Plasma fetal bile acids 7α-hydroxy-3-oxochol-4-en-24-oic acid and 3-oxachola-4,6-dien-24-oic acid indicate severity of liver cirrhosis

Two 3-oxo-Δ4 fetal bile acids, 3-oxachola-4,6-dien-24-oic acid (1) and 7α-hydroxy-3-oxochol-4-en-24-oic acid (2), occur normally in the human fetus but remain elevated in neonates and children with severe cholestatic liver disease due to an autosomal recessive inborn error of metabolism affecting Δ4-3-oxo-steroid 5β-reductase (AKR1D1). Relatively little is known about 1 and 2 in adult patients with liver disease. The chemical synthesis of 1 and 2 is therefore described and their quantitation in plasma by ultrarapid chromatography-triple quadrupole mass spectrometry. Plasma concentrations of 1 and 2 were investigated in 25 adult patients with varying degrees of liver cirrhosis with and without hepatocellular carcinoma (HCC). Highly statistically significant correlations (P < 0.0001) were found between severity of liver cirrhosis, determined by the Child–Pugh and MELD scores, with plasma 1 and 2 concentrations, both alone and combined. The presence of HCC did not influence these correlations. Plasma cholic, chenodeoxycholic, deoxycholic, lithocholic or ursodeoxycholic acids, free and as their glycine or taurine conjugates, did not correlate with Child–Pugh or MELD score when corrected for multiple comparisons. These findings demonstrate that plasma levels of fetal bile acids 3-oxachola-4,6-dien-24-oic acid and 7α-hydroxy-3-oxochol-4-en-24-oic acid and likely deteriorating AKR1D1 activity indicate the severity of liver cirrhosis measured by the Child–Pugh and MELD scores.


Results
Twenty-five patients with LC, comprising 10 with HCC and 15 without, were investigated. The severity of cirrhosis was represented across the full range of Child-Pugh scores 16 from 5 to 15 in these patients. Clinical details of the ten HCC and 15 LC patients are given in Tables 1 and 2, respectively. Table 3 shows a comparison of clinical data between LC and HCC patients. No confounders were apparent from these data. Of note is the observation that both fetal bile acids 1 and 2 were detected and quantitated in plasma from all 25 cirrhotic patients. In the current investigation, all 25 patients investigated had varying degrees of LC. Examination of their plasma by UPLC-QTOFMS yielded concentrations of fetal bile acids 1 and 2 and we analyzed the findings in relation to both Child-Pugh classes A, B and C together with the Child-Pugh scores from 5 to 15, which represent the severity of the cirrhosis 16 . Figure 2 displays the relationships between Child-Pugh class and score with plasma concentrations of fetal bile acids 1 and 2. Comparison between Child-Pugh A and Child-Pugh B/C for all cirrhotic patients with and without HCC revealed statistically significant differences for the plasma concentrations  www.nature.com/scientificreports/ of both fetal bile acid 1 (median values 49 and 249, respectively; P < 0.0001) and fetal bile acid 2 (median values 70 and 161, respectively; P = 0.002). As can be seen from these findings, plasma concentrations of both fetal bile acids correlated increased significantly with severity of cirrhosis as measured by Child-Pugh class ( Fig. 2A,B) and correlated highly statistically significantly (P < 0.0001) with severity of cirrhosis as determined by the Child-Pugh score (Fig. 2C,D). Moreover, the total fetal bile acid plasma concentration (FBA1 + FBA2) correlated highly statistically significantly (P < 0.0001) with Child-Pugh score (Fig. 2E). When nonparametric correlation was used for the data in Fig. 2C-E, Spearman rank correlation coefficients of 0.87, 0.74 and 0.87, respectively, were found, which were highly statistically significant (P < 0.0001). In contrast to the Child-Pugh score, the MELD score (Model for Endstage Liver Disease) ranges from 6 to 40 and is a validated predictor of survival in end-stage liver disease, such as decompensated cirrhosis. Plasma concentrations of 1 and 2 were statistically significantly higher in patients with MELD scores ≥ 15 than those with values < 15 17 (Fig. 3A,B). The value of 15 has been reported as a cutoff for predicting short-term survival. That notwithstanding, plasma concentrations of FBA1, FBA2 and FBA1 + FBA2 all correlated statistically significantly with MELD score (Fig. 3C-E). When nonparametric correlation was used for data in Fig. 3C-E, Spearman rank correlation coefficients of 0.76, 0.61 and 0.73, respectively, were found, which were highly statistically significant (P < 0.0001 to 0.001).
