Welcome, everyone, to February's Hepatotoxicity SIG seminar series. We are delighted to host Dr. Nagachala Sani, who is the David Krabb Professor of Medicine and the Vice President of Academic Affairs at Indiana University School of Medicine and Indiana University Health. He served as a director of the Division of Gastroenterology and Hepatology in 2020 and was the chair of the Department of Medicine from 2020 to 2021 at Indiana. He received his MBBS from Kakatiya Medical College in India and completed his internship and residency and gastroenterology and hepatology fellowship at Emory University in Atlanta, where he met one of my best friends, Jeff Kahn. His research is focused on non-alcoholic fatty liver disease and drug-induced liver injury, two highly significant public health problems. He's very well known to this audience for his work with the Dillon and his work in the field of drug-induced liver injury. He is a founding principal investigator, in addition to his work in Dilly, in the NASH CRN and is the PI for the HIV NASH CRN. And so he has a very unique perspective on drug-induced liver injury in the setting of patients who have chronic liver disease. He has authored 600 plus publications, and I really don't think I need to introduce him more thoroughly to this audience. Everybody knows who Naga is. We've all worked with him. His H index, for example, is 110. We all know his track record and his publication record. We're very excited to have you here, Naga, because you really bring a very specific and unique perspective coming from two very relevant aspects in the liver field to give this talk. So thank you. Welcome. Thank you. Really honored to be speaking. Some of the audience are the work that I've admired for the last 30 years or so. So really great to be here. Also just as a good exercise to put the slides together, refresh this area. So that's been pretty helpful. For example, as I was preparing, I was picking on Neil Kaplowitz and others for their input into some of this stuff that I'll be sharing with you. Okay, let me share the screen. Tell me I'm doing okay here. Wrong screen. Stop sharing. Share. Yep. Correct. Let me just turn the subtitles off. Okay. Once again, let me get started here. Thank you. These are my disclosures. I know this is not a CME event, but just for full disclosure, consulting agreements here, research support from some companies, I have no speaking engagements. This is the outline. I'll just touch briefly about the background of DILI. I think this audience probably have heard this multiple times, but sort of set up the talk. I will spend some amount of time on the changes in drug disposition in NASH and cirrhosis is an area that of some interest and also some relevance. We'll discuss whether mass LD is a risk factor for DILI. I think is DILI worse in patients with chronic liver disease and cirrhosis on data from the CRN as well, excuse me, from DILIN as well as from the Spanish registry. Finally, just want to touch on this very interesting experience that we have had with the medication that is selectively hepatotoxic to one type of liver disease, but has been used extensively at the same dose in another type of liver disease. Just the background, DILI, idiosyncratic DILI is rare, but when it happens, it causes a lot of impact. If you add acetaminophen, this is the most common cause in the U.S. for acute liver failure continues to be a big problem for drug development. And I think many of you know, in 2023, obeticolic acid stumbled with its NDA for NASH. I would think one of the important reasons was it's suspected of paratoxicity, continues to have black box warnings. We have seen Tolvaptan, obeticolic acid as black box. And as I said, obeticolic acid did not receive approval for NASH. One of the concerns was whether it was hepatotoxic in the 25 milligram dose. Two types, you sort of know the direct hepatotoxicity where acetaminophen, aspirin, niacin, you know, typically enzymes go up, hepatic necrosis. There are a number of other syndromes as well, sinusoidal obstruction syndrome, lactic acidosis, steatosis, hepatic dysfunction, sort of more mitochondrial toxicity. And there's sort of the nodular regenerative hyperplasia where there is sinusoidal toxicity. Whereas idiosyncratic hepatotoxicity, bulk of the talk is focused on this, is unexpected, rare, ought to be not dose related. I'll make a case that there is a dose relationship. There is sort of a threshold, which may really become important when you are developing drugs or administering drugs to people with cirrhosis and liver dysfunction. And the prototypes are listed here. Amoxicillin clavinic acid turns out to be the most common cause for DILI in the US and many of the Western registries. Isoniazid, troglitazone, I think is sort of really in 90s, early 90s, mid 90s, troglitazone, DILI requiring transplants, causing death. I think it would sort of revolutionize the field in the US. I think this is sort of what led to NIH funding the drug-induced liver injury network. And there was also another compound at the same time called bromfenac, not a type of analgesic that caused liver failure at the time. Idiosyncrasy, ideology is generally unknown. And I think, you know, I remember presenting in 2000, 1999, saying there are no good animal bottles. It's very hard to predict preclinical. Fast forward 25 years later, sort of continues to be the same, although some of the work that this organoid work that Johnny and Bob Fontana were doing, maybe for model systems to detect the DILI potential sooner than in the market. You know, interesting thing, DILI can present just about as any type of liver disease, hepatitis, hepatic necrosis, bland cholestasis, chronic hepatitis, NASH, you know, this nodular regenerative hyperplasia. Really important is if there are clinicians or trainees, when you're seeing a patient with newly diagnosed liver disease, you must keep DILI in mind. And that is sort of a key point that I would emphasize. I think, you know, eDISH, this was developed by late John Senior, and this sort of becomes a way of agencies, regulatory agencies, and even the industry looks at drug hepatotoxic potential of a compound on the x-axis is peak ALT, sort of in time's upper limit of normal, and the y-axis is peak total bilirubin times upper limit of normal. And if you have right upper quadrant is the high slot quadrant. And if there is an imbalance, that catches some attention. And for you to call somebody, a patient in that quarter, high slot, there must not be a competing etiology. Generally speaking, if you have one or two cases of high slot, qualified cases in the right upper quadrant, and the true high slot cases, meaning they've been thoroughly worked up, but there isn't a underlying, a competing etiology. For non-canceled