Embed Size (px)
Citation preview
1
1
2
3
Development and validation of a UPLC‐MS/MS method for the simultaneous 4
determination of free and conjugated Alternaria toxins in cereal‐based foodstuffs 5
Walravens J1*, Mikula H2, Rychlik M3, Asam S4, Njumbe Ediage E1, Diana Di Mavungu J1, Van 6
Landschoot A5, Vanhaecke L6 and De Saeger S1 7
1 Department of Bioanalysis, Laboratory of Food Analysis, Ghent University, Harelbekestraat 8
72, 9000 Ghent, Belgium 9
2 Institute of Applied Synthetic Chemistry, Vienna University of Technology, Getreidemarkt 9, 10
1060 Vienna, Austria 11
3 BIOANALYTIK Weihenstephan, ZIEL Research Center for Nutrition and Food Sciences, 12
Technische Universität München, Alte Akademie 10, 85354 Freising, Germany 13
4 Chair of Analytical Food Chemistry, Technische Universität München, Alte Akademie 10, 14
85354 Freising, Germany 15
5 Faculty of Bioscience Engineering, Laboratory of Biochemistry and Brewing, Ghent 16
University, Valentin Vaerwyckweg 1, 9000 Ghent, Belgium 17
6 Department of Veterinary Public Health and Food Safety, Laboratory of Chemical Analysis, 18
Ghent University, Salisburylaan 133, 9820 Merelbeke, Belgium 19
*Corresponding author: email: [email protected] 20
21
© 2014. This manuscript version is made available under the CC‐BY‐NC‐ND 4.0 license 22
http://creativecommons.org/licenses/by‐nc‐nd/4.0/ 23
24
Published in:
Journal of Chromatography A, 1372 (2014) 91–101
The final publication is available at Elsevier via https://dx.doi.org/10.1016/j.chroma.2014.10.083
2
ABSTRACT 25
A UPLC‐ESI+/‐‐MS/MS method for the simultaneous determination of free (alternariol, 26
alternariol monomethyl ether, altenuene, tenuazonic acid, tentoxin, altertoxin‐I) and 27
conjugated (sulphates and glucosides of alternariol and alternariol monomethyl ether) 28
Alternaria toxins in cereals and cereal products (rice, oat flakes and barley) was developed. 29
Optimization of the sample preparation protocol was achieved through experimental design, 30
using full factorial design for extraction solvent composition optimization and fractional 31
factorial design to identify the critical factors in the protocol. Subsequently, the method, 32
applying isotopically labelled internal standards ([2H4]‐alternariol monomethyl ether and 33
[13C6,15N]‐tenuazonic acid), was validated for several parameters such as linearity, apparent 34
recovery, limit of detection and quantification, precision, measurement uncertainty and 35
specificity (in agreement with the criteria mentioned in Commission Regulation N° 36
401/2006/EC and Commission Decision N° 2002/657/EC). During validation, quality of the 37
bioanalytical data was improved by counteracting the observed heteroscedasticity through 38
the application of weighted least squares linear regression (WLSLR). Finally, 24 commercially 39
available cereal‐based foodstuffs were analyzed, revealing the presence of tenuazonic acid 40
in both rice and oat flake samples (<LOQ ‐ 68 ± 7 µg.kg‐1) and tentoxin in rice samples (<LOQ 41
‐ 10.9 ± 2.0 µg.kg‐1). 42
43
KEYWORDS: Alternaria, (masked) mycotoxins, Ultra high performance liquid 44
chromatography/tandem mass spectrometry (UPLC‐MS/MS), survey. 45
46
3
1. INTRODUCTION 47
Mycotoxins are naturally occurring toxic secondary metabolites produced by fungi. The most 48
important mycotoxin producing fungi are Aspergillus, Penicillium, Fusarium and Alternaria.1–3 49
The fungal genus Alternaria contains numerous species that can contaminate a wide variety 50
of crops in the field and cause post‐harvest decay of various fruits, grains, and vegetables. In 51
addition to causing economic losses, several of these species (of which A. alternata is 52
regarded as the most predominant one) are able to produce a number of mycotoxins with 53
toxic properties, including alternariol (AOH), alternariol monomethyl ether (AME), altenuene 54
(ALT), altertoxins (ATX), tenuazonic acid (TeA) and tentoxin (TEN). TeA is toxic to several 55
animal species, e.g. mice, chicken and dogs. AOH, AME, ALT and ATX‐I are not acutely toxic. 56
There are several reports on the mutagenicity and genotoxicity of AOH and AME. Also, AOH 57
has been identified as a topoisomerase I and II poison which might contribute to the 58
impairment of DNA integrity in human colon carcinoma cells.4 Natural occurrence of 59
Alternaria toxins in food in European countries has been previously reviewed by Bottalico 60
and Logrieco5, Scott6, Ostry 4 and Fernandez‐Cruz et al.2. The highest levels of AOH and AME 61
were found in lentils, oilseeds, tomato paste, followed by fruit and vegetable juices, wines, 62
cereals and vegetables (tomatoes and carrots). The highest concentration of TeA was 63
reported for cereals, followed by commercial beers7. Recent data (Call for scientific data on 64
mycotoxins, EFSA, 2010) demonstrated that AOH, AME and TeA mainly occur in cereal (oats, 65
rice, barley and wheat), in tomato (products), in fruit and vegetable juices, in wines (AOH, 66
AME) and in beer (TeA).8 Maximum levels of Alternaria mycotoxins reported in marketed 67
products were in the range of 1‐103 μg.kg‐1, whereas higher levels were found in samples 68
visibly infected by Alternaria rot, i.e. in products obviously not suitable for human 69
consumption.4 70
4
There are currently no guideline limits set for Alternaria mycotoxins by regulatory 71
authorities. At the request of the European Commission, The European Food Safety Agency 72
(EFSA) provided a scientific opinion on the risks for animal and public health related to the 73
presence of Alternaria toxins in feed and food.8 Risk assessment results proved to be 74
inconclusive due to limited representatitive occurrence and toxicity data. Nevertheless, 75
application of the threshold of toxicological concern (TTC) approach9–11 indicated that there 76
might be a possible risk for human health related to the presence of Alternaria toxins in 77
foodstuffs.8 Additionally, mycotoxins can as other xenobiotics partly be metabolised, e.g. the 78
formation of conjugated toxins in living plants.12,13 In a recent definition, for structurally 79
altered mycotoxins the term “modified” has been suggested with reserving the historical 80
term "masked" mycotoxins to conjugates generated by plants 14. In the latter, the toxin is 81
usually bound to a more polar substance such as glucose, and these compounds are of 82
concern as it has been shown that they are able to release their native precursors after 83
enzymatic hydrolysis in the digestive tract of organisms (zearalenone‐4‐sulphate (ZEN4S) to 84
zearalenone (ZEN) in swine15). Historically, they are referred to be “masked” mycotoxins 85
because they have the ability to escape established analytical methods due to both 86
differences in polarity between the native precursors and their metabolites and the 87
unavailability of reference standards, and their alledged contribution to the total mycotoxin 88
content.14 In the last 2 decades, several conjugated metabolites, such as deoxynivalenol‐3‐89
glucoside (DON3G) and ZEN4S, were determined in various cereals and plants12,16–18, and 90
these observations have marked a shift towards multi‐mycotoxin methods in which toxin 91
conjugates are also included [19, 20]. With regards to Alternaria toxins, Ostry4 stated that 92
the occurrence of glucoside conjugates of AOH and AME in foodstuffs is a distinct possibility. 93
5
Therefore, it is of paramount importance to develop analytical methodologies to detect and 94
quantify these “masked” conjugates in various matrices. 95
In this research article, the development and validation of reliable, sensitive and selective 96
UPLC‐ESI‐MS/MS methods for the simultaneous determination of free and conjugated 97
Alternaria toxins in cereals and cereal based foodstuffs (rice, oats and barley) is reported. 98
Additionally, the occurrence of these (masked) toxins was assessed in a number of 99
commercially available rice and oat flakes samples. To the best of our knowledge, this is the 100
first analytical methodology aiming at the detection and quantification of both free and 101
conjugated (sulphates and glucosides of AOH and AME) Alternaria toxins in any food matrix 102
