Welcome to the ASLD Liver Meeting Digital Experience postgraduate course entitled Paradigm Shifts in Hepatology, Rethinking Conventional Wisdom. Together with my co-chairs, Kimberly Brown and Lawrence Friedman, my name is Ray Chung and I'm privileged to serve as chair of this year's 2020 postgraduate course. The practice of hepatology has undergone profound shifts driven by new pathophysiologic insights which in many cases has forced us to rethink traditional medical dogma. This is also an era of profound progress in diagnostics and therapeutics. This program explores these seismic shifts to allow attendees to re-energize their practices and embrace new goals and objectives for treating a breadth of liver diseases using these new approaches. The learning objectives of this year's course are to identify important conceptual changes that have altered our management of liver diseases, to apply insights from pathophysiology to the selection of new treatments for patients with liver diseases, and to apply new standards in the quality of care to the practice of hepatology. The course will be divided into five core thematic areas today. First, the practice of hepatology has undergone significant change. Personalized medicine, the incorporation of genetics and omics, as well as artificial intelligence have all begun to become integrated into our care of patients with liver disease. The use of non-invasive measures with increasingly accurate performance continues to render biopsies and histologic information less important to practice. The incorporation of a multidisciplinary approach to the care of end-stage liver disease, including palliative care, has transformed our care paradigm for this condition. And telemedicine, which has of course been fast-tracked by COVID-19, is now and will be a fixture in the care of our patients. Rethinking the management of alcoholic-associated liver disease, including transplantation for severe acute alcoholic hepatitis, non-alcoholic fatty liver disease, a metabolic disease meriting a multidisciplinary approach, and thinking about the contribution of the microbiome to progression of metabolic liver disease will all be major themes covered in session two. Thinking about autoimmune liver disease, we have witnessed application of new classes of therapy for disease that has been traditionally managed by the blood instrument of simple immunosuppression. New immune-mediated hepatitis has become a much more commonplace phenomenon with evolution of checkpoint inhibitor immunotherapy for cancer. The management of complications of liver disease has also undergone paradigm shifts, including changes in the management of portal hypertension, hemostasis and coagulopathy, and of our concepts of acute on chronic liver failure. Liver transplantation has been altered by new allocation rules to treat and to level the playing field, and HCC has undergone major advances with the integration of combination therapy, particularly for advanced disease. Finally, in viral hepatitis, there have been tectonic shifts in our goals. Cure is now routine for hepatitis C, but there remain important management issues. For hepatitis B, new classes of direct-acting antivirals and other immunotherapy now make functional cure a real possibility. And of course, consideration of hepatitis A, hepatitis E, and now, of course, SARS-CoV-2 effects on the liver are all important considerations for this session as well. As with the liver meeting digital experience, the postgrad course is, of course, itself gone digital. So there are a number of new flourishes that offer potential advantages to our attendees this year. First, the postgrad course is available to all registrants as a one-size-fits-all package. This will make the course available to our attendees from all over the world who in any other year may not have been able to join us for a live meeting. Second, the postgrad course content will be available on demand for five days leading up to and for another 90 days after the meeting, assuring that that content will be available to attendees whenever they want. Continuing education credits will, as traditionally, be available. OCME and ABIM MOC points will be offered and can be claimed after the conclusion of the meeting. Finally, a live question and answer period attended by each of our speakers will be broadcast live at the conclusion of the rebroadcast on Saturday, November 14th from 4.30 to 5.30 p.m. Eastern time. This will be an opportunity that we've not been able to do in prior postgrad courses of engaging our speakers directly from the audience. So we hope you have a great time with the new format, and it's my pleasure now to introduce Dr. Nancy Rowe, who will moderate Session 1. Thank you. I want to welcome you to Session 1, The Changing Practice of Hepatology. On behalf of ASLD and the organizing committee, we really want to thank you for participating in the digital postgraduate course despite these unprecedented times. I'm Nancy Rowe from Rush University Medical Center, and I have the privilege of introducing my esteemed colleagues for this session. The first lecture will be The Evolution of Personalized Medicine and Hepatology Practice, presented by Scott Friedman from the Icahn School of Medicine at Mount Sinai. The second will be A Glimpse at a Biopsy-Free World, presented by Elliott Tapper from the University of Michigan Medical Center. The third is The Care of Patients with Advanced Liver Disease. It's a team sport. This is presented by Victor Navarro from Einstein Medical Center. The fourth is Measuring the Unmeasurable, Incorporating Quality Metrics into the Care of Patients with Cirrhosis, presented by Michael Volk at Loma Linda University Medical Center. And the last is Virtual Care, The Future is Now, presented by Sanjeev Arora from the University of New Mexico. Thank you again for joining us for this session, and I hope you enjoy this. Hello, this is Dr. Scott Friedman, and I'm honored and delighted to be the lead-off speaker in this year's ASLD postgraduate course. My talk is entitled The Evolution of Personalized Medicine and Hepatology Practice, and I hope it will resonate throughout the course with many of the other lecturers. This is me. For those who don't know me, I've been an ASLD and ESL member for many years. This is actually my 37th ASLD meeting, of course, the first one not in person, like all of you. My scientific interests are listed here and include a longstanding interest in cellular pathogenesis of hepatic fibrosis and liver cancer. And in recent years, I've become extremely interested in translating basic science and hepatology into therapeutic targets, novel diagnostics, and clinical trials. And that has led to my participation in a number of industry activities that are fostering this goal. These are listed on the program website, but none of the work from these companies is reflected in any of my slides today. Here's the overview or summary of comments I'll be making. First, to define what is personalized or precision medicine and what is its promise. And here, I'm using the term personalized and precision interchangeably, although as Costas Lazaridis described in his postgraduate talk last year, they are a little bit different because precision medicine typically involves specific genomic or genetic data, whereas personalized medicine can reflect just changes or approaches to patient care that may be individualized even if they don't rely on big data. Secondly, I'll review some of the techniques that underpin personalized medicine. I'll show some applications with respect to different liver diseases. We're all familiar with. And then just comment on where the future is and where we're going with the integration of personalized medicine into clinical hepatology. So let me start with a case study. This is a 34-year-old female with abnormal AST and ALT. She has a lifelong history of elevated triglycerides that were severe enough to precipitate pancreatitis requiring pancreatectomy and splenectomy. She's hypertensive and preeclampsia developed at 29. She has no family history of liver disease, no alcohol use, and no definitive diagnosis despite 18 years of ongoing evaluation. Her physical exam was notable for hepatomegaly, but no stigmata of cirrhosis. Strikingly, while she had a normal BMI, her triglycerides were markedly elevated, her HDL was very low, and her AST and ALT were elevated modestly, as you see here. We got a liver biopsy that showed severe macrosteatosis with moderate portal inflammation, ballooning and malaried dengue bodies with stage 2 periportal and sinusoidal fibrosis. I'll return to this case in just a few minutes. The promise of personalized medicine has been with us for many years as illustrated by this New York Times Magazine issue back in 2005. We are building slowly but surely on the promise of an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person. In other words, the right treatment for the right patient at the right time. Two of the outcomes will be to establish the cause or predict the risk of illness and design a personalized healthcare plan to prevent or detect these diseases early. In the case of therapies, to use medications and other treatments that would work best for each individual and avoid adverse effects. We've expanded that definition somewhat in recent years by recognizing that health information is critical in healthy individuals as well as in disease. That's underscored by this nice review from Eric Topol and others in Cell in 2017 that remind us that an increasing number of devices can collect high definition information as a part of a prevention strategy to define baseline health risks. These can be used as a baseline should abnormalities or deviations occur that might lead to disease and the requirement for deep disease characterization followed by precision therapy. It's really not only just disease detection, but in fact, health maintenance that is becoming increasingly the domain of personalized medicine as well. Why the drive to increase precision or personalized medicine? In large measure, it's because the cost of genome sequencing has gone down dramatically far in excess of what was predicted. This is often referred to as Moore's, I should say, compared to Moore's law, which defined the increase in computer speed. The drop in genome sequencing cost has been far in excess of that as shown here, where now the genome sequencing cost has dropped from over $100 million to less than $1,000. How do we use those technologies and what are the different applications? That's described here from an article in the New England Journal that was modified by Costas Lazaridis at last year's course. Just to remind you, genes are composed of exons, which encode for proteins, and introns or intergenic regions that do not code for proteins. Based on that organization, there are different depths of information that we can yield from different techniques. First, there's a gene panel where we're just interrogating specific genes, specifically in many cases for cancer to define genetic abnormalities in cancer. This is often called a gene panel. There are a number of commercial establishments now that can offer gene panel for defining the genotypes of different cancers. Here, the cost can be $1,000 to $5,000, depending on what you find on the internet, frankly. There's also whole exome sequencing. This is now widely used, and I'll give you a beautiful example very shortly, in which we're just looking to sequence those regions that are encoded by the exons. In other words, only protein coding regions. This can be much less expensive. It requires sequencing much less of the DNA or the mRNA or cDNA, actually, and can be quite informative in disease detection or definition. Then there's whole genome sequencing, where now we sequence across the entire genome, recognizing that sometimes disease-causing variants are not in the exonic sequences, but rather in the intron or maybe intergenic regions. And this, too, has gone down precipitously in price. The big challenge here is a massive amount of data that's obtained and the need to interrogate that data to yield meaningful information. Let me show you a beautiful example described by a group from Yale in last year's Journal of Hepatology of how they used whole exome sequencing to define or discover diseases in 19 patients with idiopathic liver disease. They did deep phenotyping and whole exome sequencing, again, looking for protein coding regions. They, over a quarter of the patients, that is five out of 19, had identifiable monogenic disorders, and these were the genes affected. What you're looking at here is a summary from that nice paper by Hakim et al., in which, first of all, notice that the presenting features were typically either lean NASH or cryptogenic cirrhosis. And here are the different genetic disorders that were uncovered among the five patients that had a definitive diagnosis established by whole exome sequencing. Let me return to the case and now let you in on the secret, which is this is actually one of the cases from that study by Hakim et al., and, again, this was a 34-year-old female with abnormal AST and ALT, with severe hypertriglyceridemia, steatosis, and fibrosis. She underwent whole exome sequencing, and the investigators detected a missense mutation in the gene encoding for PPAR gamma, a glycine 161 valine substitution, and based on its location in the exon, this is predicted to disrupt the DNA binding and, therefore, completely eliminate the normal function of the PPAR gamma, which is a protein, which is a nuclear receptor that binds to specific regions in the promoters of genes. And importantly, PPAR gamma is a critical gene for the differentiation of adipocytes, so failure to have active PPAR gamma should lead to impaired adipocyte differentiation. In support of that was the fact that serum leptin levels were quite low, and, remember, leptin, while produced by many cells, is primarily a product of the healthy adipocyte or adipocyte tissues, and so in the absence of adipocyte differentiation, there was a dramatic reduction or a low level of leptin, and so this patient had a diagnosis of what's called familial partial lipodystrophy type 3, which indicates that it's due to PPAR gamma inactivating mutations. It's called familial, but in this case, the patient had no family history of liver disease. The treatment here is remarkably straightforward now that we can offer patients leptin. This patient was administered recombinant leptin, led to normalization of triglycerides, HDL, AST, and ALT, and a reduced insulin requirement. So here, as in a quarter of their cases, the implementation of whole exome sequencing revealed actionable information that made a profound difference or improvement in this patient's health. So a dramatic example for sure, but one that really speaks to the growing availability and implications of being able to sequence the exome in patients with unexplained diseases. Another approach for personalized medicine is to help define subpopulations that may have different response profiles to specific drugs. So if we look at this cluster of theoretical patients highlighted by different colors, the majority of them are green, some are blue, some are light brown. It turns out that in this theoretical example, only the green patients, those that have a base of A in this region, would respond well to a given medication. So by screening, if we know that patients who do not have this genotype will not respond, we may avoid using drug or a particular drug in a population that genetically is predisposed to be resistant to that drug. Now, if that doesn't make sense to you, let me give you a beautiful example that is near and dear to our hearts, and that is the discovery of a genotype that was associated with reduced rate of sustained virologic response of HCV after alpha interferon therapy. This was a landmark paper that explained why some ethnicities were more or less responsive to alpha interferon therapy, and it turns out that patients with a TT genotype, particularly African Americans, were more resistant to therapy for hep C, and in fact, to the point where had we maintained alpha interferon as a mainstay of therapy, we might have tried to avoid using it in these patients because of their much lower rate of responsiveness or HCV clearance. Fortunately, of course, we're now in a situation where we have direct acting antivirals, so this information becomes much less relevant to clinical practice. Nonetheless, it should underscore conceptually why genetic or genotyping can be very informative in all kinds of disease, in this case, hepatitis C, in helping us determine which drug is the right drug and which drug is likely to be most effective, and of course, this approach can also be used to predict drug toxicities in some isolated cases. That topic was reviewed in detail last year by Bob Fontana, and I encourage you to take a look at his lecture from last year. Now, genetic variants don't necessarily cause disease, but they certainly can contribute to disease phenotype, and that's nowhere better exemplified than in the case of NAFLD or non-alcoholic fatty liver disease. This is from a lovely review by Treppo and Valenti earlier this year in Journal of Hepatology that highlights how different genetic variants can affect responses and development of NASH at different functional stages. So, for example, as shown here, there are variants in these three genes highlighted in the yellow boxes that can influence the intensity of inflammation. There are other genes that can influence the response of the hepatic stellate cells to become fibrogenic, in particular, TERT and PLA3. There are genes that can regulate the health or the pathways of lipid droplet remodeling shown here, yet others that influence glucose metabolism and insulin resistance, and others still that can affect lipid catabolism, and finally, other genetic variants that may influence the vigor or oxidative stress response. And so these different variants alone or in combination may contribute to the likelihood of disease onset and disease severity, but they don't operate in isolation. They operate in conjunction with clinical factors, and that's underscored as well by the Treppo and Valenti paper, in which they looked at the impact of PNPLA3 and body mass index on hepatic triglyceride content, ALT, and risk of cirrhosis, and what you can see for each of these three features, the increasing BMI influences the propensity to develop hypertriglyceride or high triglyceride content, ALT, and cirrhosis according to the genotype of a very well-known risk gene, PNPLA3. So in patients who have the risk allele, a GG genotype, combined with high BMI, those patients are more likely to develop high triglyceride in the liver, higher ALT, and cirrhosis. So it's now the integration of both genetic factors as well as clinical factors that ultimately influence the expression of disease. But it's not only DNA sequences and RNA that can influence disease progression or disease diagnosis. Let's not forget about the microbiome. This is a summary diagram from a beautiful paper from Rohit Lumba's group published recently in Cell Metabolism that show that in fact a microbiome signature can distinguish between healthy liver fibrosis and cirrhosis in part through analysis of stool and a machine learning technology. Now, this alone doesn't mean that we need to rely on it strictly to diagnose cirrhosis, but it does speak to the idea that these signatures may have embedded within them critical information or clues that tell us how different bacteria can influence liver cell function. Another beautiful example was shown at last year's postgraduate course by Baron Schnabel highlighting his Nature paper that showed that in patients with alcoholic hepatitis, they harbored a much greater prevalence of a specific enterococcus, enterococcus faecalis, and in particular, patients with alcoholic hepatitis had a strain of this enterococcus that elaborated a cytolysin. And in fact, in this beautiful study, Berndt and his colleagues showed that using phage that target bacteria expressing this cytolysin to neutralize it could actually attenuate alcoholic liver disease in animals. Finally, liver cancer has benefited increasingly from the development of other technologies that are precision-based, but rely on analysis of circulating components. And in particular, the so-called liquid biopsy can interrogate either circulating tumor cells, exosomes, or membrane-bound particles derived from the tumors, or cell-free DNA. And so any of these can now be detected in circulating blood and may obviate the need to obtain direct tumor tissue in order to make a precision diagnosis that might influence the choice of therapy. These are data from a recently accepted study by my colleague at Malsanay, Augusto Villanueva, and his group published in Oncogene that looked at the ability of cell-free DNA analysis and mutation profiling to determine what mutations might be present in the tumor. If we focus in the right-hand side here on these genes highlighted in blue, these represent target genes in the PI3 kinase mTOR pathway. And what they showed is that mutations in elements of this pathway correlated with the lack of response to serafinib, but not nivolumab, as shown here in the blue lines in both treatment groups. And so this tells us that precision medicine may extend to detecting tumor-derived mutations in cell-free DNA that may influence the choice of which drug we can use to treat primary liver cancer. There are other great examples. I've summarized many of the examples that were highlighted in last year's postgraduate course that was focused entirely on personalized medicine. There were examples of the genetic basis of disease, genome-wide association studies to discover risk genes, and these are predominantly in NASH patients, biomarkers that may help us detect disease, predictive clinical algorithms, those variants that might influence response to therapies, and I gave you a good example with the IL-28 SNP, drug dosing and effects that may influence the susceptibility to drug-induced liver injury, and of course, the microbiome examples that I also highlighted. So this leads me to ask, well, what's next? There are a number of elements of the future that we can begin to contemplate. We can use tissue and circulating genetic information to determine disease classification drivers and targets, as well as the risk of complications, particularly in hepatocellular carcinoma arising in serotic patients. The technologies can also be used to derive new targets and drugs, particularly in the disorders listed here. Genetic efforts or precision medicine technologies will help us discover new biomarkers and validate and implement them in clinical practice, and then a major challenge will be to implement and integrate these data into clinical care using the electronic health records so that physicians