Abstract
Tolfenamic acid, an anti-inflammatory drug (NSAID), is metabolized in vivo to form several oxidative metabolites which are all conjugated with β-d-glucuronic acid. In this study, the metabolites of tolfenamic acid were identified by 1H nuclear magnetic resonance (NMR) spectroscopy in urine samples obtained on days 7 to 10 from a human volunteer after oral administration of 200 mg of the drug three times per day (steady-state plasma concentration). The metabolites of tolfenamic acid were initially concentrated by preparative solid phase extraction (PSPE) chromatography, thereby removing the endogenous polar compounds that are present in the urine. The individual metabolites were purified by preparative high performance liquid chromatography (HPLC) and then identified using 1H NMR. Both one- and two-dimensional NMR experiments were performed to identify the phase II metabolites of tolfenamic acid; the study shows the applicability of 1H NMR for the identification of drug metabolites in biological fluids. In addition to NMR analysis, two metabolites were also identified by mass spectrometry (MS). The glucuronides of the following parent compounds,N-(2-methyl-3-chlorophenyl)-anthranilic acid (T),N-(2-hydroxymethyl-3-chlorophenyl)-anthranilic acid (1),N-(2-hydroxymethyl-3-chloro-4-hydroxyphenyl)-anthranilic acid (2), N-(2-formyl-3-chlorophenyl) anthranilic acid (3),N-(2-methyl-3-chloro-4-hydroxyphenyl)-anthranilic acid (4),N-(2-methyl-3-chloro-5-hydroxyphenyl)-anthranilic acid (5), N-(2-carboxy-3-chlorophenyl)-anthranilic acid (6),N-(2-hydroxymethyl-3-chlorophenyl)-4-hydroxy-anthranilic acid (7),N-(2-methyl-3-chlorophenyl)-5-hydroxy-anthranilic acid (8),N-(2-methyl-3-chloro-4-metoxyphenyl)-anthranilic acid (9),N-(2-methyl-3-chlorophenyl)-4-hydroxy-anthranilic acid (10), andN-(2-methyl-4-hydroxyphenyl)-anthranilic acid (11) were identified. The phase II metabolites (5–11) had not previously been identified in urine from humans administered tolfenamic acid. The phase I metabolites of the glucuronides 7, 8, 10, and 11 were identified here for the first time. An HPLC method was developed that simultaneously separates all the phase II metabolites identified as well as some phase I metabolites in urine samples obtained after intake of tolfenamic acid.
In this study the metabolism of the nonsteroidal anti-inflammatory drug (NSAID)1 tolfenamic acid (N-(2-methyl-3-chlorophenyl) anthranilic acid), was investigated using 1H NMR spectroscopy. The metabolic fate of tolfenamic acid in vivo in humans is presented in fig.1 (1, 2). Tolfenamic acid is oxidized in the methyl group in the 2′-position to form a primary alcohol that is then further oxidized to the aldehyde and finally to the carboxylic acid metabolite. Additionally, hydroxylations on both aromatic rings occur, giving several phenolic metabolites. All the metabolites are excreted in urine in the form of glucuronic acid conjugates as shown in fig.1.
In a published study on the metabolism of tolfenamic acid in biofluids (1), only the phase I metabolites ofN-(2-hydroxymethyl-3-chlorophenyl)-anthranilic acid (1),N-(2-hydroxymethyl-3-chloro-4-hydroxyphenyl)-anthranilic acid (2), N-(2-formyl-3-chlorophenyl) anthranilic acid (3),N-(2-methyl-3-chloro-4-hydroxyphenyl)-anthranilic acid (4), and N-(2-carboxy-3-chlorophenyl)-anthranilic acid (6) were identified. This was achieved after samples had been treated with Helix pomatia enzyme preparation to hydrolyze glucuronic acid conjugates before HPLC-analysis. The phase I metabolites were then synthesized and identified by 1H NMR. In a previous study in which directly-coupled 1H HPLC-NMR was applied, some phase II metabolites of tolfenamic acid were analyzed and identified directly, namely metabolites T and1–5 (see fig. 1) (3). In the present studies several of the metabolites shown in fig. 1 were thus identified for the first time.