In addition, the 3-oxo-Δ 4 steroids shown in Fig. 1 are both precursors of CDCA that are metabolized by AKR1D1, 7α-hydroxy-cholest-4-en-3-one early in the neutral pathway and 7α-hydroxy-3-oxochol-4-en-24-oic acid (2) late in the acidic pathway 13 . Statistical analysis of product/substrate ratios in patient plasma, specifically, CDCA/FBA1, CDCA/FBA2 and CDCA/(FBA1 + FBA2) revealed no statistically significant correlations with Child-Pugh scores (data not shown). Importantly, the presence of HCC did not affect the plasma concentrations of the fetal bile acids, which appeared to be related to severity of cirrhosis as determined by the Child-Pugh criteria. For FBA1, median plasma concentrations with and without HCC were 53.8 and 83.1 ng/mL, respectively (P = 0.50). For FBA2, these values were 72.1 and 85.6 ng/mL, respectively (P = 0.64). Moreover, plasma www.nature.com/scientificreports/ concentrations of alpha-fetoprotein (AFP; Tables 1,2), a protein made by fetal liver and used as a diagnostic aid for HCC 18 , did not correlate with plasma FBA1 and FBA2 concentrations, either alone or combined (data not shown). AFP values, as expected were statistically significantly higher for patients with HCC (P = 0.007).
In addition to the determination of plasma fetal bile acids, 15 primary and secondary bile acids (free and glycine and taurine conjugated) were determined in patient plasma by UPLC-QTOFMS. Figure 4 depicts area responses for each of these plasma bile acids in relation to the Child-Pugh classes A, B and C.
Moreover, correlations for the 15 free and conjugated primary and secondary bile acids were also sought using the Spearman rank correlation due to the non-Gaussian distribution of the bile acid and MELD score data. After Bonferroni correction for multiple comparisons, only the taurine conjugated primary bile acids TCA (Spearman r = 0.669; adjusted P = 0.008) and TCDCA (Spearman r = 0.602; adjusted P = 0.03) correlated statistically significantly. These were stronger correlations than those with the Child-Pugh score). The other 13 bile acids did not correlate with the MELD score with Spearman r values from 0.029 to 0.469 (adjusted P values from 0.08 to 1.0). In addition, an analysis of the 15 bile acids was undertaken in relation to the presence or absence of each of ascites, encephalopathy and esophageal varices. None of these 45 analyses conducted using the nonparametric Mann-Whitney U test produced statistically significantly differences after Bonferroni correction for multiple comparisons.
As for the fetal bile acids in Figs. 2 and 3, the presence of HCC did not influence the relationship between plasma bile acids and Child-Pugh class. Although the conjugated primary bile acids appeared to increase with increasing cirrhosis severity, when P-values were adjusted for multiple comparisons using the Bonferroni correction, no statistically significant relationships remained for these primary bile acid conjugates. In addition, www.nature.com/scientificreports/ when these 15 primary and secondary bile acids were compared between liver cirrhosis patients with and without HCC, there were no statistically significant differences found (Table 4).
In summary, plasma concentrations of fetal bile acids 1 and 2 alone and combined were highly correlated with cirrhosis severity (Figs. 2, 3). This was not so for the 15 primary and secondary bile acids also determined in patient plasma (Fig. 4). It is therefore probable that the metabolic activity of AKR1D1 declines with increasing severity of cirrhosis in a manner that is unaffected by the presence of hepatic tumors.  www.nature.com/scientificreports/

Discussion
As their name suggests, fetal bile acids occur predominantly in the fetus and are known to decline significantly in the first few days after birth 9 . However, elevated concentrations of these 3-oxo-Δ 4 bile acids have been detected in neonates and children with congenital Δ 4 -3-oxosteroid 5β-reductase (AKR1D1) deficiency 10,11,19,20 . Tentative identification of fetal bile acids 1 and 2 in adults was made in a plasma metabolomics and lipid profiling investigation of HCC patients that used age-matched AML cases as controls. This identification was made on the basis of mass spectroscopic characteristics without authentic chemical standards for 1 and 2 7 . Apart from this report of 20 HCC patients vs. 22 acute myeloid leukemia (AML) controls 7 , these fetal bile acids had only previously been reported in neonates 11,20 . Because these authentic standards are not commercially available, a detailed account of their chemical synthesis is provided here, together with characterization of the intermediates by 1 H NMR and of the two final products 1 and 2 by 1 H NMR and high-resolution mass spectrometry. Determination of 1 and 2 in children and neonates with Δ 4 -3-oxosteroid 5β-reductase (AKR1D1) deficiency 10,11,19,20 has typically been performed using urinary analysis by negative ion fast atom bombardment mass spectrometry (FAB-MS) and gas chromatography-mass