Novel LC-MS-TOF method to detect and quantify ascorbic and uric acid simultaneously in different biological matrices

Abstract

Ascorbic acid (AA) and uric acid (UA) are known as two of the major antioxidants in biological fluids. We report a novel liquid chromatography–mass spectrometry with time-of-flight (LC-MS-TOF) method for the simultaneous quantification of ascorbic and uric acids using MPA, antioxidant solution and acetonitrile as a protein precipitating agent. Both compounds were separated from interferences using a reverse phase C18 column with water and acetonitrile gradient elution (both with formic acid) and identified and quantified by MS in the negative ESI mode. Isotope labeled internal standards were also added to ensure the accuracy of the measures. The method was validated for exhaled breath condensate (EBC), nasal lavage (NL) and plasma samples by assessing selectivity, linearity, accuracy and precision, recovery and matrix effect and stability. Sample volumes below 250 μl were used and linear ranges were determined between 1 – 25 and 1 – 40 μg/mL for ascorbic and uric acid, respectively, for plasma samples, and between 0.05 – 5 (AA) and 0.05 – 7.5 (UA) μg/mL for EBC and NL samples. The new method was successfully applied to real samples from subjects that provided each of the studied matrices. Results showed higher amounts determined in plasma samples, with similar profiles for AA and UA in EBC and NL but at much lower concentrations.

1. Introduction

Antioxidants such as ascorbic acid (AA) and uric acid (UA) usually coexist in human biological fluids like blood, urine or saliva. These compounds have important effects on the defense mechanisms of the body, functioning as scavengers of free radical able to moderate the oxidative stress effects of various diseases. The supplementation of these antioxidants on a diet has been proposed as a therapeutic strategy against the oxidative stress, from preventing decrease of coronary flow velocity in patients with ischemic heart disease, protection against damaging action of inhaled oxidant gases [1] or against hyperoxia-induced oxidative stress and increase in arterial stiffness in healthy humans [2].

Vitamin C is a water-soluble vitamin that is typically found in cells and biological fluids essentially as AA (95%), the reduced form, compared to the reversible oxidized dehydroascorbic acid (DHAA) form. Ascorbic acid plays a very important role in the human body as a reducing agent, nutritional factor, and enzyme cofactor that protects against diseases caused by oxidative stress; therefore, the concentration of vitamin C is measured as a sum of the contents of AA and DHA [3, 4]. Humans are not able to synthetize AA, and it must be consumed daily from a variety of dietary sources such as fruits and vegetables [5, 6]. Inadequate intake will result in the symptoms of scurvy or gingival bleeding, and an excess of AA intake will also lead to urinary stone, diarrhea and stomach convulsion [7].

Uric acid is the end product of endogenous and dietary purine metabolism in humans, playing an important role in human fluids and blood. UA is excreted in urine by the kidneys, and plasma UA levels in humans are appreciably higher as compared to those in most mammals [6]. Abnormal UA concentrations in human body can indicate the presence of numerous diseases or physiological disorders, such as renal disorder metabolic syndrome [8, 9], type 2 diabetes [10], coronary heart disease [11], head and neck cancer, obesity, hyperuricemia and gout [12, 13]. AA and UA are considered useful biomarkers to assess the health and nutritional status as well as disease diagnostic. Both compounds often appear together in biological fluids resulting in interference when attempted to be simultaneously measured. Thus, it is important to develop a simple, sensitive and simultaneous determination method of AA and UA, especially as it relates to diagnostic approaches in analytical chemistry [6, 14–16]. Moreover, we observe the reference values for AA and UA in body fluid differ significantly between authors (e.g. 10 – 90 μg/mL AA and 20 – 75 μg/mL UA in plasma) [17, 18] being the main reason for these changes because of the differences in analytical methods or sample preparation procedures.

There are several methods presented in the literature to determine ascorbic and uric acids in biological samples. These methods are usually based on enzymatic [19], spectrophotometric [20] or separation techniques like high performance liquid chromatography (HPLC) [21, 22] and capillary electrophoresis (CE) [23, 24], both mainly coupled with electrochemical (ECD), ultraviolet (UV) or fluorescence detection systems [25–27]. These methods have been used to determine only a single antioxidant and can lead to imprecise results in the case of enzymatic approaches or lack of sensitivity in the case of ECD. Since biological samples are considered a complex matrix, the determination of these antioxidants requires high sensitivity and good selectivity allowing their detection in the presence other interfering species. Moreover, if the goal is to detect AA and UA simultaneously, these problems are more concerning and few methodologies are described in literature, most of them are based on sensors [18, 28], HPLC-ECD [27, 29, 30], CE-ECD [15, 16] and HPLC-UV [6, 21, 22, 30–32]. While HPLC and CE techniques have provided the desired sensitivity, ECD and UV detection still lack enough selectivity and specificity. Hence, the use of HPLC coupled to mass spectrometry (HPLC-MS), which provides reproducibility accuracy and efficiency improvement, has been proved as a powerful tool for the determination of individual AA [3, 33, 34] and UA [35–40] in biological samples. Currently, there are no methods that determine both compounds simultaneously in these biofluids using reproducible and sensitive techniques like LC-MS.

HPLC-MS simplifies the preparation procedures, shortens analytical times and even provides information for confirmation purposes. However, we found no references in the literature that tackle both compounds in the same analysis from biological fluids. The main reason is because specific pre-analytical procedures are required for a reliable determination both antioxidants that coexists in body fluids. This can be explained by the unstable nature of AA, which is extremely sensitive to oxidation and degradation during the sampling, handling, storage and analysis of the specimens [41]. While UA is relative stable, AA is easily oxidized if not protected from natural or UV light, high temperatures, oxygen or matrix conditions (e.g. pH). A protein precipitation and AA stabilization step is typically incorporated into sample processing using acids or methanol often combined with a metal chelator such as ethylenediaminetetraacetic acid (EDTA) or diethylene-triaminepentaacetic acid (DTPA) [3, 6, 31, 33, 41–45].

