Multi-omics analysis reveals gut microbiota-metabolite interactions and their association with liver function in autoimmune overlap syndrome

Abstract

BACKGROUND

Autoimmune liver diseases, including primary biliary cholangitis (PBC), autoimmune hepatitis (AIH), and their overlap syndrome (OS), involve immune-mediated liver injury, with OS occurring in 1.2%-25% of PBC patients. OS carries a higher risk of cirrhosis, hepatocellular carcinoma, and reduced survival. While its pathogenesis remains unclear, gut microbiota dysbiosis and serum metabolite alterations may play key roles. This study uses 16S rRNA sequencing and liquid chromatography-mass spectrometry (LC-MS) metabolomics to compare gut microbiota and serum metabolites among PBC, AIH, and OS patients, and explores their associations with liver function.

AIM

To differentiate OS from PBC and AIH based on gut microbiota, serum metabolites, and liver function.

METHODS

Gut microbiota profiles were analyzed using 16S rRNA sequencing, while untargeted serum metabolomics was conducted via LC-MS. Comparative analyses were performed to identify differences in microbial composition and serum metabolite levels among PBC, AIH, and OS groups. Correlation analyses and network visualization techniques were applied to elucidate the interactions among liver function parameters, gut microbiota, and serum metabolites in OS patients.

RESULTS

Compared to patients with PBC or AIH, OS patients demonstrated significantly reduced microbial diversity and richness. Notable taxonomic shifts included decreased abundances of Firmicutes, Bacteroidetes, and Actinobacteria, alongside increased levels of Proteobacteria and Verrucomicrobia. Distinct serum metabolites, such as pentadecanoic acid and aminoimidazole carboxamide ribonucleotide, were identified in OS patients. Correlation analysis revealed that aspartate aminotransferase (AST) levels were negatively associated with the bacterial genus Fusicatenibacter and the metabolite L-Tyrosine. A microbial-metabolite network diagram further confirmed a strong association between Fusicatenibacter and L-Tyrosine in OS patients.

CONCLUSION

OS patients show decreased gut microbiota diversity and unique serum metabolites. Multi-omics linked AST, Fusicatenibacter, and L-Tyrosine, revealing OS mechanisms and diagnostic potential.

Key Words: Overlap syndrome; Multi-omics; Gut microbiomes; Metabolites; Liver function

Core Tip: This study characterizes the distinct gut microbiome and serum metabolite profiles in overlap syndrome (OS), a severe condition combining features of primary biliary cholangitis (PBC) and autoimmune hepatitis (AIH). These findings provide the first comprehensive evidence linking gut microbiota dysbiosis with specific metabolic alterations in OS, offering new insights into disease mechanisms and potential diagnostic biomarkers for distinguishing OS from PBC or AIH alone. The microbial-metabolite network discovered in this study may open new avenues for therapeutic interventions targeting the gut-liver axis in autoimmune liver diseases.

INTRODUCTION

Autoimmune liver disease (AILD) encompasses a group of chronic liver disorders caused by aberrant immune responses against hepatic or biliary antigens. The three major types of AILD are primary biliary cholangitis (PBC), autoimmune hepatitis (AIH), and primary sclerosing cholangitis[1]. These conditions share a common pathological hallmark: Immune-mediated injury to hepatocytes and/or bile duct epithelium, leading to cholangitis, hepatocellular necrosis, and consequent clinical manifestations such as jaundice and abnormal liver function. These pathological changes severely compromise patients’ quality of life[2-4]. In certain patients, features of two distinct AILDs may coexist, a condition known as overlap syndrome (OS), with the PBC-AIH variant being the most frequently observed[5].

Currently, due to the absence of large-scale epidemiological studies, data on the prevalence of PBC-AIH OS remain limited. A review of studies[6] published between 1998 and 2013, which applied varying diagnostic criteria and retrospective methodologies, reported that the prevalence of PBC-AIH OS among PBC patients ranged from 1.2% to 25%. The clinical course of OS is typically chronic and progressive. Without timely intervention, patients are at increased risk of developing liver cirrhosis or hepatocellular carcinoma. Additionally, OS is associated with a higher incidence of complications such as esophageal varices, gastrointestinal hemorrhage, and ascites, and is characterized by a relatively low 5-year survival rate[7,8].

While current research has focused primarily on the clinical characteristics, therapeutic strategies, and outcomes of PBC-AIH OS[9,10], its precise pathogenesis remains poorly understood. It is likely multifactorial, involving aberrant immune responses, genetic predisposition, and environmental triggers. Increasing evidence suggests that both AIH and PBC are associated with intestinal microbiota dysbiosis, commonly presenting with reduced bacterial diversity[11]. This implies a potentially critical role for the gut-liver axis in the pathophysiology of OS. Furthermore, microbial-derived metabolites are emerging as key mediators in modulating host immunity and may play a central role in host-microbiome interactions. Despite these insights, our understanding of OS remains incomplete. To further distinguish OS from PBC and AIH and to elucidate potential pathogenic mechanisms, this study employs 16S rRNA gene sequencing and liquid chromatography-mass spectrometry (LC-MS)-based metabolomics. Specifically, we aim to identify distinct microbial and metabolic profiles among patients with PBC, AIH, and OS, and to explore the potential interrelationships between gut microbiota, serum metabolites, and liver function parameters in OS patients.

MATERIALS AND METHODS

Study subjects and sample collection

This study was conducted at Beijing Youan Hospital, Capital Medical University, between September 2019 and September 2022. A total of 32 patients were enrolled, including 16 with PBC, 9 with AIH, and 7 with PBC-AIH OS.

The diagnosis of PBC was established according to the 2017 European Association for the Study of the Liver (EASL) Clinical Practice Guidelines[12] and the 2021 Asian Pacific Association for the Study of the Liver Clinical Practice Guidelines[13]. Diagnostic criteria included the following: (1) Clinical features: PBC primarily affects adult females and presents with cholestatic symptoms such as fatigue, pruritus, jaundice, or facial flushing, often persisting for months or years; (2) Laboratory findings: Persistent elevation of serum alkaline phosphatase (ALP) and gamma-glutamyl transferase (GGT), with mild to moderate increases in serum bilirubin; and (3) Histological features: Liver biopsy typically reveals non-suppurative cholangitis, with inflammatory infiltration surrounding the interlobular bile ducts. Additional findings may include bile duct injury, ductular reaction, and bile stasis. A diagnosis of PBC was made when patients fulfilled at least two of the above three criteria.

