And the support -- this supports our confidence in the differentiated benefits of this potential new medicine. Turning to my final Slide, 24. The strong clinical data from the ASCEND program provides us with a set of strong messages that we can bring to physicians and patients assuming, of course, that we can gain regulatory approvals. And these include a convenient oral option for the non-dialysis and dialysis patients, which may overcome the requirement for regular injections, with the ability to manage hemoglobin in a predictable manner and the potential for flexible dosing. And a demonstrated improvement in quality of life of patients. With the benefit of these compelling messages, we see a significant opportunity to transform patient lives and to drive commercial success for daprodustat. We previously highlighted a sales potential of GBP 0.5 billion to GBP 1 billion. And in this range, we assumed daprodustat would be third to market. Now clearly, the competitive situation has encouragingly moved in our favor, meaning there's certainly upside relative to our previous assumptions. We are now sharpening our focus on preparations for forthcoming regulatory interactions, and we will keep you updated as we move through that process. We now have made the decision we'll commercialize daprodustat alone. And since 2017, we've strengthened our marketing and medical capability in the specialty area by hiring over 900 employees. And we have a strong base to build off in nephrology, given the established leadership position of Benlysta in lupus nephritis. In terms of the next step, we plan regulatory submissions to the FDA and the EMA in the first half of 2022. We're excited to bring this potentially transformative medicine to market as soon as possible for the benefit of patients, so we'll be doing everything we can to accelerate our filing time line. And with that, I wanted to thank you for your time, and I'll pass it back to Hal to open up Q&A.

Thank you, Luke, and thank you, Dr. Singh. And now, operator, please, if you could open the line for Q&A. We'll try to be as brief as possible in answering the questions, but -- so that we can take as many as questions as the time permits. Thank you.

[Operator Instructions] And the first question comes from Graham Parry from Bank of America.

First question just is on FDA attitude to the endpoints you're using. So in the roxadustat briefing documents, FDA stated that it's generally asked sponsors to show noninferiority on MACE versus active comparator in dialysis and placebo in non-dialysis. And the ASCEND-ND trial was run against an ESA, which has got a box warning for cardiovascular events. So how comfortable are you that placebo-controlled isn't something the regulators are going to want to see, especially given that the hazard ratios are all trending slightly above 1 and the upper confidence intervals are also sort of pushing the 1.2 range in the primary and key secondary endpoints? And then secondly, you've highlighted a GBP 3 billion addressable market opportunity there on ESA markets in the U.S. and Europe. And your peak sales guide was GBP 0.5 billion to GBP 1 billion. Are you assuming that you can expand the non-dialysis market? And is there -- when you say there's some upside to your GBP 0.5 billion to GBP 1 billion, can we assume the GBP 0.5 billion is out of the picture and perhaps just give us some framework around where you think the GBP 1 billion could go. And then lastly, the editorial highlighted the excess cancer and GI erosion risks as problems. And also in the dialysis or the thrombosis hazard ratio is 1.81, although that wasn't statistically significant. Just given your comments around -- on the Q3 call around commercial opportunity being linked to how the editorial treated the data? And what are your thoughts now on that editorial and how that might impact your commercial opportunity?

Thanks, Graham. A lot of questions there. Maybe what we could do, Luke, if you want to take first the commercial question. And then Dr. Singh, maybe I could ask you to comment on the safety and thromboembolic events that were highlighted by Graham. And then to the extent that you could also comment on the trial design of the non-dialysis having an active comparator. And at the end, I can make a comment on the regulatory component. But Luke, do you want to start with the commercial question?

