Gut microbiome composition in patients with liver cirrhosis with and without hepatic encephalopathy: A systematic review and meta-analysis

Abstract

BACKGROUND

The gut microbiome is associated with hepatic encephalopathy (HE), but research results on the gut microbiome characteristics of patients with liver cirrhosis with and without HE are inconsistent.

AIM

To study the gut microbiota characteristics of patients with liver cirrhosis with and without HE.

METHODS

We searched the PubMed, Web of Science, EMBASE, and Cochrane databases using two keywords, HE, and gut microbiome. According to the inclusion and exclusion criteria, suitable literature was screened to extract data on the diversity and composition of the fecal microbiota in patients with liver cirrhosis with and without HE. The data were analyzed using RevMan and STATA.

RESULTS

Seventeen studies were included: (1) A meta-analysis of 7 studies revealed that the Shannon index in liver cirrhosis patients with HE was significantly lower than that in patients without HE [-0.20, 95% confidence interval (CI): -0.28 to -0.13, I² = 20%]; (2) The relative abundances of Lachnospiraceae (-2.73, 95%CI: -4.58 to -0.87, I² = 38%) and Ruminococcaceae (-2.93, 95%CI: -4.29 to -1.56, I² = 0%) in liver cirrhosis patients with HE was significantly lower than those in patients without HE; (3) In patients with HE, Enterococcus, Proteobacteria, Enterococcaceae, and Enterobacteriaceae proportions increased, but Ruminococcaceae, Lachnospiraceae, Prevotellaceae, and Bacteroidetes proportions decreased; (4) Differences in the fecal metabolome between liver cirrhosis patients with and without HE were detected; and (5) Differential gut microbiomes may serve as diagnostic and prognostic tools.

CONCLUSION

The gut microbiomes of patients with liver cirrhosis with and without HE differ. Some gut microbiomes may distinguish liver cirrhosis patients with or without HE and determine patient prognosis.

Key Words: Gut microbiome; Cirrhosis; Hepatic encephalopathy; Review; Meta-analysis

Core Tip: The gut microbiome is associated with hepatic encephalopathy (HE), but research results on the gut microbiome characteristics of patients with liver cirrhosis with and without HE are inconsistent. We aimed to conduct a meta-analysis of the gut microbiota characteristics of patients with liver cirrhosis with and without HE, to provide a reference for the targeted regulation of the gut microbiota and reduction in HE incidence.

INTRODUCTION

Hepatic encephalopathy (HE) is one of the most common complications of liver cirrhosis and a leading cause of death in patients with liver disease, with an increasing incidence rate[1,2]. In recent years, increasing research has focused on the relationships between the gut microbiome and the occurrence and development of diseases[3]. A difference in the gut microbiota has been reported between patients with cirrhosis and normal patients[4], which is also associated with complications in patients with cirrhosis[5]. Cirrhosis and portal hypertension are related to systemic inflammation and portal vein blood stasis. These conditions further damage intestinal barrier function, alter intestinal permeability, and lead to bacterial translocation[6]. Ultimately, changes in the composition and function of the intestinal microbiota are involved in the development of HE[7]. The gut microbiome may serve as a reasonable biomarker for liver cirrhosis diagnosis and prognosis[5]. However, in existing reports, there was inconsistency in the differences in the gut microbiota among cirrhosis patients with and without HE. Therefore, this study conducted a meta-analysis of the gut microbiota characteristics of patients with liver cirrhosis with and without HE, to provide a reference for targeted regulation of the gut microbiota to reduce the incidence of HE in patients.

MATERIALS AND METHODS

Data sources and search strategy

The PubMed, Web of Science, EMBASE, and Cochrane databases were searched. MeSH terms and free words were used to search for literature related to gastrointestinal microbiomes and HE. The search period ended on June 1, 2024. At the same time, a search was conducted for references. The specific retrieval strategy is shown in Supplementary Table 1. This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines and was registered in PROSPERO (CRD42024569064).

Inclusion and exclusion criteria

The inclusion criterion was as follows: Patients with liver cirrhosis with or without HE who underwent fecal gut microbiota testing. The gut microbiota diversity index and gut microbiota composition were measured. The exclusion criterion was patients with liver cancer. Patients without cirrhosis. Duplicate studies, case reports, reviews, meta-analyses, unrelated studies, and animal experiments were excluded.

Outcome measures

Main research findings: Differences in gut microbiota diversity and composition between HE and non-HE patients with liver cirrhosis. The alpha diversity of the gut microbiota included the Shannon, Simpson, and Chao 1 indices. Secondary research findings: Effects of drug therapy on the gut microbiota. Relationship between the gut microbiota and metabolism. The value of the gut microbiota in distinguishing patients with liver cirrhosis with and without HE and predicting patient prognosis.

Study selection and quality assessment

Two independent evaluators (Xiao-Tong Xu and Min-Jie Jiang) selected studies that met the criteria by screening titles and abstracts. The quality evaluation of the literature was based on the quality evaluation table, as shown in Table 1. For ambiguous parts, a third evaluator was required to participate in the discussion and make the final decision. The filtering process is shown in Figure 1.

Figure 1 Flowchart of literature screening.

Table 1 Agency for healthcare research and quality scoring table for studies.

