G Model

ARTICLE IN PRESS

CHROMA-355301; No. of Pages 10

Journal of Chromatography A, xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Multiclass analysis of mycotoxins in biscuits by high performance liquid chromatography–tandem mass spectrometry. Comparison of different extraction procedures Anna Laura Capriotti, Chiara Cavaliere, Patrizia Foglia ∗ , Roberto Samperi, Serena Stampachiacchiere, Salvatore Ventura, Aldo Laganà Department of Chemistry, “La Sapienza” University of Rome, Piazzale Aldo Moro 5, 00185 Rome, Italy

a r t i c l e

i n f o

Article history: Received 24 January 2014 Received in revised form 2 April 2014 Accepted 4 April 2014 Available online xxx Keywords: Mycotoxins Cereal-egg products Biscuits Liquid-chromatography Mass spectrometry

a b s t r a c t A sensitive, simple and rapid method for the simultaneous determination of 19 mycotoxins in biscuits (a dry matrix containing cereals and egg) has been developed using high performance liquid chromatography coupled to tandem mass spectrometry with electrospray source working in both positive and negative mode. Due to the matrix complexity and the high amount of contaminants, a solid phase extraction method using graphitized carbon black was optimized for an effective clean-up step. Accuracy was carried out in the selected matrix using blank samples spiked at three analyte concentrations. Recoveries between 63 and 107% and relative standard deviations lower than 12% were obtained. For all considered mycotoxin classes, i.e. thricotecenes A and B, zearalenone and its metabolites, fumonisins, ochratoxin A, enniatins and their structurally related beauvericin, the method was validated in terms of linearity, recovery, matrix effect, precision, limit of detection and limit of quantification. Matrix-matched calibration was used for quantification purposes, in order to compensate for matrix effect. The coefficients of determination obtained were in the range of 0.9927–1. The limits of quantification, ranging from 0.04 ␮g kg−1 for enniatin B1 to 80.2 ␮g kg−1 for nivalenol, were always lower than maximum permitted levels for every regulated mycotoxin by the current European legislation. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Mycotoxins are a group of naturally occurring toxic compounds produced by the secondary metabolism of many filamentous fungi (mainly Penicillium, Fusarium and Aspergillus genera) [1,2]. Both fungal growth and mycotoxin production depend on a variety of factors. Whenever several physical, chemical, and biological conditions take place, mycotoxin contamination may occur. High temperature processes cause varying degrees of reduction of mycotoxin concentrations, but most mycotoxins are moderately stable in most food processing systems thus they can still be found even in finished products [3]. The presence of mycotoxins in food may cause mycotoxicosis, which comprises many different adverse effects, such as the induction of acutely toxic, immunosuppressive,

∗ Corresponding author at: Dipartimento di Chimica, Università degli Studi di Roma “La Sapienza” Piazzale Aldo Moro 5, 00185 Rome, Italy. Tel.: +39 06 49913748; fax: +39 06 490631. E-mail address: [email protected] (P. Foglia).

mutagenic, teratogenic, oestrogenic and carcinogenic effects; for these reasons mycotoxin contamination of food and feed is a worldwide problem with great relevance to human and animal health [4–7]. As far as mycotoxin contamination of cereals and their products is concerned, current European Union (EU) food safety legislation regulates the content of some mycotoxins by means of the Regulation (CE) 1881/2006 [8] and its subsequent amendments (Regulation (CE) 1126/2007 and Regulation (UE) 165/2010) [9]. Ochratoxin A maximum level (ML) is set at 3 ␮g kg−1 , deoxynivalenol ML ranges between 500 and 750 ␮g kg−1 , depending on the type of cereal intended for direct human consumption (the lower level is for bread, including small bakery wares, pastries, biscuits, cereal snacks and breakfast cereals); zearalenone ML ranges between 50 and 75 ␮g kg−1 , depending on the type of cereal intended for direct human consumption (the lower level being as previously described for deoxynivalenol); fumonisins (FBs), the levels of which are regulated only for maize, maize-based breakfast cereals and maize-based snacks, have a ML set at 800 ␮g kg−1 for the sum of FB1 + FB2 . T-2 and HT-2 toxins are named, but the

http://dx.doi.org/10.1016/j.chroma.2014.04.009 0021-9673/© 2014 Elsevier B.V. All rights reserved.

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model CHROMA-355301; No. of Pages 10

ARTICLE IN PRESS A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

2

value of their limit is not indicated. All the other considered mycotoxins, such as less common trichothecenes A and B, zearalenone metabolites, enniatins and beauvericin, are not legislated yet but we decided to consider them because they can be found in maize [10] and because they can be considered as recently “emerging” mycotoxins [11,12], due to the increasing attention for their toxicity [13]. For multiclass mycotoxin analysis, in recent years there is a growing tendency to develop rapid LC-mass spectrometry (MS) methods with minimum sample treatments. Approaches commonly employ solid–liquid extraction (SLE) [14–18], solid phase extraction (SPE) [19,20], and/or liquid–liquid extraction (LLE) with the subsequent direct injection to LC-MS instrumentation and/or immunoaffinity-column (IAC) clean-up [21–28]. Nowadays, one of the main objectives in food contaminant analysis is the development of methods with minimum sample treatment. In many works the QuEChERS methodology (quick, easy, cheap, effective, rugged and safe) has also been recently employed for the determination of multiclass mycotoxin analysis in different food matrices [1,12,23,29] but its effectiveness is still under examination, with results depending on the type of matrix. The development of a multi-mycotoxin analysis method, which does not comprise any purification step for the quantification of analytes having great differences in structures and physicochemical properties, is still challenging, especially for matrices, such as cereal- and egg-derivative products, having high content of fats and other interfering substances. Therefore, despite the potentiality of LC-MS/MS techniques, in some cases a purification step cannot be omitted, in order to avoid significant matrix effects (MEs) which decrease sensitivity and could lead to quantification errors. The aim of this study was the development of a sensitive and reliable confirmatory multiclass procedure, based on SPE clean-up followed by LC/ESI-MS/MS, for the simultaneous determination of 19 selected mycotoxins in common worldwide cookies. The effectiveness of two direct methods (i.e. SLE and QuEChERS) was initially evaluated by high performance liquid chromatography (HPLC) coupled to a triple quadrupole mass spectrometry. Then we developed a method with a sample clean-up using graphitized carbon black (GCB) which allowed the selective isolation of analytes from complex matrices due to various types of interactions (i.e. hydrophobic, electronic and ion-exchange interactions). The developed method can be applied to the determination of all the selected mycotoxins and, as far as the legislated Fusarium mycotoxins (deoxynivalenol, zearalenone, T-2 toxin, HT-2 toxin and fumonisins B1 and B2) and ocrathoxin A is concerned, it allows their determination below their MLs, as regulated by EU [9,30,31].

