The gut microbial influence on cholestatic liver disease

Patients with cholestatic liver diseases like primary sclerosing cholangitis (PSC) and primary biliary cholangitis (PBC) have a different gut microbiome composition than healthy controls. In contrast with PBC, PSC has a strong association with inflammatory bowel disease and is the prototypical disease of the gut‐liver axis. Still, there are some distinct overlapping microbial features in the microbiome of patients with PSC and PBC suggesting similarities in cholestatic diseases, although the possible pathogenetic involvement of these shared microbial changes is unknown. Herein, we present an overview of the available data and discuss the relevance for potential disease relevant host‐microbiota interactions. In general, the microbiome interacts with the host via the immunobiome (interactions between the host immune system and the gut microbiome), the endobiome (where the gut microbiome contributes to host physiology by producing or metabolizing endogenous molecules) and the xenobiome (gut microbial transformation of exogenous compounds, including nutrients and drugs). Experimental and human observational evidence suggest that the presence and functions of gut microbes are relevant for the severity and progression of cholestatic liver disease. Interestingly, the majority of new drugs that are currently being tested in PBC and PSC in clinical trials act on bile acid homeostasis, where the endobiome is important. In the future, it will be paramount to perform longitudinal studies, through which we can identify new intervention targets, biomarkers or treatment‐stratifiers. In this way, gut microbiome‐based clinical care and therapy may become relevant in cholestatic liver disease within the foreseeable future.

These studies have been reviewed extensively in recent publications. 6,17,18 Overall, the main and undisputed observation is that the microbiome is different in patients with PSC compared with patients with IBD without liver disease and healthy controls.
Considering the studies of the mucosal microbiome in PSC, these may be difficult to interpret. The study-size in general is quite small (n min-max: [11][12][13][14][15][16][17][18][19][20] and few results overlap between studies, which could in part relate to differences in the control groups used. 18 The studies of the faecal microbiome in adult PSC patients published in full-length, peer-reviewed articles that have included healthy controls (Table 1)  and Veillonella. In addition, all but one study reported an enrichment of Enterococcus, which also correlates with serum levels of alkaline phosphatase (ALP), a clinical marker for cholestasis. 12 Finally, (d) IBD-status has a negligible effect on the composition of the microbiome in patients with PSC, in contrast with what is seen in patients with IBD without liver disease.
More recently, the first data on the salivary microbiome in patients with PSC was reported, 19 based on a cohort of newly diagnosed, paediatric PSC patients from Japan, where faecal microbiota composition was in line with the description above. 16 The salivary microbiome was characterized by changes in bacterial taxa that mirror the findings from faecal microbiota, with Haemophilus spp, Veillonella spp and Streptococcus spp being different between paediatric PSC patients and age-matched controls. Another novel data type, faecal bile acid profile, was recently reported in 15 adult PSC patients with IBD, compared to 15 IBD patients without PSC from Portugal, 20 and a smaller pilot-study from the United States. 21 There was limited overlap with the conclusions listed above and also between the two studies, and it thus seems important to increase sample-size to reach firm conclusions, especially regarding correlations between microbiota and bile acids. Still, taken together, thes studies TA B L E 1 Studies of the faecal gut microbiome in primary sclerosing cholangitis (PSC) published in full-length, peer-reviewed articles Kummen et al 13 Sabino et al 12   Less is known about the gut microbiome in patients with PBC than in PSC, but a few studies are worth reviewing. Similar to the studies of PSC patients, there are methodological differences, but some overlapping results ( which also showed that several genera depleted in the gut of PBC patients were negatively associated with markers of liver injury and inflammation. 25 In addition, Veillonella has also been reported to be increased in the salivary microbiome of both PBC (n = 39) and autoimmune hepatitis (AIH, n = 17) patients, and correlated positively with IL-1β, IL-8 and immunoglobulin A. 26 Overall, it is probably reasonable to conclude that there are major alterations of the gut microbiome of patients with cholestatic liver diseases. Patients with PSC show a broad depletion of bacterial diversity, which is also evident but perhaps less pronounced in PBC, but both conditions show enrichment of specific taxa, e.g. Streptococcus, Haemophilus and Veillonella. The latter, and other described changes, could be related to development of cirrhosis, but the increased abundance of Veillonella in e.g.
paediatric PSC patients and non-cirrhotic PBC patients suggests otherwise. Also, Veillonella has been associated with other noncirrhotic inflammatory and fibrotic diseases e.g., idiopathic pulmonary fibrosis, systemic sclerosis, rheumatoid arthritis, pulmonary cystic fibrosis and Crohn's disease with ileal involvement, which is associated with fibrotic stenosis in these patients. 13 How the described alterations relate to disease progression, including progression of liver fibrosis and cirrhosis, and disease prognosis, is so far not known, but the observation that disease-associated microbes change during treatment in PBC is of great interest. Given these broad range of microbiome alterations, the critical follow-up questions are then (a) Are the changes driven by the gut or liver or both? (b) Are the changes relevant for disease activity and severity? and c) what could the mechanisms be?

