Detection and Identification of Flunixin After Multiple Intravenous and Intramuscular Doses to Horses

Transcription

1 Detection and Identification of Flunixin After Multiple Intravenous and Intramuscular Doses to Horses R.A. Sams 1,*, D.F. Gerken 2, and S.M. Ashcraft 1 1Department of Veterinary Clinical Sciences, College of Veterinary Medicine, The Ohio State University, Columbus, Ohio and 2Department of Veterinary Biosciences, College of Veterinary Medicine, The Ohio State University, Columbus, Ohio [ Abstract ] The objectives of the study were to compare various methods to determine flunixin in test samples collected periodically from horses after intramuscular (IM) and intravenous (IV) dosing at the maximum recommended dosage and to document detection times for this drug in test samples. Flunixin, a nonsteroidal anti-inflammatory drug approved for use in horses, was administered to eight mares in five consecutive daily doses of 1.1 mg per kilogram of body weight by the IM or IV route. Flunixin was detected in urine samples collected at various times after drug administration by flunixin enzyme-linked immunosorbent assay (ELISA), thin-layer chromatography (TLC), high-performance liquid chromatography (HPLC), and gas chromatographic-mass spectrometric (GC-MS) methods. Detection time was defined as the time period over which flunixin was detected and was dependent on the method used. The shortest detection times were 24 to 48 h and were observed when the TIC method was used. On the other hand, detection times were as long as 15 days when HPLC, GC-MS, and flunixin ELISA methods were used. The use of these more sensitive tests to monitor official samples collected from racehorses could result in positive tests for flunixin when it is exerting no detectable clinical effects because it produces clinical effects lasting only h in horses. The disposition of flunixin in the horse has been extensively investigated (1--4). It is eliminated from plasma with a short half-life of approximately 1.6 h and has a relatively small volume of distribution. It is metabolized by oxidation to 5-hydroxyflunixin and by conjugation of the carboxylic acid moiety with glucuronic acid as shown in Figure ] (5,6). Most of the dose is excreted in the urine as flunixin or flunixin glucuronide (3,6). Various detection times have been reported for flunixin after administration to horses (2,5,6). Flunixin and 5-hydroxyflunixin were detected by gas chromatographic-mass spectrometric (GC-MS) analysis of extracts of urine samples collected for 175 h and 54 h, respectively, after IV administration of a of 1 mg/kg bw to a horse (5). In another study, flunixin was detected by GC analysis with nitrogen-specific detection in an extract from a urine sample collected from 32 to 42 h after IM administration of 1.1 mg of flunixin meglumine per kilogram bw to a standardbred gelding (6). The limit of detection of this reported method was approximately 74 ng/ml urine (6). Flunixin was detected by GC analysis with electroncapture detection from extracts of urine samples collected for approximately 48 h after a single lv dose of 1.1 mg/kg bw (2). However, no evidence of flunixin was detected in extracts of Introduction H CH 3 H CH 3 ~.N I N ~L CF3 Flunixin is an analgesic nonsteroidal anti-inflammatory drug marketed as Banamine and Finadyne by Schering-Plough Corp. for alleviation of pain and inflammation associated with musculoskeletal disorders and alleviation of visceral pain associated with colic in horses. The maximum recommended dose is 1.1 mg per kilogram of body weight (bw), once daily, for five days. The dose may be given by either intravenous (IV) or intramuscular (IM) injection. 9 Author to ~ hom cr e should he addressed HO" "O Flunixin 5'-Hydroxyflunixin Conjugate(s) Conjugate(s) Figure 1. Metabolic pathways for flunixin in the horse {5). 372 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission,