2. Experimental

standard grade) whilst the other mycotoxin reference standards were supplied as powder (Premium Quality Level and/or assay ≥98%). Sulfamethoxazole (SMX) was used as volumetric internal standard (IS). Full names, abbreviations, chemical formula and precursor ions of the selected compounds are reported in Table 1. All reagents were of analytical reagent grade, solvents were LCMS grade. Formic acid, acetonitrile, methanol, ammonium formate, dichloromethane, hydrochloric acid, magnesium sulfate (MgSO4 ), and sodium chloride (NaCl) were obtained from Sigma–Aldrich. Ultrapure water (resistivity 18.2 M cm) was obtained by an Arium water purification system (Sartorius, Florence, Italy). 2.2. Standard solutions Individual analyte stock solutions were prepared at 100 ␮g mL−1 in acetonitrile. Fumonisin B1 and B2 were dissolved in acetonitrile/water (50:50, v/v). All of the standards were stored at −20 ◦ C. A composite standard working solution was prepared considering the intensity response (i.e. sensitivity in the LC-MS/MS measurement) of the target analytes and, for the legislated ones, their allowed MLs. According to this, the composite standard working solution was prepared by combining aliquots of each individual stock solution and diluting with acetonitrile to obtain the final concentration of 5 ␮g mL−1 for DON, NIV and FUSX; 2 ␮g mL−1 for 3-ADON, 15-ADON, HT-2, ␣-ZOL, ␤-ZOL, FB1, and FB2; 0.5 ␮g mL−1 for DAS, NEO, T-2, ZEA; 0.02 ␮g mL−1 for OTA, ENA, ENA1, ENB1, and BEA. All the above solutions were stored at −20 ◦ C in amber glass vials and kept in the dark at room temperature (20–25 ◦ C) before use. Working standard solutions were prepared by suitable dilution of stocks with acetonitrile. These solutions were kept at 4 ◦ C and renewed weekly. 2.3. Samples As no blank certified reference materials are available, a number of samples of cookies (digestive type) of brands commonly found in stores and purchased randomly from Rome area retail markets, were checked to evaluate their contamination level to be used as blank material for spiking purposes (Fig. S1 in supplementary data). To ensure representative sampling, ten packages (10× 400 g) were collected and ground to a fine powder using a mortar and pestle. After that, a 500 g mycotoxin-free pooled powdered sample was randomly taken from the homogeneous fine powder and stored in the dark at 4 ◦ C until analysis and used as blank materials for the validation study. Thereafter, sub-samples of 1 g were weighed for analysis. All samples used for method development were kept in a dark and dry location at room temperature (20–25 ◦ C) until handled.

2.1. Chemical and reagents 2.4. Sample preparation Standards of (i) Fusarium toxins, major acetylated conjugates and other products of transformation, namely 3-acetyldeoxynivalenol (3-ADON), 15-acetyldeoxynivalenol (15-ADON), deoxynivalenol (DON), diacetoxyscirpenol (DAS), fusarenon-X (FUSX), HT-2 toxin (HT-2), neosolaniol (NEO), nivalenol (NIV), T-2 toxin (T-2), zearalenone (ZEA), ␣-zearalenol (␣-ZOL), ␤-zearalenol (␤-ZOL); (ii) ochratoxins: ochratoxin A (OTA); (iii) fumonisins: fumonisin B1 (FB1), fumonisin B2 (FB2); and (iv) enniatins and their structurally related beauvericin: beauvericin (BEA), enniatin A (ENA), enniatin A1 (ENA1), and enniatin B1 (ENB1) were purchased from Sigma–Aldrich (St. Louis, MO, USA). 3-ADON, 15-ADON, FUS-X, NEO and NIV (100 ␮g mL−1 in acetonitrile) were purchased in solution (analytical

2.4.1. Extraction and clean-up apparatus A model ST ultrasonic bath at a frequency of 50 ± 3 Hz from Stimin (Milan, Italy), and an ALC (Milan, Italy) multispeed refrigerated centrifuge PK131R were used. Polypropylene tubes, polyethylene frits, and a vacuum manifold were from Supelco (Bellefonte, PA, USA); Carbograph-4 was purchased by LARA (Rome, Italy). Carbograph-4 is a GCB with a surface area of 210 m2 g−1 and particle size range of 120–400 mesh, similar to Carboprep 200 (Restek, Bellefonte, PA, USA) and Envicarb X (Supelco). Carbograph-4 cartridges were prepared by placing 500 mg of the adsorbent inside 6 mL polypropylene tubes between two

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model

ARTICLE IN PRESS

CHROMA-355301; No. of Pages 10

A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

3

Table 1 Full name, abbreviation, chemical formula, molecular weight, polarity, retention time, precursor and product ions and MS parameters of the analyzed mycotoxins. Precursor ion

Precursor ion (m/z)

Product ion (m/z)

CE

S-lens

2.27

[M + HCOO]−

357.0

3.28

[M + HCOO]−

341.0

C17 H22 O8

+

4.21

[M + H]+

355.0

Neosolaniol (NEO)

C19 H26 O8

+

4.48

[M + NH4 ]+

400.1

3-Acetyldeoxynivalenol (3-ADON)

C17 H22 O7

+

5.16

[M + H]+

339.1

15-Acetyldeoxynivalenol (15-ADON)

C17 H22 O7

+

5.16

[M + H]+

339.1

Diacetoxyscirpenol (DAS)

C19 H26 O7

+

6.49

[M + NH4 ]+

384.1

HT-2 toxin (T2)

C22 H32 O8

+

7.36

[M + NH4 ]+

442.2

Fumonisin B1 (FB1)

C34 H59 NO15

+

7.49

[M + H]+

722.3

␤-Zearalenol (␤-ZOL)

C18 H24 O5

−

7.62

[M − H]−

319.17

T-2 toxin (T2)

C24 H34 O9

+

7.85

[M + NH4 ]+

484.23

␣-Zearalenol (␣-ZOL)

C18 H24 O5

−

8.13

[M − H]−

319.16

Ochratoxin A (OTA)

C20 H18 ClNO6

+

8.18

[M + H]+

404.1

Zearalenone (ZEA)

C18 H22 O5

−

8.25

[M − H]−

317.03

Fumonisin B2 (FB2)

C34 H59 NO14

+

8.35

[M + H]+

706.3

Enniatin B1 (ENB1)

C34 H59 N3 O9

+

9.80

[M + NH4 ]+

671.3

+

16 11 11 11 17 14 21 13 21 18 11 14 12 5 14 12 13 11 12 12 12 42 39 20 28 27 14 17 23 21 32 33 26 13 36 24 30 31 36 37 32 27 28 36 34 56 34 34 35 36 56