drugs, one or two cases of high slot can pose major problem in terms of receiving timely approval by the FDA and other agencies. There are studies ongoing, though, is to modify E-DISH in patients with liver disease. I know I have been part of some discussions with the sponsors as well as the agency where A-DISH has been plotted with, instead of upper limits of normal, using the baseline values to see how this plays out. And also interestingly, this, the original E-DISH Dr. Senior developed with his colleagues was for hepatocellular type of injury. Now there is an interest, and I've seen FDA asking for a E-DISH that is plotting peak alt-fos to peak total bilirubin. You know, it's an interesting, but I'm still sort of not fully convinced the significance of it. We do know the original high slot, you know, if there is substantial hepatocellular damage with jaundice, you know, 10% mortality, we don't quite know the same sort of relationship. You know, even sort of the relationship between the hepatocyte damage to causing bilirubin increase, sort of have that some estimation. Don't quite know the peak alt-fos relationship with bilirubin and then how that translates into mortality. There may be some unpublished data agency may have, but I haven't seen it, but agency continues to ask for this particular one, especially when you are in people with underlying liver disease clinical trials. Now I'd like to sort of discuss about the relationship between dose and idiosyncratic DILI. This is Craig Lamert's paper. I was the senior author on it. This is now over 15 years ago. Craig at the time was a medical student, really done fantastic career afterwards, and he's a successful colleague here at IU. This was at a meeting Jack Utrecht had said every drug hepatotoxicity is dose related. You know, he said this thing about lack of dose relationships, you know, it's just not totally foolproof. So we've come back and started to look how we can examine this. So we looked, this is sort of a database type of study. We used top 200 drugs that are prescribed or marketed at the time, and we looked at the doses of those daily doses of those compounds and the risk for DILI, either ALT, jaundice, liver failure, death, and so on and so forth. And we adjusted for, you know, median number of prescriptions. Bottom line though, there seemed to be a dose sort of a threshold, at least in this paper, we thought the 50 milligrams daily dose had sort of the threshold. And turns out, Jack made this point and Dr. Watkins had made the same point, when you look at people who have gotten liver transplant for drug induced liver failure, very rarely you see compounds that are given at, you know, one, three, four milligram doses. I think there was serostatin, and there is, by the way, there is another drug which was surprisingly given in like a one to two milligram dose. This is for, it never got approved in the US, but in Europe for fibroid bleeding. And even in a small doses, like microgram doses, it caused liver failure causally. But other than that, I think there's, there seems to be this dose relationship. This was well received. Then we have done a follow up study. This is a subsequent paper Craig published. Now he is, I think by this time he is a resident at Emory. And people have said at the time, hey, if your drug has hepatic metabolism, it may have high risk, never really been properly studied. Using the same sort of database, we have done this. Dr. Bjornsson, at the time he was in Sweden, he was part of this paper. And we've shown that a compound, if it has extensive hepatic metabolism, we defined as 50% or higher, you're more likely to have the delay signal. Interesting. And then in the same paper, Craig sort of mixed the dose and hepatic metabolism. And this is sort of, if you have a compound that you take in at a 50 milligrams or higher dose, and then also it has extensive hepatic metabolism by far that has the highest delay signal and interesting work. And, you know, though, when you look back at the literature, there are several examples of, you know, dose and, you know, the risk of delay. There are several examples of fluconazole delay. And I think Dr. Chen from the FDA pointed out, as I was preparing for this talk, there are, you know, fatal cases of fluconazole delay that are clearly linked to the concentration. You know, they had renal failure, patients were put on 400 milligrams of fluconazole and acute liver failure, the concentrations were super high. And also there are several clinical reports of incident delay in patients who are on stable doses of azathioprine, etamoxetine or duloxetine. When the dose gets doubled, boom, the delay happens. Seemed like until that point, they were tolerating it, not even with elevated liver enzymes. I think in the delay network, at least we have a dozen such cases, you know, azathioprine, stratera reported, so did duloxetine, Cymbalta as well. This sort of gives a flavor, the things that are written in review articles and in the textbooks, idiosyncratic delays, dose-related, not dose-related, is sort of not entirely true. I think there is a dose relationship, but not the same thing like intrinsic, you know, which is very predictably dose-related. And this led to work from Dr. Chen, rule of two for signifying a compound for DILI. Once again, he used multiple sources. Here, the rule of two is daily dose, sort of evolved from the work that Craig has done to daily dose more than 100 milligrams a log, P3, more than three, which is sort of lipophilicity. And if you have a compound with both of these, it's given at more than 100 milligrams and log P is more than three. The DILI's liability, these are only marketed drugs. We have no idea what goes on, company is proprietary, but this is what's available in public. OX ratio is about 14 to have a DILI concern. Whereas if you don't have either of those, the risk is much smaller. Actually, there isn't a risk. And then also, I think they had a validation cohort. It predicted not the same strength of relation with the OX ratio is about four. And then subsequently, he has gone on to develop a rule of three. This is once again, Minjin Chen from the FDA. He used daily dose, log P3, and then also used the reactive metabolite formation. So this is sort of, if you are in the audience, if there are drug developers, especially toxicology, this sort of may help. But let me show one more thing here. This is a slide from Will Proctor, I don't know where he is, but he's sort of a brilliant pharmacologist and safety expert at Genentech. He shared this slide with me. Once again, looking at compounds that are marketed, I think these may be about 119 compounds, and the C-max of this compound at a particular dose, and then delist signal. What they found though interesting, there is a