2 urine samples collected 24 h after dosing when a less-sensitive thin-layer chromatographic (TLC) method of analysis was used (2). Thus, the reported time period for detection of flunixin after single IV or IM administration depends on the method used to detect the drug. This is not surprising because the limits of detection of these methods would be expected to differ. Regulatory control of flunixin in racehorses is often accomplished by TLC analysis of urine extracts (7). Because flunixin is excreted largely as a base-labile conjugate with glucuronic acid (6), the urine sample is first treated with dilute base to hydrolyze this conjugate and thereby increase the probability of detecting flunixin. After liquid-liquid or solid-phase extraction of flunixin from the treated urine sample, flunixin is separated from other substances on a silica-gel TLC plate using an appropriate solvent-developing system. Flunixin produces a concentration-dependent quench spot upon illumination of TLC plates containing a fluorescent indicator with short-wavelength ultraviolet light and a brown spot when sprayed sequentially with Dragendorff reagent and aqueous 5% sodium nitrite reagent. The limit of detection of TLC methods for flunixin is approximately ng/ml of urine depending on the volume of urine sampled, the recovery and purity of flunixin obtained by the extraction procedure, the fraction of the extract applied to the TLC plate, and other variables. Test samples found to contain flunixin by TLC analysis are then routinely subjected to GC-MS analysis to confirm the identity of the analyte (7). An enzyme-linked immunosorbent assay (ELISA) was recently developed for detection of flunixin in horse urine samples (8). The limit of detection of the flunixin ELISA is approximately 15 ng of flunixin per milliliter of horse urine and is therefore much lower than that of TLC methods. Therefore, use of the flunixin ELISA is expected to increase the detection period of flunixin compared to the use of TLC methods. The objectives of the study were to compare various methods to determine flunixin in test samples collected periodically from horses after IM and IV dosing at the maximum recommended dosage and to document detection times for this drug in urine samples collected periodically after the last dose. Materials and Methods Formulation and route of administration Flunixin, as flunixin meglumine (Banamine, 50 mg/ml, Schering-Plough Animal Health, Union, NJ), was administered by the IV or IM route once daily for five consecutive days to horses. Animal studies Animals. Animals used in this study were eight standardbred and Thoroughbred mares aged years and weighing kg. The horses were housed in individual stalls in an enclosed barn. Horses were provided mixed grass-alfalfa hay ad libitum. The horses were judged to be in good health based on physical examination and hemogram. Horses were not administered any other drugs for three weeks before the start of the study. All studies were approved by The Ohio State University Institutional Laboratory Animal Care and Use Committee and were conducted according to the requirements of the Animal Welfare Act regulations (9CFR) and the Public Health Service Policy on Humane Care and Use of Laboratory Animals (OPRR, NIH, 1986). Multiple-dose administration and sample collection. Each of the horses was weighed and randomly assigned to one of two treatment groups consisting of four horses per group. Each horse was administered either a single IM or IV dose of flunixin as flunixin meglumine daily for five consecutive days. Urine samples were collected at 1, 2, 3, 4, 5, 7, 9, and 15 days after the last dose via an indwelling Foley urinary catheter inserted in the bladder of each horse. Urine samples were frozen at -20~ until tested. The time of collection and volume of each urine sample collected were recorded. Reagents and materials Reagents and ELISA test kits. Water and methanol were HPLC grade from Burdick & Jackson Laboratories (Muskegon, MI). N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA) was purchased from Pierce (Rockford, IL). All other reagents were analytical reagent grade or better. The flunixin ELISA test kits were purchased from Elisa Technologies, Inc. (Lexington, KY). All reagents and materials necessary to perform ELISA tests were included in the test kits. Stock solutions. Stock solutions of flunixin (Schering- Plough Corp., Union, N J) and niflumic acid (ER Squibb & Sons, Inc., Princeton, N J) were prepared in methanol at a concentration of 1.00 mg/ml. The stock solution of flunixin was diluted with methanol to produce a working standard solution at a concentration of 100 ng/ijl. These solutions were stored in the dark at 4-6~ TLC analysis for flunixin in urine sample extracts Positive control urine samples for TLC analysis. Positive control urine samples for TLC analysis were prepared by adding appropriate volumes of the stock solution of flunixin to drugfree horse urine to produce concentrations of 50.0, 100, 200, 500, 1000, and 2000 ng/ml. The positive control samples were prepared immediately before