Determination of macrolide antibiotics in meat and fish using pressurized liquid extraction and liquid chromatography–mass spectrometry

Journal of Chromatography A, 1208 (2008) 83–89

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Determination of macrolide antibiotics in meat and fish using pressurized liquid extraction and liquid chromatography–mass spectrometry夽 Houda Berrada a,∗ , Francesc Borrull b , Guillermina Font a , Rosa Maria Marcé b a b

Laboratory of Food Chemistry and Toxicology, Faculty of Pharmacy, Universitat de València, Avda Vicent Andres Estellés s/n, 46100 València, Burjassot, Spain Department of Analytical Chemistry and Organic Chemistry, Universitat Rovira i Virgili, Sescelades Campus, Marcel·lí Domingo s/n, 43007 Tarragona, Spain

a r t i c l e

i n f o

Article history: Received 29 May 2008 Received in revised form 11 August 2008 Accepted 14 August 2008 Available online 4 September 2008 Keywords: Macrolides Meat Fish Food LC–(ESI)MS PLE

a b s t r a c t We developed a method for determining the quantities of seven macrolide antibiotics in meat and fish by using pressurized liquid extraction (PLE) and liquid chromatography–mass spectrometry with electrospray ionization (LC–(ESI)MS). The PLE was optimized with regard to solvents, temperature, pressure, extraction time and number of cycles. The optimum conditions were: methanol as the extraction solvent; a temperature of 80 ◦ C; a pressure of 1500 psi; an extraction time of 15 min; 2 cycles; a flush volume of 150% and a purge time of 300 s. All recoveries for macrolide antibiotics were over 77% at 200 ␮g/kg, except for erythromycin, which was 58%. The repeatability and reproducibility on days in between, expressed as %RSD (n = 12), were lower than 10% and 12%, respectively. The quantification limits of all compounds were 25 ␮g/kg of dry weight of animal muscle except for troleandomycin (50 ␮g/kg). The method was applied to determine the pharmaceuticals in real samples taken from 18 meat and fish samples. The results showed that PLE is quantitative short time consuming technique, with use of smaller initial sample sizes. Greater specificity and selectivity in extraction and increased potential for automation were shown. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Antimicrobials are widely prescribed for both therapeutic and prophylactic reasons against microbial infections and also as growth promoting substances at sub-therapeutic levels in animal farms and aquaculture. Macrolides are a group of antibacterial compounds active against gram-positive and some gram-negative bacteria that are widely used in human and veterinary medicine. The incorrect use of these drugs can leave residues in food products and this can have undesirable effects on consumer health [1,2]. Recently, concern has been expressed that continuous sublethal levels of antibiotics in food have led to the emergence of harmful bacteria resistant to antibiotics. Some macrolides are now restricted to veterinary therapeutic use and maximum residue limits (MRLs) have been established for these substances in muscle tissue, fat, liver and kidneys. Therefore, the use of these antibiotics in foodstuffs is regulated by Council Directive 2377/90 EC [3], which describes the procedure for establishing MRLs for veterinary medicinal products in foodstuffs of animal origin.

夽 Presented at the 7th Meeting of the Spanish Society of Chromatography and Related Techniques, Granada, Spain, 17–19 October 2007. ∗ Corresponding author. Tel.: +34 96 3544958; fax: +34 96 3544954. E-mail address: [email protected] (H. Berrada). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.08.107

