Thank you. I would also like to thank the course organizers for inviting me. These are my disclosures. A 62 year old female with cirrhosis due to alcohol is hospitalized with altered mental status. So in the next few minutes, I want to highlight and explain how does a gut microbiome change in advanced liver disease? What is the role of the circulating blood microbiome in alcoholic liver disease? It's evidence that analysis of the gut microbiome can predict outcomes in phenotype of liver disease and what are the emerging data on modification of the gut microbiome as therapy for alcohol-related liver disease. I will predominantly focus on non-alcoholic and alcohol-related liver diseases today. If you look at the human gut microbiome, you'll see that various niches and organs are differently colonized by our microbes. If we focus on the intestine, you have bacteria, but we also have fungi and viruses present. We have about the same number of human cells as we have microbial cells in our body. But if you look at the genome, you have far more microbial genes that we have human genes. So what factors are now modulating the gut microbiome that allows to transition a healthy gut microbiome to a more dysbiotic state? And one of the most important drivers of gut microbiome changes is diet. We also know that medications are affecting the gut microbiome. Genetic factors based on large population studies are probably a minor factor that affect the gut microbial composition. We have altered immunity, changes in the gastric pH, intestinal motility. Liver disease etiology will also drive different compositional changes in the gut microbiome. And obviously the liver disease stage, are you in a presterotic stage here with alcohol, or are you in a serotic stage? That will also affect the gut microbiome composition, and this is probably largely driven by the altered bile flow during end-stage liver disease. We have looked at this, and we have analyzed by a 16-S sequencing of the fecal microbiome of patients with alcohol use disorder, shown in the green circle. These patients are in early stage liver disease, steatosis and fibrosis, versus the orange circle. These are alcoholic hepatitis patients, predominantly serotic, and they have a completely different gut microbial composition. One of the dominant changes that we found was that enterococcus faecalis has a 3,000-fold increase in patients with alcoholic hepatitis, but this did not correlate by whatever means with clinically severity of disease, also not with outcome of disease. We then focused more on what we have already mentioned, you know, the metabolome and what are these bacteria doing, so the functional component of this bacteria, and we found that these bacteria produce one toxin, which is called cytolysin, and you can see that the cytolysin prevalence in controls and alcohol use disorder patients is very low, while in alcoholic hepatitis patients we have about 30% of these patients are colonized by cytolysin-positive patients, and this correlates very nicely with disease outcomes, so all the cytolysin-negative patients more or less survived by the cytolysin-positive patients. 90% of them die within 180 days, so this is a very precise marker of mortality in this patient core, tacitus ratio, in this multivariate Cox regression analysis after adjustment for several factors is 14. We also looked into a patient population of NAFLD patients with various disease stages, and you can see that there is a low background of cytolysin-positivity, and this did not correlate with any of the disease outcomes, so again, this is probably very specific precisely for alcoholic hepatitis. My colleague at UCSD, Dr. Lumba, he also was able to show based on 16S sequencing and metagenomic sequencing that he can discriminate patients with NAFLD and early fibrosis versus patients with NAFLD and advanced fibrosis based on this PCA analysis. He showed that certain species in the gut microbiome are differently expressed and presented, and using this random forest model where he incorporated like 37 microbial species, BMI, age, and the diversity of the microbiome, he was able to show that with a very impressive AUC of 0.936, he can discriminate patients with advanced fibrosis from early fibrotic disease. So this seems to be actually a much better model and predictor than what we can do right now with human genes. As I have mentioned, cirrhotic patients have a different microbial composition than controls, and pre-cirrhotic patients, they have, this is shown in this paper which came out five years ago, and again, just by looking at the microbial composition, this paper was able to also discriminate with an AUC of 0.92 the cirrhotic patients from the controls. Dr. Beauchamp came up with a little bit simpler idea, this cirrhosis dysbiosis ratio. He essentially just took the ratio between some of the good bacteria and some of the bad bacteria, and if the ratio is high, as shown here in the controls, the, you know, the half indicates a healthy state versus if the disease progresses, you get increased in bad bacteria, and this ratio decreases. As I mentioned, we not only have bacteria in our body and in our intestines, but also fungi. These are sometimes overlooked, and we have analyzed this recently, and we are able to show that in a healthy control subject, not patient, you have a very nice diversity of different fungi present in the fetal mycobiome, about 25%, so one-quarter, or Canada, but independent of liver disease stage, as you can see here in this patient with alcohol use disorder or alcoholic