Ref.	Representativeness of the exposed cohort	Selection of the nonexposed cohort	Ascertainment of exposure	Demonstration that outcome of interest was not present at the start of the study	Comparability between exposed and nonexposed cohort	Assessment of outcome	Follow-up was long enough for outcomes to occur	Adequacy of follow-up of cohorts	Total score
Wang et al[9], 2023, China	1	1	1	1	2	1	1	1	9
Bajaj et al[12], 2012, United States	1	1	1	1	2	1	1	0	8
Bajaj et al[13], 2015, United States	1	1	1	1	1	1	1	1	8
Bajaj et al[14], 2019, United States	1	1	1	1	2	1	1	0	8
Bajaj et al[55], 2018, United States	1	1	1	1	1	1	1	1	8
Sung et al[15], 2019, China	1	1	1	1	2	1	1	1	9
Bloom et al[8], 2021, Japan	1	1	1	1	2	1	1	0	8
Saboo et al[56], 2022, United States	1	1	1	1	1	1	1	0	7
Yukawa-Muto et al[10], 2022, Japan	1	1	1	1	1	1	1	1	8
Li et al[57], 2022, China	1	1	1	1	2	1	1	0	8
Hua et al[58], 2022, China	1	1	1	1	1	1	1	0	7
Bajaj et al[59], 2022, United States	1	1	1	1	1	1	1	1	8
Yu et al[11], 2021, China	1	1	1	1	2	1	1	1	9
Baja et al[60], 2020, United States	1	1	1	1	2	1	1	1	9
Ahluwalia et al[16], 2016, United States	1	1	1	1	1	1	1	0	7
Zhang et al[17], 2013, China	1	1	1	1	1	1	1	1	8
Bajaj et al[61], 2012, United States	1	1	1	1	2	1	1	0	8

Data extraction

Author, country, year, age, gender, etiology, liver function score, gut microbiota detection method, sequencing region of liver cirrhosis patients with or without HE. Diversity of gut microbiota alpha diversity (Chao 1, Simpson, and Shannon indices) and the composition of the gut microbiota.

Statistical analysis

A single group rate analysis was conducted for studies that reported only the composition proportion of the fecal gut microbiota. We extracted mean values and standard deviations from studies with quantitative data on the fecal gut microbiota for analysis. The collected data were analyzed using RevMan and STATA with random effects or fixed effects.

RESULTS

Search flow and study overview

A total of 1246 articles were initially searched in the database, and after screening, 17 articles were finally included, including 782 cases of liver cirrhosis with HE and 839 cases of liver cirrhosis without HE. The literature screening process is shown in Figure 1. Two studies were conducted in Japan, 6 in China, and 9 in the United States, and relevant information was extracted from the studies. The method for detecting the fecal microbiota was 16S rRNA sequencing, among which 5 studies mentioned the amplification region, 3 were variable regions 1-2, and 2 were variable regions 3-4, as shown in Table 2.

Table 2 Basic information included in the study.