Erythromycin, spiramycin, tilmicosin and tylosin are included at annex 1 as substances with MRL values. Josamycin was included at annex III as substances with provisional MRL values. Roxithromycin and troleandomycin have not a marketing authorization for use in food-producing animals in the European Union and no MRL values could be established for these substances. In order to assess the occurrence of these residues in food, sensitive analytical methods are required which enable us to determine multiple compounds simultaneously at quite low concentration levels. One important requirement of the extraction method was that it should be fast and robust enough to allow the rational analysis of a large number of field samples. However, at the same time the method ought to allow the quantification of macrolides with sufficient sensitivity. The literature reports the determination of macrolides in urine [4], plasma [5], soil [6], animal tissues [7,8] and sludge [9,10] using such analytical techniques as liquid chromatography (LC) with ultraviolet (UV) or fluorimetric detection, LC–mass spectrometry (LC–MS) and LC–tandem mass spectrometry (LC–MS–MS) [11,12]. However, there are relatively few published multiresidue methods for determining the selected macrolides in animal tissues [13,14]. In a previous work, we used LC and a diode array detector (DAD) to successfully determine the macrolides in liver and kidney tissues from animals and we confirmed the results with electrospray mass spectrometry [8]. LC–DAD was also used to confirm the identity of the residues according to the European Commission Decision

84

H. Berrada et al. / J. Chromatogr. A 1208 (2008) 83–89

2002/657/EC [15]. An aqueous extraction of these antibiotics from liver and kidney using EDTA-McIlvaine’s buffer before cleaning these by SPE was used [8]. Another study using 0.2% metaphosphoric acid–methanol to extract antibiotics from meat and fish has also been reported [13]. However, meat is protein rich matrix that is important for those antibiotics that bind easily to proteins. Antibacterial–protein binding must be weakened by diluting the sample with a saline dilution, protein denaturation or enzymatic and chemical hydrolysis of the drug–protein complexes [16]. The quantitative determination of total antibiotics required a digestive step, which was time consuming and necessary for meat products to free protein-bound analytes. This step was delicate, with poor potential for automation. Efforts have been directed to attain high-throughput methods able to extract a large number of samples in a short time. Pressurized liquid extraction (PLE; Dionex trade name ASE for accelerated solvent extraction) is investigated here as an alternative technique to avoid a digestive step prior to solid-phase extraction to free bound protein [17]. PLE is a rather new technique that uses solvent at a relatively high pressure and temperature without their critical point being reached. This improves efficiency compared to extractions at room temperature and atmospheric pressure [18,19]. Recently, Göbel et al. have reported extracting macrolides by using rapid and simple PLE procedures with a mixture of water and methanol as the extracting solvent in sewage sludge [9]. In the same way, Schlüsener et al. [19], Jacobsen et al. [6] and Nieto et al. [10] reported using PLE with mixtures of methanol and buffered water for macrolides concentrated in soils and sludges. Other studies have reported using PLE to extract other pharmaceutical compounds from filet tissues, such as sulfonamides in meat [20] and paroxetine and fluoxetine in fish [21]. All groups remarked on the technology’s benefits in providing rapid and reliable analysis. Optimization of the extraction process generally begins by choosing an appropriate extraction solvent. Often, the same solvent used for conventional extractions is initially tested [17]. Other experimental parameters of the extraction are temperature, pressure, static time, and cell size [22–25]. The present study focuses on developing a robust, simple and practical method capable of simultaneously extracting and determining seven macrolide antibiotics which belong to different macrolide subgroups in meat and fish. The macrolide antibiotics selected were erythromycin, josamycin, roxithromycin, spiramycin, tilmicosin, troleandomycin and tylosin. To the best of our knowledge, no studies have been conducted into extracting these macrolide antibiotics from a biological matrix using a PLE.

France), nitrogen was from Carburos Metálicos (Tarragona, Spain) and phosphoric acid (97%) was from Merck (Darmstadt, Germany).

2. Experimental

2.4. Chromatographic analysis

2.1. Materials and reagents

The chromatographic system was an HP 1100 series (Agilent technologies, Waldbronn, Germany) equipped with an automatic injector, a degasser system and a single quadrupole mass detector with electrospray ionization (ESI). The chromatographic column was a Kromasil 100 C18 (25.0 cm × 0.46 cm, 5-␮m particle size) (Teknokroma). A binary mobile phase with a gradient elution was used. Solvent A was Milli-Q water with 1% acetic acid (pH 2.8) and solvent B was acetonitrile. The gradient was 20% B, and was increased to 60% in 25 min, to 100% in 35 min, held at 100% for 2 min and then returned to the initial composition in 4 min. The flow-rate was 1 ml/min, and the column temperature was 35 ◦ C. The injection volume was 50 ␮l and the compounds studied eluted within 20 min. The mass spectrometer simultaneously acquired data in full-scan and under selected ion monitoring (SIM).

