I want to welcome everyone. Thank you for joining us. We are one of the smaller SIGs at the ASLD. And I always wonder, there's a lot of interest in liver injury from drugs and herbal supplements. And I just always wonder, is it just because people haven't actually signed up? So if you're not a member of our SIG, please sign up. You'll get newsletters and information about educational programming. And the more of us there, the more presence we have, the more ASLD will give us time to do these great scientific sessions for you. I'm working on getting a roadmap for next year so that when you come to the liver meeting, you get a little brochure and know exactly which room to go to at what time for what session, like the Nash and fatty liver people do. So we're working on that. We're like the orphan. And then before I start, I just want to let you all know that right immediately following this session, in this room, we're going to have the SIG business meeting, meaning that we're going to discuss what's going to be happening next year at ASLD. And we're going to take suggestions for webinars. You can suggest yourself or other people you think are of interest to the group to hear from. So if you don't mind eating lunch a little bit late, if you stick around at 1230, we're going to talk about the program for next year. And we would really love everyone's ideas on participation. So as we're starting, we're going to start with Dr. Hufnagel, my co-moderator, a man who needs no introduction. He is the director of the liver disease research branch. At the NIDDK, his research portfolio is amazing. It spans many liver diseases from viral hepatitis, autoimmune liver diseases, fatty liver. But we all in this room probably know him best for his seminal role in championing drug-induced liver injury research and spearheading the Dillon Network to understand mechanisms of hepatotoxicity and increasing our knowledge on this topic. The title of his talk is Hepatotoxicity from Cancer Therapy, the Dillon Experience. Well, thank you for having me give this presentation. It's just 15 minutes. I'm not going to cover this topic very well, just to give you a little bit of an overview and learn how to use the slides, too. OK. Disclosures. I don't have anything disclosed, except that I work for the federal government. That's my biases. The data that I show you today will come from two major sources. One is from the Drug-Induced Liver Injury Network, and the other is from the website called Liver Talks. I supervise both of these. I'm not pressing the right button, am I? To the right. There we go. There we are. It's not. So drug-induced liver disease is a diagnosis of exclusion. We don't have an IgM, a hepatitis A antibody type test. And for cancer chemotherapy, it's very much a problem, because cancer patients have many reasons for having liver disease. I list them there. They also are usually taking more than one drug. They're taking many drugs, not just for the cancer. They might be getting three or four agents at one time, but they're also taking something for pain, for sleep, and all those things. In the Dillon Network, the cancer cases, cancer chemotherapy cases, the average number of other medications, condiment medications, is nine. Furthermore, there are so many anti-cancer agents, and they're increasing exponentially recently. Let me show you that. Let me grab this. All right, so here I show you data from liver talks, where we have basically all approved systemically available drugs approved in the United States. So all drugs that we might suspect might cause liver injury. And this is by their decade of approval. And you can see that there are more and more drugs being approved over the years. And more of these are for cancer. Originally, pre-1970, and in the 70s and 80s, about 10 drugs or 12 drugs were approved every year, only one for cancer. It's not a lot, right? That started changing in the 1990s. And it's particularly changed in the last decade, since 2010. Currently, at our rate, there's probably going to be more than 50 drugs approved per year. And quite a few of them, as many as 10, will be due for cancer. So they're one third of the drugs being approved today. So there are a lot of them, aren't there? Also, there are different types of cancer drugs, right? They're usually separated into three types. And they're not completely separable. There's some overlap. One is the cytotoxic agents that are directly cytotoxic. You can study them in cell culture. They're toxic to rapidly dividing cells. And I list the types and some examples there. Then there's the hormonal agents that don't kill cells. They turn off the production of estrogens and androgens and other hormones that the cancer is driven by. So breast cancer, ovarian cancer, prostate cancer. And the third are the exciting new drugs that are targeted. So they don't kill cells in general, but they hit a specific target. And you hope it's a target that's overexpressed or particularly active in the cancer cells and not in the rest of your body, if you're lucky, OK? And most of them are either monoclonal antibodies or small molecule inhibitors, and that includes the tyrosine kinase receptor inhibitors. And I have some examples there. I don't know what is it, but I can't seem to. All right, so there are also three types of drug-induced liver injury. And these don't line up exactly with those three types of anti-cancer agents. They're the direct agents. They are directly cytotoxic. They harm the liver, the injury is dose-related, it's reproducible, it's predictable. And the phenotypes are important. They can cause enzyme elevations, but they typically will cause an acute hepatic necrosis. Bam, the enzymes go up, and bam, they come down. It also causes, importantly, sinusoidal obstruction syndrome. They can cause acute fatty liver and nodular regeneration. There's some examples. The idiosyncratic injury is the major type of injury that we deal with in the Dillon network. And these are drugs that aren't inherently toxic. They're a little bit toxic, maybe, but not. The injury is not dose-related, and importantly, it's rare. These are not common causes. You have a cancer patient on a bunch of drugs. Don't think of idiosyncratic injury first. Think of all those other things they're getting. And the types of injury, the phenotype, is basically acute hepatitis. It's either hepatocellular or cholestatic or mixed. And it can be anecteric, just enzyme elevations. And finally, there's the indirect type of injury. It's kind of a new concept, but the concept is that it's not directly acting on the liver. It's acting through something else on the liver. For instance, a drug that causes weight loss will worsen NASH, bring out NASH. A drug that changes the immune system may bring out immune-mediated hepatitis. So that's another type. And it's actually not uncommon. It's not rare like idiosyncratic. Importantly, you can predict it, sometimes you can prevent it, and often you can treat it. So it's something new for us, right? A drug-induced liver disease, you can treat something. Wonderful, huh? All right, so here are the drug approvals that I showed you in the U.S. And here, I just showed you the cancer agents, the number, and the type of agent they are. Cytotoxic is in red. And you can see early, pre-1970s, they were all cytotoxic agents. And then the hormonal agents got introduced, along with this tamoxifen was the first. Then you see in the late 1990s, a new type of agent came out. They're targeted, and they were designed. They're designer drugs. The first one was Gleevec. Actually, it wasn't the first one, but it was one of the first ones that we think about. And you can see it more and more now. Very few cytotoxic agents are approved each year. There's always a few hormonal ones. They're usually me-too types of drugs. And the majority of them are targeted, 90%. So we're dealing with a new phenomenon, aren't we? Now, you think, well, the targeted ones will be safe, and the cytotoxic ones will be dangerous. Ha, ha, ha. Maybe not. So this is the overview of that, the 232 agents that Livertox has that are anti-cancer agents. You can see cytotoxic and targeted agents. Well, the small molecule inhibitors are about the same. Monoclonal antibodies make up about 20%. And look at the thin sliver that are checkpoint inhibitors, the monoclonal antibody checkpoint inhibitors. I think there are 10 now. And hormonal agents hold their own. Now, let me show you what Dillon has as far as drugs, anti-cancer drugs. These are the total number. They enrolled 2,500 cases. Some have had those with six-month follow-up and have been adjudicated for their causality, a few less. And those that were scored as being drug-induced liver injury is 1,900. 133 were due to anti-neoplastic agents and were used to treat cancer, okay? There are some that were used to treat autoimmune disease or something like that. So these are used in cancer patients. That's 7%. They were due to 44 different agents. So this is since 2003, okay? I'm not going back to the 70s when this would be, of course, very different. All right, here are the, what are the agents? Here are the top 18, I think, yes. And these are the ones that had two or more cases, okay? So there are 25 others, but there are only one case each of them. And look, it's very striking. The most common, three most common causes are checkpoint inhibitors. And there's some other checkpoint inhibitors further on down the line. So the checkpoint inhibitors represent, there are 55 cases. Number four is asparaginase. That's an old drug. It's still a very important drug for acute leukemia. It's a common cause of liver injury. It's classic look. You can tell when you see it. And then comes the first kinase inhibitor or small molecule inhibitor, and that's Gleevec, imantinib. It was licensed in 2001. And further down, you can see some cytotoxic agents like oxaloplatin, dimozolamide. But a lot of them are becoming either hormonal or targeted agents. So this is, Dillon has had four periods, five-year periods of funding. And you can see that cancer agents were only 2% to begin in the first five years, then 5%, then 8%, now it's 16%. So you would think with these new agents, we'd have less drug-induced liver injury. No, we're having more. That's the bad news. Now I'll tell you the good news. And now here's the display. I show on your right, the agents that are available in the United States from my analysis. And on the left, the causes of Dillon. And you can see the piece of the pie that's grown enormously are the checkpoint inhibitors, which are 42% of cases, but they're only 4% of the agents. Of course, they probably used a lot more than most. Now I