Endogenous ursodeoxycholic acid and cholic acid in liver disease due to cystic fibrosis


  • Jeffery L. Smith,

    1. Hepatic Fibrosis Group, The Queensland Institute of Medical Research, Brisbane, Australia
    2. Department of Biochemistry and Molecular Biology, The University of Queensland, Brisbane, Australia
    Search for more papers by this author
    • J.L.S. and P.J.L. contributed equally to this study.

  • Peter J. Lewindon,

    1. Hepatic Fibrosis Group, The Queensland Institute of Medical Research, Brisbane, Australia
    2. Department of Gastroenterology, Royal Children's Hospital, Brisbane, Australia
    Search for more papers by this author
    • J.L.S. and P.J.L. contributed equally to this study.

  • Anita C. Hoskins,

    1. Hepatic Fibrosis Group, The Queensland Institute of Medical Research, Brisbane, Australia
    Search for more papers by this author
  • Tamara N. Pereira,

    1. Hepatic Fibrosis Group, The Queensland Institute of Medical Research, Brisbane, Australia
    Search for more papers by this author
  • Kenneth D. R. Setchell,

    1. Division of Clinical Mass Spectrometry, Children's Hospital Medical Center, Cincinnati, OH
    Search for more papers by this author
  • Nancy C. O'Connell,

    1. Division of Clinical Mass Spectrometry, Children's Hospital Medical Center, Cincinnati, OH
    Search for more papers by this author
  • Ross W. Shepherd,

    1. Department of Pediatrics, Washington University School of Medicine, St. Louis, MO
    Search for more papers by this author
  • Grant A. Ramm

    Corresponding author
    1. Hepatic Fibrosis Group, The Queensland Institute of Medical Research, Brisbane, Australia
    • Hepatic Fibrosis Group, The Queensland Institute of Medical Research, P.O. Royal Brisbane Hospital, Herston, QLD, 4029, Australia
    Search for more papers by this author
    • fax: 61-7-3362-0191


Focal biliary cirrhosis causes significant morbidity and mortality in cystic fibrosis (CF). Although the mechanisms of pathogenesis remain unclear, bile acids have been proposed as potential mediators of liver injury. This study examined bile acid composition in CF and assessed altered bile acid profiles to determine if they are associated with incidence and progression of liver injury in CF-associated liver disease (CFLD). Bile acid composition was determined by gas–liquid chromatography/mass spectrometry in bile, urine, and serum samples from 30 children with CFLD, 15 children with CF but without liver disease (CFnoLD), and 43 controls. Liver biopsies from 29 CFLD subjects were assessed histologically by grading for fibrosis stage, inflammation, and disruption of the limiting plate. A significantly greater proportion of endogenous biliary ursodeoxycholic acid (UDCA) was demonstrated in CFnoLD subjects vs. both CFLD subjects and controls (2.4- and 2.2-fold, respectively; ANOVA, P = .04), and a 3-4 fold elevation in endogenous serum UDCA concentration was observed in both CFLD subjects and CFnoLD subjects vs. controls (ANOVA, P < .05). In CFLD, there were significant correlations between serum cholic acid and hepatic fibrosis, inflammation, and limiting plate disruption as well as the ratio of serum cholic acid/chenodeoxycholic acid to hepatic fibrosis, inflammation, and limiting plate disruption. In conclusion, elevated endogenous UDCA in CFnoLD suggests a possible protective role against liver injury in these patients. The correlation between both cholic acid and cholic acid/chenodeoxycholic acid levels with histological liver injury and fibrosis progression suggests a potential monitoring role for these bile acids in CFLD. (HEPATOLOGY 2004;39:1673–1682.)

Advances in pulmonary and nutritional strategies for patients with cystic fibrosis (CF) have significantly improved life expectancy and quality over the last 30 years. Although pulmonary complications still account for the greatest mortality and morbidity among these patients, clinically significant fibrosing CF-associated liver disease (CFLD) has a prevalence of 13%–17%.1–7 Onset and progression is during childhood; clinical signs appear late, when fibrosis is advanced. With no reliable predictors or disease markers, early interventions cannot be adequately evaluated. The abnormal CF transmembrane conductance regulator (CFTR) protein was identified in the biliary epithelium and characterized more than a decade ago,8 yet treatments to prevent progression or initiate regression of CFLD remain elusive. Bile acid therapy with ursodeoxycholic acid (UDCA) is widely used to treat cholestatic liver diseases, and improvements in serum parameters, hepatic excretory function, and liver histology9–17 have been reported in patients with CFLD. However, while the physiological benefits of UDCA are known in stimulating bile flow and non-CFTR biliary chloride channels and protecting hepatocytes from toxicity of other bile acids, the efficacy of UCDA to alter the natural history of CFLD is not proven.18

