Thank you for attending this afternoon's symposium organized by the Cholestatic and Autoimmune Liver Disease Special Interest Group. I am one of the co-moderators. My name is Mitchell Mahmood from Duke University, and my co-chair is Dr. Gideon Hestrell, Professor of Medicine at the University of Toronto. Thank you, Mitch. And it is a pleasure that you're here for the Special Interest Group. This is a session which is just slightly before the main sessions at ASLD from our Special Interest Group, which is interested in PBC, autoimmune hepatitis, and other auto-inflammatory liver diseases. We hope today to stimulate discussions around pathogenesis, diagnosis, and management of autoimmune liver diseases, including overlap syndromes. We have learning objectives which are ready to try and help you understand various phenotypes observed in autoimmune liver diseases, understand histology from the context of a non-autoimmune disease but still shows portal inflammation, and to think around overlapping features with, in particular, primary sclerosing cholangitis. Our programming includes four talks. The first talk from Dr. Lily Dara, autoimmune liver disease phenotypes across the spectrum of PBC and AIH. Then we have Dr. Alex Dmytka, how and when should you treat autoimmune features of PSE? We then have a talk from myself and Dr. David Kleiner, how immune is the histopathology of non-alcoholic steatohepatitis, a case-based discussion. We finish off with a talk from my colleague, Dr. Sonia McParland, on liver disease immunophenotyping using new technology platforms to untangle autoimmune liver disease presentations. I'll hand it back now to Professor Mahmoud just to finish off this short introduction. Thank you very much to our audience for attending this interesting symposium. There will be Q&A session at the end of this program, and we look forward to hearing from you. Once again, thank you. Hi, I'm Lily Dara from the Keck School of Medicine and the Research Center for Liver Disease at the University of Southern California in Los Angeles. Thank you for inviting me to talk about the phenotypes across the spectrum of autoimmune hepatitis and PBC. The diagnosis of AIH and PBC is based on a constellation of clinical, serologic, and pathologic findings, and there is a range to these presentations. Disease features in a single patient can change over time, and they appear to be modified by treatment. AIH has no pathognomonic feature, and the pathognomonic feature of PBC, the floor duct lesion with granuloma, is seen in up to one-third of patients and only in early-stage disease. Serologic markers are imperfect tools and must be taken into clinical and pathological context. And lastly, for both AIH and PBC, there is significant heterogeneity in pathologic findings on biopsy. The severity of the disease process varies even among classic cases. Biopsy is fraught with sampling errors, and biopsy interpretation can be subjective, as I will show you. During initial, especially acute presentation, due to the indiscriminate liver inflammation, transient hepatitis and cholangitis may coexist on biopsy. So, both AIH and PBC can display features of the other entity. In AIH, we can see an elevated alkaline phosphatase, more than twice the upper limit of normal, in 20 percent of cases. Non-separative cholangitis can be seen in 9 percent, floor duct lesions in 5 percent, and even dactopenia in 7 percent of patients. Low titer AMA positivity is seen in up to one-third of cases. These cases of AMA-positive AIH have been followed for over two decades, and they do not develop PBC. On the flip side, patients with classic, straight-up PBC can display phenotypic features normally attributed to AIH. Moderate lymphocytic interface hepatitis is seen in up to 50 percent, severe, in 25 percent. Plasma cell infiltrates are seen. ANA and anti-actin smooth muscle antibody may be seen in up to half of the patients, and anti-SLA and LP may be detected as well. Periportal, or periceptal, interface hepatitis is one of the classic features of AIH. However, it can be seen in many PBC patients on biopsy. In this PBC patient, the portal tract is in the right upper-hand corner, and you can see that the inflammation is extending beyond the limiting plate and spilling into the parenchyma. In this case, the inflammation was encompassing less than 50 percent of the area surrounding the portal tract, and it was graded as moderate interface hepatitis. This phenotypic presentation has a lot of varying on treatment decisions and outcomes, so I'd like to point out that the interpretation of moderate interface hepatitis can be subjective, and it's very helpful to have the slides reviewed by an expert liver pathologist. The modified hepatitis activity index, or HAI, by Eshack et al. is to the left. Interface hepatitis can be described as mild to moderate if there is focal inflammation in most portal areas. Moderate is described as continuous but occupying less than half of the area surrounding the portal tract. As you can see, that's very subjective. On the right-hand side, the basal luteate grading system does not define moderate, and due to the nature of sampling during the biopsy, if there's a discrepancy in two areas, the more severe lesion determines the grade. I am bringing this up to say, in order to get a true sense of the degree of inflammation, a conversation, and better yet, a review of the slides with a pathologist will be very helpful in those PBC cases with moderate interface hepatitis before treatment decisions are finalized. So, as I just showed you, interface hepatitis can be seen in classic PBC. In a French study where PBC patients were biopsied, de Gaulle and colleagues showed that over half of the PBC patients displayed moderate interface hepatitis and 25% displayed severe interface hepatitis. This is important as interface hepatitis is a predictor of liver disease progression. As you can see in the graph on the top right, the probability of cirrhosis at 10 years increases with the increasing degree of interface hepatitis detected at baseline. In the lower bar graph, you see unpublished data from USC. Our group in LA looked at Latinx patients with PBC alone who had a liver biopsy. As you can see, patients with moderate to severe interface hepatitis were less responsive to therapy, had faster progression, and worse outcomes. Another phenotypic feature that could occur across the spectrum of autoimmune liver diseases is duct injury. Intense lymphocytic inflammatory changes causing destruction of the basement membrane of bile ducts and destruction of the duct is termed florid duct lesion. An example can be seen in the duct beneath the yellow arrow. In some cases, epithelioid granulomas are present in the same biopsy. This is a classic feature of PBC. You can see an example of this in the panel on the right. However, this feature is more often present in the early stages of disease. Lymphoid aggregates with germinal centers can also be present in PBC. This intense inflammatory infiltrate that's surrounding the interlobular bile ducts includes predominantly cytotoxic T cells, but B cells, plasma cells, monocytes, and eosinophils can also be present. Although non-separative destructive cholangitis is the classic pathognomonic finding of PBC, background duct injury occurs in AIH as well. An estimated 14 to 20% of AIH patients display a degree of cholestatic phenotype. This can be seen in the initial presentation of AIH and often response to immunosuppression. This group of AIH patients with duct injury are very heterogeneous. And when you tease out the clear AIH cases from those that have features of PBC and PSC, i.e. the classic AIH cases, they alone can have ductopenia in up to 7% of the time. If you look at the table on the bottom right, patients with AIH and isolated histologic evidence of duct injury respond to steroids and have similar outcomes to those without any duct injury. In six out of eight patients who had a repeat biopsy, the duct injury disappeared after immunosuppression, indicating the attack of cholangiocytes during the course of AIH in some patients may be a temporary or transient stage and a reflection of the intense portal inflammation in acute and very active autoimmune hepatitis. From what we've discussed so far, it is clear that given the considerable number of shared features between AIH and PBC and the wide spectrum of clinical presentations for each disorder, we should not rush to label a patient with an isolated atypical histologic, biochemical, or serologic feature as having a variant syndrome. Patients displaying features of both PBC and AIH are a reflection of the wide distribution of the clinical pathological features of autoimmune liver disease. Therefore, it has been suggested that we should avoid using the term overlap syndrome as it applies a distinct and separate entity. So, I will use the term PBC-AIH variant or AIH-PBC variant depending on the dominant disease. Since autoimmune liver diseases can present as a spectrum, are there patients that present somewhere in the middle? Well, this has been a topic of interest for many years. In this table, I have listed the most frequently cited papers on this topic, and I want to bring to your attention a few issues. First, the clinical phenotypes of patients with the same diagnosis of PBC and AIH variant is very heterogeneous. These studies use different scoring systems and criteria to define patients with AIH and PBC variants. As you can see, depending on the series, the variants account for 