I'm Lori DeLev, and this is our speaker, Dr. Jack Utrecht, who has been a major force in the area of idiosyncratic drug-induced liver injury for, well, he's only 10 years old, so for at least two months now. And our introducer will be Neil Kaplowitz, who has also been a major person in the area of drug-induced liver injury for a decade or two or three or four. Good morning, everyone. My name is Neil Kaplowitz, and I'm very sorry that I could not be there in person to introduce the speaker, but I am here to introduce the High Zimmerman Hepatotoxicity Lecturer, Jack Utrecht. This lecture honors the memory and contributions of High Zimmerman, who was or is considered the founding father of the field of drug hepatotoxicity, which we now commonly refer to as DILI. Today's speaker, Jack Utrecht, is a truly innovative scientist and leader in the field of immunotoxicology. Jack is a longstanding professor of pharmacy, medicine, and pharmacology at the University of Toronto. He is the recipient of numerous awards and honors, has demonstrated a great commitment to mentoring and education, having mentored many students and postdocs, and he has published more than 200 peer-reviewed papers in the field of which he will speak. I'll tell you a little bit about his remarkable educational background. He received his M.D. from Ohio State University, his Ph.D. in organic chemistry from Cornell University, fellowship training in clinical pharmacology at Vanderbilt University, and internal medicine training at the University of Kansas, and he is a board-certified internist. His unique training that I just described has provided him with a skill set that has enabled him to study the metabolism of drugs, the identification of reactive drug metabolites, and the identification of protein adducts of reactive metabolites, and then has used this knowledge to describe the consequences, which range from no adverse effects to biochemical and innate adaptive immune responses. He has created drug-specific animal models of idiosyncratic drug reactions in various organ systems, including the liver. Particularly noteworthy has been his demonstration of the role of immune tolerance and the potential pivotal role of overcoming the liver's immune tolerance in the unmasking of the progression from minimal transient liver injury to more severe and persistent injury. I join you in looking forward to his lecture and hand this over now to President of the ASLD and my good friend and colleague, Lori DeLev, who will welcome Dr. Utrecht to the podium. Thank you very much. Okay. So, Heisman had a major effect on the field of liver injury. I mean, how many people can say they have a law named after them? And I know that John Sr. was very proud of the fact that the FDA did not have to remove any drugs from the market after they started using Heisman's law as a way of excluding drug candidates that had an unacceptable risk. One thing I think he would agree with is that any hypothesis that you propose should be consistent with the characteristics of identity in humans. And that will be a recurring theme in what I say. So, in his iconic textbook on hepatotoxicity, he divided idiosyncratic DILI into two types. Immune idiosyncrasy in which there was some clinical evidence of immune activation. And the classic example was halothane. And then there was metabolic idiosyncrasy which lacked this evidence of immune activation. So, what's the evidence that iDILI can be immune mediated? Well, there's several lines of evidence. Studies that were done a few years after that book was published showed HLA associations between the risk of iDILI for certain drugs and specific HLA haplotypes. This hasn't been done for most drugs. It's not a trivial task. But the drugs for which it hasn't been done have very similar characteristics. Another important characteristic is the liver histology which looks very much like viral hepatitis with a mononuclear infiltrate and a predominance of CD8 T cells. In some cases, we find anti-drug antibodies, a positive lymphocyte transformation test, and rapid onset on re-challenge. That only usually occurs with serious iDILI. With mild iDILI, sometimes when you re-challenge, nothing happens. But with serious cases, it's much more likely to occur, sometimes within hours of re-challenge. And just the general characteristics. An immune response is an easy way to explain the idiosyncratic nature and the delay and onset of iDILI. So the classic example of immune iDILI is halothane. And it's unique in that it's given episodically along with surgery, and it usually doesn't occur on first exposure. And presumably what happens is on first exposure, you initiate an immune response, but the duration is not long enough to get an immune response that can lead to iDILI. But you often get fever with that first challenge with halothane, and then on second or subsequent exposures, you get serious liver injury. Often preceded by fever, eosinophilia in 20% of cases. There are antibodies that are found in these patients that recognize trifluoroacetylated proteins and auto-antibodies. Lance Pohl did classic experiments on this decades ago, and it was the first indication to me that convinced me that these reactions were immune-mediated. There's no HLA association, but halothane isn't used anymore, and so there are no cases in which you could test this. One of my students produced this slide. It's very complicated. I tend to be a