Commercial macrolide standards of erythromycin, tylosin hemitartrate and spiramycin were supplied by Riedel-de Haën (Seelze, Germany); josamycin, roxithromycin, troleandomycin and tilmicosin were purchased from Sigma–Aldrich (Madrid, Spain). Standard stock solutions of individual macrolides (1 mg/ml) were prepared by dissolving 25 mg in 25 ml of methanol. These solutions were stored in dark glass bottles at 4 ◦ C and were stable for at least 3 months [8]. A final standard mixture of macrolides was prepared each week. These solutions were also stored at 4 ◦ C. Ultra pure water was obtained with a Milli-Q water purification system (18.2 M cm) (Millipore, Bedford, MA, EEUU), acetone, acetonitrile and methanol (HPLC-grade) were from SDS (Peypin,

2.2. Sample pretreatment Beef, chicken, pork, sea bream and trout filets were collected from different supermarkets in the city of Tarragona (Spain). Samples, around 500 g, were shipped to our laboratory under cool conditions in a portable refrigerator and were frozen at −30 ◦ C upon reception. Samples were lyophilized before being analyzed by the freeze dry system (Labconco, MO, USA). Then they were homogenized using a mortar and pestle and sieved to obtain particles with a diameter of less than 125 ␮m. To optimize the method, a 5 g lyophilized muscle sample was spiked with all compounds, which were dissolved in 10 ml acetone. After spiking, the samples were shaken intensively so that the compounds spread throughout the spiking solution in the sample and were in sufficient contact with the matrix. They were then evaporated to dryness at room temperature. 2.3. PLE extraction Lyophilized muscle samples were extracted using an ASE 200 PLE system (Dionex, Sunnyvale, CA, USA). A total of 5 g of the pretreated muscle was thoroughly mixed with 7 g of aluminum oxide and the mixture was put into a 33 ml stainless steel extraction cell. This cell, filled with aluminum oxide, was positioned in the PLE system connected to a four bottle solvent controller. Nitrogen at a pressure of 10 bars was supplied to assist the pneumatic system and to purge the extraction cells. The aluminum oxide had been heated at 120 ◦ C in the oven for 24 h before use. The extracting solvent was 100% methanol. The operating conditions were as follows: extraction temperature, 80 ◦ C; extraction pressure, 1500 psi; preheating period, 5 min; static extraction, 15 min; final extraction volume, 40 ml; flush volume 150% of the cell volume; nitrogen purge, 300 s; and number of extraction cycles, 2. Each PLE extract was concentrated to about 1 ml in a Büchi R200 (Labortechnik, Flawil, Switzerland) rotary evaporator set at 40 ◦ C and 250 mbar in 50 ml round-bottomed flasks. Then, the extract was transferred to a 15 ml conical tube and the round-bottomed flask was rinsed twice with 0.5 ml of methanol and evaporated to dryness using a multi-sample Turbovap LV Evaporator (Zymark, Hoptkinton, USA) provided with a nitrogen stream and a water bath at 50 ◦ C. After solvent evaporation it was reconstituted in 0.5 ml of methanol. The extract was filtered through a 0.45 ␮m nylon filter (Teknokroma, Barcelona, Spain), and then analyzed by LC–(ESI)MS.

