Glycochenodeoxycholic acid

Analysis of major bile acids in saliva samples of patients with Barrett’s esophagus using high-performance liquid chromatography-electrospray ionization-mass spectrometry

Pavol Dˇ urcˇ a,b, Veˇra Dosedeˇlová a, František Foret a,c, Jirˇí Dolina d, Štefan Konecˇný d, Markus Himmelsbache, Wolfgang Buchberger e, Petr Kubánˇ a,c,∗

a b s t r a c t

A fast, non-invasive, high-performance liquid chromatographic screening method with electrospray ion- ization mass spectrometric detection was developed for the analysis of three major glycine-conjugated bile acids in human saliva. Using a mobile phase composed of 80% methanol and 0.1% formic acid, gly- cocholic, glycodeoxycholic, and glycochenodeoxycholic acids were separated in less than 4 minutes with sensitivity in the low nM range. Bile acids are thought to contribute to the pathology of various compli- cations in gastroesophageal reflux disease, for instance, Barrett’s esophagus, which may eventually lead to esophageal carcinoma. In this pilot study, samples of saliva obtained from 15 patients with Barrett’s esophagus of various severities were compared to saliva samples from 10 healthy volunteers. Glycochen- odeoxycholic acid was significantly elevated in the patients and principal component analysis of all bile acids could distinguish the most severe Barrett’s esophagus patients. We also reported on the detection of glycochenodeoxycholic acid in exhaled breath condensate for the first time. The promising results of this pilot study warrant future investigation, aiming at non-invasive diagnostics of Barrett’s esophagus susceptibility in patients with gastroesophageal reflux disease.

Keywords:
Bile acids HPLC-MS
Saliva
Barrett’s esophagus

1. Introduction

Bile acids (BAs) are a group of compounds essential for lipid digestion and absorption with a steroid skeleton and a carboxy- late side chain usually conjugated to glycine or taurine [1]. BAs are synthesized from cholesterol in the liver, transported to the gallbladder to form bile, and secreted into the duodenum of the small intestine [2]. Part of BAs in the duodenal contents is secreted into the stomach during duodenogastric reflux that occurs phys- iologically, especially postprandially and at night. However, when duodenogastric reflux reaches the esophagus, the condition is de- scribed as duodenogastroesophageal reflux (DGER) [3]. DGER is closely linked to physiological gastroesophageal reflux (GER), i.e., a spontaneous movement of gastric contents into the esophagus. Prolonged and more frequent episodes of GER and DGER occur- ring simultaneously or individually can cause a variety of troublesome symptoms and complications that were defined as gas- troesophageal reflux disease (GERD). The most common symptoms of GERD are heartburn and regurgitation. GERD can cause com- plications like esophagitis or Barrett’s esophagus (BE) [4-6]. BE is a condition when damaged esophageal squamous cells are re- placed by intestinal-type columnar cells, however, the mechanism of this transition is unknown [7]. It was shown that patients with both GERD and BE experienced increased duodenogastric reflux [8]. Moreover, mucosal injury was most common with increased combined gastric acid and bile reflux in comparison with acid or bile reflux alone, therefore BAs together with acid were pro- posed as the major causative factors for BE [9]. While detection of acid/weakly acid reflux alone by 24-hour pH measurement or multichannel intraluminal impedance and pH measurements is a routine clinical diagnostic approach, the detection of BAs is not commonly performed due to the lack of sensitive diagnostic meth- ods. BAs can be analyzed by prolonged ambulatory aspiration tech- niques with subsequent chemical analysis or by fiberoptic probes measuring bilirubin whose absorbance correlates with the levels of BAs (Bilitec 2000 monitoring) [10]. Stein et al. compared the composition of the esophageal aspirate of healthy volunteers and patients suffering from GERD that were divided into groups according to the mucosal injury. The highest concentration of to- tal BAs was found in the aspirate of BE patients [11]. In addition, Kauer et al. determined that glycocholic acid (GCA), glycodeoxy- cholic acid (GDCA), and glycochenodeoxycholic acid (GCDCA) were the major BAs in the esophageal aspirate of patients suffering from GERD [12] and Straub et al. [13] confirmed that the induction of in- testinal type of metaplasia is most effectively induced by glycine- conjugated bile acids. Aspirate studies and Bilitec 2000 monitor- ing are, however invasive methods that can be unpleasant for pa- tients. The development of surrogate, non-invasive approaches is desirable.
In this work, we analyzed BAs in saliva because saliva col- lection is simple, repeatable, and non-invasive, and the composi- tion of BAs in saliva might have a clinical value in diagnosing BE. Salivary BAs were previously determined by enzymatic methods and in a few studies by separation methods, i.e., gas chromatography (GC) and high-performance liquid chromatography (HPLC) with mass spectrometry (MS) detection. Enzymatic methods are simple and widely used in clinical laboratories, however, they of- ten lack the necessary sensitivity and can only determine total BAs. For instance, enzymatic methods were applied to assess salivary BAs in laryngopharyngeal reflux or cystic fibrosis [14-16]. On the other hand, separation methods have high selectivity and are able to analyze individual BAs in the mixture. A typical feature of the GC-MS separation is its high sensitivity and resolution. Unfortu- nately, GC-MS analysis of BAs is limited due to exhaustive sam- ple pretreatment. Sample treatment often includes extraction, pu- rification, hydrolysis of conjugated BAs, and derivatization of po- lar groups to prepare volatile analytes [17]. In one account, Heden- borg et al. analyzed diverse salivary BAs, including glucuronidated and sulfated derivates of cholic acid (CA) and chenodeoxycholic acid (CDCA), by the combination of gas-liquid chromatography and GC-MS [18]. Since GC-MS methods are time-consuming, HPLC-MS is generally preferred to analyze BAs in biological samples even though the separation resolution is lower in comparison with GC- MS. HPLC-MS was previously employed to study BAs in serum [19, 20], plasma [1], bile [19], or urine [19, 21]. There are only a few articles describing HPLC analysis of BAs in saliva. This is most probably because saliva contains very low concentrations of BAs. Therefore solid-phase extraction (SPE) and derivatization steps were introduced to achieve better sensitivity in HPLC-MS analysis [22-24]. For instance, Guo et al. developed an HPLC-MS method for the analysis of carboxylic acids including CA and CDCA in saliva after derivatization with N-(3-dimethylaminopropyl)-Nr- ethylcarbodiimide [22]. Higashi et al. used 2-hydrazinopyridine as a derivatization reagent for the analysis of CDCA and GCDCA [23]. The above methods were used for the analysis of selected BAs, however, with no practical diagnostic application. Since BAs are thought to be involved in the pathology of BE, the development of new analysis approaches is of high priority. In this work, we devel- oped a rapid HPLC-MS screening method for comparison of three major BAs in saliva. Glycocholic, glycodeoxycholic, and glycochen- odeoxycholic acids were for the first time analyzed and their levels were compared in the saliva of patients with BE and healthy volun- teers, to determine possible applicability of saliva as a non-invasive sample in the diagnostics of BE.