A role for bile acids in the pathogenesis of CFLD was first proposed more than 30 years ago by Sandberg19 and Weber et al.20 Since then, the absence of bile acid data obtained during the critical childhood period of disease progression from children with clinically characterized liver disease and from appropriate controls compromised the power of previous studies to find significant associations. Using radioimmunoassay, Strandvik and colleagues reported increased urinary and serum primary bile acids21, 22 in patients with CF without liver disease (CFnoLD) vs. non-CF controls. Adult patients up to 27 years of age were included in both studies. Liver biopsies were performed in the CF group, and no differences were seen in urine or serum when grouped according to gross biopsy severity (“cirrhosis,” “portal fibrosis,” and “minor changes”). Using mass spectroscopy, Setchell et al. examined feces and serum from 10 children with CF without evidence of liver disease and observed increased fecal bile acid losses; however, unlike Strandvik and colleagues, they observed reduced serum levels of total bile acids and cholic acid compared with four controls.23 Nakagawa et al. analyzed duodenal bile from only five children with CF and evidence of liver disease, but without biopsy.24 Bile acids were mainly in the form of glycine and taurine conjugates; two new bile acids—homocholic acid and homochenodeoxycholic acid—were reported.

The present study was more definitive in analyzing specimens from three physiologically important compartments (serum, urine, and bile) from significant numbers of well-characterized children with CFLD and appropriate age- and sex-matched controls, which were comprised of children with CFnoLD and children without CF during the critical age period of CFLD pathogenesis. Comprehensive analysis of bile acids using sensitive and specific methods permitted comparison between the three clinical groups and within the CFLD group according to standardized histopathological severity.


CF, cystic fibrosis; CFLD, cystic fibrosis–associated liver disease; CFnoLD, cystic fibrosis without liver disease; UDCA, ursodeoxycholic acid; CFTR, cystic fibrosis transmembrane conductance regulator; LPD, limiting plate disruption; CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; LCA, lithocholic acid.

Patients and Methods

Clinical Investigations and Sample Collection

The Ethics Committees of The Queensland Institute of Medical Research and the Royal Children's Hospital, Brisbane, Australia, approved all study protocols and procedures, which conformed to the ethical guidelines of the 1975 Declaration of Helsinki. Informed consent was obtained from all patients or their next of kin.

Group 1: Children With CFLD.

This group comprised 30 children who had CF (diagnosed by sweat test) and who were undergoing clinical investigations—including liver biopsy in 29 cases—for assessment of CFLD. All children had at least two of the following clinical features: (1) hepatomegaly with or without splenomegaly; (2) persistent elevation of serum alanine aminotransferase (>1× upper limit of normal) longer than 6 months; or (3) abnormal ultrasound scan with abnormal echogenicity suggestive of fibrosis or nodular edge suggestive of cirrhosis. The relevant clinical and histopathological characteristics of the patients are shown in Table 1. Percutaneous liver biopsy was performed under general anesthetic and ultrasound control. Under the same anesthetic, fasting duodenal bile was collected by direct endoscopic intubation from 22 patients following stimulation with cholecystokinin and stored at −70°C for later bile acid analysis (see below). In this patient group the gallbladder must be made to contract, using cholecystokinin, to enable sufficient bile to be collected for analysis. Where possible, fasting serum and urine were also collected from these patients. All patients were clinically stable at the time of the procedures and apart from two of them, none had received antibiotics in the 3 weeks prior to biopsy and collections of bile, serum, and urine. None of the children had undergone oral UDCA therapy prior to evaluation.

Table 1. Clinical and Histopathological Characteristics of Patients With CFLD
Patient No.SexAge (y)Meconium Ileus?Insulin-Dependent Diabetes Melitus?Clinical Hepatomegaly ± Splenomegaly?Portal Hypertension?Scheuer Fibrosis Score (0–4)Inflammation Score (0–3)LPD Score (0–3)CF Genotype
  • Abbreviations: F, female; M, male; ΔF, ΔF508; NA, not available.