can't, I don't have time really to go much into the details of their phenotype and so forth, but this shows you the four types, cytotoxic, hormonal, and then the two types of targeted ones, the kinase inhibitors, small molecule inhibitors, and then the monoclonal antibodies, checkpoint inhibitors. Monoclonal antibodies actually are very safe. We have two cases, they're both very complicated, not clear-cut. But you can see from this that the differences between these classes, and the major difference was in severity. And our severity is marked as one, two, three, four, or five. One is anecteric, five is fatal. They die of liver disease, not of their cancer, within six months. Four is they have signs of hepatic failure, two and three are jaundiced. So we can see the cytotoxic agents are heavily to the left, causing severe disease, and a mortality rate of 17%. They qualify Rahal's High's Law, right? High mortality rate. As opposed to the hormonal agents, have quite a, you know, almost 60% are anecteric. And even though they're hepatocellular, there's no fatalities. Kinase inhibitors, they kind of fall in between. You can see the mean bilirubin for the cytotoxic was 12, but for the kinase inhibitors it was 4.3, and for the hormonal agents, 1.2. Most of them are anecteric. Now look at the checkpoint inhibitors. Very important difference. First of all, almost 70% are mild anecteric. And we have some deaths, but they were a bit complex, and probably due to their previous chemotherapy, but that can cause death. The important thing is at the very end, which is the percent that receive steroids, 87%. In fact, I can't remember one that didn't receive steroids. The fatal cases, perhaps. The fatal cases came in late. They were already intensely jaundiced when they presented. So it may be that the shift is due to interventions therapy. That's what I'm talking about. This is change that's occurred in drug-induced liver disease in America. So these drugs, they cause liver disease, but they don't cause death as much. That's pretty important, isn't it? Because you're dealing with cancer. The typical application says the side effects were common, but manageable. What does that mean? A euphemism. So I don't know how much time I have. I don't have much time. I can go through some of the phenotypic things, like with, it's now pegasparagenase. Nobody uses the regular, the pegylated form is given every two weeks, I think, or something. And it causes a classic phenotype. It's very interesting. It's different from anything else. I don't know of any other drug-induced liver injury that looks like this. Patients develop jaundice within one to three weeks very quickly, and they have a marked decrease in serum proteins. Their albumin may be two, less than two, within two weeks of receiving the drug. They have a mixed enzyme pattern, and if you do a biopsy, what does it show? Fat. Mild inflammation. The fat is visible on ultrasound as well. It has a self-limited course. The problem is it delays therapy. What is it due to? What does it cost? I think it's due to inhibition of protein synthesis because if you've taken away an essential amino acid, asparagine is not considered an essential amino acid, but if you inhibit it, if you block it, then it becomes essential. That's my theory. Someone should study this. It's very interesting. The only disease it looks like is swashiorkor. It gives you jaundice and this low protein. Okay, let me see if I can show another slide. Temozolomide is very interesting. This is an alkylating agent, but it's fairly recent. It was approved in this century. I can't remember the date, and it causes classical cholestatic hepatitis, usually within four to six weeks. It's idiosyncratic injury. It's not cytotoxic injury, and it's usually self-limited, but it's prolonged. The patients usually die of their brain tumor with an elevated alkaline phosphatase. Sometimes it can cause vanishing bile duct syndrome, but by and large, the major problem with it is you can't use temozolomide anymore. This is an alkylating agent that crosses one grade barrier, so it's valuable in that regard. Oxoplatinin is another cytotoxic agent, somewhat older agent, and it causes a classic phenotype you can't miss. What it causes is the latency is six months or more. After you finish a course, it's given for metastatic colon cancer, like monthly for six months. Then they come in at eight months, and what do they have? They have variceal bleeds and ascites. They have non-cirrhotic portal hypertensions. Wonderful model for someone to study. What does it cause by that? There's a couple posters here about it. There are interesting how to prevent this. If you prevent the injury, do you prevent its effect on the cancer? That's the issue, right? Amantanib, this is a kinase inhibitor most commonly implicated in liver injury. Interestingly, it's hepatocellular, but it's usually mild. All the cases we had recovered, most of them were anecteric. The oncologist was able to do that. The oncologists are picking up the cases because they're just seeing them regularly, so they get enzymes, and they see these mild cases that might be missed if you weren't following the patient carefully. Of course, the checkpoint inhibitors. You'll hear more about that from other speakers, and I don't need to go over it now. In the past, it was said this was hepatocellular, but it's not always hepatocellular. It can be cholestatic, and the cholestatic cases tend to be the difficult ones. They don't respond well to corticosteroids. And they can lead to, they can cause death. It's usually more prolonged, so they die outside of the six-month window, but they develop vanishing. Very interesting. What is it? It mimics autoimmune liver disease. It can mimic PBC. It can mimic PSC with abnormalities of the visible bile ducts. So as again, this is an opportunity for research that not only involves drug-induced liver disease, but involves all liver disease. So in conclusion, anti-neoplastic agents are common causes of liver injury. You'll see them on rounds, you know, consults. The differential diagnosis is challenging, but pay attention to the clinical phenotype. Knowledge of the agents that cause liver injury and their predominant clinical features are important. And this is, I think, a major use that you can apply to liver talks. Look up the drugs that the patient is on to highlight which ones that do and which ones that don't seem to cause liver injury. So let me give special thanks to the Drug-Induced Liver Injury Network. Here's a picture of all of us. And you can see that all of us are smiling. And you wonder, why are we all smiling? As a photographer told a joke, no, we're smiling because we enjoy this field so much. Drug-induced liver disease is fascinating for a little clinician because you feel the whole spectrum of types of injury. And it's like doing a crossword puzzle. What's causing this and why? And I hope people in the audience are interested in that question as well. Thank you. I'm a Mac user. I'm sorry. I don't know what difference launching us into our lineup. Next up is Dr. Kim Brower. She's a PharmD PhD, and the William Kennan Distinguished Professor in the Division of Pharmacotherapy and Experimental Therapeutics at the UNC School of Pharmacy, and the Associate Dean for Research and Graduate Education. Her research is focused on hepatobiliary drug disposition and development and refinement of in vitro model systems to predict in vivo hepatobiliary disposition. And she will be talking to us about novel bile acid-dependent mechanisms of hepatotoxicity associated with the tyrosine kinase inhibitors. Thank you, Kim, for speaking. Thank you so much for the nice introduction and also for the invitation to participate in this session on DILI and cancer therapeutics. And also thanks to AASLD for the really excellent scientific symposium yesterday focused on bile acids. Really an excellent introduction, perfect timing for this session. So these are my disclosures. I will say that the data that I'll be presenting today was generated by a former graduate student who's now gainfully employed and was funded by NIGMS, and we are grateful to all of you that pay your taxes that support our research. So as mentioned, I will be discussing the tyrosine kinase inhibitors and specifically the bile acid-dependent hepatotoxicity that we see. This is just a little preview of the novel findings from the work. And the data that we will be discussing in the next few minutes is all human hepatocyte generated, so no preclinical species involved. So as we know, the tyrosine kinase inhibitors are effective in treating various types of cancer. However, we do see a TKI associated with hepatotoxicity, which we've already heard about. It's a risk factor. In fact, six TKIs in clinical use have boxed warnings for hepatotoxicity, and 25 TKIs are labeled warnings and precautions for hepatotoxicity. And specifically, the discussion today will focus on three TKIs, the satinib, pisopinib, and serafinib, commonly associated with toxicity. The mechanisms of TKI-associated hepatotoxicity include those listed here. Of course, we're specifically interested in hepatic transporter-mediated hepatotoxicity, and many TKIs are actually found to inhibit the OATPs, specifically OATP1B1. And you can see here that OATP1B1 activity is associated with tyrosine phosphorylation, and that may be a mechanism of the TKI-mediated hepatotoxicity with 1B1. We also know, as we heard yesterday, that there is an association with BSEP inhibition, and we'll talk a little bit more about that. Now, we know that BSEP is not particularly predictive for severe drug-induced liver injury. This may be related to the particular systems that we're using that lack some of the regulatory mechanisms. As shown here, you can see that among 180-some compounds that were screened, 31% exhibited severe DILI, but had no BSEP inhibition. And also, it's notable that 21% had BSEP inhibition, but did not cause any severe DILI. So BSEP inhibition alone cannot accurately predict hepatotoxic potential of drugs. And as previously emphasized in sessions yesterday, and I think this morning as well, there are many mechanisms that may explain bile acid-dependent DILI, including the uptake transporters involved in bile acids, specifically the sodium-tarcolate co-transporting polypeptide, NTCP. And also, as bile acids are excreted by DSAP into bile, so we see inhibition of DSAP. That can be a drug inhibited by various drugs. Importantly, there are numerous regulatory mechanisms within the hepatocyte FXR-mediated processes that up-regulate these bile acid efflux transporters on the basolateral membrane, specifically OST-alpha-beta, which