76

−

281.1 311.0 265.0 295.0 229.0 247.0 175.0 305.1 185.0 215.0 231.1 203.0 261.1 279.2 297.2 307.1 247.1 349.1 263.1 215.1 323.1 334.0 352.2 275.1 160.0 187.9 305.0 215.0 185.0 275.1 160.0 130.0 238.9 358.0 220.9 175.0 131.0 160.0 336.1 318.1 354.1 196.0 214.0 244.0 262.0 134.0 210.0 228.0 209.7 228.0 100.0

Compound abbreviation

Chemical formula

Polarity

Nivalenol (NIV)

C15 H20 O7

−

Deoxynivalenol (DON)

C15 H20 O6

Fusarenon-X (FUS-X)

Retention time (min)

Beauvericin (BEA)

C45 H57 N3 O9

+

9.82

[M + NH4 ]

801.3

Enniatin A1 (ENA1)

C35 H61 N3 O9

+

9.96

[M + NH4 ]+

685.4

Enniatin A (ENA)

C36 H63 N3 O9

+

10.10

[M + NH4 ]+

699.4

polyethylene frits. Before processing samples, Carbograph-4 cartridges were attached to a vacuum manifold apparatus and washed sequentially with 5 mL of dichloromethane/methanol (80:20, v/v) containing 0.2% formic acid, 3 mL of methanol, 10 mL of 10 mmol L−1 hydrochloric acid solution, and 5 mL of Milli-Q water. Acrodisc 13 mm syringe filters with 0.2 ␮m nylon membrane (used for filtration of samples prior to the injection into the chromatographic system) were from Pall (Pall Corp., MI, USA). 2.4.2. Extraction method: SLE with clean-up step One gram of pooled ground powdered biscuit was placed into a 50 mL screw cap polycarbonate centrifuge tube, and treated with 5 mL of acetonitrile/water (80:20, v/v) containing 1% formic acid. The tube was tightly capped, vortexed for 3 min, and then placed into the ultrasonic bath at room temperature and at a frequency of 50 ± 3 Hz for 15 min. The extracts obtained were subsequently centrifuged for 15 min at 10,000 rpm, at 0 ◦ C. A 4 mL aliquot of the

64 86

99

101 101

97

96

202 137

117

137

108

136

199

148 172

148 148

supernatant was withdrawn, diluted to 250 mL with Milli-Q water and cleaned-up by a pre-conditioned Carbograph-4 SPE cartridge at a flow rate of about 20–25 mL min−1 . The bottle was washed with 50 mL of water and the washing was passed through the cartridge. Then, 300 ␮L of methanol were passed slowly (flow rate of about 5 mL min−1 ) through the cartridge to remove the residual water without eluting the analytes. Mycotoxins were eluted from the cartridge with 1 mL of methanol followed by 15 mL of dichloromethane/methanol (80:20, v/v) containing 0.2% formic acid. The eluate was collected into a 1.4 cm i.d. round-bottom glass vial, spiked with IS solution (10 ng) and evaporated to dryness at 40 ◦ C under a gentle nitrogen stream. The residue was reconstituted with 500 ␮L of methanol/water (50:50, v/v) containing 5 mmol L−1 ammonium formate, and the obtained solution was forced through a 0.2 ␮m nylon membrane syringe filter. A 5 ␮L of the final solution was analyzed by LC/ESIMS/MS.

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model CHROMA-355301; No. of Pages 10

ARTICLE IN PRESS A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

4

2.5. LC-MS/MS analysis Liquid chromatography was performed by using an Ultimate 3000 LC system (ThermoFisher Scientific, Bremen, Germany) consisted in a binary pump, equipped with a degasser, a thermostatted microwell-plate autosampler and a thermostatted column oven. A 5 ␮L aliquot of sample was injected on a Hypersil Gold (250 mm × 2.1 mm i.d., 3 ␮m particle size) equipped with a Securityguard Hypersil Gold (4 mm × 2.1 mm i.d., 5 ␮m particle size), both from Thermo Fisher Scientific. The column was maintained at 40 ◦ C. Mobile phase was water (A) and methanol (B), both with 5 mmol L−1 ammonium formate, 0.1% (v/v) formic acid, and operated at a flow rate of 300 ␮L min−1 . After an isocratic step at 50% B for 1 min, B was linearly increased to 98% within 7 min, and held constant for 7 min to rinse the column. Finally B was lowered again to 50% and the system allowed to reequilibrate. The duration of each LC run was 20 min, including reequilibration. MS detection was performed with a TSQ VantageTM triplestage quadrupole mass spectrometer (Thermo Fisher Scientific) connected with the LC system via an electrospray (ESI) source, operated both in positive and negative ionization mode by continuous positive/negative polarity switching. Fourteen mycotoxins were analyzed under positive ESI mode, while 5 compounds (i.e. NIV, DON, ␣-ZOL, ␤-ZOL, and ZEA) were analyzed under negative ESI mode (Table 1). Mass calibrations and resolution adjustments on the resolving lens and quadrupoles were automatically performed by using manufacturer solution introduced via infusion pump at 5 ␮L min−1 flow-rate, while, in order to optimize the tuning parameters for each compound, 1 ng ␮L−1 standard solutions in a suitable solvent were infused (always at 5 ␮L min−1 ). The [M + H]+ , [M + NH4 ]+ , [M − H]− and [M + HCOO]− ions were selected by the first quadrupole and fragmented in the collision cell with the appropriate collision energy (CE). From the MS/MS full-scan spectra, suitable transitions were selected for acquisition in multiple reaction monitoring (MRM) mode. The optimized MS tuning parameters independent of the ion source and chromatography selected for detection of each compound are reported in Table 1. Spray voltages of 3.2 kV for positive ionization mode and of 3.0 kV for negative ionization mode were applied. The vaporizer temperature was set to 280 ◦ C and the capillary temperature to 220 ◦ C. Sheat gas pressure, ion sweep gas pressure, and auxiliary gas pressure were set to 50, 1 and 25 (arbitrary unit), respectively. The whole LC-MS system was managed by Xcalibur software (v.2.1, Thermo Fisher Scientific). 2.6. Evaluation of method performance 2.6.1. Recovery experiments For each extraction method, recovery studies were performed. One gram of free-mycotoxin sample was placed in a flat amber glass vessel and fortified with 100 ␮L of the composite working solution diluted with 1 mL of acetone, taking care to uniformly spread it on the sample. An intimate contact between analytes and sample was obtained by mixing with a spatula for some minutes. To eliminate the organic solvent, the samples were allowed to equilibrate and air dry at 25 ◦ C in a ventilate oven for about 30 min. Then, the spiked samples were treated following the extraction procedure described above, and analyzed. Analytical recovery of each mycotoxin at any given concentration was assessed by measuring the peak area, calculating the peak area ratio relative to that of the IS, and comparing this result with that obtained for a matrix-matched solution containing the same nominal analyte quantities and the IS.