| The three biomes of microbiomehost interaction: The immunobiome, endobiome and xenobiome
In what ways could the gut microbes affect cholestatic liver disease?
Didactically, we have recently presented a conceptual framework, suggesting to categorize the disease-related host-microbiome interactions into three main "biomes", 17 as illustrated in Figure 1: digoxin and 5-Aminosalicylic acid). 2,27 In the following, we apply these categories on the available data.  to be important. 28 Alterations in bile acid homeostasis due to lack of microbes and the endobiome production of secondary bile acids was suggested to be driving the disease in this study. Interestingly, UDCA (a secondary bile acid) abrogated senescence in vitro in this study. 28 However, the benefit from UDCA in PSC is still questionable, mainly due to the lack of adequately powered clinical trials, and so far, few studies have been unable to identify differences in the microbiome between PSC patients that use UDCA and those who do not. 17 The NOD.c3c4 mice develop immune-driven cholangitis affecting both the intra-and extrahepatic biliary tree. 29 The condition has similarities to both PBC (develops antimitochondrial antibodies) and human PSC (distribution of inflammation and presence of biliary dilatation). 29 Interestingly, when NOD.c3c4 mice are raised in a germ-free environment, they show an ameliorated liver phenotype, with less extra-hepatic bile duct dilatation, less biliary inflammation and less CD3-positive cell infiltrates around the intra-hepatic bile ducts, with a similar trend seen in conventionally raised mice treated with antibiotics. 29 This indicates that the gut microbiome in the normal setting contributes to a worsening of disease, in contrast with the observation in Mdr2 -/mice, 28,29 suggesting that the gut microbiome acts differently depending on disease mechanisms. Whether the attenuation of disease in germfree NOD.c3c4 mice is caused by a general hypomorphic immune system or more specific effects of the immunobiome is therefore not known.

| The immunobiome
It is reasonable to suggest that multiple disease mechanisms likely operate at the same time. injury, which could be ameliorated with antibiotics. 30 They also isolated Klebsiella pneumoniae, Proteus mirabilis and Enterococcus gallinarum from mesenteric lymph nodes in these mice using bacterial culture, and showed that these were associated with bacterial translocation and epithelial damage. 30 If reproduced, this is a The frequency of UDCA-usage is not stated, but mean/sd. of "grams UDCA/year" is given. ing, resembling human PSC. 32,33 Interestingly, this effect was only observable in certain rat strains, and was reversed by antibiotics, but not with UDCA-or prednisone-treatment. 32 The phenotype was driven by translocation of bacterial wall peptidoglycans, in line with a role of the immunobiome and pro-inflammatory microbial peptides ( Figure 1). IBD may also induce biliary disease in vulnerable hosts.
Cystic fibrosis is known to associate with biliary disease. In mice lacking a functional cystic fibrosis gene (CFTR), which alters bile composition, induction of colitis has been shown to cause biliary inflammation that could be attenuated using antibiotics, likely caused by pro-inflammatory lipopolysaccharide (LPS). 34 Interestingly, the cholangiopathy observed in these mice can be attenuated by diet. 35 These are intriguing observations, but the circulating levels of LPS in PSC patients was not elevated compared with controls. 4

| The xenobiome
Diet is one of the exogenous factors that has the strongest impact on the gut microbiome composition and function, and the effect of carbohydrates are probably best understood among the macronutrients. 41 The collective bacterial genome encode several hundred-fold more carbohydrate-degrading enzymes than the human genome, and this xenobiome allows the use of carbohydrates that are indigestible to humans as an energy source. 41

