Journal of Analytical Toxicology,Vol. 29, March 2005 Multi-Residue Determination of Anti-Inflammatory Analgesicsin Sera by Liquid ChromatographyMass Spectrometry Irina RudikMiksa*, Margaret R. Cummings, and Robert H. Poppenga University of Pennsylvania, School of VeterinaryMedicine, Departmentof Pathobiology, New Bolton Center, Toxicology, 382 WestStreet Road, Kennett Square, Pennsylvania 19348 Abstract Non-steroidal anti-inflammatories (NSAIDs) are analgesic, antipyretic, and, as their name implies, anti-inflammatory drugs, which are widely used for the treatment of a variety of human and veterinary disease conditions in which control of pain and inflammation is desired. Acetaminophen (ACE) is a common over-the-counter analgesic. Detection of a variety of widely used NSAIDs and ACE in fluid and tissue samples is an important diagnostic tool. A sensitive and selective analytical method has been developed for simultaneous screening of 12 NSAIDs and ACE by liquid chromatographymass spectrometry with an atmospheric pressure chemical ionization interface set to operate in the negative ion mode of MS. Following sample preparation, all analytes were separated on a C18-reversed.phase column with a gradient elution of acetonitrUe and acetic acid. Full-scan mass spectral fragmentation profiles were established for each analyte and individual extracted ion chromatograms were used for quantitation. Linearity of detection was observed over the 0.05-25.0 pg/mL range of standard concentrations. The instrument limits of detection (LOD), based on an individual analyte quantitation ions, fell between 0.05 and 1.0 pg/mL for all compounds. The matrix LODs were determined to be 0.05 pg/mL for phenylbutazone (m/z 307); 0.1 pg/mL for indomethacin (m/z 312), flunixin (m/z 295), and piroxicam (m/z 330); 0.5 pg/mL for ACE (m/z150), diclofenac (m/z 250), ketoprofen (m/z 209), and mefenamic acid (m/z240); 1.0 pg/mL for oxyphenbutazone (m/z 323); 5.0 pg/mL for ibuprofen (m/z 205), salicylic acid (m/z 137), and tolmetin (m/z 212); and 10 pg/mL for naproxen (m/z 185). Introduction Non-steroidal anti-inflammatory drugs (NSAIDs)comprise a large group of therapeutic agents with anti-inflammatory, anal. Author to whom correspondence should be addressed. E-mail: [email protected]. gesic, and antipyretic actions. The pharmacologic effects of NSAIDs are primarily due to inhibition of cyclooxygenase (COX), which is responsible for the biosynthesis of prostaglandins and other related autocoids (1). Acetaminophen (ACE) also has analgesic and antipyretic actions, but achieves these effects via a different site of action than COX and is not classified as an NSAID.ACE is believed to interfere with the endoperoxide intermediates PGG2and PGH2 (2). NSAIDsand ACE are among the most commonly used over-the-counter preparations for treatment of osteoarthritis and other rheumatic diseases in human and veterinary medicine. The utilization of certain NSAIDshas also been correlated with lowering the risk of Alzheimer's desease in humans (3-5). Although generally considered safe, acute overdose and chronic use of NSAIDs are associated with significant adverse effects on the gastrointestinal, hematopoietic, and renal systems (1). Chinese patent medicines have been adulterated with NSAIDs (6) such as indomethacin, phenylbutazone, mefenamic acid, and diclofenac. In addition, several NSAIDs, such as flunixin and phenylbutazone, have been used unethically in performance horses to mask signs of inflammation and pain. ACE is hepatotoxic, and acute intoxications are common and can result from overdosage of pediatric formulations or suicide attempts (7). In veterinary medicine, cats are uniquely sensitive to the toxic effects of ACE because of their inability to glucuronidate the drug (8). ACE intoxication in cats is common because of its administration by well-intentioned owners unaware of its toxicity. For these reasons, development of an accurate, sensitive, and selective method for detection of NSAIDs in biological samples has been the goal of a number of clinical and toxicology laboratories NSAIDsare a chemically diverse group of drugs that includes salicylic acid derivatives (e.g., aspirin), para-aminophenol derivatives (e.g., ACE), indolacetic acids (e.g., indomethacin, sulindac, etodolac), arylproprionic acids (e.g., ibuprofen, naproxen, ketoprofen, carprofen), fenamates (e.g., meclofenamic acid), pyrazolones (e.g., phenylbutazone, dipyrone), enolic acids (e.g., piroxicam, tenoxicam), and heteroaryl Reproduction(photocopying)of editorialcontentof thisjournalis prohibitedwithoutpublisher'spermission. 95 Journal of Analytical Toxicology, Vol. 29, March 2005 acetic acids (e.g., diclofenac) (1). Because NSAIDs and ACE comprise a heterogeneous group of compounds, most available detection methods are optimized to detect a single drug and one or more of its metabolites. Available reports suggest that liquid chromatography-mass spectrometry (LC-MS) and LC with UV detection in conjunction with solid-phase extraction (SPE) are the most commonly used methods for detection and quantitation of single NSAIDs in biological samples (9-23). Other single-drug methods include gas chromatography-mass spectrometry (GC-MS) (24,25), spectrofluorimetric (26) and spectrophotometric methods (27), capillary electrophoresis (23,28-31), and thin-layer chromatography (32). Several multi-drug screening procedures have been developed for groups of NSAIDs based on their chemical properties. These include detection of salicylates (33-35), fenamates (36), naproxen and other naphthalene derivatives (37), anthranilic (38-40), and lecithin-based derivatives (41). Available multi-residue screening methods for NSAIDsinclude high-performance liquid chromatography (HPLC) with electrothermal or IN detection for a group of 3-16 different analytes present in a single sample with reported limits of detection (LODs) ranging from 0.01 to 3.6 lag/mL (42--45). Successful screening of up to 21 NSAIDsand