Serum bile acids in patients with liver failure supported with a bioartificial liver


Correspondence to: Dr M. Muraca, Department of Medical and Surgical Sciences, University of Padova, Via Giustiniani, 2, I-35128 Padova, Italy. E-mail:


Background : Serum bile acids are increased in liver failure, but the composition of the bile acid pool in this condition has not been studied in detail. This information is of interest because of dihydroxy bile acid toxicity.

Methods : We measured serum bile acids by gas chromatography–mass spectrometry in 13 patients with fulminant liver failure and five patients with acute-on-chronic liver failure. Furthermore, serum bile acids were analysed in the same patients after 6 h of treatment with a bioartificial liver, consisting of a hollow-fibre cartridge with microcarrier-attached porcine hepatocytes and a charcoal column.

Results : Pre-bioartificial liver serum bile acids demonstrated a high dihydroxy/trihydroxy ratio and were higher in patients with acute-on-chronic liver failure than in those with fulminant liver failure (452.8 ± 98.6 vs. 182.1 ± 39.7 µmol/L; P < 0.05). Bioartificial liver treatment decreased significantly serum bile acids in patients with fulminant liver failure (−38.8%) and acute-on-chronic liver failure (−35.8%), with a decreased dihydroxy/trihydroxy ratio. In vitro, porcine hepatocytes in the bioreactor cleared most conjugated bile acid species from pooled patient plasma.

Conclusions : Acute liver failure is associated with very high serum levels of toxic bile acids that could contribute to the pathogenesis of the syndrome. Bioartificial liver treatment reduces both serum bile acid concentrations and the hydrophobicity of the bile acid pool.


Liver failure is associated with a high mortality and, at present, the only effective treatment is liver transplantation.1, 2 The occurrence of this process in a previously healthy liver is termed fulminant liver failure, whilst the decompensation of chronic liver disease is termed acute-on-chronic liver failure. In patients with liver diseases, changes in serum bile acids are recognized as indicators of liver damage and predictors of prognosis.3, 4 However, few data are available on the serum bile acid profile in patients with liver failure. Such information could be of value, as hydrophobic bile acids could contribute to the pathogenesis of the syndrome, and various artificial devices have been proposed to remove toxic compounds from patient plasma.1 More recently, bioartificial liver support systems have emerged as a promising therapeutic modality that could substantially modify the course and prognosis of liver failure.5–7 The bioartificial liver which is furthest along the clinical testing process is the HepatAssist™ (Circe Biomedical, Inc., Lexington, MA, USA): a hybrid system for the treatment of patient plasma including a hollow-fibre bioreactor loaded with pig hepatocytes and a charcoal column.8 It has been developed at Cedars-Sinai Medical Center in Los Angeles and is the first cell-based device tested in a Food and Drug Administration-approved Phase II/III clinical trial.9 However, the effect of bioartificial liver therapy on the serum bile acid profile in patients with liver failure has not yet been studied. In this work, serum individual bile acids were accurately quantified using gas chromatography–mass spectrometry in a series of patients with liver failure, and the changes induced by treatment with a bioartificial liver were also analysed. In order to better define the effect of the bioartificial liver on the bile acid concentration, changes in bile acid levels were studied in vitro during perfusion of the hepatocyte cartridge with plasma obtained from patients with liver failure.

