Quantification of Acrylamide in Various Belgian Potato Products Using Solid Phase Extraction and Liquid Chromatography Tandem Mass Spectrometry Detection

Acry lamide (CH2=CHCONH2), a neurotoxic and potentially carcinogenic substance for human health, is in the glare of the spotlights for a few years. This is mostly due to the fact that acrylamide was found worldwide in various heated foodstuffs. Levels reported in the literature vary from 25 to 2000 μg/kg and potato products are considered as containing the highest level in acrylamide. A possible pathway of synthesis of acry lamide is the Maillard reaction between reducing sugars and the amino acid asparagine. The aim of this study was to develop a liquid chromatography/mass spectrometry method to analyse as quickly as possible acrylamide in a variety of Belgian food samples such as potatoes, French fries, crisp bread, coffee, corn-flakes, etc. The sample preparation consisted in a liquid/liquid ext raction, a centrifugation, fo llowed by purification with Solid Phase Extraction (SPE). The instruments used were a Waters 2690 Alliance HPLC system coupled to a Micromass Quattro Ultima Platinum triple-quadrupole mass spectrometer. The analysis was performed in MS/MS mode using isotopic dilution technique for quantification. An internal 13C3 labelled standard was added prior to ext raction. Quantification in MS/MS mode was calculated by reconstructing the ion current with the most abundant daughter ions for native and 13C labelled standard (ions of m/z 55 and 58).


Introduction
Since the Swedish researchers put in evidence that there was a health problem with acrylamide (AA) being present in various common foodstuffs at high concentrations [1], acrylamide is among the "hot" topics in the analysts community.
In fact, acrylamide is a co mpound qualified as neuroto xic and "p rob ab ly carcinog en ic to hu mans" by th e U.S. Environmental Protection Agency [2] and the International A gency fo r Res earch on Can cer (IA RC) [3]. A s it is mentioned in nu merous articles, acrylamide present in food is formed at h igh temperatu res by the react ion between amino acids and reducing sugars [4,5]. Th is reaction is also known as the Maillard reaction [4,[6][7][8]. Heat-treated potato products (chips, French fries) are the products containing the highest amounts of acry lamide with appro ximately 2000 µg/kg when the smallest amount is approximately 25 µg/kg in coffee or bread. Different surveys have already been conducted in different countries showing that acrylamide is widely present in various foodstuffs [9][10][11][12][13][14][15][16] . In 2003, the Scientific institute of public health confirmed that acrylamide was also present on the Belg ian market [17].
Since 2002, laboratories all over the world have begun settling analytical methods to detect that compound. Many different approaches have been exp lored but the most proposed methods were those using mass spectrometry (MS), either coupled to gas chromatography (GC) [18][19][20][21] and either to liquid chro matography (LC) [1,[18][19][20][22][23][24][25]. There might be a drawback with GC because some GC-MS methods need a derivatization of acry lamide prior to the injection. W ith regard to the extraction, solid phase extraction (SPE) is the most widespread. However, So xh let extraction [25] or Accelerated Solvent Ext raction (ASE) [26,27] are also used.
The purpose of the present work was to detect the presence of acrylamide in a variety of Belgian foodstuffs with a method using solid phase extract ion and LC with tandem mass spectrometry (M S/MS). The samp le preparation was modified fro m one provided during an acrylamide workshop at the Institute of Public Health in Brussels [28]. The optimisation of acrylamide detection and quantification by mass spectrometry will be pointed out below. So me parameters will be discussed such as collision gas pressure, LC colu mn types, matrix-matched or standard calibrat ion. Finally, results of Belgian foodstuffs study will show the overall content of acrylamide.

Instruments
A 2690 A lliance Separation Modules (Waters, Milford, MA, USA) integrated autosampler, solvent delivery system and column heater coupled to a Quattro Ultima Plat inum triple-quadrupole mass spectrometer (Micro mass, Manchester, UK) were used for LC-MS-MS analysis.

Standard Solutions
Stock solutions of acrylamide (1 mg/ mL) and 13C-labelled acry lamide (10 µg/mL), used as internal standard, were prepared by d issolving the compounds in methanol. A ll solutions were stored at 4℃.

Sample Preparation
All sa mples (raw o r fried) were then mixed with a Moulinex mixer (Germany). One g ram of sample was weighed and transferred into a 50 mL polypropylene graduated conical tube with cap. Then, sample was spiked with internal standard to achieve a final concentration of 125 µg/kg 13C-acrylamide. 10 mL of water were added. The solution was mixed a first time on a Vortex (h igh speed) during approximately one minute, then approximately 5 minutes on a rotating shaker.
The suspension was centrifuged (3700 g, 15 minutes, 5℃). If supernatant was still turbid, the centrifugation was repeated a second time. Then, 6 mL of the supernatant were filtered on a 0.20 µm syringe filter before applying on solid phase extract ion (SPE) cartridge.
For the SPE clean-up (see figure 1), the OASIS HLB SPE cartridge (Waters) was conditioned under vacuum with methanol (5 mL), and equilib rated with water (5 mL). Then, 5 mL of the filtered sample were loaded on the OASIS HLB SPE cartridge and the ext ract was allowed to pass comp letely through the sorbent material. The OASIS HLB SPE was washed with water (2 mL) and the ext ract was allo wed to pass completely through the sorbent material. Then, the cartridge was eluted with 7.5 mL of methanol. The eluent was collected.
For the second step of the clean-up, the Bond Elut AccuCAT SPE cartridge (mixed-mode SPE co lu mn consisting of a strong cation exchange and a strong anion exchange sorbent packed into one bed, Varian Inc.) was conditioned under vacuum with 5 mL of methanol. Then, the Bond Elut AccuCAT SPE cartridge was loaded with the solution fro m the prev ious step and the eluent was collected directly.
The cartridge was rinsed with methanol (1.5 mL) wh ich was combined with the eluent from the OASIS. The extract was then evaporated to dryness under N2 at 40℃. Finally, 500 µL of water were added and the solution was transferred into an injection vial.

