Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Carbamazepine (5H-dibenzo[b,f]azepine-5-carboxamide) is an effective drug in the treatment of convulsive disorders (Cereghino et al., 1973, 1974). However, carbamazepine-induced adverse reactions have been reported in as many as one-third to one-half of all patients treated with this drug (Pellock, 1987; Durelli et al., 1989; Gram and Jensen, 1989). Among these adverse reactions, 5% of them can be classified as idiosyncratic reactions (Askmark and Wiholm, 1990). These idiosyncratic adverse reactions include skin rash (Crill, 1973), blood disorders (Gerson et al., 1983) and hepatitis (Horowitz et al., 1988). A Swedish survey of 505 reports of 713 idiosyncratic reactions to carbamazepine from 1965 to 1987 reported skin reactions (48%), hematological (12%) and hepatic disorders (10%) to be the most frequent. (Askmark and Wiholm, 1990) Although the mechanism of carbamazepine-induced adverse reactions is not clear, they are thought to result from the formation of chemically reactive metabolites (Shear et al., 1988; Riley et al., 1989). An arene oxide intermediate has been postulated to be responsible for the idiosyncratic toxicity of carbamazepine. This hypothesis is mostly based on the observation that the metabolism-dependent cytotoxicity of carbamazepine in vitro can be enhanced by trichloropropene oxide (TCPO) (Riley et al., 1989), an inhibitor of epoxide hydrolase. However, this evidence is complicated by the fact that TCPO is also known to deplete glutathione (Larrey et al., 1989) and inhibit cytochrome P-450 (Ivanetich et al., 1982). Furthermore, since the arene oxide is chemically reactive, it may not reach targets, such as skin and bone marrow, distant from the liver in sufficient concentrations to induce adverse reactions. It appears that most drugs associated with bone marrow toxicity are metabolized to reactive metabolites by myeloperoxidase or other enzymes present in the bone marrow, such as alcohol dehydrogenase (Uetrecht, 1996). These considerations led us to search for alternative bioactivation pathways of carbamazepine.

Carbamazepine is extensively metabolized and more than 30 metabolites have been identified in urine of patients taking the drug (Lertratanangkoon and Horning, 1982). The major pathways involve N-glucuronidation on the carbamoyl side chain of carbamazepine; formation of carbamazepine-10,11-epoxide and hydroxylation on the aromatic rings. One of the major metabolites is 2-hydroxycarbamazepine (Eichelbaum et al., 1984), and it can be further metabolized to 2-hydroxyiminostilbene, either in the liver or in other tissues. Both 2-hydroxycarbamazepine and 2-hydroxyiminostilbene have been detected as conjugates in the urine of rats and humans (Lertratanangkoon and Horning, 1982). We hypothesized that 2-hydroxyiminostilbene can be further oxidized to a reactive iminoquinone intermediate (Fig. 1) which may be responsible for the idiosyncratic toxicity of carbamazepine.

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Proposed bioactivation pathway of carbamazepine which leads to the formation of an iminoquinone reactive intermediate.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Synthesis of the Iminoquinone Intermediate. The iminoquinone was synthesized by oxidation of iminostilbene using a modification of the method of Islam and Skibo (1991). Fremy's salt (2.6 g; Aldrich Chemical Co., Milwaukee, WI) was dissolved in 260 ml of phosphate buffer (pH 7.1). To this solution, 1 g of iminostilbene (Aldrich Chemical Co.) in 260 ml of acetone was added. The mixture was stirred at room temperature for 5 min, and the resulting red solution was extracted with 3 × 200-ml portions of chloroform. The combined extracts were washed with 3 × 100-ml portions of water and then dried over magnesium sulfate. The solvent was evaporated by a rotary evaporator. The iminoquinone was then purified by flash chromatography using hexane/ethyl acetate (7:3, v/v) as the eluting solvent. The product was a red solid with a yield of 23%, m.p. = 135-136°C; mass spectrometry (MS): m/z 208 (MH⁺) and 180 [(MH-CO)⁺]; ¹H NMR (chloroform-d):  7.94 ppm (H₆ or H₉, d, J = 7.82 Hz);  7.62 ppm (H₇ or H₈, dd, J = 7.57 Hz; 1.63 Hz);  7.59 ppm (H₄, d, J = 9.77 Hz);  7.54 ppm (H₈ or H₇, dd, J = 7.57 Hz; 1.47 Hz);  7.48 ppm (H₉ or H₆, d, J = 7.81 Hz);  6.98 ppm (H₃, dd, J = 9.76 Hz; 2.20 Hz);  6.96 ppm (H₁₀ or H₁₁, d, J = 11.00 Hz);  6.78 ppm (H₁₁ or H₁₀, d, J = 11.48 Hz);  6.58 ppm (H₁, d, J = 2.44 Hz). This iminoquinone species has been synthesized and the chemical analyses have been reported: m.p. 135-136°C; the NMR spectrum consisted only of a series of multiplets between  1.8 and 3.6; in the mass spectrum the base peak was the M-CO ion (Haque and Proctor, 1968).