Additionally, when using biologic specimens for health monitoring and diagnosis, the sample collection procedure has exceptional relevance in order to provide faster, simpler and non-invasive methodologies. In the case of plasma, sampling involves invasive procedures including a blood collection that can cause pain. Instead, the use of non-invasive methods can overcome the limitations by using other body fluids like tears, sweat, saliva, nasal lavage (NL) or exhaled breath. These methods allow a relative fast and simple collection procedure to obtain biomarkers for health screenings [25]. Several studies have proofed the usefulness of exhaled breath condensate (EBC) [46–48] and nasal lavage [49] to determine non-volatile metabolites and biomarkers used as health diagnosis. Thus, understanding the presence of certain compounds in different biological matrices can provide information for future study designs using viable non-invasive alternatives to blood sampling [50].

This paper presents a detailed method of this new approach. We develop and validate a novel method to simultaneously detect AA and UA using a high accurate and reproducible HPLC-MS-TOF method in three different matrices: plasma, nasal lavage (NL) and exhaled breath condensate (EBC). We use commercially available isotope-labeled internal standards to accurately measure antioxidants with excellent sensitivity, high specificity, high throughput, and a short run time. The novelty of this work also consists in the new analytical technique used for the determination of both compounds, paying, special attention to understand the limitations of AA stability in three different matrices that allow innovative capabilities for the studies of these antioxidants using non-invasive techniques. The presented approach was employed for the assessment of AA and UA levels in EBC, NL and plasma samples from 31 subjects with no declared chronic diseases which were treated with Vitamin C and fructose supplements in a double blinded study to determine the dietary modulation of antioxidant levels.

2. Materials and methods

2.1. Reagents and chemicals

L-ascorbic acid, uric acid, uric acid-1,3–15N2, 98% (IS-UA), meta-phosphoric acid (MPA), ammonium hydroxide (NH4OH) and EDTA were purchased from Sigma-Aldrich (St. Louis, MO). L-ascorbic acid-13C6, 98% (IS-AA) was obtained from Cambridge Isotope Laboratories (Tewksbury, MA). LC-MS grade acetonitrile, methanol, water and formic acid were obtained from Fisher Chemical (Waltham, MA).

2.2. Standards preparation, calibration and quality control (QC) samples

Stock solutions were prepared separately containing each 10 mg/mL of AA in 5% aqueous MPA and 1 mg/mL of UA in 20 mM aq. NH4OH. Internal standards were prepared at 10 mg/mL of IS-AA in 5% aqueous MPA and 500 μg/mL of IS-UA in 20 mM aq. NH4OH. The AA and IS-AA stock solutions were freshly daily prepared carefully protected from light in amber flasks during preparation and analyses, and were also maintained in ice. Intermediate solutions were also prepared individually maintaining the initial solvents from the stock solutions. All stock and intermediate solutions were stored at −20 °C. The antioxidant solution was prepared using EDTA at 0.2 mg/mL each in a solution of methanol-water (1:1, v/v). 10 μL of this antioxidant solution were spiked in all calibration, QCs and samples. The calibration standard solutions were prepared by mixing and diluting the stock solutions with water (H2O)-acetonitrile (AcN) (95:5, v/v) solution, covering the concentration ranges between 0.01 and 25 μg/mL for AA and 0.01 and 40 μg/mL for UA. Quality controls (QCs) and other spiked samples were prepared by spiking the corresponding amounts of stock/intermediate solutions into pooled EBC, NL and plasma. All calibration standards, QCs and spiked samples were stored in the freezer at −80 °C no more than one week until the analysis.

2.3. Sample collection and preparation

Plasma samples.

Blood was collected after venous puncture into EDTA vacutainer tubes. Plasma samples were generated by immediately centrifugation at 550 ×g for 20 min at 4 °C. Plasma was vortexed with cold freshly prepared 10% MPA (1:1; v/v) for 30 s and stored at−80 °C until further analysis.

EBC.

Exhaled breath condensate was collected using a Jaeger EcoScreen system. Subjects were instructed breath quietly into a mouthpiece for 20 minutes with their noses plugged using a nose clip. During the breathing maneuver, the exhaled breath passes through a mouthpiece and a trap that separates saliva and larger particles before passing through a cold trap maintained at −20° C.

Nasal lavage.

Nasal lavage samples were collected by instilling 10 ml of sterile saline into the nose and the sample collected via mild suction. The recovered sample was transferred into a sterile plastic receptacle maintained on wet ice. The lavage was self-administered by the subject after demonstration of the procedure, with the aspirate collected after instillation to limit the dwell time of the saline in the nose. In total the protocol took approximately 30 seconds. The recovered aspirate was filtered through a sterile 100-μmpore nylon filter to remove mucus aggregates before centrifugation at 550 ×g for 20 minutes (4◦C) to isolate the cell free fraction.

EBC and nasal lavage samples were similarly acidified with MPA as described for the plasma samples and the de-proteinated supernatant stored at −80◦C.

Stability is the main problem when analyzing biological samples containing AA. Plasma, NL and EBC samples were prepared with a similar procedure but with slight variations adapted to each matrix characteristics. Before sample preparation, samples were slowly thawed inside ice and protected from light. Each sample was transferred into Eppendorf tubes covered with aluminum foil in a closed/dark box and maintained in ice. For each matrix, 250 μL of sample were precipitated with acetonitrile and antioxidant solution inside Eppendorf tubes. All the tubes included 5 μg/mL of IS-AA and 10 μg/mL of IS-UA. The samples were vortexed for 3 min, kept at −20 °C for 5 min and vortexed again for 1 min. Then, the samples were centrifuged for 10 min at 17,500 ×g and 4 °C. After centrifugation, the supernatant was extracted and injected into the LC-MS-TOF system. Sample sets were prepared fresh for every analytical run.