lot simpler. So how does a drug cause iDILI? So in most cases, and I don't have a pointer here, I don't think. Well, I can't point to the screen anyhow. The mouse could do it, but the mouse is gone. Anyhow, drug forms a reactive metabolite, covalently binds to endogenous proteins, and that serves as a neoantigen. And I already mentioned that the histology and presence of antibodies suggest an adaptive immune response. And to get that adaptive immune response, you need a specific T cell that recognizes a drug-modified peptide in the context of an HLA-presented molecule on a dendritic or other antigen-presenting cell. But to get that, you need some sort of cell stress to activate these antigen-presenting cells. And the hypothesis is that most patients treated with drugs that cause iDILI have that innate immune response, even though most will not develop liver injury. Excuse me. I have a 4-year-old grandson who just started school. I didn't get COVID, but he certainly gave me a great cough that I'm getting over, but I still have some cough. So the dominant immune response in the liver is immune tolerance. And involved in that immune tolerance, some of the molecules are PD-1 and CTLA-4. And so we reasoned that even though the binding affinity is not high enough in most cases to lead to an adaptive immune response that will cause iDILI, if we suppressed immune tolerance, we could unmask the ability of a drug to cause iDILI. And I can't point, but if you look at the slide, treatment with amediacin to PD-1 knockout animals produces mild injury at about three weeks, and it resolves despite continued treatment with the drug. Very similar to what we see often in humans, because mild injury that resolves despite continued treatment is much more common than serious iDILI. But if we add anti-CTLA-4, then we get a stronger immune response that persists. It doesn't lead to liver failure, but it does lead to an increase in bilirubin. So it's sort of a high law case. Now, does this represent the mechanism of iDILI in humans? And I think the answer is yes. These immune checkpoint molecules, when used clinically, increase the risk of co-administered drugs. If you look at the histology, I'm not a very good pathologist, but my colleague says this looks exactly like iDILI in humans. And we see that if we deplete CDAT cells, it's protective. So the injury is mediated by CDAT cells, and we think the same thing is true in humans. So I think this is a good approximation to what happens in humans. And it isn't just with amediaclin. We see a similar, although not quite as severe, results with isoniazid, nivirapine, and it also differentiates troglitazone from piaglitazone. We use the same dose for both. The clinical dose of piaglitazone is much lower. We use the same dose. Troglitazone caused liver injury, and piaglitazone didn't. And I wish I could point to the appropriate bars, but I can't. What about metabolic idiosyncrasy? Well, there's actually no good evidence that there is such a thing as metabolic idiosyncrasy. The iDILI associated with many drugs appears due to a reactive metabolite, but with the possible exception of isoniazid, genetic polymorphisms in drug-metabolizing enzymes have not been found to be significant risk factors for iDILI. And even with isoniazid, it's a very minor risk factor. Metabolic idiosyncrasy isn't really a mechanism. The mechanism that was proposed is really cytotoxicity, and if that were the mechanism, we should be able to overcome it by manipulating drug metabolism and develop animal models that way, and that's never been done, at least not a type of injury that has similar characteristics as human iDILI. What about isoniazid? Well, it's true that it's not usually associated with fever or rash, but most immune-mediated pathology is not associated with fever, rash, eosinophilia. I mean, multiple sclerosis, even viral hepatitis is not usually associated with these sort of characteristics. And I think the idea that it represented, metabolic idiosyncrasy, dates back to some early studies out of the B.B. Brody lab where they treated rats with isoniazid and got liver injury, but it was acute toxicity mediated by... And the studies were really quite nice, but the characteristics of injury in that model were very different than clinical isoniazid iDILI. Willis-Mandry published some cases in which patients with severe isoniazid-induced iDILI had a recurrence within hours of re-challenge. So even though you can usually, with mild injury and starting with a low dose, not get serious injury, if you start with a full dose in somebody who's had a severe reaction, it can occur within hours. The injury is associated with a positive lymphocyte transformation test. Dean Nesbitt has done some very nice studies in this regard. We did a study in which we took patients who had a positive TB skin test and were being started on isoniazid and followed their ALT over time and also phenotyped their T cells and found that in only those patients that develop mild injury, there was an increase in Th17 cells, which are inflammatory, and also T cells that produce IL-10, which are anti-inflammatory and presumably prevent the patient from developing severe DILI. In collaboration with Will Lee, we looked at the serum of patients who had liver failure and we found antibodies that recognized isoniazid or various P450s, so very much like halothane. And finally, as I said, it induces mild liver injury