2. Material and methods

2.1. Chemicals, standards and samples

All chemicals were of reagent grade, and deionized (DI) wa- ter (18.2 M▲) was used for stock solution preparation and dilutions. Standard solutions of analytes, 10 mmol/L, were prepared from sodium salts of glycocholic acid (GCA) (p.a., Merck, Darm- stadt, Germany), glycodeoxycholic acid (GDCA) (p.a., Merck, Darm- stadt, Germany) and glycochenodeoxycholic acid (GCDCA) (≥97%, HPLC grade, Sigma-Aldrich, Steinheim, Germany). For preparation of mobile phase, methanol (HPLC grade, ≥97%, Merck, Darmstadt, Germany) and formic acid (96%, Sigma-Aldrich, Steinheim, Ger- many) were used.

2.2. Study subjects

This study was approved by the Ethics Committee of the Uni- versity Hospital Brno (Approval no. 03-200616, from June 20, 2016) and was performed in accordance with the ethical principles out- lined in the Declaration of Helsinki. Written informed consent was obtained from all study participants. We have recruited 15 pa- tients diagnosed with Barrett’s esophagus and 10 healthy controls. The classification of patients was performed based on “Prague C & M criteria” classification [25]. In this classification, the circum- ferential Barrett’s segment (C) and the longest Barrett’s segment (M) are measured. The higher the value of C and M segment, the more severe the condition of BE, as the damage extends to a larger area of the esophagus. The C-segment value can be considered the most important characteristics of the BE severity. A recent mul- ticenter study conducted by Gaddam et al. revealed that for ev- ery 1 cm extension in BE length the risk of high-grade dysplasia and esophageal adenocarcinoma increased by 21% [26]. All patients were without dysplasia or adenocarcinoma of the esophagus and have not undergone any endoscopic treatment before the study. Table 1 summarizes the criteria of the study group.