their metabolites has been reported using a variety of GC-MS procedures (46-48). Difficulties in optimization of an HPLC method for phenylbutazone and mefenamic acid and an inability to detect oxyphenbutazone have been reported (42). An involved gradient profile, a combination of organic solvents, or a buffering system are usually required in HPLC procedures to achieve the best peak separation and analyte identification. GC-MS methods, although useful for screening purposes in clinical and forensic toxicology (LODsrange 0.005-0.05 tag/mL),prove laborious because of the need for sample derivatization with methyl iodide to improve chromatographic properties (46,48). A multi-analyte method for separation of diclofenac, ketoprofen, indomethacin, and flunixin present in the same sample has recently been reported for an LC-MS system with an atmospheric pressure chemical ionization (APCI) interface set to operate in the negative ion mode (49). The use of APCI is advantageous in that it can transform fragile thermolabile species into ions without decomposition, although sometimes fragment ions are absent. The method developed for the four NSAIDsshowed that LC-MS affords high analyte specificity and short analysis time. The use of a non-routine CN-column and an organic/buffer mobile phase for elution and deprotonation of compounds might imply difficulties with such an approach. This article describes a simple, rapid, and reproducible multiresidue screening method for simultaneous detection of 12 NSAIDs and ACE in serum. Sample preparation omits any SPE columns, and drug elution is achieved on an C18reversed-phase LC column without the need for derivatization. Under the method conditions, full-scan mass spectral flagmentation patterns, obtained in the negative ion mode of APCI, are reported for each compound. Accurate identification and quantitation of all analytes are achieved over a wide range of concentrations without the need for complete chromatographic resolution. 96 Material and Methods Materials All pure analyte standards/solutions, except for flunixin, and bovine adult serum (sterile-filtered) were purchased from Sigma (St. Louis, MO). Flunixin was a gift from ScheringPlough Corp (Kenilworth, NJ). Acetonitrile (ACN), methanol (MET), sodium chloride (NaCl), and glacial acetic acid were obtained from Fisher Scientific (Pittsburgh, PA). All reagents were HPLC grade. An ultra pure water system from Millipore (Billerica, MA) was used to generate water with resistivity of 18 M~.cm (Milli-Q RG and Milli-RO10). LC autosampler vials (2 mL, wide opening) designed for a robotic arm tray with 250-1JL, high-recovery,fiat-bottom inserts were purchased from Agilent Technologies (Wilmington, DE). Preparation of standards Individual stock standards of 1000 lag/mL for flunixin, indomethacin, ketoprofen, mefenamic acid, oxyphenbutazone, phenylbutazone, and piroxicam were prepared by dissolving 10 mg of each compound in 10 mL ofACN. Naproxen and tolmetin 1000 lJg/mL standards were made with 50% ACN/water, and diclofenac standard was placed in MET.Standards for ACE, ibuprofen, and salicylic acid were purchased as solutions at 1000 l~g/mLin MET.Two 100 l~g/mLstandard mixes (A and B) were prepared. Mix A was made by combining 100 lJL of individual 1000 lJg/mL standards of ACE, salicylic acid, piroxicam, tolmetin, ketoprofen, naproxen, and diclofenac with 300 pL of ACN, and mix B was prepared by addition of 100 IJL of 1000 IJg/mL of oxyphenbutazone, phenylbutazone, flunixin, indomethacin, ibuprofen, and mefenamic acid to 400 IJL of ACN. Working standards were made by adding either 250 or 100 pL of each A and B mix to 500 IJL or 800 t~Lof ACN/water (60:40) solution for 25 and 10 IJg/mL standard mixes, respectively.All other standards were made by serial dilution of the 25 and 10 tJg/mL working standard mixes with ACN/water. All samples were stored at 4~ when not in use. Sample handling Negative control bovine serum, not containing any detectable amounts of NSAIDsor ACE, was analyzed prior to analyte fortification. Aliquots of negative control serum (0.4 mL) were spiked with standard mix solutions at 0.05, 0.10, 0.50, 1.0, 2.0, 5.0, 10.0, and 20.0 IJg/mL levels. Each fortified sample was combined with 0.05 g of NaC1and 0.8 mL of ACN.All samples were vortex mixed for 15 s and centrifuged for 10 rain at 14,000 rpm. Clear supernatants (0.5 mL) were placed into an autosampler vial and analyzed by LC-MS. LC-MS conditions A series 1100 LC-MS Hewlett-Packard system (Wilmington, DE) was equipped with a dual LC pump, a diode-array detector, a degasser, a column thermostat, and an autosampler. A Betasil C18-reversed-phase column (150 x 4.6 ram, 5-pro particle size) from Keystone Scientific (Bellefonte, PA)was equilibrated with the mobile phase (50% ACN and 50% of 0.25% acetic acid in water) at 0.5 mL/min prior to analysis. The LC profiles were developed at 40~ with a mobile phase and flow rate gradients. Journal of Analytical Toxicology, Vol. 29, March 2005 The following 17-rain program was used: 50% ACN was pumped at 0.5 mL/min for 5 rain, increased to 80% ACN over 5 rain at 0.7 mL/min, held for 5 rain, then reduced to 50% ACN over the next minute, and the column re-equilibrated for 1.0 rain. Each assay was based on the 20-1JL injections. MS was set to operate in a negative mode of APCI, detection was set with the 2-min time delay, gain at 10, fragmentor at 70, threshold of 50, and a step-size of 0.1. A full-scan MS mode was used (m/z 115-200 and rn/z 150-400) in order to obtain a complete representation of the total ion chromatogram for each compound analyzed. To optimize for ACE and salicylic acid during NSAID standard mix analyses, mass-to-charge ratio ion range of 115-200 was followed between 2 and 5.2 rain, with the rest of the compounds monitored (m/z 150--400) between 5.2 and 14.5 rain. MS was optimized for the following settings: dry gas flow rate of 3 mL/min, corona at 20 IJA,nebulizer pressure of 50 psi nitrogen, dry gas and vapor