spectrometry (GC-MS) in a complex assay that involved hydrolysis of bile acid conjugates in urine after overnight incubation with bile salt hydrolase and double derivatization by methylation and trimethylsilylation 10,11 . We describe here for the first time a rapid and simple quantitative UPLC-TQMS assay for the fetal bile acids 1 and 2, which has been utilized to investigate plasma from patients with various degrees of liver cirrhosis, both with and without HCC. This investigation established several unique conclusions. First, plasma concentration of the two fetal bile acids was not related to the presence of hepatic tumors. This could be due to small patient numbers, requiring a larger study to solve this issue. In a previous metabolomic study 7 , plasma levels of 1 and 2 were upregulated relative to AML but it was unclear if this was due to HCC or LC. Second, plasma concentrations of the fetal bile acids correlated strongly with increasing both Child-Pugh and MELD scores, indicating a relationship between these bile acids and severity of cirrhosis. Third, a further 15 both primary and secondary bile acids and their glycine and taurine conjugates were neither correlated with Child-Pugh classes A, B and C nor with the MELD score, once allowance for multiple comparisons was made. These findings were also unrelated to the presence of HCC. These results point to plasma fetal bile acids as the only indication of severity of cirrhosis known so far among the bile acid family.
The highlights of this study were that we describe a step-by-step chemical synthesis of these two fetal bile acids, which are not commercially available. In addition, we have described a simple, rapid and quantitative liquid chromatography-tandem mass spectrometry assay for their quantitation in human plasma. We have applied these novel methodologies to the investigation of hepatic cirrhosis. The limitations of our study concern this last component. First, we recruited relatively few patients (n = 25) with HCC (n = 10) and without HCC (n = 15). This group of patients was highly heterogeneous in relation to their complications arising from decompensation, with encephalopathy (n = 7), ascites (n = 8) and esophageal varices (n = 9). Notwithstanding this heterogeneity, clear correlations between each fetal bile acid and both the Child-Pugh score and the MELD score were obtained. A further limitation of this investigation is that liver tissue was not available from the patients and therefore it was not possible to determine AKR1D1 expression by qPCR or AKR1D1 protein by western blotting. This would have permitted us to confirm the role of AKR1D1 in fetal bile acid disposition, which we posited is compromised with increasing severity of cirrhosis.
Because of the large volume of research on fetal bile acids and inborn Δ 4 -3-oxosteroid 5β-reductase (AKR1D1) deficiency, it is tempting to propose that deteriorating liver function may also affect the activity of this hepatic enzyme leading to elevated fetal bile acids as a phenocopy of the inborn error of metabolism. It has been postulated 10 that excessive amounts of Δ 4 -3-oxo bile acids might also be expected to occur in children with severe liver diseases such as hepatitis B and autoimmune chronic active hepatitis but not in more benign chronic cholestatic syndromes such as biliary atresia. In this report, a single patient aged 15 months with a deficiency in fumarylacetoacetase (FAH), known as hereditary tyrosinemia type I that is associated with liver cirrhosis and fatal HCC at a young age 21 , had the highest ratio of 7α,12α-dihydroxy-3-oxo-4-cholenoic acid (a fetal bile acid) to cholic acid in a series of 25 children and neonates with various liver diseases 10 . These authors suggested that an altered redox state in hepatocytes or an unusual sensitivity of Δ 4 -3-oxosteroid 5β-reductase to liver damage might explain their findings. Elevated plasma bile acids, in particular, both free and conjugated CDCA, have been reported to be elevated in cirrhotic patients relative to healthy volunteers 22 . Total bile acids and total bile acids/cholesterol ratio were found to be elevated in 451 cirrhotic patients compared with 216 non-cirrhotic patients with non-cholestatic chronic HBV infection in China 23 . An investigation of 32 cirrhotic patients and 27 healthy volunteers examined 12 serum bile acids using UPLC-TQMS and reported that serum concentrations of the conjugated primary bile acids, taurocholic acid (TCA), glycocholic acid (GCA) and taurochenodeoxycholic acid (TCDCA) were positively associated with Child-Pugh classification 24 . Our own data would appear to corroborate these observations, with uncorrected P-values for nonparametric analyses of variance (Kruskal-Wallis) of the relationships between plasma TCA, GCA and TCDCA with Child-Pugh class. However, when corrected for false discovery, these relationships ceased to remain statistically significant, as did any relationship between non-fetal bile acids and the MELD score.