A combination of preparative solid phase extraction (PSPE), preparative HPLC, and one dimensional (1D) and two dimensional (2D) 400 MHz1H NMR analysis was applied to isolate and identify the phase II metabolites of tolfenamic acid. The metabolites were isolated from urine samples obtained on days 7 to 10 after oral administration to an individual of 200 mg tolfenamic acid 3 times per day. When unknown drug metabolites are investigated by 1H NMR in urine samples, problems often occur because these samples are dominated by thousands of signals resulting from endogenous metabolites with small molecular weights. In the case of urine samples, it is thus advantageous to simplify the NMR spectra by adding a separation step into the experiment for example by combining solid phase extraction (SPE) off-line with NMR-analysis. SPE has been shown to be very successful for the separation of endogenous compounds from metabolites of interest thereby aiding the structural elucidation of drug metabolites in complex matrices (4, 5). However, if the metabolic pattern of the drug being investigated is very complex as is the case with tolfenamic acid, it is necessary to purify the metabolites before they can be unambiguously identified. This was achieved by scaling up the solid phase extraction step followed by preparative HPLC to purify the metabolites individually.
As a result of the present studies, 15 phase II metabolites of tolfenamic acid can now be analyzed intact simultaneously in biological samples. This was finally shown by the development of an HPLC method that separates all the phase I and II metabolites of tolfenamic acid present in urine samples after intake of the drug.
Materials and Methods
Chemicals.
Acetonitrile was of UV HPLC-grade and purchased from LAB-SCAN Ltd. (Dublin, Ireland). All other chemicals were of analytical chemical grade and purchased from Riedel-de Haën (Seelze, Germany). Helix pomatia (120,600 U/ml with respect to β-glucuronidase activity) was purchased from Sigma Chemical Co. (St. Louis, MO). Urine samples obtained from a patient in steady-state treatment with tolfenamic (200 mg acid orally, three times per day) were kindly donated by GEA A/S (Hvidovre, Denmark). Reference compounds of tolfenamic acid (N-(2-methyl-3-chlorophenyl)-anthranilic acid) and the following phase I metabolites,N-(2-hydroxymethyl-3-chlorophenyl)-anthranilic acid),N-(2-formyl-3-chlorophenyl)anthranilic acid),N-(2-methyl-3-chloro-4-hydroxyphenyl)-anthranilic acid), andN-(2-carboxy-3-chlorophenyl)-anthranilic acid) were also donated by GEA A/S.
Ethics.
The urine samples were obtained from a study approved by the Local Ethics Committee of Copenhagen, which was performed by GEA A/S.
PSPE of Urine samples containing metabolites of tolfenamic acid.
Urine samples obtained on days 7 to 10 from a human volunteer after oral administration of 200 mg of tolfenamic acid 3 times per day were investigated for metabolites of tolfenamic acid. Urine (3.8 l) (adjusted to 1% HCl by addition of 12 M HCl) was applied to a 1-liter glass column (200 mm × 80 mm i.d.) packed with XAD-2 resin 0.3–1 mm particles (Serva Feinbiochemica, Heidelberg, Germany), which was previously activated with 5 liters methanol followed by 5 liters 1% acetic acid. The urine samples were acidified to prevent acyl migration of ester glucuronides during isolation of the metabolites (6). The glass column was then washed with 3.3 liters 1% acetic acid followed by 2 liters of methanol (1% acetic acid, 20:80, v/v). A stepwise elution gradient was then applied to recover the metabolites from the column (see table 1). All eluents were pumped through the glass column by a peristaltic pump delivering a flow rate of 4 ml/min. The eluent in the collected fractions was evaporated by rotary evaporation and the remaining water was removed by freeze drying. A total of 5 fractions were collected and were kept separately for further purification of metabolites by preparative HPLC.
Treatment with β-glucuronidase.
Two ml of PSPE-treated urine was diluted with 1-ml potassium phosphate buffer (0.2 M, pH 6.5), and 5 μl Helix pomatia enzyme preparation was added. The sample was incubated at room temperature for 12 hr. Before analysis, 400 μl 1.5 M sulfuric acid was added along with 3 ml methanol-acetonitrile (50:50 v/v). The sample was then vortexed for 10 min and finally centrifuged 10 min at 3000 rpm. The supernatant was then analyzed in the preparative chromatographic system (injection volume was 100 μl).
Preparative Chromatography.
