Patient characteristics

Patients undergoing hepatic vein catheterization and transjugular liver biopsy (Cohort-Ia) had a median age of 58 (48–64) years, were predominantly male (n = 35, 66%), had a median HVPG of 15 (11–19) mmHg, a median MELD score of 10 (8–14) points, and 28 (53%) patients had decompensated ACLD (dACLD). The most prevalent etiology of ACLD was alcohol-related liver disease (ALD; n = 26, 49%). Patients in Cohort-Ib had a significantly higher prevalence of Child-Turcotte-Pugh (CTP) stages B/C (Table 1).

Table 1 Patient characteristics

Patients undergoing colonoscopy (Cohort-II) had a median age of 58 (51–64) years, were predominantly male (n = 32, 86%), displayed a median MELD of 11 (10–13) points, and 24 (64%) were classified as dACLD. ALD (n = 20, 54%) and viral hepatitis (n = 9, 24%) were the most frequent etiologies of ACLD (Table 1). 28 (76%) patients had HVPG measurement within a median time interval of 4.5 (1.0–16.0) months to colonoscopy and exhibited a median HVPG of 17 (10–21) mmHg.

Characteristics of patients stratified by compensated ACLD (cACLD) and dACLD in Cohort-Ia and Cohort-II are depicted in Supplementary tables S1/S2.

Bile acid and FGF19 serum levels correlate with disease severity

BA serum levels displayed a statistically significant correlation with FGF19 serum levels (Spearman’s rs = 0.461 [0.29–0.61], p < 0.001; Fig. 1A). More than 60% of patients in Cohort-I had BA levels above the upper limit of normal (cACLD: 36%, dACLD: 82%; Supplementary Figure S2). BAs significantly increased in patients stratified by portal hypertension severity (HVPG 6–15 mmHg: 8.98 [5.20–19.4]; HVPG ≥ 16 mmHg: 40.1 [21.5–70.4]; p < 0.001) and disease stage (cACLD: 7.55 [3.38–18.6]; dACLD: 33.5 [15.4–71.4]; p < 0.001), while FGF19 levels tended to increase in patients with dACLD (cACLD: 133 [67.3–222]; dACLD: 157 [114–270]; p = 0.084) but were comparable in patients stratified by HVPG (p = 0.195) (Fig. 1B; Supplementary figure S3). Interestingly, the ratio between FGF19 and BA levels was significantly lower in patients with severe portal hypertension (HVPG ≥ 16 mmHg: 4.92 [2.13–8.79] vs. HVPG  < 16 mmHg: 12.8 [8.90–21.8], p < 0.001) and dACLD (dACLD: 5.82 [2.84–11.7] vs. cACLD: 14.0 [7.81–21.8], p < 0.001), suggesting a relative increase of BA levels as compared to systemic FGF19 levels in patients with severe portal hypertension or dACLD (Supplementary figure S4).

Fig. 1

A Correlation between bile acid (BA) and fibroblast growth factor-19 (FGF19) serum levels. B BA and FGF19 serum levels in patients with compensated and decompensated ACLD. Statistical analysis: Spearman’s correlation coefficient was calculated to assess the association between continuous variables. Mann–Whitney U test was applied to compare continuous variables between groups. BA bile acid; FGF19 fibroblast growth factor-19; c/dACLD compensated/decompensated advanced chronic liver disease; HVPG hepatic venous pressure gradient

Hepatic FXR activation in patients with cirrhosis

To determine the state of FXR expression and FXR activation in liver tissue of patients with ACLD, we assessed the expression of FXR and FXR-dependent genes (SHP, OST-α, OST-β). Hepatic FXR expression was decreased in patients with cACLD (logfold −1.49 ± 0.29; p = 0.044 vs. controls), while being similar in patients with dACLD, as compared to controls (logfold −0.61 ± 0.21, p = 0.565 vs. controls; p = 0.039 vs. cACLD). SHP expression in the liver was reduced in cACLD (logfold −2.66 ± 0.29, p < 0.001 vs. controls) and dACLD (logfold −1.41 ± 0.16, p = 0.045 vs. controls), indicating reduced FXR activation. Nevertheless, SHP expression was significantly higher in patients with dACLD as compared to cACLD (p < 0.001; Fig. 2A). Expression of the basolateral BA transporter OST-α was reduced in patients with cACLD (logfold −2.44 ± 0.23, p < 0.001 vs. controls), but was statistically similar to controls in patients dACLD (logfold −1.16 ± 0.20, p = 0.092 vs. controls). OST-α expression was significantly higher in patients with dACLD compared to cACLD (p < 0.001). Finally, hepatic OST-β expression levels were higher in patients with dACLD (logfold 1.97 ± 1.31), than in controls (p = 0.059) and patients with cACLD (logfold 0.72 ± 0.31, p = 0.016; Fig. 2B; Supplementary Table S4).