are given guidance and don't have to become geneticists. So my key takeaways are listed here. The integration of so-called omics data that includes not just genomics, but the metagenome, as well as epigenetics, which I didn't mention previously, combined with clinical information is steadily entering clinical practice and already useful in many liver diseases. Personalized medicine will slowly but surely integrate into clinical hepatology, but this will be evolutionary, not revolutionary, and we will only appreciate the impact in retrospect over the coming years. And in fact, barriers to implementation will wane as we develop better technologies to integrate information in the medical record. The cost of sequencing is dropping. Reimbursement is increasingly likely if we can prove that this actually improves health and reduces health care costs. And we will be able to disseminate and implement these new technologies over time, but this does remain a challenge. And finally, hepatologists will learn the language of personalized medicine and seek opportunities to implement this in clinical practice. This will require some level of education on the part of providers, much as we always assimilate new technologies and new language into our methods of clinical practice. So with that, I'll close. And again, thank the organizers for including me. Thank you. It's my privilege to be here at the post-grad course to talk about embracing the post-biopsy world and the conflicts of interest that do not apply to this talk. I wanted to launch in with the story of a 56-year-old woman who went to her PCP with epigastric pain with spicy food and coffee. She does not drink alcohol. She does have diabetes and a BMI of 37. Her PCP orders an ultrasound in labs. The ultrasound shows what looks like fatty infiltration and the labs are interpreted as normal save for an elevated ALT. Suspecting NASH, the PCP has read up and learned that this is a histological diagnosis and for that reason, our patient is referred to you for consideration of liver biopsy samples 150,000th of the liver. It is generally safe with a 0.1% rate of some serious adverse events, although 3 in 10 people experience pain. It is somewhat costly accounting for your or the radiologist time, the nurse, the pathologist, the bed, the facility fee, and that cost does not account for indirect costs such as the lost productivity that your patient incurs 100% of the time as well as the caregiver who shuttles them to and fro. So before you go and make an assessment of the benefits and costs, let's take a step back. Let's review the aims of liver disease evaluation. Our patient wants a diagnosis and from that diagnosis a prognosis or staging and with that information we can together make a decision about how to improve our patient's health and well-being. Let's review how biopsy fares for each of these aims in the context of today's alternatives. Diagnosis is an area where the value of biopsy has depreciated tremendously in recent years. Here's a table that we published a couple of years ago which summarizes the expected yield by diagnosis for all patients presenting to your care with elevated liver enzymes in the third column. In the fourth column what we do is we provide a pooled estimate of the diagnoses found after a patient with a negative history, physical and serological evaluation proceeds to biopsy. What you see is that the vast majority, somewhere upwards of 80% of patients will receive a diagnosis of non-alcoholic fatty liver disease or another condition which did not truly need a biopsy for diagnosis. With respect to staging or providing a prognosis, there's been lots of movement of late. The issue with biopsy here is that while a biopsy report gives you an estimate of fibrosis, in truth that estimate has confidence intervals that can be somewhat wide. There is tremendous variation between observers within patients that are compounded by technical difficulties wherein it turns out that you need a biopsy length typically greater than 2.5 centimeters to accurately rule in or out the presence or absence of advanced fibrosis. But many biopsies, particularly as we shift towards radiology, come in at less than 2 centimeters. Here is a paper from the group that established the MetaVir staging system and it's illustrative. So this is Olympic level hepatopathology and the interclass coefficient or agreement between these pathologists was pretty strong for the diagnosis of cirrhosis but not awesome for the overall score of fibrosis 0.78, steatosis 0.63 and by the time you get to portal tract inflammation it's 0.52 and there is significant variation between different experts. There is significant variation within patients. In a landmark study by Dr. Retsu in patients with NAFLD-NASH, it turns out that if you sample from the right lobe of the liver and you get a stage of 2, then the left lobe of the liver has a roughly 50% chance of yielding a different stage including 0 and 3. Both have radically different implications for patient care. And when we're talking about the stakes of that biopsy, for example inclusion into clinical trials, this problem gets magnified. In an eye-opening study published just a few weeks ago in the Journal of Hepatology, reviewing all biopsies that were used prior to inclusion in a clinical trial, it turns out that there was terrible agreement between different pathologists and that if you went based on one pathologist assessment, one in two patients could have been excluded due to under staging or under grading of their histologic activity. This results in both misclassification and a diminished treatment effect on trial. Finally, staging has been functionally dichotomized. So while there might be five stages 0 to 4, it turns out that understanding stages in terms of advanced fibrosis or not, F3, F4 or not, is what we really need to provide our patient with a prognosis. This was originally revealed to us by Dr. Paul Angulo and validated in multiple other data sets including outside the United States in Europe. So it turns out that there are alternatives to liver biopsy which can be dichotomized in a much more efficient fashion. Here is a picture from a paper that where we reviewed non-invasive methods and I will tell you a couple of things. There are two main ways to non-invasively assess liver disease severity using blood tests or imaging. The basic premise of blood tests rely on the notion that as one's liver disease progresses, the AST rises and the AST to ALT ratio also rises. Also, the platelet count falls as a function of decreased thromboepoietin or platelet sequestration. So for example, the fibrosis 4 index, which includes age or time with the diagnosis, the AST to ALT ratio and the platelet count, very powerfully communicates one's risk of advanced fibrosis. We also have imaging techniques which use shear waves to assess the viscoelastic properties of the liver. There is a point-of-care test called FibroScan or Vibration Controlled Transient Telestrography and Magnetic Resonance Elastography. While MRE gives better resolution across the spectrum of liver disease stages, there is equal ability for both of these techniques to identify cirrhosis, roughly equal to identify advanced fibrosis. And while FibroScan has an advantage of being performed in the clinic, MRE has an advantage of being more technically capable in people with markedly advanced BMI, for example, greater than 40. The availability of these tests have eroded the value of biopsy by efficiently excluding the number of people who need staging by biopsy. And it has raised the threshold to proceed to biopsy, reserving for that procedure only patients with indeterminate or confusing values. And this has tremendous implications for the future ability of tests to validate themselves against biopsy as a standard. We simply lack equipoise over biopsy versus non-invasive techniques at this point. Now when you're going to choose a test, I have highlighted the issues with the biopsy standard. But of course, these non-invasive tests have their own flaws. Of course, there are things that will reduce a platelet count or raise the AST that are not liver disease. And there are things such as hepatic inflammation, which influence liver stiffness, that do not reflect liver fibrosis. However, these things are not black boxes, and we understand their sensitivity and specificity. And with that information, we can use positive and negative likelihood ratios to guide our decision making in the context of the use of these tests. The basic currency of liver disease evaluation is confidence. We want something that provides us with clinical certainty, and we have to know the ins and outs and the pitfalls of these tests before we apply them. But we also have to think about how useful additional testing is based on the amount of certainty that we can provide for a patient in the context of their goals of care. Once we understand a patient's goals of care, we can establish for us and the patient the acceptable level of uncertainty that we need to move forward with managing their diagnosis based on the information in front of us. But the key thing is how do we get to that place? For example, our patient, with likely NASH, shows up with an epidemiology that yields a two to five percent pre-test probability of the presence of advanced fibrosis. That is high, but it's not high enough to start screening for varices. We need to get to a place that brings us towards enough certainty to manage her. And some tests only bring us part of the way, and others will finish the process. Before I get into exactly how these tests influence decision making, let's take a step back and review the Fibrosis-4, which is probably at this point the canonical serologic non-invasive test. This was a test that was validated against biopsy, but because everyone has these conventional lab values, you can use clinical outcomes to validate the cutoffs, as was done by Angulo in 2013, and Hagstrom very recently in gastroenterology. Cutoffs such as a Fibrosis-4 of 2.67 very efficiently identify whom among our patients with fatty liver disease are at risk for liver related events. Furthermore, because these tests are so widely available, they can be applied to populations in a way that elastography, and certainly biopsy, cannot. In fact, they are cannot. Dr. Canwall presented a tremendous study where she looked at the Veterans Affairs patients with non-alcoholic fatty liver disease, and provided for you and your patients an estimate of their risk of hepatocellular carcinoma as a function of their Fibrosis-4 and diagnosis codes. But in clinic, for the patient in front of you, we have to understand the test characteristics in order to apply them. The first thing that I will tell you is that if your index includes age, then that is clearly a confounding factor. And the test performance of this index is going to be best in people of middle age, but it will be lower in people who are older, age 65, or greater, as was shown in a nice paper a few years ago. Once you apply the cutoff, you can have a sense of how certain you can be about a patient's risk of advanced fibrosis. And in many cases, even when the test performs excellently in people aged 45 to 55, you still don't get to a place where you might be confident in their stage. For that reason, you have to apply a second test. And I'll leave this figure here for you to review to understand how a sequential test influences your new post-test probability of the presence or absence of advanced fibrosis. And you can see for each of the different pre-test probabilities how these imaging tests influence your certainty. And the trade-offs between 89 and 95 percent might not be so great, so it's easy to choose the cheapest or most available one. But the trade-offs between 24 and 42 percent for people who come in at relatively low risk are more obvious. The way that we apply these tests clinically follows in this algorithm. You ought to apply the cheapest, most widely available test. So the first thing that you do is you exclude patients who are low risk from further evaluation. If they have a low FIB4, there is no value in a FibroScan or MRE. But if they have a high one or an indeterminate one, you need to do another test to get towards clinical certainty. So you choose the best test based on the chance that it will succeed for your patient. And then you have to counsel your patient and anticipate and strategize for hiccups, whether or not you need a liver biopsy or you're going to proceed to manage the patient based on what their most likely health state is with appropriate screening or lifestyle management. One of the advantages of non-invasive tests is that it does not simply limit us to four stages in the way that a liver biopsy does. Because it turns out that these are continuous values, and the higher that you go, the higher the risk of adverse events. So we know that one of the best ways to gauge a patient's prognosis is through hepatic wedge pressures, which is invasive and costly and has its own technical difficulties. But it turns out that a liver stiffness obtained by FibroScan of greater than 21 efficiently identifies patients who are at risk of liver related events. And because this contrast with biopsy, it's something that we can repeat every year, we can follow changes when a patient responds to lifestyle management, stops drinking, or is cured of hepatitis C and we can tell that patient accurately what their prognosis is in a dynamic fashion. But biopsy has still carved out its own niche and I think it's worth talking about the few places where we are still recommending liver biopsy and how strong those recommendations might be. For the purpose of diagnosis our guidelines still say that we need a biopsy to diagnose autoimmune hepatitis but that does not mean that we are free from the difficulties of decision-making in this context. It turns out that the histological features which diagnose autoimmune hepatitis might be somewhat specific 81% but they are far from sensitive 40% and so in many cases we will still be left considering the diagnostic test characteristics of this procedure. It is still recommended that when a patient has identified hemochromatosis with a ferritin greater than 1,000 that we proceed with liver biopsy for staging and while guidelines and data may not have caught up all the way, it is conceivable that something like transient allostrography will be able to replace liver biopsy for the diagnosis of advanced liver disease. One of the issues to consider here is that fibroscan might be effective but magnetic resonance allostrography because it relies on MRI sequences might not be able to accurately identify advanced fibrosis because the iron interferes with signal acquisition. And even though our trials require liver biopsy because the FDA has set that improvement of liver biopsy is the best surrogate that we can use, i.e. improvement of NASH without worsening fibrosis, there is intense work underway to use our non-invasive tests and condition them on the criteria set forth by the FDA. So for example the fibroscan which can assess both liver stiffness and liver fat can be combined with the AST level to condition the value on the diagnosis of FDA inclusion criteria. Now whether or not this ultimately is accepted as a viable surrogate will be dependent on the kind of research that gets performed to associate tests like this with hard clinical outcomes. Additionally the repeated biopsy is also a barrier to trial inclusion and there has been a lot of work on assessing whether change in MRI indices correlates with improvement of histology. A recent systematic review has suggested that fat fraction as determined by the MRI decreases in a way consistent with overall histological activity of NASH. But yet this still requires validation against clinical outcomes likely to convince the FDA of its utility. The key takeaways of this talk will be that we will never fully replace biopsy. It will always be there to sort out difficult cases. But we will continue as we have to reframe the role of biopsy for today's patients because all tests are useful to the extent that they can influence decision-making case-by-case. Now we as doctors, clinicians make the decisions not tests and every piece of information that we receive is somehow flawed. We have to consider that when managing our patients. When using non-invasive tests we have to use sequential testing in order to stage disease and arrive at clinical certainty and every test, biopsy or not, is a surrogate that must be interpreted in the context of a patient's competing risks and outcomes. Thank you for your attention. Hello and thanks for joining me for this talk entitled Care of the Patient with Advanced Liver Disease, a Team Sport. My name is Victor J. Navarro. I'm the Paul J. Johnson Chair for the Department of Medicine and I'm also a transplant hepatologist with the Einstein Healthcare Network and a professor of medicine at the Sidney Kimmel Medical College of Thomas Jefferson University in Philadelphia, Pennsylvania. I'm funded by the Patient-Centered Outcomes Research Institute to conduct research that's described within this presentation. I have no other disclosures. During this talk I'd like to touch on four major topics beginning with the increasing burden of liver disease. Then I'll move on to discuss the impact of disease on patients and families followed by a discussion of teamwork and its value and end with some of the barriers that we face and how to overcome them in implementing the best teamwork approach to care for patients with advanced liver disease. The group at the University of Michigan collected data from persons aged 18 to 64 from a commercial claims database spanning six years, 2009 and 2015, and importantly what the group found is that there was an increase in the overall prevalence of patients with cirrhosis increasing from 0.2% up to nearly 0.3% for the year's study. Also very important in their work they identified that there was an increase in the costs associated with caring for patients with cirrhosis. In this important 2018 publication a group of investigators led by Baylor in the Dallas-Fort Worth area also looked at claims data and they found an increase in chronic liver disease related hospitalizations. This figure however is displayed to point out that as portrayed by DRG categories of illness and in the solid black line the most severe and extreme, patients who presented with liver disease were actually increasingly ill. So patients were presenting with more severe liver disease over time. So the main messages from the epidemiologic studies is that advanced liver disease is increasing in prevalence and patients are presenting with more severe advanced liver disease and health care costs attributed to treating patients with advanced stage liver disease are increasing. I'd like to discuss now the impact of liver disease. We're all familiar with very important data on survival. Compensated patients, that is prior to the onset of complications, have a median survival with cirrhosis of about 12 years with a 95% one-year 90% two-year survival. But after complications which develop at a rate of 5% per year that one-year survival drops to about 60% two-year to 45%. Importantly though survival is just one albeit important aspect that we should consider as an outcome that affects our patients. In fact there are other outcomes that greatly impact our patient's well-being. These include symptoms, quality of life, depression and distress, caregiver burden, finances and employment. I looked in the palliative care literature and found this important study of the symptom prevalence and quality of life of patients with end-stage liver disease. This was a systematic review and meta-analysis but an important survey and I adapted that the data in that paper into this figure to point out that patients who have end-stage liver disease are really very similar to other patients who have cancer, heart failure and end-stage renal disease as regards pain, insomnia, fatigue, depression and anxiety. Patients with advanced liver disease portrayed here in the aqua colored bars. Also we look in the liver literature and the group at Inova did a very nice cross-sectional study using the medical expenditure panel survey and their findings highlight that patients with chronic liver disease are less likely to be employed, have higher health care expenses, more distress and depression and lower quality of life. So the main messages from these outcome studies can be listed as follows. Advanced liver disease complications portend reduced survival. As in other chronic illnesses patients bear a significant burden, liver disease also causes financial, economic, emotional burden on patients and unfortunately time didn't allow me to show you some other wonderful work, caregivers as well. Quality of life is negatively impacted by liver disease. I'm going to talk now about the concept of teamwork in healthcare and I thought first it ought to be defined as best I could in looking at the literature and some years of experience and caring for patients with liver disease. I thought the best way to think about teamwork in healthcare, particularly in caring for patients with advanced liver disease, is providing care that meets all needs, very simply stated. We learned very much from the group in Dallas at UT Southwestern and in fact we've known for a long time that it's very useful to care for patients with advanced liver disease in a team-based format, bringing many different disciplines to the table. But in this important study it's confirmed that teams that we've built in our institutions have an important benefit to patients and in this case survival. What the group did is looked at their patients before and after the institution of a multidisciplinary cancer clinic and they found that in patients who were cared for after the institution of the clinic there was an improved survival compared to patients in the gold line that were treated before the institution of clinic. But really this is the epitome of team-based care as many clinics like this incorporate various stakeholders including not only we hepatologists but also our advanced practice clinicians, frequently interventional radiologists, oncologists, surgeons, case managers, and in some instances even supportive and palliative care providers. So the important point here is that multidisciplinary clinics for HCC now are shown to improve overall from a team-based format. I'd also like to discuss for a few minutes some very important work coming from Texas in which from the VA Medical Center in Houston a group of authors describe a conceptual framework for an integrated patient-centered care to approach. I think this is an excellent way to conceptualize the principles one might even say requirements of teamwork and I extracted this these quotes from the article because I believe they really form the basis