1 to 24 percent of the total cases. That's the column on the right. Second, the studies are all retrospective series, and most have small numbers of PBC-AIH variants. Even the often cited classic Chazollière study, which is the basis of the Paris criteria, records only 11 PBC patients with features of AIH. Third, when the Paris criteria are applied in many, especially older studies, the presence of moderate to severe interface hepatitis is not mandated, and cutoffs for lab values of the limit of normal are not specified. Lastly, most studies are of predominantly PBC cohorts. Therefore, we have more data on the patients on the PBC end of the spectrum, with some features of AIH. The clinical management prognosis and outcomes of these patients is not necessarily the same as AIH patients presenting with features of PBC. The Paris criteria was first described in a 1998 paper where the authors categorized 11 PBC patients meeting two of three criteria for either disease as having a PBC-AIH variant. These criteria have recently been modified in that for the AIH portion of the criteria, moderate to severe interface hepatitis on histology must be met, according to the ASLD and ESL. Therefore, a liver biopsy is always necessary. The other two criteria for AIH are an ALT over five times the upper limit of normal, and an IgG over twice the upper limit of normal, or the detection of anti-smooth muscle antibody. Please note the upper limit of normal in this study. For the PBC component, two of the following three criteria must be met, an OCFOS over twice the upper limit of normal, or a GGT over five times, an AMA over 140, and the presence of a florid duct lesion on biopsy. So there are issues with the uniform application of the Paris criteria, and there remain uncertainties on how to apply them. The upper limit of normal of ALT in the 1998 study by Chaz Ulier was 35 units per liter. Therefore, an ALT of over 175 would get one point. However, in the current era, our definition of a normal ALT in females is less than 19, and applying this as the upper limit of normal for Paris would mean that an ALT of more than 95 units per liter would qualify as one point. I think you would all agree that a moderate amount of transaminase elevation is not at all unusual in PBC, and most often responds to ERSA. The guidelines do not specify which cutoff to use as the upper limit of normal, so this causes some confusion. As previously discussed, the quantification of the degree of interface hepatitis is subjective, prone to sampling error, and classic PBC, especially when untreated, can present with moderate to severe interface hepatitis. And third, on the flip side, since only a floor duct lesion counts histologically towards the histologic PBC component, some have argued that later stages of PBC, such as in cases of bile duct loss, may be missed and result in underdiagnosis of AIH patients that may have AIH-PBC variant and benefit from UDCA. So moving on, what is the immune phenotype of patients with features of both PBC and AIH? Elevated serum IgM is a feature of PBC. Nearly every PBC patient, including AMA-negative cases, have elevated IgM. Memory B cells derived from patients with PBC and stimulated in vitro with a bacterial product, such as CPG, have been shown to secrete greater quantities of IgM compared to controls, both healthy and diseased. This seems to be a unique feature of PBC. Older studies using peripheral blood T lymphocytes from PBC patients co-cultured with memory B cells have suggested that abnormal T suppressor cell function, rather than hyperactivation of B cells, is responsible for this hyper-secretory IgM state. Either way, the plasma cells in PBC seem to be loaded with IgM. There are IgG-positive plasma cells as well. But it's important to note that there are features to the plasma cells in each disorder. For example, in AIH, plasma cells are abundant in the area of interface hepatitis. As you can see, the brown staining next to the black stars. In PBC, in the middle, you can see plasma cells show a coronal arrangement surrounding the intrahepatic bile ducts, especially those with chronic non-separative destructive cholangitis. These are depicted by the arrows in the middle panel. And in the last image, you can see double immunostaining for the plasma cell marker in blue and IgM in brown. Many of these plasma cells are actually IgM-positive plasma cells. Previous studies have shown that plasma cells in liver biopsies of AIH patients are predominantly loaded with IgG, whereas PBC patients have an abundance of IgM-positive plasma cells. IgG-positive plasma cells are also seen in PBC, but the issue is the ratio. Most of the studies that have looked at this before had small numbers, and they didn't have a lot of variant patients in them. And when they did, they didn't use the Paris criteria. We stain liver biopsies from patients with AIH and PBC, and those with PBC-AIH as defined by the Paris criteria, including the histologic criteria, for IgG and IgM. Dr. Gary Cano, a liver pathologist blinded to the clinical diagnosis, counted the number of cells per portal tract. Plasma cell IgG to IgM ratios were calculated by dividing the total IgG plasma cells by the total IgM plasma cells per portal tract for each patient. As you can see in the representative photomicrographs, AIH patients displayed a lot more IgG-positive plasma cells compared to IgM, whereas in both PBC and the PBC-AIH variant patients, there was an increase in the number of IgM-positive plasma cells, resulting in a much lower IgG-to-IgM ratio, close to one. Therefore, the patients who met Paris criteria for PBC-AIH variant more closely resembled the PBC patients. Like the group in Miami, we have seen in our Latinx patients a high prevalence of PBC with features of AIH. These patients present with more advanced fibrosis and have a faster disease progression and worse outcomes. You know, the question is really, are these PBC-AIH variants just a severe form of PBC, as others before us have suggested? In this slide, you can see the outcomes of the USC-Los Angeles County autoimmune liver disease patients. The PBC patients with AIH features had a worse response to therapy, increased progression and cirrhosis, and a higher percentage of death and transplant compared to PBC-alone and AIH-alone patients. In summary, there are a considerable number of similarities in the histologic, biochemical, and serologic presentations of AIH and PBC. The term overlap syndrome should be avoided, as these patients do not have a separate disease entity. And we now recognize that there is a continuum of clinical pathological features that can be used to determine the severity of AIH and PBC. In summary, there are a considerable number of similarities to the continuum of clinical pathological features across the spectrum of AIH and PBC. Moderate to severe interface hepatitis is seen in PBC and is an independent predictor of progression. Non-separative cholangitis and even dactypenia can be seen within the spectrum of AIH. Pathology is key, and all cases of suspected PBC with features of AIH should be biopsied. Most patients will display a dominant phenotype, and this should guide therapy. Current studies are too small to determine how a diagnosis of PBC with AIH feature is different from uncomplicated PBC. And importantly, my pet peeve, we should define subjective features such as moderate interface hepatitis more clearly and agree on cutoff values for labs when applying the Paris criteria. PBC patients of Latinx and Hispanic ancestry present with more features of AIH have poor response to therapy and worse outcomes. Further studies are needed to determine if this severe presentation and these poor outcomes are a function of hereditary factors and ethnicity or socioeconomic disparities or both. Our group has showed that the immune phenotype of plasma cells and the ratio of IgG positive to IgM positive plasma cells on liver biopsy suggests our PBC AIH variants more closely resemble PBC. Obviously, this is a very descriptive finding, but nonetheless intriguing. More detailed immunophenotyping of these patients by transcriptomics such as single-cell RNA-seq or cytometric and cell sorting methods such as cytometry by time of flight or CyTOF and image-mass cytometry or IMC may help in defining the immune phenotype spectrum of disease. Finally, I want to end by saying thank you to our funding and to my colleagues and collaborators at Keck, including Jeff Kahn, who I work with closely on the autoimmune projects, Neil Kaplowitz, my wonderful mentor, to Brian Lee, pictured on the right, my former fellow and very talented physician-investigator and current faculty at Loma Linda University who really drove this project forward and did all the analysis for it. And a big thank you to the amazing Gary Kainil who sits behind the multi-headed microscope every Friday afternoon during Reynolds rounds and reviews our biopsies with us. His expert input is invaluable to us and to the patients. Thank you for your attention. Hi, everyone. I would like to, first of all, thank the organizers of the Cholestatic and Autoimmune Liver Disease Seek for inviting me to present on such a controversial topic as how and when should you treat the autoimmune features of PSC. Nothing to disclose which is relevant to this talk. I would like to start out with a case followed by a discussion on definition, diagnosis, prognosis, treatment of PSC autoimmune heterohydritis overlap, and then finally just highlight a few research opportunities