Analysis of phenylcarboxylic acid-type microbial metabolites by microextraction by packed sorbent from blood serum followed by GC–MS detection

Highlights

• •Phenylcarboxylic acids (PhCAs) are a potential diagnostic marker of sepsis.
• •Microextraction by Packed Sorbent (MEPS) for PhCA analysis in serum samples.
• •MEPS investigated as alternative to Liquid-Liquid extraction for PhCA analysis.
• •Pilot study on 4 patients supports PhCA profiling using MEPS for sepsis diagnosis.

Abstract

A method for analysis of 8 phenylcarboxylic acids in blood serum was developed based on the coupling of microextraction by packed sorbent, derivatization and GC–MS detection. These compounds are low molecular weight aromatic microbial metabolites that are proven and prospective indicators of sepsis in critically ill patients. Recoveries of the phenylcarboxylic acids from serum samples using microextraction by packed sorbent were 30–70%. The present method was linear (R² ≥ 0.9981) over a clinically significant range of concentrations (94–2250 µg L⁻¹/0.5–18 µM). The limits of quantification for the optimized method were 60–100 µg L⁻¹/0.4–0.7 µM for phenylpropionic, phenyllactic, 4-hydroxybenzoic and 4-hydroxyphenylacetic acids, and 160 µg L⁻¹/0.9–1.3 µM for benzoic, 4-hydroxyphenyllactic, homovanillic and 4-hydroxyphenylpropionic acids. The developed conditions were used to determine concentrations of the phenylcarboxylic acids in the most complicated matrix – serum samples of critically ill patients. Results were compared with liquid-liquid extraction and revealed a reduction in the time for sample preparation (45 min vs. 6 min) and serum (200 µL vs. 80 µL) volume. The combination of microextraction by packed sorbent and GC–MS methods, especially when fully automated could be a powerful tool for the clinical diagnosis of sepsis.

1. Introduction

Metabolomic research into the diagnosis of various diseases is ongoing and important. There is a strong demand for the identification, evaluation and clinical verification of biomarkers. Such markers will be important components of the future advancement of medical care. New solutions for qualitative and quantitative chemical analysis are actively being used in the identification of metabolomic biomarkers [1], [2].

Despite rapid advances in medicine, early diagnosis of sepsis is not yet a reality. The number of deaths due to sepsis remains high [3], [4], [5]. Identification of phenylcarboxylic acids (PhCAs), which are byproducts of the bacteria that cause sepsis, in blood serum is a new approach to the diagnosis of the early stages of sepsis [6]. Microbiological experiments have demonstrated that the main representatives of human microbiota and pathogens of nosocomial infections produce PhCAs [7], [8], [9]. Concentrations of three aromatic metabolites (phenyllactic acid (PhLA), 4-hydroxyphenyllactic acid (p-HPhLA), and 4-hydroxyphenylacetic acid (p-HPhAA)) have been shown to correlate with both the severity and symptoms of sepsis. The concentration of these compounds, which are also called ‘sepsis-associated aromatic microbial metabolites’ [7], increases in the blood serum of patients in concert with the development of postoperative complications, pneumonia, or sepsis, while the level of the PhCAs in the blood serum of unaffected patients remains within a normally expected physiological range [10], [11]. The normal range of concentrations for some PhCAs, such as benzoic acid (BA), phenylpropionic acid (PhPA), PhLA and p-HPhLA have been determined in the serum of healthy people [12], [13]; their ubiquitous, low level presence reflects the integration of the human metabolism with its microbiota [8]. However, there were several PhCAs (i.e., 4-hydroxybenzoic acid (p-HBA), 4-hydroxyphenylpropionic acid (p-HPhPA) and homovanillic acid (HVA)) that were found in the serum of critically ill patients [13], whose clinical significance remains to be clarified.

Currently, liquid–liquid extraction (LLE) with diethyl ether (Et₂O) is used for the isolation of target compounds from serum samples [13], [14]. Derivatization with N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) forms the volatile trimethylsilyl derivatives of the PhCAs [15], [16]. Then volatile derivatives of the PhCAs are detected by gas chromatography coupled with mass spectrometry (GC–MS) [13], [15] or flame ionization detector (GC-FID) [14], [16]. However, the LLE procedure is complicated and time-consuming. Therefore, another method of extraction was explored.

Microextraction by packed sorbent (MEPS) is a promising modern technique for SPE [17], [18], [19]. This method is widely used and has the following advantages:

• •the simplicity of the equipment;
• •the availability of a wide range of sorbents;
• •the use of small volumes of analyzed sample and organic solvents;
• •the possibility of varying the parameters at each stage of sample preparation.