  • *

    Steatosis was the only histological abnormality.

1F16.4 YesYesYes410ΔF/ΔF
2F9.2  YesYes423ΔF/ΔF
3M10.3  YesYes311?
5F14.8  YesYes312ΔF/ΔF
6F9.9Yes Yes 312ΔF/ΔF
7M15.6Yes Yes 311ΔF/ΔF
8F11.4 YesYes 222ΔF/−
9F10.9    211?
10M9.9  YesYes211ΔF/ΔF
11M1.3    211ΔF/−
12M13.4Yes YesYes211ΔF/ΔF
13F8.9Yes YesYes223ΔF/−
14M12.1  YesYes211ΔF/−
15M15.4  YesYes222?
16M18.7    2NANA?
17M9.1  Yes 1NANAΔF/ΔF
18M10Yes   121?
19F8.7    112?
20M2.7    111ΔF/ΔF
21F17 YesYes 111?
22F10.5  Yes 1NANAΔF/−
23F11.3Yes Yes 100?
24F6.5    000ΔF/ΔF
25M10.8Yes Yes 0*00?
26F7.9    000?
27F8.9    000ΔF/ΔF
28M8Yes   0NANAΔF/ΔF
29F15.6  Yes 0*00ΔF/ΔF
30F10Yes Yes NANANA?

Group 2: Children With CFnoLD.

This group comprised 15 children with CF and no clinical, biochemical, or ultrasonographic evidence of CFLD. Ultrasound examination was normal in all cases and did not show any signs of steatosis. Fasting duodenal bile was obtained only from those patients undergoing endoscopy for other indications, such as gastroesophageal reflux (n = 11). Where possible, fasting serum and urine samples were also collected. Liver biopsy was not performed on any patient in this group.

Group 3: Controls.

This group comprised 43 children without CF who were admitted for a variety of benign conditions (e.g., minor elective surgery). Samples of blood and urine were obtained where possible. In 14 of these children, fasting duodenal bile was collected during endoscopy for conditions unrelated to the enterohepatic circulation of bile acids (e.g., gastroesophageal reflux). Liver biopsy was not performed on any patient in this group.

Histological Assessment

Liver biopsy tissue was handled as previously detailed25 and the severity of fibrosis was assessed using the staging system of Scheuer.26 The degree of inflammation was scored as previously described.26 A semiqualitative assessment of ongoing hepatic injury was performed, as recently described in detail,25 by assessing the limiting plate disruption (LPD) and modified for use in CFLD in recognition of the variability in severity within the same biopsy core.

Determination of Bile Acid Composition

Bile acids were quantitatively extracted from serum (0.1–0.5 mL), urine (5–20 mL), and duodenal bile (0.1–0.5 mL), after addition of the internal standard nordeoxycholic acid (0.1–10 μg), using reverse-phase octadecylsilane-bonded silica cartridges (Bond Elut C18) as previously described.27 The bile acid extract was solvolyzed and enzymatically hydrolyzed with cholylglycine hydrolase, and the resulting hydrolyzed, unconjugated bile acids were isolated by solid-phase extraction and lipophilic anion exchange chromatography. The bile acids were converted to methyl ester-trimethysilyl ethers for analysis by gas spectrometry/mass spectroscopy as previously described.28 In bile and urine, individual bile acid results are presented as a percentage of the total bile acid concentration (i.e., proportion) due to inherent problems associated with collection (e.g., duodenal bile consists of a mixture of gallbladder and hepatic bile, together with intestinal secretions). All samples were analyzed blinded.

Statistical Analysis

Results are represented as mean ± SEM. Where variables were normally distributed, differences between groups were analyzed using ANOVA. Where data could not be normalized, groups were compared using Kruskal-Wallis ANOVA and the Mann-Whitney U test; P < .05 was considered significant. Correlations between bile acid species and the grade of histological fibrosis, inflammation scores, and LPD scores were determined using Spearman's rank correlation for discontinuous variables.


Clinical and Histopathological Characteristics

Group 1: Children With CFLD.