Two preparative HPLC columns were used A: a Knauer column 16 × 250 mm i.d., (Berlin, Germany) packed with Spherisorb ODS-2 (Phase Separations Ltd., Deeside, UK) (10 μm particles), was used with a flow rate of 7.0 ml/min and B: a Knauer column 16 × 250 mm i.d., packed with RoGel RP (Chemie Uetikon AG, Uetikon, Switzerland) (10 μm particles) using a flow rate of 8.0 ml/min. The injection volume used was 2 ml and the UV-detector was operated at 280 nm.
Preparative HPLC of the first PSPE-eluted fraction (Fraction 1) was performed using column A, and the eluent initially consisted of acetonitrile-methanol in 0.1% trifluoroacetic acid (TFA) (11.6:8.4:80; v/v). A linear gradient was applied over 42 min, resulting in a mobile phase composition of acetonitrile-methanol in 0.1% TFA (42:30.5:27.5; v/v). For fraction 2 the eluent initially consisted of acetonitrile-methanol in 0.1% TFA (23.2:16.8:60; v/v), and a linear gradient was applied over 26 min resulting in a final mobile phase composition of acetonitrile-methanol in 0.1% TFA (42.5:30.5:27.5; v/v); again column A was used. Forfraction 3 the eluent initially consisted of acetonitrile-methanol in 0.1% TFA (34.8:25.2:40; v/v). A linear gradient was used over 32 min, resulting in a final mobile phase composition of acetonitrile-methanol (42:58; v/v); column A was applied. For fraction 4 the eluent initially consisted of acetonitrile-methanol in 0.1% TFA (33.6:46.4:20; v/v). A linear gradient was applied over 13 min, resulting in a final mobile phase composition of acetonitrile-methanol in 0.1% TFA (55.8:40.4:3.8; v/v); column A was used. The individual metabolites were purified by applying these gradients, depending on within which PSPE fraction they were found and until appropriate purity for 1H NMR analysis was obtained. Two metabolites co-eluted (metabolite 4a and metabolite 6) and were purified applying column B. The eluent was isocratic and consisted of acetonitrile-methanol in 0.1% TFA (34.8:25.2:40; v/v).
Analytical Chromatography.
The analytical column was an YMC-pack ODS-AQ chromatographic column (4.6 × 250 mm i.d., 5 μm particles) (YMC Co., Kyoto, Japan); the flow rate was 2 ml/min. A gradient system was applied where eluent A consisted of acetonitrile-methanol (40:60 v/v), and eluent B consisted of phosphoric acid (15 mM). The mobile phase initially consisted of 70% eluent B for 15 min. The amount of eluent A was then increased to 50% with a linear gradient in the next 20 min; the amount of eluent A was further increased to 53% using a linear gradient in the following 15 min; the amount of eluent A was finally increased to 70% in 1.7 min and was kept at 70% for 6.3 min. Column temperature was set at 40°C and UV detection was at 280 nm.
Preparation of urine samples, obtained from a human volunteer 2–4 hr after oral intake of one dose of 200 mg of tolfenamic acid, was performed by SPE. Bond Elut C18 solid phase extraction columns (Varian, Harbor City, CA), 100 mg, were initially conditioned with 2 × 1.0 ml methanol, then 2 × 1.0 ml 1% acetic acid. One ml of urine sample was applied and the columns were then washed with 5 × methanol-1% acetic acid (20:80, v/v). The metabolites of tolfenamic acid were finally eluted with 500 μl methanol, and the eluate was collected and diluted 1:1 with 1% acetic acid before HPLC analysis.
1H NMR Analysis.
The one-dimensional (1D) NMR data of the purified metabolites were acquired using a Bruker AMX-400 spectrometer (Rheinstetten, Germany)1H NMR spectra were obtained at 400.14 MHz. Free induction decays (FIDs) were collected into 32 K computer data points with a spectral width of 5050.5 Hz, 90° pulses were used with an acquisition time of 2.58 sec, and the spectra were acquired by accumulation of 64–128 scans. Prior to Fourier transformation, an exponential apposition function was applied to the FID corresponding to a line broadening of 0.3 Hz.
