Clinical characteristics of the study population

In this study, we analyzed the gut microbiota of 56 healthy controls and 106 patients with HBV-related liver diseases. Among these patients, 14 had resolved HBV, 58 had CHB, and 34 had advanced liver disease (15 had LC, whereas 19 had HCC). Of the patients with HCC, 12 (63%) had a history of LC. The baseline characteristics of the participants are summarized in Table 1. The median age was 28 years for healthy controls, 57 years for resolved HBV patients, 51 years for CHB patients, and 62 years for advanced liver disease patients (P < 0.0001); the corresponding proportions of men were 32%, 50%, 66%, and 82% (P < 0.0001), respectively. In general, the levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), blood urea nitrogen, creatinine, and fasting blood glucose were higher in patients with resolved HBV, those with CHB, and those with advanced liver disease than in healthy controls (all P < 0.0001). Patients with advanced liver disease had lower platelet counts than did those with resolved HBV or CHB (P = 0.0019).

Table 1 Demographics and clinical characteristics of the study populationOverview of the sequencing data

The 16S ribosomal RNA (rRNA) gene V3–V4 amplicon sequencing yielded 15,801,571 reads, and the average number of sequence reads per sample was 97,540 (minimum: 46,986; maximum: 183,523). To account for sample size–related variations, rarefaction was performed to normalize the reads to 46,986 reads per sample. A total of 7,504 amplicon sequence variants (ASVs) were retained after rarefaction.

Differences in gut microbiome structures

The common and unique ASVs among the different stages of liver diseases and the number of observed ASVs are presented in Fig. 1A, B. Alpha diversity matrices were calculated on the basis of rarefied ASVs. The estimated values of Simpson and Shannon diversities revealed higher degrees of bacterial richness and evenness in patients with resolved HBV than in healthy controls (Fig. 1C, D). Chao1 and Chao2 indices indicated a higher degree of species richness in patients with HBV-related liver disease than in healthy controls (all P < 0.05; Fig. 1E, F).

Fig. 1

Comparisons of bacterial diversity and richness between healthy controls and patients with resolved HBV, CHB, and advanced liver disease. A A Venn diagram displays the unique and shared ASVs among the four groups. B The healthy controls harbored the lowest observed ASVs. Alpha diversity indices including C Shannon diversity, D Simpson diversity, E Chao1 index, and F Chao2 index revealed the differences in richness and evenness between groups. * means P < 0.05 and ** means P < 0.01

Principal coordinate analysis (PCoA) performed using weighted and unweighted UniFrac distances revealed distinct clusters between healthy controls and the aforementioned three patient groups (Fig. 2A, B). The adonis2 function, which was used to evaluate between-group dissimilarities and adjust for the possible confounding effect of age in the multivariate model, revealed a nonsignificant effect of age (P = 0.671 for weighted UniFrac distance and P = 0.146 for unweighted UniFrac distance) on microbiota composition compared with the significant effect of disease status (P = 0.001 for both distances). Pairwise comparisons performed using the permutational multivariate analysis of variance (PERMANOVA) indicated significant differences in bacterial composition between patients with resolved HBV and healthy controls, between patients with CHB and healthy controls, and between patients with advanced liver disease and healthy controls (all q < 0.05 for weighted and unweighted UniFrac distances; Additional file 1). The composition of fecal microbiota communities was relatively similar among the three patient groups.

Fig. 2

Beta diversity indices of the fecal microbiota. A PCoA plot of bacterial beta diversity based on the weighted UniFrac distance B PCoA plot of bacterial beta diversity based on the unweighted UniFrac distance. Beta diversity analyses revealed a distinct clustering pattern between healthy controls and patients with HBV-related liver disease

Bacterial abundance and taxonomic distribution

The most abundant phyla in the study cohort were Bacteroidetes (44.7%), Firmicutes (41.4%), Proteobacteria (6.0%), Actinobacteriota (5.5%), and Verrucomicrobiota (1.0%). The most abundant genera were Bacteroides (30.3%), Prevotella (6.5%), Bifidobacterium (3.4%), Faecalibacterium (3.3%), Blautia (3.1%), Parabacteroides (3.1%), Escherichia-Shigella (2.6%), and Megamonas (2.0%). The taxonomic distribution of the predominant bacteria among the study groups is shown in Fig. 3A–E. The ratio of Firmicutes to Bacteroidetes (F/B ratio) was higher in all three patient groups than in healthy controls (all P < 0.001; Fig. 3F).