The diagnosis of AIH was based on the criteria established by the International AIH Group[14]. The diagnostic framework includes the following components: (1) Clinical presentation: Patients commonly experience persistent fatigue, malaise, right upper quadrant discomfort, and occasionally left upper abdominal pain. Additional symptoms may include anorexia, weight loss, esophageal and gastric varices, and hepatosplenomegaly; (2) Laboratory findings: Serum levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) are persistently or recurrently elevated. Immunological markers such as antinuclear antibody, anti-smooth muscle antibody, anti-liver kidney microsome type 1 antibody, and anti-soluble liver antigen are frequently positive; and (3) Histopathology: Liver biopsy typically reveals characteristic features such as interface hepatitis, hepatocyte necrosis, and lymphoplasmacytic infiltration. Other liver diseases must be excluded, particularly those with lesions restricted to small lobular areas.

Currently, there is no universally accepted diagnostic criterion for OS. However, diagnosis generally relies on a combination of clinical, laboratory, and histological features based on established criteria for both PBC and AIH. In this study, we adopted the Paris criteria[15,16], which are widely used and recognized by the EASL and the American Association for the Study of Liver Diseases. According to these criteria, OS is defined by the presence of at least two out of three key diagnostic criteria for both PBC and AIH. Additionally, the presence of interface hepatitis on liver histology is essential to confirm the diagnosis of OS.

Fresh morning fecal samples (approximately 3 g) were collected from all participants and immediately transferred into microbial preservation tubes containing stabilization buffer to ensure sample integrity. Concurrently, 3-5 mL of fasting venous blood was drawn from each patient. Blood samples were placed in sterile centrifuge tubes and centrifuged at 3000 rpm for 10 minutes. The supernatant was carefully transferred to sterile Eppendorf (EP) tubes and centrifuged again to remove residual cell debris and particulates, yielding purified serum for subsequent metabolomic analysis.

This study was approved by the Ethics Committee of Beijing Youan Hospital, Capital Medical University (approval No. LL-2024-026-K) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to inclusion.

Microbial characteristics analysis

Total genomic DNA was extracted from fecal samples using the QIAamp DNA stool mini kit (Qiagen, Germany). Extracted DNA was stored at -20 °C until further analysis. For amplification of the bacterial 16S rRNA gene, polymerase chain reaction (PCR) was performed in 20 μL reaction volumes, each containing 30 ng of genomic DNA, 0.4 μL of each fusion primer (341F: 5’-CCTACGGGNGGCWGCAG-3’; 806R: 5’-GGACTACHVGGGTWTCTAAT-3’), and 12.5 μL of 2 × Taq PCR Master Mix (Vazyme Biotech, China), with double distilled water added to volume. The thermal cycling conditions were as follows: Initial denaturation at 94 °C for 30 seconds; 42 cycles of denaturation at 95 °C for 5 seconds, annealing at 55 °C for 15 seconds, and extension at 72 °C for 10 seconds, followed by a final extension at 72 °C for 5 minutes. Amplification products (approximately 460 bp) were verified by 1.5% agarose gel electrophoresis[17]. PCR products were purified using Agencourt AMPure XP magnetic beads and eluted in the buffer. The sequencing libraries were constructed, labeled, and assessed for fragment size and concentration using the Agilent 2100 Bioanalyzer. Libraries meeting quality criteria were sequenced on the Illumina HiSeq 2500 platform.

Raw sequencing data underwent quality control procedures, including adapter trimming, filtering, deduplication, base correction, and chimera removal, resulting in high-quality sequences (clean data) for downstream analysis. Paired-end reads were merged into Tags based on sequence overlap, which were then clustered into operational taxonomic units (OTUs). Taxonomic classification was performed using the RDP Classifier (v2.13) against the SILVA 138 database (release 138.1), using an 80% confidence threshold. Parallel annotation was conducted using the Greengenes database (version 13_8) with BLASTn (e value < 1 × 10^-5; identity > 97%). Only taxonomic assignments with complete hierarchical classification from Kingdom to genus were retained. Alpha diversity metrics and Venn diagrams were used to assess species richness and within-sample microbial diversity. Beta diversity and between-group differences were visualized using dimensionality reduction techniques such as principal coordinates analysis and sample clustering dendrograms. To compare microbial community structures among disease groups, OTUs were classified, and relative abundances were visualized using bar plots at the phylum and genus levels. Taxa with an average relative abundance below 0.5% or unannotated taxa at a given level were grouped into “others” for visualization. Differential taxa between groups were identified using the Wilcoxon rank-sum test, and P values were calculated accordingly. Sankey diagrams were constructed to illustrate the hierarchical relationships between taxa at the phylum and genus levels in OS patients. OTU data were further transformed into a matrix format suitable for redundancy analysis (RDA), performed using the vegan package (v2.6-10) in R[18], to assess the influence of environmental variables on microbial composition. Finally, Spearman correlation analysis was used to explore associations between dominant microbial taxa and liver function parameters, and the results were visualized using a heatmap generated in R[19].

LC-MS for serum metabolomics detection

For serum metabolomics analysis, 100 μL of each serum sample was transferred into EP tubes and resuspended in pre-chilled 80% methanol containing 0.1% formic acid. Samples were vortexed thoroughly, incubated on ice for 5 minutes, and subsequently centrifuged at 15000 g at 4 °C for 20 minutes. A portion of the resulting supernatant was diluted with LC-MS grade water to reach a final concentration of 53% methanol. The diluted supernatant was transferred to a fresh EP tube and subjected to a second centrifugation under the same conditions. The final supernatant was then used for LC-MS/MS analysis.

Ultra-high performance LC-MS/MS analysis was conducted using a Vanquish Ultra-high performance LC system (Thermo Fisher Scientific, Germany) coupled to an Orbitrap Q ExactiveTM HF-X mass spectrometer (Thermo Fisher Scientific, Germany) at Novogene Co., Ltd. (Beijing, China). Chromatographic separation was achieved on a Hypersil GOLD C18 column (100 mm × 2.1 mm, 1.9 μm) with a 17-minute linear gradient and a flow rate of 0.2 mL/minute. For analysis in positive ion mode, the mobile phases consisted of eluent A (0.1% formic acid in water) and eluent B (methanol). For negative ion mode, eluent A was 5 mmol/L ammonium acetate (potential of hydrogen = 9.0), and eluent B was methanol. The solvent gradient was programmed as follows: 2% B for 1.5 minutes; A linear increase to 100% B over 12.0 minutes; 100% B held for 2.0 minutes; Return to 2% B at 14.1 minutes; Re-equilibration at 2% B until 17 minutes. The mass spectrometer was operated in both positive and negative ionization modes with a spray voltage of 3.2 kV, a capillary temperature of 320 °C, a sheath gas flow rate of 40 arb, and an auxiliary gas flow rate of 10 arb.

