**Development of a High-Throughput Screening Analysis for 195 Pesticides in Raw Milk by Modified QuEChERS Sample Preparation and Liquid Chromatography Quadrupole Time-of-Flight Mass Spectrometry**

**Xingqiang Wu 1, Kaixuan Tong 1, Changyou Yu 2, Shuang Hou 2, Yujie Xie 1, Chunlin Fan 1, Hui Chen 1,\*, Meiling Lu <sup>3</sup> and Wenwen Wang <sup>3</sup>**


**Abstract:** This study aimed to develop a simple, high-throughput method based on modified QuECh-ERS (quick, easy, cheap, effective, rugged, and safe) followed by liquid chromatography quadrupole time-of-flight mass spectrometry (LC-Q-TOF/MS) for the rapid determination of multi-class pesticide residues in raw milk. With acidified acetonitrile as the extraction solvent, the raw milk samples were pretreated with the modified QuEChERS method, including extraction, salting-out, freezing, and clean-up processes. The target pesticides were acquired in a positive ion electrospray ionization mode and an All ions MS/MS mode. The developed method was validated, and good performing characteristics were achieved. The screening detection limits (SDL) and limits of quantitation (LOQ) for all the pesticides ranged within 0.1–20 and 0.1–50 μg/kg, respectively. The recoveries of all analytes ranged from 70.0% to 120.0% at three spiked levels (1 × LOQ, 2 × LOQ, and 10 × LOQ), with relative standard deviations less than 20.0%. The coefficient of determination was greater than 0.99 within the calibration linearity range for the detected 195 pesticides. The method proved the simple, rapid, high throughput screening and quantitative analysis of pesticide residues in raw milk.

**Keywords:** raw milk 1; pesticides 2; screening 3; QuEChERS 4; high-throughput 5

#### **1. Introduction**

Milk is considered an important part of a healthy diet, providing essential nutrients and energy. High-quality raw milk is required by dairy factories to make dairy products, such as cheese, yogurt, and cream [1]. Once the raw milk is defective, it cannot be improved in the subsequent processing, which may have far-reaching effects. Currently, China is one of the world's largest producing and consuming countries of milk and dairy products, with the per capita consumption of milk in China increasing from 4.89 kg in 1997 to 19.2 kg in 2019 [2]. The quality and safety of milk and its products are of a great concern to both the government and consumers [3]. Meanwhile, the contamination of milk with pesticide residues is a severe concern in many countries [4–6]. Pesticide residues in milk may come from direct or indirect sources such as feeding animals from contaminated forage grass, feeding and drinking water, and various pesticides used to treat pests, pathogens, and fungal diseases [7]. Through the above pathways, these pesticide residues inevitably accumulate in animals. They are transferred to secreted milk, with serious health hazards

**Citation:** Wu, X.; Tong, K.; Yu, C.; Hou, S.; Xie, Y.; Fan, C.; Chen, H.; Lu, M.; Wang, W. Development of a High-Throughput Screening Analysis for 195 Pesticides in Raw Milk by Modified QuEChERS Sample Preparation and Liquid Chromatography Quadrupole Time-of-Flight Mass Spectrometry. *Separations* **2022**, *9*, 98. https:// doi.org/10.3390/separations9040098

Academic Editor: Chiara Emilia Cordero

Received: 28 March 2022 Accepted: 11 April 2022 Published: 12 April 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

likely to occur as humans consume contaminated milk or dairy products [8,9]. Hence, it is necessary to ascertain pesticide residues in milk to ensure safe dietary intake.