All 30 children were from families of northern European descent and were predominately df508 genotype (14 homozygotes and five heterozygotes). The mean age was 10.8 years (range: 1–18) and the sex ratio was 14 male/16 female. No child with CFLD had end stage disease, and all had normal synthetic function. Only one child was pancreatic-sufficient; the others were on replacement pancreatic enzymes in recommended doses. None had undergone hepatobiliary surgery. There was clinical evidence of hepatomegaly with or without splenomegaly in 20 (67%); definite portal hypertension in 10 (33%); and a history of meconium ileus at birth in 10 (33%). Only three (10%) had a limited resection of the terminal ileum; four (13%) had insulin-dependent diabetes. Although patients with diabetes are known to have an altered lipid metabolism, including biliary lipids, in this study duodenal bile was not available for analysis from the four patients with diabetes and thus did not confound the results.

Histopathological findings for each of the 29 CFLD patients undergoing liver biopsy are listed in Table 1. Six biopsies (21%) scored 0 for fibrosis, of which two revealed steatosis only. The remainder revealed the usual spectrum of cholestasis, bile duct proliferation, steatosis, portal inflammation, and fibrosis, ranging from Scheuer grades 1–4. Sixteen patients (55%) had fibrosis of moderate severity (Scheuer grades 1–2) and seven patients (24%) had advanced fibrosis (Scheuer grade 3–4). The severity of LPD was variable and was not associated with the severity of established fibrosis. The predominant inflammatory cells present were neutrophils and accompanied the proliferating bile ductules indicative of biliary obstruction. No other histopathological diagnoses were found to account for the liver disease in these patients.

Group 2: Children With CFnoLD.

All 15 children were of northern European descent and had a comparable age match with a mean of 10.9 years (range: 3–16) and a sex ratio of 9 male/6 female. Five patients had meconium ileus (33%); only two had a limited resection of the terminal ileum (7%). Only one patient was receiving insulin for diabetes. Four patients were DF/DF, four were DF/−, one was −/−, and six were unknown genotypes.

Group 3: Controls.

All 43 controls were of northern European descent, with a mean age of 8.6 years (range: 3–16) and a sex ratio of 27 male/16 female.

CF Bile Acid Composition in Duodenal Bile, Serum, and Urine

The primary bile acids, cholic acid (CA) and chenodeoxycholic acid (CDCA), were the most abundant bile acid species in duodenal bile and serum, with CA being the most abundant in bile and CDCA predominating in serum (Table 2). In serum, the total bile acid concentration in both CFLD (5.56 ± 0.87 μmol/L) and CFnoLD (4.80 ± 0.89 μmol/L) patients was significantly higher than in non-CF controls (3.18 ± 0.60 μmol/L; ANOVA, P < .001). This difference was also reflected in increased levels of CA and CDCA, and, to a lesser extent, increases in deoxycholic acid (DCA) and UDCA levels in CF patients compared with controls. The total bile acid concentration of duodenal bile in CFnoLD and CFLD was approximately 10-fold greater than in controls, while levels in CFnoLD were approximately 40% greater than in CFLD (CFLD, 13.7 ± 2.4 mmol/L; CFnoLD, 19.7 ± 2.1 mmol/L; controls, 1.5 ± 0.3 mmol/L).

Table 2. Bile Acid Composition in Duodenal Bile, Serum, and Urine in Patients With CFLD and CFnoLD and in non-CF Controls
Patient GroupBile Acid Composition
  • NOTE: Results are represented as the mean ± SEM. Statistical difference between groups was analyzed by ANOVA.

  • Abbreviation: ND, not detectable.

  • *

    Other bile acids are metabolites/derivatives of primary and secondary bile acids.

  • P = .06.

  • P = .02.

  • §

    P = .04.

  • P < .001.

  • P = .001.

  • #

    P < .05.

Duodenal bile (% of total)      
 Non-CF (14)43.1 ± 2.927.6 ± 2.19.6 ± 2.41.6 ± 0.62.6 ± 0.814.8 ± 2.2
 CFnoLD (11)45.4 ± 2.033.0 ± 2.05.7 ± 1.30.3 ± 0.15.6 ± 1.2§11.4 ± 1.4
 CFLD (22)52.1 ± 2.1231.6 ± 1.97.0 ± 1.40.2 ± 0.12.3 ± 0.48.6 ± 1.4
Serum (μM)      
 Non-CF (29)0.72 ± 0.241.35 ± 0.270.33 ± 0.060.33 ± 0.040.16 ± 0.03#ND
 CFnoLD (7)1.52 ± 0.262.75 ± 0.560.76 ± 0.260.16 ± 0.070.61 ± 0.27ND
 CFLD (14)1.71 ± 0.292.83 ± 0.420.43 ± 0.090.11 ± 0.020.48 ± 0.16ND
Urine (% of total)      
 Non-CF (6)4.32 ± 1.257.29 ± 2.377.22 ± 2.159.22 ± 2.428.44 ± 3.4363.51 ± 2.58
 CFnoLD (8)5.15 ± 0.995.56 ± 2.205.56 ± 1.631.13 ± 0.825.56 ± 1.5277.03 ± 3.86
 CFLD (19)7.89 ± 1.567.02 ± 1.354.42 ± 1.280.90 ± 0.264.99 ± 4.0074.76 ± 4.75