which come up in the near future. Our case is an 18-year-old male who presented to a local ER with a rash which was subsequently diagnosed as HSP and he was noted to have highly elevated AST, ALT, GGT, alkaline phosphatase. The review of system was positive for diarrhea. He was a pretty healthy male, 86 kilo with no stigmata of chronic liver disease. When he was seen by a local GI doctor, he was found to have elevated IgG, ANA, anti-smooth muscle antibodies, and p-ANCA. A liver biopsy showed interphase hepatitis with plasma sets, prominent ductive proliferation, and mild periportal fibrosis. The MRCP was negative for cholangiopathy and the colonoscopy showed diffuse erythema and ulceration with vectors bearing consistent with a PSC IBD type ulcerative colitis. He was diagnosed with type 1 autoimmune hepatitis with overlap with small duct PSC and ulcerative colitis. He was started on prednisone 60 milligrams daily, which was weaned over six months, azathioprine, ursodiol, and salomon for his gut disease. Within six months, his LFDs completely normalized, his IgG normalized. Then unfortunately, nine months after the diagnosis, he had a flare in his ASD, ALT, alkaline phosphatase, which all rose again. IgG was slightly higher from the baseline at six months. A repeat liver biopsy showed cholangitis, pericholangitis, periductus sclerosus, lymphocytic infiltrate, progression of fibrosis, but not much features of interphase hepatitis or plasma sets. The colonoscopy was also repeated and the colitis was in remission. A repeat MRCP was still negative for cholangiopathy. Now, he was diagnosed with small duct PSC as a predominant phenotype with autoimmune hepatitis, overlap, and ulcerative colitis. It was decided not to escalate immunosuppression therapy, but to start him on oral vancomycin for his pediatric PSC in addition to azathioprine, ursodiol, and salomon. Within two months, his LFDs again normalized and his IgG normalized. Now, here are the questions. Many questions. I won't be able to answer all of them, but should he be continued on immunosuppression with azathioprine? That was obviously the first question, which I was asked when I saw him for a second opinion. Should even prednisone be resumed, knowing of that overlap? And is long-term immunosuppression necessary, even in the absence of prominent autoimmune features on repeat evaluation? There is no definition for PSC autoimmune hepatitis overlap. It is a description which is usually done in the context of comparing it against typical autoimmune hepatitis and typical PSC. And usually, patients who are diagnosed with PSC autoimmune hepatitis overlap are expected to meet the diagnostic criteria for both diseases, PSC and autoimmune hepatitis. There was a nice consensus position statement by the International Autoimmune Hepatitis Group, which has a controversial issue already in its title, and I will frequently refer to that publication. And that publication also included five theories on a PSC autoimmune hepatitis overlap. Is that a condition which is a sequenced presentation of two distinct disorders, which you could argue was our case, first with autoimmune hepatitis, subsequently with PSC? Is it a concomitant presentation of two distinct autoimmune liver diseases? Is it a continuum of disease without strict boundaries? Or is overlap itself a distinct disease entity compared with PSC and autoimmune hepatitis? The authors of that position statement suggest that most investigators favor the fifth option, which is overlap is one primary disorder that also has characteristics for the other. And there is some truth to it, at least what we find in animal models. As many of you know, the MDR-2 knockout mice is a very well-described toxic bile-induced model of PSC. And if you observe these mice long enough, up to 25 weeks, you will find that these mice expand the B cells in the liver, and that is associated with a rise in IgG levels and occurrence of antinuclear antibodies. This autoimmune phenomenon is not inconsequential for the disease, because if you deplete those B cells, you can hold the progression of the fibrosis. Our group also started to investigate PSC autoimmune hepatitis overlap in an animal model. And we used the DDC model, and there are two presentations here. The ASR-D1 is a poster presentation by Ramesh Kudera, and the other one is a talk at the fibrosis session on our single-cell RNA-seq and the TEC-seq data in this model. What we did is we exposed these mice to two weeks of DDC, which is a xenobiotic which induces a brisk biliar injury, and then we observed them for four weeks afterwards during the resolution of the biliar injury. And we found that alkaline phosphatase levels following the injury quickly reserved to baseline values. But during that recovery period for the four weeks, we were surprised to find that the fibrosis actually progressed. And that progression of fibrosis following the biliar injury was accompanied by a B cell expansion in the liver by a rise in serum IgM levels. We did single-cell RNA-seq studies and the TEC-seq studies to investigate the cellular crosstalk between all these cell populations, and we also depleted the B cells to look for the functional relevance. And we did find that B cell depletion halted the progression of fibrosis, really suggesting that autoimmunity in that model is a secondary process which follows biliar injury, but it is not inconsequential for the disease. Coming back to the position statement and why we don't have diagnostic criteria, the problem is that the diagnostic criteria for autoimmune hepatitis and PSC are not particularly specific. Particularly looking at the autoimmune hepatitis criteria, as we all know, hypergamma globulinemia is required to make the diagnosis of autoimmune hepatitis, but elevated IgG levels are found in up to 61% of patients with typical PSC. ANA can be elevated in up to 77% of patients with typical PSC, and not even interface hepatitis is particularly specific for one of these conditions. So if you don't have diagnostic criteria, how can you develop the diagnostic criteria for the overlap syndrome? So what can we learn from prior studies in pediatric and adult patients with PSC autoimmune hepatitis overlap? Let's start out with the largest study, which is a retrospective study from the Pediatric PSC Consortium in which 781 patients were identified with PSC. Of them, 33% of the patients, 260 patients were designated with PSC autoimmune hepatitis. You see all the different phenotypes of PSC in the table below. On the very right, we have the PSC autoimmune hepatitis overlap patient compared to those PSC patients with autoimmune hepatitis. You can see that those patients had more females in that group, and you see that more of them had no IBD associated with the disease. When studies were performed in adults with PSC, the frequency was much lower. The retrospective study from the Mayo Clinic and 211 patients with PSC found overlap with probable autoimmune hepatitis using the revised international autoimmune hepatitis group score of only 6%. A study from Amsterdam and Netherlands found a slightly higher rate of 8% of overlap within 113 patients with PSC. How about the other perspective, looking from autoimmune hepatitis and investigating how many patients with autoimmune hepatitis develop overlap features? That was done by a prospective study at King's College and published in 2001, 20 years ago. 55 patients were enrolled with suspected autoimmune hepatitis and underwent ERCP, intraoperative cholangiogram, liver biopsy, and flexible sigmoidoscopy. Using these three diagnostic modalities, 49% of the patients were diagnosed with a PSC autoimmune hepatitis overlap. It was found that, interestingly, these two groups did not really differ by any biochemical features. The alkaline phosphatase or GGT levels were not higher in the group with overlap compared with those with pure autoimmune hepatitis. When that study was reproduced in adults, it was interesting to see that there were a much lower prevalence of PSC overlap. 59 adult patients with autoimmune hepatitis underwent liver biopsy and MRCP, while 14 of these patients had mild intrahepatic cholangiopathic changes. Only one of these patients was ultimately diagnosed with PSC, which gives us a rate of 1.7% of overlap among adult patients with autoimmune hepatitis. When those patients with mild changes on MRI were compared with those patients without any changes, it was noticed that these patients had advanced fibrosis with a lower platelet count and an advanced fibrosis stage on the liver biopsy, suggesting that those intrahepatic biliary changes may be secondary to the fibrosis. What do our societies suggest? How do we make the diagnosis? The recently issued ASAD guidelines for autoimmune hepatitis states that the diagnosis of autoimmune hepatitis PSC overlap syndrome should be considered in all patients with autoimmune hepatitis and chronic ulcerative colitis, unexplained cholestatic laboratory findings, or non-response to conventional corticotherapy. The British PSC guidelines came from a perspective of PSC, and they suggested that liver biopsy is recommended for those patients with significantly elevated transaminases, immunoglobulin levels, or positive autoantibodies. What do we know about the prognosis? Let's start again with a pediatric PSC study. Of these 781 patients, 38% of the patients developed complication of portal hypertension over 10 years, 25% biliary complication, 14% of 113 patients underwent liver transplantation. If