H. Berrada et al. / J. Chromatogr. A 1208 (2008) 83–89

85

Table 1 Method validation parameters for the entire beef analytical method PLE–LC–(ESI)MS Antibiotic

Linear range (␮g/kg)

r2

Ions

LOD (␮g/kg)

MRLsa (␮g/kg)

Erythromycin A Josamycin Roxithromycin Spiramycin Tilmicosin Troleandomycin Tylosin

25–400 25–400 25–400 25–400 25–400 50–400 25–400

0.990 0.993 0.992 0.994 0.992 0.992 0.993

735(100), 576(60), 158(55) 829(100), 174(15) 838(100), 414(80), 679(35) 422(100), 174(35), 843(15) 435(100), 869(20), 174(18) 772(100), 435(10), 814(10) 916(100), 174(15), 772(10)

41 27 35 20 23 51 18

200 – – 200, 250b 50, 75c – 100

a b c

Maximum residue limits established for all food producing species in annexes I–IV of Council Regulations 2377/90. At porcine muscle. At poultry muscle.

The average conditions selected for the optimum performance of the ESI interface in the positive mode were: nebulizer pressure 40 psi, drying gas flow-rate 12 l/min, drying gas temperature 350 ◦ C and capillary voltage 3500 V. Different fragmentation voltages were studied to find spectra with three ions for most of the studied compounds. Fragmentation voltages were defined individually and the values were: 75 V for tilmicosin and troleandomycin and 125 V for erythromycin, josamycin, roxithromycin, spiramycin and tylosin. The ions selected for quantifying the samples are shown in Table 1. The most abundant ion was used for quantification and the second and third ones were used for confirming the results except in the case of josamycin, which has spectra with only two ions. 2.5. Method validation The whole PLE procedure was optimized with beef meat. Carefully checked lyophilized muscle samples were used as blanks before they were spiked with 200 ␮g/kg of each compound for the PLE. To validate the analytical method, beef meat spiked at five different concentrations – 25.0 (50.0 ␮g/kg for troleandomycin), 100.0, 200.0, 300.0 and 400.0 ␮g/kg – was used for matrix-matched calibration standard curves (Table 1). These spiked samples containing macrolide standards were processed through the complete extraction procedure. To make sure no interfering substances were present around the retention time of analytes, we analyzed three meat samples from five different species (beef, chicken, pig, trout and sea bream fish). In all of them, 15 filet samples were analyzed. Repeatability and recovery were assessed by performing tests on blank samples of beef. Six aliquots were fortified at 50, 100 and 200 ␮g/kg. These levels are around the MRLs set by the EU, which is more or less equivalent to MRL/2, MRL, and 2 MRL established levels. As mentioned above, there is no directive in existence regulating tolerance levels of josamycin, roxithromycin and troleandomycin in meat. In this case, analyte recoveries were estimated by following the same procedure as reported above, i.e., at 50, 100 and 200 ␮g/kg. We analyzed each level three times to find the mean concentration and the relative standard deviation (%) of the fortified samples.

These spiked samples were also extracted and injected on different days to find the inter-day precisions (n = 4 × 3). 3. Results and discussion 3.1. PLE optimization The initial conditions for optimizing the PLE were taken from a previous study which determined some of the studied macrolides, among other pharmaceuticals, in sewage sludge samples [10]. These conditions were: a solvent that was made up of water/methanol (50:50, v/v), 1500 psi, 100 ◦ C, 2 cycles, extraction time 15 min, 300 s of purge time, 5 g of dry sample and finally 150% of flush volume. The first challenge when developing the PLE method is choosing an appropriate extraction solvent [26–28]. Acidic water at pH 3.6, methanol, acetonitrile and mixtures of methanol and water were tested to choose the best extracting solvent. Pure organic solvents, such as acetonitrile or methanol, as well as aqueous phosphoric acid/methanol mixtures, were able to extract the studied macrolides from beef meat to a certain degree (Table 2). Acidic water itself showed extraction efficiencies of around 55% for most of the macrolides except for tylosin, which showed 77%. When mixtures of water with methanol were used, lower extraction efficiencies (on average 20% lower) were observed for the compounds compared with single methanol as the extracting solvent. Similar conclusions for some macrolides in sewage sludge were made by Göbel et al. [9]. Based on results presented in Table 2, methanol was chosen as the solvent for extracting macrolides from beef muscle since it recovered over 75% of them, except for erythromycin. These results are comparable with those obtained using other techniques such as liquid–liquid extraction, matrix solid-phase dispersion and solid-phase microextraction, whose recovery values were from 60 to 120% [11], although a larger volume of solvents, and in some cases, several solvents, must be used. According to Lou et al. [27] time, temperature and pressure are the most important variables that can affect PLE efficiency. In order to see if extraction could be improved, we decided to use methanol in various experiments, changing the temperature (40, 60, 80, 100 and 120 ◦ C) and pressure (500, 1000, 1500, 2000 and 2500 psi).