Two-dimensional COSY and NOESY experiments were performed on the purified metabolites of tolfenamic acid to improve signal assignment of the complex aromatic region. The parameters for the COSY experiments were as follows: The number of scans per increment was 16, the spectral width was 5881.62 Hz, and 512 increments were performed in the F1 dimension. The FIDs were collected into 1-K computer data points. The relaxation delay between successive pulses was 1.5 sec. The parameters for the NOESY experiments were as follows: The number of scans per increment was 32, the spectral width was 4901.96 Hz, and 512 increments were performed in the F1 dimension. The FIDs were collected into 1 K computer data points, the delay between successive pulses was 2 sec, and the mixing time was 0.53 sec. The data were zero filled by a factor of 2 prior to Fourier transformation.
All NMR spectra of the purified glucuronides were recorded in CD3OD.
Identification by Mass Spectrometry (MS).
Mass spectra were obtained on a Finnigan MAT LCQ instrument (San Jose, CA) using atmospheric pressure chemical ionization in the negative mode. The analytes dissolved in methanol were injected (10 μl/min) into a stream (0.2 ml/min) of acetonitrile—0.1 M ammonium acetate (50:50 v/v). The evaporation temperature was 450°C, the capillary temperature was 200°C with −9V applied, and the sheath gas flow (N2) was set to 45 (arbitrary units).
Results and Discussion
Purification of Phase II Metabolites of Tolfenamic Acid.
The pooled urine samples were applied to the PSPE procedure described in Materials and Methods.. It was found by analyzing the eluate from the glass column before and after β-glucuronidase treatment that the glucuronides of tolfenamic acid exclusively eluted in the fractions containing 50% and 80% methanol, corresponding to fractions 1–4. Only phase II metabolites of tolfenamic acid were isolated and identified in the present investigations. Fraction 1 was found by treatment with β-glucuronidase to contain 7 glucuronides of tolfenamic acid which were later identified to be metabolites 2, 4a, 5b, 6, 7, 8 and 11 (see fig. 1). Fraction 2 contained 10 glucuronides. In addition to the glucuronides found in fraction 1, this fraction contained the glucuronic acid conjugates 1, 4, and 5a. Fraction 3 contained glucuronides T, 1, 3, 4, 4a, 5a, 6, 9, and 10. Fraction 4 only contained 2 glucuronides, namely, T and 9. The glucuronides found in the 4 fractions were further purified by preparative HPLC as described in Materials and Methods.. Five of the glucuronides (1, 3, 4, 9 and T) were purified by one purification step; the other metabolites were purified by repeated chromatography using the preparative HPLC-method until satisfactory purity for NMR analysis was obtained.
NMR for Identification of Tolfenamic Acid Metabolites.
The structures of the purified metabolites of tolfenamic acid were investigated using 1- and 2-D 1H NMR spectroscopy. The chemical shift values obtained for glucuronic acid conjugates of previously known phase I metabolites were compared with the chemical shifts published earlier for tolfenamic acid and its phase I metabolites and from 1H chemical shifts obtained from NMR spectra acquired of the authentic phase I metabolites (1-3). For the unknown metabolites a combination of 1D 1H NMR, 2D COSY, and NOESY experiments were used to identify the metabolites. In addition to NMR analysis, two metabolites, i.e. metabolite 5b and metabolite 11, were examined by MS.
The chemical shifts and the respective coupling patterns of the glucuronides of tolfenamic acid and the glucuronides of its oxidative phase I metabolites obtained in the present investigation are summarized in table 2. The numbering scheme for the protons on the tolfenamic acid structure is shown in fig. 1.
The characteristics of the NMR spectra for all the phase II metabolites of tolfenamic acid are that the H3-proton has resonance at δ7.92–8.09 and gives rise to a double doublet (J = 8 and 2 Hz) corresponding to an ortho and a meta coupling (unless oxidation has taken place in ring A, see fig. 1). The protons of the CH3-group in the 2′-position gives rise to a singlet at δ2.10–2.30. When hydroxylation takes place in this group, the protons shift to δ4.58–4.80. As a consequence of further oxidation to the aldehyde, a singlet arises at δ10.51. Ester glucuronides are characterized by the doublet at δ5.65–5.78 which corresponds to the anomeric proton of the glucuronic acid moiety. Ether glucuronides, on the other hand, give rise to a doublet at δ4.95–4.96. The phase II metabolites T, 1, 2, 3, and 4 have previously been identified by directly-coupled HPLC-NMR3, and the assignments will not be discussed in detail here.
