Introduction

Cirrhosis is an important factor in increasing mortality and disability1,2. Recently, the significant role of disturbances in the composition of the gut microbiota (gut dysbiosis) in the progression of cirrhosis has been shown. Cirrhosis is associated with an increase in the gut abundance of harmful facultative anaerobes (Streptococcaceae, Enterobacteriaceae, Enterococcaceae, Bacilli, Proteobacteria), capable of penetrating into intestinal wall, ascitic fluid, liver and systemic circulation (bacterial translocation), and a decrease in the abundance of beneficial bacteria (Faecalibacterium, Ruminococcaceae, Clostridia) that form short-chain fatty acids strengthening the intestinal barrier. These adverse changes increase liver damage and portal hypertension, worsening the short- and long-term prognosis of cirrhosis. This is called the gut-liver axis3,4,5,6,7. However, the mechanism of development of gut dysbiosis in cirrhosis has not yet been established. The main manifestations of cirrhosis are represented by two syndromes, namely portal hypertension and decreased liver function. It was hypothesized that it is portal hypertension, which can impair the motor and protective function of the intestine, that leads to the development of cirrhosis-specific dysbiosis3,4,5,6,7. To test this hypothesis, we compared the gut microbiota composition between cirrhosis patients and patients with non-cirrhotic portal hypertension due to non-cirrhotic portal vein thrombosis (NCPVT) and preserved liver function. In addition to cirrhosis, portal vein thrombosis (PVT) can be a cause of portal hypertension. If thrombotic masses are not lysed, but are organized with or without the development of cavernous transformation of the portal vein, this is chronic extrahepatic portal vein obstruction (EHPVO). These patients have portal hypertension, which is manifested by the development of ascites, shunt hepatic encephalopathy and esophageal varices, but usually there are no liver dysfunctions, namely, hypoalbuminemia, hyperbilirubinemia and others. They formed the NCPVT group. PVT can also complicate the course of cirrhosis8,9,10,11,12. Therefore, the gut microbiota of patients with cirrhosis without PVT (CirNoPVT), cirrhosis with chronic PVT (CPVT), and NCPVT should be compared to suggest which changes in its composition are associated with portal hypertension itself and which are associated with PVT and cirrhosis alone.

Materials and methods

Patients

This was a non-interventional cross-sectional study that included patients with portal hypertension of various origins: non-cirrhotic portal hypertension due to post-PVT EHPVO (group NCPVT), cirrhosis without PVT (group CirNoPVT), cirrhosis with PVT (group CPVT). The control group consisted of clinically healthy individuals who came to the clinic for a preventive examination. All patients signed voluntary informed consent. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Local ethics committee of Sechenov University (20.10.2020, ref: 20–22). The NCPVT and CPVT groups comprised the overall group of patients with PVT. The CirNoPVT and CPVT groups comprised the overall group of patients with cirrhosis.

Inclusion criteria: age 18–70 years, signing of informed consent, presence of chronic portal hypertension according to the contrast-enhanced computed tomography scans (one or both signs: the increase of diameter of portal, splenic, superior, or inferior mesenteric veins; portosystemic collaterals), in combination with one or more of the following signs: esophageal or gastric varices, splenomegaly, ascites. The duration of portal hypertension before inclusion must be at least 6 months.

Exclusion criteria: taking drugs that affect the gut microbiota (pro-, pre-, syn- or antibiotics) or alcohol for 6 weeks before inclusion in the study, the presence of gut diseases or other diseases that significantly affect the composition of the gut microbiota including solid tumors.

The diagnosis of cirrhosis was established by biopsy or a combination of clinical, laboratory and instrumental data including transient elastography according to current guidelines13.