Raw MS data files were processed using Compound Discoverer 3.1 (Thermo Fisher Scientific) for peak detection, alignment, and quantification. Key processing parameters were set as follows: Retention time tolerance, 0.2 minutes; Mass accuracy tolerance, 5 ppm; Signal intensity tolerance, 30%; Signal-to-noise ratio ≥ 3; and Minimum peak intensity of 100000. Peak intensities were normalized to the total ion current for each sample. Subsequent metabolite identification was performed by predicting molecular formulas based on adduct ions, molecular ion peaks, and fragmentation patterns. Identified peaks were matched against the mzCloud (https://www.mzcloud.org/), mzVault, and MassList databases for high-confidence compound annotation and relative quantification. Statistical analyses were conducted using R software (version 3.4.3), Python (version 2.7.6), and the CentOS operating system (release 6.6). For data that did not conform to a normal distribution, normalization was attempted using the area normalization method prior to further analysis.

Identified metabolites were annotated using several databases, including the Kyoto Encyclopedia of Genes and Genomes (https://www.genome.jp/kegg/pathway.html), Human Metabolome Database (https: //hmdb.ca/metabolites), and the LIPIDMaps database (http: //www.lipidmaps.org/). Multivariate statistical analyses, including principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA), were performed using the metaX toolkit (v1.6.0; http://metax.genomics.cn), a comprehensive platform for metabolomic data analysis[20]. To identify significantly altered metabolites, univariate analysis was performed using the Student’s t-test. Metabolites were considered differentially expressed if they met the following criteria: Variable importance in projection (VIP) score > 1 from the PLS-DA model, P value < 0.05, and a fold change (FC) ≥ 2 or ≤ 0.5. Volcano plots were constructed based on the log₂ (FC) and -log₁₀ (P value) of each metabolite to visually identify potential biomarkers.

For hierarchical clustering, data were normalized using z scores of the intensity values for differential metabolites. Heatmaps were generated using the pheatmap package in R. Pearson correlation analysis was applied to evaluate the relationships among differential metabolites using R, and correlation matrices were visualized with the corrplot package. The significance of correlations between differential metabolites and environmental or clinical parameters was assessed using the cor.mtest function in R. A P value < 0.05 was considered statistically significant.

Microbe-metabolite correlation network analysis

Prior to integrative analysis, the microbiome and metabolome datasets were independently screened to identify significantly altered microbial taxa and differentially expressed metabolites across disease groups[21]. Following this, correlations between microbial abundance and metabolite expression levels were assessed using either Spearman or Pearson correlation coefficients to identify significantly associated microbe-metabolite pairs[22]. In the constructed network graph, the fundamental components include nodes and edges. Each bacterial genus or metabolite is represented by a node, with node size indicating its relative abundance or intensity. Nodes are connected by edges, which represent significant correlations; the thickness of an edge reflects the strength of the correlation between two nodes. Edges can be classified as directed or undirected, depending on whether the directionality of the relationship is specified. In some instances, multiple edges may connect the same pair of nodes, with distinctions made through variations in attributes such as direction, line type, color, and thickness to represent different types of associations[23].

Correlation coefficients were calculated using R, and only those pairs meeting the thresholds of |r| ≥ 0.8 and P < 0.05 were retained for network construction. The processed data were formatted into node and edge tables and imported into Gephi64 software for visualization. In the dynamic network visualization, node size reflects the degree-the number of edges connected to that node. Nodes with higher degrees appear larger and are considered more central or influential within the network. Each node thus represents either a microbial genus or a metabolite, and its degree indicates the extent of its connectivity. The degree metric serves as a basic indicator of node importance in terms of network topology and potential biological relevance[24].

To further assess the importance of individual nodes within the network, various centrality metrics were employed to quantify node influence and structural significance. Commonly used centrality measures include betweenness centrality, closeness centrality, and eigenvector centrality. Betweenness centrality quantifies the number of shortest paths that pass through a given node. Nodes with high betweenness centrality serve as critical connectors or bridges, playing pivotal roles in maintaining network cohesion and information flow. Closeness centrality is defined as the reciprocal of the average shortest distance from a node to all other nodes in the network. Nodes with high closeness centrality are well-positioned to rapidly disseminate information and exert influence over the network. Eigenvector centrality evaluates a node’s importance based not only on its number of connections, but also on the importance of its neighbors. A high eigenvector centrality score indicates that a node is highly connected to other influential nodes, signifying its central role within the broader network structure[25].

Based on the correlation coefficients between differentially abundant bacterial taxa and metabolites, we constructed a comprehensive network graph to identify core microbial genera and key metabolites. This integrative analysis revealed potential functional relationships between critical microbial components and metabolite biomarkers.

Statistical analysis

All data analyses were conducted using R software (version 3.4.1). Continuous variables were presented as mean ± SD and compared between groups using the Student’s t-test for normally distributed data. Categorical variables were expressed as frequencies and percentages, and intergroup comparisons were performed using the χ² test. For microbial and metabolomic data, which often deviate from normal distribution, non-parametric statistical methods were employed. Specifically, the Wilcoxon rank-sum test and the Kruskal-Wallis test were used to evaluate differences in species abundance and metabolite expression among groups. A P value < 0.05 was considered statistically significant.

RESULTS

Comparison of general information among patients in different groups

A total of 32 patients were included in this study, comprising 16 with PBC, 9 with AIH, and 7 with OS. All diagnoses were confirmed through liver histopathological examination. The demographic and clinical characteristics of each group are summarized in Table 1. Consistent with the known epidemiology of AILDs, all three conditions predominantly affected female patients. Among the laboratory indicators, white blood cells and red blood cells counts were significantly different between the PBC and OS groups (P < 0.05). Notably, platelets counts were generally higher in OS patients compared to those with PBC or AIH. Liver enzyme levels also varied across groups. Patients with AIH exhibited markedly elevated ALT and AST levels relative to those with PBC or OS. In contrast, patients with PBC showed significantly higher levels of total bilirubin (TBIL), ALP, and GGT, all of which are indicative of cholestasis.

Table 1 Demographic and clinical data of patients in the study.

