Dilution of QuEChERS Extracts Without Cleanup Improves Results in the UHPLC-MS/MS Multiresidue Analysis of Pesticides in Tomato

Tomato is well-known to be one of the most cultivated and consumed vegetables worldwide and frequently contain pesticide residues. Therefore, a simple multiresidue method was established and validated to determine 129 pesticides and metabolites in tomato samples using a modified acetate QuEChERS without cleanup for sample preparation and determination by ultrahighperformance liquid chromatography tandem mass spectrometry (UHPLC-MS/MS). Dilution of the raw extract in different proportions of mobile phase was evaluated and a dilution of 10 times presented adequate results improving analysis performance while minimizing the matrix effect. Validation performed according to SANTE guideline presented satisfactory results. Practical method limit of quantification was 0.01 mg kg for most compounds. Recoveries between 70 and 120% with precision ≤ 20% were found for most compounds and spike levels evaluated. Matrix effect results were not significant for most part of compounds. Method proved to be simple, robust, and effective to be applied in routine analysis. Method applicability was performed by analysis of samples commercialized in Brazil and positive results were found demonstrating the importance of the proposed method.


Introduction
According to the International Union of Pure and Applied Chemistry (IUPAC 1997), pesticides are substances or mixture of substances used in the production, harvest, or storage of food. In agricultural production, the use of pesticides has been recommended, mentioning economic benefits, as they are used to control pests and diseases that affect crops, ensuring increased production. However, when used in an irregular and indiscriminate manner, they are also responsible for the contamination of food, water sources, and soil, resulting in human health risks, besides the environmental consequences (Prestes et al. 2009). Food chain, air, water, soil, flora, and fauna are the main routes of human exposure to pesticides (Kim et al. 2017). As the production of fruits and vegetables is higher, it is estimated that these foods contribute to a greater daily intake of pesticides, when compared to other plant origin food groups such as breads and cereals.
In Brazil, the National Health Surveillance Agency (ANVISA) is the responsible organization that establishes the maximum residue limit (MRL) values of pesticide residue officially accepted in food after being treated with the recommended dose in the prescription of each product/crop case and observed the pre-harvest interval. In an international level, MRL values are established by the Codex Alimentarius for food trade.
Pesticide residues can be present in food samples since they are used to control pests and diseases that affect crops, ensuring the quality of production. In Brazil, the National Health Surveillance Agency (ANVISA) is the responsible organization that establishes the maximum residue limit (MRL) values of pesticide residues officially accepted in food samples. In an international level, MRL values are established by the Codex Alimentarius for food trade.
Tomato (Solanum lycopersicum; Lycopersicun esculentum or Lycopersicum lycopersicum) plays a significant role in human diet. It is one of the most cultivated and consumed vegetables worldwide, with high pigment content, and large amount of water (Melo et al. 2012).
Considering the complex food composition, the determination of pesticide residues requires sensitive and selective quantification techniques, due to matrix components, which may act over the analyte signal response, and the low level of concentration of these residues that also require effective sample preparation (Kemmerich et al. 2015;Ferreira et al. 2016;May et al. 2017). Currently, the QuEChERS method (of the achromous quick, easy, cheap, effective, rugged, safe) and its modifications are applied to a wide variety of matrices and compounds. Various sorbents and different sorbent combinations can be used depending on the type of matrix, chromatographic technique, and the compounds to be analyzed (Prestes et al. 2009). However, some sorbents used for the cleanup step should be carefully used because it can remove some pesticides resulting in low recoveries (Perestrelo et al. 2019). However, methods that use only one step of extraction, dispensing the cleanup step, have also been promisingly reported for rice , and urine, serum, tomato, soil, commercial dosage form, and water samples (Gebrehiwot et al. 2019). Lehotay (2019) reported that dispersive solidphase extraction (d-SPE) cleanup in QuEChERS method for determination of pesticide residues in fruits and vegetables helped reduce the amount of co-extracted matrix from the extracts and the matrix effect; however, the extent of cleanup was too small to justify this additional step. This cleanup step can lead to loss of analytes and increased time and costs with reagents. Considering the higher sensitivity of the current instrumentation, the dilution of the extract is an adequate solution to minimize matrix effects and increase the robustness of the methods.
Therefore, the aim of this work was to test the hypothesis that dilution of extracts will yield greater accuracy for wider scope of pesticides than analysis of more concentrated extracts after cleanup, achieving low limits of quantification without extra instrument maintenance needs. A simple and effective sample preparation procedure was established for multiclass determination of pesticide residues in tomato by UHPLC-MS/ MS analysis using a triple quadrupole (QqQ) system.