Ref.	Amplified region/reference databases	Characteristics of gut microbiota/alpha diversity indices
		Age	Gender (male/female)	MELD	Child-Pugh score/class	Etiology	PPI
Wang et al[9], 2023, China	N/A, N/A	Cirrhosis with HE (n = 30): (1) Higher of Burkholderiaceae, Ralstonia, Parabacteroides-distasonis; (2) Phylum: Firmicutes (58.08%), Proteobacteria (18.83%), Bacteroidetes (18.19%), and Actinobacteria (2.5%); (3) Genera: Bacteroides (10.59%), Enterococcus (22.98%), Prevotella (2.21%), Escherichia-Shigella (11.9%), Ruminococcus-gnavus (2.13%), and Streptococcus (5.9%); and (4) Shannon (3.84 ± 2.01), Simpson (0.78 ± 0.19), and Chao 1(231.26 ± 66.41)
		56.13 ± 13.20	19/11	19.67 ± 5.95	Child A: 0, Child B: 9, Child C: 21	HBV (9)/HCV (1)/alcohol (12)/HBV + alcoholic (1)/HCV + alcohol (1)/others (6)	N/A
		Cirrhosis without HE (n = 30): (1) Higher of Aegypius-monachus, Erysipelatoclostridiaceae, and Erysipelotrichales; (2) Phylum: Firmicutes (58.42%), Proteobacteria (15.08%), Bacteroidetes (20.14%), and Actinobacteria (1.6%); (3) Genera: Bacteroides (14.80%), Enterococcus (16.36%), Prevotella (2.55%), Escherichia-Shigella (2.55%), Ruminococcus-gnavus (5.35%), and Streptococcus (3.82%); and (4) Shannon (4.11 ± 1.61), Simpson (0.81 ± 0.18), and Chao 1 (211.16 ± 56.35)
		52.23 ± 9.00	20/10	13.90 ± 6.51	Child A: 9, Child B: 7, Child C: 14	HBV (14)/HCV (1)/alcohol (8)/HBV + alcoholic (3)/HCV + alcohol (1)/others (3)	N/A
Bajaj et al[12], 2012, United State	N/A, RDP	Cirrhosis with HE (n = 17): Leuconostocaceae (2.19 ± 1.08), Clostridiales-Incertae Sedis XIV (0.99 ± 0.21), Ruminococcaceae (5.68 ± 1.42), Lachnospiraceae (9.54 ± 3.72), Enterobacteriaceae (10.02 ± 4.13), Streptococcaceae (4.05 ± 1.98), Fusobacteriaceae (1.36 ± 0.95), Alcaligenaceae (2.61 ± 0.74); N/A
		56 ± 3	16/1	MELD17 ± 6	N/A	Alcohol (10)/others (7)	16
		Cirrhosis without HE (n = 8): Leuconostocacea (1.69 ± 0.95), Clostridiales_Incertae Sedis XIV (1.29 ± 0.35), Fusobacteriaceae (2.75 ± 2.75), Lachnospiraceae (12.08 ± 2.47), and Ruminococcaceae (9.04 ± 2.62); N/A
		55 ± 5	7/1	MELD 12 ± 5	N/A	Alcohol (3)/others (5)	7
Bajaj et al[13], 2015, United State	N/A	Cirrhosis with HE (n = 43): Clostridiales XIV (4.5%), Lachnospiraceae (16%), Ruminococcaceae (7.4%), Fusobacteriaceae (1.0%), Prevotellaceae (5.0%), Enterococcaceae (1.0%), Enterobacteriaceae (1.0%), Erysipelotrichaceae (0.6%), Bacteroidaceae (24.9%), and Streptococcacae (1.9%); N/A
		56 ± 16	34/9	17.2 ± 7.2	9.7 ± 4.4	HCV (20)/alcohol (6)/HCV + alcohol (9)/NASH (7)/others (1)	19
		Cirrhosis without HE (n = 59): ClostridialesXIV (6.6%), Lachnospiraceae (21.3%), Ruminococcaceae (8.7%), Fusobacteriaceae (0%), Prevotellaceae (5.3%), Enterococcaceae (0%), Enterobacteriaceae (3%), Erysipelotrichaceae (1.9%), Bacteroidaceae (24.5%), and Streptococcacae (4.4%); N/A
		54 ± 13	50/9	8.6 ± 2.5	6.2 ± 5.1	HCV (27)/alcohol (14)/HCV + alcohol (7)/NASH (7)/others (4)	21
Bajaj et al[14], 2019, United State	N/A, N/A	Cirrhosis with MHE (n = 122): (1) PHES: Higher of Lactobacillales, Bacilli, and Micrococcaceae; (2) ICT: Higher of Lactobacillales, Enterobacteriales, Bacilli, Enterobacteriaceae, Protebacteria, Gammaprotebacteria, Micrococcaceae, Eubacteriaceae, Streptococcaceae, Actinomycetales; (3) Stroop: Higher of Lactobacillales, Lactocillaceae, Actinobacteria; and (4) Shannon MHE on PHES (1.9 ± 0.7), MHE on ICT (2.0 ± 0.7), and MHE on Stroop 1.9 ± 0.6
		60.1 ± 7.0	122/0	11.3 ± 4.7	N/A	N/A	77
		Cirrhosis without HE (n = 145): (1) PHES: Higher of Clostridiales, Incertae Sedis XI and XIII, Pasteurellales, Pasteurellaceae, Lachnospiraceae, and Clostridia; (2) ICT: Higher of Bacteroidaceae, Bacteroidales, Bacteroidia, Bacteroidetes; (3) Stroop: Higher of Eubacteriaceae, GP1, Telmatobacter, Breoghania, Caulobacterales, Telmatobacter, Acidobacteria, proteobacteria-incertaesedis, Hyphomonadaceae, and Listeriaceae; and (4) Shannon non-MHE on PHES (2.2 ± 0.6), non-MHE on ICT (2.1 ± 0.6), and non-MHE on Stroop (2.1 ± 0.6)
		59.4 ± 9.1	85/60	14.3 ± 7.1	N/A	N/A	54
Bajaj et al[55], 2018, United State	V1-2/RDP	Cirrhosis with prior HE (n = 10): Higher abundance of Erysipelothriceaea, Erysipelotrichia, and Erysipelotrichales; N/A
		N/A	N/A	13.2 ± 3.1	N/A	N/A	N/A
		Cirrhosis without prior HE (n = 14): Higher abundance of Selomonadales and Negativicutes; N/A
		N/A	N/A	7.8 ± 1.7	N/A	N/A	N/A
Sung et al[15], 2019, China	V3-4, N/A	Cirrhosis with HE (n = 62): (1) Firmicutes (41.4%), Bacteroidetes (36.5%), Proteobacteria (15.8%) and Actinobacteria (3.4%); (2) Schlegelella (1.172 ± 2.924), Megasphaera (1.105 ± 1.518), Veillonella (11.007 ± 12.473), Enterococcus (1.369 ± 5.038), Lactobacillus (1.086 ± 3.184), XI (1.603 ± 7.644), Prevotella (1.254 ± 4.101), Sedis (0.401 ± 0.923), Bacteroides (1.780 ± 4.062), and Parabacteroides (0.788 ± 2.946); and (3) Shannon (2.6 ± 0.59) and Chao 1 (146.83 ± 47.35)
		56.7 ± 12.2	49/13	18.5 ± 6	Child A: 0; Child B: 22; Child C: 40	HBV (18)/HCV (19)/alcoholic (36)/others (4)	22
		Cirrhosis without HE (n = 15): (1) Firmicutes (on average 332%), Bacteroidetes (47.2%), Proteobacteria (15.5%), and Actinobacteria (1.4%); (2) Schlegelella (0.029 ± 0.085), Megasphaera (1.031 ± 1.861), Veillonella (6.015 ± 7.01073), Enterococcus (0.02 ± 0.031), Lactobacillus (0.011 ± 0.017), XI (0.039 ± 0.081), Prevotella (0.022 ± 0.084), Sedis (1.788 ± 2.968), Bacteroides (1.607 ± 2.290), and Parabacteroides (1.334 ± 2.269); and (3) Shannon (2.64 ± 0.45) and Chao 1 (194.53 ± 52.93)
		58.1 ± 12.3	12/3	16.8 ± 5.58	Child A: 0; Child B: 8; Child C: 7	HBV (6)/HCV (2)/alcohol (8)/others (0)	4
Bloom et al[8], 2021, Japan	N/A, NCBI	Cirrhosis with HE (prior or current) (n = 33): (1) Higher of Bifidobacteriaceae, Enterococcaceae, Strptoccaceae, Lactobacteriaceae, and Enterobacteriaceae; and (2) Simpson (4.2 ± 2.64)
		57 ± 10	19/14	21 ± 8	N/A	Alcohol (16)/NASH (10)/viral and alcohol (4)/viral (1)/others (2)	28
		Cirrhosis without HE (n = 16): (1) Higher of Lachnospiraceae, Bacteroidaceae, Ruminococcaceae, Veillonellaceae, AKKermansiaceae, and Eubacteriaceae; and (2) Simpson (10.88 ± 9.46)
		59 ± 10	11/5	16 ± 10	N/A	Alcohol (6)/NASH (2)/viral and alcohol (3)/vral (4)/others (1)	11
Saboo et al[56], 2022, United State	V1-2/RDP	Cirrhosis with HE (n = 57): Lower of Lachnospriaceae, Bacteroidaceae, and Ruminococcaceae; N/A
		59.6 ± 6.4	46/11	15.1 ± 7.2	N/A	Alcohol (20)/HCV (22)/NASH (12)/others (3)	28
		Cirrhosis without HE (n = 125): Higher of Lachnospriaceae, Bacteroidaceae, and Ruminococcaceae; N/A