2.3. Instrumentation

2.3.1. HPLC/MS

An HPLC/MS system consisted of an Agilent 1260 Infinity Binary Pump, a 1260 Infinity Sampler, a 1260 Infinity Thermostat, and a 1260 Infinity Thermostatted Column Compartment (Agilent Tech- nologies, Santa Clara, CA, USA) connected to an Agilent 6560 Ion Mobility Q-TOF (Agilent Technologies, Santa Clara, CA, USA) that was run in Q-TOF mode.
Chromatographic separations were performed on a reversed phase column (Poroshell 120 EC-C18 2.7μm, 3.0×150 mm) main- tained at 40°C. The injection volume was 10 μL. The separation of analytes was achieved using isocratic elution in a mobile phase consisting of 80% methanol and 0.1% formic acid. The analysis was completed in less than 4 min using a flow rate of 0.6 mL/min. For overnight storage, the separation column was flushed with 100% methanol.
The mass spectrometer was operated in a negative ionization mode, and data were acquired using Agilent MassHunter B.08.00 software. The parameters for the Dual AJS ESI source were as fol- lows: gas temperature 300°C, drying gas flow 11 L/min, nebulizer pressure 25 psig, sheath gas temperature 350°C, sheath gas flow 12 L/min, VCap voltage 4.5 kV and nozzle voltage 1.5 kV. The high- est signals of analytes were achieved when the fragmentor voltage was set to 375 V and the collision energy to 15 V. The mass spectra were obtained for the m/z range 100-1700 with an acquisition rate of 1 spectrum/s. During all measurements, the reference mass so- lution was sprayed and reference masses of m/z 112.9856 and m/z 966.0007 were used for correction of the measured masses.

2.4. Sample collection

Saliva samples were collected from healthy individuals and BE patients by spitting into 50 mL polypropylene containers. Samples were stored at -80°C. In addition to saliva samples, exhaled breath condensate (EBC) samples were also collected. We collected 8 EBC samples from 4 BE patients and 4 healthy controls using a simple EBC sampler developed in our group earlier [27]. EBC samples were also stored at -80°C.

2.5. Sample preparation

2.5.1. Saliva samples

Frozen saliva samples were defrosted at room temperature. For the quantitative analysis of saliva samples, a combined standard addition method and protein precipitation were used to eliminate the matrix effect of saliva. The standard addition method consisted of spiking 300 μL aliquots of the saliva sample with 900 μL of GCA, GDCA and GCDCA standard mixture dissolved in methanol to yield the final concentrations of each standard in the sam- ples at three concentration levels of 0 nmol/L, 10 nmol/L and 50 nmol/L. Methanol acted simultaneously as the protein precip- itation reagent. The spiked samples were vortexed for 10 minutes and sonicated for another 10 minutes. Afterwards, they were cen- trifuged for 10 minutes (4149 g) and 1 mL of the supernatant was transferred into a new vial. Supernatants were dried under a stream of nitrogen, while the temperature was maintained at 40°C. Dried samples were redissolved in 300 μL of 80% methanol, vor- texed and sonicated, each for 10 minutes. Due to the undissolved content, the samples were further centrifuged (10 minutes, 4149 g) and clean supernatants were finally analyzed by HPLC-MS.

2.5.2. EBC samples

Frozen EBC samples were defrosted at room temperature. 300 μL of EBC sample was dried under the stream of nitrogen, while the temperature was maintained at 40°C. Dried samples were re- dissolved in 300 μL of 80% methanol, vortexed, and sonicated, each for 10 minutes. The samples were then analyzed by HPLC-MS.

