Quantification of Several Acidic Drugs in Equine Serum Using LC MS-MS

Journal of Analytical Toxicology Advance Access published August 27, 2013 Journal of Analytical Toxicology 2013;1 5 doi:10.1093/jat/bkt069 Special Issue Quantification of Several Acidic Drugs in Equine Serum Using LC MS-MS Brendan Heffron, Lisa Taddei, Marc Benoit and Adam Negrusz* Department of Biopharmaceutical Sciences, Animal Forensic Toxicology Laboratory, College of Pharmacy, University of Illinois at Chicago, 2242 West Harrison Street, Chicago, IL 60612, USA *Author to whom correspondence should be addressed. Email: anegrusz@uic.edu The use of nonsteroidal antiinflammatory drugs in racehorses is allowed under most jurisdictions. Furosemide is administered to treat exercise-induced pulmonary hemorrhage. To help distinguish between therapeutic and illegal uses, racing regulatory bodies have set thresholds in serum for several drugs. The method for the simultaneous detection and quantification of furosemide, flunixin, ketoprofen, phenylbutazone and oxyphenbutazone using 500 ml of serum, and liquid extraction using diethyl ether : hexanes : dichloromethane followed by liquid chromatography tandem mass spectrometry quantitation, was developed and validated. Method validation included inter- and intraday precision and accuracy. Method validation also included bench-top, freeze thaw, processed and long-term storage stability testing. For all stability testing, the compounds showed a breakdown of <15%. Inter- and intraday precision for all compounds was found to be within the acceptance interval of +15% [+20% at the lower limit of quantitation (LLOQ)]. Accuracy data for all compounds were within the acceptance interval of +15% (+20% at the LLOQ). Uncertainty was calculated using the simplified Guide to the Expression of Uncertainty in Measurement approach and was <30% for all drugs at 95% confidence level. The method was found to be both robust and accurate for all tested drugs. Introduction Nonsteroidal antiinflammatory drugs (NSAIDs) are commonly used to manage pain and lameness in the racehorse by inhibition of cyclooxygenase (COX) enzymes 1 and 2 (1 3). While the inhibition of COX 2 reduces the horse s immune response, it is the inhibition of COX 1 that is responsible for the increase in ulcer episodes and hemorrhaging that is associated with long-term use of NSAIDs (4). As such, NSAIDs are classified as restricted substances by horse racing authorities, but are allowed for therapeutic purposes. The Association of Racing Commissioners International lists most NSAIDs as Class 4 drugs (5). A majority of racing jurisdictions have limited the drugs allowed in the racehorse at post time and have set concentration thresholds for them in blood. The NSAIDs, which require concentration monitoring in blood, are ketoprofen, flunixin, phenylbutazone and its active metabolite oxyphenbutazone. Current thresholds established by the Illinois Racing Board for these drugs in blood are 10, 20 ng/ml, 2 and 2 mg/ml, respectively (6). Furosemide is used to treat exercise-induced pulmonary hemorrhaging in the racehorse (7), although there has been some research into whether it is effective in this regard (8). Furosemide is a diuretic causing massive urination and loss of water (9). This dilution of the urine can mask the presence of other drugs that are regulated by a majority of racing jurisdictions. Furosemide can cause the elevation of carbon dioxide concentration in plasma, another common area of screening by racing authorities (9, 10). Because of the potential benefit to the horse, furosemide is allowed in the USA. The threshold concentration in the blood has been set by racing jurisdictions to ensure it is not given too close to the race or in concentrations that are high enough to significantly decrease the specific gravity of the urine. Under Illinois regulations, the threshold for furosemide in blood is 100 ng/ml (6). Previously our laboratory conducted screening for these drugs using enzyme-linked immunosorbent assays (ELISA), which although sensitive, is a poor indication of their concentrations. In addition, many false positives from the screening process had to be confirmed using more definitive and accurate instrumentation, such as gas or liquid chromatography. Because the amount of serum available for testing is significantly limited, the use of ELISA followed by one or two instrumental analyses can quickly deplete the amount of serum available. Also, because of the high volume of samples, this approach can be both laborious and expensive for a laboratory to successfully screen and confirm the presence of these drugs. Methods allowing the quantitation of some of the drugs studied in this paper have been published, but have not been attempted to quantitate all in a single injection (11). While this method is accurate, the limited amount of serum available requires that multiple drugs be quantitated in a single extraction. The aim of this study was to develop and validate both selective in screening and accurate in quantitating analytical method for the qualitative and quantitative determination of furosemide, flunixin, ketoprofen, phenylbutazone and oxyphenbutazone (Figure 1) in equine serum