SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, 1999

Transcription

1 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, DRUGS, COSMETICS, FORENSIC SCIENCES Determination of Amoxicillin in Trout by Liquid Chromatography with UV Detection after Derivatization SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, 1999 LAMBERT K. SØRENSEN, HELGA HANSEN, and LENA SNOR Steins Laboratorium, Ladelundvej 85, 6650 Brørup, Denmark A liquid chromatographic (LC) method based on solid-phase extraction was developed for determination of amoxicillin in muscle tissue of rainbow trout. The compound was extracted in an aqueous solution by precipitation of organic material with a mixture of sulfuric acid and sodium tungstate. The extract was processed by solid-phase extraction on an end-capped phenyl sorbent, and concentrated on a divinylbenzene-co-n-vinylpyrrolidone polymeric sorbent. The extract was derivatized and analyzed by reversed-phase gradient LC on a C 18 column with UV detection at 323 nm. The method detection limit was 2.9 µg/kg. Mean recovery in muscle was 80.5% (range µg/kg). The method was applied to fillets from trout offered feed containing amoxicillin in an aquaculture pilot plant. Amoxicillin was detected in muscle tissue shortly after administration but not 3 weeks later. The relative repeatability standard deviation for incurred residues in muscle tissue was 6.4% (range µg/kg). Amoxicillin is widely used to treat bacterial infections in mammals and for protection of cultured fish against bacterial infections. However, amoxicillin residues in food can cause allergic reactions in humans and may increase the development of resistant bacterial strains responsible for human infections. Few analytical methods have been reported for determination of amoxicillin in animal tissue. A liquid chromatographic (LC) method reported for determination of amoxicillin in bovine muscle tissue based on on-line dialysis and solid-phase extraction (SPE) is relatively insensitive, with a detection limit of 200 µg/kg (1). A sensitive LC method with fluorescence detection using C 18 SPE and repeated diethyl ether liquid liquid extraction has been reported for determination of amoxicillin in catfish and salmon (2). A practical drawback of this method is the extensive use of diethyl ether, requiring special apparatus and laboratory facilities. An LC method described for determination of amoxicillin in animal muscle and liver using cation exchange and porous graphitic Received March 9, Accepted by JM June 23, carbon SPE for cleanup showed relatively low mean recoveries ranging from 37 to 58% at a level of µg/kg (3). A sensitive analytical method was developed for the determination of amoxicillin residues in rainbow trout (Oncorhynchus mykiss) using standard analytical equipment and avoiding liquid liquid extraction with ozone-depleting agents or highly flammable solvents. METHOD Reagents (a) Stock solution of amoxicillin (1000 g/ml). Dissolve a quantity of amoxicillin (Cat. No. A-8523, Sigma Chemical Co., St. Louis, MO) corresponding to g anhydrous amoxicillin (C 16 H 19 N 3 O 5 S) in water and dilute to 100 ml. The solution is stable for at least 2 weeks when stored at 5 ±2 C (4). (b) Standard solution of amoxicillin (2 g/ml). Dilute ml stock solution of amoxicillin to 100 ml with water. Prepare fresh each day. (c) Solvents. Acetonitrile and methanol, chromatography grade. (d) Water. Purified through a Milli-Q Plus system (Millipore, Bedford, MA). (e) Sulfuric acid (0.7M). Dissolve 7.2 g H 2 SO 4 (95 97%) in water and dilute to 100 ml. (f) Sodium hydroxide solutions. Prepare 0.5 and 5M solutions by dissolving 2.0 and 20 g NaOH, respectively, in water, and dilute to 100 ml. (g) Sodium tungstate (0.7M). Dissolve 22.4 g Na 2 WO 4 2H 2 O in water and dilute to 100 ml. (h) Phosphate buffer (ph 9.0). Prepare 25 mm buffer by dissolving 0.34 g KH 2 PO 4 in ca 80 ml water; adjust ph to 9.00 ± 0.05 with sodium hydroxide solution. Dilute to 100 ml. (i) Phosphate buffer for LC mobile phases (ph 6.5). Prepare 0.10M phosphate buffer by dissolving 9.94 g Na 2 HPO 4, g NaH 2 PO 4 H 2 O, and 4.96 g Na 2 S 2 O 3 in water, and dilute to 2000 ml. (j) Benzoic anhydride solution (0.20M). Dissolve 1.13 g benzoic anhydride (C 14 H 10 O 3 ) in acetonitrile and dilute to 25 ml. Prepare fresh each day. (k) Mercury solution (0.026M). Dissolve 0.35 g HgCl 2 in 50 ml water. Stable for at least 1 month at C.

