Anal. Chem. 2002, 74, 1393-1401
Quantitative Analysis of Amoxycillin and Its Major Metabolites in Animal Tissues by Liquid Chromatography Combined with Electrospray Ionization Tandem Mass Spectrometry Siegrid De Baere,* Marc Cherlet, Kris Baert, and Patrick De Backer
Faculty of Veterinary Medicine, Department of Pharmacology, and Pharmacy and Toxicology, Ghent University, Salisburylaan 133, B-9820 Merelbeke, Belgium
A sensitive and specific method for the quantitative determination of amoxycillin and its major metabolites (amoxycilloic acid, amoxycillinpiperazine-2′,5′-dione) in animal tissue samples using liquid chromatography combined with positive electrospray ionization tandem mass spectrometry (LC/ESI-MS/MS) is presented. A liquid extraction using an aqueous 0.01 M potassium dihydrogenphosphate solution as the extraction solvent was performed for a preliminary sample cleanup. The extracts were further purified by a solid-phase extraction using an octadecyl (C18) column. Ampicillin was used as the internal standard. Chromatographic separation of the analytes of interest was achieved on a reversed-phase Hypersil column (100 × 3 mm i.d., dp, 5 µm), using a mixture of 9.6 mM pentafluoropropionic acid in water and acetonitrile as the mobile phase. Gradient elution was performed. To obtain as high sensitivity and selectivity as possible, the mass spectrometer was operated in the multiple reaction monitoring mode. The method was validated for the analysis of amoxycillin and its investigated metabolites in various porcine tissues, kidney, liver, muscle, and fat, according to the requirements defined by the European Community. Calibration graphs were prepared for all tissues, and good linearity was achieved over the concentration range tested (25-500 ng/g, r g 0.9974, and goodness of fit e9.6). A limit of quantification of 25 ng/g was obtained for amoxycillin and its metabolites in all tissues, which corresponds to half the maximum residue limit for amoxycillin. Limits of detection ranged from 2.3 to 12.0 ng/g for amoxycillin and from 1.1 to 15.1 ng/g and 0.2 to 2.4 ng/g for amoxycilloic acid and amoxycillinpiperazine-2′,5′-dione, respectively. The results for the within-day precision and the trueness fell within the ranges specified. The method has been successfully used for the quantitative determination of amoxycillin and its major metabolites in tissue samples from pigs medicated via the drinking water, proving the usefulness of the developed method for application in the field of residue analysis. Amoxycillin, a broad spectrum R-amino-substituted β-lactam antibiotic, is widely used for human and animal therapy of primary 10.1021/ac010918o CCC: $22.00 Published on Web 02/14/2002
© 2002 American Chemical Society
respiratory, gastrointestinal, urogenital, and skin bacterial infections and of secondary infections caused by amoxycillin-sensitive bacteria, following a viral disease. However, improper use of antibiotics may result in undesirable residues in edible animal tissues and consequently pose a human health hazard. For example, trace amounts of antibiotics or their metabolites in food may cause allergic reactions in some individuals. In addition, the wide use of antibiotics also may reduce their efficacy to treat diseases, due to the occurrence of new strains of bacteria that are resistant to many types of antibiotics. 1 Therefore, regulatory agencies within the European Union (EU) have defined a maximum residue limit (MRL) of 50 ng/g for amoxycillin in muscle, liver, kidney, and fat tissues of all foodproducing animals. 2 The chemical analysis of amoxycillin (Figure 1A) in complex matrixes such as tissues is not straightforward and presents particular difficulties: its amphotheric nature and high polarity may cause it to elute among other endogenous, polar substances and precludes the use of standard liquid extraction steps. In addition, amoxycillin is unstable in media of high or low pH.3 As a consequence, methods available in the literature for the determination of amoxycillin are limited. So¨rensen et al. described a multiresidue high-performance liquid chromatographic (HPLC) method for the analysis of seven penicillins in muscle, liver, and kidney tissues of cattle and pigs. Sample preparation consisted of protein precipitation and extraction of the penicillins in aqueous medium, followed by solid-phase extraction on divinylbenzene-co-N-vinylpyrrolidone polymeric sorbent and a final cleanup using diethyl ether. Ultraviolet (UV) detection was performed at 323 nm after derivatization of the analytes with benzoic anhydride and 1,2,4,-triazolemercury(II) reagent. The limit of detection (LOD) was ∼10 ng/g, but a limit of quantification (LOQ) was not mentioned.4 * To whom correspondence should be addressed: (phone) 0032 9 264 73 46; (fax) 0032 9 264 74 97; (e-mail)
[email protected]. (1) Ang, C. Y. W.; Luo, W.; Hansen, E. B.; Freeman, J. P.; Thompson, H. C. J. AOAC Int. 1996, 79 (2), 389-396. (2) Commission Regulation (EC) 508/1999 of 4/3/1999 amending Annexes I-IV to Coucil Regulation (EEC) 2377/90 laying down a Community procedure for the establishement of maximum residue limits of veterinary medicinal products in foodstuffs of animal origin. Off. J. Eur. Commun. 1999, L60, 16. (3) Menelaou, A.; Somogyi, A. A.; Barclay, M. L.; Bochner, F. J. Chromatogr., B 1999, 731, 261-266.