you look at the patients with PSC autoimmune hepatitis in that top table on the very right, compared with those without autoimmune hepatitis, there's not a significant difference in the occurrence of these liver-related endpoints, similar rate of portal hypertension, biliary complications, and five-year event-free survival. In a multivariate analysis, to predict event-free survival, small duct PSC and presence of IBD were found to be protective, where there was no predictive power of autoimmune hepatitis. So, what is the prognosis of patients with overlap when compared with autoimmune hepatitis? Again, going back to the prospective study from 20 years ago in pediatric patients, autoimmune hepatitis and sclerosing cholangitis, both groups of patients were treated with prednisone and azathioprine. The ASC group was also treated with ursodiol. We see that the ASC patients more quickly normalized the alkaline phosphatase levels compared with pure autoimmune hepatitis. But importantly, while there was a response in those biochemistry, there was a trend towards worse clinical outcomes because four patients in the ASC group required a liver transplant, whereas none of the pure autoimmune hepatitis patients required a transplant. Those findings were actually reproduced in an adult cohort from the same institution. Out of about 350 patients with autoimmune hepatitis, the King's College group identified 16 patients with PSC autoimmune hepatitis overlap, which is 6%, much less, of course, than what we find in these pediatric patients. They were younger, only 27 years. It's a median age compared with 46. And they did have a lower rate of complete response to immunosuppressive therapy. And most importantly, they had a higher incidence of liver-related deaths or transplantation within that PSC autoimmune hepatitis overlap group. An elevated ALP-AST ratio was found to be increasing the odds for reduced survival in a multivariate analysis. So how about the treatment? We obviously have a problem here. Immunosuppression is known to prolong event-free survival in autoimmune hepatitis. Immunosuppression is not proven to be safe or efficacious in the treatment of PSC. Withholding immunosuppression from patients with PSC autoimmune hepatitis overlap may therefore allow progression of the hepatitis if autoimmune hepatitis is a driving factor. On the other side, overtreatment with immunosuppression may expose patients with PSC who are already at risk for cholangiocarcinoma, colon cancer, and metabolic bone disease to the adverse effect of chronic corticosteroid exposure. So do we have any recent results on studies? Again, the King's College Group recently published on their experience in pediatric patients with IBD who were found to have a diagnosis of sclerosing cholangitis based on MRCPs. Interestingly, in that cohort, 72 percent, 59 out of the 82 patients were diagnosed with autoimmune sclerosing cholangitis because these patients had an international autoimmune hepatitis group score of seven or above. It is worth mentioning that the disease severity seems to be rather mild based on alkaline phosphatase levels, but only 162. GGT levels only 112 compared with 259 GGT levels in the Mark DeNose retrospective PSC cohort of 780 patients. I would also like to draw your attention to the fact that pANCA were considered as autoantibodies in the diagnosis of autoimmune hepatitis overlap. Given the high rate of autoimmune hepatitis overlap, it is not surprising that 77 percent of these patients with sclerosing cholangitis and IBD were treated with prednisone and azathioprine for their liver disease and also diet. The authors showed pretty good outcomes in that particular cohort. None of these patients developed a complication of portal hypertension over a four-year, five-year observation period compared with 21 percent in the retrospective large multicenter cohort study, only four percent really a complication, and none of the patients required a transplant. Experience in adult patients with PSC autoimmune hepatitis overlap is again limited by a small patient population. There's a retrospective study from Sweden in which there were seven patients with small duct PSC and autoimmune hepatitis overlap versus 19 patients with large duct. Immunosuppressive therapies seem to result in a biochemical response, especially in the patients with large duct PSC, not so much at all in patients with small duct PSC, but we still had a pretty high liver transplant and death rate in patients with large duct PSC treated with immunosuppression. What are the ASAD guidelines for the management of autoimmune hepatitis? It is suggested to use empiric data and use glucocorticoids or glucocorticoids in combination with azathioprine or ursodiol. The statement also suggests to direct the therapy to the predominant manifestations of the overlap syndrome. The British PSC guidelines suggest that those patients with good evidence of PSC and additional features of autoimmune hepatitis should be treated similarly to those with classic autoimmune hepatitis. The choice of the most appropriate systemic study is not clear. What do I do? I don't search for an overlap. I identify the predominant phenotype of autoimmune hepatitis or PSC. I use liver biopsy and MRCP at the beginning of the diagnosis of autoimmune liver disease in all of my patients. I do not include p-anchor and consideration for autoimmune hepatitis diagnosis. If patients start out with autoimmune hepatitis, I do search for overlap with PSC in those patients who are male, who have associated IBD, who have pruritus, who have elevated cholestasis markers. If we find overlap with PSC, I tend to start them on ursodiol or vancomycin in patients with IBD before escalating immunosuppression therapy if the LFTs are elevated. In patients with PSC who have atypical feature like female sex, absence of IBD, highly elevated IgG, ANAR can, I do consider the diagnosis of overlap. We have as research opportunities two abstracts from the pediatric PSC consortium. We have the children's study which is an NIH-supported multi-center study. We start collecting data on prospectively enrolled PSC patients across 12 centers. The overlap with autoimmune hepatitis is one of the key research questions. Our group in Cincinnati joined a learning health network which will enroll patients with PSC or autoimmune hepatitis to assess treatment protocols, side effects, and response rate in these patients with overlap syndrome. The key takeaways are frequency of PSC overlap is more common in children. That's why in children, we may have an opportunity to really assess prospectively the efficacy and safety of immunosuppressive therapy in these patients. And giving the paucity of prospective data, it is reasonable to direct the therapy to the predominant phenotype in an individualized fashion in children and adults. I thank you for your attention, and I'm looking forward to questioning this controversial subject. So, hello. My name is Dr. Gideon Hirshfield. My colleague presenting today is Dr. David Kleiner from the Laboratory of Pathology and National Cancer Institute. We appreciate this is a Zoom meeting now that ASLD is not in person. But when we planned this presentation, we hoped to have a two-way discussion between clinician and pathologist to fit within the special interest group meeting on autoimmune and inflammatory disease. So, the purpose of the next 20 minutes is to try and generate some thoughts and discussions about how immune is the histopathology of non-alcoholic steatohepatitis, a case-based discussion. These are our disclosures. What I'd like to do is give a general overview of the topic and then ask Dr. Kleiner to help us with some real examples of histopathology that are challenging, and to use those to challenge you about what you think the histopathology of NASH should look like, and how features of autoimmune injury may be seen in patients that you're biopsying, and what this means for clinical practice and what this means for our scientific understanding of progressive liver diseases. As you know, when we think about progressive liver diseases, we see them into very clear boxes, viral, metabolic, alcohol, autoimmune, and genetic. But of course, when we look at the features of these diseases and how they present to our clinic, nothing is as discrete. And particularly, as you've heard already in this session around autoimmune liver disease, where we don't know the etiology, and therefore we are very dependent on descriptions. When we do think about autoimmune liver disease and why our patients get autoimmune liver disease, we recognize that behind the scenes, there is a genomic effect, an epigenome, the exposome, and the microbiome. And all of these different factors come to play in an individual in the presentation of one of the major autoimmune liver disease syndromes, PBC, autoimmune hepatitis, PSC, or the autoallergic disease IgG4-related cholangitis. But fundamentally for autoimmune liver disease, we don't have an etiology, so we use descriptions and we use exclusion to reach our diagnosis. And it's for that reason it's not a surprise that features can overlap. And when I think about how features can overlap, I think about the concept of, for example, interface hepatitis across the three different autoimmune liver diseases, and how I recognize that interface hepatitis is frequently moderate to severe in autoimmune hepatitis. But if you look hard enough, you'll see interface hepatitis in PSC and you'll see interface hepatitis in primary