A novel method for the simultaneous determination of

This article was downloaded by: [FDA Library] On: 19 December 2014, At: 12:45 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Food Additives & Contaminants: Part A Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tfac20

A novel method for the simultaneous determination of 14 sweeteners of regulatory interest using UHPLC-MS/ MS a

a

a

a

Romina Shah , Samantha Farris , Lowri S. De Jager & Timothy H. Begley a

Food and Drug Administration, Center for Food Safety and Applied Nutrition, College Park, MD, USA Accepted author version posted online: 06 Dec 2014.Published online: 17 Dec 2014.

Click for updates To cite this article: Romina Shah, Samantha Farris, Lowri S. De Jager & Timothy H. Begley (2014): A novel method for the simultaneous determination of 14 sweeteners of regulatory interest using UHPLC-MS/MS, Food Additives & Contaminants: Part A, DOI: 10.1080/19440049.2014.994111 To link to this article: http://dx.doi.org/10.1080/19440049.2014.994111

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

Food Additives & Contaminants: Part A, 2014 http://dx.doi.org/10.1080/19440049.2014.994111

A novel method for the simultaneous determination of 14 sweeteners of regulatory interest using UHPLC-MS/MS Romina Shah*, Samantha Farris, Lowri S. De Jager and Timothy H. Begley Food and Drug Administration, Center for Food Safety and Applied Nutrition, College Park, MD, USA

Downloaded by [FDA Library] at 12:45 19 December 2014

(Received 23 September 2014; accepted 28 November 2014) An improved, efficient, sensitive method for the determination of 14 non-nutritive sweeteners in food products was developed using electrospray ionisation (ESI) ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) in the negative-ion mode. Fourteen sweeteners and three internal standards were separated on a reversed-phase UHPLC column using a simple gradient programme. Analyte quantitation and confirmation were performed with data collection in multiple reaction monitoring (MRM) mode. Limits of detection (LODs) were determined in a representative drink, candy and yogurt sample and ranged from 0.1 to 1.8 ng ml–1 (drinks) and from 0.1 to 2.5 ng g–1 (candy and yogurt). Repeatability at the limit of quantitation (LOQ) ranged from 1% to 13% relative standard deviation (RSD). Twenty-seven commercially available food products were tested using the optimised method showing that the majority of products contained sweetener concentrations below their assigned maximum usable dose. Recovery studies were performed and accuracy data are presented. Keywords: non-nutritive sweeteners; UHPLC-MS/MS; foods; challenging target analytes

Introduction Non-nutritive sweeteners are commonly used in foods as alternatives to sugar to provide sweet taste with little or no calories (Zygler et al. 2012). They are an important class of food additives added to foods to cause a technical effect such as sweetening (Wasik et al. 2007). The list of allowable sweeteners varies among nations worldwide (Yang & Chen 2009). For example, cyclamate (CYC) and neohesperidine dihydrochalcone (NHDC) are not approved for use as food additives in the United States but are authorised in the European Union (Lim et al. 2013). Sweeteners are often used in combination to enhance sweetness and limit undesirable aftertastes (Zygler et al. 2009). Food products containing sweeteners are heavily promoted as beneficial for the treatment of obesity and management of diabetes (Zygler et al. 2009). Sweeteners can be found in a large number of food products including carbonated and non-carbonated beverages, hard candies and dairy products such as yogurt (Zygler et al. 2011). There is considerable controversy surrounding the adverse health effects of low-calorie sweeteners. Consumers worldwide have reported side-effects linked to sweetener consumption including mood and behavioural changes, skin irritations, headaches, allergies, respiratory difficulties and cancer (Zygler et al. 2009). As such, it is important to monitor and control the concentration of sweeteners in foods to ensure compliance with countryspecific regulations. The European Union limits the amount

of sweeteners added to food and sets a maximum usable dose (MUD) for specific food commodities (Zygler et al. 2011). In order to ensure that products are in compliance with regulations, it is necessary to have a reliable, robust and quantitative method for the simultaneous determination of several commonly used sweeteners in a single analysis. Although several analytical methods for the determination of artificial sweeteners have been published, none is appropriate for routine regulatory analyses. The USFDA currently uses the AOAC Official Method #969.27, a thin layer chromatography method for the determination of some non-nutritive sweeteners in food samples (AOAC 2012). This method lacks specificity and is limited to the qualitative determination of a select few sweeteners. In addition, this method lacks confirmation criteria compatible with today’s standards. An ion chromatography (IC) method with suppressed conductivity detection has been developed and used within the agency, but it lacks selectivity in certain matrices such as those that contain citric acid. The authors report significant interference from a very large citric acid peak in this anion-exchange separation which can adversely impact target analyte determinations (Hawk et al. 2007). Furthermore, the scope of the method is narrow and does not incorporate all sweeteners of regulatory interest (Hawk et al. 2007). Various other methods for the determination of nonnutritive sweeteners have been reported in the literature. The most common multi-sweetener methods involve

*Corresponding author. Email: [email protected] This work was authored as part of the Contributor’s official duties as an Employee of the United States Government and is therefore a work of the United States Government. In accordance with 17 U.S.C. 105, no copyright protection is available for such works under U.S. Law.