2.5.3. Statistical evaluation of the results

The calculations of mean, median and relative standard devia- tion values were performed in Microsoft Excel software (Microsoft Corporation, Redmond, WA, USA). The between-group compari- son (patients vs. controls) of bile acid concentration values was performed using the Mann–Whitney U test. P-value <0.05 was considered as significant. The principal component analysis (PCA) was performed using the Microsoft Excel add-on software XlStat (Addinsoft, Paris, France). 2.6. Method validation The fit-for-purpose validation of the standard addition method for the determination of major bile acids in saliva samples was per- formed with consideration to ICH guidelines [28]. Selectivity was evaluated by comparing the peaks of analytes in the chromatograms of standards, samples and also spiked samples. A 5-point calibration curves (0, 5, 10, 20, and 100 nmol/L of each analyte), obtained by dilution of stock solutions of standards in DI H2O and blank DI H2O, were used for the linearity study. Linear calibration curve parameters were obtained from the plot of the analyte peak area against analyte concentration using a least square regression analysis. Precision, expressed as relative standard deviation (RSD), was determined by repetitive measurements (n = 3) of a standard mix- ture (GCA, GDCA, and GCDCA) at medium (10 nmol/L) concentra- tion level of each analyte. The limits of detection (LOD) of the method for GCA, GDCA, and GCDCA were determined from the standard deviations of y- intercepts of the regression lines of the calibration curves [28]. The combined matrix and recovery effect was calculated by di- viding the slopes of the calibration curve using the standard ad- dition method (the standards were added to the saliva samples at the beginning of the whole procedure) and the slopes of the cali- bration obtained using standard mixtures. 3. Results and discussion 3.1. HPLC-MS method development 3.1.1. Optimization of the mobile phase A successful separation of all analytes on the reversed phase column was achieved when an isocratic elution in the mobile phase composed of 80% methanol and 0.1% formic acid was used. The role of formic acid in the mobile phase was to suppress the dissociation of the analyzed bile acids that are weakly acidic. Dis- sociation is an undesirable process during separation on the re- versed phase column, decreasing the separation efficiency. Differ- ent concentrations of formic acid in the mobile phase were tested to improve the shape of the peaks up to the upper concentration limit of 0.1%. Above this concentration of formic acid, the ioniza- tion of bile acids in the negative ionization MS detection mode was decreased, resulting in the lower signal of the analytes. Finally, 0.1% concentration of formic acid was selected as a compromise between the maximum separation efficiency and detection sensi- tivity. Due to the complex matrices of saliva samples injected onto the column, a cleaning step was performed after each isocratic run. This step consisted of raising the concentration of methanol in the mobile phase to 100% after 3.5 minutes and holding it for 3 minutes to remove all impurities from the column. Afterwards, the column was equilibrated with a mobile phase consisting of 80% MeOH/0.1% formic acid for 3.5 minutes, prior to the next analysis. 3.1.2. Optimization of MS conditions The Q-TOF mass spectrometer was operated only in a single stage MS (MS1) negative ionization mode. To increase the detec- tion sensitivity for the bile acids that are present in saliva sam- ples in nanomolar concentrations, the parameters of the ion source were optimized by analyzing 100 nmol/L standard mixture solu- tions. The signal intensities of negative single charged ions of standards of the three bile acids (GCA [M-H]− ion at m/z 464.3018, GCDCA [M-H]− ion at m/z 448.3068, GDCA [M-H]− ion at m/z 448.3068) were selected for optimization. Changing the fragmentor voltage had the most profound effect on the signal increase. Fig. 1A shows the dependence of signals of the ions of interest on the frag- mentor voltage with maximum signal intensity between 300 and 400 V. Another parameter that was optimized was the collision en- ergy (Fig. 1B). When the collision energy was increased to 15 V, a decrease in the noise of the background in a single stage MS was observed. This phenomenon resulted in higher signal to noise ra- tio of the analyte ions. However, when the collision energy was increased above 15 V, fragmentation of molecule ions of glycine- conjugated bile acids occurred. The optimum fragmentor voltage and collision energy were thus set to 375 V and 15 V, respectively. 3.2. Analytical parameters of the method – method validation Under the optimized separation and detection conditions, the main analytical parameters of the developed HPLC-MS method were assessed and are shown in Table 2. For the determination of dead retention time (needed for calculation of retention factors) a 0.01% solution of thiourea was injected. All analytes were baseline separated in under 4 minutes, while the lowest retention factor was for GCA (k = 1.4). The developed HPLC-ESI-MS method showed very good linearity over the tested concentration range of 0-100 nmol/L for all analyzed bile acids. The selectivity was assessed by comparison of the peaks of an- alytes in standard solution, non-spiked and spiked saliva samples and showed that there were no interfering matrix components. The resolution (Rs) between the most critical peaks at the high- est concentration level of analytes (100 nmol/L) was higher than 1.5 (Rs = 1.85). A 