Table 2 Solvent influence on the extraction recovery of studied macrolides from beef meat (n = 3) Extraction solvent

% Recovery (% RSD) Erythromycin A

Acetonitrile Methanol Methanol/water (3:1) Methanol/water (1:1) Methanol/water (1:3) Water pH 3.6

32 (15) 58 (8) 51 (8) 50 (6) 47 (11) 42 (6)

Josamycin 54 (11) 77 (6) 72 (9) 67 (11) 53 (15) 43 (7)

Roxithromycin

Spiramycin

Tilmicosin

Troleandomycin

Tylosin

48 (10) 78 (6) 69 (6) 59 (10) 54 (13) 50 (10)

60 (9) 80 (5) 74 (9) 59 (7) 53 (9) 48 (12)

54 (12) 82 (5) 75 (7) 69 (12) 63 (15) 56 (10)

48 (13) 74 (10) 67 (15) 63 (16) 61 (13) 57 (9)

72 (8) 90 (6) 83 (8) 79 (7) 77 (12) 78 (15)

86

H. Berrada et al. / J. Chromatogr. A 1208 (2008) 83–89

Fig. 2. Recoveries obtained by successive extractions. For experimental conditions, see text.

3.2. Method validation

Fig. 1. (A) Effect of the temperature on the extraction efficiency. (B) Effect of the pressure on the extraction efficiency. For experimental conditions, see text.

As temperatures increase, interactions between analytes and matrix components are weakened and viscosity and surface tension are decreased. As can be seen in Fig. 1a, the best results were obtained at 80 ◦ C, with recoveries ranging from 58 to 90% and %RSDs from 4 to 12% (n = 3). At 100 ◦ C the color, cloudy suspension and %RSDs increased because compounds of high molecular mass were co-extracted. We used methanol in a PLE at different pressures to extract macrolides from beef muscle. As Fig. 1b shows, the highest recovery of macrolides was obtained at 1500 psi. Good recoveries are also obtained at low pressure, close to 500 psi, which subsequent analysis shows is the lowest pressure possible. However, the system becomes unstable (overfilled collections vials) because of difficulties in maintaining the set pressure. High pressure during extraction keeps the solvent in a liquid state when working at temperatures at or above boiling point. Extraction is also more efficient because it forces the solvent into areas where it would not normally go under atmospheric conditions [28,29]. Lengthy exposure to solvents allows the matrix to swell, thus improving the penetration of the solvent into the sample interstices and the contact of the solvent with the analytes [30]. The number of cycles was optimized and four consecutive extractions from the same sample were made. Significant amounts of analytes were found in the second extract but the recoveries for all the compounds were considered negligible in the third cycle, as Fig. 2 shows. For this reason, two cycles were considered optimum and this allowed us to introduce fresh solvent and maintain a favorable balance between solvent and sample. We checked the percentage of flush from 150% down to 50% to see if this increased the preconcentration factor. However, recoveries were higher with 150% flush volume and so this was chosen as the optimum.