hepatitis. First of all, the fungi diversity decreases, and essentially Canada, and especially Canada albicans, essentially takes over the entire microbiome of these patients, and interestingly, the alcohol use disorder patients and the alcoholic hepatitis patients, shown in the red and blue, you cannot discriminate them based on the fungi profile in their feces, as opposed to what I have shown you before on the bacterial profile, but it's clearly distinct from healthy subjects. Moving away now from the fetal microbiome to more systemic markers, which are also easily accessible to us via just a simple blood draw. We also looked into the fungi response to certain fungi. We measured the ASCA, which is the Antisaccharomyces cerevisiae antibody, that one of the main immunochains, in fact, is Canada, and you can see that alcoholic hepatitis patients have increased ASCA levels as compared to the alcohol use disorder patients and control patients. If you use mathematical modeling, you can see that patients with a low ASCA, or lower than 34 units per ml, have a better survival than the patients with higher ASCA levels, and if you add the AUC from the ASCA to the MELD, you can get this additive model, where you have an AUC of 0.849 to predict the 19th day mortality in patients with alcoholic hepatitis. We know that patients, and especially late-stage cirrhotic patients, they have increased intestinal permeability, commonly also called gut leakiness, and this is characterized by translocation not only of viable bacteria, but also of fungi products and microbial products, what we call pathogen-associated molecular patterns, and what we believe is that in decompensated cirrhosis, this translocation is increased as compared to the compensated cirrhotic patients. All of these microbial products bind to certain receptors, and they can initiate an inflammatory response, and this inflammatory response contributes to end-organ function in decompensated cirrhotics, including hepatic encephalopathy and kidney dysfunctions. This was very nicely highlighted by older papers, where you have patient cohort with low or normal LPS binding protein, so surrogate marker for bacterial translocation, and if you have a cirrhotic patient group with high bacterial translocation, if you add ciprofloxacin to these patients, you can see that you decrease the inflammatory response in the serum, but also circulating monocytes, suggesting that, in fact, if you eradicate the gut microbiome or at least reduce a gut microbiome, you can reduce inflammatory markers, and this systemic inflammatory response in these patients. A recent paper also sequenced, now, the circulating DNA in patients with alcohol use disorder and alcoholic hepatitis, so just by looking at the circulating gut microbiome in the serum, you can see that you can also discriminate controls from alcohol use disorder patients and alcoholic hepatitis patients, and this genus chantinobacterium was inversely correlated with the MELD score. Another very simple marker for gut microbial activity and metabolism are bile acids. As you know, bile acids are synthesized and secreted from the parasites into the bile ducts, and they reach the intestine, where they can affect, first of all, the gut microbiome, but then also the gut microbiome modulates the primary bile acids into secondary bile acids, and they are being reabsorbed in the portal vein. If you look at systemic, so the serum bile acid profile, again, a very easy readout for microbial activity, you can see that in patients, obviously, with alcoholic hepatitis, which we understand is a cholestatic disease, have very high total bile acid levels, but if you look at the microbial, at the bile acid profile in the serum, alcoholic hepatitis patients, but also, to some degree, patients with alcohol use disorder have a completely different bile acid composition, and we were also able to show that this taurine-conjugated kinodeoxycholic acid also correlates with MELD in these patients with alcoholic hepatitis. So, coming back to our case, what we can probably say is that this patient with advanced liver disease, she has probably an increase in what, quote-unquote, bad bacteria and kind of the overgrowth. Certain microbial metabolites, like bile acids, like LPS, and are associated with clinical outcomes, but again, this is very early stages. Most of these studies have not been really reproduced, and this requires validation in large prospective studies. So, in my second part of my presentation, I want to briefly focus on precision approaches. How can we use a gut microbiome to restore host microbiome homeostasis, and in particular, how can we treat liver disease by focusing on the gut microbiome? And one of the concepts that we have developed over the time is, you know, we can use bugs as drugs. So, we can essentially just analyze what is different between the healthy and deceased patients, and once we have found these changes, we can then use defined mixtures of microorganisms to reprogram the microbial ecology, so that you then get more of this diverse gut microbiome. One of the most primitive and crude methods to do this is fecal microbiota transplantation. This is a very small study coming out from India with NIS8 patients of these are steroid ineligible patients with alcoholic hepatitis. They received a daily FMT for eight days via a nasotitunal tube, and these authors were able to show that the survival improved as compared to historical controls. In another study, where patients were subjected to fecal microbiota transplantation via an enema, and this was