Development of a new screening method for the detection of antibiotic residues in muscle tissues using liquid chromatography and high resolution mass spectrometry with a LC-LTQ-Orbitrap instrument

A liquid chromatography–high resolution mass spectrometry (LC–HRMS) method was developed for screening meat for a wide range of antibiotics used in veterinary medicine. Full-scan mode under high resolution mass spectral conditions using an LTQ-Orbitrap mass spectrometer with resolving power 60,000 full width at half maximum (FWHM) was applied for analysis of the samples. Samples were prepared using two extraction protocols prior to LC–HRMS analysis. The scope of the method focuses on screening the following main families of antibacterial veterinary drugs: penicillins, cephalosporins, sulfonamides, macrolides, tetracyclines, aminoglucosides and quinolones. Compounds were successfully identified in spiked samples from their accurate mass and LC retention times from the acquired full-scan chromatogram. Automated data processing using ToxId software allowed rapid treatment of the data. Analyses of muscle tissues from real samples collected from antibiotic-treated animals was carried out using the above methodology and antibiotic residues were identified unambiguously. Further analysis of the data for real samples allowed the identification of the targeted antibiotic residues but also non-targeted compounds, such as some of their metabolites.

Keywords: LC/MS; veterinary drug residues, antibiotics; animal products, meat

To date, for screening residues in food of animal origin, antimicrobial agents used in veterinary medicine are mainly detected by microbiological assays using plate test bacterial growth inhibition techniques, such as the four-plate test or the STAR test (Kilinc et al. [15]; Pikkemaat [20]; Gaudin et al. [6]). Microbiological methods offer low-cost analysis and do not require expensive equipment. However, due to their mode of detection, even if it is possible to identify one or more families of antibiotics using post-screening microbiological methods, they always lack specificity. It means that using such microbiological methods, it is not possible to discriminate one antibiotic from another. Moreover, they are quite often not sensitive enough to reach the maximum residue limits (MRLs) set by the European Commission regulation (EU) 37/2010/EC.

It is well known that, for confirmatory purpose in chemical residue testing, mass spectrometry is the technique of choice. A chemical approach based on mass spectrometric detection brings the specificity needed to chemically identify an antibiotic compound, even at the screening step. In the last decade, many analytical methods based on (very high pressure) liquid chromatography coupled to tandem mass spectrometry instruments (VHP) LC–MS/MS have been developed for multi-antimicrobial residue screening (Granelli and Branzell [9]; Kaufmann et al. [13]; Bohm et al. [1]; Gaugain-Juhel et al. [7]). The use of HPLC (or U-HPLC) coupled to tandem MS combines chemical separation of analytes with the selectivity and the sensitivity of mass detection achieved in multiple reaction monitoring (MRM) mode. To enhance confidence in molecular identification, the analytes are usually detected by monitoring the ionic signal of at least two mass transitions, in combination with determining chromatographic retention time. This approach is considered "pre-target screening," because analytes are pre-selected prior to their mass signal acquisition. This analytical technique in MRM mode needs selection of the compounds supposed to be monitored prior to the analysis and an optimization on each compound is necessary to fully determine the parameters of detection of their mass transitions (i.e. precursor ion, product ions, collision energy and voltages of the source). Consequently, only compounds included in the MRM method will be possibly detected, and other contaminants could never be detected, even though they were actually present in the sample.

More recently, new approaches using high resolution mass spectrometry (HRMS) have been reported for screening residual compounds with equipment such as time-of-flight mass detectors (TOF) or Orbital trap mass detectors (Orbitrap). These instruments allow full-scan acquisition of all signals obtained from the ionisation source, without pre-selecting any compounds. This approach in screening for trace amounts of chemicals is considered "post-target screening." Analytes are searched for after their mass acquisition. The selectivity is obtained from a full-scan acquisition of signals by extracting the ion chromatogram of the accurate mass of the target ions, thanks to filters based on narrow mass windows (3–20 ppm). This option also offers the possibility to retrospectively analysing the whole set of acquired data, without limiting in the number of compounds to be searched. This post-target approach has been applied recently for screening of marine toxins (Skrabakova et al. [22]; Gerssen et al. [8]), pharmaceuticals in waste water (Petrovic et al. [19]), veterinary drugs (Kaufmann et al. [13], [14]; Ortelli et al. [17]; Peters et al. [18]; Stolker et al. [23]) or pesticides in the environment, food and biological matrices, and also for screening drug abuse during horse-doping control (Moulard et al. 2011).