temperatures of 300~ and capillary voltage of 2500V. Linearity, precision, and instrument limit of detection (ILOD) Linearity of instrument standard response was determined for all analgesics over the range of concentrations: 25.0, 10.0, 5.0, 1.0, 0.50, 0.25, 0.10, and 0.05 I~g/mL. Individual calibration curves were constructed for every analyte based on peak areas under the corresponding Table I. Chromatographic and Fragmentation Behavior of NSAIDs and ACE individual extracted current-ion profiles. A plot Iont (%) RT* TRw of concentration versus peak area resulted in a Name MW* (m/z) Abundance (min) (pg/mL) linear curve for individual drugs based on triplicate analysis (n = 3) at corresponding exACE 151.2 150# 100 3.20 28 (canine) tracted ion profiles. This resulted in average correlation coefficients (R2) between 0.9930 Diclofenac 318.1'* 250# 100 12.1 2.5 (humans,150 rag)*t 294 55 and 0.9998 for each standard curve repeated in triplicate for all mass-to-charge ratios listed in Flunixin 296.3 295# 100 11.2 1.1-2.2 (bovine) Table I (quantitation and confirmation ions). 303 35 0.25-1 (canine) The percent standard deviation (%SD = 251 15 [square root of {nSx2 - (Sx)2/n(n - 1)}]*100%, Ibuprofen 206.3 161 100 12.7 12-15 (humans) where n is the number of measured values for 10 (canine) 205# 85 particular measurements, x) for all the curves Indomethacin 357.8 312# 100 12.0 was between 0.005 and 0.514%, indicating that 0.5-3.0 (humans) 356 30 detection was linear from 0.05 to 25.0 l~g/mL. Precision of standard mixes and fortified Ketoprofen 254.3 209~ 100 9.30 1.1 (equine) serum was determined by three consecutive 253 10 injections (n = 3) of 20-1JL aliquots of the 1.0 197 10 ~g/mL standard mix and 20 ~g/mL serum Mefenamic 241.3 240# 100 13.4 10-20 (humans, lg)*t spikes. Peak area of each extracted ion was esacid 196 10 tablished by a Hewlett-Packard integration alNaproxen 230.3 185# I00 9.60 2-5 (canine) gorithm and quantitation of repeatable 229 10 10 (equine) injections was evaluated for % recovery based 170 50 on calibration curves constructed between 0.05 and 25 lJg/mL. Data were observed to be Oxyphen324.4 323# 100 9.60 Not available reproducible and %SD values fell below 10% butazone 280 15 (n = 3) for ACE, naproxen, phenylbutazone, 204 25 salicylic acid, ketoprofen, oxyphenbutazone, Phenyl308.4 307# 100 12.9 5-10 (canine) and piroxicam in a standard mix for all quanbutazone 1-4.4 (equine) titation ions (Table I). The %SD values for di10-20 (bovine) clofenac, flunixin, ibuprofen, indomethacin, Piroxicam 331.4 330# 100 8.10 0.3-1.0 (canine) mefenamic acid, and tolmetin in a standard 266 25 mix fell between 10.2 and 14.4%. For more 210 5 precise quantitation of a 1.0 IJg/mL standard, Salicylic acid 138.1 137# 100 4.95 < 20% (humans, topical) calibration curves based on lower standards (0.05-5 lJg/mL) alone were constructed for every analyte at its ion of quantitation (Table Tolmetin 257.3 212# 100 8.30 40 (humans, 400 rag)t* I). A range of percent recoveries observed was * Molecular weight. between 79 and 93% with %SD range of ~"All available mass fragment ions observed under method's conditions. 1.0-14.7 for all analytes. Quantitation of all * Retention time. Therapeutic range (2,7). analytes in a 201Jg/mL fortified serum repeat# Ion of quantitation. able injections was evaluated for all ions listed ** DIC sodium complex. tf Therapeutic range has not been established. Reported are the peak plasma concentrations at highest oral with mass-to-charge ratios in Table I. The dosage. %SD values fell between 2.6 and 19.3% based 97 Journal of Analytical Toxicology, Vol. 29, March 2005 on all quantitation ions and 5.8 and 16.3% for all confirmation ions listed with mass-to-charge ratios. Between-day variability was evaluated for a 10 pg/mL fortified serum spike based on a three-day minimum design. The %SD of a three-day analysis for all ions listed with mass-to-charge ratios, except for indomethacin (Table I), was 1.2-10.0%. Detection of indomethacin showed slightly higher between-day variation with %SD of 13.2 and 13.8% at m/z 312 and 356, respectively. ILOD was defined as the lowest concentration of each analyte present in a standard mix that could be detected and expressed as a concentration at signal-to-noise ratio (S/N) of 3:1 based on the individual extracted-ion chromatograms. Analyte recovery, LOD, and limit of quantitation (LOQ) Extraction efficiency for every NSAID and ACE was evaluated at 8 fortified serum concentrations in replicates of 4 (n = 4). The amount of drug extracted during the purification procedure as compared to the concentration used to fortify the sample was expressed in terms of percent drug recovered evaluated at each ion listed with mass-to-charge ratio. Individual extracted ion (re~z) peak areas were plotted onto individual A Results Chromatography and MS fragmentation A complete chromatographic and fragmentation pattern was collected for all 12 individual 10 pg/mL NSAID and ACE standards under given conditions in order to establish the correct peak assignments and mass spectral behavior of each analyte when present in a mix. A different degree of fragmentation was seen for the 13 drugs analyzed. The full-mass spectral fragmentation profile for ACE, phenylbutazone, and salicylic acid MEFA B IBU 161.2 100 standard calibration curves, and the corresponding concentration of each drug in fortified samples was determined. The ratio of two concentrations (recovered vs. fortified) times 100% was used to report % drug recovery. Sample LODs were defined as the lowest concentration of each analyte that can be detected in a fortified serum sample at 3:1 S/N. The lower sample LOQ was defined as the lowest measured concentration (based on LOD) of each analyte that can be recovered from the fortified serum sample within 20% of true value based on an individual ion of quantitation (Table I). 100 240.1 [M-H] 205.0 [M-H] 80 80NH ~ 6o "1= = 60 < ,.o ,< 40 OH 40 m/c 161 20 OH 2o-~ d 2~ ,,., ,1!,I,I!