2.4. Instrumentation and data analysis

Sample analyses were carried out with an Agilent 1290 series ultrahigh-performance LC system coupled with an Agilent 6230 time-of-flight (TOF) mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). Two different types of reversed phase columns were initially tested for AA and UA separation from the other matrix components: Zorbax Stable Bond (SB) C18–3.5 μm (150 mm × 4.6 mm ID) coupled with Zorbax SB C18–3.5 μm (15 × 4.6 mm) guard column and Zorbax Eclipse XDB-C18-1.8 μm (50 mm × 2.1 mm) coupled to a Zorbax XDB-C18-1.8 μm (15 × 2.1 mm) guard column (Agilent Technologies, Santa Clara, CA, USA). The best resolution was investigated using a mobile phase containing water (A) and acetonitrile (B), both with 0.1% of formic acid. The flow rate was 0.6 mL/min and column temperature was kept at 28 °C. Vials were kept at 4 °C in an autosampler before the injection of 20 μL. A variable wavelength DAD detector was used at 245 and 265 nm just for confirmatory purpose.

An electrospray ionization (ESI) source with an Agilent Jet Stream nebulizer was used in negative mode, with acquisition in a mass range between 50 and 750 (m/z). The source was set with the following parameters: gas temperature at 350 °C, gas flow at 8.5 L/min, nebulizer pressure at 40 psi, capillary voltage at 2000 V, fragmentor voltage at 130 V and nozzle voltage at 0V. Since TOF provides the accurate exact mass of the ions, each compound was monitored for m/z 175.0248 and 181.0444 for AA and IS-AA, and 167.0211 and 169.0146 for UA and IS-UA, respectively. Finally, the data were processed and analyzed by Agilent MassHunter Workstation software.

AA and UA concentrations in human plasma, NL and EBC were quantified by plotting nominal concentrations versus the response factor of each analyte (ratio Peak Area [Analyte]/Peak Area[IS]) using a linear regression. All calculations were further processed using Microsoft Excel.

2.5. Method validation

The proposed method was validated by considering linearity and specificity within an expected concentration range, precision, accuracy, recovery and matrix effect, limits of detection and stability of each compound. All parameters were determined by each matrix defined in the study.

2.5.1. Selectivity

The selectivity is the ability to differentiate and quantify analytes of interest in the presence of other components from the sample. This method was verified by separating each compound peak in the chromatogram of each of the blank matrices of the study (EBC, NL and plasma), a standard of the calibration mixture and blank matrix samples spiked with internal standards (ISs) and standard mixture.

2.5.2. Linearity, specificity, and sensitivity

The linearity of the method was determined by using the response ratios (ratios of peak areas of analytes and IS) from freshly made AA and UA standard solutions within the expected and exceeding concentration ranges. Since the different matrices contain different amounts of each compound, we worked with 7–8 levels of calibration and concentration ranges adapted to the expected concentrations. A linear function of concentrations of each analyte was determined obtaining the slope (m), the intercept (b), and the correlation coefficient (r2) from the regression analysis, using a weighting factor of 1/concentration weighting (x). Standards were run in three separate analytical runs and the calibration model was accepted if the back-calculated concentrations of the calibration standards were within 20% deviation of the nominal value and within 25% deviation for the lower limit of quantification (LLOQ). At least 75% of the calibration standards should meet this criterion and the correlation coefficient should also be greater than 0.99.

The lower limit of quantification (LLOQ) was established as the lowest calibration standard with an accuracy and precision less than 20%. Each compound signal of the LLOQ sample should be at least 5 times the signal of the blank sample. The specificity of the method was evaluated by analyzing processed standards (in mobile phase), both blank and spiked with AA and UA. The sensitivity was determined by detecting the limit of detection (LOD) and the limit of quantification (LOQ) using a lower analyte concentration that would yield a signal-to-noise ratio (S/N) of 3 and 10, respectively.

2.5.3. Precision and accuracy

The precision (CV, %) and accuracy (%) of the method were evaluated by performing intra- and inter-day injections of multiple replicates at low- and high-level of the concentration for AA and UA in all the biological matrices. Precision was checked through percentages of the coefficient of variation (CV) of the peak area ratio and accuracy was evaluated by determining the recovery (ratio of the determined and the nominal values of concentrations multiplied by 100%) of AA and UA in samples with known amounts of these two antioxidants. In both cases, values from non-spiked (blank) samples were subtracted from amounts determined in spiked samples with known added antioxidants. These parameters were obtained by measuring seven times in one day and repeating three injections on five separate days/runs. The maximum tolerated CVs were 20% and values of intra- and inter-day accuracy should be within 85 – 115% for not less of 75% of the QCs in any batch.

2.5.4. Recovery and matrix effect

The process recovery and the matrix effect were determined in different spiked plasma, NL and EBC samples with QC concentrations for each AA and UA. To determine the recovery, two nominal concentrations of the standard mixtures (high and low-level) were spiked into each matrix samples (n = 3), then processed, and analyzed with replicate injections. The calculated concentrations of the pooled non-spiked samples (blanks) and the values were subtracted from the calculated concentrations form the spiked samples. The difference was multiplied by 100 to get the percent recovery. The percent matrix effect was studied by analyzing the ratio of the peak area response in the presence of matrix (plasma, NL and EBC spiked with AA, UA and ISs) to the peak area in absence of the matrices (pure standard in mobile phase). A CV of ≤15% across all QC concentrations was set as the level of acceptance for both recovery and matrix effect. Carry over was also investigated by injecting blank samples after high concentration standards (25 μg/mL for AA and 40 μg/mL for UA). The peak area detected at the specific retention time should be lower than 20% of LLOQ of the method.

