Application of an Accurate and Validated Method for Identification and Quantification of Acrylamide in Bread , Biscuits and Other Bakery Products Using GC-MS / MS System

A gas chromatography tandem mass spectrometry has been developed and validated for the separation, detection, identification and quantification of acrylamide in bread, biscuits and similar products. The method showed good precision with values lower than 6%. A good sensitivity was achieved for bread with 2.41 and 7.23 μg kg limit of detection (LOD) and limit of quantification (LOQ), respectively, while for biscuits, LOD and LOQ were 4.63 and 13.89 μg kg, respectively. Accuracy obtained through the bias of 2 certified reference materials (“crisp bread ERMBD272” and “rusk ERM-BD274”) gave a value below 1.68-2.52%. The method was applied by analyzing 49 types of bread, biscuits and other similar products. The results showed different levels of acrylamide in bread (values ranged between 7.6 and 165.6 μg kg), biscuits (between LOD and 2405.0 μg kg), sandwich biscuits with cream (112.6-570.4 μg kg), biscuits for infants and young children (between LOD and 801.7 μg kg), gingerbread (349.5-955.5 μg kg) and crackers (347.8-366.1 μg kg).


Introduction
Acrylamide (AA) is a chemical that can be formed in some foods during certain types of high-temperature cooking.The neurotoxicity of AA in humans is well established from occupational and accidental exposures, and experimental studies have shown reproductive, genotoxic and carcinogenic effects in animals.AA has been classified as probably carcinogenic to humans by the International Agency for Research on Cancer (IARC). 1 The possible public health risks from dietary exposure to AA attracted concern by public authorities all over the world. 2 Based on the expert committee evaluation of FAO/ WHO, 1 µg kg -1 body weight (bw) day -1 is considered as an average exposure to AA while 4 µg kg -1 bw day -1 is considered as high exposure to AA. 3 The mean dietary exposure range to AA is 0.2-1.0 µg kg -1 bw day -1 for the general adult population while 95th-percentile range is 0.6-1.8µg kg -1 bw day -1 . 4In 2013, the U.S. Food and Drug Administration (FDA) 5 issued draft guidance to help the food industry to reduce the amount of AA in certain foods.
Commission Recommendation 2013/647/EU established, based on the monitoring results received from the Member States during the period 2007-2012, the "indicative values" for AA in different foods.The "indicative values" are not safety thresholds, but their aim is to indicate the need for an investigation in the case the values are exceeded and to take appropriate measures to control the formation of AA. 6 Taking into account that the presence of AA in food products could increase the risk of cancer in consumers of all ages, 7,8 an internationally priority is to develop reliable analysis method to quantify low concentrations of AA in foods (dozens of ppb) and to find solutions to reduce this contaminant.][11] Although in the last decade were published a large number of analytical methods based on gas chromatography (GC) [12][13][14][15][16][17] and liquid chromatography (LC) coupled with mass spectrometry (MS; MS/MS), [18][19][20][21][22] the methods are mainly used by private laboratories and official control authority for routine analysis of AA in food products.Last publications apparently confirm the validity of LC and GC techniques, which enables to guarantee the determination of AA with quantification limits from 30 to 50 µg kg -1 for HPLC-MS, 4 to 30 µg kg -1 for GC-MS and LC-MS/MS.Commission recommendation of 2 June 2010, states that for ensuring comparability of analytical results, methods that can achieve a limit of quantification (LOQ) of 30 µg kg -1 for bread and foods for infants and young children and 50 µg kg -1 for potato products, other cereal products, coffee and other products should be chosen. 23he aim of this study is to present the results obtained on internal validation of GC-MS/MS method for determination of AA in bread, biscuits and other similar products with LOQ below the above recommended values.Moreover, assessing the level of AA in some bakery products from the Romanian market was performed.

