Suppression of bile acid synthesis as a tipping point in the disease course of primary sclerosing cholangitis

Abstract

Background & aims: Farnesoid X receptor (FXR) agonists and fibroblast growth factor 19 (FGF19) analogues suppress bile acid synthesis and are being investigated for their potential therapeutic efficacy in cholestatic liver diseases. We investigated whether bile acid synthesis associated with outcomes in 2 independent populations of people with primary sclerosing cholangitis (PSC) not receiving such therapy.

Methods: Concentrations of individual bile acids and 7α-hydroxy-4-cholesten-3-one (C4) were measured in blood samples from 330 patients with PSC attending tertiary care hospitals in the discovery and validation cohorts and from 100 healthy donors. We used a predefined multivariable Cox proportional hazards model to evaluate the prognostic value of C4 to predict liver transplantation-free survival and evaluated its performance in the validation cohort.

Results: The bile acid synthesis marker C4 was negatively associated with total bile acids. Patients with fully suppressed bile acid synthesis had strongly elevated total bile acids and short liver transplantation-free survival. In multivariable models, a 50% reduction in C4 corresponded to increased hazards for liver transplantation or death in both the discovery (adjusted hazard ratio [HR] = 1.24, 95% CI 1.06-1.43) and validation (adjusted HR = 1.23, 95% CI 1.03-1.47) cohorts. Adding C4 to established risk scores added value to predict future events, and predicted survival probabilities were well calibrated externally. There was no discernible impact of ursodeoxycholic acid treatment on bile acid synthesis.

Conclusions: Bile acid accumulation-associated suppression of bile acid synthesis was apparent in patients with advanced PSC and associated with reduced transplantation-free survival. In a subset of the patients, bile acid synthesis was likely suppressed beyond a tipping point at which any further pharmacological suppression may be futile. Implications for patient stratification and inclusion criteria for clinical trials in PSC warrant further investigation.

Lay summary: We show, by measuring the level of the metabolite C4 in the blood from patients with primary sclerosing cholangitis (PSC), that low production of bile acids in the liver predicts a more rapid progression to severe disease. Many people with PSC appear to have fully suppressed bile acid production, and both established and new drugs that aim to reduce bile acid production may therefore be futile for them. We propose C4 as a test to find those likely to respond to these treatments.

Keywords: 7α-Hydroxy-4-cholesten-3-one; AOM, Amsterdam–Oxford model; ASBT, apical sodium-dependent bile acid cotransporter; Biliary disease; C4; C4, 7α-hydroxy-4-cholesten-3-one; CYP7A1, cytochrome P450 family 7 subfamily A member 1; Cholestasis; Cholestatic liver disease; FGF19, fibroblast growth factor 19; FXR, farnesoid X receptor; GUDCA, glycooursodeoxycholic acid; HR, hazard ratio; IBAT, ileal bile acid transporter; Liver transplantation; Liver transplantation-free survival; MELD, model for end-stage liver disease; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; STROBE, Strengthening the Reporting of Observational Studies in Epidemiology; TUDCA, tauroursodeoxycholic acid; UDCA, ursodeoxycholic acid; UPLC-MS/MS, ultraperformance liquid chromatography–tandem mass spectrometry; Ursodeoxycholic acid; c-index, concordance index; liver.

Graphical abstract

Fig. 1

Cholestasis-driven suppression of bile acid synthesis is evident in PSC. (A and B) Circulating levels of C4 and total bile acids in healthy controls (n = 100) and patients with PSC in the discovery (n = 191) and validation (n = 139) cohorts. Individual data points, boxplots (white dots indicate the median), and probability densities are shown for each category. (C) Untransformed bivariate plot of C4 and total bile acids, coloured by whether the samples were drawn from individuals with or without PSC and cohort affiliation. Two outliers with C4 >300 nmol/L are not shown. To the upper right, the same data points are plotted on log₁₀-transformed axes with smooth loess lines fitted for each cohort with PSC. To the lower right, an estimated log-linear regression model is plotted on the original scale (zoomed in for clarity) generated from all samples from patients with PSC pooled together. The adjusted R² value from linear regressions fitted to each cohort separately is indented. C4, 7α-hydroxy-4-cholesten-3-one; PSC, primary sclerosing cholangitis.

Fig. 2

The limited apparent effect of UDCA on bile acid-mediated regulation of bile acid synthesis. (A) Concentrations of total bile acids in circulation, grouped by phenotype, cohort, and UDCA use. (B) Fractions of circulating UDCA and UDCA-derived bile acids relative to the total circulating bile acids (UDCA enrichment; UDCA + isoUDCA + GUDCA + TUDCA/total bile acids). Median enrichment (IQR) is annotated to the right of each bar. (C) Concentrations of C4 in circulation, grouped by phenotype, cohort, and UDCA use. (D) Bivariate log–log plots of C4 and total bile acids stratified by cohort affiliation and UDCA treatment status. The strengths of the associations between paired samples were tested using Spearman’s rank-order correlation (r_s). C4, 7α-hydroxy-4-cholesten-3-one; GUDCA, glycooursodeoxycholic acid; TUDCA, tauroursodeoxycholic acid; UDCA, ursodeoxycholic acid.

Fig. 3

C4 associates with liver transplantation-free survival. Liver transplantation-free survival curves of patients with PSC in the (A) discovery cohort and (B) validation cohort, calculated using the Kaplan–Meier method. Patients were categorised by the boundaries determined by quartiles of C4 in the discovery cohort. In both cohorts, patients were censored at 10-year follow-up. Comparisons of the survival distributions were tested using log-rank tests (p values indented). The number at risk and number censored are shown for each indicated time point. Smoothed calibration curves illustrating the agreement between the estimated predicted event-free probabilities from the full Cox model (C4 + Mayo PSC score) and the observed event-free fractions in the external validation cohort at the (C) 5-year and (D) 8-year time horizons. The 45° dashed line indicates perfect calibration. One-dimensional histograms of the predicted event-free probabilities are shown on the top of each plot to illustrate their distributions. C4, 7α-hydroxy-4-cholesten-3-one; PSC, primary sclerosing cholangitis.