Additionally, MEPS can be combined on-line with various analytical methods, such as GC–MS or HPLC-MS. This leads to reduced human labor input, which improves analytical productivity and reproducibility [20], [21], [22], [23].

C18 sorbent is actively used for the extraction of organic compounds containing alkyl and aryl groups from various biological matrices [22], [23], [24], [25], [26]. This sorbent is characterized by a high rate of sorption equilibration and the complete desorption of sorbed analytes by using small portions of organic solvents. At the same time, we found no publications on the extraction of the PhCAs from blood serum using SPE with C18 sorbent. Only two studies were performed using MEPS with C18 [22], [23], but there is no information about the determination of two of the most polar and clinically significant aromatic metabolites – PhLA and p-HPhLA.

Bustamante et al. [22] evaluated the analysis of PhCAs (including PhPA, p-HBA, p-HPhAA and HVA) in plasma using MEPS (C18 and C8) and GC–MS/MS after a deconjugation step. Despite sample preparation, only PhPA was detected in the plasma samples of healthy volunteers at the level of 350 µg L⁻¹. The recoveries of p-HBA and p-HPhAA were 10–30%; recoveries for PhPA and HVA were 30–80%. The obtained data and relative standard deviation (RSD) were compared with LLE. MEPS pretreatment presented a better overall RSD then LLE, but only 27 of 40 studied analytes were detected, while LLE allowed the detection of all 40 phenolic compounds.

The authors of another study [23] performed analysis of PhCAs (including BA, p-HPhPA, HVA) in plasma after a deconjugation step and used MEPS with C18 and GC–MS coupled on-line to in-liner derivatization. The ion-pairing reagent (tetrabutylammonium hydroxide) was used to improve extraction yields of the more polar analytes, as well as in the derivatization step. The LODs for the majority of analytes were 10 µg L⁻¹ and for the derivatives of BA they were 100 µg L⁻¹. The authors point out that they could not detect any PhCAs in plasma without the deconjugation step because these types of analytes are present in plasma in their conjugated forms only. The authors managed to identify several PhCAs in the plasma samples of healthy volunteers after deconjugation, including HVA, but not BA. This result is questionable. There are several studies where the concentration of BA was measured without the deconjugation step [13], [14] in the blood serum of healthy volunteers at the level of 0.3–0.7 µM (37–85 µg L⁻¹).

The necessity of the deconjugation step is an important issue. Most phenolic compounds are conjugated [27] since this is a mechanism of toxin neutralization that also improves the solubility of the non-polar compounds. At the same time, there is no information about the conjugation of septic-associated PhCAs, such as PhLA, p-HPhAA and p-HPhLA. The hypothesis of Beloborodova et al. [11] is based on biological and pathophysiological activity of the PhCAs, which could be performed by the free unconjugated forms of the PhCAs, including HVA, which are involved in the pathogenesis of septic shock. According to these data we decided not to use the deconjugation step in our research. Despite the fact that the literature contains information on the detection of a number of the PhCAs in serum or plasma samples, the data were primarily obtained using healthy volunteers. Information on the use of MEPS for the extraction of target compounds from the blood serum of critically ill patients was not found. The blood serum of such patients is a much more complicated matrix for extraction, since there are many other extraneous components in the blood of this group of patients that are absent in the serum of healthy people and can significantly distort the results of sorption/desorption. The goal of the research presented herein was to use MEPS with C18 to extract PhCAs from serum, including those from critically ill patients, followed by derivatization of the PhCAs to obtain volatiles that could then be detected using GC–MS.