In a study of alcoholic cirrhosis in heavy alcohol drinkers, striking increases in taurine conjugated bile acids in serum were observed (50-to 130-fold) with more moderate increases in glycine conjugated bile acids (10-to 20-fold). Increased levels of glycochenodeoxycholic acid (GCDCA), TCDCA, GCA, and TCA were positively correlated with disease progression from Child-Pugh A to C 25 . These dramatic findings were of a much higher order than those in our study. It would be of great interest to know how fetal bile acids concentrations were modified in such patients. www.nature.com/scientificreports/ Plasma or serum AFP concentrations are generally regarded as a useful addition to surveillance by ultrasound, CT or MRI imaging for the detection of early HCC and this has been established by a meta-analysis of 32 studies involving 13,367 patients 26 . Perusal of the raw data establishes that the AFP values for our patients are not diagnostic for the presence of HCC. AFP was detected in 9/15 cirrhotic patients without HCC with a median value of 2.6 ng/mL compared with 9.0 ng/mL for those with HCC. AFP is well known to vary in HCC from normal values (< 10 ng/mL) to extremely high values (> 100,000 ng/mL). When used diagnostically, an AFP of 20 ng/mL shows good sensitivity but low specificity and in contrast a value of 200 ng/mL has a high specificity but a sensitivity of only 22% 27 . It is interesting to propose that deteriorating liver function in the direction of HCC represents an evolving fetal phenotype, particularly when considering the fetal bile acid changes reported here. However, there was a lack of correlation between AFP and fetal bile acid concentrations. In addition, the inborn error of metabolism involving Δ 4 -3-oxosteroid 5β-reductase (AKR1D1) deficiency that leads to the accumulation and urinary excretion of fetal bile acids 1 and 2 11,19 , has not been investigated in relation to increased postnatal serum AFP concentrations. The role of an evolving fetal phenotype in LC that may lead to HCC is an open question.
The issue remains how does severity of liver cirrhosis influence the plasma concentration of fetal bile acids 1 and 2? A generalized increased bile acid synthesis, as eluded to above, may result in elevated plasma concentrations of the intermediates 1 and 2. Alternatively, as is the case with the inborn error of metabolism with genetically impaired Δ 4 -3-oxosteroid 5β-reductase deficiency, AKR1D1 activity may be impaired by an as yet undefined mechanism, leading to worsening cirrhosis. In human liver biopsies taken from 34 obese patients, AKR1D1 mRNA expression was reported to decrease with advancing steatosis, fibrosis and inflammation 14 . Glucocorticoids, including endogenous cortisol, both downregulate AKR1D1 and are metabolized by it 28 . A reduced plasma clearance of cortisol in cirrhosis has long been observed 29 , congruent with a slower metabolic turnover of cortisol in cirrhotic patients 30 . Taken together, these and other reports may provide an biochemical basis for the association of increased fetal bile acids in patients with advancing cirrhosis.

Materials and methods
Reagents. All reagents for the chemical syntheses that were conducted in South Korea were purchased from Sigma-Aldrich Korea Ltd. and Tokyo Chemical Industry Co., Ltd.
General methods for chemical purification and analysis. Product purification was conducted by preparative flash column chromatography using ZEOprep 60 unbonded silica (230-400 mesh). All NMR experiments were carried out using a Bruker Avance III 400 MHz NMR spectrometer equipped with a 5-mm broadband-observed probe head. Chemical shifts were measured relative to residual solvent peaks as an internal standard set to δ 7.26 (CDCl 3 ). Multiplicities in the 1 H NMR spectra as described as s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br. s = broad singlet. Liquid chromatography-mass spectra (LCMS) were measured in positive electrospray ionization (ESI) mode on a Waters Xevo G2-XS instrument. Figure 5 shows the scheme for the synthesis of fetal bile acids 1 and 2.

Synthesis of fetal bile acids.