we think of as a safety valve, often overlooked, but a very important mechanism for excreting bile acids out of the hepatocyte, as well as the MRPs, MRP3 and MRP4. And when there are loss of function of these mechanisms, or some altered homeostasis mechanism, this could lead to hepatotoxicity. So the first question that Chitra wanted to answer was whether the TKIs that we were planning to investigate were, in fact, bile acid-dependent, exhibiting cholestatic DILI. And to answer this question, we used a C-DILI assay, which is a in vitro assay based on sandwich-cultured human hepatocytes, where the regulatory mechanisms are intact for both the regulation of the metabolic systems, as well as the transporters. And so this assay integrates both BCEP inhibition, as well as the inhibition of the basolateral efflux transporters, and also FXR antagonists. And the hepatocytes are incubated for 24 hours after four days in culture, with either a standard or sensitization media, shown here as open and closed bars, respectively. There's a negative control, which is cyclosporine, where we see no change in ATP or LDH levels in the various hepatocyte lots, as shown here. Of course, imatinib, which is, as we've already learned, a direct toxicity control, inhibits, shows a decrease in ATP, as well as an increase in LDH levels. And finally, troglitazone, which is our positive control DILI drug, shows a decrease in ATP levels only in the sensitization media, which is a media that contains both bile acids and fatty acids. And also, this sensitization media, as previously mentioned, does upregulate OST alpha beta, this safety valve mechanism. So looking first at the TKIs under investigation, incubating with this sandwich culture to hepatocyte system for 24 hours, we do, in fact, see bile acid-dependent toxicity, as noted here, divergence between the ATP levels and LDH levels in this standard versus the sensitization media, the concentration, the lowest concentration at which we see toxicity, as shown here by these red arrows. And actually, when we look at the toxic concentrations for dasatinib, pisopinib, and serafinib, as shown here, we can see that those are, in fact, concentrations within the range of the CMAX for the use clinically for these drugs. Now, we also were interested in the intracellular concentrations of the TKIs, and so we were able to measure those in the sandwich culture to hepatocytes. And you can see here that the concentrations, when we correct for the unbound fraction, as shown here, these unbound concentrations within the hepatocytes are below the IC50 value reported for BSAP, more than 10-fold lower in some patients. And so this actually suggests that there may be mechanisms other than the BSAP inhibition for the toxicity attributed to these TKIs. Now, we already heard yesterday and again this morning about the various mechanisms involved in synthesis of bile acids, CYP7A1 being the first step in the synthesis. We also know that cytochrome P458B1 and 27A1 are involved, and then we have the conjugation enzymes SPAT and VAX. We also know that with respect to cytotoxicity, not all bile acids are created equal. And so the chenodeoxycholic acid and the glycine-conjugated chenodeoxycholic acid are definitely more hydrophobic than the more hydrophilic bile acids. And so measuring these bile acids in vitro models to really look at hepatocellular concentrations is very important. To do this, we can use the sandwich culture to hepatocyte system where we actually measure the cell plus bile accumulation in the standard buffer versus cellular accumulation in the calcium-free buffer. And in side-by-side incubations, because when we remove the calcium, we in fact are disrupting the canalicular networks as depicted here in the images. You can see the accumulation in the canalicular networks that form in vitro in this sandwich configuration. We can calculate the mass in the bile compartment and report that by measuring both cell plus bile or cell accumulation. And I'll show you data with the kinase inhibitors in a couple of minutes. The biliary excretion index can be calculated here as a mass in the bile compartment normalized for cell plus bile accumulation. And of course, we can also calculate the in vitro uptake and biliary clearance. So looking first at the effect of the kinase inhibitors on mRNA levels of the CYP7A1 and other cytochrome P450s, we do see upregulation at the mRNA level of CYP7A1, suggesting induction of bile acid synthesis, which was a little bit of a surprise to us. We did not see any other effects on mRNA levels for CYP8B1 or 27A1 or cytochrome P453A4 or the conjugation enzymes. Also note here in these two different lots of the human hepatocytes that there was no effect of the TKIs on the NTCP mRNA levels. But surprisingly, based on Western blot, you'll notice that the protein levels were significantly increased by all three of these kinase inhibitors. So there was more NTCP on the membrane in the sandwich cultured human hepatocytes treated with these TKIs, but this was not due to increased mRNA levels. So perhaps there was increased stabilization of that protein on the membrane by the TKIs. Now, as we've learned, it's very important to think about the bile acid concentrations within the hepatocyte. And so we measure those by mass spec after 24-hour exposure in the sandwich cultured human hepatocytes. Shown here on the left is a heat map showing the full change in the bile acid concentrations as depicted here with the treatments for the three TKIs. And the significant changes, if you're interested in the more detailed analysis here for each of the lots of hepatocytes. I think overall, what we tend to see is with the dasatinib exposure, there is a significant increase in bile acid concentrations in the media. And with pisopinib, there was a significant increase in cellular bile acid concentrations. I think one other interesting finding was with serafinib, the GCDCA3-glucuronide within the hepatocyte was significantly increased as well as in the media as shown here on the right. So if we look from a functional standpoint, so so far we've looked at the mRNA and the bile acid concentrations, but looking at the uptake of a prototypical bile acid substrate, the tritiated tarcolate in these hepatocytes, we see in fact the dark bars here showing the cell plus bile accumulation of the tarcolate that across all three lots of human hepatocytes, dasatinib significantly increases bile acid accumulation. The tarcolate uptake clearance also significantly increased relative to DMSO control. And also we see the same effect with pisopinib. So consistent with the finding that there was more NTCP on the membrane and increased uptake of bile acids certainly makes sense. Also, if we calculate the biliary excretion index, so this would be related to BSEP inhibition, we see that in fact dasatinib significantly decreased relative to control values here, the tarcolate biliary excretion index. And interestingly, there was a increase in biliary excretion of the tarcolate by pisopinib, suggesting perhaps some stimulation of BSEP-mediated mechanisms. So in, and also with serafinib, there was no change in BEI or the uptake clearance of the tarcolate. In summary then, what we've shown is that for these three tyrosine kinase inhibitors, in fact, we do see a bile acid associated a drug-induced liver injury based on the C. dilly assay. And in fact, there are many mechanisms that appear to be linked to this toxicity based on the human patasite data in addition to BSEP alterations. We saw an increase in NTCP on the membrane resulting in increased uptake clearance of bile acids. We saw increased synthesis of the CYP7A1, increasing bile acid concentrations in the cell and or in the media. There was an apparent increase by dasatinib in the efflux of these bile acids and also serafinib appeared to increase the glucuronidation of the GCDCA. So I think that key takeaways from this study, certainly from a clinical perspective, thinking about co-administration of other medications that an individual might be taking at the time of these kinase inhibitors that could impact these various mechanisms would be very important to consider. In addition, I think from a drug development perspective, thinking about use of in vitro systems that have regulatory machinery that is intact rather than just the screening assays that have been traditionally used for BSEP inhibition are really important because you obviously would need the metabolic and transporter regulation mechanisms to identify these potential interactions. And with that, I would like to thank Chitra and the other graduate students and postdocs in our laboratory, our collaborators and our funding sources and I look forward to the panel discussion and answering any questions afterwards. Thank you. Next, it is my pleasure to introduce Dr. Michael Lee. He's an Associate Professor and Transplant Hepatologist at UCSF. His translational and clinical research program focuses on immune-mediated liver injury from checkpoint inhibitors or LEC. I'm trying to get that to catch on. I know some people are in favor of Chili. We're calling it LEC, and Mike will take it away. Thanks so much for that introduction. Thank you all for being here, and I appreciate being invited to speak today. So my name's Mike. I'm a transplant hepatologist at UCSF. Despite what the program says, I'm not going to spend the entire time talking just about steroid dosing and immune-checkpoint inhibitor hepatitis. Instead, as Drs. Dar and Ding suggested, I'll be giving a broader overview of the management of immune-checkpoint inhibitor hepatitis and also spending some time discussing the distinct clinical entity that is immune-checkpoint inhibitor cholangitis. No disclosures. So here's an outline of what I'll be talking about today. The first four topics are specifically related to, I say, hepatitis, while the last is a brief overview and discussion about the relatively limited literature around cholangitis or cholangiopathy. For the purposes of this talk, I'll be using the nomenclature, I say, hepatitis or cholangitis. I understand there's- You'll have an argument later. There's a lot of ongoing debate, including from some reviewers in my papers, about the most appropriate nomenclature to use, but I hope that everyone's okay with me just choosing these words for the purposes of this session. So we'll start with discussing first-line corticosteroid therapy. So based on cancer and GI society guidelines, for more mild cases of ICI hepatitis, oncologists typically hold ICI therapy and start half to one milligram per kilogram per day of prednisone if liver tests do not improve after several days. For