C-max of a flora, about 1.1 in one paper of micromoles. The other one, I think his paper is about 1.3. Once again, sort of giving this signal from available databases, sort of a, you know, translational epidemiology showing compounds that give plasma concentrations that exceed one micromole in humans presents a risk for delete, as well as other adverse events as well. So this is sort of the body of work to suggest that the dose is important, the dose or the concentrations are important. Why is this important for the talk that I'm giving? You know, in liver disease, there is just extensive alterations in drug metabolism, and can lead for the parent compound doses to be higher, concentrations to be higher, which may just make this relationship between concentration and delete far more prominent in people who have underlying liver disease. So our first experience with this dose thing is now goes back to about 15 years ago. Ravi Jullery was a postdoc at the time. We were studying the effect of high dose salicylate on endothelial dysfunction in HIV-infected individuals. I think this was, interestingly, when we have done a dose about 11 patients, four out of 11 had very high transaminases, including two patients were symptomatic, and everybody got the same dose, a gram three times a day. In people who had hepatotoxicity, their concentration was quite high. And in this dose, we were able to re-dose at a lower dose and complete the study. But this was sort of first instance where we saw, you know, this dose for those individuals was high, for whatever reason, led to higher concentrations. And this is the next, this is published by Min Jun as well as Poonam Misra from the FDA, a nice article in gastroenterology making, once again, this actually now talking about underlying liver disease, when the direct acting antiviral agents were developed, and really a lot of excitement, but some things were causing acute liver failure, especially in people with liver disease. Y-Q-Repak had, if you're giving this to, if you're giving this to patients with cirrhosis, it was, it had the DILI liability, acute liver failure. And in that particular, as you may or may not remember, I've forgotten why, what Y-Q-Repak has, but it has these three, ombicivir, provetoprevir, and ritonavir, and also a tablet of decibir. So, interestingly, the provetoprevir is thought to be the responsible compound in this fixed dose package, Y-Q-Repak, and in Child's B, the exposure to provetoprevir was high, is increased by 62% and severe hepatic impairment by 1000%. And, and authors made a case that the cases of acute liver failure were related to this exposure. This is sort of the first time coming from the agency side, the concentration relationship, and, you know, I've had some discussions with Mt. Vavagon. I think they see, you know, not, you know, have exposure to concentration in their experience as they evaluate the new drug applications, and I think they see instances, examples. The same sort of phenomena was evident with Zepatier, which is L-Basavir, as well as GrazoPrevir, and severe hepatic impairment increased the AUC for Grazo by about 12-fold. And so even the authors made a case, even the Zepatier hepatotoxicity was also exposure dependent. Then I think some of you know beta-cholic acid experience with, of course, marketing for PBC. This was also dose related, especially this approved dose 5 and 10 milligrams. If it was given to cirrhotic patients and decompensated cirrhotic, the risk of cholestatic delay was quite high. And then I think sort of same thing came up with, in the NASH program as well. The one case that became a major problem for the sponsor with DILI, it may have been a fatal DILI, the concentration was significantly high, like what you would expect in a child's class C patient. This is, you know, this is all circumstantial, you know, human work. You can't, it's hard to show cause and effect, except you do a randomized controlled trial. But a lot of things pointing to, in some instances, concentration becomes important and these concentrations can go up for some compounds if you have underlying liver disease. You know, just the, I think some of you may realize in NASH or MASH or, you know, this metabolic dysfunction associated liver disease also includes sort of, even in the alcohol space, there are profound alterations in drug metabolizing enzymes. This is a figure that I borrowed from Nathan Sherrington, really has done good work in describing these. You don't have to go into it just to show that the cytochrome P450, there is upregulation, downregulation, the same thing with the efflux and influx transporters and uptake. So these can affect the exposure. And, you know, some of the, just to continue on some of the work, this was when Dr. Rupalanchi, who's now a leader in the field, when he was a postdoctoral trainee, has done the work of showing NAFOD reduces hepatic 3A activity. Mind you, this is not just the message RNA or just the western blots here. This is actually the activity from liver biopsies using just the microsomes. And as you could see, as you have moderate, the bar graph there, if you have moderate hepatic steatosis, you know, 40, 50 percent, you know, the activity was down by about 75 percent. This is important though, is if you're giving a compound that is a hep 3 substrate, hep 3A substrate, you know, the concentration of the compound can go up quite a bit. And, you know, just show the continuing, you know, what happens with the drug-metabolizing enzymes in alpha-D-NASH. This is a paper that we published last year in Naturecom. In a, started with an RNA-seq, bulk RNA-seq and biopsies from NASH. I think we had about 93. And then we also included meta-analysis of data from the public and public domains in about five RNA-seq data sets, as well as a number of microarray data to show what is happening in NASH and in the expression of drug-metabolizing enzymes. Here, if you could see sort of the volcano plaque showing fair degree of alkylation when you plot against NAFLD activity score. And Nick Powell, who is a brilliant clinical pharmacology colleague, and when he plotted some of these enzymes, proteins, and alkylations, you could just see striking relationships. More importantly, what he was able to plot out is the impact on different drugs. And these are regression plots are adjusted for Bonferroni. And then the same thing with fibrosis. You know, once again, a long list of the genes that are dysregulated and strong relationship to hepatic fibrosis as well. And these are, you know, once again, is relevant to the number of compounds that one takes. You may just ask, you know, what is all this to do with DILI and, you know, chronic liver disease? And I think I'm trying to make a connection that underlying liver disease