2. Materials and methods

2.1. Samples and materials

Benzoic acid (BA, ≥99.5%), 2,3,4,5,6-D₅-benzoic acid (D₅-BA, ≥99 atom % D, ≥99%), phenylpropionic acid (PhPA, ≥99%), phenyllactic acid (PhLA, ≥98%), 4-hydroxybenzoic acid (p-HBA, ≥99%), 4-hydroxyphenylacetic acid (p-HPhAA, ≥98%), 4-hydroxyphenylpropionic acid (p-HPhPA, ≥98%), homovanillic acid (HVA, ≥97%), 4-hydroxyphenyllactic acid (p-HPhLA, ≥97%), N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA, contains 1% trimethylchlorosilane, 99% BSTFA), formic acid (≥95%), methyl-tert-butyl ether (MTBE, ≥99.8%), hexane (≥97.0%), methanol (≥99.9%) were obtained from Merck (Germany); sulfuric acid, hydrochloric acid, acetic acid, acetone, diethyl ether (Et₂O), sodium chloride were Laboratory Reagent grade and obtained from Khimreactiv (Russia). Serum samples from a healthy volunteer were collected in Federal Research and Clinical Center of Intensive Care Medicine and Rehabilitology (Moscow, Russia). Serum samples from critically ill patients were provided by City Clinical Hospital No. 81 (Moscow, Russia). Local ethical committee approval was received and the informed consent of all participating patients was obtained. All serum samples were frozen and kept at −18 °C. Serum samples were defrosted at room temperature prior to use.

2.2. GC–MS analysis

All GC–MS analyses were performed on a GC–MS (Trace GC 1310 gas chromatograph equipped with ISQ LT mass spectrometer) using the capillary column TR-5 ms (95% poly(dimethylsiloxane) + 5% phenyl polysilphenylene-siloxane phase, 30 m × 0.25 mm, df = 0.25 µm) obtained from Thermo Scientific (Thermo Electron Corporation, USA). The column flow was constant at 1.5 mL min⁻¹ with helium as the carrier gas, split 1:20. The GC analysis was performed in 25 min with a starting oven-temperature of 80 °C (hold time 4 min) and a single ramp of 10 °C min⁻¹ to 250 °C (hold time 4 min). The injector temperature was 200 °C and the injection volume was 1 µL. Full-scan mass spectra were recorded with m/z range 50–480 in electron-impact mode at 70 eV, using Xcalibur 2.2 software. The MS source was 200 °C and the GC–MS interface was kept at 250 °C. Scan rate was 3 scans/sec; cathode delay time 4 min; and cathode turn-off time 20 min. Trimethylsilyl derivatives of the PhCAs were identified using retention times and characteristic m/z. Mass spectra data for the trimethylsilyl derivatives of the PhCAs were proved by the NIST mass spectra library. Retention times and characteristic m/z values of trimethylsilyl derivatives of the PhCAs have been described in detail in our previous papers [13], [28]. Quantitative analyses were carried out using D₅-BA as the internal standard. All peak areas of the PhCAs (A_PhCAs) were divided by the D₅-BA area (A_St). The PhCA recoveries were calculated using formula (1):

$$\text{Recovery}\left(\%\right)=\left(A_{\mathit{PhCAs}}/A_{\mathit{St}}\right)/{\left(A_{\mathit{PhCAs}}/A_{\mathit{St}}\right)}_{100\%}\times 100,$$ (1)

• (A_PhCAs∕A_St) – relative signal obtained after sample preparation;

• (A_PhCAs∕A_St)_100% – relative signal obtained without any kind of sample preparation (except derivatization).

Statistical processing of data was performed based on the results of three parallel experiments using Microsoft Excel 2013. All data represented the mean recoveries with a confidence interval (P = 0.95, n = 3). RSD (%) was used to compare the reproducibility of the results.

2.3. Preparation of model solutions

Firstly, PhCAs and internal standard were dissolved in acetone. The mixture of all PhCAs in acetone (6 mg L⁻¹) and the acetone solution of the internal standard (7.5 mg L⁻¹) were kept at + 4 °C and used to perform the experiments without any kind of sample preparation (except derivatization). For derivatization, a 5 µL aliquot of the acetone solution of the PhCAs and a 5 µL aliquot of the acetone solution of internal standard were dissolved in 70 µL of Et₂O. The concentration of each PhCA in this solution was equal to 375 µg L⁻¹ (2–3 µM), the internal standard concentration – 469 µg L⁻¹. The solution was evaporated at 40 °C. A 20 µL aliquot of BSTFA was added to the precipitate to form trimethylsilyl derivatives (15 min, 80 °C). The mixture was cooled at +4 °C, dissolved with 60 µL of hexane and analyzed by GC–MS.

To investigate PhCA extraction from serum samples, aqueous stock solutions of each PhCA (6 µg mL⁻¹) and internal standard (7.5 µg mL⁻¹) were prepared. These stock solutions were then used to prepare solutions with concentrations of 94, 375, 750 and 2250 µg L⁻¹ to create respective calibration curves. The least squares method was used to calculate the equations of the calibration curves.