Non-target screening, looking for unknowns without any previous information on their chemical identity, can also be implemented from the full-scan mass acquisition data using the selectivity of high resolution mass spectrometry (HRMS) and adding the power of extractive/statistical software. Processing data from the full-scan chromatograms can eventually lead to extraction and chemical recognition of new biomarkers or trace compounds. This smart approach has recently been used by Hogenboom et al. ([10]) for environmental samples to search for emerging water contaminants and by Dervilly-Pinel et al. ([4]) to search for metabolic biomarkers adapted to the screening of anabolic steroid treatment in calves. This mass spectrometry-based "metabolomic" strategy opens a new trend in the field of veterinary drug residue control.

In our laboratory, pre-target screening using LC–MS/MS in MRM mode has been developed and validated for the identification of 60 antibiotics, all belonging to the main antimicrobial families (i.e. cyclines, penicillins, cephalosporins, macrolides, aminoglycosides, sulfonamides and quinolones), in pig muscle tissues and in cows milk. This method monitors these antimicrobials at their MRL level, employing simple and fast extraction (Hurtaud-Pessel et al. [11]; Gaugain-Juhel et al. [7]). This method is currently being collaboratively assessed in France to be proposed either as a post-screening method leading to formal molecular identification or as an alternative to direct screening with microbiological tests in the national monitoring plan for antibiotic residues in meat products. The objective of the work reported here is, first, to assess the transfer of this MRM-based method to LC–HRMS and, second, to develop a "post-target" screening method primarily dedicated to penicillins, cephalosporins, sulfonamides, macrolides, tetracyclines, quinolones and aminoglycosides in muscle tissues.

Our work demonstrates that some modifications in sample preparation are necessary to achieve adequate sensitivity of the HRMS signals at the MRL level for some of the tested compounds. The sensitivity of the method for the whole set of 60 antimicrobials was assessed through analysis of spiked samples. Automatic data processing using specific software (ToxId®) was implemented to allow the automatic identification of the compounds through the evaluation of their respective exact mass in combination with their retention times.

All reagents and solvents used were of analytical- or HPLC-grade. Methanol, trichloroacetic acid (TCA) (analytical grade), formic acid (98–100% for analysis) and ammonium acetate were purchased from Merck (Darmstadt, Germany). Acetonitrile was obtained from Fisher Scientific (St. Quentin Fallavier, France). Heptafluorobutyric acid (HFBA) was obtained from Fluka (St. Quentin Fallavier, France). Water was purified using a Milli-Q-System (Millipore, Molsheim, France).

The standards were obtained from different companies: marbofloxacin, norfloxacin, ciprofloxacin hydrochloride, enrofloxacin, difloxacin hydrochloride, oxolinic acid, nalidixic acid, flumequine, spiramycin, tylosin tartrate, tilmicosin, erythromycin, josamycin, amoxicillin, ampicillin sodium, penicillin-G sodium (=benzylpenicillin), penicillin V (=phenoxymethylpenicillinic acid potassium salt), oxacillin sodium, cloxacillin sodium, dicloxacillin sodium, nafcillin sodium, cephapirin sodium, cefquinome sulfate, cefazolin sodium, cefalonium hydrate, cephalexin hydrate, ceftiofur, cefoperazone sodium, oxytetracyclin hydrochloride, chlortetracyclin hydrochloride, tetracyclin hydrochloride, spectinomycin dihydrochloride, streptomycin sulfate, dihydrostreptomycin sesquisulfate trihydrate, kanamycin sulfate, gentamicin sulfate, neomycin trisulfate hydrate, sulfaphenazole, sulfaguanidine monohydrate, sulfadiazin sodium, sulfathiazole, sulfamethazine, sulfamethoxypyridazin, sulfamonomethoxine, sulfadoxine, sulfaquinoxalin sodium, sulfadimethoxin sodium, sulfamethoxazole and sulfamerazine were purchased from Sigma (St, Quentin, Fallavier, France). Sarafloxacin hydrochloride, doxycyclin hyclate, paromomycin sulfate and apramycin sulfate were obtained from Cluzeau Info Labo (Courbevoie, France); danofloxacin mesylate, tulathromycine and tulathromycine marker from Pfizer (Amboise, France); neospiramycin from Wako (Neuss, Germany) and tylvalosin (=3-O-acetyltylosin) from Eco (London, UK).