alters drug metabolism, may increase concentration, exposure, and in some circumstances, especially if the parent compound has hepatic toxicity potential, that may just cause, increase the risk for DILI. And in this paper, what Nick picked up was, one of the most striking things is CYP2C19. This is clopridogrel, is sort of the classic substrate for this. This is a meta-analysis that he published in this Nature Communication paper, showing that when you compare to NASH versus controls, there is, you know, nearly one, sort of a 1.13-fold lower expression of CYP2C19. And if you plot against advanced fibrosis, stage three or stage four, it's about 1.22 consistent. And I think he has some data, I don't think we included in this paper, for a number of other enzymes as well. So, you know, just to sum up, what are the mechanisms behind altered pharmacokinetics in cirrhosis? There are a number of intrinsic capacity, blood flow, shunting, protein binding, delivery of oxygen, transporters. And interestingly, even some of you may realize there is a small intestine has a fair bit of cytochrome P453A. Interestingly, in cirrhosis, nothing happens to gut 3A, but if you do shunts, that 3A virtually goes away. There was a paper that we published years ago, TIPS dramatically reduces cytochrome P453A and increases the Cmax and AUC quite dramatically for 3A substrates. So, you know, coming to sort of the core of the talk, you know, dealing patients with chronic liver disease, this sort of the million dollar question is, are individuals with chronic liver disease, are they at a higher risk for idiosyncratic delay? The second is that I think is an area of interest. I think that's sort of where I'm going in terms of making a connection with the exposure and concentration, et cetera. The second, I think for clinicians and drug developers and the scientists, if delay were to happen in patients with liver disease, does it have worse outcomes? And the third I'm not gonna touch is the causality. It's, you know, the instruments have been developed for people without underlying liver disease. So when you have people with underlying liver disease, you know, how does RUCAM, RACAM, those sort of things play out? I don't think that's clear either. If you look at the literature though, this is a really fantastic review article written by Jim Lewis from Georgetown and Jonathan Stein cites really well. These are listed here are the compounds. If you look in the literature, you will find the reports that these compounds have a delay potential, higher increased delay potential in people with underlying liver disease. I think many of us sort of accept isoniazid, rifampin, pyrazinamide, that combo, or at least isoniazid increased risk in liver disease patients. And I think this is worth scrutinizing, you know, whether this is the case or not. Heart, methimazole, methotrexate, I think it's been scrutinized and I think the verdict, the most recent data are, you know, methotrexate and the risk for fibrosis in people with underlying liver disease, I think the data are pretty soft. Convincing though is the tamoxifen. If you have underlying NAFLD or NACLD, and if you give tamoxifen, I think there are a number of well-documented cases of rapid progression. This is sort of a classically discussed paper in this context, almost 23 years old. This paper published in Hepatology made the conclusion that if you have underlying hepatitis B and if you get anti-TB treatment, your risk for DILI is significantly higher. And you can see here in a column here, if you have anti-TB treatment and you are a B carrier, your risk of suspected DILI was about 35% compared to significantly lower in two or control groups. Problem with this though is the DILI was defined as ALT 1.5 times upper limit of normal. I just think in today's world, we take five times upper limit of normal based on just the definition that Guru Aital has published. The second thing, interesting thing, the most episodes of ALT elevation were also associated with an increase in hep B DNA irrespective of hep E antigen status. I think today if we were to adjudicate these cases, I just think these at the most become possible cases. So I think this particular highly cited paper in today's world, I just don't think makes the case that underlying hep B is a risk factor for anti-TB DILI. And then the same sort of staff story with chronic hep C for heart therapy, when people started to use this highly active antiretroviral agents, there was just a fair bit of co-infection and there was this risk of DILI. But I think Ken Sherman and others have made a case that this actually is immune reconstitution and hep C is getting activated and replication of hep C and those sorts of things rather than heart hepatotoxicity. The last I spoke with these experts, they tend to believe that this is not necessarily a risk factor any longer. In terms of underlying liver disease compounds, risk of causing chronic DILI, these are the examples here. I think some are, I think most of us would agree. Tamoxifen, I've already touched on it. I think majority would agree if there is underlying fatty liver NASH, there is a heightened risk. There may be some geographic susceptibility. And I think the literature suggests that the Japanese may be at a higher risk for this progression. We haven't deeply studied it. Imidazole, it's stated in the textbooks, but not so sure it's been well scrutinized whether underlying fatty liver disease is a risk factor. Methotrexate, obesity, hep C, heavy alcohol consumption are thought to be risk factors and eroticon as well as tibidine. And shown in the bullet here, interesting, the two, Eli Lilly, a company doing quite well in a couple of miles from where I work, had two programs, both abandoned, in phase three, mind you, probably costed more than a billion dollars for a company. One was a novel glucagon receptor antagonist. The other one is a basal pegalyspro. Both compounds were discontinued because both caused hepatic steatosis and unexpected. You know, they were doing MRIs thinking at least for the glucagon receptor antagonist, thinking that it may show an improvement in fat as a NASH drug. Exact opposite showed up. Interestingly, in both these papers, I was a co-author on PNPLA-3, which is sort of a risk factor for mast cell D, sort of the dominant genetic risk, was associated with drug-induced hepatic steatosis with these two compounds. Interesting observation. Continuing this work, this is a paper that I published 20 years ago. Dr. Kaplowitz wrote a letter to the editor saying, you know, he picked up some nuanced problems with the design. Nonetheless, we showed in this paper published in Gastroenterology at the time, if you have elevated ALT at baseline, which many of these people with dyslipidemia have, and you