Characteristic	PBC (n = 16)	AIH (n = 9)	OS (n = 7)
Age (years)	58.00 (42.25-61.00)	52.00 (28.00-57.50)	57.00 (42.00-60.00)
Sex (male/female)	2/14	1/8	1/6
RBC (× 1012/L)	3.89 (3.67-4.04)	3.98 (3.32-5.19)	4.40 (4.23-4.55)a
WBC (× 109/L)	4.17 (3.05-5.73)	4.74 (3.58-6.69)	6.09 (4.75-8.37)a
PLT (× 109/L)	138.50 (46.00-219.00)	132.00 (60.00-245.00)	181.00 (111.00-281.00)
ALT (IU/L)	57.50 (31.75-76.00)	109.00 (39.00-275.00)	82.00 (16.00-259.00)
AST (IU/L)	95.00 (63.00-126.00)	97.00 (39.00-275.00)	52.00 (49.00-199.00)
TBIL (μmol/L)	35.05 (24.80-49.60)	31.60 (16.85-54.95)	30.80 (19.50-33.90)
GGT (IU/L)	190.50 (93.75-485.00)	66.00 (57.00-132.5)	186.00 (29.00-494.00)
ALP (IU/L)	324 (176.50-583.00)	138.00 (94.00-294.50)	216.00 (90.00-310.00)
ALB (g/dL)	37.9 (32.68-45.00)	38.40 (26.20-42.70)	37.30 (34.50-46.80)
TG (mmol/L)	0.97 (0.76-1.24)	1.27 (0.73-1.70)	1.02 (0.64-1.36)
CHOL (mmol/L)	4.63 (3.59-8.12)	4.48 (2.94-5.49)	5.14 (4.03-7.22)
TBIL (μmol/L)	35.05 (24.8-49.6)	31.60 (16.85-54.95)	30.80 (19.50-33.90)
INR	1.04 (0.98-1.13)	1.13 (1.05-1.24)	1.00 (0.91-1.14)
IGA (mg/dL)	3.59 (3.17-4.40)	2.49 (2.12-3.91)	3.50 (2.70-7.87)
IGG (mg/dL)	16.85 (14.63-20.70)	16.00 (15.00-29.55)	17.05 (13.80-18.10)
IGM (mg/dL)	4.45 (2.75-5.26)	1.34 (0.83-2.38)	1.85 (1.06-4.48)
ANA (positive/total)	16/16	8/9	6/7
AMA (positive/total)	12/16	2/9	5/7
ACA (positive/total)	4/16	1/9	1/7

^aP < 0.05.

P vs primary biliary cholangitis. PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome; RBC: Red blood cell; WBC: White blood cell; PLT: Platelets; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; TBIL: Total bilirubin; ALB: Albumin; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase; TG: Triglyceride; CHOL: Cholesterol; INR: International normalized ratio; IGA: Immunoglobulin A; IGG: Immunoglobulin G; IGM: Immunoglobulin M; ANA: Antinuclear antibody; AMA: Antimitochondrial antibody; ACA: Anticardiolipin antibody.

Alterations in microbial richness and diversity in OS patients

To evaluate the gut microbiota profiles of patients with PBC, AIH, and OS, 16S rRNA gene sequencing was performed on fecal samples from all participants. An OTU table was generated based on 97% sequence similarity. A total of 552 OTUs were identified across the three groups. Among them, 188 OTUs were unique to the PBC group, 47 were unique to the AIH group, and only 5 were unique to the OS group (Figure 1A). Rarefaction curves indicated that sequencing depth was sufficient, as the number of observed OTUs reached saturation across all samples. Alpha diversity analysis revealed a significant reduction in microbial diversity and richness in the OS group compared to PBC and AIH. Specifically, the Sobs, Ace, Shannon, and Chao indices were significantly decreased in OS patients, while the Simpson and Coverage indices were elevated (Figure 1B-G). These findings indicate notable microbial dysbiosis in OS, characterized by reduced species richness and evenness[26,27]. Beta diversity analysis, although showing some separation between groups, did not yield statistically significant differences (Figure 1H). To further assess microbial community structure, we performed PLS-DA. The PLS-DA model exhibited good explanatory power (R²X = 0.13, R²Y = 0.11), and the clear separation among the three groups in the score plot suggests distinct microbial compositions (Figure 1I).

Figure 1 Alterations in gut microbial richness and diversity in primary biliary cholangitis, autoimmune hepatitis, and overlap syndrome patients. A: Venn diagram; B: ACE index; C: Chao1 index; D: Shannon index; E: Good coverage metric; F: Simpson index; G: Sobs; H: Boxplot visualization of β-diversity differences; I: Principal coordinates analysis plot. PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome; PCoA: Principal coordinates analysis.

Taxonomy biomarkers in OS patients

At the phylum level, the gut microbiota across all three patient groups was primarily composed of Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, and Verrucomicrobia. Compared with patients with PBC and AIH, OS patients exhibited a marked decrease in the relative abundances of Firmicutes, Bacteroidetes, and Actinobacteria, alongside a notable increase in Proteobacteria and Verrucomicrobia (Figure 2A and Table 2). At the genus level, OS patients showed a significant increase in the relative abundance of pathogenic genera such as Escherichia and Klebsiella. Conversely, beneficial genera, including Bacteroides, Faecalibacterium, Gemmiger, Prevotella, and Clostridium XIVa, were significantly reduced. These compositional shifts suggest a disruption of gut microbial homeostasis in OS patients, potentially contributing to disease onset and progression[28]. Moreover, Bacteroides and Prevotella were predominantly enriched in PBC patients, whereas Gemmiger was more abundant in AIH patients (Figure 2B).

Figure 2 Alterations in microbial structure and composition among primary biliary cholangitis, autoimmune hepatitis, and overlap syndrome patients. A: Phylum level; B: Genus level. PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome.

Table 2 Top 10 phylum-level shifts in gut microbiota composition among three patient groups.

Gut microbiome	AIH (%)	PBC (%)	OS (%)
Firmicutes	61.287655	53.938722	49.503265
Proteobacteria	11.205756	19.142401	28.905284
Bacteroidetes	20.802816	21.928249	15.823387
Actinobacteria	5.998074	2.806389	1.049066
Verrucomicrobia	0.516567	1.83022	3.527271
Fusobacteria	0.04227	0.275033	1.138819
Candidatus Saccharibacteria	0.037415	0.019463	0.017548
Synergistetes	0.005823	0.011598	0.012459
Cyanobacteria	0.004831	0.002569	0.009026
Tenericutes	4.87 × 10-4	9.63 × 10-4	0.0

PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome.

To visualize microbiota structure and inter-genus relationships, a Sankey diagram was constructed. This allowed for the identification of co-occurrence patterns and potential ecological interactions between genera. In OS patients, the gut microbiota was characterized by a specific dominance of Escherichia and Klebsiella. In contrast, Prevotella, Bacteroides, Veillonella, Escherichia, Faecalibacterium, Gemmiger, Lachnospiraceae incertae sedis, and Phascolarctobacterium were predominant in PBC patients, while Gemmiger, Faecalibacterium, Clostridium XIVa, and Prevotella were more enriched in AIH patients. Interestingly, genera such as Faecalibacterium, Gemmiger, Prevotella, and Lachnospiraceae incertae sedis were common to all three groups, yet their abundances were significantly reduced in OS patients. Additionally, OS patients exhibited a relative enrichment in Klebsiella, Escherichia, Akkermansia, Phascolarctobacterium, Veillonella, and Dialister, further indicating a distinct microbial signature associated with OS (Figure 3).