sclerosing cholangitis. And when you think about how severe a feature of any disease will be, there's usually some form of normal distribution. So there'll always be patients with whichever inflammatory liver disease we're seeing who have perhaps a more extreme presentation of one of the features that we measure in clinical practice. When you evaluate the patient for autoimmune liver disease, we have essentially a very important cognitive step to decide whether the liver tests are cholestatic or hepatitic. But we have a very significant challenge, which is really where the discussion comes today, around excluding common liver disease. And when we think about common liver disease, we now really are facing a very high rising tide of metabolic and fatty liver disease. So this really is something that is under the microscope, both clinically and from a clinical trials perspective. Now, the definition of fatty liver is evolving. This is the latest proposal for so-called MAFLD, but in essence, we're looking at patients with either obesity, a lean weight, or type 2 diabetes, who then have, in addition, added metabolic risk abnormalities. And the reason I show you this is because this is clearly very common, even in a population of patients with autoimmune liver disease. And therefore, not just clinically, but scientifically, we can expect there to be some commonality in pathways. But additionally, when we think about what is NASH, we then, as we understand it more, have also realized that NASH, or NAFLD, or MAFLD, whatever you want to call it, is not just a metabolic process, but it is inherently an inflammatory process as well. And we've learned that inflammation amongst genetics, insulin resistance, diabetes, the microbiome, is also very important in how the disease presents, how the disease progresses, and potentially how the disease may be treated. This is the sort of classical, simple histopathology for clinicians around what NASH looks like. And I just show you this to anchor some of the things that we're going to talk about, steatosis, ballooning, lobular inflammation, and fibrosis. It's, however, I think, very neat, but not necessarily very real-world. And this is a very important paper from Elizabeth Brandt and Dr. Kleiner and others from 2011, where they just showed you, actually, the spread of features in patients who've got histiate hepatitis, but then when you look at histology variables. So before I ask Dr. Kleiner to show us some interesting cases which challenge us, I'd just like to ask him to just comment a little bit about portal inflammation and lobular inflammation in NASH. And if he does remember going back 10 years ago to when it must have been a momentous amount of work to put this table together, what he takes away from the concept of inflammation in NASH. So one of the things that has colored the whole field of NASH and steatohepatitis was the invention of this score that we came up with, the NAFLD activity score. And in the conception of that score, portal inflammation was left out. And because it was left out, it kind of left the imagination of researchers and clinicians. They thought it wasn't important, which wasn't really the point of the score, that the score was more for monitoring change over time. But it didn't come out in our analysis of what mattered in terms of our diagnosis. And so from that point, it was left out. But when we go back and we look at steatohepatitis, both as pathologists and as members of this clinical research network, chronic portal inflammation is present in most of the biopsies that we see. You can see that down in the middle of the table. And it worsens as the disease worsens. So this paper and one of the other papers that we published on portal inflammation in NASH both make the point that severity of portal inflammation increases as this overall disease severity increases. So as you go from no steatohepatitis to definite steatohepatitis, and not shown in this table, but what was also clear from our other analyses is that it increases as the fibrosis increases. So as you get more portal fibrosis, as it becomes bridging and finally cirrhotic, generally speaking, you can see more portal inflammation. Nothing is absolute though. So you have cases at the high end of steatohepatitis or fibrosis that don't have much in portal inflammation and the reverse. Occasionally you can see an early case with a lot of portal inflammation. So that's, I think, the important takeaway here. So I think that's a perfect segue. We've now got three cases, which I'm gonna ask Dr. Kleiner to present. And then we'll have a little bit of discussion around based on how much time we have to just to highlight some of the features. So these are really great cases that Dr. Kleiner's chosen that I think really make you think. Okay, so this first case is someone who developed steatohepatitis in the face of primary biliary cholangitis. And this was a patient who was followed for a long time in the liver disease clinic at the NIH, took part in our methotrexate study in the early 90s, but then was switched over to ERSO, soon after that study ended actually, and was on it for a long time. And this was in fact the fifth biopsy that she had gotten. And the first four biopsies really showed no steatosis at all, or a little bit in the fourth. At the time of this biopsy, and I don't know the reasons for doing the biopsy at this point, but at the time of this biopsy, the transaminases were normal, the LFOS was minimally elevated, there was still a positive anti-mitochondrial antibody and anti-nuclear antibody. And you can see that the IgM is a little bit elevated, but there were also indications that she had metabolic liver disease. So here's a low power picture, and you can see that there is portal inflammation, most prominently sort of center right in the picture, and then steatosis upwards and left of that. And you can actually see some ballooning, balloon cells also in that same area of steatosis. The steatosis is very much zonal, just like it is in fatty liver disease that's caused by metabolic disorders. And then as we move on, so despite the fact that she was on ERSO for 11 years, she still has a fair amount of portal inflammation in a number of portal areas. And you can see the kind of interface hepatitis that we see in PBC. And down in the inset in the lower right, you can see keratin 7 stain, and the bile ducts are the very dark outlines in on the left panel, but on the right are hepatocytes that are positive for keratin 7. And this is an indication that there is still chronic cholestatic change, still injury going on, despite the fact that she's being treated for all of these years. These are the NASH features. So you see closer up, you can see the balloon cells more clearly, and steatosis also around the balloon cells, and a little bit of lobular inflammation. In this context, the lobular inflammation is not specific for, and doesn't really help us in the diagnosis. Really what we focus on is the distribution of the fat, the presence of the balloon cells, and malary bodies if we see those as well. So I think that's the last. So can I ask you a question then? We've got just maybe 30 seconds, 45 seconds before we move on. So how would you tell the difference between the etiology of the portal inflammation from the portal inflammation of NASH and the portal inflammation of PBC, if you didn't know that she was AMA positive? Right, so you can't really. So what you can do in this case, because we had the prior biopsies, is you can look back at the biopsy that preceded this, and look at the amount of portal inflammation there when there was clearly no fatty liver disease present, and compare it to the current one. And is there an increase? Does it look like there's more? But in general, if this were the only biopsy that we had on this patient, the inflammation and which lymphocyte belongs to which disorder is not really something that we can distinguish. So that's really interesting. So I'd like to ask another, this is a much harder question. So from your experience of looking at a lot of histopathology, what then do you think is the mechanism for portal inflammation in NASH? I mean, I kind of understand what's going on in PBC. I have a biliary epithelium under attack by the immune system. Why do you think you get portal inflammation in NASH? So that's a good question. And I don't know that from a pathophysiologic standpoint, we really understand that. There is lymphocyte recruitment to the liver because of the injury that's going on in the parenchyma. And some of it accumulates in the portal areas, and I guess does not proceed into the parenchyma. But this seems to be a real constant across all liver diseases that you get both some degree of portal inflammation and even interface hepatitis and lobular inflammation. And the precise character may differ between liver diseases, but they overlap a lot sort of in their general features. And do you think it might, it relates fundamentally to the blood supply and the sort of portal triads and the tracts. Is that the reason it's so focused there? Yes, well, yes. So that's usually the first exit point for at least some proportion of the lymphocytes coming out of the thin walled portal vein and also the capillaries that supply both the parietal biliary plexus and feed into the sinusoids. Great, that's really interesting. Well, we don't really know the direction of liver inflammation. It's quite fascinating. It's probably bi-directional, isn't it? So I think we should move