Sure, Hal. So I think on the editorial overall, I mean, I'll leave it to Dr. Singh and Hal to comment further, but I think it was relatively balanced. Clearly, a lot of mentions of vadadustat, which probably reflects the background of the contributor, the author. And I think it's probably worth explaining in the Q&A the intention to treat versus on-treatment, particularly in terms of the comments around cancer, but I'll leave that for the other speakers. I think for -- in terms of the forecast, again, we won't update today, but the key assumption is that it was third to market, that's now clearly changed. I think at a macro level, the way I look at this, is you've got -- when you look at differentiation versus options available in terms of epo, I mean you've got relative simplicity with dapro. I think in the non-dialysis population, of course, it's oral versus injection or infusion. There's no cold chain. We don't expect -- and again, this is subject to regulators. We don't expect restrictions on complement oral iron or phosphate binders, nor expect no DDI with statins. And again, I think you've got this high-term morbidity in complex conditions with CPD and Type 2 diabetes. And I think the other thing is which we can supply the papers behind the data that we've quoted. You've got a large proportion of stage 3 to 4 patients being managed by primary care physicians in the U.S. And that even in the nephrology setting, it's a relatively undertreated patients who have anemia are being undertreated. In dialysis, home dialysis, clearly, when we talk to the dialysis providers, they're trying to increase the level of home dialysis. And I mentioned the hypo-responders can also be managed. It's interesting they're around 10% to 12% of patients and around 40% of the costs, which obviously has consequences when you look at the bundle. So net-net, I think it's fair to say probably 2/3 of the volume will be in the non-dialysis setting, again, subject to the final label that we see.

Thanks, Luke. Dr. Singh, do you want to comment on the safety profile that Graham brought up? And then I'll conclude with some of the regulatory component.

Yes. Yes. So I think the 2 points are -- first is with respect to thromboembolism. There were somewhat disparate results between the D and the ND trial. The D trial showed essentially that there were a higher number of thromboembolic events in the patients that were treated with darbe versus those treated with dapro with the dialysis patients, and that was mostly being driven by number of vascular access, the thrombosis events. In dialysis, as you know, vascular access thrombosis is much more common because the access has been used and there's a higher -- not hugely higher, but somewhat higher prevalence of grafts because these grafts are placed in and around the time these patients are started on dialysis. In the ND trial, I think as you cited, there was a numerically higher number of thromboembolic events, non-favorable events in the patients who were randomized to dapro and slightly less so in darbe. For the Journal, the reviewers, the nephrology community as a whole, I think, it didn't -- this wasn't a particularly problematic thing. I think these are very complex patients, both dialysis and non-dialysis patients. Neither of those results reached statistical significance and particularly with respect to when you're looking at the way you analyze the data, doing analysis for the co-primary, which didn't show statistically different results, which were now, I think it's very important not to over-interpret any of the analyses of any of the components or any of the principal secondary endpoints. So that's number one. I think with respect to the -- you're raising the issue around cancer in the ND trial, I mean, honestly, I'm not exercised by it at all because, firstly, the total number of events were relatively small. Secondly, it was conflicting between the 2 trials that like with the thromboembolism between the D and the ND. In the D, cancer rate were somewhat a little bit favored dapro as compared to darbe, whereas in the ND it favored darbe as compared to dapro. Remember that these are not the intention-to-treat analyses. The intention-to-treat analyses, what you do as trialists and that is the important information to bear in mind. If you go into the presentation and to the slides, I think there's some backup slides that perhaps Mick can show. He's showing them right now. I can see them. You see that the treatment of emergent definitions, which are defined using the on-treatment approach. And then on the right-hand side, you see it with the ITT definition, which is what patients really care about, right? And you see that there's really no difference. And so I'm not exercised about this at all and nor were the reviewers, nor were the editors, nor were the people who listened in on the presentation and wrote comments and questions at the ASN. So I think -- I hope that answers the question. I can't speak to the regulatory stuff. I'm going to turn that back over to Hal.