2.4. Preparation of serum samples

All experiments for the development of MEPS conditions were carried out using serum samples from one healthy volunteer. Some PhCAs are ubiquitous in the serum of healthy individuals, which was taken into account during the experiments. Aliquots (5 µL) of aqueous solutions of the PhCAs (6 µg mL⁻¹) and standard (7.5 µg mL⁻¹), concentrated sulfuric acid (2.5 µL) and distilled water (70 µL) were added to 80 µL of serum sample. A 5 µL aliquot of aqueous standard solution, a 2.5 µL aliquot of concentrated sulfuric acid and a 75 µL aliquot of distilled water were added to 80 µL of serum sample during the experiments with critically ill patient serum.

2.5. Liquid-liquid extraction (LLE)

The conditions for LLE of the PhCAs have been previously described [13]. Briefly, a 200 µL aliquot of serum and a 5 µL aliquot of aqueous solution of internal standard (7.5 mg L⁻¹) were diluted with 800 µL of distilled water. Solid sodium chloride (0.3–0.5 g) and a 15 µL aliquot of concentrated sulfuric acid were added. An extraction with Et₂O was carried out (2 × 1 mL). The ether extract was evaporated at 40 °C and derivatized with BSTFA, as described previously.

2.6. Microextraction by packed sorbent (MEPS) procedure

MEPS was performed using a 50 µL volume syringe coupled with barrel insert and needle assemblies (BINs) packed with ∼4 mg of C2, C8 and C18 (SGE Analytical, Melbourne, Australia). The influence of the most parameters (pH of model solution, conditioning solutions, procedure of sample loading, sorbent washing, elution, etc.) on the PhCA extraction from model solutions was studied previously [28]. The parameters that were investigated here included: type of sorbent, acetone concentration in model solution and conditions of sorbent washing. Details on the MEPS procedure are presented in the Results and Discussion section. Final optimized conditions for sample preparation from model solutions:

• •sorbent: C18;
• •acetone concentration in model solution: 0.2%;
• •pH of model solution: 2;
• •conditioning solutions: CH₃OH, water, 1% HCOOH, (3 × 50 μL, the speed of the sample through the sorbent – 900 μL min⁻¹);
• •sample loading: 15 × 50 μL, 300 μL min⁻¹;
• •sorbent washing: 0.1% HCOOH (2 × 20 μL, 500 μL min⁻¹);
• •drying: air (5 × 50 μL^*, 900 μL min⁻¹);
• •elution: Et₂O (10 × 10 μL, 600 μL min⁻¹);
• •derivatization: complete drying of the eluate, 20 μL of BSTFA, 15 min, 80 °C; cooling during 30 min at +4 °C, 60 μL of hexane, GC–MS analysis of 1 μL.

^*50 µL is the volume of the syringe which was filled with air during the sorbent drying.

3. Results and discussion

3.1. Extraction of the PhCAs from model solutions

Blood serum is a complex aqueous mixture of low and high molecular weight components, so aqueous solutions were chosen as the matrix for the initial development of MEPS conditions. A detailed description of the experiments with C18 sorbent in model solutions, which included the development of the basic parameters, such as the conditions of adsorption/desorption, has been reported previously [28]. The recoveries of the PhCAs reached 20–65% for hydroxylated acids (PhLA, p-HBA, p-HPhAA, p-HPhPA, HVA and p-HPhLA) and 100% for more nonpolar acids (BA and PhPA).

All experiments with model solutions (more than 100) were carried out using one syringe with C18, so we decided to compare this syringe with a new one. Three replicates were performed using each syringe. Results were compared statistically using F-test and Student’s t-test (Table 1).

Table 1.

Statistical data for F- and Student’s t-tests of the equality of two series of experiments (acetone content of 12.5% in model solutions) with old (more than 100 experiments) and new syringes with C18 sorbent for MEPS (concentrations of PhCAs – 375 μg L⁻¹, P = 0.95, n = 3).

Parameter		BA	PhPA	PhLA	p-HBA	p-HPhAA	p-HPhPA	HVA	p-HPhLA
Syringe 1	Mean recovery	103	101	53	34	31	61	55	23
	Variance	15	21	39	9	2	6	24	5
Syringe 2	Mean recovery	96	102	52	34	30	59	53	21
	Variance	41	14	40	13	6	5	37	2
F-test		0.40	1.48	1.02	1.39	2.51	1.30	1.56	2.72
t-test		1.70	0.26	0.10	0.18	0.59	1.23	0.62	1.37