whatsoever. 103
104
2. MATERIALS AND METHODS 105
2.1 Reagents and chemicals 106
AOH (1 mg, solid standard) was supplied by Fermentek (Jerusalem, Israel) and dissolved in 1 107
mL of methanol:dimethylformamide (60:40, v/v). Certified reference standards of AME 108
(100.6 µg, dried down), TeA (100.0 µg, dried down) and TEN (102.0 µg, dried down) were 109
purchased from Biopure Referenzsubstanzen GmbH (Tulln, Austria). TeA and TEN were 110
reconstituted in 1 mL of acetonitrile, whereas AME was reconstituted in 1 mL of 111
methanol:dimethylformamide (60:40, v/v). ALT (1 mg.mL‐1, in methanol) was purchased 112
from the Institut für Organische Chemie (KIT, Karlsruher Institut für Technologie), whereas 113
ATX‐I (200 µg.mL‐1, in acetonitrile) was kindly provided by Dr. Michele Solfrizzo (ISPA‐CNR, 114
Bari, Italy). Reference standards of conjugated Alternaria toxins (AOH‐3‐sulphate [AOH3S], 115
AOH‐3‐glucoside [AOH3G], AME‐3‐glucoside [AME3G], AME‐3‐sulphate [AME3S]) were 116
synthesized and provided by the Institute of Applied Synthetic Chemistry (University of 117
6
Technology, Vienna, Austria). Stock solutions of AOH3S (1 mg.mL‐1) and AME3S (1 mg.mL‐1) 118
were prepared in methanol, whereas stock solutions of AOH3G (1 mg.mL‐1) and AME3G (1 119
mg.mL‐1) were prepared in acetonitrile. Stock solutions in acetonitrile were stored at 4°C, 120
whereas stock solutions in other solvents were stored at ‐18°C. The isotopically labeled 121
internal standards [2H4]‐AME and [13C6,15N]‐TeA were synthesized and provided by the Chair 122
of Analytical Food Chemistry (Technische Universität München (TUM), Freising, Germany). 123
Stock solutions of 40 µg.mL‐1 were prepared in methanol and stored at ‐18 °C. 124
Working solutions of TEN, TeA, ATX‐I, AOH3G and AME3G (10 µg.mL‐1) were prepared in 125
acetontrile and stored at 4 °C; whereas working solutions of AOH, AME, ALT, AOH3S and 126
AME3S were prepared in methanol and stored at ‐18 °C. All working solutions were renewed 127
monthly. Working solutions of [2H4]‐AME and [13C6,15N]‐TeA (1 µg.ml‐1) were prepared in 128
methanol and stored at ‐18 °C. 129
Water was obtained from a Milli‐Q® SP Reagent water system from Millipore Corp. (Brussels, 130
Belgium). Methanol and acetonitrile (both absolute, LC‐MS grade) and glacial acetic acid 131
(ULC/MS) were purchased from BioSolve BV (Valkenswaard, The Netherlands), whereas 132
methanol, acetonitrile and n‐hexane (all HiPerSolv Chromanorm HPLC grade) were obtained 133
from VWR International (Zaventem, Belgium). Acetic acid (glacial, 100%), ammonium acetate 134
and ammonium carbonate were supplied by Merck (Darmstadt, Germany). Acetone (99.5+% 135
for analysis) was purchased from Acros Organics (Geel, Belgium). Disinfectol® (denaturated 136
ethanol with 5% ether) was supplied from Chem‐Lab NV (Zedelgem, Belgium). 137
2.2 Synthesis of masked Alternaria mycotoxins: sulfates and glucosides of AOH and AME 138
Following the procedure recently reported by Mikula et al.21, sulfates and glucosides of AOH 139
and AME were prepared by total synthesis. Starting from commercially available compounds, 140
the synthetic intermediates alternariol‐7,9‐dimethyl ether (ADME) and alternariol‐7,9‐141
7
dibenzyl ether (ADBE) were obtained applying palladium‐catalyzed C‐H‐activation as key step 142
(Figure 1a). Furthermore, the mono‐silylated AOH derivative ASE was prepared by silylation 143
and subsequent demethylation of ADME. Glucosylation of all intermediates was performed 144
using Königs‐Knorr conditions (Aceto‐bromo‐α,D‐glucose, Ag2O) followed by deprotection of 145
acetyl, benzyl and/or silyl groups, respectively, to afford the desired conjugates AOH‐9‐146
glucoside, AOH‐3‐glucoside and AME‐3‐glucoside (Figure 1b). Sulfation of ADME and ADBE 147
using the sulfuryl imidazolium salt SIS followed by deprotection of benzyl groups or the 7‐148
methyl group, respectively, and subsequent cleavage of the 2,2,2‐trichoroethyl group of the 149
sulfate moiety led to the successful preparation of AOH‐3‐sulfate and AME‐3‐sulfate (Figure 150
1c), both as ammonium salts. All compounds were obtained in milligram quantities as white 151
solids, characterized by NMR spectroscopy and HPLC and stored under argon at 20°C. 152
2.3. Synthesis of labeled alternariol monomethyl ether ([2H4]‐AME) and tenuazonic acid 153
([13C6,15N]‐TeA). 154
Deuterium‐labeled AME was synthesized by palladium‐catalyzed protium‐deuterium 155
exchange from the unlabeled toxin. After stirring an aliquot of AME in a mixture of [2H8]‐156
dioxane and deuterium oxide for 4 days at 160 °C in the presence of palladium on barium 157
sulfate, a mixture of isotopologues [2H3‐7]‐AME was recovered, of which [2H4]‐AME was the 158
predominant form (59 %). Further details about the synthesis and the nuclear magnetic 159
resonance spectroscopic and mass spectrometric characterization of the internal standard 160
can be found elsewhere.22 161
The internal standard [13C6,15N]‐TeA was synthesized from [13C6,
15N]‐isoleucine by 162
Dieckmann intramolecular cyclization after acetoacetylation with diketene. This three‐step 163
synthesis involved the preparation of the methyl ester of [13C6,15N]‐isoleucine with 164
thionylchloride in methanol, the subsequent N‐acetoacetylation of the protected amino acid 165
8
with the diketene‐releasing reagent 2,2,6‐trimethyl‐4H‐1,3‐dioxin‐4‐one in toluene, and the 166
Dieckmann intramolecular cyclization with sodium methoxide in methanol to yield labeled 167
TeA together with its 5R epimer. Further details about the synthesis and the characterization 168
of the internal standard can be found elsewhere.23 169
2.4 LC‐MS/MS methodology 170
Analysis was performed on a Waters Acquity UPLC ‐ Quattro Premier XE mass spectrometer 171
(Waters, Milford, MA, USA), equipped with MassLynx and QuanLynx® version 4.1. software 172
(Micromass, Manchester, UK) for data acquisition and processing. 173
Chromatographic separation was achieved using an Acquity UPLC HSS T3 (1.8 µm, 2.1 x 100 174
mm) column (Waters, Milford, MA, USA). Column and autosampler temperature were set at 175
35°C and 10°C, respectively. A mobile phase consisting of eluents A [ultrapure water/acetic 176
acid (99:1, v/v)] and B [acetonitrile/acetic acid (99:1, v/v)] was used at a flow rate of 0.4 177
ml.min‐1. A gradient elution was applied as follows: 95% mobile phase A for 0.5 min, 178
followed by a linear increase from 31 to 38% mobile phase B between 0.6 and 4.25 min. An 179
isocratic phase of 90% mobile phase B was initiated at 4.35 min and maintained for 0.85 min, 180
after which column reequilibration took place, resulting in a total run time of 7 min. 181
In order to achieve optimal sensitivity and selectivity of the MS conditions, data acquisition 182
was performed applying multiple reaction monitoring (MRM) with an optimized dwell time 183
(0.07 to 0.25 s). For each target analyte, two precursor to product ion transitions were 184
determined. In accordance with Commision Decision (EC) N° 657/2002, laying down the 185
performance criteria of analytical methods, the use of two transitions allows determination 186
of the ratio between these two transitions (relative ion intensity). This ratio, together with 187
the relative retention time and signal‐to‐noise ratio (S/N), allows the confirmation of the 188
identity of the detected compound.23 MS/MS instrumental parameters were optimized via 189
9
direct infusion (flow rate of 10 µL.min‐1) of a tune solution (10 ng.µL‐1 of each analyte, 190
dissolved in methanol/ultrapure water (50/50; v/v, 0.1% formic acid). MRM transitions, the 191
optimum cone voltages and collision energies selected for each transition are given in Table 192
1, as well as the indicative retention times on the column. The first transition, which 193
corresponds to the most abundant product ion was used for quantification, and the second 194
one for confirmation purposes. 195
The mass spectrometer was operated both in the positive (ESI+) and negative (ESI‐) 196
electrospray ionization mode. Instrumental parameters were set as follows: ESI source block 197