The proportion of CA in the bile was significantly higher in CFLD compared with both CFnoLD and non-CF controls (ANOVA, P = .02); however, there was no significant difference between CFnoLD and controls (see Table 2). Serum CA concentration was significantly elevated in both CFLD and CFnoLD compared with controls (ANOVA, P < .001), but there was no difference between CFLD and CFnoLD (see Table 2). Urinary CA proportions were not different among the three subject groups (see Table 2).


There was no significant difference in the proportion of CDCA in the bile between CFLD, CFnoLD, and controls (see Table 2). Serum CDCA concentration was significantly increased in both CFLD and CFnoLD compared with controls (ANOVA, P < .001); however, there was no difference between CFLD and CFnoLD. There were no significant differences in urinary CDCA proportions between the three groups.

Lithocholic Acid and DCA.

Patients with CF had a markedly decreased proportion of the hepatotoxic bile acid lithocholic acid (LCA) in samples of duodenal bile and serum (Fig. 1), and in urine compared with non-CF patients; the proportions ranged between 50%–90% of controls (see Table 2). In contrast, while DCA tended to be lower in bile and urine samples from patients with CF, the differences were not statistically significant (see Table 2).

Figure 1.

Biliary LCA (percentage of total bile acids) and serum LCA (μM) in CFLD patients, CFnoLD patients, and controls. (a) The proportion of LCA in the bile was decreased in both CFLD patients and CFnoLD patients compared with controls, with a trend toward statistical significance (ANOVA, P = .06). (b) Serum LCA levels were significantly decreased in both CFLD patients and CFnoLD patients compared with controls (ANOVA, P = .001). Abbreviations: CFLD, cystic fibrosis–associated liver disease; CFnoLD, cystic fibrosis without liver disease.

UDCA Levels in CF.

In duodenal bile, there was a marked increase in the proportion of endogenous UDCA in patients with CFnoLD compared with both CFLD and controls (ANOVA, P = .04) (Fig. 2a). Biliary UDCA concentrations were also likely to be significantly increased in CFnoLD, because the total bile acid concentration was approximately 40% greater in CFnoLD vs. CFLD patients. Mean UDCA proportions in CFnoLD patients were approximately 2.4-fold higher than in controls and 2.2-fold higher than in patients with CFLD. There was no statistical difference between biliary UDCA proportions in CFLD patients vs. controls. In serum, there was also a marked increase in endogenous UDCA levels in CFnoLD patients vs. controls (3.8-fold) and in CFLD patients vs. controls (3-fold) (ANOVA, P < .05) (Fig. 2b); however, there was no significant difference between the CFnoLD and CFLD patient groups. There was a significant decrease in urinary UDCA composition in both CF groups when compared with controls (ANOVA, P < .001) (see Table 2).

Figure 2.

Biliary UDCA (percentage of total bile acids) and serum UDCA (μM) levels in CFLD patients, CFnoLD patients, and controls. (a) The proportion of UDCA in the bile was significantly increased in CFnoLD patients vs. both CFLD patients and controls (ANOVA, P = .04). (b) Mean serum UDCA levels were significantly increased by 3.8-fold in CFnoLD patients and 3-fold in CFLD patients compared with controls (ANOVA, P < .05). Abbreviations: CFLD, cystic fibrosis–associated liver disease; CFnoLD, cystic fibrosis without liver disease.

