Liquid Chromatographic Determination of Amoxicillin Concentrations in Bovine Plasma by Using a Tandem Solid-Phase Extraction Method H

Home , Ciclacillin

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Sept. 1992, p. 1859-1863 Vol. 36, No. 9 0066-4804/92/091859-05$02.00/0 Copyright X 1992, American Society for Microbiology Liquid Chromatographic Determination of Amoxicillin Concentrations in Bovine Plasma by Using a Tandem Solid-Phase Extraction Method H. J. NELIS,' J. VANDENBRANDEN,2 A. DE KRUIF,2 AND A. P. DE LEENHEERl* Laboratoria voor Medische Biochemie en voor Klinische Analyse, Rijksuniversiteit Gent, Harelbekestraat 72,1 and Kliniek voor Verloskunde, Voortplanting en Bednjfsdiergeneeskunde,2 B-9000 Ghent, Belgium Received 21 August 1991/Accepted 18 May 1992 We report a means ofdetermining amoxicillin in bovine plasma by liquid chromatography with UW detection at 235 nm. Purification and concentration of extracts were accomplished by a tandem solid-phase extraction procedure with two reversed-phase columns. Separation of amoxicillin from interferences was improved by the incorporation of a crown ether in the solvent systems used both for the solid-phase extraction and the final high-pressure liquid chromatography. Cefadroxil was added as an internal standard. The average recovery of amoxicillin from plasma (n = 23) was 78.2 + 3.0%, and the within-run and between-run coefficients of variation ranged from 1.8 to 7.0%. The detection limit was estimated at 0.1 ,ug/ml. This method was used to determine amoxicillin in bovine plasma after intramuscular administration of the drug.

The high-pressure liquid chromatographic (HPLC) deter- mg, 2.8 ml) came from Baker (Phillipsburg, N.J.) or Analyt- mination of amoxicillin in plasma represents a greater ana- ichem (Harbor City, Calif.). All other reagents were analyt- lytical challenge than does HPLC determination of ampicil- ical grade and were obtained from UCB (Braine-l'Alleud, lin. A sensitive ampicillin assay with detection in the low UV Belgium) or Merck (Darmstadt, Germany). region (210 nm) requires special approaches for sample Samples. Plasma samples were taken from three lactating pretreatment and chromatography, to eliminate the interfer- cows. They were given a single dose of 10 mg of amoxicillin ence from polar plasma constituents (11). Amoxicillin exhib- per kg of body weight by a deep intramuscular injection in its a maximum atA230, but it is considerably more polar than the neck region (23.5, 26, and 30.5 ml, respectively, of a 20% ampicillin and, consequently, is more susceptible to such preparation [Alfasan, Woerden, The Netherlands] divided interferences. To improve the selectivity of amoxicillin over two injection sites). Blood was sampled at 0, 1, 2, 3, 4, assays, the compound can be chemically (1, 2, 5, 9) or 8, 12, and 24 h postinjection by means of a catheter in a electrochemically (8) converted to derivatives that absorb jugular vein. The blood was collected in Venoject Li heparin UV light at higher wavelengths (1, 5) or that fluoresce (2, 8, tubes (Terumo Europe, Leuven, Belgium), and the plasma 9). was separated by centrifugation at 1,600 x g and was kept In our previous ampicillin assay (11), enhanced selectivity frozen at -80°C until analysis. and sensitivity were not derived from improved detection Apparatus and chromatographic conditions. The liquid but from the use of a tandem solid-phase purification and chromatograph consisted of a Varian 5020 pump (Varian concentration of the sample. The performance of a single Palo Alto, Calif.), an N60 valve injector (Valco solid-phase extraction method is apparently insufficient to Associates, separate amoxicillin efficiently from interfering plasma con- Houston, Tex.) fitted with a 100-,ul loop, an LC-95 variable- stituents (6, 7). The principle of two consecutive solid-phase wavelength detector (Perkin-Elmer, Norwalk, Conn.) set at extractions was maintained in the amoxicillin assay reported 235 nm (0.02 absorbance units full scale), and an SP 4100 here. However, the original approach devised for ampicillin integrator (Spectra-Pysics, San Jose, Calif.) set at attenua- did not prove to be transferable to amoxicillin. The modified tion 16. The column (15 by 0.46 cm) packed with 5-,um method uses two different reversed-phase extractions rather Hypersil ODS (Hichrom, Reading, United Kingdom) was than one ion-exchange and one reversed-phase step. A key eluted with 0.005 M potassium dihydrogen phosphate-0.1 M factor for obtaining the required degree of chromatographic 18-crown-6 in acetonitrile (84:16; vol/vol). The flow rate was selectivity both on the extraction columns and in the final 1 ml/min, and the temperature was ambient. HPLC separation is the incorporation of a crown ether in the Sample preparation. (i) Double reversed-phase extraction. respective solvent systems. To 0.5 ml of plasma was added 50 pl of the internal standard solution (cefadroxil; 15 ,ug/ml), 5 ml of 1.8% sodium chloride, and 0.1 ml of 1 M 18-crown-6 in methanol. This mixture MATERIALS AND METHODS was applied on top of a CH (cyclohexyl) extraction column Chemicals and reagents. Amoxicillin trihydrate and ce- that had been preconditioned successively with methanol (5 fadroxil (for structural formulas, see Fig. 1) were purchased ml) and water (5 ml). The column was washed with 1 ml of from Sigma (St. Louis, Mo.). HPLC-grade acetonitrile was water-methanol (9:1; vol/vol) containing 0.1 M 18-crown-6. from ROMIL (Loughborough, United Kingdom). Crown Elution was carried out with 3 ml of methanol-water (45:55; ether (18-crown-6) was obtained from Janssen Chimica vol/vol). The eluate was transferred to a conical evaporation (Beerse, Belgium). Bond Elut CH (cyclohexyl) and (500 tube, and the methanol was evaporated by applying vacuum C18 (water aspiration) under continuous vortexing of the tubes (Rotary Evapo-Mix; Buchler Instruments, Fort Lee, N.J.). * Corresponding author. To the aqueous residue was added 0.1 ml of 1 M 18-crown-6 1859 1860 NELIS ET AL. ANTIMICROB. AGENTS CHEMOTHER.