use. Extraction procedure. Aliquots (2 ml) of test samples collected from horses and positive control samples were mixed with 1.0 ml of 0.1 N sodium hydroxide solution in 16 x 125- mm glass test tubes and allowed to stand for 10 rain at ambient temperature. At the end of this time, 4 ml of ph 3.3 phosphate buffer (a saturated solution of potassium dihydrogen phosphate adjusted to ph 3.3 with 6 N hydrochloric acid) was added to each tube and vortex mixed for 3 s. Then 5 ml of dichloromethane/petroleum ether (10:1, v/v) was added to each tube, and the contents of the tubes were mixed by end-over-end rotation at 20 rpm for 5 min. After centrifuging at rpm for 5 min, the aqueous layers were removed by aspiration and discarded. The remaining organic phases were decanted into new 16 x 125-mm tubes and evaporated to dryness under nitrogen on a water bath. The residue from each sample extract was dissolved in 10 ~L of dichloromethane for spotting on the TLC plate. 373

3 Journal of AnalvticaF Toxicology, Vol. 23, September 1999 Spotting, developing, and spraying procedure. A 2-1~L aliquot of each test sample extract and positive control sample extract was spotted 1.0 cm from the bottom edge ofa 10-cm x 10-cm silica gel TLC plate with fluorescent indicator (EM Reagents, 0.25-mm layer thickness, silica gel 60, F-254). The plates were developed in the Davidow solvent-developing system (ethyl acetate/methanol/concentrated ammonium hydroxide solution, 85:10:5, v/v/v) to a distance of 5.0 cm above the origin. The plates were then viewed under long-wavelength (365 nm) and short-wavelength (254 nm) UV light; fluorescent and quenching spots were marked. The plates were heated on a warming tray in a fume hood for 5 min to remove residual developing solvents and then sprayed sequentially with Dragendorff and aqueous 5% sodium nitrite reagents. Flunixin produced a brown spot at a relative migration distance (RF value) of approximately 0.20 under these conditions. The RF values, observed reactions, and relative intensities of all spots produced by positive control samples and test samples were recorded. The test was considered positive for flunixin if a spot was observed at the RF value of flunixin. High-performance liquid chromatographic (HPLC) analysis for flunixin in urine sample extracts Urine calibrators for HPLC. Urine calibrators were prepared by adding appropriate volumes of the stock solution of flunixin to drug-free horse urine to produce concentrations of 50.0, 100, 300, 500, 700, and 1000 ng/ml. All calibrators were stored frozen at-17~ A complete set of calibrators consisting of one at each concentration was processed with each set of test sampies. Quality-control samples for HPLC. Quality-control samples were prepared by adding the appropriate volume of a separately prepared stock solution of flunixin to drug-free horse urine to produce quality-control samples at concentrations of 100 ng/ml and 1000 ng/ml. Quality-control samples were stored frozen at-17~ One quality-control sample at each concentration was processed and analyzed with each set of test samples. Sample preparation. A 1.0-pL aliquot of the stock solution of niflumic acid (for use as the internal standard) and 0.50 ml of horse urine (calibrator, quality-control sample, or test sample) were added to 16 x 100-rnm screw-cap tubes. Urine samples were then mixed with 0.10 ml of 1.0 N sodium hydroxide solution for 1 h at room temperature to hydrolyze base-labile conjugates of flunixin. After the addition of 0.10 ml of aqueous 5% zinc sulfate solution to each tube, the contents of the tubes were mixed by vortex mixing. The tubes were then centrifuged for 5 rain, and the supernatant solutions were transferred to 16 x 100-mm screw-cap tubes. A 1.0-mL aliquot of 8.5M acetic acid was added to each tube, the contents were briefly mixed, and 5.0 ml of diethyl ether was added. The contents of each tube were mixed by end-over-end rotation for 10 min, centrifuged for 5 min, and the supernatant solutions transferred to 16 x 100-mm screw-cap tubes. A 1.0-mL aliquot of a saturated aqueous solution of sodium bicarbonate was added to each tube, and the tubes were then rotated end-overend at 20 rpm for 5 min and centrifuged for 5 rain. The su- pernatant solutions were transferred to 16 x 100-ram screwcap tubes and evaporated to dryness at 40-45~ under a stream of dry nitrogen. The residues remaining after evaporation were each dissolved in 0.50 ml of water, and 40 ~L was injected into the HPLC for analysis. HPLC