Fig. 3

Distribution of the predominant bacterial taxa at different taxonomic levels. A, B, C, D, E represent phylum, class, order, family, and genus levels, respectively. F The ratio of Firmicutes to Bacteroidetes (F/B ratio) was increased in patients with resolved HBV, CHB, and advanced liver disease compared to healthy controls. *** means P < 0.001

Differences in gut microbiota compositions

Linear discriminant analysis (LDA) effect size (LEfSe) was used to identify specific taxonomic biomarkers for the study groups. The cutoff value of the logarithmic linear discriminant analysis (LDA) score was set at 3.0 to determine the major taxonomic differences between the following groups: (1) healthy controls and patients with resolved HBV, (2) healthy controls and patients with CHB, (3) healthy controls and patients with advanced liver disease, (4) patients with CHB and those with resolved HBV, and (5) patients with CHB and those with advanced liver disease. Significant differences were observed between healthy controls and patients with HBV-related liver disease (Fig. 4A–E).

Fig. 4

Differentially abundant genera between groups identified by LEfSe (only logarithmic LDA scores > 3.0 are shown). A Resolved HBV vs. Healthy controls. B Chronic hepatitis B vs. Healthy controls. C Advanced liver disease vs. Healthy controls. D Resolved HBV vs. Chronic hepatitis B. E Advanced liver disease vs. Chronic hepatitis B

Regarding phylum-level comparisons, Firmicutes, Verrucomicrobiota, and Fusobacteria were enriched and Bacteroidetes was depleted in the three patient groups compared with the findings in healthy controls. Furthermore, Proteobacteria and Actinobacteriota were enriched in patients with advanced liver disease. The abundance of Proteobacteria was higher in patients with advanced liver disease than in those with CHB (Additional file 2).

Genus-level comparisons between patients with resolved HBV and healthy controls (Fig. 4A) revealed that the most enriched genera in patients with resolved HBV were Veillonella (LDA score = 4.08; P < 0.001), Blautia (LDA score = 4.05; P = 0.003), Akkermansia (LDA score = 4.01; P = 0.004), Streptococcus (LDA score = 3.74; P < 0.001), and Clostridia UCG-014 (LDA score = 3.73; P = 0.017), whereas the most enriched genera in healthy controls were Bacteroides (LDA score = 4.94; P = 0.004), Parabacteroides (LDA score = 4.04; P = 0.008), and Sutterella (LDA score = 3.84; P < 0.001).

Genus-level comparisons between patients with CHB and healthy controls (Fig. 4B) revealed that the most enriched genera in patients with CHB were Streptococcus (LDA score = 4.15; P < 0.001), Blautia (LDA score = 4.01; P = 0.001), Veillonella (LDA score = 3.87; P < 0.001), Fusobacterium (LDA score = 3.87; P = 0.007), and Akkermansia (LDA score = 3.66; P = 0.012), whereas the most enriched genera in healthy controls were Bacteroides (LDA score = 4.89; P < 0.001), Megamonas (LDA score = 4.17; P = 0.005), Parabacteroides (LDA score = 3.95; P < 0.001), Sutterella (LDA score = 3.65; P = 0.001), and Lachnoclostridium (LDA score = 3.44; P = 0.005).