and desolvation temperatures: 125 and 350°C, respectively; capillary voltage: 3.2 kV (ESI+) 198
and 2,95 kV (ESI‐); argon collision gas pressure: 9.1E‐3 bar; cone nitrogen gas flow and 199
desolvation gas flow: 50 and 800 L.h‐1, respectively. 200
2.5. Collection of samples of commercial foodstuffs 201
A total of 24 commercially available cereal based food products (rice, n=16 and oat flakes, 202
n=8) were obtained from several supermarkets in Belgium between February and July 2013. 203
In accordance with Commission Regulation (EC) N° 2006/401, laying down the method of 204
sampling for the official control of the maximum levels established for aflatoxins, ochratoxin 205
A and Fusarium toxins in cereals and cereal products, the weight of the aggregate sample 206
(which represents the combined total of all the incremental samples) at retail stage must be 207
at least 1 kg.24 Therefore, in this case, several retail units were combined to obtain a total 208
sample size of at least 1 kg. After thorough homogenization and grinding, a 2,5000g (± 209
0,0025) laboratory sample was stored (4°C) until further analysis. 210
2.6. Sample preparation and extraction methodology 211
Rice, oat flakes and barley samples were thoroughly ground using the M20‐grinder (Ika 212
Werke, Staufen, Germany). After each milling step, the device was thoroughly cleaned and 213
10
decontaminated using water, soap and disinfectol. Before weighing, the ground material was 214
vigorously homogenized. 215
Homogenized sample (2,5000g ± 0,0025) was spiked with labeled internal standards [2H4]‐216
AME and [13C6,15N]‐TeA at concentrations of 12 µg.kg‐1 and 8 µg.kg‐1, respectively. After 20 217
sec of vortex mixing, the samples were allowed to equilibrate in the dark for 15 min. 218
Samples were extracted for 60 min with 10 mL of extraction solvent 219
(acetonitrile/water/acetic acid [79/19.5/1.5, v/v/v]), combined with a hexane defatting step 220
using an Agitelec overhead shaker (J. Toulemonde & Cie, Paris, France). Sample extracts 221
were centrifuged (12 min, 3000g) and after removal of the upper hexane layer, the 222
supernatant was filtered through a paper filter (MN 617, Ø 110mm, Machery Nagel, 223
Germany). Subsequently, the filtrate was evaporated to dryness under a gentle nitrogen 224
stream at 40°C using an evaporator module (Grant Instruments Ltd, Cambridge, UK). Finally, 225
the residue was redissolved in 100 μL of injection solvent, consisting of mobile phase A and 226
mobile phase B (70/30, v/v), vigorously vortexed and subjected to ultracentrifugation 227
(Ultrafree®‐MC centrifugal device; Millipore, Bedford, MA, USA) for 10 min at 10000g prior 228
to LC‐MS/MS analysis. An LC‐MS/MS chromatogram of a spiked (40 µg.kg‐1) oat flakes 229
sample is shown in Figure 2. 230
2.7. Method validation 231
Due to the unavailability of certified reference material, optimization and validation of the 232
multi‐mycotoxin analytical methods for cereal (barley) and cereal products (rice, oat flakes) 233
were performed using spiked blank samples. Performance criteria that were in compliance 234
with Commission Decision (EC) N° 2002/657 and Commission Regulation (EC) N° 401/2006 235
were assessed for the following validation parameters: specificity, linearity, limit of 236
detection (LOD), limit of quantification (LOQ), apparent recovery, repeatability (intraday 237
11
precision; RSDr), reproducibility (intermediate precision; RSDR) and expanded measurement 238
uncertainty.24,25 Limit of detection (LOD) and limit of quantification (LOQ) were assessed 239
according to ICH harmonized guidelines [26]. All validation parameters were calculated using 240
the response which is calculated as the ratio of the peak area of the analyte to the peak area 241
of the isotopically labeled internal standards ([13C6,15N]‐TeA was used for AOH3G, ALT, 242
AME3G, TeA, ATX‐I, AOH, TEN and AME3S, whereas [2H4]‐AME was used for AME). 243
2.7.1. Specificity, matrix effect, linearity, LOD and LOQ 244
For every matrix, specificity and selectivity were tested by the analysis of 10 representative 245
blank samples and 3 blank samples fortified with chemically related compounds (fumonisin 246
B1 and deoxynivalenol at 100 µg.kg‐1, and ochratoxin A and aflatoxin B1 at 20 µg.kg‐1). 247
Matrix effects were assessed according to Sulyok et al.17: for each analyte in each 248
investigated matrix, calibration curves in neat solvent were constructed by plotting the 249
signal intensity versus the analyte concentration. Likewise, signal intensities obtained from 250
spiking the samples before and after extraction were plotted against the actual spiking levels. 251
Subsequently, using the slopes of the resulting functions, the efficiency of the extraction 252
step (EE) and the signal suppression/enhancement (SSE) due to matrix effects were 253
calculated. 254
Linearity was evaluated using matrix matched calibration (MMC) curves, by spiking blank 255
samples at six concentration levels (5, 10, 20, 40, 60 and 80 µg.kg‐1) for the three different 256
matrices. In addition to the commonly reported regression coefficients (R2), a lack‐of‐fit test 257
(IBM SPSS 21) was also performed to assess the adequacy of the linear model. Furthermore, 258
it is recognized that the condition of equal variances, termed homoscedasticity 259
(homogeneity of variance), should be tested in any linear regression analysis. When the 260
assumption of homoscedasticity is not met for analytical data, an effective way to 261
12
counteract the greater influence of the greater concentrations on the fitted regression line is 262
to use weighted least squares linear regression (WLSLR).27–29 263
LOD and LOQ were determined using MMC curves (in triplicate), by spiking blank samples at 264
8 concentrations levels (0.1 to 10 µg.kg‐1) for each investigated matrix. For every analyte in 265
each matrix, the standard error of the y‐intercept, as well as the slope of the corresponding 266
calibration curve (lower level equals concentration for which S/N ≥ 3 for both product ions, 267
and upper concentration level equals 10 µg.kg‐1) were calculated using the linest function 268
(Microsoft Excel 2010). LOD equaled the concentration corresponding to the blank response 269
plus three times the standard error of the y‐intercept, whereas LOQ equals two times the 270
LOD value. 271
2.7.2 Apparent recovery, repeatability, reproducibility and measurement uncertainty 272
Apparent recovery, repeatability, intermediate precision and expanded measurement 273
uncertainty were all calculated from a combined sample plan, which comprised of the 274
analysis of blank samples of each matrix spiked in triplicate with the different mycotoxins at 275
concentration levels of 5, 40 and 80 µg.kg‐1 on three consecutive days. 276
The apparent recovery was determined using MMC curves (six concentration levels, 5 to 80 277
µg.kg‐1). The measured concentration of the spiked blank samples (for each matrix) was 278
determined by plotting the observed signal (expressed as the relative peak area) into the 279
corresponding MMC curves. IUPAC defines the apparent recovery as the ratio of the 280
predicted value obtained from the MMC curves divided by the actual/theoretical value.17 281
Repeatability (RSDr) and intermediate precision (RSDR) relative standard deviations were 282
calculated using one‐way analysis of variance (ANOVA). 283
The combined standard uncertainty (uc) is an estimated standard deviation equal to the 284
positive square root of the total variance obtained by combining the intralaboratory 285
13
reproducibility (RSDR) and the bias of the analytical method, which consists of the 286
uncertainty associated with the purity of the standards (U[Cref]), the accuracy of the bias 287
(Sbias) and the root mean square of the bias (RMSbias). The expanded measurement 288
uncertainty (U) is obtained by multiplying uc by a coverage factor of 2, which gives a level of 289
confidence of approximately 95%. 290
2.8. Statistical analysis 291
Data processing and calculations were performed using Microsoft Office Excel 2010 and IBM 292
SPSS Statistics 21. 293
294
3. RESULTS AND DISCUSSION 295
3.1. Optimization of the MS/MS conditions 296