2 1346 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, 1999 Figure 1. ph profiles for adsorption of amoxicillin to SPE end-capped phenyl sorbent ( ) and SPE polymeric sorbent ( ). The profiles were obtained on sample extracts from muscle tissue spiked with 100 ng amoxicillin. Note: Mercuric chloride is highly toxic. Avoid contact. Consult Material Safety Data Sheets for its handling and disposal. (l) 1,2,4-Triazole reagent (2.0M). Dissolve 3.45 g 1,2,4- triazole (C 2 H 3 N 3 ) in 15 ml water. Add 2.5 ml mercury solution and adjust ph to 9.00 ± 0.05 with sodium hydroxide. Dilute to 25 ml with water and equilibrate for 1 h. Centrifuge for 10 min at 1500 g and use clear supernatant. Prepare fresh each day. (m) LC mobile phases. A, Dilute 100 ml acetonitrile to 1000 ml with phosphate buffer for LC. B, Mix 160 ml acetonitrile and 140 ml methanol and dilute to 1000 ml with phosphate buffer for LC. C, Mix 300 ml acetonitrile and 200 ml methanol and dilute to 1000 ml with phosphate buffer for LC. Apparatus (a) LC system. Waters pump gradient system 600, Waters 996 photodiode-array detector, Waters 717 autosampler (Milford, MA) and a circulator water bath (40 ±1 C) for temperature control of analytical column or equivalent. (b) Analytical column. Nova-Pak C 18, 4 µm, mm id (Waters Corp., Milford, MA). (c) Homogenizer. Ultra Turrax with 18 mm shaft tube (Ika, Staufen, Germany). (d) Centrifuge. Sigma centrifuge model 2-15 (Osterode, Germany). (e) Solvent evaporator. With nitrogen flow (Mikrolab Aarhus, Aarhus, Denmark). (f) SPE columns. Phenyl end-capped, 500 mg, 10 ml (Cat. No H, International Sorbent Technology, Hengoed, UK); divinylbenzene-co-n-vinylpyrrolidone polymeric sorbent, 60 mg, 3 ml (Oasis HLB Cat. No. WAT094226, Waters Corp.). (g) Vacuum manifold for SPE. Polypropylene luer fittings and needles. (h) Syringe filters. Acrodisc 13 PVDF, 13 mm 0.45 µm, disposable (Gelman Sciences, Ann Arbor, MI). (i) Glass fiber filters. Whatman GF/B, 47 mm (Maidstone, UK). (j) Containers. Polypropylene centrifuge tubes with screw cap, 15 and 50 ml (Sarstedt, Nümbrecht, Germany); polypropylene volumetric flasks, and LC vials. Sample Preparation Clean fresh trout and cut out fillets without skin. Pack fillets into plastic bags and seal hermetically. Store at maximum 18 C until analysis. Do not store samples more than 1 week before analysis to avoid significant loss of amoxicillin. Thaw fillets and homogenize in a food processor just before extraction. Extraction Weigh 2.5 g homogenized tissue in 50 ml polypropylene centrifuge tube. Mix sample with 20 ml water for 1 min, us-

3 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, Figure 2. Liquid chromatogram of a blank control muscle tissue. Figure 3. Liquid chromatogram of a blank control muscle tissue spiked to a level of 10 µg/kg with amoxicillin before extraction.