Procedures a and b for the synthesis of 4 from 3 (chenodeoxycholic acid) were conducted as follows: a. Compound 3 (2.80 g, 7.13 mmol), formic acid (11 mL) and 70% perchloric acid (0.4 mL, 4.93 mmol) were mixed and the reaction stirred at 55 °C for 1.5 h. To the reaction mixture was slowly added acetic anhydride (9 mL) at 40 °C and then stirred at 55 °C until bubbling ceased. The mixture was then cooled to RT and slowly added to water and extracted with ethyl acetate. The separated organic phase was washed with brine, dried over anhydrous MgSO 4 , filtered and concentrated under reduced pressure. The resulting residue was purified using flash column chromatography (SiO 2 , 5% methanol in dichloromethane) to afford the 3,7-diformyl intermediate (0.57 g, 48%). b. To the 3,7-diformyl intermediate (0.22 g, 0.49 mmol) in acetone (4 mL) was slowly added 0.2 M sodium hydroxide (5 mL, 1.0 mmol) and the reaction mixture stirred at RT for I h, neutralized to pH 7 with acetic acid, diluted with water and extracted with dichloromethane. The separated organic phase was washed with brine, dried over anhydrous MgSO 4 , filtered and concentrated under reduced pressure. The resulting residue was purified using flash column chromatography (SiO 2 , 30% ethyl acetate in n-hexane) to afford compound 4 (0.11 g, 53%).
Procedure c for compound 5 was conducted as follows: To compound 4 (0.11 g; 0.26 mmol) in acetone (11 mL) was slowly added Jones' reagent (0.3 mL) at 4 °C and the reaction stirred at this temperature for 2 h, then at RT for 1 h, whereupon methanol (0.6 mL) was slowly added to the reaction mixture to form a suspension, which was filtered and washed with acetone. The combined filtrate was concentrated under reduced pressure and the resulting residue purified using flash column chromatography (SiO 2 , 30% ethyl acetate in n-hexane) to afford compound 5 (0.05 g, 46%).
Procedure d for compound 6 was conducted as follows: To compound 5 (0.55 g, 1.31 mmol) and p-toluenesulfonic acid (PTSA; 0.04 g, 0.22 mmol) in DMF (5 mL) was slowly added bromine (0.24 g, 1.50 mmol) in DMF (1 mL) at RT and stirred for 20 h. The mixture was then slowly added to water and extracted with ethyl acetate. The separated organic phase was washed with brine, dried over anhydrous MgSO 4 , filtered and concentrated under reduced pressure. The resulting residue was purified using flash column chromatography (SiO 2 , 40% ethyl acetate in n-hexane) to afford compound 6 (0.57 g, 87%).
Procedure e for compounds 1 and 7 was conducted as follows: A solution of semicarbazide.HCl (0.25 g, 2.24 mmol) and sodium acetate (0.18 g, 2.19 mmol) in water (1 mL) was added dropwise to compound 6 (0.41 g, 0.82 mmol) in acetic acid (8 mL) and the mixture stirred at 60 °C for 0.5 h, then stirred at RT for 1 h to form a precipitate. This solid was filtered, washed with pure water and used without further purification. This solid, acetic acid (9 mL), pyruvic acid (0.9 mL) and water (2.1 mL) were stirred at RT for 20 h. Precipitated solid was filtered off and the filtrate extracted with ethyl acetate, which was washed with brine and dried over anhydrous MgSO 4 , filtered and evaporated under reduced pressure. The resulting residue was purified using flash column www.nature.com/scientificreports/ chromatography (SiO 2 , 50% ethyl acetate in n-hexane) to afford compound 1 (0.025 g, 8%) and compound 7 (0.085 g, 26%). Procedure f for compound 2 was conducted as follows: Compound 7 (0.085 g, 0.21 mmol) was dissolved in 0.5 M sodium hydroxide (10 mL) and stirred at RT for 1.5 h. The reaction mixture was acidified to pH 1-2 with 1 M HCl and extracted with ethyl acetate. The combined organic phase was washed with brine, dried over anhydrous MgSO 4 and evaporated under reduced pressure. The resulting residue was lyophilized without further purification to afford compound 2 (0.075 g, 90%).
