Madrigal Pharmaceuticals, Inc. (MDGL) Earnings Call Transcript & Summary

So thank you all for coming. And for those who have joined on the webcast, welcome. With me today, I have Dr. Becky Taub, President of R&D and Chief Medical Officer of Madrigal; and Stephen Harrison needs no introduction, lead investigator of our Phase III MAESTRO studies, resmetirom. Please see today's press release, unless you want to memorize this for our forward-looking statements and I ask you to look at the press release today and our SEC filings for forward-looking statements and risk factor considerations associated with these statements and our business. The EASL 2022 International Liver Congress has been a busy time for Madrigal. Earlier today, we presented late-breaking data from our Phase III MAESTRO-NAFLD-1 safety study of resmetirom. This was the first presentation of the double-blind results in a major scientific meeting and we're going to review some of that data shortly. We and our collaborators presented 3 additional oral abstracts and 2 posters among the oral abstracts or analyses of new data from the open-label cirrhosis population studied in MAESTRO-NAFLD-1, I think very encouraging data. And that data support our decision to initiate a new outcome study in patients with compensated cirrhosis due to NASH, obviously, a population with a very high unmet need and also a method of achieving full approval in the NASH space earlier than we would with the ongoing study, the biopsy study, which would take some years to complete. We were gratified speaking of the cirrhosis study to see the EASL that EASL highlighted that abstract in a press conference yesterday. As we approach the NASH biopsy study readout later this year, I believe that our development program has contributed a renewed sense of optimism and momentum in the NASH field. At least I hope it's stimulated that I heard from many people today who would actually be considered our competitors under normal circumstances, and they are really wishing us and I believe it, to make it over the finish line to try to reinvigorate the field. Now I'd like to thank the team at Madrigal, the many authors who contributed to the abstracts, the MAESTRO study investigators and the patients participating in clinical trials for making this possible. I'm going to turn the meeting over to Becky and Dr. Harrison now to review the data that we presented at EASL, and we'll follow that with a Q&A session.

Thank you, everyone. So you can hear me okay? I think it's coming through this mic. So what we're going to do tonight is show an abbreviated version of 2 of the presentations that were focused on the clinical studies, which were the MAESTRO-NAFLD primary data presentation that was a late-breaker and the NASH cirrhosis trial, which was discussed this afternoon. And in the interest of the fact that we had a lot of presentations, we won't discuss the other presentations. I think it's -- a list is on the next slide. We can come back to this. So basically, we're going to focus on the late-breaker and the cirrhosis trial. We also showed -- had a presentation from the work done at HistoIndex to look at the impact of liver volume and steatosis, correction on fibrosis using artificial intelligence, evaluation of liver biopsies. And we showed this morning the utility of FIB-4 thresholds based on screening data from MAESTRO-NASH. So in that talk, we showed a lot of actual screening data from MAESTRO-NASH and discussed various noninvasive parameters that can be used to successfully identify NASH patients even without a liver biopsy. So that's going to be very important feature. And -- we'll take questions on those topics, but not actually show the slides. And then I think some of you saw an important poster, particularly from a different AI company called PathAI and NASH analysis of our Phase II liver biopsies, which showed excellent concordance with detecting NASH resolution is the same rate as our two pathologists from our Phase II study. And both HistoIndex and PathAI will form part of the team that reviews the MAESTRO-NASH liver biopsies, 52 liver biopsies that we will be reading out the data from in the upcoming months to present the final results in the fourth quarter of this year. So given that introduction, I'll actually go back to this slide, just to remind you that Dr. Harrison is going to present data from the mice, the third arrow here from the MAESTRO-NAFLD trial first from the double-blind and open-label non-cirrhotic arm of that study, and then from the ongoing cirrhosis arm of that study, the first cohort that has finished week 52. And again, we're in the -- we're continuing to recruit MAESTRO-NASH to get up to 2,000 patients that will form the basis for the final clinical outcome portion, but we announced almost a year ago that we had -- or about a year ago that we had completed the Subpart H recruitment, and we'll be reporting out the data on that 900 patient cohort. And then we also have an open-label extension of MAESTRO-NAFLD that's ongoing. It's a 52-week extension study. as well as you heard, our intention to start a second outcome study, which is MAESTRO-NASH outcomes in well-compensated NASH cirrhosis. So we'll go ahead and have Stephen present some of the data from MAESTRO-NAFLD and the NASH cirrhosis arm.