PVT and NCPVT were diagnosed based on the data of the contrast-enhanced computed tomography scans by the presence of a contrast defect in the portal vein trunk or its cavernous transformation. The duration of these changes before inclusion must be at least 6 months. These patients were included in the CPVT group, if their liver stiffness corresponded to stage F4 fibrosis at the time of detection of PVT, and in the NCPVT group, if their liver stiffness corresponded to the stage of F0-F3 fibrosis at the time of the first detection of PVT and at the time of screening for participation in the study. In other cases patients (acute PVT, progression to F4 stage during follow-up) were excluded from the study to eliminate group assignment bias.

All patients underwent standard tests: complete blood count, blood chemistry, abdominal ultrasound, gastroscopy, neurological and physical examination. In addition, their morning stool was taken for the 16S rRNA gene sequencing.

Sample preparation and bacterial 16S rRNA gene sequencing

DNA was extracted using MagPure Stool DNA LQ Kit (Magen) according to the manufacturer’s instructions. A QuDye dsDNA HS Assay Kit (Lumiprobe) and Fluo-200 fluorometer (AllSheng) were used to measure the DNA concentration.

DNA libraries were prepared using PCR amplification with gene-specific 515 F Forward Primer 5’Phos-GTGYCAGCMGCCGCGGTAA14 and 806R Reverse Primer 5’Phos-GGACTACNVGGGTWTCTAAT15 for the V4 regions of 16S rRNA. 290 bp PCR products were amplified by Tersus Plus PCR kit (Evrogen). Thermocycler conditions were 95° C – 3 min, 30 repeat cycles: 95 °С – 30 s, 50 °C–30 s, 72 °C–60 s, then 72 °C–5 min.

Quality of products was analyzed through electrophoresis in 1.5% agarose gel and purified by VAHTS DNA Clean Beads (Vazyme) according to the manufacturer’s instructions. The adaptor ligation reactions included 100 ng 290 bp amplicon, 1 mkl 10x ligase buffer, 0.5 mkl T4 DNA ligase (Genterra), 0.5 mkl index adaptor (VAHTS RNA Adapters set 8 for MGI, Vazyme), water to 10 mkl. The incubation of the reaction mixtures were at 16 degrees overnight. Ligation reaction products were purified by VAHTS DNA Clean Beads (Vazyme) according to the manufacturer’s instructions.

The Phanta Max Super-Fidelity DNA Polymerase kit (Vazyme) was used for the library amplification reaction, primers were Ad153_PCR2_2 (TGTGAGCCAAGGAGTTG) and Ad153_PCR2_1 (5’Phos-GAACGACATGGCTACGA)16.

The quality of the prepared libraries was analyzed on the Qsep100 capillary electrophoresis Analyzer (Bioptic) using the Standart Cartridge kit (Bioptic). A QuDye dsDNA HS Assay Kit (Lumiprobe) and Fluo-200 fluorometer (AllSheng) were used to measure the DNA concentration. Single-stranded circular DNA library for MGI sequencers from PCR products was made by MGIEasy Circularization Kit (MGI) according to the manufacturer’s instructions.

High-throughput Sequencing Kit (MGI) was used to prepare the DNA for sequencing. The samples were sequenced using the DNBSEQ-G50 platform (MGI) according to the manufacturer’s protocol with paired-end reads and 150 bp read length.

Bioinformatic analysis

Microbiome bioinformatic analysis was performed using QIIME2 v.2023.9.217. Raw paired-end reads were filtered by quality, trimmed, merged and denoized to amplicon sequence variants (ASVs) using DADA218. The taxonomic assignment of ASVs was performed using pre-trained Naïve Bayes classifier implemented in q2-feature-classifier plugin19 with human stool-specific taxonomic weights20. The classifier was trained on the Greengenes 13_8 99% operational taxonomic unit (OTU) sequences trimmed to 515 F/806R region. Then ASVs with undefined phylum annotation were filtered out. ASVs that were present in less than 3 samples and had overall frequency below 10 were also removed. For alpha- and beta-diversity analyses, all samples were rarefied to the same sequencing depth (5,500 sequences per sample). Shannon diversity index, Bray-Curtis dissimilarity index, rarefaction curves for alpha- and beta diversity indices were calculated using q2-diversity plugin. Group comparisons of alpha diversity were performed with Kruskal-Wallis test. Group comparisons of beta diversity were performed with nonparametric PERMANOVA21 and PERMDISP. Beta diversity was visualized using principal coordinates analysis (PCoA) plots using qiime2R R package. The Bonferroni correction was used to account for multiple testing. Taxa that were differentially abundant across groups of patients were identified using LEfSe v.1.1.0122 with on-line resource http://www.bic.ac.cn/BIC/#/.