The F calculated from the data for all PhCAs was not greater than the critical value of the F-distribution (F(0.95; 2; 2) = 19) and, hence, the two variances are considered equal. The t calculated from the data for all PhCAs was not greater than critical value of the t-distribution (t(0.95; 4) = 2.78) and, hence, the two means are considered equal. These results demonstrate the high efficiency of sorbent regeneration after more than 100 experiments with the quality of the results obtained being the same as when using new sorbent. Moreover, linear correlations for all PhCAs (R² ≥ 0.9982) were determined over a clinically significant range of concentrations – 94–2250 μg L⁻¹ (BA – 0.8–18 μM, PhPA – 0.7–17 μM, PhLA and p-HPhPA – 0.6–14 μM, p-HBA – 0.7–16 μM, p-HPhAA – 0.6–15 μM, HVA and p-HPhLA – 0.5–12 μM).

There were several parameters that are important for further experiments with serum samples that we have not investigated during our work with model solutions. The first parameter is acetone concentration in the model solutions. Acetone was used to dissolve the PhCAs and its concentration in the model solutions of 12.5% led to the PhCAs recoveries presented in Table 2. The results obtained using MEPS were compared with the LLE that is currently used for PhCAs extraction in routine analyses. The recoveries obtained by MEPS were higher for PhLA, p-HPhPA and HVA than by LLE, and the reproducibility of the results was much better for MEPS (RSD_(MEPS) = 3–10%, RSD_(LLE) = 10–30%). However, blood proteins denature in the presence of organic solvents, such as acetone, methanol etc. For this reason, the acetone concentration was reduced from 12.5 to 0.2%. The recoveries of BA and PhPA were decreased by 30–35%; p-HPhLA recovery increased by 15% using MEPS – an important fact, as this acid is one of the most clinically significant. A decrease of the acetone concentration leads to the change in the PhCA solubility. Solubility of BA and PhPA were reduced as these acids are more nonpolar than p-HPhLA. Furthermore, the recoveries of BA and PhPA were higher using LLE than MEPS.

Table 2.

Recoveries (%) of the PhCAs from model solutions and serum samples using LLE and MEPS (concentrations of PhCAs – 375 μg L⁻¹, P = 0.95, n = 3).

PhCA	Model solutions				Blood serum of healthy volunteer
	12.5% acetone		0.2% acetone
	MEPS	LLE	MEPS	LLE	MEPS	LLE
BA	95 ± 10	80 ± 20	60 ± 5	100 ± 10	50 ± 10	100 ± 30
PhPA	100 ± 10	100 ± 30	70 ± 5	100 ± 5	70 ± 10	80 ± 20
PhLA	60 ± 5	30 ± 10	50 ± 10	60 ± 10	50 ± 10	60 ± 10
p-HBA	35 ± 5	30 ± 10	30 ± 5	60 ± 10	30 ± 10	40 ± 10
p-HPhAA	30 ± 10	40 ± 20	40 ± 10	85 ± 20	40 ± 10	40 ± 10
p-HPhPA	65 ± 10	20 ± 10	50 ± 5	75 ± 30	50 ± 10	60 ± 20
HVA	55 ± 5	20 ± 10	50 ± 5	65 ± 30	50 ± 10	50 ± 10
p-HPhLA	20 ± 5	20 ± 10	35 ± 5	40 ± 10	45 ± 5	30 ± 20

The second stage for optimization is sorbent washing. This stage is necessary for the desorption of non-target components from the sorbent surface. Acid solutions are most often used for these purposes; we investigated the following conditions: 2 × 20 µL or 5 × 20 µL of 0.05% sulfuric acid, 2 × 20 µL or 5 × 20 µL of 0.1% or 1% formic acid. The recovery results for 1% formic acid were the lowest with recoveries for the most components less than 20%. For sulfuric acid, the results were not reproducible because of the formation of solid precipitate (the product of BSTFA hydrolysis). This could be explained by the lower volatility of sulfuric acid than formic acid. The results for 0.1% formic acid were reproducible and did not depend on the number of washing cycles. Hence, for the following experiments 2 × 20 µL of 0.1% formic acid was used.

Using the final optimized conditions for the PhCA extraction from model solutions, we performed a series of experiments with C2 and C8 sorbents. Only BA was extracted using C2 sorbent; C8 showed higher recovery than C18 for PhLA (65 ± 5 vs. 50 ± 10%), but lower recoveries for BA (50 ± 5 vs. 60 ± 5%). Nevertheless, C8 sorbent could also be used for the PhCA extraction to be studied in detail.

