**1. Introduction**

The mycotoxins nivalenol (NIV); deoxynivalenol (DON) and its acetyl derivatives, 3-acetyl DON (3-AcDON) and 15-acetyl DON (15-AcDON); T-2 toxin (T-2) and its metabolite HT-2 toxin (HT-2); and zearalenone (ZEN) are produced by various *Fusarium* species. Cereals such as wheat, maize, barley, oats, rye, and relevant derived products are most likely to be affected [1]. The toxic effects of *Fusarium* toxins on human and animal species have been extensively studied [2–5]. Because of its ability to induce acute vomiting in pigs, DON has also been given the trivial name "vomitoxin". Its acute effects in humans are similar to those observed in animals. The most common effects of long-term dietary exposure to DON are weight gain suppression and anorexia [2]. Similarly, toxicity studies have shown that NIV has anorectic effects upon short-term exposure, as well as immunotoxic and hematotoxic effects [5]. T-2 and HT-2 are known to inhibit the synthesis of protein, DNA, and RNA, and to have immunosuppressive and cytotoxic effects [4]. Estrogenic activity is the critical mode of action of ZEN and its main reductive metabolites. Adverse effects have been reported on the reproductive tract either in male and female animals, including testosterone synthesis, spermatogenesis, fertility, and embryo survival [3]. According to the available toxicological data concerning carcinogenicity in humans, ZEN, DON, and NIV were included by the World Health Organization's (WHO) International Agency for Research on Cancer (IARC) in the Group 3 as not classifiable regarding their carcinogenicity to humans [6].

To ensure the safety of cereals and derived products, the European Commission (EC) set maximum levels for mycotoxins, including DON and ZEN [7]. Regarding T-2 and HT-2, indicative levels to stimulate data collection for the sum of these *Fusarium* mycotoxins in cereals and cereal-derived products were published in the Commission Recommendation 2013/165/EU [8]. When evaluating the risks to human and animal health related to the presence of DON, its acetylated derivatives (3- and 15-AcDON) and the modified form DON-3-glucoside, as well as T-2 and HT-2 in food and feed, the European Food Safety Authority (EFSA) recommended the interlaboratory validation and standardization of LC–MS/MS methodologies for the simultaneous quantification of DON and its derivatives and analytical methods with an appropriate sensitivity for T-2 and HT-2 toxins in food and feed commodities. In addition, the collection of analytical data on the co-occurrence of the above-mentioned toxins was requested [2,9]. Even though not regulated yet, available studies on the occurrence and toxicity of NIV have been evaluated by the EFSA, highlighting the need for validated methods suitable for the determination of NIV in the low μg/kg range to provide reliable data for risk assessment [5,10].

The establishment of standardized methods of analysis is of the utmost importance to guarantee a uniform application of the European legislation in all member states and contribute to maintaining a high level of food and feed safety. The results of collaborative studies and proficiency tests (PTs) made available in the last decade have provided evidence that liquid chromatography-mass spectrometry (LC-MS)-based methods of analysis for (multiple) mycotoxins in food/feed can be actually fit for the purpose of the enforcement of legislative limits [11–13]. At the EU level, standardized methods of analysis are established by the European Committee for Standardization (CEN). The first standard LC-MS-based methodologies for the determination of mycotoxins in foods were published in 2017 by the CEN and cover ZEN [14] and T-2/HT-2 toxins [15] determination. These two methods were validated within the M/520 standardization mandate [16] by which the Commission invited CEN to establish European Standards/Technical Specifications that provide standardized methods of analysis for mycotoxins in food. Six of the 11 methods of analysis listed in this mandate were specifically requested to be based on LC-MS.

The work described herein addresses item 4 of the standardization mandate, aiming at optimizing and validating an analytical method for the simultaneous determination of NIV, DON and its acetyl derivatives (3-AcDON and 15-AcDON), T-2, HT-2, and ZEN in cereals and cereal products by liquid chromatography-tandem mass spectrometry (LC-MS/MS). According to the EC sector information and statistics on cereals (https://ec.europa.eu/info/food-farming-fisheries/plants-and-plant-products/ plant-products/cereals\_en), more than 50% of the cereals grown in the EU are wheat. It was therefore agreed within the working group (CEN TC275/WG5 biotoxins) to select wheat and wheat products as representative target commodities for the validation study. The results of the full collaborative study, involving seventeen international laboratories, are reported together with information relevant to the preparation and characterization of the test materials, the protocol of analysis, and the statistical analysis of the results. The derived method has been recently adopted as a CEN standard [17].