2

R. Shah et al.

HPLC with various types of detection (Zygler et al. 2011). An HPLC-UV method is reported for the determination of CYC, saccharin (SAC) and aspartame (ASP), which does not require derivatisation. However, in order to achieve baseline resolution of CYC and SAC, the pH of the phosphate buffer mobile phase needs to be maintained at 2.5, which could severely compromise the integrity of a reversedphase column (Croitoru et al. 2011). Furthermore, many foods and beverages contain UV-active species that could interfere with the analysis if chromatographic separation was not achieved. HPLC with evaporative light scattering detection (ELSD) has been published for the determination of nine sweeteners (Wasik et al. 2007). This type of detection technique may lack selectivity for target analytes, especially among interferences in the matrix. A multi-sweetener method using HPLC with mass spectrometric (MS) detection has been reported; however there were disadvantages with this method (Zygler et al. 2011). Dulcin (DUL) could not be directly detected and was determined as a formic acid adduct. In addition, three commonly used sweeteners acesulfame potassium (ACS-K), SAC and sucralose (SCL) gave nonlinear responses in the tested calibration range. An HPLC-MS method has been published for the simultaneous determination of seven non-nutritive sweeteners in foods using single ion monitoring (SIM) (Yang & Chen 2009). Nine sweeteners were determined in foods using HPLC-ESI/MS/MS in MRM with improved MS confirmation data (Lim et al. 2013). But recovery studies were performed at spiking concentrations significantly below what would be expected in real samples, limiting the value of the accuracy data. The current paper describes an improved, efficient UHPLC-MS/MS method to determine simultaneously 14 sweeteners of regulatory interest in foods and beverages in a single analysis. Target analytes include CYC, SAC, SCL, DUL, ASP, NHDC, ACS-K, alitame (ALI), neotame (NEO), rebaudioside A (REB A), stevioside (STV) and the three sugar alcohols, erythritol (ERY), xylitol (XYL) and maltitol (MAL). This method allows quantitation and MRM confirmation of all target analytes using three isotopically labelled internal standards. Tested matrices included carbonated and non-carbonated beverages, hard candies and yogurt samples. Yogurts were processed using a solid phase extraction (SPE) method but minimal sample clean-up was required to analyse beverages and hard candies. Negative-ion UHPLC-ESI-MS/MS analyses enabled the quantitation and confirmation of 14 commonly used sweeteners in 27 commercially available products.

Materials and methods Reagents and column Sweetener standards were obtained from various sources. Saccharin (≥ 99%), xylitol (≥ 99%), meso-erythritol

(≥ 99%), sucralose (≥ 98%), maltitol (≥ 98%), neohesperidine dihydrochalcone (≥ 95%) and acesulfame-K (≥ 99%) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Cyclamate and aspartame standards were obtained from Supelco (Bellefonte, PA, USA). USP reference standards were purchased for alitame, neotame and rebaudioside A with use at 0.878, 0.954 and 0.975 mg, respectively. Dulcin standard was obtained from Wako Chemicals USA Inc., (Richmond, VA, USA). The reference standard for stevioside (99.3% purity) was purchased from ChromaDex (Irvine, CA, USA). The deuterium-labelled internal standards d-sorbitol-1- 13 C (99%) was purchased from Sigma-Aldrich and sodium cyclamate d-11 (≥ 99%) and saccharin d-4 (≥ 99%) were purchased from C/D/N isotopes Inc. (Pointe-Claire, QC, Canada). All sweetener stock standards were prepared in 100% H2O with the exception of aspartame which was dissolved in MeOH/ H2O (80/20, v/v), due to low solubility in H2O. All solvents were Optima LC-MS grade from Fisher Scientific (Pittsburgh, PA, USA). Formic acid (98% purity) was purchased from Fluka (Milwaukee, WI, USA). Ammonium acetate (NH4OAc, purity 98%) was obtained from Sigma-Aldrich. N,N-diisopropylethylamine (DIPEA, purity ≥ 99%) was purchased from Fisher Scientific. Chromatographic separations were performed on a Waters (Milford, MA, USA) Acquity UPLC BEH C18 analytical column (1.7 µm, 2.1 × 100 mm) with a Vanguard pre-column (1.7 µm, 2.1 × 5 mm). All samples were filtered using Titan 3, 17 mm, 0.20 µm PTFE membrane syringe filters (Rockwood, TN, USA). Instrumentation An Applied Biosystems Sciex (Foster city, CA, USA) 4000 Q-trap LC-MS/MS system interfaced with an Agilent (Agilent Technologies, Santa Clara, CA, USA) 1290 series UHPLC was used in all experiments. Nitrogen gas was delivered to the LC-MS system by an in-house nitrogen system. Data acquisition and instrument control were accomplished using Analyst software version 1.5.2. A Fisher Scientific Marathon 21000R centrifuge, Branson 3510 sonicator (VWR International, Radnor, PA, USA) and Glas-col digital vortex shaker (Wilmad Labglass, Vineland, NJ, USA) were used during the sample preparation. SPE Chromabond ® C18ec 3 ml cartridges packed with 500 mg sorbent bed were purchased from Macherey-Nagel Inc. (Bethlehem, PA, USA). Standard solution preparation Analytical stock solutions were prepared by weighing approximately 10 mg of analyte and dissolving in a 10 ml volumetric flask made up to volume with 100% H2O or MeOH/H2O (80/20, v/v). Working stock solutions were made by further diluting the stock solutions 10-fold with 100% H2O. Stock solutions and working stock

Food Additives & Contaminants: Part A solutions were stored at 4°C in amber 10 ml screw cap vials with PTFE septa in the cap closure, when not in use. Stock solutions were used for up to 6 months and working solutions for up to 7 days. Mixed calibration standards were made using working stock solutions of the individual sweeteners. Calibration standards were prepared fresh daily and contained all 14 analytes of interest and three internal standards. Autosampler vials were amber 1.5 ml screw cap with PTFE septa in the cap closure. Quantitation was based on using an external calibration curve that was generated as the ratio peak area of analytes versus peak area of internal standards. The equation of the calibration curve was obtained using least squares regression.