To ensure food safety, several organizations and countries, such as the European Commission [10] and China [11], have established maximum residue limits (MRL) for various pesticides in milk. Therefore, to meet these requirements, there is an increasing need for an effective analytical method for simultaneous qualitative and quantitative screening of pesticide residues in milk. The current reported methods for the analysis of multiresidue pesticides in milk use different detection techniques, such as high-performance liquid chromatography with diode-array detection (HPLC-DAD) [12], gas chromatography– electron capture detection (GC-ECD) [13], gas chromatography–mass spectrometry (GC-MS) [14], gas chromatography–tandem mass spectrometry (GC-MS/MS) [15,16], and liquid chromatography–tandem mass spectrometry (LC-MS/MS) [17–19]. Recently, liquid chromatography coupled with high-resolution mass spectrometry techniques (LC-HRMS) had been applied to determine pesticide residues in milk matrices [20,21]. LC-HRMS offered the ability to collect full scan spectra and accurate masses while acquiring and reprocessing data without prior compound-specific adjustments, enabling retrospective data analysis [22]. Hence, LC-HRMS has a strong competitive advantage compared with low-resolution mass spectrometry in the multi-residue analysis of compounds and has demonstrated great potential for non-targeted detection.

Although LC-HRMS demonstrates high sensitivity and accuracy in developing analytical methods, selecting a suitable sample preparation method is an important prerequisite for achieving multi-residue analysis. Milk is a complex matrix in which interfering components (e.g., proteins, fatty acids, and pigments) may play a role in suppressing the signal of pesticide residues. Therefore, effectively reducing matrix interference is crucial for determining pesticide residues in milk [23]. Different sample preparation methods for extracting pesticides from milk have been explored. These methods mainly include liquid–liquid extraction (LLE) [19,24], gel permeation chromatography (GPC) [15], solid-phase extraction (SPE) [5,25], dispersive solid-phase extraction (d-SPE) [21], and the QuEChERS (quick, easy, cheap, effective, rugged, and safe) method [13,14,16,18]. Among them, GPC and SPE are tedious and time-consuming to operate, which do not facilitate the processing of a large number of samples. Meanwhile, LE and d-SPE methods have a large background interference of the sample matrix after pretreatment, which causes a decrease in detection sensitivity of the analytical instrument [26]. QuEChERS is fast, safe, and low-cost in the aforementioned techniques, including extraction and purification steps. Compared to other sample preparation techniques, QuEChERS is simple to use and has efficiency improvement with good reproducibility and stability. The QuEChERS method has been widely used for the high-throughput analysis of chemical contaminants in various food products [27].

This work aimed to establish a simple and efficient pretreatment method for the simultaneous detection of multi-pesticide residues in raw milk using an advanced LC-Q-TOF/MS technique. The pretreatment procedure was optimized, including different extraction salts, purification sorbents, and freezing times. Meanwhile, this method's linearity, sensitivity, accuracy, precision, and matrix effect were fully evaluated. Finally, a simple and effective sample preparation procedure was established to determine 195 pesticide residues in raw milk combined with LC-Q-TOF/MS. Moreover, the validated method was employed to screen pesticide residues in actual raw milk samples from dairy farms.

#### **2. Materials and Methods**

#### *2.1. Instrumentation*

The liquid chromatography quadrupole time-of-flight mass spectrometry (1290–6550) was from Agilent Technologies (Santa Clara, CA, USA). Chromatographic separation was achieved on a chromatographic condition: equipped with a reversed-phase chromatography column (ZORBAX SB-C18 column 2.1 mm × 100 mm, 3.5 μm; Agilent Technologies, Santa Clara, CA, USA); mobile phase A is 5 mM ammonium acetate-0.1% formic acid-water; mobile phase B is acetonitrile; gradient elution program, 0 min: 1%B, 3 min: 30%B, 6 min: 40%B, 9 min: 40%B, 15 min: 60%B, 19 min: 90%B, 23 min: 90%B, 23.01 min: 1%B, run after 4 min. The flow rate was set at 0.4 mL/min. The column temperature was 40 ◦C. The injection volume was 5 μL.

An Agilent Dual Jet Stream electrospray source was used on the Q-TOF in positive ionization mode. The conditions for mass spectrometry were set as follows: Scan mode: All ions MS/MS; capillary voltage was 4 kV; nebulizer gas was 0.14 MPa; drying gas temperature was at 325 ◦C with a flow rate of 12.0 L/min; sheath gas temperature was set at 375 ◦C with a flow rate of 11.0 L/min; Fragmentation voltage at 145 v. All Ions MS/MS mode parameter settings: acquisition range was *m*/*z* 50–1000, data acquisition rate is four spectra/s; collision energy was 0 eV at 0 min, and collision energy was set to 0, 15, and 35 eV in consecutive order after 0.5 min.