		60.8 ± 7	97/28	9.6 ± 3.8	N/A	Alcohol (54)/HCV (45)/NASH (21)/others (5)	50
Yukawa-Muto et al[10], 2022, Japan	N/A	Decompensated cirrhosis with HE (n = 26): (1) Higher of Streptococcus salivarius, Lactobacillus spp., Bifidobacterium dentium, Lactobacillus mucosae, Lactobacillus fermentum, Veillonella atypica, Veillonella parula, and Lactobacillus rhamnosus; and (2) Shannon (6.47 ± 0.38)
		68.4 (52-85)	15/11	11.7 (6-19)	N/A	HBV (3)/HCV (9)/NASH (7)/alcohol (3)/PBC (3)/AIH (1)	N/A
		Decompensated cirrhosis without HE (n = 27): (1) Lower of Streptococcus salivarius, Lactobacillus spp., Bifidobacterium dentium, Lactobacillus mucosae, Lactobacillus fermentum, Veillonella atypica, Veillonella parula, and Lactobacillus rhamnosus; and (2) Shannon (6.90 ± 0.5)
		67.6 (54-81)	16/11	8.2 (6-12)	N/A	HBV (0)/HCV (26)/NASH (1)/alcohol (0)/PBC (0)/AIH (0)	N/A
Li et al[57], 2022, China	V3-4/RDP	Cirrhosis with HE (n = 33): (1) Higher of Ruminococaceae, Lachnospiraceae, Ruminococcus, Rumi_uncultured, Roseburia, Dialister, Dialister, Coprococcus, Blautia, and Anaerostipes; and (2) Shannon (2.84 ± 0.45) and Chao 1 (221.39 ± 39.36)
		51 (44-58)	19/14	16.84 (13.6, 19.27)	8 (7-11); Child A: 3; Child B: 21; Child C: 9	HBV + HCV (25)/others (8)	N/A
		Cirrhosis without HE (n = 73): (1) Lower of Ruminococaceae, Lachnospiraceae, Ruminococcus, Rumi_uncultured, Roseburia, Dialister, Dialister, Coprococcus, Blautia, and Anaerostipes; and (2) Shannon (3.04 ± 0.39) and Chao 1 (244.24 ± 58.72)
		52 (45-58)	46/27	14.34 (12.15, 16.47)	7 (6-8); Child A: 34; Child B: 35; Child C: 4	HBV + HCV (50)/others (23)	N/A
Hua et al[58], 2022, China	N/A	Cirrhosis with HE (n = 5): (1) Higher of Pasteurellales, Pasteurellaceae, Haemophilus, and Selenomonas; (2) Higher of Veillonella (4.12 ± 6.09), Pasteurellales (2.39 ± 4.04), Pasteurellaceae (2.39 ± 4.04), Haemophilus (2.34 ± 3.96), Selenomonas (0.016 ± 0.04), Lachnospiraceae (11.48 ± 6.53), Coprococcus (0.311 ± 0.292), and Dorea (0.855 ± 0.716); and (3) N/A
		60.11 ± 9.54	4/1	N/A	N/A	HBV (5)	N/A
		Cirrhosis without HE (n = 15): (1) Higher of Veillonella; (2) Higher of Veillonella (4.40 ± 6.13), Blautia (1.36 ± 1.88), Turicibacterales (0.012 ± 0.039), Turicibacter (0.012 ± 0.039), Turicibacteraceae (0.012 ± 0.039), Adlercreutzia (0.053 ± 0.014), Holdemania (0.009 ± 0.016), and Coprococcus (0.305 ± 0.43); (3) Lachnospiraceae (9.08 ± 6.38); and (4) N/A
		61.73 ± 13.55	11/4	N/A	N/A	HBV (15)	N/A
Bajaj et al[59], 2022, United State	V1-2/RDP	Cirrhosis with MHE (n = 180): Higher of Pediococcus, Lacticaseibacillus, Pseudomonas, Doudenibacillus, Lactobacillus, Enterococcus, Escherichia-Shigella, Limosilactobacillus, Klebsiella, Sutterella, Megamonas, Veillonella, Absiella, and Prevotellamassila; N/A
		N/A	N/A	N/A	N/A	N/A	N/A
		Cirrhosis without MHE (n = 141): Higher of Enterobacter, Weissella, Anaerobutyrium, Romboutsia, and Citrobacter; N/A
		N/A	N/A	N/A	N/A	N/A	N/A
Yu et al[11], 2021, China	N/A, NCBI	Cirrhosis with HE (n = 5): (1) Higher of Bacteroidetes and Proteobacteria; (2) Lower of Firmicutes, Candidatus Saccharibacteria, and the Verrucomicrobia phyla; and (3) Shannon (1.78 ± 0.52), Simpson (0.71 ± 0.15), and Chao 1 (48.42 ± 17.78)
		58.2 ± 7.04	3/2	N/A	N/A	N/A	N/A
		Cirrhosis without HE (n = 16): (1) Lower of Bacteroidetes and Proteobacteria; (2) Higher of Firmicutes, Candidatus Saccharibacteria, and the Verrucomicrobia phyla; and (3) Shannon (2.23 ± 0.4), Simpson (0.82 ± 0.07), and Chao 1 (68.14 ± 20.05)
		55.33 ± 5.86	10/6	N/A	N/A	N/A	N/A
Bajaj et al[60], 2020, United State	N/A, MYSQL	Cirrhosis with HE 16S rRNA (n = 29): (1) Bacteroidaceae (35%), Prevotellaceae (0%), Lactobacillaceae (1%), Lachnospiraceae (11.3%), Ruminococcaceae (4.4%), Streptococcaceae (1%), Veillonellaceae (0%), and Enterobacteriaceae (1%); and (2) Shannon (1.49 ± 0.37)
		58.8 ± 11.7	20/9	14.8 ± 7	N/A	N/A	28%
		Cirrhosis without HE (n = 51): (1) Bacteroidaceae (33%), Prevotellaceae (0%), Lactobacillaceae (1%), Lachnospiraceae (26%), Ruminococcaceae (10%), Streptococcaceae (0.1%), Veillonellaceae (0%), and Enterobacteriaceae (0%); and (2) Shannon (1.6 ± 0.4)
		57.9 ± 7.4	36/15	9.5 ± 3.4	N/A	N/A	18%
Ahluwalia et al[16], 2016, United State	N/A	Cirrhosis with prior HE (n = 85): (1) Phylum: Firmicutes (24 ± 20), Bacteroidetes (24 ± 26), Proteobacteria (7 ± 15); (2) Family: Bacteroidetes-Bacteroidaceae (18.2 ± 25.2), Bacteroidetes_Porphyromonadaceae (3.9 ± 7.7), Bacteroidetes-Prevotellaceae (2.9 ± 8), Bacteroidetes-Rikenelleaceae (0 ± 2.4), Firmicutes_Lactobacillaceae (4 ± 9.5), Firmicutes-Enterococcaceae (3.6 ± 10.9), Firmicutes-Clostridiales XIV (1.4 ± 3.5), Firmicutes-Lachnospiraceae (6.3 ± 11), Firmicutes-Ruminococcaceae (3.0 ± 5.4), Firmicutes-Veillonellaceae (2. 0 ± 4.8), and Proteobacteria_Enterobacteriaceae (5.5 ± 14.8); and (3) N/A
		56.2 ± 13.5	16/69	15.5 ± 5.2	N/A	HCV (27)/alcohol (16)/HCV + alcohol (20)/NASH (12)/others (10)	N/A
		Cirrhosis without prior HE (n = 62): (1) Phylum: Firmicutes (28 ± 19), Bacteroidetes (33 ± 30), and Proteobacteria (3 ± 6); (2) Family: Bacteroidetes-Bacteroidaceae (22.2 ± 24), Bacteroidetes-Porphyromonadaceae (2.1 ± 5.2), Bacteroidetes-Prevotellaceae (4.5 ± 13.5), Bacteroidetes-Rikenelleaceae (2.0 ± 5.5), Firmicutes-Lactobacillaceae (2.0 ± 8.5), Firmicutes-Enterococcaceae (1.1 ± 6.7), Firmicutes-Clostridiales XIV (3.2 ± 4.7), Firmicutes-Lachnospiraceae (10.5 ± 8.4), Firmicutes-Ruminococcaceae (5.5 ± 6.2), Firmicutes-Veillonellaceae (1.7 ± 2.9), and Proteobacteria-Enterobacteriaceae (2.4 ± 5.9); and (3) N/A
		54.2 ± 11.6	46/16	11.0 ± 4.2	N/A	HCV (21)/alcohol (15)/HCV + alcohol (3)/NASH (15)/others (8)	N/A
Zhang et al[17], 2013, China	N/A	Cirrhosis with MHE (n = 26): Higher of S. salivarius; Chao 1 (276 ± 104)
		53 ± 12	13/13	N/A	Child A: 3; Child B: 16; Child C: 7	AIH (1)/HBV (13)/PBC (7)/PSC (0)/alcohol (5)	N/A
		Cirrhosis without MHE (n = 25): Lower of S. salivarius; Chao 1 (250 ± 139)
		59 ± 11	16/9	N/A	Child A: 12; Child B: 11; Child C: 2	AIH (2)/HBV (12)/PBC (3)/PSC (1)/alcohol (7)	N/A
Bajaj et al[61], 2012, United State	N/A	Cirrhosis with HE (n = 19): Incertae Sedis XIV_Blautia (1.6%), Incertae Sedis XIV_other (1.4%), and Fusobacteriaceae_other (1.1%); N/A
		N/A	N/A	N/A	N/A	N/A	N/A
		Cirrhosis without HE (n = 17): Leuconostocaceae_Leuconostoc (1.0%), Bacteroidales incertae sedis other (1.0%), and Alcaligenaceae other (0.8%); N/A
		N/A	N/A	N/A	N/A	N/A	N/A