Analytical method Gradient separations were performed within 30 min at a flow rate of 0.25 ml min−1 and column temperature of 40ºC (Table 1). The injection volume was 5 µl. Mobile phase A consisted of 10 mM NH4OAc in H2O/MeOH (98/2, v/v) and mobile phase B was 10 mM NH4OAc in H2O/MeOH (1/99, v/v). After MS optimisation, the method was run with a turbo heater temperature of 400ºC and an ion spray voltage of −4000 V for all analytes. The nitrogen flow rate for the curtain, nebulising and desolvation gases was set to 20, 60 and 60 ml min−1 with a medium strength setting for the collision gas. The collision cell exit potential was −15 V with an entrance potential set to −10 V for all analytes. Samples tested All food products tested listed specific added sweeteners on their labels. Four types of hard candy products were obtained from local grocery stores. All candies 1–4 were individually wrapped as single-serving packages. A total of 15 commercially available packed drink products were obtained for testing. These included non-carbonated drinks (drinks 1–6) and carbonated drinks (drinks 7–15). In addition, eight yogurt products were purchased from local grocery stores. Table 1. Gradient elution timetable used for the UHPLC-MS/MS method. Time (min) 0 2 7 15 25 25.1 30

% MP A

% MP B

100 100 50 0 0 100 100

0 0 50 100 100 0 0

Note: MP A = 98:2 10 mM NH4OAC:MeOH; MP B = 1:99 10 mM NH4OAC:MeOH.

3

Sample preparation Individual hard candies were accurately weighed between approximately 3 and 6 g and dissolved in 10 ml H2O. Samples were shaken using the digital vortex shaker for 30 min at 500 rpm and then diluted with H2O between 2-and 50-fold in order to obtain analyte concentrations within the linear range of the calibration curve (0.025– 5.0 µg g−1) followed by the addition of the three internal standards. This procedure produced complete dissolution of the candy samples, resulting in transparent solutions with no visual insoluble material remaining after shaking. Drink samples were diluted with H2O and filtered through a PTFE syringe filter directly into HPLC autosampler vials. Candy and drink samples were analysed in triplicate, and in cases where the product was packaged in individual servings (candy) three separate packages were analysed. Carbonated beverages were sonicated for 5 min to remove dissolved gases. Degassed beverages, along with non-carbonated beverages, were diluted between 10- and 500-fold with H2O for quantitation followed by the addition of the three internal standards. A certified reference material SRM-3282 low-calorie cranberry juice cocktail obtained from the National Institute of Standards and Technology (NIST) was diluted 50-fold and analysed in this study. Yogurt samples of approximately 5 g were accurately weighed into centrifuge tubes. A 50 ml volume of 0.075% formic acid adjusted to pH 4.5 by the addition of approximately 3 ml of DIPEA (extraction buffer) was added to the centrifuge tubes. The tubes were shaken using the digital vortex shaker for 30 min at 500 rpm. Following centrifugation at 4ºC for 10 min at 4000 rcf, a 5 ml aliquot of the resultant supernatant was subjected to further clean-up by SPE. A 3 ml C18ec cartridge was conditioned with 1.5 ml of MeOH and 3 ml of extraction buffer. Following this, 2.5 ml of sample supernatant were loaded onto the cartridge and the cartridge was then washed with 1.5 ml of extraction buffer. Analytes were eluted from the cartridge with 1 ml of MeOH into a 5 ml volumetric flask. A second elution with 1 ml of MeOH was done after 10 min had elapsed in order to enable column equilibration and improved recoveries. In the final step, the eluate containing 5 ml flask was brought up to volume with H2O. It was very important to prevent the cartridge from drying out during the course of the SPE procedure.

Method validation In this study, the developed method was validated using USFDA guidelines for single laboratory validation. The linear calibration range was experimentally determined along with detection and quantitation limits for each target analyte. Method precision and accuracy were determined by conducting repeatability and recovery experiments. Both intra- and inter-day results are presented.