#### **2. Results and Discussion**

#### *2.1. Pre-Trial Results*

Prior to the full validation study, laboratories had to participate in a pre-trial study. The aims of the pre-trial were to let the participants become familiarized with the correct execution of the method protocol and to optimize the LC-MS/MS conditions for mycotoxin detection, as well as to pre-check the method interlaboratory performances.

The results forms were returned by twelve out of fifteen laboratories participating in the pre-trial. One laboratory was removed because of the extremely low values reported for all target mycotoxins. A root cause analysis performed in collaboration with the participant laboratory lead us to identify a possible problem in the extract clean-up, leading to very high matrix effects which could not be properly compensated by the internal standards. Two laboratories reporting valid results were excluded from the statistical evaluations for 3-AcDON and 15-AcDON since they reported values as the sum of 3-AcDON and 15-AcDON.

The recoveries from spiked wheat flour were in the range of 78–102%, with relative standard deviations for repeatability (RSDr) ranging from 4.4% to 12% and relative standard deviations for reproducibility (RSDR) ranging from 8.9% to 25%. For contaminated wheat samples, RSDr and RSDR ranged from 3.2% to 17% and from 7.5% to 29%, respectively.

The Horwitz ratio (HorRat) values, calculated as the ratio between the RSDR obtained during the pre-trial study and the predicted RSDR calculated by the modified Horwitz equation [18,19], were between 0.5 and 2.1 for all mycotoxins in the three test materials. In all the considered cases, the method performances fell within the criteria established by the European Commission for the acceptability of methods for mycotoxin determination, set for DON, ZEN, and T-2/HT-2 [20,21], proving the candidate method to be suitable for the full collaborative study.

Some issues related to the method protocol execution arose and were discussed with the participants. Two participant laboratories were not able to achieve a full chromatographic separation of 3-AcDON and 15-AcDON. Two participants reporting poor peak shapes for NIV were identified as outliers by the Cochran test. Therefore, improvements of the LC conditions were suggested by the interlaboratory study coordinator and examples of suitable LC-MS/MS settings were provided (Figure 1A,B).

#### *2.2. Full Collaborative Study*

Seventeen laboratories participated in the full collaborative study, and 15 report forms were collected. Eleven participants used triple quadrupole mass detectors (in a wide range of manufacturers and models), whereas four used hybrid quadrupole-OrbitrapTM detectors. Given the limited number of participants using high-resolution MS (HRMS) detectors, low- and high-resolution MS data were evaluated as a unique set.

Three participants were not able to resolve 3- and 15-AcDON peaks and reported results relevant to the sum of the two isomers. Two participants could report data for 3-AcDON only, since 3-AcDON and 15-AcDON were detected in negative and positive ion mode, respectively, in two separate chromatographic runs, and it was not possible to quantitate 15-AcDON with the response ratio of the isotopically labelled 3-AcDON. Moreover, two participants using a triple quadrupole MS detector with polarity switching between defined retention time windows did not detect ZEN. This was probably due to slight retention time shifts leading ZEN to elute out of the dedicated retention time period.

**Figure 1.** (**A**) Selected ion chromatogram (quantifier SRM–selected ion monitoring-transitions) and LC-MS settings of a wheat sample spiked with 120 μg/kg nivalenol (NIV), 600 μg/kg deoxynivalenol (DON), 75 μg/kg 3- and 15-AcDON (acetyl deoxynivalenol), 25 μg/kg HT-2 toxin (HT-2) and T-2 toxin (T-2), 50 μg/kg zearalenone (ZEN), and relevant isotopically labeled internal standard (ISTD—upper lines); (**B**) extracted ion chromatogram (quantifier ions) and LC-MS settings of a mycotoxin standard solution containing 1.87 μg/mL NIV, 9.37 μg/mL DON, 1.12 μg/mL 3- and 15-AcDON, 0.37 μg/mL HT-2 and T-2, 0.75 μg/mL ZEN, and relevant ISTD (upper lines).

Some data were considered "invalid" and were excluded from the statistical evaluation. Specifically, one laboratory reported problems in chromatography (poor peak shape) and provided extremely high values for the majority of the data. One laboratory using HRMS reported extremely high values in all samples for early eluting toxins (NIV and DON), probably due to poor separation from the solvent front. One participant faced sensitivity issues in all the samples but calibrants and reported extremely low values for the majority of the toxin/sample combinations. The cause was not clearly identified, but it was reasonably attributed to problems occurring during the clean-up procedure.