MS/MS instrumental parameters were optimized via direct infusion (flow rate of 10 µl.min‐1) 297
of a tune solution (10 ng.µL‐1 of each analyte) in the inlet of the mass spectrometer. During 298
30s, both in positive and negative ionization mode, a diagnostic MS spectrum was obtained, 299
from which a precursor ion for each analyte was selected. After optimization of the cone 300
voltage, the precursor ion for each analyte was mass‐selected by the first quadrupole and 301
subjected to collision‐induced dissociation (CID) by applying different collision energies, 302
yielding a product ion mass spectrum. To ensure high specificity and selectivity, for each 303
analyte, two product ions were selected and inserted in the final MS method. The optimized 304
parameters for every analyte (ionization mode, cone voltage, collision energy, precursor and 305
product ions) are shown in Table 1. Generally, it is recognized that operating in the negative 306
electrospray ionization mode (ESI‐) leads to lower signal intensity, but is also less prone to 307
matrix interferences, yielding a lower background signal in sample matrix. ESI‐ mode was 308
selected for all but one of the Alternaria toxins present in the LC‐MS/MS method.22,23,30–32 309
14
However, for ALT, ESI+ mode was selected, because CID in ESI‐ mode did not yield a 310
satisfactory product ion mass spectrum. 311
3.2. Optimization of the LC conditions 312
Optimization of the LC conditions aimed at both retention and separation of 10 analytes in a 313
fast chromatographic run. Due to differences in polarity of the analytes, the selection of a 314
suitable stationary phase was imperative. For all analytes, preliminary experiments with 315
various gradient elution programmes of ultrapure water and MeOH or ACN pointed out that 316
ACN led to narrower peak shapes compared with MeOH.29 The performance of 1 HPLC 317
(Zorbax Eclips XDB‐C18 [3,5 µm, 2.1x100 mm]) and 2 UPLC (Acquity UPLC HSS T3 [1.8 µm, 318
2.1x100 mm]; UPLC BEH C18 [1.7 µm, 2.1x100 mm]) chromatographic columns was assessed. 319
As previously shown19, the Eclips XDB‐C18 column column proved to be suitable for the 320
simultaneous determination of the polar conjugates and less polar mycotoxins. 321
Chromatographic separation was achieved in an analytical run of 14 minutes. However, UPLC, 322
which is characterized by column packing material with sub 2 micron particle size, leads to a 323
significant increase in speed, resolution and sensitivity as compared to HPLC. Indeed, for 324
both UPLC columns, chromatographic separation was achieved in less than 10 minutes. 325
Furthermore, an improved peak shape for TeA was observed with the HSS T3 column, 326
making this the column of choice for further optimization.29 Ten mobile phase modifiers (no 327
modifier; formic acid (FA) (0.1‐0.3%); acetic acid (AA) (0.5‐1%); 0.1% FA/ammonium formate 328
(2mM‐5mM); 0.5% AA/ ammonium acetate (2mM); 1% AA/ammonium acetate (5mM); 329
ammonium hydrogen carbonate (5mM)) were tested in triplicate. In the absence of any 330
modifier, the highest absolute intensity for the majority of analytes (especially the sulphates) 331
was observed, but the resulting pH (close to neutral) led to an impaired peak shape for TeA. 332
For all analytes, average overall results (intensity, peak shape) were satisfactory with 1% 333
15
glacial acetic acid (pH 2.9) as a modifier. A flow rate of 0.400 ml.min‐1 and a column 334
temperature maintained at 35°C resulted in retention times ranging from 1.81 (AOH3G) to 335
5.25 min (AME). Finally, to increase the sensitivity and selectivity of the MS conditions, the 336
MRM method was ultimately divided into 6 data acquisition windows (based on these 337
retention times), and the dwell time for each window was optimized (0.07 to 0.25 s). 338
3.3. Optimization of the sample preparation 339
Starting from an in‐house developed sample preparation and extraction methodology19, 340
design of experiments (DOE) was applied for optimization of the sample preparation for rice 341
and oat flakes matrices. The intuitive approach (Changing one separate factor at a time, 342
COST) to experimental work often does not lead to the real optimum, and gives different 343
implications with different starting points. The solution is to construct a carefully prepared 344
set of representative experiments, in which all relevant factors are simultaneously varied. In 345
contrast with COST, DOE enables the scientist to deal with factor interactions and 346
differentiate between systematic (real effects) and unsystematic (noise) variability. 347
Furthermore, DOE allows an interpretation of the results through the generation of response 348
contour plots (MODDE® software).31 349
In a first phase, for both matrices, the extraction solvent composition was optimized using a 350
three‐level full factorial design (33) with three factors, i.e. type of organic solvent (methanol, 351
acetontrile or acetone), percentage of organic solvent (50, 70 or 90) and pH (3, 6 or 9). 352
Interpretation of the generated response contour plots (data not shown) indicated that the 353
type of organic solvent proved to be a critical factor for multiple analytes, both in rice and 354
oat flakes matrices. With the exception of ALT, TeA and AME, acetonitrile resulted in far 355
higher absolute responses for all analytes. The percentage of organic solvent also proved to 356
be a critical factor. For 2 analytes, i.e. AME and ALT in both matrices, 50% of acetonitrile 357
16
resulted in the highest absolute areas. However, for all other analytes, highest absolute 358
areas were observed using 90% of acetontrile. In case of conflicting outcomes, which is to be 359
expected in the development of multi‐mycotoxin analytical methodologies, a compromise 360
was sought to determine the favourable level of a factor. Finally, an acidic pH resulted in the 361
highest absolute areas for ALT, TeA, AOH and AME in both matrices, whereas pH proved to 362
be a non‐critical factor for the other analytes. Ultimately, for both rice and oat flake matrices, 363
considering the wide range of polarities of the analytes, the best compromise for the 364
extraction solvent composition was as follows: acetonitrile/water/acetic acid (79/19.5/1.5, 365
v/v/v), which is in good agreement with previously reported studies.1,13,18,20,34,35 366
Next, using the optimal extraction solvent composition, the whole extraction procedure for 367
both matrices was subjected to screening using a three‐level fractional factorial design with 368
resolution IV (37‐3). These designs allow y factors to be examined on x levels with xy−p 369
experiments, where p is the number to which factors or interactions are confounded, i.e. 370
cannot be estimated independently. When a factor proved to be significant after statistical 371
analysis (p=0.05), the level of the factor corresponding to the highest absolute areas was 372
built in to the protocol. If a factor proved to be noncritical, the least time‐consuming or 373
cheapest level of the factor was built into the protocol. Fractional factorial designs give little 374
to no insight in possible interactions between the factors. For this reason, full factorial 375
designs were set up with the factors that proved to be critical in the fractional factorial 376
designs. Investigated factors were: extraction solvent volume, liquid‐liquid extraction (LLE) 377
with different hexane volumes, duration of vortex‐mixing, duration of the extraction, 378
duration of centrifugation, evaporation temperature and duration of ultracentrifugation. For 379
the majority of analytes in both matrices, extraction solvent volume, LLE with hexane and 380
duration of the extraction were factors that proved to be critical, and were subsequently 381
17
optimized using a three‐level full factorial design (33). No interaction effects were observed 382
from the full factorial designs with the critical factors. Finally, applying the optimized sample 383
preparation and extraction methodology for rice and oat flakes to another cereal‐based 384
matrix, i.e. barley, also generated satisfactory results. 385