| Clinical opportunities and perspectives for the future
Some of the first treatment trials in PSC utilized antibiotics, with the first case series emerging in the late 1950s. 6,7,46 Both vancomycin and metronidazole have later been used in subsequent treatment trials in PSC with effect on e.g. ALP and gamma-glutamyl transferase (GGT) levels, but so far without effect on harder end-points like liver transplantation-free survival. 6,7 Vancomycin is still considered promising, 46 and more trials will be performed in PSC before and after transplantation with this drug. Similar studies in PBC are scarce, except that the antibiotic rifampicin could be used to alleviate pruritus in PBC, although the exact mechanisms involved here are not clear. Changing the microbiome with more specific pro-or pre-biotics than what is used today could also be a possibility. Current bacterial strains in use are of limited value, and in line with this, no effect has been observed F I G U R E 1 The three-biomes of microbiome-host interaction: The immunobiome, endobiome and xenobiome. Experimental and human observational data suggest that a diverse set of microbial functions may be relevant for microbiome-host interaction, including endogenous molecules produced by the microbiota (the endobiome), bacterial processing of pharmacological agents or dietary compounds (the xenobiome) and specific bacterial molecules or metabolites driving the immune process (the immunobiome). ASA, aminosalicylic acid; HLA, human leukocyte antigen; Ig, immunoglobulin; MAIT, mucosal-associated invariant T; MR1, MHC-related protein 1; NKT; natural killer T; SCFA, short-chain fatty acid; TMA, trimethylamine; TMAO, trimethylamine-N-oxide in small trials in PSC. 47 Another possibility is to perform interventional proof-of-concept trials targeting the microbiome. One example of this is FMT, which has attracted interest also in PSC. 48  They reported an increase in bacterial diversity, and the data also suggested that engraftment correlated with improvement in ALP levels, and importantly, they observed no adverse events. If found effective, such studies may be used to identify specifically which microbes or microbial metabolites that are involved in the disease process, or help us to stratify which FMT-donor is best suited for which recipient. The latter is maybe the most important among the methodological hurdles that need to be addressed in FMT-treatment.
As we enter the era of personalized medicine it will be important to consider the role of microbiota composition in determining individual efficacy and safety of several drugs. Interestingly, the majority of new drugs currently in clinical testing in both PBC and PSC that show promising results act on bile acid homeostasis, 48 where the endobiome is important; including pharmacological compounds affecting the farnesoid X-receptor (FXR) e.g. obeticholic acid (OCA) and all-trans retinoic acid (ATRA), fibroblast growth factor 19 (FGF19) analogs, nor-ursodeoxycholic acid (nor-UDCA), activators of the peroxisome proliferator-activated receptor (PPAR), e.g. fibrates like bezafibrate and fenofibrate, and selective α-and/or δ-receptor isoform activators e.g. seladelpar. 48 How these different agents affect the microbiome, or vice versa (involving the xenobiome), is not wellestablished. 2 One important example illustrating basic principles is metformin, a common drug that has been used for a long time in diabetes type 2. Metformin has recently in been shown to act through the endobiome (also here the bile acid-FXR axis is involved). 50 There is also a potential role of the microbiome in the clinic as biomarker of disease, disease severity and activity, and as predictor of treatment efficacy. Longitudinal studies of the microbiome and the related-biomes are needed if we are to understand how they relate to disease progression. Microbial profiles could help in diagnosing biliary disease or relevant subgroups or complications in a biliary disease (e.g. cholangiocarcinoma in PSC). 13,14,17 Finally, as shown by a series of publications from oncology, the microbiome can be used to identify patients responding to treatment, 51 or who are at risk of severe side effects. 51

| CON CLUS IONS
The gut microbiome and the liver are part of an integrated metabolic machinery, which is altered in patients with cholestatic diseases like PSC and PBC, but how and in what way microbes and related molecules are involved in disease initiation and/or progression of disease is so far not well-established. Experimental and human observational data suggest that a diverse set of microbial functions may be relevant, including endogenous molecules produced by the microbiota (endobiome), bacterial processing of pharmacological agents or dietary compounds (xenobiome) and specific bacterial molecules or metabolites driving the immune process (immunobiome). A better understanding of the host-microbial interactions in cholestatic diseases may therefore greatly improve clinical care in these conditions.

CO N FLI C T O F I NTE R E S T
There was no conflict of interest.