The mass spectrum information of 195 pesticide databases is shown in Table 1. PL602- L electronic balance was purchased from Mettler-Toledo Co., Ltd. (Zurich, Switzerland); N-112 Nitrogen evaporator concentrator was obtained from Organomation Associates (EVAP 112, Worcester, MA, USA); SR-2DS oscillator was obtained from Taitec company (Saitama, Japan); KDC-40 Low-speed centrifuge was obtained from Zonkia Group Corp., Ltd. (Hefei, China); Milli-Q ultrapure water machine was obtained from Millipore Co., Ltd. (Milford, MA, USA).

#### *2.2. Reagents and Materials*

Raw milk samples were collected from local dairy farms. All pesticide standards (purity grade, >98%) were obtained from Alta Company (Tianjin, China). Formic acid, ammonium acetate, acetonitrile, methanol (all LC-MS grade), and toluene (HPLC grade) were obtained from Fisher Scientific, Inc. (Fair Lawn, NJ, USA). Analytical grade forms of acetic acid, sodium chloride, anhydrous Na2SO4, trisodium citrate, disodium citrate, and anhydrous MgSO4 were obtained from Shanghai Anpu Experimental Technology (Shanghai, China). The cleanup absorbents as octadecylsilane (C18) and primary secondary amine (PSA) were obtained from Tianjin Agela Technology (Tianjin, China).

#### *2.3. Preparation of Standard Solutions*

Standard stock solutions of individual pesticides were prepared in acetonitrile, methanol, or water to a concentration of 500–1000 mg/L. All stock solutions were stable for 6 months in a closed tea-colored volumetric flask at −20 ◦C. The 10 mg/L intermediate working solution and the working internal standard solution (Atrazine-D5) were prepared by diluting the stock solution with methanol. Working solutions were prepared daily by diluting a stock solution with all pesticides and used immediately after preparation.

#### *2.4. Sample Preparation*

The QuEChERS procedure entailed the following steps: 2.0 g of raw milk sample were weighed into the 50 mL tube. 16 mL of 1% acetic acid acetonitrile (*v*/*v*) was added, followed by EN salt (4 g MgSO4, 1 g NaCl, 0.5 g disodium citrate, and 1 g trisodium citrate), vortexed for 1 min, and shaken for 2 min. After that, the sample tubes were frozen at <sup>−</sup><sup>20</sup> ◦C for 0.5 h and then centrifuged (4200 rpm) for 5 min. 5 mL of supernatant was again pipetted into a 15 mL clean-up tube (containing 500 mg MgSO4 and 200 mg C18). The clean-up tube was vortexed for 5 s and then shaken for 2 min, followed by centrifugation at 4200 rpm for 5 min. Subsequently, 2 mL of the supernatant from the clean-up tube was pipetted into a 10 mL glass tube and evaporated to dryness in a 40 ◦C water bath with a gentle stream of nitrogen. Finally, 1 mL of acetonitrile/water (3:2, *v*/*v*) solution was used to redissolve the solution and pass it over the membrane for LC-Q-TOF/MS analysis.








**Table 1.** *Cont.*














RT: retention time; SDL: screening detection limit; LOQ: the limit of quantification; MRL: maximum residue limits; R2: coefficient of determination. "—" means no MRL value.

#### *2.5. Validation of the Method*

The method was validated in the raw milk matrix by evaluating the following parameters: screening detection limit (SDL), the limit of quantification (LOQ), linearity, matrix effect, accuracy, and precision. To define the SDL, refer to the European SANTE/12682/2019 guidelines [28]. LOQs were assessed by determining the lowest concentration of spiked samples where recovery and precision were satisfactory (70–120% and less than 20%, respectively). Calibration curves were investigated by determining the results of a series of standard addition recovery experiments (1–200 μg/kg) of blank matrix extract solutions before injection. Matrix effects were evaluated by comparing the slope of the matrix-matched calibration curve with the solvent calibration curve. To validate the accuracy and precision of the established method, recovery studies were performed for each substrate in six replicates for three spiked levels at 1 × LOQ, 2 × LOQ, and 10 × LOQ.