3.4. Method validation 386
The UPLC‐ESI+/‐‐MS/MS method was successfully validated for the (masked) Alternaria toxins 387
AOH3G, ALT, AME3G, TeA, ATX‐I, AOH, TEN, AME3S and AME in various matrices, i.e. rice, 388
oats and barley, based on the performance criteria stipulated in Commission Regulation (EC) 389
N° 401/200624 and Commission Decision (EC) N° 2002/65723. 390
Specificity concerns the ability of the analytical method to distinguish between the analyte 391
being measured and other substances or interferences.23 In order to determine the 392
specificity/selectivity, for every matrix, 10 representative blank samples and 3 blank samples 393
fortified with chemically related compounds were subjected to analysis, revealing no 394
interfering peaks (S/N ≥ 3) in the 2.5% margin of the relative retention time for all target 395
analytes. 396
Matrix effects result from co‐eluting residual matrix components affecting (suppression or 397
enhancement) the ionization efficiency of target analytes. In extreme cases, these matrix 398
effects can result in complete suppression of the analytical signal.17 Since matrix effects may 399
exert a negative effect on important method performance parameters and subsequently 400
lead to erroneous quantitative results, they have to be tested for and evaluated during 401
method development and validation. EE values varied from 35% (TeA in all three matrices) 402
to 98% (AME in oats matrix). Strong ion suppression (<10%) was observed for the glucosides 403
(AOH3G, AME3G), TeA and AME in oats and barley matrix, while only limited ion suppression 404
(>70%) could be observed for the sulphates (AOH3S, AME3S) in all three matrices. 405
18
Furthermore, comparison by visual inspection of the slopes confirmed the presence of 406
matrix effects, and indicated the need to use matrix specific (since, for all target analytes, 407
construction of a calibration curve in rice, oats or barley matrix resulted in significantly 408
different slopes, i.e. non‐parallelism of the curves confirmed by t‐test25) MMC curves for 409
quantification purposes (Figure 3). 410
Linearity was evaluated using MMC curves (in triplicate), by spiking blank samples at six 411
concentration levels (5‐80 µg.kg‐1) for each investigated matrix. In addition to the commonly 412
reported regression coefficients (R2)34, the lowest observed value being 0.951 for AME3G in 413
oats (Table 3), a lack‐of‐fit test was also performed to assess the adequacy of the linear 414
model35. The linearity of the relative peak areas for all analytes in each investigated matrix 415
was excellent in the applied range, with p‐values greater than 0.05, which demonstrated no 416
lack of fit and thus good adequacy of the models in predicting linearity. 417
Since analyte concentrations in unknown rice and oat flake samples will be evaluated by 418
using the regression results obtained from calibration curves, a well‐designed and 419
interpreted calibration curve is essential, and contributes to the quality of bioanalytical 420
data.26 Since the condition of equal variances, termed homoscedasticity or homogenity of 421
variance, is frequently not met for bioanalytical data, larger deviations present at larger 422
concentrations tend to influence the regression line more than smaller deviations associated 423
with smaller concentrations, and thus the accuracy in the lower end of the range is 424
impaired.25–27 Therefore, the homoscedasticity assumption should be tested in any linear 425
regression analysis. It can be performed by plotting residuals versus concentration and by 426
applying an F‐test in accordance with the following statistics: 427
, ; ,
19
where the experimental F‐value is expressed as the ratio between the variances obtained at 428
the lowest (s12) and at the highest (s2
2) concentration level of the calibration range, and the 429
tabled F‐value is obtained from the F‐table at the confidence level of 99% for f1=f2=(n−1) 430
degrees of freedom. If variance is constant over the whole calibration range, residuals will 431
fall more or less randomly around the x‐axis and Fexp will be lower than Ftab, indicative for a 432
situation of homoscedasticity.26 As an example, the residual versus concentration plot of TEN 433
in oat flakes clearly shows that the error is not randomly distributed around the 434
concentration axis (Figure 4). The F‐test (data not shown) also revealed a significant 435
difference between the variances, when the experimental F‐value (Fexp = 31.90) was 436
compared to the tabled one (Ftab = 6.03). 437
An effective way to counteract the observed heteroscedastic situation is to use weighted 438
least squares linear regression (WLSLR).26 The first step in WLSLR is the choice of the 439
weighing factor, wi, and possible empirical weighing factors to be studied are 1/x, 1/x2, 1/x1/2, 440
1/y, 1/y2 and 1/y1/2. The best weighing factor is chosen according to a percentage relative 441
error (%RE), which compares the regressed concentration (Cfound) calculated from the 442
regression equation obtained for each wi, with the nominal standard concentration (Cnom): 443
%
Calculating ∑%RE, defined as the sum of absolute %RE values, is a useful and sensitive 444
indicator of goodness of fit in the evaluation of the effectiveness of a weighing factor for 445
WLSLR.26 In Table 2, the sum of the relative errors (∑%RE) and accuracy (in terms of bias, %) 446
at low (5 µg.kg‐1), medium (40 µg.kg‐1) and high (80 µg.kg‐1) concentration level obtained by 447
using unweighted (wi = 1) and weighted (wi = 1/x2) linear regression for all target analytes in 448
rice and oats matrix are displayed. Compared to the other possible empirical weighing 449
20
factors (1/x, 1/x1/2, 1/y, 1/y2 and 1/y1/2), the weighing factor 1/x2 produced the least ∑%RE 450
for this data set, providing the most adequate approximation of variance. Additionally, 451
wheighing factor 1/x2 drastically improved the accuracy (expressed as bias, in %) at the 452
lowest concentration level of the calibration curve (Table 2). 453
Limit of Detection (LOD) and Limit of Quantification (LOQ) were determined using MMC 454
curves, by spiking blank samples at 8 concentrations levels (0.1‐10 µg.kg‐1) for each 455
investigated matrix. Although no regulatory limits have been set for Alternaria toxins in 456
cereals and cereal‐based foodstuffs, the developed UPLC‐MS/MS methodology allowed for 457
the determination of all target analytes at a low parts per billion (ppb) level. In Table 3, LOD 458
and LOQ values are represented for every analyte in each investigated matrix. LOQ values 459
ranged from 0.93 to 4.99 μg.kg‐1 except for AME3S in rice (8.32 μg.kg‐1). For all analytes in 460
each investigated matrix, the calculated LODs and LOQs were also verified by the S/N ratio, 461
which should be more than 3 and 10 according to the IUPAC (International Union of Pure 462
and Applied Chemistry) settings36,37. EFSA states that the database concerning toxicological 463
effects of Alternaria toxins in experimental animals and/or in humans is currently too limited 464
to be used as a basis for identification of reference points for different toxicological effects 465
(EFSA). In the absence of established tolerable daily intake (TDI) values, a risk assessment 466
based on TDI values for T‐2/HT‐2 toxins38 was performed, indicating that when the 467
incorporated amount of Alternaria toxins in grains and grain‐based products would result in 468
a concentration at LOQ level, the dietary exposure to humans, based on standard food 469
basket consumption figures39, would be negligible with regard to toxicity. 470
A combined sample plan, which comprised of the quantitative analysis (using MMC curves 471
with six concentration levels, 5 to 80 µg.kg‐1) of spiked blank samples (concentration levels 472
of 5, 40 and 80 µg.kg‐1) in triplicate on three consecutive days for each investigated matrix, 473
21
was used to determine the apparent recovery, repeatability, intermediate precision and 474
expanded measurement uncertainty. Results for each investigated matrix are displayed in 475
Table 4. 476