Statistical analysis

The values of the quantitative variables are presented as median [interquartile range]. Qualitative data were expressed as counts and percentages. Comparisons between multiple groups were performed using the Kruskal-Wallis test. Comparison of continuous variables between the 2 groups was carried out using the Mann-Whitney test. Categorical and qualitative variables were compared between multiple groups by chi-squared test. Categorical and qualitative variables were compared between 2 groups by Fisher’s exact test. The level of significance was set at p < 0.05. The analysis was carried out using STATISTICA 10 (TIBCO Software, Palo Alto, CA).

Results

The CPVT and NCPVT patients from a previously collected database were screened to participate in the study. The study also included CirNoPVT patients who came to the Clinic for periodic examination during the recruting of patients with PVT. Twenty two clinically healthy individuals who came to the Clinic for a preventive examination were taken as a control group. Unfortunately, 6 sequenced samples with low sequence depth (less than 4,000 ASV per sample) and we were forced to exclude these patients from the study (Fig. 1).

Fig. 1
figure 1

Study flowchart.

Full size image

There was no significant difference in the etiology of cirrhosis between patients with and without PVT (p = 0.830; Supplementary Table 1). The causes of EHPVO in the NCPVT group were hematologic malignancies, mutation in the thrombin gene, deficiencies of natural anticoagulants, portal tributary phlebitis due to omphalitis or acute pancreatitis in past, taking combined oral contraceptive pill and unclear (Supplementary Table 2).

There was no significant difference among NCPVT, CPVT, CirNoPVT, and control groups in age, body mass index, diet, and gender distribution (Supplementary Table 3). There were no significant differences among NCPVT, CPVT, and CirNoPVT groups in main co-morbidities (Table 1).

Table 1 Main characteristics of included patients.
Full size table

The CPVT and CirNoPVT groups did not differ significantly in the manifestations and severity of cirrhosis, as well as in the drugs used, with the exception that patients with PVT were more likely to use anticoagulants than CirNoPVT patients (Table 1).

Patients with NCPVT had preserved liver function, which was reflected in the absence of hepatic hypoalbuminemia and hyperbilirubinemia, higher albumin levels and lower total bilirubin levels than in both groups of the patients with cirrhosis. Four patients with NCPVT had hyperbilirubinemia due to indirect fraction caused by hematologic complications of hereditary thrombophilias (microangiopathic hemolytic anemia) or ineffective hematopoiesis due to myeloproliferative diseases. All these patients had normal serum levels of albumen, prothrombin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma glutamyl transferase (GGT), and alkaline phosphatase (ALP), as well as liver stiffness corresponded to fibrosis stage ≤ F2, except for the only case of minor hypoalbuminemia that was associated with the complications of myeloproliferative disease (Table 1).

The severity of ascites and hepatic encephalopathy in NCPVT patients was less than in patients with cirrhosis. The severity of esophageal varices did not differ significantly between all 3 patient groups. Patients with NCPVT had normal values ​​for biomarkers of liver damage (ALT and AST), which were higher in patients with cirrhosis. Patients with NCPVT had less severe thrombocytopenia than patients with cirrhosis. At the same time, the level of red and white cells in the blood of patients of all three groups did not differ significantly (Table 1).