Materials and methods

In vitro study

Pooled plasma was obtained from patients with fulminant liver failure subjected to plasma exchange and from patients with acute exacerbation of chronic liver disease (acute-on-chronic liver failure). Hollow-fibre Zymax modules were purchased from Microgon, Inc. (Rancho Dominguez, CA, USA). A polycarbonate cylinder (inside diameter, 29.1 cm; outside diameter, 31.2 cm) contained 670 cellulose nitrate/cellulose acetate fibres (inside diameter, 635 µm; outside diameter, 760 µm; wall thickness, 62.5 µm; overall length, 510 mm; potted length, 445 mm) and had an extra-fibre volume of 177 mL. The total fibre internal surface area was 5850 cm2, external surface area was 7010 cm2 and the pore diameter in the semipermeable fibre wall was 0.2 µm. Porcine hepatocytes were prepared by two-step liver digestion, as described previously.10 The viability of the cells was > 90%, as measured by the trypan blue exclusion test. Five billion hepatocytes were incubated with collagen-coated dextran microcarriers (Cytodex-3, Pharmacia Fine Chemicals, Uppsala, Sweden) and inoculated into the extra-fibre compartment of the Zymax module, as described previously.11 In control experiments, plasma was perfused through the hollow-fibre modules filled with microcarriers only. Bioreactors were incorporated into a sealed perfusion circuit, consisting of a reservoir, a roller pump (Masterflex) and silicone gas-permeable tubing, submerged in a warm (39 °C) water bath aerated with 30% O2−5% CO2. Plasma (500 mL) was recirculated at 400 mL/min for 6 h. Plasma samples were collected from the reservoir at hourly intervals and stored at −25 °C until analysis.12, 13 The high bile acid levels in pooled plasma allowed the analysis of the samples by high performance liquid chromatography, in order to monitor changes in the concentrations of individual conjugated bile acids. Plasma samples (0.1 mL) were diluted with an equal volume of methanol, and 20 µL aliquots were injected into the liquid chromatograph and analysed as described previously.12 Samples of pooled plasma were also analysed by gas chromatography–mass spectrometry at both baseline conditions and at the end of the experiment. The concentration of each bile acid species measured by thelatter method was compared with the sum of the concentrations of the corresponding glycine and taurine conjugated moieties determined by high performance liquid chromatography. Differences in individual bile acid concentrations between the two procedures were in the range 6–11% in basal plasma and 5–13% at the end of the experiment.

In vivo study

Thirteen patients with fulminant liver failure (median age, 35 years; range, 16–56 years; male/female ratio, 5/8) and five patients with acute-on-chronic liver failure (median age, 55 years; range, 39–70 years; male/female ratio, 3/2) were enrolled. Patients entered the study when stage III–IV encephalopathy developed whilst they were receiving standard medical therapy. All patients with fulminant liver failure developed stage IV encephalopathy and were candidates for urgent liver transplantation. The aetiology of fulminant liver failure included viral hepatitis B (n = 3), acetaminophen poisoning (n = 2), ischaemic liver insufficiency (n = 2) and fulminant liver failure of an indeterminate aetiology (n = 6). The term ‘indeterminate’ refers to patients who clinically appear similar to those with fulminant liver failure of viral aetiology, but in whom no viral markers nor other possible aetiological factors can be identified. Ischaemic aetiology refers to low flow to the liver secondary to dehydration due to heatstroke and pancreatitis; in the two patients in the present series, it developed 3 and 4 days, respectively, before bioartificial liver treatment. The two patients with toxic aetiology had taken 50 g of acetaminophen 2 and 5 days, respectively, before bioartificial liver treatment. In the patients with viral or indeterminate aetiology, the interval between jaundice and encephalopathy ranged between 20 and 28 days. Seven patients with fulminant liver failure underwent monitoring of intracranial pressure, which ranged between 20 and 38 mmHg. The acute-on-chronic liver failure group included three patients with virus-related liver cirrhosis (two hepatitis C virus and one hepatitis B virus), one patient with primary biliary cirrhosis and one patient with autoimmune-related liver cirrhosis. In this group, stage III–IV coma unresponsive to conventional medical treatment developed 24–48 h before bioartificial liver treatment.