Good evening, everybody. How is it going? Nice long day. Just doing a little check here on what we have. Okay. So as Becky mentioned, just to go through some of the data that we presented. So this is the Phase III trial design for the MAESTRO-NAFLD-1 study. And as you guys can see, there are 3 arms, the 100-milligram, 80-milligram, 100-milligram open label and a fourth placebo arm treatment is for 52 weeks, randomization is a 1:1 to 1:1. Now to get into the trial, you had to have at least 3 metabolic risk factors. But specifically, you don't have to have a liver biopsy, and that's the unique feature of this particular study. It's a presumed NASH population. So understanding that it's a milder population than what we enrolled in MAESTRO-NASH, which looked at NASH with NAS 4 more in F2 or F3 fibrosis. This one, you just had to have a FibroScan of 5.5 to less than 8.5 and a CAP greater than or equal to 280. So there's a little typo there, the less and should be greater than on 5.5. So really, we're looking for kind of a finite group of patients. And the reason the CAP was at 280 was because we needed an MRI of greater than 8% or greater than equal to 8%. And so there's a very nice correlation between CAP score and an MRI score of greater than or equal to 8. So the screen fail rate on MRI is around 20%. When you prescreen with a FibroScan CAP and you have a score greater than equal to 280, it's very good at predicting liver fat greater than or equal to 8%. So that's -- there were a couple of other ways to get in. There was a little asterisk there. And so maybe I didn't go into detail with that today, but I'll go into detail a little more with that for you guys. So in sites that didn't participate in the MAESTRO-NASH study, where you had to have a FibroScan greater than 5.5 or you were allowed to enroll into MAESTRO-NAFLD if your FibroScan was greater than 5.5. Does that make sense? Okay? So the majority of patients got in using the Orange criteria. There were some that got in -- if they didn't participate in MAESTRO-NASH, they could get in by doing a FibroScan and it being greater than 5.5. So it could be 12 and they could be in the study. In addition, if you did a liver biopsy and a screen failed for MAESTRO-NASH, at the biopsy stage, they could roll over into this trial. Now when we look at enrollment, there was a significant interest in this particular trial. And so enrollment went very, very fast. And ultimately, about 1,143 patients were enrolled, 972 into the double-blind arm, so the 3 arms at the top, the placebo, 100-milligram and 80 and then 171 into the open-label arm. So the objective there was a safety objective that was primary, the 52 weeks, and that was basically how well was the drug tolerated and what was the safety around both doses. And then looking at key secondary objectives, again, hierarchical control, LDL cholesterol at week 24, ApoB, which is a very important marker for cardiovascular risk at week 24, hepatic fat fraction percent change from baseline at week 16, triglycerides in a subgroup of patients that had baseline values greater than equal to 150 at week 24 and then FibroScan. Additionally, we did look at MR as well. So just to jump into some of the baseline characteristics. I didn't spend a lot of time on this, this morning, but maybe a little more time here with you guys tonight. This is, I think a very good representation of what we see in the clinic. So mean age around 55. So where do we -- what do we know about that? Well, and for my prevalent study I did in San Antonio and published in JHEP this year or late last year, the mean age of our population with NAFLD was 55. Prevalence of NAFLD was 38% by MRI. And on liver biopsy, 14% of that population had fatty liver, NASH, and 6% had F2 or greater fibrosis. So this population is exactly the same age. So theoretically, you could speculate that some of the similarities that existed from my paper exist in this population. I didn't editorialize that in my comments today, but you get that as a bonus for coming tonight. So as in most studies, and everyone that I saw here, the majority of enrolled patients are female. BMI was a robust 35 to 36 and half of them are diabetic. Importantly, the ASCVD risk score is not low. It's between 11% and 13.5%, which is an at-risk population. Liver stiffness measurements are what you would expect in a milder population between 7 and 77, lots of fat here. When you select studies, just think in your mind when you look at all these other trials, when you select studies based on MRI of greater than 10%, you'll see mean PDFFs in the low 20s. When you pick it at 8, you'll see them in the high teens. That's what you see here, 17, 18. MRE range between 2.6 and 2.8. If you look at data that's been generated in independent patient data meta analyses or systematic reviews, 2.6 to 2.8 gets you into a NASH with early fibrosis area. So the rest here, I think, is relatively straightforward. The other important point is ALT in this population as a mean, is only around 36 to 38, which by almost every standard is still in the normal range. It's abnormal, but if you look at Quest or LabCorp, upper limit of normal is around 40. So this isn't a population that has a really high ALT. So keep that in mind when we look at some of those results. So when we look at safety, I think that's important. It is the primary endpoint. You don't see any difference between groups for treatment emergent adverse events. There's no difference in the severity grade I, grade II or treatment-emergent adverse events greater than or equal to grade III severity, very well balanced between 80, 100 and placebo. When you look at, at least one serious treatment adverse event, the same. AE discontinuation rate, a little bit higher in the drug group, but those related to discontinuations, not a big difference here. And when we look at the GI-related adverse event discontinuations, it's 5, 6 and 2 with the 5 and 6 and 80 and 100 milligrams being loose stools, but most of those actually are mild and patients just chose to discontinue treatment. It's not like they were admitted to the hospital with profuse watery diarrhea and had to be hydrated. So -- what we don't report here in the table format is the 100-milligram open-label arm. So I have that highlighted at the bottom. 94% and 89% completed key efficacy endpoints at week 24 and 52, respectively. And the dropout rate due to adverse events was overall 1.2%, so very low. Now this is kind of a busy slide, but it is chock-full of information that I think you guys will like to see. So on the left is lipids. And on the right is MRI. So just to orient you, anything in bright red is the 100-milligram open-label arm. The orange is the 100-milligram arm, and the black here in this slide, I think it was brown and the ones we presented. This one is black was the 80-milligram dose. You don't see a lot of blue because that's placebo. So first look at LDL. You see with 100-milligram open-label arm around the 21% change from baseline, 14% for the 100-milligram arm. And you see in every one of those examples, the 100-milligram open-label arm did a little better than 100-milligram blinded arm. And the question was raised today, why is that? Well, remember that the double-blind arms, we're in the middle of the COVID pandemic. And on average, those patients missed about 2 months of dosing out of the 1 year. So it's interesting to me, I guess, maybe it speaks to the potency of the drug that despite missing 2 months of dosing on average, we still saw a nice result. Now importantly, when we look at P values, that's the first key bullet point, key secondary endpoints were achieved for both the 80 and 100-milligram dose with highly significant statistical significance for all of the variables listed across the top. So not just the lipids, but MRI and FibroScan as well. So when we -- you can see the values, they're very similar for LDL, for ApoB. The ApoB number, I find very interesting. If you look at -- and I'm not a lipidologist, but I know people that are in talking with them, I ask what's a good number for ApoB that makes sense that we can kind of stick our stake in the ground and say, this has some implications cardiovascularly, it's around 15% reduction. So you can see that, that's looking good. On MRI, 49%, 48%, 41% for the dosing arm, 6% for placebo, and at 52 weeks, it's a little bit better, but about the same to 53%, 48%, 43%. And so Dr. Noureddin asked me the question today, what do I make of the fact that it didn't continue to drop? Have we reached our plateau? And I don't know the answer to that fully because we didn't do biopsies in these patients, and I don't know what the steatosis looked like at the end. But if you can look in the middle here, we deliver volume-corrected PDFF change from baseline. This is the open-label arm, 100 milligram. Each one of those patients, we looked at their liver volume reduction and put in a correction factor for what their PDFF should be. And it actually is 61% instead of 49% when you correct for liver volume reduction. So if you applied that same corrective factor to 52 weeks, you would see that the numbers would be a little bit -- would be different. They might also be different at week 16 for that matter. And FibroScan had a nice dose response as well. So I think that's the main points from this slide and with a little bit of editorial that I didn't give in today's presentation. So again, just looking at the ALT, and I made this point earlier that most of the patients had ALT values, the mean value was less than 40%, it was in the high 30s, 38%. I think. And so we wanted to look at the change in ALT and we selected people that had baseline ALTs greater than 30%, greater than or equal to 30% because a lot of people came into the trial with lower than 30%. So when you select it higher than 30%, we saw a very nice dose, actually not a dose response, the 80 and 100 behave very similarly compared to placebo for ALT and AST, a slight separation between 80 and 100 for GGT, but really no difference out to week 36 and 52. And these variables were highly significant relative to placebo and consistent with what we saw in the open-label arm. And again, let me remind you that most patients missed about on average 2 months of dosing. Now there were some ALT increases greater than or equal to 3x the upper limit of normal and a little more than 0.5% of the 80-milligram dose, 1/3 of 1% in the 100-milligram and 1.6% in placebo. And that is not inconsistent with what we see with many other drugs being developed. Now this may be the busiest slide, but the most interesting slide all at the same time. So -- and this is chock-full of data. I mean, like you want to just chew on this for a little bit because if you look at the - what we looked -- here we looked at MRE and FibroScan. But we only took people with baseline liver stiffness on MRE greater than or equal to 2.9 and FibroScan greater equal to 7.2. Now in this study, most patients didn't have baseline values that met this criteria. So you see the ends here are 88, 60, 44, not the 300 in each arm that we talked about at the beginning. So just keep in mind, the numbers are a bit smaller, although still relatively decent sized. And despite the low numbers, look at this resmetirom versus placebo significant for a change in MRE even in this population of relatively mild NASH patients. Now FibroScan, you don't see a P value here, and that's because although directionally showing a resmetirom treatment effect at 100 milligram, the mean change wasn't significantly different in part because the variance of FibroScan from -- in one patient over time is high. It's about 30% to 40%. So instead, we looked at what we call a responder analysis. Now for MRE, it's a little bit easier to pick the number because we have QIBA guidance from the radiologist to tell us what the coefficient of variance is for MR elastography, and it's 19%. So if you want to have truth, you pick that variance, and you say, how many had improvement greater than 19%. And you see 24% if you combine the 100-milligram open label and 100-milligram blinded arm, you see 24%. For 80, it's 22% for placebo 11. So you're doubling the rate of MRE hitting that responder analysis. And conversely, if you look at the worsened, you're cutting in half the number that worsen when you dose with resmetirom. Now with FibroScan, it's a little bit more arbitrary as to what we pick as a responder analysis. So we chose to do it 2 different ways. First, on the right, a 30% change from baseline, kind of picking that middle ground around what coefficient of variance is because it's a little bit varied depending on what population you study. And if you use the M probe or the XL probe, but here, you see a nice dose response relationship, 45, 36, 26 versus placebo at 21. And then when you look at improved by greater than or equal to 2 kPa, 57, 43, 32 and 25 and the P values for those are below. So 100-milligram open-label, highly significant, 100-milligram double-blind, significant, no significance with the 80-milligram compared to placebo. So the conclusion from this late-breaker top line data from MAESTRO-NAFLD-1. And I have it and highlighted in red because I think the FDA needs to see that, resmetirom was safe and well tolerated, which in the therapeutic index equation is half the battle, safe and well tolerated at the top dose of 100 milligrams as well as the 80-milligram dose. And so this met the primary endpoint. That was the primary endpoint. Now when you look at key secondary endpoints, Remember, we did a hierarchical analysis, LDL at the top. All of them were achieved. So safety and efficacy are in line with what we