a patient cohort, again, very small, NIS10, with decompensated cirrhosis and hepatic encephalopathy, these patients received initially five days of antibiotics, then an FMT, and they were followed and compared to a standard of care group. If you look at the day 20 after FMT, you can see that the hepatic encephalopathy score significantly improved in the FMT group, but not in the standard of care group. I'm sure many of you have seen this recent report from the New England Journal of Medicine, where drug-resistant E. coli have been transmitted by a fecal microbiota transplantation, and actually one patient died from this complication. So I think in the future, we also need to redefine the screening methods for donors for fecal microbiota transplantation. Now, moving towards more precision medicine, what else can we do instead of, you know, wiping out the entire gut microbiome by FMT? What else can we do more precisely? And there is this very nice concept that we can essentially engineer microbes to restore microbial ecology, and we can engineer these microbes to either make good metabolites that are beneficial for the host or, in fact, probably also for the gut microbiome, or we can make bacteria to metabolize one of the bad products. In this case, one of the prominent examples was that, at least in the preclinical literature, that engineered bacteria were made to reduce intestinal ammonia, and this had very good effect in preclinical mouse models. However, this is a press release from the company who made this bacteria, and it looks like that in this randomized double-blind placebo-controlled study with 23 patients with cirrhosis and elevated blood ammonia levels, there was no evidence of blood ammonia lowering or any changes in the primary and secondary outcomes relative to placebo. So, we are waiting for the publication to see why this failure happened. Okay, thank you. Another, again, moving towards more precision medicine, what else can we use to, in this case, not replace or add some bacteria? In this case, how can we selectively eliminate deleterious microbes in our gut microbiome? Again, very crude method is to use antibiotics. We can use bacteriocins. These are small molecule produced by bacteria that affect the gut microbiome, but we can also use bacteriophages to now selectively eliminate these deleterious microbes and to restore this intestinal homeostasis. And I want to show you one example that we have recently undertaken and shown to be very effective in a preclinical mouse model of humanized mice. So we can use bacteriophages. You can see here a cartoon. They have a head and a tail, and this is an electron microscopy image. The tail actually docks onto bacteria, injects the DNA. These bacteriophages replicate inside the bacteria, and then they lyse them, so they destroy the bacteria, and this happens very specifically. So one bacteriophages only recognize certain strains of bacteria. We have used this now in a mouse model of alcoholic hepatitis, where we used germ-free mice. We colonized them with donor stool from this, what I have mentioned before, cytolysin-positive alcoholic hepatitis patients, and then we created bacteriophages against this cytolysin-producing bacteria, and when we gave this, we have two patients here, or mice that have been colonized with donor stool from these two patients. You can see that we can improve liver injury, that we can improve liver steatosis, but most importantly, we can now reduce this toxin in the liver that comes from the microbiome, and this exerts liver disease, and this will be published next Wednesday. So this is my summary slide, so translating, how can we translate microbiome research into clinical practice? I think this, if you focus on the gut microbiome, we should not only look at the composition of the gut microbiome and the metabolome, but we also need to consider barrier integrity, systemic immunity, and host genomics. We need to classify more microbiota-driven subtypes of disease. I'm convinced, and there is good evidence, that not every, for example, NAFLD patient is driven by this microbiome changes in the intestine. We need to validate new targets in host microbiota interactions with the aim of restoring host microbiome homeostasis, and the precision medicine comes now. We can use, and I've shown you examples for this, we can use bugs as drugs, we can use phages as drugs, we can use, and I've not shown you, but there is very good evidence from preclinical mouse models that we can actually use drugs from bugs. So, for example, indole derivatives are very beneficial for the disease, and we can, in fact, also drugs a bug. So if we know that drugs make certain metabolites, we can use chemical inhibitors to interfere with this production and can reduce this deleterious metabolites. However, also, as a hepatologist, I want to emphasize, don't forget to treat the etiology of the liver disease. You know, without your liver disease etiology and without injury, you would not get liver disease just from the microbiome itself. Key takeaway points, liver disease is associated with changes in the gut microbiota and mycobiota, which is the fungi part of the microbiome. Microbial parameters can be used to predict outcomes in phenotype of liver disease, but this requires validation in large prospective studies, as I have mentioned before, and I believe and I'm convinced that intestinal dysbiosis represents a very attractive target for therapy, but this needs more precision. Thank you.

gut microbiome

alcoholic liver disease

Enterococcus faecalis

FMT

bile acids