analysis. The HPLC instrument consisted of a constant metering pump (Consta Metric III G, Milton Roy, Riviera Beach, FL) and autosampler (WISP model 712 B, Waters, Milford, MA). Ultraviolet absorption of the sample at 327 nm was measured with an ultraviolet absorption detector (Spectroflow model 773, Applied Biosystems, Ramsey, N J). The detector output was recorded on a reporting integrator and peak heights were reported (model 4270, Spectra Physics, San Jose, CA). Chromatograms were recorded at a chart speed of 0.25 cm/min. Separations were carried out on a 4.6 x 150-ram column (Zorbax, Mac-Mod Analytical, Inc., Chadds For.d, PA) packed with 5-1Jm RX-Cls chromatographic medium. The guard column (Uptight, Upchurch Scientific, Inc., Oak Harbor, WA) was packed with Jm Perisorb RP-18 chromatographic medium. The mobile phase was 0.03M phosphoric acid/methanol (35:65, v/v) at a flow rate of 1.0 ml/min. Under these conditions, retention times of flunixin and niflumic acid were approximately 6.50 and 10.4 min, respectively. Calibration curves. The ratios of the peak heights of flunixin and niflumic acid in urine calibrators were calculated and plotted against the corresponding concentration of flunixin. The slope, intercept, and correlation coefficient of each calibration curve were determined by linear regression analysis with equal weighting of the data. The concentrations of flunixin in test samples and quality control samples were calculated from the slope and intercept of the calibration curve. For those test samples found to contain flunixin at a concentration greater than 1000 ng/ml (i.e., the upper limit of the calibration curve), the sample was retested after diluting an aliquot of the test sample with drug-free horse urine to reduce the concentration to less than 1000 ng/ml. ELISA analysis of flunixin in urine samples Urine calibrators for flunixin ELISA. Urine calibrators were prepared by adding appropriate volumes of the stock solution of flunixin to drug-free horse urine to produce concentrations of 1.0, 10.0, 100, and 1000 ng/ml The calibrators were prepared and used immediately for ELBA analysis. Urine sample aliquots (negative control, flunixin calibrator, and test samples collected from horses) were diluted 1:9 with the system buffer provided in the test kit, and the test was performed on 20 IJL of the resuiting mixture according to the instructions provided by the manufacturer. Collection and analysis of data. Absorbances of calibrators and test samples were determined at 650 nm using a microplate reader (model EL 312, Bio-Tek Instruments, Inc., Winooski, VT). Apparent flunixin concentrations were calculated from a five-parameter logistic model of the calibrator data (StatLIA software, version , Brendan Scientific, Oak Park, MI). GC-MS analysis of flunixin in urine sample extracts Positive control samples for GC-MS analysis. Positive control samples were prepared by pipetting appropriate volumes 374

4 of flunixin stock and working standard solutions into negative control horse urine to produce concentrations of 10, 50, 100, 250, 500, 1000, 2500, 5000, and 10,000 ng/ml. Positive control samples were prepared and processed immediately. Extraction procedure. Aliquots (2 ml) of test samples and positive control samples were mixed with 1.0 ml of 0.1 N sodium hydroxide solution in 16 x 125-mm glass test tubes and allowed to stand for 10 min at room temperature. At the end of this time, the solutions were adjusted to ph 2 by the addition of 1.0M hydrochloric acid solution. Each solution was extracted twice with 5 ml of diethyl ether; the contents of the tubes were mixed by end-over-end rotation at 20 rpm for 5 rain and centrifuged at rpm for 5 min; the aqueous layers were removed by aspiration and discarded. The organic phases were decanted into 16 x 125-mm tubes; the second extract was combined with the first extract. The ether extracts were then washed with 1 ml of a saturated aqueous solution of sodium bicarbonate and evaporated to dryness under nitrogen on a water bath. Derivatization procedure. Residues of sample extracts were dissolved in 20 t~l of BSTFA; the tubes were capped and then heated at 60~ for 20 min. The reaction mixtures were allowed to cool to room temperature, and 20 ~L of ethyl acetate was added to each tube. The contents were transferred to autosampler vials containing a 100-1~L glass insert (Hewlett- Packard part # ) for GC-MS analysis. GC-MS instrument. The GC-MS system consisted of a capillary GC (Hewlett-Packard, model 