saw from the Phase II trial we published in Lancet. And what are consistent with what we previously had presented in the randomized parallel open-label 100-milligram arm. Are there limitations to this study? Sure. The biggest one being we didn't enroll a whole lot of advanced fibrotic patients. But that gets me to my next talk, which is cirrhosis, where we did evaluate them. So we'll talk about that in a minute. And so we really believe, and I think in talking to many of my colleagues that belief is really not just mine or not just the leadership at Madrigal, but we really believe these positive results support the conviction that resmetirom has the potential to be the first medication approved for the treatment of patients with NASH and fibrosis. And I've been around this since the early back of the napkin days with Becky in 2015 and we designed the Phase II. And back then, it was like seventh or eighth in line, simtuzumab, selonsertib, obeticholic acid, elafibranor, [indiscernible], all that was in front of resmetirom. And all those have kind of reported out already and either struggling to get to the finish line or aren't making it. So there's been a lot of interest in the thyroid hormone receptor beta class of drugs, particularly this one to get to what we are at today, reading the safety and tolerability of this drug and now the anticipation turns toward the MAESTRO-NASH trial for efficacy. So I'll stop there and well, I really won't stop there. I'll just keep going. So just keeping it in the tune of today being a Madrigal Presentation Day at EASL, there was another presentation. It's not often I give a talk at a late-breaker and say, stay tuned. I'm not going to share all the information about this trial in one particular talk, wait 2 hours, go to a different room and you'll hear more. Yes, there's more. And so tonight, for you guys, you don't have to wait 2 hours. I'm going to roll right into the next presentation. Same study, so we can probably go faster. In fact, I did, you did because you cut a slide out. So -- you don't need me to show you the whole Phase II or Phase III design again. So I'm not. I'm just going to show you the 80-milligram open-label arm that was a nice addition to this trial. When we designed the MAESTRO-NAFLD-1 study, I think it's fair to say, Becky, this arm was not part of the study. This was -- this was generated through seeing cirrhotic patients. Can I say this, seeing cirrhotic patients in clinic and thinking about maybe there's a group there we need to study. And so we designed a cohort to look at this in more detail. And to do that, we said, let's set the same 3 metabolic risk factors, but let's focus on well compensated NASH patients, not child B, not child C patients, although we have hepatic impairment data in those patients as well. Let's focus on the Child-Pugh A, 5 and 6 point group with no history of decompensation. So to get into the trial, give me a biopsy that shows cirrhosis, give me a biopsy that shows F2, F3 disease somewhere in their past and now clinically, they progress to cirrhosis, but still compensated. And in a very few number of cases, there were some just clinical evidence of cirrhosis that were included. But the vast majority had a liver biopsy that showed either cirrhosis or close to cirrhosis with clinical progression to cirrhosis. Now there were 2 cohorts. The first cohort enrolled 105 patients, and that's what we reported on today. There is a second cohort of an additional 75 patients that is still ongoing. And there were a couple of additional tests that were included in that cohort. So other than the addition of a week 2 visit, this study schema is exactly the same as what I've already shown you. Why is there a week 2 visit? Well, we did PK sampling and we could adjust the dose up to 100 milligrams based on exposure. And I think around 10% or 15% of patients actually increased the dose. And if I'm wrong on that, Becky, you can correct me. So looking at baseline characteristics, I find this fascinating because we think about finding the right patients based on the liver biopsy and our clinical data. But all of this data, a lot of this data, particularly on the right, was generated after the patients were screened into the trial. And on the right, you see FIB-4 of 2.9. Just for a little bit of flavor and maybe some cool like trivia, the semaglutide data presented today by my colleague, Rohit Loomba, in a cirrhotic population had a mean FIB-4 at 2.4, which was below the cutoff for advanced fibrosis of 2.67. I don't know if you noticed that. I didn't want to say anything about it. But I thought that was interesting because here, we're squarely in the advanced fibrosis camp based on FIB-4 alone. And when you look at PRO-C3 there, very, very high as well. ELF scores puts you into the greater than 9.8 by the way, is the cut point for risk of progression to cirrhosis. And so we're really close on that one. 11.3 actually defines those at risk of decompensation. So we're between 9.8 and 11.3. Liver stiffness measurement of 24.3. If you look at most of the criteria, that's a 25, puts you solidly in a cirrhotic camp. If you have a platelet count of less than 150 in a stiffness measurement, more than 20, that's Baveno VI criteria for varices and that sort of thing. So this is a very -- what I call a very, very fibrotic group of patients. And the MREs kind of seals the deal. Anything more than 5 kPa cirrhosis, and you're at 5.74. So on the left, it's what you would expect. These people are older, 63, 2/3 of them are female. Again, females generally are the greater percentage of enrollment in these trials. That is even more so in advanced trial, advanced fibrosis. But look at the diabetes, 71% are diabetic. What was it in MAESTRO-NAFLD-1, do you remember? It was 48%. So kick it up a notch or a couple of notches, the 71. The ASCVD risk or remember from the -- from MAESTRO-NAFLD-1 is around 11 or 12. It's 16 here. So this population is not only more advanced histopathologically, but they are at greater risk of ASCVD as well. And just -- if you don't remember anything else from tonight, the ALT/AST ratio, when you see a ratio of 1, that's advanced liver disease. And here, it's 1. and platelet count kind of seals at 158. So as a mean, that's a relatively low number. And then the albumin tells us that this is a well-compensated population. So all in all, I think this kind of verifies that we're dealing with cirrhosis. Now this may be the coolest slide of the entire EASL. So here, we broke this down into people that had greater than or equal to 8% liver fat and those that had less than 5% liver fat and everybody in between. So 3 cohorts. Yes, the numbers get small, 31, 28 and 40. But look at what we see here. And the red numbers, I think, are the most important. But I don't think -- I think we should mention the PDFF. You go from 12.8 to 6.8 to 3.8 and FIB-4 goes from 2.5 to 2.7 to 3.7. In MRE, 5.5, 5.8, 6.1. MELD score 7.5, 8.0, 8.8, platelet count drops like a rock, 175, 152, 133. And now that really, really cool stuff, liver volume. So when you look at on the right, 2563. Normal liver volume is around 1,400 ccs. From the Phase II trial that we published in Lancet that really got us going in this space, liver volumes are around 2,300 and fat fraction on MRI-PDFF is around 17 to 18 in that trial. So 2,300, PDFF 17 to 18. But look here, -- if you put all these 3 together, the mean liver volume is around 2,300, the same as the Phase IIb non-cirrhotic cohort. But the fat is a lot less. So there's something more contributing to enhanced liver volume besides just fat, okay? Now you do see that as you go from a fatty liver cirrhotic to a non-fatty liver cirrhotic, there is a drop-off of around 400 ccs, but these are small numbers as well. it's still large though, even in the more advanced population. Now look at what happens to spleen volume. It goes from 484 to 546 to 649. And I think that's interesting. We know in 1 paper that has been published that's listed at the bottom by you in PLOS One that spleen volume is actually a good predictor of high-risk varices. In fact, the negative predictive value of spleen volume is around 95% for predicting the absence of lots of varices or high-risk varices. Now that's about the extent of our knowledge in the literature around spleen volume and liver volume. But it is quite interesting that you see a more advanced portal hypertensive population, as you lose the liver fat, and we can begin to really quantify that a bit. Now let's look at the first results. The first results I show you are FibroScan. Now in MAESTRO-NAFLD-1, if you're paying attention, I showed you that I picked a 30% change as the responder analysis. Here, we picked 25%. And we're toying around with both cutoffs because we don't really know the ideal responder analysis for FibroScan. Here, we pick 25%. And what you can see is at least 40% of the cirrhotic population achieved that 25% reduction change from baseline. It was more so in those that had less fat -- in fact, the largest reduction in VCTE of a mean 9 kPa was in the most advanced group of patients. And when you look at worsening, there really was no change between the low fat, high fat or low fat -- I'm sorry, no fat or any fat versus all. And then when we looked at MRE, we used the 19%, just like I did before because that's the QIBA analysis for coefficient of variance. And here, you see about 23%, 22% of patients improved or achieved that level of improvement versus half that number. You know what this looks like. This looks like the regenerate fibrosis response rate. It was 23% for drug. It was 11% for placebo or 12% for placebo and a treatment effect delta of 11%. That was statistically significant for the REGENERATE trial. And we see very similar, obviously, non-histopathologic but we see MRE data that looks -- that's what stuck out at me. Now this gets just more and more interesting. And I think it triggered in my colleagues, a lot of thought today around this idea of a change in baseline in liver volume and then a change in spleen volume. So just to walk you through this. On the left is MRI. Again, it's all patients. And then the middle red bar, it's those that don't have fat and then on the right orange, it's those that have fat. So 36% reduction in liver fat in all comers. Obviously, if you didn't have fat in your liver, it's hard to reduce it, 26%, we're able to take it down even more than 5%. And then in those that had more than 5%, 38% relative reduction. When we look at spleen volume, it's about a 20% drop. And that's at 52 weeks. It was around 16% at around -- 15% around 16 weeks and continue to fall at 52 weeks by another 5% or 6%. Importantly, 73% of these people, regardless of their baseline severity, had at least a 15% reduction in liver volume at week 52. So when we look at the middle bars, now we're looking at a responder analysis in spleen volume. Why? Because the variance in spleen volume is pretty significant. So we wanted to kind of put a specific change in there. We picked 10%. So here, around 1/3 of those patients, 37% had at least a 10% or greater change in spleen volume. Now here, there was a difference, although not statistically significant, I don't believe, between those with no fat and those with that as far as the percent that increased in spleen volume actually as opposed to decrease in spleen volume. So we can have a discussion around what that means, but I think it's very, very interesting. Maybe the data that's the most difficult to interpret, but quite interesting to me is the correlation between change in spleen volume and change in liver volume. And you see that illustrated on the X and Y axis with all the dots. And the correlation is very strong between liver and spleen volume change. So very quickly, just skimming over the liver enzymes. These reductions in liver chemistry test, ALT, AST and GGT, were similar across all patient subgroups. And then when we look at lipids in this cirrhotic population that has a high ASCVD risk, 16%. We see nice reductions in ApoB here, a 23% change from baseline and also LDL, triglycerides and lipo (a) is really down 42%. And some changes you see in body weight and systolic and diastolic blood pressure as well, and know that's not related to an alpha effect, which was a question that came up in the presentation today. So in conclusion, very, very excited that this population of well compensated NASH cirrhotic patients in dosing with resmetirom, safe and well tolerated. And I think that's understated. This is a cirrhotic population, 105 patients treated for a year, safe and well tolerated. And there was additional reductions in MRI fat, LDL cholesterol, ApoB, lipo (a), all the atherogenic lipids that we would measure. Furthermore, reductions in elastography, whether that's FibroScan or MR elastography. And yes, this is novel, it's innovative, and we don't really know what to make of it, but statistically significant reductions in liver volume by an average of 20%. Now we could talk more about what that means, but I also gave another presentation with HistoIndex looking at accounting for this change in volume and the change in steatosis and how that impacts what pathologists look at under the microscope. And it's pretty interesting data. Now limitations of the study include lack of a placebo control. We didn't have a control group here. But the good news here is that we think based on everything you see here above that it gives us a good foundation for MAESTRO-NASH outcomes, which, as you know, is a Phase III trial that's about to kick off, looking at well-compensated Child-Pugh A patients. Okay. Now we can get to the meat of why you're here.