The water losses in samples during drying were determined by lyophilisation and this information was taken account to express obtained results at whole (wet) samples. The effect of lyophylisation was also checked and the recoveries were calculated by comparing the peak area obtained from meat samples spiked at 200 ␮g/kg before and after the sample preparation. The data from both pretreatment procedures were compared with a t-test for both compounds. They were not significantly different for a confidence interval of 95%. All research was done on beef meat for validative purposes. The linear range was calculated by matrix-matched calibration standard curves and linearity was good, between 25 and 400 ␮g/kg for most of the compounds with the determining coefficient values (r2 ) above 0.99, except for troleandomycin, which showed good linearity between 50 and 400 ␮g/kg. The recovery and repeatability of the LC–(ESI)MS method was evaluated by spiking six blank samples of beef meat in triplicate with 50, 100 and 200 ␮g/kg of each macrolide. The values were higher than 78% at 200 ␮g/kg and all %RSDs were below 15% (n = 3) except for erythromycin A, whose average recovery was 58% and whose RSD was 12%. For inter-day assays, recovery data were also satisfactory. The inter-day repeatability was determined in triplicate on 4 successive days. The analytical results are summarized in Table 3. When we analyzed samples spiked with three different concentrations (50, 100 and 200 ␮g/kg), recoveries were higher than 68%. To evaluate the ionic suppression effect, data obtained from spiked samples and those obtained from standards at the same concentration as the spike were compared with a t-test. They were not significantly different for a confidence interval of 95%. Ionic suppression was not observed and the experiments proved that recoveries were independent of applied fortification levels. The whole procedure was applied to 15 blank samples of different origins (beef, chicken, pig, sea bream and trout) to verify the specificity of the method and it was shown that no interference was detected around the retention time of the seven analytes in any of the samples analyzed. The recovery values of the macrolides extracted from different matrices were similar (the differences found were less than 15%), indicating that the matrix effects of the origin meat were minimal (Table 4). The method allows recoveries comparable to the most commonly applied extraction methods [8,11,13], without the need for further clean-ups or for increasing throughput due to the high PLE automation grade. The specified PLE extraction procedures generated interferencefree chromatograms at the retention times of the macrolides

H. Berrada et al. / J. Chromatogr. A 1208 (2008) 83–89

87

Table 3 Inter- and intra-day recovery of the method at three levels of spiking beef meat (n = 3) Antibiotic

Spiking level (␮g/kg)

Intra-day recovery (%)

RSD (%)

Inter-day recovery (%)

RSD (%)

Erythromycin A

50 100 200

48 54 58

13 11 8

45 59 57

15 14 12

Josamycin

50 100 200

69 74 77

9 7 6

65 72 74

11 10 8

Roxithromycin

50 100 200 50 100 200

69 73 78 72 84 82

10 6 6 10 7 5

65 69 69 77 81 80

13 12 9 12 8 7

Tilmicosin

50 100 200

73 78 82

7 7 5

67 74 76

14 10 11

Troleandomycin

50 100 200

71 80 84

11 8 4

68 73 74

15 11 7

Tylosin

50 100 200

72 83 90

11 7 4

76 80 87

14 11 7

Spiramycin

Table 4 Recoveries and %RSD (n = 3) of macrolides from various meat and fish matrices at fortification level of 200 ␮g/kg Antibiotic

Bovine meat

Porcine meat

Poultry meat

Sea bream fish

Truite Fish

Erythromycin A Josamycin Roxithromycin Spiramycin Tilmicosin Troleandomycin Tylosin

58 (8) 77 (6) 78 (6) 82 (5) 82 (5) 74 (4) 90 (4)

63 (6) 84 (4) 81 (9) 75 (8) 83 (5) 72 (6) 88 (5)

59 (9) 82 (5) 86 (5) 80 (5) 85 (4) 70 (6) 86 (4)

68 (6) 90 (5) 83 (6) 84 (5) 87 (3) 75 (7) 84 (4)