and the platform for which we could build a team-based approach to caring for patients with liver disease. These requirements as I see it are that we have informed patients and caregivers and that we have receptive and responsive clinicians essentially engaged caregivers patients and doctors and providers and that we have a health care environment that supports the collaborative efforts of patients caregivers and their clinicians. I'll show you an adaptation from a figure taken in the paper. This conceptualizes the interplay of these three quotes these three elements and how they support what the authors call collaborative treatment planning for integrated advanced liver disease care. To me as I said I think this forms an excellent way to conceptualize team-based care for patients with advanced liver disease. For one we can achieve our patients being fully informed with accurate prognostic and therapeutic information and we have providers that give that information in an honest assessment on an ongoing basis. We create the collaborative treatment planning for integrated advanced liver disease care that these investigators were describing. And also when you get to that point you're able much more smoothly to discuss with patients the transition which unfortunately frequently occurs from curative treatments to supportive and palliative treatments. Now given that we can accept that there's value in team-based care I think it's important now to discuss what barriers exist that a lot that prevent us from optimally implementing it. To discuss with this coordination and fragmentation. Very simply stated patients can suffer from duplicative testing polypharmacy hopefully minimized now with interoperability improvements and conflicting treatment approaches. In the perfect situation we could provide perfect continuity of care much like what we might see in a solo practitioner's office where there's one practitioner one patient and family but in reality we know that in an end-stage liver disease where many providers are involved in many health systems patients are subject to fragmentation in their care and that leads to great frustrations on the part of the provider as well as the patient as I'm sure we're all aware. The frustrations on the part of the providers patients and delays in care are very important that aside it was the group at Michigan who provided us with this eye-opening study of 26,000 Medicare enrollees and they found that poor coordination of care was associated with increased mortality and hospitalization and higher likelihood of 30-day readmissions. An additional barrier to care is access. Access can be hindered based upon distance that patients travel. It may be the unavailability of hepatological care within their traveling distance or their institution. It may be also a hesitancy from their primary care or even gastroenterologist to refer to gastro to a hepatologist. Well this study I think was very valuable again by the group at Michigan that confirmed that access to care for chronic liver disease patients has deleterious effects when limited. In this study the group that looked at the VISN VA cohort data with liver disease patients over about 10 years and they compared survival among patients who did and did not have an outpatient GI visit and as you can see in the solid line survival was better compared to patients who had no GI visit. Finally as I move on to think of the various barriers to effective teamwork I think of debility. It's a phrase or term that I apply that I think captures three important concepts. Malnutrition, poor intake or uptake of nutrients, sarcopenia, reduced muscle mass and frailty. To me when I think of a patient and the commonality of all these problems I think put together they really lead to what I call debility. Now debility I think directly limits patient engagement. They don't feel well, often they have trouble traveling and this leads to less engagement in health care with our hepatologists. These three major areas frailty, sarcopenia, malnutrition, there's great data research done such as work done by Dr. Lai and colleagues that shows that frailty is extraordinarily common in patients awaiting liver transplantation and she's shown that this adds to the mortality risk and has been able to quantitate this for us. Sarcopenia, present in over half of patients with advanced cirrhosis, a very important predictor of mortality and also associated with increased health care costs. And finally nutrition, extraordinarily common, it negatively impacts outcomes. Lastly I want to talk about the concept of supportive care and lack thereof. As I reflect upon my experience as a clinician it's clear to me that supportive care really requires a team and that team includes not only the hepatologist and advanced care provider but also the social worker, dietician, behavioral therapist, physical therapist and pharmacist, the team that really can address all the patient's needs. The supportive care elements that this group is able to address is really quite broad, not only management of symptoms but also achieving better communication and engagement with patients. I find our social workers to be tremendously important in this area. Understanding the disease process and prognosis is our responsibility but also to give that information and understanding to the other team members so that the message can be repeated to the patients and families. And finally advanced care planning, understanding that in many situations there is the unfortunate outcome of death and in order to prepare patients for that we really ought to be discussing advanced care planning. And finally we must be prepared to manage caregiver and bereavement support. Well we've discussed this under the heading of supportive care but really if you look at what palliative care is they truly are one and the same. Palliative care is supportive care. So let's look to the palliative care literature and say well what is the value in palliative care even if we call it something like supportive care. Well patients with palliative care are less likely to be admitted to the intensive care unit or emergency room and there are good data now to show that palliative care improves important outcomes, that is outcomes that are important to patients, the quality of life, symptoms, and mood. And also palliative care is very helpful to reduce costs. I think it's important to highlight another very important potential outcome from palliative care utilization. This is from the New England Journal of a very important study conducted by Jennifer Temel at Mass General, a single center study which randomized about 150 patients to receive all received oncological care for non-small cell cancer but patients were randomized to receive early palliative care in their treatment or standard care and as you can see patients who received early palliative care had an improved survival. In fact the median survival was 30 percent longer for the palliative care group but there was also an improvement in other parameters. So we take away from this important study that early palliative care has the potential to improve quality of life, mood, and prolonged survival in advanced non-small, excuse me, non-small cell lung cancer and we hope that this could be extrapolated to other diseases. Against that backdrop of the potential value of palliative care we have to ask ourselves how we're doing in treating patients with end-stage advanced liver disease in the United States. This was a retrospective nationwide cohort analysis conducted by the group at Toronto using the nationwide inpatient sample. There were over 55 million hospitalizations that were looked at during six years between 2006 and 2012 and importantly as you can see in the solid line they compared to the liver disease patients to cancer patients there was an increase in the rate of palliative care referral from about one percent up to over seven percent by 2012, still lower than we see in patients with cancer. So it's important to say that access to palliative care is increasing for patients with advanced liver disease but given the potential benefits more work needs to be done in this area. So if we survey the various barriers to palliative care I think we can summarize them with this list. There are preconceived notions about palliative care likely being hospice care and this tends to drive patients and even providers away from palliative care, and there's a preferential focus on treatment. There's a discomfort to discuss it, there's lack of providers who are skilled in delivering it, and there also may be payment limits as well as barriers, both socioeconomic and ethnic or cultural. Given the complex needs of patients with advanced liver disease, best met with teamwork and care coordination, but also the value of supportive and palliative care which is underutilized in patients with advanced stage liver disease. My colleague Dr. Manisha Verma and I wrote the Palliver study. This study, funded by the Patient-Centered Outcomes Research Institute, is introducing palliative care within the treatment of end-stage liver disease, a randomized controlled trial. The Palliver study is a five and, thanks to COVID, now a six to seven year study in which our sites are divided between VA and non-VA sites, and within each group there's randomization to sites being assigned to its hepatologist rendering palliative care after the appropriate training, to patients who meet our entry criteria with advanced liver disease, and will be compared to sites that render palliative care through trained palliative care providers. Our main outcome is to show an improvement in quality of life among the patients who are treated by hepatologists rendering palliative care. The Palliver study comprises 18 sites around the United States, with our main site at Einstein and our coordinating center at the Duke Clinical Research Institute. Our main aim through this study is to demonstrate the value of supportive or palliative care in the treatment of patients with advanced liver disease, and hope that we can incorporate such an approach into routine liver care. I'd like to summarize some key takeaways. Increase in liver disease and its complexity levies great burden on patients and caregivers. Collaborative treatment planning, as described by the group at the VA Medical Center in Houston, is facilitated by educating patients and caregivers and discussing prognosis early, and teamwork facilitates collaborative treatment planning. As clinicians, we should aim to minimize fragmentation and optimize care coordination. Teamwork is the embodiment of care coordination, and we must find ways to incorporate supportive or palliative care earlier in the care of patients with advanced liver disease. Again, my name is Victor Navarro, and I'd like to thank you for your attention. Hi, this is Michael Volk. I'm going to talk to you about measuring the unmeasurable, incorporating quality metrics into the care of patients with cirrhosis. This is my bio. I work at Loma Linda, and I'm a health services researcher who focuses on improving health care delivery for patients with chronic liver disease. Here are my disclosures. Now, you might be wondering, why should I care about quality metrics? Is this going to be one more bureaucratic hassle I have to deal with when I'm just trying to take care of my patients? Well, first of all, let me tell the notion that American health care is the best in the world, necessarily. This graph shows life expectancy per country by per capita health care spending in international dollars. And you can see that for countries like Sierra Leone, Namibia, Iraq, you spend a little bit more money, and you get better life expectancy. Then you get to the more industrialized countries like Singapore and Japan, where spending more money does not necessarily yield gains in life expectancy. Finally, you see the outlier here, the United States. We spend three times what most countries spend on health care, and yet have lower life expectancy than many. By most metrics, we underperform compared to other industrialized countries. For example, medical errors kill more people than breast cancer. Less than half of patients receive proven treatments for common diseases. The problem is not with the people, it's with the delivery system. We have an acute care model that's poorly suited for chronic disease and prevention. We're not well-coordinated. There are multiple providers taking care of each patient in different systems with poor communication. We're inefficient. We do things manually that other industries have automated. And we're focused on disease treatment, not health. For example, in the hospital, it's easier for me to get an MRI than to get the patient out of bed. We're also suffering from burnout. Health care providers have burnout that's higher than the rates in the general population in other careers, and it's worsening with time. But when you think about it, the inherent job stressors haven't changed. Being a health care provider in 1970 versus 2020, we still had to deal with patients who were dying, life and death decisions. Colleagues were stressed out. All that's unchanged. So what has changed? My hypothesis is that it's the ratio of patient to non-patient time. Even in the last 15 years of my career and talking to people who are older, the amount of time that we spend actually taking care of patients is getting swallowed up by all the administrative hassles we have to deal with, driven by all the billing and compliance requirements, and exacerbated by electronic medical records that seem to dump a lot of the tasks that used to be taken care of by our staff back onto us. And so in my mind, the root cause is the financing system. The current fee-for-service structure limits autonomy and imposes administrative hassles related to billing, compliance, reimbursement. One solution would be population medicine. Imagine if we could serve as captains of the ship with a multidisciplinary team that delivers care in whatever way that we want to structure it. But the devil is in the details. We learned in the 90s that capitation doesn't work because it provides an incentive for physicians to reduce the amount of care that patients get rather than improve the quality. And so what we really need is a value-based reimbursement system. But in order to do that, we need to be able to measure quality, which is one component of value. So how do we do that? So what is quality health care? It's difficult to define, sort of like pornography. There's a famous Supreme Court case in 1962 in which the Chief Justice got very flustered when he was asked to define pornography. And eventually he said, well, I know it when I see it. But unlike pornography, people don't always know quality health care when they see it. So much of the way we think about defining and measuring quality was created by a guy named Avidis Danavidian. He was a University of Michigan faculty who divided quality health care into structure, process, and outcome. Examples in cirrhosis structure, a program that has transplant might be assumed to have better quality liver disease care than a program that does not, although that doesn't always translate into better patient outcomes. An example of process measure would be HCC surveillance. Do the health care providers send the patient for AFP and ultrasound every six months or not? Outcome measures can be divided into clinical, such as readmissions, mortality, or patient reported outcomes, such as ascites control or quality of life. It turns out that cirrhosis is an ideal target for quality improvement. It's a well-defined disease that's common, kills about as many people each year as diabetes or kidney disease, and costs as much as heart failure. There are many evidence-based treatments that we can apply to improve outcomes in patients with cirrhosis, but we don't do that well in delivering them. For example, only about 20% of patients get appropriate prophylaxis for fiercely bleeding, only about a third get HCC screening, 30% get secondary prophylaxis against SBP, only half get antibiotics in the setting of GI bleeding, and only about half get their antibiotics within six hours of getting diagnosed with SBP. A common case example that I'm sure you've seen is a 47-year-old man with hepatitis C cirrhosis who's getting weekly paracentesis, is not getting albumin with it. He's on 80 milligrams of Lasix, but only 25 milligrams of aldoctone. His ammonia levels are elevated, so he's on a protein-restricted diet, and thus getting increasingly sarcopenic. He is told to avoid Tylenol, but he's on high doses of NSAIDs. Not surprisingly, readmissions are common in cirrhosis. About 14% are admitted within a week, 37% within a month of discharge, and many of these are preventable. Finally, cirrhosis is a classic high beta condition. So these are conditions for which a small proportion of the overall patient population accounts for the vast majority of the cost. So because of all this, the practice metrics committee of the ASLD set out to develop a series of quality measures. And so we first did a scoping literature review to find what's in the literature. We then convened a multidisciplinary panel to review each of the measures and rate them according to a number of different qualities. One was, how well is the measure supported by the literature? Two is, what is the gap currently? So in other words, if a measure is something that patients are already receiving at high rates, then there's not much point in tackling that. Three is, what is the reach? So a measure that affects only a small proportion of the population is not as important as one that affects a large proportion. And then we also looked at things like potential adverse outcomes or unintended consequences. And then we looked at feasibility of measurement. We then convened patient focus groups to look at patient reported outcomes and rate those in terms of their importance to the patients. Things like fluid in the legs, in the belly, confusion, concentration, memory, itching, muscle cramps, etc. Now once we developed these series of measures, the measures now have to be specified. So for example, HCC screening, which is a process measure. You have a numerator, which in this case is patients receiving at least one abdominal imaging in the past year. Now why, you might ask, not every six months? Well, I think we'd all agree that if a patient gets it every seven months, that's adequate. So we have to define a cutoff that's less frequent than what the measure actually is. Then we have a denominator, which in this case is patients with cirrhosis. And then we have denominator exclusions. For example, patients with life expectancy less than one year due to non-liver related causes and perhaps others. Now some would argue that patient non-compliance or patient unwillingness to undergo screening could be, or should be, a denominator exclusion. But when you think about the reasons why patients are quote-unquote non-compliant, it may be that the test is being ordered at a location far from where they live, it's not being coordinated with their other visits, their co-pay is too high, etc. And so ideally, to support quality improvement rather than physician profiling, we want to take responsibility for the full scope from the intent to order a test to when the patient actually receives it. And then finally, ideally, each data element would exist in the electronic medical record in discrete format, meaning not free text, for ease of data collection. So once we developed these series of quality measures, the next step is to create a system for data management and quality improvement. And thus we created the Cirrhosis Quality Collaborative, or CQC, which is funded by the AASLD and currently involves 10 sites across the country. We have a software registry that allows data entry from each of the sites to be fed to a centralized location. And the data entry is via web forms from patients and also automated interface with the EMR in order to decrease reporting burden. In other words, the last thing we want to do is dump additional burden onto the clinicians. So this data goes into the registry and then spits out in the form of monthly QI and data quality reports. So you have things like run charts and statistical process control charts that facilitate the work of quality improvement, allow sites to compare between each other, and also dive down to the individual patient level to figure out which patients are not receiving the care they need. This registry also supports population management, being able to look at entire population, see how are we doing on different measures and what could we do differently, and things like pre-visit planning to maximize the impact of those in-person visits, and finally additional outputs for down the road such as research. So the next steps are expansion of the CQC, validation of the QI measures with the data we're collecting now. We then hope to start collecting cost data for the other part of the value equation. We're convening an ASLD industry colloquium with insurance companies and other relevant parties, probably actually because of COVID, it won't be until early 2022. And then we're partnering with local insurance companies to pilot test value-based reimbursement mechanisms. An example in the IBD world would be Project Sonar. So what can you do at your institution? Well, number one, you can foster a culture of safety and quality. When I lecture to the fellows every year on health care quality and safety, I use a series of examples of errors that I've committed throughout my career, and the purpose of that is to make it clear that it's okay, we're human, everyone makes mistakes. The key is reporting your mistakes and also trying to build systems to prevent the mistakes from harming patients. The next thing you can do is to start to build multidisciplinary teams. We know that multidisciplinary tumor boards really improve the care for patients with HTC. We know from transplant that multidisciplinary teams can really impact the outcome. And so you can start doing that in other areas of your practice. And then start collecting data on quality metrics. It doesn't have to be fancy. You can start with very simple things and then build from there. And finally, hopefully, join the CQC. We anticipate having open enrollment in early 2021. Thank you. Virtual care, the future is now. My name is Sanjeev Arora, and I serve as a professor at the University of New Mexico. I have no commercial interests related to the information presented here. In March of 2020, healthcare services were disrupted all over the world. Half the countries surveyed had partially or completely disrupted services for hypertension, diabetes complications, cancer treatment, cardiovascular emergencies. Rehabilitation services were disrupted in almost two-thirds of countries. And 94% of countries surveyed by the WHO felt that their workers