2.5.5. Stability studies

We also know that the pH of the solution, temperature and light are the most decisive factors influencing the AA solutions stability. With special attention to the well-known influencing factors, for the validation of the method the following tests were evaluated: long term stability of the analyte in matrix stored in the freezer and stability of the processed sample at autosampler temperature. For the method validation, the stock solutions were processed in amber flasks and standard solutions were ready to be injected within 1–2 h. The stability of AA and UA QC samples was tested for different intervals of time and freeze spiked samples (−80°C) were stored for maximum three days using the proposed HPLC-MS method.

2.6. Application of the method

The developed method was applied to detect and quantify AA and UA in a series of EBC, NL and plasma samples collected from 31 healthy participants on a double-blinded randomized study. All samples were collected in a fixed order of blood draw (plasma), EBC, nasal lavage within a 30-minute period between each procedure.

The use of human subjects was approved by the University of California Davis Institutional Review Board (IRB 776722). 31 healthy adults (18–35 years, either gender) who were non-smokers served as subjects. All subjects first completed an orientation session, in which human subject consent was obtained, baseline pulmonary function was performed, and a respiratory health questionnaire answered. After the subject’s respiratory health status had been cleared by a designated project physician, the subject returned to the laboratory to complete study orientation.

Vitamin C and Uric Acid Dietary Modulation.

In order to maintain a normal but constant antioxidant intake three days prior to each sample collection, enrolled subjects were given instructions on how to keep a one-week dietary record. Subject’s dietary records were analyzed using a standard nutrient database (NDRS) for antioxidant intake. Subjects were then instructed on how to maintain a constant normal intake of antioxidants for three days prior to each sample collection. In addition, subjects were asked to abstain from strenuous exercise, non-steroid anti-inflammatory drugs, tea, alcoholic beverages, and caffeine- or theobromine-containing foods for 24 h before each sample collection. Each subject completed both the Antioxidant Modulation (AM) and Placebo (P) regimen. One hour prior to the AM protocol subjects drank a fructose/vitamin C beverage (267 g/L fructose, 3 ml/kg plus 1 g Vit C). The P protocol consisted of the same regimen except the subject consumed a glucose beverage (267 g/L glucose, 3ml/kg, no Vit C). The timing of supplementation was based on the previous dosing with vitamin C [1] and the time course of changes in plasma uric acid following fructose intake [2]. The order of AM and P regimes were randomized and double-blinded.

3. Results and discussion

3.1. LC-MS-TOF separation

The separation of the standards was initially tested with two different columns, both rapid resolution reverse phases C18 recommended for acidic highly polar compounds, but with different bond (SB and XDB) and dimensions. The first column, Zorbax SB, had special smaller porous (1.8 μm), length (50 mm) and ID (2.1 μm) to allow ultra-fast separation. Peak separations are shown (Figure 1). Fig 1a shows all peaks eluting at the same retention time for Zorbax SB column, not being able to retain any of the compounds from the standard mixture (elution time before 1 min). However, using the same flow and mobile phase conditions, but with column Zorbax XDB (Fig 1b), peaks in the mixture were separated showing good resolution. This column, with bigger porous (3.5 μm), length (150 mm) and ID (4.6 μm) was able to retain AA and UA ant elute them separately before 4 min.

Figure 1.

LC-MS peaks separation using Zorbax SB (a) and XDB (b) columns.

To select the MS parameters, a mixture of AA, UA and their ISs of 5 μg/mL each was injected into the system in full scan in positive and negative modes. ESI negative mode showed the best response, where the deprotonated molecular ions [M-H]- (Table 1) were detected very well on the MS chromatograms. Typical LC-MS chromatograms and the corresponding mass spectra for AA, UA and ISs are shown (Figure 2). Retention time for AA and IS-AA was 2.90 min and for UA and IS-UA was 3.12 min (Table 1). Figure 2 shows the mass spectrum obtained at 2.90 min (Fig 2 left) with most abundant ions at m/z 175 and 181, corresponding to [M−H]− for AA and IS-AA, respectively. Correspondingly, Fig 2-right shows the mass spectrum obtained at 3.12 min with higher ions at m/z 167 and 169 from [M-H]- for UA and IS-UA, respectively. Additionally, the mass of the deprotonated dimers [2M-H]- was also detected for all the compounds as confirmatory information.

Table 1.

Main properties AA, UA and their ISs. LC-MS retention time (RT), exact mass, molecular ions ([M-H]− and [2M-H]−) and molecular formula.

Compound	RT (min)	M (g/mol)	[M-H]−	[2M-H]−	Formula
Ascorbic acid	2.90	176.03	175.0248	351.0564	C6H8O6
IS-AA	2.90	182.05	181.0444	363.0966	13C6H8O6
Uric acid	3.12	168.03	167.0211	335.0489	C5H4N4O3
IS-UA	3.12	170.02	169.0146	339.0370	13C5H4N4O3

Figure 2.

Typical LC-MS chromatograms from Extracted Ion Chromatograms in negative mode (-EIC) corresponding to AA and IS-AA (left) and UA and IS-UA (right). Mass spectra extracted at the specific times is also presented at 2.90 min (left) and 3.12 min (right)

The identification of the peaks in the matrices was done by comparison of the retention times in the pure standards using in the same analytical conditions and extraction of exact masses for each compound (Table 1). Matrix samples identifications and interferences were studied with spiked commercial standards and ISs to EBC, NL and plasma samples. Representative LC-MS-TOF chromatograms of each matrix are shown (Figure 3). The retention times for AA and UA were maintained from the pure standards run. When matrix was included, AA maintained a good resolution and peak shape, however, UA and IS-UA lost some resolution, mainly when the concentrations were low. Although a perfect resolution was not easily achieved for these compounds, peak shape was enhanced when samples and standards were freshly prepared every day.

Figure 3.

LC-MS chromatogram from the analysis of EBC (a), NL (b) and plasma (c). Samples are run with Zorbax SB C18–3.5 um column. Ascorbic acid (AA) and its internal standard IS-AA appear at 2.9 min and uric acid (UA) and its internal standard (IS-UA) appears at 3.1 min.