Bioartificial liver therapy

All procedures were conducted in full compliance with the standards of the Institutional Committee for the Protection of Human Subjects, in accordance with the 1975 Declaration of Helsinki. Blood was removed from the patient at 90–100 mL/min through a double-lumen catheter placed in the superficial femoral vein. Blood was separated into blood cells and plasma using a Spectra plasma separator (Cobe, Lakewood, CO, USA). Plasma entered the bioartificial liver system (HepatAssist 2000; Circe Biomedical, Lexington, MA, USA) at 50 mL/min; the high-flow plasma recirculation loop consisted of a reservoir, a column loaded with activated cellulose-coated charcoal (Adsorba 300C; Gambro, Hechingen, Germany), an oxygenator (Capiox 308; Terumo Inc., Tokyo, Japan), a heater (Temp Marg, Marquest, Lakewood, CO, USA) and a hollow-fibre module. Five billion viable porcine hepatocytes were aggregated using collagen-coated dextran microcarriers (Cytodex 3; Pharmacia, Piscataway, NJ, USA) and seeded into the extra-fibre space of the hollow-fibre cartridge. Plasma was circulated through the fibres at 400 mL/min. Sodium citrate was used as an anticoagulant to prevent thrombosis within the circuit. Plasma ionized calcium levels were measured hourly during bioartificial liver treatment to prevent ionized hypocalcaemia due to the calcium-chelating effect of citrate. Calcium chloride infusion was initiated at the start of bioartificial liver treatment. After return from the bioartificial liver, plasma and blood cells were reconstituted and returned to the patients by way of the double-lumen venous catheter. Blood samples were obtained before and immediately after bioartificial liver treatment, centrifuged at 3000 r.p.m. for 5 min and stored at − 25 °C until analysis.

Gas chromatography–mass spectrometry analysis of serum bile acids

The extraction of bile acids from serum was performed as described by Rodriguez & Setchell.13 After the addition of the internal standard, nordeoxycholic acid (2.5 µmol), 1 mL of serum was diluted with 4 mL triethylamine sulphate (0.5 m, pH 7.0). After heating at 64 °C for 20 min, bile acids were extracted with a Bond-Elut C18 cartridge (Analytichem, Harbor City, CA, USA) and recovered with methanol (5 mL). Solvolysis, hydrolysis and preparation of methylesters–trimethylsilyl ethers was carried out as described by Setchell et al.14 with minor modifications. Solvolysis was carried out in a mixture of methanol (1 mL), distilled tetrahydrofuran (9 mL) and 1 m trifluoroacetic acid in dioxane (0.1 mL) heated at 45 °C for 2 h. After evaporation, the residue was hydrolysed by overnight incubation with 54 units of choloylglycine hydrolase (Sigma Chemical Co., St. Louis, MO, USA) in 2.5 mL of 0.2 m phosphate buffer, pH 5.6, at 37 °C in a shaking bath. Hydrolysed bile acids were extracted with a Bond-Elut C18 cartridge and eluted with methanol (5 mL). Unconjugated bile acids were further purified of other contaminating sterols by diluting the extract with water to 72% methanol and by passage through a column of Lipidex-DEAP (inside diameter, 0.4 cm; length, 13 cm), followed by elution with 0.1 m acetic acid in 72% ethanol (7 mL) and evaporation under a nitrogen flow. The methyl esters were prepared by reacting the extract with 0.1 mL CH3-HCl for 20 min at 60 °C. After evaporation, the extract was further reacted overnight with 0.1 mL freshly prepared bis(trimethylsilyl)trifluoroacetamide–dimethylformamide at 50 °C. The methyl esters–trimethylsilyl ethers of the bile acids were analysed by gas chromatography–mass spectrometry using an MD-800 instrument (Fisons Instruments, Manchester, UK) housing a 30-m DB-1 (inside diameter, 4 µm; film thickness, 0.25 mm) fused silica capillary column (J & W Scientific Inc., Rancho Cordova, CA, USA), and with the use of a temperature programme operated from 220 to 265 °C in increments of 10 °C/min, followed by increments of 30 °C/min up to 290 °C, and a final isothermal period of 3 min. Helium was used as the carrier gas at a flow rate of 2 mL/min. Electron ionization (70 eV) mass spectra were recorded over the mass range 50–800 Da by repetitive scanning of the eluting components.

Statistical analysis

Data were analysed using the two-sided Wilcoxon matched-pair signed-rank test, the Mann–Whitney rank sum test and the Kruskal–Wallis non-parametric analysis of variance (anova) test, when appropriate. P values of ≤ 0.05 were considered to be significant. Data are presented as the mean ± standard error of the mean (S.E.M.).