Sample Preparation Procedure Evaluation
Tomato blank samples were acquired from a local organic producer from Santo Ângelo (Brazil). A portion of 1 kg was processed according to the Codex Alimentarius (CODEX 2020) with shell and seeds, and stored in freezer at − 10°C in identified PP flasks until use. Three different QuEChERS versions as original (Anastassiades et al. 2003), acetate (Lehotay et al. 2005), and citrate (Anastassiades et al. 2007) were evaluated for the extraction of target analytes from tomato sample. These tests were conducted in blank samples spiked at 0.05 mg kg −1 . The methods of extraction performance were evaluated, following the experimental conditions described below, based on the recovery (%) and relative standard deviation (RSD%). Original QuEChERS: 10 g sample were weighed in 50 mL PP disposable extraction tubes. After that, 10 mL of acetonitrile were added and the mixture was vigorously shaken for 1 min. For the partitioning step, 1 g of NaCl and 4 g of anhydrous MgSO 4 were added. The tubes were shaken for 1 min again and the tubes were centrifuged for 8 min at 2140 g. Acetate-buffered QuEChERS: 10 g sample were extracted with 1% (v/v) of acetic acid in acetonitrile. Then, 1.7 g of anhydrous sodium acetate and 4 g of anhydrous MgSO 4 were added and the tubes were shaken and centrifuged for 8 min at 2140 g. Citrate-buffered QuEChERS: 10 g sample were extracted with 10 mL acetonitrile. After, were added 1 g of NaCl, 4 g of anhydrous MgSO 4 , 1 g trisodium citrate dehydrate, and 0.5 g disodium hydrogen citrate sesquihydrate and the tubes were shaken and centrifuged for 8 min at 2140 g.
The three variations of QuEChERS method were first evaluated using dispersive solid-phase extraction (d-SPE) as follows: 150 mg of MgSO 4 and 25 mg PSA (QuEChERS original); 150 mg of MgSO 4 and 50 mg PSA (QuEChERS acetate); and 150 mg of MgSO 4 and 25 mg PSA (QuEChERS citrate) per mL of extract, respectively. Tubes were centrifuged for 10 min at 2140 g; the supernatant was filtered through a PTFE syringe filter (0.2 μm) and transferred to autosampler vials for UHPLC-MS/MS analysis. In order to evaluate different d-SPE sorbents and the injection without cleanup, two different tests were performed to improve the cleanup step: (i) d-SPE was evaluated with PSA (125 mg), C18 (250 mg), GCB (50 mg), Florisil (deactivated and activated) (250 mg), chitosan (125 mg), alone or in combination. The results were evaluated based on the extracts visual characteristics, considering those that presented cleaner extracts, with less turbidity and coloration. To improve the cleanup efficiency, the quantities of the sorbents that showed the best performance were optimized through central composite design (CCD) using the Statistic 7.0 software; (ii) to evaluate the injection without cleanup, the extract was diluted in different rates (2-, 2.5-, 3-.33-, 5-, and 10-fold, v/v) with mobile phase before injection.