The Apparent recovery (RA) at and above LOQ level ranged from 94.4% to 106.5% for all 477
analytes in all three matrices (Table 4). These results are in good agreement with the 478
guideline ranges (80‐110%) imposed by Commission Decision N° (EC) 2002/65723. 479
Repeatability (RSDr) and intermediate precision (RSDR), calculated using 1‐way ANOVA, 480
ranged from 2.0% to 14.1% and 2.2% tot 14.1%, respectively (Table 4). According to 481
Commission Decision N° (EC) 2002/657, repeatability and intermediate precision values for 482
the repeated analysis of fortified material, under repeatability and reproducibility conditions, 483
shall not exceed the level calculated by the Horwitz Equation (RSDr = ⅔(2[1 – 0,5 log C]); RSDR = 484
2[1 – 0,5 log C]). However, for mass fractions lower than 100 μg.kg‐1 the application of the 485
Horwitz Equation gives unacceptable high values.23 Therefore, according to an in‐house 486
developed standard operating procedure on analytical method validation, RSDr and RSDR 487
values for concentrations lower than 100 μg.kg‐1 should be lower than 20% and 25%, 488
respectively. 489
Uncertainty of the measurement is defined as a parameter, which is associated with the 490
results of a measurement and characterizes the dispersion of the values that could 491
reasonably be attributed to the measurand.40,41 The expanded measurement uncertainty (U) 492
is an estimated standard deviation equal to the positive square root of the total variance 493
obtained by combining the intralaboratory reproducibility (intermediate precision; RSDR) and 494
the bias of the analytical method, multiplied by a coverage factor of 2, giving a level of 495
confidence of approximately 95%. U was calculated at three concentration levels (5, 40 and 496
80 µg.kg‐1) for all analytes in all three matrices, and ranged from 6.1% to 52.3% (Table 4). In 497
22
general, lower U values were obtained for AOH, AME, TeA and TEN (values ranging from 6.1 % 498
to 22.4%) except for AME and TEN at medium (40.5% and 30.2% respectively) and high (26.7% 499
and 32.0% respectively) concentration levels in barley matrix. The availability of reference 500
standards with a certificate of analysis for these analytes indeed leads to a significantly lower 501
value for the U[Cref] term in the calculation of U. 502
3.5. Analysis of commercially available cereal based foodstuffs 503
Subsequent to the optimization and validation of these analytical methods, commercially 504
available rice (n=16) and oat flake (n=8) samples were sampled (between March and June 505
2013) and subsequently subjected to analysis. Analytical results are reported as x ± U 506
whereby x is the analytical result and U is the (concentration dependent) expanded 507
measurement uncertainty.24 A total of 75% (12/16) of rice samples, and 38% (3/8) of oat 508
flake samples were contaminated with TeA, with concentrations in the range of 1.9 ± 0.1 to 509
68 ± 7 µg.kg‐1 and 2.1 ± 0.2 to 39 ± 5 µg.kg‐1, respectively. Furthermore, 25% (4/16) of rice 510
samples were contaminated with TEN (4.1 ± 0.7 to 10.9 ± 2.0 µg.kg‐1), yet no TEN was found 511
in oat flake samples. Although the prevalence of TeA and TEN in these samples is higher, the 512
observed levels are in good agreement with the mean lower bound (LB) ‐ upper bound (UB) 513
concentrations for TeA and TEN in grain and grain‐based products reported by EFSA8. None 514
of the other target analytes (AOH, AME, ALT, ATx‐I) nor AOH and AME conjugates (AOH3G, 515
AOH3S, AME3G, AME3S) were detected in any of the samples analyzed. 516
517
4. CONCLUSIONS 518
A UPLC‐ESI+/‐‐MS/MS method for the simultaneous determination of both free and 519
conjugated Alternaria toxins in cereal and cereal based foodstuffs was successfully 520
developed and validated, using Regulation 401/2006/EC and Decision 2002/657/EC as 521
23
guidelines24,25 . Accordingly, these methods, applying isotopically labeled internal standards, 522
were validated for several parameters and proved to be fit‐for‐purpose. Subsequently, a 523
mini‐survey, conducted between March and June 2013, revealed the presence of TeA in 524
both rice and oat flake samples and TEN in rice samples. The validated analytical 525
methodologies will be further used to analyze cereal‐based foodstuffs within the framework 526
of a research project funded by the Belgian Federal Public Service of Health, Food Chain 527
Safety and Environment (Contract RF 12/6261 ALTER). The principal aim of this project is to 528
gather quantitative occurrence data on free and conjugated Alternaria toxins in various 529
foodstuffs available on the Belgian market. 530
531
ACKNOWLEDGMENTS 532
This study was funded by the Federal Public Service of Health, Food Chain Safety and 533
Environment (Contract RF 12/6261 ALTER). 534
24
Figure 1 ‐ (a) Chemical structures of AOH derivatives, (b) chemical glucosylation and deprotection, (c)
chemical sulfation and deprotection.
25
Table 1 ‐ Optimized LC‐MS/MS instrumental parameters for all analytes, including the isotopically
labeled internal standards [13C6,15N]‐TeA and [2H4]‐AME.
Analyte (a) Precursor ion
(m/z) Molecular ion
Cone
voltage (V)
Product ions (b)
(m/z)
Collision
energy (eV)
Retention
time (min)
AOH3G 419.0 [M‐H]‐ 65 256.1 32 1.81 228.1 40
AOH3S 337.1 [M‐H]‐ 34 213.0 38 2.40
257.0 24
ALT 292.9 [M+H]+ 23 257.1 14 2.62
239.0 11
AME3G 433.0 [M‐H]‐ 63 270.1 31 2.97
227.1 45
TeA 196.1 [M‐H]‐ 55 139.0 20 3.28
112.0 24
ATX‐I 351.2 [M‐H]‐ 35 315.1 18 3.73
333.1 12
AOH 257.1 [M‐H]‐ 54 213.0 24 3.97
215.0 26
TEN 413.3 [M‐H]‐ 44 141.0 24 4.33
271.2 18
AME3S 351.1 [M‐H]‐ 34 271.1 24 4.66
256.1 38
AME 271.1 [M‐H]‐ 48 256.0 24 5.25
228.0 30
[13C6,15N]‐TeA 202.9 [M‐H]‐ 52 141.9 20 3.28
112.9 24
[2H4]‐AME 274.9 [M‐H]‐ 57 260.0 24
5.25 232.0 30
(a) AOH: alternariol ‐ AME: alternariol monomethyl ether ‐ ALT: altenuene ‐ TeA: tenuazonic acid ‐ TEN: tentoxin ‐ ATX‐I: altertoxin‐I
‐ AOH3S: AOH‐3‐sulphate ‐ AOH3G: AOH‐3‐glucoside ‐ AME3S: AME‐3‐sulphate ‐ AME3G: AME‐3‐glucoside.
(b) The underlined product ion is the most abundant and thus used for quantification purposes
Figure 2
Figure 3
‐ LC‐MS/MS
‐ Compariso
solution, r
chromatogr
n of the slop
revealing the
am (quantifie
es of calibrat
e presence of
er ions displa
tion curves o
f varying (ma
ayed) of a 40
of altertoxin‐
atrix depende
0 μg.kg−1 spik
I in rice, oats
ent) signal su
ked oat flakes
s, barley and
uppression.
26
s sample.
standard
Figure 4
The resid
Table 2 ‐
(80 µg.
AOH
AOH
AL
AME
Te
ATX
AO
TE
AM
AM
AOH: a
4 ‐ Residuals
dual plot clea
‐ Sum of the
kg‐1) concent
RICE
wi
H3G 1
1/x2
H3S 1
1/x2
LT 1
1/x2
E3G 1
1/x2
eA 1
1/x2
X‐I 1
1/x2
OH 1
1/x2
EN 1
1/x2
E3S 1
1/x2
ME 1
1/x2
alternariol ‐ AME
AOH3S: AO
s plotted agai
arly shows th
i
relative erro
tration level
regressio
E
∑%RE
218.1 2 197.1
398.4 2 271.6
248.7 2 199.0
233.1 2 199.9
186.1 2 149.8
247.4 2 212.9
266.4 2 209.6
222.4 2 172.0
493.7 2 349.3
243.2 2 187.4
: alternariol mon
OH‐3‐sulphate ‐ A
inst TEN conc
hat the error
ndicative of h
rs (∑%RE) an
obtained by
on for all targ
Bias (
low mediu
15.6 3.8
0.9 0.9
12.4 5.7
1.4 1.3
9.9 3.5
1.9 0.3
3.2 3.5
1.3 0.8
4.6 1.9
0.4 0.7
10.8 2.6
2.3 1.8
13.6 4.5
2.0 0.9
6.9 3.3
5.3 0.9
16.6 14.2
3.1 0.4
2.8 0.9
0.1 1.9
nomethyl ether ‐ A
OH3G: AOH‐3‐glu
centrations (
is not random
heterogeneit
nd accuracy a
using unwei
get analytes i
(%)
um high
8 3.4
9 0.5
7 4.1
3 0.1
5 4.1
3 0.3
5 4.7
8 1.1
9 0.1
7 0.9
6 4.8
8 0.9
5 6.4
9 1.2
3 4.7
9 0.1
2 1.6
4 0.2
9 0.6
9 0.1
ALT: altenuene ‐
ucoside ‐ AME3S
oat flakes) fo
mly distribut
ty of variance
t low (5 µg.k
ghted (wi = 1
in rice and oa
OATS
wi
1
1/x2
1
1/x2
1
1/x2
1
1/x2
1
1/x2
1
1/x2
1
1/x2
1
1/x2
1
1/x2
1
1/x2
TeA: tenuazonic
: AME‐3‐sulphate
or the inter‐d
ted around th
e.
kg‐1), medium
1) and weight
ats matrix
∑%RE low
393.6 15.0
257.6 1.5
504.8 14.5
315.4 0.9
403.2 16.1
264.9 11.5
944.8 14.4
665.8 3.4
334.1 17.6
206.3 1.5
434.7 18.2
273.4 1.5
498.9 17.5
281.3 1.2
426.1 18.6
254.6 2.8
561.6 0.4
322.6 0.9
421.5 11.0
260.5 2.4
acid ‐ TEN: tento
e ‐ AME3G: AME‐
day validatio
he concentra
m (40 µg.kg‐1)
ted (wi = 1/x
Bias (%)
medium
4.6
3.2
0.8
0.6
1.5
0.1
4 1.1
2.0
1.6
0.1
4.9
3.4
4.9
3.2
2.6
0.8
0.5
0.4
0.6
1.7
oxin ‐ ATX‐I: altert
‐3‐glucoside.
