Targeting the gut-liver axis in cholangiocarcinoma: mechanisms, therapeutic advances, and future directions

Authors

Lu Wang

Weiwei Qiao

Xiaowen Zhen

Yeqiong Zhang

Zhiwei Dong

Doi

10.3389/fonc.2025.1646897

PMID: 25608663 · DOI: 10.3389/fonc.2025.1646897 · Journal: Frontiers in Oncology (2025)

TL;DR

This narrative review synthesizes evidence that gut-liver axis dysregulation drives cholangiocarcinoma (CCA) progression through three mechanisms: gut microbiota dysbiosis, bile acid (BA) metabolic disturbances, and immune microenvironment remodeling. The authors compile clinical observations of altered gut and biliary microbial composition in CCA, dysregulated BA signaling via NR1H4 (FXR), GPBAR1 (TGR5), and S1PR2 receptors, and an LPS/TLR4-driven immunosuppressive niche. They evaluate emerging therapeutics — antibiotics (vancomycin, neomycin), probiotics, fecal microbiota transplantation, FXR agonists (obeticholic-acid/INT-747), S1PR2 antagonists (JTE-013), and TGR5 antagonists (SBI-115) — and propose a three-phase translational roadmap integrating multi-omics, AI, and engineered microbiota for precision CCA management.

Cohort & data

  • Article type: Narrative review (no primary cohort).
  • Scope: Cholangiocarcinoma (intrahepatic iCCA, perihilar pCCA, and extrahepatic eCCA/dCCA), with comparisons to hepatocellular carcinoma and gallbladder cancer.
  • Synthesized data sources cited in the review:
    • 16S rRNA sequencing studies (Zhang Q, Ito Z, Deng T, Zhang T, Jia X) of CCA versus benign biliary disease (BBD)/healthy controls.
    • ITS2 rDNA mycobiome sequencing (Zhang L) profiling fungal dysbiosis in iCCA.
    • Mendelian randomization analyses leveraging MiBioGen gut-microbiota GWAS and IEU GWAS data (Zhang Y, Chen Z).
    • Targeted bile acid metabolomics by HPLC/MS, UPLC-MS/MS (15–30 BA species; Prounjai S, Zhang X, Rejchrt S, Wang W, Farhat Z).
    • Single-cell RNA sequencing with consensus clustering (Deng M) classifying CCA into BA-active vs BA-inactive metabolic subtypes.
    • Oral microbiota diagnostic studies (Rao et al.) with three-bacterial-biomarker classifier (AUC 0.981) for iCCA vs HCC.
  • See PMID:25608663, Tables 1, 2, 3, and 4.