In vivo study

Figure 1 shows a representative gas–liquid chromatogram of the methyl ester–trimethylsilyl ether derivatives of serum bile acids in a patient with fulminant liver failure. Bile acids could be well characterized from their mass spectra (two representative spectra are shown in Figure 2 ). Although serum bile acid levels were elevated in all patients studied, they were significantly higher in patients with acute-on-chronic liver failure than in those with fulminant liver failure (Table 1). However, the bile acid pool composition was similar in the two groups (Table 2). After bioartificial liver therapy, serum bile acid levels were reduced significantly and to a similar degree in both groups of patients (−38.8 ± 8.1% in fulminant liver failure and −35.8 ± 13.2% in acute-on-chronic liver failure; Table 1). In patients with fulminant liver failure, all bile acid species were significantly reduced after bioartificial liver treatment, with a significant decrease in the ratio of dihydroxy to trihydroxy bile acids (Table 1). The mean concentrations of all individual bile acids were also reduced in patients with acute-on-chronic liver failure, but the difference was significant only for lithocholic acid (Table 2). In these patients, the ratio of dihydroxy to trihydroxy bile acids was also reduced, although the difference did not reach the level of statistical significance, probably due to the small sample size (Table 1).

Figure 1.

Capillary gas–liquid chromatography profile of serum bile acids in a patient with fulminant liver failure. The peaks denote the following compounds as identified by gas–liquid chromatography–mass spectrometry: NDC, nordeoxycholic acid; LC, lithocholic acid; DC, deoxycholic acid; CDC, chenodeoxycholic acid; C, cholic acid; UDC, ursodeoxycholic acid.

Figure 2.

Mass spectra of cholic and deoxycholic acids.

Table 1.  Total serum bile acids and dihydroxy/trihydroxy (DiOH/TriOH) ratio in patients with acute liver failure before and after treatment with a bioartificial liver (BAL)
 Fulminant liver failure (n = 13)Acute-on-chronic liver failure (n = 5)
Serum bile acids (µmol/L)DiOH/TriOH ratioSerum bile acids (µmol/L)DiOH/TriOH ratio
  • *

    Different from fulminant liver failure, P  < 0.05.

  • Different from basal values, P  < 0.05.

Basal182.1 ± 39.74.85 ± 0.83452.8 ± 98.6*4.76 ± 1.93
After BAL111.5 ± 26.33.18 ± 0.44290.5 ± 64.0*3.35 ± 0.22
Table 2.  Individual serum bile acids in patients with fulminant liver failure (FLF) and acute-on-chronic liver failure (AoCLF) before and after treatment with a bioartificial liver (BAL)
  • C, cholate; CDC, chenodeoxycholate; DC, deoxycholate; LC, lithocholate; UDC, ursodeoxycholate.

  • *

    Different from basal values, P  < 0.05.

FLF (n = 13)
Basal14.70 ± 2.7311.10 ± 5.72  129 ± 30.9529.18 ± 9.09 0.58 ± 0.58
After BAL 6.62 ± 2.03* 3.39 ± 0.55*82.12 ± 21.83*20.03 ± 5.26* 0.49 ± 0.49*
AoCLF (n = 5)
Basal35.64 ± 11.06 8.14 ± 1.66281.6 ± 87.83100.7 ± 43.7214.70 ± 2.628
After BAL 7.50 ± 0.78* 5.62 ± 2.85237.1 ± 59.2483.96 ± 21.67 6.62 ± 2.027

In vitro study

Plasma bile acid levels ranged from 153 to 216 µmol/L. After 6 h of perfusion through the hepatocyte module, significant reductions in the levels of taurocholate (P < 0.01), taurochenodeoxycholate (P < 0.001), taurodeoxycholate (P < 0.05) and glycochenodeoxycholate (P < 0.05), with respect to baseline, were observed (Table 3). Furthermore, the fall in plasma bile acid concentrations was more pronounced during the initial 3 h than in the second half of the experiment. During control perfusions through the modules filled with microcarriers only, no significant changes in bile acid profile were observed (Table 3).

Table 3.  Bile acid levels (expressed as a percentage of the initial concentration) during perfusion of hollow-fibre modules loaded with microcarrier-attached hepatocytes ( n  = 5) or with microcarriers alone ( n  = 5)
  • GC, glycocholate; GCDC, glycochenodeoxycholate; GDC, glycodeoxycholate; TC, taurocholate; TCDC, taurochenodeoxycholate; TDC, taurodeoxycholate.