27
n assay.
ation axis,
and high 2) linear
high
2.0
3.9
0.2
0.3
1.9
4.1
2.3
1.1
2.4
5.4
2.1
4.3
2.3
4.4
1.0
3.6
0.9
0.2
0.3
2.5
toxin‐I ‐
28
Table 3 ‐ R² values and ranges of the matrix‐matched calibration curves in rice, oats and barley,
supplemented with limits of detection (LOD) and limits of quantification (LOQ) for all analytes (µg.kg‐1)
Rice Oats Barley
Range R² LOD LOQ Range R² LOD LOQ Range R² LOD LOQ
AOH3G 5‐80 0.992 0.82 1.65 5‐80 0.992 0.74 1.49 5‐80 0.986 2.11 4.23
AOH3S 5‐80 0.989 1.54 3.08 5‐80 0.991 1.24 2.48 5‐80 0.988 2.41 4.82
ALT 5‐80 0.991 1.63 3.26 5‐80 0.990 1.19 2.38 5‐80 0.984 2.21 4.43
AME3G 5‐80 0.993 0.77 1.53 5‐80 0.951 2.46 4.91 5‐80 0.984 2.50 4.99
TeA 5‐80 0.996 0.93 1.86 5‐80 0.996 0.61 1.21 5‐80 0.995 1.18 2.35
ATX‐I 5‐80 0.990 1.01 2.02 5‐80 0.992 1.75 3.49 5‐80 0.983 1.81 3.62
AOH 5‐80 0.992 0.80 1.60 5‐80 0.991 0.50 1.00 5‐80 0.987 0.85 1.70
TEN 5‐80 0.994 0.91 1.82 5‐80 0.992 0.46 0.93 5‐80 0.981 0.86 1.71
AME3S 10‐80 0.974 4.16 8.32 5‐80 0.988 0.79 1.58 5‐80 0.985 0.84 1.68
AME 5‐80 0.993 0.67 1.35 5‐80 0.991 1.25 2.50 5‐80 0.975 1.32 2.52
AOH: alternariol ‐ AME: alternariol monomethyl ether ‐ ALT: altenuene ‐ TeA: tenuazonic acid ‐ TEN: tentoxin ‐ ATX‐I: altertoxin‐
I ‐ AOH3S: AOH‐3‐sulphate ‐ AOH3G: AOH‐3‐glucoside ‐ AME3S: AME‐3‐sulphate ‐ AME3G: AME‐3‐glucoside.
29
Table 4 ‐ Repeatability (RSDr), intermediate precision (RSDR), apparent recovery (RA, %) and expanded measurement uncertainty (U, %) values for all analytes on
low, medium and high concentration level in rice, oats and barley
AOH3G AOH3S ALT AME3G TeA
RSDr RSDR RA U RSDr RSDR RA U RSDr RSDR RA U RSDr RSDR RA U RSDr RSDR RA U
Rice 5 µg.kg‐1 2.5 3.0 99.1 35.63 2.6 3.2 101.4 36.54 3.3 3.5 98.1 36.18 2.8 2.8 98.7 35.59 2.2 2.2 100.4 6.07
40 µg.kg‐1 4.0 4.0 100.9 36.31 3.2 4.5 101.3 37.81 4.3 4.3 99.7 36.70 5.0 5.0 100.8 37.07 5.8 5.8 99.3 14.75
80 µg.kg‐1 4.1 5.2 100.5 37.37 5.8 7.6 99.9 39.53 3.6 4.9 99.7 36.98 3.1 4.4 101.1 36.61 3.2 3.6 100.9 10.34
Oats 5 µg.kg‐1 2.7 2.7 98.5 35.54 4.1 4.3 100.9 34.03 4.7 4.7 98.5 36.88 10.6 10.6 96.6 45.94 2.8 2.8 101.5 8.40
40 µg.kg‐1 7.0 7.3 103.2 40.83 8.3 8.3 99.4 40.73 7.8 7.8 100.1 40.07 6.4 11.3 98.0 45.27 4.1 4.4 99.9 12.47
80 µg.kg‐1 5.2 5.2 96.1 37.71 8.9 9.1 99.7 41.24 5.1 5.1 95.9 38.01 10.9 14.6 101.1 52.30 3.5 3.5 105.4 14.77
Barley 5 µg.kg‐1 5.9 5.9 101.6 38.02 5.6 6.0 101.8 38.02 7.6 7.6 99.5 40.25 7.2 7.2 102.1 39.79 2.5 2.5 101.4 7.38
40 µg.kg‐1 7.8 9.5 99.3 43.16 7.9 7.9 99.3 41.23 6.8 7.2 101.9 40.27 8.0 9.1 100.6 42.75 5.5 5.5 99.4 13.99
80 µg.kg‐1 8.8 8.9 101.9 43.12 8.6 9.4 101.3 42.12 5.0 5.0 99.6 36.74 8.3 8.3 103.1 42.03 3.8 3.8 102.7 11.78
ATX‐I AOH TEN AME3S AME
RSDr RSDR RA U RSDr RSDR RA U RSDr RSDR RA U RSDr RSDR RA U RSDr RSDR RA U
Rice 5 µg.kg‐1 2.5 3.4 97.7 36.07 2.6 2.6 98.0 7.84 2.0 2.3 104.7 18.15 2.9 4.2 103.1 36.89 2.4 2.5 100.1 7.37
40 µg.kg‐1 4.4 5.5 101.8 37.88 5.9 5.9 100.9 15.26 4.5 4.5 100.9 12.91 8.4 9.2 100.4 42.73 5.2 7.7 101.9 20.95
80 µg.kg‐1 2.4 4.4 99.1 36.44 3.7 5.7 98.8 15.16 2.8 5.9 100.1 15.10 5.8 8.0 99.8 42.53 3.7 3.7 100.1 10.21
Oats 5 µg.kg‐1 4.7 4.7 98.5 36.68 3.4 3.4 98.8 9.91 3.3 3.3 97.2 10.26 3.7 3.7 100.9 36.03 4.0 4.0 102.4 11.47
40 µg.kg‐1 5.5 6.0 103.4 39.13 8.3 8.3 103.2 22.44 6.0 6.0 100.8 16.29 9.3 9.3 100.4 42.73 5.9 5.9 101.7 16.36
80 µg.kg‐1 6.7 6.7 95.7 39.84 7.3 7.3 95.6 20.83 4.4 5.1 96.4 15.58 9.0 9.0 99.8 42.53 4.9 6.7 102.5 18.85
Barley 5 µg.kg‐1 5.2 6.4 101.0 38.85 5.3 5.3 104.0 16.79 4.5 4.5 102.5 13.79 3.2 3.2 100.9 35.95 6.7 6.7 102.4 19.35
40 µg.kg‐1 6.0 6.2 98.0 38.91 6.2 6.2 103.3 17.29 11.0 11.0 102.3 30.21 6.5 6.5 98.7 39.30 14.1 14.1 94.4 40.54
80 µg.kg‐1 8.1 8.1 103.7 41.08 6.9 6.9 103.4 19.69 8.7 11.3 104.0 31.94 9.7 9.8 106.5 46.72 2.5 10.0 104.0 26.66
AOH: alternariol ‐ AME: alternariol monomethyl ether ‐ ALT: altenuene ‐ TeA: tenuazonic acid ‐ TEN: tentoxin ‐ ATX‐I: altertoxin‐I ‐ AOH3S: AOH‐3‐sulphate ‐ AOH3G: AOH‐3‐glucoside ‐ AME3S: AME‐3‐sulphate ‐
AME3G: AME‐3‐glucoside.
30
REFERENCES
1. Shephard, G. S. Determination of mycotoxins in human foods. Chem. Soc. Rev. 37, 2468–77 (2008).
2. Fernández‐Cruz, M. L., Mansilla, M. L. & Tadeo, J. L. Mycotoxins in fruits and their processed products: Analysis, occurrence and health implications. J. Adv. Res. 1, 113–122 (2010).
3. Köppen, R. et al. Determination of mycotoxins in foods: current state of analytical methods and limitations. Appl. Microbiol. Biotechnol. 86, 1595–612 (2010).
4. Ostry, V. Alternaria mycotoxins: an overview of chemical characterization, producers, toxicity, analysis and occurrence in foodstuffs. World Mycotoxin J. 1, 175–188 (2008).
5. Bottalico, A. & Logrieco, A. in Mycotoxins Agric. Food Safety. Eds Sinha KK Bhatnager D, Marcel Dekker, New York 65–108 (1998).