Correlation Between Bile Acid Composition and Histological Parameters of Injury in CFLD

As reported in Table 2, only CA and UDCA were found to be significantly different between the CFnoLD and CFLD patient groups. Likewise, only CA and UDCA (as a proportion of the total bile acids), showed significant association with histological markers of liver injury (Table 3). In duodenal bile, the proportion of UDCA significantly correlated with both the inflammation score (r = 0.56, P = .019) and the LPD score (r = 0.63, P = .007) (Fig. 3), but had no statistically significant correlation with fibrosis stage (r = 0.26, P = .26). In contrast, serum CA was positively correlated with stage of fibrosis (r = 0.61, P = .036) and both inflammation (r = 0.68, P = .031) and LPD scores (r = 0.64, P = .047), albeit with relatively small numbers of CFLD sera available for these analyses (see Table 3). By comparison, biliary CA was negatively associated (r = −0.50, P = .04) with LPD score (see Table 3).

Table 3. Correlation Between Bile Acid Species and Grade of Fibrosis, Inflammation Score, and LPD Score
Bile Acid SpeciesSpearman's Rank Correlation (r) and Statistical Significance (P)
Grade of FibrosisInflammation ScoreLPD
  1. NOTE: Bold correlations reflect significant associations.

CA0.06 (.79)0.61 (.04)0.34 (.18)−0.18 (.49)0.68 (.03)0.095 (.79)−0.5 (.04)0.64 (.05)−0.03 (.94)
CDCA−0.11 (.63)−0.31 (.32)−0.03 (.9)0.21 (.43)−0.37 (.29)−0.017 (.96)0.42 (.09)−0.32 (.36)−0.12 (.75)
DCA−0.32 (.16)−0.08 (.8)−0.14 (.58)−0.48 (.5)0.11 (.77)−0.007 (.98)0.37 (.15)0.06 (.87)0.04 (.91)
LCA0.22 (.32)−0.22 (.49)0.02 (.94)0.29 (.25)−0.05 (.89)0.06 (.87)0.18 (.5)−0.18 (.62)−0.11 (.76)
UDCA0.26 (.26)−0.18 (.57)0.33 (.2)0.56 (.02)−0.31 (.38)0.04 (.91)0.63 (.007)−0.22 (.55)0.2 (.58)
CA/CDCA0.12 (.6)0.80 (.003)0.14 (.7)−0.22 (.39)0.89 (.001)0.13 (.7)−0.49 (.05)0.84 (.004)0.27 (.42)
Figure 3.

Correlation between biliary UDCA levels in patients with CFLD and histological parameters of hepatic injury. Biliary UDCA (percentage of total bile acids) was significantly correlated with (a) inflammation score (r = 0.56, P = .019) and (b) LPD score (r = 0.63, P = .007). Abbreviation: LPD, limiting plate disruption.

CA/CDCA Ratio in Predicting Liver Injury Progression in CFLD

There was no significant difference in the CA/CDCA ratio between the three subject groups in bile, serum, or urine (results not shown). However, strong correlations were demonstrated between the serum CA/CDCA concentration ratio and fibrosis stage (r = 0.80, P = .003) and both inflammation (r = 0.89, P = .001) and LPD scores (r = 0.84, P = .004) in subjects with CFLD (Fig. 4). In addition, biliary CA/CDCA showed a significant negative correlation with LPD score (r = −0.49, P = .05), but not with fibrosis or inflammation score (see Table 3). No correlations were observed for urinary CA/CDCA (see Table 3).

Figure 4.

Correlation between serum CA/CDCA ratio and histological parameters of hepatic injury in patients with CFLD. Serum CA/CDCA ratio was significantly correlated with (a) fibrosis stage (r = 0.80, P = .003), (b) inflammation score (r = 0.89, P = .001), and (c) LPD score (r = 0.84, P = .004). Abbreviations: CA, cholic acid; CDCA, chenodeoxycholic acid; LPD, limiting plate disruption.


This comprehensive study reports the bile acid composition of three physiologically relevant body fluids—duodenal bile, serum, and urine—from children with CF. It compares bile acid composition in CF patients with and without liver disease with age-matched pediatric controls. These bile acid profiles are correlated with the severity of histologically confirmed liver disease. We report a significant 2.2- to 2.4-fold elevation in the proportion of endogenous UDCA in the bile of children with CFnoLD compared with CF children with liver disease or controls. This marked elevation in UDCA was also apparent in the circulation, as evidenced by a 3.8-fold increase in serum concentrations in all patients with CF.