Figure 3 Sankey diagram depicting microbial taxonomic shifts (phylum to genus) in three disease groups. Flow width corresponds to relative abundance. PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome.

Characteristics of gut microbiota related to liver function in OS patients

To investigate the relationship between gut microbiota composition and liver function in OS patients, RDA was performed. The RDA plot revealed a clear association between microbial community structure and clinical liver function parameters. Overall, the model explained 40.74% of the total variation in microbial composition, with the first and second axes accounting for 22.79% and 17.95% of the variability, respectively (F = 1.3402, P = 0.01). Among the liver function indicators, AST was significantly associated with microbial variation (P = 0.0436). The robustness of the RDA results was further confirmed through 999 Monte Carlo permutation tests, yielding a significant P value (< 0.01). Further analysis identified a significant negative correlation between the relative abundance of Fusicatenibacter and AST levels (P < 0.05). AST is a key hepatic enzyme involved in amino acid metabolism and is commonly released into the circulation during liver injury, suggesting a potential link between gut microbial alterations and liver dysfunction in OS patients (Figure 4A).

Figure 4 The correlation between the gut microbial communities and liver function. A: Redundancy analysis across three groups; B: Heatmap in overlap syndrome patients. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. RDA: Redundancy analysis; PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; TBIL: Total bilirubin; ALB: Albumin; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase; TBA: Total bile acid; TG: Triglyceride.

Additionally, Spearman correlation analysis was conducted to explore associations between gut microbial genera and liver function markers. Notably, AST, ALT, and albumin levels were negatively correlated with Fusicatenibacter (P < 0.05), reinforcing its potential relevance to hepatic impairment. The albumin/globulin ratio showed positive correlations with Ruminococcus, Streptococcus, and Clostridium XIVa, and a negative correlation with Blautia (P < 0.05). Moreover, TBIL levels were positively associated with Blautia (P < 0.01), while GGT was positively correlated with Phascolarctobacterium (P < 0.05) (Figure 4B).

Characteristics of serum metabolomics in OS patients

A non-targeted serum metabolomics analysis was conducted on 32 subjects using a high-resolution LC-MS/MS platform (Figure 5). PCA revealed that the three groups of quality control samples clustered closely together, indicating good instrument stability and data reproducibility (Figure 5A). The first two principal components, principal component 1 and principal component 2, jointly explained 29.45% of the total variance. To further differentiate the metabolomic profiles among the PBC, AIH, and OS groups, OPLS-DA was performed. Clear metabolic separation was observed between OS and PBC patients, as well as between OS and AIH patients (Figure 5B and C). The OPLS-DA models showed high explanatory power, with R2 values of 0.92 and 0.96, respectively, suggesting excellent model fit. However, the Q2 values were -0.42 and -0.338, indicating limited predictive capability and suggesting that further model optimization may be necessary (Figure 5D and E). Volcano plots were used to identify differentially expressed metabolites. Compared with OS patients, PBC patients exhibited 10 upregulated and 16 downregulated metabolites (Figure 5F and Table 3), while AIH patients had 4 upregulated and 14 downregulated metabolites (Figure 5G and Table 4). Cluster analysis based on the OPLS-DA model and visualized through a heatmap identified 26 differential metabolites with VIP > 1. Significant differences in metabolite profiles were observed among the three groups. Specifically, AIH patients demonstrated elevated levels of metabolites such as caprylic acid and trans-petroselinic acid. In OS patients, key enriched metabolites included pentadecanoic acid, whereas, in PBC patients, major metabolites such as 1,3-benzothiazol-2-ol were more prominent.

Figure 5 Serum metabolomic characteristics of three groups. A: Principal component analysis score plot; B: Orthogonal partial least squares-discriminant analysis (OPLS-DA) for primary biliary cholangitis (PBC) and overlap syndrome (OS); C: OPLS-DA for autoimmune hepatitis (AIH) and OS; D: Assessment of predictive performance and stability of (B); E: Assessment of predictive performance and stability of (D); F: Volcano plot for PBC and OS; G: Volcano plot for AIH and OS; H: Sample clustering heatmap for PBC, AIH, and OS. PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome; PC: Principal component; VIP: Variable importance in projection.

Table 3 Significantly altered metabolites in overlap syndrome compared to primary biliary cholangitis.

Name	FC	P value	VIP
4-(anilinomethylidene)-3-methyl-4,5-dihydroisoxazol-5-one	1.552696	0.000721	2.639853
2-hydroxy-6-[(8Z,11Z)-pentadeca-8,11,14-trien-1-yl]benzoic acid	0.370792	0.003842	1.818916
Ergosterol peroxide	3.118903	0.004243	2.925869
Cholest-4-en-3-one	0.40443	0.004499	1.984777
Propionylcarnitine	1.925051	0.004683	2.61661
Fludrocortisone acetate	0.248968	0.005297	1.854885
N-Methyldioctylamine	1.585333	0.005887	2.162885
Proscillaridin A	3.2096	0.009332	2.315147
(+/-)-CP 47,497-C7-hydroxy metabolite	0.544798	0.012931	2.197074
PC (16:1e/22:1)	1.747451	0.014132	2.307943
Octadec-9-ynoic acid	0.440369	0.014988	2.076044
PC (16:1e/22:0)	0.48038	0.016027	2.072384
17beta-Trenbolone	0.574612	0.01804	2.224429
SM (d14:0/14:1)	2.716597	0.018685	2.053585
alpha-farnesene	0.647487	0.022544	1.773355
Trigonelline	0.421173	0.023757	1.652406
LPE 18:2	1.567382	0.025836	1.859335
9-HOTrE	0.616145	0.029622	1.97427
Verapamil hydrochloride	0.563206	0.036	1.803812
Cer-NS (d18:1/16:0)	0.549991	0.037082	1.795717
1,3-dimethyluracil	0.471792	0.037551	1.671353
6-(7-methyloctyl)-1H,3H,4H,6H-furo[3,4-c]furan-1-one	1.971003	0.037965	2.152845
(+/-)5(6)-DiHET	0.366881	0.042184	1.46635
Pilocarpine	0.489733	0.043371	1.922503
PC (3:0/16:4)	1.628991	0.044805	1.521527
Testosterone	0.573249	0.046918	1.426691

FC: Fold change; VIP: Variable importance in projection; PC: Phosphatidylcholine; SM: Sphingomyelin; LPE: Lyso-phosphatidylethanolamine; Cer-NS: Ceramide n-acyl-sphingosine.

Table 4 Significantly altered metabolites in overlap syndrome compared to autoimmune hepatitis.