MELD: Model for end-stage liver disease; RDP: Ribosomal database project; HE: Hepatic encephalopathy; HBV: Hepatitis B virus; HCV: Hepatitis C virus; PPI: Proton pump inhibitor; AIH: Autoimmune hepatitis; NASH: Nonalcoholic steatohepatitis; PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; ICT: Inhibitory control test, MHE: Minimal hepatic encephalopathy; NCBI; National Center for Biotechnology Information; PHES: Psychometric hepatic encephalopathy score; N/A: Not applicable.

Gut microbiota diversity

The alpha diversity of the gut microbiota can be represented by the Shannon, Simpson, and Chao 1 indices. A total of 7 studies compared the Shannon index between HE and non-HE in cirrhosis (4 studies reported no significant difference between the two groups, and 3 studies reported lower values in cirrhosis with HE), 3 studies reported the Simpson index between HE and non-HE in cirrhosis (1 study reported no significant difference between the two groups, and 2 studies reported lower values in cirrhosis with HE), and 5 studies reported the Chao 1 index between cirrhosis with HE and without HE (1 study reported no significant difference between the two groups, 3 studies reported lower values in cirrhosis with HE, and 1 study reported higher values in cirrhosis with HE), as shown in Figure 2.

Figure 2 Comparison of alpha diversity indices between patients with hepatic encephalopathy and without hepatic encephalopathy in cirrhosis. HE: Hepatic encephalopathy.

Seven studies extracted Shannon data, in one study, three different results were obtained based on different diagnostic criteria for minimal HE (MHE). The final analysis revealed that the Shannon index in cirrhosis patients with HE was significantly lower than that in patients without HE (P < 0.00001), and there was good consistency between studies (I² = 20%), as shown in Figure 3A. Three studies extracted Simpson data, and the analysis results revealed no significant difference between patients with liver cirrhosis with HE and those without HE (P = 0.47), but there was significant heterogeneity among the studies (I² = 76%) as shown in Figure 3B. Given the significant heterogeneity, we conducted a sensitivity analysis by omitting one study at a time. These analyses indicate that the heterogeneity contribution of the Bloom et al[8] to the Simpson meta-analysis was the greatest. When we excluded this study from the analysis, there was still no significant difference in Simpson between the two groups (P = 0.16, I² = 0%, n = 2 studies). As shown in Supplementary Figure 1. Five studies extracted the Chao 1 index, and the results revealed no significant difference in the Chao 1 index between patients with liver cirrhosis with HE and those without HE (P = 0.16), but there was a certain degree of difference among the studies (I² = 66%), as shown in Figure 3C. Given the significant heterogeneity, using the same method as before, Wang et al[9] contributed the most to heterogeneity, and after deletion, heterogeneity significantly decreased (I² = 38%). However, when we excluded this study from the analysis, the Chao 1 index of cirrhosis patients with HE was significantly lower than that of cirrhosis patients without HE (P = 0.004, n = 4 studies), as shown in Supplementary Figure 2.