So we're going to take questions. Okay. And Paul, if you want to help out with, that's fine. So Ritu, go ahead.

Ritu Baral from Cowen. A question for you and Dr. Harrison. As we look at the MAESTRO-NAFLD data and you factor in missed doses, the 2 months of potentially missed doses. And the liver volume, which Dr. Harrison, you highed highlighted for steatosis, but could also affect fibrosis reads, how does holistically everything that you've presented figure into your expectations for fibrosis for Phase III MAESTRO-NASH, that data coming up? And also, if you could address what your expectations are for the placebo arm of that trial, given that we've sort of consistently seen 20% one stage of fibrosis improvement, but the more recent trials have been creeping up for whatever reason?

But before I answer that long-winded, very thought-provoking question, [indiscernible]. All right. So -- so part of this is your question, I think, is a good one because it makes me think about everything that I learned at the meeting this week. And I'm reflecting actually on what Rohit presented from the semaglutide data with PathAI. And so what you saw with ordinary path is a placebo response rate that was hitting 20 or so in the cirrhotic arm, right? But when you applied PathAI to that, it kind of didn't change the readout on the drug-treated arm, but it dropped placebo down to a meaningful 6%, 7%. So there was no difference between the 2. And we've seen that over and over and over again now in presentation after presentation and in publication after publication, AI digital path kind of normalizes -- it kind of gives us what [indiscernible] ground truth for placebo response rate. And so I think when you begin to think about how we evaluate this moving forward, I think there is a role for AI digital path to help us have ground truth. Where do we really see placebo response rates being? So then your -- next question or your -- you remind me of like Ferris Bueller's Day Off where you had just like 1 question in 27 parts or something. So you asked about MAESTRO-NAFLD-1 changes we saw in liver volume in...

What you learned from liver volume and missed doses? How would you interpret what you presented today [indiscernible] everything you've learned today from the NAFLD -- MAESTRO-NAFLD fibrosis biomarkers, incorporate the missed doses, which you said was 2 months, incorporate potential liver volume changes on fibrosis rates [indiscernible]. What do you expect from the MAESTRO-NASH data?

