In chronic liver disease (CLD), liver injury results in progressive fibrosis, culminating in cirrhosis, liver failure, and associated complications. Both the gut microbiota (1–4) and intestinal barrier dysfunction (1) are implicated in the pathogenesis of liver injury. Dysbiosis of the stool microbiota has been reported in CLD (5–9) with a reduction in the abundance of commensal microbiota and a relative increase in potentially pathogenic ones (10). This may in part be due to invasion of the distal gut by “oral type” microbiota (5). The mechanisms by which gut microbial dysbiosis contribute to liver disease in humans are not clear, but it is thought that an increase in intestinal permeability (11,12), leading to the translocation of microbial products, plays a major role.
Although the prevalence of small intestinal bacterial overgrowth is increased in CLD (13), very little is known about the small intestinal mucosal microbiota. Most studies have examined stool microbiota (5–7,14), which are not necessarily representative of small intestinal mucosa-associated microbiota (MAM) due to regional variation in the bacterial populations throughout the gut (15). In animal models, mucosal bacteria are more likely to translocate across the gut wall compared to luminal bacteria (16). The MAM from the proximal small intestine (SI) drains into the portal vein and are well placed to exert an influence on the liver. Emerging data suggest that there may be dysbiosis of the duodenal MAM in cirrhosis (17), although how this relates to the pathogenesis of cirrhosis is unclear. Additionally, the associations between small intestinal dysbiosis, small intestinal permeability, and liver injury in humans have not been examined in detail.
The aims of this study were to characterize the duodenal MAM in patients with CLD in comparison to a well-characterized, disease-free control group, and to examine relationships with both small intestinal permeability and liver disease severity. We hypothesized that CLD would have dysbiosis of the duodenal MAM, which would be associated with increased small intestinal permeability, serum alanine aminotransferase (ALT), and hepatic fibrosis.
MATERIALS AND METHODS
Subjects with CLD and a healthy control group, both who underwent routine upper endoscopy, were included. Indications for endoscopy were iron deficiency, surveillance of noninflamed Barrett's esophagus (both cohorts), or variceal surveillance (CLD subjects only). Subjects were recruited over a 24-month period from a tertiary care liver center (Princess Alexandra Hospital, Brisbane, Australia). CLD was defined as liver injury for greater than 6 months, as determined by clinical history, liver biochemistry, imaging, and/or histology. Etiologies of CLD were hepatitis C virus (HCV) and hepatitis B virus (HBV), nonalcoholic fatty liver disease (NAFLD), alcoholic liver disease (ALD), or autoimmune liver disease. Informed consent was obtained from all participants and ethical clearance was granted by the Metro South Human Research Ethics Committee (HREC/12/QPAH/083) and University of Queensland Medical Research Ethics Committee (Approval number 2013000800) before commencement. The study was performed in accordance with the principles of the Declaration of Helsinki.
Strict exclusion criteria were applied to minimize confounding factors. Exclusion criteria were as follows: A documented history of gastrointestinal disease (other than noninflamed Barrett's esophagus) including, where available, endoscopic evidence of esophagitis, gastritis, duodenitis, peptic ulcer disease, coeliac disease, or inflammatory bowel disease; or any of the following within 6 weeks of the study: symptoms of abdominal pain, diarrhea, vomiting; or use of aspirin, nonsteroidal anti-inflammatory drugs, proton pump inhibitors, bile acids or bile acid sequestrants, oral iron therapy, antibiotics, prebiotics or probiotics; or significant alcohol intake. Specifically, subjects with ALD were not drinking for 6 weeks before the study. For the control group, subjects with the metabolic syndrome (MetS, diagnosed by the IDF/AHA/NHLBI 2009 consensus criteria (18)) were excluded, whereas for the CLD group, subjects with ascites were excluded.