Key findings

  • Gut microbiota dysbiosis is a recurrent CCA signature. Across the cited studies, iCCA patients show depletion of beneficial taxa (Faecalibacterium, Roseburia, Bacteroides) and enrichment of pathobionts (Escherichia-Shigella, Enterobacteriaceae). One machine-learning–driven random forest classifier using eight bacterial genera achieved AUC 0.92–0.99 differentiating CCA from HCC (Deng T et al., reviewed). See PMID:25608663, §3.1 and Table 1.
  • Mendelian randomization implicates causal microbial taxa. Elevated Eubacterium hallii group, Candidatus Soleaferrea, and Flavonifractor abundances confer increased biliary tract cancer risk, while Dorea and Lachnospiraceae ND3007 are protective. Veillonellaceae, Alistipes, Enterobacteriales, and Firmicutes are causally implicated in iCCA. See PMID:25608663, §3.1.
  • Fungal dysbiosis tracks tumor stage. Candida albicans overgrowth and reduced Saccharomyces cerevisiae in iCCA gut mycobiota correlate with advanced TNM stage (III–IV). See PMID:25608663, §3.1 and Table 1 (Zhang L reference).
  • Bile acid metabolic reprogramming is diagnostic and prognostic. Conjugated primary BAs (glycocholic acid GCA, taurochenodeoxycholic acid TCDCA) accumulate in CCA serum and bile; secondary conjugates (GLCA, GUDCA) decrease. A CDCA+TCDCA panel reportedly outperformed CA19-9 (AUC = 0.95) for CCA vs BBD/HCC (Zhang X). A four-BA panel (hyodeoxycholic acid, isoLCA, bCDCA, DCA) achieved sensitivity 0.933 and specificity 0.867 (Wang W). See PMID:25608663, §4.1 and Table 2.
  • BA-active CCA subtype has worse outcomes. Single-cell RNA-seq plus consensus clustering (Deng M) classified CCA into BA-active and BA-inactive subgroups; the active subtype showed shorter OS and immunotherapy resistance, with SLCO1B3 and CEACAM1 as prognostic markers. See PMID:25608663, §4.3 and Table 2.
  • LPS/TLR4 axis is a unifying oncogenic node. Gram-negative dysbiosis raises portal LPS, activating hepatic Kupffer cells via TLR4/MyD88 and IL-6/STAT3 signaling. LPS-induced METTL3/PI3K/AKT signaling promotes cholangiocarcinoma migration and invasion (Ke J 2024). See PMID:25608663, §3.3.
  • BA receptor signaling has dual roles. NR1H4 (FXR) activation by CDCA or GW4064 suppresses proliferation via SHP-mediated STAT3 inhibition and downregulation of BCL2L1 (Bcl-xL). GPBAR1 (TGR5) drives ERK1/2 phosphorylation via Src/EGFR transactivation and confers apoptosis resistance through PKA-mediated CD95 inactivation. S1PR2 activation by taurocholic acid triggers ERK/AKT/NF-kB cascades, upregulating PTGS2 (COX-2). See PMID:25608663, §4.2.
  • Vancomycin shows immunomodulatory benefit in PSC. Multiple completed and active clinical trials (NCT01322386, NCT01802073, NCT02137668, NCT03710122, NCT06197308) report improved liver function tests (GGT, ALP, ALT normalization rates of 39%/22%/55.9% within 6 months) in PSC patients on oral vancomycin, supporting microbiota modulation as adjuvant therapy. See PMID:25608663, §6.1.1 and Table 4.
  • FMT in PSC restores biochemistry. A pilot single-FMT colonoscopy study in 10 PSC patients showed safety, increased bacterial diversity, and reduced serum ALP correlated with engraftment; a phase IIa RCT (NCT06286709) is ongoing. See PMID:25608663, §6.1.3 and Table 4.
  • IDH1 mutations indirectly impair BA homeostasis. Frequent IDH1/IDH2 mutations in iCCA drive 2-hydroxyglutarate accumulation, epigenetically suppressing BA synthesis genes (e.g. CYP7A1) via DNA hypermethylation. Ivosidenib (AG-120) is highlighted as the first targeted therapy approved for IDH1-mutant CCA based on the ClarIDHy phase III trial. See PMID:25608663, §6.2.