4

R. Shah et al.

Results and discussion


Method development Separation Non-nutritive sweeteners are a class of compounds that have significantly different physical and chemical properties. This makes it very challenging to develop a single method for their separation and isolation from matrix interferences. Most have poor chromophoric properties and detection by HPLC-UV lacks specificity, especially in real matrices. Additionally, the analytes in this study encompass a wide range of polarities and molecular size with very different pKa values that makes chromatographic separation difficult. For example, ERY is a very small highly polar compound compared with REB A which is considerably larger and relatively more hydrophobic (Figure 1). Several analytical columns were tested in the current study in order to achieve a single separation of the target analytes. Hydrophilic interaction chromatography (HILIC) was attempted using an amino phase column. This resulted in some sweeteners of regulatory importance such as ASP and SCL to be unretained and eluting in the column void volume. Two mixed-mode columns were tested; however, the pH required for retention of all analytes (pH about 9) was outside the stability range for these columns. Reversed-phase separations have been previously reported for the separation of sweeteners in

favour of HILIC. At pH values needed for separation of analytes by HILIC, significant peak tailing and broadening were seen with some sweeteners (Ordóñez et al. 2012). This was also the case in the current investigation and thus a reversed-phase separation mechanism was chosen. Furthermore, reversed-phase separations are commonly used in food analysis and can easily be applied to routine regulatory analyses. Several types of reversed-phase columns including C8 and C30 were tested during the current investigation for a single analytical separation of these different analytes. Some of these columns had polar embedded functional groups including the Dionex Acclaim PA, Agilent Zorbax SB-Aq and Phenomenex Synergi Fusion-RP columns. These were thoroughly tested in an attempt to increase retention of the highly polar sugar alcohols. These proved unsuccessful for peak shape and retention of these sweeteners. After several columns were tested, all 14 sweeteners were readily separated by an Acquity UPLC BEH C18 analytical column with a mixture of aqueous and organic solvents as the mobile phase. UHPLC provides an improved separation compared with HPLC since analytes elute faster, with increased sensitivity and efficiency due to the smaller particle size of the stationary phase in the analytical column. Acetonitrile was attempted as an organic modifier, but MeOH provided better chromatographic peak shape and was used for all separations.

Figure 1. Chemical structures of the non-nutritive sweeteners of varying molecular size and polarities. ERY, erythritol; XYL, xylitol; CYC, cyclamate; DUL, dulcin; SAC, saccharin; ACS-K, acesulfame potassium; ASP, aspartame; ALI, alitame; MAL, maltitol; NEO, neotame; SCL, sucralose; NHDC, neohesperidine dihydrochalcone; STV, stevioside; REB A, rebaudioside A; MW, molecular weight.


Food Additives & Contaminants: Part A

5

Figure 2. (colour online) MRM chromatogram of a 1000 µg ml−1 standard solution of 14 sweeteners and three internal standards. The overlay of the chromatograms represents the most intense MRM transition (quantitation) and the second transition (confirmation). Complete experimental conditions are listed in the Analytical method section.

Other LC conditions, including mobile phase composition, flow rate and gradient, were specifically optimised. Separation of the 14 analytes of interest was accomplished within a 30-min chromatographic run (Figure 2). Target analytes eluted within 13 min, followed by a 10-min high organic wash and 5-min equilibration time. The extended high organic wash was necessary to prevent carryover of sweeteners between injections. Additionally, it was critical to maintain the column temperature at 40ºC in order to minimise increased back pressure on the UHPLC system from the MeOH gradient (Aburjai et al. 2011). Under these conditions, retention times proved very consistent and stable enabling use of a scheduled MRM method.

used for identity confirmation of sweeteners in standards and samples. Internal guidelines require the signal-tonoise ratio for analytes detected in a sample to be ˃ 10:1 for the quantitative transition and ˃ 3:1 for the confirmatory transition for accurate quantification and quantitation/ confirmatory ion ratio in samples must match within ±10% (absolute) of the ion ratio found from standards. Retention times for analytes detected in a sample must be within ±5% of the retention times for analytes in standards (US Food and Drug Administration 2003). The two MRM transitions from precursor to product ions for all sweeteners and internal standards are listed in Table 2 with corresponding transition specific MS conditions and retention times.

Mass spectrometry

Selection of internal standards

MS parameters were optimised using flow injection analysis (FIA) and ESI in negative mode. Signal response for the sweeteners in negative mode was significantly higher than in positive mode with no detectible adduct formation, which is consistent with a previous study (Yang & Chen 2009). The current method generated two structurally significant MS product ions for each sweetener and all three internal standards allowing for more selectivity and confirmation of target analytes, while providing an important advantage over previously reported methods (Yang & Chen 2009; Zygler et al. 2011). The most intense MRM transition was used for quantitation and the second for confirmation. Ion ratios of the two MS transitions were

Generally, it is important to have internal standards for quantitation to account for possible ion suppression from matrix interferences in the complex composition of foods (Yang & Chen 2009). Although it is ideal to have labelled standards for each compound being analysed, these are sometimes unavailable and cost prohibitive. Therefore, we chose three compounds to serve as internal standards for the 14 sweeteners in this study. These had similar chemical and physical properties to the target analytes and were readily available. Saccharin-d4 is used to quantitate SAC, ACS-K, ALI, NEO, REB A and STV. Sodium cyclamate-d11 served as the internal standard for CYC, NHDC, ASP, DUL

6

R. Shah et al.


Table 2. UHPLC-MS/MS parameters for sweeteners and internal standards. Sweetener

Retention time (min)

Precursor ions (m/z)

ACS-K ALI ASP CYC CYC-d11 DUL ERY MAL NHDC NEO REB A SAC SAC-d4 Sor-13 C SCL STV XYL

3.0 8.7 7.8 6.0 5.9 8.0 1.1 1.1 9.4 11.4 12.1 5.0 5.1 1.1 7.2 12.1 1.1

162.0 330.1 293.0 178.1 189.1 179.1 121.0 343.1 611.2 377.1 965.4 182.0 186.0 182.1 395.0, 397.0 803.3 151.0

Table 3.