get a statin, your risk of hepatotoxicity isn't any higher than liver disease controls not getting a statin. And this was an important paper at the time because we clinicians were trying to figure out what to do. We were getting a number of referrals. They had baseline elevations in ALT, elevations in baseline ALT. They had dyslipidemia, and they needed to get put on lower statin and so forth. And this sort of gave some comfort. Maybe we can try. And this paper is now followed by at least a half a dozen, including an RCT from Jim Lewis, showing statins are really remarkably safe. You may have a very rare John this case, but we're talking one in 250 to 300,000 prescriptions given. This is more recent. Craig again at IU and I published this. Very interesting though. Let me walk you through this. This is an epidemiological study. Our hypothesis was individuals with chronic liver disease are not at an increased risk for delay due to common hepatotoxic agents. We have a really great health information exchange. What we were able to do is we constructed a chronic liver disease group. These people had sort of a quasi-NAFLD. People had at least ALT two times, 45 or higher at least on two occasions. They didn't have B, have C, alcohol, or hypotension. And we had about 25,000 such chronic, sort of a suspected NAFLD group. And we had two control groups with normal liver tests. One was sort of less than 45 control group was rigorously constructed ALT less than 31 in men and less than 19. And then what we've done is we looked at suspected DILI in these cases and controls after receiving one of 10 most common drugs from DILI, like Augmentin, nitrofurantoin, isoniazid, number of planolones, et cetera. And they must not have this competing etiologies. Once again, mind you, this is a epidemiological study. Dr. Kaplan would call this a sort of a porous study design because there are a lot of gaps. Shown here in the middle column CLV, the risk of suspected DILI in people with CLD is about 2.1% from those 10 compounds within 12 weeks after they received any one of them. And the control group one and control group three is listed here, 0.4% versus 0.3%. And this turned out to be the frequency of suspected DILI in chronic liver disease was about six fold higher than control group one and control group two. And interestingly, this sort of tracked even for risk of death or liver transplant. Of course, the frequency goes down. That gave validation, right? That there is DILI in underlying liver disease, the risk is higher. So this was an interesting epidemiological study. What we found in the literature though, evidence, similar evidence, there is a South Korean study that was published subsequently. It was also in the same journal and letter showing that NAFLD diagnosis, that there was an association with the DILI, almost this IRR about 7.7, interesting design. And there is a old Italian prospective study also showing NAFLD was associated with the DILI by about an odds ratio of four. So all these things sort of land around four odds ratio and all these, mind you, are epidemiological studies in NASH, but I don't know how else you can sort of test this. Many years ago when I was picking Dr. Kaplowitz's brain about how chronic liver disease can cause DILI, he said there are a number of reasons. This is a paper, I think, maybe Dominic, this may be from your group, showing how NAFLD NASH can cause increased susceptibility to DILI, increased reactive oxygen species, you know, TNF alpha, ER stress. Clearly there is mitochondrial dysfunction and as I've shown, altered transporter expression. So a number of mechanisms. So I think there is a theoretical basis and I think we are seeing the epidemiological signal that NAFLD NASH may be a risk factor for DILI. Then now moving on to sort of, if you have underlying liver disease, what does that mean for outcomes? This is Marvin Gabriel, a brilliant colleague from here, shown in a paper that was published in Gastroenterology. Underlying liver disease is actually, interestingly, far more important predictor of six month mortality than malder albumin, almost five folds. And very interesting. Two sets of data. The bottom one is Maribel Lucena, really great friend and brilliant group from Malaga. A Spanish registry showing if you have underlying liver disease, liver related death is significantly higher. And the top box or table is from the Dillon group that we published about nearly a decade ago showing the same thing. If you have underlying liver disease, known underlying liver disease, all cause mortality is about three to four fold higher. Interesting though, generally speaking, you know, idiosyncratic DILI, paracellular DILI is what we worry about. That's the one that you sort of get the Heim's law liver failure. But in chronic, in patients with chronic liver disease, cholestatic DILI becomes equally important. I think this is what we have really learned from the obeticolic acid experience. In the PBC, obeticolic acid, you know, the cholestatic DILI is what's causing to sort of liver failure and that sort of that decompensation to ignore that amount of reserve is sort of causing fatalities. And this is looking back at the data from the 2015 Dillon data. This is in people, 90 people who had underlying liver disease. We saw cholestatic DILI actually had numerically higher overall mortality compared to paracellular, 25%. And this is something we're planning to collaborate with the Spanish registry to highlight once again. So cholestatic DILI thought to be less problematic in general population, but I think a suggestion that even a cholestatic DILI becomes important if you have underlying liver disease, cirrhosis primarily. What does this mean? This means that in a clinical development, when you're monitoring for DILI, not only you're looking out for hepatocellular DILI, but you're also looking carefully for cholestatic DILI. I think this is sort of the basis for FDA asking that E-DISH that plots, you know, alkaline phosphatase and total bilirubin. Just the last two, three slides, bear with me. This is sort of very interesting phenomena. In the paper that we published, Roger Pallante is the first author. Saroblitazar, we are developing, we are leading a program as a therapy for primary biliary cholangitis. In a phase two study, by the way, this is approved in India. Millions have taken for diabetic dyslipidemia. It's also been given for NASH. Dr. Davor Bhabhi can vouch for it. And no reports of DILI in the NASH space, dyslipidemia space, given at four milligrams. But in the trial we've done, I think we only had 36 patients, multiple groups, and we had four cases of biochemical DILI with the high AST and ALT. And I think all of these were probable or highly likely cases. Of course, thankfully, when the drug was