Agilent Mass Hunter (version B. 08.00) software was used to analyze the data based on the self-built database. To ensure the accuracy of target pesticide identifications, the specific settings of the corresponding screening parameters included the retention time offset threshold (≤ 0.15 min), the co-exist score (≥15), the signal-to-noise ratio (≥3), the mass deviation (≤10 ppm), and the number of characteristic ions in the qualitative identification of compounds (5:2). The data results were analyzed and summarized by Microsoft Excel 2016 (Seattle, WA, USA) software, and the analysis of graphs was drawn by Origin 2018 software.

#### **3. Results**

#### *3.1. Optimization of the QuEChERS Procedure*

The QuEChERS procedure was evaluated due to the possibility of matrix interferences influencing the identification of compounds, which are the most challenging situations in high-throughput screening and are also required to validate quantitative determination. For this reason, different procedures based on the QuEChERS method have been evaluated as follows.

#### 3.1.1. Optimization of the Extraction Solvent Volume

This study used acetonitrile with 1% acetate as an extraction solvent because it can extract various compounds with different polarity ranges and is the most effective organic solvent in multi-residue methods [17,18,20]. The volumes of extraction solution, such as 10 mL, 16 mL, and 20 mL of acetonitrile with 1% acetate, were compared to improve the extraction efficiency. In the spiked level of 100 μg/kg, the detected pesticides were 170, 173, and 166, respectively, using 10 mL, 16 mL, and 20 mL of acetonitrile with 1% acetate for raw milk. By 10 mL of the extraction solution, the final sample solution contains a high matrix background interference, affecting the definitive identification of compounds under the same purification conditions. Moreover, when the extraction solution volume was 20 mL, the sample solution was diluted by a factor of five, which noticeably reduced the sensitivity of the compound detection. Ultimately, the relatively good experimental results could be found when the volume of the extraction solution was 16 mL. Considering the response of the target pesticide and background interference, 16 mL acetonitrile with 1% acetate was selected for the extraction solvent.

#### 3.1.2. Optimization of the Type of Extraction Salt

The matrix environment, especially pH, may play an essential role in extracting some pesticides during the extraction process. Therefore, the effect of pH on pesticide recovery has been frequently investigated in many studies [27]. Extraction salts could adjust the pH of the matrix and affect the extraction efficiency by reducing the solubility of the target pesticides in an aqueous solution and enhancing their transfer into the extraction solution. To assess the extraction salt, the various compositions of salt pocket from the initial method (4 g anhydrous MgSO4 and 1 g sodium chloride), the AOAC method (6 g anhydrous MgSO4 and 1.5 g sodium acetate), and the EN method (4 g anhydrous MgSO4, 1 g anhydrous NaCl, 1 g dihydrate trisodium citrate, and 0.5 g disodium citrate) [29] were compared. As shown in Figure 1, the number of pesticides with the recovery in 70–120% by the EN method was slightly higher than the other two methods. This is because citrate buffering (EN) gently adjusts the pH of the matrix to between 5.0 and 5.5, enabling the satisfactory recovery of some sensitive pesticides under acidic or basic conditions. The results also verified that pH-sensitive pesticides, such as carbofuran and carbofuran-3-hydroxy (carbamate pesticides), had good performance and stability effects through EN buffer salts. Therefore, the EN method salt pocket was selected.

**Figure 1.** Recoveries (%) obtained for various salt pockets methods; (**A**) 4 g anhydrous MgSO4,1g sodium chloride, (**B**) 4 g anhydrous MgSO4, 1 g anhydrous NaCl, 1 g dihydrate trisodium citrate and 0.5 g disodium citrate, and (**C**) 6 g anhydrous MgSO4, 1.5 g sodium acetate.