Sampling
Seventeen types of bread and similar products, 18 types of biscuits, 5 types of biscuits for infants and young children, 4 types of sandwich biscuits, 2 crackers and 3 gingerbread were studied.Samples were either purchased in local market (various types produced in different countries) or produced on a pilot plant of the National Research & Development Institute for Food Bioresources (Romania).
Carrez I and II solutions were prepared by dissolving 28.8 g of potassium hexacyanoferrate and 57.6 g of zinc sulfate in 100 mL of water, respectively.
Reference test materials ERM  -BD272 crispbread (AA content of 980 ± 90 µg kg -1 ) and ERM  -BD274 rusk (AA content of 74 ± 7 µg kg -1 ) were certified at the BAM Federal Institute for Materials Research and Testing (Berlin, Germany) to verify the accuracy of the method.

Preparation of stock and working solutions
Stock solution of AA (100 mg L -1 ) and IS (100 mg L -1 ) were prepared in amber vials by dissolving in ultrapure water.Working solutions I, II and III (10, 1 and 0.1 mg L -1 of AA, respectively) and working solution I (10 mg L -1 ) of IS were prepared by diluting the stock solution with ultrapure water.All stock and working solutions were kept in a refrigerator at 4 °C.

Sample preparation
The procedure of sample preparation for AA quantification in bread, biscuits and other bakery products was according to previous study [13][14][15] with some modifications 24,25 regarding sample preparation (sample preparation for soft bread, sample weight taken for analysis, quantities and volumes for reagents) and AA analysis by GC-MS/MS in selected reaction monitoring (SRM) acquisition mode, respectively.Briefly, the steps were as follow: Drying, milling and weighting Only fresh bread samples were dried in an oven at 90 °C, for 120 min, while biscuits and other similar products were used as they were.The samples were fine milled using a Retch GM 200 mill (Germany) and an ultra centrifugal ZM 200 mill (Germany).In a centrifuge vial of 50 mL, 3 g of "bread and other similar products" and 1.5 g from "biscuits and other similar products" category were weighed for AA analysis, respectively.

AA extraction in water
Over the sample weighed, working solution I of the IS of 10 mg L -1 (110 µL) and 30 mL ultrapure water at 60 °C were added.Extraction in water at pH 4-5 was achieved by addition of 20-40 µL of glacial acetic acid, followed by deproteinization with 400 µL Carrez I and II solutions and centrifugation (6000 × g) at 5 °C, for 30 min.

AA derivatization
The supernatant obtained was derivatized with 7.5 g KBr, 40-100 µL HBr (pH 1-3), 10 mL saturated bromine-water solution (around 1.6%) on a shaking water bath below 4 °C, for at least 2 h.After the end of the derivatization reaction, bromine in excess was removed by adding around 1-2.5 mL of 1 mol L -1 sodium thiosulfate, until the yellow color disappeared.

2,3-DBPA
The concentration was achieved in a first step using a vacuum evaporation system (Rotovapor R-210, BUCHI Labortechnik AG, Germany) till 2 mL, followed in a second step by dryness under a stream of nitrogen.

2,3-DBPA purification and concentration
The residue dissolved in 50 mL hexane was purified on a glass column filled with activated Florisil and calcinated sodium sulfate, previously conditioned with 20 mL hexane.The 2,3-DBPA derivative was eluted with acetone, then concentrated till dryness and the residue was redissolved in 400 µL ethyl acetate and 40 µL triethylamine.The final solution was filtered through a 0.2 µm regenerated cellulose microfilter (17 mm diameter, Spartan 13RC, Whatman GmbH, Dassel, Germany) directly in a vial and analyzed by GC-MS/MS in SRM mode.

Preparation of the calibration solutions
A blank and 7 calibration levels in the range 0.1-7.5 mg L -1 were prepared in flasks of 250 mL.In each flask, the following were added: 100 mL ultrapure water, 110 µL IS of 10 mg L -1 and working solution II/I of AA (1 mg L -1 per 10 mg L -1 ) according to the calibration level.These solutions were derivatized in the same manner as the derivatization steps described for bread, biscuits and other similar samples, without purification and the final residue was redissolved in 1000 µL ethyl acetate and 100 µL triethylamine.