~.,,,,,h ,,i,!,1h!H I,h,. ......=.......,m,~,.~...,.=,~,,,,I.............~, ,:.,,;, 0 t ........... ,l 175 200 m/z 250 C 100 m/z 225 200 DIC H 2C~--I~ O H D 100 = 60 irdZ 312 311.6 o H20/LIOHI--U [ND 80 8O r 250 m/z - O II 249.8 I! 96.2 250 252,1 293.7[M-H]" .~: 60 <~'=:40 I c=o 313.9 [~ <~ 40 I 96.1 20 J I I ~, .............................. 260 m/~ IIJ, 280 I 20 3SS.8[M-H]I cl 0 r .... 320 m/Z [ ' , r I ,I 340 Figure I. Full-scan mass spectrum profiles generated by ibuprofen, mefenamic acid, didofenac, and indomethacin (5 pg/mL) in MS with negative mode of APCI, Fragmentation of ibuprofen (A), diclofenac (C), and indomethacin (D) generated molecular species (all at 100% abundance) corresponding to the loss of one CO2 molecule from each parent ion resulting m/z 161,2.50, and 312, respectively. An ion at m/z 196 (10% abundance) was seen for the loss of CO2 from the parental structure of mefenamic acid (B), Deprotonated molecular ion was seen for every compound ([M-H]-), Complete chemical structure, fragment generation, and the resulting mass-to-charge ratio values are shown for every drug in their corresponding panels, 98 Journal of Analytical Toxicology, Vol. 29, March 2005 under negative APCI resulted in generation of molecular ions only ([M-H]-]) at m/z 150, 307, and 137, respectively (data not shown). Fragmentation spectra for tolmetin resulted in a single ion at m/z 212 corresponding to the loss of CO2from the molecular structure (not shown), ([M-H]-) ion for tolmetin at m/z of 256 was not detected. Fragmentation of ibuprofen, mefenamic acid, diclofenac, and indomethacin present in a 5 IJg/mL standard mix resulted in two mass-to-charge ratio ions for each molecule (Figure 1). In addition to the [M-H]- ions for each of these drugs at m/z 205 (85% abundance, ibuprofen), 240 (100% abundance, mefenamic acid), 294 (55% abundance, diclofenac), and 356 (30% abundance, indomethacin), further fragmentation resulted in the loss of one CO2 unit from each of ibuprofen (m/z 161), mefenamic acid (m/z 196), diclofenac (m/z 250), and indomethacin (m/z 312) as shown in Figures 1A-1D, respectively. The five remaining NSAIDsproduced three possible detectable ions upon fragmentation (Figure 2). Each compound resulted in [M-H]- ion at m/z 330, 323, 253, 229, and 295 for piroxicam, oxyphenbutazone, ketoprofen, naproxen, and flunixin, respectively. Loss of one CO2 unit from ketoprofen, naproxen, and flunixin structures generated ions at m/z 209, 185, and 251, respectively. Further fragmentation of ketoprofen (Figure 2C) and naproxen (Figure 2D) resulted in the additional loss of a methyl unit ([[M-H]-(CO2)-(CH3]]-) and produced ions at m/z 197 (10% abundance; ketoprofen) and 170 (50% abundance; A naproxen). The third detectable ion of flunixin at m/z 303 (Figure 2E) could be attributed to the formation of a mixed anhydride between the parental structure and the acetic acid present in the mobile phase as proposed in Figure 3A. The loss of two fluoride units is not surprising because fluoride is considered to be a 'good' leaving group. As in the case of flunixin, pattern of fragmentation of piroxicam generated three ions, formation of one of which could be attributed to the presence of the acetic acid in the mobile phase. The observed ion at m/z 210 in the fragmentation pattern of piroxicam can be explained by the loss of (O=C-NH-CsH4N)unit from the parental structure as shown in Figure 2A. Further loss of H20, one CH3 unit, and a reaction between the newly formed piroxicam fragment and the acetic acid would lead to a new and stable mixed anhydride with m/z 266 as suggested in Figure 3B. The two additional ions observed for oxyphenbutazone were possibly due to the initial loss of a 3-carbon unit (CH3-CH2-CH2)from the parental ion resuiting with m/z 280 followed by the loss of the phenyl group (C6H6)with the final m/z 204 (25% abundance) as proposed in Figure 2B. A complete summary of elution times for each individual drug present in a standard mix, their corresponding mass-to-charge ratio values and mass spectral abundances are reported in Table I. Multi-residue detection and ILOD Mass spectral data for a mixture of 12 NSAIDsand ACE was OH 329,7 0 /H2 H2 --C --CH 3 ~176 9 80 21o 60 CHz;C 100- 8O r- m/z 280 B [M-H]- 100 322.7 [M-H]" u '~ 60 2 .D < 40 .....2,!0 ' 250m/Z m/z 1 9 7 - ~ =) 80 200 184.8 100- '~60 < < 40 170.1 ~, m/z 185 3 c ~ ~H H 40 20 253.2 " 0 229~-H]" . 200 225 m/z 250 150 J. . . . 200 7 _300 o! [oo 294.7 H~C CF3 [M-HI" ~ 40 j m/z251 303.2 'o' ............II........ m& 1 20 m/z El00 1 OCH3 i_ ~ ,_%7-0 ,~ 6o I 360 D 209.0 204.1 20" .................................................................................... 200 ClO o 4O 266.1 20 I 20 i 251 L m/z 250 275 m/z 300 Figure 2. Complete MS fragmentation data obtained for piroxicam, oxyphenbutazone, ketoprofen, naproxen, and flunixin in a full-scan negative monitoring mode of APCI. All spectra were collected for 5 tJg/mL individual standards under identical conditions. Structures of parental compounds and MS generated daughter ions corresponding to the detected mass-to-chargeratio ion fragments are shown by arrows in corresponding panels. [M-H]- ion was found to be most abundant (100%) for piroxicam (A), oxyphenbutazone (B), and flunixin (E). Lossof one CO2 unit generateddaughter ions at 100% abundance for ketoprofen (m/z 209) (C) and naproxen (m/z185) (D). 