Figure 3 Alpha diversity indices of patients with liver cirrhosis with or without hepatic encephalopathy. A: The Shannon index results; B: The Simpson index results; C: Chao 1 index results. HE: Hepatic encephalopathy; CI: Confidence interval.

Differences in microbial composition

Based on the linear discriminant analysis effect size or relative abundance data of cirrhosis patients with or without HE mentioned in the literature, the same bacteria are summarized in Table 3. Ruminococcaceae, Lachnospiraceae, Prevotellaceae, and Bacteroidetes tended to decrease in patients with liver cirrhosis with HE. The abundances of Enterococcus, Proteobacteria, Enterococcaceae, and Enterobacteriaceae tended to increase in patients with liver cirrhosis with HE. There were five literature reports on the relative abundance of microbiota at the phylum and family levels, but there were only two studies for each bacterium. Single-group rate analysis was conducted on various bacteria, as shown in Figure 4. The results revealed that at the phylum level in cirrhosis with HE vs cirrhosis without HE, the relative abundance of Firmicutes was 0.497 [95% confidence interval (CI): 0.334-0.661] vs 0.458 (95%CI: 0.211-0.705), that of Bacteroidetes was 0.271 (95%CI: 0.092-0.451) vs 0.335 (95%CI: 0.070-0.600), that of Proteobacteria was 0.172 (95%CI: 0.120-0.224) vs 0.153 (95%CI: 0.103-0.203), and that of Actinobacteria was 0.029 (95%CI: 0.006-0.052) vs 0.015 (95%CI: -0.002 to 0.032), as shown in Figure 4A. At the species level, in cirrhosis with HE vs without HE, the relative abundance of Lachnospiraceae was 0.133 (95%CI: 0.086-0.180) vs 0.235 (95%CI: 0.176-0.294), that of Ruminococcaceae was 0.055 (95%CI: 0.024-0.087) vs 0.093 (95%CI: 0.053-0.133), that of Prevotellaceae was 0.050 (95%CI: 0.007-0.093) vs 0.053 (95%CI: 0.009-0.097), that of Enterobacteriaceae was 0.010 (95%CI: -0.004 to 0.024) vs 0.030 (95%CI: -0.003 to 0.063), that of Bacteroidaceae was 0.297 (95%CI: 0.199-0.396) vs 0.285 (95%CI: 0.202-0.369), and that of Streptococcaceae was 0.013 (95%CI: -0.003 to 0.029) vs 0.018 (95%CI: -0.023 to 0.059). Fusobacteria was 0.010 (95%CI: -0.004 to 0.025) in cirrhosis patients with HE, as shown in Figure 4B. The specific values of the gut microbiota were included in four studies. Lachnospiraceae (-2.73, 95%CI: -4.58 to -0.87, I² = 38%) and Ruminococcaceae (-2.93, 95%CI: -4.29 to -1.56, I² = 0%) at the family level were significantly lower in patients with liver cirrhosis complicated with HE compared to those without HE. Veillonella at the genus level showed no significant difference between the two groups (3.02, 95%CI: -0.72 to 6.77, I² = 43%) (Figure 5).

Figure 4 Comparison of the relative abundance of bacteria between cirrhosis patients with or without hepatic encephalopathy. A: Comparison at the phylum level; B: Comparison at the species level. HE: Hepatic encephalopathy; CI: Confidence interval.

Figure 5 Comparison of the gut microbiota between patients with liver cirrhosis with or without hepatic encephalopathy. A: Comparison of Lachnospiraceae; B: Comparison of Ruminococcaceae; C: Comparison of Veillonella. HE: Hepatic encephalopathy; CI: Confidence interval.

Table 3 Differential microbiota between liver cirrhosis with and without hepatic encephalopathy.

Bacterium	Ref.1	Ref.2
Ruminococcaceae	/	[8,12,13,16,56,60]
Lachnospiraceae	[57,58]	[8,12-14,16,60]
Clostridiales XIV	/	[13,16]
Prevotellaceae	/	[13,16,60]
Bacteroidaceae	[13,16,60]	[8,14,56]
Eubacteriaceae	[14]	[14,18]
Bacteroidetes	[11]	[9,14-16]
Lactobacillus	[15,59]	/
Enterococcus	[9,15,59]	/
Veillonella	[15,59]	[58]
Lactobacillaceae	[16,60]	/
Veillonella	[15,59]	[58]
Ruminococcus	[57]	[9]
Firmicutes	[11,15]	[9,16]
Proteobacteria	[9,11,15,16]	/
Coprococcus	[57,58]	/
Actinobacteria	[9,15]	/
Bacteroides	[15]	[9]
Prevotella	[15]	[9]
Escherichia-shigella	[9,59]	/
Erysipelotrichales	[55]	[8]
Enterococcaceae	[8,13,16]	/
Enterobacteriaceae	[8,14,16,60]	[13]
Streptococcaceae	[14,60]	/
Fusobacteriaceae	[13]	[12]
Pasteurellaceae	[58]	[14]
Pasteurellales	[58]	[14]
Veillonellaceae	[16]	[8]

¹Higher in cirrhosis with hepatic encephalopathy compared with cirrhosis without hepatic encephalopathy.

²Lower in cirrhosis with hepatic encephalopathy compared with cirrhosis without hepatic encephalopathy.

HE: Hepatic encephalopathy.