2.6.2. Calibration curve and method validation The optimized method was validated for linearity, accuracy, precision (intra- and inter-day precision) and sensitivity. Instrumental linearity was evaluated in separate experiments in which calibration lines were constructed in a concentration range wider than that used for quantification of samples. To evaluate linearity of the method, under the instrumental conditions reported in Section 2.5, two sets of calibration lines, named “standard” and “matrix-matched”, respectively, were constructed. The standard calibration curve was constructed by diluting appropriate volumes of the composite working standard solution in water/methanol (50:50, v/v), while the matrix-matched calibration curve was prepared by spiking blank sample extracts with known and appropriate volumes of the composite working standard solution. In both cases the same amount of IS (10 ng) was added. Both standard and matrix-matched solutions were prepared at six concentration levels, ranging between 0.25 and 30 ␮g kg−1 for NIV, DON and FUS-X; 0.1–12 ␮g kg−1 for FB1, FB2, HT-2, ␣-, ␤-ZOL, 3- and 15-ADON; 0.025–3 ␮g kg−1 for ZEA, NEO, DAS, T2; 0.001–0.12 ␮g kg−1 for OTA, ENA1, ENA, ENB1 and BEA. Each point of the calibration curves was prepared in duplicate and was injected in each batch starting from zero up to the highest calibration concentration. For each analyte, the combined ion current profile for the selected transitions was extracted from the LC-MRM dataset, and the peak area plot versus injected amount was obtained by measuring the resulting peak area and relating this area to that for the IS. All samples were run in triplicate and results averaged. Unweighted regression lines for standard and matrix matched calibration curves were calculated and compared. For each compound, accuracy was evaluated by the recoveries of analyte-free samples spiked at three different levels using 50, 100, 150 ␮L of the composite working standard solution (so that the compounds subject to legislation was assessed at 50%, 100% and 150% of their legislative limit) and calculated with respect to the IS. Precision was measured as within laboratory precision. Intraday precision was assessed by the relative standard deviation (RSDr ) calculated from results generated under repeatability conditions of six replicates per each concentration in a single day. Inter-day precision was calculated by the relative standard deviation (RSDR ), calculated from results generated under reproducibility conditions by three determination per concentration on 6 days. Precision was then calculated using matrix-matched calibration. Sensitivity was evaluated by method detection limit (MDL) and method quantification limit (MQL). MDL was extrapolated as the concentration of the analyte giving a signal-to-noise ratio (S/N) = 3 for the second most intense transition, MQL as the concentration of the analyte giving S/N = 10 considering the sum of all transition ion currents for each analyte. Therefore, the extrapolated concentrations were used to prepare spiked samples then analyzed to verify that the above-mentioned S/N ratios were respected.

2.6.3. Matrix effect ME was assessed for each analyte by comparing the slope of the standard calibration line (astandard ) with that of the matrixmatched calibration line (amatrix ), for the same concentration levels, according to the formula [1 − (amatrix /astandard )] × 100.

3. Results and discussion Due to the high fat content and some co-extracting interferents, a direct method for the simultaneous detection of several mycotoxins with different polarity might be very difficult to develop and not

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model CHROMA-355301; No. of Pages 10

ARTICLE IN PRESS A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

suitable for very complex matrices such as biscuits, especially for analytes present at low concentrations. In fact, although LC-MS/MS is a powerful technique and direct analysis or methods with little sample treatment are possible, results are affected by a significant ME, which could lead to lower sensitivity for some compounds and inaccuracy, also if correction for ME is done. Moreover, even though direct analysis is possible without clean-up, dirty extracts can result in progressive signal loss and subsequent need of frequent cleaning of the instrument ion source. To evaluate these statements, two extraction procedures were initially tested for extraction and direct quantification of 19 mycotoxins without any clean-up procedure: (a) QuEChERS based on partitioning via salting-out extraction, and (b) SLE. Afterwards a clean-up procedure was considered and optimized. 3.1. Evaluation of extraction procedures A modified QuEChERS procedure [23,26] was initially investigated. Briefly, 1 g of pooled ground powdered biscuit samples was placed into a 50 mL screw cap polycarbonate centrifuge tube, and treated with 5 mL of water containing 0.1% formic acid. The tube was vortexed for 3 min. Afterwards, 5 mL of acetonitrile was added shaking by vortex for 3 min. Two grams of MgSO4 plus 0.5 g of NaCl was added and the mixture shaken for 1 min and centrifuged for 15 min at 10,000 rpm, at room temperature. A 2 mL aliquot of the upper acetonitrile layer was removed and forced through a 0.2 ␮m nylon membrane syringe filter. Five ␮L of the final solution was analyzed by HPLC/ESI-MS/MS. Despite the high fat content of the matrix, primary secondary amines were not used for extract purification because fumonisins, which have an acidic nature, could be retained on the sorbent [26]. SLE was then evaluated as “dilute-and-inject” method. Briefly, 1 g of pooled ground powdered biscuit samples was placed into a 50 mL screw cap polycarbonate centrifuge tube, and treated with 5 mL of acetonitrile/water (80:20, v/v) containing 1% formic acid. The tube was tightly capped, shaken by vortex for 3 min, and then placed into the ultrasonic bath at room temperature for 15 min. The extracts obtained were then centrifuged for 15 min at 10,000 rpm, at 0 ◦ C. A 2 mL aliquot of the supernatant was withdrawn, diluted twice with water, and forced through a 0.2 ␮m nylon membrane syringe filter and 5 ␮L of the final solution was analyzed by LC/ESIMS/MS. Preliminary experiments were carried out to find the most appropriate organic solvent mixture and acidic conditions for achieving acceptable recoveries for each analyte. The extraction efficiency was tested using acetonitrile/water (80:20, v/v), acetonitrile/water (80:20, v/v) containing 0.1% acetic or formic acid or acetonitrile/water (80:20, v/v) containing 1% acetic or formic acid conditions. We focused on acetonitrile as organic phase because it is already known that using other organic solvents, such as methanol, some mycotoxins cannot be extracted or have a low recovery [32]. The increase of the percentage of formic acid improved both recoveries and peak responses of fumonisins, although it did not have such positive effect on NIV and DON, the signals of which were not always reproducible. However, we selected acetonitrile/water (80:20, v/v) with 1% formic acid as the best extraction mixture for evaluation of the direct quantification. Moreover centrifugation at 0 ◦ C was applied to precipitate part of the fat fraction present in the matrix. 3.1.1. Recovery and matrix effect QuEChERS and SLE procedure recoveries were obtained by spiking, before and after the extraction step, analyte-free samples with 100 ␮L of composite working standard solution in order to have contamination at the level of lowest limits for the analytes