Daily macronutrient intake, anthropometric measurements, assessment of hepatic fibrosis, peripheral endotoxemia, and standard blood markers of liver enzymes, fasting blood glucose, insulin, and lipid profiles were measured as outlined in Supplementary Methods (see Supplementary Digital Content 1, http://links.lww.com/CTG/A68). Small intestinal permeability was assessed using a dual lactulose:rhamnose (L:R) assay, described in detail in Supplementary Methods (see Supplementary Digital Content 1, http://links.lww.com/CTG/A68).
Assessment of duodenal mucosa-associated microbiota
During endoscopy, a 2 × 1 mm specimen of mucosal tissue was collected from the second part of the duodenum opposite the ampulla of Vater using standard biopsy forceps. To minimize oropharyngeal contamination, the endoscope was advanced straight to the duodenum, and mucosal samples were obtained before suction of oropharyngeal or gastric contents. From the duodenal tissue, microbiota profiles were generated for each patient. DNA was extracted from the mucosal biopsies and amplified using barcoded primers targeting the V6–V8 region of the 16S ribosomal ribonucleic acid (rRNA) gene. Resultant libraries were sequenced by MiSeq and bioinformatics analysis performed using Quantitative Insights Into Microbial Ecology. Full details and primers are described in Supplementary Methods (see Supplementary Digital Content 1, http://links.lww.com/CTG/A68).
Normally distributed variables were presented as mean and standard deviation, and analyzed with Student's t test or analysis of variance (with Tukey's multiple comparison). L:R ratios were log transformed to normalize their distribution. Liver stiffness measurements were analyzed as a continuous score. Non-normally distributed variables were presented as median and interquartile range, and analyzed with the Mann-Whitney U, rank test, or Kruskal-Wallis tests; or Spearman's rho for correlations. Statistical significance was defined as P < 0.05. Statistical analyses were performed using IBM Statistical Package for Social Sciences version 22, 2013.
Microbiota data were analyzed using Calypso web-based program for microbial ecology, Calypso v8.2 (cgenome.net/calypso) (19), as outlined in the Supplementary Methods (see Supplementary Digital Content 1, http://links.lww.com/CTG/A68). In brief, alpha diversity was assessed using the Shannon index, and differences in the relative abundance of taxa were analyzed by both differential expression analysis based on the negative binomial distribution and analysis of composition of microbiomes (ANCOM) statistical tests. Analyses were adjusted for multiple comparisons testing. For both differential expression analysis based on the negative binomial distribution and ANCOM analyses, taxa with a P < 0.05 with a false discovery rate q value <0.05 were highlighted as significant. Beta diversity was assessed by principal co-ordinate analysis (PCoA), using the Bray-Curtis distance metric and adonis nonparametric statistic. To assess the overall contribution of taxa to the patient cohorts, linear discriminant analysis effect size and the constrained multivariate mixOmics method, sparse partial least squares discriminant analysis (20) were used. Predicted function of the genome based on the 16S rRNA sequence data was determined using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) (http://picrust.github.io/picrust) (21).
Study population characteristics
A total of 46 subjects (35 with CLD and 11 controls) met inclusion and exclusion criteria and had duodenal MAM profiles available for analysis. When comparing CLD and control subjects, age, gender, body mass index, and dietary intake of major macronutrients (fat, protein, and carbohydrate) were similar (Table 1). As expected, CLD subjects had a lower platelet count and serum albumin concentration. Cirrhosis were present in 23 (65%) CLD subjects, 18 (78%) of whom had Child's A cirrhosis. The MetS (an exclusion criteria for controls) was present in 17 (46%) of CLD subjects. Among the different etiologies of CLD, 15 patients had HCV (all HCV RNA positive), 9 NAFLD, 7 ALD, 3 autoimmune liver disease, and 1 patient had HBV. When comparing the 3 most prevalent etiologies (HCV, NAFLD, and ALD), HCV subjects had higher serum ALT, but other baseline characteristics were similar (Table 1). All except for 2 subjects in the entire cohort were Europids (Australian or European ethnicity).