#### 3.1.3. Optimization of the Freezing Temperature

The low-temperature precipitation step enables the removal of a large proportion of interfering substances, such as lipids, fats, and proteins that may be extracted along with the target pesticide residues. The significant advantage of this purification technology is that it is simple to operate and does not require specialized equipment [30]. The main components of milk are protein and animal oil esters. Therefore, it was necessary to use a low-temperature precipitation method for the raw milk to reduce the co-extracts in the extracts. As shown in Figure 2, the TIC chromatograms of different experimental groups overlapped, indicating a significant reduction in the signal intensity of co-extractives and matrix-derived interferences under low-temperature conditions. Meanwhile, the results showed that the recovery and precision of pesticides frozen at −20 ◦C for 0.5 h were better than those of the experimental group without freezing. Still, the results were similar to those of the experimental group frozen for 1.0 h. Thus, a freezing time of 0.5 h was chosen in the final method.

**Figure 2.** LC-Q-TOF/MS Total ion chromatogram overlap showing the effect of freezing (Blueline: without freezing; Redline: freezing 0.5 h; Greenline: freezing 1.0 h).

#### 3.1.4. Optimization of the Purification Adsorbent

Despite the sample solution being frozen-out to remove most of the interfering substances, the remaining matrix components may still interfere with the determination and contaminate the LC-Q-TOF/MS system, so it is necessary to develop an additional efficient clean-up step. Sorbents play a crucial role in the QuEChERS method. Various sorbents such as primary secondary amines (PSA) and octadecyl (C18) are often used for sample clean-up in pesticide residue analysis. C18 is a reversed-phase adsorption material that removes non-polar impurities such as lipids, cholesterol, and lipophilic compounds. PSA is a weak anion exchange sorbent that could adsorb polar molecules and effectively remove co-extracted components from the matrix, such as organic acids and sugars [27].

Raw milk is a complicated matrix with high lipid, fat, and protein intensities. Thus, the optimization of the purification step is achieved by different adsorbent combinations and dosage variables. In the present experiment, 500 mg of anhydrous magnesium sulfate was applied to remove the residual water. In addition, five different types of sorbents (100 mg of C18, 200 mg of C18, 300 mg of C18, 50 mg of PSA, and 50 mg of PSA + 200 mg of C18) were tested to investigate the influences on recoveries in raw milk.

According to SANTE/12682/2019 guidelines, the acceptable recovery interval is 70–120%, with an RSD less than or equal to 20% for multi-residue methods. As shown in Figure 3, the most significant number of pesticides with satisfactory recoveries and RSDs were found when 200 mg of C18 was used, along with better peak shapes and less matrix interference for some drugs, such as thiophanate-methyl. It may be that 200 mg of C18 can remove more interfering substances without affecting the pesticide detection, but excessive use of C18 will adsorb pesticides to reduce the recovery. Meanwhile, PSA adsorbent alone could not effectively remove lipids and proteins, which affected the detection of target pesticides. Finally, based on these results, 200 mg of C18 was selected as the sorbent to clean-up raw milk samples in this study.

**Figure 3.** Comparison of different sorbents for dispersive-SPE clean-up of analytes in raw milk. (**A**): 100 mg C18; (**B**): 200 mg C18; (**C**): 300 mg C18; (**D**): 50 mg PSA; and (**E**): 50 mg PSA+ 200 mg C18.

#### *3.2. Matrix Effect*

The co-eluting components, such as lipids, fats, and proteins in raw milk interfere with the ionization of pesticides with the suppression or the enhancement of the response. The formula evaluated the matrix effect in raw milk: the matrix effect (ME, %) = (slope of the matrix standard curve/slope of the solvent standard curve − 1) × 100. Matrix effects can be classified into three categories based on the results of the calculated data (Strong matrix effect: |ME| ≥ 50; Medium matrix effect: 20 < |ME| < 50; and Weak matrix effect: |ME| ≤ 20) [23]. As shown in Figure 4, more than 89.2% of the pesticides had a weak matrix effect in raw milk. The data results indicate that the method accurately analyzes trace pesticide residues in milk.