Analytical Chemistry, Vol. 74, No. 6, March 15, 2002 1393
Figure 1. Chemical structure of amoxycillin (A), 5R,6R-amoxycilloic acid (B), 5S,6R-amoxycilloic acid (C), and amoxycillinpiperazine-2′,5′dione (D).
Rose et al. reported a HPLC-UV method for the determination of amoxycillin and ampicillin in liver and muscle tissues. The sample preparation was even more laborious than that applied by So¨rensen et al. and consisted of a liquid extraction and protein precipitation step, followed by two solid-phase extraction (SPE) steps (using cation exchange and porous graphitic carbon extraction columns) and a derivatization step (using acetic anhydride and triazole/mercuric chloride). Detection was performed at 325 nm. No LOD and LOQ values were mentioned. 5 The HPLC-fluorescence method reported by Ang et al. was very sensitive (LOD ) 0.5 ng/g and LOQ ) 1.2 ng/g in catfish tissue). The sample preparation, however, was also time-consuming (liquid extraction and protein precipitation, followed by SPE extraction using C18 columns and additional cleanup by liquidliquid extraction with diethyl ether). The extracts had to be derivatized with formaldehyde and trichloroacetic acid to convert amoxycillin to a fluorescent derivative. 1 The main disadvantage of this method, however, is the specificity. From preliminary experiments, which were performed at our laboratory, it was clear that the major metabolites of amoxycillin, 5R,6R- and 5S,6Ramoxycilloic acid (Figure 1B and C, respectively), were converted to the same fluorescent derivative as amoxycillin. Hence, no distinction between amoxycillin and the amoxycilloic acid metabolites could be made during chromatographic analysis. A liquid chromatography/electrospray ionization (ESI) mass spectrometric method was described by Tyczkowska et al. The method was able to detect several β-lactam antibiotics in bovine milk, including amoxycillin, down to 100 pg entering the mass spectrometer. However, no validation results were reported as the method was only used for confirmatory purposes.6 (4) So¨rensen, L. K.; Snor, L. K.; Elkaer, T.; Hansen, H. J. Chromatogr., B, 1999, 734, 307-318. (5) Rose, M. D.; Tarbin, J.; Farrington, W. H. H.; Shearer, G. Food Addit. Contam. 1997, 14 (2), 127-133.