stopped, it rapidly improved to three or in four milligrams. One is in a two milligram dose. The compound now has been developed at a one milligram dose at the moment. And I think they've enrolled close to 140 patients in their subsequent PBC trial. So far, no DILI signal, but very interesting. This is the first instance that I know where a compound is given safely in one type of liver disease, but a different type of liver disease, it may cause a better toxicity. And these are sort of the profiles. Look at the patient one. I think the ALT went up to 1,000. And fortunately, you know, the bilirubin was flat in all these no-highs-low cases. You know, 600, I think 500. So impressive, you know, sort of interesting. I'd be curious if anyone in the audience, you know, Chris, I know you have a fair bit of experience in this. I'd be curious to hear if you have seen anything like this. So to summarize, you know, liver disease is associated with profound changes in DNA expression and activity. I think I'm becoming convinced that concentration appears to be important for a better toxicity, at least in some cases. And then I've made a case to you, but I think most of you would agree that advanced liver disease offers serum concentration for some parent drugs. Therefore, in some scenarios, their higher exposure to a parent compound is a better toxic. And then, you know, whether it is the dose or, you know, the reactive metabolites or sort of the lipophilic, then that compound may become selectively a better toxic in people with advanced liver disease than in preserved liver function. I think this is sort of what we may have seen, PBC and obeticolic acid story. You know, I think they can tolerate five, 10 milligrams, but when it sort of, when patients with cirrhosis take it, you know, just same dose, concentration, higher exposure, you know, DILI signal. And then I've shown epidemiological studies suggesting mass LD may increase the risk for all-cause DILI, but clearly more evidence is needed. And I think most of us sort of, you know, clinically we know, but there is now evidence in the literature that DILI outcome is worse in patients with chronic liver disease and cirrhosis. You know, I have one recommendation though. There are clinical trialists here. I was discussing this with Dr. Skip Hayashi as well. In clinical trials, in cases of suspected DILI, you know, I think many companies do this structured evaluation for competing ideologies, close follow-up, but I think a blinded examination of the random serum concentration of the investigational product, I think is really important. I think FDA is revising the guidance, hepatotoxicity guidance from 2008. I think they will be recommending storing samples, but I'll go a step beyond. And I think it would be really important to get the concentration. This is going to help the adjudication committees as well as the Data Safety Monitoring Board. I think that is my last slide. And I want to really thank for the opportunity and also thank you for your attention. I'm going to stop sharing, Lily. Yeah, wonderful talk. Naga, very interesting topic that I know everyone in this field thinks about and talks about a lot. And I'm going to open it up to discussion. But one thing I want to clarify, because I know Jim has some opinions here too as a toxicologist. Yeah, I'm calling on you because you're texting me in the background and telling me all this history about dose response. And he's literally drawing graphs about dose response curves and texting them to me. So we teach, most of us teach about DILI, and we always discuss about this whole direct versus idiosyncratic dose response. So we usually discuss this threshold dose. So just to clarify, I think because of the HLA Association, the way you're thinking about the idiosyncratic DILI and the way the, I think, the correct way to think about it probably is that you need a certain threshold dose to generate enough antigen, hapten, whatever you want to call it, to elicit an immune response. Is that what you're saying, Naga? Possibly. That is possibly. Also remains the possibility, though, the concentration exposure and the reactive metabolites, et cetera, may be directly a better toxic. It's almost like this sort of, due to concentration, this intrinsic is sort of becoming, excuse me, this idiosyncratic is becoming intrinsic. And for example, I think with the dose doubling as a thioprene, they're on stable 100 milligrams, and you go up to 200, boom, the DILI comes up. And there's nothing to, I mean, they always had the HLA. So just even the HLA binding part, there may be sort of a critical threshold. But I don't have, other than these observations, I don't really have a model that connects all the dots. I'm not quite sure where to put the HLA. I haven't taught that in detail. Yeah, it's, yeah, right, that's the piece that kind of throws me off. But it's interesting, because when you see all these preclinical studies, I think there's a bunch of them from France, and Neil knows all about this, he talks about this all the time, where you see mitochondrial toxicity, direct toxicity from all these idiosyncratic drugs when you put them directly on hepatocytes or on mitochondria. So that's always been something I can't reconcile. I don't know if Neil's up for turning on his mic and commenting on that, but it's just, squaring the two is very, very difficult in my head. I don't disagree with you, but I'm just trying to understand it. I'll tell you, though, all these sort of preclinical in vitro methods, too, and all these, when you talk to really expert toxicologists who are really brilliant, for every instance, there is a paper suggesting there's a paper on the same thing saying, yeah, it doesn't make sense. So the rule of two and three sort of makes sense for my story, so I put it up. There are a number of papers, especially the proprietary datasets. People have shown, Chris and maybe Jim, you know, they didn't play out. So I think there is, the story is still evolving, but I think, though, based on robotic folic acid, other things we've seen, concentration, at least in some instances, when it is the parent drug sort of makes sense. Can I just say, first of all, I think, Naga, you did a fantastic job summarizing a very complex field, and, you know, I'm always struck, as Lily was alluding to, to the fact that, in general, drugs with idiosyncratic DILI liability cause some kind of direct stress, and that, you know, is presumably related, it could be related to effects of concentration, drug metabolism, reactive metabolites, but ultimately, the stress responses that occur also play into this question of DILI susceptibility and DILI severity. And so it's a very, you know, it's, you know, it's a very complicated scenario, and I, you know, I think, you