E 2 A B NH2 O/-N-.- -COOH

H w 1 u z

B H H H S HO ---CO.NH 0

NH2 o CH3

COOH FIG. 1. Structural formulas of amoxicillin (A) and cefadroxil (internal standard) (B). 3 in methanol and sufficient water to give a total volume of 16 16 MIN approximately 3 ml. This mixture was applied on top of a C18 TIME extraction column that had been preconditioned with meth- FIG. 2. Chromatogram of bovine plasma containing cefadroxil anol (5 ml) and water (5 ml). The column was washed with 1 (internal standard; peak 1) and amoxicillin (peak 2). Peak 3 coeluted ml of water-methanol (9:1; vol/vol) containing 0.1 M 18- with amoxicillin. (A) Plasma sample taken 2 h postdose; (B) blank crown-6 and was eluted with 3 ml of methanol. The eluate plasma sample. was evaporated to dryness by using the Evapo-Mix apparatus, the residue was redissolved in 0.25 ml of the chromatographic solvent, and a 100-pl aliquot was injected. (one sample a day) to calculate between-run precision. A (ii) Ion-exchange plus reversed-phase extraction. A 0.5-ml control sample containing 1.60 ,ug/ml was analyzed in one sample of bovine plasma supplemented with 50 ,ul of the run (n = 8) to test within-run precision. In addition, the internal standard solution (cefadroxil; 15 ,ug/ml) was mixed day-to-day variation on the slope of the calibration curves with 0.075 ml of 0.15 M phosphoric acid and 3 ml of was calculated. methanol. After centrifugation, the supernatant was isolated (iii) Recovery. The recovery was determined by repeti- and concentrated under vacuum (Rotary Evapo-Mix). To the tively analyzing plasma supplemented with known amounts aqueous residue was added 4 ml of 0.01 M hydrochloric acid, of amoxicillin (0.42, 1.66, and 4.16 ,ug/ml), but with the and the mixture was applied on top of a propylsulfonic acid addition of the internal standard after the evaporation of the extraction column that had been preconditioned succes- methanolic eluate of the C18 extraction column. A mixture of sively with methanol (5 ml) and 0.01 M HCl (5 ml). The the same amount of amoxicillin and the internal standard column was washed with 2 ml of 0.01 M HCl and was eluted served as a reference. Similarly, the recovery of the internal with 3 ml of 0.067 M potassium dihydrogen phosphate. After standard cefadroxil was evaluated, with amoxicillin added at the addition of 0.1 ml of 0.1 M 18-crown-6 in methanol and 2 the end. ml of water, the solution was transferred to a C18 extraction (iv) Detection limit. The detection limit was estimated from column that had been preconditioned successively with meth- the size of the smallest amoxicillin peak obtained in plasma anol (5 ml) and water (5 ml). The remainder of the procedure extracts. It was defined as the concentration in plasma that was as described above for double reversed-phase extraction. resulted in a detectable peak of approximately four times the Quantitation. Stock solutions of amoxicillin in 0.067 M noise level. KH2PO4 contained 4.4 and 44 ,ug/ml (expressed as anhy- Pharmacokinetic analysis. Model-independent parameters drous amoxicillin). The concentration of the internal stan- were evaluated. The peak concentration of drug in serum dard (cefadroxil dissolved in 0.067 M KH2PO4) was 15 and the time at which it was reached were the observed ,ug/ml. All solutions were stored at 4°C and were controlled values. The elimination half-life was calculated from the daily by chromatography. Samples (0.5 ml) of blank plasma terminal log-linear phase of the plasma concentration-time were supplemented with known amounts of amoxicillin (0.18 curve. The area under the curve was calculated by the to 4.4 ,ug/ml) and the internal standard (1.5 ,ug/ml) and were trapezoidal rule. analyzed as the unknowns. Calibration curves were constructed by plotting peak height ratios (amoxicillin/ce- RESULTS fadroxil) versus the corresponding amoxicillin concentrations. The amoxicillin concentrations in unknown samples Chromatography and plasma profiles. Figure 2A and B were calculated by extrapolation from the standard curves. represent chromatograms of postdose bovine plasma and Method validation. (i) Linearity. The slopes, intercepts, blank plasma, respectively. Amoxicillin and the internal and correlation coefficients of the calibration curves were standard cefadroxil eluted after 5.6 and 4.9 min, respec- calculated by linear regression analysis. tively. An acceptable blank trace was obtained, but a minor (ii) Reproducibility. Three plasma pools supplemented peak coeluted with amoxicillin (Fig. 2B). This interference with amoxicillin (nominal values, 0.42, 1.66, and 4.16 ,ug/ml) had to be taken into account in the standardization. were divided into 0.7-ml aliquots and frozen at -80°C. Sample preparation. There was virtually no difference in Samples (0.5 ml) were repetitively analyzed on different days the overall cleanliness of the extracts obtained with the two VOL. 36, 1992 HPLC DETERMINATION OF AMOXICILLIN IN PLASMA 1861

TABLE 1. Reproducibility of amoxicillin determination by TABLE 3. Recovery of amoxicillin and cefadroxil (internal analysis of control samples standard) from bovine plasma Precision Mean + SD Coefficient of No. of Recovery No. of concn (,ug/ml) variation (%) samples Compound Concn (pg/ml) (%[mean ±+ SD]) samples Within-run 1.596 ± 0.029 1.8 8 Amoxicillin 0.42 77.5 ± 3.6 7 Amoxicillin 1.66 80.2 ± 3.7 8 Between-runa 0.445 ± 0.031 7.0 10 Amoxicillin 4.16 76.7 ± 2.7 8 1.780 ± 0.071 4.0 9 Amoxicillin Avg recovery 78.2 ± 3.0 23 4.391 ± 0.175 4.0 10 Cefadroxil 1.5 74.6 ± 3.8 10 a For the between-run precision values, theoretical (nominal) values were 0.416, 1.664, and 4.160 pg/ml, respectively.