Product ions (m/z) 82.1, 311.9, 260.8, 80.0, 80.0, 150.0, 89.1, 179.0, 303.1, 200.1, 803.4, 106.0, 106.0, 89.1, 359.1, 641.3, 89.1,

Collision energy (V) −20, −20, −16, −35, −35, −16, −14, −20, −50, −20, −38, −25, −25, −20, −16, −40, −16,

77.7 294.8 200.1 96.0 97.0 107.0 101.1 89.1 124.8 345.0 317.2 62.0 62.0 101.1 361.0 317.2 100.7

Declustering potential (V)

−30 −25 −20 −30 −30 −29 −12 −30 −60 −20 −96 −31 −35 −20 −18 −70 −20

−65 −70 −60 −95 −95 −55 −50 −65 −120 −80 −150 −70 −70 −60 −75 −150 −60

Method limit of detection (LOD) for sweeteners in foods using UHPLC-MS/MS in MRM mode.a

LOD Drink (ng ml–1 ) Candy (ng g–1) Yogurt (ng g–1)

ACS-K

ALI

ASP

CYC

DUL

ERY

MAL

NHDC

NEO

REB A

SAC

SCL

STV

XYL

0.62 0.55 1.3

1.5 1.8 n.d.

0.73 0.66 1.0

0.1 0.1 n.d.

1.8 1.9 n.d.

0.45 0.42 n.d.

1.1 0.90 n.d.

0.98 0.85 n.d.

1.1 0.99 n.d.

1.2 1.4 2.5

1.5 1.9 n.d.

1.4 1.7 1.9

1.2 1.1 n.d.

0.51 0.31 n.d.

Note: aLODs were estimated at three times the S/N ratio (peak to peak) from concentrations within 10 times the LOD.

and SCL; with D-sorbitol-1- 13 C as the internal standard for the three sugar alcohols ERY, XYL and MAL.

Method validation Method performance Calibration curves were constructed using the ratio of peak areas analytes versus internal standards obtained by MRM analysis of multi-analyte standard solutions. During sample analysis, five-point calibration curves were produced daily using weekly prepared standards (0.025– 2.5 µg ml−1). The curves using least squares regression produced good linearity (R2 > 0.99) over the concentration range 0.025–2.5 µg ml−1. Matrix-matched calibration curves were also produced using fortified food samples with no detectable sweeteners. These included a beverage, candy and yogurt fortified to concentrations ranging between 0.025 and 2.5 µg ml−1. Regression lines were linear over this concentration range (R2 > 0.99). Comparison between the calibration data from the fortified samples and standard solutions indicated that analyte accuracy was within the experimental error. Therefore, calibration curves using standard solutions were used for quantitation in samples. The LOD in representative matrices for each analyte was determined using

the standard graphical method (Table 3) (Namiesnik 2009). Three replicate standard solutions were prepared at three different concentrations bracketing the estimated LOD for each analyte. For example, if the estimated LOD for a compound was 0.025 µg ml−1, the three standard solutions prepared were 0.02, 0.025 and 0.03 µg ml−1. Six injections were performed for each of the three concentrations. All values were plotted as concentration on the x-axis and ratio of peak area analyte versus internal standard on the y-axis. A linear least squares line was fitted to the points and the slope determined. The LOD was calculated as three times the average standard deviation of all 18 measurements divided by the slope of the line. The LOQ was calculated as 10 times the average standard deviation of all 18 measurements divided by the slope of the line. LODs ranged from 0.1 to 1.8 ng ml−1 in drinks, from 0.1 to 1.9 ng g−1 hard candies and from 1.0 to 2.5 ng g−1 in yogurt. LOD determinations for the yogurt samples were made specifically for the analytes most likely to be present within these dairy products. Overall, these concentrations are significantly lower than would be expected in commercial products containing sweeteners. Instrumental method precision was evaluated using standards at the LOQ concentrations. Analysis of these standards on three consecutive days resulted in %RSD

Food Additives & Contaminants: Part A values ranging from 1 to 13. Additionally, five-point multi-day calibration curves were constructed over the linear concentration range in order to confirm the accuracy and precision of the calibration data. The %RSD values ranged between 2% and 5% (n = 3), which supports the data obtained at LOQ concentrations.


