The Development and Validation of an LC–Orbitrap–HRMS Method for the Analysis of Four Tetracyclines in Milk and Its Application to Determine Oxytetracycline Concentrations after Intramuscular Administration in Healthy Sarda Ewes and Those Naturally Infected with Streptococcus uberis

Abstract

An LC–Orbitrap–HRMS method was developed and validated for the simultaneous determination of four tetracyclines—oxytetracycline (OTC), tetracycline (TC), doxycycline (DC), and chlortetracycline (CTC)—in milk. This method involves sample extraction with McIlvaine–EDTA buffer solution (pH 4) and solid-phase extraction (SPE) with Oasis HLB cartridges, followed by the evaporation of the extract and its reconstitution with a 14% methanol aqueous solution before injection into the instrumental system. This method has been validated in terms of linearity, sensitivity, selectivity, precision, and accuracy, in accordance with Commission Decision 2002/657/EC requirements. Compared to existing methods, this approach optimally combines a quantitative procedure for extracting analytes from the milk of different species, including sheep, bovines, and goats, with a very short LC–Orbitrap–HRMS instrumental analysis time (only 8 min), simultaneously ensuring high precision, sensitivity, and applicability as a rapid confirmation method in official food control laboratories. The proposed method was applied to determine the concentration levels of OTC in milk samples derived from healthy Sardinian sheep and those naturally infected with Streptococcus uberis, after the intramuscular administration of an antibiotic, in order to evaluate how much of the drug was “subtracted” during penetration from blood into milk, with a potential effect of reducing its therapeutic efficacy.

1. Introduction

In veterinary and human medicine, antibiotics are very important tools for the treatment of many diseases caused by bacterial pathogens. These substances were widely used in the past in veterinary medicine due not only to their therapeutic effect in the treatment of infections but also for prophylactic and metaphylactic purposes, as well as their role as growth promoters [1]. Among these antibiotics are tetracyclines, a group of compounds characterized by a broad spectrum of activity and an excellent cost/effectiveness ratio compared to that of other drugs, and among the over 20 variants included in their chemical family, four molecules (Figure 1), namely, tetracycline (TC), chlortetracycline (CTC), oxytetracycline (OTC), and doxycycline (DC), are the most administered in veterinary medicine, being suitable for a wide range of animal species such as cattle, pigs, horses, dogs, cats, poultry, rabbits, and fish [2]. Their use is also prevalent in the breeding of small ruminants and in dairy farming [3,4,5,6].

The use of these antimicrobials in the food sector, with the expansion and intensification of global animal farming for food production, is monitored according to regulatory standards aimed at promoting their rational and prudent use in order to minimize the development of antimicrobial resistance and preserve their effectiveness and availability over time, reducing the impacts on human health [7,8,9]. Growing concern about potential health risks associated with tetracycline residues in meat, milk, eggs, and other food products of animal origin has led to the definition of permissible residue concentrations in foods, which vary significantly between regions such as the United States and the European Union. Indeed, both the United States Food and Drug Administration (FDA) and the European Commission have established acceptable daily intake values based on toxicological data and the current performance of analytical technology [7] and produced official documents for the regulation of veterinary drugs in foods to protect consumers. For example, the European Commission Regulation 2009/470/EC [10] establishes rules and procedures for determining the Maximum Residue Limits (MRLs), in terms of concentrations, of a pharmacological substance that may be permitted in foods of animal origin.