Small intestinal permeability is higher in CLD
Small intestinal permeability was significantly higher in CLD subjects compared to controls (mean ± SD log [L:R × 100] 1.41 ± 0.37 vs 1.09 ± 0.26, P < 0.05, Table 1). Among CLD subjects, permeability was highest in those with HCV. Endotoxemia was present in 5 subjects, all with CLD. For these subjects, intestinal permeability did not differ when compared to nonendotoxemic CLD subjects (log [L:R × 100] 1.32 ± 0.40 vs 1.42 ± 0.34, P = 0.6), or controls (P = 0.29).
Duodenal MAM composition
Altogether, 109 different operational taxonomic units were identified in the duodenal MAM (at 97% sequence identity threshold) derived from 5 bacterial phyla and differentiated into 9 classes, 11 orders, 20 families, and 25 genera (see Supplementary Table 1, Supplementary Digital Content 2, http://links.lww.com/CTG/A69). The average number of quality reads was 13,204 per CLD subject, and 14,480 per control subject.
Overall, Firmicutes and Bacteroidetes were the most abundant phyla, together accounting for 85% of detected bacterial operational taxonomic units. The most abundant genera were Streptococcus, followed by Prevotella, Veillonella, and Actinomyces (see Supplementary Table 2, Supplementary Digital Content 3, http://links.lww.com/CTG/A70). Rarefaction analysis of the cumulative sum-scaled data showed that sufficient coverage of the bacterial diversity present in all samples was achieved (see Supplementary Figure 1, Supplementary Digital Content 4, http://links.lww.com/CTG/A71).
Both composition and diversity of the duodenal MAM differ between CLD and controls
The duodenal bacterial community in CLD differed to that of healthy controls on PCoA (see Supplementary Figure 2, Supplementary Digital Content 5, http://links.lww.com/CTG/A72). Using both a constrained multivariate model (Figure 1a) and linear discriminant analysis effect size (Figure 1b), it was possible to identify those taxa that were most discriminatory of CLD and control subjects. Both these analyses indicated that the MAM of CLD subjects was characterized by Streptococcus-affiliated lineages, whereas the control subjects were shaped by smaller contributions from a variety of taxa (Neisseria, Haemophilus, Porphyromonas, Veillonella, Moryella, and Prevotella spp.). The duodenal MAM of CLD subjects also exhibited less alpha diversity than controls (Shannon index, Figure 1c).
Significant differences in the relative abundance of particular taxa were observed between CLD subjects and controls (Figure 2). At the phylum level, CLD subjects had a greater abundance of Firmicutes and lower abundance of Actinobacteria (Figure 2a and see Supplementary Table 2A, Supplementary Digital Content 3, http://links.lww.com/CTG/A70). Decreases in the abundance of the genera Moryella, Porphyromonas, and Veillonella (Figure 2b, and Supplementary Table 2B, Supplementary Digital Content 3, http://links.lww.com/CTG/A70) and Actinomyces-affiliated taxa (see Supplementary Table 2C, Supplementary Digital Content 3, http://links.lww.com/CTG/A70) were observed in CLD subjects. These differences were also validated by the ANCOM statistical test, which showed the CLD subjects had a reduction in the relative abundance of Moryella, Porphyromonas, and Actinomyces (see Supplementary Tables 2D and E, Supplementary Digital Content 3, http://links.lww.com/CTG/A70), and some Veillonella-affiliated taxa (see Supplementary Table 2F, Supplementary Digital Content 3, http://links.lww.com/CTG/A70).
Microbial composition or diversity did not differ by sex when examined across all participants (P = 0.5 and P = 0.4 respectively, see Supplementary Figure 3, Supplementary Digital Content 6, http://links.lww.com/CTG/A73), or when this was confined to CLD subjects (P = 0.3 and P = 0.5 respectively, data not shown).
Functional properties of the microbiota distinguish CLD and control subjects
PICRUSt analysis, to infer functional properties of the microbiota, indicated that the microbial communities of the CLD subjects possessed a greater abundance of pathways related to sugar and carbohydrate metabolism, glycan degradation, and bacterial toxin production. Control subjects had a greater abundance of lipopolysaccharide biosynthesis and amino acid-related pathways (see Supplementary Table 3, Supplementary Digital Content 7, http://links.lww.com/CTG/A74, and Supplementary Figure 4, Supplementary Digital Content 8, http://links.lww.com/CTG/A75).