The results of the remaining laboratories were subjected to statistical analysis for the identification of outliers by the Cochran and Grubbs test to remove laboratories showing a significantly higher variability among replicates and extreme values, respectively [22]. For all toxin/material combinations, the number of identified outliers was lower than 20%, and thus below the maximum of 2/9 laboratories that could be removed as recommended in the AOAC (Association of Official Analytical Chemists) International guidelines for conducting interlaboratory studies [22].

Full collaborative study results are reported in Table 1. The method was validated with the perspective to be applied for compliance testing by official food control laboratories. Therefore, it was aimed at achieving method performance characteristics to meet the provisions in the Commission Regulation No. 401/2006/EC and Commission Recommendation 519/2014/EU [20,21]. The mycotoxin levels in contaminated materials were set in order to encompass legal limits (were available), which were also chosen as spiking levels for recovery assessment (Table 1).

The recoveries were in the range of 71–97%, and the lowest values (71–78%) were obtained for NIV, the most polar mycotoxin. The relative standard deviations for repeatability (RSDr) were in the range of 2.2–34%, and the relative standard deviations for reproducibility (RSDR) were between 6.4% and 45%. Almost all the data fulfilled the EC acceptability criteria [20,21]. In a few cases, RSDR values higher than 40% were obtained—namely, for 3-AcDON at 75 μg/kg (wheat flour), 15-AcDON at 36 μg/kg (wheat A), 24 μg/kg (wheat flour B), 11 μg/kg (wheat crackers A), and for T-2 at 48 μg/kg (wheat flour B). Overall, the worst precision values were obtained for 15-AcDON. This could be attributed either to the lower number of valid results or the use of 13C17-3-AcDON as an internal standard.

The expected HorRat values, based on historical performance data and used as a guide to determine the acceptability of the precision of a method, are between 0.5 and 1.5, although the limits for performance acceptability are 0.5–2.0 [22]. In this study, the HorRat values ranged from 0.4 to 2.0, being in some cases slightly better than the expected ones on the basis of historical data (Table 1 and Figure 2). The performances of this collaborative study are in line with the data reported for other very recent collaborative studies for the validation of multimycotoxin LC-MS/MS methods [13,23], reporting HorRat values down to 0.2. Besides the expertise of the participating laboratories, the improved performances of the LC-MS/MS multimycotoxin methods can be attributed to the use of isotopically labelled internal standards, which improved the method precision by compensating for LC-MS signal drift over long analysis sequences as well as for matrix effects.

**Figure 2.** Reproducibility data obtained for all trichothecenes and ZEN in wheat, wheat flour, and cracker test materials in the mass fraction ranges 0–120 μg/kg and 120–2500 μg/kg.

Overall, the method validation results indicated that the focused mycotoxins can be reliably detected at levels encompassing the legal limits.

Figure 2 depicts plots of RSDR vs. mass fraction in the range 0–120 μg/kg (a) and 120–2500 μg/kg (b). The RSDR values were mostly between 22% and 45%.



#### *Toxins* **2020** , *12*, 786

 set


**Table 1.** *Cont.*


*Toxins* **2020** , *12*, 786

> **Table 1.** *Cont.*


**Table 1.** *Cont.*

a

Looking at the overall results, the main factors influencing recoveries and precision turned out to be the chemical nature of the toxin. Lower but still acceptable recoveries (≥71%) were obtained for NIV. With this being the most polar toxin targeted by the method, some analyte losses (higher than other mycotoxins) could occur in the solid phase extraction (SPE) clean-up loading/washing steps. Furthermore, a good chromatographic separation between 3- and 15-AcDON was confirmed to be critical to achieve satisfactory method performance results for both isomers. On the other hand, no clear influence of the matrix was observed on recoveries and precision, suggesting that the method can be considered "horizontal"—i.e., applicable to different commodities without changing/adapting the sample preparation and analysis procedure.