Name	FC	P value	VIP
All-Trans-13,14-dihydroretinol	0.155118	0.002634	2.637447
Alpha-farnesene	0.544136	0.001544	2.616726
PC (14:1e/18:0)	2.137244	0.003936	2.522067
(+/-)-CP 47,497-C7-hydroxy metabolite	0.473641	0.01921	2.325252
PC (16:1e/22:1)	1.939505	0.00826	2.30595
Methyl nicotinate	0.063976	0.045424	2.14672
Octadec-9-ynoic acid	0.502862	0.011726	1.993845
Homo-gamma-linolenic acid (C20:3)	0.49057	0.027711	1.964745
PC (16:1e/22:0)	0.585272	0.012906	1.906264
PC (16:0e/16:0)	1.635313	0.029478	1.902553
Linolelaidic acid (C18:2N6T)	0.649017	0.029729	1.902188
Homoarginine	0.481561	0.016852	1.82961
Mupirocin	0.563021	0.046233	1.817926
Phenylglyoxylic acid	0.286324	0.03155	1.786545
Tetranor-12(S)-HETE	0.433141	0.004515	1.741955
N-Methyldioctylamine	1.530444	0.024254	1.708007
Jasmonic acid	0.545306	0.024253	1.438671
(2S)-2-(2-hydroxypropan-2-yl)-2H,3H,7H-furo[3,2-g]chromen-7-one	0.647968	0.021451	1.313058

FC: Fold change; VIP: Variable importance in projection; PC: Phosphatidylcholine.

Serum metabolomic characteristics related to liver function in OS patients

To further explore the relationship between serum metabolites and liver function in OS patients, a RDA was performed. The RDA model explained 76.65% of the total variance in metabolite composition based on liver function indicators (F = 1.5187, P = 0.109). The first and second axes accounted for 63.07% and 13.58% of the variation, respectively. Notably, AST levels were significantly associated with overall metabolite composition (P = 0.0139) (Figure 6A). Subsequent heatmap analysis provided a more detailed view of these correlations. AST and TBIL levels were negatively correlated with several metabolites, including L-phenylalanine, L-tyrosine, and 2-hydroxycinnamic acid (HCA) (P < 0.05), while albumin levels showed a positive correlation with these metabolites. Additionally, both TBIL and total bile acid levels were positively correlated with DL-tryptophan and indole-3-acrylic acid (P < 0.05) (Figure 6B).

Figure 6 The correlation between the metabolisms and liver function. A: Redundancy analysis for three groups; B: Heatmap for overlap syndrome patients. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. RDA: Redundancy analysis; PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; TBIL: Total bilirubin; ALB: Albumin; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase; TBA: Total bile acid; TG: Triglyceride.

Multi-omics network analysis reveals the relationship between gut microbiota and serum metabolites in OS patients

To investigate the potential interactions between gut microbiota and serum metabolites in OS patients, we performed a correlation analysis integrating microbial and metabolomic data[29]. At the phylum level, in both PBC and OS groups, Verrucomicrobia displayed significant positive correlations with several metabolites, including trehalose, (-)-epigallocatechin, oxytetracycline, trigonelline, and hydroxyprogesterone caproate. In contrast, Bacteroidetes showed significant negative correlations with hydroxyprogesterone caproate and taurochenodeoxycholic acid (Figure 7A). In the AIH and OS groups, Actinobacteria were positively correlated with metabolites related to phosphatidylcholine and Tenericutes, while Firmicutes demonstrated strong negative correlations with the same metabolite classes (Figure 7B).

Figure 7 Heatmap showing the correlation between differential microbiota and differential metabolites in overlap syndrome compared to primary biliary cholangitis or autoimmune hepatitis. A: Compared to primary biliary cholangitis; B: Compared to autoimmune hepatitis. PBC: Primary biliary cholangitis; AIH: Autoimmune hepatitis; OS: Overlap syndrome.

To further elucidate these relationships, we constructed a multi-omics correlation network integrating 16S rRNA gene sequencing data with untargeted metabolomics (Figure 8). In this network, nodes represent bacterial genera or serum metabolites, and edges represent statistically significant correlations between them, determined using Spearman correlation analysis with a false discovery rate threshold of < 0.05. The resulting network is relatively sparse, reflecting the stringency of the correlation criteria. The final integrated network consisted of 48 nodes, including 27 microbial genera and 20 metabolites, with a total of 558 edges. Among these, several metabolite nodes demonstrated high closeness centrality, including phosphatidyl ethanolamine (14:0e/22:2), mycophenolic acid, indole-3-acrylic acid, lysopa 16:0, 5-hydroxyomeprazole, methyl-2-aminobenzoate, omeprazole sulphone, tripeptide glutamyl-methionyl-histidine (EMH), 1-benzyl-4-(4-nitrophenyl) piperazine, L-tyrosine, L-phenylalanine, and homoarginine. These nodes had shorter average distances to microbial taxa, indicating a broader influence within the network. Metabolites with high closeness centrality can be considered functionally important, as they may facilitate rapid information transfer and exert regulatory effects on the microbial community.

Figure 8 Network diagram of differential metabolite-differential microbiota correlations in overlap syndrome patients. OS: Overlap syndrome.

The clustering coefficient is an important metric that reflects the subgroup structure of nodes within a network and characterizes the degree of clustering or modularity. In the multi-omics network constructed in this study, several tightly connected microbial-metabolite clusters were observed, indicating intensive local information exchange. Notably, Akkermansia and Parabacteroides (represented by blue edges) formed strong correlations with metabolites such as DL-tryptophan, L-phenylalanine, indole-3-acrylic acid, and feruloyl putrescine, suggesting active metabolic interaction within this cluster. Likewise, microbial genera including Gemmiger, Lactobacillus, Rosebuia, Eubacterium, Faecalibacterium, Veillonella, Enterococcus, Ruminococcus, Megamonas, Lachnospiracea incertae sedis (represented by green edges) exhibited tight associations with metabolites such as stercobilin, methyl nicotinate, methyl-2-aminobenzoate, and EMH, indicating functional coherence within this subgroup. Another cluster composed of Blautia, Butyricicoccus, Streptococcus, Phascolarctobacterium, and Escherichia (represented by orange edges) showed strong connectivity with lysopa 16:0 and phenylglyoxylic acid. Furthermore, a densely connected module (highlighted by purple edges) involving Ruminococcus, Fusicatenibacter, Dialister, Megasphaera, Clostridium XIVa, Prevotella, Klebsiella, Citrobacter, Bacteroides, and Bifidobacterium demonstrated robust correlations with a wide range of metabolites, including mycophenolic acid, phosphatidyl ethanolamine (14:0e/22:2), 1-benzyl-4-(4-nitrophenyl)piperazine, L-tyrosine, omeprazole sulphone, 5-hydroxyomeprazole, 3-[(3-chloro-4-hydroxy-5-methylanilino)methylidene]pentane-2,4-dione (Supplementary Table 1). This cluster exhibited high network density and clustering, suggesting intense microbial-metabolite interactions and efficient information flow within this community.