3.2. Sample preparation

For a reliable and correct chromatographic separation of AA and UA from the studied matrices, it is required to take specific pre-analytical procedures since stability is a crucial problem for these antioxidants. Real samples and standards must be kept in the dark and in ice, since temperature, light, and the presence or absence of oxygen the antioxidants degradation. Plasma, NL and EBC samples were acidified with 5% aqueous MPA to stabilize the sample before storage at −80 °C. [3, 6, 31, 33, 41–45]

Additionally, just before sample preparation each matrix was spiked with an organic modifier and a stabilizer (EDTA solution) to precipitate proteins and protects the molecule from oxidation, particularly when they are employed together. Two different modifiers were tested in the study with plasma samples: acetonitrile and methanol (Figure 4).

Figure 4.

LC-MS data from plasma sample and plasma sample spiked with UA and AA standards and ISs treated with acetonitrile (a) and methanol (b) as modifiers to precipitate proteins. TIC (Total ion chromatogram) and EIC (Extracted Ion Chromatogram).

It was found that methanol addition to plasma samples created split peaks with a big shoulder when the specific masses were extracted for AA, UA and their corresponding ISs. Moreover, the AA and UA co-eluted at the same retention time with methanol, while acetonitrile maintained good resolution and separation of both chromatographic peaks [42]. Acetonitrile was selected as a solvent to spike in the samples and QCs.

For NL and EBC, an additional sample preparation procedure was tested to improve the analytes responses [47, 48]. Since these two matrices contain lower amount of antioxidants than plasma, compounds are usually very diluted in the matrix and usually requires pre-concentration treatments to enhance signals. Pre-concentration by lyophilization was applied in clean EBC and NL samples (blanks) that were previously spiked with antioxidant solution mix and ISs, as well as additional spikes of commercial standards (Spk 1 and Spk2). A total of 500 μL of pooled samples were mixed, spiked and vortexed. Lyophilization samples were freeze dried during 24h and the dried extract was reconstituted in a small amount of mobile phase (25 μL) before LC-MS-TOF analysis. Another batch of samples were diluted with solvent were spiked with 500 μL of acetonitrile (dilution AcN), vortexed, incubated and centrifuged. The supernatant was run in the LC-MS. Quantification results are summarized (Table 2). When compared both treatments, we could observe that AA (and IS-AA) disappeared during the freeze-drying process. Maybe the time required to sublimate the water in the sample could degrade this compound. Although AA was not detected in Blank EBC samples for the ‘dilution AcN’ process, we were able to detect the IS-AA in all cases and it achieved good recovery in both matrices (92–103% for EBC and 96–101% for NL). Differently, UA was detectable with all treatments with good recovery in all cases (106 – 125% EBC and 97 – 126% NL). However, UA was not detected in EBC when diluted due to the low concentration of those compounds in this matrix. Because of that, we reduced the dilution amount in final sample preparation with acetonitrile to enhance the compounds signal and avoided lyophilization to determine these antioxidants.

Table 2.

Concentrations detected in EBC and NL pooled samples spiked with AA and UA standards. Results are expressed in mean concentration (μg/mL injected sample) ± standard deviation (SD), and recovery (R, %) based on expected concentration.

Compound	Sample	EBC				NL
		Lyophilization		Dilution AcN		Lyophilization		Dilution AcN
		Mean ± SD	R (%)	Mean ± SD	R (%)	Mean ± SD	R (%)	Mean ± SD	R (%)
Ascorbic acid	Blank	n.d	-	n.d.	-	n.d	-	0.46 ± 0.07	-
	Spk 1	n.d	-	9.25 ± 0.03	92	n.d	-	10.03 ± 0.13	96
	Spk 2	n.d	-	5.10 ± 0.05	103	n.d	-	5.45 ± 0.05	101
Uric acid	Blank	0.25 ± 0.01	-	n.d.	-	80.53 ± 1.32	-	1.09 ± 0.04	-
	Spk 1	10.60 ± 0.36	106	17.96 ± 0.06	114	90.24 ± 2.57	97	19.47 ± 0.25	119
	Spk 2	22.67 ± 0.44	113	9.96 ± 0.09	125	105.21 ± 1.98	123	13.58 ± 0.13	126

n.d: not detected peak; Blank: Clean matrix with only ISs; Spk 1: Add IS and 10 μg/mL of AA and 20 μg/mL of UA; Spk 2: Add IS and 5 μg/mL of AA and 10 μg/mL of UA

3.3. Method validation

3.3.1. Selectivity

Selectivity of the method was proved in previous sections (3.1 and 3.2). We were able to separate each compound peak in the chromatogram from pure standard mixes (Figure 1), each of the blank matrices of the study (Figure 3) and blank matrices spiked with commercial standards. The method was selective enough to enable efficient extraction and separation at 2.9 and 3.12 min to detect AA, UA and their corresponding ISs without interference from solvent, EBC, NL and plasma.

3.3.2. Linearity and LLOQ

The validating parameters for the calibration curves: slope (m), the intercept (b), the correlation coefficient (r2) and the linear range for the instrumental response is shown in Table 3. Since expected concentrations varied for the matrices, two calibration ranges were assessed considering plasma and NL or EBC samples. The calibration curve linearity was injected two times on three different days and was determined for AA between 25 and 1 μg/mL (8 levels) for plasma samples and between 5 and 0.05 μg/mL (7 levels) for EBC and NL samples. And, similarly, UA calibration curves ranged between 40 and 1 μg/mL (7 levels) for plasma and between 7.5 and 0.05 μg/mL (8 levels) for EBC and NL. Results indicated a good linearity for both analytes over the referred concentration ranges with correlation coefficients (r2)≥0.99. The LOD of the assay (at a signal-to-noise ratio of 3) for AA and UA was 0.01 and 0.005 μg/mL, respectively, while the LOQ (a signal-to-noise ratio of 10) were 0.025 μg/mL in both cases.