Thank you, Dr. Singh. Let me just highlight the intent-to-treat definition. Remember, intent to treat is when a patient gets randomized to 1 arm versus the other, and then you follow up for the entire study period. And you can see, I think, quite clearly that the difference in cancer rates on the side, of the right side of the slide are almost identical. And this is, of course, the primary endpoint in the way trials should be examined. And I just want to put an exclamation mark on that because the 1.03 really, I think, highlights the key here that there is no difference. Let me just comment on your question about the active control in the non-dialysis. I think it's important to remember that many patients, as we highlighted, who are -- have anemia from chronic disease but are not in the dialysis setting have significant symptoms from their anemia and would be candidates for epo. But because of the need for subcutaneous administration or many of them do not get any ESAs, and it's those patients that we think really would benefit from dapro, and therefore, the design is as such. And I should also highlight that while we don't comment on any specific conversations we've had with the regulators, I'd like to reiterate that the design of both studies, the sample size, the analyses, particularly the primary endpoint of being intent-to-treat, confidence intervals, et cetera, were all done in strong collaboration with the regulators. And so we're very confident both in the data and the design. Why don't we turn over...

I'll just add something, Hal, for a second. I was involved in the CHOIR trial where we had an active comparator as well as being involved in the executive committee of the TREAT trial, where we tried to do a placebo-controlled trial. It is important to know that in the TREAT trial, we had an enormous difficulty in getting that trial done. And I think the general feeling is that the standard of care, if you actually want to randomize patients to treatment is to treat them in the control arm with ESA. I think it's an important point. And we had a lot of difficulty with -- and I think that's why the regulators essentially asked you guys to design trials which had an active comparator.

Yes. thank you for that extra point. Okay, operator, let me -- let's see if we can get to the next question.

Yes, certainly. Your next question comes from the line of Laura Sutcliffe from UBS.

First question is on your dialysis trial. I think if I look at the equivalent vadadustat trials, there are a greater proportion of patients in your trial having events in both arms. So could you perhaps just talk to the aggregate risk profile of the patients in your trial and whether you think they look like normal "dialysis patients"? Second question is on the commercial strategy in dialysis. This is a market where it's quite possible that a large chunk of it could be off limit. Can you -- if daprodustat is approved, does your success depend on you getting non-dialysis on the label? Or is that the wrong way to think about it? And then finally, just on the hypo-responders piece. If I remember correctly, the FDA took quite a dim view of how this was presented in the roxa package. I think they said there was no specific evidence supplied or something like that. So could you just maybe talk about how you plan to get around that problem? Do you have enough there that classed as specific evidence in the hypo-responder population?

Thanks, Laura. Dr. Singh, maybe you could take the baseline covariant differences question from Laura, as well as maybe briefly comment on the hypo-responders, and then we'll turn it over to Luke for his thoughts on the commercial strategy in non-dialysis.

Yes. So I think your -- the simple answer to your question is, are they typical of the sort of patients we take care of on dialysis? The answer is yes. But let me be a little bit more specific. We had a good representation, I think, an appropriate representation of people who are African-Americans. As you know, in the U.S., there is a significant number of patients who are African-American. And in fact, our trial, I think we pride ourselves on the fact that we had really good representation of African-Americans in the trial. First -- in the trial as a whole, 15%, roughly the same between the 2 arms were black. And it was higher when you just looked at the U.S. population. When you look at the dialysis type of randomization, about -- close to 90%, specifically, 89% were on hemodialysis and 11% on peritoneal dialysis, very, very much similar to what you see in -- on patients who are being treated with different dialysis modalities. When you look at history of cardiovascular disease, about 45% in both arms had a history of cardiovascular disease, very similar. Over 90% had a history of hypertension, and about 40% of patients had a history of diabetes. Remember that when we take care of patients, about 35% to 40% of patients are diabetic as the cause of kidney failure. And so I think -- and this was, again, obviously reviewed carefully by the Journal, I think the population is very typical of what you see. I can't speak to the other trial, the vada trial because I wasn't involved in that. But I think, very comfortable with the idea that our population are typical of what we see in dialysis patients. But I see when I run in the dialysis unit in Boston or whether you run in the dialysis unit in London, in the U.K. And I think that therefore, these results are generalizable. With respect to ESA hypo-responders, about 12% of our patients were hypo-responsive to ESA. And that is roughly what you see now in the population on dialysis. There are many different ways of defining hypo responsiveness and I won't bore you with all the different definitions. But we used 2 definitions, and we presented data on ESA hypo responsiveness. There are additional papers that will be coming out that go into that in much more detail. We found that the drug -- we found that dapro treatment of ESA hypo-responders resulted in a similar hemoglobin response to that of patients who are treated with darbepoetin. The only little difference was that there seem to be less utilization of iron in the ESA hypo-responders who were randomized to dapro than to those who were randomized to darbe. And I think this data is sort of in the supplementary section, it's discussed in the primary paper. But is in the supplementary appendices that are presented with the paper and are published on the Journal's website.