Patients with cirrhosis were more likely to use spironolactone and loop diuretics than patients with NCPVT. Patients with PVT (CPVT + NCPVT) used anticoagulants more often than patients with cirrhosis without PVT. There was no significant difference in beta-blocker use among patient groups (Table 1).

There was no significant difference among groups in Shannon gut microbiota alpha-diversity index (p = 0.710; Supplementary Fig. 1). There were no significant differences in gut microbiota beta-diversity among the groups of cirrhosis without PVT, CPVT and NCPVT. However, beta-diversity within all 3 groups of patients with portal hypertension significantly differed from the group of healthy controls (Fig. 2, Supplementary Table 4).

Fig. 2
figure 2

Principal coordinate analysis (PCoA) plot illustrating Bray-Curtis beta-diversity between samples of patients with portal hypertension (cirrhosis without portal vein thrombosis [CirNoPVT], cirrhosis with portal vein thrombosis [CPVT], non-cirrhotic portal vein thrombosis [NCPVT]), and healthy controls.

Full size image

The general group of patients with portal hypertension (cirrhosis with and without PVT + NCPVT) had in the gut microbiota a higher abundance of Streptococcus, Veillonella, Escherichia, Enterococcus, Streptococcaceae, Enterobacteriaceae, Enterococcaceae, Bacilli, Gammaproteobacteria, Proteobacteria, and a lower abundance of Roseburia, Faecalibacterium, Parabacteroides, Gemmiger, Methanobrevibacter, Ruminococcaceae, Methanobacteriaceae, Clostridia, Methanobacteria, Firmicutes and Euryarchaeota (Fig. 3a). Patients in the CirNoPVT and CPVT groups had largely similar differences from the group of healthy individuals in the gut microbiota composition (Fig. 3b-d).

Fig. 3
figure 3

Histograms of the LDA scores produced by LEfSe analysis reveal differentially abundant taxa in the gut microbiota between the following groups: (a) portal hypertension and healthy controls; (b) cirrhosis and healthy controls; (c) cirrhosis without portal vein thrombosis (CirNoPVT) and healthy controls; (d) cirrhosis with portal vein thrombosis (CPVT) and healthy controls.

Full size image

Patients with NCPVT, on the one hand, as well as patients with cirrhosis, had a higher abundance of Streptococcus, Escherichia, Enterococcus, Enterobacteriaceae, Enterococcaceae, Streptococcaceae, Bacilli, Gammaproteobacteria, Proteobacteria, and a lower abundance of Roseburia, Faecalibacterium, Methanobrevibacter, Ruminococcaceae, Methanobacteriaceae, Clostridia, Methanobacteria, and Euryarchaeota as they were compared with healthy individuals. On the other hand, they had a higher abundance of Bifidobacterium, Bifidobacteriaceae, Actinobacteria, and a lower abundance of Gemmiger and Catenibacterium compared to healthy individuals, which was not observed in the cirrhosis groups (Fig. 4a).

Fig. 4
figure 4

Histograms of the LDA scores produced by LEfSe analysis reveal differentially abundant taxa in the gut microbiota between the following groups: (a) non-cirrhotic portal vein thrombosis (NCPVT) and healthy controls; (b) NCPVT and cirrhosis without portal vein thrombosis (CirNoPVT); (c) NCPVT and cirrhosis with portal vein thrombosis (CPVT); (d) CPVT and CirNoPVT.

Full size image

Patients with NCPVT, compared with cirrhotic patients without PVT, had a higher content of Bifidobacterium, Bifidobacteriaceae and Actinobacteria and a lower content of Parabacteroides, Weissella, Lactococcus, Porphyromonadaceae and Leuconostocaceae in the gut microbiota (Fig. 4b). Patients with NCPVT compared to patients with CPVT had higher levels of Bifidobacterium, Collinsella, Bifidobacteriaceae, Coriobacteriaceae, Actinobacteria and Coriobacteriia, and lower levels of Coprobacillus (Fig. 4c).