1394
Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
From the above it is clear that there was need for a method for the determination of amoxycillin in animal tissues that combined a straightforward and less time-consuming sample preparation procedure, a high selectivityswith regard not only to endogenous and analogous compounds but also to metabolitess and a high sensitivity, allowing the quantitation of amoxycillin at levels that are as low as half the MRL (25 ng/g). The present paper describes the development and validation of a liquid chromatographic tandem mass spectrometric (LC/MS/ MS) method that takes into account the three mentioned objectives. First, a sample preparation procedure was developed, which was easy to perform and which yielded sufficiently clean extracts to allow for proper quantification of amoxycillin at the aimed LOQ of at least 25 ng/g and to allow for a reasonable HPLC column lifetime and overall LC/MS/MS system stability. Second, special care has been devoted to the choice of the mobile phase in order to enhance the retention and separation of amoxycillin and its major metabolites (5R,6R- and 5S,6R-amoxycilloic acid and amoxycillinpiperazine-2′,5′-dione, cf. Figure 1). Third, by using ESI-MS/ MS, no time-consuming derivatization step was needed and a high specificity and sensitivity was obtained. Finally, the method was fully validated by a set of parameters that is in compliance with the requirements as defined in the Rules Governing Medicinal Products in the European Community (linearity, trueness, and precision, LOD, LOQ, specificity and recovery).7 Moreover, the validated method was tested on incurred tissue samples, which were obtained during a residue depletion study, showing its applicability in real circumstances. To our knowledge, this is the first time that a fully validated LC/ESI-MS/MS method for the quantitative determination of amoxycillin and its major metabolites in tissue samples at levels that are as low as half the MRL (25 ng/g) has been reported. EXPERIMENTAL SECTION Standards and Chemicals. Sodium amoxycillin was a Chemical Reference Substance (CRS) of the European Pharmacopoeia (Strasbourg, France). Ampicillin, which was used as the internal standard (IS), was obtained from Sigma Aldrich Chemie (Steinheim, Germany). The standards of 5R,6R- and 5S,6R-amoxycilloic acid and of amoxycillinpiperazine-2′,5′-dione were a gift from SmithKline Beecham (Worthing, United Kingdom). Stock solutions of 1 mg/mL of each compound in water were prepared and stored at e-75 °C for 6 months. By diluting each stock solution with water, appropriate working solutions were obtained (concentration of amoxycillin and metabolites 10, 5, 2, 1, and 0.5 µg/mL; ampicillin concentration 10 µg/mL). The working solutions were stored at e-75 °C and were replaced every month. To prevent deterioration of the analytes of interest (due to repeated freeze-thaw cycles), the stock and working solutions were divided into small portions of (1 mL and stored at e-75 °C. Each analysis day, fresh portions of the appropriate stock and working solutions were thawed and used for the preparation of calibration curve or quality control samples. After use, the stock and working solutions were discarded. All products and solvents used for the extraction were of analytical grade and were obtained from Fluka (pentafluoropro(6) Tyczkowska, K. L.; Voyksner, R. D.; Straub, R. F.; Aronson, A. L. J. AOAC Int. 1994, 77 (5), 1122-1131.
pionic acid (PFPA); Buchs, Switzerland), Merck (potassium dihydrogen phosphate or KH2PO4, trichloroacetic acid (TCA), and concentrated HCl; Darmstadt, Germany) and Panreac (methanol; Barcelona, Spain). All solvents used for the mobile phase (acetonitrile and water) were of HPLC grade and were obtained from Acros (Geel, Belgium). Bond Elut LRC octadecyl (C18) solid-phase extraction columns (500 mg/10 mL) were purchased from Varian. Biological Samples. Known β-lactam-free tissue samples (muscle, kidney, liver, fat) were obtained from pigs that did not receive any medication. Incurred tissue samples were obtained from pigs that received amoxycillin via the drinking water. The pigs were slaughtered at different time points after cessation of medication. After slaughtering, the tissue samples were minced and homogenized using a Moulinette mixer (Moulinex, Paris, France). The samples were transferred into plastic bags and stored at e-75 °C until analysis. The animal experiments were performed according to the Rules Governing Medicinal Products in the European Community, Vol. VI.7 Tissue Extraction. One gram each of kidney, liver, muscle, and fat tissue was weighed into a 50-mL polypropylene centrifuge tube, and 25 µL of the IS working solution was added. After vortex mixing for 15 s, 5 mL of an aqueous 0.01 M KH2PO4 solution were added and the samples were extracted for 10 min by rotation on a homemade apparatus. After centrifugation for 10 min at 3500 rpm, the supernatant was transferred into another centrifuge tube. The same amount of 0.01 M KH2PO4 solution was added to the remainder of the tissue samples, and the extraction procedure was repeated. The supernatants from the two extraction steps were combined, and 1 mL of a 20% TCA solution in water was added. The samples were vortex mixed, followed by a centrifugation step of 10 min at 3500 rpm. The supernatant layer was further purified by solid-phase extraction within 20 min after the addition of TCA to the samples. Solid-Phase Cleanup. A C18 column was installed on a vacuum manifold (Alltech, Laarne, Belgium) and preconditioned with, respectively, 5 mL of methanol, 5 mL of water, and 3 mL of a 2% TCA solution. Thereafter, the aqueous tissue extract was allowed to pass slowly through the C18 column. The column was washed with 2 mL of a 2% TCA solution and with 2 mL of water and dried for 10 min. The analytes were eluted using 2 mL of acetonitrile. The eluate was concentrated to dryness under a gentle stream of nitrogen at a temperature of ∼40 °C. The dry residue was redissolved in 300 µL of water, vortex mixed for 15 s, and transferred to an autosampler vial. A 25-µL aliquot was injected onto the HPLC. Chromatography. The HPLC system consisted of an Alliance type 2690 separations and a column heater module, both from Waters (Milford, MA). Chromatographic separation was achieved using a reversed-phase Hypersil column (100 × 3 mm i.d.; dp 5 µm) in combination with a guard column of the same type (10 × (7) Notice to applicants for the establishment of maximum residue limits (MRLs) for residues of veterinary medicinal products in foodstuffs of animal origin by the European Community in accordance with Council Regulation (EEC) 2377/90, Vol. VI, pp 102-111. Commission of the European Communities, Brussels-Luxembourg, 1991.