5-point calibration curves (0, 5, 10, 20, and 100 nmol/L of each analyte) were obtained from the plot of the analytes peak ar- eas against analyte concentration using a least square regression analysis. The calibration curves equations showed excellent linearity (R2 ≥ 0.997). Although the R2 values (shown in Table 2) showed excellent linearity over the specified range, the real saliva samples were analyzed using a standard addition method to remove the matrix effect. Precision, expressed as RSD was evaluated by repetitive mea- surements (n = 3) of the standard mixture (GCA, GDCA, GCDCA) at medium (10 nmol/L) concentration level of each analyte. The re- peatability of retention times was below 0.2% RSD, while the RSDs for peak areas were below 5.1%. Low (LODs were obtained, ranging from 0.3 to 0.5 nmol/L. The LODs are comparable to the LODs in other publications that used LC-MS with extensive sample pretreatment (SPE) and/or tandem MS/MS detection. Xiang et al. [1] achieved LODs 0.7 to 0.9 nmol/L for the same analytes in blood plasma using SPE preconcentration. Steiner et al. [20] reported the LOQs for GCA, GDCA, and GCDCA as 20 nmol/L, corresponding LODs were thus 6 nmol/L. Mochizuki et al. [24] reported the LODs for CDCA in saliva after derivatization to be 0.1 nmol/L. In the work of Higashi et al. [23], SPE extrac- tion of GCDCA from saliva resulted in LOD of 0.03 nmol/L, which is about 10-fold lower than in our work but without the use of any SPE preconcentration. The LODs achieved in this work could be decreased by using a lower volume of methanol to redissolve the dried samples or by application of additional sample pretreat- ment steps, such as SPE. The repeatability of retention times was below 0.2% RSD, while the RSDs for peak areas were below 5.1%. The combined matrix and recovery effect was calculated by di- viding the slopes of the calibration curve using the standard ad- dition method (the standards were added to the saliva samples at the beginning of the whole procedure) and the slopes of the cal- ibration obtained using standard mixtures. The combined matrix and recovery effect was within the range of 71.1 to 93.9% for the majority of the samples and analytes, however, in some cases, the number was much lower (i.e. 6.1-54%), especially for GDCA (GCA 74.5-91.2%, GCDCA 54.0-92.8%, GDCA 6.1-93.9%). The combined matrix and recovery effect of each sample is shown in Table S2 in the Supplementary material. Additionally, the R2 values of the obtained calibrations after standard addition of the standards are shown in Table S2. Since all R2 values, except one, are between 0.9951 and 1.000, the standard addition method was considered suitable for the quantitative evaluation. 3.3. Analysis of saliva samples The determination of bile acid concentrations in saliva samples was performed using the standard addition method, as described in section 2.5.1. Examples of the extracted ion chromatograms (EIC) [at m/z 464.3018, 448.3068] obtained from the HPLC-MS analysis of aliquots of the saliva sample obtained from a patient with BE are shown in Fig. 2. All analyzed bile acids were detected in this sam- ple. It can be observed that the peaks of GCA, GCDCA and GDCA are increased with an addition of the corresponding bile acid stan- dard at concentration levels of 10 nmol/L (chromatogram b) and 50 nmol/L (chromatogram c). In EIC (at m/z 448.3068) of saliva samples of some individuals, an unidentified peak with the reten- tion time of 1.7 minutes was observed. This unknown compound did not interfere with the bile acids, because it eluted much ear- lier than the analyzed bile acids. The concentrations of the three bile acids were evaluated in all saliva samples. The concentrations of GCA in saliva samples (Fig. 3A) of 10 healthy individuals ranged from 0.00 to 4.81 nmol/L (average 1.54 nmol/L, median 1.00 nmol/L), while in saliva sam- ples of 15 BE patients, the GCA concentration ranged from 0.58 to 13.13 nmol/L (average 2.89 nmol/L, median 1.81 nmol/L). Statisti- cal comparison of GCA concentrations between these two groups of subjects using Mann-Whitney U test revealed that there was no statistically significant difference at α = 0.05 (p-value 0.1585). The concentrations of GCDCA in saliva samples (Fig. 3B) of 10 healthy individuals ranged from 0.95 to 4.16 nmol/L (average 2.62 nmol/L, median 2.79 nmol/L), in saliva samples of 15 BE patients the con- centrations of GCDCA ranged from 2.15 to 50.26 nmol/L (average 7.86 nmol/l, median 4.54 nmol/L). There was a statistically signif- icant difference between these two groups at α = 0.01 (p-value 0.0051). The concentrations of GCDCA in saliva samples of BE pa- tients were higher compared to healthy controls. It is worth not- ing that patient P7 had the highest concentrations of all BAs in the analyzed samples. It might therefore contribute significantly to the higher average concentration of BAs in the patient sam- ples. However, even when this sample was removed for statistical testing purpose, the concentrations of GCDCA in saliva samples of BE patients were statistically higher compared to healthy controls (Mann-Whitney U test, p = 0.0078). The concentrations of GDCA in saliva samples (Fig. 3C) of 10 healthy individuals ranged from 0.00 to 1.51 nmol/L (average 0.47 nmol/L, median 0.00 nmol/L), in saliva samples of 15 BE patients the concentrations of GDCA ranged from 0.00 to 15.92 nmol/L (average 2.07 nmol/L, median 0.00 nmol/L). There was no statistically significant difference between these two groups (p-value 0.3320) at α = 0.05. The concentrations of bile acids in the saliva samples of all healthy