3.2. Extraction of the PhCAs from serum samples

Blood serum and plasma are generally diluted with water or formic acid solution before MEPS to reduce the viscosity of the sample [18], since this acts to prolong the sorbent lifetime. We explored the efficacy of using water, sulfuric acid, solid sodium chloride, and combinations thereof.

Sodium chloride increases the ionic strength of the solution leading to a decrease in the protein and peptide solubility and reversible precipitation. Sulfuric acid lowers the pH whereby proteins and peptides are irreversibly denatured and precipitated. Addition of sodium chloride or sodium chloride and sulfuric acid to serum caused formation of a white precipitate. The MEPS procedure with serum samples was carried out by passing the supernatant through the sorbent cartridge. The analyte recoveries appeared to be low. Similar results were obtained when we added sodium chloride and diluted with water (1:1 v/v).

The PhCAs were extracted from the serum after the samples were acidified with concentrated sulfuric acid to pH 2 or acidified with concentrated sulfuric acid to pH 2 and diluted with water (1:1 v/v). The preferred method of sample preparation was acidification with concentrated sulfuric acid followed by dilution with water in order to prolong the lifetime of C18 sorbent. The final conditions of the sample preparation of serum using MEPS and the PhCA recoveries are presented in Fig. 1 and Table 2, respectively. The results are also illustrated by the mass chromatogram (Fig. 2) where all PhCAs are clearly identified. In general, there were no significant matrix effects on the PhCA extraction when using MEPS. Only the recovery of BA was higher after LLE.

Fig. 1.

Scheme of the conditions for the MEPS of the PhCAs from blood serum.

Fig. 2.

Mass chromatogram of a serum sample from a healthy volunteer with the addition of the PhCAs (concentration of each – 375 μg L⁻¹) obtained using m/z values of their trimethylsilyl derivatives (104, 105, 179, 192, 193, 223, 267, 296, 326).

The limits of quantification of the optimized method for PhPA, PhLA, p-HBA and p-HPhAA were 60–100 µg L⁻¹/0.4–0.7 µM; for BA, p-HPhLA, HVA and p-HPhPA were 160 µg L⁻¹/0.9–1.3 µM. Calibration curves were used to calculate the PhCA content in blood serum in the 94–2250 μg L⁻¹ concentration range, where ≤94 μg L⁻¹ (0.5–0.8 μM) is the PhCA content in a healthy person's serum [13] and 375–2250 μg L⁻¹ (2–18 μM) is the most clinically significant concentration range of the PhCAs. This range allows one to make conclusions about the severity of a patient's condition and the presence of an infectious complication and/or multiple organ failure. The calibration equations for each of the PhCAs are linear functions in the considered concentration range (R² ≥ 0.9981) (Table 3).

Table 3.

The calibration curve equations (94–2250 µg L⁻¹/0.5–18 µM) for PhCAs (matrix – serum of healthy volunteer).

PhCA	The equations of calibration curves	Approximation coefficient, R2
BA	y = (2.17 ± 0.03) × 10−3x − (0.16 ± 0.03)	0.9997
PhPA	y = (2.44 ± 0.08) × 10−3x − (0.11 ± 0.09)	0.9981
PhLA	y = (1.06 ± 0.02) × 10−3x + (0.01 ± 0.03)	0.9989
p-HBA	y = (1.06 ± 0.02) × 10−3x + (0.01 ± 0.03)	0.9989
p-HPhAA	y = (0.61 ± 0.01) × 10−3x + (0.04 ± 0.01)	0.9997
p-HPhPA	y = (2.27 ± 0.05) × 10−3x + (0.11 ± 0.06)	0.9990
HVA	y = (1.22 ± 0.04) × 10−3x + (0.08 ± 0.04)	0.9982
p-HPhLA	y = (0.94 ± 0.01) × 10−3x + (0.01 ± 0.03)	0.9984

We can conclude that both of techniques (LLE and MEPS) provide comparable results for determination of PhCAs in serum samples in the most clinically significant concentration range. The RSD in determining the clinically significant concentration of 375 μg L⁻¹ (2–3 μM) is less than or equal to 22%. It is noteworthy that only 80 μL of blood serum is needed for sample preparation with MEPS while 200 μL is needed for LLE. The MEPS procedure takes 6 min with the entire analysis (including GC–MS) taking 45 min; meanwhile, LLE takes 20 min with the entire analysis taking more than 1 h. One syringe with C18 sorbent could be used for, at least, 20 sample preparations of serum. The MEPS technique requires several hundred microliters of organic solvent while LLE requires, at minimum, 2 mL of Et₂O.