know, many studies have shown that, you know, particularly at high doses, exposure in vitro to various DILI candidates, you know, verified candidates, induces hepatocellular stress, and it induces hepatocellular death in many cases. So, you know, there is certainly, whether the intrinsic toxicity is really just the fact that there's greater stress, which then, you know, activates injury-promoting mechanisms or adaptive mechanisms, which dampen the whole thing, becomes a very important issue. But I think it's, I think you bring up a very important point that there is this interface between intrinsic toxicity and immune-mediated liver injury that really are integrally related. I was wondering though, Neil, the idiosyncratic, maybe there are sort of the one where you have this hypersensitivity, like, you know, with the skin reaction that is very different than sort of this metabolic type related to dose and reactive metabolites. That is a continuum with intrinsic, wonder. Well, I don't know, but, you know, certainly in so many instances where, you know, HLA associations have been sought in idiosyncratic dilly, there is a clear risk oftentimes restricted to a specific HLA. So I think that's hard to deny. And although there may be, you know, intrinsic toxicity in some cases that the vast majority appear to be adaptive immune-mediated. So I have, Jay has a question and I have two toxicologists here. I have Mitch McGill and Jim Loyendyke. So Jay, you go first, and then I want to turn it to the toxicologists because they both have comments on this. Well, actually, Hy Zimmerman said just what Neil said, that drugs that cause idiosyncratic injury usually cause a very mild, direct toxicity. And I think this is what faces the FDA when they see a drug that causes transaminitis. Are they the drug that's going to cause the bad, so-called Hy's Law case? But I want to say, first of all, we really need to think about the cause of the liver injury. For instance, you talked about methotrexate and amiodarone. Those are direct toxins. Largely, methotrexate is very interesting toxicity. I don't think anybody quite understands what it is. I think it's actually vascular more than hepatocytic. But anyways, and then there's this type of toxicity that I have been stressing of indirect, where the drug doesn't harm the liver directly. It causes another effect, like weight gain. Drugs that cause weight gain can be accused of causing dilly because the patient develops a NASH-like picture after they gain 50 pounds on some antidepressant or antipsychotic medication. And that goes for hepatitis B. When someone, these studies of hepatitis B from Taiwan, I agree, I don't really quite believe them because it may be just hepatitis B reactivating for whatever reason on these drugs. When you alter the disease that you're treating, things happen. So, and it's very clear that people with idiosyncratic injury, let's say from azithromycin, if they have cirrhosis or significant liver disease, they don't suck, they don't tolerate the liver injury as well as a healthy person. It goes for hepatitis B and C as well. They are worse in people with cirrhosis and quite dangerous. So this goes completely, is completely compatible with what we know of liver disease in general. I'll turn it over to you, Lily, to ask other people to speak. Thank you. Naga, if you want to say something before. No, no, no. I think it's, you know, Jim made great points. So, and so did Neil Kaplowitz. So I look forward to, you know, Jim and others. So- I was just gonna jump in and just say that your comments on dose, right? It's like, sometimes we go in and we lecture to our students and we say things like idiosyncratic, you know, liver injury is not dependent on dose. And I'm sitting here thinking, I learned that everything was dependent on dose. Like, that's like this axiom that I, and I just, I loved the way you presented. I haven't thought about this in a long time since my PhD work, but just reminding us that it probably is connected to dose, but it's just how it's connected to dose is not readily apparent. And it's like obscured by what other susceptibility factors and that may vary from drug to drug. And I just, the takeaway for me was there's still a lot of work to do. So it was just a great talk. And yes, I was texting with Lily and throwing out ideas. It was awesome. Really great though, is I was in preparation. I was talking to Paul Watkins. Sorry, he couldn't join. And then even if it is concentration, I mean, how do we connect this increased concentration to hepatocytes? I mean, they're shunting and the sinusoids are like, they lose the penetrations and all kinds of stuff. And so I think there are just a lot of gaps. It's sort of interesting. I actually am here, Naga, by the way. Listened to the whole thing, thought it was great. Thank you. The only comment that I will make, if I can make one comment, is that when you get a vaccine, you don't get a robust response unless your arm is sore. And your arm is sore because there are things put in there to create inflammation. So you have to have, I think, or probably have to have inflammation in the liver in order to be eligible for a robust adaptive immune attack. And that naturally follows that if you could design drugs that didn't do that, you wouldn't have to worry about an HLA adaptive immune response. Now that's largely hypothetical, but the data that I look at generally supports it. Yeah, I think I've actually talked about this with you, Paul and Jack, that like when you want to produce commercial antibody, you can't just present the antigen. You don't just inject the protein and boom, you get antibodies. You have to stimulate an immune response. So you get something like Freund's adjuvant. And so it's possible that it's interesting to me because it's possible that that initial mitochondrial toxicity or something that you get damp release and that sort of thing is actually necessary for that to occur, for that adaptive immune response to then happen. But it's also interesting to me because I've always had this idea that like, well, because there's no immune system in vitro models, in vitro models are irrelevant for most idiosyncratic hepatotoxicity. But it seems maybe that's not the case. I mean, if you see cell death, mitochondrial damage in an in vitro model, from my perspective as a scientist, that might actually be important for these studies of idiosyncratic delay. And then it's also kind of interesting to me because it's odd as Naga presented that, you know, from Min Jun's work, for example, that strongly suggests that hepatic metabolism with P450s, for example, increases risk of idiosyncratic delay. And yet in patients with chronic liver disease, they typically have depressed P450s, at least many P450 isoforms and may still