Food analysis The main goal of this method is to be able to determine whether or not a sample contains sweeteners as well as to authenticate the presence and concentrations of these analytes in various foods. Drinks, candies and yogurts were chosen because sweeteners are widely used in these commonly consumed products (Gardner et al. 2012). Recovery studies were performed in representative foods to test the method accuracy and precision. Finally, a total of 27 products were analysed using the final method. Due to the high concentrations of the sweeteners in food products, the prepared samples needed to be significantly diluted to bring the analyte concentrations within the linear dynamic range of the method. Recovery studies One of the biggest challenges in food analysis is the effect of matrix composition on the performance of the analytical method. In order to determine method accuracy and selectivity, a representative from each food commodity containing no target analytes was fortified with known amounts of sweeteners. The sweeteners chosen for spiking experiments encompassed the most polar, intermediate and most non-polar compounds in this study. Sugar alcohols were not spiked in the yogurt matrix because using the current SPE procedure it is not possible to isolate these analytes from these samples due to their small size and very polar nature. This is of no concern as sugar alcohols are not expected to be found in yogurts. Each representative of a food product was fortified in triplicate at three different concentrations, 10%, 50% and 100%, of MUD values specific for beverages and analysed with an unfortified sample. Some of the sweeteners do not have assigned MUD values and therefore spiking concentrations were chosen as follows. Sugar alcohols have been given quantum satis meaning harmless enough to have no specific quantity restriction (Mortensen 2006). Thus, 10%, 50% and 100% of 20 µg ml−1 were assigned as the spiking concentrations. In the case of REB A, a spiking concentration of 65 µg ml−1 was chosen as it is the average concentration of this analyte found in drinks in a previous study conducted in our laboratory (Shah et al. 2012). Since CYC is banned for use as a food additive in the United States an average concentration of 250 µg ml−1 found in foods worldwide was chosen for spiking purposes (World Health Organization 2009). Recoveries

7

Table 4. Results of accuracy and precision from recovery experiments of the sweeteners at 10%, 50% and 100% MUD concentrations in a representative drink, candy and yogurt matrix by UHPLC-MS/MS (n = 3). MUD Drink ACS-K ASP ERY SCL REB A Candy ACS-K ASP ERY SCL REB A Yogurt ACS-K ASP ERYa SCL REB A

10% MUD

50% MUD

100% MUD

350 600 20 300 65

97 93 77 106 94

(5.9) (2.9) (2.2) (5.3) (2.7)

95 101 74 114 103

(1.7) (1.5) (9.4) (3.4) (3.2)

86 106 70 90 108

(1.0) (3.4) (2.8) (7.5) (13)

350 600 20 300 65

98 78 95 72 76

(8.1) (2.4) (10) (6.5) (7.5)

85 96 70 84 88

(3.1) (15) (4.2) (4.4) (11)

85 (3.1) 77(2.1) 75 (2.9) 93 (6.6) 102 (2.0)

350 600 20 300 65

81 (4.2) 98 (1.2) n.d. 83 (3.8) 86 (4.2)

80 (1.7) 100 (3.9) n.d. 83 (13) 93 (11)

82 (4.7) 90 (13) n.d. 84 (3.9) 89 (1.4)

Notes: aSpike and recovery testing for ERY was not determined (n.d.) in yogurt. Values in italics represent the %RSD of three replicate samples.

ranged from 70% to 114% with %RSD values ranging from 1 to 15, as listed in Table 4. Inter-day precision was tested by analysing one replicate of each fortified product at the three spike concentrations over three consecutive days. These %RSD values ranged from 2% to 12%. These recovery data and %RSD values indicate that good method accuracy and precision are achieved. This demonstrates that for the majority of samples, matrix effects are minimal for quantitation of target analytes due to the use of internal standards and the significant dilution needed for quantitation in the linear range.

Drinks Fifteen commercially available drink products labelled as ‘low calorie’ were analysed using the optimised method. Matrix components did not interfere with the quantitation of analytes, as previously seen from drink analyses in our laboratory (Shah et al. 2012). The data from the drink analysis show significant differences in the sweeteners detected in drinks, with concentrations ranging from 3 to 2800 µg ml−1 (Table 5). With the exception of SAC in drinks 6 and 13, all sweeteners were detected below their assigned MUD concentrations in beverages. Drinks 1–3 contained very high concentrations of ERY in combination with one or two other sweeteners. This is an interesting finding because ERY has no assigned MUD concentration in any food category or specific quantity restriction (quantum satis) as governed by the European Union (Mortensen 2006). ERY

n.d. n.d. n.d. n.d. 56 (13) 414 (11) 148 (2.8) 30 (15) 356 (6.5) 83 (9.4) 180 (1.5) 191 (3.5) n.d. n.d. 94 (3.8) 0.38 (5.8) n.d. n.d. n.d.

ASP 2800 (4.0) 1044 (14) 1643 (14) n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

ERY n.d. n.d. 3 (4.4) n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

NHDC n.d. 26 (3.5) 105 (2.3) n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

REB A

Note: Values in italics represent the %RSD of three replicates from one bottle or can (drinks), and three replicates of individually wrapped candy samples.

48 (4.2) n.d. n.d. 20 (2.0) 48 (6.9) n.d. 76 (1.0) n.d. n.d. 51 (5.3) 52 (4.0) 69 (7.8) n.d. 49 (2.2) n.d. n.d. n.d. 21 (6.6) n.d.

Drink 1 Drink 2 Drink 3 Drink 4 Drink 5 Drink 6 Drink 7 Drink 8 Drink 9 Drink 10 Drink 11 Drink 12 Drink 13 Drink 14 Drink 15 Candy 1 (µg Candy 2 (µg Candy 3 (µg Candy 4 (µg

g–1) g–1) g–1) g–1)

ACS-K

(µg ml–1 )

n.d. n.d. n.d. n.d. n.d. n.d. n.d. 156 (5.5) n.d. n.d. n.d. n.d. n.d. n.d. 87 (4.7) n.d. n.d. n.d. n.d.

SAC

Table 5. Results from the analysis of 15 drinks and four hard candies carried out in triplicate; quantification and confirmation was done in MRM mode.