Commission Regulation 2010/37 [11] is the official regulatory document that reports the explicit MRLs set for pharmacologically active substances, and therefore tetracyclines, in foods of animal origin. Furthermore, guidelines regarding the performance of analytical methods, their validation, and the interpretation of analytical results in official food control laboratories are described in Commission Decision 2002/657/EC [12]. Given the importance of milk and dairy products in human nutrition, the systematic monitoring and evaluation of antibiotic residues in milk are crucial to guarantee the appropriate use of antimicrobials, helping to reduce the consumption of antibiotics and consequently their residues in food and deter bacterial resistance emergence. Therefore, the development of specific analytical methods is necessary to achieve more accurate quantitative data collection. In recent years, significant progress has been made in the detection and quantification of tetracyclines, degradation products, and metabolites in milk [13,14,15,16]. Conventional screening tests used for the detection of antimicrobials can be classified into instrumental, microbial [17,18], and immunological methods [19,20]. Analytical instrumental methods are often based on high-performance liquid chromatography (HPLC) [21,22] coupled with different types of detection such as spectrophotometry, fluorometry, and mass spectrometry [23]. The results deriving from the use of the LC-MS technique are useful both for qualitative screening (for the identification of substances) and confirmation purposes (for the determination of analyte concentrations) [24,25,26,27,28,29], and this technique is usually employed when high selectivity and sensitivity are required.

In addition to the analytical techniques, the sample pretreatment step is of great importance because it is essential for the development of any successful quantitative analysis method.

In recent decades, many types of materials have been studied and used for the implementation of sample preparation procedures, and, in the most recent scientific literature, there have been works describing the advantages offered by a new class of magnetic ionic liquids (MILs) [30], which are comparable to conventional ones, in sample preparation procedures. MILs exploit the advantageous properties of ionic liquids and magnetic properties, creating an extremely effective combination in a very promising research field, with increasing applications of MILs in sample preparation steps, particularly in microextractions such as liquid–liquid microdispersion extraction (DLLME), solid-phase matrix dispersion, and two-phase aqueous extraction [31,32]. These advanced pretreatment techniques involve simple operating procedures requiring minimal quantities of environmentally polluting organic solvents compared to traditional methods, but, along with their benefits, there are also disadvantages due to the still rather high cost of ionic liquids and the complexity of their synthesis, which are not always within the reach of public laboratories. Solid-phase extraction, in our opinion, is the optimal choice due to its balance between efficiency, selectivity, and environmental impact. The main reasons why SPE was chosen in this study are as follows:

-

It allows the effective removal of sample impurities, a critical step for the accurate and sensitive detection of tetracyclines;

-

It entails a reduced use of organic solvents compared to LLE, which translates into greater respect for the environment and a reduction in costs;

-

The high efficiency and selectivity of SPE cartridges such as Oasis HLB provide high recovery rates and reliable results;

-

It has high reproducibility, which makes it suitable for routine analyses in laboratories, as demonstrated by applying the method to the analysis of over 400 milk samples as part of a research project aimed at studying the pharmacokinetics/dynamics (PK/PD) of OTC in milk from healthy and infected half udders of dairy ewes after antibiotic administration.

The aim of this study was to develop an easily applicable analytical method using ultra-high performance liquid chromatography coupled with Orbitrap high-resolution mass spectrometry (UHPLC–Orbitrap–HRMS) technology for the determination of four tetracyclines (i.e., TC, DC, CTC, and OTC) in bovine, ovine and caprine milk. This method was validated according to CD 2002/657/EC [12], with the primary objective of applying it to the control and monitoring of tetracycline drug residues in milk within the National Residue Control Plan (PNR) of the Italian Ministry of Health.

2. Materials and Methods

2.1. Reagents and Materials

Oxytetracycline (OTC, purity ≥ 95.7%), tetracycline (TC, purity ≥ 96.8%), doxycycline (DC, purity ≥ 97.0%), and chlortetracycline (CTC, purity ≥ 93.3%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Methanol (purity ≥ 99.8%), HiPerSolv CHROMANORM (VWR International Srl, Milano, Italy) acetonitrile, and LC-MS-grade formic acid were obtained from Fisher Scientific and ultrapure water was produced using the Milli-Q Water Advantage System (Millipore, Billerica, MA, USA).

Disodium hydrogen phosphate, citric acid, EDTA disodium salt, phosporic acid, and sodium hydroxide were purchased from Carlo Erba Reagents Srl. (Cornaredo (MI), Italy).