Synthesis of 2-Hydroxyiminostilbene. 2-Hydroxyiminostilbene was synthesized by reduction of the iminoquinone using the method of Chang (1983). A solution of 40 mg of iminoquinone in chloroform was extracted with freshly prepared sodium hydrosulfite solution (excess, Aldrich Chemical Co.) until the organic layer changed from red to yellow. The chloroform layer was washed with water and dried over magnesium sulfate. The solvent was then removed by a rotary evaporator to give 36 mg (89% yield) of 2-hydroxyiminostilbene as a yellow solid, m.p. = 225 - 226°C; MS: m/z 210 (MH⁺); ¹H NMR(chloroform-d):  7.05 ppm (H₇ or H₈, dd, J = 7.50 Hz; 1.41 Hz);  6.90 ppm (H₆ or H₉, d, J = 7.57 Hz);  6.85 ppm (H₈ or H₇, dd, J = 7.45 Hz; 1.60 Hz);  6.55 ppm (H₃, dd, J = 8.3 Hz; 2.93 Hz);  6.54 ppm (H₉ or H₆, d, J = 7.81 Hz);  6.44 ppm (H₄, d, J = 8.3 Hz);  6.42 ppm (H₁₀ or H₁₁, d, J = 12.00 Hz);  6.41 ppm (H₁, d, J = 2.68 Hz);  6.32 ppm (H₁₁ or H₁₀, d, J = 11.72 Hz) and two broad peaks in the range of  4.9 ppm to  4.4 ppm due to the proton on the nitrogen and the proton of the hydroxyl group. The chemical analyses of 2-hydroxyiminostilbene reported in the literature were: m.p. 225-226°C; NMR (chloroform-d/dimethyl sulfoxide-d₆):  8.27 ppm (s, 1H, -OH),  7.5 to 6.2 ppm (m, 9H, aromatic and olefin),  5.41 (s, broad, 1H, -NH); MS: m/e 209 (M⁺) (Chang, 1983).

Synthesis of N-acetylcycteine (NAC) Adduct of the Iminoquinone. A methanol solution (4.8 ml) of the iminoquinone (10 mg) was added to NAC (4 g) in 43.2 ml of phosphate buffer (pH 8). The mixture was stirred for 10 min at room temperature and then concentrated on a rotary evaporator. The NAC adduct of the iminoquinone was purified by silica gel thin-layer chromatography (TLC) developed with a solvent of ethyl acetate/methanol (7:3, v/v). The TLC band (R_F = 0.3) that contained the iminoquinone-NAC adduct was scraped off the TLC plate and the adduct was extracted with ethyl acetate. The yield was approximately 33%. The mass spectrum of the iminoquinone-NAC adduct included a MH⁺ ion at m/z 371.

Synthesis of 2-Hydroxycarbamazepine. The hydroxy group on 2-hydroxyiminostilbene was first protected using a modification of the method of Kendall et al. (1979). Imidazole (12 mg, 0.18 mmol; Aldrich Chemical Co.) was added to a solution of 2-hydroxyiminostilbene (18.4 mg, 0.09 mmol) and tert.-butyldimethylchlorosilane (26.4 mg, 0.18 mmol; Aldrich Chemical Co.) in dry N,N-dimethylformamide (0.1 ml). The mixture was stirred for 3 h, diluted with 5% aqueous sodium bicarbonate (0.5 ml), and extracted four times with hexane/dichloromethane (9:1, v/v). The combined extracts were washed with water, dried, and evaporated by a rotary evaporator.

Treatment of the Urine Sample. Random urine samples were obtained from two male patients (ages, 93 and 83, respectively) who had received carbamazepine (250 and 200 mg/day, respectively) for more than 1 year. The urine sample (~100 ml) was concentrated to about 5 ml by lyophilization. The concentrated urine sample was acidified to pH 4.5 and then purified by C-18 Prep-Sep solid-phase extraction column (Fisher Scientific, Unionville, ON). The sample was loaded onto the column followed by washing with 2 × 10 ml portions of water. The desired urinary metabolites were then eluted from the column by 2 × 10-ml portions of methanol. The methanol effluents were combined and concentrated to ~1 ml. Aliquots (10 µl) were analyzed by liquid chromatography (LC/)MS. Enzymatic hydrolysis was carried out by incubating the urine samples with -glucuronidase (100 U; Sigma Chemical Co., Oakville, ON) in 0.2 ml of pH 5.1 buffer) for 20 h at 37°C. The carbamazepine metabolites were then extracted with ethyl acetate and the extracts were evaporated (N₂ stream), dissolved in methanol, and aliquots of 10 µl were analyzed by LC/MS.