higher-grade liver toxicity, which is the focus of my talk today, guidelines suggest immediate initiation of high-dose corticosteroids of between one to two milligrams per kilograms per day of methylprednisone equivalents. However, these guidelines remain based on expert opinion, and the AGA guidance even specifically states that these doses were empirically determined and have not been rigorously examined. Well, so we think that steroid dosing matters not just with regards to treating the underlying immune-mediated hepatitis. There are clear dose-related adverse effects of systemic corticosteroid treatment that I won't delve too far into. In addition, it's plausible that steroid treatment could negatively impact cancer outcomes as the immune suppression is theoretically counteracting the effects of the immune checkpoint inhibitor. So here's a graphical abstract from a paper that we published where we tried to compare clinical outcomes of patients with high-grade immune checkpoint inhibitor hepatitis who received one versus two milligrams per kilograms per day of methylprednisone equivalents. This was a study of 215 patients, 128 of whom received a median dose of one milligram, and 87 of whom received a median dose of two milligrams per kilogram. So the major issue in any retrospective pharmacoepi study, as in this case, is confounding by indication. So what we did is we performed inverse probability of treatment weighting to balance pretreatment covariates that were imbalanced between the two groups. And so we developed a propensity score model to predict risk of being in the higher steroid dose group, and then inverse probability weights were calculated from the propensity score and used in regression. And so what we found is that there wasn't any difference in risk of development of steroid refractory disease, and that there was no difference in time to biochemical resolution of their hepatitis between the two groups. However, the patients in the higher dose group were more likely to develop steroid-related complications, including hyperglycemia requiring treatment, as well as certain infections. So based on this data, we suggest that a starting dose of one milligram per kilogram of methylprednisolone equivalents is sufficient for most patients with high-grade ICA hepatitis, and that if there's insufficient biochemical improvement at that dose, it's reasonable to add a second immunosuppressive agent for management of steroid refractory disease, which occurred in about 30% of our cohort, much more to come very shortly on steroid refractory disease. One major caveat is that most patients in our study did not have true liver dysfunction in terms of bilirubin or INR elevations, and in patients who have evidence of especially liver synthetic dysfunction, more aggressive upfront steroid dosing might be reasonable. So since up to a third of patients won't respond to steroid treatment alone, this leads us nicely into the next topic of how we define and manage steroid refractory ICI hepatitis. So the problem is that the definition of steroid refractory disease is very unclear right now. So all the major societies recommend escalation of immunosuppression after only two or three days of steroid treatment. They also don't give clear guidance about thresholds for improvement or non-response. So clearly we need to find a way to define steroid refractory disease in a scientific way. So here's some data that I'm presenting from our group that was presented in abstract form yesterday. We found that unsurprisingly, the percent change in ALT after seven days of steroid treatment differed between the steroid refractory and steroid responsive patients. So on the box plot shown here, you can see that the median ALT reduction in the blue group, which is the steroid responsive group, was a decrease in 58%. So here's 0%, meaning no ALT change. Here's negative 50%, meaning a 50% reduction in ALT. And here's positive 50, 100, 150%. While the median ALT reduction was only 12% in the red steroid refractory group as shown here. So as a single predictor of steroid refractory disease, the percent change in ALT after seven days of steroid treatment had a C-statistic of 0.83, which is quite good discrimination. And we found that the optimal cutoff based on maximizing the J-statistic, or the Uden's index, was an ALT reduction of 30%, which is indicated by this orange arrow on the ROC curve, as well as the orange arrow here on the box plot, where you can sort of visually see that this cut point separates the two groups reasonably well. When splitting patients into those who received early additional immunosuppression within seven days versus after seven days of steroid treatment, early additional immunosuppression was associated with faster biochemical response. And so we therefore proposed that patients who don't experience at least a 30% reduction in ALT after seven days of steroid treatment should be considered steroid refractory. At that point, additional immunosuppression should be initiated. And so in terms of the additional immunosuppression, mycophenolate is generally considered to be second line therapy at this point. In our cohort of about 300 patients, 91% of the steroid refractory patients ended up getting mycophenolate after they received steroids. There are case reports of other immunosuppressive or immunomodulatory agents that have been shown to be effective, including azathioprine, tacrolimus, either in and of itself or in conjunction with mycophenolate, tocilizumab, and even ATG. Notably, while steroid refractory patients do take longer to achieve biochemical remission, as shown by the Kaplan-Meier curve here on the right, almost all of them do actually achieve this biochemical remission. And so our suggestion to escalate immunosuppression based on labs after seven days of steroid treatment is definitely more lax than what the guidelines currently recommend, which is two to three days. Which feels short. But because relatively few patients with even high-grade immune checkpoint interval hepatitis develop true liver dysfunction, a less aggressive approach may be reasonable while still setting a clinically meaningful boundary in terms of when to initiate mycophenolate or other immunosuppressive agents. So a related topic is the role of the hepatology specialist, which we'll discuss next. So our group published a study assessing the impact of early liver specialist consultation. And so unsurprisingly, early consultation had no effect in steroid-responsive patients. This is likely because in almost all cases, oncologists are initiating appropriate steroid therapy very promptly without those patients ever reaching us. However, in steroid refractory patients, early consultation was actually associated with faster biochemical resolution of their hepatitis, as shown by the Kaplan-Meier curve here on the right. And the adjusted hazard ratio for this effect of early consultation on time to ALT normalization was 1.89. In addition, we found that steroid refractory patients in the early consult group received earlier additional immunosuppression than other steroid refractory patients. And so based on this last finding, we performed a mediation analysis to determine if the beneficial effect of early liver specialist consultation was mediated by quicker initiation of additional immunosuppressive agents. And so you can see here, both early consultation and time to additional immunosuppression were significantly associated with a time to ALT normalization when in separate Cox regression models. So here's model one, here's model two. However, when you include both of those variables in the same model in model three, early consultation was no longer significantly associated with the outcome while time to additional immunosuppressive agents was still significantly associated with that outcome, which indicates that this beneficial effect that we saw is primarily mediated by how quickly patients receive mycophenolate or other agents. So I think that liver specialists, which are many people in the room today, should play two major roles in immunochemical neurohepatitis. The first is to confirm the diagnosis, in particular by ruling out other possible etiologies of liver injury, especially before considering escalating to second line immunosuppressive agents. Liver biopsy can certainly be considered in atypical cases or in patients not responding to steroid therapy. And then in addition, we can also assist with treatment, both in terms of guiding steroid dosing and by influencing the decision of when to start additional immunosuppression. We speculate that early exclusion of alternative etiologies by specialists who routinely evaluate liver injury and their experience with immunosuppressive medications for management of other autoimmune liver diseases may actually enable this earlier initiation of additional immunosuppression as was seen in our study. So next we'll discuss the safety of re-challenging patients with ICI therapy once they've recovered from their initial insult, once they've been treated appropriately. So the various guidelines, again, suggest permanent discontinuation of ICI therapy following high-grade ICI hepatitis, depending on which society guideline you look at. It's either grade three versus grade four. But there's emerging data over the past several years suggesting that re-challenge may be feasible in patients, even with higher-grade ICI hepatitis. So this was a study published in JAMA Oncology by a group who used the World Health Organization VigiBase database. And out of 642 cases of ICI hepatitis, 31 were re-challenged, all with the same agent, with a recurrence rate of 29%. And as you can see here, risk of hepatitis recurrence was intermediate compared to risk of other IRE recurrences, with arthritis and dermatitis being the most common to recur, as seen down here. In terms of hepatitis-focused or hepatitis-specific re-challenge studies, these two studies evaluated a total of 73 patients with hepatitis who were re-challenged, with a total recurrence rate of similar, of 27%. Oh, whoops. In our study of 31 melanoma patients who were re-challenged, the recurrence rate was 12%. This lower rate is likely because most melanoma patients are initially receiving combination PD-1 and CTLA-4 therapy and that most of these patients then received PD-1 monotherapy re-challenge, which is a lower-risk regimen compared to the combination regimen. It was unclear whether this re-challenge actually improved clinical outcomes, though we found a signal towards survival benefit in the small study, but again, it was just a very small study of 31 patients. So overall, if the treating oncologist believes that a