#### **3. Conclusions**

A method for the simultaneous determination of zearalenone and the major trichothecenes—namely, nivalenol, deoxynivalenol and its acetyl derivatives (3-acetyl DON and 15-acetyl DON), and HT2-and T-2 toxins—was validated in an international collaborative study involving 15 participant laboratories from 10 countries according to the AOAC/IUPAC (International Union of Pure and Applied Chemistry) International Harmonized Protocol [22]. The method was based on LC-MS/MS detection after SPE extract clean-up. The results of the collaborative study met the required method performance criteria, with HorRat values in the range of 0.4–2.0. Data from the collaborative study as well as feedback from participants confirmed that the chromatographic separation of 3-AcDON and 15-AcDON remains one of the challenging steps of the method, as well as achieving satisfactory recoveries for NIV, which were higher than 70%. Even though affecting the cost-effectiveness of the method, the use of isotopically labeled internal standards is essential to manage the matrix effect, contributing to achieve a satisfactory precision. Finally, the applied generic sample preparation procedure, based on polymeric SPE columns, has been already demonstrated to be applicable to a wide range of mycotoxins and food commodities [24], suggesting possible further extensions of the method scope. The method complies with the provisions of the commission regulations setting performance criteria for mycotoxin methods [20,21] and has been recently adopted as a CEN standard [17].

#### **4. Materials and Methods**

#### *4.1. Chemicals and Reagents*

Participants in the collaborative study were requested to use only reagents of recognized analytical grade and specifically the following: water, acetonitrile, and methanol of HPLC quality; ammonium acetate for mass spectrometry, c(CH3COONH4) <sup>≥</sup> 99.0%; solid-phase extraction columns Oasis® HLB 3 mL, 60 mg (Waters, Milan, Italy), or equivalent; standard NIV, DON, 3Ac-DON, 15Ac-DON, HT-2, T-2, ZEN crystalline, as a film or as certified standard solution; isotopically labelled internal standards ( 13C15-NIV, 13C15-DON, 13C17-3-AcDON, 13C22-HT-2, 13C24-T-2, 13C18-ZEN) as acetonitrile solutions (25 μg/mL).

### *4.2. Test Materials*

Three commodities—namely, common wheat (cereal), wheat flour (cereal flour), and wheat crackers (cereal-based food)—were selected. For each food commodity, one blank (containing mycotoxins at levels below the detection limit of the in-house validated method—i.e., 10 μg/kg NIV; 5 μg/kg DON; 1 μg/kg 3-AcDON and 15-AcDON; 0.5 μg/kg HT-2, T-2, and ZEN), one blank for spiking purposes (for recovery evaluation) and three contaminated materials (low, medium, and high level) with target mycotoxins were prepared. Desired mycotoxin levels in contaminated test materials were within a contamination range encompassing the EU maximum permitted levels (where applicable) (Table 1). All the materials were dispensed in plastic bottles (approx. 30 g each container) that were labeled, sealed, and stored at −20 ◦C until analysis for homogeneity and stability study or dispatch for collaborative trial testing.

#### 4.2.1. Preparation of Whole Soft Wheat and Soft Wheat Flour Test Materials

Uncontaminated common wheat kernels and soft wheat flour (about 10 kg each) were selected among a set of samples purchased from the Italian retail market. The absence of contamination was verified by analysis for the mycotoxin content according to relevant validated/reference methods [25–27]. Wheat kernels were milled by an ultra-centrifugal mill (ZM 200, Retsch, 500 μm sieve). About 4 kg of each commodity were used as blank, whereas the remaining 6 kg were used for the preparation of the contaminated materials.

Since naturally contaminated materials containing all focused mycotoxins at the desired levels were unavailable at the time of this study, the preparation of contaminated test materials was performed according to a previously developed protocol [12]. Briefly, contaminated test materials were prepared by mixing and homogenizing blank material (common wheat or common wheat flour) with culture extracts of *Fusarium* toxigenic species (deposited at the Institute of Sciences of Food Production collection, http://www.ispa.cnr.it/Collection) grown on wheat kernels. Fungal cultures of *F. graminearum* ITEM 126 (producing DON, ZEN, and 3-AcDON) and *F. sporotrichioides* ITEM 707 (producing T-2 and HT-2) were dried, ground, extracted, and analyzed for the mycotoxin content according to relevant validated/reference methods [25–27]. Subsequently, a multi-mycotoxin spiking solution containing NIV, DON, 3- and 15-AcDON, T-2, HT-2, and ZEN was prepared by mixing adequate amounts of culture extracts and standard mycotoxin solutions (for NIV and 15-AcDON). For each contamination level (low, medium, high), 2 kg aliquots of blank material were split into 0.5 kg portions and fortified with the multi-mycotoxin solution. After solvent evaporation (overnight at room temperature), the 0.5 kg portions were pooled and passed through an ultra-centrifugal mill (ZM 200, Retsch, 500 μm sieve), then homogenized by a laboratory mixer for 24 h.

This procedure resulted in three contaminated wheat materials (low, medium, and high level) and three contaminated wheat flour materials (Table 1).