Fig. 2

Hepatic expression of FXR and FXR-dependent genes patients with compensated and decompensated ACLD. Statistical analysis: Ordinary one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test was applied to compare continuous variables between groups. FXR farnesoid X receptor; SHP small heterodimer partner; OST-α/-β organic solute transporter-α/-β; CON control group; c/dACLD compensated/decompensated advanced chronic liver disease

Hepatic FXR-FGF19 feedback signaling and BA synthesis

We investigated whether the expression of important genes for hepatic BA synthesis and FGF19-dependent feedback signaling (i.e., CYP7A1, CYP8B1, and FGF receptor 4 [FGFR4]) was linked to disease severity, FXR activation, or FGF19 and BA serum levels in patients undergoing liver biopsy (Cohort-Ia). Expression of CYP7A1 (p = 0.514), the main enzyme for de-novo BA synthesis, and FGFR4 (p = 0.156), the receptor mediating FGF19-dependent feedback signaling for BA synthesis, displayed no significant difference between controls and patients with ACLD. Conversely, CYP8B1 expression, an enzyme for the alternative pathways regulating the composition of the BA pool, was lower as compared to controls (cACLD: logfold −2.04 ± 0.20, p < 0.001; dACLD: logfold −1.55 ± 0.19, p < 0.001; Fig. 3A).

Fig. 3

A Hepatic expression of FXR-FGF19-regulated genes for bile acid synthesis, and B correlation between serum bile acid and FGF19 levels and hepatic expression of FXR-regulated genes in patients with ACLD. Statistical analysis: ordinary one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test was applied to compare continuous variables between groups. Spearman’s correlation coefficient was calculated to assess the association between continuous variables. SHP small heterodimer partner; FGFR4 fibroblast growth factor receptor-4; CYP7A1 cholesterol 7 alpha-hydroxylase; CYP8B1 sterol 12-alpha-hydroxylase; BA bile acid; FGF19 fibroblast growth factor-19

Hepatic SHP displayed a significant positive correlation with FXR (rs = 0.774, 0.63–0.87, p < 0.001), OST-α (rs = 0.754, 0.60–0.85, p < 0.001), OST-β (rs = 0.435, 0.18–0.64, p = 0.001), and also an association with CYP8B1 (rs = 0.663, 0.47–0.80, p < 0.001) and FGFR4 expression (rs = 0.437, 0.18–0.64, p = 0.001), whereas SHP was not linked to CYP7A1 expression (rs = 0.096, p = 0.494; Supplementary Figure S5). CYP7A1 expression showed a significant negative correlation with systemic BA (rs = −0.487, −0.67–[−]0.24, p < 0.001) and FGF19 serum levels (rs = −0.512, −0.71–[−]0.25, p < 0.001). Conversely, neither systemic BA nor FGF19 levels were linked to hepatic expression of CYP8B1 or SHP (Fig. 3B). Systemic BA exhibited a positive association with hepatic OST-α (rs = 0.246, −0.03 to 0.49, p = 0.076), and particularly with OST-β expression (rs = 0.358, 0.09–0.58, p = 0.009; Supplementary Figure S5).

Based on conflicting data from studies whether the liver expresses FGF19 under pathological conditions [2], we evaluated hepatic FGF19 expression in liver biopsies. FGF19 mRNA was only detected in 57% (n = 28/49; n = 4 not analysed due to limited cDNA availability) of liver biopsies. BA serum levels were significantly higher in patients with detectable hepatic FGF19 mRNA (25.5 [11.6–62.2] vs. 7.30 [4.06–16.2] μmol/L without detectable FGF19 mRNA; p = 0.001), while no difference of FGF19 serum levels was noted (p = 0.655; Supplementary Figure S6; Supplementary Table S4).