Patients with CPVT compared with cirrhotic patients without PVT had lower abundance of Leuconostocaceae and Weissella in the gut microbiota (Fig. 4d).

Comparing 4 groups at the level of the main taxa (Supplementary Tables 5 and 6), the abundances of Proteobacteria and Bacilli were increased in all groups with portal hypertension compared to the norm, but without significant differences among groups with portal hypertension (Fig. 5a-b). The abundance of Clostridia and Euryarchaeota was reduced in all groups with portal hypertension compared to controls, but without significant differences among groups with portal hypertension (Fig. 5c-d). The abundance of Actinobacteria was increased only in the NCPVT group (p = 0.008), while it was not significantly different from controls in both groups with cirrhosis (Fig. 5e). Moreover, the abundance of these bacteria in NCPVT was significantly higher than in both groups of patients with cirrhosis (p < 0.001 and p = 0.010; Fig. 5e). The abundance of Porphyromonadaceae (and the genus Parabacteroides included in this family) was significantly reduced in both groups with PVT (p = 0.003 and p < 0.001), while it was not significantly different from normal persons in the group of cirrhosis without PVT (p = 0.089; Fig. 5f).

Fig. 5
figure 5

Comparison of the relative abundances of the main significant taxa of the gut microbiota between patients with non-cirrhotic portal vein thrombosis (NCPVT), cirrhosis with portal vein thrombosis (CPVT), cirrhosis without portal vein thrombosis (CirNoPVT) and healthy individuals: (a) Proteobacteria, (b) Bacilli, (c) Clostridia, (d) Euryarchaeota, (e) Actinobacteria, (f) Porphyromonadaceae.

Full size image

There was no significant difference between patients with cirrhotic and non-cirrhotic PVT (CPVT vs. NCPVT) in the extent of thrombosis and the incidence of cavernous transformation of the portal vein (Supplementary Table 7). There was no significant difference in the beta-diversity of the gut microbiota between patients with different extension portal vein thrombosis among the CPVT and NCPVT groups (Supplementary Table 8, Supplementary Fig. 2). There was no significant difference in the beta-diversity of the gut microbiota between patients with and without cavernous transformation of the portal vein among the CPVT and NCPVT groups (Supplementary Table 9, Supplementary Fig. 2). Among the overall group of patients with PVT, patients with cavernous transformation of the portal vein had higher levels of Pasteurellaceae and Haemophilus in the gut microbiota compared to patients without this portal vein change. Among the general group of patients with PVT, patients with thrombosis extending to both portal vein sources (superior mesenteric and splenic veins) had higher levels of Actinobacteria, Turicibacteraceae, Klebsiella, and Turicibacter in the gut microbiota compared to patients without splenic vein involvement (Supplementary Fig. 3). There was no significant difference in the gut microbiota composition between patients with PVT alone or with simultaneous portal and splenic vein thrombosis (Supplementary Fig. 3). The size of these subgroups in the CPVT and NCPVT groups was too small to conduct adequate LEfSe analyses within these groups.

Discussion

The aim of our study was to analyze whether non-cirrhotic portal hypertension with preserved liver function causes the same adverse changes in the composition of the gut microbiota as cirrhosis. We received a positive answer to this question. All the main manifestations of cirrhosis-specific gut dysbiosis (increases in the Proteobacteria and Bacilli abundances and a decrease in the Clostridia abundance) were also observed in patients with portal hypertension without cirrhosis. This indicates that portal hypertension, and not a decrease in liver function, can be the main factor in the development of harmful cirrhosis-specific gut dysbiosis, which in our study manifested in the same way as in most previously published papers4.