HLB Oasis SPE cartridges (3 mL/60 mg/80 Å) were purchased from Waters (Milford, MA, USA).

An Acrodisc^® GHP 0.2 µm Syringe Filter (VWR International Srl, Milano, Italy) was employed.

An ACQUITY UPLC BEH C18 chromatographic column and an ACQUITY UPLC BEH C18 VanGuard Pre-column (Waters, Millford, MA, USA) were also used.

2.2. Standard Working Solutions

McIlvaine Buffer: We dissolved 28.41 g of anhydrous dibasic sodium phosphate in distilled water in a 1 L volumetric flask, diluted it to volume, and mixed it. We dissolved 21.01 g of citric acid monohydrate in distilled water in a 1 L volumetric flask, diluted it to volume, and mixed it. We combined 1 L of citric acid solution with 625 mL of phosphate solution in a 2 L flask. We then checked the pH, which, in this case, should be 4.00 ± 0.05.

McIlvaine/0.1 M EDTA Solution (pH 4): We added 60.49 g of disodium EDTA dihydrate to 1.625 L of McIlvaine buffer.

In this study, 1000 mg L⁻¹ single standard stock solutions of TC, DC, CTC, and OTC were prepared in methanol and stored in dark glass bottles at −18 °C.

The working mix of tetracycline standard solution at 10 µg mL⁻¹ was prepared by diluting 100 µL of each stock solution to 10 mL in a volumetric flask with methanol.

2.3. Sample Preparation

A total of 2.0 g of milk was weighed into 30 mL fluorinated ethylene propylene (FEP) tubes, 10 mL of McIlvaine–EDTA buffer solution (pH 4) was added, and the mixture was shaken for 10 min. After it was centrifuged at 18,457 RCF (with 12,000 rpm being selected for a rotor with a radius = 11.5 cm) for 10 min at 0 °C, a portion of at least 5 mL of extract (1 mL/min flow rate) was loaded onto an Oasis HLB SPE cartridge (3 mL, 60 mg) previously conditioned with 3 mL of methanol and 2 mL of ultrapure water. After it was washed with 1.5 mL of 5:95 methanol/water solution, the sample was eluted with 2 mL of methanol. The purified extract was evaporated under a gentle nitrogen stream and reconstituted with 2 mL of 14% (v/v) methanol in water, filtered, and analyzed via UPLC-HRMS. A representative diagram of this procedure is shown in Figure 2.

2.4. Chromatographic and Spectrometric Conditions

Instrumental analyses were performed using the Dionex Ultimate 3000 cromatographic system coupled with a Q–Exactive Orbitrap High Resolution Mass Spectrometer equipped with a heated electrospray ionization (HESI-II) source (Thermo Fisher Scientific, Waltham, MA, USA). Chromatographic runs were conducted on an ACQUITY UPLC BEH C18 column (130 Å, 1.7 µm, 2.1 mm × 100 mm) preceded by ACQUITY UPLC BEH C18 VanGuard Pre-column (130 Å, 1.7 µm, 2.1 mm × 5 mm (Waters, Millford, MA, USA)) operating at 45 °C. Mobile phase A was H₂O with 0.1% formic acid and mobile phase B was CH₃CN. Elution started with 20% of B until 5 min had elapsed, followed by a linear gradient toward 100% of B up to 6.5 min and a rapid return to initial conditions at 6.6 min, with equilibration time spanning until 8 min had passed, at a flow rate of 0.400 mL/min and with a 5 µL injection volume. According to this gradient, the entire separation of all analytes was completed within 8 min (Figure 3).

Analyses were performed under heated positive electrospray ionization conditions. The acquisition process involves a full MS¹ scan, followed by a data--dependent scan (dd–-MS²) based on predefined m/z criteria for the precursor ions indicated in a specific inclusion list.