Lower microbial diversity correlates with increased small intestinal permeability and serum ALT in CLD
Among the CLD subjects, microbial diversity inversely correlated with both small intestinal permeability (r = −0.41, P = 0.02, Figure 3a), and with serum ALT (r = −0.35, P = 0.04, Figure 3b). Among those genera that had lower abundance in CLD, the abundance of Porphyromonas inversely correlated with increased small intestinal permeability (see Supplementary Table 4, Supplementary Digital Content 9, http://links.lww.com/CTG/A76).
There were no associations observed between the duodenal MAM and hepatic stiffness for either microbial diversity (r = 0.16, P = 0.4, see Supplementary Figure 5, Supplementary Digital Content 10, http://links.lww.com/CTG/A77) or the relative abundance of any taxa (see Supplementary Table 5, Supplementary Digital Content 11, http://links.lww.com/CTG/A78). There were no overall differences in community composition between CLD subjects with and without cirrhosis (see Supplementary Figure 6A, Supplementary Digital Content 12, http://links.lww.com/CTG/A79), including those with compensated and decompensated cirrhosis (see Supplementary Figure 6B, Supplementary Digital Content 12, http://links.lww.com/CTG/A79). In addition, the duodenal MAM was not significantly different between CLD subjects with and without the MetS (see Supplementary Figure 7, Supplementary Digital Content 13, http://links.lww.com/CTG/A80). We were unable to identify any associations between dietary macronutrient intake and abundance of any microbial taxa (P > 0.05 or false discovery rate q value >0.1, data not shown).
Etiologies of CLD
Among the 3 main etiologies of CLD, the alpha diversity for both the HCV and NAFLD subjects was significantly less than the control group (P < 0.01 and P < 0.05 respectively, see Supplementary Figure 8A, Supplementary Digital Content 14, http://links.lww.com/CTG/A81). Notably, the duodenal MAM of the HCV group was the least diverse, whereas this group also had the highest intestinal permeability and serum ALT. On PCoA, the microbiota profiles of HCV and NAFLD, but not ALD, were different to the control subjects (P < 0.01 and P < 0.05, respectively; see Supplementary Figure 8B, Supplementary Digital Content 14, http://links.lww.com/CTG/A81); however, CLD etiologies did not differ to each other (P = 0.4). HCV RNA viral loads were not associated with taxa abundance (see Supplementary Table 6, Supplementary Digital Content 15, http://links.lww.com/CTG/A82), microbial diversity (see Supplementary Figure 9A, Supplementary Digital Content 16, http://links.lww.com/CTG/A83), or small intestinal permeability (see Supplementary Figure 9B, Supplementary Digital Content 16, http://links.lww.com/CTG/A83).
To date, most of the evidence for intestinal microbial dysbiosis in CLD is based on the stool samples in cirrhotic patients. We have demonstrated that the duodenal mucosal microbiota of CLD subjects is not only different to controls, but this dysbiosis is associated with increased small intestinal permeability. The CLD MAM was less diverse overall, and specifically had a reduction in the abundance of Moryella, Porphyromonas, and Veillonella and Actinomyces taxa. Certain Streptococcus-affiliated lineages were increased in CLD, and these largely accounted for the differences in separation between the duodenal MAM communities of CLD and control subjects. In CLD, lower microbial diversity correlated with both increased small intestinal permeability, and serum ALT, suggesting an association with gut barrier dysfunction and hepatic inflammation.