5890) connected to a mass selective detector (Hewlett-Packard, model 5970). The GC was equipped with an autosampler, splitless injector, and a Microseal Septum (Merlin Instrument Company, Half Moon Bay, CA). The column was a 15 m x 0.25-mm i.d. DB-1 capillary column with a Jm film thickness (J&W Scientific, Folsom, CA). GC conditions were as follows: injection port temperature, 280~ initial column oven temperature, 150~ column oven program: temperature maintained at 150~ for Table I. Results of TLC Analysis of Extracts of Urine Samples Collected from Horses Administered Five Consecutive Daily IM or IV Doses of 1.1 mg of Flunixin as Banamine Solution per Kilogram Body Weight 1.0 rain after injection and increased at a rate of 20~ to 220~ and then maintained for 14 rain. Aliquots of I ~L were injected by autosampler injector using the splitless mode. The carrier gas was helium at a flow rate of 1.0 ml/min. Ions monitored for mass spectral analysis of flunixin as its mono-tms derivative were: , , and amu. The mass spectrometer was tuned and calibrated with PFTBA before each set of analyses. The mass spectrometer was operated in the selected ion monitoring mode under electron impact ionization conditions at 70 ev. Criteria for positive identification of flunixin. Ion area ratios for the three characteristic ions of the flunixin mono-tms derivative were calculated by dividing each ion area by the ion area of the most intense ion. The retention time and ion area ratios were compared with those for the flunixin mono-tms derivative from a urine flunixin calibrator processed and analyzed in sequence with the test samples. Positive identification of flunixin in extracts of test samples was reported if each ion ratio from the test sample was within 20% of the respective ion ratio from the calibrator and the retention time of each ion peak from the test sample was within 1.0% of the respective retention time from the calibrator. If the identification criteria were satisfied, then the sample was reported positive for flunixin. Internal Standard Flunixin Internal Standard Horse/Administration route IM IM IM IM IV IV IV IV Day 1 +* _t + Day Day Day Day Day Day Day *+ = Flunixin detected by TLC analysis. ~- = Flunixin not detected by TLC analysis. Time (min) Figure 2. HPLC chromatogram of an extract of a negative control urine sample (left) and that of a urine sample collected from a horse 3 days after the last of five daily IM doses of 1.1 mg flunixin as Banamine solution per kilogram bw (right). Retention times of flunixin and the internal standard were 6.93 and min, respectively. 375

5 Results TLC analysis for flunixin Flunixin was extracted from urine by a liquid-liquid extraction procedure and separated from other components of the extract on silica-gel TLC plates containing fluorescent indicators. The limit of detection of flunixin by this method was approximately 1.0 IJg/mL. Flunixin was readily detected (Table I) by TLC analysis of extracts of urine samples collected 24 h after the last IV dose and for 48 h after the last IM dose. Urine flunixin concentrations obtained 24 h after the last IM dose ranged from 5.56 to 13.0 IJg/mL and were less than the limit of detection in all urine samples collected 15 days after the last dose. Flunixin was detectable in two samples collected 9 days after the last IM dose. On the other hand, urine flunixin concentrations at the end of 24 h after the last IV dose ranged from 1.52 to 6.95 IJg/mL and were greater than the limit of detection in 2 of 4 urine samples collected 15 days after the last dose. Urine flunixin concentrations in all samples are reported in Table II. HPLC analysis for flunixin Flunixin was detected and measured in extracts of urine samples by a HPLC method. Typical chromatograms showing the results of analysis of an extract of a negative control urine sample and that of a urine sample collected three days after the last dose are shown in Figure 2. The limit of detection of flunixin by this HPLC method was approximately 50 ng/ml, based on a signal-to-noise ratio of 3. Table II. Concentrations of Free Plus Conjugated Flunixin (pg/ml) Determined by HPLC Analysis of Extracts of Urine Samples Collected from Horses Administered Five Consecutive Daily IM or IV Doses of 1.1 mg of Flunixin as Banamine Solution per Kilogram Body Weight* Horse/Administration route IM IM IM IM IV IV IV IV Day Day Day < Day Day I) < < Day l <0.0-; <0.050 {1.298 <0, Day <0, <0.050 <0.050 < , Day 15 <0.050 <0,050 <0.050 <0.050 <0,050 < * Limit