were partially or fully reassigned to support COVID-19. Hepatitis elimination programs all over the world have slowed or stopped. Investigators from Europe calculated that a one-year delay in hepatitis diagnosis and treatment could result in additional 45,000 liver cancers and 72,000 excess liver-related deaths in the coming decade. Most mistreatments would be in lower-middle-income countries, but deaths would occur in high-income countries too. It was clear that COVID-19 is having an impact on much more than just deaths from infections. For cancers such as colorectal and breast, tens of thousands of excess deaths are expected between 20 and 30 because of delays in diagnosis and screening. Virtual care is needed now. This is a picture of live face-to-face consultation, what we call telemedicine. This is e-consultation where a specialist doesn't actually see the patient, but looks at the information about a patient and provides an opinion. This is remote monitoring where we can get data about a patient remotely without the patient coming to your clinic. And on the fourth, here is the case-based telementoring, also seen in the ECHO project. Telemedicine has been around for many decades, but yet its adoption has been slow. This study from the University of Pennsylvania tells us why. They did a telehepatology program which was piloted, and the key takeaways they had were that there was strong partnership between academic and community-based providers. Patients were satisfied and they had little hesitation to participate in telemedicine. Referring providers also felt that telemedicine was efficient and valuable, but the program could not continue because they did not see a business case, and that they were unable to bring the program to scale in the reimbursements that were offered by the government. Despite the lack of an adequate business case, there was widespread adoption of telemedicine as COVID-19 came along. In-person visits fell precipitously in March of 2020. For Medicaid beneficiaries, half of the visits occurred with telemedicine, compared to less than 0.1% prior to COVID-19. In the EPIC electronic medical provider database, 69% of visits occurred with telemedicine. These have been steadily declining as the economies have been opening up. In mid-July, it had declined to 21% of total encounters. What enabled this adoption was, of course, the desire to prevent virus transmission, but key waivers provided by the federal government of the United States and private payers played a catalytic role. There were no geographic restrictions for telemedicine. Pre-COVID, all states in the U.S. required telemedicine providers to have a state license. Patients and providers could do telemedicine from their homes, which was different from before. Reimbursement rates were raised so that providers got the same reimbursement as they would get for an in-person consultation. And phone-only consultation rates were also increased. There is, of course, significant uncertainty as to whether these reimbursement changes will persist after the COVID-19 pandemic is over. We know that telemedicine significantly engages patients who have barriers to care, patients who cannot travel. This is a study from Canada where they demonstrated that more indigenous patients got treatment, more patients with an history of injection drug use and incarceration could be treated with telemedicine with the same SVR rates. Another type of virtual care is electronic consultation. In this, the specialist does not actually see the patient. What happens is the primary care physician sends the information electronically to the hepatologist who can either render an opinion, direct a course of action, order additional tests, or request a face-to-face visit. E-consults have been a major priority of the American Association of Medical Colleges called Project Core. We have seen that e-consults result in effective triage and more efficient use of specialist resources. It offers great convenience for patients by avoiding travel to see a specialist, helps overcome long waits for specialty visits, reduces wait times, but also improves access. CMS now pays 0.7 relative value unit. This reimbursement is seen as low and has been a barrier for widespread adoption of e-consultation. There is another larger elephant under the table. This is that there is a global shortage of specialists, which is not resolved by virtual care. Deploying a specialist over a camera does not overcome this shortage. The American Association of Medical Colleges has estimated that in the U.S. in 2020, there is a shortage of more than 40,000 specialty physicians, which will go up to 65,000 by 2025. Globally, the shortage of specialists is much more severe. We believe that as many as 6 billion people in the world don't have access to the right knowledge at the right place at the right time by not having access to an appropriate specialist. In the U.S. alone, we see that this specialty demand is accelerating. In the last decade, there has been a three-and-a-half-fold increase in need for specialty consultation for Medicare patients. In safety net hospitals in the U.S., wait times are 6 to 12 months for an initial visit with a patient. For hepatologists, we see increasing rates of obesity, NASH, cirrhosis, hepatocellular cancer, and there are only about 1,000 hepatologists in the U.S. How will we meet this demand? The lack of the right knowledge at the right place at the right time can be fatal, not just for liver disease. 1.6 million people die of TB every year, despite the widespread availability of free medicines and testing in many, many countries all over the world. In Africa, a million people die of bronchitis and pneumonia. 1,800 children die every single day of diarrheal diseases. When we know that oral rehydration and some simple antimicrobials can prevent these deaths, unfortunately, the knowledge is not available where it is needed. Progress of hepatitis C elimination programs can also be accelerated if we figured out a way to democratize the knowledge of specialists. The World Health Organization reports that only a minority of the 71 million people living with hepatitis C know their diagnosis. Less than a fifth of those diagnosed have been treated. We know they're effective, and in 99 countries, the prices are very affordable and the prices are declining all over the world. Yet our progress has been slow. We have an opportunity to improve the diagnosis and access to treatment and save millions of lives. The severe shortage of hepatologists lead to many additional deaths globally. We know that primary care clinicians need training and support for interpreting hepatitis testing, for accurately making the diagnosis of cirrhosis and distinguishing between the various causes, for screening protocols for liver cancer, for management of the complications of cirrhosis, and assessment for the need for liver transplantation. Knowledge shifting is the only way to ensure that patients have access to the right knowledge at the right place at the right time. We have to train primary care clinicians and nurses to have the knowledge they need to provide the care. Project ECHO's strategy is to move knowledge instead of moving patients and providers. ECHO was founded in 2003. I was facing an eight-month wait to see me in my hepatology clinic at the University of New Mexico. Out of more than 30,000 patients in New Mexico with hepatitis C, less than 1,500 had been treated and primary care clinicians were unwilling to use regulated interferon and ribavirin. We set up ECHO based on four key ideas. First amplification to use the technology to leverage scarce resources, which were a hepatologist, a psychiatrist, and a pharmacist. We shared our best practices and set up 21 new centers of excellence for treating hepatitis C in New Mexico prisons and rural areas. We gave them our protocols, the AASLD protocols, but they were reluctant to give interferon in their rural clinics. So I asked myself, how did you become an expert in treating hepatitis C? When I did my fellowship in Boston, I would see a patient present to my professor, see another one present to my professor, and two years later they started calling me a gastroenterologist I said, aha, I'm going to use this model to create new hepatitis C experts in the United States, in rural New Mexico. And we would use a web-based database to monitor outcomes. In ECHO we trained physicians, physician assistants, nurse practitioners in hepatitis C, trained them to use our web-based software. And in 2003, we started conducting these teleacquisitions, which were knowledge networks in which all teach and all learn, very much like my fellowship program. This is a typical tele-ECHO clinic where all 21 of these primary care clinicians would join and present cases to each other. And we would all discuss them together and learn from each other. And we would use 15 minutes of our two hours to give them a lecture on hepatitis C. What we found was that in 12 months, they all became experts through a process called a learning loop. They learned from our advice, from our lectures. They learned from each other, but mostly they learned by doing through the guided practice model. ECHO is different from telemedicine. In telemedicine, a specialist sees a patient over a camera and helps manage the patient. It does not, however, increase the capacity for more and more patients to be seen. The purpose of a tele-ECHO clinic is capacity expansion, where an expert hub team trains primary care clinicians all over a rural area, who in turn then help manage hundreds and then thousands of patients. We also demonstrated in other studies published in hepatology that ECHO improves self-efficacy of clinicians. In question number three here, but what is your ability to treat hepatitis C and manage side effects? Where one, I have no skill and seven, I'm an expert who can teach others. The score goes from two out of seven to 5.2 in 12 months. And you as a primary care clinician now serve as a consultant within your clinic, 2.4 to 5.6. This was so effective that the weight in my clinic fell from eight months to two weeks over an 18 month period. Their overall competence went from 2.8 to 5.5. We knew clinicians were benefiting because they told us that achieving competence in caring for hepatitis C patients was beneficial to them. It had diminished their professional isolation. It had enhanced their professional satisfaction, 4.8 out of five, and had expanded access to HCV treatment in their community. In the New England Journal of Medicine, we published evidence that these rural providers and prison primary care clinicians could produce the same level of care as university specialists. For genotype one, two, and three with pegylated interferon driver virin and minorities had much greater access to treatment. We then expanded ECHO to a large number of other disease areas like diabetes, cardiology, chronic lung disease, bone health, HIV, high risk pregnancy. We currently have ECHO for about a hundred different healthcare conditions with learners in 165 countries. The goal of ECHO is force multiplication. To overcome the great shortage of expertise in the world, we want a nurse practitioner to be able to provide the same level of care of a specialist for the most common complex problems affecting their societies. For hepatitis C, we expanded our hepatitis C network, but much work needs to be done. We currently have 31 hubs in the United States linked to thousands of clinics and millions of patients could be treated if we expanded ECHO all across the United States. In the world currently, we have 48 hubs in all different continents just for hepatitis C treatment. For our other disease areas, we have 246 hubs in the United States. Most of the major universities in the United States have adopted ECHO. Worldwide, we have 401 academic hubs connected to learners, about 750,000 learners in 165 countries. We know that ECHO works for hepatitis C. This is evidence from Argentina, where they mentored 25 providers in all different parts of Argentina, 437 patients were treated, cure rates were 96.4% in the ECHO patients, and there was a significant improvement in all the evaluated skills and abilities of participants. The Veterans Administration published evidence that the rate of primary care provider initiated antiviral medication was 21.4% among patients that participated with an ECHO provider versus 2.5% amongst unexposed patients with similar SVR rates. A really important paper was published in Hepatology for patients with cirrhosis of the liver. When ECHO was used, patients exposed to ECHO had a 50% mortality compared to those that were taken care of by providers not participating in ECHO. This is the Indian National Association for the Study of Liver Disease with PGI Chandigarh Doing ECHO where they set up 25 district hospitals to treat hepatitis C with the generic direct antiviral agents, the overall SVR rate was 91%. More recently, they have told us that about 88,000 patients were treated this way, published in Journal of Hepatology in 2019. We also have launched ECHOs in prisons where we have trained 1,000 prison peer educators to prevent the transmission of hepatitis C in the correction system. Overall, we have about 290 peer-reviewed publications showing that doctors like to participate, 67 showing that their competence improves, 33 that patient health outcomes improve. ECHO has the potential to improve quality safety, rapid learning, reduce variation in care, improving access for rural and underserved patients, improving professional satisfaction. What we need to do is to democratize the knowledge of the hepatologists in the room today, the hepatologists attending this conference, so we can help many more patients around the world. There has been a massive growth of ECHO since it started. In 2019, we had 237,000 mentees in the 165 countries we participated in. In 2020, the first nine months, that number is close to 750,000. What makes ECHO work is team-based care. Technology reduces the cost of collaboration. We need to demonopolize knowledge of experts in an all-teach-all-learn platform. Task shifting makes all healthcare workers work at the highest level of their human potential with interprofessional consultation, guided practice, and mentor-mentee relationships. ECHO also produces joy of work by building communities of practice. In summary, what I want to say is that virtual care, e-consults, telemedicine are here to stay. We cannot put the genie back into the bottle. Although telemedicine visits are experiencing a decline from the peak of utilization, it is likely to stabilize at much higher visits levels than at baseline. Further refinement of the business case is necessary to facilitate widespread adoption of e-consults and telemedicine all over the world. And lastly, my favorite subject, it is essential to democratize the expertise of hepatologists around the world to improve care for liver disease and accelerate the progress of hepatitis elimination programs worldwide. Thank you for your attention. Hello. I'm Mara Rinell, and I'm going to be moderating the next session, session two, entitled New Concepts in Metabolic and Toxic Liver Disease. There will be three talks in this session. The first of which will be called NAFL Beyond Obesity. It will be presented by Kathleen Corey of Mass General Hospital. The next talk, entitled Rethinking Management of Alcohol-Associated Liver Disease, the other fatty liver epidemic, will be given by Mack Mitchell from UT Southwestern. And then to close the session, Dr. Anna May Deal from Duke University will give a talk on the gut liver access, what the clinician needs to know. Thank you. I hope you enjoy it. Hello. My name is Kathleen Corey, and I'm the director of the Massachusetts General Hospital Fatty Liver Program and an assistant professor at Harvard Medical School. Today I'll be talking to you about non-alcoholic fatty liver disease beyond obesity. I want to thank the organizers for the opportunity to speak with you today, and thank you all for being here. This is my biography, and next are my disclosures. So non-alcoholic fatty liver disease and obesity often go hand in hand. We know that individuals with obesity have significantly increased risk of NAFL development, and we know there's a dose-dependent relationship between body mass index, or BMI, and NAFLD. So our current paradigm often centers around obesity leading to NAFLD and NAFLD treatments focusing on obesity. However, when we think about lean NAFLD, we're forced to change our paradigm. It requires us to broaden our approach to NAFLD, because while obesity is one of the many risk factors for NAFLD, and especially in lean NAFLD, it is not required for NAFLD development or progression. NAFLD is frequent in lean individuals, and it highlights the many metabolic contributors to NAFLD and the metabolic consequences of NAFLD. So let's start with a case. We have a 37-year-old man with a BMI of 22.8 kilograms per meter squared, so within the normal weight category. He presents with elevated liver enzymes and increased echogenicity on ultrasound consistent with fatty infiltration of the liver. His workup is negative for secondary causes of fatty liver disease, causes of elevated liver enzymes, including alcohol use, and he's diagnosed with primary non-alcoholic fatty liver disease. For staging, he undergoes a vibration-controlled transient elastography, VCTE, which finds a liver stiffness measurement of 10 kilopascals, which is elevated. An evaluation for metabolic syndrome and NAFLD risk factors is notable for dyslipidemia, characterized by a low HDL, elevated triglycerides, and an elevated LDL, an A1c of 5.8%, putting him in the prediabetic range, and a blood pressure of 133 over 84, consistent with mild hypertension. So during this talk, we'll talk about several things, how to define lean NAFLD, how prevalent is lean NAFLD in adults and adolescents, what metabolic conditions should we be concerned about in our patients with NAFLD, and what complications should we be worried about, including liver-related and cardiovascular disease-related outcomes. There's several lean NAFLD definitions. In general, lean NAFLD is NAFLD that occurs in adults or adolescents without overweight or obesity. But as you'll see in the literature, the definition of lean can vary significantly. Some groups define lean as non-obese, with a BMI of less than 30. Some groups define lean as non-obese and non-overweight, with a BMI of less than 25. The World Health Organization defines non-obese for Asian adults as a BMI of less than 23, and for non-Asian adults as a BMI of less than 25. How prevalent is lean NAFLD? Well, two recent systematic reviews and meta-analysis evaluated the lean NAFLD prevalence using that World Health Organization definition, so less than 23 kilograms per meter squared for Asian adults, and less than 25 kilograms per meter squared for non-Asian adults. The overall global prevalence of lean NAFLD was 5.1%. Among the lean population, this rose significantly to 10.6%. And the prevalence of lean NAFLD among all those with NAFLD is nearly 1 in 5, so 19.2% of those with NAFLD actually have lean NAFLD. What about in adolescents and young adults? Well, studies vary, but a recent meta-analysis found approximately 2.3% of normal weight adolescents have NAFLD, while a study from NHANES found the prevalence of lean NAFLD to be as high as 8%. In this study from NHANES, when they compared lean NAFLD to their non-NAFLD counterparts, those with lean NAFLD had significantly more metabolic disease. They were older. They had lower HDLs, more frequently had high triglyceridemia, and had more insulin resistance. They also noticed that non-Hispanic white lean adolescents were more likely to have NAFLD than their non-Hispanic black counterparts. Several studies have evaluated lean NAFLD by region. Lewinol evaluated the prevalence of lean NAFLD in North America, Europe, Asia, and Oceania. They found the lowest prevalence of lean NAFLD in Europe at 2.2% and the highest prevalence in Asia at 4.8%. This second study, by Yi et al., was able to look more granularly at the presence of non-obese NAFLD by country. Now make note that I'm saying non-obese NAFLD for this study, as they defined non-obese or lean NAFLD as a BMI of less than 30, which is different from many studies. But you'll see that they found very high prevalence levels, as high as nearly 70% in Mexico and in Sweden, and as high as 44% and 43% in mainland China and the US, respectively. So again, emphasizing that we're seeing increasing levels of NAFLD, even in those without obesity. What's very important to note for our patients with lean NAFLD is that metabolic complications that can lead to serious morbidity and mortality are very frequent in this population. So patients with lean NAFLD have a prevalence of type 2 diabetes of 12%. More than a third have hypertension. 