Optimized Method
The optimized sample preparation procedure (Fig. 1) consisted of 10 g of homogenized sample in a 50 mL PP tube, spiked with atrazine-d5 (0.05 mg kg −1 ) as surrogate standard. The sample was extracted with 10 mL of acetonitrile with 1% (v/v) of acetic acid by manually shaking for 1 min. After this step, 4 g of MgSO 4 and 1.7 g of CH 3 COONa were added. Then, the mixture was manually shaken for 1 min and centrifuge at 2420 g for 8 min for the partitioning step. The supernatant was filtered in PTFE filter (0.2 μm) and diluted 10-fold d i l u t i o n ( 1 + 9 , v / v ) w i t h t h e m o b i l e p h a s e . Triphenylphosphate, used as internal standard (IS), was added at 0.02 mg L −1 .

Validation Procedure
The method established in this work was validated in accordance with SANTE (2019) guideline based on selectivity, linearity, matrix effect, limit of detection (LOD), limit of quantification (LOQ), accuracy, and precision (repeatability and intermediate precision) which were evaluated. Selectivity was evaluated by comparing the chromatograms obtained from matrix blank and spiked samples. Linearity was checked from six concentration levels (0.001, 0.002, 0.004, 0.008, 0.02, and 0.05 mg L −1 ) with criteria of residuals < 20%. Matrix effect (ME) was estimated comparing the slopes of curves prepared in matrix blank extract and solvent (acetonitrile). ME results were expressed in percent and the criteria for considering the effect significant was larger than ± 20%. Furthermore, identification criteria for MS/MS was Fig. 1 Representation of the modified QuEChERS sample preparation step proposed considered as a minimum of two product ion for compound, signal-noise (S/N) ≤ 3 and ion ratio variation between ± 30%. Accuracy with criteria of 70-120% of acceptance and associated repeatability (RSDr) and intermediate precision (RSDip) values ≤ 20% was adopted as criteria of acceptance.
In this study, for accuracy and precision evaluation, recovery assays at the levels 0.01, 0.02, 0.04, and 0.5 mg kg −1 (n = 7) were performed. Intermediate precision was evaluated in different days at the spike level 0.04 mg kg −1 (n = 7). Method LOQ was established as the lowest concentration of the analyte with accuracy of 70-120% and associated precision as repeatability (RSDr) ≤ 20% was adopted. LOD was calculated dividing the LOQ by 3.3.

Application in Commercial Samples
The proposed method was applied for pesticide residues determination in sixteen samples obtained from farmers and local markets of the state of Rio Grande do Sul (Brazil). Samples of 1 kg of tomato were collected, homogenized with dry ice, and extracted as described in this work.

Determination of Pesticide Residues by UHPLC-MS/MS
The determination of the target pesticides in tomatoes using QuEChERS method without cleanup step was possible due to the good selectivity and sensitivity of the UHPLC-MS/MS system. The mobile phase and gradient program promoted a proper separation of compounds. These conditions were already used in our laboratory by Martins et al. (2017) andde Matos et al. (2019), providing adequate analytical signals for all the compounds selected in our study. The chromatographic conditions provided high resolution and good peak shape for all the evaluated compounds. UHPLC-MS/MS retention times (t R ), SRM transitions, capillary voltages (CV), and collision energies (CE) were optimized for each analyte and are showed in Table 1. The transition with the highest intensity was selected for quantification, and the transition with the second highest intensity was used for identification.