LC-ES I-MS -MS
For analytical separation, an Allt ima HP C 18 amide column (250 x 2.1 mm I.D., 3 µm, A lltech, Deerfield, IL, USA) and Alltima HP C18HL (250 x 2.1 mm I.D., 3 µm, Alltech, Deerfield, IL, USA) were tested. The elution mode was isocratic using water with 0.1 % acetic acid as LC solvent. The flow-rate was 0.2 mL/ min and the injection volume was 20 µL. The total run time was 8 min. The column and samp les temperatures were set to 40℃ and 10℃, respectively. The mass spectrometer was operated in positive-ion mode, nitrogen as cone gas (~100 L/h) and desolvation gas (~680 L/h). Argon was used as collision gas. The source and desolvation temperatures were set to 125 and 250 ℃ , respectively. Multip le reaction monitoring (MRM ) t races  Table 1.
The capillary and cone voltages were set to 3.2 kV and 43 V, respectively.

Potato Samples
Samples used for optimising the method were purchased in different supermarkets in Liège, Belgiu m.
Then, the optimised method was applied to potato samples of different kinds provided by a specific Belg ian industry: normal and "wavy" French fries, cube-shaped and sliced potatoes. Those samples (internal codes fro m 112 to 118) were prepared industrially with different amounts of sugar for the reason that they were destined for diverse countries.
All samp les were analysed before and after frying, wh ich was realised in a deep fat fryer with beef tallow, at 180℃. The fat used was analysed at three defined times: before first sample was fried, in the middle of the series of samples and after last sample.
Quality Controls (QC) have been realised with commercially available Belgian raw potatoes.

Collision Gas Pressure
In order to reach a good sensitivity for acry lamide, the collision gas pressure was optimized. Acrylamide is a low mass molecule so the frag mentation is d ifficu lt to be achieved with the trad itional value of 2.5 10-3 mba r. Six injections of the same solution were done at five pressure values: 1.5x10 -3 , 7.5x10 -3 , 1.6x10 -2 , 3.1x10 -2 and 2.4x10 -2 mbar.  Figure 2 shows the variation of signal with the co llision gas pressure.
As it can seen on Figure 2, the best peak area for the transition 72>55 is obtained for a collision gas pressure of 1.6x10-2 mbar. This is far fro m the values usually used in ESI-M S methods but this will assure a better response for acrylamide analysis.

LC Columns
Two different LC co lu mns were tested (Alltech Alltima HP C18HL 3µm and Alltech Alltima HP C18 amide 3µm) using the same solutions and analytical p rotocol. Tests were realised on raw potatoes spiked with a 250 ppb acrylamide solution and 125 ppb internal standard solution. The chromatograms obtained for acry lamide 12C and 13C are shown in Figure 3. LC analysis realised with the amide colu mn (b ) gives better results than with the HL colu mn (a), as well as for the area of the peaks than for their shape. What's more, the retention times are increased to 4.85 minutes with the amide column instead of 1.85 with the HL co lu mn.

Matrix-Matched Cali bration Versus Standard Cali brati on
In order to check matrix effects, we co mpared the results of a calibration curve realised with standards solutions at different concentrations with the results of raw potato samples spiked at different concentrations. Those samples were used as quality controls. The concentration of acrylamide (in ppb) versus the response is shown in Figure 4. The standard solutions and the spiked samples are represented by dots and triangles respectively. Figure 4 shows that the response obtained with each potato sample matches the values on the calibration curve realised with the standard solutions.
There is no significant difference between a calibrat ion curve realised with or without matrix. Fro m now, this will allo w using curves realised without mat rix and then the analysis will need less samples, less solvents and materials and therefore induce lower costs.
Limit of quantificat ion (LOQ) has been established at 50

Real Samples Study
The optimised method was applied to potato samples. All of them were analysed before and after fry ing.
Results of the content in acrylamide detected in the potato samples are shown in Table 2.
The concentrations obtained are in concordance with the values found in literature.
What's more, the results of the three analyses of frying oil were all negative, meaning that the amount of acrylamide found in a sample is co ming neither fro m potatoes previously fried, nor fro m the oil itself. It seemed that the quantity of sugar present in a variety of French fries is not correlated to the concentration of acrylamide formed during frying.

Conclusions
A simp le and rap id method for the quantification of acrylamide in a variety of foodstuffs has been developed. The optimisation was done focusing on the collision gas pressure and on the analytical colu mn. This method has been tested with success on different Belg ian products (potatoes, French fries, bread …).