Multiclass Method for the Determination of Anthelmintic and Antiprotozoal Drugs in Livestock Products by Ultra-High-Performance Liquid Chromatography–Tandem Mass Spectrometry

The following standards were purchased from Sigma-Aldrich (St. Louis, MO, USA): 5-hydroxy thiabendazole, albendazole sulfoxide, bithionol, carbendazim, fluazuron, keto triclabendazole, isometamidium, ternidazole, thiophanate, and toltrazuril sulfone. Arprinocide, benznidazole, diethylcarbamazine, and halofuginone were purchased from Toronto Research Chemicals (Toronto, ON, Canada). Emamectin b_1a (emamectin) and ornidazole were purchased from ChemService (West Chester, PA, USA) and StordSynthesis (Hebei, China), respectively. The rest of the 42 compounds (abamectin, albendazole, albendazole sulfone, etc.) were purchased from Dr. Ehrenstorfer (Augsburg, Germany). Methanol (MeOH) and acetonitrile (MeCN) were purchased from Merck (Darmstadt, Germany). Dimethyl sulfoxide (DMSO), formic acid, MgSO₄, and NaCl were purchased from Sigma-Aldrich, and PSA was purchased from Agilent Technologies (Santa Clara, CA, USA). Ammonium formate was purchased from Alfa Aesar (Ward Hill, MA, USA) and C₁₈ (55–105 μm, 125 Å) was purchased from Waters (Milford, MA, USA). A syringe filter from Teknokroma (Barcelona, Spain) was used by incorporating it into polytetrafluoroethylene (PTFE) membrane filters (0.2 μm). For albendazole, albendazole sulfone, albendazole sulfoxide, buquinolate, flubendazole, oxfendazole, oxfendazole sulfone, oxibendazole, mebendazole, mebendazole amine, standard solutions (1,000 μg/mL) were prepared in MeOH/DMSO (1:1, v/v). Similarly, MeCN was used as the solvent to prepare a standard stock solution of guaifenesin (1,000 μg/mL), while DMSO was used to prepare the stock solutions for fenbendazole, 5-hydroxy mebendazole, methylbenzoquate, and nicarbazin (1,000 μg/mL). The corresponding standard stock solutions (1,000 μg/mL) were prepared at MeOH in all other compounds. All standard stock solutions were stored in amber bottles at −20°C.

Beef (n=10), pork (n=10), and chicken (n=10) were purchased from local markets in Korea. Each sample was homogenized and stored in a freezer (−20°C) until required for further use. Thus, each homogenized sample (2 g) was weighed into a 50 mL centrifuge tube and then extracted using 0.1% formic acid in MeCN/MeOH (95:5, v/v, 10 mL) and water (10 mL) under shaking for 5 min. Subsequently, MgSO₄ (4 g) and NaCl (1 g, original QuEChERS salt) were added to the sample. After, shaken for 5 min, and subjected to centrifugation at 4,700×g (4°C, 10 min). The supernatant was then transferred to a 50 mL centrifuge tube containing C₁₈ (150 mg), PSA (150 mg), and MgSO₄ (900 mg). And then, the obtained mixture was shaken for 5 min and centrifuged at 4,700×g (4°C, 5 min). The obtained supernatant (5 mL) was transferred to a new centrifuge tube, DMSO (20 μL) was added, and the solvent was evaporated under a stream of N₂ at 40°C. Afterwards, the residue was dissolved in a mixture of MeOH and water (1:1, v/v, 1 mL), and the extract was subsequently filtered through a 0.2 μm PTFE filter before analysis (Kim et al., 2021).