148 (5.9) n.d. n.d. 21 (4.3) 70 (14) n.d. n.d. n.d. n.d. 58 (8.7) n.d. n.d. 72 (2.1) 22 (2.1) n.d. 59 (5.0) 347 (17) 255 (5.1) 101 (1.4)

SCL

8 R. Shah et al.



9

Figure 3. (colour online) MRM chromatogram of a 1/50-fold dilution of drink 15. Complete experimental conditions are listed in the Analytical method section. The overlay of the chromatograms represents the most intense MRM transition (quantitation) and the second transition (confirmation).

is classified as a ‘bulk’ sweetener as it is less intense compared with sucrose in sweetening capacity and is often added to foods to mask the bitter aftertastes of other sweeteners (Mortensen 2006). For all drinks one to three sweeteners were found (Figure 3). Examination of the data shows that seven commonly used sweeteners were found in the drink products. Seven other sweeteners in this study were not detected in any of the drinks. The three most prevalent sweeteners found in the drink products analysed in this study were ACS-K, ASP and SCL. There are considerable differences in the concentrations of sweeteners in the drinks. Beverage manufacturing processes may contribute to these variations. Differences are most likely due to the varying sweetening strengths of these compounds relative to sucrose. Therefore, various amounts of sweeteners are added to produce the desired sweetening effect (Mortensen 2006). Also, there are significant differences in chemical properties among these compounds such as solubility and thermal stability (Mortensen 2006). As such some sweeteners are better functioning in certain food types while others are best suited to use in drinks. For the most part the sweeteners found in the current study matched what was declared on the drink label as an ingredient. However, in drink 3 we found a low concentration of NHDC, a sweetener not approved for use as a food additive in the United States. The ingredient label on this drink did not list NHDC. Drink 12 listed ASP in the

ingredient list on the label, but none was detected upon analyses. CYC is banned for use in the United States as a food additive and was not found in any of the drinks. In order to assess the between-package variability of the products, one drink product was run in triplicate using three separate bottles. The between-package concentrations found for ACS-K and SCL were very consistent and reproducible with reported values of 17 (±1.1) and 139 (±2.5) µg ml−1. The within-package %RSDs ranged from 1 to 15, demonstrating good method precision. In order to confirm further that our method was valid and accurate for its designed purpose, we obtained and analysed a standard reference material (SRM) from NIST. Although there were no certified values for any sweeteners assigned to this material, the sample (SRM-3282) was labelled as a low-calorie cranberry juice cocktail. This implied a high likelihood that the sample would contain one or more sweeteners in our study. Upon triplicate analyses, ACS-K and SCL were found in the SRM at concentrations of 68 and 90 µg ml−1 , with %RSDs values of 10 and 4.6, respectively.

Hard candies A total of four hard candy products were tested. All candies were analysed in three replicates of individually wrapped candy. SCL was found in all the candy products at

10

R. Shah et al.

concentrations ranging between 59 and 347 µg g−1. The SCL concentrations found in the candy samples were higher than the drinks. Candies 1 and 3 contained one other sweetener present in combination with SCL, but candies 2 and 4 contained only SCL. In candy 2 the SCL concentration exceeded its assigned MUD concentration. The relative concentrations of the sweeteners varied significantly within both drink and candy samples. Because of this it was necessary to analyse several dilution concentrations to quantitate all the sweeteners. The sweetener concentrations found in each of the four candy samples tested are listed in Table 5.


Yogurts There are many components in food matrices that have similar polarities to the target analytes, most of which are water soluble, except DUL and NHDC. Therefore, it is very difficult to isolate sweeteners from foods. Yogurts represent a much more complex mixture of ingredients than beverages or hard candies, thus requiring a thorough sample clean-up prior to UHPLC-MS/MS analyses (Huvaere et al. 2012). This ensures better long-term performance of the instrument and minimises ion suppression effects. A previous SPE method was modified and optimised for our purposes (Zygler et al. 2010). Several SPE parameters were tested, including sorbent phase type, cartridge size, sample load volume and extraction buffer. The most critical factor affecting analyte recoveries was the composition of the extraction buffer, as seen previously (Zygler et al.

Table 6. Average concentrations (µg g–1) of sweeteners in eight yogurt samples (n = 3). (µg g–1) Yogurt Yogurt Yogurt Yogurt Yogurt Yogurt Yogurt Yogurt

1 2 3 4 5 6 7 8

ACS-K

ASP

SCL

REB A

n.d. n.d. 195 (4.9) n.d. 72 (3.6) 89 (4.9) 145 (5.8) 128 (2.1)

n.d. 274 (7.9) n.d. n.d. 231 (3.0) n.d. n.d. n.d.

n.d. n.d. 55 (3.8) n.d. n.d. 107 (4.4) 174 (5.3) 183 (3.9)

29 (8.9) n.d. n.d. 24 (6.9) n.d. n.d. n.d. n.d.

Note: Values in italics represent the %RSD of three replicate weightings.