5

subjected to legislation. In particular, contamination levels were 500 ␮g kg−1 for NIV, DON and FUS-X, 200 ␮g kg−1 for FB1, FB2, HT2, ␣-, ␤-ZOL, 3- and 15-ADON, 50 ␮g kg−1 for ZEA, NEO, DAS, T2, and 2 ␮g kg−1 for OTA, EnA1, EnA, EnB1 and BEA. In these studies any IS was used. In fact, due to diversity of chemical structures and properties of each analyte, finding appropriate non-isotopic IS in a multi-component analysis is very challenging. Moreover, a single IS cannot compensate the encountered MEs, as it could be different with each analyte in the considered matrix. Isotopically labeled standard could be used but the cost of multi-mycotoxins analysis would be too high. The experiments were conducted in triplicate. Due to the presence of several co-extracted contaminants, in particular a large amount of fats, a ME was evaluated. At this step ME was calculated for each mycotoxin as ME (%) = (1 − (Am )/As ) × 100), where As is the average of the peak area of mycotoxins standard in solvent and Am is the average of the peak area of mycotoxins in matrix-matched solution at the same concentrations. Recoveries and MEs for the two direct methods are shown in Fig. 1. QuEChERS procedure gave unsatisfactory low recoveries, ranging between 50 and 70% for the majority of the analyzed compounds. Moreover OTA showed a recovery of about 11% and the most polar compounds, such as NIV, DON and FUS-X, were completely lost. In addition, also a high ME was observed for all the analyzed mycotoxins. As previously reported [29] sometimes soaking dry food in water at the beginning of the extraction could help to weaken the interaction of the analyte with the matrix components and assist the extraction. To improve recoveries, biscuits were soaked in water for half an hour prior to add the organic phase. This modification, however, did not lead to any significant improvement and gave irreproducible recoveries. Due to this reason this procedure was probably not suitable for the considered matrix and then it was not longer considered. Regarding to SLE procedure, we tested the dilute-and-inject method. To reduce the percentage of the organic phase and the ME, a two-fold dilution of an aliquot of the extract was performed before the injection. As shown in Fig. 1, although this method gave satisfactory recoveries for the majority of the analyzed mycotoxins, it was not possible to calculate the recovery for NIV and DON because they showed a low peak response. Moreover, it can be seen that there is a severe ME for all the target mycotoxins (higher than 90%) and therefore this methodology was not acceptable for quantification purposes. To reduce ME and ion suppression, further dilution of the matrix can be used, therefore different diluted extracts (1:3 and 1:5) were also evaluated (data not shown), resulting in a good reduction of ME but leading to the loss of many signals. 3.2. Optimization of extraction procedure with clean-up step Based on recovery and ME results it can be said that direct injection of the extract is not a suitable method for simultaneous detection and quantification of all considered mycotoxins in matrix such as biscuits and similar products having endogenous compounds that could be co-extracted and compromise the analyte response. Then a clean-up step becomes necessary. Due to the high content of fats in biscuits, a sorbent specifically suitable for defatting, namely SupelTM QuE Z-Sep+ (500 mg in 12 mL polypropylene tube from Supelco), was initially tested to purify the extract prior to injection. This sorbent consists of both C18 and zirconia bonded to the same silica particles capable of binding fats through hydrophobic interaction, and of attracting compounds with electron donating groups. ME was quite reduced for almost all mycotoxins, but it did not allow the recovery of fumonisins, because they were retained in the sorbent structure and completely lost. Moreover, NIV and DON still did not have acceptable signal

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model CHROMA-355301; No. of Pages 10 6

ARTICLE IN PRESS A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

Fig. 1. Recoveries (upper) and matrix effects (ME – lower) for QuEChERS and SLE methods in blank samples spiked at contamination levels of 500 ␮g kg−1 for NIV, DON and FUS-X, 200 ␮g kg−1 for FB1, FB2, HT-2, ␣,␤-ZOL, 3- and 15-ADON, 50 ␮g kg−1 for ZEA, NEO, DAS, T2, and 2 ␮g kg−1 for OTA, EnA1, EnA, EnB1 and BEA.

responses, thus this defatting procedure resulted unsuitable in this study. Clean-up step was then carried out using GCB. As formerly reported [33], due to the presence of positively charged chemical heterogeneities on its surface, GCB can be considered to be both reversed-phase (RP) and anion exchangers sorbent. For this reason GCB exhibits the advantage to be an adsorbent suitable for both polar and non-polar compounds; in addition it is suitable for holding many triglycerides and phospholipids. Its employment for cleaning-up the extract before LC/ESI-MS/MS determination of mycotoxins has already been discussed in our previously published papers [4,19,20,34]. Briefly, to avoid breakthrough for the less retained compounds, 4 mL of the entire extract, added to 250 mL of water, was submitted to the clean-up procedure. As formerly reported [4,19,20,34], GCB proved to be suitable for mycotoxin application in terms of selectivity, even though its well-known loadability limitation must be taken into account [35]. Therefore, by using sorbent amount smaller than 500 mg, the most polar mycotoxins were not adequately retained (in particular NIV and DON exhibited low recoveries). were eluted from GCB using Mycotoxins dichloromethane/methanol (80:20, v/v) containing 0.2% formic acid, because it was able to elute strongly retained compounds. Moreover, the addition of formic acid allowed the recovery of compounds such as fumonisins and OTA, which are known to

establish electrostatic interactions with the GCB surface. Also in this step, to evaluate clean-up efficiency, the extract was spiked with 100 ␮L of composite working standard solution before and after the clean-up step. In this way, the clean-up effects on total recovery can be isolated from ME and evaluated by comparing the peak areas for the same compound in samples spiked ante and post clean-up step. In this procedure SMX (which has a good and stable signal response) was used as volumetric IS and added before the evaporation step. Using GCB allowed detecting and recovering all the analyzed mycotoxins. Recoveries (n = 6) were carried out spiking each sample before and after the entire procedure and the results, shown in Table 2 (level x), were compared with those obtained by the direct injection method. It can be seen that recoveries obtained using SLE without (Fig. 1) and with clean-up procedure (Table 2) were comparable for the majority of the mycotoxins, however the employ of a purification step with GCB has also allowed the identification and the subsequent quantification of the most polar compounds (NIV, DON and FUS-X); moreover, ME was statistically lower (for ME results see Table 3). After these results, the choice of a GCB material for cleaning-up the extract was due to the better selectivity, lower ME and time-savings. For comparison purpose, we chose one of the most commonly used SPE cartridge, the Oasis HLBTM cartridges (Waters, Milford, MA, USA), filled with hydrophilic–lipophilic balanced sorbents (divinylbenzene/N-vinylpiyrrolidone, 1:1) both testing 200 and