DISCUSSION

In recent years, increasing attention has been directed toward understanding the role of the gut microbiome in AILDs. Most existing studies have focused on comparing microbiome alterations between healthy controls and patients with PBC or AIH[11,30,31]. However, in clinical settings, OS often presents with ambiguous disease boundaries. Some patients may initially exhibit features of PBC or AIH, later developing characteristics of the other condition. Moreover, certain PBC patients display autoimmune traits consistent with AIH, yet do not fulfill the diagnostic criteria for OS[8]. These complex and evolving disease phenotypes pose significant challenges for both diagnosis and management and underscore the need for deeper investigation into the distinct gut microbiome features of OS patients. Although PBC and AIH have been shown to exhibit unique microbiota alterations characterized by reduced microbial diversity and depletion of specific taxa, such microbiome signatures remain largely uncharacterized in OS[30,32]. In parallel, numerous studies have explored the correlation between abnormal liver enzyme levels and liver disease severity, using non-invasive approaches to stratify patients and monitor disease progression[33,34]. While substantial work has been done to characterize the gut microbiome composition in PBC and AIH, more recent research has begun to integrate multi-omics approaches, particularly metabolomics[30,35,36], to explore the functional consequences of microbiota alterations and their interactions with host metabolism.

Our research design allowed us to directly investigate the relationship between the gut microbiome and metabolome in OS patients, offering novel insights into this understudied overlap condition. Previous studies have reported distinct microbial alterations in PBC and AIH when compared to healthy individuals. For instance, PBC patients show an increased abundance of genera such as Haemophilus, Veillonella, Clostridium, Lactobacillus, Streptococcus, Pseudomonas, Klebsiella, and several unclassified taxa within the Enterobacteriaceae family, while beneficial genera like Bacteroides, Sutterella, Oscillospira, and Faecalibacterium are significantly reduced[29,37,38]. Similarly, AIH patients exhibit increased levels of Veillonella, Klebsiella, Streptococcus, and Lactobacillus, along with a notable depletion of Bifidobacterium[39-41]. In this study, OS patients demonstrated a reduced relative abundance of Firmicutes, Bacteroidetes, and Actinobacteria, alongside a significant increase in Proteobacteria. It is noteworthy that reductions in Firmicutes and Bacteroidetes are not exclusive to liver dysfunction, as similar patterns are observed in other metabolic and inflammatory conditions, such as non-alcoholic fatty liver disease (NAFLD), reflecting a more general state of gut microbial dysbiosis[42,43]. Our findings suggest that the abundance of these major phyla may be influenced by the patient’s metabolic status, rather than being uniquely associated with OS. In contrast to these established patterns, our study reveals that OS presents unique microbial features. The relative abundance of Proteobacteria and Verrucomicrobia was significantly higher in OS patients than in those with PBC or AIH alone. Proteobacteria, the largest and most metabolically diverse phylum of gram-negative bacteria, has been extensively implicated in inflammatory bowel disease. Elevated levels of Proteobacteria genera such as Escherichia and Klebsiella both of which possess adhesive properties toward intestinal epithelial cells can compromise intestinal barrier function, alter microbial diversity, and trigger inflammation through modulation of host inflammatory gene expression[44]. Given that OS results from the convergence of two AILDs, these findings further emphasize the pro-inflammatory milieu and disrupted immune-microbiota interactions characteristic of the condition. Moreover, our Sankey diagram analysis revealed that several genera from the Firmicutes and Bacteroidetes phyla-including Faecalibacterium, Lachnospiraceae incertae sedis, Gemmiger, and Prevotella-were commonly enriched in all three disease groups (PBC, AIH, and OS), but were significantly depleted in OS patients.

ALT and AST are key enzymes commonly used to evaluate liver injury and dysfunction[45]. ALT is predominantly localized in hepatocytes, and its presence in the bloodstream is a direct indicator of hepatocellular damage[46]. AST, while also present in the liver, is distributed in other tissues such as the heart, skeletal muscle, and kidneys, and therefore may reflect both hepatic and extrahepatic injury. Albumin, synthesized primarily by hepatocytes, serves as an important marker of both liver synthetic capacity and overall nutritional status[47]. Previous studies have shown that alterations in gut microbiota composition are associated with liver dysfunction. In line with these findings, we observed that serum levels of AST, ALT, and albumin were negatively correlated with the abundance of Fusicatenibacter in our cohort. Notably, this inverse relationship suggests a potential link between liver injury and depletion of this beneficial microbial genus in the context of OS. Fusicatenibacter is a genus of anaerobic clostridia considered to be beneficial or associated with low-grade inflammatory responses. It metabolizes glucose to produce a range of anti-inflammatory short-chain fatty acids (SCFAs), including lactate, acetate, and succinate. These SCFAs play key roles in maintaining intestinal health and modulating inflammation. As the liver becomes increasingly compromised reflected by elevated ALT, AST, and decreased albumin such probiotic species may be depleted, further aggravating the gut-liver axis imbalance. Importantly, Fusicatenibacter has been shown to stimulate colonic lamina propria mononuclear cells to produce interleukin-10, an anti-inflammatory cytokine that promotes gut immune homeostasis and protects against intestinal inflammation[48,49]. Moreover, our analysis further revealed a previously unreported positive correlation between Blautia abundance and TBIL levels in OS patients. Blautia is known for its ability to perform 7α-dehydroxylation, a key enzymatic step in the conversion of primary to secondary bile acids in the intestine[50]. Previous studies have shown that depletion of Blautia is associated with liver inflammation in IgG4-related sclerosing cholangitis, a chronic bile duct disorder[51]. Furthermore, Blautia abundance appears to be influenced by bile acid dynamics. For instance, when bile acid secretion into the small intestine is inhibited, Blautia levels have been observed to increase[52], and animal models demonstrate that bile acid supplementation results in elevated Blautia abundance in the gut[53]. Taken together, these findings suggest that elevated Blautia levels in OS patients may represent a compensatory microbial response aimed at enhancing bile acid metabolism and excretion in the setting of bile acid retention or cholestasis.