  • *

    P  < 0.05,

  • P  < 0.01,

  • P  < 0.001 (comparison between microcarrier-attached hepatocytes and microcarriers alone).

Microcarrier-attached hepatocytes
6046.9 ± 4.556.4 ± 2.480.4 ± 2.9*80.4 ± 12.386.3 ± 5.975.6 ± 10.3
12040.4 ± 4.353.5 ± 4.773.1 ± 2.9*75.9 ± 14.062.9 ± 3.2*59.2 ± 11.3
18041.9 ± 4.836.7 ± 5.665.3 ± 5.0*71.8 ± 16.958.0 ± 6.6*59.1 ± 11.7
24036.6 ± 8.632.4 ± 5.260.8 ± 6.8*68.3 ± 17.849.6 ± 4.758.2 ± 12.8
30034.7 ± 7.927.8 ± 4.452.4 ± 6.784.2 ± 10.454.8 ± 5.163.0 ± 11.9
36035.3 ± 8.321.2 ± 4.247.9 ± 10.0*73.1 ± 17.160.4 ± 9.2*64.7 ± 11.2
Microcarriers alone
6092.5 ± 4.387.6 ± 3.988.5 ± 2.791.4 ± 8.386.2 ± 4.484.5 ± 8.3
12087.1 ± 6.986.7 ± 3.284.8 ± 3.687.9 ± 7.281.9 ± 5.482.1 ± 3.7
18087.1 ± 3.789.0 ± 2.783.1 ± 3.486.7 ± 4.579.2 ± 3.886.5 ± 6.2
24089.5 ± 2.686.6 ± 6.290.9 ± 3.187.4 ± 6.883.5 ± 5.786.5 ± 6.5
30090.9 ± 3.491.0 ± 2.491.4 ± 4.188.7 ± 6.386.3 ± 3.687.0 ± 4.4
36091.2 ± 1.890.8 ± 4.191.1 ± 4.091.0 ± 5.188.8 ± 3.288.2 ± 3.9


According to the American Liver Foundation, 5 million Americans have viral hepatitis, and liver failure is the seventh leading cause of death in the USA (43 000 deaths in 1998). The vast majority of these events are due to acute decompensation of chronic liver disease, whilst fulminant liver failure is responsible for approximately 2000 cases. Although the two syndromes differ markedly in terms of clinical presentation and pathogenesis, in both conditions the loss of hepatocellular function and the accumulation of toxins in blood and tissues are believed to be the main factors leading to multiorgan failure and death.1 Despite conflicting evidence on the role of individual putative toxins, experimental evidence in animals suggests that at least some of them can act synergistically to produce organ damage.1

Endogenous bile acids accumulate in the body in both acute and chronic liver diseases, and are known to be toxic to hepatocytes, bile duct epithelium and other organ systems.15–17 Moreover, these compounds can inhibit DNA synthesis and liver regeneration.18 Normal serum bile acid concentrations, as determined by gas chromatography–mass spectrometry, range between 1.5 and 2.4 µmol/L,19 and their blood and tissue concentrations can increase dramatically in some forms of liver disease.3, 4, 20, 21 The levels of serum bile acids seem to correlate with the severity of the disease,22–25 although patients with liver failure have not been included in these and similar studies. The major determinant of the increased serum bile acid levels in advanced hepatocellular disease seems to be the altered distribution of hepatic blood flow in the presence of chole-hepatic and porto-systemic shunts.26, 27 Altered hepatocyte function, resulting in decreased hepatic uptake and biliary excretion, can also play a role.26, 28, 29 The loss of functional liver mass per se does not significantly affect serum bile acid levels until very late in the course of the disease.30 Other abnormalities in advanced liver disease include a decreased bile acid pool size, mainly due to a reduced amount of cholic acid.31–34 This results in an increased dihydroxy/ trihydroxy ratio and a higher detergency and toxicity of serum and tissue bile acids.35, 36 A dihydroxy bile acid, such as glycochenodeoxycholic acid, can induce apoptosis of cultured hepatocytes at concentrations ranging between 50 and 200 µm, whilst higher concentrations result mainly in cell necrosis.16 Moreover, chenodeoxycholic acid aggravates bilirubin-induced apoptosis in cultured rat neural cells.37