It's hard to say what liver volume change means on fibrosis. I mean we know there's more to liver volume change than just steatosis change. What -- so to add to that, a little more color that I -- if you did a -- who came to the last presentation. Okay. So I got asked by [indiscernible], did the pemvidutide data have any liver volume change? And I said, "I don't know". I didn't know we did it. And I finished the presentation, you're at the back of the room and the Chief Medical Officer says, I forgot to tell you, we did measure and it's 20%. So 20% reduction with pemvidutide, where you saw a 90% regression of fat at 6 weeks. So it's not all about the fat, right? Because in ours, it's 30% -- it's 49%, whatever and a 20% change in volume. And other data have shown that it's not just the fat. There's something else to it. When you look at what it could be, to me, edema is what makes liver stiffer. It's -- it's associated with inflammation. So anytime you have an inflamed organ of any type, you have more water in it. And so if you're fixing inflammation, if you're reducing inflammation, liver stiffness generally goes down. Think of a hepatitis C patient that you get a FibroScan on and it's 15, 16 kPa, you treat their hep C, you show you cured their hep C 12 weeks later, you repeat the FibroScan and now it's like 6 or 7. You didn't like knocked fibrosis out of the park in 12 weeks. There's a lot of that change in liver stiffness that's due to resolution of inflammation, not even steatosis, just resolution of inflammation. But I don't think that we have data linking that to fibrosis per se. So I don't think I can say liver volume change gives me tons of confidence that we're going to magically have our answer on fibrosis. Having said that, other biomarkers, we did obtain that we did talk about today, give me more enthusiasm. And that's the change in MR elastography, which was significant between drug treated and placebo when we looked overall and then it was even more enhanced when you looked at those that had the 19% or greater response. So you're overcoming the coefficient of variance of the test and now you're telling me who had a response, and it was twice as good as placebo. So yes, this is a relatively mild fibrotic group of patients. But even in that group, we were seeing meaningful changes in MR elastography, which by all accounts may be the best -- the best noninvasive assessment of outcomes for patients and maybe the best noninvasive assessment of fibrosis that we have. So that's my answer.

Whoever 1, 2, 3 or 1, 2, 3. Go ahead Jay.

Jay Olson from Oppenheimer. Thank you for the summary of the EASL presentations and thanks for taking the question. I had 2 questions about the cirrhotic cohort of MAESTRO-NAFLD-1. Can you talk about how the reduction in liver and spleen volume might correlate with improvements in symptoms and patient's quality of life? And then related to that, did you see any improvements in portal hypertension? And do you think that resmetirom could be a potential treatment for portal hypertension secondary to NASH cirrhosis?

Okay. So we didn't measure portal hypertension. So I didn't -- that came up today in the Q&A at the meeting. We didn't do EGDs, we didn't do HVPGs, we didn't do HepQuant in these people, although the second cohort of patients that we haven't talked about today, we will be getting some functional assessment in those patients. So really, all we have is what I showed you, which was the very nice correlation with liver volume reduction and spleen volume reduction and an inverse correlation with improvement in platelet count. Now we were talking about this earlier. I'm not sure that you fully agree with what I said about what we do know about this, but I'll say it because I think there is some inferences. What we know about spleen volume is that some of the vasodilators and resmetirom is not a vasodilator. But when we talk about treating variceal bleeds, we use terlipressin in Europe. And we use octreotide and midodrine in the U.S., and those are vasodilators. And there is mixed data, but there is some data that you actually change spleen volume when you do that. So there's a correlation between stopping a variceal bleed and improving spleen volume. But sort of that's all hypothesis-generating type questions. So we're the first to show in a public forum like this that we've shown that correlation. So what that means? I think I would be foolish to tell you I know what that means. It's good, I think. It's not bad. But how good? I don't know. I can't correlate that to a certain HVPG improvement. We're not taking people from clinically significant portal hypertension to no hypertension, right? In fact, we probably don't have much -- or we don't have clinically significant portal hypertension, probably based on the cohort that we enrolled. We would -- we wanted to exclude those people.

So I would probably say something because I personally believe that HPVG doesn't work in fatty liver patients very well. And it's just a technique that is invasive and why use it. But what we do know is if you look at that third cohort, their platelet counts are lower and their spleens are bigger. This is perfect for portal hypertension. And there are other -- they have other indices that suggest increased portal hypertension as well with the liver function abnormalities, et cetera, et cetera. So I think it's fair to say they're more progressed. What does more progressed mean? It means more portal hypertension. And I think at some point, we'll have HepQuant data that may confirm some of this to measure HepQuants -- measure of portal hypertension. So I don't like invasive procedures to measure portal hypertension. That's not our plan. So we actually think this is a novel way that's very simple and straightforward to come up with a way of measuring change in portal hypertension that will be a lot more appealing to patients and be convincing, but we have to validate it. This is exploratory at this point.

There is a way to -- it's small numbers, right? I mean they were talking 30-something people in that based on liver fat less than 5% category. But you can take Baveno VI, which is a kPa greater than 20 platelet count less than 150, that's a high-risk category for varices -- and how many of those did we flip -- did we take and reverse, drop their kPa and actually raise their platelet count. And it's small numbers, but maybe that's a way we could get it some kind of inference of what we're doing.

And so just to say that, one, that group, that group that seems to have more portal hypertension, bigger spleens was 1 that dropped their kPa from 27 on average to 18 and had a 0.8 correlation with decrease in spleen volume and decrease in liver. So it is very early data, but I think pretty interesting.

I don't think we did PROs in that population. We didn't do [indiscernible]. Do we have any data yet? Okay. So you're asking a question about patient-reported outcomes, where apparently, we did the data, but we don't have -- we haven't mined that data yet.

That's super helpful. And maybe just one additional question. What do you suppose drove the placebo response on FibroScan? I know it's relatively low and the numbers are small. But were you surprised by the placebo response? And what do you think?

No. That's the coefficient of variance of this test. I mean it's just a -- it's all over the map. FibroScan is used for its negative predictive value. I don't really make a big [indiscernible] of FibroScan change. I'm much more enthusiastic about the MRI changes that we're seeing. But again, when you're combining the change in platelet count with the change in kPa, I mean there's data, Baveno has worked on that for a long time. I think in the combination, it makes more sense. When you look at a single variable like that, I don't make a whole lot of it.

Ed Arce with H.C. Wainwright. So wanted to continue, I guess, the discussion and the threat we're on. I suppose this might be a little early as we're just really discovering and thinking about the implications of what liver volume and spleen volume reductions might mean for fibrosis improvement, specifically within the F4 patients. But ultimately, all of this has to distill into what benefit this could have in terms of the overall risk-benefit equation that the FDA would look to. And so the question is, have you begun to have discussions about what this means with the FDA? Do they have a position yet on this? And what could future work look like to help improve the visibility to that?

So I think first of all, FDA is very careful about what data they accept and what they consider to be validated, and they look to their biomarker group to validate certain data. So they're going to tend to believe that what we're doing is exploratory, even talking about MREs or FibroScan in their mind, it's like you need a very well-established data set to get beyond that word exploratory. And so -- at the same time, they described us as pioneers, people in this space, and they're enthusiastic about us moving in a direction. And I think, Stephen, you can also comment on that.

To me, Ed, it really sets us up nice for the MAESTRO outcomes trial. We begin to think about the changes we're seeing in this open-label arm and how we feel that, that can help us enroll the right patients and get to an outcome quicker to show a difference between drug and placebo. So how much it helps us sway the FDA towards this population, I don't think it does a lot. I think it's more -- it's like the group we presented to today. All the KOLs you know, saw that data for the first time, and you could see little bubbles going off above their head going. That's interesting. And they're all racking their brains going, where have I seen this before? And they're all coming up empty because they haven't seen it before. So it's -- it's proof of concept. It's thought-provoking. It's innovative. And where we go with it, I think there's a lot we can do with it. But one of the best things we can do with it is take away from the fact that we're actually moving the needle in a positive direction in an advanced population of patients, where before that data we didn't know that we were doing a whole lot with cirrhotics. We hadn't studied cirrhotics outside of hepatic impairment, right? So that's just PK trying to figure out how much dose levels change in patients depending on the severity of disease. But now we actually have data that says, not only are we dropping liver fat, which we did, not only you're dropping ALT, not only are we improving MR elastography, but we're actually making structural changes to organs in the livers of these patients. And again, what that means it's a bit of a guess, but it's all good news, right? I mean it's -- these changes are not changes that we look at and say, that doesn't mean anything or that's bad. It probably means something good. If everything else is moving in the right direction, and we're seeing these changes at the organ level, it doesn't mean we're necessarily going to see changes at the biopsy level, but I'd advocate you don't need an ever biopsy as cirrhotic anyway. You just need to look for an outcome. So I don't have great -- I can't -- I know what you're asking for it. I just can't give you in a granular detail what this means relative to anything else that's in the published literature doesn't exist.