The optimized HESI II conditions were as follows: sheath gas, 40 arbitrary units; auxiliary gas 10 (positive ionization, +), with a spray voltage of 4.0 kV (+); capillary temperature, 300 °C; heater temperature, 330 °C; and S–-lens RF level, 50. In full MS/dd–-MS² mode, the Q-Exactive HRMS instrument performed a full scan followed by a dd-MS² scan. The full-scan mass range was set to 150–500 m/z with a resolution of 70,000 (FWHM at 200 m/z). The optimized automatic gain control (AGC) target was assigned a value of 3.0 × 10⁶ (the maximum number of ions filling the C-Trap) with a maximum ion injection time (IT) of 300 ms.

For the ddMS² scans, the Orbitrap resolution was set to 17,500 FWHM, the AGC target was set to 2.0 × 10⁵, and IT was set to 75 ms for the ddMS² scan period. The precursor quadrupole isolation window was set to 2.0 m/z. The default charge state was set to 1 and loop count was set to 12. The data–dependent settings’ apex trigger was set in the range of 1 to 2 s charge exclusion; peptide match, excluded isotopes, and dynamic exclusion were disabled.

Fragmentation of precursors was optimized as normalized collision energy (NCE) via the direct injection of all the tetracycline standard solutions at 10 ng/mL into the mass spectrometer. Identification and detection of analytes were based on matching the exact masses of the protonated/deprotonated molecular ions and at least two corresponding fragments at the retention times of target compounds (all these data are reported in

Table 1. List of analytes, chemical formulae, retention times, and exact masses of precursors and fragment ions, with normal collision energy applied.

Analyte	Chemical Formula	RT	Exact Mass	Ion Transitions	NCE
		(min)	[MH] + (m/z)		(eV)
Oxytetracycline (OTC)	C22H24N2O9	4.94	461.1555	426.1183	30
				381.0605
				337.0707
Doxycycline (DC)	C22H24N2O8	6.56	445.1605	428.1340	30
				410.1234
				154.0498
Tetracycline (TC)	C22H24N2O8	5.20	445.1605	410.1234	30
				337.0703
				154.0498
Chlortetracycline (CTC)	C22H23N2O8Cl	6.23	479.1216	444.0485	26
				371.0317
				154.0498

below). The mass spectrometry device was calibrated regularly with the standard solution LTQ ESI Positive Ion supplied by Thermo Fisher.

2.5. Experimental Design of Method Validation

In order to guarantee adequate method performance according to CD 2002/657/EC [12], a conventional validation approach was implemented. This validation study was designed to reduce the overall number of required samples through a combination of certain validation experiments, including 20 different batches summarizing and taking into account three fortification levels, 0.5, 1.0, and 1.5 MRL, for each analyte (50, 100, and 150 µg kg⁻¹). The performance characteristics that needed to be evaluated were linearity, selectivity, precision, trueness, decision limit (CCα), detection capability (CCβ), and matrix effect. Due to the unavailability of certified reference materials for tetracyclines in milk, complete validation sets were performed using raw milk samples from different animal species (bovines, sheep, and goats).

An overview is provided in

Table 2. Overview of the batches of milk samples required for the validation study.

Validation series 1	6 batches for selectivity and fortification: 24 milk samples
	(quantification with standard calibration)
Validation series 2	6 batches for selectivity and fortification: 24 milk samples
	(quantification with standard calibration)
Validation series 3	6 batches for selectivity and fortification: 24 milk samples
	(quantification with standard calibration)
Validation series 4	20 batches, 20 milk samples
	(for absolute recovery and relative matrix effect)

.

Validation series 1, 2, and 3 were accomplished on three different days, applying the same extraction and instrumental methodology for all samples but varying the batches of solvents, vials, pipettes, and laboratory conditions.

Table 3. Experiments required for each validation series and relative measurement.

Validation	Fortification Level	N of Experiments	Performance Characteristic
Standard solution points	Five different levels	5	Linearity range
1 aliquot of 6 different blank batches	No fortification	6	selectivity
1 aliquot of 6 different blank batches	Level 1	6	CCα, CCβ, trueness, intraday repeatability interday repeatability
fortified prior to extraction	0.5 MRL
1 aliquot of 6 different blank batches	Level 2	6
fortified prior to extraction	1.0 MRL
1 aliquot of 6 different blank batches	Level 2	6
fortified prior to extraction	1.5 MRL

provides a description of the series.