Table 3.

Regression parameters, linear range and limits of detection and quantification of plasma, NL and EBC for ascorbic and uric acid

Regression parameters	Ascorbic acid		Uric acid
	Plasma	NL and EBC	Plasma	NL and EBC
Linear range (μg/mL)	25 – 1	5 – 0.05	40 – 1	7.5 – 0.05
Slope (m)	0.258	0.198	0.103	0.080
Intercept (b)	−0.0348	−0.0010	−0.0147	−0.00006
Correlation coefficient (r2)	0.991	0.990	0.990	0.994
Limit of quantification, LOQ (μg/mL)	0.025		0.025
Limit of detection, LOD (μg/mL)	0.01		0.005

3.3.3. Precision and accuracy

Intra and inter-day precision and accuracy was evaluated by preparing daily calibration standards of low and high levels of concentrations. Three replicates standards were run over five separate days (inter-day) and seven replicates were also injected consecutively within the same run (intra-day). Results obtained for precision, accuracy and calculated concentration are presented in Table 4. The intra-day and inter-day precision, expressed as the coefficient of variation (CV, %) were below 20%, being values at low-level of concentration for inter-day study the ones with higher variation (18.6 and 12.7% CV for AA and UA, respectively). Accuracy data, expressed as the bias from the nominal spiked concentration (RE, %) was maintained between 85 – 115%, except for low levels of UA, which were 76 and 79% for intra- and inter-day accuracy, respectively. These results suggest that the developed method is precise, accurate and reproducible.

Table 4.

Inter- and intra-day precision (expressed as coefficient of variation, CV, %) and accuracy data (%) for the calculated concentration (mean ± SD) at two different concentration levels for AA and UA acids.

Compound	Level	Spiked NC	Intra-day (n = 7)			Inter-day (n = 3)
			Mean CC ± SD	Precision (CV, %)	Accuracy (%)	Mean CC ± SD	Precision (CV, %)	Accuracy (%)
Ascorbic acid	High-level	6.15	5.42 ± 0.07	1.3	88	5.65 ± 0.26	4.5	92
	Low-level	0.308	0.261 ± 0.007	2.5	85	0.286 ± 0.039	18.6	93
Uric acid	High-level	9.93	10.52 ± 0.11	1.0	100	9.93 ± 0.43	4.2	95
	Low-level	0.263	0.200 ± 0.010	5.1	76	0.207 ± 0.026	12.7	79

n: number of replicates, NC: Nominal concentration (μg/mL), Mean CC: Mean Calculated Concentration, SD: standard deviation, CV, %: (SD/mean) × 100; Accuracy, %: (determined conc./nominal conc.) × 100

3.3.4. Recovery and matrix effect

The extraction recovery of the method was determined by two different spiked concentrations in each of the studied matrices: 5 and 2.5 μg/mL of AA and 10 and 5 μg/mL of UA in plasma and 5 and 10 μg/mL of AA and 10 and 20 μg/mL of UA in EBC and NL (Table 5). We could observe that blank samples had a concentration around 0.45 (AA) and 1.10 (UA) μg/mL for nasal lavage, and 1.79 (AA) and 12.51 (UA) μg/mL for plasma, however, no AA and UA were detectable (< LOD) for any of the blank EBC samples. When subtracted the blank values to the spiked calculated concentrations, we determined recovery. Recovery achieved ranged between 88 and 102 % for all the cases, except low level spiked NL samples, that reached values of 131 and 126 % for AA and UA, respectively.

Table 5.

Study of the recovery and matrix effect for each of the spiked matrices at two levels of concentration (low- and high-level)

Compound	Parameter	EBC		NL		Plasma
		Low	High	Low	High	Low	High
Ascorbic acid	Spiked NC	5	10	5	10	2.5	5
	Blank CC ± SD	n.d.		0.45 ± 0.02		1.79 ± 0.05
	Spiked CC ± SD	5.10 ± 0.05	9.25 ± 0.03	6.98 ± 0.07	10.03 ± 0.13	4.19 ± 0.07	6.67 ± 0.12
	Recovery (%)	102	92	131	96	96	98
	Matrix effect (%)	69	94	102	102	131	101
	CV (%)	0.9	0.31	1.4	1.3	1.1	1.4
Uric Acid	Spiked NC	10	20	10	20	5	10
	Blank CC ± SD	n.d.		1.10 ± 0.04		12.51 ± 0.29
	Spiked CC ± SD	9.96 ± 0.09	17.96 ± 0.06	19.47 ± 0.25	13.58 ± 0.13	17.00 ± 0.30	21.35 ± 1.20
	Recovery (%)	100	90	126	92	90	88
	Matrix effect (%)	102	87	131	91	132	119
	CV (%)	0.86	0.3	1.3	1.2	6.6	14

n.d.: not detected (< LOD), NC: Nominal concentration (μg/mL), CC: Calculated Concentration, SD: standard deviation,

Recovery (%): [Spiked CC-Blank CC]/Spiked NC ×100

Matrix effect (%): Ratio of peak area response [matrix/pure standard] × 100

CV (%): coefficient of variation from the recovery calculated ([SD/mean] × 100)

When matrix effect was studied values ranged from 69 to 132%, with some effects when low-level QCs tested (most QCs between 80–120%). Although there was an extent to the matrix affect for AA and UA in the present conditions, it did not influence the accurate determination of these antioxidants in the studied biological matrices and the ion suppression or enhancement from each matrix was negligible for this method. In all cases the variation was acceptable with values CV < 15%. No carry over was detected after injecting the higher concentrations for AA and UA.

3.3.5. Stability studies

Samples were stabilized with the addition of MPA before sample/QC storage and also the spike of antioxidant solution containing EDTA in methanol:water (1:1) before sample preparation. It can be affirmed that the sample preparation steps and the LC–MS determination must be performed as fast as possible. The validated method has 20 min analytical run and AA and UA elute before 4 min, requiring the additional time to clean the sample from matrix interferences a to stabilize the system.