Thank you, Dr. Singh. And just to put a point on that, pretty clearly in the ASCEND trials, the patients on dapro performed as well as the general population, even if they were hypo-responders, which we think is an important point. But Luke, do you want to comment on the clinical...

Sure. Yes, I'll just do that quickly. Yes. I mean as I said earlier, Laura, and thanks for your question. I mean we see 2/3 of the revenue at peak, that being non-dialysis, so clearly, that's a very important component for the product. In terms of the LDOs in the U.S., I mean, obviously, subject to approval, but we've already initiated discussions and they're going quite well. I think you should assume they are in the bundle. I think we get a lot of questions around TDAPA time frames and [ ATPCS ] time frame. So I think if you just, as a rule of thumb, assume around 9 months or 3 calendar quarters after the FDA approval.

Thanks, Luke. I know it's the top of the hour, but let's see if we can get a few more questions in just because we are running a little late. We have 6 more queued. So operator, can we have the next question.

Your next question comes from Emmanuel Papadakis from DB.

I'll try to keep short. Dr. Singh, thanks for the presentation. Perhaps a quick one for you, based on the totality of evidence, to what extent do you believe the safety profile of daprodustat is definitively differentiated to roxadustat and vadadustat? We've seen a lot of data for 3 of those molecules now. And what do you think the probability of an advisory committee is recommending this for non-dialysis approval? And assuming it does, perhaps you could just give us what either in terms of your practice or you think representatively in the U.S., you think the likely percentage utilization in that population would be if it has a black box?

Why don't we just go right to Dr. Singh on the totality of data relative to what's out there? I should just say before we start, we're -- given it's a first-in-class potential approval, we are expecting that there'll be an advisory committee, so we're preparing for that. But Dr. Singh, I don't know if you have any comments on how this data does or does not compare with prior trials and prior agents in a similar class. And then you can follow-up on how you see this, if approved, impacting your practice?