of quant[tation = 50 ng/ml. Flunixin EI.ISA analysis Urine samples collected from horses were tested using a commercially available flunixin ELISA test kit. The limit of detection of the ELISA test kit for flunixin was approximately 15 ng/ml of urine, based on 1-50 data provided with the kit. The presence of flunixin (or flunixin metabolites) was detected by ELISA assay of all samples tested through 9 days after the dose and 6 of 8 samples collected 15 days after the dose. Results of ELISA tests performed on the samples from the two treatment groups are reported in Table III. GC-MS analysis for flunixin The electron-impact ionization mass spectrum of the mono-tms derivative of flunixin is shown in Figure 3. The molecular ion at m/z 368 and major fragment ions at m/z 353 and 263 were used to identify flunixin in extracts of urine samples. Flunixin was identified from extracts of 51 of 64 samples collected from horses through 15 days after IV and IM flunixin. Flunixin was identified from an extract of one sample collected 15 days after the IV doses, The limit of detection of flunixin was approximately 50 ng/ml under these conditions, and the results of analyses are reported in Table IV. Table III. Apparent Flunixin Concentrations (pg/ml) in Urine Samples Collected from Horses Administered Five Consecutive Daily IM or IV Doses of 1.1 mg of Flunixin as Banamine Solution per Kilogram Body Weight* Horse/Administration route IM IM IM IM IV IV IV IV Day 1 >10 >10 >10 > > Day > Day Day Day Day Day Day 15 < < * Tests were performed using flunixin ELISA. Discussion and Conclusions Flunixin was detected by TLC, flunixin ELISA, HPLC, and GC-MS methods for different detection periods after five, consecutive, daily IM or IV doses at a dosage of 1.1 mg of flunixin per kilogram of bw. The detection period for flunixin by TLC analysis was 24 h after IV administration and 48 h after IM administration. This time period was approximately equal to the time period during which flunixin has been shown to exert its clinical effects (9-11). Therefore, use of a TLC procedure to detect flunixin in extracts of horse urine permits detection of the drug during the time that it exerts its clinical effects. The iden- 376

6 tity of flunixin was readily confirmed by GC-MS in 100% (11 of 11) of the samples in which it was detected by the TLC method. Flunixin was detected by flunixin ELISA in 96.9% (62 of 64) of urine samples collected through 15 days after flunixin administration if the ELISA was performed without a threshold. The limit of detection of the flunixin ELISA was approximately 15 ng/ml and was much lower than that of the TLC, HPLC, and GC-MS methods. Therefore, use of the flunixin ELISA resulted in detection of flunixin or its metabolites in urine samples collected through 15 days after multiple IV and IM doses of flunixin and greatly extended the detection time relative to that when TLC methods were used. Use of this screening test therefore resulted in detection of flunixin beyond the time that it is believed to exert a clinical effect (9-11). The identity Table IV. Results of GC-MS Analysis of Extracts of Urine Samples Collected from Horses Administered Five Consecutive Daily IM or IV Doses of 1.1 mg of Flunixin as Banamine Solution per Kilogram Body Weight* Horse/Administration route IM IM IM IM IV IV IV IV Day 1 +* Day Day * + + Day Day Day Day Day * + = flunixin identified in extract of urine sample collected on day indicated after last dose. "~ - = flunixin not identified in extract of urine sample collected on day indicated after last dose. c t,,-.o r n m/z Figure 3. Full-scan electron-impact mass spectrum of the mono-tms derivative of flunixin I [...!~g..., of flunixin was confirmed by GC-MS analysis of extracts in 82.3% (51 of 62) of urine samples in which it was detected by flunixin ELISA at apparent flunixin concentrations greater than 15 ng/ml. If a threshold concentration of 100 ng/ml was used for the flunixin ELISA, then 82.8% (53 of 64) of the urine samples were positive by flunixin ELISA and the presence of flunixin was confirmed in 92.5% (49 of 53) of these samples by GC-MS. Flunixin was also readily detected by the HPLC method reported in this study. The limit of detection of the HPLC method for flunixin was approximately 50 ng/ml and was much lower than that of the TLC method but higher than that of the ELISA assay. Therefore, screening of extracts of urine samples for flunixin by HPLC methods substantially increased the detection period over that which