4 1348 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, 1999 Table 1. Relative repeatability standard deviation (RSD r ), relative intralaboratory reproducibility standard deviation (RSD R,intra ), and recovery of amoxicillin from spiked muscle tissue of rainbow trout a Rec., % Spiked level, µg/kg RSD r, % RSD R,intra, % Mean SD b a b Determination of amoxicillin concentration was conducted in duplicate at each spiked level. One double determination was performed at each level on each of 8 different days. Blank sample material used for spiking was different from day to day. Standard deviation of 8 mean results obtained by double determinations. ing Ultra Turrax homogenizer. Add 2 ml 0.7M sulfuric acid and 2.25 ml 0.7M sodium tungstate solution, and shake vigorously for 1 min. Check ph of mixture. If outside the range of , repeat extraction of new sample with adjusted volume of sodium tungstate. After sample with appropriate ph is prepared, centrifuge mixture for 10 min at 1500 g. Decant supernatant into 50 ml polypropylene centrifuge tube. Suspend pellet in 20 ml water and repeat extraction using 2.00 ml 0.7M sulfuric acid and 2.85 ml 0.7M sodium tungstate solution. Combine supernatants and filter through glass fiber filter. Solid-Phase Extraction Condition a phenyl SPE cartridge with 5 ml methanol followed by 10 ml water. Pull extract through cartridge with flow rate of 2 3 ml/min. Collect eluate in 50 ml polypropylene centrifuge tube. Adjust ph of eluate to with 0.5 5M sodium hydroxide. The sodium hydroxide solution Figure 4. One hour stability of amoxicillin in trout fillet extracts of different ph. Samples were extracted and cleaned up on phenyl sorbent as described in the method. The ph values of the eluates were then adjusted to various levels. After 1 h the ph was adjusted to and samples were cleaned up on polymeric sorbent and derivatized as described in the method.

5 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, Figure 5. Stability at 20 ±2 C of homogenized muscle tissue spiked to a level of 80 µg/kg (n = 2). must be added in drops to keep ph fluctuations below ph Condition polymeric SPE cartridge with 5 ml methanol followed by 5 ml water. Pull sample solution through polymeric cartridge with flow rate of 1 2 ml/min. Wash cartridge with 2 ml phosphate buffer ph 9.0 and dry by suction for 1 min. Tare a 15 ml polypropylene centrifuge tube. Add 150 µl phosphate buffer solution (ph 9.0) to tube. Elute polymeric cartridge with 1.5 ml acetonitrile, collecting eluate in 15 ml centrifuge tube. Evaporate eluate to ca 100 µl at 50 ±2 C under stream of nitrogen and dilute to mass of 500 mg with phosphate buffer, ph 9.0. Derivatization Add 75 µl benzoic anhydride solution to extract. Vortex-mix for 30 s and allow to react at C for 10 min. Add 450 µl 1,2,4-triazole reagent, cap tube, Vortex-mix for 60 s, and allow to react in water bath at 55 ±2 C for 30 min. Cool to ca 20 C in cold water and filter through 0.45 µm PVDF syringe filter. Standard Preparation Prepare calibration standards containing 100, 200, 300, 400, and 500 ng amoxicillin by transferring 50, 100, 150, 200, and 250 µl standard solution containing 2 µg/ml amoxicillin to 15 ml tared polypropylene centrifuge tubes. Dilute to mass of 500 mg with phosphate buffer solution (ph 9.0), and add 75 µl benzoic anhydride solution. Vortex-mix for 30 s and allow to react at C for 10 min. Add 450 µl 1,2,4-triazole reagent to each solution, cap tubes, and Vortex-mix for 60 s. Place solutions in water bath at 55 ±2 C for 30 min. Cool sample to ca 20 C in cold water and filter through 0.45 µm PVDF syringe filter. Perform derivatization of calibration standards and test samples in parallel. LC Analysis Inject 150 µl standard solution or sample extract into LC system running at column temperature 40 ±1 C, flow rate 1.0 ml/min, and 100% eluent A. Increase linearity to 100% eluent B for 20 min, and keep constant for 7 min. Then increase linearity to 100% eluent C for 1 min, and keep constant for 8 min. Return system to 100% eluent A for 1 min, and condition for 11 min before next injection. Perform UV detection at 323 nm, and collect data in time window min after injection. Calculations Determine content of amoxicillin (ng) in final sample extract from linear regression analysis with line forced through the origin. The concentration in tissue (µg/kg) is obtained via division by the sample weight (g).