Table 1. Individual MS/MS Settings for Amoxycillin, 5R,6R- and 5S,6R-Amoxycilloic Acid, Amoxycillinpiperazine-2′,5′-dione, and Ampicillin
compound
MRM transition (m/z)
collision energy (eV)
amoxycillin 5R,6R- and 5S,6R-amoxycilloic acid amoxycillinpiperazine-2′,5′-dione ampicillin
365.8 > 207.9 383.8 > 189.0 365.8 > 160.0 349.9 > 106.0
12 20 18 17
2 mm i.d.) from Chrompack (Middelburg, The Netherlands). The mobile-phase solvent A was a solution of 9.6 mM PFPA in water, while the mobile-phase solvent B consisted of equal parts of acetonitrile and water (50/50, v/v) with the same PFPA concentration. The mobile phase was delivered to the HPLC column at a flow rate of 0.2 mL/min. A gradient elution was performed to elute the analytes of interest and some late-eluting endogenous compounds from the column (0-5 min, 60% A, 40% B; 6-12 min, 100% B; 13-20 min, 60% A, 40% B). Mass Spectrometry. The HPLC column effluent was pumped to a Quattro Ultima triple quadrupole mass spectrometer (Micromass, Manchester, United Kingdom), equipped with an ESI Z-spray source, which was used in the positive ion MS/MS mode. The instrument was calibrated with a solution of sodium iodide according to the manufacturers’ instructions. Thereafter, a tuning was performed for each analyte of interest by direct infusion of a 1-µg/mL solution at a flow rate of 9 µL/min using a Hamilton syringe (200 µL, Bonaduz, Switserland) and a Harvard pump 11 apparatus. The following tune parameters were used for all analytes: capillary, 3.5 kV; cone, 30 V; source temperature, 120 °C; desolvation temperature, 250 °C; cone gas flow, (100 L/h; desolvation gas flow, (700 L/h; resolution (LM1, HM1, LM2, HM2), 14.0; ion energy 1, 2.0; ion energy 2, 1.0; entrance, -2; exit, 2; multiplier, 650 V; Pirani pressure, (2.3 × 10-3 mbar; dwell time, 0.5 s. The optimal settings for collision energy, corresponding to a (nearly) 100% fragmentation of the molecular ions (or precursor ions), were different for each analyte and are presented in Table 1. For our quantitative purposes, the instrument was operated in the multiple reaction monitoring (MRM) mode, using for each analyte one precursor ion > product ion transition as shown in Table 1. Quantification was done with the Masslynx software (Micromass) using the above-mentioned MRM transitions. Method Validation. The proposed method for the quantitative determination of amoxycillin and its major metabolites in animal tissues was validated by a set of parameters that is in compliance with the recommendations as defined by the European Community and with our own criteria which were based on the literature. 7-14 (8) Commision Decision 93/256 laying down the methods for detecting residues of substances having a hormonal or thyrostatic action. Off. J. Eur. Commun. 1993, L118. (9) Heitzman, R. J., Ed. Veterinary Drug Residues, Report Eur. 14126-EN, Commission of the EC, Brussels-Luxembourg, 1994. (10) Knecht, J.; Storck, G. Z. Anal. Chem. 1974, 270, 97-99. (11) Mehta, A. C. J. Clin. Pharm. Ther. 1989, 14, 465-473 (12) Van Bavel, M.; Boot, A.; Bravenboer, J.; De Goey, F.; Maas, C.; Van Der Putten, A.; Verwaal, W. De Ware(n) Chemicus 1996, 26, 1-16.
Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
1395
Table 2. Extraction Recovery for Amoxycillin, 5R,6Rand 5S,6R-Amoxycilloic Acid, and Amoxycillinpiperazine-2′,5′-dione in the Various Tissue Matrixes concentration (%, n ) 3) compound
kidney
liver
muscle
fat
amoxycillin 5R,6R-amoxycilloic acid 5S,6R-amoxycilloic acid amoxycillinpiperazine-2′,5′-dione
69.3 72.9 58.5 91.5
39.2 42.8 67.5 89.0
82.9 75.6 99.3 104.0
43.9 80.9 98.6 103.2
The linearity of the method was evaluated by analyzing calibration curve samples, which were prepared by spiking blank tissues from pigs that did not receive any medication, with amoxycillin or its metabolites. The addition of 50 µL of the abovementioned standard working solutions resulted in analyte concentrations of 25, 50, 100, 250, and 500 ng/g. The standard working solutions were directly applied onto the homogenized tissue samples using a calibrated micropipet (Socorex Calibra 10100 µL, Lausanne, Switserland), followed by a vortex-mixing (30 s) step. After 5 min of equilibration, the extraction procedure was started. The calibration curve samples were treated in a way similar to the unknown samples. Peak area ratios between the analytes of interest and the IS were plotted against the concentration ratios. The correlation coefficients (r) and the goodness-offit coefficients (g) were determined. Within-day precision was evaluated by analyzing at least six blank tissue samples, which were spiked with amoxycillin at three different concentrations (MRL, double the MRL, and half the MRL) on the same day. In the case of the amoxycillin metabolites, the within-day precision was evaluated at only one concentration level, i.e., at 25 ng/g. The limit of quantification (LOQ) was defined as the lowest concentration of amoxycillin or its metabolites for which the method was validated with a trueness and precision that fall within the recommended ranges (trueness: -20 to +10%, precision, relative standard deviation (RSD) < RSDmax with RSDmax ) 2(1-0.5logConc) × 2/3). The LOQ was also established as the lowest point of the calibration graph. The LOQ was determined by analyzing at least six blank tissue samples, which were spiked with amoxycillin or its metabolites at a concentration of 25 ng/g. The LOD was defined as the lowest concentration of amoxycillin and its metabolites that could be recognized by the detector with a signal-to noise ratio of g3. The LOD values were calculated, using spiked tissue samples (25 ng/g). The selectivity of the method was demonstrated by analyzing blank tissues from pigs that did not receive any medication (interference of endogenous compounds) and by injecting standard solutions of several β-lactam antibiotics (dicloxacillin, ox(13) Draft Commission Decision laying down performance criteria for the analytical methods to be used for detecting certain substances and residues thereof in live animal and animal products according to Council Directive 96/23/EC repealing Commission Decision 90/515/EC, 93/256/EC, and 96/ 257/EC, European Commission, Directorate General for Public Health and Consumers Protection, SANCO/1805/2000, 18.07, 2000. (14) Draft Vol. 8, Notice to applicants, Veterinary medicinal products: Establishment of maximum residue limits (MRLs) for residues of veterinary products in foodstuffs of animal origin: Development and validation of a proposed regulatory method, EMEA/CVMP/573/00, 2000.