Genus-level comparisons between patients with advanced liver disease and healthy controls (Fig. 4C) revealed that the most enriched genera in patients with advanced liver disease were Escherichia-Shigella (LDA score = 4.26; P = 0.018), Veillonella (LDA score = 4.13; P < 0.001), Streptococcus (LDA score = 4.10; P < 0.001), Blautia (LDA score = 4.04; P < 0.001), and Collinsella (LDA score = 3.96; P = 0.002), whereas the most enriched genera in healthy controls were Bacteroides (LDA score = 4.87; P = 0.001), Prevotella (LDA score = 4.56; P < 0.001), Megamonas (LDA score = 4.09; P = 0.003), Parabacteroides (LDA score = 3.98; P < 0.001), and Faecalibacterium (LDA score = 3.92; P = 0.002).

Genus-level comparisons between patients with resolved HBV and those with CHB (Fig. 4D) revealed that Clostridium sensu stricto 1 (LDA score = 3.39; P = 0.012) and Romboutsia (LDA score = 3.20; P = 0.041) were enriched in patients with resolved HBV, whereas no taxon was enriched in patients with CHB.

Finally, genus-level comparisons between patients with CHB and those with advanced liver disease (Fig. 4E) revealed that Prevotella (LDA score = 4.36; P = 0.013), Faecalibacterium (LDA score = 3.81; P = 0.026), Dialister (LDA score = 3.44; P = 0.003), and Eubacterium ruminantium group (LDA score = 3.15; P = 0.039) were enriched in patients with CHB, whereas Lactobacillus (LDA score = 3.60; P = 0.028) and CAG-352 (LDA score = 3.03; P = 0.027) were enriched in patients with advanced liver disease.

Fig. 5 shows the mean relative abundance of the following 11 selected genera: Bacteroides, Megamonas, Parabacteroides, and Sutterella, which were enriched in healthy controls; Akkermansia, Blautia, Lactobacillus, Streptococcus, and Veillonella, which were enriched in the patient groups (all P for trend < 0.01); and Clostridium sensu stricto 1 and Romboutsia, which had higher abundances in patients with resolved HBV than in those with CHB.

Fig. 5

Mean relative abundance of selected genera

Functional analysis based on 16S rRNA sequencing data

We identified 45 metabolic functions that were classified as Level 2 of the Kyoto Encyclopedia of Genes and Genomes (KEGG) Orthologs (Additional file 3). A metagenome function prediction analysis performed using the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2 (PICRUSt2) tool revealed alterations in bacterial functions with disease progression. A total of 28 differentially expressed pathways were identified (Additional file 4). A heatmap was generated to visualize the differential expression levels of Level 2 KEGG pathways (Fig. 6). The pathways enriched with disease progression were involved in membrane transport, cancer, bacterial infection, viral infection, transcription, endocrine and metabolic disease, and cell motility. By contrast, pathways enriched in healthy controls were involved in cell growth and death, endocrine system, carbohydrate, glycan, lipid, terpenoids, and polyketides metabolism, transport and catabolism, energy metabolism, circulatory system, nervous system, and immune disease. Notably, pathways involved in the metabolism of cofactors and vitamins and the development of antineoplastic drug resistance exhibited the highest levels of enrichment in patients with resolved HBV.

Fig. 6

Differences of predicted microbial functions with PICRUSt2 analysis based on 16S rRNA sequencing. The microbial functions were clustered based on the similarities of expression pattern

Subgroup analysis in patients with CHB

Patients with CHB were further divided into the following four groups (representing the four phases of HBV natural history) according to the presence of the hepatitis B e antigen (HBeAg) and the level of ALT: HBeAg (+) chronic HBV infection, HBeAg (+) chronic hepatitis B, HBeAg (−) chronic HBV infection, and HBeAg (−) chronic hepatitis B. The alpha and beta diversities were compared between the four groups and patients with resolved HBV. The values of the alpha diversity indices, such as Shannon diversity, Simpson diversity, Chao1, and Chao2, and the number of observed ASVs were lower in patients with HBeAg (−) chronic hepatitis B (Additional file 5). PCoA plots generated on the basis of weighted and unweighted UniFrac distances revealed no distinct separation between the aforementioned four groups and patients with resolved HBV (all P > 0.05; Additional file 6 and Additional file 7).