The present study is the first to employ gas chromatography–mass spectrometry analysis for the accurate identification and measurement of individual serum bile acid species in patients with liver failure. In the past, an increase in total serum bile acids of up to 150 µmol/L was observed in 10 patients with fulminant liver failure, and chenodeoxycholic acid accounted for 80% and cholic acid for 9% of the total bile acids in the plasma ofa single patient with fulminant hepatitis, as measuredby gas–liquid chromatography without mass spectrometry.38, 39 In the present series of patients with fulminant liver failure, serum bile acid levels were approximately 200 µm, with a striking prevalence of dihydroxy bile acids (dihydroxy/trihydroxy ratio, 4 : 1). In particular, chenic acid concentrations averaged 130 µm, a level which can induce apoptosis in 60% of hepatocytes after 4 h of incubation in vitro.16 An increase in total serum bile acids38 and chenodeoxycholic acid39 in fulminant hepatitis was observed using non-chromatographic methods. In the present work, gas chromatography–mass spectrometry analysis of serum revealed bile acid concentrations much higher than those reported previously; in particular, very high levels of the most toxic compounds, deoxycholic and lithocholic acids, were observed. To our knowledge, the pattern of serum bile acids has never been reported in patients with acute-on-chronic liver disease. Concentrations of dihydroxy bile acids, similar to those observed in the present series of patients with acute-on-chronic liver failure, can readily induce necrosis in cultured hepatocytes.16 Thus, in patients with liver failure, the serum bile acid levels could further aggravate the disease, leading to a vicious circle similar to that implicated in cholestasis.40 As noted above, serum bile acid levels were about twice as high in patients with acute-on-chronic liver failure as in those with fulminant liver failure. The reasons for this difference are not clear. In decompensated cirrhosis, deficient hepatocyte function with reduced transport of bile acids from plasma to bile might be the main mechanism leading to the accumulation of circulating bile acids. In fulminant liver failure, in turn, the production of bile acids could be lowered due to a more profound reduction in functional hepatic mass, resulting in relatively lower serum bile acid concentrations.

The reduction of serum bile acids after bioartificial liver treatment could have partially contributed to the clinical improvement observed with the HepatAssist system.8, 9 As the HepatAssist is a hybrid system, the relative contributions of pig hepatocytes and of the charcoal column to the observed clearance of serum bile acids cannot be determined. Although the binding of serum bile acids to charcoal has been well documented,41, 42 no changes in bile acid profile were observed during perfusion through a charcoal column.43 Moreover, in the present in vitro studies, the charcoal column was not included in the plasma circuit. Thus, the significant decrease in bile acid levels observed during perfusion of hepatocyte-loaded modules in vitro can be attributed to hepatocyte bile acid handling only. Finally, we have recently reported that porcine hepatocytes in the bioartificial liver clear cholate and taurocholate from plasma and conjugate cholic acid.44 Based on these findings, we conclude that, although the charcoal column can increase the detoxifying capacity of the bioartificial liver system, porcine hepatocytes seem to play a pivotal role in the clearing of bile acids from the circulation and in decreasing the hydrophobicity of the bile acid pool.

In conclusion, this study demonstrated the presence of very high levels of toxic bile acids in patients with liver failure. In theory, this could have aggravated liver injury and contributed to the pathogenesis of the syndrome. Treatment with a bioartificial liver lowered serum bile acids and decreased the dihydroxy/trihydroxy ratio, thereby reducing the toxicity of circulating bile acids. However, additional studies are needed to determine whether the beneficial effects of bioartificial liver therapy in patients with liver failure could be attributed, at least to some degree, to the removal of toxic compounds, such as detergent bile acids, from the blood circulation.


This work was financed in part by the Fondazione per l'Incremento dei Trapianti d'Organo and by the Regione Veneto.