UDCA is a secondary dihydroxylated bile acid reported to be present in humans as 1%–3% of the total bile acids (although higher amounts are often reported) and is formed in the gut by intestinal bacteria by epimerization of 7β-hydroxyl of the primary bile acid CDCA.29, 30 However, there is some evidence that UDCA can also be synthesized in the liver, because 7β-hydroxylated bile acid intermediates are produced in human liver.31 Because endogenous biliary UDCA is significantly increased in CFnoLD patients compared with both CFLD patients and controls, we hypothesize that more UDCA may be synthesized and secreted into the canaliculus in CFnoLD. UDCA is presumably secreted into the canaliculus through the bile salt export pump32, 33; to our knowledge, no specific UDCA transporter has been described. UDCA administration has been shown to increase the expression of the bile salt export pump34, 35 and thus confers some protective benefits.36 This increase in endogenous UDCA in children with CFnoLD may provide some protection to hepatocytes and cholangiocytes from the effects of cholestasis, in what is otherwise a rather hydrophobic bile acid pool. Why only some patients with CF appear to benefit from this effect is unclear, but further investigation is warranted.

Few patients with CF require limited intestinal resection at birth as a complication of meconium ileus. Although meconium ileus was once considered a risk factor for development of CFLD,6 a limited resection of terminal ileum is unlikely to significantly impact on the enterohepatic circulation of UDCA, because unconjugated UDCA can also be absorbed by passive nonionic diffusion, mainly in the proximal intestine. It is also noteworthy that the frequency of meconium ileus and the specific occurrence of surgical resection between those children with CF both with and without CFLD were similar.

The effects of oral UDCA have been extensively reviewed.29, 30, 36, 37 However, several are worthy of highlighting in the context of CFLD,9–17 a disease associated with abnormal chloride secretion causing malhydration of epithelial secretions, cholestasis, and retention of more hepatotoxic bile acids than normal.38 UDCA directly stimulates a secondary, non-CFTR, calcium-sensitive chloride channel in biliary epithelium39 and up-regulates the expression of the chloride-bicarbonate anion exchanger isoform-2.40 UDCA stimulates bile acid secretion through the bile acid–dependent and bile acid–independent (organic anion) transport pathways to stimulate bile flow. In this regard, it attempts to redress the basic defect in CFTR to hydrate secretions and improve bile flow. UDCA has direct cytoprotective effects on hepatocytes and bile duct epithelial cells, particularly in response to cellular disruption from bile acids; as a bile acid with low toxicity, it can serve to displace and dilute other more toxic hydrophobic bile acids. UDCA therefore seems well suited to offer hepatoprotection to children with CF for whom retained, toxic bile acids may contribute to the evolution of CFLD. Although there is supportive clinical evidence for the use of UDCA in CFLD, no study has been able to demonstrate improvement in the natural history of CFLD—particularly amelioration of the complications of portal hypertension, need for liver transplantation, or a measurable survival benefit. The hurdles of small patient numbers, disease heterogeneity, long natural history, imprecise monitoring tools, and lack of early markers for CFLD detection make such studies elusive. The study of Colombo et al.9 of UDCA treatment in patients with clinical CFLD demonstrated the necessity to enrich biliary bile with UDCA to magnitudes over 30% of the total bile acid pool to achieve optimal improvement in biochemical indices of CFLD. However, our observation of a significant increase in endogenous UDCA in the bile of at-risk patients with CF who have avoided detectable liver disease is significant. To provide a protective effect in CFnoLD, it is not necessary to postulate the need to reach biliary enrichment levels required for therapeutic effect in patients with established liver disease. We speculate that the early, endogenous increase in UDCA may be hepatoprotective, even at levels considered subtherapeutic in those with established CFLD. Clearly further studies are required to fully understand the pathophysiological mechanisms associated with the pathways of hepatic UDCA uptake, primary UDCA synthesis, intracellular transport, and biliary excretion.