Previous studies demonstrate that the stool, distal colonic mucosal, and salivary microbiomes of cirrhotic subjects have different profiles (10,22), but also share some common disease-associate signatures—primarily overabundance of potentially pathogenic endotoxin-producing Enterobacteriaceae (7,10,22), Enterococcaceae (10,22), and Streptococcaceae (5,7) as well as lower abundance of the autochthonous taxa Ruminococcaceae and Lachnospiraceae (7,10,22). Streptococcaceae in particular are a key component of the oral microbiome (10), and Qin et al. (5) also reported an overrepresentation of Streptococci in the stool of cirrhotic subjects compared to healthy controls, inferring that oral commensals appear to “invade” the gut in cirrhosis. Our observation that Streptococcal-affiliated lineages were largely contributory to the distinct CLD MAM supports this notion. Qin et al. (5) also showed an overall reduction in microbial gene richness of stool microbiota in cirrhotic subjects, in line with our findings of overall decreased diversity and abundance of Moryella, Porphyromonas, and Veillonella and Actinomyces taxa in CLD.
Duodenal dysbiosis in CLD could occur as a result of disturbances in the small intestinal niche caused by alterations in bile acid flow (23) and motility (24). The number of human studies in CLD that focus specifically on the duodenal MAM is very limited. Chen et al. (17) observed dysbiosis in a study of duodenal MAM from cirrhotic subjects with chronic HBV and primary biliary cholangitis. Taxonomic differences in the CLD subjects between their study and ours are likely explained by demographic differences between the study populations, and variations based on the 16S rRNA gene variable region selected for amplification. In our study, HCV subjects had the most pronounced dysbiosis. This may be linked to overall disease severity, with the higher ALT levels implying a greater degree of hepatic inflammation. Although the virus itself may influence B and T lymphocyte function (25,26), there was no association between HCV viral load and duodenal MAM or permeability.
Increased permeability of the SI renders the liver susceptible to the effects of the intestinal microbiota or their products. We demonstrated that reduced microbial diversity was associated with increased small intestinal permeability and hepatic inflammation (based on the serum ALT). Collectively, these data suggest that enriched microbial diversity in the SI may be a protective factor, and loss of this may compromise intestinal homeostasis and integrity. Notably, endotoxemia, which is a recognized feature of advanced liver disease, was not associated with small intestinal permeability. This may be explained by the low overall abundance of Gram-negative bacilli in the duodenal MAM. It is likely that a variety of microbial metabolites in addition to endotoxin may be important in disease processes in CLD (27) and that endotoxemia is not necessarily driven by the small intestinal microbiota.
The duodenal MAM of CLD subjects displayed a greater abundance of pathways related to carbohydrate metabolism and glycan degradation. Dietary carbohydrate intake was similar between CLD and controls, implicating an alternative carbohydrate source such as gut mucins is being metabolized by these bacteria. Primary biliary cholangitis spp., which shaped the CLD MAM, are highly functioning in the SI and are dependent on their capacity to utilize carbohydrates, for which they compete with the host (28). Gut mucins are a key factor in maintaining intestinal barrier function (29), and primary biliary cholangitis spp. produce mucin-degrading enzymes (30). Although we did not find a correlation between the abundance of primary biliary cholangitis and small intestinal permeability, more robust metagenomics and metabolomics analyses are required to interrogate this relationship more fully.
Although there was a clear association between microbial diversity and serum ALT, there was no association between liver stiffness and the microbiota. This may be a limitation of the small sample size, particularly among noncirrhotic subjects, and the heterogeneity of the population. As our primary aim was to evaluate the microbiota in CLD compared to non-CLD subjects, there may have been insufficient power for detailed analysis by grouped stages of hepatic fibrosis, and disease etiology. Hepatic fibrosis and liver disease severity are also a product of disease duration, and longitudinal studies are needed to interrogate this more robustly. However, the novelty of this study is the analysis of microbiota derived from a very limited resource, namely duodenal tissue, from subjects who have no other duodenal pathology, in relation to intestinal permeability. To our knowledge, his has not been done before.
In summary, this study demonstrates that CLD has dysbiosis of the duodenal mucosal microbiota, characterized by a distinct community clustering and reduced bacterial diversity. Of significance, reduced duodenal bacterial diversity is associated with increased small intestinal permeability, and serum ALT. These findings prompt for further mechanistic and longitudinal studies, rather than cross-sectional studies, which would enable the effects of differences in disease pathogenesis on duodenal MAM to be interrogated further.
