3286
J. Agric. Food Chem. 2004, 52, 3286−3291
Simple and Rapid Liquid Chromatography−Tandem Mass Spectrometry Confirmatory Assay for Determining Amoxicillin and Ampicillin in Bovine Tissues and Milk SARA BOGIALLI, VITTORIO CAPITOLINO, ROBERTA CURINI, ANTONIO DI CORCIA,* MANUELA NAZZARI, AND MANUEL SERGI Dipartimento di Chimica, Universita` “La Sapienza”, Piazza Aldo Moro 5, 00185 Roma, Italy
A simple specific and rapid confirmatory method for determining the two amphoteric penicillins, that is, amoxicillin and ampicillin, in bovine muscle, liver, kidney, and milk is presented. This method is based on the matrix solid-phase dispersion technique with hot water as extractant followed by liquid chromatography (LC)-tandem mass spectrometry. With this instrumentation, the selected reaction monitoring acquisition mode with two fragmentation reactions for each analyte was adopted. After acidification and filtration of the aqueous extracts, 25 µL of the tissue final extracts and 50 µL of the milk final extract were injected into the LC apparatus. Absolute recovery of the two analytes in any biological matrix at the 50 ppb level in tissues and the 4 ppb level in milk was 74-95% with relative standard deviations (RSDs) of no larger than 9%. When penicillin V was used as surrogate internal standard, relative recovery of the targeted compounds present in bovine tissues and milk at, respectively, 25 and 2 ppb levels ranged between 100 and 106% with RSDs of no larger than 11%. When fractionation of analytes by using a short chromatographic run was attempted, remarkable signal weakening for the two analytes was experienced. This effect was traced to polar endogenous coextractives eluted in the first part of the chromatographic run that interfered with the gas-phase ion formation of the two penicillins. Slowing the chromatographic run eliminated this unwelcome effect. Limits of quantification of the two analytes in bovine milk were estimated to be 65 °C was attempted, a steady decrease of the extraction yield was experienced. This loss was traced to hydrolytic attack of the unstable four-term ring of penicillins. Thus, an extraction temperature of 65 °C was used for subsequent experiments. Effect of Extractant Volume on Analyte Recoveries. For the reason mentioned above, recovery studies were conducted to find the minimum water volume capable of efficiently extracting the analytes and the surrogate internal standard from all of the biological matrices considered. Experiments were performed by spiking milk and muscle tissue samples with the analytes at 100 ppb level and the surrogate internal standard at the 150 ppb level, and extracting. A 2-mL aliquot followed by two 1-mL aliquots of the aqueous eluent coming out from the extraction cell were separately collected and analyzed. For each matrix, experiments were performed in triplicate. For the sake of conciseness, only results relative to milk and muscle are shown in Figure 3. As can be seen, the ability of water to remove penicillins from biological matrices was matrix-dependent, as different volumes of the extractant needed to have satisfactory recovery of the analytes and the surrogate internal standard. One could speculate that this effect is due to different interactions between the analytes and some nonextracted matrix components. Anyway, 3 mL of heated water sufficed to extract
Figure 3. Elution histogram obtained on extraction of three penicillins
from (A) milk and (B) bovine muscle by passing through the extraction cell water heated at 65 °C and collecting separately the first 2 mL and then two 1-mL aliquots of the extracts. Table 2. Absolute Recoveries of Amoxicillin and Ampicillin by
Extraction (n ) 5) from Various Biological Matrices (Spike Levels ) 50 ppb in Tissues and 4 ppb in Milk) recovery, % (RSD, %) compound
muscle
kidney
liver
milk
amoxicillin ampicillin
93 (6) 92 (7)
88 (9) 89 (8)
74 (7) 74 (7)
94 (8) 95 (6)
the analytes from any matrix considered, and they were used in the subsequent experiments. Effect of Extractant Flow Rate on Analyte Recoveries. We evaluated the influence of the flow rate at which water passed through the extraction cell on the extraction efficiency. For this experiment, a bovine muscle sample spiked with the analytes at 100 ppb level and with penicillin V at 150 ppb was submitted to the extraction procedure by passing water through the cell at flow rates ranging between 0.5 and 2 mL/min. Results (not shown here) from duplicate experiments at each flow rate selected evidenced that the analyte extraction yield was substantially not dependent on the extractant flow rate. We chose to extract penicillins at a flow rate of 1 mL/min because at a 2 mL/min flow rate the extraction cell sometimes clogged, especially when analytes were extracted from milk samples. Extraction Efficiency. Under conditions reported within Experimental Procedures, we evaluated the ability of water to extract amoxicillin and ampicillin from the various matrices considered. This study was conducted by spiking tissue and milk samples with the analytes at the tolerance levels set by the EU, that is 50 and 4 ppb, respectively. Absolute recoveries were estimated by adding penicillin V after extraction and before extract filtration. For each matrix, five experiments were performed, and results are shown in Table 2. The extraction
3290
J. Agric. Food Chem., Vol. 52, No. 11, 2004
Bogialli et al.