Separation was conducted on a Shimadzu UHPLCMS 8060 triple quadrupole mass spectrometer (MS, Shimadzu, Kyoto, Japan) equipped with a Waters X-SELECT HSS C₁₈ column (2.1 mm×150 mm, 3.5 μm particle size, Waters, Dublin, Ireland). Data processing used LC solution software version (5.99) from Shimadzu. Gradient separation was performed using a binary gradient composed of water containing 0.1% formic acid and 2 mM ammonium formate (mobile phase A) and MeCN containing 0.1% formic acid (mobile phase B). The gradient profile was as follows: 0 min, 15% B; 2 min, 15% B; 12.5 min, 95% B; 17.0 min, 95% B; 17.1 min, 15% B; 20.0 min, 15% B. The injection volume was 5 μL, and a flow rate of 0.3 mL/min was used under argon gas. The MS source settings were as follows: capillary voltages=4.0 kV (positive) and 2.8 kV (negative); capillary temperature=350°C, auto-sampler temperature=15°C, column temperature=40°C, and cone voltage=30 kV. The MS instrument was operated in the electrospray ionization (ESI) mode with positive and negative switching modes, and scheduled multiple reaction monitoring (MRM) was employed for all target compounds.

The linearity, accuracy, precision, limit of detection (LOD), and limit of quantification (LOQ) of the developed method were determined according to the CODEX guidelines (CAC/GL71-2009) and the Ministry of Food and Drug Safety (MFDS) of Korea guidelines (FAO and WHO, 2009; MFDS, 2016). More specifically, the accuracy and precision were determined by analyzing negative samples at three different concentrations, i.e., spiking at 0.5, 1, and 2× the MRL. In addition, the analysis included the determination of toltrazuril (toltrazuril sulfone), emamectin (emamectin B1a), and nicarbazin [N,N’-bis(4-nitrophenylurea)], based on the specified marker residues. The matrix-matched standards for the calibration curves were prepared using a six-point range of target concentrations (i.e., 0.25, 0.5, 1, 2, 4, and 8× the MRL). The LODs and LOQs were defined as the concentrations at which the signal-to-noise ratios were ≥3 and ≥10, respectively.

To determine the degree of the matrix effect for each system, the matrix-matched curve of a post-extraction spiked sample and the solvent standard curve were compared at the same concentration (in the case set MRL value; 0.25, 0.5, 1, 2, 4, and 8× the MRL and in the case not set MRL value; 2.5, 5, 10, 20, and 40 μg/kg), as outlined in Eq. (1) below. In general, the matrix components of a sample can either increase or decrease ionization efficiency due to interfering substances, i.e. salts, lipids, and peptides (Antignac et al., 2005).

The MS parameters were determined using individual standard solutions and were optimized based on the mass spectra of all compounds. Using the ESI mode with positive and negative switching and MRM, bithionol, chlorfluazuron, oxyclozanide, keto triclabendazole, nicarbazine, niclosamide, and toltrazuril sulfone were detected as their corresponding [M−H]⁻ species, semduramyicin was detected as [M+NH]⁺, and all other compounds were detected as [M+H]⁺ mode. Using standard solutions diluted in MeOH/water (1:1, v/v), the MS parameters were optimized using a cone voltage of 30 V. The parent and daughter ions were selected by optimizing the collision energy. Furthermore, daughter ions with higher intensities and better peak shapes were selected as quantitative ions. Most compounds possessed one parent ion and either one or two daughter ions. The optimized precursor ions, daughter ion collision energies, and retention times of all compounds are listed in Table 1. A reversed-phase X-SELECT HSS C₁₈ column was used to separate the various veterinary drugs examined herein. This column was selected because the multi-residue analysis was previously performed using a C₁₈ column (Dasenaki and Thomaidis, 2015). It was found that mobile phase A improved sensitivity and reduced peak tailing (Chang et al., 2019; Clarke et al., 2013; Frenich et al., 2014), while mobile phase B produced a better peak shape than a mixture of MeOH and 0.1% formic acid in MeCN (Zrnčić et al., 2014). Supplementary Fig. S1 shows the extracted ion chromatograms of the target compounds. These chromatograms were observed by injecting an aliquot (5 μL) of the desired standard solution into the beef sample at a concentration of 100 μg/kg (fenbantel and isometamidium) or 10 μg/kg, which corresponds to spiking of 1× the MRL.