Individual stock standard solutions (0.5 mg/ml) were prepared by dissolving the appropriate amount of each standard into water or methanol according to their solubility, i.e. each penicillin compound in 100% water; each cephalosporins and aminoglycosides compound in water/methanol (1/1; v/v); each compound from tetracycline, macrolide and sulphonamide families in 100% methanol. Each quinolone compound stock solution was prepared in 1 N sodium hydroxide/methanol (1:24, v/v). All stock solutions were stored in a dark place at +4°C, except the methanolic solutions which were stored at −20°C. For spiking, dilute composite standard solutions were also prepared in ultra-pure water to obtain the desired concentrations.

A 1-mM HFBA and 0.5% formic acid solution was prepared by diluting 0.065 ml of HFBA and 2.5 ml of formic acid (100%) to 500 ml of water. A 0.5% formic acid solution in methanol/acetonitrile (1:1; v/v) was prepared by diluting 2.5 ml of pure formic acid to 500 ml with methanol/acetonitrile (1:1; v/v). These two solutions were employed as the LC mobile phases A and B, respectively.

A 5% TCA solution in acetonitrile was prepared by dissolving 10 g of trichloroacetic acid in a 10-ml volumetric flask and adjusting the volume with water, then transferring 2.5 ml of this solution to 45 ml of acetonitrile in a 50 ml volumetric flask and adjusting the volume with acetonitrile.

A 5% TCA solution in water was prepared by dissolving 5 g of trichloroacetic acid in a 100-ml volumetric flask and adjusting the volume with water. A 2 -M ammonium acetate solution was prepared by dissolving 15.4 g of ammonium acetate in 100 ml of water. This solution was then diluted 10 times to obtain a 0.2 -M solution.

To allow extraction of all families of studied compounds, two sample preparations were carried out. Twice, a 2 -g amount of minced muscle tissue per sample was accurately weighed and placed into 16-ml centrifuge tubes. Internal standard solution (200 µl of sulfaphenazole at 1 mg l−1) and 800 µl of water were added to each tube.

In the first tube, 8 ml of acetonitrile were added to the sample. After rotary-stirring for 10 min at 100 rpm and centrifugation at 14,000 g for 5 min, 9 ml of the supernatant were transferred into a clean tube and were evaporated to dryness under a nitrogen stream at 50°C. The remaining residue was dissolved in 0.5 ml of 0.2 M ammonium acetate, mixed by vortexing and then filtered onto a 0.45-µm PVDF Millex HV (Millipore) filter of 13 mm diameter prior to injection.

In the second tube, 0.5 ml of 5 % TCA solution in water and 7.5 ml of 5% TCA solution in acetonitrile were added to the sample. After stirring for 10 min and centrifugation at 14,000 g for 5 min, 7.5 ml of the supernatant was transferred into a new tube and 6–7 drops of 12.5 % NH4OH solution were added for neutralization (pH 7). After centrifugation at 14,000 g for 5 min, 7.5 ml of supernatant was transferred for evaporation at 50°C under a nitrogen stream until reducing the volume to about 1 ml. At this step, another centrifugation at 14,000 g for 5 min was performed before continuing the evaporation under nitrogen flow at 50°C till 50–100 µl. The remaining residue was dissolved in 1 ml of water and loaded onto a preconditioned C18 solid-phase extraction cartridge (Bond-Elut®, 200 mg). After washing the cartridge with 1 ml of water, the elution was carried out with 2 × 0.7 ml of methanol. The methanolic solution was evaporated to dryness under a gentle stream of nitrogen at 50°C and the bottom residue was dissolved in 0.5 ml of 0.2 M ammonium acetate. The final solution was filtered onto a 0.45-µm PVDF Millex HV (Millipore) filter of 13 mm diameter prior to LC injection.