Metabolite 4a.
A singlet NMR-signal at 2.15 ppm corresponded to an intact CH3 group in the 2′-position. The integrals of the aromatic region revealed that hydroxylation had taken place as only 6 proton signals were found. The aromatic signals were assigned from the 2D-COSY spectrum, and point of hydroxylation was found to be the 4′-position. The H5′ and H6′ protons gave rise to two doublets at 7.13 and 7.17 ppm, respectively, with a coupling constant of 10 Hz corresponding to ortho-couplings, whereas no meta-couplings could be observed. Finally, metabolite 4a was an ether glucuronide as the anomeric proton on the glucuronide ring gave rise to a doublet at 4.96 ppm. This metabolite was thus the ether glucuronide ofN-(2-methyl-3-chloro-4-hydroxyphenyl)-anthranilic acid.
Metabolite 5a.
A singlet at δ2.22 indicates an intact CH3 group in the 2′-position. Again the integrals of the aromatic region revealed the presence of only 6 protons, indicating hydroxylation on one of the aromatic rings. The aromatic signals were assigned by a COSY-experiment, and it was found that hydroxylation had taken place at the 5′-position. The H4′ and H6′ protons gave rise to two doublets with coupling constants of 2.50 Hz corresponding to meta couplings. The metabolite was an ether glucuronide according to the resonance of the anomeric proton at 4.95 ppm. Metabolite 5a was thus the ether glucuronide ofN-(2-methyl-3-chloro-5-hydroxyphenyl)-anthranilic acid.
Metabolite 5b.
The 1H NMR spectrum of this metabolite was equivalent in the aromatic region to metabolite 5a and also contained an intact CH3 group in the 2′-position. However, in addition to the anomeric proton at 4.95 ppm corresponding to an ether glucuronide, the NMR spectrum of metabolite 5b had a doublet at 5.78 ppm corresponding to an ester glucuronide. The integrals of the protons on the glucuronic acid moiety (δ3.6–4.1) also indicated that this metabolite was indeed the double glucuronide ofN-(2-methyl-3-chloro-5-hydroxyphenyl)-anthranilic acid. A mass spectrum was recorded of the metabolite to confirm that the double glucuronide was intact. A molecular ion was observed with a mass of 628.0 m/z, and a MS-MS experiment showed that this molecular ion gave two fragments with the masses of 452.5 m/z and 276.4 m/z, each resulting from loss of one glucuronic acid moiety. Additionally, the masses corresponding to the presence of chlorine isotopes were observed (i.e. M-1/M+1); thus, a molecular ion was observed with a mass of 630.0 m/z with a peak height that was 1/3 of that of the molecular ion with a mass of 628.0 m/z. Similarly fragments with masses 454.3 m/z and 278.7 m/z were observed during the MS-MS experiment. The mass spectrometry experiments confirmed the structure of metabolite 5b as a diglucuronide.
Metabolite 6.
No signals in the 1D NMR spectrum corresponding to the protons in the 2′-position were observed. This indicates that the methyl group is oxidized all the way to carboxylic acid. The metabolite is a β-1-O-acyl glucuronide identified by the doublet arising at 5.65 ppm. This metabolite is thus the β-1-O-acyl glucuronide of N-(2-carboxy-3-chlorophenyl)-anthranilic acid. The aromatic protons were assigned by a 2D-COSY experiment. In addition, the chemical shift values and coupling patterns were in agreement with the nonglucuronidated form of metabolite 6 that was available as a reference compound.
Metabolite 7.
The H3-proton of this metabolite is shifted 0.5 ppm up-field (δ7.52). This was confirmed by the assignments of the aromatic protons in the 2D COSY spectrum. Additionally, the coupling pattern of this proton changes from a double doublet to a doublet with a coupling constant of 4.5 Hz corresponding to a meta coupling. There are only 6 protons on the aromatic rings, and it was concluded that hydroxylation had taken place at the 4-position on aromatic ring A (see fig. 2). Also, the CH3-group in the 2′-position is hydroxylated to a primary alcohol, giving rise to a singlet at 4.80 ppm. The glucuronide is a β-1-O-acyl glucuronide according to the 1′β-proton signal at 5.75 ppm. This metabolite is thus the β-1-O-acyl glucuronide ofN-(2-hydroxymethyl-3-chlorophenyl)-4-hydroxy-anthranilic acid.