Right. So I'm going to -- I'm not trying to make comparisons because I think that it's -- these are different molecules and their trials were designed differently. And there isn't enough time for me to go through every single distinction on these. But I think we're very comfortable with our data suggesting that we -- our goal was to demonstrate noninferiority to comparator ESA. And I think we did that and I'm very proud of it. And I think I'm proud of the fact that we did really well conducted trials where we essentially nailed down virtually every patient. So we didn't really have unknown information sitting out there, so we don't know what happened to patients, et cetera, that were randomized in the trial. So I think all of that, to me, suggests that we have a very robust program, and we presented, I think, very detailed information both in the papers, but also in the supplementary appendices. What do I think is the -- what -- if I had a crystal ball and said what would -- as a nephrologist, would I think would be the place of these agents? I think it's very important to know that we are really undertreating our non-dialysis population. It's very difficult. And I'm saying this now putting a clinical hat on, right? And you ask me what do we do in Boston. So we don't -- patients have a hard time getting into -- to get a physician-supervised anemia treatment. That's a fact. And this is particularly -- so in the population that seems to be most impacted. African-Americans, people who are of low socioeconomic class, et cetera. And so I think there's an opportunity, at least in the U.S., I can't speak for everyone in the world, but at least in the U.S., for us to be able to now start treating these patients who should be treated for their anemia. That's important. It's also important to note that when these patients' anemia is not treated, they have quality of life issues, right? And if you can get their hemoglobins up, patients are going to feel better and they're going to function more like their normal selves. So I think there's an opportunity there, just speaking as a nephrologist with regards to what I see. The second is I do think that there's an opportunity potentially to have an agent that might have potential or might have potentially influence iron parameters. We did see that. It was -- it's out there. Their hepcidin levels go down by about 30% very soon after these patients are randomized to dapro and TIBC levels go up. We didn't see a difference with regards to iron utilization. I think iron utilization is very complex. I will be straightforward with you. There was an iron protocol that was applied. And most of these patients are not treated with oral iron, which is where you'd see an effect, an hepcidin-induced effect on iron absorption, which -- so I think that more studies need to be done there. Thirdly, and I think -- I know we're running out of time. But thirdly, the experiment where you are able to normalize hemoglobin by stimulating natural production of erythropoietin has not been done. That is the next step, right? And so post-marketing studies, other studies will need to be done, say, is it really appropriate to undertreat dialysis, non-dialysis patients' sub-normal hemoglobins? Are we going to accept that? Or will we ever accept undertreating or inappropriately treating diabetics to higher blood sugars than what normal people have? I don't think that's acceptable. And I think in the long term, with my crystal ball, and forgive me for saying it, with a crystal ball-type of metaphor, I think that you're going to want to see whether we can normalize the hemoglobin. And I think the best way to normalize hemoglobin is to use an agent that might naturally stimulate erythropoietin production. Now remember, I was the PI and first author of the CHOIR study that resulted in a black box warning that the FDA put into the label. I'm convinced that these patients who are being treated had those side effects, not because their hemoglobins rose too high, although that's certainly one of the hypotheses. But because they're exposed to large doses of exogenous erythropoietin. And so I think that, that experiment has to be done where we try to normalize with an agent that doesn't result in very high levels of erythropoietin. I apologize for speaking longer than I should, but I needed to get that out there.

Thank you. Thanks, Dr. Singh. Emmanuel, let me just add one other comment about the research focus that we've had over the years that John Lepore, Head of Research, has really focused on. There are actually 3 different isoforms of the HIF prolyl hydroxylase, PHD1, 2 and 3. And each regulates different sets of genes and different tissues. Remember, this is a transcription factor modulating agent, which is quite unique. And the main isoform that regulates epo production in the kidney is PHD 1 -- sorry, PHD2 I think. The very small molecules developed that are HIF PH1 modifiers, inhibitors have different specificity against each of the PHD isoforms. And dapro has a very high selectivity for PHD2. So in addition, we're very careful to ensure that dapro was selected for the HIF PH as that they're involved in erythropoiesis, not if they are not HIF-regulated prolyl hydroxylase, such as collagen prolyl hydroxylase, et cetera. And that might be in addition to the very, very robust development program that we put forward, things that will differentiate our safety profile from other agents. But both the very robust development plan and the unique aspects of the molecule probably have had an impact on our ability to see so clearly the effect of this impressive agent.

Your next question comes from Simon Baker from Redburn.

Two, if I may, please. Firstly, Dr. Singh, I just wondered if you could give us a little bit more color on the manifestation and severity of the esophageal and gastric erosion that was observed. And then secondly, the editorial talked about the regional variations on the safety endpoint for vadadustat. I wonder if you could discuss the regional variations that you saw in these studies. It looks like in ASCEND-D, the MACE performance in Western Europe was particularly good. And for ASCEND-ND, the U.S. upper confidence interval looked like it was about 1.54. I just wonder if you could discuss the implications of that.