in our impaired immune tolerance model. But there isn't an HLA association, and why is that? Well, with abacavir, the parent drug binds to a specific HLA, and if you don't have that HLA, you won't have an immune response. With most drugs, the reactometabolite binds to a couple proteins, and so the HLA association is weaker. But with isoniazid, it's a hard electrophile. It binds to lysine, and this is an immunoblot of the proteins modified by isoniazid, and it basically binds to everything. And so it's much less likely that there would be a strong HLA association in such a case. And 20 percent of patients treated with isoniazid will develop mild liver injury. It's been proposed that inhibition of the mitochondrial electron transport chain is responsible for idyllic and culturally reported a study in which he took a combination of rotenone and isoniazid at sub-toxic dose, but combined, it was toxic to hepatocytes. I'm always nervous about killing cells in a test tube as a model for idiosyncratic drug reactions. But he proposed that isoniazid, ideally, is due to underlying abnormalities in mitochondrial function. Well, we tested this in vivo, and the combination of rotenone and isoniazid promptly killed all the animals. So we decreased the dose a bit until the animals survived, but we didn't see any evidence of liver injury. And then we went to the impaired immune tolerance model, and we see liver injury, but it's not increased by rotenone. How about in general? What's the involvement of mitochondria in idyllic? Well, I think for valproic acid-induced idyllic, there's clear evidence of mitochondrial involvement. The characteristics are different. Infants are at much higher risk. Dehistology is usually dominated by microvesicular steatosis, sometimes hyperaminemia. Patients with Albers syndrome, which have decreased mitochondrial DNA, are at much increased risk, and some variants also may be at increased risk. So valproic-induced idyllic may, in fact, be an example of metabolic idiosyncrasy. I don't think there's enough evidence to be confident. Liver injury caused by other drugs, like acetaminophen, thiouridine, linazolid, also involve mitochondria, but they're not idiosyncratic. But I think the most important data to indicate that simple inhibition of the mitochondrial electron transport chain is not a mechanism of idyllic is given by fenformin and metformin. Fenformin strongly inhibits the mitochondrial electron transport chain and was withdrawn from the market because it often caused lethal lactic acidosis because of this inhibition. But it rarely, if ever, caused idiosyncratic dilly. Metformin, which replaced fenformin, is safer because the kinetics are much more predictable, but it can also cause fatal lactic acidosis, rarely, if ever, causes idilly, and it doesn't even increase the risk of co-administered drugs, which you would expect if that was part of the mechanism, like the rotenone experiment. I think that clinical data is much more important than the data that we produced in animals. So going back to the overall mechanism, again, it's presumably idiosyncratic in most cases because of this requirement for an HLA and T cell receptor. But most patients, as I mentioned before, should have an innate immune response because it doesn't depend on these factors. And if you look at early changes, this is RNA-seq data from an animal treated with nevirapine, and you see all sorts of changes. And years ago, I thought that such omics data would lead us to the mechanism of idilly. After a lot of money, I found that it really doesn't. It can provide some clues, but it really doesn't define the mechanism. I think a much better way to investigate the mechanism is to have specific hypotheses and try to test them. Most of the hypotheses will be wrong, but I think that's a much more effective way of trying to understand the mechanism. One of the important parts of the innate immune response is inflammasome activation. So I wish I had a pointer again, but you have DAMPs that bind to toll-like receptors, activate inflammasomes, that converts pro-CasPase 1 to active CasPase 1, and that converts the pro-IL-18 and pro-IL-1 beta to their active forms. And so to test this, we treated THP1 cells with clozapine, and in fact, it activates inflammasomes with the production of IL-1 beta, and then you see the last green bar where we add a CasPase inhibitor, and that prevents it. In contrast, olanzapine, which has a very similar structure, also forms a reactive metabolite, doesn't cause liver injury or agranulocytosis. It doesn't activate inflammasomes. Now clozapine is fairly unique in that it's not only bioactivated by P450, but also by myeloperoxidase, and macrophages and THP1 cells have myeloperoxidase, and we see covalent binding of clozapine to the THP1 cells. But nivirapine is different. It requires P450 to activate inflammasomes, and if you look on the left-hand side, there's no activation of inflammasomes. The far right is a positive control. So what we did is take hepatocyte spheroids and incubate them with nivirapine, and take a supernatant from that and add it to the THP1 cells, and that activated inflammasomes. And it was inhibited by a CasPase inhibitor, inhibitor of sulfatransferase, because one of the reactive metabolites of nivirapine is a benzylic sulfate, didn't have any effect, but an inhibitor of P450 did block the effect. And I should have mentioned, can I go back one, we