2010). The use of formic acid and DIPEA at pH 4.5 yielded the best recoveries (Table 4) for the sweeteners from yogurts. DIPEA is an ion-pairing agent that allows for improved recoveries compared with triethylamine (TEA) as it enables a stronger hydrophobic interaction between the sorbent bed and sweeteners (Zygler et al. 2010). As a result, this enables better retention of the sweeteners on the SPE cartridge, especially ACS-K and CYC. Recoveries listed in Table 4 indicate good SPE method accuracy. Table 6 lists the results from the analyses of eight yogurt samples. All yogurts were analysed in triplicate and all four sweeteners found in the yogurt samples were below their assigned MUD concentrations. The yogurt products contained either one or two sweeteners in combination (Figure 4). The concentrations of the sweeteners ranged from 24 to 274 µg g−1. The %RSDs were in the range of

Figure 4. (colour online) MRM chromatogram of a 1/25-fold dilution of yogurt 7. Complete experimental conditions are listed in the Analytical method section. The overlay of the chromatograms represents the most intense MRM transition (quantitation) and the second transition (confirmation).



2.1–8.9, indicating that good precision is achieved for the entire method, including the SPE procedure. The sweeteners detected by our method were consistent with those listed as ingredients on the yogurt labels. Conclusions An UHPLC-MS/MS method for the determination and quantitation of 14 non-nutritive sweeteners in foods was developed and single-laboratory validated. This method requires minimal sample clean-up for beverages and hard candies and provides accurate identification and quantitation of all target analytes in a single analytical procedure. Yogurt samples are processed using SPE with high and consistent recoveries. This method demonstrates good precision and accuracy. It will be useful for industry and regulatory authorities in order to monitor sweetener concentrations in commercial products to ensure compliance with country-specific regulations. References Aburjai T, Alzweiri M, Al-Hiari YM. 2011. Temperature and pressure behaviours of methanol, acetonitrile/water mixtures on chromatographic systems. Am J Anal Chem. 2:934–937. AOAC. 2012. AOAC Official Method 969.27. Nonnutritive Sweeteners in Nonalcoholic Beverages. Qualitative Thin Layer Chromatographic Method, AOAC Official Method of Analysis 19th Edition Croitoru MD, Fülöp I, Ajtay MK, Balogh C, Dogaru MT. 2011. Direct Hplc-UV determination of cyclamate, saccharine and aspartame from soft drinks. Acta Aliment. 40:459–465. Gardner C, Wylie-Rosett J, Gidding SS, Steffen LM, Johnson RK, Reader D, Lichtenstein AH et al. 2012. Nonnutritive sweeteners: current use and health perspectives: a scientific statement from the American Heart Association and the American Diabetes Association. Diabetes Care. 35:1798–1808. Hawk H, Graham L, Fong A, Phillips B. 2007. Identification and quantitation of artificial sweeteners in drinks using ion chromatography. FDA Lab Inf Bull. #4392:1–12. Huvaere K, Vandevijvere S, Hasni M, Vinkx C, Van Loco J. 2012. Dietary intake of artificial sweeteners by the Belgian population. Food Addit Contam A. 29:54–65.

11

Lim H-S, Park S-K, Kwak I-S, Kim H-I, Sung J-H, Jang S-J, Byun M-Y, Kim S-H. 2013. HPLC-MS/MS analysis of 9 artificial sweeteners in imported foods. Food Sci Biotechnol. 22:233–240. Mortensen A. 2006. Sweeteners permitted in the European Union: safety aspects. Scand J Food Nutr. 50:104–116. Namiesnik PKAJ. 2009. Quality assurance and quality control in the analytical chemical laboratory. Boca Raton (FL): Taylor & Francis Group. Ordóñez EY, Quintana JB, Rodil R, Cela R. 2012. Determination of artificial sweeteners in water samples by solid-phase extraction and liquid chromatography-tandem mass spectrometry. J Chromatogr A. 1256:197–205. Shah R, De Jager LS, Begley TH. 2012. Simultaneous determination of steviol and steviol glycosides by liquid chromatography-mass spectrometry. Food Addit Contam A. 29:1861–1871. US Food and Drug Administration. 2003. Guidance for industry 118, mass spectrometry for confirmation of the identity of animal drug residues. Rockville (MD): Center for veterinary Medicine. Wasik A, McCourt J, Buchgraber M. 2007. Simultaneous determination of nine intense sweeteners in foodstuffs by high performance liquid chromatography and evaporative light scattering detection – Development and single-laboratory validation. J Chromatogr A. 1157:187–196. World Health Organization G. 2009. Safety evaluation of certain food additives. Prepared by the seventy-first meeting of the Joint FAO/WHO Expert Committee on Food Additives (JECFA). pp. 35–36. Yang D-J, Chen B. 2009. Simultaneous determination of nonnutritive sweeteners in foods by HPLC/ESI-MS. J Agric Food Chem. 57:3022–3027. Zygler A, Wasik A, Kot-Wasik A, Namieśnik J. 2011. Determination of nine high-intensity sweeteners in various foods by high-performance liquid chromatography with mass spectrometric detection. Anal Bioanal Chem. 400:2159–2172. Zygler A, Wasik A, Kot-Wasik A, Namieśnik J. 2012. The content of high-intensity sweeteners in different categories of foods available on the Polish market. Food Addit Contam A. 29:1391–1401. Zygler A, Wasik A, Namieśnik J. 2009. Analytical methodologies for determination of artificial sweeteners in foodstuffs. Trac-Trend Anal Chem. 28:1082–1102. Zygler A, Wasik A, Namieśnik J. 2010. Retention behaviour of some high-intensity sweeteners on different SPE sorbents. Talanta. 82:1742–1748.