**Figure 4.** Matrix effect distribution of pesticides in raw milk analysis methods.

#### *3.3. Method Validation*

The linearity, SDL, LOQ, accuracy, and precision were determined to evaluate the performance of the modified QuEChERS method. The linearity was selected in the 1–200 μg/kg concentration range. As presented in Table 1, the coefficients of determination (R2) were higher than 0.99 for the pesticides in different linear ranges.

The sensitivity of the method was performed by SDL according to SANTE/12682/2019. SDLs were determined by spiking a series of mixed standard solutions in 20 blank samples and the lowest level at which pesticides had been screened in at least 95% of the samples [28]. As shown in Figure 5A, the percentage of pesticides with SDLs no more than 10 μg/kg was 93.3% for raw milk. LOQs were determined as the lowest validated spike level based on the recovery results by spiking a series of mixed standard solutions in blank samples. For raw milk, the LOQs were in the range of 0.5–50 μg/kg, and more than 87.2% of pesticides were less than or equal to 10 μg/kg, as shown in Figure 5B. The details of the SDLs and LOQs are listed in Table 1.

**Figure 5.** The distribution of the screening and quantification limits of pesticides in raw milk: (**A**) SDL distribution of pesticides in raw milk; (**B**) LOQ distribution of pesticides in raw milk.

For the accuracy and precision assessment, six replicates at three spiked levels were used, including 1 × LOQ, 2 × LOQ, and 10 × LOQ. The overall accuracy values for quantifying target pesticides in raw milk through recovery experiments ranged between 70.0% and 119.8%. The lowest accuracy value was relative to aminopyralid (70.0%). Thus, the method's precision can be considered appropriate (SANTE/12682/2019). For 195 pesticide residues, the RSD values ranged from 0.5 to 20.0% under in-laboratory conditions in all recovery experiments, indicating that the method's precision was acceptable. Therefore, it could be concluded that the modified QuEChERS method was sufficiently sensitive to determine the residues of the investigated pesticides in raw milk samples. The experimental results of the method performance evaluation, including recovery values (Rec, %) and RSD (%), are shown in Table 1.

#### *3.4. Analysis of Real Samples*

The established method was applied to 21 actual raw milk samples collected from local dairy farms in China (six batches from the Inner Mongolia Autonomous Region, six batches from Shaanxi Province, six batches from Shandong Province, and three batches from Hebei Province). Raw milk samples were collected at the dairy farm, transported to the laboratory using the cold chain, and stored at −20 ◦C. Samples need to be thawed to room temperature before analysis. To guarantee the accuracy and reliability of the experimental results, the spiked samples were tested simultaneously. The samples were pretreated according to the preparation section and then analyzed by LC-Q-TOF/MS. The results obtained showed that no pesticides were detected in the actual samples. The recovery results of the quality control samples met the analytical requirements, indicating that the values were accurate and reliable.

#### **4. Conclusions**

A high-throughput screening method based on modified QuEChERS and LC-Q-TOF/MS was established to analyze multi-residue pesticides in raw milk rapidly. The modified QuEChERS sample preparation method used an EN salting agent, followed by a freezing treatment, and then a purification treatment with C18 adsorbent, which effectively removed interference and reduced the matrix effect of multiple pesticide residues in raw milk. Overall, 195 pesticides passed the validation with satisfactory recoveries (70−120%) and an RSD of ≤20%. The method exhibited a good sensitivity to milk matrices, and the percentage of pesticides with SDL and LOQ values not exceeding 10 μg/kg for the established method were 93.3% and 87.2%, respectively. These results show that the method is cost-effective, convenient, and reliable for the routine screening of pesticide residues in raw milk and fully complies with the requirements of relevant regulations.

**Author Contributions:** Conceptualization, X.W.; Data curation, K.T.; Formal analysis, X.W., K.T. and Y.X.; Investigation, Y.X.; Methodology, H.C.; Project administration, C.F.; Resources, C.Y., S.H. and W.W.; Software, K.T., M.L. and W.W.; Supervision, C.F. and H.C.; Validation, C.Y. and M.L.; Writing—original draft, X.W.; Writing—review & editing, H.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was financially supported by the Science and Technology Project of the State Administration for Market Regulation (2021MK165).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