pretreatment could not be used. The original eluent contain- tandem solid-phase extractions (double reversed-phase or ing 0.067 M KH2PO4, methanol, acetonitrile, and a crown ion-exchange plus reversed-phase procedure). However, the ether failed to separate amoxicillin from large interfering total recovery of amoxicillin in the double reversed-phase plasma peaks, even when the percentage of acetonitrile was procedure was about 20% higher than that in the ion- lowered to accommodate for the higher polarity of amoxicil- exchange plus reversed-phase procedure (78% versus lin. Furthermore, amoxicillin also could not be separated <60%). In the latter procedure, the recovery from the from cefadroxil, which is structurally the most straightfor- reversed-phase extraction column was nearly quantitative, ward choice as an internal standard. Two modifications were but approximately 20% of the amoxicillin was not retained necessary. First, a decrease of the molarity of potassium on the PRS column. The loss of another 20% of the amoxi- phosphate from 0.067 to 0.005 M led to a considerable cillin before the application of the deproteinized extract improvement in resolution between amoxicillin and the could not exactly be accounted for, but it occurred in interferences, mainly because of a specific increase in the connection with the protein precipitation and/or evaporation retention of amoxicillin. This is consistent with the enhanced of the methanol. interaction between crown ethers and aminopenicillins at Method validation. The linear range of the standard curves lower concentrations of potassium salts (10). Second, by extended to at least 17 p,g/ml. However, in the daily routine, omitting the methanol from the eluent, i.e., by dissolving the standard curves were constructed only in the range of 0.17 to crown ether in the acetonitrile, amoxicillin and cefadroxil 4.4 ,ug/ml. Linear regression analysis of standard curves became separated at the baseline. always yielded correlation coefficients that exceeded 0.999. Unlike for ampicillin, the original tandem solid-phase Data with regard to within-run and between-run precisions extraction gave a recovery of amoxicillin from plasma that are given in Table 1, and day-to-day variations on the slope did not exceed 30%. Most of the loss (>50%) occurred on of the standard curves are given in Table 2. Table 3 lists the ion-exchange extraction column, which gave poor reten- recovery data for amoxicillin and the internal standard. The tion, although the pK. of amoxicillin is nearly identical to detection limit was set equal to the concentration corre- that of ampicillin (2). This anomalous behavior could be sponding to the small interfering plasma peak, i.e., roughly attributed to the ionic strength of the mixture applied on the 0.1 ,ug/ml. PRS column. With an amoxicillin standard dissolved in 0.01 Application. The method was applied to a small-scale M HCI, the recovery from PRS was 95%, whereas 0.05 M pharmacokinetic study of amoxicillin in three cows. Figure 3 HCl lowered this value to 46%. However, the recovery of shows a representative plasma amoxicillin-versus-time plot. mixed with 0.01 M HCI was nil Basic pharmacokinetic parameters for the three cows are amoxicillin from plasma listed in Table 4.

DISCUSSION tig/mll The same two principles that formed the backbone of our previous ampicillin assay (11), i.e., the incorporation of a crown ether into the chromatographic eluent and a tandem solid-phase extraction, also were the basis of the determination of amoxicillin. However, in both respects amoxicillin showed an erratic behavior compared with that of ampicillin, so that the original chromatographic system and sample

TABLE 2. Day-to-day variation on the slope of the calibration curvesa Slope Coefficient of (mean + SD) variation (%) 0.381 0.016b 4.1 8 16 24 hours 0.381 ± 0.007c 1.8 TIME a A total of 11 samples were tested. FIG. 3. Levels of amoxicillin in the plasma of a nonlactating cow b Obtained on spiked samples. (cow 2) as a function of time after administration of a dose of 10 c Obtained by direct injection of standards. mg/kg of body weight. 1862 NELIS ET AL. ANTIMICROB. AGENTS CHEMOTHER.