Stability tests were performed by preparing spiked QC samples one day, storing them in the −80 °C for different times and by run samples after being stored inside the autosampler (AS) of the instrument for different amounts of time (Table 6). We could determine that keeping the samples inside the AS at +4 °C for 30 h did not influence the amount recovered of the spiked AA and UA in each of the matrices. Accuracy ranged between 75 – 125% for both antioxidants, except for spiked plasma samples with AA values at 129%. Variation was also stable with CV < 20%. Also, spiked samples stored at short term in the −80 °C freezer maintained the stability for 36 h, with accuracy around 80 and 120% for AA and UA in most cases. However, some stability problems were found when spiked plasma was stored for the longest times (36h). With these studies, we decided to prepare all samples within an interval of 1–2h and run them before 3 days after their preparation. Samples were always kept in the −80 °C and transferred into the instrument AS to run before 24 h.

Table 6.

Stability results for the determination of AA and UA in EBC, nasal lavage and plasma. Results are expressed by accuracy (%) from a spiked amount in a matrix and the coefficient of variation (CV, %) of n = 3 replicates for each sample.

Condition	Matrix	Time (h)	Ascorbic acid				Uric acid
			Spiked NC	Mean CC ± SD	Accuracy (%)	CV (%)	Spiked NC	Mean CC ± SD	Accuracy (%)	CV (%)
AS storage at +4 °C	EBC	2	5	4.60 ± 0.33	92	7.1	10	8.99 ± 0.66	90	7.3
		10	5	4.45 ± 0.29	89	6.5	10	8.64 ± 0.52	86	6.1
		20	5	4.37 ± 0.25	87	5.7	10	8.54 ± 0.49	85	5.7
		30	5	4.14 ± 0.33	83	7.9	10	8.11 ± 0.63	81	7.8
	NL	2	5	4.07 ± 0.20	81	4.8	10	10.06 ± 0.49	101	4.9
		10	5	4.04 ± 0.61	81	15.1	10	9.68 ± 1.73	97	17.8
		20	5	4.43 ± 0.19	89	4.4	10	9.08 ± 1.21	91	13.3
		30	5	4.09 ± 0.20	82	4.9	10	10.11 ± 0.50	101	4.9
	Plasma	02	2.5	2.43 ± 0.01	97	0.6	5	6.04 ± 0.41	121	5.8
		12	0.75	0.58 ± 0.24	76	8.8	1.83	2.32 ± 1.86	117	20.7
		15	2.5	2.29 ± 0.29	92	12.5	5	6.47 ± 0.10	129	1.5
		32	1.83	1.72 ± 1.84	93	10.4	7.9	9.37 ± 3.88	119	17.4
Freezer storage at −80 °C	EBC	0	5	4.43 ± 0.12	89	2.7	10	8.64 ± 0.22	86	2.6
		12	5	4.17 ± 0.35	83	8.4	10	8.16 ± 0.68	82	8.3
		24	5	4.66 ± 0.39	93	8.4	10	9.55 ± 1.83	96	19.2
		36	5	4.27 ± 0.37	85	8.6	10	8.37 ± 0.71	84	8.5
	NL	0	5	3.97 ± 0.27	79	6.9	10	9.82 ± 0.69	98	7.0
		12	5	4.06 ± 0.45	81	11.0	10	10.04 ± 1.13	100	11.2
		24	5	3.90 ± 0.38	78	9.8	10	8.41 ± 2.01	84	24.0
		36	5	4.14 ± 0.26	83	6.3	10	10.24 ± 0.66	102	6.5
	Plasma	0	5	4.93 ± 0.09	99	1.8	10	10.96 ± 2.10	110	19.2
		12	5	4.86 ± 0.01	97	0.2	10	11.91 ± 4.48	119	23.0
		36	15	14.03 ± 8.06	91	11.9	34.1	47.88 ± 17.66	135	-
QC stored at −80 °C	Solvent	0	6.15	5.61 ± 0.03	91	0.55	10.5	9.71 ± 0.02	92	0.20
		24	6.15	5.37 ± 0.03	87	0.53	10.5	9.90 ± 0.00	94	0.02
		5 days	6.15	5.04 ± 0.10	82	1.92	10.5	9.42 ± 0.04	90	0.38
		6 days	6.15	4.97 ± 0.09	81	1.90	10.5	9.52 ± 0.28	91	2.97
		10 days	6.15	4.78 ± 0.01	78	0.25	10.5	9.97 ± 0.04	95	0.42
		11 days	6.15	5.41 ± 0.21	88	3.87	10.5	9.63 ± 0.06	92	0.57
		12 days	6.15	5.42 ± 0.07	88	1.28	10.5	10.52 ± 0.11	100	1.00
		15 days	6.15	5.36 ± 0.05	87	0.92	10.5	9.71 ± 0.09	93	0.96

AS: AutosSampler, NC: Nominal concentration (μg/mL), Mean CC: Mean Calculated Concentration, SD: standard deviation

Accuracy (%): (determined conc./nominal conc.) × 100, CV (%): (SD/mean) × 100

Also, a longer-term stability study was prepared by store solvent QC spiked samples during max 15 days. Those QCs were prepared before any sample batch was processed and stored in the −80 °C. Then, two replicates were run (one before and one at the end of the sequence) with every batch of processed samples during the following two weeks. Those QCs showed a clear stability with high accuracy (< 78%) and low variation between replicates was also appreciated (< 5%).