by liquid liquid extraction followed by liquid chromatography tandem mass spectrometry triple quad technology. Furthermore, this method was put into place at our laboratory in January 2012 and has been found to be successful in daily screening and quantitating these drugs in a timely and accurate manner. Experimental Reference compounds Flunixin meglumine was obtained from USP (Rockville, MD, USA). Furosemide, phenylbutazone, oxyphenbutazone, ketoprofen and flunixin-d 3 were obtained from Sigma (St Louis, MO, USA). Standard stock solutions were prepared at 1 mg/ml in methanol for all compounds for use in the preparation of the calibration curve. Separate control stock solutions were prepared at a concentration of 1 mg/ml for all compounds for use in the creation of the quality control preparations. A standard stock solution of flunixin-d 3 was prepared at 1 mg/ml. All stock solutions were stored at 4 + 28C. Reagents Hexanes, ethyl ether, dichloromethane, acetonitrile, formic acid, potassium phosphate, phosphoric acid and water were obtained from Fisher Scientific (Hanover Park, IL, USA) and were all HPLC # The Author [2013]. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Table I Product ions of acidic drugs after electrospray ionization (ESI) and collision-induced dissociation Compound Ionization Precursor ion Mass 1 resolution Fragmentation voltage Collision energy (ev) Product ions (m/z) Furosemide Negative 329 Unit 160 15 78 a Furosemide Negative 329 Unit 160 20 205 Furosemide Negative 329 Unit 160 25 285 Oxyphenbutazone Positive 325.2 Unit 110 48 81.1 a Oxyphenbutazone Positive 325.2 Unit 110 64 77.1 Oxyphenbutazone Positive 325.2 Unit 110 100 51.1 Phenylbutazone Positive 309.2 Unit 80 32 81.1 a Phenylbutazone Positive 309.2 Unit 80 56 77.1 Phenylbutazone Positive 309.2 Unit 80 100 51.1 Flunixin Positive 297.1 Unit 110 36 109.1 a Flunixin Positive 297.1 Unit 110 56 264.1 Flunixin Positive 297.1 Unit 110 84 83.1 Ketoprofen Positive 255 Unit 140 10 105 a Ketoprofen Positive 255 Unit 140 20 209 Ketoprofen Positive 255 Unit 140 40 77 Flunixin- d3 Positive 300.1 Unit 140 36 112 a Flunixin- d3 Positive 300.1 Unit 140 48 236 Flunixin- d3 Positive 300.1 Unit 140 56 264 a Marks the quantitation ion. Figure 1. Structures for drugs in the study. grade or better. Drug-free horse serum was obtained from Bioreclamation LLC (Westbury, NY, USA). Phosphate buffer ( ph 3) was prepared by adding potassium phosphate monobasic to 500 ml of water and stirring until saturation of the solution indicated by crystals stopping dissolution. The ph of the solution was adjusted to 3 with phosphoric acid. Twenty-five microliters of flunixin-d 3 standard solution was then added and mixed giving a concentration of 50 ng/ml. The solution was stored at 4 + 28C and was stable for 1 month. 0.2% Formic acid was prepared by adding 2 ml of formic acid to 998 ml of water and mixing. This solution was stored at room temperature and is stable for 1 month. Standard curve and quality control preparations For all drugs, six point standard curves were prepared in drugfree equine serum. Quality control preparations at two different concentrations were prepared in naïve equine serum. For furosemide, the standard curve concentrations were 40, 80, 160, 320, 400 and 600 ng/ml. Quality control serum preparations were 50 and 200 ng/ml. For flunixin, the standard curve concentrations were 10, 20, 40, 80, 100 and 150 ng/ml. The quality control concentrations were 15 and 60 ng/ml. For ketoprofen, the standard curve concentrations were 5, 10, 20, 40, 50 and 75 ng/ml. The quality control concentrations were 10 and 40 ng/ml. For phenylbutazone and oxyphenbutazone, the standard curve concentrations were 1, 2, 4, 8, 10 and 15 mg/ml. The quality control concentrations were 1.5 and 6 mg/ml. hundred microliters of potassium phosphate buffer with flunixin-d 3 was added to all tubes. Three milliliters of dichloromethane : hexanes : ethyl ether (1 : 1 : 1) mixture was added to each tube. The samples were capped, rotoracked for 10 min and centrifuged for 10 min at 1,500 g. The solvent layer was transferred into a clean tube and dried using nitrogen and a water bath at 378C. Twohundredandfiftymicrolitersof0.2%formicacidinwaterwas added to each tube. The tubes were vortexed and transferred to HPLC vials and 5 ml were injected into the LC MS-MS system Instrumentation and conditions An Agilent 1200 Series HPLC coupled with an Agilent 6000 Series Mass Spectrometer Triple Quadruple (LC MS-QQQ) (New Castle, DE, USA) in both positive and negative ionization modes was used for all analyses. An Agilent Poroshell 120 EC-C 18 2.1 100 mm, 2.7 mm column was used. The temperature of the column was set at 308C, and the mobile phase flow was 500 ml/min. The mobile phase was 0.2% formic acid in water (A) and 10% water in acetonitrile (B). Initial conditions were 55% B kept for 1 min. Between 1 and 2 min, a gradient was 55 100% B and held at 100% B for 3 min. For the mass spectrometer, the nitrogen drying gas temperature was 3508C. The gas flow was 11 L/min, and the nebulizer pressure was 40 psi. Electrospray ionization was employed for all drugs. The collision gas was high-purity nitrogen. To increase sensitivity, multiple time segments