This study also demonstrated a reduction in the secondary bile acid LCA in all three body compartments of children with CF, both CFLD and CFnoLD, vs. controls. Low LCA levels contributed to the increase in the proportion of primary bile acids CA and CDCA in both duodenal bile and serum of children with CF, both CFLD and CFnoLD, vs. controls. Decreased LCA may be associated with the increased use of antibiotics in patients with CF, although most patients were not receiving antibiotics immediately prior to sample collection. Furthermore, the increased synthesis of primary bile acids secondary to the known increased fecal bile acid losses in CF41 may have contributed to a dilution of this bile acid. However, both factors would have led to a commensurate decrease in DCA, the most common secondary bile acid, and this was not seen. A more satisfactory explanation would be the effect of relative cholestasis interrupting the enterohepatic circulation and failure of some primary bile acids to gain access to the intestinal lumen for bacterial dehydroxylation to secondary bile acids. This would expect to increase as cholestasis increases and is indeed seen in this study, with increased CA proportion correlating with increasing liver damage. Thus our data do not support the role of the secondary and more hepatotoxic bile acids, LCA and DCA, in the pathogenesis of fibrosis in CFLD. Bile acids other than those commonly found in bile, such as metabolites and derivatives of primary and secondary bile acids (see Table 2), may also be important in our understanding of the pathogenesis of CFLD, but this requires further investigation.

In serum, the total bile acid concentration in children with CF, both CFLD and CFnoLD, was nearly twice that of non-CF controls. This increase was mainly due to higher levels of CA and CDCA in CF patients and is similar to that of Robb et al.,38 who demonstrated increased total serum bile acids in a cohort of 55 children with CF, particularly older children, vs. controls, principally due to increased CA. Strandvik and Samuelson22 reported that serum CA and CDCA were significantly increased in CF patients having normal or nonspecific changes in liver pathology vs. healthy controls (CA, 0.94 ± 0.95 μM vs. 0.40 ± 0.0.24 μM; CDCA, 2.28 ± 1.80 μM vs. 0.99 ± 0.70 μM). These differences were also evident in their cohort of CF patients having fibrosis or cirrhosis vs. controls. The results of the present study (see Table 2) resemble those reported by Strandvik and Samuelson.22 In contrast to these findings, early studies by Sandberg19 in patients with CF showed no difference in serum bile acid concentrations vs. non-CF controls. Setchell et al.23 also reported no differences in total serum bile acids in 10 CFnoLD patients vs. four controls, but reported a significantly lower CDCA (0.98 ± 0.51 μM vs. 1.69 ± 0.0.84 μM). These differences between studies are likely to be due to differences in study populations and sample size. The strengths of the present study are its larger number of patients, better characterization of liver disease, and choice of appropriate control groups. Increased total serum bile acid levels are a marker of cholestasis and suggest that children with CFnoLD also have a degree of cholestasis, which is supported by the primary bile acid predominance in this cohort.

In patients with CFLD, only CA and UDCA levels were significantly associated with indicators of liver injury on biopsy (Scheuer's fibrosis stage, inflammation score, or LPD score). The significant association between the proportion of CA in the serum and all three parameters of liver injury indicate that serum CA may potentially be a good indicator of predicting hepatic injury. Support for a role for CA in monitoring liver disease progression comes from studies in animal models of cirrhosis. Azer and colleagues42 found CA increased with progression of hepatic injury and that the ratio of CA/CDCA was a sensitive indicator of cirrhosis development. This group43 also demonstrated that serum CDCA levels and CA/CDCA were good indicators for predicting the course (survival or need for transplantation) of chronic cholestatic liver disease. In our study, although there were no differences in serum CA/CDCA between the three groups (CFLD, CFnoLD, and non-CF), we did demonstrate a significant association between serum CA/CDCA (each expressed as a percentage of the total) and both the stage of fibrosis and the inflammation score. These data suggest that the CA/CDCA ratio may be useful in predicting the severity of hepatic fibrosis in CFLD. However, these results are based on a relatively small patient population; thus longitudinal studies involving larger patient numbers are required to better evaluate the use of serum CA and CA/CDCA as effective markers of CFLD progression.

In conclusion, this study has characterized the bile acid profiles of children with CF and CFLD and demonstrated significant alterations in bile acid composition from normal control children. Elevated endogenous UDCA in children with CF without liver disease suggests the possibility of a successful adaptive response to cholestasis with a protective role against liver injury in these patients. If so, it provides indirect support to the therapeutic exogenous enrichment of CF bile with UDCA in patients with CFLD, especially if it is detectable and treated early. The correlation between both CA and CA/CDCA levels with histological liver injury and fibrosis progression suggests a potential monitoring role in CFLD.


We are grateful for the assistance provided by Mr. Mike Franklin for his very kind gift of the Bond Elut C18 columns and expertise used in bile acid extraction from human samples. We also acknowledge Sonia Greco for the collection, processing, and storage of serum, urine and bile samples.