I just have one follow-up, if I may. Given that we're at the end of this conference, had lots of discussions. And in the thyroid hormone receptor beta class, you're clearly, the pioneer and the leader, but they're positive data invites competitors, right? And so I have a question because it came up in fact, over the last couple of days, this idea in both drugs from [indiscernible] Aligos, this idea that the PK variability of resmetirom is rather wide. I'd like to invite to answer sort of how you think about that in general? And then more specifically, what's the implication, if any, that with dose titration?

We would like to know the PK variability of Lipitor or rosuvastatin, a drug that's taken up into the liver so well. And actually, it's a disadvantage for a drug that's taken up in the liver to have a high plasma half-life. You will see variability. People have different sizes of livers, people have genetic levels of transporters that can be different. But we've found that it's safe and that we can use doses without titration very, very well. And as far as drug interactions, that was another thing that was broad. It doesn't have drug interactions, okay? It doesn't inhibit any [indiscernible] of consequence. And it's not -- its exposure is not affected by other drugs. So I think that it makes sense for people who are trying to follow on to an active program to try to find something to improve on. That's normal. But why would we be this far if there was a major significant liability related to something like that.

[indiscernible]. So I'll just speak [indiscernible] when you look at sex hormone binding globulin from among patients, the error bars are not very great. And it means that the drug dwells in the hepatocyte had relatively comparable concentrations from patient to patient. So it's kind of a bogus claim that there could be an advantage if you had a drug where the serum levels don't change. One of the issues you could have with that with a long half-life is you get a slow leakage across the blood-brain barrier, and you end up with hypothalamic beta-agonism, which will change your TSH levels that get you into trouble and not allow you to go above a certain dose. And if you look at, for example, turn, they only go to 10 milligrams, they can probably go further. But there already was a statement about variability of T3 and TSH in the patients that they looked at. So it may not be an advantage to have the drug circulating for long periods of time.

Yes. I mean we don't know that. We're just saying you can always find things to talk about that [indiscernible]. I think, Jon, before you start, I just wanted to answer one question you had earlier, and that had to do with when Stephen referred to that there was 2 months on average less dosing due to COVID in the double-blind compared to the open-label arms. And I don't expect you guys to remember this. But there was a drug kit supply, blister kit supply issue at a certain -- at the peak of COVID, which I think was December 2020 and -- and so the manufacturer of these kits, which we needed for the double-blind arms had a production issue, and that's why the double-blind arms had an issue with drug supply from open-label arms because it was open label, they just got bottles of pill and they didn't need a drug kit. So that was the basis of that comment. But go ahead with yours.

Jon Wolleben, JMP. Two for me and kind of [indiscernible] on the sex hormone binding globulin levels. I appreciate the subset analyses today. Have you identified what patients have better exposures throughout the studies that you've run so far? And then second question, do we know what happens when we stop resmetirom, do we have any follow-up when patients stop the drug and what happens to liver fat or any of these other noninvasive markers?

Yes. So I mean, as Paul mentioned, sex hormone. So this is a very high response drug. So we're -- we have a very few percent of patients with a low sex hormone binding response. This is like a 90% response drug. And the most common reason to have 2 or 3 patients or a handful of patients out of 300 with low sex hormone binding globulin is poor adherence to taking the drug. And I think you're going to see that in any kind of clinical trial. And this is very minor in this study. And it's just because we -- sex hormone binding globulin is so highly specific that you can see -- you don't have to measure PK, you can see that. So I don't think there's much variance at all, actually, with the response to sex hormone binding globulin.

And that's a follow-up. And a question about follow-up, if you go off the follow-up.

Oh, the follow-up. If you go off drug, we have a -- even in MAESTRO-NAFLD, for example, we have a 4-week follow-up and then they can go into the open-label extension whatever. We've never seen anything in the follow-up when they take a drug holiday or they came off drug during when the kits weren't available or anything like that. But we -- I think, in my view, I felt like the PDFF response was somewhat durable because when we just looked across the board at PDFF independent in the study, independent of whether they were on drug at week 16 or week 52 as opposed to trying to do a statistical correction, it really didn't make a difference. So we still saw pretty much the same effect. That wouldn't be true for lipids. So it's well-known that for statins or any other lipid-lowering drug, if you're off that lipid-lowering drug for a week, most of the effect would attenuate. That does not appear to be the case. So it's more durable with the PDFF. And I think that's been true of other PDFF reduction as well.

Tom Smith at SVB Securities. Thanks for the presentation, and thanks for taking the question. Just one on safety. I know there were blinded independent panels that were tasked with adjudicating cardiovascular and the hepatic at every 3 months. Can you comment on whether there were any cardiovascular hepatic events that were adjudicated as related to treatment of resmetirom arms? And then just on -- one on efficacy, on the fibrosis-related imaging and the biomarkers. I appreciate the cuts of the imaging data, but I don't think you talked about any changes on PRO-C3. Can you just give us any color on any changes that were observed there?

Yes. So I heard the first part of your question, and there was actually a mistake in the slides. So it was a DMC that met every 3 months. The adjudication by a blinded committee was going on for the entire trial. And it wasn't like a 3-month interval whatever. And there were no related either of those type of events, either a hepatic or a cardiac.

And then the PRO-C3?

And it was a low rate in all of the arms.

Yes. So the PRO-C3, so with PRO-C3, just to give you a global perspective on PRO-C3. Any company that looked at PRO-C3 like a year ago, they used an ELISA assay. And then they switched to Roche COBAS Amplicor assay. And so the ranges are different. So what you know to be true from ELISA as a mean norm or a predictor of fibrosis is not the same with the Roche Amplicor assay. So a lot of this has kind of had to be redone. So you might see a lot of companies holding back on the PRO-C3 data because like you get reported the ELISA data at baseline, you get reported the Roche data at end of treatment, and you can't make any inferences of that. So you have to have one or the -- you have to have ELISA beginning and end or Roche beginning or end. So a lot of that data has kind of had to undergo delay as it was being sorted out. Now I'll let Becky comment on where we are with PRO-C3?

I mean -- so first of all, I think there's another aspect of PRO-C3 and it was something that we commented on in our -- when we had our Phase II data that we showed that the biggest difference in PRO-C3 between placebo and drug treated was in our F3 population because they actually have relevant PRO-C3. We didn't have F3 patients in MAESTRO-NAFLD. And the one thing we learned about Roche, when Roche did the assay is they looked at -- it was a much more robust validation and they looked at lots of normals. And what we found was the overlap between normals and early fibrosis stage is almost complete. You cannot tell the difference between normal people with no liver pathology. And these minor elevations in PRO-C3 that you get with these very, very early fibrosis stages. And I think a lot of people are like, -- and it's literally the same with ELF and other -- some others of these markets.