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model

ARTICLE IN PRESS

CHROMA-355301; No. of Pages 10

A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

7

Table 2 Accuracy and precision of the LC/ESI-MS/MS method for determining mycotoxins in biscuits spiked at three different concentration levels. Mycotoxin (abbreviation)

Recovery % (RSD%)

NIV DON FUS-X NEO 3-ADON 15-ADON DAS HT-2 FB1 ␤-ZOL T2 ␣-ZOL OTA A ZEA FB2 ENB1 BEA ENA1 ENA

0.5× (␮g kg−1 )

× (␮g kg−1 )

1.5× (␮g kg−1 )

70.8 (7.1) 95.3 (9.6) 105.5 (7.3) 114.9 (9.6) 105.2 (10.2) 101.2 (10.2) 98.2 (7.4) 104.4 (5.4) 88.5 (8.5) 97.3 (6.4) 104.1 (11.1) 103.0 (5.6) 76.7 (3.8) 105.1 (8.1) 89.3 (5.0) 100.6 (7.4) 81.3 (6.3) 107.1 (12.0) 99.9 (5.3)

63.9 (6.8) 88.2 (8.1) 96.6 (5.5) 89.4 (6.4) 88.9 (8.6) 89.2 (8.3) 89.6 (6.6) 96.1 (4.5) 84.7 (5.7) 95.3 (5.9) 90.2 (9.2) 96.3 (5.8) 83.2 (4.2) 95.1 (6.3) 92.6 (5.2) 99.0 (5.7) 83.7 (5.5) 92.3 (6.2) 88.4 (5.4)

63.1 (6.4) 97.7 (8.3) 95.3 (7.2) 97.9(9.0) 98.3 (8.6) 100.0 (10.3) 96.7 (9.1) 97.5 (4.3) 80.2 (5.4) 97.9 (6.3) 95.5 (9.3) 94.9 (4.6) 85.7 (3.5) 94.3 (6.5) 84.2 (6.3) 86.5 (5.3) 84.9 (6.6) 89.3 (6.4) 85.6 (3.9)

For recovery experiments, samples were spiked at three different concentration: 0.5×, ×, and 1.5× where concentration at level x was: 500 ␮g kg−1 for NIV, DON and FUS-X, 200 ␮g kg−1 for FB1, FB2, HT-2, ␣,␤-ZOL, 3- and 15-ADON, 50 ␮g kg−1 for ZEA, NEO, DAS, T2, and 2 ␮g kg−1 for OTA, ENA1, ENA, ENB1 and BEA. For each concentration level, mean recovery and RSD (in bracket) were calculated on n = 6. Regarding the mycotoxin contamination of cereals and their products, current European Union (EU) food safety legislation regulates the content of some mycotoxins [8,9]. OTA maximum level (ML) is set at 3 ␮g kg−1 , DON ML ranging between 500 and 750 ␮g kg−1 depending on the kind of cereals intended for direct human consumption; ZEA ML ranging between 50 and 75 ␮g kg−1 depending on the kind of cereals intended for direct human consumption; T-2 and HT-2 toxins are named present, but the value of their limit is not indicated. All the other considered mycotoxins, such as trichothecenes A and B, ZEA metabolites, ENs and BEA are not legislated.

500 mg of sorbent in 6 mL polypropylene tubes. Before processing samples, Oasis HLB cartridges were attached to a vacuum manifold apparatus and washed with 5 mL of dichloromethane/methanol (80:20, v/v) followed by 2 mL of methanol and 5 mL of Milli-Q water. After that, a 4 mL aliquot of the extracts was diluted to 100 mL with water and passed through the pre-conditioned Oasis HLBTM cartridge; the cartridge was then washed with 10 mL of water followed by 200 ␮L of methanol, to eliminate residual water, and finally left to dry. Analytes were eluted with 1 mL of methanol followed by 10 mL of dichloromethane/methanol (80:20, v/v), and the rest of

the procedure described for GCB was followed. Either using 200 or 500 mg of sorbent, NIV, DON and OTA were insufficiently recovered. Recovery of all other mycotoxins ranged between 50 and 80% but stronger ME was found. Moreover the entire procedure was more time consuming than the one with GCB. 3.3. Analytical condition optimization The developed method allowed the simultaneous determination of 19 mycotoxins in a single analysis using an Ultimate 3000 LC

Table 3 Validation parameters of the proposed LC/ESI-MS/MS method for selected mycotoxins in biscuits. Mycotoxin (abbreviation)

R2

a

NIV DON FUS-X NEO 3-ADON 15-ADON DAS HT-2 FB1 ␤-ZOL T2 ␣-ZOL OTA A ZEA FB2 ENB1 BEA ENA1 ENA

0.9997 0.9990 0.9993 0.9991 0.9999 0.9998 0.9927 1.0000 0.9996 0.9999 0.9987 0.9999 0.9998 0.9994 0.9995 1.0000 0.9997 0.9997 0.9998

(mean)

MEb (%)

Intra-day precisionc RSDrd (%)

Inter-day precisionc RSDR e (%)

MDLf (␮g kg−1 )

MQLg (␮g kg−1 )

37.6 35.3 23.5 12.3 18.9 24.7 11.5 13.3 9.0 4.0 9.0 −0.2 −2.0 −3.7 4.3 −1.9 0.4 4.0 11.6

5.8 7.9 6.3 7.6 8.3 8.9 7.2 4.5 5.6 5.2 8.7 4.9 3.5 6.3 4.6 5.2 5.7 6.3 4.2

8.9 9.5 8.4 9.3 9.9 9.3 8.6 5.2 6.9 9.2 9.9 9.1 5.6 8.6 6.8 6.1 7.9 8.2 5.6

42.9 10.8 8.0 0.4 2.1 4.3 1.3 1.9 1.0 3.1 0.3 2.2 0.1 0.3 0.7 0.04 0.04 0.1 0.06

80.2 41.5 16.6 1.7 6.2 6.2 1.3 4.8 3.4 24 1.2 6.0 0.2 0.7 2.1 0.04 0.05 0.1 0.06

a

Coefficient of determination of matrix-matched calibration regression line. Matrix effects (ME%): [1 − (slopematrix-matched calibration solution /slopestandard calibration solution ) × 100. c Intra- and inter-day precision were determined at medium level: 500 ␮g kg−1 for NIV. DON and FUS-X. 200 ␮g kg−1 for FB1. FB2. HT-2. ␣.␤-ZOL. 3- and 15-ADON. 50 ␮g kg−1 for ZEA. NEO. DAS. T2, and 2 ␮g kg−1 for OTA. ENA1. ENA. ENB1 and BEA. d RSDr was calculated on n = 6. e RSDR was calculated on n = 3 in six different days. f Method detection limit (S/N = 3 for the second most intense transition in MRM). g Method quantification limit (S/N = 10 for the sum of all the transitions in MRM). b