3.3. The PhCAs in critically ill patients

A number of studies were performed to identify differences in the PhCA levels between healthy people and critically ill patients [6], [10], [11], [12], [13]. The highest concentrations of the sepsis-associated aromatic microbial metabolites were measured in patients with sepsis who had an unfavorable outcome, with their total level providing useful prognostic value [10]. The clinical and diagnostic significance of the PhCAs is currently evidential. All results were obtained using LLE as a sample preparation method, which was found to be reliable.

MEPS has also been shown to be an accurate and reliable method for extraction of components from the plasma and serum samples of mostly healthy people [22], [23]. However, when the septic process takes place in the patient, the content of peptides, fatty acids, etc., increases in blood serum due to protein synthesis abnormalities and other biochemical processes. Increasing the number of non-target components complicates analysis. Thus, the primary concern was how MEPS would cope with the task of PhCA extraction from serum samples of critically ill patients. To assess the applicability of the developed technique and also to compare against LLE, PhCA levels were determined in 4 critically ill patients at different stages of sepsis and one healthy volunteer. The results are presented in Table 4 in μM [μg L⁻¹], which is the standard accepted unit in medical practice [6], [7], [8], [10], [11], [12], [13], [14], [15]. Information concerning the disease history of these patients is presented below to highlight differences in the severity state.

Table 4.

The concentrations of the PhCAs in serum samples of healthy volunteer and critically ill patients after MEPS and LLE, µM [μg L⁻¹].

Serum samples		BA		PhLA		p-HBA		p-HPhAA		HVA		p-HPhLA
		MEPS	LLE	MEPS	LLE	MEPS	LLE	MEPS	LLE	MEPS	LLE	MEPS	LLE
Healthy volunteer		1.6 ± 0.1 [200 ± 10]	1.7 ± 0.1 [210 ± 10]	0	0	1.0 ± 0.5 [130 ± 80]	0.9 ± 0.3 [120 ± 40]	0	0	0	0	2.1 ± 0.3 [380 ± 40]	1.8 ± 0.4 [330 ± 70]
Patient 1		1.8 ± 0.1 [220 ± 10]	1.8 ± 0.1 [220 ± 10]	1.5 ± 0.2 [250 ± 30]	1.7 ± 0.4 [280 ± 60]	1.2 ± 0.5 [160 ± 80]	1.5 ± 0.3 [200 ± 40]	4.1 ± 0.2 [620 ± 30]	3.8 ± 0.4 [580 ± 60]	0	0	6.8 ± 0.4 [1230 ± 70]	7.4 ± 0.5 [1350 ± 90]
Patient 2		1.5 ± 0.1 [180 ± 10]	1.6 ± 0.1 [200 ± 10]	9.3 ± 0.4 [1550 ± 60]	10 ± 1 [1600 ± 200]	0	0	>15* [>2250]	29 ± 3 [4400 ± 500]	6.3 ± 0.4 [1150 ± 70]	5.7 ± 0.4 [1030 ± 70]	>12* [>2250]	43 ± 3 [7800 ± 500]
Patient 3	2nd day	2.4 ± 0.1 [290 ± 10]	2.2 ± 0.1 [270 ± 10]	0	0	0	0	0	0	5.0 ± 0.4 [900 ± 70]	6.1 ± 0.5 [1110 ± 90]	0.8 ± 0.2 [140 ± 40]	0.9 ± 0.2 [160 ± 40]
	6th day	2.2 ± 0.1 [270 ± 10]	2.0 ± 0.1 [250 ± 10]	6.2 ± 0.3 [1040 ± 50]	7.1 ± 0.8 [1100 ± 100]	0	0	3.3 ± 0.2 [500 ± 30]	3.1 ± 0.4 [490 ± 60]	0	0	> 12* [> 2250]	39 ± 3 [7100 ± 500]
Patient 4	2nd day	2.0 ± 0.1 [250 ± 10]	1.9 ± 0.1 [230 ± 10]	1.7 ± 0.2 [290 ± 30]	2.0 ± 0.4 [330 ± 60]	0	0	1.1 ± 0.1 [170 ± 20]	0.9 ± 0.2 [140 ± 30]	0	0	11.2 ± 0.5 [2040 ± 90]	12 ± 1 [2200 ± 200]
	3rd day	3.9 ± 0.2 [470 ± 20]	3.6 ± 0.2 [440 ± 20]	1.6 ± 0.2 [270 ± 30]	1.6 ± 0.3 [270 ± 50]	0	0	4.6 ± 0.2 [700 ± 30]	3.9 ± 0.5 [590 ± 80]	0	0	6.7 ± 0.4 [1230 ± 70]	6.7 ± 0.4 [1230 ± 70]