66 (7) 91 (4) 85 (5) 82 (6) 79 (3) 69 (6) 90 (5)

studied and Fig. 3 shows LC–(ESI)MS chromatograms of spiked beef meat samples at 100 ␮g/kg. The detection limits were experimentally calculated from the analysis of beef samples spiked with a standard mixture of the analytes at serially diluted concentrations as the minimum concentration of an analyte. These were given a signal to noise ratio of 3, as set in the instrument’s software package. The detection limits were lower than 15 ␮g/kg for all compounds and the quantification limits, considered as being at the lower point of the linear range, were 25 ␮g/kg for most compounds. This showed that the method could be useful for determining macrolide residues in contaminated meat and fish tissues. The analytical limits CC␣ and CC␤ were determined as required by Commission Decision 2002/657/EC. Table 5 shows the CC␣ and CC␤ with an error of 5%, considering the experimental standard deviation of within-laboratory reproducibility at the adequate contamination level. In the case of erythromycin, spiramicin, tylosin and tilmicosin, which have an established MRL, CC␣ and CC␤ were calculated by Table 5 CC␣ and CC␤ values for studied macrolides in beef meat Antibiotic

CC␣ (␮g/kg)

CC␤ (␮g/kg)

Erythromycin A Josamycin Roxithromycin Spiramycin Tilmicosin Troleandomycin Tylosin

208 6 8 204 53 12 102

211 15 17 206 54 35 104

analyzing 20 blank beef meat, all fortified with the analyte at the maximum permitted limit according to the EU criteria. However, since josamycin, roxithromycin, and troleandomycin are not licensed for use in veterinary products, it does no have an established MRL and as a consequence, food for human consumption should be free from. CC␣ was calculated for these macrolides by analyzing 20 blanks to be able to calculate the signal to noise ratio at the time window in which the analyte is expected. Three times to the signal to noise ratio can be used as decision limit and CC␤ was calculated by analyzing 20 blanks spiked at the decision limit. The values of the decision limit plus 1.64 times the standard deviation of the within-laboratory reproducibility of the measured content equals the detection capability (␤ = 5%). 3.3. Method application This method has been used to detect the presence of macrolides in raw meat and fish bought from the market and from butchers’ shops in Tarragona. Four pieces of beef, four of chicken, four of pig, three trout filets and three sea bream filets were purchased to investigate the occurrence of macrolides. Among 18 analyzed food samples, three sea bream samples were positive to LC–MS. Only erythromycin A was found at 87, 69 and 58 ␮g/kg with RSDs (%) (n = 3) of 12, 15 and 16%, respectively. Fig. 4 shows the extracted ion chromatogram of a sea bream sample collected in September 2006. The total level of residues found did not exceed 100 ␮g/kg, the set MRL. However the presence of erythromycin A in the edible filets, a compound that should only be administered by injection or in feed, involves the routine control of the antibiotics in food.

88

H. Berrada et al. / J. Chromatogr. A 1208 (2008) 83–89

Fig. 3. SIM chromatograms corresponding to an extract of a meat sample with the compounds added at 100 ␮g/kg.

Fig. 4. SIM chromatogram corresponding to an extract of a sea bream filet sample where erythromycin was found at 87 ␮g/kg.

4. Conclusions This work described for the first time how PLE was used to extract seven macrolide antibiotics from meat and fish filets and how these were simultaneously and efficiently determined using LC–(ESI)MS. The method was precise with good recovery and low quantification limits. Samples were lyophilized before analysis, which gave clean extracts and avoided further clean-ups being needed. The limits of quantification were lower than the MRL and the efficiency of PLE was also comparable to that of conventional techniques. Acknowledgements The authors wish to thank the Ministerio de Ciencia y Tecnología (Spain) for funding this study through the Project AGL