were created. Time Segment 1 scanned for furosemide in the negative ionization mode. Time Segment 2 scanned for all other drugs in the positive ionization mode. Data analysis was done using the Agilent Masshunter Qualitative and Quantitative software. Data were collected in multiple reaction monitoring (MRM) mode. Product ions and their respective fragmentation voltages and collision energies for the drugs and the internal standard were determined using the Agilent Optimizer software and are listed in Table I. Extraction method For all curve and quality control samples, 500 ml of equine serum was transferred into a 16 125 mm glass screw cap tube. Five Selectivity Selectivity was tested by analyzing drug-free equine serum samples from three different manufacturing lots. These samples 2 Heffron et al.

underwent extraction and then were analyzed for the presence of product ions for all analytes within a timeframe of 0.2 min based on injected standard retention times. Carryover Carryover and contamination were evaluated by analyzing three samples of naive serum fortified at the highest curve point concentration followed by three injections of 0.2% formic acid in water. Carryover was said to have happened if the signal-to-noise (S/N) ratio was higher than 5 : 1 for any of the ions for the drug of interest. Limit of detection and quantitation Serial dilutions of naı ve serum fortified with the drugs were analyzed using the Agilent Masshunter software. The limit of detection (LOD) was determined to be the lowest concentration where the signal-to-noise of all target ions was at least 5 : 1. The limit of quantitation was determined to be the lowest concentration where S/N of all target ions was at least 10 : 1. Long-term stability Stability testing under long-term conditions was performed on three samples of drug-free serum, which were fortified at the high quality control concentration and then frozen. These samples were stored at 2208C for 90 days. On the day of analysis, the samples were thawed and three aliquots of drug-free serum were fortified at the same concentration. All samples were then extracted and analyzed. Ion suppression and enhancement Ion suppression and enhancement were evaluated by preparing two sets of three samples at the high concentration controls of the drugs. One set consisted of neat standards at a concentration of the high quality controls. A second set consisted of naive serum fortified after extraction at the concentrations at the high quality control preparations. These ratios calculated as peak areas of quantitation ions from the second set divided by the peak areas from the first set were expressed as a percentage and 100 was subtracted. Values below zero represent ion suppression and values above zero, ion enhancement. Stability testing For the stability testing, the area of the peak for the quantitation ion (Table I) for each of the drugs of interest was used in the calculations for the following tests. Bench top samples Stability testing for bench top samples was performed on three samples of drug-free serum that were fortified at the high quality control concentration and left at room temperature for 24 h before being extracted and analyzed. These samples were fortified to yield final concentrations at the high-level control preparations for all drugs. On the day of analysis, three separate samples of naive serum were fortified with drugs studied at the same concentrations. All serum samples were then extracted and analyzed. Processed samples Stability testing for processed samples was performed on three aliquots of drug-free serum, which were fortified at the high quality control concentration for all drugs, extracted and prepared for analysis according to the previously described procedure but then left at room temperature for 48 h before being analyzed. On the day of analysis, three samples of naive serum were fortified at the same concentrations, extracted and analyzed. Freeze thaw samples Stability testing under freeze thaw conditions was performed on three samples of naive serum, which were fortified at the high quality control concentration and then frozen. These samples were stored at 2208C for 48 h, and then thawed at room temperature. The freeze thaw cycle was repeated twice. On the day of analysis, the samples were thawed and three aliquots of drug-free serum were fortified at the same concentration. All samples were then extracted and analyzed. Extraction recovery For the calculation of extraction recovery, three drug-free samples were fortified at the high quality control concentrations before extraction. Next, three drug-free serum samples were extracted and then fortified at the high quality control concentrations after extraction. The average responses of the peaks of the two sets of samples were compared and expressed as a percentage. Accuracy and precision Accuracy and precision for the analytical method were determined by using a total of 30 replicates at both low and high control serum preparations. Six samples at each concentration were run on 5 separate days. The inter- and intraday precisions and accuracies were then calculated. Results and discussion The method for simultaneous screening and quantitation