99 Journal of Analytical Toxicology, Vol. 29, March 200g collected for a range of standard concentrations (0.05 to 25.0 pg/mL) in a full-scan mode of MS with negative APCI (m/z 115-400). The total ion chromatogram for a 5 pg/mL standard mix resulted in 11 well resolved MS peaks as shown in Figure 4. Nine out of 12 NSAIDs compounds were well separated and showed good chromatographic behavior. All available peaks were assigned to the individual analyte with observed co-elution of oxyphenbutazone and naproxen at 9.5 min and co-elution of indomethacin and diclofenac at 12 rain. For unambiguous identification and quantitation of each compound present in a mixture, extracted current-ion profiles for all analytes were obtained for every available mass-to-charge ratio mass fragment. Chromatographic co-elution of diclofenac and indomethacin could be resolved based on the two non-shared individual extracted ions. Indomethacin fragmentation resulted in ions at m/z 356 and 312, not shared by diclofenac with mass-to-charge ratio fragments of 294 and 250 (data not shown). The same reasoning was applied to oxyphenbutazone and naproxen where three different ions can be used for identification and confirmation of each analyte (Table I). In the case of diclofenac, the two characteristic ions at m/z 253 and 251 (protonation products ofm/z 250) appear in the extracted current ion profiles of ketoprofen and flunixin (data not shown). However, it is clear that elution of ketoprofen and flunixin mass spectral peaks occur 1-2 min earlier as compared to diclofenac. Thus, clean peak separation allows for correct mass-to-charge ratio assignments at the two ions m/z 253 and 251, rendering them useful for confirmation of ketoprofen and flunixin in a sample. Extracted-ion chromatograms for m/z 210 (piroxicam) and m/z 303 (flunixin) was problematic at or below 5 pg/mL, suggesting that their presence cannot be used in confirmation of piroxicam and flunixin. ILODs were determined for all compounds present in a standard mix based on individual quantiation ions (Table I) with values of 0.05 pg/mL for indomethacin, phenylbutazone, and piroxicam; 0.1 pg/mL for diclofenac, flunixin, ketoprofen, mefenamic acid, and oxyphenbutazone; 0.5 pg/mL for ACE, ibuprofen, salicylic acid, and tolmetin; and 1.0 pg/mL for naproxen. This range of ILODs values (0.05-1.0 pg/mL) falls well below the established therapeutic range for all the analytes as summarized in Table I. Although higher ILODs were obtained for the available confirmation ions with the range of 0.5-5.0 pg/mL, the presence of most analytes could still be confirmed at or below the individual therapeutic ranges. The list of all the available therapeutic ranges for 13 different analytes, their mass-to-charge ratio quantitation ions, and all available confirmation ions are listed in Table I. Detection of analytes in fortified sera and matrix LOD and LOQ Negative control bovine serum was fortified at 8 different concentrations of a standard mix of 12 NSAIDs and ACE: 0.05, 0.10, 0.50, 1.0, 2.0, 5.0, 10.0, and 20.0 pg/mL. Samples were prepared and analyzed as described in Methods. Total mass spectral profile was obtained for a negative serum sample and for a 20 pg/mL serum spike (data not shown). No characteristic ions of possible analytes were observed in the total MS profile of the drug-free sample at the corresponding reten- A o mc\ __N m/z __N Cn3On O o , H~C I~C o//\\o 0 !1 296 /cH~ H CH2F re~Z238 O O / CFa H N~ " G--O--C--CH m/z 252 II II o o m/z 94 a o. m/z o O "~ 302 0 0 m/z 266 Figure 3. Proposedchemical reactionsfor generation of observed mass-to-chargeratio daughter ions of flunixin (A) and piroxicam (B) under chromatographic conditions with negative ionization mode of APCI. Parental structure of flunixin can undergo two different modification pathways, generating ion m/z 252 for the loss of COx or reacting with the acetic acid molecule prior to the lossof CO2 forming a mixed anhydride at m/z 302 (A). Piroxicam (B) can undergo one proposed type of the chemical modification, loosing a CH~OH unit and reacting with the acetic acid molecule leading to the loss of a phenyl amine structure (m/z 94). 100 Journal of Analytical Toxicology, Vol. 29, March 2005 tion times. The total MS ion chromatogram of a fortified sample contained 7 out of 11 analyte peaks that were clearly assigned to the specific analgesics based on a response of a standard mix (Figure 4). This demonstrates that the peak behavior and the elution patterns of most compounds are similar to a standard mix. Peak assignment was not possible for salicylic acid and ACE in the total MS of the fortified sample. The loss of clear resolution between piroxicam and tolmetin, and ketoprofen, oxyphenbutazone and naproxen was observed and showed the effect of matrix interference. However, mass spectral comparison of a fortified sample to a control serum demonstrated the specificity of the method. Difficulties in peak assignment for fortified samples were overcome by the data obtained in the extracted current-ion profiles of all analytes. Figure 5 shows the absence of matrix interferences and the clean peak resolution observed in the extracted single current-ion profiles for 13 analytes at the corresponding quantitation ions (Table I). '~ x 10es ,Nos.~c 35 30 PBU't ~ : : PIR OPSUTR4Ap FLU KET Extracted ion current profiles for the confirmation ions were the same (data not shown) and could be used for verification of drug identity. Sample analyte recoveries and the corresponding %SD values were determined for n = 4 replicates over a range of concentrations, 0.05-25.0 lag/mL (data not shown). All analgesics in fortified samples were quantitated based on extracted single current-ion profiles at corresponding mass-to-charge ratio values (Table I). Matrix LODs were determined for all available mass-to-charge ratio ions, expressed in units of concentration and defined as the lowest ion signal seen for a spiked serum at S/N 3:1. LOQs were based on all data as described in Methods. Values of LODs and LOQs are reported for all available mass-tocharge ratio ions and summarized in Table II. As seen in Tables I and II, most of the analgesics analyzed could be detected, quantitated, and confirmed when present in a fortified serum at or below their corresponding therapeutic ranges. Confirmation above therapeutic levels could be obtained for ibuprofen, ketoprofen, and naproxen. Average recoveries at individual LODs for all analytes (n = 4 replicate analysis) were expressed in % recovery (• %SD and summarized in Table II. MEFA Discussion 15 10 The development