TABLE 4. Model-independent pharmacokinetic parameters for spiked samples and those obtained by direct injection (Table amoxicillin in nonlactating cowsa 2). Previous amoxicillin assays do not include internal standardization (1, 2, 4, 5, 7, 9, 13, 14) or use a compound that is Animal no. t h C. T h AUC (wt [kg]) r/ (h) (Lg/mi) m (h) (Ag h/ml) structurally unrelated to the drug (3, 6). The new method described here was applied to studies of 1 (470) 6.0 3.1 3 29.0 the pharmacokinetics of amoxicillin in cows. A total of 20 2 (520) 6.1 3.4 2 27.6 samples, including 5 standards and 1 control sample, could 3 (610) 5.9 3.1 2 25.5 be analyzed in a normal working day. Because of the a The cows were given a dose of 10 mg of amoxicillin per kg of body weight complexity of the sample pretreatment, the present method by intramuscular injection. tl,2, elimination half-life; Cm,,,,, peak concentration has the same limitations as our corresponding ampicillin of drug in serum; Tm.., time to peak concentration of drug in serum; AUC, assay (11) in terms of sample throughput. Advantages of the area under the concentration-time curve. method include its minimal technical requirements (no need for chemical or electrochemical derivatization or column switching) and HPLC column lifetime. The latter is likely to because the acid was too dilute to overcome the buffering be reduced considerably by repeated injections of impure action of the plasma. The final pH was only 3.5, which is well extracts obtained by a simple deproteinization of plasma. above the pK. of amoxicillin. The removal of the plasma Compared with other such assays, the one described here proteins partly overcame this problem, but it could not offers the advantage of enhanced selectivity and sensitivity. prevent the persistent loss of about 40% of amoxicillin (20% on the PRS column and 20% elsewhere). Because the double ACKNOWLEDGMENTS reversed-phase cleanup method gave extracts of comparable and it was also At the time of the experimental work, H.J.N. was a research purity but higher yields, because simpler (no associate from the Belgian Foundation for Scientific Research need for protein precipitation and subsequent evaporation of (NEWO). This work was conducted in cooperation with Alfasan methanol), the double reversed-phase procedure was se- Nederland BV, Woerden, The Netherlands. lected for final use. Initially, for application on the cyclo- We are indebted to J.H. Albracht for support. We thank Pascale hexyl column, the plasma sample was diluted only with Hoste for outstanding technical assistance and F. Belpaire for water and was mixed with the methanolic solution of the calculating some pharmacokinetic parameters. crown ether. The mean recovery was 80.0% + 5.0% (t = 0.88 ,ug/ml; n = 6). All samples in connection with the REFERENCES pharmacokinetic study in cows were analyzed under those 1. Carlqvist, J., and D. Westerlund. 1979. Determination of amox- conditions, without any anomalous results. However, when icillin in body fluids by reversed phase liquid chromatography the method was again used more than a year later, occa- coupled with a post-column derivatization procedure. J. Chro- sional abnormal losses of the internal standard were ob- matogr. 164:373-381. 2. Carlqvist, J., and D. Westerlund. 1985. Automated determina- served. The problem was readily solved by including sodium tion of amoxycillin in biological fluids by column switching in chloride in the sample before its application on the cyclo- ion-pair reversed-phase liquid chromatographic systems with hexyl cartridge. Sodium chloride reportedly increases the post-column derivatization. J. Chromatogr. 344:285-296. recovery of aminopenicillins in the presence of crown ethers 3. Erdmann, G. R., K. Walker, G. S. Giebink, and D. M. Canafax. on Sep Pak C18 extraction columns (12). To avoid any 1990. High performance liquid chromatographic analysis of further inconsistencies in recovery, possibly caused by amoxicillin in microliter volumes of chinchilla middle ear effu- batch-to-batch variations in the extraction columns, sodium sion and plasma. J. Liq. Chromatogr. 