CONFLICTS OF INTEREST
Guarantor of the article: Ashok Sridar Raj, MBChB, PhD, FRACP.
Specific author contributions: A.S.R.: Study concept and design, patient recruitment, data acquisition and analysis, writing and editing of manuscript. G.A.M.: Study concept and design, data analysis, writing and editing of manuscript. E.R.S.: Methodology, bioinformatics analysis, data analysis, writing and editing manuscript. C.D.T.: Methodology (high-performance liquid chromatography), data analysis, review of manuscript. P.B.: Supervision of methodology, review of manuscript. D.A.V.: Methodology (endotoxin assay), review of manuscript. L.M.F.: Supervision of methodology, review of manuscript. M.M.: Supervision of methodology, data analysis, editing of manuscript. G.J.H.: Study concept, review of manuscript.
Financial support: This project was supported in part by the Princess Alexandra Hospital Research Foundation Small Project Grant and A.S.R. was supported by an Australian Postgraduate Award. Special thanks to Ms Marrianne Black (Princess Alexandra Hospital) for performing transient elastography, and Dr Justin Scott and Dr Anne Bernard (QFAB Biostatistics) for statistical support. Sequencing was performed at the Australian Centre for Ecogenomics, the University of Queensland.
Potential competing interests: None.
WHAT IS KNOWN
- ✓ CLD is associated with dysbiosis of the stool microbiota.
- ✓ Little is known about the small intestinal mucosal microbiota of CLD and how it relates to intestinal permeability and disease severity.
WHAT IS NEW HERE
- ✓ The duodenal mucosa-associated microbiota of CLD differs from that of healthy controls, with distinct community clustering and lower diversity.
- ✓ In CLD, lower microbial diversity correlates with increased small intestinal permeability and serum ALT.
- ✓ Strategies to ameliorate duodenal MAM dysbiosis could improve intestinal barrier dysfunction and liver injury in CLD.
- ✓ The duodenal MAM warrants further attention as a potential therapeutic target to restore intestinal barrier dysfunction in CLD.
1. Gäbele E, Dostert K, Hofmann C, et al. DSS induced colitis increases portal LPS levels and enhances hepatic inflammation and fibrogenesis in experimental NASH. J Hepatol 2011;55:1391–9.
2. Seki E, De Minicis S, Osterreicher CH, et al. TLR4 enhances TGF-beta signaling and hepatic fibrosis. Nat Med 2007;13:1324–32.
3. Le Roy T, Llopis M, Lepage P, et al. Intestinal microbiota determines development of non-alcoholic fatty liver disease in mice. Gut 2013;62:1787–94.
4. Chen P, Torralba M, Tan J, et al. Supplementation of saturated long-chain fatty acids maintains intestinal eubiosis and reduces ethanol-induced liver injury in mice. Gastroenterology 2015;148:203–14.e16.
5. Qin N, Yang F, Li A, et al. Alterations of the human gut microbiome in liver cirrhosis. Nature 2014;513:59–64.
6. Mouzaki M, Comelli EM, Arendt BM, et al. Intestinal microbiota in patients with nonalcoholic fatty liver disease. Hepatology 2013;58:120–7.
7. Chen Y, Yang F, Lu H, et al. Characterization of fecal microbial communities in patients with liver cirrhosis. Hepatology 2011;54:562–72.
8. Boursier J, Mueller O, Barret M, et al. The severity of nonalcoholic fatty liver disease is associated with gut dysbiosis and shift in the metabolic function of the gut microbiota. Hepatology 2016;63:764–75.
9. Bajaj JS, Sterling RK, Betrapally NS, et al. HCV eradication does not impact gut dysbiosis or systemic inflammation in cirrhotic patients. Aliment Pharmacol Ther 2016;44:638–43.
10. Bajaj JS, Betrapally NS, Hylemon PB, et al. Salivary microbiota reflects changes in gut microbiota in cirrhosis with hepatic encephalopathy. Hepatology 2015;62:1260–71.