Sample Preparation Procedure
In this work, sample preparation strategies based on QuEChERS method (original, acetate, and citrate versions) were evaluated based on recoveries and precision results. All 129 compounds were extracted with the three evaluated methods, but with different recovery efficiencies. The acetate QuEChERS method extracted more compounds (119) between 70 and 120% in comparison with original QuEChERS (113) and citrate QuEChERS (105). Therefore, the acetate method was chosen as the most efficient method and the next evaluations were performed with this procedure.
The efficiency of the different sorbents for tomato pigments removal was also evaluated using chitosan (1); Strata-X (2); PSA (3); PSA + C18 (4); PSA + deactivated Florisil (5); PSA + activated Florisil (6), and PSA + GCB (7). Figure 2 presents the visual characteristics of each test and showed an important difference in the pigmentation between the extracts. The combination of PSA and GCB (test 7) showed a good alternative when compared to the other combinations used, since it was possible to remove more pigments.
From the results obtained with these preliminary tests, an optimization using central composite design (CCD) and response surface methodology (RSM) based on the work presented by Rizzetti et al. (2016) was performed to establish the best cleanup conditions. The amounts of the two factors PSA and GCB were optimized testing the amounts corresponding to the four corners of the square represent the factorial design points, four star points of the axial (± alpha) design points, and the replicated center point, as described in Table 2.
In order to verify the influence of the PSA and GCB variables, Pareto graphs (Fig. 3a) was obtained and the response surfaces, considering the peak area of the analytes, were generated to indicate the best conditions. Results for the optimization of the amount of these sorbents carried by CCD are presented in Fig. 3b.
The compounds azinphos-ethyl, carbendazim, chlorpyrifos, deltamethrin, phosalone, phosmet, terbufos, terbuthylazine, tebuconazole, thiabendazole, among others, showed a decrease in areas when using GCB, whereas for the other compounds there is no difference in terms of response. This is because the compounds mentioned are planar compounds and are known to be retained in the GCB. The influence of the GCB on the response is confirmed when the response surfaces are evaluated (Fig. 3b), which indicate that better peak areas are obtained when low amounts of GCB are used. On the other hand, the surface indicates that the amount of PSA does not influence the areas obtained. Thus, we opted to use the amounts of 5 mg of GCB and 50 mg of PSA for cleaning 1 mL of extract to obtain clean extracts with better response.
In order to evaluate the need of a cleanup step, we also performed dilution tests of the raw extract, obtained by acetate QuEChERS method, with mobile phase. The extracts were diluted in different rates 2-, 2.5-, 3.3-, 5-, and 10-fold. The evaluation of the dilution rates was performed based on the chromatographic peak shape and matrix effect for each dilution. The matrix effect was obtained by comparing the areas from the analytical solutions in solvent (acetonitrile) and the solutions prepared in matrix matched extract as described by Pucci et al. (2009). Figure 4 shows the matrix effect graphs for the different dilutions rates tested. Values between − 20 and + 20% is not considered an important matrix effect, since this variation is close to the repeatability values. Values between − 50 and − 20% or + 20 and + 50% represent a medium matrix effect and values below − 50% and above + 50% indicate a strong matrix effect (Ferrer et al. 2011). As can be observed, a higher number of compounds was distributed in the range of − 20 and + 20% for the extracts with 10-fold dilution. This dilution proportion showed a better chromatographic peak shape, when compared to the other dilutions. Therefore, the 10 times dilution was defined as the most proper for the sample preparation step. In addition, the efficiency of 10-fold dilution showed similar results, when compared to the use of sorbents for cleanup the extract. Since that, we decided to skip the cleanup step in the sample preparation to simplify the method. In this sense, Lehotay et al. (2010) pointed out that the inclusion of some steps of the QuEChERS methods, such as in relation to pH adjustments, different amounts of sorbents in the cleanup step, and the use of a freezing step, did not provide any considerable benefit, increasing the time, complexity, and cost of the method. The reduction of matrix effects was evaluated by Stahnke (2012) at 10 levels of dilution up to 1000-fold for the commodities avocado, black tea, orange, and rocket (arugula) spiked with 39 pesticides and extracted by the QuEChERS citrate method. Results showed that a factor of 25-40 reduces ion suppression to less than 20% if the initial suppression is ≤ 80%. For stronger matrix effects or complete elimination of suppression, higher dilution factors were needed. In the case of extracts obtained by the QuEChERS method, suppressions between 25 and 50% may be eliminated by 10-fold dilution, the same factor used in our work. Yang et al. (2015) evaluated the relationship between matrix effects, chromatographic separations, and elution patterns of pesticides and of matrix components of avocado, spinach, honey, orange, and hazelnut at dilution factors up to 100-fold of the extracts obtained by the QuEChERS acetate method. Results showed that the higher dilution factor could minimize matrix effects for most pesticides and allow analysis without cleanup to improve data quality and operation efficiency. Honey and spinach presented high percentage of pesticides free of matrix effects already at 10-fold dilution, demonstrating that this procedure is effective for these matrices. Ferrer et al. (2011) investigated dilution of extracts of orange, tomato, and leek obtained with the QuEChERS citrate to reduce matrix effects of 53 pesticides. For tomato extracts without dilution, almost 30% of the pesticides showed medium signal suppression, and 2% of the selected pesticides had a strong matrix effect. Dilution factors from 2 to 100 were evaluated and a dilution factor of 15 demonstrated to be effective to eliminate most of the matrix effects, although a dilution factor of 10 presented similar results.