And remember, we had a very mild fibrotic population. So I'm not sure alpha PRO-C3 is the ideal thing to be testing in that population. Probably more so from MAESTRO-NASH and for MAESTRO-NAFLD-1.

Andrea Tan, Goldman Sachs. Just a question for you. There were some comments earlier today about potentially even 5 to 8 years to see meaningful fibrosis improvement. Just curious if you agree with that timeframe? And then in terms of responder analysis for fibrosis, what proportion of responders would you consider to be clinically meaningful?

Well, so -- those people that are saying it takes 5 to 8 years to see fibrosis improvement. I think you're in an academic institution and they're waiting for tenure. So they need to say that to have a job. I heard that comment today actually by one of the panelists or presenters. And it's -- I don't agree with it. But again, it's just my opinion. It's not stated in hard truth or fact. Having said that, I can quote you a number of studies that I personally have done where I've shown fibrosis regression with drugs as early as 12 weeks. Let's talk about aldafermin, 12-week data. I did call the cowboy for doing liver biopsies 12 weeks apart, okay? But in that study, I published it in hepatology. I showed meaningful change in fibrosis as early as 12 weeks with every NIT consistently moving in that direction. Fast forward, efruxifermin, 16 weeks of treatment in a cohort of cirrhotic patients, 1/3 of them resolved cirrhosis or went from F4 to F3. Now you could argue that sampling variability. And maybe it is, but every NIT moved in the right direction to go along with that. So I don't think that fibrosis is so resilient that it can't move before 5 to 8 years. Having said that, I'll flip the coin and say that if I'm dealing with a very fibrotic F4 population, and as you know, cirrhotics are not all created equal. That's why you have the Ishak score, not just the NASH CRN, because we take a 4 and we say is it a 6 or a 5, right, because there's different levels of collagen deposition even amongst F4 patients. And so it might be that if you have an Ishak 5 or 6, it might take 5 years to show a meaningful change on an ordinal score. It doesn't mean you'll take 5 or 6 years on a continuous score like qFibrosis or PathAI or even if you correct for volume reduction or change in steatosis, like we showed with HistoIndex. But they're quoting data from hep B and from hep C, showing that it takes a while on liver biopsy to show meaningful change in fibrosis when you cure the disease or you suppress [indiscernible]. And again, it is data that's published, and it's based on ordinal scoring systems. It doesn't mean that if you have an F2 NASH patient and you completely eliminate the drivers of fibrosis that you won't see meaningful ordinal change much quicker than 5 years, in fact, much quicker than 1 year. So I don't think in any way, shape or form, this meeting should dissuade your opinion or thought process around how quickly you can resolve fibrosis. You shouldn't take away anything negative about the fact that MAESTRO-NASH is 1 year in F2, F3, because I still think we can see meaningful change in that cohort in that shorter period of time. Remember, go back to the Lancet paper that we published the Phase II data in. And if you look at the F3 population, and when we apply a continuous variable to fibrosis change, 47% of the F3 population had improvement in fibrosis and none of the placebo did with 36 weeks of treatment with a relatively lower dose of resmetirom at 60 milligrams for the most of the patients compared to 80 or 100 with 12 more weeks of treatment in MAESTRO-NASH.

[indiscernible] you also have shown bariatric surgery results in NASH patients where, yes, at 5 years, you have much more fibrosis regression. But at 1 year, you had a significant change from [indiscernible].

That's right. Yes. The one thing that there's -- that I always point out on that particular study is there's 5 cirrhotic patients that don't really change over that 5-year period of time. And again, that's my point about we shouldn't be biopsying NASH cirrhotics anyway, but we have -- we can get them to an outcome much quicker. I mean we can show that we're preventing progression to an outcome and that sort of thing. It's the non-cirrhotics that I think the real issue is in how quickly can you resolve fibrosis. And even the comments I had with Novo Nordisk at this meeting, when you look at their Phase IIb trial, they didn't hit stat sig on fibrosis and they're arguing 72 weeks isn't enough. Well, I don't buy that. I mean when you drill down on their data, they're showing prevention of progression of fibrosis. And I think it's just the way the histology was read on an ordinal score.

Yes, I mean I think this is a very important point. I mean, if you think about a cardiovascular outcome study, you're doing a study in patients who have established cardiovascular disease and you're looking at time to a cardiovascular event. And what you're really trying -- you're not saying you're reversing their cardiovascular disease. You're saying, you're preventing it from progressing to that point when they have a heart attack. And in this case, that's what we're doing. We're slowing down the progression of their cirrhosis and their portal hypertension, so that there's [indiscernible] events or fewer variceal bleeds.

Yes. So you can have a patient walk into my clinic and I see clinic quite often. And a lot of my patients, I get referred or my 40 GI partners that don't want to deal with the complicated liver cases. And so I get a lot of what I call train wrecks. And many of them, believe it or not, feel fine. They [indiscernible] normal jobs, but they've been told they have cirrhosis. And I ask them, how do you feel? Do you feel okay? Yes, Doc, I feel fine. No fatigue? No fatigue. I work hard every day. No mental status changes. I'm not slowing. I feel good. I don't turn yellow, and their biochemistries and their liver function tests are all normal. If I could keep that patient there for the rest of their life. Yes, I need to screen them for liver cancer and for occasional varices. But if I could keep them there from a quality of life perspective, they're very happy with that. I don't need to show that I took them from an F4 to an F1 or a F2. I just need to prevent them from getting worse? Would it be great if I could regress it? Yes. But from a patient quality of life perspective, I just need to keep them where they're at.

[indiscernible]

[indiscernible] always agrees with me. No. I think the FDA has -- they've been a little -- and to their defense, they haven't been approached with drugs that really move the needle significantly. The only one that they've been presented with is obeticholic acid, which has an 11% treatment effect delta on fibrosis, and it comes with some baggage that gives them a relatively low therapeutic index, and so they didn't get approved. I think the drugs that are in the pipeline behind that, including resmetirom, have a much better shot on goal, improving histopathology. And I think mindsets are being willing to be mended and changed. If you go back to NASH-TAG this year, where we did the fireside chat, I mean I had [indiscernible] on the phone, going, yes, we'll listen. We want to be adaptive. We want to -- we're interested in listening to alternatives. And to me, that's a little bit like how many psychiatrists does it take to change a lightbulb, one, but the lightbulb has to want to change, right? So I mean, with the FDA, they have to be willing to listen and willing to adapt. And I'm hearing that is -- they're interested in that, but they need -- I think they're handcuffed by bureaucracy and by -- they have a data-driven approach. So we need to generate the appropriate data. And I think when you think about resmetirom, the slide that we always show first is the totality of all the studies that are being done. I mean who else has 3 Phase III trials to support the data, hepatic impairment studies, open-label extension arms in cirrhotics, a Phase III trial about to kick off in well-compensated cirrhotics, the safety trial we just presented today in over 1,000 patients treated for 1 year and a registration trial. So that REGENERATE didn't have that. [indiscernible] aren't doing that right now. So I mean it's kind of -- this is a huge amount of data that's being collected and kudos to Madrigal for ordering almost every noninvasive test that's out there in the market today where we can begin to mine this stuff collectively. So I think change -- they're willing to change. We need to present them data and we're collecting the data. It's just a matter of time to get it, mine it, refine it and present it.

So -- so do we have any questions that [indiscernible]?