individuals and BE pa- tients are shown in the supplementary material (Table S1). As can be seen in Table S2, the combined matrix effect and recoveries were low for GDCA in some samples. In most of these samples, the calculated concentration of GDCA was below the LOD. 3.4. Principal component analysis of bile acids in saliva samples Principal component analysis (PCA) is a statistical procedure that converts a large number of variables into a smaller number of so-called principal components (PCs). It is able to reveal strong patterns in a dataset. In here, we applied PCA to classify the pa- tients and healthy controls based on bile acids concentration in saliva samples, and to test whether PCA is able to distinguish pat- terns (cluster formation) within the patients based on the severity of BE. Bile acids concentrations (GCA, GCDCA, and GDCA) of all par- ticipants were combined into a data matrix with 25 rows (samples of different individuals) and three columns of corresponding bile acids concentrations. The table was subjected to a mean-centered PCA procedure. The “leave-one-out” cross-validation procedure was used to determine the number of significant principal components. This procedure revealed two significant principal components that accounted for 98.9% of the total variance. The score plot is shown in Fig. 4A. For clarity, healthy individuals are depicted by green squares, patients with C segment < 10 are depicted by orange tri- angles, and the patients with C segment ≥ 10 are depicted by red circles. The PCA plot shows that most of the patients with a severe condition of BE (C segment ≥ 10) can be clearly distinguished from other patients mainly in the direction of the axis of principal com- ponent F1 (patients P7, P10, P2). The fourth patient (P4) with C ≥ 10 is however outside this cluster. It can be also observed (Fig. 4B) that all other BE patients with less severe condition C < 10 (orange triangles) can still be slightly separated along the F1 axis from the healthy individuals (green squares) by being shifted towards the right-hand side. The above findings of this pilot screening study provide a promising starting point for a deeper investigation including a larger number of participants. We are aware of the fact that the number of participants in this study was limited. Additionally, the sample collection from each individual patient and healthy con- trol was performed only once and at different times of the day. It would be of interest to collect multiple saliva samples from each individual and assess the variability of the bile acids during differ- ent times a day or within several days. Another limitation of this study was that the patients were not instructed to avoid eating, drinking, or brushing their teeth prior to sampling and had not dis- continued the proton pump inhibitor medication. We do not know whether a strict control of these parameters would improve the method accuracy, but it would be worth investigating in the fu- ture, because standardization of sampling is of high importance, for instance to compare the results obtained with other research laboratories. 3.5. Analysis of EBC samples EBC samples were of additional interest to us because the pres- ence of bile acids in EBC might point to the possibility of reflux penetration into the upper respiratory system, a condition that can be encountered in extraesophageal reflux. An initial HPLC-MS screening of the selected bile acids (GCA, GCDCA and GDCA) in EBC samples of 4 healthy controls and 4 patients (3 BE patients with C > 10, one patient with C < 10) was performed. This experiment was designed to test for the presence of these acids in EBC. While GCA and GDCA were not detected in any of the samples, we have detected the most abundant BA (GCDCA) in EBC samples of all pa- tients and healthy controls. We compared the obtained peak ar- eas of GCDCA between the patient group and the healthy control group but found that there was no statistically significant difference (p = 0.2515) between patients and healthy controls at a sig- nificance level of α = 0.05. Although these results are only prelim- inary, they prove that HPLC-MS can be used also for the analysis of EBC samples and the detection of GCDCA in EBC was reported for the first time in this work. 4. Conclusions In this work, we have detected three selected glycine- conjugated bile acids in saliva samples of BE patients and healthy controls. GCDCA was elevated in saliva samples from a group of patients diagnosed with BE in comparison to the healthy control group. The observation of elevated bile acid concentrations in the saliva samples of patients corresponds well with the hypothesis that bile acids are one of the main causative factors of BE pathol- ogy. The highest concentrations of GCDCA were found in saliva samples of patients with the most severe BE condition, and those patients were also clearly separated from the other patients and healthy controls using PCA. The analytical method could be fur- ther simplified by employing the isotope dilution method, as iso- topically labeled standards of bile acids are commercially available. Hence, large volumes of both saliva and EBC samples would not be required, however, this approach would increase the cost of anal- ysis. We also found GCDCA in the collected EBC samples. To our best knowledge, bile acid was detected in the EBC sample for the first time. The results of this pilot study give promise to the fu- ture investigation of a large patient cohort, aiming at non-invasive diagnostics of Barrett’s esophagus Glycochenodeoxycholic acid susceptibility in GERD patients.