Analytical Methods. The analyses of carbamazepine urinary metabolites were performed by interfacing an Ultracarb C-18 column (2 × 100-mm; Phenomenex, Torrance, CA) to a Sciex API III mass spectrometer (Perkin-Elmer, Sciex, Thornhill, Ontario, Canada). Aliquots of a urine sample (10 µl) were eluted with a solvent consisted of water, acetonitrile, and acetic acid (49:50:1, v/v) including 1 mM ammonium acetate at a flow rate of 0.2 ml/min. When the metabolites with shorter retention times (e.g., the metabolite with MH⁺ ion at m/z 432) were analyzed, the eluting solvent consisted of water, acetonitrile, and acetic acid (74:25:1, v/v) including 1 mM ammonium acetate. When -glucuronide conjugates were analyzed, the eluting solvent consisted of water, acetonitrile, and acetic acid (84:15:1, v/v) including 1 mM ammonium acetate. A splitter was used to decrease the flow to the LC/MS interface at ~10 µl/min. The collisional activation spectra were obtained by using the LC/MS/MS mode, with argon as the target gas.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Oxidation of 2-Hydroxyiminostilbene. 2-Hydroxyiminostilbene was readily oxidized to the iminoquinone intermediate by HOCl. When 2-hydroxyiminostilbene was reacted with HOCl at equal concentrations, the reaction mixture changed instantaneously from yellow to red and a new peak due to the iminoquinone was observed by high-performance liquid chromatography analysis. Figure 2 shows the UV absorption spectra of 2-hydroxyiminostilbene and the reaction mixture after HOCl was added. The UV absorption spectrum of the oxidation product was similar to that of the iminoquinone. H₂O₂ was also able to oxidize 2-hydroxyiminostilbene to the iminoquinone species, however, at a much slower rate than HOCl. It was also observed that 2-hydroxyiminostilbene underwent autoxidation by air at room temperature to the iminoquinone intermediate. The half-life of 2-hydroxyiminostilbene was approximately 2 h in phosphate buffer (pH 7.4).

View larger version (16K): [in this window] [in a new window]   Fig. 2.   The UV absorption spectra of the synthesized iminoquinone (dashed line) and 2- hydroxyiminostilbene before (0 s) and after (10 s) reaction with HOCl. The concentration of both 2-hydroxyiminostilbene and HOCl were 0.1 mM.

Reactivity of the Iminoquinone Intermediate. The reactivity of the iminoquinone as an electrophile was studied. It was found to react with sulfhydryl-containing nucleophiles, such as glutathione (GSH) and NAC, to form conjugates. The iminoquinone-NAC conjugate was isolated and analyzed by NMR (Fig. 3). The ¹H NMR (D₂O) consisted of:  7.17 ppm (H₇ or H₈, dd, J = 7.9 Hz; 1.70 Hz);  6.99 ppm (H₆ or H₉, d, J = 7.48 Hz);  6.97 ppm (H₈ or H₇, dd, J = 7.80 Hz; 1.65 Hz);  6.91 ppm (H₃, d, J = 2.77 Hz);  6.89 ppm (H₉ or H₆, d, J = 7.56 Hz);  6.51 ppm (H₁₀ or H₁₁, d, J = 11.72 Hz);  6.44 ppm (H₁, d, J = 2.78 Hz);  6.40 ppm (H₁₁ or H₁₀, d, J = 11.75 Hz); NAC-CH,  4.28 ppm (1H, dd, J = 7.05 Hz, 3.66 Hz); NAC-CH₂,  3.34 ppm (1H, dd, J = 14.11 Hz, 3.84 Hz); NAC-CH₂,  3.13 ppm (1H, dd, J = 14.10 Hz, 7.05 Hz); NAC-CH₃,  1.67 ppm (3H, s). The ¹H NMR data, combined with the findings from the ¹H-¹H and ¹H-¹³C correlation experiments, confirmed the structure of the conjugate where the sulfur of NAC was substituted in the 4-position on the aromatic rine (Fig. 4).

View larger version (8K): [in this window] [in a new window]   Fig. 3.   Proton NMR of the iminoquinone-NAC conjugate. The integrals are shown at the bottom of the spectrum.