of phenylbutazone, oxyphenbutazone, flunixin, furosemide and ketoprofen was successfully validated and was found to provide accurate qualitative and quantitative information for the selected drugs in a high-throughput scenario. Naive serum extracts were shown to be drug-free. No detectable peaks for the product ions were detected within 0.2 min of the expected retention time. None of the drugs showed carryover from one injection to the next. The curves were linear over the range of concentrations tested, with the correlation coefficients no less than 0.92 using a calibration model with linear fit and weighting of 1/X. The correlation coefficient values for furosemide, flunixin, ketoprofen, phenylbutazone and oxyphenbutazone were 0.972, 0.999, 0.978, 0.999 and 0.928, respectively. During the validation process, the stability testing showed that all drugs were stable throughout the conditions to which a sample could normally be exposed, and the results are presented in Table II. For long-term storage and processed sample testing, Quantitation of Acidic Drugs in Serum 3

Table II LOD, lower limit of quantitation (LLOQ), matrix effect and stability for the drugs of interest Drug LOD (ng/ml) LLOQ (ng/ml) Matrix effect (%) Freeze thaw stability (%) Long-term stability (%) Processed sample stability (%) Bench top stability (%) Extraction recovery (%) Furosemide 10 15 215 90 104 97 87 54 Oxyphenbutazone 200 500 25 108 106 99 96 60 Phenylbutazone 300 500 22 95 89 101 96 54 Flunixin 0.5 1 210 103 95 102 88 50 Ketoprofen 1 3 210 90 92 102 91 44 Table III Precision and accuracy data Drug Quality control (QC) high QC low Precision Accuracy Precision Accuracy Interday Intraday Total Interday Intraday Total Furosemide 9.8 3.7 10.8 4.2 13.1 6.2 14.9 3.4 Oxyphenbutazone 11.7 3.3 12.6 4.8 17.4 6.8 19.3 9.3 Phenylbutazone 11.7 1.9 12.3 1.5 18.5 4.2 19.6 4.0 Flunixin 5.8 1.9 6.3 8.0 12.8 7.3 15.1 6.0 Ketoprofen 9.1 7.2 11.9 11.5 16.7 6.8 18.0 4.0 Table IV Expanded uncertainty (95% confidence level) for the drugs of interest Drug Expanded uncertainty (%) Ketoprofen 14 Phenylbutazone 26 Oxyphenbutazone 22 Flunixin 18 Furosemide 20 Table V Reported samples for year 2012 of 7,167 samples tested Drug Number of reported positives Concentration range Furosemide 9 115 384 ng/ml Flunixin 13 25 121 ng/ml Phenylbutazone and metabolite 18 2.4 8.5 mg/ml degradation was,5% with the exception of phenylbutazone, which showed degradation of 12% during long-term storage. Ion suppression was shown to be minimal with all samples showing a suppression of 2 15% (Table II). The use of a single internal standard reduces cost and is successful most likely due to the fact that all analytes have similar extraction recoveries of 44 60% (Table II). The precision and accuracy results are in the acceptable ranges and are presented in Table III. At the high concentration, all calculated parameters were within 15% of their expected values for both inter- and intraday precision and accuracy. At the low concentration, all calculated concentrations were within 20% of their expected values for precision and were within 10% for accuracy. The total run time was 6 min per sample, which allows a full tray (96 samples) to be analyzed and data collected in around 10 h. Because this method was intended to be used as an identifying and confirmatory procedure for regulatory purposes, the expanded uncertainties were calculated according to the simplified Guide to the Expression of Uncertainty in Measurement (GUM) (12). For the compounds in this study, a confidence level of 95% was chosen and the results can be found in Table IV. As can be seen, using a 95% confidence level, the level of uncertainty for all drugs was,30%. This method was put into place in the laboratory at the beginning of 2012 and during that year 7,167 post race serum samples were tested. Of 7,167 samples, 0.25% exceeded the threshold for phenylbutazone or its metabolite, 0.12% were above the threshold for furosemide and 0.18% were in excess for flunixin (Table V). Except for phenylbutazone and its metabolite, there were no multidrug violations. This represents a positive rate of 0.55%. These drugs represented 42% of the total reported positives or overages in our laboratory. Conclusion The method was found to be both robust and accurate for all tested drugs. In a high-throughput toxicology screening environment, it is imperative that results be obtained using as little sample as is necessary to ensure accuracy. The use of LC MS QQQ helps to address these needs by providing fast and accurate analysis including qualitative and quantitative data. This reduction in analysis time can lower both the reporting period and the cost per sample. The ability of the instrumentation to analyze for multiple compounds in a single injection provides the opportunity for more drugs to be added to the method with a minimal drop in sensitivity. Further experimentation should explore the possibility of adding other drugs to the screening procedure. Funding This project was funded by the Illinois Racing Board. References 1. Foreman, J.H., Grubb, T.L., Inoue, O.J., Banner, S.E., Ball, K.T. (2010) Efficacy of single-dose intravenous phenylbutazone and flunixin 4 Heffron et al.