Chromatographic separations were performed on an Accela liquid chromatography U-HPLC system (ThermoFisher, Bremen, Germany) equipped with a RP18e Purospher column (125 × 3 mm; 5 µm particle size) from Merck (Darmstadt, Germany) protected by a RP18e guard column (4 × 4 mm, 5 µm particle size). The column was kept at a temperature of 25°C. The flow-rate used was 500 µl min−1, and the injection volume was 25 µl. The mobile phase A consisted of 1 mM HFBA in 0.5% formic acid solution and B of 0.5% formic acid solution in methanol/acetonitrile (50:50; v/v) The elution gradient was linearly ramped from 10 to 95% of eluent B for 12 min and held at 95% for 3 min (12–15 min). Then, the elution gradient was linearly ramped down to 10 % B for 1 min and maintained for 6 min to allow column conditioning for the next injection.

Mass spectral analysis was carried out on LTQ-Orbitrap mass spectrometer XL MS (Thermofisher, Bremen, Germany) equipped with an electrospray ionization interface (ESI) and operated in the positive ion mode. The instrument was calibrated using the manufacturer's calibration solution consisting of three mass calibrators (i.e. the caffeine, the tetrapeptide MRFA and Ultramark®) to reach mass accuracies in the 1–3-ppm range. Parameters of the ion source were as follows: capillary voltage 35 V, ion spray voltage 4.3 kV, tube lens 125 V, capillary temperature 350°C, sheath gas flow 40 (arbitrary units), auxiliary gas flow 10 (arbitrary units) and sweep gas 0 (arbitrary units). Nitrogen was used as the sheath and auxiliary gas in the ion source. The instrument was operated in full-scan FTMS over a m/z range of 100–1200 Da at a resolving power of 60,000 (full width at half maximum). The eluent was directed into the source of the mass spectrometer from 1 to 20 min by using a divert valve.

At the screening step, there are at least two issues of significance for successful implementation of the method: first, the preparation of the sample and second the detection technique. The very first challenge is to develop a generic non-selective extraction able to cover a wide range of compounds of different chemical properties. At the same time, this extraction must demonstrate a high rate of efficiency in order to give sufficient sensitivity and to reach the required detection limits. This efficient sample preparation must then be combined to a detection technique which is not restrictive, i.e. sufficiently fit for all possible compounds, and which can lead to a response for all compounds at their required target limit. LC–HRMS can match with these requirements for detection. The full-scan MS is not restrictive. The only limitation the mass spectrometer holds is the capacity of the compounds forming ions in the ionization source. Of course, the best settings for the ionizing conditions in the source (temperature of source or capillary, flow of gases...) considering a multi-residue method are not those generally proposed to optimize for specific compounds but those which allow satisfactory medium conditions for ionizing all separated compounds entering into the source. Chromatographic separation of the compounds can also become of strategic importance. In our study, the target compounds were all antibacterial veterinary drugs. Among them, penicillins, cephalosporins, sulfonamides, macrolides, tetracyclines, and quinolones, are easily ionizable compounds. Many liquid chromatographic conditions take advantage of a formic acid or an acetic acid solution as the aqueous phase and of MeOH or ACN as the organic phase to separate these compounds through reversed phase LC analytical columns. On the other hand, aminoglycosides are not easily separated in these previously notified conditions and it is one of the reasons why some multi-residue methods developed for monitoring antimicrobial veterinary drug residues do not cover aminoside compounds (Kaufmann et al. [14]). The use of ion pairing agents diluted into the LC mobile phase is a common way for increasing the retention of these compounds on a reversed-phase LC column (Inchauspe et al.[12]). For this purpose, we previously proposed that the separation of all antibacterials could be achieved by adding pentafluoropropionic acid (PFPA) as the aqueous mobile phase instead of formic or acetic acids (Hurtaud-Pessel et al. [11]; Gaugain-Juhel et al. [7]). In the present method, another ion-pairing agent was chosen, heptafluorobutyric acid (HFBA). It is widely accepted by LC–MS analysts that the use of PFPA or HFBA may decrease the sensitivity of signals entering the mass spectrometer detector compared to the use of formic acid or acetic acid. Nevertheless, it is also one of the compromises we proposed to provide fairly good detection for all the targeted antibacterials.