32% have the metabolic syndrome. More than half have dyslipidemia. And 42% have central obesity. It's been shown that lean NAFLD has altered high-risk cardiometabolic profiles, including atherogenic dyslipidemia, that are similar to those in NAFLD with obesity. What about the prevalence of NASH and fibrosis among those with lean NAFLD? Well, these data are more limited because of the requirements of a liver biopsy. But they do show a high prevalence of both NASH and fibrosis. Among lean NAFLD patients who underwent biopsy, between 39% and 50% in some studies had NASH. And 29% had a fibrosis stage of two or more. Now, there has been some controversy about whether lean NAFLD was associated with a higher risk of liver-related mortality or liver events. Hagstrom et al. evaluated this and did find that lean NAFLD tended to develop more severe or advanced liver disease. However, they did not find an increase in liver-related or overall mortality in lean NAFLD compared to matched controls. They did, however, find that older age, the fibrosis stage, and the modifiable risk factor hypertension were associated with increased mortality in lean NAFLD, things that we should be watching for in our patients with lean NAFLD to help us predict risk and also modify that risk. So how do we treat lean NAFLD? One question that we often get is, if our patients are a normal BMI, should we be asking them to lose weight? Is there any benefit to weight loss? And we certainly don't want people losing so much weight that they are at an unsafe or unhealthy BMI. This study by Wong et al set out to answer that question. This is a randomized controlled trial of an intensive lifestyle intervention in both lean and non-lean NAFLD. They included 154 NAFLD patients who were randomized to an intervention or a control group. And among the control group and the intervention, 39 individuals each had a BMI of less than 25. So they were able to analyze these by lean NAFLD. The patients were randomized to lifestyle intervention program for 12 months. They received individual education weekly for four months and then monthly for the additional eight months. They were counseled on increasing their energy expenditure, how to reduce their caloric intake. And they all received personally designed exercise programs. Now, as you can see here, as we see in many lifestyle intervention studies, there's a wide range of the amount of weight that was lost, ranging from less than 3% of total body weight to up to 10% weight reduction from baseline. What we see in the light blue lines are those with a BMI of less than 25. And we find, just like in obese NAFLD, a stepwise increase in NAFLD remission radiographically as weight loss increases. Overall, they found that for those with a BMI of less than 25, those in the intervention group, 67% had NAFLD remission compared to 18% in the control group. And in the intervention group, there was a significant decline in intrahepatic triglyceride levels when compared to those in the control group. Finally, although not statistically significant, there was a difference in liver stiffness measurement with a decline in the intervention group and a slight uptick in the control group, suggesting that perhaps longer duration therapy or increased weight loss may be associated even in lean NAFLD with an improvement in fibrosis. We see here that patients who lost 3% to 5% of their body weight, 50% had resolution of NASH. And for those who were able to achieve at least a 10% total body weight loss, 100% had remission of their NAFLD. So what this study did show was that even in lean patients with normal BMI, weight reduction is both possible and beneficial. It improved steatosis in a stepwise manner with the amount of weight loss. And there was a trend toward an improvement in liver stiffness measurement. And so what we advise our patients to start with is to aim for at least a 3% to 5% total body weight loss. This may seem daunting, again, for patients who are not overweight. But if we're thinking about a 170-pound patient, their goal weight loss is not dramatic. It's between 5.1 and 8.5 pounds. And that could lead to 50% NAFLD resolution. So asking patients to lose between 5 and 10 pounds may significantly help their lean NAFLD. What other recommendations do we make for patients with lean NAFLD? Well, we have them target a hypocaloric diet, again, with a targeted weight loss threshold of 3% to 5%. We ask them, like our patients with NAFLD and obesity, to follow a Mediterranean diet, high in omega-3 fatty acids, fruits, vegetables, lean protein, low-fat dairy, and minimized carbohydrates. We also ask them to minimize saturated fat intake, including red and processed meat, and ask them to minimize commercially produced fructose consumption, including sugar-sweetened beverages. Regular activity is also an intervention that may be particularly helpful for those with lean NAFLD. And we recommend up to 300 minutes of moderate-intensity exercise per week, or between 75 and 150 minutes of vigorous-intensity aerobic exercise weekly. Now, switching gears, what about cardiovascular disease in lean NAFLD? We know that NAFLD is an independent risk factor for cardiovascular disease, and is one of the leading causes of mortality in NAFLD overall. But is this something that we need to be worried about in our patients with lean NAFLD? Well, our group sought to evaluate whether NAFLD overall was associated with major adverse cardiovascular disease events in the PROMIS trial. This was a nested cohort study that we conducted from the PROMIS trial. And the PROMIS trial was a study of patients without known coronary artery disease who had stable chest pain, who were then randomized to receive coronary artery CT, or CTA, versus functional testing, and then followed forward for events. Those who underwent CTA, a proportion of them also had liver spleen imaging. We were able to capture who had steatosis at baseline versus a normal liver. Not surprisingly, we found that about 25.5% of the population whose liver was imaged did have steatosis. And these were compared to those with normal liver for the rates of MACE, which was defined as death, myocardial infarction, or unstable angina. What we found was that steatosis was associated with higher rates of major adverse coronary events, even after adjusting for baseline levels of coronary heart disease. We saw an overall rate of major adverse coronary events of 3.1%, and they were significantly higher in those who had steatosis at baseline compared to those without. But importantly, and for our purposes today, when this relationship was adjusted for the presence of obesity, it was not attenuated. So obesity did not seem to be mediating the relationship between steatosis and cardiovascular disease, suggesting that even in lean patients, cardiovascular disease, lean NAFLD patients, that cardiovascular disease may be an important cause of morbidity and mortality. So overall, hepatic steatosis, as assessed by non-contrast CT, was associated with a 70% increase in the risk of major adverse cardiovascular events, independent of traditional cardiovascular risk factors, and independent of obesity. We also conducted a similar study in a cohort of patients with biopsy-proven NAFLD to identify predictors of incident cardiovascular disease among adults with biopsy-proven NAFLD. We looked at 285 adults who were free from cardiovascular disease at baseline and had baseline biopsy-proven NAFLD and followed them forward for a mean of 5.2 years to assess for the presence of incident cardiovascular disease. Incident cardiovascular events occurred in about 9% over that time. And while smoking, albumin, and advanced fibrosis on multivariate analysis, as well as ALT, were significant predictors for future incident cardiovascular disease, again, obesity was not a risk factor for cardiovascular disease in NAFLD, suggesting that it is not obesity that's mediating or causing the increased risk of cardiovascular disease we see in our patients with NAFLD, and further suggesting that we need to be vigilant about monitoring for and risk stratifying our patients with lean NAFLD for cardiovascular disease. And finally, our group did compare the prevalence and incidence of cardiovascular disease by obesity status in NAFLD in an abstract that's being presented at this year's liver meeting. We compared CVD prevalence and incidence in NAFLD patients with obesity, 233, or NAFLD patients who were lean or overweight. We found that those with obesity versus those that those with obesity versus those without obesity had a similar prevalence of cardiovascular disease that was not statistically significant with an adjusted odds ratio of 0.97. There was also no difference in the incidence of cardiovascular disease by obesity status. Obesity then was not shown to be a significant predictor of incident cardiovascular disease. The only predictors identified for incident cardiovascular disease were age in the entire cohort, and importantly, in the non-obese cohort, LDL. So this work suggests that those with lean NAFLD are not protected from cardiovascular disease that is associated with NAFLD and obesity, and we need to be vigilant about risk stratification and treatment. So managing CVD risk in lean NAFLD includes lifestyle intervention, such as weight loss, diet, and exercise, statin therapy for all with a treatment indication, antihypertensive therapy, specifically ARBs and ACE inhibitors for those with hypertension who fail lifestyle intervention, and aggressive management of prediabetes and diabetes in close contact with endocrinology, as well as close interdisciplinary work with cardiology to reduce cardiovascular disease risk. So lean NAFLD helps us make some subtle paradigm shifts in our thinking about NAFLD. We need to make sure that we're considering NAFLD and even NASH and progressive fibrosis in lean individuals. And similar to those with obesity, lean NAFLD patients are at equivalent risk for liver and cardiovascular disease-related outcomes, and multidisciplinary teams are valuable to help manage the metabolic comorbid disease seen in lean individuals. Thank you. Welcome to the next session of the postgraduate course for the American Association for the Study of Liver Diseases. Dr. Ranilla, thank you for your introduction. I'd also like to thank the organizers of the postgraduate course, Dr. Kim Brown, Dr. Ray Chong, and Dr. Larry Friedman for inviting me to speak about rethinking management of alcohol-associated liver disease, the other fatty liver epidemic. My name's Mack Mitchell, and I'm joining you today virtually from the University of Texas Southwestern Medical Center in Dallas, Texas. Here are my disclosures. So I want to begin my talk today by describing a typical patient who any of us might see in our practice. The patient was a 49-year-old man with a BMI of 29 who had a 20-year history of drinking more than eight ounces of whiskey almost daily. On Christmas Eve in 2013, he was admitted and treated in the emergency department for an ankle fracture, and he was noted to have minimal elevation in his serum bilirubin that was felt to represent alcohol-associated liver disease. He was advised to stop drinking, or at least to cut down, and was sent home for follow-up with orthopedic surgery. I'll come back to this patient in just a little while, but first, as the title of my talk implies, alcohol-associated liver disease may be the other fatty liver epidemic, but what I hope to show you today is that ALD is a more common cause of both morbidity and mortality from liver disease than its close cousin, non-ALD. And its close cousin, non-alcoholic fatty liver disease. While liver-related deaths from hepatitis C have decreased steadily due to the availability of highly effective and well-tolerated antiviral therapy, the deaths from NAFLD increased by 11%, and the deaths from alcohol-associated liver disease increased by 5% over the last few years. As you can see on this slide, the absolute number of deaths due to alcohol-associated liver disease is more than nine times greater than the number of deaths from non-alcoholic fatty liver disease. So what are the similarities, if any, between these two types of liver disease? First, obesity increases the risk of both alcohol-associated liver disease and non-alcoholic fatty liver disease. Plus, both the histological features and the genetic risk factors are similar. For example, a single nucleotide polymorphism in patatin-like phospholipase 3, or PNPLA3 as it's usually known, increases the risk of both NAFLD and cirrhosis due to alcohol. Likewise, a splice variant in 17-beta hydroxysteroid dehydrogenase 13 protects against both non-alcoholic fatty liver disease and cirrhosis due to alcohol. It's interesting to note that both of these proteins coat the lipid droplet, which may be providing a genetic clue regarding the mechanism underlying fat accumulation in both conditions. So finally, when we try to separate the two conditions, it can be difficult in obese patients. By definition, patients with non-alcoholic fatty liver disease cannot have significant alcohol consumption, which is usually defined as drinking more than 14 drinks per week for women and more than 21 drinks per week for men. Several studies have actually shown a direct relationship between alcohol consumption and the risk of chronic liver disease. These two graphs illustrate the relationship between the average daily consumption of alcohol and the hazard ratio for the risk of developing cirrhosis. Unfortunately, there doesn't appear to be a threshold safe level of alcohol consumption in women. And in men, the threshold appears to be less than one drink per day. Now, if we superimpose the levels of alcohol consumption used to separate NAFLD from ALD, we can see that the risk of cirrhosis has already increased, even with the low amounts of alcohol consumption that are considered to be below the level required to diagnose alcohol-associated disease. This observation raises concern that even low levels of daily alcohol consumption may increase the risk of cirrhosis, particularly in obese patients. So we all recognize the different stages of pathology due to alcohol. But what's worth noting is that while 80% of heavy drinkers develop steatosis, a much smaller percentage, somewhere around 30% to 40%, develop steatohepatitis. And even fewer, around 20%, develop cirrhosis. So much of the current research in this area is centered on identifying those factors that lead to the progression from steatosis to steatohepatitis and cirrhosis, and ultimately, even hepatocellular carcinoma. Steatosis is actually a relatively common finding in patients who drink at hazardous levels. But as you can see in this table, the annual rate of progression to cirrhosis is only 3% compared to a rate of 10% in those with steatohepatitis. The total annual mortality rate in patients with steatosis is similar to that seen in patients with cirrhosis. But the cause of death is actually far less likely to be related to liver disease and far more likely to be related to cardiovascular or other causes. However, the annual total mortality is much higher in patients with steatohepatitis, with a similar percentage of those patients dying from liver disease as we see in the patients with cirrhosis. So steatohepatitis is really an asymptomatic condition that develops slowly in response to a variety of injuries, including alcohol. On the other hand, alcohol-associated hepatitis or alcoholic hepatitis, as it was previously known, is a systemic condition that develops on a background of steatohepatitis. Often with advanced fibrosis or cirrhosis. These patients have features of the systemic inflammatory syndrome, or SIRS, even in the absence of infection. In those patients with severe cases of alcohol-associated hepatitis, there appears to be a failure of liver-specific functions, such as the excretion of bilirubin, the synthesis of albumin, the activation of clotting factors. And there's also evidence for impairment in liver regeneration. And although the reasons for this are unclear, there is accumulating evidence to suggest that both of these may be linked to ongoing inflammation and the effects of the pro-inflammatory cytokines, such as interleukin-1 beta. Several years ago, a working group of investigators conducting trials that were sponsored by the National Institute of Alcohol Abuse and Alcoholism reviewed the criteria for diagnosing alcohol-associated hepatitis so that these could be used as inclusion criteria in clinical trials. These criteria actually serve as a useful way to identify patients with alcohol-associated hepatitis. However, confounding factors will still lead to a need for liver biopsy to confirm the diagnosis in around 15% to 20% of those patients with suspected alcohol-associated hepatitis. So let's come back now to our patient. Six weeks later, he was admitted with severe jaundice, malaise, nausea, and anorexia, despite a slight reduction in his drinking. So in addition to his jaundice, he had low-grade fever, tachycardia, and an elevated white cell count in the absence of infection, all of these being manifestations of SERS. So he met criteria for severe alcohol-associated hepatitis. But unfortunately, despite therapy, his condition worsened. And he eventually succumbed to multi-organ dysfunction syndrome, a common reason for death in patients with severe alcohol-associated hepatitis. So let's now consider what might have changed in those six weeks since his previous emergency room visit, and what options, if any, exist for his treatment. So the pathogenesis of alcohol-associated liver disease is very complex, however, translational studies in humans and animal models have both helped our understanding and the identification of potential targets for therapy. We've known for a long time that ethanol is a direct hepatotoxin, and although the mechanism of injury is uncertain, oxidant stress, mitochondrial injury, and ER stress have all been demonstrated to occur. Ethanol also directly damages the gut barrier, increasing permeability and altering the microbiome. The release of endotoxin and other bacterial products stimulates the innate immune response in the liver, which results in the release of pro-inflammatory cytokines such as IL-1 beta and TNF-alpha into the circulation. Glucocorticoids, which are the current standard of care for treating alcohol-associated hepatitis, are potent but somewhat nonspecific anti-inflammatory agents that have a lot of side effects that limit their long-term use, and they also increase the risk of infections. Although anti-TNF therapies were shown to increase the risk of infection in patients with alcohol-associated hepatitis, interleukin-1 receptor blockers such as anikinra may carry a lower risk of infection, particularly when they're used in combination with zinc. Zinc deficiency is actually quite common in alcohol-associated liver disease, and supplementation can potentially improve the gut barrier function, since it is actually zinc metalloproteins that form the tight junctions. So other strategies include enhancing repair from injury and promoting regeneration. GCSF reduces the risk of infection during chemotherapy-induced neutropenia, but it also stimulates the release of progenitor cells that can repopulate the injured liver, potentially enhancing liver regeneration. Interleukin-22 is an anti-inflammatory cytokine that increases the synthesis of metallothionines in the liver and has anti-infective properties. Modification of the microbiome, either through fecal microbial transplantation or through the use of probiotics, has been tested in some small trials in patients with alcohol-associated hepatitis. Perhaps even more exciting news comes with the use of bacteriophages that are highly selected for certain pathogenic bacteria. These viruses actually have been shown in animal models to be efficacious in eliminating toxin-producing strains of Enterococcus, and although studies have not been conducted in patients with liver disease, bacteriophages have been used to treat antibiotic-resistant bacterial infections in patients. So the treatment of alcohol-associated liver disease really does begin with abstinence, which is unquestionably the most important factor in the long-term progression of disease and the risk of death. In the first 90 days after acute decompensation with alcohol-associated hepatitis, mortality is due to the severity of the underlying liver disease. But after 90 to 180 days, a relapse to hazardous drinking is the most significant determinant of survival. It's also important to note that improvement in alcohol-associated hepatitis develops steadily over a period of some 6 to 12 months, reaching a plateau around 12 months following diagnosis, even in those patients who are abstinent. The STOP-AH trial included over 1,000 patients with alcohol-associated hepatitis who were treated with prednisolone, pentoxyphilin, or a combination in addition to supported standard care. This important trial showed very clearly that pentoxyphilin was no better than standard care alone. Likewise, the trial confirmed that there was some benefit in survival at 28 days for patients treated with prednisolone, but that benefit was not sustained beyond those first 28 days. In fact, most of the mortality occurred within the first 28 days, but in those who were treated with prednisolone, there was an increased rate of infections, including fungal infections that began after the treatment was initiated. Whereas earlier studies had shown some benefit of pentoxyphilin in preventing hepatorenal syndrome, there wasn't any difference noted in the incidence of acute kidney injury in this study. The DASH trial enrolled 103 patients who received either methylprednisolone or a combination of anakinra, an interleukin-1 receptor antagonist, zinc to improve the gut barrier function, and pentoxyphilin, presumably to prevent acute kidney injury. However, as you can see, the 30-day survival was similar, both in those treated with the combination therapy and in those treated with glucocorticoids. But the combination group actually showed a trend toward better survival at 90 and 180 days. And even though the overall rate of infections in these two groups was similar, there were no fungal infections that were seen in the group treated with combination therapy, whereas four of the 50 patients treated with steroids developed incurable fungal infections. These promising results are now being extended in a multicenter trial that's sponsored by the NIAAA and is now enrolling at eight different sites in the United States. So other recent studies have evaluated treatments that improve either the repair from injury or that enhance regeneration. Interleukin-22 is an anti-inflammatory pro-regenerative cytokine, and it was given on days one and seven to patients with both moderate and severe alcohol-associated hepatitis. The seven-day allele score was used as the primary endpoint in the study in order to predict what the survival might be at 180 days. The allele scores were favorable in some 83 percent of the moderate and severe groups when combined, compared to only favorable scores in 56 percent of a group of prospective but not randomized controls who were treated with steroids. However, only 12 percent of the historical controls that were treated with standard care had favorable allele scores. So in addition to improvement in the allele scores, the MELD scores also improved in both the moderate and the severe patients at day 42. And although these are preliminary results, there's reason for optimism that IL-22 may offer some benefit in alcohol-associated hepatitis, not only based on its mechanism of action, but also on the lack of serious side effects and the anti-infective properties that IL-22 has. Granulocyte colony stimulating factor, or GCSF, has been used for decades to reduce the risk of infection in patients with chemotherapy-induced neutropenia. In this study, patients with severe alcohol-associated hepatitis who had an unfavorable allele score following seven days of treatment with steroids were enrolled in a trial to study GCSF. And as you can see, the patients who were treated with GCSF had improved survival compared with those who were given pentoxifilin alone. However, last year at ASLD, there was a different trial of patients with severe AH that failed to show that GCSF improved survival at 30 days, 90, or 365 days. Nevertheless, GCSF is continuing to be evaluated, both as a single agent and in combination with other agents, such as the antioxidant n-acetylcysteine, in patients with severe alcohol-associated hepatitis. So there have been some case reports of the successful use of fecal microbial transplantation in patients with severe alcohol-associated hepatitis, and there are some trials that are ongoing. This small, non-randomized study of patients with GI bleeding or infections who were deemed to be ineligible to receive steroids showed that there was improved survival