11. Pijls KE, Jonkers DM, Elamin EE, et al. Intestinal epithelial barrier function in liver cirrhosis: An extensive review of the literature. Liver Int 2013;33:1457–69.
12. Miele L, Valenza V, La Torre G, et al. Increased intestinal permeability and tight junction alterations in nonalcoholic fatty liver disease. Hepatology 2009;49:1877–87.
13. Shah A, Shanahan E, Macdonald GA, et al. Systematic review and meta-analysis: Prevalence of small intestinal bacterial overgrowth in chronic liver disease. Semin Liver Dis 2017;37:388–400.
14. Bajaj JS, Heuman DM, Hylemon PB, et al. Altered profile of human gut microbiome is associated with cirrhosis and its complications. J Hepatol 2014;60:940–7.
15. Aguirre de Cárcer D, Cuív PO, Wang T, et al. Numerical ecology validates a biogeographical distribution and gender-based effect on mucosa-associated bacteria along the human colon. ISME J 2011;5:801–9.
16. Wang L, Fouts DE, Starkel P, et al. Intestinal REG3 lectins protect against alcoholic steatohepatitis by reducing mucosa-associated microbiota and preventing bacterial translocation. Cell Host Microbe 2016;19:227–39.
17. Chen Y, Ji F, Guo J, et al. Dysbiosis of small intestinal microbiota in liver cirrhosis and its association with etiology. Sci Rep 2016;6:34055.
18. Alberti KG, Eckel RH, Grundy SM, et al. Harmonizing the metabolic syndrome: A joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 2009;120:1640–5.
19. Zakrzewski M, Proietti C, Ellis JJ, et al. Calypso: A user-friendly web-server for mining and visualizing microbiome-environment interactions. Bioinformatics 2017;33:782–3.
20. Lê Cao KA, Costello ME, Lakis VA, et al. MixMC: A multivariate statistical framework to gain insight into microbial communities. PLoS One 2016;11:e0160169.
21. Langille MG, Zaneveld J, Caporaso JG, et al. Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat Biotechnol 2013;31:814–21.
22. Bajaj JS, Hylemon PB, Ridlon JM, et al. Colonic mucosal microbiome differs from stool microbiome in cirrhosis and hepatic encephalopathy and is linked to cognition and inflammation. Am J Physiol Gastrointest Liver Physiol 2012;303:G675–85.
23. Kakiyama G, Pandak WM, Gillevet PM, et al. Modulation of the fecal bile acid profile by gut microbiota in cirrhosis. J Hepatol 2013;58:949–55.
24. Chang CS, Chen GH, Lien HC, et al. Small intestine dysmotility and bacterial overgrowth in cirrhotic patients with spontaneous bacterial peritonitis. Hepatology 1998;28:1187–90.
25. Yao ZQ, Prayther D, Trabue C, et al. Differential regulation of SOCS-1 signalling in B and T lymphocytes by hepatitis C virus core protein. Immunology 2008;125:197–207.
26. Douam F, Bobay LM, Maurin G, et al. Specialization of hepatitis C virus envelope glycoproteins for B lymphocytes in chronically Infected patients. J Virol 2015;90:992–1008.
27. Krautkramer KA, Dhillon RS, Denu JM, et al. Metabolic programming of the epigenome: Host and gut microbial metabolite interactions with host chromatin. Transl Res 2017;189:30–50.
28. Zoetendal EG, Raes J, van den Bogert B, et al. The human small intestinal microbiota is driven by rapid uptake and conversion of simple carbohydrates. ISME J 2012;6:1415–26.
29. McGuckin MA, Lindén SK, Sutton P, et al. Mucin dynamics and enteric pathogens. Nat Rev Microbiol 2011;9:265–78.
30. Derrien M, van Passel MWJ, van de Bovenkamp JHB, et al. Mucin-bacterial interactions in the human oral cavity and digestive tract. Gut Microbes 2010;1:254–68.