Starting from the sample preparation previously developed in our laboratory (Hurtaud-Pessel et al. [11]; Gaugain-Juhel et al. [7]), two extraction routes were finally implemented to cover all 60 antibacterials. The first extraction with acetonitrile followed by an evaporation step was tested and found suitable for macrolides, sulfonamides, penicillins and cephalosporins. The second extraction with 5% TCA did not fit because the sensitivity of the signals was too low for some of the analytes from the tetracycline and the aminoglycoside families. A concentration step was, therefore, introduced. Extraction in acidic medium, with precipitation of proteins using TCA dissolved in acetonitrile, was chosen to continue with a concentration step by evaporation of the ACN. Neutralization was then necessary and a further clean-up using SPE was undertaken to reach the target detection level for tetracyclines, aminoglycosides and quinolones.

The list of the monitored compounds is given in Table 1. The identification of the compounds is based on their exact mass in positive mode and their corresponding retention time. The high resolving power of the Orbitrap, combined with high mass accuracy, leads to the requested selectivity to identify a compound using its exact mass. In this method, a resolving power of 60,000 FWHM was chosen for the full-scan mass acquisition. This resolution was excellent, even though decreasing it to 30,000 FWHM could also give satisfactory results. When the sample was collected from a complex biological matrix, bringing signals to a high background created from a huge number of matrix-generated ions, then the specific extracted ion mass chromatogram obtained from the full-scan chromatogram using a narrow mass window (5 ppm) provided a sharp peak representative of the specific compound alone without any other interference. If a higher mass window was used, for example 50 or 200 ppm, then many interfering ions appeared on the extracted ion mass chromatogram (Figure 1).

Graph: Figure 1. Extraction ion chromatograms of ampicillin (MH+ at m/z 350.11690) spiked in bovine muscle at 50 µg/kg with different extraction mass windows: (a) 200 ppm, (b) 50 ppm and (c) 5 ppm.

Table 1. List of compounds with molecular formula, exact mass of MH+, expected retention time and level of fortification.

To evaluate the performance of the LC–HRMS screening method developed in our study, some characteristic parameters have been determined. In the field of veterinary drug residues, Commission Decision [2]/657/EC lays down criteria for the validation of analytical methods used for screening or confirmatory purpose. In 2010, a new guideline dedicated to the validation of screening methods for the monitoring of residues from veterinary medicines has been edited to technically complete the Commission Decision [2]/657/EC (European Union [5]). According to the Commission Decision [2]/657/EC, the characteristics of performance to be determined specifically for a qualitative screening method are the detection capability of the method, also called CCβ, its selectivity/specificity against various interferences and its applicability/ruggedness/stability. Moreover, it is stated that a method utilized for screening purposes should display a false compliant sample rate lower than 5%. CCβ is the concentration at which only ≤5% of false compliant results remain possible. In case of analytes with an established regulatory limit (MRL for instance), CCβ must be less or equal to the regulatory limit. During the validation period, to demonstrate that the CCβ of the method is in full accordance with the regulatory/action limit, a minimum of 20 different representative samples and a maximum of 60 of them should be tested, depending on the level of sensitivity of the method. The more sensitive the method, the less number of samples to validate. In our study, no complete validation as stipulated in the guidelines has been implemented yet, but a pre-validation study was undertaken. Only a small number of different bovine muscle samples (<5) have been selected when 20–60 different samples should have been taken from different food-producing animal species.