patient would potentially benefit from further immunotherapy, a trial of re-treatment is reasonable, which is different from many other drug-induced liver injuries, but exceptions to this, again, are likely patients who develop steroid refractory hepatitis or the rare patients who develop true liver dysfunction, as measured by bilirubin or INR abnormalities. So finally, we'll pivot over to discussing ICA cholangitis, which, as I mentioned, is much less studied, much less well-characterized than sort of the purer form of hepatitis specifically. So in terms of large duct involvement, that's quite rare, with a reported range of less than 0.1% to maybe 0.5% of all ICA-treated patients. Compare this to the one to 10% of ICA-treated patients who will develop a grade three or higher hepatitis, and then if you expand that to all grades, up to 20% of ICA-treated patients will develop a hepatitis. Small duct cholangitis is much more common, and this also accounts for a proportion of patients that we previously called as having hepatitis. So there are two studies that sort of nicely demonstrated this. One was a study out of MGH that looked at 60 patients, all who were biopsied, and 16 of those 60 patients, or 27%, had a predominantly cholangitic pattern on their histology. And then there was a European study looking at 117 patients of whom 43 of them had a predominantly cholestatic pattern based on labs, not necessarily histology, although a subset of those patients with a cholestatic pattern of liver injury did also receive a liver biopsy and had sort of similar features of a cholangitic pattern. So patients with large duct involvement can have essentially sclerosine cholangitis findings, as seen on MRCP, although again, large duct involvement is much less common than small duct involvement. And then sort of nonspecific lymphocytic and neutrophilic cholangitis are shown in the biopsy slides B and C. So there are also multiple case reports of vanishing bile duct syndrome attributed to ICI therapy, as shown here. So here's a liver biopsy from one of these cases. So slides B and C show severe cholestasis, duct irregularity, ductular reaction. And slide D is showing a CK7 staining, which shows ductal proliferation and the appearance of some biliary epithelial cells, but you'll note that there aren't any organized bile ducts in that picture. So first-line management remains corticosteroids. There's relatively good, though small, studies suggesting that immune checkpoint heterocholangitis is more likely to be steroid refractory than pure hepatitis. Ursodiol is sometimes used to treat nonspecific cholestatic liver injury, is the treatment of choice for PBC, is, I suppose, the treatment of choice for vanishing bile duct syndrome for what it's worth in that scenario. And given that ursodiol is well-tolerated, is effective in other cholestatic liver diseases, and has been used in small case series successfully, albeit mostly in conjunction with steroid therapy, I think that ursodiol is a regional adjuvant therapy. Whoops. So here are a few take-home points. So first, in terms of first-line steroid therapy, one milligram per kilogram per day of methylprednisolone equivalents seems to be sufficient for most patients with high-grade ICA hepatitis. We propose the following definition for steroid refractory hepatitis, which is less than 30% reduction in ALT after seven days of steroid treatment, and patients meeting these criteria should be started on mycophenolate or other agents at the seven-day time point. Hepatology specialists may improve outcomes in steroid refractory patients, and we suggest that when consulted, our role should be to help rule out other causes of liver injury, and to initiate additional immunosuppression promptly at the seven-day mark. Next, recurrence rates after ICA rechallenge are in the 20 to 30% range, meaning that in patients who may benefit from additional immunotherapy retreatment can be attempted, although with careful liver test monitoring. And finally, ICA cholangitis is poorly characterized and studied compared to hepatitis, and patients with cholangitis may be more likely to be steroid refractory, ursodial can be considered as therapy in addition to standard first-line steroid treatment. So I'd just like to acknowledge my many mentors across multiple institutions, my collaborators, and then ACG for their funding of my work. Thank you. Great talk, great session so far. Let me introduce you. Thank you. So the next talk is Dr. Lucia Possamae. She's a consultant hepatologist, an honorary clinical senior lecturer at St. Mary's Hospital in London, at the Imperial College of London, right? Her research interests are in the immunology of drug-induced liver injury and autoimmune liver diseases. The title of her talk is The Role of the Adaptive and Innate Cells in Immune-Mediated Liver Injury from Checkpoint Inhibitors. Thank you very much for the invitation to speak at the meeting today. I'm delighted to be here with you in Boston. I think my introductions have already been made, so I'll move on. I have no disclosures to make. So although my presentation is very much on the basic immunopathology of immune-mediated liver injury due to checkpoint inhibitors, I wanted to start with a brief case vignette just to highlight the clinical relevance, explain the motivation for why we'd want to better understand this condition, and focus the questions I will cover. So this H&E section was presented at our histology meeting within the last two weeks in London. It was an 83-year-old patient who was being treated for metastatic endometrial cancer with the anti-PD-1 drug nivolumab, without complication for six months from the beginning of 2023. Over the summer, she developed a grade three elevation in transaminases and was treated according to guidelines with IV methylprednisolone, stepping down onto oral prednisolone. And after a brief flare in July necessitating dose escalation, she eventually weaned down and discontinued steroids. Unfortunately, she'd experienced quite severe neuropsychiatric side effects from steroids and a steroid-induced myopathy, which was impacting on her mobility and her independence. She then unfortunately re-presented just in the last few weeks at the end of October and was admitted on the hepatology team with a rebound in her ALT to over 1,700. This biopsy was taken and showed marked lobular inflammation with very dense aggregates of immune cells in places coalescing into a granulomatous hepatitis, which you can see in that section there, a small granuloma. So when we see this histology at the most basic level, the questions we need to ask, what are the identity of those liver-infiltrating immune cells? What's their function? How do they get there? And how are they associated with the hepatocellular damage that's reflected by an ALT of 1,700? And then the motivation in asking those questions and in better understanding the immunopathology of this condition is the development of smarter and more targeted therapies or preventative strategies to avoid the reliance on steroids with all their attendant side effects, as demonstrated in our case and we heard about in Michael's talk. Also for the treatment of steroid refractory patients. So just a few general principles or key principles. The liver injury associated with checkpoint inhibitors is just one of a myriad of immune-related adverse effects of these medications. The immune-activating effects of checkpoint inhibitors are systemic, and yet organ inflammation is local and contingent on the particular interplay of factors within the tissue microenvironment and the impact of checkpoint blockade. And so while commonalities certainly exist between the different IRAEs in the systemic compartment, we should consider them tissue-specific and distinct. The thyroiditis and myasthenia, for example, are antibody-mediated, B-cell-mediated, and therefore respond to B-cell-depleting strategies quite unlike the hepatitis, as we'll see. And as we've just heard, even at the level of the single organ in the liver, we recognise two clinically and pathologically distinct forms of inflammation. The classical hepatitis, as seen in our case, and a cholangitis, particularly seen with anti-PD1s, and as we've just heard, runs a slightly different clinical course. Today I will be focusing very much on the hepatitis, although there's undoubtedly, as we've heard, a clinical need for better understanding of CPI-related cholangitis. So moving on to clinical studies of the immunology of this condition, we published in the Journal of Hepatology in 2021 a study of patients' immunophenotypes with checkpoint-inhibitor-related hepatitis, as we decided to call it in this manuscript, CPI hepatitis, the patients here in the light blue, and compared those with groups of healthy patients in the white, those exposed to checkpoint inhibitors but without liver inflammation in dark blue. We first looked at the monocyte populations within the peripheral blood and demonstrated a highly activated monocyte phenotype with high levels of CD163 expression on intermediate and classical monocyte subsets. The cleaved product, soluble CD163, was also elevated in patients' sera compared with healthy controls. This is a marker that's been associated with macrophage-driven inflammation in a number of other conditions, including liver conditions, and in a few other IRAEs. The degree of monocyte CD163 cell surface expression, as shown in the top right, correlated with the severity of liver injury, and I haven't shown the graph, but it also decreased longitudinally over time as the hepatitis settled. We also demonstrated circulating monocytes express very high levels of CCR2, a tissue homing receptor, which is particularly important for monocyte recruitment to the liver, and more on that later. The lymph node homing receptor CCR7 was conversely downregulated. So in terms of cells of the adaptive immune system, we looked at CD4, CD8, and B cell populations and found the most marked changes in the CD8 compartment. CD8 T cells demonstrated enhanced cell surface markers of activation in the top two panels here with increased HLA, DR, and ICOS compared with both healthy control and CPI-exposed patients without hepatitis. In the bottom two panels, we show the intracellular cytotoxic effector molecules, granzyme B and perforin. These are the molecules responsible for the cytotoxicity of the CD8 population, but also significantly elevated in the context of checkpoint inhibitor-related liver injury. These findings of an activated peripheral CD8 T cell population have also been shown in other studies, studied by chemetal and mixed IRAEs. They showed an HLA, DR high, CD38 high