Method Validation Results
The optimized method was validated for 131 compounds in accordance with SANTE (2019) guideline. Figure 5a shows a comparison of the chromatograms obtained from blank and spike tomato sample at the level 0.01 mg kg −1 for flutolanil indicating that the method is very selective. No interferences for the selected compounds were observed.

Application to Commercial Samples
The validated method was applied to commercial samples from the state Rio Grande do Sul. Table 3 presents the result of the method application in sixteen commercial samples to evaluate method performance in routine analyses. Ten pesticides were found in the samples with concentrations between 0.011 and 0.238 mg kg −1 . Six samples presented no detectable residues of the evaluated compounds. The compounds found were chlorantraniliprole (0.011 mg kg −1 ), chlorpyrifos (0.017 and 0.022 mg kg −1 ), d i f l u b e n z u r o n ( 0 . 0 6 3 m g k g − 1 ) , f a m o x a d o n e (0.023 mg kg −1 ), imidacloprid (0.098, 0.016, 0.012, and  Being that, the fungicide pyraclostrobin, famoxadone, and the insecticide diflubenzuron are present in sample above the MRL established by European Union and are not authorized in Brazil. However, the compounds azoxystrobin and clothianidin remained with concentrations below the LOQ (0.010 mg kg −1 ). Chlorantraniliprole was found in one sample and is not authorized in Brazil for tomato. Figure 5b presents the chromatogram comparing the signal of the analysis of the blank of solvent and reagents, the blank tomato sample spiked with pyraclostrobin, and a tomato sample containing this pesticide.

Conclusion
In this study, a simple and fast modified QuEChERS method without cleanup was proposed for extraction of pesticide residues from tomato samples for the quantification by UHPLC-MS/MS after a dilution of the final extract in the mobile phase. This is extremely important during the development of the work, since an extract with a lower amount of interferents is essential for the maintenance of the UHPLC-MS/MS system, in addition to influencing the recovery of the analytes. Thus, the use of a largest dilution factor (10×) increased the detectability and also streamlined the extraction procedure. Satisfactory validation results were obtained for the proposed  MRL maximum residue limit, NA not authorized for tomato, n.d. not detected, < LOQ below the limit of quantification (LOQ) method that presented sufficient sensitivity to be applied in samples from routine analysis and monitoring programs based on the low method LOQ of 10 μg kg −1 obtained for the majority of compounds. The method was validated according to SANTE (2019). The parameters selectivity, linearity, matrix effect, accuracy, and precision were evaluated. Of the 131 applicable pesticide residues, two did not present acceptable recovery and/or RSD results, being acetamiprid and imazapyr. Therefore, these compounds were not considered validated. Finally, samples analyzed by the proposed method showed residues of ten pesticides in concentration ranging from 0.011 to 0.238 mg kg −1 . Therefore, the proposed method, suitable for 129 compounds, is quick and simple to perform, allowing easier implementation in the laboratory routine than other methods currently available.