GC-MS/MS-SRM analysis
The calibration solutions and the derivatized sample extracts were analyzed using a gas chromatograph (TRACE GC ULTRA) coupled with triple quadrupole mass spectrometer (TSQ Quantum XLS) from Thermo Fisher Scientific (USA).The analysis was performed in the electron impact positive ionization mode (EI + ); acquisition mode: "selected reaction monitoring-SRM" and ion scanning mode: "product".
The determination was carried out with a capillary column based on polyethylene-glycol (30 m × 0.25 mm internal diameter; 0.25 µm) (TraceGOLD™ TG-WaxMS, Thermo Fisher Scientific, USA).The mobile phase was helium with a constant flow rate of 1.6 mL min -1 . 1 µL sample was injected in a Right PTV type injector using TriPlus AS autosampler (Thermo Fisher Scientific, USA), in a split mode with a split ratio of 1:10 and injector temperature of 220 °C.Under these conditions, the retention time of AA and of the IS was of 10.65 ± 0.5 min.AA and IS were identified by the ion fragments corresponding to the derivatized ions, 2-BPA (2-bromopropenamide) and 2-BP( 13 C 3 )A.The fragmentation of the precursor ions with m/z 151 and 154 was achieved with argon (1 mTorr), leading to the formation of product ions (daughter) with m/z 70 (2-BPA) and 73 (2-BP( 13 C 3 )A), being used for quantification.The calculation of the AA concentration in the test samples was based on the ratio of the peak area corresponding to the product ions, with m/z 70 and 73 for 2-BPA and 2-BP( 12 C 3 )A, respectively.
For bread samples and other similar products, AA concentration was calculated based on equation 1, while for biscuits and other similar products using equation 2.
where C is the 2-BPA concentration measured by the instrument in the food sample (mg L -1 ), C calc. is the 2-BPA concentration calculated for the food sample (µg kg -1 ), U 1 is the moisture content of the bread sample before drying (%), U 2 is the moisture content of the bread sample after drying (%), w is the weight of the sample (g), 440 is the final volume of the sample extract (µL).

Calibration curve and linearity
Calibration curve and linearity were verified by the method of least squares in the range 0.1-7.5 mg L -1 , choosing the method of the internal standard.The calibration curve was obtained based on derivatized standard solutions and plotting the ratio of the peak area of 2-BPA and 2-BP( 13 C)A against the concentration of 2-BPA.The calibration model was considered correct if relative standard deviation (RSD) in repeatability conditions was within the limits of ± 15% for all the levels investigated.The correlation was assessed to be linear for a value greater than 0.99 for the correlation coefficient (R).

Linearity range
In order to establish higher values for the linearity range, the samples were spiked with working solution I of native AA (10 mg L -1 ) at different levels of concentration: 40; 75; 80; 150; 225 and 300 µg kg -1 (bread and other similar products) and 250; 500; 750; 1000 and 2000 µg kg -1 (biscuits and other similar products).The ratio of the peak area of 2-BPA and 2-BP( 13 C)A against the concentration of 2-BPA from the food samples studied was plotted.The correlation was assessed to be linear for a value greater than 0.99 for the correlation coefficient (R).

Sensitivity
Sensitivity was characterized by limit of detection (LOD) and limit of quantification (LOQ).Assessment of LOD and LOQ for AA analysis in bread and similar products by GC-MS/MS were performed applying IUPAC approach, while ICH approach was used for biscuits and other similar products.The criteria imposed for LOQ were: precision (expressed as relative standard deviation, RSD ≤ 20%) and accuracy (expressed as recovery in the range 80-120%).

Selectivity
Selectivity was verified by observing the possible interference given by AA in bread, biscuits and other similar products. 32Specificity was achieved by SRM detection.