^*The concentration is beyond the considered range of linearity – 94–2250 μg L⁻¹ (BA – 0.8–18 μM, PhPA – 0.7–17 μM, PhLA and p-HPhPA – 0.6–14 μM, p-HBA – 0.7–16 μM, p-HPhAA – 0.6–15 μM, HVA and p-HPhLA – 0.5–12 μM).

Healthy volunteer: All biochemical parameters were normal; PhCA concentrations were less than 2 μM. The PhPA concentration was (0.8 ± 0.2) μM/(120 ± 40) μg L⁻¹ and (1.0 ± 0.3) μM/(150 ± 50) μg L⁻¹ for MEPS and LLE, respectively. PhPA is a characteristic acid found in healthy people and is generally absent from the serum of critically ill patients; accordingly, it is not presented in Table 4.

Patient 1: 64 years old, admitted to intensive care unit (ICU) in severe state after surgery for acute peritonitis (perforation of the intestine). Despite intensive therapy, multiple organ failure was progressing, which led to death on the second day in ICU. The PhCA concentrations were measured in the serum sample which was collected a day before death and showed increased levels of PhLA, p-HBA, p-HPhAA and p-HPhLA in comparison with normal levels.

Patient 2: 60 years old, admitted to ICU with pneumonia in critical state. Despite complex therapy (i.e., infusion, antibacterial, vasopressor support, artificial ventilation, etc.), multiple organ failure and systemic inflammatory reaction progressed. Deterioration of the condition led to death on the sixth day of treatment in the ICU. A serum sample was collected the day before the death and the concentrations of PhLA, p-HPhAA, HVA and p-HPhLA greatly exceeded normal levels (Fig. 3).

Fig. 3.

Mass chromatogram of serum sample of critically ill patient 2 (collected a day before death) obtained using m/z values of trimethylsilyl derivatives of the PhCAs (104, 105, 179, 192, 193, 223, 267, 296, 326). The relative abundance of the p-HPhLA peak is 37 (t_r = 18.03).

Patient 3: 60 years old, admitted to ICU with pneumonia and an intestinal tumor. On the second day, the patient’s state was stable; however, progression of the infectious process and multiple organ failure (sepsis) led to death on the seventh day. The PhCA determination was carried out on blood samples collected on the second and sixth days of treatment. During this period, the content of PhLA, p-HPhAA and p-HPhLA changed from normal to critical values; the concentration of HVA decreased.

Patient 4: 46 years old, admitted to ICU with chronic alcohol intoxication, episyndrome and gastrointestinal bleeding. Due to worsening of the patient’s state and despite the ongoing therapy, death occurred on the fourth day. The PhCA determination was performed on blood samples collected on the second and the third days. The concentrations of BA and p-HPhAA were increased and p-HPhLA concentration was decreased.

This investigation demonstrates the comparability of MEPS and LLE results and the possibility to analyze complex matrices (critically ill patient serum) using MEPS with C18 sorbent. This allows us to recommend MEPS, instead LLE, for PhCA determination in serum samples.

If we compare the obtained data, in particular the concentration of the PhCAs in a healthy volunteer, with the data of other authors [22], in which only PhPA was determined in healthy volunteer samples after deconjugation (p-HBA was found in trace amounts), and others [23], where even BA was not detected without a deconjugation step, we managed to quantify BA, PhPA and p-HBA without an additional deconjugation stage. Moreover, a large volume injection technique with PTV injection was applied in one study [23] and MS/MS detection in another [22]. In both cases, the detection limits should have been lower than in our study. According to these data, we can conclude that the deconjugation step possibly influences the matrix effect by increasing the content of non-target components (i.e., the products of deconjugation).

4. Conclusions

Conditions for microextraction of 8 PhCAs from model solutions and serum samples using C18 packed sorbent were developed. MEPS is a potential analytical method that could be applied to the rapid and reliable early diagnosis of sepsis. MEPS requires relatively low volumes of serum and organic solvents. It is a relatively rapid method that is also amenable to automation as of a GC–MS workflow.

Declaration of Competing Interest

None.

Appendix A. Supplementary data

The following are the Supplementary data to this article: