Comparison of Sample Preparation and Determination of 60 Veterinary Drug Residues in Flatfish Using Liquid Chromatography-Tandem Mass Spectrometry

This study was performed to optimize the analytical method for multi-residues of 60 compounds in flatfish samples. Three sample preparation methods were tested to identify the optimal recovery conditions for target analytes. As a result, 10 mL of water/acetonitrile (1:4, v/v) was used to extract analytes from fish samples. For purification, C18 and 10 mL of acetonitrile saturated hexane were used to treat the samples. After evaporation and reconstitution, the fish samples were analyzed by ultra-performance liquid chromatography-tandem mass spectrometry. The proposed method was validated according to the CODEX guidelines (CAC/GL-71). Our results showed the recoveries of 73.2–115% and coefficients of variation of 1.6–22.1%. The limit of quantification was 0.0005–0.005 mg/kg in the fishery products. In analysis of real samples, no samples exceeded the limit of quantification. This analytical method can be used for multi-residue screening and confirmation of the residues of veterinary drugs in fishery products.


Introduction
The aquaculture industry continues to expand, providing fishery products for human consumption. Aquatic products are a main source of animal proteins for the global population [1]. The use of chemicals such as antibiotics, probiotics, and other feed additives is essential for improving the productivity and commerciality of aquaculture [2,3]. Veterinary drugs such as fluoroquinolone and penicillin are widely used to prevent diseases in aquaculture. Anthelminitic drugs, such as benzimidazole, are mainly used to treat parasitic infections. β-Lactam antibiotics such as penicillin and cephalosporin are used as growth promoters and to treat bacterial infections such as respiratory or skin infections [4][5][6][7]. However, the intensive use of antibiotics in aquaculture can affect microbial populations in the aquatic environment and further promote the spread of drug-resistant bacteria and resistance genes [8,9]. These changes are important to human health because environmental microorganisms are a source of various genes that have transformed into virulence factors for acquisition by many human pathogens. In addition, several drugs, such as penicillin and quinolone, are classified as critically important for human medicine [10]. Thus, excessive use of these medicines should be reduced in animal products.
Previous monitoring studies report that a variety of veterinary drugs have been frequently detected in aquaculture animals in Korea. Fluoroquinolones were detected in 7.5% of 268 freshwater and seawater fish samples in 2011 [11]. Sulfadiazine, erythromycin, and trimethoprim have been

Comparison of Sample Preparation Methods
Methods of preparing samples have been developed for analyzing a wide range of veterinary drug residues [13,20,[22][23][24][25]. In this study, we reviewed three methods of sample preparation ( Table 2) Recovery tests in the flatfish tissue were performed to compare the three methods according to CODEX guidelines (70-120%). In Method 1, the extraction step was performed using EDTA as a chelating agent to improve the extraction recovery and prevent rapid chelation with metal ions [26,27]. A combination of formic acid and ammonium formate was used to improve peptide separation in the samples [13]. In Method 2, acidified acetonitrile was used as an extraction agent to eliminate interference from the matrix. Acidified acetonitrile is widely used to extract veterinary drugs from animal tissues [28,29]. Turnipseed et al. (2016) suggested that formic acid (0.2-1%) in acetonitrile may affect the degradation of several β-lactams. Formic acid in water causes a rapid degradation of monobasic penicillins. Thus, acetic acid was used to increase the acidity of the acetonitrile extractant [25]. In Method 3, water/acetonitrile solution was used for extraction because many compounds with different chemical groups and different physicochemical properties were present in the mixtures [30,31]. In addition, acetonitrile is typically preferred for precipitating proteins in tissue [25].
The extracted solutions in each method were purified using n-hexane, primary secondary amine (PSA), and octadecylsilane (C 18 ). Animal tissues are rich in fats, lipids, and amino acids. Lipids can interfere with the analysis of some substances in animal tissues and contaminate the HPLC column [32]. Fats (specifically phospholipids) have been shown to cause significant matrix effects on ESI in the APCI MS analysis [33]. n-Hexane was added to eliminate some residual interference without the loss of target compounds. PSA and C 18 absorbents have been used to prevent the co-extraction of interference compounds. The amino groups on the PSA can form strong hydrogen bonds with carboxylic acids and other polar organic acids [34,35]. C 18 has been reported to allow the removal of oil and pigments [36].
Comparison of the three methods showed that Method 3 exhibited the highest recovery of 88% of the target compounds ( Figure 1). In our previous method (Method 1), the recovery was below 50% for 22 analytes in fish samples. In addition, 50 compounds were affected by substances that interfered with the matrix. The linearity of the seven compounds was below the CODEX guidelines (r 2 > 0.98). For Method 2, 19 compounds were not properly validated because of their poor peak shapes and high relative standard variations in fish samples. The validation results of Method 3 showed that only 8 compounds were incorrectly identified. Another sample preparation step may be required to simultaneously analyze 7 compounds (decoquinate, diminazene, novobiocin, phenylbutazone, robenidine, triclabendazole, and keto triclabendazole).
Based on the results obtained by comparison of the three methods, we selected and optimized Method 3. Briefly, acetonitrile/water solvents were used as the extract solution. A clean-up step was carried out by adding C 18 and n-hexane. A concentration step was performed to improve the signal intensity of the compounds. A combination of methanol and water was used for residue analysis [24,25]. We evaporated the extract by placing the samples in a 40 • C water bath and dissolved the final residue in methanol/water (1/1, v/v). The temperature was controlled to maintain the target compounds under stable conditions.

Method Validation
Sixty-five compounds were initially selected for analysis based on the regulation of fishery products in Korea. Except for seven analytes, 60 compounds (47 drugs and their metabolites) were validated in the fishery products. The proposed method was evaluated according to selectivity, specificity, and linearity proposed by the CODEX guidelines. Selectivity was measured using blank samples of the spiked target compounds in fishery products. The chromatograms of the spiked sample solution and the standard mixture solution of target analytes were compared. The chromatograms of target compounds are shown in Figure S1. Specificity was evaluated using non-spiked blank samples at the same retention time for each analyte. Good linearity was observed with the regression coefficients (r 2 ) of ≥ 0.98 for all compounds. Our results revealed good linearity of the targets within the target concentrations. Recovery was tested five times at three different concentrations (0.005 mg/kg, 0.01 mg/kg, and 0.02 mg/kg) in the fishery product samples. The recovery values were found to be between 73.2% and 115% and the coefficient variation (CV) ranged from 1.6% to 22.1%, meeting the guidelines. The recovery and CV at the target testing levels have been summarized in Table 3. The limit of detection (LOD) and the limit of quantification (LOQ) were calculated based on the signal to noise ratio (S/N) of the target compounds; S/N was ≥ 3 and ≥ 10 for LOD and LOQ, respectively. The matrix effects and LOQ values are presented in Table 4. In addition, inter-laboratory validation was conducted by three different institutions. As a result, the linearity of the calibration curve showed an r 2 ≥ 0.98. Recovery of the analytes ranged from 64.3% to 115% and repeatability ranged from 1.1% to 22.2%. The results of inter-lab validation demonstrate the validity of the proposed method.

Matrix Effect
The samples were analyzed by LC-MS/MS. The sample matrix has been reported to affect quantification of the target analytes because of ionization suppression or enhancement for the analyte/matrix combination [28]. The main sources of these effects were endogenous substances such as ionic species (salt) and various organic molecules (lipid, peptide, and metabolites with a chemical structure close to the target analyte structure) [37]. In this study, the matrix-matched calibration curves were applied to adjust the matrix effects. The matrix-matched and the solvent standard curve were compared to evaluate the matrix effects (ME) which were calculated as follows: A matrix effect enhances the ionization efficiency of the target compounds, whereas a negative effect indicates the suppression of ionization. As shown in Table 4, ionization suppression occurred owing to the matrix effects for most of the target drugs. Ion enhancement of the signal was observed for several compounds (azithromycin, cefapirin, desacetylcefapirin, cefoperazone, imidocarb, and tildipirosin) in the fish samples including eel and shrimp (data was not shown). Gbylik et al. (2013) obtained similar results, although different multi-residue analysis was performed [38]. Tildipirosin is a high-signal polar compound that has been shown to be affected more by the sample matrix effects than by non-polar molecules [39]. Dickson et al. (2014) showed that tildipirosin had slightly larger matrix effects than other macrolide compounds because it co-eluted with other polar compounds [40]. Therefore, in the case of tildipirosin, analytical methods using an internal standard may be useful for correcting analytical differences during sample analysis.
Although d-SPE treatment was used to reduce the matrix effect, a negative matrix effect was observed in the fish samples. The matrix effect can be major issue for the development of a multi-residue method using LC-MS/MS. Therefore, optimizing sample preparation and dilution procedures, and manipulating LC and MS conditions, are needed to reduce the matrix effect [41]. Isotope-labeled internal standards also can be alternative method to significantly reduce the matrix effect [42]. Further investigation is required for optimization of simultaneous determination of veterinary drug residues in fish matrices.

Application to Real Samples
The fishery product samples (n = 102) were collected and monitored to evaluate veterinary drug residues using the proposed method. Seven kinds of fishery products were collected from a domestic market in 2019. The fishery product samples were analyzed and checked to ensure the retention time, quantification ion, confirmation ion, and ion ratio were consistent with standards. The detected sample did not exceed the LOQ and Korean maximum residue limits. Tildipirosin was detected at 0.001 mg/kg in one sample of catfish. Further studies are needed to control the veterinary drug residues in many samples of fishery products using the proposed method.

Chemicals, Materials, and Solution
The chemical standards (Table S1)  PSA was purchased form Agilent technologies (Santa Clara, CA, USA). All glass apparatus used in the experiment was cleaned and dried with cleaning solution, methanol, and tertiary distilled water (Arium 61316 composing Arium 611VF). Sartorius (Gottingen, Germany) and centrifuge tube of Corning (NY, USA) were used. The filter was purchased and used with a size of 0.2 µm polytetrafluoroethylene (PTFE) material from Teknokroma (Barcellona, Spain). For the mixed extraction of the sample, we used a mixer (MMV-1000W, Eyela, Tokyo, Japan) and the centrifuge (HERAEUSE Megafuge 16R, ThermoFisher Scientific, Waltham, MA, USA). The standard stock solutions were prepared for each compound in methanol, 50% methanol in water (v/v) and dimethyl sulfoxide (100 mg/L). All working solution mixtures of 1 µg/mL were diluted in 50% methanol in water (v/v). Standard solutions and working solutions were stored at −20 • C.

LC-MS/MS Analysis
The instrumental analysis was performed using ultra-performance liquid chromatography (xevo TQ-S) with X-SELECT C 18 (2.1 mm × 150 mm × 3. The source and desolvation temperatures were set at 150 • C and 500 • C, respectively, and desolvation gas (nitrogen) flow rate was 600 L h -1 . The column and auto sampler temperature were maintained at 40 • C and 15 • C, respectively. The collision argon gas was set at a pressure of 4 × 10 −3 mbar. Triple quadrupole tandem mass analysis was performed in positive (ESI+) and negative (ESI−) mode. Data collection was performed in MRM mode using MassLynx software (Waters, UK).

Sample Preparation
Flatfish samples were purchased from a market in the Republic of Korea. They were homogenized and stored in the freezer at −20 • C until use in the experiment. Two grams of the homogenized sample was weighed and placed into a 50 mL centrifuge tube. Ten milliliters of acetonitrile/water (4/1, v/v) was added and mixed with the sample for 5 minutes. The tube with the sample was centrifuged at 4500× g and 4 • C for 10 min. The centrifuged extract was acquired and transferred to a new centrifuge tube with 500 mg C 18 powder. Then, 10 mL of acetonitrile saturated hexane was added to the tube and shaken for 1 minutes. After centrifuging for 5 minutes at 4500× g and 4 • C, 5 mL of the bottom solution below the hexane layer was transferred to a new centrifuge tube. Subsequently, the extraction solution (5 mL) were evaporated at 40 • C under N 2 gas. The residue was then dissolved with 1 mL of methanol/water (1/1, v/v), filtered through 0.2 µm PTFE membrane filter, and placed in a pp-vial.

Method Validation
The proposed method was validated according to the procedures described in the Codex Alimentarius Commission guidelines (CAC/GL-71) [17]. In terms of selectivity, linearity, accuracy, precision, LOD, and LOQ, the results for the method were obtained. Accuracy and precision were expressed as recovery and CV. The target concentration (TC) was set at 0.01 mg/kg, and validation was performed for concentration level 0.5 × TC, 1 × TC, and 2 × TC. The calibration curves were obtained by matrix-matched standard solution at six points (0.0025, 0.005, 0.01, 0.015, 0.02, and 0.03 mg/kg). The validation was conducted by spiking blank fish with a mixed working solution at three concentration levels. Recovery and CV were obtained through five repeated experiments. The LODs and LOQs were defined with signal-to-noise ratios (S/N ratio) ≥3 and ≥10, respectively.

Conclusions
The quantitative multi-residue determination of 60 veterinary drugs in fish tissue by LC-MS/MS as developed. Three methods of sample preparation were compared. The extraction solvent of a mixed solution of water and acetonitrile was selected to dissolve various substances. C 18 and hexane were used to remove interfering substance in flatfish. This multi-residue method showed clear sample cleanup efficiency and demonstrated satisfactory recovery, accuracy, and precision for 60 veterinary drugs in flatfish samples. The proposed method can be applied to the monitoring of real samples. This study may be used to determine the concentration of veterinary drug residues in fishery products.

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