6. Scott, P. Analysis of agricultural commodities and foods for Alternaria mycotoxins. J. AOAC Int. 84, 1809–1817 (2001).
7. Siegel, D., Merkel, S., Koch, M. & Nehls, I. Quantification of the Alternaria mycotoxin tenuazonic acid in beer. Food Chem. 120, 902–906 (2010).
8. (EFSA) European Food Safety Authority. EFSA on Contaminants in the Food Chain (CONTAM); Scientific Opinion on the risks for animal and public health related to the presence of Alternaria toxins in feed and food. EFSA J. 9, 97 (2011).
9. Kroes, R. & Kozianowski, G. Threshold of toxicological concern (TTC) in food safety assessment. Toxicol. Lett. 127, 43–6 (2002).
10. Kroes, R. et al. Structure‐based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65–83 (2004).
11. Kroes, R., Kleiner, J. & Renwick, a. The threshold of toxicological concern concept in risk assessment. Toxicol. Sci. 86, 226–30 (2005).
12. Berthiller, F. et al. Masked mycotoxins: determination of a deoxynivalenol glucoside in artificially and naturally contaminated wheat by liquid chromatography‐tandem mass spectrometry. J. Agric. Food Chem. 53, 3421–5 (2005).
13. Berthiller, F., Sulyok, M., Krska, R. & Schuhmacher, R. Chromatographic methods for the simultaneous determination of mycotoxins and their conjugates in cereals. Int. J. Food Microbiol. 119, 33–7 (2007).
14. M. Rychlik, H.U. Humpf, D. Marko, S. Danicke, A. Mally, F. Berthiller, H. Klaffke, N. Lorenz, Proposal of a comprehensive definition of modified and other forms of mycotoxins including “masked” mycotoxins, Mycotoxin Res. 30 (2014) 197–205.
15. Gareis M, Bauer J, Thiem J, Plank G, Grabley S, G. B. Cleavage of zearalenone‐glycoside, a masked mycotoxin, during digestion in swine. J Vet Med B – Zentralblatt fur Vet. R. B – Infect Dis Vet Publ Hlth. 236–240 (1990).
16. Engelhardt, G., Zill, G. & Wohner, B. Transformation of the Fusarium mycotoxin Zearalenone in maize cell‐suspension cultures. Naturwissenschaften 75, 309–310 (1988).
31
17. Schneweis, I., Meyer, K., Engelhardt, G. & Bauer, J. Occurrence of zearalenone‐4‐beta‐D‐glucopyranoside in wheat. J. Agric. Food Chem. 50, 1736–8 (2002).
18. Sulyok, M., Berthiller, F., Krska, R. & Schuhmacher, R. Development and validation of a liquid chromatography / tandem mass spectrometric method for the determination of 39 mycotoxins in wheat and maize. Rapid Commun. Mass Spectrom. 20, 2649–2659 (2006). doi:10.1002/rcm
19. Vendl, O., Berthiller, F., Crews, C. & Krska, R. Simultaneous determination of deoxynivalenol, zearalenone, and their major masked metabolites in cereal‐based food by LC‐MS‐MS. Anal. Bioanal. Chem. 395, 1347–54 (2009).
20. De Boevre, M. et al. Development and validation of an LC‐MS/MS method for the simultaneous determination of deoxynivalenol, zearalenone, T‐2‐toxin and some masked metabolites in different cereals and cereal‐derived food. Food Addit. Contam. Part A. Chem. Anal. Control. Expo. Risk Assess. 29, 819–35 (2012).
21. Mikula, H., Skrinjar, P., Sohr, B., Ellmer, D. & Hametner, C. Total synthesis of masked Alternaria mycotoxins d sulfates and glucosides of alternariol ( AOH ) and alternariol‐9‐methyl ether. Tetrahedron, 69, 10322–10330 (2013).
22. Asam, S., Konitzer, K., Schieberle, P. & Rychlik, M. Stable isotope dilution assays of alternariol and alternariol monomethyl ether in beverages. J. Agric. Food Chem. 57, 5152–60 (2009).
23. Asam, S., Liu, Y., Konitzer, K. & Rychlik, M. Development of a stable isotope dilution assay for tenuazonic acid. J. Agric. Food Chem. 59, 2980–7 (2011).
24. 2002/657/EC. COMMISSION DECISION 2002/657 of 12 August 2002 implementing Council Directive 96/23/EC concerning the performance of analytical methods and the interpretation of results. 8–36 (2002).
25. 401/2006/EC. COMMISSION REGULATION (EC) 401/2006 of 23 February 2006 laying down the methods of sampling and analysis for the official control of the levels of mycotoxins in foodstuffs. (2006).
26. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, Validation of Analytical Procedures: Text and Methodology Q2(R1), 2005.
27. Almeida, A., Castel‐Branco, M. & Falcão, A. Linear regression for calibration lines revisited: weighting schemes for bioanalytical methods. J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci. 774, 215–22 (2002).
28. J. Van Loco. in Determ. Chem. Elem. Food Appl. At. Mass Spectrom. 135–163 (2007).
29. Singtoroj, T. et al. A new approach to evaluate regression models during validation of bioanalytical assays. J. Pharm. Biomed. Anal. 41, 219–27 (2006).
30. Lau, B. P. et al. L iquid chromatography – mass spectrometry and liquid chromatography – tandem mass spectrometry of the Alternaria mycotoxins alternariol and alternariol monomethyl ether in fruit juices and beverages. J. Chromatogr. A 998, 119–131 (2003).
31. Noser, J., Schneider, P., Rother, M. & Schmutz, H. Determination of six Alternaria toxins with UPLC‐MS/MS and their occurrence in tomatoes and tomato products from the Swiss market. Mycotoxin Res. 27, 265–71 (2011).
32
32. Asam, S., Lichtenegger, M., Liu, Y. & Rychlik, M. Content of the Alternaria mycotoxin tenuazonic acid in food commodities determined by a stable isotope dilution assay. Mycotoxin Res. 28, 9–15 (2012).
33. Eriksson, L., Johansson, E., Kettaneh‐Wold, N., Wikström, C. & Wold, S. Design of Experiments: Principles and Applications. Umetrics Acad. (2008).
34. Sulyok, M., Krska, R. & Schuhmacher, R. Application of an LC–MS/MS based multi‐mycotoxin method for the semi‐quantitative determination of mycotoxins occurring in different types of food infected by moulds. Food Chem. 119, 408–416 (2010).
35. Varga, E. et al. Development and validation of a (semi‐)quantitative UHPLC‐MS/MS method for the determination of 191 mycotoxins and other fungal metabolites in almonds, hazelnuts, peanuts and pistachios. Anal. Bioanal. Chem. 405, 5087–104 (2013).
36. Sonnergaard, J. M. On the misinterpretation of the correlation coefficient in pharmaceutical sciences. Int. J. Pharm. 321, 12–7 (2006).
37. Ediage, E. N., Di Mavungu, J. D., Monbaliu, S., Van Peteghem, C. & De Saeger, S. A validated multianalyte LC‐MS/MS method for quantification of 25 mycotoxins in cassava flour, peanut cake and maize samples. J. Agric. Food Chem. 59, 5173–80 (2011).
38. Monbaliu, S. et al. Development of a multi‐mycotoxin liquid chromatography / tandem mass spectrometry method for sweet pepper analysis. Rapid Commun. Mass Spectrom. 23, 3–11 (2009). doi:10.1002/rcm
39. Monbaliu, S. et al. Occurrence of mycotoxins in feed as analyzed by a multi‐mycotoxin LC‐MS/MS method. J. Agric. Food Chem. 58, 66–71 (2010).
40. 2013/165/EC. COMMISSION RECOMMENDATION of 27 March 2013 on the presence of T‐2 and HT‐2 toxin in cereals and cereal products. 9, 2011–2014 (2013).
41. (EFSA) European Food Safety Authority. The EFSA Comprehensive European Food Consumption Database ‐ Chronic food consumption statistics reported in grams/kg body weight/day. (2011). at <http://www.efsa.europa.eu/en/datexfoodcdb/datexfooddb.htm>
42. Hund, E., Massart, D. L. & Smeyers‐verbeke, J. Operational definitions of uncertainty. Trends Anal. Chem. 20, 394–406 (2001).
43. Hund, E., Massart, D. L. & Smeyers‐verbeke, J. Comparison of different approaches to estimate the uncertainty of a liquid chromatographic assay. Anal. Chim. Acta, 480, 39–52 (2003).