Yes. So I think there are 2 points, right? One is that -- and again, I'm not going to dwell on other people's data, but just very briefly with respect to design, we use the same hemoglobin, we use the same hemoglobin targets across the whole world, right? And whereas in the vada program, they use different hemoglobin targets in different parts of the world. So that was an important geographic difference. We did look at geography, and we did stratify for geography. And there wasn't really anything that I think when you look at U.S. versus ex-U.S., there wasn't any difference. So -- and everything else pretty much is sort of didn't reach statistical significance. So I'm not convinced actually that there's any geographical difference. And just as a nephrologist, I'm not sure I could explain why you made interpretation. Similarly I think that if you use the same hemoglobin, if you use the same hemoglobin target, I think, as we did, I'm not surprised at all with that data that shows no difference between the U.S. and ex-U.S. or outside U.S. With respect to your other question with regards to esophageal and gastric erosions, I think it's important to note that these were relatively mild. The effect of what happened in these patients, these erosions were mild and they were self-limited. And when I looked at the data, thinking of what are the sort of things that happen in these patients, it was all -- across the map, there were lots of little, little things. Some patients had upper GI-type dyspepsia. Other patients had noticed that there was some GI bleeding. There wasn't any pattern to it. I do think that we will need additional investigation here because the numbers of patients were relatively small. And I think that we pointed that out that -- and it was disparate, right, between the 2 trials, between the D trial and the ND trial. So I think it was seen. I think it was important to state what we saw. And I think it's important to also say that we need additional studies to investigate that further.

Thank you very much, Dr Singh, and I'd just like to point out that with the gastric erosions too, the small numbers and desperate data in each of the trials within the dialysis, there was a trend towards it being less actually with -- if I recall correctly. So it is small numbers and a bit wobbling, but something to look into.

The next question comes from Kerry Holford from Berenberg.

One for me. So on commercial infrastructure, Luke, you mentioned you would leverage your Benlysta position. Should we expect any significant incremental spend required to launch dapro, assuming it's approved? And can I ask for clarification on the number of reps you mentioned, I think it was 900. Are those already on board promoting Benlysta or is that the total you will need to then increase to? And my second question is just on the rate of increase in hemoglobin. So you referenced what I saw in the paper a faster increase in hemoglobin noted in the dialysis trial versus ESA between the first 4 weeks of therapy. Is that an important differentiated point in the clinic, do you expect that to work in your favor in discussions with regulators? Or are there, in fact, any potential risks associated with such rapid change in hemoglobin?

Thanks, Kerry. Luke, why don't you go ahead with the sales force and...

Yes, so 900, Kerry, is the total number of specialty care hires that we've made, so that includes Benlysta, the oncology business, expansion to Nucala. So essentially, the argument there is we've been able to rapidly increase the number of people who are highly qualified versus a history in selling small molecules in plastic containers in your primary care physicians. And in terms of spend, I think in aggregate on the P&L, no, it's not going to be an impact. But clearly, we'll reallocate between resources that we have and other products. But I would expect that we would invest heavily behind this product at the time of launch to ensure maximum uptake.

Thanks, Luke. And Dr. Singh, could you briefly comment on the importance of the rapid rate of increase in hemoglobin relative to epo in the studies?

Yes. So the important point is we did not see a rapid rate of rise in hemoglobin in our patients in the D and ND trial between the 2 groups. And I think in part that likely reflects the fact that we have a protocol that was really effective in controlling hemoglobin in patients who are randomized to either dapro or darbe. And obviously, I can't speak to what other trials did. I will point out that 10 years ago, I published an editorial in CJASN, which in very great detail discussed the hemoglobin rate of rise issue when the FDA commented about this with respect to the -- to this with the CHOIR data, where I published the CHOIR data. And it was analyzed in the briefing document, and I got into lots of discussions with the FDA, and you can see that in the editorial. It was important to note that it's not just the rate of rise of hemoglobin, it's hemoglobin flexes. And so rate of rises was -- the FDA invoked that as potentially very important, but also rate of fall in hemoglobin could be important. It's complex because these are post-hoc analyses, they have not been published generally, and they're open to a lot of confounding and confounding especially by indication. So I think we will need a lot more studies. And I think clearly, discussion on rate of rise. But it wasn't seen in our study, in the data that we've reviewed so far.

Your last question comes from James Gordon from JPM.