The pathogenic role of cirrhosis-associated gut dysbiosis can be briefly presented as follows4,22. Beneficial Clostridia form short-chain fatty acids and other beneficial products that strengthen the intestinal barrier. Therefore, a decrease in the abundance of these bacteria in cirrhosis is associated with an increase in the permeability of the intestinal barrier. Harmful Proteobacteria form endotoxin, and together with Bacilli, being facultative anaerobes, are able to penetrate into living tissues of the human body (bacterial translocation), leading to a systemic inflammatory reaction. The latter contributes to the development of systemic vasodilation, fluid retention, hyperdynamic circulation, aggravating portal hypertension4,22,23.

Apparently, we were the first to find a decrease in the content of methane-forming archaea in the intestine in portal hypertension, both cirrhotic and non-cirrhotic in origin4. The significance of these changes in the pathogenesis of cirrhosis remains to be determined. The level of methane-forming archaea is increased in obesity and decreased in anorexia, which may be associated with body weight control24. Cirrhosis is characterized by sarcopenia, a decrease in muscle and total body weight, which may be associated with a deficiency of archaea in the intestine25. In addition, methane-producing archaea are reduced in inflammatory bowel disease, which may also indicate a protective role against the development of intestinal damage24.

Our study also revealed that the patients with non-cirrhotic portal hypertension, in contrast to patients with cirrhosis, had the level of beneficial Bifidobacteria from the Actinobacteria phylum significantly increased. Perhaps this is a compensatory reaction and is somehow related to the preservation of liver function in these patients. Testing this hypothesis is a task for subsequent researchers.

An interesting finding was a decrease in the Porphyromonadaceae abundance of the gut microbiota in PVT, regardless of its genesis, which was not detected in cirrhosis without PVT. The exact role of these bacteria in the human intestine has not been established, but it is known that their level is increased in the intestines of long-livers26. Further research is needed to clarify this issue.

How does portal hypertension lead to the development of cirrhosis-specific gut dysbiosis? We can assume that the cause of this is portal gastropathy and enteropathy, which develop as reactions to stagnation of venous blood in the stomach and intestine. Possibly this leads to a decrease in the protective function of the stomach and small intestine and a slowdown in small intestine mobility27,28,29. Harmful bacteria, the growth of which characterizes cirrhosis-specific gut dysbiosis, most likely have an oral origin. Normally, they die in the stomach and small intestine. However, these protective mechanisms may be impaired with portal hypertension. This, together with a slowdown in small intestinal motility, can contribute to their excessive growth and suppression of beneficial bacteria. New experimental and clinical studies are needed to test this hypothesis.

The detection of a disturbance in the composition of the gut microbiota in portal hypertension can be an indication for the prescription of drugs that affect the gut bacteria community, including probiotics. Recent reviews have shown a positive effect from these drugs in cirrhosis30, as well as in critically ill patients31.

The strength of our work is that it is the first to study the composition of the gut microbiota in non-cirrhotic portal hypertension and compared it with the composition of the gut microbiota in cirrhosis, confirming the hypothesis of the similarity of gut dysbiosis in these pathological conditions, which strengthened the theory about the critical role of portal hypertension in the development of harmful changes in the composition of the gut microbiota in cirrhosis.

A limitation of our study is the small number of patients examined, which is associated with the relative rarity of NCPVT, and also the fact that we were not able to conduct additional tests, namely, measure the hepatic venous pressure gradient, assess the state of the intestinal barrier and small intestinal motility, as well as the secretion of antibacterial molecules, which remains a challenge for future research. Analysis of the correlation between the composition of the gut microbiota and these indicators in portal hypertension of various origins remains a task for future studies. The next limitation of our study was that we were unable to obtain groups that were equal in the number of participants and on all the indicators of interest, which may also be a source of biases. Further studies with more evenly distributed groups are needed to verify our findings.

Conclusion

Our data suggest that cirrhosis-specific gut dysbiosis, consisting in an increase in the abundances of harmful Proteobacteria and Bacilli and a decrease in the abundance of beneficial Clostridia, is associated with portal hypertension, and is not with decreased liver function. We acknowledge that our data are preliminary and need to be verified by further studies.