To assess the method, all targeted antimicrobial compounds in Table 1 were tested. The compounds were divided into several groups sorted per family and were spiked at a screening target concentration, which corresponds to the MRL level or any other level of interest especially for compound bearing no MRL. Four repetitions were performed for each group. An internal standard, sulfaphenazole, was spiked to each sample prior to the extraction, to evaluate the extraction efficiency and to control the retention time and the mass accuracy. From these experiments, all retention times were found stable. For example, the relative standard deviation (n = 56) calculated for the retention time of the sulfaphenazole internal standard is 0.34%. The mass accuracy also showed good stability. The accurate mass measurement of the internal standard sulfaphenazole (m/z 315.09102) was operated for all the extracted samples (n = 56) and the deviation of the measured accurate mass ranged from −2.0 to 0.03 ppm over a period of 5 days of validation. These mass measurement errors show the high stability of the mass spectrometer and, thus, allow the use of a narrow mass extraction window of 5 ppm. This range of experimental mass errors fits quite well with the specifications of the LTQ-Orbitrap given by the manufacturer using external calibration (3 ppm).

The data were further processed with ToxId® software, using a previously created searched list of compounds. This allowed identifying automatically each compound using the theoretical exact mass with mass windows of 5 ppm and the expected retention time. All compounds were positively identified in each spiked sample using ToxID when the following criteria were met: RT in accordance with the expected RT, measured accurate mass in accordance with the expected accurate mass with a tolerance of 5 ppm, and peak intensity higher than an arbitrary threshold of 10,000. This arbitrary threshold has been established examining chromatograms of blank samples and was the limit chosen to distinguish positive from negative samples. With an intensity lower than 104, the peak is considered an artefact. For sulfonamides, two pairs of isobaric compounds, sulfamethoxypyridazine and sulfamonomethoxine, displaying the same elemental composition, have the same exact mass MH+ at m/z 281.07028. These compounds are differentiated only by their respective retention time at 5.5 and 6.1 min, respectively, as shown in Figure 2. The same situation occurred with sulfadoxine and sulfadimethoxine at m/z 311.08085 (Figure 2), and with tetracycline, epi-tetracycline and doxycycline at m/z 445.16054 (Figure 3). Flumequine at m/z 262.08739 and oxolinic acid at m/z 262.07099 are easily differentiated with a resolution of 60,000 FWHM (Figure 4). No antibacterial compounds were detected in any of the blank muscle tissue samples.

Graph: Figure 2. (a) Total ion chromatogram obtained from LC–HRMS analysis of a spiked muscle with 12 sulfonamides at 100 µg/kg. (b) Extracted ion chromatogram of sulfamethoxypyridazine and sulfamonomethoxine at m/z 281.07028 with extraction window of 5 ppm in spiked muscle at 100 µg/kg. (c) Extracted ion chromatograms of sulfadimethoxine and suldafoxine at m/z 311.08085 with extraction window of 5 ppm in spiked muscle at 100 µg/kg.

Graph: Figure 3. (a) Total ion chromatogram obtained from LC–HRMS analysis of a spiked muscle with tetracyclines and epi-tetracyclines compounds at 100 µg/kg. (b) Extracted ion chromatogram of ion MH+ at m/z 445.16054 with extraction window of 5 ppm in spiked muscle at 100 µg/kg, corresponding to tetracycline, epi-tetracycline and doxycycline.

Graph: Figure 4. (a) Total ion chromatogram obtained from LC–HRMS analysis of a spiked muscle with quinolones compounds at level betweewwn 100 and 300 µg/kg. (b) Extracted ion chromatogram of flumequine at m/z 262.08739 with extraction window of 5 ppm in spiked muscle at 200 µg/kg. (c) Extracted ion chromatogram of oxolinic acid at m/z 262.07099 with extraction window of 5 ppm in spiked muscle at 100 µg/kg.