population. They also interestingly showed an increased CHI67 staining reflecting increased proliferation in the periphery in that CD8 T cell compartment. There was an abstract presented at DDW last year by a group from Nottingham in the UK that used CyTOF and also found an HLA, DR high CD8 phenotype as a hallmark of liver injury in this condition. So it's reassuring these findings are reproducible across different groups and using different immunological techniques. So coming back to the histology, there have been a number of published descriptions of CPI hepatitis going back to 2012 in which CD8 T cell-rich infiltrates have been noted. There was an excellent abstract poster yesterday in the poster session abstract 1704, if you want to look it up, from a Japanese group who described the histological features of CPI hepatitis in a cohort of 32 patients and found CD8 T cells in all samples, CD4 T cells via immunohistochemistry in 66 and B cell staining only in 13%. We also used immunohistochemistry in our study to identify cells composing those dense lobular infiltrates that I've discussed earlier and we found CD8 T cells with granzyme positive staining. Also CD68 positive myeloid cells which by their CCR2 and CD163 expression are likely to represent relatively recently recruited monocyte-derived macrophages rather than the resident Kupfer cell population. Co-localization of these cells was demonstrated. Draw your attention to the top right panel where the CD68 in red and the CD8 in green markers co-localization of cytotoxic T cells with monocyte-derived macrophages in clusters. And so at the end of this paper we proposed this descriptive model in which changes in the peripheral blood activated and elevated tissue homing markers on monocytes and activated and enhanced cytotoxicity of T cells was mirrored within the hepatic compartment by clusters of those cells co-localized and adjacent to areas of tissue damage. Just a word on CD4 cells, although in our cohort we didn't see or demonstrate any significant hepatitis-related changes in the CD4 compartment, other studies have noticed these. The Resch et al. paper, the figure here on the left, highlighted an increase in CD4 HLA-DR expression and an increase in CD4 affected memory T cells associated with mixed IRAEs which included a significant proportion of hepatitis patients. Hutchinson et al. in this fascinating Nature Communications paper from 2021 which I recommend which linked CMV reactivation with triggering of CPI-related liver injury also noticed an activated CD4 T affected memory phenotype present at baseline which actually predicted subsequent development of toxicity. So clearly some relevant and significant changes in the CD4 compartment as well. So having described the pathological changes in cell type seen in patient populations there are a number of mechanistic questions that arose and we're very keen to move on to be able to answer those questions and look at the relative contributions of innate and adaptive immune cells in this form of liver injury and so turned to murine models. Like the majority of humans, mice do not develop hepatitis spontaneously in the context of CTLA-4 and PD-1 blockade and so an additional priming or inflammatory stimulus seems to be required to generate reproducible models of checkpoint inhibitor-related hepatitis. Matushi et al. in 2015 used amodiquin which is an antimalarial with some direct hepatotoxic potential in their model and Llewellyn et al. used an IDA-1 inhibitor as a third checkpoint in order to generate a murine model. We've developed a model using TLR9 agonist as a primer but also have experience using TLR4 and acetaminophen as just an inflammatory trigger in order to generate models of CPI-related hepatitis. So here's the experimental strategy by dosing wild-type mice with combination checkpoint inhibitors, so both anti-CTLA-4 and anti-PD-1s and adding a day-one priming dose of a TLR9 agonist. We've been able to demonstrate the consistent development of a lobular hepatitis in mice with a mixed CD8 T-cell and monocyte-derived macrophage infiltrate spatially associated with areas of hepatocyte death and with elevated serum markers of liver injury across a seven-day time course. So looking at the histology by day seven in the mice treated with checkpoint inhibitors if you look at the bottom right panel there you can see times 400 magnification these dense aggregates and clusters of cells which are positive for CD8 T-cell marker and F480 which is a murine macrophage marker and I hope you'll agree that's certainly reminiscent of the histology that I showed earlier from the human study. So then we moved on to selectively deplete CD4 and CD8 T-cells in order to look at the relative impact of those two cell populations. The CD4 depleted mice are shown in the pink bars here and the CD8 depleted mice in the purple compared with checkpoint inhibitor treated but non-depleted mice in the blue. The mice with CD4 depletion still had quite a similar extent of tissue injury histological inflammation and serum markers of liver injury so didn't appear to have much impact on the development of liver injury in the model however CD8 depletion despite there still being a monocyte infiltrate to the liver had no evidence of liver injury and no serum markers of liver damage. Coming back to Llewellyn et al. I mentioned their model earlier they used a PD1 genetic knockout mouse treated with anti-CTLA-4 and an IDA-1 inhibitor and showed a very similar impact to the selective depletion of CD8 T-cells with the complete attenuation of liver injury in their mice. The panels on the left here showing serum markers of liver injury GLDH and ALT with their CD8 depleted mice in the blue triangles showing baseline levels of serum markers of liver injury and the panels on the right show the histological impact with an absence of periportal or pannulobular inflammation with CD8 depletion so very similar to our findings in a different model. Having looked at the CD8 T-cells we then went on to look at the relevance of the monocyte infiltrate in the liver and initially used genetic disruption of CCR2 mediated monocyte recruitment to the liver using a CCR2 RFP-RFP mouse. This is a mouse that lacks functional CCR2 protein and therefore monocyte recruitment to the liver is inhibited. These mice not only showed a reduction in the monocyte-derived macrophage infiltrate in the liver as you would expect from the genetic knockdown but mice showed no increase in CD8 T-cells within their liver and no evidence of tissue injury. So given the therapeutic potential benefit of inhibiting CCR2 mediated monocyte recruitment we then went on to try an oral CCR2 inhibitor, Senequiviroc or CVC on mice run through this same time course with the experimental protocol depicted in this schematic on the top left there. We demonstrated very similar benefits to genetic disruption with prevention of CD8 T-cell accumulation and liver injury to the CVC-treated mice in the green bars there showing decrease in CD8 T-cells, Granzyme B and Perforin positive T-cells, a reduction in serum markers of liver injury, the cytokeratin-18 fragments and normal liver histology at day 7. So bringing this to a summary, I've slightly clumsily adapted the earlier schema with the additional data. So what can we say about liver injury induced by checkpoint inhibitors from an immunological point of view? Well, it certainly is associated with cells of both the innate and adaptive immune system. The clinical studies demonstrate a circulating population of highly activated and proliferative CD8 T-cells which are mirrored in the liver by aggregates of CD8 T-cells associated with hepatocyte injury. And in two independent murine models, selective depletion of those cytotoxic T-cells was sufficient to completely prevent liver injury, suggesting these may be non-redundant effector cells. Cells of the innate immune system clearly play a vital supporting role with the patient histology demonstrating monocyte-derived macrophages co-localized with those CD8 T-cells and activated circulating monocytes with enhanced tissue homing markers. Genetic disruption or pharmacological inhibition of CCR2-mediated monocyte recruitment to the liver in a mouse model was able to prevent CD8 recruitment and attenuated liver injury. So coming back to those questions that I initially asked, I hope I've been able to at least partially, if not completely, answer them. We're certainly starting to understand better the phenotype and identity of cells responsible for immune-mediated liver injury and there's certainly convincing evidence for direct CD8-mediated hepatocyte injury. Potential candidate pathways, including CCR2-mediated monocyte recruitment, some cytokine and chemokine pathways that I haven't had time, unfortunately, to show you are starting to emerge. Hopefully, over the next few years, we may be able to move towards more targeted therapies in this condition. There's obviously a need for much more research and more questions evolve the more you look at this. I'd certainly like to highlight the crosstalk between the immune cells, the differences between the periportal and lobular infiltrates and the relatively understudied arductular injury as areas of particular research need in the future. I'd just like to acknowledge my colleagues at Imperial College London and the Royal Marsden Hospital, in particular Dr Catherine Good, our post-doctoral scientist who's performed a lot of the experimental work you've seen today. Thank you very much for your attention. Everybody should join us up here. Okay. Okay, we're going to start the question and answer session. A reminder that we have Wait, Wait, Don't Dilly at 4 p.m. It's going to be a lot of fun. If you could just walk up to the microphone, introduce yourselves, and ask your question. Please go ahead. Well, maybe I'll go first if my microphone is working. Sure. Hi, I'm Liz Rand from Children's Hospital of Philadelphia. As a pediatric hepatologist, I was intrigued by the Dilly drug injury types and by the various bile acid transporter genes, and I wonder if people with apparent idiosyncratic reactions to drugs may actually be heterozygotes for some of the genes that cause PFIC or BSEP mutations and so forth, and I wonder if somebody has a bunch of DNA on those cases and has been checking for that. Yes, that is currently under investigation, and there is data suggesting that for some of the loss of function or partial loss of function that that is a risk factor for Dilly, certainly with BSEP as well as, I think, some of the other transporters. Please go ahead. Hi, I'm Lynn Wong from MD Anderson. My question is for Dr. Lee. What are your thoughts on the