Yes, we do. And 1 question that I've received from multiple participants online is regarding the missing dosage in the MAESTRO-NASH study. Does it resemble that of MAESTRO-NAFLD study?

Yes. So as we said, we already presented this that it does not. We said that -- we said this, I think, in January. MAESTRO-NASH did not have a blister kit shortage. And every effort was put in place to make sure that the drug supply was protected for MAESTRO-NASH, and there were no significant issues at all. And the -- in addition, the enrollment and the continuation rate are exactly what you'd expect from serial liver biopsy study, a very low dropout rate, a very high completion rate for the week 52 liver biopsy.

Very good. Thank you. First question from Liisa Bayko of Evercore. There are actually 2 questions, both of them related to liver volume. First one, which I think was already addressed on clinical implications of liver and spleen volume reduction.

Yes. We spent a bit of time on that. So I think...

As I just wanted to read the question that was submitted. And the second one is related to necessary correction for fibrosis data for reduction in liver volume. Is that necessary?

Is it necessary? Not if your drug works really well. I mean you can see large changes on ordinal scoring systems. I think where it becomes a challenge is where there's shades of gray. Remember with ordinal scoring, I always like to use this example. Let's say you have fibrosis that really is on a continuous scale, 2.9. Well, the pathologist is going to read that as a 2. And let's say after a year of treatment, it improved to 2.1 on a continuous scale. The pathologist is going to read that as a 2, right? So there wasn't enough collagen change to take you from a 3 to a 2. It was still in the 2 range or maybe it's in the 3 range, right? So it's the shades of gray where you're not moving whole points on ordinal scores where I think AI pathology, AI digital path and maybe applying corrections will help us.

Yes. And I would also say that fibrosis is not just how concentrated fibrosis is, it is a little bit that way for the earlier stages F1 and F2. But the F3 stage, where you have the bridging architecture that's been -- well been used that term has been used for 50 years. I mean this is a really well-known principle and the F3 response was there with no correction with resmetirom. So I think, Stephen, we're agreed on this, and we will have in addition to pathologist reading our biopsies, we will have AI to help support the analyses.

The other thing is that the majority of the patients are F3. So the majority of the patients are F3 here in Phase II, which was not large enough to find a fibrosis effect anyway unless it was a miraculous effect, there were -- the percentage of patients who were at F3 was much lower. F3, you see the effect without the volume correction. But a rational FDA person, there may be one, would look at the AI data that we have and understand the significance of it. So we don't need it for F3s, but for F2s, it would be helpful.

We don't need it, but I think it's additive for sure. And again, just speaking to derisking the study as much as possible. I mean you're getting -- you're getting everything that could be brought to bear to help adjudicate the results of the trial being done.

And one last question. Were there any stat sig differences between resmetirom and placebo for any of their safety metrics?

No. I mean that when you look at the table on adverse events in the MAESTRO-NAFLD-1,it's almost identical. There's no differences between groups for treatment-emergent adverse events. There's no difference between groups for grades of severity for the number of severe treatment emerging for SAEs, for severe treatment-emergent adverse events. The only numerical difference was in discontinuation rates, where it was 5, 6 and 2 between 80, 100 and [indiscernible]

Numeric differences.

And there's no signal that there's something that's rare that's going on in the -- that's what the question deals [indiscernible] there's no cases of pneumocystis carinii or something like that.

Yes, there's no repeat AEs that are unusual [indiscernible].

As there are actually additional questions. Maybe we'll take a couple of questions from Yasmeen Rahimi from Piper. Her question is -- can you help us understand how we can take the MRE data and other double-blind randomized portion of the NAFLD study, and predict histological responses in fibrosis?

I can't really help Yasmeen with that. I mean, we don't have a great deal of data around change in MRE and improvement in histology. Alina Allen, probably at Mayo Clinic is the best at reporting data, speaking to this. And again, we think that the change of more than 19% or 20% by MR elastography, gets you over a coefficient of variance threshold. So anything above that change is a meaningful change. But linking what a delta treatment effect delta is above that to a one-stage improvement in fibrosis, we really don't have enough data for me to stand up here and risk my reputation. I'm telling you that there's a 40% chance that you're going to have a histopathologic response. I just can't do that. To me, it's promising data, it's encouraging data. It's moving in the right direction relative to what's happening with placebo. There has been nothing that we've reported with resmetirom that is negative. I think that speaks volumes. How many other trials have you seen reported where you see a little bit of this, but some of this? A little bit of this, but I'm not sure about this, right? Or if you drill down on one particular variable and you get down far enough, you find something that makes sense. No. I mean everything we've reported with MAESTRO-NAFLD-1 has been positive data, safety, tolerability and every secondary endpoint we looked at hit statistical significance.

I think you've already answered your second question that -- about the cirrhosis cohort that does the data give you confidence that resmetirom would reverse fibrosis in cirrhotic patients, which, to some extent, you answered it, that everything is pointing to the right direction.

Right. So there's no -- as Stephen and I would also say there's no correlation with any change in fibrosis noninvasive testing with improvement in fibrosis on biopsy because there's just not much data. Now there's more data that suggests that worsening of some of these measures correlates with progression, but they're just simply -- the data don't exist at this point.

But logic would suggest that if you got that much better, you would see an improvement.

Yes but remember, but I mean let's try not to overread the data. I mean what we're seeing in the cirrhotic trial is very encouraging. It's not a randomized arm. There's not a randomized arm, but I don't know what placebo would have done in that situation. We can hypothesize that they wouldn't have changed any. But you're asking me to give you an answer that I just can't -- I mean there's where the plank ends and where I'm kind of walking on. I've spent too many years in the military to know better than to do that. So we need to understand the limitations of the data that we're dealing with and get excited about what we are showing.

Very good. Perhaps we take one more question. It comes from Mayank Mamtani of B. Riley. It's related to ALT data. What does the ALT data overall in cirrhotic and non-cirrhotic tell you about the drug -- what the drug is doing on advanced liver disease patients?

Well, again, when we look at the data, we don't have a lot of patients with elevated ALT. I mean ALT was 40 in that cohort. AST was 39. So we're not dealing with a very high mean value, but we still saw meaningful reductions in ALT, AST and gamma GT. I think it's very, very positive. But again, I don't -- I can't link that to a number, right? I would prefer to link it to other positive data sets like change in MR elastography or positive changes that we've seen in other noninvasive tests. It's to me, any one of these drugs you look at, it's the [indiscernible] IT response that gives me the best probability of success, not just one, not just ALT, not just MRE, but where you're seeing multiple of the same thing happening to PDFF response, right? We know PDFF response of 30%, we can predict what that means on NASH resolution based on data Rohit mined that included resmetirom data. So we have PDFF responses in the cirrhotic population as a mean that met that threshold for improvement. So I mean it's PDFF, it's MRE, it's ALT, it's all of those that begin to exude a level of confidence that we're doing something. But keep in mind, this trial, what we reported today is all about safety and tolerability. Can we begin to get some secondary endpoints that give us confidence that MAESTRO-NASH is going to be positive? Yes. But that's really the extent of where we are with that.

Very good. And then one last question was on the weight loss for the cirrhotic cohort. Any data available on that?

There was no weight loss. It was 1.5%. That's not significant. It was consistent with all our other data.

Very good. Well, thank you, everyone.

Yes. So we ran over, but I think we had a good discussion. And thank you all again for coming. And if the MAESTRO-NASH results go the way we are expecting [indiscernible] we'll move very quickly to get our Subpart H [indiscernible] prepare for the future. Thanks very much for coming. Appreciate it.