one year following fecal microbial transplantation compared to the historical controls who were treated with pentoxifilin alone. It's interesting to note that the microbiota shifted from the pathogenic to non-pathogenic species following FMT, and there was documented coexistence of both donor and recipient microbial species at one year, which is consistent with what has been observed in other FMT studies. And while the study has obvious methodological limitations, it does demonstrate that you can have lasting modifications to the microbiota. Finally, when medical therapies fail, which is all too often the case in patients with severe alcohol-associated liver disease, liver transplantation offers a real option for improving long-term survival. There are many issues to resolve, such as how best to manage the underlying alcohol use disorder, but a lot of promising new therapies are being developed for alcohol use disorders. So these two figures show that over the last 15 years, the proportion of patients transplanted for both alcohol-associated liver disease and even alcohol-associated hepatitis has increased steadily. Transplantation for alcohol-associated hepatitis without a period of six months of abstinence is a subject of active discussion. However, it's important to note that the proportion of all transplants for alcohol-associated hepatitis is still less than 2% of the overall number of transplants. So the graph in patient survival following liver transplantation for alcohol-associated hepatitis is similar to that which has been seen in other transplant recipients, despite the fact that the average MEL score in this group was much higher than it was for other liver diseases, such as non-alcoholic fatty liver disease, hepatitis C, and even cholestatic liver disease. While relapse to drinking remains a serious concern, a well-prepared multidisciplinary team, including addiction specialists, is the best way to manage these patients following transplantation. Most of the major societies actually develop similar guidelines when it comes to liver transplantation for alcohol-associated hepatitis. But since this is the AASLD postgrad course, I'll call your attention to the first line in this table. Liver transplantation may be considered in carefully selected patients who understand their diseases related to excess consumption of alcohol and are willing to engage in long-term treatment of the underlying alcohol use disorder and who have not responded to medical therapy. So finally, the key takeaways here are to remember that hazardous levels of drinking cause fatty liver, steatohepatitis, and fibrosis, and the mortality is most closely related to the extent of fibrosis. Alcohol-associated liver disease and alcohol-associated hepatitis is a systemic disorder that develops in patients who have steatohepatitis and often cirrhosis with features of the systemic inflammatory syndrome, even in the absence of infection. The pathogenesis of alcohol-associated hepatitis and alcohol-associated liver disease include gut dysbiosis, decreased gut barrier function, and direct toxic effects of ethanol that can cause hepatic inflammation and injury. Inflammation is the cornerstone of treatment and must be emphasized both to patients and to physicians. Other treatments, including glucocorticoids, anti-inflammatory drugs, and agents that restore gut barrier functions such as zinc and possibly fecal microbial transplantation, probiotics, and bacteriophages in the future may be valuable. In severe AH, there's a loss of liver-specific functions and there's also impaired regeneration. Pro-regenerative and repair strategies such as interleukin-22 and GCSF may also be effective future strategies. Finally, in very carefully selected patients, liver transplantation may be life-saving for alcohol-associated hepatitis. With that, I'd like to conclude my talk and thank you for your attention and invite you to attend the question and answer period. If you do have questions about the management of patients with alcohol-associated liver disease and alcohol-associated hepatitis, thank you. Good morning. I'm Anna Mae Deal from Duke University and I'm excited to talk to you today about the microbiome, what the clinician needs to know. The gut microbiome is a mixed population of microbes in our guts. It includes bacteria, fungi, and viruses. It's shared. It's similar to our housemates. It's receptive. We can mold it by our diet, drinking behavior, and disease. Furthermore, it's very dynamic and malleable. In fact, the microbiome changes when we take a long airplane flight or after we have meals and we know that antibiotics can influence it. Not only can we modify the microbiome, but the microbiome can modify us by the products that it makes. These include secondary and tertiary bile acids, short-chain fatty acids, TMA, ethanol, and ethanol metabolites. Basically, we have a two-way conversation going on with our gut and it's the interactions between us and them that govern health. Examples of the importance of the gut-liver-brain axis are illustrated by these lists of pathophysiologic sequelae of microbial activities. For example, ammoniogenesis is thought to contribute to hepatic encephalopathy. The production of trimethylamines depletes choline, and that might be important for NAFL pathogenesis. On the other hand, the trimethylamines that are produced in the gut go into the liver where they're converted to TMAO, and we know that this is important for coronary artery disease. LPS from the gut causes endotoxemia, and this can lead to activation of the inflammasome, and metabolic syndrome, alcoholic and non-alcoholic fatty liver disease. Conversely, biosynthesis of short-chain fatty acids is protective. It maintains the intestinal barrier and protects us from leaky gut, but it does inhibit gut motility and promote nutrient absorption, and so contributes to obesity. Bile acid metabolism generates secondary and tertiary bile acids in the gut, and these interact with receptors like FXR and TGRI, leading to the production of hormones, FGF15 and FGF19, that impact lipid and glucose homeostasis, and this is important for NAFL, NASH, cirrhosis, and liver cancer. All of this has led to what I call the bad bug hypothesis, and the idea here is that there are certain pathogenic microbes that alone or in combination can cause specific diseases or specific stages of disease. So for example, if you were carrying microbes A, B, and C, you might get NASH, or if you have microbes X, Y, Z, you develop cirrhosis. Now there is some evidence to support the bad bug hypothesis. The microbial diversity generally decreases as diseases progress, suggesting that either loss or gain of specific bugs might be bad, and I'm going to show you now one very elegant example of how a specific microbe has been proven to be pathogenic for NASH. In fact, this work actually fulfills Cox's postulants about how an infection causes a disease. A patient came into a hospital in China with very high blood ethanol levels and was found to have steatohepatitis. However, he denied ethanol consumption. This led his physicians to evaluate his fecal microbiome, and they found an ethanol-producing strain of klebsiella pneumonia in his stool. So they isolated the putative pathogen from a diseased individual. They considered that it could be pathogenic because of a plausible biological mechanism. It's well known that ethanol and ethanol metabolites can cause steatohepatitis. They then went on to prove that the pathogen causes the disease by transferring either the feces or directly transferring the bacteria itself in the mice and eliciting steatohepatitis. Even more, they deleted the ethanol-producing enzyme from the bacteria, and the mice were protected. Finally, they validated the association of the pathogen in independent cohorts. In this paper, they report that ethanol-producing strains of klebsiella are enriched in Chinese with NAPL, and other workers have independently corroborated that blood ethanol is increased in non-drinking NASH subjects, including obese mice and children. So this is a nice example where we have a bacteria that is specifically involved in the pathogenesis of the disease. However, I would say that the preponderance of evidence actually goes against the bad bug hypothesis. We know that disease and stage-associated microbial signatures are highly variable, and disease-associated products seem to more consistently track with a particular kind of disease. So for example, NASH is commonly associated with endotoxin, ethanol, ethanol metabolites, TMA, and choline depletion. In contrast, severe ASH, associated with acute liver failure, is also associated with LPS, branched-chain amino acid degradation, methane production. So those observations have led to a modification of the bad bug hypothesis, where it's now thought that there are a variety of pathogenic microbes that alone or in combination can give rise to disease- or stage-specific metabolites, and that it's the gain of bad products or the loss of good products that actually causes disease. Now, the evidence for this is pretty good. We know that fecal microbial transfer, or FMT, can induce disease. When we give an FMT from a diseased mouse with NAPL or alcohol-induced liver disease to a healthy mouse, we can elicit the disease in the recipient. Similarly, if we do an FMT from people who have NASH or ASH and transfer that into a mouse, the mouse developed a disease. However, all bugs are not bad. Some bugs seem to be good, because we know that germ-free mice are actually more vulnerable to liver disease, they get worse alcohol-induced liver disease, and they develop a sclerosing cholangitis-like syndrome with age. But importantly, there are implications of this bad bug hypothesis, because the idea here is that it's the metabolites, either gain or loss of bad products of bacteria, that need to be rectified. And so the objectives of treatments now are to deplete the bad products and replenish the good products. There's a number of strategies that have been developed to reconfigure micro-real communities to be more healthy. And these include giving probiotics, prebiotics, antibiotics, or FMT. So let me walk you through these. Prebiotics are poorly absorbed sugars, fibers, or adsorbents. Examples include lactulose, cholecystironine, activated charcoals. And the idea here is that you stimulate a bloom of good bugs, and then this suppresses and counteracts the bad bugs. Or you directly prevent the absorption of the bad bug products. Probiotics, on the other hand, are actually mixes of good bugs. Examples are VSL-3 or the lactobacilli in yogurt. And the idea here is to overwhelm or replace the bad bugs. Finally, we're well aware of poorly absorbed antibiotics. And the idea here is to delete bad bugs. And with fecal microbial transfer, you're replacing a whole bad community with a presumed healthy community. So let me walk you through some trials that have actually been done in NAPL, NASH, and alcohol-related liver disease. First, the ones in NAPL and NASH. Remember that most of these studies have been done in non-hospitalized, relatively compensated patients. Rifaximin, which is, of course, an antibiotic, has been tried in NAPL and NASH. A couple of trials showed some benefit, but they're relatively small. In total, only 92 patients. Both of them gave the same dose of Rifaximin, 1,200 milligrams a day. One was short-term, lasted a month. The other was a bit longer, six months. One decreased LPS, aminotransferases. The other improved inflammatory cytokines, liver enzymes, and the liver fat fraction on non-invasive imaging. However, another trial showed no benefit. It was placebo-controlled, but it was of intermediate duration, six weeks. And it was about half the dose of Rifaximin as the effective trials. Probiotics have also been used in NAPL and NASH. There are five trials, different probiotics, fewer than 200 patients total. Four of them showed some benefit, as evidenced by decrease in liver enzymes or evidence of oxidative stress. One was actually a randomized controlled trial in children, which showed a decrease in NAPL and improvements in BMI and GLP-1. Another one was in adults that showed a decrease in liver enzymes and fibrosis score. However, one trial actually showed an increase in liver fat. Prebiotics are also being used in NAPL and NASH. One study included an oligofructose versus placebo for about nine months in a very small group of patients, but it did show some benefit, increasing fecal bifidobacteria, decreasing serum LPS, improving steatosis in NASH, and a different oligosaccharide was also found to improve insulin resistance and intestinal barrier function and enrich the stool with good bacteria in a pilot study. Absorbents are being tried in NAPL and NASH. There's one called YAK001, and there's a picture of what that is. It's activated carbon on somebody's fingertip. So far, it's been mostly used in rodent studies. It's thought to absorb cytokines and hydrophobic bile acids and other bacterial products. This has shown some improvements in NASH and animal models, and it's actually undergoing clinical trial in Europe. Finally, FMT has been tried in NAPL and NASH. There's a sort of positive pilot study from ZUADAL, and there are three ongoing clinical trials in the U.S. Turning to ASH, it's important to recognize that most of the studies here have been done in hospitalized and decompensated patients. There was a study of probiotics versus placebo that was open-label, went for seven days, and did show improvements in endotoxemia, inflammation, and liver function. Another placebo trial that was open-label showed an increase in neutrophil phagocytic activity and some improvement in liver enzymes. This is an interesting study, a couple of fecal microbial transplant studies, one done in steroid-ineligible patients with severe acute alcoholic hepatitis, and the comparison group was historical controls. The actively treated FMT patients showed normalization of fecal microbiota, improved metabolic profiles, and significantly improved one-year survival, but remember, the comparison group here is historical controls. Another study compared FMT to other kinds of treatment for SAH. It was retrospective, and three-month survival was the endpoint. The group that got FMT had a 75% three-month survival versus 30% to 40% in the patients treated with either steroids, nutrition, or contact spilling. So what are the take-home messages? Well, concerning the gut microbiome, I think we can say it's malleable. The composition is determined by us, our housemates, our diet, and microbial competitors. That we and they are interactive. Normal hosts have normal microbiome, and abnormal hosts have abnormal microbiomes, and vice versa. Microbial products are what trigger responses in us. A few of them have been characterized like short-chain fatty acids, bile acids, LPS, TMA, ethanol, but most of them are unknown, and it's suspected that it's really the combined actions of multiple metabolites that cause the ultimate effect. And finally, we now understand that multiple bugs can produce the same metabolites, but the relative contributions are likely context-dependent. Regarding take-home messages for therapy that's targeted at the microbiome, the idea here is to exploit the microbiome malleability to fix the host. Several approaches can be used to achieve that objective, increasing the good bugs with pre- and probiotics, decreasing bad bugs with antibiotics, replacing the whole thing and starting over with FMT, or let it be, but sop up the toxic products by giving absorbents. The problem here, though, is that although all of this is very attractive, there are significant barriers to success, and the biggest one is deficient knowledge about a lot of important stuff, such as what are the optimal microbial metabolites in a soup that we need for health? What are the microbial sources of these desirable metabolites? Are there disease- and state-specific metabolite signatures? What's the optimal route of administration, PO versus correcta? How durable is the response if initial recalibration is actually successful? And what is the risk of triggering transfer of bad bugs? So the bottom line, I think, is that the microbiome is a very worthy therapeutic target, but at the present time, microbiome-targeted therapies are not ready for prime time. Thank you. Hello, I am Larry Friedman from Boston, and it's my great pleasure to introduce the three speakers for session three of the 2020 AASLD postgraduate course. This session is entitled The Immune System and the Liver. The first speaker is Marlon Mayo from the University of Texas Southwestern Medical Center, and she will discuss evolving management of autoimmune liver disease beyond conventional immunosuppression. The second speaker is Michael Dugan of Massachusetts General Hospital. He will discuss liver disease related to cancer immunotherapy. The third speaker is Paul Watkins of the University of North Carolina at Chapel Hill. He will discuss drug-induced liver injury as an immune disorder of the liver, implications for management. I also want to remind you that there will be a live question and answer session from 4.30 to 5.30 on Saturday, November 14th, and we look forward to seeing you there. So I'd like to turn the program over now to Dr. Mayo. Thank you. Hello, cyber world. My name is Marlon Mayo, and I would like to thank the organizers for giving me the honor of contributing to the first all-virtual AASLD postgraduate course. And these are my disclosures. So the therapeutic targets in autoimmune liver diseases, including primary biliary cholangitis, primary sclerosing cholangitis, and autoimmune hepatitis, are classified into those that are directed at altering the immune response, fibrosis, biophysiology, or the microbiome. And I've listed under each heading many, but not all of the agents that have reported results from clinical trials. Although some of these drugs could definitely fit under more than one category. So today I'm going to give you a broad overview of each of their mechanisms of action and efficacy. So we have three diseases, 23 drugs, and 20 minutes. So grab a cup of joe and let's get started. So based on what we know about the pathophysiology of autoimmune liver diseases, immunosuppression is the most logical therapy pathway. Although there's been more success using this approach in autoimmune hepatitis than there has been in the cholestatic liver diseases. So corticosteroids such as prednisone, prednisolone, and budesonide inhibit signal transduction through the IL-2 receptor, and that's how they suppress lymphocyte activation. We know they are highly effective in autoimmune hepatitis from the first randomized controlled trial of steroids, which demonstrated a dramatic improvement in survival compared to placebo, and no placebo controlled trial has been considered ethical ever since. So more recently, budesonide was shown to be able to induce remission without steroid side effects better than prednisone in non-steroidic patients. But despite their proven efficacy, it's not 100% still, and the long-term compliance with steroid therapy is poor due to weight gain, acne, diabetes, and osteoporosis. So corticosteroids appear to have a mild calming effect in PBC. Studies adding budesonide to ursodiol in PBC have mixed results. The Leuschner trial showed a reduction in alkaline phosphatase that was not replicated in the Angulo study, and the latter study also showed worsening of bone density. So this risk has generally been considered to outweigh the potential mild benefit. In PBC, corticosteroids have an even louder effect with a very small and variable improvement in alkaline phosphatase, and there remains concern of risk of sepsis from cholangitis. So azathioprine and mycophenolate are cell cycle inhibitors that inhibit purine synthetase. So they broadly reduce lymphocyte proliferation. These drugs do not induce remission like steroids and autoimmune hepatitis, but they are capable of maintaining remission. So azathioprine can maintain remission in up to about 80% of patients for up to a decade. In those who fail azathioprine, either due to intolerance or poor efficacy, mycophenolate is a valuable alternative. So in this recent meta-analysis, the pooled response of mycophenolate in azathioprine failure was 58%. The side effects of azathioprine are increased rates of skin cancer and lymphoma, and mycophenolate is a terrible teratogen, so we are still in need of better options. Azathioprine was eliminated as a contender for PBC monotherapy by a landmark trial in 1976 by Jenny Heathcote that found no effect of azathioprine on alkaline phosphatase or survival. And this illustration from that paper shows you that they followed alkaline phosphatase for four years and found no difference with placebo. So we have no randomized trials in PSC. However, a case series of 13 patients with 25-year follow-up was recently published, and they saw no potential sign of efficacy. Clinical outcomes were similar to historical controls. It was, however, well-tolerated with no increase of cholangiocarcinoma, which is a theoretical concern. So methotrexate is a folate antagonist and adenosine agonist, so it inhibits cell proliferation, but it also has intracellular inflammatory pathway inhibition. It is also a powerful abortant, which has prevented anyone from performing a randomized controlled trial in autoimmune hepatitis, because this is a disease that affects young females. There is some suggestion of efficacy, however. A case series a couple of years ago showed a complete response rate of 54.5%. The effect of methotrexate in PBC was controversial for many years until Combs published the results of a randomized controlled trial of methotrexate or placebo added to ursodiol that was not able to detect any difference in cirrhotic decompensation or transplant-free survival after eight years of treatment. It's worth noting, however, that that study was stopped for futility because the number of endpoints that accrued in both groups was small. I think above all else, this study taught us the near impossibility of conducting a prospective trial with mortality as an endpoint in a group of early stage PBC patients on ursodiol. Now, interestingly, the same investigator group recently reanalyzed the data presented at last year's meeting and showed that the subgroup of patients in that trial who had an incomplete biochemical response to ursodiol, which was about half of the study population, they achieved a 21% reduction out on phosphoenol. And a 2.1% reduction out on phosphatase with the addition of methotrexate. Now, in PSC, there is one randomized trial of methotrexate, which also found significant reductions in out on phosphatase. But again, the effect on clinical outcomes was not seen. And they also looked at ERCP findings. This was pre-MRCP era, as well as histology. So finishing off our general immunosuppressant section, tacrolimus