A couple on pre-dialysis and make safety, and then just one about the pitch and dialysis. In the pre-dialysis, why do you think the dapro does look a little bit worse than the ESA on safety? And then the other way around dialysis, do you think there is a real difference in terms of how patients are pre-dialysis or dialysis patients actually respond to the drug? Or do you think it's just noise and should be ignored? And just any regulatory concerns about the on-treatment analysis, where that 1.4 hazard ratio, could that be a concern as well? That was the questions on safety. The other question was just in dialysis, if you're a patient who's not an ESA hypo-responder and is not a home dialysis patient, what's the pitch to be using HIF in that population, please?

Dr. Singh, do you want to take the difference in safety profiles in non-dialysis versus dialysis? And maybe just reiterate some of the comments made earlier on the on-treatment analysis. And then Luke, I don't know if you want to make any final comments on that commercial aspect of the last component and then we'll wrap up.

Yes. I mean, I never want to say never to anything. You made a very strong statement about -- I think the most important thing to know is that we -- the trials were designed to demonstrate noninferiority in an ITT population. And in the intention to treat population, which is the way these trials are generally done and certainly were done with respect to what the regulators wanted. And what I think the Journal is now clearly indicated the way to do these trials that we achieved noninferiority. And I think that's important. And then now just to compare what you're seeing in non-dialysis and dialysis, I don't want to say anything is noise. I mean I never say that. But it doesn't concern me because I think that meant noninferiority. That's why you do these trials. The -- with respect to the on-treatment, I think we've gone through that in some detail. I think on-treatment is not the same as intention to treat. Intention to treat is the way you do these trials so that you -- so you look at -- you do randomization, you balance the confounding effects and then you look at the impact of the intervention. When you balance these other confounding factors and covariance, and I think when you do it that way, you don't see the results you see with on-treatment. I think that's important to realize. It wasn't something that the reviewers or the journals were concerned about. And I think that's also important to know. So I'm pretty sanguine about the fact that dapro is comparable to ESA in both populations. And as well it was tolerated but no additional safety signals. I think that's my take-home message from looking at the data in detail as someone who's been doing this for many years.

Thank you, Dr. Singh. Luke, do you want to comment on the non-dialysis rationale?

Sure. I think it was the dialysis if I heard James correctly, and feel free to correct me if I'm wrong. Look, I think James, it's important to look at the total context here. I mean, if you look at the visit payment under the bundle, it's about $231. So if you do have 1 in 10 or 12% of patients who are using very high volumes there, that is going to have an economic consequence for these LDOs. And of course, we can contract across all patients. So there is, I think we, as our initial discussion, can construct something quite attractive there in terms of relative to epo. I think if you look at versus vadadustat and roxadustat, and I suspect this drives some of the uptake in Japan, you've got the simplicity of switching. But you've also got the DDI. If you look at roxa, and again, this is not necessarily saying this will happen in the U.S., but you've got to avoid oral iron and phosphate binders 1 hour before with statins as well. 2 hours with vada between phosphate binders and/or iron. And you need to do liver monitoring with statins and [indiscernible]. So these are all elements we think we can assemble arguments for those LDOs to contract and be interested in interacting with us.

Thank you, Luke. Why don't I just wrap up? Thanks, everyone, for attending. And just to say, again, we're very excited about the dapro program, the dapro molecule. It is a very unique molecule, as I highlighted, a very robust development program, quite compelling data. And most importantly, I think the -- as you've heard repeatedly, the unmet need is quite significant, both in dialysis and particularly in non-dialysis for those patients with chronic kidney disease and anemia who actually get nothing and have a significant fatigue. I want to end by thanking our own internal team for a terrific job over many years on bringing this to fruition. And thank Dr. Singh for a great job presenting the data and for helping us with the design and implementation of what will likely be a very transformational medicine. So great job to everybody. Thank you very much for taking the time, and we'll end on this note.

Thank you to all the speakers, and thank you, everyone. That concludes your conference call for today. You may now disconnect. Thank you for joining, and enjoy the rest of your evening.