2.6. Method Application

The application of the validated method was planned for the analysis of ovine milk samples derived from 10 Sardinian sheep that were healthy and 10 that had been naturally infected with Streptococcus uberis mastitis. Both groups of sheep were treated with 5 mL of an oxytetracycline dihydrate (OTC) formulation (217.4 mg/mL, equivalent to 200 mg/mL of OTC corresponding to 20 mg/kg of body weight), inoculating them via intramuscular injection in the semitendinosus/semimembranosus muscle according to the manufacturer’s instructions (Pfizer Inc., New York, NY, USA) at Time 0.

Animal testing was approved by the Italian Ministry of Health (Authorization n. 916/2017-PR 22-11-2017).

Milk samples were collected every 12 h for the first 7 days, and then once daily until the 20th day after administration, and stored in the laboratory at −20 °C until UPLC-HRMS analysis was conducted. In total, 250 milk samples from healthy sheep and 197 milk samples from infected sheep were analyzed.

The normality distributions of the OTC concentrations were assessed using the Shapiro–Wilk normality test. Median and interquartile range were used to describe quantitative variables. A nonparametric statistical analysis (i.e., the Wilcoxon matched-pair signed-rank test) was performed to investigate intergroup differences in milk antibiotic concentrations between healthy and infected sheep during follow-ups.

3. Results

3.1. Linearity

The linear dynamic range for each compound was investigated using five-point solvent calibration solutions (1.0, 5.0, 10.0, 50.0, and 200.0 ng mL⁻¹) in three replicates. In data processing, weighted linear regression (1/x) and the exclusion of the origin of the axis were applied. All four tetracyclines showed the same linear range, with coefficients of determination (R²) > 0.999.

3.2. Selectivity

To evaluate the selectivity of the method, 20 blank sheep milk samples were analyzed for the presence of any interferences (signals, peaks, or traces of ions) around the retention time of each analyte (Rt ± 2.5% min). For confirmation purposes, the unambiguous single analyte-specific fragmentation pattern (see Table 1) was used to distinguish target substances from matrix interferences.

3.3. Precision

The precision of the method was assessed in terms of interday and intraday laboratory repeatability. The analysis was performed on 18 blank sheep milk samples (the same used for the selectivity study) spiked with all the analytes at each of the three specified levels (50, 100, and 150 µg kg⁻¹) for three days. The interday repeatability results were also representative of intralaboratory reproducibility because they were obtained by following the same protocol but with two different operators performing the analysis using different sets of reagents and laboratory equipment. The relative standard deviation (RSD%) was determined in both cases and assessed considering the criteria established by CD 2002/657/EC [12]: at the MRL level (mass fraction = 100 µg kg⁻¹), the RSD values for all the analytes ranged from 4.6 to 15, well below the coefficient of variation CV = 23 obtained from the Horwitz equation.

3.4. Trueness

Due to the unavailability of certified reference materials containing information on OTC, DC, CTC, and TC at the same time, the trueness of the method was assessed by performing recovery tests. Sample recovery was determined with blank milk samples spiked with analytes at 50, 100, and 150 µg kg⁻¹ (six replicates) and their recoveries were calculated using the following equation:

$$Recovery\left(\%\right)={\displaystyle \frac{Cmeasuredcontent}{Cfortificationlevel}}\times 100$$

(1)

The recovery values used as references, recommended by the CD 2002/657/CE [12], varied within the range of 70% to 120% according to the mass fractions of analytes.