of an analytical method using LC-MS with negative APCI for simultaneous separation and identification of 0 13 analgesics is reported. All compounds were separated using 4 6 8 10 12 14 Time (min) a traditional C18 column and an organic/acetic acid/water moFigure 4. Total ion full-scan mass spectral profile for a 5 pg/mL mix of 12 bile phase. APCIwas chosen for this multi-residue screening beNSAIDs and ACE standards analyzed with negative APCl. All peaks are cause of its ability for sensitive determination of analytes with labeledwitha 3-lettercodeas designatedforeachdrugdescribedin this moderate polarity and molecular mass. In the case of the 12 study.Oxyphenbutazone/naproxenand indomethacin/diclofenacshows NSAIDs and ACE included in this method, the mass-to-charge the co-elution of thesetwo pairs of compounds at 9.6 and 12 rain, reratios for [M-HI- ions ranged between 138 and 356 (Table I). spectively. Full-scan mass spectra examined by means of negative APCI under the conditions of the method produced deprotonated molecular ions ([M-H]-) for all el0 ~ el0 ~ the compounds except for tolmetin. Fragmen,n/z 150 m/z 240 [ tational data was specific for each drug without 50 6 8 10 12 14 16 recourse to HPLC separation allowing clear 4 6 $ 10 12 14 16 ndz 185 .' identification of individual components of the m/z 250 I', 200 i;, J 4 6 8 10 12 14 16 mixture. Detection and quantitation of 12 4 6 $ 10 12 14 16 i m/z 323 /l NSAIDs and ACE was based on individual ex500 J m/z 295 ': tracted ion chromatograms and identification 4 6 g I0 12 14 16 o i i- i , o by their fragment-ion spectra. The most com,' ~ L ''Vz 330 4 6 8 10 12 14 16 monly generated fragments for NSAIDs were m/z 205 20 "g ndz 137 those resulting from the loss of one CO2 unit as 50 [ ,' 4 6 g 10 12 14 16 observed for ibuprofen, mefenamic acid, dim/z 312 ;, clofenac, indomethacin, flunixin, naproxen, ke] m/z 212 toprofen, and tolmetin. The loss of a methyl 4 6 g 10 12 14 16 4 6 8 10 12 14 16 group generated a third fragment ion for i m/: 209 "=r 20000 L ~ z307 . . . . I00 i naproxen and ketoprofen (Figure 2). The pres4 6 g 10 12 14 16 4 6 8 10 12 14 16 ence of the acetic acid in the mobile phase for Time (min) Time (rain) LC elution did not pose any problems in ion Figure 5. Extracted single current-ion profiles obtained in a negative mode of APCI for a fortified generation or interpretation. Addition of acetic serum sample at 10 pg/mL. Chromatograms of 13 possible quantitation ions are shown at their acid across parental structure was seen only corresponding mass-to-charge ratios. No matrix interferences were observed. For analyte idenfor flunixin and piroxicam with suggested tification, refer to Table I. mechanisms shown in Figure 3. In compar5 101 Journal of Analytical Toxicology, Vol. 29, March 2005 ison with the previously published data on multi-residue detection of diclofenac,flunixin, indomethacin, and ketoprofen by negative APCI (49), the present paper shows that the highest abundance can be achieved for ions other than [M-H]-, especially in the case of diclofenac, indomethacin, and ketoprofen (see Table I), and an [M-H]- ion can be observed for flunixin under present conditions. [t is often suggested that the MS detector should be set to operate in the selected ion monitoring mode (SIM)in order to improve sensitivityand specificityof the method. Running analysis in a full-scan monitoring mode is known to produce less sensitive LODs as compared to SIM. However, detection of 12 NSAIDs and ACE by the present method was shown to be reproducible and specific for each analyte. The chromatography allowed optimization of the low molecular weight compounds (ACEand salicylicacid)with detection of two ranges of mass-tocharge ratio ions at different time ranges, m/z 115-200 from 2 to 5.2 rain and rn/z 150--400 from 5.2 to 14.5 rain. The re- Table II. Matrix Recoveries,tODs, and tOQs Name (m/z) LOD* (pg/mL) ACE 150 0.5 1.0 81.8 12.9 Diclofenac 250 294 0.5 1.0 0.5 1.0 92.2 93.8 13.5 3.22 Flunixin 295 251 0.1 2.0 0.1 2.0 95.1 85.3 14.2 10.1 Ibuprofen 205 161 5.0 10 82.1 52.9 23.6 9.49 Indomethacin 312 356 0.1 0.5 Ion LOQ* %Recoveryt (pg/mL) at LOD 10 20 0.1 0.5 Ketoprofen 209 253 [97 0.5 10 10 20 20 20 Mefenamic acid 240 196 0.5 5.0 Naproxen 185 170 10 10 Oxyphenbutazone 323 280 204 1.0 10 5.0 Phenylbutazone 307 0.05 0.05 Piroxicam 330 266 0.1 0.5 0.5 1.0 Salicylic acid 137 5.0 Tolmetin 212 105 116 %SD 3.20 18.2 43.8 58.5 50.6 6.54 6.66 12.4 107 79.2 9.19 5.94 20 20 30.4 38.2 4,20 7.81 1.0 10 5 75.4 70.4 64.8 0.5 5.0 20 103 73.2 71.9 52.9 19.6 14.3 24.4 4.79 13.9 14.3 Reference 1. L.J. Roberts, II and J.D. Morrow. Analgesic-antipyretic and antiinflammatory agentsand drugs employed in the treatment of gout. In Goodman & Gilman's The PharmacologicalBasis of Therapeutics, 2. 3. 9.40 4. 5.0 20 62.2 17.8 * LODs and LOQs were determined as described in Methods. t Percent recoveries are reported based on an average of n = 4 replicates for all analytes at corresponding individual LODs. 102 suiting ILODs, matrix LODs, and LOQswere at or below the therapeutic ranges for most drugs under present conditions (Tables I and II). Other advantagesof a full-scan modewere considered for this initial study. Full-scan mass spectra generated for pure drugs with different degrees of fragmentation could enhance screening capabilities and prove more useful over SIM for identification of a particular NSAID in a sample with an unknown drug. Fragment ion spectra of standards can be overlaid with full-scan mass spectra generated for the unknown sample leading to unambiguous drug identification. Quantitation can then be followedbased on the extracted ion chromatograms. The method described in this report was applied to a clinical case investigated by the authors involvinga ferret suspected of aspirin intoxication. Serum was not available fromthis animal, but a submitted whole blood sample was analyzed for the 12 NSMDs and ACE. Sample clean-up procedure described previously was used