In addition, we found that OS patients exhibited significantly higher levels of two serum metabolites pentadecanoic acid and 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR) compared to patients with other AILDs such as PBC or AIH. These observations are particularly noteworthy, as previous studies have linked pentadecanoic acid to intestinal barrier protection and anti-inflammatory effects in nonalcoholic steatohepatitis models[54,55], while AICAR has been implicated in immune regulation, including suppression of Th1/Th17 responses[56]. However, their specific elevation in OS patients, and potential interactions with gut microbiota and dietary factors, represent novel findings from our study. Furthermore, compared to patients with PBC or AIH, our analysis revealed a novel observation in OS patients: Serum levels of AST and TBIL were negatively correlated with circulating concentrations of L-phenylalanine, L-tyrosine, and HCA. This inverse relationship represents a distinct metabolic profile that has not been previously reported in OS patients. In previous study, plasma concentrations of aromatic amino acids, including phenylalanine and tyrosine, are significantly elevated in PBC patients relative to healthy controls[57]. L-phenylalanine is an essential aromatic amino acid with important physiological roles, serving as a precursor for adrenaline, thyroid hormones, and melanin pigments. It is irreversibly converted to L-tyrosine via the action of phenylalanine hydroxylase, with tetrahydrobiopterin acting as a cofactor[58]. L-phenylalanine and L-tyrosine are predominantly metabolized in the liver, their accumulation in systemic circulation is generally considered a marker of hepatic dysfunction. In advanced liver disease, impaired metabolic capacity often leads to elevated plasma levels of these aromatic amino acids. However, in early or compensated stages of liver impairment, hepatic hydroxylation activity may remain intact, allowing for continued efficient clearance of these metabolites[59]. Interestingly, our study revealed a negative correlation between these amino acids and liver injury markers (AST and TBIL) in OS patients. One possible explanation for this discrepancy is that the majority of OS patients in our cohort had relatively preserved hepatic function, potentially due to effective disease management and pharmacological interventions. Under these conditions, the liver may still maintain sufficient metabolic capacity to convert and clear these amino acids efficiently. The observed elevations in AST and TBIL, while indicative of liver inflammation, might reflect an adaptive hepatic response rather than outright metabolic failure, thereby contributing to the inverse correlation with these amino acids. Notably, we also found that HCA, a phenolic compound with known anti-inflammatory properties, exhibited a similar negative correlation with AST and TBIL levels. HCA has been widely studied for its ability to reinforce intestinal barrier integrity, accelerate mucosal repair, and modulate gut microbiota by promoting the growth of commensal bacteria while suppressing pathogenic species[60]. Preclinical evidence suggests that HCA can alleviate intestinal inflammation in models of inflammatory bowel disease, suggesting potential systemic benefits in immune-mediated disorders. Therefore, the observed negative correlations between AST, TBIL, and the metabolites L-phenylalanine, L-tyrosine, and HCA may reflect both preserved metabolic activity of the liver and protective host-microbiota interactions in OS patients during a compensated or partially treated disease state.

There exists a close relationship between the gut microbiota and serum metabolomics, reflecting the complex interplay between microbial communities and host metabolic processes. The gut microbiota is capable of metabolizing both dietary components and host-derived secretions, generating a wide array of metabolites, including SCFAs (such as butyrate, propionate, and acetate), bile acids, amino acid derivatives, vitamins, and other bioactive compounds[61,62]. These microbial metabolites can cross the intestinal barrier and enter the circulatory system, thereby influencing the composition of the serum metabolome and modulating the host’s systemic metabolism, immune responses, and disease progression[63]. In our study, integrative multi-omics analysis uncovered distinct associations between gut microbial taxa and serum metabolites in OS patients. Of particular interest, we identified significant correlations between Akkermansia and Parabacteroides and several aromatic amino acid metabolites, a finding that has not been previously reported in this patient population. Previous studies have shown that dietary interventions, such as the ketogenic diet, significantly affect amino acid metabolism and are associated with increased abundances of both Akkermansia and Parabacteroides[64]. Recent findings suggest that microbial metabolism of aromatic amino acids functions as a key signaling mechanism facilitating communication between the host and microbiota. The synthesis and regulation of tryptophan, phenylalanine, and tyrosine, all aromatic amino acids are influenced by the gut microbiota, and fluctuations in their circulating levels may impact intestinal permeability and systemic immune responses[65]. Therefore, Akkermansia and Parabacteroides may play a pivotal role in aromatic amino acid metabolism, actively responding to and influencing changes in the host’s internal environment.

In our study, we identified a novel correlation between AST levels in OS patients and both Fusicatenibacter and L-tyrosine. Fusicatenibacter is known to contribute to the production of SCFAs[66]. In studies on NAFLD, a reduction in Fusicatenibacter has been negatively associated with liver function, potentially due to decreased SCFA production, as SCFAs have been shown to attenuate the progression of NAFLD[67]. Similarly, in liver cancer, a negative correlation between Fusicatenibacter and AST levels has also been reported[68]. Independent studies on early-stage liver cirrhosis have found that the liver’s hydroxylation capacity and tyrosine turnover are not diminished, but rather increased during this compensatory phase[59], suggesting that L-tyrosine may serve as a marker of early-stage liver functional adaptation. Based on these findings, we hypothesize that SCFAs may play a critical role in the pathogenesis of OS and influence liver function through microbial-host metabolic interactions. Future studies employing targeted metabolomics approaches are warranted to validate this hypothesis and further elucidate the role of SCFAs in OS progression.

Despite the insights gained, this study has several limitations. First, as a single-center study with a relatively small sample size, the findings should be interpreted with caution. A key limitation is the limited number of OS patients (n < 10), which may partly explain the absence of statistically significant differences in β-diversity across the three groups. Therefore, future studies with larger, multicenter cohorts and multivariate statistical analyses are needed to validate our findings and hypotheses. Nevertheless, we employed rigorous inclusion criteria, ensuring that all enrolled patients had histologically confirmed diagnoses of PBC, AIH, or OS. Liver biopsy specimens were evaluated by experienced pathologists blinded to clinical data, thus enhancing diagnostic accuracy and reducing intra-group variability. While dietary habits and lifestyle factors may have influenced microbiome and metabolomic profiles, we attempted to minimize such confounding by selecting patients from geographically similar regions. It is also worth noting that this study did not include a healthy control group, focusing instead on patients with PBC, AIH, and OS to investigate disease progression rather than healthy-to-diseased transitions.

In summary, this study is the first to integrate gut microbiota, metabolomic, and liver function data to compare patients with PBC, AIH, and OS. By leveraging a multi-omics approach, we aimed to gain a deeper understanding of the transition from PBC or AIH to OS, improve the differential diagnosis of these conditions, and lay the groundwork for further research into the roles and mechanisms of the gut microbiota and their metabolites in AILD progression.

CONCLUSION

Compared with patients with PBC and AIH, those with OS exhibit further reductions in intestinal microbiota diversity and richness, along with distinct serum metabolite profiles. Multi-omics integration revealed a specific association between AST levels, Fusicatenibacter, and L-tyrosine in OS patients, potentially providing novel insights into the pathophysiological mechanisms underlying this complex disease.

ACKNOWLEDGEMENTS

The authors highly appreciate all patients who participated in the study.