see the same sort of results with amediacuin, carbamazepine, troglitazone, again it differentiates troglitazone from pioglitazone. One weird finding was treatment with nivirapine led to lymphopenia, and we were trying to figure out what the mechanism of that might be. And one obvious possibility was cortisol, or the rodent equivalent is corticosterone. It's known that corticosteroids cause lymphopenia. And sure enough, nivirapine treatment leads to a release of corticosterone. Now does that have anything to do with the mechanism? I'm not sure. Then we looked, and again I wish I had a pointer, but we actually see covalent binding of nivirapine to the adrenal glands, and the two bars on the right with the highest binding are the ones that had the highest corticosterone levels. Again, what this has to do with the mechanism of ideally? It may be that this increase in corticosterone decreases the immune response and helps to prevent more serious injury. But it's too early to tell. I should mention that clozapine also causes a release of corticosterone. So what are the risk factors for high DILI? Well, as I said, the major risk factors appear to be HLA association and T cell receptors, but there must be other risk factors. Genetic factors such as polymorphisms in drug metabolizing enzymes and BCEP do not appear to be significant risk factors. As Heisman said, pre-existing liver disease or even other inflammatory conditions such as inflammatory bowel disease don't appear to increase risk. You would expect that they would and they don't, and I think that the reason for this is whenever you induce an inflammatory response, the immune system also produces anti-inflammatory molecules to help keep from an overreaction. And for all I know, patients with ulcerative colitis may actually be at decreased risk because of these anti-inflammatory molecules that are also produced. Other factors like the microbiome are being studied. They may be important, but I think we don't have enough information yet. Is it possible to decrease risk? Well, even though HLA associations can be important, with few exceptions, they're not strong enough to be useful clinically. Unfortunately, other risk factors just aren't sufficiently defined to be useful. The strength of the early innate immune response may predict risk, and I'll give you evidence why this might be the case in a minute, and a slow titration of the dose may induce immune tolerance that decreases the dose, or the risk as well, and we were able to show that with Sonobamate. So there's an interesting animal model of idyllia, and that's penicillamine. It's idiosyncratic. It doesn't occur in mice. It doesn't occur in most strains of rats, only in brown Norway rats, and even in brown Norway rats, it only occurs in about 50% of the animals. It sort of looks more like dress, because there's also a skin rash, et cetera, and when we looked at the early time points, 24 hours, there was a spike in serum IL-6, and it was a much bigger spike in animals that later went on to develop liver injury than in animals that didn't develop liver injury, and if you look at the right graph, which again I wish I could point to, the incidence of liver injury increases at about three weeks, and about 50% of the animals develop liver injury. If for two weeks we treat with a low dose of penicillamine, it totally prevents the reaction. If we treat for only five days, it partially prevents, but doesn't completely prevent. If we treat for two weeks and then do nothing for four weeks, and then treat with high dose, it totally prevents, so this is a long-lasting, I think, immune tolerance to the drug, and again with sinalbomate, a drug that caused dress in early clinical trials, I suggested that they start with a low dose and slowly titrate up, and in a study of over 1,000 patients, there were no cases of dress with that slow titration, and the drug was able to be approved on that basis. So finally, key takeaways, any hypothesis should be consistent with clinical data, and where possible, actually tested in humans. Clinical data strongly suggests that most, and there may very well be exceptions, ideally is mediated by the adaptive immune response, especially CD8 T-cells. The adaptive immune response requires an innate immune response, and that's unlikely to be idiosyncratic, and I think we can study the early steps in the evolution of the immune response. When you start taking a drug and develop liver injury after several weeks, it's not like nothing happens between when you start the drug and you develop liver injury. There must be an evolution of the immune response, and that's less likely to be idiosyncratic, and we should be able to study that. If you ask me to guess what might be able to predict the risk that a drug candidate would cause, ideally, I would say the release of DAMPs, especially in exosomes that can activate in flammasomes, but it's way too soon, and I'm probably wrong. But that would be my best guess at this point. I think that it may be possible with a better understanding of the mechanism to be able to prevent, ideally, but it's going to take a lot more work before we're in that position, and I thank you for your attention, and I'm sorry that we got a late start. I don't know if we have time for questions.

idiosyncratic drug-induced liver injury

immune responses

immune idiosyncrasy

metabolic idiosyncrasy

immune tolerance

risk factors

inflammasomes