The sensitivity of the method was very high for macrolides, quinolones and lincosamides, high for sulfonamides and tetracyclines. For penicillins, cephalosporins and aminoglycosides, the sensitivity was lower but no problem of identification occurred except for penicillin V, for which a weak signal is observed. Figure 5 displays the intensity of the signal for the whole set of compounds. The arbitrary threshold set at 10,000 was the minimum intensity expected for a possible identification using automatic processing with ToxID. The limits of detection (LOD) were calculated from each compound comparing the intensity of the signal obtained for the spiked samples at the target screening concentration to the threshold of 10,000 (Table 1).

Graph: Figure 5. Mean signal intensity obtained from each compound spiked in muscle samples (n = 4) at level of validation.

The applicability of the method was tested on some incurred samples of muscle tissues collected from cows and swine administered veterinary antibiotic treatments. The same samples were also analysed using the LC–MS/MS method in MRM mode. Samples were extracted, analyzed using the LC–HRMS method and processed using ToxId software. In these different samples, sulfadimethoxine, doxycycline, penicillin G, DHS and tulathromycine were detected both using LC–HRMS method and LC-QqQ. However, no quantification was made in the various samples, as the objective of the method was only for screening, even though quantification using LC–HRMS with the Orbitrap system was feasible. The additional advantage of the LC–HRMS method was the opportunity offered to search for the presence of additional compounds retrospectively from the full-scan spectrum. For example, in one beef muscle, sulfadimethoxine was found and identified using retention time and exact mass. In this sample, comparing chromatograms to the chromatogram of a blank muscle tissue, one other compound was selectively detected at 7.9 min and m/z 353.09142. The identification as N4-acetyl-sulfadimethoxine was further confirmed by CID fragmentation experiments and by comparison with chemical standard. This compound was then added to our ToxId processing file. Even if this compound is not regulated and is not displaying antibacterial activity, its detection in animal tissues is the undoubted evidence of an administration of the parent drug to the animal.

From screening, the further step for definitively confirming an antimicrobial compound has been developed using the LTQ-Orbitrap LC–FTMS instrument. Indeed, the LTQ-Orbitrap XL offers some other possibilities; for example, to operate fragmentation of a selected precursor ion either in the linear ion trap (CID) or in the High Collision Dissociation cell (HCD). The detection of product ions can also be performed using either the linear ion trap detector or the Orbitrap detector. Therefore, there are at least three possible ways of obtaining further confirmation of a detected compound:

• 1. CID with detection in ion trap leading to low resolution mass measurement of products ions.
• 2. CID with detection in Orbitrap leading to high resolution mass measurement of products ions.
• 3. HCD with detection in Orbitrap leading to high resolution mass measurement of products ions.

The LC–HRMS method reported here has been successfully pre-evaluated for the screening of at least 63 antimicrobial compounds in muscle tissue. In comparison with the targeted LC–triple quadrupole method currently used for screening in our laboratory, this approach using full-scan mass acquisition offers the possibility to analyse retrospectively sets of data. The application of the method to real-life contaminated samples showed that veterinary drug metabolites, which are proof of veterinary treatment, can easily be searched from the data by extracting the exact mass ion chromatograms. Of course, these metabolites have to be confirmed and could further be included in the extending list of searched compounds in our ToxID file. We intend to open this method to other classes of veterinary drugs, such as NSAIDs, antiparasitic or anticcoccidial drugs. There is theoretically no limit to the number of compounds to be acquired. The limitation in developing a unique multi-class multi-residue screening method is sample preparation, achieving suitable ionization of the compounds and the sensitivity of the signals obtained.

The next issue remaining unsolved is to determine whether the exact mass combined with retention time are sufficient to unambiguously confirm a compound. Using an LTQ-Orbitrap, fragmentation is possible either through CID or through HCD devices and the measurement of fragment ions either with low or high resolution. To date, there are no criteria laid down in any international Guidelines or in Commission Decision [2]/657/EC for these new approaches using new HRMS instruments, such as TOF or Orbitrap system.