and cyclosporine inhibit calcineurin, which is a regulator of T cell activation and proliferation genes. And both of these drugs are effective for autoimmune hepatitis, as illustrated by multiple small studies that are summarized in these meta-analysis figures. Tacrolimus, in fact, was superior to mycophenolate as a second line therapy in two head-to-head studies. However, the recently published AASLD guidelines for autoimmune hepatitis favors mycophenolate over tacrolimus, really based upon ease of use and lower toxicity. And more details are explained in the paper cited here. Cyclosporine was tested for PBC in the 1980s in two large randomized control trials and found to be reasonably effective. Alkaline phosphatase reduction of several hundred units occurred, and there was also lower rates of histological progression and lower liver-related mortality. Unfortunately, the rate of nephrotoxicity was high, and then ursodiol came along about the same time, which had no toxicity, and so cyclosporine was abandoned as a treatment. In a similar vein, tacrolimus has been tested in PSC and shows some efficacy, but it's poorly tolerated. So in this one-year study, alkaline phosphatase dropped significantly, but 31% of patients withdrew due to GI and neurological toxicities. So our antifibrotic section is gonna be brief because there are only two drugs to cover, and neither one works. The colchicine binds tubulin and inhibits microtubule formation. It is also anti-inflammatory. It is ineffective in PBC. We know this from 10 randomized trials which have been combined and showed a net no difference in mortality, liver transplantation, biochemistry, or liver histology. Symtuzumab is a monoclonal antibody against lysyl oxidase-like 2, which is an enzyme that catalyzes the cross-linkage of collagen and elastin. And it was tested in PSC, which is arguably the most fibrotic of the autoimmune liver diseases with its characteristic fibrosing lesions of the bile ducts. But neither low nor high-dose symtuzumab laid to any difference in collagen content of the liver as compared to placebo. For the cholestatic liver diseases, PBC and PSC, we've been more successful with drugs that modify biophysiology than we have with immunosuppressants. So these drugs have not really been explored for autoimmune hepatitis, so all the data I will show you is from PBC and PSC. In addition to ursodiol and norsodil, I've included FXR agonists and fibrates in this section because they do alter biophysiology, although they're really pleiotropic, as we will discuss. So ursodiol increases the hydrophilicity of bile and increases Ni-exchanger 2-protein expression, which leads to an increase in fluidity of the bile and protects the bile epithelial cells and the hepatocytes from bile acid toxicity. Norsodil is a derivative of urso with a shortened side chain, which resists conjugation. And thus, it doesn't follow the typical enterohepatic circulation pathway. Instead, it's reabsorbed by bile epithelial cells into the sinusoids and hepatocytes, which concentrates in the liver and virtually eliminates any chance of conversion to other bile acids by colonic bacteria. Now, urso is the standard of care now in PBC. It was FDA-approved in the United States, and I won't show you all the data, but multiple large randomized controlled trials have shown that it improves biochemical tests, delays histological progression, it prolongs expected survival without liver transplantation, so survival compared to prediction models, Mayo R-score, GLOBE score, MELD score, compared to historical controls. And what we've observed since the introduction of ursodil is that there is a reduction in the need for liver transplants. It doesn't, however, completely eliminate the need for liver transplants, and so additional therapies are needed. Now, in PSC, the use of urso has been more controversial. In the high-dose urso trial of 28 to 30 milligrams per kilogram of urso for PSC, they witnessed reductions in alkaline phosphatase of 160 units, which was double what was seen in the placebo group. However, much to everyone's surprise, there was an increase in combined outcome of death, transplant, and particularly development of new varices, and this led to early discontinuation of the study. Post-hoc analysis found high levels of lithocholic acid in these patients, implying that perhaps the excess urso was being converted by gut bacteria into a more toxic bile acid. And thus, nor-urso, due to its cholepatic shunting, may be a safer option in PSC. An initial study has shown biochemical efficacy. It reduced alkaline phosphatase by 121 units compared to a slight rise in alkaline phosphatase in the placebo group. And a phase three study is ongoing in Europe, which also includes histology. So farnesoid X-receptor, or FXR agonists, are nuclear hormone receptor agonists that have a plethora of potentially beneficial effects in cholestasis. They inhibit de novo bile acid synthesis, increase excretion, and decrease reuptake of bile acids. They're anti-inflammatory and anti-fibrotic. And obeacolic acid was the first-in-class FXR agonist, which obtained conditional approval from the FDA in 2016 for use in PBC with incomplete response to ursodiol, based on these data from their phase three trial that showed a mean reduction of alkaline phosphatase of about 113 units. Now, subsequently, other FXR agonists have been evaluated for their efficacy and safety in phase two trials. Silofexor, for example, led to a 40-point drop in alkaline phosphatase as compared to a three-point rise in the placebo-treated patients. And FXR agonists also show some efficacy in PSC with both obeacolic acid and silofexor reducing alkaline phosphatase in phase two trials. Obeacolic acid led to about an 83-point reduction and silofexor to a 73-point reduction. In PSC, however, it should be noted that only about half of these patients were also treated with ursodiol versus the PBC studies where the vast majority of patients were concomitantly taking ursodiol. Now, throughout this presentation, I've referred to change in alkaline phosphatase as a reduction in absolute units of alkaline phosphatase rather than percent drop. That's just so that you can compare apples to apples. So the improvements in biochemistries with FXR agonists are rapid and they're durable. What we still don't know is the impact of FXR agonists on clinical outcomes, such as development of varices, ascites, jaundice, and encephalopathy, need for transplantation. There's an ongoing study with obeacolic acid that's designed to look at these outcomes, which was a mandate by the FDA at the time of their conditional approval. It is encouraging, however, that we're now seeing reductions in liver stiffness after several years of therapy. And, of course, if you improve alkaline phosphatase, AST and bilirubin, you will improve expected survival as calculated by mathematical models such as GLOBE score and APRI. Now, what we've learned about FXR agonists in cholestatic patients is that they can cause itching as a side effect, and at least in the case of obeacolic acid, there can be rare episodes of liver toxicity, especially in child's B and C cirrhotic patients. And also by increasing cholesterol saturation and FGF-19 levels in bile, we may be predisposing these patients to gallstones and or cholecystitis. So PPAR agonists, commonly referred to as fibrates, increase secretion of biliary phospholipids, which protects the bile epithelial cells from bile acid damage. They're also anti-inflammatory through inhibition of NF-kappa B, and they regulate lipid metabolism. In PBC, there have been several PPAR agonists that have been tested as add-on therapy to ursodiol when the biochemical response is not complete, and they show as good or better reduction in alkaline phosphatase than we have seen with the FXR agonists. So here you can see in the phase three study of visafibrate, there was a 146-point drop in alkaline phosphatase as compared to no reduction with placebo. In the phase three study of celadelpar, whose top-line results were released in August, and the detailed results are gonna be presented at this meeting as a late-breaker on Monday, there was a 122-point reduction in alkaline phosphatase compared to just two points in the placebo group. Now, visafibrate is only available outside the United States, and celadelpar is still under clinical development, but phenofibrate is available as a generic in the United States. So each of these drugs targets a different PPAR, so they're not exactly interchangeable, and they're expected effects, although they all seem to demonstrate some biochemical efficacy in PBC. And here you can see the pooled effect of six open-label studies showing 114-point reduction in alphos with phenofibrate. What we see as a broad class of drugs is that the improvements in biochemistries are rapid and durable. We still don't know what the impact on clinical outcomes are going to be. This is under investigation, but it is encouraging, at least in the Phase III study of bisofibrate. They saw reductions in liver stiffness after two years of treatment as compared to a rise in liver stiffness in the placebo group. In addition, at the recent Digital International Liver Conference, Dr. Tanaka presented from Japan a retrospective cohort study, so not randomized, in which about 6,000 people received ursodiol only, and almost 1,000 received the combination therapy of ursodiol and bisofibrate. They found that the addition of bisofibrate to ursodiol further reduced the risk of death of transplant with an adjusted hazard ratio of 0.23. In PSC, we have no randomized control data. There are some small open-label studies from Japan which hint that there may be at least a biochemical effect. So the last therapeutic approach I want to discuss is alteration of the microbiome. This approach has really only been considered and tested in PSC, so that's all the data that we have. So, antibiotics and fecal transplant correct the dysbiosis that has been documented in patients with PSC. This theoretically will reduce portal entry of bacteria and triggers of autoimmunity. There have been several antibiotics that have been tested in small clinical trials. The largest drop in alkaline phosphatase is seen with vancomycin. You can see a 521-unit drop versus a 61-unit drop in placebo. There's also some reduction in alkaline phosphatase with metronidazole, but there was also a small head-to-head comparison between vancomycin and metronidazole that showed the alkaline phosphatase reduction was indeed greater with vancomycin. Again, clinical outcomes data are lacking. The fecal microbiome of PSC patients is characterized by a lack of diversity, and if we restore that diversity with fecal transplant, that could have a beneficial effect on disease course maybe. It's an interesting concept. It was evaluated in a small study of 10 patients with PSC who got fecal transplant from a single donor. Nine out of 10 successfully increased their fecal microbiome diversity. Alkaline phosphatase reductions were seen in some of the patients, but they're not as large as what we've seen in the vancomycin studies. So I've given you a rapid flyover survey of a lot of different medications, both historical and those that are under development for autoimmune liver diseases, and I've tried to classify them by mechanism of action. My key take-home points are actually philosophical, but they are derived from the aggregate data that we reviewed today. So first of all, novel therapies for autoimmune liver diseases have shifted away from broad immune suppression towards other modalities, such as targeting bile acids in the microbiome. And with multiple different targets, the future is bright, and it's likely to include some combination therapies. Secondly, we've shifted our focus towards looking at alkaline phosphatase improvement in PBC and PSC, and there are many exciting drugs that improve alkaline phosphatase. And the other surrogate endpoints are just beginning to emerge, things like liver stiffness and globe score and even patient symptoms. But we need to keep in mind that biochemical improvement was the same endpoint we historically observed and discounted because we had a hard time linking it to better clinical outcomes, and this was seen with ursodiol, methotrexate, as well as cyclosporine. And then lastly, I know I speak for many when I say thank you to the patients, the investigators, the FDA, the NIH, the pharmaceutical industry, the investors who continue to run in this marathon of discovery for better patient outcomes for patients with autoimmune liver diseases. The Boston Marathon may be canceled this year, but this marathon is not, and we are clearly out of the starting block and partway around the track, but we need to keep going in order to win this race. And I also want to thank you for your attention. Hi, my name is Michael Dugan, and I'm going to talk to you today about liver disease related to cancer immunotherapy. I'm currently an assistant professor of medicine at Massachusetts General Hospital and Harvard Medical School. Here are my disclosures. The learning objectives for today are to understand the prevalence of liver injury from immunotherapy and the relationship to specific immunotherapies used, to understand the differential diagnosis and initial workup of a patient with suspected hepatitis from immunotherapy, which we call checkpoint hepatitis, to understand the role of liver biopsy in diagnosing hepatitis from immunotherapy, and to understand the management of confirmed to suspected checkpoint hepatitis and the role of secondary immunosuppression in treatment. We're going to start with a case, a 62-year-old woman with resected stage 3b melanoma who was started on adjuvant immunotherapy with the drug called ipilimumab in 2017. She completed cycle 2 on July 19th, and on August 4th, developed fevers, chills, and body aches. Cycle 3 was held at that point, and monitoring labs showed an increase in transaminases and alkaline phosphatase concerning her inflammatory hepatitis from her immunotherapy. Before we talk about her case further, we should talk a little bit about how T-cells become activated. Naive T-cells are activated by two signals. The first signal comes from the T-cell receptor recognizing a foreign peptide in the context of MHC. This is a specificity signal. The second signal signifies danger. This comes from CD80 or CD86 on the antigen-presenting cell, binding to the T-cell co-stimulatory receptor CD28 on the T-cell. When a naive T-cell receives both of these signals, it becomes fully activated. But the signaling is modified by many other co-stimulatory and co-inhibitory receptors. The co-inhibitory receptors that modify T-cell activation are collectively referred to as immune checkpoint receptors. Two of these pathways, the CTLA-4 pathway and the PD-1, PD-L1 pathway, are manipulated for cancer immunotherapy. CTLA-4 and PD-1 are both expressed on the T-cell, and PD-L1 is expressed on the tumor cell or on an antigen-presenting cell. CTLA-4 can be blocked with the drug ipilimumab. There is only one CTLA-4 inhibitor. There are now three approved PD-1 inhibitors and three approved PD-L1 inhibitors for a total of seven different immunotherapy monoclonal antibodies currently in clinical use. Immunotherapy has been highly successful. I'm showing you survival curves for metastatic melanoma. On the left, patients treated with ipilimumab, and on the right, patients treated with nivolumab, one of the anti-PD-1 inhibitors. Currently, metastatic melanoma has a three-year survival that is around 40% compared to less than 5% before the advent of immunotherapy. And immunotherapy is now approved to treat more than 14 different kinds of cancer. Alongside the tremendous success of immunotherapy has come a host of inflammatory toxicities which can affect any organ system in the body, though most prominently barrier organs such as the skin, GI tract, and liver, and the lungs. I'm showing you here a list of common toxicities and highlighting the toxicities in the GI tract, and then further highlighting hepatitis in particular. Checkpoint hepatitis is also referred to as immune checkpoint inhibitor or ICI hepatitis or immune-related IR hepatitis. It occurs in somewhere between 2% and 5% of patients on single-agent treatment, but is more common in patients on dual immunotherapy, occurring in almost a quarter of patients who receive combination immunotherapy. It's also more common in patients who receive combination therapies with targeted therapy or chemotherapy for their cancer, which has opened up the possibility that immunotherapy may, in general, increase the risk for a drug-induced liver injury. It can occur any time during the treatment course for immunotherapy, but typically occurs within the first couple of months. It's typically diagnosed incidentally based on surveillance labs, and rarely it can cause fulminant hepatic failure. Patients on immunotherapy for cancer are at risk for multiple etiologies of liver injury, however. Hepatic disease can cause injury to the liver either by compression of the bile ducts or external compression, as well as through parenchymal invasion. These patients are at increased risk for vascular injury, either through treatments such as selective internal radiation therapy or CERT therapy, as well as elevated risk for clotting in general. They are at elevated risk for drug-induced liver injury, for example, because of their chemotherapies or targeted therapies that they may be on at the same time, or due to other concomitant medications. They are at risk for viral infections. Many of these patients are still immunosuppressed, either because of chemotherapy or because of their tumor itself, which puts them at increased risk for EBV and CMV, as well as reactivation of latent hepatitis B. And patients with latent viral infections, like hepatitis B and C in the liver, can also have more robust inflammatory responses to them in the context of immunotherapy. These patients obviously do have a risk of autoimmune hepatitis, although this is distinct from classic autoimmune hepatitis, as we'll discuss a little bit later. And they may be at increased risk for metabolic injury to the liver, though this is an area of active research. The role of the gastroenterologist or hepatologist is often to broaden the differential in these patients and make sure that we rule out etiologies besides checkpoint hepatitis. Checkpoint hepatitis is graded in the oncologic literature using the Common Terminology Criteria for Adverse Events, or CTCAE criteria, and this is commonly what's used in practice by oncologists as well. This grading system rates checkpoint hepatitis based on the transaminase elevation and prescribes certain general strategies for treatment based on the degree of transaminase elevation. There are limitations to the CTCAE criteria. It doesn't take into account duration of injury. It does not capture markers of synthetic dysfunction or clinical symptoms, and it does not correlate with responsiveness to treatment. So returning to our case, this patient had grade one checkpoint hepatitis. However, because of the systemic symptoms as well as the relatively modest degree of transaminase elevation, we weren't completely confident in the diagnosis and we decided to send the patient for liver biopsy, which happened four days later, and as you can see, just as the transaminases and alkaline phosphatase are beginning to rise. So what is the workup for a patient with suspected checkpoint hepatitis? First of all, we do send viral serologies and make sure this isn't a viral infection. Autoimmune serologies are probably not helpful in these patients. They have no real correlation with checkpoint hepatitis, unlike classic autoimmune hepatitis. Imaging is very important, so ultrasound to exclude vascular occlusion as well as bile duct obstruction. For patients who have a cholestatic pattern of liver injury and a normal ultrasound, consider getting an MRCP or an EUS in order to absolutely exclude the possibility of bile duct obstruction from malignancy or from an enlarging lymph node. And then cross-sectional imaging is often helpful in these patients given their elevated risk for metastatic disease in the liver, which sometimes initially presents as changes in the transaminases. Liver biopsy has a less clear role in the management of these patients. Current oncology guidelines do disfavor biopsy for diagnosis of checkpoint hepatitis. It is unknown if the pathology actually correlates with treatment outcomes, and currently there really is no study of how biopsy affects management decisions. We consider liver biopsy in patients who have grade 2 or higher hepatitis, patients who have systemic symptoms, or who have evidence of synthetic dysfunction, all situations where we want to make sure that we do have the correct diagnosis. But biopsy is certainly not mandatory in any of these cases. So we did send this patient to liver biopsy, and what we found was moderate pan-lobular hepatitis with lobular and periportal mixed inflammatory infiltrates with lymphocytes, eosinophils, histiocytes, and small epithelioid granulomas. This is a subtype of the hepatitic pattern of immunotherapy-induced liver injury. So what are the typical findings for immunotherapy-induced hepatitis? The largest case series is from MGH, 60 patients recently published. The median age in this patient population was 62 years with significant range. Males and females were relatively well-balanced, and most patients had melanoma, which is consistent with the use of immunotherapy to treat most patients with melanoma. Combination therapy was the most common therapy, and that is also consistent with the known increased risk of hepatitis in patients who are receiving combination therapies. And the median time to detection of the hepatitis was two and a half months with, again, a substantial range. Most of the patients had relatively severe hepatitis, grade 3 or grade 4, and the most common pattern was hepatitic, although that was really relatively well-balanced with cholestatic pattern in the initial labs. When we looked at the biopsies themselves, most patients, almost half, had a hepatitic pattern of liver injury. One of the more prominent subtypes of that is the pan-lobular hepatitis, which I'm showing you in the picture. A separate study looked at a little more detail at what was in that inflammatory infiltrate. So it's typically a lymphocytic and histiocytic infiltrate. Plasmacytic infiltrates are actually uncommon, which is different from autoimmune hepatitis, as you know, and microgranulomas are commonly seen. When we do immunohistochemistry, we see that most of the lymphocytes are cytotoxic CD8 T-cells with rare CD4 T-cells. There are also a substantial fraction of patients, about a quarter in the above case series, that had a cholangitic pattern of inflammation. So once we have a confirmed diagnosis, what are our treatment