on to the next case because that really highlights a really helpful point for us. So this is the second case. And this is a patient who also took part in one of our studies. This was the P-glitazone pilot study that we did in the early 2000s. There are two biopsies I'm gonna show you. The first biopsy was the patient's study entry biopsy. And the portal inflammation was increased, but there was bridging fibrosis. So it was certainly within the spectrum I would see for NASH. And the second biopsy was done a year after the end of study. And there was more inflammation and interface activity. So here's the basic sort of laboratory information at the time of the first biopsy. So the ALT was elevated. AST was a little bit elevated depending on what the upper limit of normal was. Alkaline phosphatase was normal. The AMA was negative. There were no other autoantibodies that I could find at the time. The ITG was actually a little elevated according to the upper limit subnormal at the time as was the IGM, which is kind of interesting. I was not expecting that when I looked this up. But the biopsy shows portal inflammation and this is quite reasonable portal inflammation for somebody with an advanced stage of NASH. And you can see both aggregation of lymphocytes in the center of the portal area on the left, as well as disruption of the limiting plate, at least focally on the right with more dispersed inflammation. And then here's a little bit higher power picture of the interface activity at the time of the first biopsy. And you can see this interaction along the edge. The limiting plate should be nice and straight. And so the fact that it goes in and out and you have these lymphocytes snuggling up against the hepatocytes indicates that there is some ongoing injury. And then a typical sort of balloon cell on the left. So then the second biopsy is done actually two years later. So one year off pioglitazone, we did an end of study biopsy, but then did some follow-up biopsies in some of the patients. And at this time, the ALT and the AST were higher actually. So clearly above normal. I don't know whether this patient responded or not, but their transaminases are up higher. So autoantibodies at the time were negative, but three years later, interestingly, the ANA was noted to be elevated. Again, the immunoglobulins didn't really change much, but our upper limits of normal changed. So they're normal now. So here's the biopsy from this one. The amount of portal inflammation is about the same, but more portal areas were involved. And it's hard to show that on the screen here, so you'll have to kind of take my word, but the interface hepatitis was definitely more prominent. And you can see complete disruption of the edge on the right-hand picture with lymphocyte sort of freely infiltrating the limiting plate. And then a couple more pictures of interface hepatitis. Again, this was present in several portal areas and there was enough here that I took note of it. I didn't really suggest another diagnosis at the time, but it was certainly commented on in the report. And then we still have the typical features of steatohepatitis. So there's some ballooning on the left and a ubiquitin stain on the right showing these inclusions, which are the Mallory bodies. And so that brings us to the end of this case. We've only got a few minutes left. I would like to ask one question before we do the last case, even if we overrun slightly, Dr. Kleiner. Why does PORT inflammation get worse as the stage of the disease progresses in NASH? That's a good question. You know, I would have to hypothesize about this, but as you distort the liver anatomy more, there are other things that start to happen. You get shunts forming. You might get actually a little bit of cholestatic change as it becomes more difficult to get the bile salts out of the liver with a distorted liver. So you may get other kinds of injury occurring besides the injury that we associate with steatohepatitis. So- But would it be fair to say that that's not a feature that you associate with PBC, PSC, or AIH? You don't see when someone has more advanced cirrhotic autoimmune liver disease that PORT inflammation tracks that? Um, actually it might, because I haven't, I'd have to go back and actually look at the data that I've accumulated over the years, but I have looked at, I know that that's true in chronic viral hepatitis. That's interesting. Yeah. So it really comes down to 3D architecture. It seems to be very important to what you're looking at as a pathologist. Okay, I think we should move to the last case, because we've just got a few minutes left. Okay, this was a case where I did raise a question at the time of the biopsy about something else going on. So you can see the transaminases and the ALKFOS were not really particularly remarkable for NASH. Autoantibodies were negative, except for an elevated ANA. And again, the immunoglobulins at the time were normalish, except for this slightly elevated IgM. So again, kind of like the last biopsy, there was a lot of portal inflammation, even more so than that last biopsy. And what I've highlighted here with the arrows are plasma cells that are involved in the interface activity. And that's always something that you kind of take note of as a possible autoimmune hepatitis feature. More portal inflammation in these pictures, and actually a substantial perivenular inflammation on the right. That's actually around a vein, not a portal area. And you can see this sort of confluent inflammation, which we do see again from time to time in NASH, but not common. And it had typical NASH features. So again, ballooning in both photos indicated on the right with the arrow. I don't know that I had a ubiquitin stain on here, but I think it would have been positive. And so that's this case. It was just, it was enough portal inflammation in this situation for me to raise the question in the report about something else going on. 20 minutes sufficiently. So I'm gonna just finish off with one slide, which I found really informative, Dr. Kleiner, because this is 10 years on from all the work that you and Dr. Brunt did. In the first paper in 2011, and it's sort of an update in 2021 about how to describe what you see, I would also say with some very, very important footnotes, which are relevant, which I don't want people to miss. But again, it's coming up around inflammation, lobular and portal, and the fact that that is a feature of steatohepatitis and that how the pathologist reports that I think is relevant. So maybe if you just wanted to give a little bit of narrative to this one slide and then I'll bring the session to an end. Right, well, so this table is very familiar to me because I was responsible for it. But I think what we tried to do here was highlight these situations where it might be good to call out that something else might be going on. And so, yes, when you look at the footnotes for the inflammation section, which aren't shown on this slide, they're actually a little bit below the text at the bottom there. But you look there, we do say that when you get marked or severe portal inflammation, or you start seeing plasma cells, these sorts of other features, you need to start thinking about other diseases. Typically, clinicians will rule out viral hepatitis, but they may or may not look for other things. And it's just worthwhile bringing this to the attention of your clinician as a pathologist and saying, hey, maybe there's something more going on here than just metabolic liver disease. Great. And so look, I mean, we would love to have done this in person. It would have been fantastic. But really, I think what we tried to do today was just make people aware of how inflammatory NASH can be with a disease process that's very distinct to autoimmune liver disease, but how we can learn about some of those features in terms of the biology. And then to give the high level conclusions is that liver histopathology is still a fundamental contributor to managing patients with liver disease. And it helps describe patterns of liver injury and guides choice of therapy. But those patterns of liver injury need to be understood in the context of what you know about that patient, both definable triggers and other features which describe it. And just as with clinical presentation, serum liver tests and serology findings are heterogeneous. And it's just a small sample. I'm reminded of a recent paper from Germany of a laparoscopic biopsy. In one biopsy from one lobe, there's autoimmune hepatitis. In the other biopsy from another lobe, there's NASH. And this is the same patient. And the biopsies are taken at exactly the same time. And really, we want to recognize that the hepatic pathologist is a consultant with expertise in interpreting biopsy findings in the light of clinical information. So you really do need to talk to your pathologist to make the most of your patient's liver biopsy. And with that, I'd like to thank ASLD and I'd like to thank Dr. Kleiner for finding the time to really dig out some really fantastic cases. And I hope you've enjoyed the session and it's prompted some thought for your own practice in science. Thank you. Thank you. Thank you to the organizers for inviting me to present on our efforts to profile the liver in health and disease using single cell transcriptomic technology. The work that I'm going to describe today has been performed at the Esmere Transplant Center in Toronto General Hospital, in which I'm a scientist, as well as at the University of