Genes & alterations

  • NR1H4 (FXR) — downregulated in primary CCA; agonists obeticholic-acid (OCA/INT-747) and GW4064 reactivate SHP/LRH-1 to suppress CCA proliferation and migration. Heterogeneous expression across subtypes (pCCA/dCCA H-score <120) suggests DNMT-inhibitor combinations are needed.
  • GPBAR1 (TGR5) — context-dependent: hepatoprotective in early cholestasis (M2 macrophage polarization, NF-kB suppression) but pro-tumorigenic in advanced CCA (apoptosis resistance via PKA/CD95). Antagonist SBI-115 reduces pro-fibrotic and metastatic signaling.
  • S1PR2 — taurocholic-acid–activated; triggers ERK/AKT/NF-kB and PTGS2 (COX-2) induction, suppressing CD8+ T-cell activity and recruiting regulatory T cells. JTE-013 antagonist suppresses tumor invasiveness in preclinical models.
  • IDH1 / IDH2 — frequent in iCCA; gain-of-function drives 2-HG accumulation, which suppresses BA biosynthesis genes including CYP7A1. Ivosidenib (AG-120) was approved on the basis of the ClarIDHy phase III PFS benefit.
  • FGFR2 — recurrent fusions/rearrangements in iCCA; futibatinib is cited as an FGFR2 inhibitor synergizing with cisplatin.
  • METTL3 — LPS-inducible m6A methyltransferase that promotes CCA migration and invasion via PI3K/AKT (Ke J 2024, cited).
  • STAT3 — downstream of IL-6 inflammatory signaling and of FXR–SHP repression; loss of FXR activity unleashes STAT3 phosphorylation and BCL2L1 (Bcl-xL)-mediated survival.
  • CXCL16 — hepatic sinusoidal endothelial-cell chemokine reduced by secondary BA DCA, impairing CXCR6+ NKT-cell recruitment.
  • CXCL1 — TLR4-driven hepatocyte chemokine that recruits PMN-MDSCs to accelerate CCA progression; neomycin blocks the CXCL1–PMN-MDSC axis.
  • ABCB11 (BSEP) and HNF4A — bile-salt export pump and hepatobiliary master regulator implicated in BA homeostasis maintenance.
  • FGF19 (FGF15 in mouse) — intestinal FXR target induced by Lactobacillus rhamnosus GG (LGG) to suppress BA synthesis.
  • SLCO1B3, CEACAM1 — prognostic markers of the BA-active CCA subtype (Deng M et al., reviewed).
  • PDCD1 (PD-1) — immune checkpoint whose blockade efficacy in BTC is modulated by gut microbiota composition (e.g. Bacteroidetes enrichment improves anti–PD-1 outcomes; Proteobacteria dominance reduces sintilimab+anlotinib efficacy).
  • TGFB1 — pro-inflammatory cytokine measured as a vancomycin-responsive immunomodulatory marker in the NCT03710122 trial.
  • KRAS, TP53 — cited in the context of murine KRAS/p53 knockout models that the authors flag as inadequately recapitulating human cholestatic pathophysiology.
  • EGFR — transactivated by TGR5/Src signaling to amplify ERK1/2 proliferative signaling in cholangiocytes.

Clinical implications

  • BA-panel diagnostics outperform CA19-9 in distinguishing CCA from BBD/HCC in the reviewed studies and warrant prospective multicenter validation. See PMID:25608663, §4.3.
  • Oral vancomycin is a tested adjuvant in PSC that may slow progression to CCA via TH17 suppression and CXCL16-dependent NKT recruitment; trial data so far are confined to PSC, not human CCA. See PMID:25608663, §6.1.1 and Table 4.
  • Ivosidenib is endorsed as a precision option for IDH1-mutant iCCA based on the ClarIDHy phase III readout. See PMID:25608663, §6.2.
  • Futibatinib is cited as an FGFR2-targeted backbone for chemotherapy-combination strategies in FGFR2-rearranged iCCA. See PMID:25608663, §1.
  • Combination strategies the authors propose include FMT + anti-PD-1 for dysbiosis-index–high tumors, and DNMT inhibitors + OCA for FXR-methylated subtypes — both currently hypothesis-stage. See PMID:25608663, §7.2.
  • Caution on antibiotics: prolonged broad-spectrum antibiotic use correlates with reduced survival in HCC patients on anti–PD-1 therapy, so antibiotic-microbiota strategies require careful sequencing and adjunctive probiotics/FMT. See PMID:25608663, §6.1.1.