was obtained by TLC methods. In fact, flunixin was detected in 75.0% (48 of 64) of the extracts of urine samples collected for 15 days after flunixin administration and in extracts of two of four urine samples collected 15 days after the last dose of the IV multiple-dose regimen. Use of this screening test, therefore, resulted in detection of flunixin beyond the time that it is believed to exert a clinical effect (9-11). Confirmation of the identity of flunixin by GC-MS analysis was readily accomplished in 97.9% (47 of 48) of urine samples in which flunixin was detected by HPLC analysis. The presence of flunixin in one sample in which it was detected by HPLC was not confirmed by GC-MS analysis. This agreement between methods was attributed to the comparable lower limits of detection. Apparent flunixin concentrations determined by flunixin ELISA were greater than concentrations of total flunixin determined by HPLC, suggesting that the immunoassay may have detected one or more flunixin metabolites not measured by HPLC. The only identified metabolites of flunixin are 5-hydroxyflunixin and conjugated flunixin (5,6). The total flunixin concentration determined by HPLC represented flunixin excreted in the urine as unchanged drug and its conjugates. On the other hand, 5-hydroxyflunixin was reported to be detectable for shorter times than flunixin (5), suggesting that this metabolite was not contributing to the apparent flunixin concentrations deter- 3s3 mined by the flunixin ELISA assay at later collection times. Thus, the apparent flunixin concentration may have been higher than the concentration of flunixin determined by HPLC because of cross-reactivity of conjugated flunixin with the ELISA assay. Flunixin was identified by GC-MS analysis of extracts of 88% (7 of 8) of urine samples 368 collected 9 days and 13% (1 of 8) of urine samples collected 15 days after the last dose of a multiple-dose regimen of flunixin. The limit of detection of the GC-MS method for flunixin was approximately 50 ng/ml and was similar to that of the HPLC method. This method was also suitable for confirming the identity of flunixin detected by the flunixin ELISA tests if a threshold concentration of 100 ng/ml was used for the flunixin ELISA

7 Table V. Detection Periods for Flunixin after Single- and Multiple-Dose Administration Dose Method of Detection (mg/kg) Route analysis time Reference IV TLC* 48to>216h IM GC-NPD 32 to 42 h PO, IV GC-ECD 48 h IV GC-MS 175 h IV GC-MS 7 to > days IM, IV Flunixin ELISA IV: >15 days this study multiple dose IM: >15 days 1.1 IM, IV TLC IV: 24 h this study multiple dose IM: 48 h 1.1 IM, IV HPLC IV: >15 days this study multiple dose IM: 9 days 1.1 IM, IV GC-MS lv: >15 days this study multiple dose IM: 9 days * Abbrevations: TLC = thin-layer chromatography; HPLC = high-performance liquid chromatography; GC-NPD = gas chromatography with nitrogenphosphorus detection; GC-ECD = gas chromatography with electron-capture detection; GC-MS = gas chromatography-mass spectrometry. The detection times reported in this study are generally equal to or longer than those previously reported from single dose studies (Table V). The longer detection times compared to those reported after a single-dose may be due to different lower limits of detection or to accumulation of flunixin during multiple-dose treatment. However, a previous study reported that accumulation did not occur during multiple-dose administration of flunixin (12). Because the limit of detection reported in one study (13) was lower than that of any previously published method as well as that reported in the current study, it is possible that even longer periods of detection are possible if this method of detection is used to test samples collected after multiple-dose administration. The method with a lower limit of detection was a GC-MS method with selected ion monitoring (13) as was that reported by us. However, the lower limit of detection previously reported (13) may have resulted from their use of no additional criteria such as peak-area ratios to confirm proof of identity. Because such criteria are now commonly required, it is possible that the limit of detection that they reported cannot be duplicated. The detection of flunixin after a has been reported previously by several groups of investigators (2,6). Flunixin could not be detected in extracts of urine by GC methods employing electron-capture detection for more than 48 h after a single oral or IV dose of 1.1 mg of flunixin as Banamine per