Metabolite 8.
The H3-proton is shifted upfield and gives rise to a doublet (δ7.48, J= 8.0 Hz), indicating hydroxylation but in this case at the H5-position. The aromatic protons were assigned from a 2D COSY experiment. The CH3 group in the 2′-position is intact, giving rise to a singlet at 2.15 ppm. Finally, the anomeric proton of the glucuronic acid moiety has resonance at 5.65 ppm, and this metabolite is thus the β-1-O-acyl glucuronide ofN-(2-methyl-3-chlorophenyl)-5-hydroxy-anthranilic acid.
Metabolite 9.
In the 1D NMR-spectrum of this metabolite a singlet at 2.30 ppm indicates an intact CH3 group in the 2′-position. In addition, a singlet was observed at 3.77 ppm corresponding to a methoxy-group. This methoxy-group was found to be positioned in the 4′ position based on the coupling pattern of the 5′ and 6′ protons, namely two doublets (H5′ at 7.20 ppm and H6′ at 7.26 ppm) with an ortho-coupling constant of 8 Hz. The methoxy-group could be positioned at the 6′ position; however, this position is sterically hindered, and it is therefore unlikely that substitution takes place there. None of the NMR experiments performed in the present investigations did, however, certify this fact. The doublet at 5.77 ppm indicates an ester glucuronide and metabolite 9 is thus believed to be the β-1-O-acyl glucuronide ofN-(2-methyl-3-chloro-4-metoxyphenyl)-anthranilic acid.
Metabolite 10.
This metabolite was also hydroxylated in ring A of tolfenamic acid (see fig. 2) in agreement with the fact that the proton signal of the H3 proton shifts upfield and gives rise to a doublet with a meta coupling constant (δ7.51, J = 3.2 Hz). On the other hand, the H5 proton gives rise to a double doublet with a meta as well as an ortho coupling (δ6.93, J = 3.2 and 8.8). All the aromatic protons were assigned by the 2D COSY-experiment. The CH3-group in the 2′-position is intact (δ2.30) and the glucuronide is a β-1-O-acyl glucuronide (δ5.76). Metabolite 10 is thus the β-1-O-acyl glucuronide ofN-(2-methyl-3-chlorophenyl)-4-hydroxy-anthranilic acid.
Metabolite 11.
Metabolite 11 was also an ester glucuronide with the resonance of the anomeric proton at 5.78 ppm. The CH3 group in the 2′-position was intact in agreement with the singlet at 2.13 ppm. The coupling pattern of the aromatic protons in ring A (see fig. 2) indicated no hydroxylation in this ring (as assigned by the 2D-COSY experiment). The integrals of the protons on ring B corresponded to 3 protons.3 A NOESY-experiment showed long range coupling between the CH3-group in the 2′-position and a H3′-proton. This H3′-proton gave rise to a doublet with a small coupling constant (δ6.74, J = 3 Hz), indicating that the ortho position (H4′) was hydroxylated. The H5′ and H6′ protons gave rise to a double doublet (δ6.64) and a doublet (δ6.74), respectively. In conclusion, this metabolite must be dehalogenated at the 3′-position. To confirm that the metabolite was dehalogenated, a mass spectrum of metabolite 11 was recorded. A molecular ion with a mass of 418.2 m/z was observed which indeed correlates with a metabolite of tolfenamic acid that is hydroxylated and has lost a chlorine atom. MS-MS analysis yielded a fragment with a mass of 242.4 m/z in agreement with loss of a glucuronic acid moiety. Also no masses were observed resulting from the chlorine isotope (i.e. m + 2). Metabolite 11 was thus found to be the β-1-O-acyl glucuronide ofN-(2-methyl-4-hydroxyphenyl)-anthranilic acid. Dehalogenation of aromatic drugs with a hydroxyl-group adjacent to the halogen has been reported (7, 8).
A Separation Method for the Phase II Metabolites of Tolfenamic Acid in Biological Fluids.