References

[1] X.Q. Xiang, Y. Han, M. Neuvonen, J. Laitila, P.J. Neuvonen, M. Niemi, High per- formance liquid chromatography-tandem mass spectrometry for the determi- nation of bile acid concentrations in human plasma, J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 878 (2010) 51–60 https://doi.org/10.1016/j.jchromb. 2009.11.019.
[2] K.W. Romanski, The role and mechanism of action of bile acids within the di- gestive system – bile acids in the liver and bile, Adv. Clin.Exp. Med. 16 (2007) 793–799.
[3] P.O. Katz, The role of non-acid reflux in gastro-oesophageal reflux disease, Al- iment. Pharm. Ther. 14 (2000) 1539–1551 https://doi.org/10.1046/j.1365-2036. 2000.00875.x.
[4] J. Sarosiek, R.W. McCallum, Mechanisms of oesophageal mucosal defence, Best Pract. Res. Clin. Gastroenterol. 14 (2000) 701–717 https://doi.org/10.1053/bega. 2000.0119.
[5] P.J. Kahrilas, GERD pathogenesis, pathophysiology, and clinical manifestations, Cleve. Clin. J. Med. 70 (Suppl 5) (2003) S4–19 https://doi.org/10.3949/ccjm.70. Suppl_5.S4.
[6] N. Vakil, S.V. van Zanten, P. Kahrilas, J. Dent, R. Jones, G. Global Consensus, The montreal definition and classification of gastroesophageal reflux disease: A global evidence-based consensus, Am. J. Gastroenterol. 101 (2006) 1900–1920 https://doi.org/10.1111/j.1572-0241.2006.00630.x.
[7] S. Jurgens, F. Meyer, S.J. Spechler, R. Souza, The Role of Bile Acids in the Neo- plastic Progression of Barrett’s Esophagus – A Short Representative Overview, Z, Gastroenterol 50 (2012) 1028–1034 https://doi.org/10.1055/s-0032-1312922.
[8] S.E.A. Attwood, T.R. Demeester, C.G. Bremner, A.P. Barlow, R.A. Hinder, Alkaline gastroesophageal reflux – Implications in the development of complications in Barrett’s columnar-lined lower esophagus, Surg 106 (1989) 764–770 https://doi.org/10.5555/uri:pii:0039606089901414.
[9] D.S. Oh, J.A. Hagen, M. Fein, C.G. Bremner, C.M. Dunst, S.R. DeMeester, J. Lipham, T.R. DeMeester, The impact of reflux composition on mucosal in- jury and esophageal function, J. Gastrointest. Surg. 10 (2006) 787–796 https://doi.org/10.1016/j.gassur.2006.02.005.
[10] P. Bechi, F. Pucciani, F. Baldini, F. Cosi, R. Falciai, R. Mazzanti, A. Castagnoli, A. Passeri, S. Boscherini, Long-term ambulatory enterogastric reflux monitor- ing. Validation of a new fiberoptic technique, Digest. Dis. Sci. 38 (1993) 1297– 1306 https://doi.org/10.1007/BF01296082.
[11] H.J. Stein, H. Feussner, W. Kauer, T.R. Demeester, J.R. Siewert, Alkaline gas- troesophageal reflux: assessment by ambulatory esophageal aspiration and pH monitoring, Am. J. Surg. 167 (1994) 163–168 https://doi.org/10.1016/ 0002-9610(94)90068-x.
[12] W.K.H. Kauer, J.H. Peters, T.R. DeMeester, H. Feussner, A.P. Ireland, H.J. Stein, R.J. Siewert, Composition and concentration of bile acid reflux into the esopha- gus of patients with gastroesophageal reflux disease, Surg 122 (1997) 874–881 https://doi.org/10.1016/S0039-6060(97)90327- 5.
[13] D. Straub, R. Elferink, P.L.M. Jansen, J. Bergman, K. Parikh, K.K. Krishnadath, Glyco-conjugated bile acids drive the initial metaplastic gland formation from multi-layered glands through crypt-fission in a murine model, Plos One 14 (2019) https://doi.org/10.1371/journal.pone.0220050.
[14] E. De Corso, S. Baroni, S. Agostino, G. Cammarota, G. Mascagna, A. Mannocci, M. Rigante, J. Galli, Bile acids and total bilirubin detection in saliva of patients submitted to gastric surgery and in particular to subtotal Billroth II resec- tion, Ann. Surg. 245 (2007) 880–885 https://doi.org/10.1097/01.sla.0000255574. 22821.a1.
[15] M. Sereg-Bahar, A. Jerin, I. Hocevar-Boltezar, Higher levels of total pepsin and bile acids in the saliva as a possible risk factor for early laryngeal cancer, Ra- diol. Oncol. 49 (2015) 59–64 https://doi.org/10.2478/raon-2014-0020.