Toronto. So our group is fascinated with the cellular biology of the liver, and importantly, the fact that the liver has such an immense regenerative capacity, which is impaired on a background of liver disease. And so what I'm going to talk to you about is a very simple story, which provided the basis for our four way into looking at the liver at a single cell level. And then I will talk about some unpublished data towards our efforts to expand our understanding of the liver using single cell technologies, as well as to study liver disease. And so what I'm showing you here is a schematic of the repeating units of the liver. It's called the liver lobule. And what you see here is the periportal area, including the bile duct, the portal vein, as well as the hepatic artery and the central vein. And we'll circle back to these repeating units in the liver as we talk about adding spatial context to transcriptomic signatures within the liver. So my group is very interested in not just the parenchymal cells, the hepatocytes, the cholangiocytes, the endothelial cells, but also the immune cells that cycle through the liver and that are recruited into the liver. And so we're interested in these populations because of the fact that in models of partial hepatectomy or regeneration models in animals, if you delete out macrophages, what you see is that this regeneration is abrogated. So we're interested in understanding the role of macrophages in health and disease and their ability to regulate or support the regenerative processes. And so today, what I first just want to touch on very briefly is a simple study that was our first examination of the cellular biology of the liver using single cell transcriptomics. And the question that we asked was, what is the cellular makeup of the healthy human liver? And the objective of this study was to create a comprehensive cellular map of the human liver, which would provide a foundation for single cell liver disease studies. And this work is a wonderful collaboration between my lab, where a liver immunology lab, Ian McGilvrey's lab, he is a surgeon scientist at Toronto General, and Gary Bader's lab, and he is a computational biologist at University of Toronto. And so the first question we're really asked before we started our study, and it's a question that we'll continue to ask as I discuss our efforts to look at the liver, the cellular biology of liver disease, is how to appropriately handle and dissociate liver tissue. And so for this study, we had access to human caudate lobes. So this was discarded tissue, which was from a healthy individual and being transplanted into a recipient with end-stage liver disease. And having access to this tissue allowed us to cannulate this tissue with its own vasculature. And that allowed us then to introduce enzymes into the tissue to gently dissociate the tissue and to release the tissue without the need for mechanical dissociation. And so what we're able to do is then open the mesenchymal layer and release those cells and profile them by single cell RNA sequencing. And what this allowed us to do is to efficiently capture hepatocytes because the hepatocytes were the most highly impacted by tissue manipulation. And so what this meant was that we had a soup of cells that included both damaged and non-damaged cells. And what we saw in our optimization was that the more we manipulated the tissues to enhance for viable cells, the more we lost hepatocytes. So what we did is we used our bioinformatics pipeline to select cells that had a high library size and a low mitochondrial transcript ratio to generate our map. And so this gave us a map of 20 populations from five healthy livers, and this data is publicly available. And what was interesting was that we saw multiple populations of hepatocytes, which we'll talk about a little bit later in terms of our understanding of whether these different populations of hepatocytes are differentiated because of their zonation or whether this is a dissociation bias. We also saw multiple populations of tissue-resonant macrophages in the human liver. And I just want to touch on how we annotate those populations as we talk about the framework for looking at liver disease. So firstly, what we did was we looked at these inflammatory macrophages and non-inflammatory macrophages based on their defining genes. And what we really wanted to look for was whether the differences between these genes were associated with either cell death or a dissociation bias. And what we saw were the defining genes that differentiated these populations included inflammatory genes such as S100A8 and A9, CDE74 and IL-18. And as well, we saw non-inflammatory genes including VCIG4 and hemoxy differentiated or that characterized our non-inflammatory macrophages. And then what we needed to do then was to look at the function of these populations. And so we used gene set enrichment analysis to look at the active pathways in populations that we annotated as inflammatory or non-inflammatory. And what we saw was that these populations that were more inflammatory in their gene signature were active in immune responses as well as bacterial defense, whereas the non-inflammatory macrophages were active in myeloid differentiation or inhibition of myeloid differentiation and cytokine suppression. And this is something we could confirm with an in vitro assay, a functional assay. And what we did was we differentiated the inflammatory versus non-inflammatory macrophages based on their expression of the protein MARCO. And what we saw was when we stimulated these populations with lipopolysaccharide and interferon gamma to assess their inflammatory potential, we ended up seeing that the MARCO negative population, which were these inflammatory cells, in fact, did have a higher inflammatory potential and secreted more TNF alpha. And we saw that the MARCO positive ones that were the more immunoregulatory ones secreted less TNF alpha. So this allowed us to really look at the liver on a single cell level and really start to understand some of the populations. And just before I go into our unpublished data and our efforts to look at liver disease, I just want to share the fact that all these protocols are available on protocols.io. And there's a nice chat section in which we discuss kind of some of the pitfalls we encountered while generating that protocol. And the data set itself is available through an interactive human liver web tool that you can browse without the need for significant coding in R. And so as we start to look in autoimmune liver disease, we encountered various challenges. And these centers are on three main points. First of all, diseased tissue itself is challenging to dissociate to a single cell level because of fibrosis. Number two, disease itself can be quite patchy. So a biopsy might not capture the most diseased areas of each liver. And then number three, the spatial context of transcriptional patterns need to be captured within the diseased liver. And so in order to address these challenges, we adapted, adopted additional approaches or additional platforms. So what we realized or what we hypothesized is that multiple approaches would be required to really look at liver disease based on these challenges of the ability to access cells within these tissues and the ability to spatially profile them. So I spoke to you already about single cell RNA sequencing. And what we can also do using the single cell RNA-seq platform is perform site-seq in which we use oligo-labeled antibodies that allow us to detect simultaneously transcripts as well as surface proteins that are encoded by those transcripts. And these are picked up using ADT-labeled antibodies. And so what I want to now talk about are two additional approaches that are particularly helpful in light of the challenges of looking at liver disease. And this is single nucleus RNA sequencing and spatial transcriptomics. And so we perform spatial transcriptomics with the desire to really map out the transcripts that we found in both the healthy and then the diseased human liver along the periportal to central venous axis of the repeating units of the liver. And so we've used the 10X genomics platform. We're able to section human liver tissue at a thickness of 16 microns. And each of the spots captured by this technology captures 55 microns of cells. So it's a bulk RNA-seq in these small 55 micron areas. And so what we're able to do then is capture the mRNA from the adjacent cells in the attached tissue. And then what we can do is within each spot, we can deconvolve the data with our single cell maps that we've generated. And so the important thing that we need to do in advance of this is to landmark the human liver tissue using HNA. And this is work that's been done by Cornelia Thione. And what you see is that we can map out the central veins in C and the periportal regions and map out hexagonal units throughout the liver. And we use this as a guide to look at zonation within the liver. And so what we see when we perform spatial transcriptomics to look at hepatocyte zonation is that if we consider the hexagonal units within the liver and we map them out with our pathologist, we see that when we look at the clusters that are generated from the healthy human liver, the largest dimension of variation or the highest variance explained in this data is related to periportal versus pericentral genes. And this is work that was really spearheaded by a brilliant postdoc in my group, Tallulah Andrews, as well as Katya Persiani. So