Limitations & open questions

  • No primary cohort. This is a narrative review; all numerical claims (sample sizes, AUCs, hazard ratios) derive from cited primary studies and inherit their limitations.
  • Small-sample, observational design of most cited microbiome studies precludes causal inference and may reflect confounders (recent antibiotics, diet, cirrhosis severity) rather than CCA-specific dysbiosis. The authors call for large multicenter prospective cohorts with longitudinal monitoring.
  • Geographic restriction, particularly to East Asian and European cohorts, limits generalizability; the authors propose tripartite cohort stratification (Southeast Asian, European/North American, latent-risk diaspora).
  • α-diversity heterogeneity across CCA subtypes (some studies report reduced, others elevated) remains unresolved.
  • TGR5 functional duality (hepatoprotective vs pro-tumorigenic) lacks defined stage thresholds, complicating agonist/antagonist selection.
  • FXR systemic loss vs tissue-specific activation carries opposing oncogenic risks (β-catenin-driven GI tumorigenesis on hyperactivation), requiring long-term safety studies.
  • Murine model translational gap. Species-specific BA metabolism (e.g. CYP8B1 deficiency) and absent immune components limit KRAS/p53 KO and other mouse CCA models. The authors flag the need for humanized or organoid systems.
  • Synthetic-biology biocontainment. Engineered E. coli Nissle 1917 chassis demonstrates 10⁻³–10⁻⁵ CFU/recipient horizontal gene transfer, raising concerns for engineered-microbiota clinical use.
  • FMT protocol standardization and unclear pathogenic thresholds limit reproducibility.
  • Antibiotic-induced immunotherapy resistance is a known confounder; integration with PD-1 blockade needs careful design.

Citations from this paper used in the wiki

  • “Cholangiocarcinoma (CCA), an aggressive malignant tumor of the bile ducts, exhibits distinct epidemiological trends across subtypes… CCA is anatomically classified as intrahepatic (iCCA) or extrahepatic (eCCA), with eCCA further subdivided into perihilar and distal subtypes” — §1 Introduction.
  • “patients with iCCA exhibited significantly reduced abundances of beneficial gut bacteria Bacteroides, Faecalibacterium, and Roseburia, along with increased levels of pathogenic taxa Escherichia-Shigella and Subdoligranulum, compared to healthy controls” — §3.1, citing Zhang Q et al.
  • “a machine learning-driven approach leveraging a random forest algorithm identified eight discriminative bacterial genera (e.g., Faecalibacterium, Klebsiella) to develop a tripartite microbial signature-based classifier, achieving exceptional diagnostic accuracy (AUC: 0.92–0.99)” — §3.1.
  • “Conjugated bile acids exhibit marked accumulation, particularly primary conjugated BAs (glycocholic acid [GCA], taurochenodeoxycholic acid [TCDCA]) showing marked increases in both serum and bile compared to BBD and healthy controls” — §4.1.
  • “Multi-omics studies further identified a quadruple diagnostic panel comprising hyodeoxycholic acid, isoLCA, bCDCA, and DCA, achieving 93.3% sensitivity and 86.7% specificity” — §4.3.
  • “FXR activation induces small heterodimer partner (SHP) expression, which suppresses STAT3 phosphorylation to downregulate Bcl-xL in BTC cells, ultimately triggering apoptosis and inhibiting proliferation” — §4.2.
  • “frequent IDH1/2 mutations in iCCA drive 2-hydroxyglutarate accumulation, which epigenetically impairs cell differentiation and may suppress BA synthesis genes (e.g., CYP7A1) via DNA hypermethylation. Ivosidenib (AG-120), a selective IDH1 inhibitor, demonstrated improved progression-free survival in the phase III ClarIDHy trial” — §6.2.
  • “Ninety-six percent, 81.3%, and 94.9% experienced a reduction of GGT, ALP, and ALT, respectively. Thirty-nine percent, 22%, and 55.9% experienced normalization of GGT, ALP, and ALT, respectively, within the first 6 months of OV [oral vancomycin] treatment” — Table 4, NCT01802073.

This page was processed by crosslinker on 2026-05-14.