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model CHROMA-355301; No. of Pages 10 8

ARTICLE IN PRESS A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

Fig. 2. LC-MRM chromatograms of a blank sample spiked with the analyzed mycotoxins at contamination levels of 500 ␮g kg−1 for NIV, DON and FUS-X, 200 ␮g kg−1 for FB1, FB2, HT-2, ␣,␤-ZOL, 3- and 15-ADON, 50 ␮g kg−1 for ZEA, NEO, DAS, T2, and 2 ␮g kg−1 for OTA, EnA1, EnA, EnB1 and BEA. 1: NIV; 2: DON; 3: FUS-X: 4: NEO; 5: 3- and 15-ADON; 6: DAS; 7: HT-2; 8: FB1; 9: ␤-ZOL; 10: T2; 11: ␣-ZOL; 12: OTA A; 13: ZEA; 14: FB2; 15: ENB1; 16: BEA; 17: ENA1; 18: ENA

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model CHROMA-355301; No. of Pages 10

ARTICLE IN PRESS A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

system coupled to a TSQ VantageTM triple-stage quadrupole mass spectrometer. In water/methanol or water/acetonitrile mixtures, both 0.1% (v/v) formic acid, and operating in positive mode, the sodiated adduct prevailed on proton adduct formation. The addition of ammonium formate reversed this behavior in-source and the signals for [M + H]+ or [M + NH4 ]+ increased. Also in negative mode the use of this modifier gave a good signal. Additionally, all mycotoxins gave better signals in a methanolbased mobile phase than in acetonitrile, therefore the former was used. As a consequence, LC mobile phase with water (A) and methanol (B), both at the concentration of 5 mmol L−1 ammonium formate, 0.1% (v/v) formic acid, was chosen to enhance sensitivity while reducing the sodium adduct [M + Na]+ formation. The temperature of the chromatographic column was set at 40 ◦ C to achieve better quantification limits. Five microlitres were injected, because larger injection volumes gave rise to peak broadening as well as stronger signal suppression, especially for the most polar analytes. The mass-chromatogram of all the examined mycotoxins is shown in Fig. 2. Each peak is relative to the sum of the selected transitions for each mycotoxin in an analytes-free biscuit sample spiked with 100 ␮L of composite standard working solution. The 3- and 15- ADON had the same retention time, however, we found different MS/MS product ions that could be separately considered for quantification purposes.

3.4. Method validation Method validation was performed in terms of linearity, accuracy, intra- and inter- day precision, MDL and MQL for all analyzed mycotoxins. Under the instrumental conditions reported in Section 2, the linear dynamic range was estimated for all the analytes for both standard and matrix-matched calibration curves. The calibration curves were created from six concentration levels. Each point was prepared in duplicate and injected in triplicate starting from zero up to the highest calibration concentration. Coefficients of determination R2 of matrix-matched calibration regression line, shown in Table 3, were all in the range from 0.9927 to 1, showing that mycotoxin analytical responses were linear over the studied ranges. Linearity of the method was also tested by using appropriate volumes of the composite working standard solution added to six aliquots of free-mycotoxin biscuit sample to obtain the same six concentration levels in the considered range described above, and submitting the samples to the whole procedure. The obtained intercepts did not differ significantly from that of matrixmatched calibration line obtained as described above, and, also considering the data on ME shown in Table 3, matrix-matched calibration was used for quantification purposes. Accuracy was evaluated from the analytical recovery in which biscuits were spiked at three concentration levels (so, basing on European Commission Decision 2002/657/EC, the compounds subject to legislation was assessed at 50%, 100% and 150% of the their legislative limit). Six replicates were prepared for each experiment in accordance with EU guidelines [8]. Spiked concentration (␮g kg−1 ), recovery and RSD are shown in Table 2. Recovery ranged between 63.1 and 107.1% with a RSD below 12.0%. Precision was evaluated as within laboratory precision. The RSDs (RSDr and RSDR ) were determined for data obtained by spiking biscuits at one level (100% legislative limit) repeated on six days. Data are reported in Table 3 and results are conformed to the requirements of Regulation EC 401/2006 [36]. Sensitivity was estimated by MDL and MQL. When using a MS detector, the first condition that has to be satisfied for ascertaining the targeted compound presence is that the precursor ion and

9

at least two product ions (i.e., two MRM transitions) produce signals distinguishable from the background ion current. According to this, for each analyte, MDLs and MQLs were estimated from the LC/ESI-MS/MS MRM chromatogram analysing spiked samples, at a concentration close to 3 and 10 times signal to noise, respectively, and correcting the obtained values for calculated procedure recovery (see Table 3). The ion currents of selected transitions relative to each compound were extracted from dataset, and the resulting traces were smoothed by applying the automatic processing smoothing method (Xcalibur) using Gaussian type and 7 points. Then, always applying the automatic processing, the peak height-to-averaged background noise ratio (S/N) was measured. The background noise estimate was based on the peak-to-peak baseline near the analyte peak. MDL was extrapolated as the concentration of the analyte giving a signal-to-noise ratio (S/N) = 3 for the second most intense transition, MQL as the concentration of the analyte giving S/N = 10 considering the sum of all transition ion currents for each analyte.

3.5. Real sample analysis The developed method was applied to different types of biscuit for a total of 20 samples from widespread brands. The results show that 2 out of 10 samples (about 20% of incidence) were contaminated with all the 4 ENs, but they were found in the range between MDL and MQL. All the other mycotoxins were not detected in all the analyzed samples. Although some studies showed extensive contamination of cereal grains with enniatins [12 and its related references] few information are available on the contamination of biscuits. To the best of the authors’ knowledge, the number of publications regarding mycotoxin in biscuits is still very limited, and nowadays there is only one work on this matrix [27]. According to our results, also Beltrán et al. [27] did not detect any of the analyzed mycotoxins in biscuit samples. However, according with Malachova et al. [12], Beltrán et al. [27] found mycotoxins in the other analyzed bakery products (wheat, bread, sliced bread, cereals bread, pasta, breakfast cereals and processed baked goods). As a whole, these results show how mycotoxin contamination is strongly dependent on the matrix, the type of ingredients and the processing techniques used for their production (including cooking).