3.4. Application of the method to different matrices

The validated method was finally applied for the quantification of AA and UA in three biological matrices: EBC, nasal lavage and plasma. The matrices were obtained for the same subjects (n = 31) which provided two samples from a double-blinded random antioxidant regime. Each participant took antioxidant supplements (antioxidant modulation, AM) and placebo two separate days during the study. The new method was applied to detect and quantify AA and UA in the matrices and to compare the effect of the antioxidant regime in the subject for the different matrices. Duplicate samples were prepared and mean concentrations (μg/mL in the matrix) and the standard deviations (SD) for all the subjects are listed in Supplementary Material (Table S1). Summarized data is presented in Table 7 with ranges of AA and UA concentrations found in each matrix and mean concentrations calculated by ‘Total’ of participants and by each group treated with different regime: antioxidant modulation (AM) and placebo (P).

Table 7.

Range and final concentrations (μg/mL in the matrix) of ascorbic and uric acids and the standard deviations (±SD) from the subjects collected during the all the study (Total) and the two regimes: antioxidant modulation (AM) and placebo (P). n is the number of values at higher concentrations than the LOQ defined in the method

Compound	Matrix	Range	Concentration ± SD (n)
			Total	AM	P
Ascorbic acid	EBC	0.03 – 1.25	0.14 ± 0.27 (28)	0.09 ± 0.12 (11)	0.06 ± 0.07 (14)
	NL	0.22 – 5.47	1.10 ± 1.00 (63)	1.14 ± 0.72 (30)	0.78 ± 0.44 (29)
	Plasma	6.77 – 32.68	20.33 ± 4.95 (63)	21.32 ± 4.85 (30)	19.05 ± 4.91 (29)
Uric Acid	EBC	0.03 – 3.08	0.18 ± 0.48 (61)	0.11 ± 0.20 (29)	0.09 ± 0.13 (28)
	NL	0.05 – 10.70	2.58 ± 2.10 (64)	2.81 ± 1.77 (30)	1.87 ± 1.13 (30)
	Plasma	16.76 – 81.32	47.83 ± 14.32 (63)	50.95 ± 14.04 (30)	43.97 ± 14.02 (29)

The values detected for both antioxidants were within the reference range in plasma reported in literature (10 – 90 μg/mL AA and 20 – 75 μg/mL UA [11, 51]). Low variations (SD) were calculated within replicates (Table S1). AA concentrations were found between 0.03 – 1.25 μg/mL for the EBC, 0.22 – 5.5 μg/mL for NL and 6.8 – 32.7 μg/mL for the plasma samples. For UA values ranged between 0.03 – 3.1 μg/mL for EBC, 0.05 – 10.7 μg/mL for NL and 16.8 – 81.3 μg/mL for plasma. When matrices are compared, and despite a higher dilution applied on plasma sample preparation, these samples contained much higher final concentrations for both antioxidants compared to NL and EBC. In fact, EBC and NL samples contain much lower amounts and are often below the limits of quantification of the method, especially in EBC. Nasal lavage AA and UA concentrations more easily detectable and quantifiable than in EBC. This is in accordance with the generalized idea of high dilution for the compounds present in EBC samples. Mean concentrations detected through all the participants (Table 7, ‘Total’) move from low values in EBC (0.14 and 0.18 μg/mL AA and UA, respectively) to slightly higher concentrations in NL (1.10 and 2.6 μg/mL). Plasma concentrations averaged around 20 μg/mL for AA and almost 50 μg/mL for UA.

When the amount of detected AA and UA for each subject was compared (Figure 5), we could observe the same profile throughout the samples within the same matrix. This means that concentrations for AA had similar variations through participants. Also, a good correlation between the different matrices could be observed, mainly for NL and plasma samples. The main limitation is the lack of EBC quantifiable values (<LOQ), primarily for ascorbic acid. EBC antioxidants concentrations showed different variations when compared to the other matrices.

Figure 5.

Comparison of amounts (μg/mL) of ascorbic acid (AA) and uric acid (BB) detected in the three matrices of the study: EBC (a), NL (b) and plasma (b). All participants are presented in X-axis and split between antioxidant modulation (AM) regime (left side and cercles) and placebo regime (right side and triangles)

Finally, when the regimes administered to the participants were compared both antioxidants showed a tendency to higher concentrations in the three matrices when AM (supplementation with fructose and Vitamin C) was administered (Fig. 5, circles). In Table 7 higher AA and UA concentrations in NL and plasma are clearly listed when the AM is applied, where AA goes from 0.8 (P) to 1.1 (AM) μg/mL in NL and from 19 to 21.3 μg/mL in plasma; and UA goes from 1.9 to 2.8 μg/mL in NL and from 44 to 51 μg/mL in plasma. However, these averaged concentrations had high standard deviations in all cases and a paired t-test at α = 0.05 showed statistical differences just for NL when detecting AA and NL and plasma for UA (Figure 6). AA in EBC was difficult to compare due to the high number of samples with concentrations below the LOQ. AA in plasma and UA in NL showed no statistical differences between regimes (p-value > 0.05). These results showed the effectiveness of the dietary antioxidant modulation in subjects that received a fructose and Vitamin C supplementation by showing an increasing of AA and UA concentrations UA in some of the matrices studied.

Figure 6.

Boxplots for ascorbic acid (AA) on NL (a), and uric acid (UA) for NL (b) and plasma (b) comparing antioxidant modulation (AM) and placebo (P) regimes.

4. Conclusions

We developed and validated a fast LC-MS-TOF method for the simultaneous detection and quantification of ascorbic and uric acid present in three different human biological matrices: EBC, nasal lavage and plasma. The sample preparation is fast and simple, and the run time allow the analysis of several numbers without degradation responses of the targeted compounds. The analytical method showed the ability to separate AA and UA form other interferences in the matrix with good resolution. The method offers good sensitivity, precision, and accuracy for the determination of both antioxidants with low limit of quantification (0.025 μg/mL).

This method has been successfully applied to real samples, and reliable results were obtained for plasma and NL samples. However, AA and UA detection in EBC samples still have some limitations due to the high diluted presence of these compounds in these breath related matrices. Some further studies aimed to enhance the pre-concentration procedure for these samples are still necessary.