we're able to use this as a platform then to look at the zonation of transcripts within a healthy and diseased liver. And so what we wanted to also examine as we establish our platform for looking at immunoliver disease was how to answer the following two questions. Number one, diseased tissue is challenging to dissociate to a single syllable because of fibrosis, and disease itself can be quite patchy. So a biopsy might not capture the most disease areas. And so in order to answer this question, we established a collaboration between our groups, as well as the Regev and Rosen group at the Broad Institute. And we asked the question of whether we can systematically compare single cell RNA sequencing and single nucleus RNA sequencing for their ability to recover diverse cell populations within the healthy human liver. And then we would use this as a platform to look at the diseased liver. And this is a preprint that is accessible through bioRxiv. And so the important point to note is that we both accessed fresh and frozen tissue from the same patient. We used the dissociation protocol I described in the earlier slides for isolating cells from fresh tissue, but we also used detergent buffers to extract the nuclei from cells from frozen tissue. And we then performed a protocol evaluation comparing the cells and transcripts isolated by single cell and single nucleus RNA sequencing. And we validated these with the visuospatial transcriptomics, as well as the human protein Alice. And what we saw was that single nucleus RNA sequencing captured all the populations we would expect in the human liver based on our understanding of the first map. However, there were certain populations like sinusoidal endothelial cells, stellate cells, and cholangiocytes that were difficult to release in the original map that were captured quite nicely in the using single nucleus RNA sequencing. And this is work that was done again by Tallulah Andrews, a PhD student in my group, Doria Atif, and Jeff Liu, who's a postdoc in Gary Bader's lab. And so what we're able to also look at are where the immune cells within the liver and by single cell or single nucleus RNA sequencing. And what we saw was all immune cells were better captured by single cell RNA sequencing, as opposed to single nucleus RNA sequencing. And we attribute that to the fact that many of the genes that are used to annotate immune cells in the liver are found in the cytoplasm, and they wouldn't be captured by single nucleus RNA sequencing. And the second reason is that single nucleus RNA sequencing is more likely to capture all the cells in the liver, so it relatively dilutes out the immune fraction. And so this allowed us to use some of these tools we developed in Healthy Liver to examine liver disease. And so one of the key points we collected was that we really need to have histology, we need to snap free tissue for a single nucleus RNA sequencing, take fresh cells for single cell RNA sequencing, and then safe cells for flow cytometry validation, as well as spatial transcriptomics. And so what I'll describe is a wonderful team grant in which we're defining and targeting autoimmune liver disease with a particular focus on primary sclerosome cholangitis. And this is a collaboration and a really great effort between the Edgemere Transplant Center, as well as the Toronto Center for Liver Disease with Gideon, Aaliyah, and colleagues in that group. And so really what we want to look at is the explanted tissue, as well as biopsies from patients with primary sclerosome cholangitis. And what we want to perform are various evaluations of the histology, as well as the interpathic heterogeneity, and then perform single nucleus RNA sequencing and spatial transcriptomics to understand and to be able to then validate signatures associated with early and late primary sclerosome cholangitis. And so what we were able to do with the help of our pathologists is to look at tissue structure, fibrosis, bile duct proliferation, as well as copper excess, T-cells, and B-cell infiltrations, as well as macrophage infiltrations. And what we really saw when we started to look within the same liver, so we took multiple wedges from the same liver, were that when we looked by H&E, trichrome staining, as well as CK17 to look at bile duct proliferation within the same patient, we saw quite a bit of heterogeneity in the different wedges we were collecting from these patients. And what we wanted to ask the question, first of all, could we profile these tissues by single nucleus RNA sequencing after we see the histological pattern, and whether there would be kind of a threshold for which the tissue would be too damaged to profile with a single nucleus RNA sequencing. And so what we saw was that primary sclerosin cholangitis samples, as expected, had more mitochondrial RNA than healthy individuals, so suggesting there is more cell death. And this was to be expected. But when we looked at our less serotic versus our more serotic tissue that we saw in the previous slides, what we saw was that even in the serotic tissue from patients that were being explanted for PSC, we could see structure in the data. We could capture cholangiocytes, macrophages, stellate cells, and LCECs, as well as multiple populations of hepatocytes. However, what we did see was that there was a point in which we really couldn't capture very much structure in the data at all. So when we tried to annotate these populations based on our five liver map, we weren't able to capture multiple populations. Most of the transcripts maybe were ambient RNA or hepatocyte-related genes. However, we're able to integrate the healthy tissue with the tissue from the patient with PSC that was the wedge region, which was a little less serotic. And what we could see was that we had multiple populations of cholangiocytes, stellate cells, macrophages, as well as some disease-specific hepatocyte clusters within these patients, within this area of this patient. Finally, we were able to examine the healthy versus disease liver using spatial transcriptomics. So what we're able to use is nanostream digital spatial profiling. And what we wanted to look at was we wanted to examine the healthy liver versus the liver of patients with primary sclerosis and cholangitis. And what this technology allows us to do is to take a 660-diameter circle of tissue, and we're able to create or perform a bulk RNA on this tissue in the context of what we see via pan-CK staining, which stands for the epithelium, DAPI staining, which stands for the nucleus, the CD68 staining, which stands for macrophages, and CD45 staining, which just stands for general immune cells. And what we see when we look at the healthy liver versus the liver of a patient with primary sclerosis and cholangitis, so we see the biliary proliferation, we see immune cell infiltration and macrophage infiltration compared to the healthy liver in which we see minimal or very little biliary epithelium, and we see a very nice structure of the central vein and moderate infiltration of macrophages as would be expected. And so what we found were that these PST ROIs, before we performed the bulk RNA seqs, were enriched with macrophages, immune cells, and biliary proliferation. And then when we deconvolute this data with our healthy liver map, we're able to see actually that inflammatory macrophages, plasma cells, alpha beta T cells, and cholangiocytes are enriched in most PST samples, and certainly in the PST samples that have, that show the most disease. And then finally, we're able to look at human, normal human liver versus PST scars, and again, I'll remind you of the profile of the healthy human liver, and what we see is that lovely lobular pattern in which the genes are, the greatest variation in the genes we're detecting are associated with the central vein to periportal axis within those repeating units. And what we're seeing when we look at the PST liver is that, and this is the HNE staining versus the transcriptoma profile, is that the transcript, the major differences in the transcripts are associated with scar versus relatively normal liver with, or more normal liver in which we see nuclei. And so these populations are populations we're now pulling out, we're deconvolving them with multiple data sets, and we're functionally validating them. And so in summary, I hope I have convinced you that single cell transcriptomics represents a powerful tool to understand the complexity of human liver. However, a multi-pronged approach may be required to really fully map the disease liver as well as the healthy human liver. And I'd like to acknowledge the wonderful team that's involved in this work. Dr. Tallulah Andrews did quite a bit of the transcriptomics work, as well as Jawaria Atif, Dr. Katya Persiani, as well as the McGilvery lab, the Bader lab, all the surgical fellows, the members of the UHN transplant program, the Princess Margaret Genomics Center, our wonderful human immunology team, the PSC team from the Toronto Centre for Liver Disease with Gideon, Aliyah, Adam, Bettina, and Jordan, as well as the SickKids hepatology team and the TGH pathology group, as well as PSC partners who've been involved heavily in the aspects of this work. And I'd be happy to answer questions.

autoimmune liver diseases

PBC

autoimmune hepatitis

diagnosis challenges

management challenges

biopsy interpretation

overlap syndromes

histopathology

primary sclerosing cholangitis

autoimmune injury features

single cell transcriptomic technology

cellular maps