3.5. Limit of Detection (LOD) and Limit of Quantification (LOQ)

The limit of detection (LOD) and limit of quantification (LOQ) were calculated as described by Miller and Miller [33]:

$$LOD=3.3\times {\displaystyle \frac{SDa}{b}}$$

(2)

and

$$LOQ=10\times {\displaystyle \frac{SDa}{b}}$$

(3)

Here, SDa is the standard deviation of the intercept a and b is the slope of the calibration line. To verify the LOD and LOQ values, the milk samples were spiked with tetracyclines at the same levels. The results obtained for all the validation parameters are summarized in

Table 4. LOD, LOQ, recovery, and intra- and interday repeatability for TCs in milk.

Analyte	LOD	LOQ	Concentration Level (µg kg−1)	Recovery (%)	Repeatability
	(µg kg−1)	(µg kg−1)		(n = 18)	Intraday	Interday
					(RSD% n = 6)	(RSD % n = 18)
OTC	3	10	50	100.8	9.61	8.57
			100	102.5	6.95	6.01
			150	104.4	4.32	4.60
DC	3	11	50	77.9	7.83	8.58
			100	77.4	4.35	6.43
			150	77.8	5.27	5.27
TC	2	7	50	102.6	12.95	12.91
			100	105.0	5.27	6.44
			150	110.5	7.46	7.55
CTC	2.5	8	50	97.6	5.60	15.00
			100	101.0	6.82	15.06
			150	101.8	5.48	12.67

.

3.6. Detection Limit (CCα) and Detection Capability (CCβ)

The detection limit, CCα, means “the limit beyond which one can conclude with an error a probability (5% for authorized substances) that a sample is not compliant” while detection capability, CCβ, means “the smallest content of the substance that can be detected, identified and/or quantified in a sample with a probability of error equal to β (5% per authorized substances)” (CD 2002/657 EC Annex 1, 1.11 and 1.12).

CCα was determined at the MRL validation level for OTC, TC, CTC and for DC, without regulatory limit, at the LOD level, as follows (α = 5%):

$$\mathrm{C}\mathrm{C}\mathsf{\alpha }=n\mathrm{C}\mathrm{x}+1.64\times \mathrm{S}\mathrm{D}\mathrm{i}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}-\mathrm{l}\mathrm{a}\mathrm{b}\mathrm{r}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{i}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{a}\mathrm{t}\mathrm{C}\mathrm{x}$$

Here, Cx = MRL for OTC, TC and CTC and Cx = LOD for DC.

CCβ was determined at the CCα level for OTC, TC, CTC, and DC as follows (β = 5%):

$$\mathrm{C}\mathrm{C}\mathsf{\beta }=\mathrm{C}\mathrm{C}\mathsf{\alpha }+1.64\times \mathrm{S}\mathrm{D}\mathrm{i}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}-\mathrm{l}\mathrm{a}\mathrm{b}\mathrm{r}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{i}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{a}\mathrm{t}\mathrm{C}\mathrm{C}\mathsf{\alpha }$$

The CCα and CCβ results are reported in

Table 5. Regulatory limits [11] and CCα and CCβ values determined for OTC, DC, TC, and CTC.

Analyte	MRL [11]	CCα	CCβ
	(µg kg−1)	(µg kg−1)	(µg kg−1)
OTC	100	110.41	120.47
DC	3 (LOD)	5.04	12.06
TC	100	111.91	125.22
CTC	100	129.38	164.35

.

3.7. Matrix Effect (ME)

The sample type can influence the analytical signal through an ion suppression or enhancement mechanism (reducing or enhancing a detector’s ability to detect analytes) when matrix components coelute with target analytes and compete in the ionization process at the ESI source. The milk treatment procedure was designed and optimized with the aim of minimizing potential matrix effects, mainly attributable to the lipid component of the milk, whose level increased when passing from bovine milk to sheep’s milk, and the SPE cleaning phase made it possible to eliminate the chemical–physical differences between milk samples from different animal species.

To evaluate the ME, the slopes of the matrix-matched calibration curves (Slope A) and of the standard calibration curves (Slope B) were compared (Figure 4).