An HPLC-method was developed by which both phase I and phase II metabolites of tolfenamic acid could be analyzed simultaneously in urine samples. Fig. 2A shows the chromatogram of a mixture of all the purified phase II metabolites and some phase I metabolites when injected into the chromatographic system. The phase I metabolites are marked by an asterisk in the chromatogram. Fig. 2B shows the chromatogram of a urine sample that was obtained in the interval 2–4 hr after oral intake of 200 mg of tolfenamic acid by a human volunteer. The urine sample was subjected to SPE before HPLC analysis. This chromatogram is compared with the chromatogram of a blank sample (fig.2C). As it is seen from the chromatograms in fig. 2, it was possible to separate all metabolites of tolfenamic acid observed in the urine samples except for metabolite 10 and the nonglucuronidated form of metabolite 6. In addition, most of the metabolites are excreted in the form of glucuronides. The only nonglucuronidated metabolites we have observed in urine samples after single dose and repeated dose intake of tolfenamic acid are the phase I metabolites of 1, 3, 4, and 6.
Conclusions
PSPE in combination with preparative HPLC and NMR analysis is an efficient means by which unknown drug metabolites can be unambiguously identified in biofluids. This is true especially if the drug under investigation possesses hydrophobic properties (as for the metabolites of tolfenamic acid) so they can easily be separated from the endogenous metabolites using solid phase extraction columns packed with polystyrene packing material. The experiments described show that, if the metabolic pattern is complex, as was the case with tolfenamic acid where the different positions of hydroxylation on the aromatic rings give rise to a highly complex 1H NMR pattern in the aromatic spectral region (δ6.3–8.5), then 2D NMR experiments must be used to identify the metabolites. In the present investigation, we found that the combination of 1D 1H NMR, 2D COSY, and 2D NOESY experiments were suitable experiments for the unambiguous identification of unknown metabolites of tolfenamic acid.
Furthermore, this study describes the identification of the intact glucuronic acid conjugates of tolfenamic acid in biofluids. Several phase II metabolites were identified for the first time namely the ester glucuronides ofN-(2-carboxy-3-chlorophenyl)-anthranilic acid (6),N-(2-hydroxymethyl-3-chlorophenyl)-4-hydroxy-anthranilic acid (7),N-(2-methyl-3-chlorophenyl)-5-hydroxy-anthranilic acid (8),N-(2-methyl-3-chloro-4-metoxyphenyl)-anthranilic acid (9),N-(2-methyl-3-chlorophenyl)-4-hydroxy-anthranilic acid (10) and N-(2-methyl-4-hydroxyphenyl)-anthranilic acid (11) and the ether glucuronides ofN-(2-methyl-3-chloro-4-hydroxyphenyl)-anthranilic acid (4) andN-(2-methyl-3-chloro-5-hydroxyphenyl)-anthranilic acid (5a) as well as the double ester and ether glucuronide ofN-(2-methyl-3-chloro-5-hydroxyphenyl)-anthranilic acid (5b). In previous investigations only the phase I metabolites of tolfenamic acid were identified after β-glucuronidase treatment of the biological samples was investigated; however, the phase I metabolites of 7, 8, 10, and 11 were not reported. The dechlorinated metabolite (11) is evidence of a rarer metabolic pathway.
Even though several of the phase II metabolites identified are ester glucuronides, no positional isomers resulting from acyl migration were observed during the 1H NMR analysis. Such positional isomers give rise to characteristic shifts in the proton signals resulting from the glucuronic acid moiety.9 Acyl migration was prevented in the present investigations because the urine samples were acidified before isolation and all isolation steps were performed under acidic pH. From unpublished investigations we know that no acyl migration of the ester glucuronide of tolfenamic acid (T) takes place at acidic pH and that acyl migration of T is very slow even at pH values above 7.
The metabolites purified here can be used as reference compounds for further investigations of the metabolism of tolfenamic acid. A chromatographic method was developed for simultaneous separation of all the metabolites isolated and identified in the present investigations.
Footnotes
-
Send reprint requests to: Ulla Grove Sidelmann, Department of Analytical and Pharmaceutical Chemistry, The Royal Danish School of Pharmacy, Universitetsparken 2, Dk-2100 Copenhagen, Dennmark.
- Abbreviations used are::
- NSAID
- nonsteroidal anti-inflammatory drug
- TFA
- trifluoroacetic acid
- PSPE
- preparative solid phase extraction
- Received December 4, 1996.
- Accepted March 18, 1997.
- The American Society for Pharmacology and Experimental Therapeutics