Spectral data for synthetic fetal bile acids 1 and 2 were as follows: (R)-4-((8S,9S,10R,13R,14S,17R)-10,13-dimethyl-3-oxo-2, 3,8,9,10,11,12,13,14,15,16,17-   www.nature.com/scientificreports/ diluted with deionized water (100 μL) and 10μL injected into the UPLC-TQMS system, which comprised an Acquity I-Class UPLC and a Xevo-TQ-S Micro mass spectrometer (Waters Quantitation of primary and secondary bile acids in plasma by ultraperformance liquid chromatography-triple quadrupole mass spectrometry (UPLC-TQMS). Human plasma (100 μL) was protein precipitated with methanol (400 μL) and centrifuged at 25,000×g for 3 min. Resultant supernatant (100 μL) was diluted with deionized water (100 μL) and 10 μL injected into the UPLC-TQMS system (see above). UPLC separation was achieved on a CORTECS T3 2.7 μm (2.1 × 30 mm) analytical column with a flow rate of 1.3 mL/min. Mobile phases A and B were 0.01% aqueous formic acid containing 0.2 mM ammonium formate and isopropanol/acetonitrile (50:50 v/v) containing 0.01% formic acid and 0.2 mM ammonium formate, respectively. After 0.1 min at 20% B, the bile acids were eluted with a gradient of 20-55% B over 0.7 min, followed by a 0.9 min column wash with 98% B and re-equilibrated to initial conditions. The analytical column was maintained at 60 °C. MRM analyses were performed as above. Retention times together with MRM transitions, cone voltages and collision energies are shown in Supplementary Table S1. Comparison of patients for these 15 bile acids was made from the peak area responses in plasma without calculation of formal concentration using calibration curves 34 . However, in the case of CDCA, the primary bile acid most closely related to FBA1 and FBA2, a calibration curve was constructed as for the fetal bile acids above. For CDCA, the calibration curve was linear (R 2 = 0.998; P < 0.0001).

Patient investigations.
Patients were recruited from the Octavian Fodor Institute for Gastroenterology and Hepatology Hospital, Cluj-Napoca, Romania. All patients gave their written informed consent prior to study inclusion. The study was performed in accordance with the 1975 Declaration of Helsinki and approved by the institutional review board of the Regional Institute of Gastroenterology and Hepatology "Prof. Dr. Octavian Fodor", Cluj-Napoca, Romania (approval no. 2445/12.02.2016). All patients included in this study were hospitalized, having been admitted for liver cancer screening, because of a first decompensation event (ascites, encephalopathy, jaundice or variceal bleeding), because of newly diagnosed focal liver lesions on a background of liver cirrhosis or because of referral for treatment (e.g. ablation, surgery or TACE). HCC patients (n = 10) were consecutively enrolled into the study. During the same period, consecutive patients with liver cirrhosis (LC) but no HCC (n = 15) were included. The diagnosis of HCC was made according to non-invasive criteria or based on histology in patients with atypical lesions on a background of liver cirrhosis. LC patients included those on the first diagnosis of cirrhosis based on ultrasound, laboratory data, Fibroscan, CT/MRI and/or hepatic venous pressure measurement with liver biopsy or patients who were known with a history of liver cirrhosis and presented with cirrhosis decompensation (hepatic encephalopathy, ascites, variceal bleeding). The exclusion criteria were: (1) previous or concurrent cancer; (2) active infections other than chronic hepatitis B or hepatitis C virus; (3) history of organ allograft; (4) pregnancy or breastfeeding; and (5) refusal to sign for informed consent. Clinical details of HCC patients and LC patients are given in Tables 1 and 2 respectively. For the HCC patients, 3/10 were female and for the LC patients, 3/15 were female. Child-Pugh class 16 for the HCC patients (number/total) were A (5/10), B (4/10) and C (1/10) and for the LC patients were A (12/15), B (2/15) and C (1/15), corresponding to Child-Pugh scores 16 of 5-11 for HCC and 5-15 for LC. MELD scores for the HCC patients ranged from 6 to 18 and for the LC patients ranged from 6 to 27. Peripheral venous blood was drawn into heparinized tubes for each patient and spun at 1300 g to prepare plasma, which was aliquoted and frozen at − 80 °C prior to shipment to Switzerland by courier on dry-ice. Plasmas were stored at − 80 °C in Bern before forwarding of aliquots by courier on dry-ice to Wilmslow, UK where they were analyzed by UPLC-TQMS for fetal bile acids 1 and 2, together with 15 free and conjugated primary and secondary bile acids as described above.
Data analysis. Statistical analysis was performed using GraphPad Prism 9.1.0 (San Diego, CA, USA) by nonparametric statistics because of the general non-Gaussian distribution of the data. Comparisons between two groups of data were analyzed using the Mann-Whitney U test and between three or more groups using the Kruskal-Wallis test. Correlations between variables were made by Spearman's rank correlation. All values are expressed as medians. P < 0.05 was considered statistically significant. The Bonferroni correction was used when multiple comparisons were made. www.nature.com/scientificreports/