be at increased risk. So that kind of presents this possibility that this paradigm of haptic production and forming neo antigens and you get an immune response may not be as relevant as we always thought, at least with newer drugs. And so, for example, we've just published some data almost simultaneously with a group from the FDA showing that with TKIs, actually P450 metabolism, which was always thought to produce reactive metabolites with TKIs, it's actually protective. You actually get less liver injury with greater P450 activity. So I don't know, maybe it goes back to kind of what Jay was saying about there's actually something else. It's not just the drug itself, the drug is having some other effects. Yeah, I would just, the only thing I would add though, Mitch nicely said is somehow integrating the transporters into the cytochrome P450 story. I think Paul brings this up on multiple occasions and I think that's sort of an important, yeah, you know, what happens and then, you know, parent compound as opposed to metabolite, it's just complex. You know, a simple story would be tough. I have a comment from Christine Hunt. Christine, I don't know if you have any questions or you just wanted to congratulate Naga. Oh, Naga, I just wanted to say really brilliant talk. Oh, thank you. I particularly appreciate your deep dive and open-minded approach to look at a large swath of clinical, basic and epidemiologic research. And it really has helped elucidate for me dealing in chronic liver disease. Thank you very much. Thank you, really. Great work. Raul, you were mentioned a few times and your studies were cited. I don't know if you have a different perspective or how you think about this. No, no, I have to congratulate Naga as well. It was fantastic talk. Yes, actually he published the only clinical study that links, for example, NASH with a risk of chronic liver disease in clinical gastroenterology and optilly, sorry, in clinical gastroenterology and hepatology. But I was wondering if that in this, these studies are, you know, need validation as he mentioned. And also, it's difficult to identify it is NASH, you know, the condition that may prompt the patient to dilly or there are associated condition that are actually the risk factor, for example, obesity or diabetes type two, diabetes mellitus. So we need more, more, more research in the field. We need probably prospective collaborative research in order to clarify these points. But anyway, thank you very much. And I think it was an excellent, excellent talk. Thank you, Raul. I would actually think though is when I was getting into the dilly field, Hashim and El-Sarag had a paper showing in the VA database, diabetes was a risk factor for acute liver failure from drugs. And then a bunch of literature in the pharmacology space, clinical pharma space, obesity modifies, you know, the drug metabolism, PK, so on and so forth. I thought, I actually thought the NAFLD is sort of the unifying story. The diabetes causes NAFLD and the obesity causes, and then sort of some of the work we have done to show the changes in drug metabolism, et cetera. So I thought that direction rather than the comorbidity is somehow explaining, but totally agree. Somehow we have to figure out how to, you know, sort this out. Jay. Yeah, I think many drugs have both effects. They're a bit direct and indirect. And the statins are a good example of that. You know, they do cause transaminitis and it's dose related. And it's insignificant, but they also cause idiosyncratic injury. And you're right, it's very rare, maybe anywhere from one to 50,000 to one to 200,000. People who get a statin will get injury. And that's the problem. Idiosyncratic injury is very rare. And what usually people are looking at are just the transaminitis that are caused by these drugs. Let me say something about your PBC experience with the PPAR antagonist, I guess they are, right? Agonist, PPAR, alpha-gamma agonist. Yes, right. So the interesting thing was with methotrexate and PBC, okay, it caused the same thing. It caused a transaminitis, very acute, ALTs of the five and 600s. But the interesting thing at the same time, the alkaline phosphatase fell and we continued people on therapy and the ALT fell to normal too. And those were actually the responders to treatment. And I have no explanation for that. I wish I had done liver biopsies during the time to see what this release of transaminases was. So I think it's quite vague. And then in your slides too, the ALT-BAS was falling as the ALT and AST went up. Very interesting. What is happening? I have no idea. Excellent discussion. I have comments from Julio Gutierrez and Einar. I don't know if there's no more questions. Einar? Yeah, I agree with everybody else. It was a very interesting talk, but I was just wondering, I mean, we have all these cirrhotic patients. They receive all kinds of drugs without hesitation, but we very rarely see any acute liver injury. Why is that? Maybe it's more difficult. The causality assessment is more difficult, but in these registries, there are relatively few patients with cirrhosis. Is that correct in the Dillon? What's the proportion of- In Dillon, I think about 10% have underlying liver disease. I suspect cirrhosis is less frequent. I would think 2% to 3%. I think Spanish registry is about the same, right? Not very many people with underlying. No, you make a great point. People with Nash's cirrhosis, polypharmacy, I mean, at this rate, they should be walking around with Dillon all the time. We just don't see, right? Yeah, there are a lot of holes to this story. Well, you still need the HLA. I think that's, it's like, you need multiple things. It's kind of necessary, but not sufficient. Neil has this beautiful graph that I always use when I give talks. It's about the susceptibles, then there are adapters, and then there's those that get the injury. So I think that's the story. There's multiple layers, that whole Swiss cheese hypothesis where multiple things have to line up. All right. Once again, it's been fun putting the talk together and speaking to you all, you know, such distinguished audience here. So thank you. Thank you, everyone. Thank you, guys. Very nice to see you. It was great to see everybody. See you next time. Nice talk, Nadenh. Nice talk. Yes, congratulations. Fantastic talk. Thank you, Paul. And amazing attendance too. I'm gonna send the recording out. Linda will, ASLD will send it out. Uh-oh. I hope you can't remove the board. We can trim it, but it's great. The discussion is the fun part. Bye, everyone. Bye-bye.

Hepatotoxicity

Drug-induced liver injury

Chronic liver diseases

NAFLD

Cirrhosis

Idiosyncratic DILI

Drug metabolism

Pharmacokinetics

Dose-response relationship

Drug-Induced Liver Injury Network

Regulatory challenges