meglumine before, during, and after exercise in an experimental reversible model of foot lameness in horses. Equine Veterinary Journal: Supplement, 38, 601 605. 2. Foreman, J.H., Ruemmler, R. (2011) Phenylbutazone and flunixin meglumine used singly or in combination in experimental lameness in horses. Equine Veterinary Journal: Supplement, 40, 12 17. 3. Soma, L.R., Uboh, C.E., Maylin, G.M. (2012) The use of phenylbutazone in the horse. Journal of Veterinary Pharmacology and Therapeutics, 35, 1 12. 4. Marshall, J.F., Blikslager, A.T. (2011) The effect of nonsteroidal antiinflammatory drugs on the equine intestine. Equine Veterinary Journal: Supplement, 39, 140 144. 5. Association of Racing Commissioners International, Inc., Drug Testing Standards and Practices Program Model Rules Guideline Version 5.00. Uniform Classification Guidelines for Foreign Substances and Recommended Penalties and Model Rule. Association of Racing Commissioners International, Inc. http://arcicom.businesscatalyst. com/assets/uniform-classification-guidelines%2c-version-05-00 - zilpaterol%2c-tetramisole%2c-ambroxol.pdf (11 March 2013, date last accessed). 6. Joint Committee on Administrative Rules Administrative Code. Illinois Racing Board Rules and Regulations. Section 603.60 Permitted Use of Foreign Substances and Threshold Levels. Effective 28 November 2012. http://www.ilga.gov/commission/ jcar/admincode/011/011006030002100r.html (11 March 2013, date last accessed). 7. Hinchcliff, H.W., Morley, P.S., Guthrie, A. (2011) Use of furosemide for exercise-induced pulmonary hemorrhage in racehorses. Journal of American Veterinary Medical Association, 239, 1407. 8. Vengust, M., Kerr, C., Staempfli, H.R., Pringle, J., Heigenhauser, G.J., Viel, L. (2011) Effect of frusemide on transvascular fluid fluxes across the lung in exercising horses. Equine Veterinary Journal, 43, 451 459. 9. Dirikolu, L., Lehner, A.F., Hughes, C. (2003) Detection, quantification, and pharmacokinetics of furosemide and its effects on urinary specific gravity following IV administration to horses. Veterinary Therapeutics: Research in Applied Veterinary Medicine, 4,350 363. 10. Cohen, N.D., Stanley, S.D., Arthur, R.M., Wang, N. (2006) Factors influencing pre-race serum concentration of total carbon dioxide in thoroughbread horseracing in california. Equine Veterinary Journal, 38, 543 548. 11. You, Y., Uboh, C.E., Soma, L.R. (2009) Screening, quantification, and confirmation of phenylbutazone and oxyphenbutazone in equine plasma by liquid chromatography-tandem mass spectrometry. Journal of Analytical Toxicology, 33, 41 50. 12. LeBeau, M.A. Measuring and reporting uncertainty. In: Watts, J., Moffat, A.C., Osselton, M.D., Widdop, B. (eds). Clarke s Analysis of Drugs and Poisons. Vol. 1. 4th edition, Chapter 23. Gurnee, IL: Pharmaceutical Press, 2011, 371 389. Quantitation of Acidic Drugs in Serum 5