1396 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
acillin, penicillin procaine, phenoxymethylpenicillin) onto the HPLC column at a concentration of 1 µg/mL (interference of analogous compounds). The chromatographic conditions were the same as those used for the analysis of amoxycillin and its metabolites. The recovery was tested by analyzing blank tissue samples that were spiked with only amoxycillin or its metabolites at a concentration level of 250 ng/g. Prior to LC/MS/MS analysis, the IS was added to the sample extracts. After LC/MS/MS analysis, the peak area ratios between the analytes of interest and the internal standard were calculated and were compared with those in a blank sample extract, to which the analytes of interest and the IS were added just prior the LC/MS/MS analysis. RESULTS AND DISCUSSION Isolation of the Compounds. During preliminary experiments, several extraction procedures already described in the literature1,4,5 were tried. These experiments showed that the method of Sørensen et al.4 was very time-consuming. Two pH adjustment steps were needed, and the final volume of the tissue extract after the liquid extraction step was 100 mL. Hence, it took nearly 1 h to apply the tissue extract on the Oasis HLB solidphase extraction column. Experiments using cation exchange columns (SCX, PRS, HCX) were not successful (extraction recovery for amoxycillin of 160.0). In addition, it was observed that amoxycillinpiperazine-2′,5′-dione does not show a peak in the MRM trace of amoxycillin. This can be attributed to the lack of presence of a product ion at m/z ) 207.9 in the amoxycillinpiperazine-2′,5′-dione spectrum. By adding a volatile ion-pairing reagent, 9.6 mM PFPA, to the mobile phase, the problem of separation between amoxycillin and Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
1397
Figure 3. LC/ESI-MS/MS chromatogram of a mixture of standards of amoxycillin (A), 5R,6R- and 5S,6R-amoxycilloic acid (B, C), amoxycillinpiperazine-2′,5′-dione (D), and ampicillin (E).
its metabolite could be solved as shown in Figure 3. However, in real incurred samples and in the function over the column lifetime, the separation between amoxycillin and its piperazine-2′,5′-dione metabolite decreases, resulting at the end in coelution. To prevent long column equilibration times after gradient elution, the ion-pairing reagent was added to both the aqueous and organic parts of the mobile phase (solvents A and B), resulting in a constant PFPA concentration during a chromatographic run. Mass Spectrometry. Figure 2 shows the full-scan MS/MS spectra of amoxycillin, its metabolites, and ampicillin after direct infusion of 1 µg/mL standard solutions of each compound in the ESI source, using the tune parameters as mentioned above and the collision energy as shown in Table 1. The positive ion mode was selected on the basis of the presence of the free NH2 function in the chemical structures. The negative ion mode was also tried out, but it proved to be less sensitive for all compounds, which has also been reported by Straub and Voyksner.15 The ions with an m/z value of 365.8 (Figure 2A), 384.3 (Figure 2B), and 350.0 (Figure 2D) correspond to the molecular ions of amoxycillin, amoxycilloic acid, and ampicillin, respectively. The molecular ion of amoxycillinpiperazine-2′,5′-dione is the same as for amoxycillin (m/z ) 365.8), but it has nearly completely disappeared under the MS/MS conditions used (Figure 2C). The product ion at m/z ) 160 could be determined in all spectra and corresponds to a cleavage product that is common (15) Straub, R. F.; Voyksner, R. D. J. Chromatogr. 1993, 647, 167-181.
1398 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
for all β-lactam antibiotics ([Fz + H]+, Figure 4. It is formed by opening of the β-lactam ring.6,15 The ion at m/z ) 114 ([Fz COOH]+) is formed after a further loss of [COOH] and was clearly present in the MS/MS spectra of amoxycillin and amoxycillinpiperzine-2′,5′-dione (Figure 2A and C). Other ions corresponding to the common part of all β-lactam antibiotics are found at m/z ) 174 (ampicillin, [Fy + H - CO]+, Figure 4D) and at m/z ) 189 (amoxycilloic acid metabolite, [Fx - COOH]+, Figure 4B). Fragments formed by the cleavage of the amide moiety are generally more specific for the different β-lactam compounds.15 Ampicillin showed an ion at m/z ) 192 ([F1 - H]+) and m/z ) 106 ([F2]+, Figure 2D and 4D, respectively). As the latter ion was the base peak of the MS/MS spectrum, it was selected for the MRM transition which was used for quantitative purposes (Table 1). In the MS/MS spectrum of amoxycillin, two relevant ions were observed at m/z ) 349 and 208 (Figure 2A), corresponding to the [M + H - NH3]+ and the [F1 + H]+ fragment, respectively (Figure 4A). Although the ion at m/z ) 349 (resulting from the loss of NH3) was the base peak of the mass spectrum, it was not selected for quantitation. The ion at m/z ) 208, which corresponds to the amide part of the molecule, was estimated to be more specific and was therefore used for quantitation. For the amoxycilloic acid and amoxycillinpiperazine-2′,5′-dione metabolites, similar fragmentation patterns were observed (Figure 2B and C, Figure 4B and C). In both cases, the base peak ions were used for quantitative purposes.