kilogram of bw (2). An extractive alkylation method with GC separation and nitrogen-specific detection for determination of flunixin from equine urine was reported to have a lower limit of detection of 74 ng/ml (6). Flunixin was detected in extracts of urine samples collected through h after IM adminis- tration of 1.1 mg of flunixin per kilogram of bw to a standardbred gelding (6). Thus, most previous studies indicated detection of flunixin no longer than 48 h after a single IV or IM dose of 1.1 mg of flunixin per kilogram bw to horses. The only exceptions reported are studies in which flunixin was detected in extracts of horse urine collected 175 h after a of 1 mg of flunixin as Finadyne per kilogram bw (5) and one in which flunixin was detected from 48 h to more than 9 days after a of mg of flunixin as Banamine per kilogram bw (13). Both studies used GC-MS methods, and the limit of detection of the first method was not reported (5), whereas that of the second method was 5 ng/ml (13). Results of various studies of the detection times for flunixin after single and multiple doses are summarized in Table V. These studies have demonstrated that commercial flunixin ELISA, HPLC, and GC-MS methods detect flunixin in urine samples collected from horses after multiple doses of flunixin for a much longer period of time than the time during which clinical effects are observed. Furthermore, these studies show that multiple doses of flunixin may result in detection of flunixin for more than two weeks after the last dose. Analysts in racing commission laboratories use TLC, HPLC, and ELISA methods to screen samples for the presence of flunixin (7). The choice of screening method affects the detection time since limits of detection of these methods vary widely. Furthermore, analysts for racing commissions may change methods without notice to regulators or veterinarians. Therefore, racing regulators and veterinarians involved in race track practices should be aware that detection times may be as long as two weeks or more if the samples are screened for flunixin by ELISA or HPLC methods. References 1. l.w. Houdeshell and P.W. Hennessey. A new nonsteroidal antiinflammatory analgesic for horses. J. Equine Med. Surg. I : (1977). 2. S. Chay, W.E. Woods, T. Nugent, J.W. Blake, and T. Tobin. The pharmacology of nonsteroidal anti-inflammatory drugs in the horse: flunixin meglumine (Banamine). Equine Practice 4:16-23 (1982). 3. UR. Soma, E. Behrend, I. Rudy, and R. Sweeney. Disposition and excretion of flunixin meglumine in horses. Am. J. Vet. Res. 49: (1988). 4. S.D. Semrad, R.A. Sams, O.N. Harris, and S.M. Ashcraft. Effects of concurrent administration of phenylbutazone and flunixin meglumine on pharmacokinetic variables and in vitro generation of thromboxane 82 in mares. Am. J. Vet. Res. 54: (1993). 5. P. Jaussaud, D. Courtot, J.L. Guyot, and J. Paris. Identification of a flunixin metabolite in the horse by gas chromatography-mass spectrometry. J. Chromatogr. 423: (1987). 6. M. Johansson and E.-U Anl~r. Gas chromatographic analysis of flunixin in equine urine after extractive methylation. J. Chromatogr. 427:55-66 (1988). 7. T.M. Dyke and R.A. Sams. Analytical methods used by racing laboratories: an international survey. 10th International Conference of Racing Analysts and Veterinarians, Stockholm, Sweden, 1994, pp Edgar, T. Tobin, H.H. Tai, S. Bass, ).T. Ice, and D.E. Schroedter. 378

8 ELISA assay for flunixin. 11th International Conference of Racing Analysts and Veterinarians, Queensland, Australia, 1996, pp P. Lees, A.J. Higgins, A.D. Sedgewick, and S.E. May. Applications of equine models of acute inflammation. Vet. Rec. 120" (1987). 10. S.D. Semrad, G.E. Hardee, M.M. Hardee, and J.N. Moore. Flunixin meglumine given in small doses; pharmacokinetics and prostaglandin inhibition in healthy horses. Am. J. Vet. Res. 46: (1985). 11. P.L. Toutain, A. Autefage, C. Legrand, and M. Alvinerie. Plasma concentrations and therapeutic efficacy of phenylbutazone and flunixin meglumine in the horse: pharmacokinetic/pharmacodynamic modelling. J. Vet. Pharmacol. Ther. 17" (1994). 12. J. Rudy, J. Fegley, L.R. Soma, and C. Uboh. Effects of multiple doses of flunixin on plasma and urinary concentrations in horses. Proceedings of the 7th International Conference of Racing Analysts and Veterinarians, Louisville, KY, 1988, pp D. Crone and D.K.K. Leung. Flunixin administration, unpublished results, Hong Kong Jockey Club, Hong Kong. Manuscript received August 3, 1998; revision received December 28,