Impact of drugs on the gut microbiota

The subjects in the 7 studies had previously used proton pump inhibitors (PPIs), ranging from 7% to 94%, but only one study reported the effect of PPIs on the gut microbiota of patients with cirrhosis[10]. Cirrhosis patients without HE who used PPIs experienced an increase in the abundance of Streptococcus salivarius (S. salivarius) compared with that in liver cirrhosis patients with HE who used PPIs. Three studies reported the effects of rifaximin or lactulose on the gut microbiota. A comparison of the gut microbiota before and after using rifaximin in patients with liver cirrhosis and HE revealed no significant change in the total amount of fecal bacteria[10]. However, compared with patients who did not respond to rifaximin, patients who responded to rifaximin had significantly lower S. salivarius levels. Moreover, lactulose increased the abundance of L. salivarius[10]. Yu et al[11] reported that the use of rifaximin affected the abundance of the gut microbiota in patients with cirrhosis, leading to a decrease in the abundance of Clostridium perfringens, and an increase in the number of anti-inflammatory bacteria such as Faecalibacterium prausnitzii. However, rifaximin had no significant effect on the diversity of the gut microbiota in patients with liver cirrhosis without HE and had a slight effect on the diversity of the gut microbiota in patients with liver cirrhosis with HE; however, a clear trend was not detected. Overall, rifaximin can prevent HE by regulating the abundance of different bacterial strains in the gut microbiota, reducing the abundance of harmful bacteria, and maintaining the abundance of beneficial bacteria. Bajaj et al[12] reported the effect of lactulose on the gut microbiota and reported that after lactulose was discontinued in patients with liver cirrhosis with HE, the abundance of Faecalibacterium spp. decreased (6% vs 1%, P = 0.026), and that of Veillonella spp. also decreased (2% vs 0%, P = 0.07), but that of other bacteria did not significantly change.

Differences in the metabolism of the fecal microbiota (fatty acids and amino acids)

Yu et al[11] analyzed the differential metabolic pathways of the gut microbiota in cirrhosis patients with and without HE that were involved mainly in the synthesis of phosphopeptidic acid, peptidoglycans, amino acids, and fatty acids; the degradation of alcohols, rockulose and rhamnose; and the synthesis and degradation of coenzymes. Wang et al[9] also analyzed metabolites in feces and reported no difference in the concentration of short-chain fatty acids between the HE and non-HE groups. However, the average concentrations of acetic acid, propionic acid, butyric acid, and nonanoic acid in the feces of cirrhosis patients with HE is lower than those in normal patients. Moreover, the concentration of tryptophan in patient feces was detected, and the level of indole-3-acetic acid - aminolevulinic acid in the feces of the HE group was significantly lower than that in the feces of the cirrhosis without HE group. There were significant differences in the levels of short-chain fatty acid and tryptophan metabolites in fecal between cirrhosis patients without HE and normal patients. Bloom et al[8] reported that the concentration of short-chain fatty acids in patients with liver cirrhosis with HE was lower than that in patients without HE. Moreover, the gut microbiota is related to metabolites, such as a negative correlation between Enterococcus and acetic acid, propionic acid, isobutyric acid, and butyric acid[9]. Bajaj et al[13] reported that the microbial community in the feces of patients with cirrhosis may be related to vitamin and nucleic acid metabolism.

Gut microbiota and metabolism for the differential diagnosis and prediction of prognosis

Wang et al[9] reported that metabolites such as acetic acid, propionic acid, and butyric acid can be used to distinguish between the HE and healthy groups, with area under the curve values of 0.923, 0.917, and 0.865, respectively. It can also be used to distinguish between the cirrhosis and the healthy groups, with area under the curve values of 0.947, 0.89, and 0.927, respectively. Bajaj et al[14] reported via logistic regression that the Lachnospiraceae genus (Ruminococcus and Clostridium XIVb) in feces was associated with good cognition but not with clinical variables. Sung et al[15] reported that Bacteroides, Clostridium XI, and others are associated with the recurrence and overall survival of HE patients within 1 year in patients with liver cirrhosis. Another study by Bloom et al[8] revealed that eight bacterial species, including Clostridium species and Ruminococcus species, were associated with a history of HE. Yu et al[11] reported that Prevotellaceae, Ruminococcaceae, and Lachnospiraceae can reduce hospitalization risk regardless of the presence of model for end-stage liver disease (MELD) and ascites. Prevotellaceae can also reduce the risk of MHE. Ahluwalia et al[16] reported that Lachnospiraceae, Ruminococcaeae, and Incertae Sedis XIV were negatively correlated with MELD. Enterococcus was positively correlated with the MELD score. Zhang et al[17] reported that S. salivarius may become a potential biomarker for ammonia lowering therapy in patients with liver cirrhosis and MHE.

DISCUSSION

HE is still a common complication and cause of death in patients with end-stage liver disease, affecting the quality of life of patients and their families. HE can be divided into covert HE and overt HE (OHE), with covert HE including MHE[2]. Compared with patients without MHE, cirrhosis patients with MHE have more frequent episodes of OHE[18], and the mortality rate of OHE patients is also greater. Without liver transplantation, the 1-year survival rate after the first episode is 42%, and the 3-year survival rate is 23%[19]. The gut-liver axis has become a focus of chronic liver disease, prompting more research on the role of the gut microbiota in cirrhosis[20]. In patients with cirrhosis, changes in the structure and function of the gut microbiota are closely related to clinical prognosis[5,21,22] and are also closely associated with the occurrence of HE[6,23-25]. Changes in the gut-liver-brain axis are considered core pathogenic factors in the cognitive impairment associated with liver cirrhosis. Most treatments related to HE focus on the gut[14,26,27].

Therefore, this study searched for studies on the relationship between the gut microbiota and HE. During the screening process, two studies were excluded because one focused on patients with alcoholic hepatitis and the other focused on patients with chronic acute liver failure. Eventually, 17 studies were included. First, the data on the alpha diversity of the gut microbiota were analyzed, and 7 studies reported the Shannon index. A meta-analysis revealed that it was significantly reduced in patients with liver cirrhosis with HE, and the heterogeneity was small. After the Chao 1 and Simpson indices were compared, there was no significant difference between the two groups, and there was a significant heterogeneity, which may be attributed to the limited research reporting on these two indices.