Figure 4. Suggested fragmentation pattern of the molecular ion of amoxycillin (A), amoxycilloic acid (B), amoxycillinpiperazine-2′,5′-dione (C), and ampcillin (D).
Method Validation. Linearity. The linearity was evaluated for amoxycillin and its major metabolites in all tissue matrixes (kidney, liver, muscle, fat). The calibration curves were linear over the range tested (25-500 ng/g for all compounds in all tissues). Each time, the correlation coefficients (r) and the goodness-of-fit coefficients (g)10 were determined. The results fell within the ranges specified (r g 0.99, g e 10%) and are shown in Table 3. For all calibration curves, the y-intercepts were virtually zero, indicating the absence of endogenous interferences. The regression slopes, however, showed a remarkable between-day variation. Hence, it was decided to construct a new calibration curve each analysis day. Precision and Trueness. Within-day precision and trueness were thoroughly investigated for the analysis of amoxycillin in all tissue matrixes at three different concentrations: the MRL (50 ng/g), half the MRL (25 ng/g), and double the MRL (100 ng/g). The results are presented in Table 3. The relative standard deviations (RSD) ranged from 5.4 to 13.7% (25 ng/g), from 3.9 to 7.4% (50 ng/g), and from 4.0 to 12.7% (100 ng/g) and were below the maximum RSD values of 18.6, 16.7, and 15.1%, respectively.9 The between-day precision was evaluated using spiked QC samples (concentration level 50 ng/g) which were analyzed together with each batch of samples. The results are not shown in detail, but the criteria as mentioned above were also met.
The trueness fell in all cases within the accepted range of -20 to +10% of the target (or spiked) values.9 For the analysis of the major metabolites of amoxycillin, a minor validation was performed at a concentration of 25 ng/g of each compound in all tissues. The RSD values ranged from 6.0 to 9.7% and from 6.4 to 13.7% for the amoxycilloic acid and piperazine2′,5′-dione metabolites, respectively. Values for trueness fell between -10.3 and +10.4% and between -4.2 and +9.5%, respectively. Hence, it can be concluded that the criteria for both the precision (i.e., RSD < RSDmax) and trueness (-20 to +10% of target value) are met for the analysis of the metabolites of amoxycillin. Limit of Quantification. The LOQ for the determination of amoxycillin and its metabolites was established at a value of 25 ng/g in all tissues. As can be seen from Table 3, both the precision and trueness fell within the recommended ranges.13 The LOQ was also established as the lowest point of the calibration graphs. Limit of Detection. Inspection of the peaks of target analytes in the chromatograms of the lowest calibration samples (concentration level of 25 ng/g) revealed the above-mentioned S/N ratios. From these results, the LOD values were calculated and the results are shown in Table 3. The LOD values for amoxycillin were lower than those reported by Sørensen et al.4 Ang et al.1 reported LOD values of 0.5-0.8 ng/g in muscle tissue, which were Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
1399
Figure 5. LC/ESI-MS/MS selected ion chromatograms of a blank kidney sample (A) and of two incurred kidney samples that were taken at 12 (B) and 36 h (C) after cessation of medication. Trace: (1) amoxycillin, (2) 5R,6R- and 5S,6R-amoxycilloic acid, (3) amoxycillinpiperazine2′,5′-dione, and (4) ampicillin.