As shown in Figure 4, for all the tetracyclines, slope A was smaller than slope B, indicating ion suppression effects, which were quantified according to the following equation:

$$ME\left(\%\right)=\left({\displaystyle \frac{SlopeB-SlopeA}{SlopeB}}\right)\times 100$$

(4)

For all substances, ME% was within the tolerance limit of 20%, with 18.51% for CTC, 13.05% for DC, 8.19% for TC, and 6.12% for OTC; therefore, thanks to adequate sample cleanup, it was reasonable to quantify the tetracyclines based on the solvent calibration curve.

3.8. Monitoring Quality Assurance

The developed method was periodically subjected to proficiency tests (Test Veritas) in order to monitor its reliability and effectiveness for application to the routine analysis of milk samples arriving at the Drug Residue Laboratory of the Experimental Zooprophylactic Institute of Sardinia for the regional and national control plan for drug residues in food.

In the two-year period from 2023 to 2024, this method was used in participation in two proficiency tests organized by Test Veritas, Progetto Trieste—Proficiency Testing Schemes for the agri-food sector. In the analyzed bovine milk samples, OTC and DC were correctly identified and quantified, with z-scores (the difference between the assigned and determined concentration value for one or more analytes in the proficiency test sample) for quantification lying between −1 and 2.

3.9. Method Application: OTC Determination in Sheep Milk

The validated method was applied to obtain the quantitative concentration of OTC in the udder milk of 10 healthy Sardinian sheep and 10 such sheep naturally infected with Streptococcus uberis mastitis, the most common disease in sheep dairy farms associated with the use of antimicrobials and the related risk of developing resistance to antimicrobial drugs. Twenty analytical sessions, one for each sheep included in the experiment, were carried out for the analysis of over 400 sheep milk samples delivered to the laboratory.

In each session, the analytical series were planned and conducted as follows: a five-point OTC standard solution calibration curve preceded and was followed by a blank reagent; then, a procedural milk blank sample (the negative control, NC) was inserted every ten real samples, along with three milk blank samples (the positive control, PC) fortified at different OTC concentration levels, also preceded and followed by negative control samples.

The blank milk samples were always represented by the milk of each ewe at time T0, before OTC administration. In this way, the method’s specificity was verified by the absence of interfering species in the milk chromatograms at the OTC retention time and confirmed by the instrumental specificity of high-resolution mass spectrometry since, in all positive samples, the FS-ddMS2 mass spectra always showed the typical OTC fragmentation pattern (Table 1) with a mass accuracy <5 ppm. On the other hand, the presence of the positive control samples allowed us to monitor the accuracy of the method in its application via continuously calculating the recovery.

The results regarding the OTC concentrations over time are graphically represented in Figure 5.

The normality distributions of the OTC concentrations were assessed with the Shapiro–Wilk normality test. Median and interquartile range were used to describe quantitative variables. A nonparametric statistical analysis (i.e., the Wilcoxon matched-pair signed-rank test) was performed to investigate intergroup differences in milk antibiotic concentrations between the healthy and infected sheep during a follow-up (Figure 5), for which a two-tailed p-value < 0.05 was considered statistically significant.

4. Discussion

4.1. Optimization of Sample Treatment and Instrumental Analysis

Sample preparation is a crucial issue in food analysis because it can be a source of inaccuracies and limit the development of high-throughput methods [34]. Among foods, milk has a very complex and heterogeneous matrix, and its composition varies depending on the species that produced it. In particular, in sheep’s milk, there can be significant matrix effects that are difficult to avoid during analytic extraction. The development of the milk sample preparation method essentially involved two steps:

• Removal of potential interferences;
• Concentration of the analytes.

In this work, a combination of extraction with McIlvaine–EDTA buffer and SPE purification followed by the evaporation of the extract allowed us to obtain quantitative recoveries of four tetracyclines—TC, OTC, DC, and CTC—in different types of milk, working with minimal sample quantities and an injection volume of 5 µL in the chromatographic system, almost completely eliminating the matrix effect.