[16] K. Blondeau, L.J. Dupont, V. Mertens, G. Verleden, A. Malfroot, Y. Vanden- plas, B. Hauser, D. Sifrim, Gastro-oesophageal reflux and aspiration of gas- tric contents in adult patients with cystic fibrosis, Gut 57 (2008) 1049–1055 https://doi.org/10.1136/gut.2007.146134.
[17] W.J. Griffiths, J. Sjovall, Bile acids: analysis in biological fluids and tissues, J, Lipid Res 51 (2010) 23–41 https://doi.org/10.1194/jlr.R001941-JLR200.
[18] G. Hedenborg, A. Norlander, A. Norman, Bile acids in serum, ultrafiltrate of serum and saliva from patients with cholestatic jaundice, Scand. J. Clin. Lab. Inv. 47 (1987) 83–89.
[19] S. Perwaiz, B. Tuchweber, D. Mignault, T. Gilat, I.M. Yousef, Determination of bile acids in biological fluids by liquid chromatography-electrospray tandem mass spectrometry, J. Lipid Res 42 (2001) 114–119.
[20] C. Steiner, A. von Eckardstein, K.M. Rentsch, Quantification of the 15 major human bile acids and their precursor 7 alpha-hydroxy-4-cholesten-3-one in serum by liquid chromatography-tandem mass spectrometry, J. Chrom. B An- alyt. Technol. Biomed. Life Sci. 878 (2010) 2870–2880 https://doi.org/10.1016/j. jchromb.2010.08.045.
[21] B.S. Kumar, B.C. Chung, Y.J. Lee, H.J. Yi, B.H. Lee, B.H. Jung, Gas chromatography-mass spectrometry-based simultaneous quantitative analyti- cal method for urinary oxysterols and bile acids in rats, Anal. Biochem. 408 (2011) 242–252 https://doi.org/10.1016/j.ab.2010.09.031.
[22] C. Guo, D.M. Li, C.M. Liu, Z.P. Guo, Y. Chen, Facile derivatization of ultratrace carboxylic acids in saliva for quantification by HPLC-MS/MS, Anal. Bioanal. Chem. 410 (2018) 4293–4300 https://doi.org/10.1007/s00216-018-1078- 7.
[23] T. Higashi, Y. Shibayama, T. Ichikawa, K. Ito, T. Toyo’oka, K. Shimada, K. Mi- tamura, S. Ikegawa, H. Chiba, Salivary chenodeoxycholic acid and its glycine- conjugate: Their determination method using LC-MS/MS and variation of their concentrations with increased saliva flow rate, Steroids 75 (2010) 338–345 https://doi.org/10.1016/j.steroids.2010.01.015.
[24] Y. Mochizuki, S. Inagaki, M. Suzuki, J.Z. Min, K. Inoue, K. Todoroki, T. Toyo’oka, A novel derivatization reagent possessing a bromoquinolinium structure for bi- ological carboxylic acids in HPLC-ESI-MS/MS, J. Sep. Sci. 36 (2013) 1883–1889 https://doi.org/10.1002/jssc.201300083.
[25] P. Sharma, J. Dent, D. Armstrong, J. Bergman, L. Gossner, Y. Hoshihara, J.A. Jankowski, O. Junghard, L. Lundell, G.N.J. Tytgat, M. Vieth, The develop- ment and validation of an endoscopic grading system for Barrett’s esopha- gus: The Prague C & M Criteria, Gastroenterol 131 (2006) 1392–1399 https://doi.org/10.1053/j.gastro.2006.08.032.
[26] S. Gaddam, P.E. Young, B.R. Alsop, N. Gupta, H. Gavini, A.D. Higbee, S.B. Wani, M. Singh, A. Rastogi, A. Bansal, B.D. Cash, D.A. Lieberman, R.E. Sampliner, G.W. Falk, P. Sharma, Relationship Between Barrett’s Esophagus (BE) Length and the Risk of High Grade Dysplasia (HGD) and Esophageal Adenocarcinoma (EAC) in Patients With Non Dysplastic Barrett’s Esophagus Results From a Large Multicenter Cohort, Gastroenterol 140 (2011) S81 -S81 https://doi.org/10.1016/ S0016-5085(11)60329- 6 .
[27] M. Gregus, F. Foret, P. Kuban, Single-breath analysis using a novel simple sam- pler and capillary electrophoresis with contactless conductometric detection, Electrophoresis 36 (2015) 526–533 https://doi.org/10.1002/elps.201400456.
[28] , ICH Harmonised Tripartite Guideline: Validation of analytical procedures: Text and Methodology Q2(R1), International Conference of harmonization of technical requirements for registration of pharmaceuticals for human use, 2005 https://database.ich.org/sites/default/files/Q2_R1 Guideline.pdf accessed 7 February 2020.