use of budesonide instead of methylprednisone or prednisone? Great question. I know that there are a couple of case reports of budesonide. So for the audience, budesonide is a steroid that undergoes extensive first-pass metabolism in the liver. It's used. The benefit is that it has minimal systemic effects because... Oh, sorry, is the mic not on? It's okay. I'll keep talking. Oh, I just have to lean forward. Sorry. So the benefit is that because of this extensive first-pass metabolism that budesonide has pretty minimal systemic effects. Budesonide has also been shown to be effective in autoimmune hepatitis, which is a lot of how we base our sort of management decisions on immune checkpoint in our hepatitis. I think that that's a very interesting sort of avenue of exploration. I've actually thought about trying to figure out how to come up with some sort of clinical trial to study budesonide versus prednisone. I think the limitation is that I don't know what it's like at MD Anderson, but at least where I've trained and where I currently work, the oncologists are starting prednisone themselves, and so I think it would be a much more difficult situation to sort of corral people across multiple specialties to try to pull off a trial of prednisone versus budesonide. I do think that in patients, especially those who may have reasons not to receive prednisone therapy or reasons to avoid systemic corticosteroid therapy, that a budesonide trial is totally reasonable. We would love to collaborate. We've actually been clinically using budesonide as first line for some time. Oh, wonderful. So come talk to us. Is there a difference between IV and oral prednisone? You didn't seem to be separating. Yeah, so I didn't show this data, but we basically converted a solumetral dosing, which is IV, or I guess technically we converted oral prednisone dosing into solumetral or methylprednisolone equivalents, and so there is a little bit of a difference in terms of oral bioavailability of prednisone where the exact dose of prednisone, if you converted that to methylprednisolone, would actually be weaker by a factor of around 15% or 20%. Go ahead. Yeah, Eleonore Demartin, Paul Bruce Hospital in Paris, France. I have two questions for Dr. Li. So we previously published in a small series that up to 50% of the patient with a grade 3 or above hepatitis can spontaneously improve without corticosteroids, and I can tell you in larger court it's exactly the same. So as you said, even the grade 3 or above is not a very severe hepatitis, no hyperbilirubinium, no coagulopathy. So I'm questioning even the one milligram, kilogram day, I think is like over-treating these patients. We use this dose for acute severe hepatitis, so this is my first question. And the second is about the second-line treatment. So you gave MMF after seven days. in my experience, sometimes takes longer to get a serious response. So how many patients on your series that received the MMF later did not respond? So, I mean, my thought about this is like, we are overtreating these patients, so. Yeah, I agree with that. It's just hard to, it's impossible to determine the counterfactual in patients retrospectively, right? So all of my data is retrospective. I think, again, the difficulty is that we as hepatologists are not typically the ones who are initiating treatment. And so typically by the time a hepatologist actually sees one of these patients, they've already been started on the prednisone. I agree that this is a sort of a very unique type of liver injury in that very few patients develop true liver synthetic dysfunction, which is the thing that we would care most about. And that there is probably a substantial proportion of even high-grade immune checkpoint at our hepatitis patients who would, in the counterfactual situation where they did not receive prednisone, would get better on their own. So I also agree that we are probably overtreating. I think sort of the first step is to try to, at least to step back on the prednisone dosing at first. And then I think in the future, again, this would require a lot of collaboration across different disciplines between hepatology, oncology, is to move towards whether we can actually perform a trial just like in budesonide versus prednisone. We could also potentially study prospectively treatment with steroids versus observation at first. Right. We need to change the guidelines. We need to make our colleagues change the guidelines. Yes. I think that's the start point. Yeah, the guidelines are very aggressive. And especially I think, you know, my sort of cautiously worded suggested changes are about stepping back on the dose of the prednisone, about giving more time than just the two or three days of prednisone before you escalate to mycophenolate. I think two of the three days is clearly an inadequate duration of time. You could argue even seven days may be inadequate, but at least based on our data, that seemed like a better cutoff. Thank you. Kevin Cornwell from Washington University in St. Louis. Dr. Hofnagel, most of what makes its way to liver tox are going to be where the elevations are more dramatic, at least three or at least five times the upper limits of normal. But some of the TKIs in particular tend to cause sort of a low level elevation, and particularly some of the TKIs, say for L-positive cancer, are taken over the course of many years. So is there any data about the long-term consequences of these type of low level elevations in patients on TKIs that tend not to develop resistance and tend to be taken over a course of many years? I don't think I do have any such information about. What we focus on is clinically apparent disease, and that's kind of a mishy-mashy distinction between just enzyme elevations and what we consider drug-induced liver disease. And that's not always clear in the literature. Even in Dillon, it's a soft area. It is important to look at the cases that are jaundiced and compare them in that way. Guru. Hi, I'm Guru Aithal from Nottingham. Your animal model I was just asking a question about, you had to give two checkpoint inhibitors plus TLR9 to generate inflammation. It's similar with the previous model where they had to give amidequin to induce something. So what is in human where their own checkpoint inhibitors equivalent to TLR9 that you gave to animals or amidequin to another model? So there's something else other than the drugs which is triggering the first initiating event. What do you think it is? Absolutely, it's an excellent question. We know not all patients exposed to dual checkpoint inhibitors do develop hepatitis, and my hypothesis would be that those who do have a subclinical triggering event, and there's three interesting pieces of evidence. The CMV reactivation paper that I mentioned in Nature Communications linked viral reactivation as a trigger for CPI hepatitis. So that's one potential trigger. The second is other minimally hepatotoxic drugs. So there's some epidemiological evidence that patients taking statins have higher rates of checkpoint inhibitor related hepatitis. We know from the co-administration of TKIs, for example, in patients who've previously been exposed to checkpoint inhibitors, that they have higher rates of inflammation. There's a particularly interesting case of somertinib, a TKI which is minimally hepatotoxic on its own, but when given to previously checkpoint inhibitor exposed patients, then caused quite significant severe rates of hepatitis. But when you looked at the histology, it was a checkpoint inhibitor related hepatitis. It just seemed to have been triggered by the TKI administration, so that's the second. And the third, I suspect the microbiome, the data, there's good data now in colitis and just a few early studies looking at the role of different bacterial families in the initiation. So I suspect it's another triggering event, some local damp, some pathogen associated molecular patterns from the gut is just providing enough stimulus in order to trigger inflammation in an environment in which the checkpoint pathways, which are normally so important for limiting hepatic inflammation and engendering that tolerant environment in the liver, have been blocked, and then you just have a trigger and you get this unchecked inflammation. Thank you. I think a similar question to bile salt and export pump inhibitors. You mentioned in the last slide, or one before last, that co-medication should be really something you need to look out for. So is there any data experimentally where two medications in together, other than checkpoint inhibitors, would generate a bile salt-induced liver injury? Yes, excellent question. So there is in vitro data where co-administration of two different compounds, one that actually inhibits a transporter or competes for that transporter at various concentrations can show this enhanced toxicity in vitro. So I think compounds that might affect the regulation or the efflux of the bile acids, the MRP3 for possibly OST alpha beta, for example, as well as BSAP, of course, would be important. Yes, good question. Go ahead, this microphone. Okay, thank you. Morven Canningham, Toronto. A brief comment and a question, if I may. Just following on from Dr. DeMartin and your comments about high doses of steroids, if you look at the most recent updates for the ESMO guidelines, in general, there's a real push towards early treatment with high-dose steroids because in all comers, that seems to reduce severity of adverse events. So I totally agree that we over-treat a lot of patients, but I think we have our work really cut out for us to convince the oncologists of that, and I think that's gonna require a lot of good data and collaboration, and Toronto's up for some collaboration as well, if anyone's interested. The question, very clinically focused. More and more patients are coming to my clinic on a checkpoint inhibitor and other anti-cancer therapies that could potentially also be hepatotoxic. How on earth, in the real world, do you practically start to tease apart whether these patients have checkpoint inhibitor hepatitis or whether they have another form of DILI and how you treat them appropriately because they want rapid resolutions they can get back on their anti-cancer therapy? Great question, also. Very, very great observation. So I think that the role of a liver biopsy is still actively debated. I think some groups think that liver biopsy can predispose to either holding off on steroids or even treating with reduced steroids. At the very least, my clinical practice is that if there's any sort of atypical situation, such as receiving treatment with an immune checkpoint inhibitor in addition to some other