13:3339-3350. chloride was maintained in the definitive procedure. 4. Fouistone, M., and C. Reading. 1982. Assay of amoxicillin and The extent of purification was sufficient to permit concen- clavulanic acid, the components of augmentin, in biological fluids with high-performance liquid chromatography. Antimi- tration of the extract (final volume before injection, 250 ,ul), crob. Agents Chemother. 22:753-762. so that the equivalent of 200 ,ul of plasma was injected onto 5. Haginaka, J., and J. Wakai. 1985. High-performance liquid the column. In less selective assays that use UV detection, chromatographic assay of ampicillin, amoxicillin and ciclacillin this quantity is usually limited to 2.5 to 20 ,ul and the in serum and urine using a pre-column reaction with 1,2,4- detection limits are on the order of 0.5 ,ug/ml (3, 4, 7, 13, 14). triazole and mercury (II) chloride. Analyst 110:1277-1281. The limit of detectability for the method described here was 6. Jonkman, J. H. G., R. Schoenmaker, and J. Hempenius. 1985. about 0.1 jig/ml, which is a factor of 10 higher than the Determination of amoxycillin in plasma by ion pair column corresponding value for our ampicillin assay (11). The dif- extraction and reversed-phase ion pair high-performance liquid ference was mainly due to the presence of the small endog- chromatography. J. Pharm. Biomed. Anal. 3:359-365. enous that coeluted with 7. Lee, T. L., and M. A. Brooks. 1984. High-performance liquid plasma peak amoxicillin. However, chromatographic determination of amoxicillin in human plasma although this peak proved to be the sensitivity-limiting using a bonded-phase extraction. J. Chromatogr. 306:429-435. factor, it occurred in a reproducible way. That is, the peak 8. Mascher, H., and C. Kikuta. 1990. Determination of amoxicillin height ratio (unknown versus internal standard) was fairly in plasma by high-performance liquid chromatography with constant (0.048 ± 0.0057) and the peak was present in all fluorescence detection after on-line oxidation. J. Chromatogr. samples, including those of human plasma. To compensate 506:417-421. for the positive error, each run was standardized with 9. Miyazaki, K., K. Ohtani, K. Sunada, and T. Arita. 1983. drugfree plasma supplemented with amoxicillin and always Determination of ampicillin, amoxicillin, cephalexin, and ceph- included the analysis of a blank sample. radine in plasma by high-performance liquid chromatography The choice of cefadroxil, the cephalosporin equivalent of using fluorometric detection. J. Chromatogr. 276:478-482. 10. Nakagawa, T., A. Shibukawa, and T. Uno. 1982. Liquid chro- amoxicillin, as an internal standard has the same rationale as matography with crown ether-containing mobile phases. II. the use of cephalexin in our ampicillin assay (11). Both Retention behaviour of 3-lactam antibiotics in reversed-phase compounds behave in a similar way in the assay, as evi- high-performance liquid chromatography. J. Chromatogr. 239: denced by their comparable recoveries (Table 3) and the 695-706. equal slopes of the standard curves obtained by analyzing 11. Nelis, H. J., J. Vandenbranden, A. De Kruif, D. Mattheeuws, VOL. 36, 1992 HPLC DETERMINATION OF AMOXICILLIN IN PLASMA 1863

and A. P. De Leenheer. 1992. Liquid chromatographic determi- lichaamsvloeistoffen. Praktische analysevoorschriften VIII. nation of ampicillin in bovine and dog plasma using a tandem Amoxycillin, atenolol, clobazam and cytarabine. Ziekenhuisfar- solid-phase extraction method. Antimicrob. Agents Chemother. macie 4:137-140. 36:1606-1610. 14. Vree, T. B., Y. A. Hekster, A. M. Baars, and E. Van Der KleUn. 12. Terada, H., and Y. Sakabe. 1985. Studies on residual antibac- terials in foods. IV. Simultaneous determination of penicillin G, 1978. Rapid determination of amoxycillin (ClamoxylR), and penicillin V and ampicillin in milk by high-performance liquid ampicillin (PenbritinR) in body fluids of man by means of chromatography. J. Chromatogr. 348:379-387. high-performance liquid chromatography. J. Chromatogr. 145: 13. Uges, D. R. A. 1988. De bepaling van enige xenobiotica in 496-501.