Table 3. Accuracy and Precision Data from Analysis of Penicillins in Two Selected Matrices at Concentrations Equal or Close to Maximum Residue
Limits (MRLs) Set by the European Union (EU) accuracy, % (RSD, %) muscle
kidney
liver
milk
compound
MRL/2a
MRL
2 MRL
MRL/2
MRL
2 MRL
MRL/2
MRL
2 MRL
MRL/2
MRL
2 MRL
amoxicillin ampicillin
103 (5) 102 (6)
107 (8) 103 (10)
110 (11) 106 (7)
101 (6) 100 (4)
104 (5) 102 (5)
103 (7) 105 (8)
100 (9) 105 (6)
105 (8) 108 (9)
103 (9) 105 (4)
106 (7) 104 (11)
105 (3) 106 (6)
108 (9) 109 (7)
a
Tolerance levels of amoxicillin and ampicillin set by the EU are 50 ppb in bovine tissues and 4 ppb in milk.
Table 4. Limits of Detection (LOD) and Quantification (LOQ) of the
Method for Determining Penicillins in Milk and Tissues milk
tissues
compound
LOD,a ppb
LOQ, ppb
LOD, ppb
LOQ, ppb
amoxicillin ampicillin
0.5 (366 f 114) 0.1 (350 f 192)
0.8 0.2
2.1 (366 f 114) 0.5 (350 f 192)
3.1 0.8
a
m/z values of the transitions giving the worst S/N ratios are reported in parentheses.
Figure 4. SRM LC-MS/MS trace resulting from analysis of amoxicillin
and ampicillin at 4 ppb level in bovine milk.
yields varied from 74 to 95% with relative standard deviations (RSDs) not higher than 9%. Accuracy and Precision. For amoxicillin and ampicillin, the EU has set maximum residue limits (MRLs) of 50 ppb in edible animal tissues and 4 ppb in bovine whole milk. Following criteria reported in the EU guidelines (12), this method was validated at three different concentrations corresponding to half of the MRL, the MRL, and 2 times the MRL. At any analyte concentration and for any matrix considered, five measurements were performed with the criterion of adding the surrogate internal standard (penicillin V) before analyte extraction. Results are presented in Table 3. To check that the accuracy of the method was not dependent on the analyte concentration and/or the type of matrix, mean accuracy data for each matrix were compared among them by using the two-way analysis of variance (ANOVA) test at the P ) 0.05 significance level. In both cases, the calculated F2,6 and F3,6 values (4.18 and 4.54, respectively) were lower than the critical values (5.14 and 4.76), indicating that the extraction method was not influenced by either the concentration of the analyte or the nature of the matrix. The accuracy data varied between 100 and 110% with RSDs not higher than 11%. Thus, this method meets requirements reported in the EU guidelines (12) indicating that a method can be considered to be accurate and precise when accuracy data are between 80 and 110% with RSDs not higher than 20%. Limits of Detection (LOD) and Quantification (LOQ) of the Method. LOQs of the method were estimated from the SRM LC-MS/MS chromatograms resulting from analyses of 4 and 25 ppb of the two penicillins in, respectively, milk (please, note that the resulting chromatographic trace is shown in Figure 4)
and tissue samples. After extraction of the sum of the ion currents of the two molecular ion-to-daughter ion transitions relative to each analyte, the resulting trace was two-times smoothed by applying the mean smoothing method. Thereafter, the peak height-to-averaged background noise ratio was measured. The background noise estimate was based on the peakto-peak baseline near the analyte peak. LOQs were then calculated on the basis of a minimal accepted value of the signalto-noise ratio (S/N) of 10. These data are listed in Table 4. In the same table, LODs of the method are also presented. When using an MS/MS detector in the SRM mode, the most important condition to be satisfied for ascertaining the presence of a targeted compound is that at least two signals produced by the decomposition reactions are distinguishable from the background ion current. Accordingly, a definition of LOD (S/N 3) of each analyte was adopted considering the transition giving the worst S/N. As can be read, amoxicillin and ampicillin could be detected in milk at levels of