with minor modifications.The bloodsample (0.4 mL) was treated with NaCI and 2 mL of ACN, mixed, and centrifuged. The clear supernatant was collected and the pellet was re-washedwith another 2 mL of ACN.The two supematants were combined and reduced to dryness under nitrogen, and resuspended in 0.4 mL of ACN. Sample analysisand the extracted current-ion profilesfor this caseshowed the presence of salicylic acid quantitated at 72 12g/mL(data not shown). In order to determine if the above sample handling modificationswas applicable for the established procedure, a 25 t2g/mLsalicylic acid fortified serum was prepared in the same manner as the above blood sample. The analysis of the spiked sample resulted in 72% salicylicacid sample recovery,suggesting that the procedure can be applied to blood samples as well as serum. In summary, performing the analysis in a full-scan mode of MS and exposing primary ions to negative APCI produced sufficient spectral fragments for each compound to achieve nearly unambiguous interpretation and identification of all NSAIDs and ACE present in the fortifiedserum sample at a range of concentrations. Future work will concentrate on improving sample extraction procedures and applying the complete multi-residue technique to detection of analgesics in more complexbiological samples such as liver and kidney. 5. 10th ed. McGraw-Hill, New York, NY, 2001, pp 687-731. D.M. Boothe. Anti-inflammatory drugs. In Small Animal Clinical Pharmacology and Therapeutics. W.B. Saunders Company, Philadelphia, PA, 2001, pp 281-311. J.R. Burke and J.C. Morgenlander. Update on Alzheimer's disease. Promising advances in detection and treatment. Postgrad. Med. 106:85-94 (1999). B. De Strooper and G. Konig. An inflammatory drug prospect. Nature 414:159-160 (200~). E.D, Agdeppa, V. Kepe, A. Petri, N. Satyamurthy,J. Liu, S.C. Huang, G.W. Small, G.M. Cole, and J.R. Barrio. In vitro detection of (S)naproxen and ibuprofen binding to plaques in the Alzheimer's brain using the positron emission tomography molecular imaging probe 2-(1 -[6-[(20[(18)FIfluoroethyl)(methyl)amino]-2-naph- Journalof AnalyticalToxicology,Vol. 29, March 2005 thyl]ethylidene)malononitrile. Neuroscience 117:723-730 (2003). 6. E.J.Ernst.Adulteration of Chinese herbal medicines with synthetic drugs: a systematic review. Intern. Meal. 252:107-113 (2002). 7. A. Anker. Toxins in depth. In Clinical Toxicology, 1st ed. M.D. Ford, K.A. Delaney, L.J. Ling, and T. Erickson, Eds. W.B. Saunders Company, Philadelphia, PA, 2001 pp 265-280. 8. R.K. Sellon. Acetaminophen. In Small Animal Toxicology, M.E. Peterson and P.A. Talcott, Eds.W.B. SaundersCompany, Philadelphia, PA, 2001, pp 388-395. 9. S.M.R. Stanley, N.A. Owens, and J.P.Rodgers. Detection of flunixin in equine urine using high-performance liquid chromatography with particle beam and atmospheric pressure mass spectrometry after solid-phase extraction. J. Chromatogr. B 667:95-103 (1995). 10. P. Van Eenoo, F.T. Delbeke, K. Roels, and K. Baert. Detection and disposition of tolmetin in the horse. J. Pharm. Biomed. Anal. 31: 723-730 (2003). 11. P.A. Asea, J.R. Patterson, G.O. Korsrud, P.M. Dowling, and J.O. Boison. Determination of flunixin residues in bovine muscle tissue by liquid chromatography with UV detection. J. AOAC Int. 84:659-665 (2001). 12. C. Cristofol, B. Perez, M. Pons, J.E. Valladares, G. Marti, and M. Arboix. Determination of indomethacin residues in poultry by high-performance liquid chromatography. J. Chromatogr. B 709: 310-314 (1998). 13. A.D. De Jager, H. Ellis, H.K.L. Hundt, K.J. Swart, and A.F. Hundt. High-performance liquid chromatography determination with amperometric detection of piroxicam in human plasma and tissue. J. Chromatogr. B 729:183-189 (1999). 14. A. Doliwa, S. Santoyo, M.A. Campanero, and P. Ygartua. Sensitive LC determination of piroxicam after in vitro transdermal permeation studies. J. Pharm. Biomed. Anal 26:531-537 (2001). 15. S.B. Clark, S.B. Turnipseed, G.J. Nandrea, M.R. Madson, J.A. Hurlbut, and J.N. Sofos. Confirmation of phenylbutazone residues in bovine kidney by liquid chromatography/mass spectrometry. J. AOAC Int. 85:1009-1014 (2002). 16. M. Yritia, P. Parra, J.M. Fernandez, and J.M. Barbanoj. Piroxicam quantitation in human plasma by high-performance liquid chromatography with on- and off-line solid-phase extraction. J. Chromatogr. A 846:199-205 (1999). 17. E.J.De Veau. Determination of non-protein bound phenylbutazone in bovine plasma using ultrafiltration and liquid chromatography with ultraviolet detection. J. Chromatogr. B 721:141-145 (1999). 18. W.R.G. Baeyens, G. Van der Weken, J. Haustraete, H.Y. AboulEnein, S. Corveleyn, J.P.Remon, A.M. Garcia-Campana, and P. Deprez. Direct HPLC analysis of ketoprofen in horse plasma applying an ADS-restricted access-phase. Biomed. Chromatogr. 13: 450-454 (1999). 19. T.H. Eichhold, R.E. Bailey, S.L. Tanguay, and S.H. Hoke, II. Determination of (R)- and (S)-ketoprofen in human plasma by liquid chromatography/tandem mass spectrometry following automated solid-phase extraction in the 96-well format. J. Mass Spectrom. 35: 504-511 (2000). 20. S. Liu, M. Kamijo, T. Takayasu, and S. Takayama. Direct analysis of indomethacin in rat plasma using a column-switching highperformance liquid chromatographic system. J. Chromatogr. B 767:53-60 (2002). 21. A.I. Gasco-Lopez, R. Izquierdo-Hornillos, and A. Jimenez. LC method development for ibuprofen and validation in different pharmaceuticals. J. Pharm. Biomed. Anal. 21: 143-149 (1999). 22. J. Klimes, J. Sochor, P. Dolezal, and J. Korner. HPLC evaluation of diclofenac in transdermal therapeutic preparations. Int. J. Pharm. 217:153-160 (2001). 23. M.S. Aurora-Prado, M. Steppe, M.F. Tavares, E.R. Kedor-Hackmann, and M.I. Santoro. Comparison between capillary electrophoresis and liquid chromatography for the determination of diclofenac sodium in a pharmaceutical tablet. J. AOAC Int. 85: 333-340 (2002). 24. J.Y.Kim, S.J.Kim, K.