This study was conducted according to the modified QuEChERS procedure. The sample extraction and clean-up conditions were optimized based on a previously used method for multiclass drug analysis (Kim et al., 2021). Kim et al. used a modified QuEChERS type extraction method and added the concentration step. The original QuEChERS method consists of two steps an extraction/partitioning step with the addition of salts, and a clean-up step that uses d-SPE. More specifically, sample extraction was carried out using 0.1% formic acid with MeCN/MeOH (95:5, v/v), which had been previously demonstrated to yield a high recovery rate (Clarke et al., 2013; Lopes et al., 2012). In addition, MgSO₄ was employed due to its good water absorbency properties, which permits salting-out, while NaCl was added to increase the polarity of the extraction solvent and enhance the extraction selectivity (Rejczak and Tuzimski, 2015). For sample clean-up, PSA was used to remove fatty acids and organic acids, MgSO₄ was used to remove water, and a C₁₈ absorbent was used to remove non-polar components (Anastassiades et al., 2003; Wilkowska and Biziuk, 2011). DMSO was added prior to sample concentration to enhance the sample recovery to act as a keeper during evaporation (Kim et al., 2021; Whelan et al., 2010).

Method validation was performed in terms of linearity, accuracy, precision, LOD, and LOQ. All compounds exhibited a best linearity, with correlation coefficients (r²) exceeding 0.98 at matrix-matched calibration at six points. The accuracy, expressed as the recovery, ranged from 60.2% to 119.9%, and the coefficients of variation (CV) ranged from 1.2% to 31.5% for the three determined levels. In the chicken samples, the average recovery of 2-amino albendazole sulfone was ~111.1%, which was considered unacceptable based on the recovery limit of 110% specified by the CODEX guidelines at a concentration of 200 μg/kg; all other compounds satisfied the CODEX guidelines. In addition, inter-laboratory (n=2) validation was conducted according to CODEX guidelines (CAC/GL-71) and the results were satisfied with the guideline. Table 2 lists the accuracies and precisions obtained of all compounds following their analyses in the three matrices. In addition, the LOD values ranged from 0.3 to 3 μg/kg, while the LOQ values ranged from 1 to 10 μg/kg, which are lower than the corresponding values of the Korean MRLs. It should be noted that, in general, the LOQ values were ~1 μg/kg; however, the corresponding values for imidocarb and pyrantel in beef were 10 μg/kg, respectively, while oxfendazole had a LOQ value of 10 μg/kg. The LOD, LOQ, and Korean MRL values for the three matrices (beef, pork, and chicken) are given in Table 3, wherein it can be deduced that the obtained values were satisfactory. Thus, the developed method appeared to demonstrate an acceptable analytical performance for residue control in livestock products.

The matrix effects observed for the various samples and compounds are presented in Fig. 1 and Table 4. Fig. 1 shown that the positive and negative matrix effects were observed for the livestock products examined herein. These effects were classified into five groups, namely high signal suppression (ME<−50%), moderate suppression (ME<−10% to −50%), no matrix effect (ME>−10% to <10%), moderate signal enhancement (ME>10% to <50%), and high signal enhancement (ME>50%; Chatterjee et al., 2016). It was found that the matrix effects varied in the range of −95%–56%, wherein high matrix effects were observed for 11 compounds (20.0%) in the beef matrix, nine compounds (16.4%) in the pork matrix, and seven compounds (12.7%) in the chicken matrix. No matrix effect was observed for 9 compounds (16.4%) in beef samples, 17 compounds (30.9%) in the pork samples, and 15 compounds (27.3%) in the chicken samples. In addition, moderate matrix effects were observed for 35 compounds (63.6%) in the beef matrix, 29 compounds (52.7%) in the pork matrix, and 33 compounds (60.0%) in the chicken matrix. Most compounds (beef, 47; pork, 40; chicken, 47) exhibited signal suppression, while a few (beef, 6; pork, 14; chicken, 7) exhibited signal enhancement. Pyrantel is neither in both suppression and signal enhancement in beef. The beef and chicken matrices were largely responsible for signal suppression, while the pork matrix led to both signal suppression and enhancement. The greatest suppression and enhancement were observed for the pork matrix, and these corresponded to −95% and 56% for isometamidium and halofuginone, respectively. These variable matrix effects are likely due to the complexity of the tissue matrix. Although the most effective means to compensate for matrix effects is to use an internal standard (Yin et al., 2016), internal standards are expensive, and the corresponding compounds for the various target compounds are often unavailable. Thus, the current study was performed using a matrix-matched standard curve.