Precision
Precision was determined by repeatability (injection and analysis repeatability), inter-laboratory reproducibility and intermediate precision.Injection repeatability was achieved by carrying out between 9 and 10 consecutive injections in a short period of time (the same day).Analysis repeatability (intra-day) was determined by 5-6 repeated analyses of the same sample, in the same day by the same analyst and under the same experimental conditions.Inter-laboratory reproducibility (inter-day) was achieved through Food Analysis Performance Assessment Scheme (FAPAS) 33 of the Central Science Laboratory (FERA).Intermediate precision (inter-day) was carried out by repeated analyses of the same sample in 3 different days by the same analyst and under the same experimental conditions.Repeatability and intermediate precision was expressed as RSD (%), while interlaboratory reproducibility as z-score.

Accuracy
The accuracy of the method was evaluated with recovery and bias measurements for 2 reference test materials.The recovery percentage (R%) of the method was established from minimum 6 repeated analysis of the same sample fortified with working solution II of AA at different concentrations.

Robustness
To assess the robustness of the method developed in the laboratory, the following parameters were varied: volume of hexane used in the purification of the extracts, derivatization time and mobile phase flow.It was considered that the method is robust for the parameters investigated, whether obtained under repeatability conditions RSD or reproducibility was below 2%.

Measurement uncertainty
The uncertainty sources were identified and analyzed and the uncertainty budget was estimated, according to the reference documents. 34,35Expanded uncertainty (U ∆ ) was calculated by multiplying the combined standard uncertainty (U c ) with a coverage factor (k = 2) for a confidence level of 95%.

Method application
The validated and developed method was applied by analyzing different products presented in Tables 1-8.The AA level was also studied in different batches of the same brand (3 batches × 2 types of breads; 2 batches × 5 types of biscuits).

Statistical analysis
AA content was expressed as mean ± standard deviation.The differences among sample groups were analyzed by one-way ANOVA (analysis of variance) followed by Tukey's test, p < 0.05 was considered statistically significant.
Regarding validation parameters, average for concentration values, C calc (µg kg -1 ), standard deviation in repeatability SD(r) and reproducibility conditions SD(R) in µg kg -1 as well as relative standard deviation in repeatability RSD(r) and reproducibility conditions RSD(R) in % were calculated using Microsoft Excel.The repeatability and reproducibility limits in µg kg -1 were also considered, where the ratio between repeatability and reproducibility is 2.8 × SD(r)/SD(R) for a confidence level of 95%.

Results and Discussion
An overview of the test samples together with their coding and AA content is given in Tables 1-8.

Calibration curve and linearity
Linearity was investigated with the aid of a regression line with 7 calibration levels by the method of least squares.
The calibration curve for the determination of AA in bread, biscuits and other similar products by GC-MS/MS analysis (y = 0.8613x + 0.0332) was linear over the range of 0.1-7.5 mg L -1 .R was between 0.9996-1, which proves a high linearity degree.The results showed that RSD(r) for all the levels of concentration studied ranged between 0.1 and 7.8%, being within the limits of ± 15%.
The linearity range was verified according to ISO 36 for bread, biscuits and other similar products, respectively.The linearity range of the method for AA analysis was in the range of: 7.23-312.88µg kg -1 (R > 0.998) for bread and similar products (n = 23) and 17.59-2191.41µg kg -1 (R > 0.999) for biscuits and similar products (n = 42), respectively, n denotes the number of the analyzed samples.The results obtained are in line with a recent study, 37 which reported an LC-MS/MS method for AA determination in different types of breads, with a linear range of up to 750 µg kg -1 food and a determination coefficient of 0.999.

Sensitivity
For bread and similar products, LOD and LOQ obtained were 2.41 and 7.23 µg kg -1 , respectively.For biscuits and similar products, LOD and LOQ were: 4.63 and 13.89 µg kg -1 , respectively.LOQ values fulfilled the two criteria (RSD < 10% and recovery between 93.68 and 102.93%).
LOQ value for the present method was below 30 µg kg -1 which fulfils the criteria defined in European Commission recommendation on the monitoring of AA levels in food. 23lso, the results obtained are in accordance with a previous GC-MS study, which reported LOD and LOQ values of 2 and 5 µg kg -1 , respectively for bread samples. 14he working range reached from 2.41-2191.41µg kg -1 AA in a food sample.