Owing to the lack of detailed information on the quantitative relative abundance of various gut bacteria between patients with liver cirrhosis with HE and those without HE in some studies, data on linear discriminant analysis effect size were available. Therefore, we summarized the gut microbiota mentioned in the included literature. Several intestinal bacteria, including Ruminococcaceae, Lachnospiraceae, Prevotellaceae, and Bacteroidetes, which showed a decreasing trend in patients with liver cirrhosis with HE, were consistent across multiple studies. The abundances of Enterococcus, Proteobacteria, Enterococcaceae, and Enterobacteriaceae increased in HE patients with liver cirrhosis. Afterwards, we analyzed the intestinal bacteria that could extract accurate data and found that Lachnospiraceae (-2.72, 95%CI: -4.58 to -0.87, I² = 38%) and Ruminococcaceae (-2.93, 95%CI: -4.29 to -1.56, I² = 0%) showed lower levels in liver cirrhosis patients with HE at the family level, but Veillonella at the genus level was not significantly different between the two groups (3.02, 95%CI: -0.72 to 6.77, I² = 43%).

Lachnospiraceae, Ruminococcaceae, and Clostridiales XIV are beneficial local taxa that can produce short-chain fatty acids and increase the content of Bifidobacterium and Lactobacillus bacteria, maintaining the integrity of the intestinal barrier. In previous studies, these bacteria were generally found to be relatively abundant in patients with good cognition[28]. Other studies have also shown that the use of materials from healthy donors rich in Lachnospiraceae and Ruminococcaceae for fecal transplantation can effectively reduce the recurrence of HE and the occurrence of serious adverse events[29], which are negatively correlated with the Child-Pugh score[30] and increase the level of short-chain fatty acids in plasma[31]. Therefore, increasing these beneficial bacteria can contribute to the overall improvement of cognitive function and may prevent hospitalization related to the liver. However, these bacteria were reduced in patients with liver cirrhosis with HE, and their reduction also led to relative overgrowth of potential pathogenic taxa. Enterobacteriaceae and Enterococcus are increased in patients with liver cirrhosis with HE, and they increase intestinal permeability, which is related to disease progression and endotoxemia[5,30,32]. Moreover, Fusobacteriaceae, Veillonellaceae, and Enterobacteriaceae are positively correlated with inflammation[33].

With respect to the metabolic differences between HE patients and non-HE patients with in cirrhosis, all three articles mentioned short-chain fatty acids, but there were some differences in the research results. Previous studies have shown that cirrhosis patients have a decreased ability to produce butyric acid bacteria[34]. Based on the previous analysis of the differences in the gut microbiota between cirrhosis patients with HE and without HE, bacteria that produce short-chain fatty acids are reduced in liver cirrhosis patients with HE, and supplementation with bacteria that produce short-chain fatty acids can help improve cognitive function. Therefore, it was speculated that the level of short-chain fatty acids was lower in patients with liver cirrhosis with HE. The addition of short-chain fatty acids to strengthen the intestinal barrier and reduce the relative abundance of potential pathogenic taxa may be used to treat HE. Some studies have also shown that the gut microbiota is influenced by bile acid metabolism[35].

Owing to the inclusion of very few studies on the impact of PPIs on the gut microbiota, no meta-analysis has been conducted. The literature mentioned that after applying PPIs, the abundance of S. salivarius increased, and this bacterium contained urease operons encoding ammonia-producing enzymes. This may be due to reduced gastric acid secretion in liver cirrhosis, making it easier for S. salivarius to transfer from the mouth to the intestine through the weakened gastric acid barrier[36-38]. Previous studies revealed that the use of PPIs increased the abundance of Streptococcaceae in both cirrhotic and noncirrhotic patients[4,39,40], and that certain Streptococcal species can cause spontaneous bacterial peritonitis[41]. A meta-analysis revealed a decrease in Ruminococcaceae and Lachnospiraceae after the application of PPIs[42], considering the long-term inhibition of gastric acid secretion and disruption of the gastric acid barrier, which inhibited the growth of bacteria producing butyric acid[43]. The use of PPIs may increase the risk of HE in patients with cirrhosis[44].

Previous studies have shown that lactulose and rifaximin have almost no effect on the gut microbiota[45-47]. In mouse studies, rifaximin did not affect the composition of the gut microbiota but rather beneficially altered the production of intestinal ammonia by regulating the expression of intestinal glutamine enzymes[48]. However, the two articles included in this study revealed that rifaximin can affect the abundance of certain intestinal bacteria to some extent, such as reducing Veillonellaceae and increasing Eubacteraceae[47]. Overall, the improvement in cognitive function and reduction in endotoxemia caused by rifaximin were beneficial to the regulation of the metabolic characteristics of the gut microbiota rather than a significant adjustment of the gut microbiota[49]. The effect of lactulose on the fecal microbiota depends on both the patient’s health status and the dosage and duration of lactulose use[50].

Although some studies have suggested that the gut microbiota is relatively stable within 6 months[5], the influence of the gut microbiota is significant, and there are significant differences in reported microbiota in the literature, with some inconsistencies in research results. However, several bacteria still presented consistent trends, indicating that the gut microbiota may become a biomarker for predicting the occurrence and prognosis of liver cirrhosis with HE[51-54]. Some studies have also suggested that these microbial changes are not related to the diagnostic criteria for HE; however, the existence of specific bacterial taxa is related to the understanding of this population[14]. This study has limitations in terms of heterogeneity: The number of included studies was limited, and the data that can be extracted for quantitative analysis were also limited. Therefore, subgroup analysis was not conducted on the basis of demographic characteristics such as race, age, or degree of HE.

CONCLUSION

The diversity of the gut microbiota in patients with liver cirrhosis with HE was reduced. There are characteristic changes in the gut microbiota in patients with liver cirrhosis with HE, and these changes in the gut microbiota composition and function can help predict the occurrence of HE and predict the prognosis of patients to some extent. However, there were significant differences between studies, and more large-scale studies are needed.

ACKNOWLEDGEMENTS

The authors would like to thank all the individuals who participated in this study.