-J. Paeng, and B.C. Chung. Measurement of ketoprofen in horse urine using gas chromatography-mass spectrometry. J. Vet. Pharmacol. Ther. 24:315-319 (2001). 25. I.A. Wasfi, A.A. Hadi, N.A. Alkatheeri, I.M. Barezaiq, M. EIGhazali, N.S. Boni, and O. Zorob. Identification of a flunixin metabolite in camel by gas chromatography-mass spectrometry. J. Chromatogr. B 709:209-215 (1998). 26. G.M. Escandar.Spectrofluorimetric determination of piroxicam in the presence and absence of [3-cyclodextrin. Analyst 124:587-591 (1999). 27. M.I. PascuaI-Reguera, M.J. Ayora-Canada, and M.S. Castro Ruiz. Determination of piroxicam by solid-phase spectrophotometry in a continuous flow system. Eur. ]. Pharm. Sci. 15:179-183 (2002). 28. C. Albrecht and W. Thormann. Determination of naproxen in liver and kidney tissues by electrokinetic capillary chromatography with laser-induced fluorescence detection. J. Chromatogr. A 802: 115-120 (1998). 29. M.S. Prado, M. Steppe, M.F. Tavares, E.R. Kedor-Hackman, and M.I. Santoro. Method validation for diclofenac sodium in pharmaceuticals by capillary electrophoresis. J. Capillary Electrophor. 6:125-129 (1999). 30. Q. Dong and W. Jin. Monitoring diclofenac sodium in single human erythrocytes introduced by electroporation using capillary zone electrophoresis with electrochemical detection. Electrophoresis 22:2786-2792 (2001). 31. C. Simo, A. Gallardo, R.J. San, C. Barbas, and A. Cifuentes. Fast and sensitive capillary electrophoresis method to quantitatively monitor ibuprofen enantiomers released from polymeric drug delivery systems. J. Chromatogr. B 767:35-43 (2002). 32. L.G. Lala, P.M. D'Mello, and S.R. Naik. HPTLC determination ofdiclofenac sodium from serum. J. Pharm. Biomed. Anal. 29:539-544 (2002). 33. A.W. Abu-Qare and M.B. Abou-Donia. A validated HPLC method for the determination of pyridostigmine bromide, acetaminophen, acetylsalicylic acid and caffeine in rat plasma and urine. J. Pharm. Biomed. Anal. 26:939-947 (2001). 34. S. Zaugg, X. Zhang, J. Sweedler, and W. Thormann.Determination of salicylate, gentisic acid and salicyluric acid in human urine by capillary electrophoresis with laser-induced fluorescence detection. J. Chrornatogr. B 752:17-31 (2001). 35. R.I. Catarino, M.B. Garcia, R.A. Lapa, J.L. Lima, and E. Barrado. Sequential determination of salicylic and acetylsalicylic acids by amperometric multisite detection flow injection analysis. J. AOAC Int. 85:1253-1259 (2002). 36. T. Perez-Ruiz, C. Martinez-Lozano, A. Sanz, and E. Bravo. Determination of flufenamic, meclofenamic and mefenamic acids by capillary electrophoresis using beta-cyclodextrin. J. Chromatogr. B 708:249-256 (1998). 37. E. Mikami, T. Goto, 1. Ohno, H. Matsumoto, and M. Nishida. Simultaneous analysis of naproxen, nabumetone and its major metabolite 6-methoxy-2-naphthylacetic acid in pharmaceuticals and human urine by high-performance liquid chromatography. J. Pharm. Biomed. Anal. 23:917-925 (2000). 38. H.S. Lee, C.K. Jeong, S.J. Choi, S.B. Kim, M.H. Lee, G.I. Ko, and D.H. Sohn. Simultaneous determination of aceclofenac and diclofenac in human plasma by narrowbore HPLC using columnswitching. J. Pharm. Biomed. Anal. 23:775-781 (2000). 39. E. Mikami, T. Goto, T. Ohno, H. Matsumoto, K. Inagaki, H. Ishihara, and M. Nishida. Simultaneous analysis of anthranilic acid derivatives in pharmaceuticals and human urine by high-performance liquid chromatography with isocratic elution. J. Chromatogr. B 744:81-89 (2000). 40. M. Polasek, M. Pospisilova, and M. Urbanek. Capillay isotachophoretic determination of flufenamic, mefenamic, niflumic and tolfenamic acid in pharmaceuticals. J. Pharm. Biomed. Anal. 23:135-142 (2000). 41. V. Ioffe, T. Kalendarev, I. Rubinstein, and G. Zupkovitz. Reverse phase HPLC for polar lipids. Simple and selective HPLC procedures for analysis of phospholipids-based derivatives of valproic acid and various non-steroidal anti-inflammatory drugs. J. Pharm. Biomed. Anal. 30:391-403 (2002). 42. P. Lapicque, P. Netter, B. Bannwarth, P. Trechot, P. Gillet, H. Lambert, and R.J.Royer. Identification and simultaneous determination 103 Journal of Analytical Toxicology,Vol. 29, March 2005 43. 44. 45. 46. 104 of non-steroidal anti-inflammatory drugs using high-performance liquid chromatography. J. Chromatogr. B 496:301-320 (1989). A.G. Kazemifard and D.E. Moore. Liquid chromatography with amperometric detection for the determination of non-steroidal anti-inflammatory drugs in plasma. J. Chromatogr. B 533:125-132 (1990). P. Gowik, B. Julicher, and S. Uhlig. Multi-residue method for nonsteroidal anti-inflammatory drugs in plasma using high-performance liquid chromatography-photodiode-array detection. Method description and comprehensive in-house validation. J. Chromatogr. B 716:221-232 (1998). A. 8akkali, E. Corta, L.A. Berrueta, B. Gallo, and F. Vicente. Study of the solid-phase extraction of diclofenac sodium, indomethacin and phenylbutazone for their analysis in human urine by liquid chromatography. J. Chromatogr. B 729:139-145 (1999). G. Gonzalez, R. Ventura, A.K. Smith, R. de la Torre, and J. Segura. Detection of non-steroidal anti-inflammatory drugs in equine plasma and urine by gas chromatography-mass spectrometry. J. Chromatogr. A 719" 251-264 (1996). 47. H.H. Maurer. Systematic toxicological analysis procedures for acidic drugs and/or metabolites relevant to clinical and forensic toxicology and/or doping control. J. Chromatogr. B 733:3-25 (1999). 48. H.H. Maurer, F.X.Tauvel, and T. Kraemer. Screening Procedures for detection of non-steroidal anti-inflammatory drugs and their metabolites in urine as part of a systematic toxicological analysis procedure for acidic drugs and poisons by gas chromatography-mass spectrometry after extractive methylation. J. Anal. Toxicol. 25" 237-244 (2001). 49. M.E. AbdeI-Hamid, L. Novotny, and H. Hamza. Determination of diclofenac sodium, flufenamic acid, indomethacin and ketoprofen by LC-APCI-MS. J. Pharm. Biomed. Anal 24:587-594 (2001). Manuscript received November 20, 2003; revision received March 1,2004.
© Copyright 2026 Paperzz