4. Conclusion The LC-MS/MS method developed in this work, specifically studied for confirmatory analysis purpose of 19 mycotoxins in biscuits, fulfilled EC Commission Decision 2002/657 [37] and the current EU food safety legislation [8,9]. An important advantage of the proposed method is the clean-up for target mycotoxins using an alternative adsorbent instead of expensive immunoaffinity columns. Considering the current trend of analysis of multiple food contaminants while maintaining high throughput and simple sample preparation, direct analysis of the sample might seem the most suitable option. However, in this case, direct injection of the matrix in the chromatographic column was not feasible because of its very high complexity. Direct injection also provided poor detectability of target analytes due to high matrix interference. In addition to these limitations, direct injection also lowers the analytical column lifetime and ion source contamination rapidly occurs. The proposed method, thanks to the properties of GCB, overcame severe ME problems and allowed the identification and quantification of all analyzed mycotoxins in a very complex matrix.

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009

G Model CHROMA-355301; No. of Pages 10 10

ARTICLE IN PRESS A.L. Capriotti et al. / J. Chromatogr. A xxx (2014) xxx–xxx

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma. 2014.04.009. References [1] P. Zollner, B. Mayer-Helm, J. Chromatogr. A 1136 (2006) 123. [2] A.L. Capriotti, G. Caruso, C. Cavaliere, P. Foglia, R. Samperi, A. Laganà, Mass Spectrom. Rev. 31 (2012) 466. [3] L.B. Bullerman, A. Bianchini, Int. J. Food Microbiol. 119 (2007) 140. [4] A. Bacaloni, C. Cavaliere, F. Cucci, P. Foglia, R. Samperi, A. Laganà, J. Chromatogr. A 1179 (2008) 182. [5] European Food Safety Authority (EFSA). Available from: http://www.efsa. europa.eu/en/topics/topic/mycotoxins.htm and http://www.efsa.europa.eu/ en/dataclosed/call/datex101020b.htm [6] G.S. Shepard, Food Addit. Contam. 25 (2008) 146. [7] J.L. Richard, Int. J. Food Microbiol. 119 (2007) 3. [8] Regulation 1881/2006, Off. J. Eur. Commun. L 364 (2006) 5. [9] Commission Regulation 1126/2007, Off. J. Eur. Union L 255 (2007) 14. [10] C. Cavaliere, G. D’Ascenzo, P. Foglia, E. Pastorini, R. Samperi, A. Laganà, Food Chem. 92 (2005) 559. [11] C. Juan, J. Manes, A. Raiola, A. Ritieni, Food Chem. 140 (2013) 755. [12] A. Malachova, Z. Dzuman, Z. Veprikova, M. Vaclavikova, M. Zachariasova, J. Hajslova, J. Agric. Food Chem. 59 (2011) 12990. [13] M.E.J. Pronk, R.C. Schothorst, H.P. van Egmond, RIVM Report 388802024/2002, 2002, Available from: http://rivm.openrepository.com/rivm/bitstream/10029/ 9184/1/388802024.pdf [14] N.A. Lee, S. Wang, R.D. Allan, I.R. Kennedy, J. Agric. Food Chem. 52 (2004) 2746. [15] K.R. Reddy, N.I. Farhana, B. Salleh, J. Food Sci. 76 (2011) 99. [16] M. Sulyok, R. Krska, R. Shumacher, Anal. Bioanal. Chem. 389 (2007) 1505. [17] M.C. Spanjer, P.M. Rensen, J.M. Scholten, Food Addit. Contam. 25 (2008) 472.

[18] A.L. Capriotti, C. Cavaliere, S. Piovesana, R. Samperi, A. Laganà, J. Chromatogr. A 1268 (2012) 84. [19] A. Faberi, P. Foglia, E. Pastorini, R. Samperi, A. Laganà, Rapid Commun. Mass Spectrom. 19 (2005) 275. [20] C. Cavaliere, P. Foglia, E. Pastorini, R. Samperi, A. Laganà, Rapid Commun. Mass Spectrom. (2005) 2085. [21] A.M. Cheraghali, H. Yazdanpanah, N. Doraki, G. Abouhossain, M. Hassibid, S. Aliabadi, M. Aliakbarpoor, M. Amirahmadi, A. Askarian, N. Fallah, T. Hashemi, M. Jalali, N. Kalantari, E. Khodadadi, B. Maddah, R. Mohit, M. Mohseny, Z. Phaghihy, A. Rahmani, L. Setoodeh, E. Soleimany, F. Zamanian, Food Chem. Toxicol. 45 (2007) 812. [22] J. Stroka, R. van Otterdijkb, E. Anklama, J. Chromatogr. A 904 (2000) 251. [23] J. Rubert, Z. Dzuman, M. Vaclavikova, M. Zachariasova, C. Soler, J. Hajslova, Talanta 99 (2012) 712. [24] A. Garrido-Frenich, R. Romero-González, M.L. Gómez-Pérez, J.L. Martínez-Vidal, J. Chromatogr. A 1218 (2011) 4349. [25] R. Romero-González, J.L. Martínez-Vidal, M.M. Aguilera-Luiz, A. GarridoFrenich, J. Agric. Food Chem. 57 (2009) 9385. [26] M. Zachariasova, O. Lacina, A. Malachova, M. Kostelanska, J. Poustka, M. Godula, J. Hajslova, Anal. Chim. Acta 662 (2010) 51. ˜ [27] E. Beltrán, M. Ibánez, T. Portolés, C. Ripollés, J.V. Sancho, V. Yusà, S. Marín, F. Hernández, Anal. Chim. Acta 783 (2013) 39. [28] E. Varga, T. Glauner, F. Berthiller, R. Krska, R. Schuhmacher, M. Sulyok, Anal. Bioanal. Chem. 405 (2013) 5087. [29] P. Yogendrarajah, C. Van Poucke, B. De Meulenaer, S. De Saeger, J. Chromatogr. A 1297 (2013) 1. [30] Commission Recommendation 576/2006, Off. J. Eur. Union L 229 (2006) 7. [31] Commission Decision 657/2002, Off. J. Eur. Union L 221 (2006) 8. [32] C. Juan, A. Ritieni, J. Manes, Food Chem. 134 (2012) 2389. [33] C. Crescenzi, A. Di Corcia, G. Passariello, R. Samperi, M.I. Turnes Carou, J. Chromatogr. A 733 (1996) 41. [34] A. Laganà, R. Curini, G. D’Ascenzo, I. De Leva, A. Faberi, E. Pastorini, Rapid Commun. Mass Spectrom. 17 (2003) 1037. [35] J.P. Antignac, B. Le Bizec, F. Monteau, F. Andre, Anal. Chim. Acta 483 (2003) 325. [36] Commission Regulation 401/2006, Off. J. Eur. Commun. L 70 (2006) 12. [37] Commission Decision 2002/657/EC, Off. J. Eur. Union L 221 (2002) 8.

Please cite this article in press as: A.L. Capriotti, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.04.009