cancer agent that could potentially also cause a non-immune mediated drug-induced liver injury, that at that point, a liver biopsy can be helpful, with the caveat that while, yes, there are some very more or less specific findings of immune checkpoint inhibitor hepatitis on histology, oftentimes, the most common pattern is a very nonspecific lobular inflammation, which also could potentially be seen if you biopsied somebody with a drug-induced liver injury. And I think at that point, in my experience, oncologists are hoping for more rapid resolution, as you mentioned, or at least a reduction of their liver injury to at least grade one, where they can be retreated or re-challenged, depending on the agent. And I acknowledge that that's a very difficult situation. I think a biopsy's reasonable in those cases, but a biopsy's also not. I'm gonna quickly chime in here that in my mice model, it's different than your model. I see the granulomas, and also, I've had discussions with Dr. Kleiner, who's the Dillon pathologist, and in humans, in the Dillon samples, we see granulomas, we see these macrophages. So although there's no pathognomonic definition, histologically, there seems to be, and I don't know, David, if you wanna chime in here. This is an open discussion. It's a friendly group. But if there seems to be anything you can hang your hat on that this is more checkpoint versus a TKI, because that's a common scenario. Just a quick comment. TKIs tend to give you a pretty aggressive-looking hepatitis, although I saw that early on with checkpoint inhibitors. I think they step on them so quickly, or if they do do a biopsy, they do it quickly. And it does tend to be kind of a mild hepatitis with granulomas in the hepatocellular injury pattern. So while I agree it can be tricky, I think you can get hints sometimes from the biopsy that it's more one than the other. Yeah, and then this also brings up the issue of that I think Guru was also alluding to, is that are we seeing, and that's why I like the term illicy, because it kind of explains the mechanism better. This isn't checkpoint inhibitor-induced hepatitis. This is immune-mediated liver injury from a checkpoint inhibitor. It's an indirect injury. That's why that, you know what I mean? That is a better term for it. And because we all write papers and grants and we need short terms, I like that it's only four letters, right? As in CPI-hep, you know? Also, I know you're gonna argue that chili's short too. Where's Raul? So that's the issue. Like, is this sensitization to another agent or is it actually from just breaking immune tolerance and then a second hit, like a PAMP or a DAMP? It's really complicated to try to decipher that. And as you, I think, had in your background, the Mitushi paper, Jack Utrecht's work, where he basically uses this as a model for idiosyncratic DILI. He breaks immune tolerance and then gives INH or gives nevirapine and attributes it to the nevirapine and the INH. It's very complicated. Go ahead. Hi, my name is Tilly Verghese. I work at Arbutus Biopharma. So I had a question about more of the risk factors that predispose people to ICI-related hepatitis or immune-related hepatitis. So in terms of dose or frequency of these drugs, does that affect it? Or has there been any studies looking at specifically receptor occupancy of some of these types of targets that we're looking at? Is that maybe a reason why people are getting it? Because as we know, something like nivolumab can last in terms of like a half-life that's pretty long, right, like at least three, four weeks. So do you think that has a major risk factor for the development or it's not related at all? Yeah, so there's very, the incidence of liver injury is pretty well characterized across all the major immune checkpoint inhibitors. So we very clearly see that combination PD-1 with CTLA-4 therapy, by far the highest risk. That's what typically is done in melanoma patients, which is the patient population that immune checkpoint inhibitors and immunotherapy has been studied on the longest period of time. So we have the most data in that patient population. And then next step down is CTLA-4 monotherapy, which does not usually happen very much to my knowledge these days. And then one step below that is PD-1 or PD-L1 monotherapy. And so the reported rates of high-grade immune checkpoint inhibitor hepatitis are up to 10% with combination therapy, probably somewhere in the three to 5% for CTLA-4 monotherapy, and then right around one to 2% for all the PD-1 or PD-L1 monotherapies. Depending on what reference you look at. Yeah. It's all over the place, the rates at this point. But yeah. So do you think if the receptor occupancy of these drugs were decreased, you would have a decrease in immune-related events, or is it time-dependent? So that's also a really good question because while we sort of characteristically know that most immune-related adverse events happen within the first several months, and each immune-related toxicity affecting different organ systems are sort of a pattern to some that happen a little bit earlier, some that happen later. Hepatitis is sort of in the intermediate to somewhat later phase of when you consider all common immune-related toxicities. But there are plenty of patients who develop liver injury months, year after starting their immunotherapy. And even patients who have discontinued immunotherapy for three to six months, there have also been a reasonable number of patients reported to develop immune-trapped coronary hepatitis, which I guess sort of points towards the idea of the immunotherapy sort of priming the liver for some sort of injury, and then some additional hit sort of finally triggers the response. Thank you. I'll point out that the corticosteroids are not for the liver. They're for the immune system that you're treating, not the liver. Also, it's difficult to understand when you stop the checkpoint inhibitor. They're given every month. So you say, well, he got two cycles, and the month later he comes in, how long has he been off? He's not been off at all, because you give it every month. If you gave it every day, it's the same. So when you say when a drug stops, it should be the first day you didn't give it. That's the date. So this is a question. If you just got one shot of a checkpoint inhibitor, could you see injury three or four months later at the same rate as if you gave it every month? Because one of the interesting things is if when you treat it and they resolve the hepatitis, often they don't need further therapy. It's been done. The immune system, whatever you did to it, is done. It's the same with the checkpoint inhibitor. It doesn't go to the liver. It goes to the immune system. Very strange, isn't it? Yeah. So while, yes, patients can develop liver injury after just one dose, I don't actually know if people have looked at whether patients can receive one dose, come off, and then develop liver injury months later. Well, that's the issue with the TKI things that follow on. Is that due to TKI or to the shot that they received several months before? It would also be pretty, I think, unusual for a patient to receive only one dose unless there was a specific toxicity reaction. That's usually why people only receive one dose. Either that or they're, because of that, they cancel. I've seen a bunch of nivolumab cases, one dose. I think at least two. But obviously you see that within two weeks. So it's, you know, usually it's stopped. I don't have re-challenged data on that. Please. Sorry, changing the subject. Mark Furse from Imperial College. So Dr. Brewer, given the known mechanism, well, the proposed mechanism for TKI toxicity that you described, have you looked at peripheral bile acid profiles to see if that potentially can help unravel what is sometimes quite complicated, as we were talking about right at the beginning of this session, sorting out the cause of liver injury in these patients is somewhat challenging. Yes, excellent question. So you're talking about inpatients that have received the TKIs. Yes, we have not done that. We have, so it's very challenging because we know that diet influences bile acid concentrations. And even if you control for diet, still there's a lot of individual variability in the bile acid concentration systemically. But I do think this is an area that could be explored further because we would expect there to be some potential predictors, and we have looked at this in the NASH population, where there is some implications that patients with NASH may be more susceptible to DILI. And in fact, there are particular profiles of more toxic bile acids in the circulating, in a diet-controlled study. But we have not done that with the TKIs in patients. Whether it's more common is unclear. It is more severe if you have an underlying liver disease, and that's the problem. So I think it's early days in using the bile acids as biomarkers for toxicity. There is some work ongoing to look at bile acids as biomarkers for OATP function, for example. You may be familiar with that. Future work, always more to do. A practical question. In steroid non-responders, if you add the mycophenolate mufetil, when do you stop the treatment? When do you stop? Stop the treatment for steroid and mycophenolate. So my clinical practice, and this may not be the clinical practice of everyone here, is that once patients are shown to be responding to the mycophenolate, I start tapering the steroids first. I want to get the steroids off before anything else. And I keep the mycophenolate dose stable until the steroids are tapered off completely. Assuming the patient's liver tests are normal at that point, that's when I ease off on the mycophenolate. Okay, thank you everybody. I think we're gonna start the... Oh, usually the steroid taper occurs over a period of five to six weeks. Yeah, I would say a little longer than a month for me too. It takes a month to come off the steroid if they respond, and then a couple weeks. I don't immediately stop the mycophenolate. Yeah. I don't want to make official recommendations. I think the guidelines were written by Bob there. Bob, every other day, mycophenolate, and then you stop it? Over two weeks? Yeah, and a taper. I wouldn't just stop it immediately, even the mycophenolate, but like every other day, and then maybe stop it after a week or two.

immune-checkpoint inhibitor hepatitis

corticosteroid therapy

steroid-refractory disease

hepatology specialists

ICI therapy rechallenge

immune-related liver injury

hepatotoxicity forms

histological ambiguity

two-hit model of injury

risk factors

management strategies

bile acid profiles

mycophenolate mofetil

combination therapy