Fig. 1. Sample matrix effects of the target compounds in the beef, pork, and chicken matrices.

Download Original Figure

To demonstrate the applicability of our method, the livestock samples (n=30) collected from Korean local markets were analyzed. Among these samples, toltrazuril sulfone was detected in pork and chicken samples at concentrations of 1 μg/kg in pork (2 samples) and 5 μg/kg in chicken (1 sample); however, it should be noted that their concentrations were lower than the Korean MRL (Table 3). Toltrazuril is a triazine-based antiprotozoal that is commonly used in pigs and chicken turkeys (Mehlhorn et al., 1988). Although toltrazuril sulfone is reportedly more effective in smaller amounts than toltrazuril, it is highly toxic and can cause side effects if consumed by humans through the food chain (Franklin et al., 2003; Lindsay et al., 2000).

In a previous study, toltrazuril and toltrazuril sulfone were detected in frankfurter sausages at a concentration of 2 μg/kg (Martínez-Villalba et al., 2010). Indeed, the detection of anthelmintic and antiprotozoal drugs in livestock samples has been widely reported (Adesiyun et al., 2021; Pawar et al., 2021; Yoo et al., 2021). Ai et al. (2011) detected diclazuril in rabbit muscles (n=10), while monensin (1.4–22 ng/g, n=42) and ractopamine (0.6–64 ng/g, n=15) were detected in bovine liver, and monensin (0.8 and 1.1 ng/g, n=2) and ractopamine (1.8–6.3 ng/g, n=12) were detected in bovine muscle. Ractopamine (0.5–67 ng/g, n=7) was detected in bovine kidney, while monensin (2.0 ng/g, n=1), decoquinate (150 ng/g, n=1), lasalocid (1.5 and 14 ng/g, n=2), narasin (4 ng/g, n=1), and N,N′-bis(4-nitrophenylurea; 190 ng/g, n=1) were detected in chicken muscle (Matus and Boison, 2016). Furthermore, according to Kang et al. (2015), acetyl salicylic acid (12–576 μg/kg; n=28, 50–53 μg/kg; n=1) was detected in pigs and chickens, paracetamol (28–381 μg/kg, n=15) was detected in pigs, clopidol (9–4,614 μg/kg) was detected in chickens (n=28) and ducks (n=6), while diclazuril and amprolium were detected in chicken livers (104–525 μg/kg, n=8; and 195–196 μg/kg, n=2, respectively). Moreover, toltrazuril and its metabolites (toltrazuril sulphone and toltrazuril sulfoxide) were detected in chicken liver (n=29) at concentrations of 161–469, 67–1,822, and 209–760 μg/kg, respectively, while phenylbutazone and its metabolite (oxyphenylbutazone) were detected at levels of 247 and 15 μg/kg in cattle liver (n=1), respectively, and nicarbazin was detected at a concentration of 0.05 μg/kg in eggs (n=1; Kang et al., 2015).

Overall, this study shows that the detection amount is smaller than in previous studies (Ai et al., 2011; Matus and Boison, 2016). Therefore, monitoring results shows that livestock products are a safe level of residues. Therefore, the UHPLC-MS/MS method established in this study can be used as a reliable method for the detection of anthelmintic and antiprotozoal drug residues.