Chromatographic conditions were designed to improve the sensitivity and resolution of all the analytes in the positive ionization mode. Several experiments were performed to optimize the mobile-phase conditions. The best response and separation for tetracyclines—especially for TC and DC, which have the same molecular mass and the same fragmentation pattern, so the distance of retention times of which is very important—was obtained with a rapid gradient using formic acid in water and acetonitrile as mobile phases. UPLC columns and gradient profiles were also evaluated to further optimize the chromatographic separation of the analytes. Three columns were tested, including the Agilent Infinity Lab Poroshell 120EC-C18 (150 mm × 3.0 mm, 2.7 µm), Waters ACQUITY BEH C18 (100 × 2.7 mm, 2.1 µm), and Phenomenex F5 (150 mm × 2.1 mm, 2.6 µm). Optimal results were obtained from the ACQUITY BEH C18 column (100× 2.7 mm, 2.1 µm) with the gradient described in Section 2.3.

4.2. OTC Concentration in Milk

No statistical differences were observed in the median [interquartile range] OTC concentration in the udder at 12 h after inoculation between the healthy (2.7 [2.2–3.0] mg kg⁻¹) and infected (2.7 [1.6–3.9] mg kg⁻¹) ewes. A significantly higher (p < 0.05) OTC content in the milk from infected animals starting from 24 up to 204 h after antibiotic administration was noted; in particular, a prolonged plateau-like curve was observed in the infected mammary glands from 12 h to 48 h before the onset of the decrease while the drop curve started after 12 h for the healthy animals. At 48 h, the recorded OTC concentrations were 2.9 [1.4–4.6] mg kg⁻¹ and 0.5 [0.5–0.6] mg kg⁻¹, constituting the largest difference between the experimental groups; however, the wider intragroup variability in antibiotic distribution in the infected animals compared to the healthy ones must be highlighted. The significant difference in OTC concentration observed in half-udder milk between the healthy and infected sheep could be associated with the pH value increasing in mastitic milk [35] and with the increased permeability of the blood/milk barrier due to damage to mammary gland tissue.

5. Conclusions

In this study, we developed and validated a UPLC–Orbitrap–HRMS method for the quantification of the four tetracyclines oxytetracycline, doxycycline, chlortetracycline, and tetracycline, widely employed in veterinary medicine in the milk of different animal species including bovines, sheep, and goats. The developed method showed good validation performance in terms of sensitivity, trueness, and precision, with recoveries ranging from 77.9 to 110.5% according to CD 2002/657/EC [12] requirements. The validation study allowed us to assess the LOD and LOQ (2.0 ÷ 11.0 µg kg⁻¹) and CCα and CCβ in MRL concentrations.

A simple sample extraction, the same used for milk of different species, i.e., sheep, bovines, and goats, followed by SPE purification and the concentration of the analytes, was combined with an instrumental analysis time of just 8 min via the LC–Orbitrap–HRMS technique. This method is suitable for use for rapid confirmation in official food control laboratories due to its high precision, sensitivity, and applicability. The sample extraction procedure reduces the matrix effect, which could lead to strong signal suppression in the case of high-fat milks such as sheep’s milk. In this way, the analytes can be quantified using a solvent calibration curve, ensuring a further benefit in terms of analysis time.

Its analytical performance is fully in line with regulatory criteria, meaning that the UHPLC-HRMS methodology, with a fast analysis time of 8 min using the Orbitrap Exactive analyzer, is an efficient and useful analytical tool for the quantitative confirmation of the presence of four tetracycline antibiotics in milk, selecting, as an effective sample treatment procedure, a simple and rapid combination of extraction with EDTA–McIlvaine buffer at pH 4 and purification with HLB SPE.

Finally, the validated method was successfully applied to determine the OTC concentration levels in over 400 half-udder milk samples from healthy Sardinian sheep and those naturally infected with Streptococcus uberis after the intramuscular administration of antibiotics for an experimental study on the pharmacokinetics/pharmacodynamics of oxytetracycline in sheep, demonstrating its possible application in a wide range of research fields for pharmaceutical, biological, and food safety analysis.