Distinct clusters of bacterial and fungal microbiota in end-stage liver cirrhosis correlate with antibiotic treatment, intestinal barrier impairment, and systemic inflammation

Abstract

Decompensated liver cirrhosis (dLC) is associated with intestinal dysbiosis, however, underlying reasons and clinical consequences remain largely unexplored. We investigated bacterial and fungal microbiota, their relation with gut barrier integrity, inflammation, and cirrhosis-specific complications in dLC-patients. Competing-risk analyses were performed to investigate clinical outcomes within 90 days. Samples were prospectively collected from 95 dLC-patients between 2017 and 2022. Quantitative metagenomic analyses clustered patients into three groups (G1-G3) showing distinct microbial patterns. G1 (n = 39) displayed lowest diversity and highest Enterococcus abundance, G2 (n = 24) was dominated by Bifidobacteria, G3 (n = 29) was most diverse and clustered most closely with healthy controls (HC). Of note, bacterial concentrations were significantly lower in cirrhosis compared with HC, especially for G1 that also showed the lowest capacity to produce short chain fatty acids and secondary bile acids. Consequently, fungal overgrowth, dominated by Candida spp. (51.63%), was observed in G1. Moreover, G1-patients most frequently received antibiotics (n = 33; 86.8%), had highest plasma-levels of Zonulin (p = 0.044) and a proinflammatory cytokine profile along with numerically higher incidences of subsequent infections (p = 0.09). In conclusion, distinct bacterial clusters were observed at qualitative and quantitative levels and correlated with fungal abundances. Antibiotic treatment significantly contributed to dysbiosis, which translated into intestinal barrier impairment and systemic inflammation.

Keywords: Intestinal microbiome; cirrhosis-related complications; dysbiosis; fungi; gut barrier; leaky gut; liver cirrhosis; metagenomics; systemic inflammation.

Graphical abstract

Figure 1.

Bacterial communities of patients and healthy controls (HC). Panel (a) Shows hierarchical clustering result along with composition on the genus level for all samples yielding three distinct patient groups (G1–G3) and a separate HC group. Below, bacterial cell concentrations measured by flow cytometry (bacterial concentration; khaki color indicates no data available) as well as relative abundances of pathways for the synthesis of butyrate, propionate (two pathways) and secondary bile acids are given. In panel (b) Network of bacterial species based on correlation analyses is given, where node size refers to mean abundance of each species and colors of edges represent positive (blue) and negative (red) correlations. Species most abundant in a specific group are color coded. Panel (c) Shows phylogenetic trees for genomes of Enterococcus and Bifidobacteria assembled from metagenomes of patient samples along with their basic quality parameters completeness and contamination. Group origins of genomes are indicated by color and presence of genes encoding vancomycin resistance (vanABC) in respective samples is given; The violet triangle signifies vanC detection on genomes.

Figure 2.

Fungal communities of patients and healthy controls (HC). Results in panel (a) are displayed according to groupings based on hierarchical clustering result of bacterial communities that are shown in Figure 1. Three different methods were applied based on (i) Amplification of ITS region, (ii) Custom metagenomic analysis and (iii) MetaPhlan4. Fungal concentrations measured by reads associated with fungi derived from the individual methods are given as black bars, whereas composition on the genus level is given in the three barplots below (same order as above). In panel (b) A network of fungal species and associated bacterial species based on correlation analyses is given, where node size refers to mean abundance of each species and colors of edges represent positive (blue) and negative (red) correlations. Bacterial species primarily associated with a specific group are color coded (compare Figure 1).

Figure 3.

Zonulin levels were higher in patients than in HC (p = 0.023) (a) Among cirrhosis patients, G1 displayed significantly higher zonulin concentrations than the other patients (p = 0.044) (b) Concentrations of sCD163 in cirrhosis patients exceeded those of HC (p < 0.001) (c), with highest values in G1 compared to the other patients (p = 0.019) (d).

Figure 4.

Concentrations of cytokines that differed between groups based on hierarchical clustering on microbiota composition (a). Correlation analyses between cytokines (n = 48) and main microbiota components displayed significant correlations between eleven cytokines (light-green) and Enterococcus abundance (b). Colors of edges represent positive (blue) and negative (red) correlations.

Figure 5.

Competing risk analyses of liver transplantation (LTx)-free survival (a), Infections (b), Fungal infections (c), Overt hepatic encephalopathy (d) (oHE) and acute-on-chronic liver failure (ACLF) (e). Group 1 (G1) was treated as reference group.

Figure 6.

Proposal of an integrated model linking observed results of medication, gut microbiota, markers of intestinal barrier and inflammation and clinical characteristics in cirrhosis patients. Colors of lines represent positive (blue) and negative (red) correlations.