comparable to those described in the present method, taking into account the difference in the initial amount of sample to be analyzed (i.e., 5 g vs 1 g). 1400 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
Specificity. The described method proved to be specific for amoxycillin and its investigated metabolites with respect to the interference of endogenous compounds in all tissues (cf. Figure
5A). The method can also be considered specific with regard to other β-lactam antibiotics, such as dicloxacillin, oxacillin, penicillin procaine, and phenoxymethylpenicillin, since these compounds were not detected under the specified chromatographic and mass spectrometric conditions. Recovery. The results of the recovery experiments are shown in Table 2. For amoxycillin, the recovery ranged from 39.2 to 82.9%. This wide range in the recovery can be explained by matrix effects. The results for muscle tissue (82.9%) correspond with those found by other authors (cf. So¨rensen et al., ∼75%, and Ang et al., ∼75-80%). For the analysis of the amoxycillin metabolites, the highest recoveries were observed for amoxycillinpiperazine2′,5′-dione (89.0-104.0%). Analysis of Biological Samples. To evaluate the applicability of the method, incurred tissue samples (kidney, liver, muscle, fat) from pigs, which were slaughtered at different time points after cessation of medication (12, 36, 60 and 108 h), were analyzed. The initial aim of the study was to determine the withdrawal time for amoxycillin, but the quantitative determination of the amoxycilloic acid and piperazine-2′,5′-dione metabolites was also a point of interest. As is shown in Figure 5, the method presented gave the opportunity to quantitate amoxycillin and its metabolites in incurred samples by performing one single sample preparation step and chromatographic run. However, within one analytical batch of samples, the 5R,6R- and 5S,6R-amoxycilloic acid metabolites were only partially resolved in some samples (cf. Figure 5C). This can possibly be attributed to matrix effects. For the acceptance of a sample within an analytical batch, the following criterion was handled at our laboratory. The relative retention time (Tr) of the analytes of interest (i.e., Tr analyte/Tr IS) in the incurred samples must be within a (3% range of those in the standard samples, analyzed in the same analytical batch. As the relative retention times for the 5R,6R- and 5S,6R-amoxycilloic acid metabolites in a standard mixture were 0.397 (0.385-0.409) and 0.429 (0.416-0.442), respectively, the relative retention times of both metabolites in the incurred sample shown in Figure 5C (i.e., 0.409 and 0.421, respectively) fell within the acceptance criterion. In addition, some interesting results were seen during sample analysis: the amoxycillin concentrations were >10 times above the MRL in kidney and already around or below the MRL in all other tissues at 12 h after cessation of medication. At 36 h, no amoxycillin was detected in almost all tissue samples. The amoxycilloic acid metabolites, however, remained much longer in kidney and liver tissues (cf. Figure 5B and C) at concentrations that were much higher than 50 ng/g, the MRL for amoxycillin. In muscle and fat tissues, the presence of these metabolites was negligible. The amoxycillinpiperazine-2′,5′-dione metabolite showed to be of minor importance and had nearly disappeared in all tissues within 36 h (concentration 5 times the concentration of amoxycillin in kidney samples at the 12-h point and >10 times the MRL of amoxycillin in liver samples at the same time point), has led to some important questions regarding the risk assessment.16 In addition, the usefulness of determining only amoxycillin as the marker residue for MRL can be questioned. With regard to the above assumptions, it must be clear that further investigation is certainly necessary. However, the above results demonstrate the extreme value of the developed LC/ESI-MS/MS method for application in the field of residue analysis of β-lactam antibiotics.
CONCLUSIONS We succeeded in the development of a highly sensitive and specific LC/ESI-MS/MS method for the quantitative determination of amoxycillin and its major metabolites (5R,6R- and 5S,6Ramoxycilloic acid, amoxycillinpiperazine-2′,5′-dione) in complex matrixes such as animal tissues (kidney, liver, muscle, fat). The method was completely validated for amoxycillin in all tissues according to EU regulations (linearity, precision, trueness, LOQ, LOD, specificity), and good results were obtained. For the amoxycillin metabolites, a shortened validation procedure was performed. Incurred tissue samples from pigs that were slaughtered at different time points after cessation of medication were quantitatively analyzed using the described method. The obtained results did prove the usefulness of the method for the application in the field of residue analysis. In addition, some important questions were raised regarding the allergenic potential of large amounts of amoxycilloic acid metabolites in animal food products (kidney, liver) and the relevance of amoxycillin as sole marker residue for MRL. Further investigation regarding these topics is necessary and will be performed at our laboratory in the future. We can say that this issto our knowledgesthe first time that a fully validated LC/ESI-MS/MS method for the quantitative determination of amoxycillin and its major metabolites in animal tissue samples, at levels that are as low as half the MRL of the marker amoxycillin (25 ng/g), has been reported.
Received for review August 14, 2001. Accepted November 29, 2001. AC010918O (16) Dewdney, J. M.; Maes, L.; Raynaud, J. P.; Blanc, F.; Scheid, J. P.; Jackson, T.; Lens, S.; Verschueren, C. Food Chem. Toxicol. 1991, 29 (7), 477-483.
Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
1401