options? And I'll say at this point, nothing is evidence-based. We typically treat as a variant of autoimmune hepatitis or of transplant rejection. And many of the guidelines are actually based on guidelines for treatment of other checkpoint blockade toxicities. The goal of this treatment is to minimize the morbidity and mortality of the checkpoint toxicity without inhibiting antitumor immunity, which is the more difficult part. If you're using immune suppression, it often does interfere with antitumor immunity. This is going to require finding ways of avoiding systemic corticosteroids. The most important thing we can do is make sure that we're always treating checkpoint hepatitis and not treating another etiology of liver injury with steroids. And then eventually, we need to develop therapies that we can use alongside immunotherapy to treat toxicities while immunotherapy is being continued. So why do we need to avoid steroids? Well, first I'll say that we know that steroids don't eliminate the benefit of checkpoint blockade. This is data from Sloan Kettering, published in 2015, looking at patients who developed a toxicity compared to those who didn't, patients who received steroids compared to those who didn't. And you can see the curves are statistically overlapped. The problem with these data is that patients only received steroids if they had an adverse event. Anyone with a serious adverse event got steroids. So if there's a mechanistic link between antitumor responses and toxicities, we may have masked it with the steroids. Another way of showing this is, what if the yellow line could have been up there? We dragged it down by giving steroids to these patients. There is data to support this hypothesis. So this is data from a colleague of mine, Alexander Fagy, who looked at patients with metastatic melanoma treated with ipilimumab, anti-CTLA-4, who developed pituitary inflammation, apophysitis. And these patients were treated with one of two treatment paradigms chosen by the endocrinologist. One was to suppress pituitary inflammation with some dose of prednisone in excess of 20 milligrams of prednisone a day. The other was just to replace the adrenocortical axis with prednisone doses less than 7 milligrams of prednisone a day. And what Alex found is there was no difference in pituitary outcomes between these two treatment approaches, but there was a huge difference in overall survival and time to treat failure. So patients who got high-dose steroids did about as well as you would expect someone to do if they were treated with ipilimumab. So again, steroids didn't remove the baseline treatment benefit. The difference is that the low-dose steroid patients did exceptionally well, suggesting that indeed there is a correlation between a better antitumor response and the development of this toxicity, and that giving steroids reduced that benefit. Of course, this is still all retrospective data, and none of this is definitively demonstrated at this point. There is some more retrospective data from Sloan Kettering that was published a little more recently, looking at patients with lung cancer who were treated with either immunosuppressive doses of prednisone, the red line, or no prednisone, or replacement dose prednisone. You can see again, immunosuppressive doses of prednisone were associated with reduced responses to the immunotherapy. That said, if you have the appropriate diagnosis, first-line treatment for checkpoint hepatitis is systemic corticosteroids. We just want to use these appropriately, not treat patients who have other diagnoses with prednisone. So optimal dosing is really unknown at this point, but we typically treat between 0.5 and 2 milligrams per kilogram of prednisone daily. I usually start with 60 milligrams of prednisone daily, regardless of the severity, and adjust from there. Tapers are usually 6 to 8 weeks. This form of hepatitis often resolves more slowly than other forms of immunotherapy toxicities and slower than autoimmune hepatitis, but responses should begin to be seen within about three days. If patients do not respond within three days, or they recur during the steroid taper, then consider secondary immunosuppression. These are mycophenolate mofetil, azathioprine, and tacrolimus. None of these have been directly compared to each other in terms of their efficacy, and there are no formal studies of the efficacy of any of them individually. For acute liver failure, we have other options, such as ATG, abatacept, or JAK inhibitors. These should be considered last-line therapies because they are likely, through their mechanism of action, to inhibit antitumor responses as well. There is no evidence that infliximab or other anti-TNF agents are effective for treating checkpoint hepatitis. You may be asked about this by oncology teams because infliximab is effective for some of the other checkpoint blockade toxins. Antibiotics, fecal microbiota transplant, and ursidioxycholic acid have not been adequately studied to make any recommendation about their use in checkpoint hepatitis at this point. So going back to our case, our patient was started on benazone initially at 60 milligrams a day as the alkaline phosphatase and transaminases rose. The patient initially leveled off, but then during the steroid taper had a substantial drop in both alkaline phosphatase and the transaminases and eventually returned to normal. So the key takeaways from today's talk. Treatments on immunotherapy for cancer have multiple risk factors for liver injury and checkpoint hepatitis should be considered in a broad differential. Hepatitis viral serologies and imaging, ultrasound and CT scan are important parts of the initial workup for hepatitis on immunotherapy. MRCP could be considered in patients who have evidence of an obstructive etiology based on their liver chemistries. Liver biopsy can be helpful in clarifying uncertain diagnoses but is not routine for all patients with suspected checkpoint hepatitis. Systemic steroids, somewhere between 0.5 and 2 milligrams per kilogram of prednisone or the equivalent, tapered over six to eight weeks, are first-line treatment for checkpoint hepatitis. Second-line immune suppression includes mycophenolate mofetil, azathioprine, and acrolimus. Now with that, I'd like to thank you for your attention and I will be happy to address any questions during the question period later on. I'm Paul Watkins. I'm at the University of North Carolina in Chapel Hill and these are my disclosures. You can see I do consult widely for the pharmaceutical industry and I'd like to start my talk off by giving a case, which I got from LiverTalk's website, of a 77-year-old man with metastatic renal cell carcinoma who developed jaundice six weeks after starting pazopanib. He had no previous history of liver disease and had normal routine liver tests before therapy and after two weeks of treatment. He had no risk factors for viral hepatitis or history of alcohol abuse and was receiving no other potentially hepatotoxic medications. So a clean case. And what's shown here are his liver chemistries as a function of time with pazopanib starting at time zero along the x-axis and you can see that his liver chemistries did not change before at the time of starting pazopanib or two weeks on treatment. But then at six weeks, the serum alkaline phosphatase, the blue square, and the ALT, the red circle, were elevated substantially. This was a mixed injury. The drug was stopped and the ALT fell back towards to normal levels quickly. The alkaline phosphatase also fell, but then stopped and plateaued. And during this time, the serum bilirubin, which is the green triangles, went up and peaked at almost 25 milligrams per deciliter, but remained highly elevated at 13 milligrams per deciliter. So this is a clear-cut case. Pazopanib was stopped. In this case, actually, the patient didn't survive. But in most drug-induced liver injuries, you know, stopping the drug does cause gradual improvement. We also know from the Dillon Network and the European and Spanish networks that about 20 percent or one in five people who develop acute idiosyncratic dilly still have evidence of ongoing liver injury at six months and about half that at one year. So this prolonged recovery period is classic. And during that time, and certainly in this case, where the person still brightly jaundiced, you would be hesitant to treat him for any medical condition. And unfortunately, in this case, is metastatic renal cell cancer. And this is one of the real problems in drug-induced liver injury. Although people tend to recover, during recovery, they can have limitations on treatment of their active medical problems. So it'd obviously be desirable to have a way to speed recovery. So can we effectively treat these idiosyncratic dilly events? And we now know, over the last 10 years, that idiosyncratic dilly is frequently and maybe usually an immune disease. And the best evidence for that is the strong associations with HLA molecules, HLA alleles, with risk. And you can see I've listed drugs here. I think it's a current list of drugs and associated risk HLA alleles. I've just limited this to Caucasians. There are other alleles in other populations. But for simplicity and also to fit everything on the slide, I've just stuck with Caucasians. And you can see pisaponib, our index, implicated drug here, has a known HLA association with an odds ratio of about 2. I'd like to point out that pisaponib, as well as the other four drugs that are shown in red here, the actual risk alleles were identified not post-marketing, as is the case in the drugs that are shown in black, but during the clinical trials where gene profiling was compared in those treated patients who developed ALT elevations versus those patients in the same trial that had no evidence of ALT elevations. And in most of these cases, these ALT elevations were asymptomatic, and we know they would usually go back to normal even with continued treatment, and that was observed in these cases. What this says is that the adaptive immune attack occurs much more frequently than the serious outcome of DILI, and this appears to be due to immune tolerance. That is, the adaptive immune attack is dampened down and does not progress on to clinically important liver injury. What's going on? Well, you've got the drug exposure and you've got an adaptive immune attack on the liver, and we know this can sometimes go on to the clinically important idiosyncratic DILI, but all evidence suggests when this occurs, the usual thing to happen is that the immune response gets dampened down and there's immune tolerance. Now, we just heard a talk about checkpoint inhibitors, which I couldn't hear because this is all pre-recorded, but I'm sure they discussed the promising role of steroids in treating immune checkpoint inhibitors, and that appears to be, you know, until the drug is gone and immune tolerance is restored, that the immune attack on the liver can be effectively treated by steroids. And if that was the entire story with idiosyncratic drug-induced liver injury, steroids should work as well. Unfortunately, the evidence that steroids are effective in idiosyncratic DILI is really lacking, and some trials have actually reported adverse events. These are small, anecdotal, sometimes retrospective studies, but it's not at all clear that steroids work in DILI. And that's probably because there's more to the story than just immune attack and lack of immune tolerance. It's believed that these events have to start at the level of the hepatocyte with some degree of stress, injury, and perhaps sometimes death, but not always of hepatocytes. And there has to be a neoantigen form, that is, there has to be something that's expressed on the surface of hepatocytes that is the focus of the adaptive immune attack, that is not part of the repertoire or the inventory that's been taken of the immune system in your body. However, that is not enough, it's thought, to initiate an adaptive attack. You need the release from liver cells of what are called danger signals, predominantly damage-associated molecular patterns or DAMPs, and these need to initiate an innate immune response or inflammation. Just as vaccines are not very effective unless they contain adjuvants to cause, actually, tissue damage and inflammation, it appears that the liver must have evidence of ongoing injury disease. The adaptive immune system must recognize this pathology in order to kick in the attack on the neoantigen. And this, as I say, may be the reason why steroids aren't remarkably effective, if they're effective at all in this condition. And, of course, the things we know how to do is to stop the drug, get rid of the drug, and in the case of leflutamide and teriflutamide, that are fat-soluble drugs that undergo enteropathic cycling, it's been shown you can accelerate the removal by feeding people cholestyramine, and this leads to more rapid resolution of the liver injury. There also are approaches that would reduce hepatocyte stress that may be useful. N-acetylcysteine, of course, was shown in the acute liver failure network to be useful in treating acute liver failure, particularly early in the course, and the idiosyncratic DILI group was the group that appeared to benefit the most. And several societies have come out recommending early NAC treatment and early acute liver failure. There are a variety of antioxidants that have been proposed to treat DILI in addition to NAC. SAMe would be an example. There are some herbal products in China. There's one even approved by the China FDA to treat basically ALT elevations that are observed during treatment. And then the unique case of L-carnitine and Velproate, hepatotoxicity. But the role of steroids is really been unclear, and I'll just say that although there have been several guidelines published, the most recent by this international organization, CIAMS, reported in conjunction with discontinuation of suspected drug, it is unclear what kind of therapy should be undertaken to modify the natural course of DILI. Well, can we use these HLA associations potentially to prevent DILI? And it's not as complicated as you'd think. It turns out, I'm listing here the total number of alleles, again in Caucasians, that have been identified as risk factors for DILI, and they're in five genes, HLA-A, B, DRB1, DQA, DQB. And there are only a handful, two or three, several risk alleles identified. And now with next generation sequencing, most academic centers can do this genotyping fairly efficiently, and my understanding is at a reasonable cost. Of course, they do the whole battery for bone marrow transplantation and solid organ transplantation, but they're able to do focus. If you have a question of does this patient have a particular allele, they can do that and do it relatively quickly. So could this be used to actually screen people before they get the drug? And with lumiracoxib is an example of a drug where this was tried, and it was a COX-2 inhibitor. It was approved on worldwide markets, but then withdrawn because of severe liver injuries, including liver transplants. But the company noted when they went back and did genotyping in their clinical trials, again, ALT elevations versus no ALT elevations on drug, they found a strong association with the class 2 haplotype actually, but allele, that predicted ALT elevations. In fact, if you didn't carry this particular haplotype, your incidence of ALT elevations was extremely low and similar to ibuprofen and naproxen, which were the comparators in the trial. So they proposed to reintroduce to the market lumiracoxib with this genotyping. And this is my favorite headline, Novartis aims to resurrect zombie drug lumiracoxib in EU after four years. This was unsuccessful. And I believe both the EMA and I'm quite confident the FDA believed this would work, that if you screened out these individuals, you would have a relatively safe drug for the liver comparable to other NSAIDs. However, it was pointed out if this is in your medicine cabinet for aches and pains, people other than those who are genotyped are likely to take it. There wasn't felt there was a dramatic unmet medical need and the associations had not been established in other ethnic racial groups. And this is a problem because the associations found in Caucasians don't necessarily apply to other populations. But the concept was accepted. And I think what we will see in the future are drugs introduced onto the market that have liver safety liabilities. But with screening, with HLA genotyping, they'll be viewed as safe. They'll probably be new drugs, they'll probably be patented drugs to really justify the expense and the research that will need to go into it. Can we use HLA associations in causality assessment? This I think will be a more immediate application or in the near future than pre-screen. So I go back to my original patient, which was a clear cut case of pisapinib toxicity. But what if this patient had had a one-week course of amoxicillin clavulonate, also called amoxiclav or Augmentin, just prior to the discovery? And we know, if you can look in liver tox to see, this is classically a cholestatic reaction in a man of this age group and could produce exactly what you're seeing. If it was possible to be very confident that it was the Augmentin that caused this injury, I think that might have implications to treatment. First of all, if the patient did well on pisapinib, he could get it again. But if the cancer was progressing, I think you might be quicker to retreat with pisapinib and maybe other drugs during the recovery phase. So let's talk about that. So I already told pisapinib, twofold increase, that might be worth genotyping. But there are two HLA risk alleles, one class 1, class 2. Let's look at the effect of just trying to add to causality assessment the class 2 allele. You ask your lab to genotype for it, and you find out, yes, this patient had. So how good is, how helpful is genotyping this class 2 allele? Well, it's present in about two-thirds of the cases of Augmentin dilly in the registries. But it's also present in up to 20% of patients who don't have Augmentin dilly. So on the surface, that may not seem very useful. But in fact, it's in line with other diagnostic tests we routinely order. So for instance, in idiopathic autoimmune hepatitis, an ANA greater than 1 to 60 is present in up to three-quarters. But it's also in about one in four controls. An anti-smooth muscle antibody is present in about two-thirds of cases, but also up to 43% of controls. And I've taken this information from a very nice manuscript that's in Journal of Hepatology by Guru et al.'s group. So I think it's in line with what we routinely use. But how about if you combine it with other risk factors? So what I've shown here is the effect of having both the class 2 allele, which I just showed you, and the class 1 risk allele. And that's that top row. And you can see that doubles the odds ratio to about seven and a half. And if you have both the class 1 and class 2 risk alleles and this variant in a gene called PTPN22, you double it again and getting up around 14%. And in the Dillon Network, we have a large cohort, a discovery cohort and a validation cohort now for Augmentin Dilly cases. And we're finding additional variants that significantly contribute to risk. And realizing that the genetic risk is multifactorial and moving towards what's called a polygenic risk score, hopefully will be a way to really accurately assess risk of Augmentin Dilly. But we're not there yet. I'd like to draw your attention to the last line here, which is a report in a very recent publication from Jay Hufnagel in the Dillon Group that has found a strong association between Dilly due to green tea, or more specifically green tea extract, that ends up being a common constituent in herbal and dietary supplements, particularly designed for weight loss. That association of that Dilly with the HLA-B3501. And a very interesting observation that was made in this manuscript, which is again just recently in EPUB form, and this is a figure from it, is looking at the carriage rate of this risk allele as a function of causality assessment. As you can see from the green tea cases here, causality assessment was made according to the Dillon scale without knowledge of the HLA-B3501 status of the individuals. But you can see that in the two subjects that were considered to definitely have Dilly due to green tea, both had carried the HLA risk allele and one was homozygous for it. That's the green bar. And then in the cases, the nine cases deemed highly likely due to green tea, essentially 90% carried the allele. And in the probable cases, 64% carried the allele. But once you looked at the possible cases, the incidence, 18%, was very comparable to the carriage rate in Dilly due to other HDS products in the Dillon registry, 15%, or Dilly due to the other drugs in the 1,113 other cases in the Dillon network. And then very similar to the carriage rate in a very large population of controls, the emerge population. But in the unlikely category, you can see that the carriage rate was relatively high, suggesting that some cases may in fact be due to green tea. Furthermore, probable in the Dillon scale is only greater than 50% likelihood. So identification of carriage of the risk allele should move individuals from probable causality to the higher categories that has improved confidence of the association. It should also be noted that the same allele HLA-B3501 has been identified and validated as a risk factor to Dilly due to polygonum multiforum in Asian populations. So it may be that this is a more general risk factor or there's some common element which has not yet been identified that is the real culprit in these products for Dilly. So I'm going to close just by thanking the Dillon network. It's been funded since 2003. This is 20 years. It'll be funded till 2023. We're certainly hoping it gets funded for another cycle. I think it's been very productive, but there's much work to be done. These are the currently funded centers as well as the principal investigators. If you are near these sites and see a patient with Dilly and it's possible to get them to visit these sites, we would love to enroll them. It is also or it may be possible to admit very good cases remotely through what we call our retrospective study. So if you have good Dilly cases, feel free to contact each of these investigators. And here's a picture of the HAPI group taken a couple years ago. And with that, I will close and say thank you and wait for the question and answer period.

Quality measurement in healthcare

Process measures in healthcare

Outcome measures in healthcare

Donabedian model

Quality metrics

Data classification

Statistical interrogation

Clinic performance analysis

Liver diseases

Gut microbiome

Non-alcoholic steatohepatitis (NASH)

Drug-induced liver injury (DILI)

HLA genotypes

N-acetylcysteine

Personalized medicine