Selectivity
Selectivity was demonstrated by chromatographic separation of AA in the presence of other components from the food matrix (bread, biscuits and similar products).
For specificity, SRM detection and the use of IS method lead to a specific analysis.The retention time of AA within laboratory condition for the samples was almost identical with the retention time of the IS, which fits in the tolerance of ± 0.5% imposed by Commission Decision (2002/657/CE). 30For AA quantification, ion m/z 70 (2-BPA) and ion m/z 73 for IS (2-BP( 12 C 3 )A) were used.The values obtained for injection and analysis repeatability, intermediate precision are presented in Table 9.
RSD(r) was between 0.71 and 3.30% for injection repeatability and between 0.44 and 4.90% for repeatability analysis, respectively.The RSD(R) intermediate precision was between 0.35 and 5.83%.The applied method showed a good precision, RSD < 10% for AA analysis in bread, biscuits and similar products and the results were within the limits set by the following regulations: VICH GL49, 28 CAC/GL 16, 29 EC. 30 The AA result for crispbread sample from the FAPAS proficiency test (laboratory No. 24) showed a value of 193.77 µg kg -1 (z-score of 0.2), which fulfilled the acceptability criteria from the organization. 33

Method accuracy
The bias obtained for the two reference materials were: 1.68% (n = 39) for ERM  -BD272 crispbread and 2.52% (n = 30) for ERM  -BD274 rusk, respectively.The described method demonstrated a satisfactory accuracy, between 99.59 and 102.28% for bread and similar products and between 95.44 and 104.63% for biscuits and similar products.AA: acrylamide; data shown as mean ± SD (n = 3) for the amount of AA; means followed by different letters are significantly different (p < 0.05); b OF: oat flour; c WF: wheat flour; d WWF: whole wheat flour; e the AA levels were higher than the maximum value of the working range, but in order to quantify it, the samples weight taken for analysis was lower (0.75 g instead of 1.5 g).

Robustness
The small variations in the volume of hexane used in the purification of the extracts, derivatization time and mobile phase flow are not susceptible to variations in the area corrected.Standard deviation was between 1.79 and 2.07%.

Measurement uncertainty
The uncertainty of results for AA determination in bread, biscuits and similar products by GC-MS/MS was ± 21% as estimated by budget uncertainty.

Table 1 .
Bread and similar products produced in the pilot plant

Table 2 .
Bread and similar products (origin: retail store, produced in Romania) a AA: acrylamide; data shown as mean ± SD (n = 3) for the amount of AA; means followed by different letters are significantly different (p < 0.05).

Table 4 .
Biscuits samples (origin: retail store, produced in different countries)

Table 3 .
Biscuits samples produced in the pilot plant a AA: acrylamide; data shown as mean ± SD (n = 3) for the amount of AA; means followed by different letters are significantly different (p < 0.05); b RF: rice flour; c WF: wheat flour; d OF: oat flour; e LOQ: limit of quantification.

Table 7 .
Crackers samples (origin: retail store) shown as mean ± SD (n = 6) for the amount of AA; values with the same letter denote no significant difference (p > 0.05) between samples; b WF: wheat flour.
a AA: acrylamide; data

Table 6 .
Sandwich biscuits with cream filling (origin: retail store) a AA: acrylamide; data shown as mean ± SD (n = 5) for the amount of AA; means followed by different letters are significantly different (p < 0.05); b WF: wheat flour.

Table 5 .
Biscuits for infants and young children (origin: retail store, produced in different countries) ; data shown as mean ± SD (n = 6) for the amount of AA; means followed by different letters are significantly different (p < 0.05); b WF: wheat flour; c WWF: whole wheat flour; d LOQ: limit of quantification.
a AA: acrylamide