2006-04438/ALI. H. Berrada also thanks the Universitat de València for a mobility grant. References [1] T. Heberer, Toxicol. Lett. 131 (2002) 5. [2] S. Omura (Ed.), Macrolide Antibiotics: Chemistry, Biology and Practice, second ed., Academic Press, Orlando, FL, 2002. [3] Council Directive 2377/90 and the later modifications 1570/98, 2593/99, 2338/2000 and 1181/2002, Commission of the European Communities, Brussels, 1990, 1998, 1999, 2000, 2002. [4] M.J. González de la Huebra, G. Bordin, A.R. Rodríguez, Anal. Chim. Acta 517 (2004) 53. [5] F. Kees, S. Spangler, M. Wellenhofer, J. Chromatogr. A 812 (1998) 287. [6] A.M. Jacobsen, B. Halling-Sørensen, F. Ingerslev, S.H. Hansen, J. Chromatogr. A 1038 (2004) 157. [7] M. Horie, K. Saito, R. Ishii, T. Yoshida, Y. Haramaki, H. Nakazawa, J. Chromatogr. A 812 (1998) 295. [8] H. Berrada, J.C. Moltó, F. Borrull, G. Font, R.M. Marcé, J. Chromatogr. A 1157 (2007) 281.

H. Berrada et al. / J. Chromatogr. A 1208 (2008) 83–89 [9] A. Göbel, A. Thomsen, C.S. McArdell, A.C. Alder, W. Giger, N. Theiß, D. Löffler, T.A. Ternes, J. Chromatogr. A 1085 (2005) 179. [10] A. Nieto, F. Borrull, R.M. Marcé, E. Pocurull, J. Chromatogr. A 1174 (2007) 125. [11] M.J. González de la Huebra, U. Vincent, J. Pharm. Biomed. Anal. 39 (2005) 376. [12] M.J. González de la Huebra, U. Vincent, C.V. Holst, J. Pharm. Biomed. Anal. 43 (2007) 1628. [13] M. Horie, H. Takegami, K. Toya, H. Nakazawa, Anal. Chim. Acta 492 (2003) 187. ˜ [14] M.A. García-Mayor, R.M. Garcinuno, P. Fernández-Hernando, J.S. DurandAlegría, J. Chromatogr. A 1122 (2006) 76. [15] EC Decision 2002/657, Off. J. Eur. Commun. L221 (2002) 8. [16] K. Ridgway, S.P.D. Lalljie, R.M. Smith, J. Chromatogr. A 1153 (2007) 36. [17] R. Carabias-Martínez, E. Rodríguez-Gonzalo, P. Revilla-Ruiz, J. HernándezMéndez, J. Chromatogr. A 1089 (2005) 1. ˜ [18] J.A. Mendiola, M. Herrero, A. Cifuentes, E. Ibanez, J. Chromatogr. A 1152 (2007) 234. [19] M.P. Schlüsener, M. Spiteller, K. Bester, J. Chromatogr. A 1003 (2003) 21.

[20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

89

E.L. McClure, C.S. Wong, J. Chromatogr. A 1169 (2007) 53. S. Chu, C.D. Metcalfe, J. Chromatogr. A 1163 (2007) 112. G. Font, A. Juan, Y. Picó, J. Chromatogr. A 1159 (2007) 233. K. Adou, W.R. Bontoyan, P.J. Sweeney, J. Agric. Food Chem. 49 (2001) 4153. A. Gentili, D. Perret, S. Marcheses, M. Sergi, C. Olmi, R. Curini, J. Agric. Food Chem. 52 (2004) 4614. O.P. Heemken, N. Theobald, B.W. Wenclawiak, Anal. Chem. 69 (1997) 2171. R.M. Alonso-Salces, E. Korta, A. Barranco, L.A. Berrueta, B. Gallo, F. Vicente, J. Agric. Food Chem. 49 (2001) 3761. X.W. Lou, H.G. Janssen, C.A. Cramers, Anal. Chem. 69 (1997) 1598. J. Gan, S.K. Ppiernik, W.C. Koskinen, S.D. Yates, Environ. Sci. Technol. 33 (1999) 3249. H. Giergielewicz-Mozajska, L. Da Browski, J. Namiesnik, Crit. Rev. Anal. Chem. 31 (2001) 149. B.E. Richter, B.A. Jones, J.L. Ezzel, N.L. Porter, N. Avdalovic, C. Pohl, Anal. Chem. 68 (1996) 1033.