6 1350 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, 1999 Figure 6. Liquid chromatogram of muscle tissue of trout fed with amoxicillin. The measured concentration was 143 µg/kg. Validation (a) Limits of detection and quantification. The limits of detection (LODs) for muscle tissue were determined on 20 rainbow trouts of different age and from different fish farms. To obtain a realistic LOD, samples were spiked with amoxicillin to a peak height on chromatograms corresponding to 3 times the baseline variation in a time window of 2 min, centered at the retention time on chromatograms from control tissue. Thus, samples were spiked with the compound to a level of 1.6 µg/kg before extraction. The LODs were determined as the average of 20 measurements plus 3 times the standard deviation (SD). The limits of quantification (LOQs) were calculated as the average plus 6 times the SD of 20 measurements. (b) Precision and recovery on spiked samples. The repeatability, intralaboratory reproducibility, and recovery of the method were determined on homogenized samples of blank control muscle spiked with amoxicillin to levels of 10, 30, 100, and 200 µg/kg. Volumes of 100 µl standard solution containing 0.25, 0.75, 2.50, and 5.00 mg/l of amoxicillin were added to 2.5 g samples in polypropylene centrifuge tubes. The spiked samples were stirred thoroughly with a glass spatula and equilibrated at 5 ±2 C for at least 1 h before extraction. Analyses at each level were performed in duplicate on each of 8 different days. The blank sample material was different from day to day. The 8 duplicate analyses at each level were divided equally between 2 trained analysts. Repeatability was calculated in accordance with ISO standard describing a basic method for determination of repeatability and reproducubility of standard measurement methods (5). The intralaboratory reproducibility was calculated by the same principle used to determine reproducibility. (c) Ruggedness. The ph dependence of adsorption of amoxicillin on the phenyl sorbent was tested on sample extracts spiked with 100 ng amoxicillin. After aspiration of the extract, the phenyl sorbent was eluted with 4 ml acetonitrile. The eluate was diluted with phosphate buffer, concentrated, and cleaned up by liquid liquid extraction with methylene chloride (4). The ph dependence of adsorption of amoxicillin on the polymeric sorbent was tested using sample eluates from phenyl sorbent, spiked and ph-adjusted prior to cleanup on polymeric sorbent. The elution profile of amoxicillin from polymeric sorbent was tested on trout muscle spiked to a level of 80 µg/kg. Volumes of 0.5 ml acetonitrile were used for elution. The sensitivity of amoxicillin to high ph values was tested on sample eluates from phenyl sorbent that were adjusted to different ph values from 8.5 to 12.0 prior to spiking with 200 ng amoxicillin. After 1 h, the ph was lowered to and the extract was cleaned up on the polymeric sorbent. The stability at 10 ±2 C of final sample extracts from trout spiked to a level of 40 µg/kg was tested over 12 days by measuring the peak area obtained by LC. Two different ex-