Intestinal FXR activation, its link to FGF19

We investigated whether FXR activation and FGF19 expression in the ileum was linked to disease severity and FGF19 serum levels in patients with ACLD (Cohort-II). FXR expression decreased in patients with cACLD (logfold −2.14 ± 0.46, p = 0.025) and dACLD (logfold −2.49 ± 0.35, p = 0.004), whereas SHP expression was similar in patients with cACLD (p = 0.207) and showed a statistical trend to decrease in dACLD (logfold −1.97 ± 0.40, p = 0.107), as compared to controls. Expression of OST-α and OST-β decreased in the ileum mucosa of both patients with cACLD (OST-α: logfold −3.16 ± 0.50, p = 0.006; OST-β: logfold −3.21 ± 0.42, p = 0.002) and dACLD (OST-α: logfold −4.12 ± 0.42, p < 0.001; OST-β: logfold −3.91 ± 0.40, p < 0.001) as compared to controls (Fig. 4A). The expression of SHP correlated with FXR (rs = 0.864, 0.75–0.93, p < 0.001), OST-α (rs = 0.825, 0.68–0.91, p < 0.001), OST-β (rs = 0.885, 0.78–0.94, p < 0.001), and with FGF19 (rs = 0.602, 0.32–0.79, p < 0.001) in the ileum (Supplementary Figure S7).

Fig. 4

A Expression of FXR and FXR-dependent genes in ileum biopsies of patients with ACLD. B Expression of FGF19 in ileum biopsies and its correlation with serum FGF19 levels. Statistical analysis: ordinary one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test was applied to compare continuous variables between groups. Spearman’s correlation coefficient was calculated to assess the association between continuous variables. SHP small heterodimer partner; OST-α/-β organic solute transporter-α/-β; BA bile acid; FGF19 fibroblast growth factor-19

FGF19 expression was similar between controls and patients with cACLD (p = 0.216), interestingly however, it decreased in patients with dACLD (logfold −4.26 ± 0.67, p = 0.023). Of note, FGF19 mRNA was not detected in three patients (8%; n = 1 cACLD, n = 2 dACLD). No FGF19 gene expression in the ileum mucosa was neither linked to serum FGF19 levels (p = 0.502) nor serum BA levels (p = 0.170). Notably, intestinal FGF19 expression showed a significant negative correlation with MELD (rs = −0.447, −0.69 to −0.12, p = 0.008; Fig. 4B; Supplementary Figure S8; Supplementary Table S4).

Intestinal FXR signaling and mucosal defence in cirrhosis

Since FXR signaling regulates intestinal barrier integrity, which is believed to be impaired in the setting of cirrhosis [2], we determined whether FXR activation in the ileum was associated with the expression of genes involved in the mucosal barrier in patients with ACLD. Intestinal expression of FXR and SHP was linked to the expression of the tight junction proteins zonula occludens-1 (ZO-1; FXR: rs = 0.836; SHP: rs = 0.830) and occludin (OCLN; FXR: rs = 0.896; SHP: rs = 0.826), and the antimicrobial peptide alpha-5-defensin (DEFA5; FXR: rs = 0.820; SHP: rs = 0.691; all p < 0.001; Supplementary figure S7). Importantly, the expression of ZO-1 and OCLN decreased in patients with dACLD (ZO-1: logfold −1.41 ± 0.32, p = 0.099; controls vs. ACLD p = 0.07; OCLN: logfold −1.96 ± 0.36, p = 0.025; controls vs. ACLD p < 0.05), while DEFA5 expression was significantly lower in both patients with cACLD and dACLD as compared to controls (cACLD: logfold −3.25 ± 0.88, p = 0.017, dACLD: logfold −3.11 ± 0.38, p = 0.013; controls vs. ACLD p < 0.01; Fig. 5; Supplementary Table S4).

Fig. 5

Expression of genes related to intestinal barrier and mucosal defence in the ileum of patients with ACLD. Statistical analysis: ordinary one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test was applied to compare continuous variables between groups. ZO-1 zonula occludens-1; OCLN occludin; DEFA5 α5-defensin; CON control group; c/dACLD compensated/decompensated advanced chronic liver disease