7 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, tracts were tested. The stability at 20 ± 2 C of homogenized muscle tissue spiked to a level of 80 µg/kg was tested over 78 days. Homogenates of 2 different fish were stored in 2.5 g portions in capped 50 ml polypropylene centrifuge tubes. (d) Test on trout treated with amoxicillin. Rainbow trout each weighing ca 100 g were kept in a pilot plant consisting of basins each containing ca 350 L water. Thirty trout were placed in each basin. The water temperature was 11.8 ± 0.5 C, ph was 7.51 ± 0.14, and oxygen content was >8.5 mg/l. Amoxicillin-containing feed pellets made for the experiment (BioMar, Brande, Denmark) were administered to the trout for 5 days. The feed was administered in the basins from 8:30 a.m. to 4 p.m. by automatic machines. The feed level was 0.5% of the biomass in the administration period and 1.4% before and after this period. The concentration of amoxicillin in the feed pellets was 20 g/kg, corresponding to a dose of 100 mg amoxicillin/kg body weight/day. Trout were sampled before and after the administration period, and then cleaned and filleted. Each fillet was cut into 2 subsamples which were stored separately in hermetically sealed plastic bags at 20 ±2 C. One of these subsamples was analyzed shortly after sampling. The other was analyzed after 10 months of storage. (e) Precision for incurred residues. Two trout fed with amoxicillin as described in (d) were homogenized separately. Ten subsamples of each homogenate were analyzed under repeatability conditions. (f) Selectivity. The specificity of the method was tested on trout muscles spiked with ampicillin, penicillin G, penicillin V, oxacillin, cloxacillin, nafcillin, and dicloxacillin each to a level of 800 µg/kg before extraction. Results and Discussion Adsorption of the amphoteric amoxicillin to reversed-phase sorbents can be strongly dependent on ph and sorbent material. This property was used in the development of a pure SPE-based sample cleanup technique for determination of amoxicillin in trout muscle. Under acid conditions (ph ) interfering compounds were retained in the raw extract on an end-capped phenyl sorbent without retaining amoxicillin (Figure 1). The analyte could, on the contrary, be retained completely on polymeric sorbent under basic conditions (ph ) which were used for further sample cleanup and concentration (Figure 1). More than 95% of the adsorbed amoxicillin was eluted from the polymeric sorbent with 0.5 ml acetonitrile. No amoxicillin was observed in the third elution volume of 0.5 ml acetonitrile. The sample extract was derivatized before LC. The NH 2 group was derivatized with benzoic anhydride to increase retention on C 18 columns, and the wavelength of detection was moved to the higher UV region by complexing amoxicillin with the 1,2,4-triazole mercuric reagent, using a previously described procedure (4). Typical chromatograms of muscle from blank control fish and equivalent spiked with amoxicillin to a level of 10 µg/kg are shown in Figures 2 and 3, respectively. From the analyses of 20 samples of trout muscle fortified with amoxicillin to 1.6 µg/kg, a mean of 1.65 µg/kg and a SD of 0.40 µg/kg were obtained. The calculated LOD and LOQ were 2.9 and 4.1 µg/kg, respectively. The results from the precision and recovery study on spiked muscle tissue are summarized in Table 1. The relative repeatability standard deviation (RSD r ) and the relative intralaboratory reproducibility standard deviation (RSD R,intra ) for single measurements were independent of concentration level in the range of µg/kg (Table 1). The mean RSD r and RSD R,intra were 3.71 and 5.37%, respectively. The mean recovery of 80.5% was independent of concentration level (Table 1). The calibration curves were linear with the coefficient of determination (R 2 ) exceeding The mean retention time (RT) of amoxicillin during the precision study was 25.9 min with an SD of 0.19 min. Amoxicillin in sample extracts is not stable under strongly basic conditions. At ph values above 10.5, degradation was observed after1hofstorage (Figure 4). Adjustment of ph to before cleanup on polymeric sorbent must therefore be performed slowly to avoid the ph exceeding In general, polypropylene materials were used in the handling of solutions containing amoxicillin to eliminate the risk of adsorption effects. Adsorption effects have been reported to nonsilanized glassware (4). However, the glass fiber filter used did not show adsorption effects. The selectivity of the method was tested on trout muscle spiked to a level of 800 µg/kg with penicillin G, penicillin V, ampicillin, oxacillin, cloxacillin, nafcillin, and dicloxacillin. The RTs of these penicillins in standard solution were min. All penicillin peaks were fully separated from the amoxicillin peak (RT 26.0 min). Only ampicillin was detected in sample extracts cleaned up by SPE (RT 35.0 min). The method was tested on trout that had been given feed pellets containing amoxicillin. Amoxicillin was detected in all muscle tissue samples h after the administration period (n = 10). The concentration range was µg/kg, with an average of 1770 µg/kg. The large variation in residue concentration may be related to stress factors which often occur in small scale studies, leading to an uncontrolled feed uptake between individuals. No residues could be detected before (n = 2) or 3 weeks after the administration period (n = 10). The content of amoxicillin in trout fillet tissue determined by analysis declined during storage at 20 ±2 C, indicating degradation of the compound. The intact fillets of trout given amoxicillin were reanalyzed after 10 months of storage at 20 ±2 C, and a mean content of 199 µg/kg was found, corresponding to 11% of the original level. Even higher losses were observed with homogenized samples of control fish spiked to a level of 80 µg/kg and stored at 20 ±2 C (Figure 5). The relative repeatability standard deviations for incurred residues were 6.1 and 6.7% at measured concentration levels of 11 and 143 µg/kg, respectively. A typical chromatogram is shown in Figure 6.

8 1352 SØRENSEN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 6, 1999 Final sample extracts of muscle tissue spiked to a level of 50 µg/kg were stable for at least 12 days when stored at 10 ±2 C. References (1) Snippe, N., van de Merbel, N.C., Ruiter, F.P.M., Steijger, O.M., Lingeman, H., & Brinkman, U.A.T. (1994) J. Chromatogr. B 662, (2) Ang, C.Y.W., Luo, W., Hansen, E.B., Jr, Freeman, J.P., & Thompson, H.C., Jr (1996) J. AOAC Int. 79, (3) Rose, M.D., Tarbin, J., Farrington W.H.H., & Shearer G. (1997) Food Addit. Contam. 14, (4) Sørensen, L.K., Rasmussen, B.M., Boison, J.O., & Keng, L. (1997) J. Chromatogr. B 694, (5) ISO Standard (1994) International Organization for Standardization (ISO) Geneva, Switzerland