View larger version (10K): [in this window] [in a new window]   Fig. 4.   Oxidation of 2-hydroxyiminostilbene to the iminoquinone intermediate and the reaction of the iminoquinone with NAC.

Patient Urine Sample Analysis. When the urine sample was analyzed by LC/MS, neither the GSH nor the NAC conjugate of iminoquinone was detected. However, a metabolite with MH⁺ ion at m/z 432 (R_t = 5.7 min) was detected. This metabolite has the same molecular weight as the glucuronide conjugate of 4-methylthio-2-hydroxyiminostilbene, which is a probable further metabolite of the iminoquinone-GSH conjugate. The MS/MS of the metabolite with MH⁺ ion at m/z 432 showed a fragment ion at m/z 256 which corresponded to the loss of dehydroglucuronic acid moiety ([M + 1-176]⁺) (Fig. 5). When the urine sample was hydrolyzed by -glucuronidase, a peak with MH⁺ ion at m/z 254 was detected by LC/MS. The molecular weight of this species was 2 mass units less than that of 4-methylthio-2-hydroxyiminostilbene suggesting that it could be a further oxidized product of 4-methylthio-2-hydroxyiminostilbene.

View larger version (11K): [in this window] [in a new window]   Fig. 5.   MS/MS spectrum of the metabolite, which has the same molecular weight as that of 4-methylthio-2-hydroxyiminostilbene, in its glucuronide form with a MH+ ion at m/z 432.

View larger version (24K): [in this window] [in a new window]   Fig. 6.   Selected ion current of the urinary metabolites from the LC/MS of a patient's urine: 2-hydroxyiminostilbene-glucuronide (A), carbamazepine-glucuronide (B), monohydroxycarbamazepine-glucuronide (C), and 10,11-dihydrodiol- carbamazepine (D). The ion current for each ion is provided in the upper right corner of each trace.

View larger version (13K): [in this window] [in a new window]   Fig. 7.   MS/MS spectrum of the glucuronide conjugate of 2-hydroxyiminostilbene.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   LC/MS ion current at m/z 208 corresponding to the iminoquinone intermediate in a urine sample before (A) and after (B) hydrolysis by -glucuronidase.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The mechanism(s) of the idiosyncratic reactions associated with carbamazepine therapy are poorly understood. It has been postulated that bioactivation to a reactive arene oxide metabolite is a prerequisite to toxicity (Shear et al., 1988; Pirmohamed et al., 1992) and a risk factor for the adverse reactions is a deficiency of epoxide hydrolase (Spielberg et al., 1981) However, two good studies have failed to find a consistent mutation, or pattern of mutations, in the microsomal epoxide hydrolase gene which is common in patients with a history of carbamazepine hypersensitivity reactions (Gaedigk et al., 1994; Green et al., 1995). Carbamazepine has been shown to be bioactivated to a protein-reactive metabolite by human liver microsomes; however, the covalent binding of [¹⁴C]carbamazepine to human liver microsomes was relatively low (Pirmohamed et al., 1992). The formation of the arene oxide metabolite has only been inferred from the urinary excretion of unquantified phenols and catechols of carbamazepine (Lertratanangkoon and Horning, 1982; Regnaud et al., 1988). However, the aromatic hydroxyl metabolites are not necessarily derived from an arene oxide intermediate (Hanzlik et al., 1984). A study by Lillibridge et al. (1996) using mouse liver microsomes has shown that a quinone intermediate may also be formed in carbamazepine bioactivation. Furthermore, it has been demonstrated that for bromobenzene, which does form an arene oxide, most of the covalent binding is due to a quinone (Hanzlik et al., 1984; Rietjens et al., 1997). Although the detection of GSH conjugate of the postulated arene oxide intermediate in rodents has been reported (Madden et al., 1996; Amore et al., 1997), in humans, the same GSH conjugate was not detected nor were any of its further metabolized products, such as the NAC, cysteinyl, or thiomethyl derivatives (Maggs et al., 1997). In addition, due to their reactivity, arene oxides are short-lived and probably would not be able to reach sites (e.g., bone marrow or skin) distant from the major site of production, i.e., liver, at sufficient concentrations to cause toxicity. Therefore, the objective of this work was to investigate alternative bioactivation pathways where reactive metabolites might be generated at the targets of toxicity.

One of the major routes of carbamazepine metabolism in the liver is hydroxylation of the aromatic ring to yield one of the major carbamazepine metabolites, 2-hydroxycarbamazepine, which can be further hydrolyzed to 2-hydroxyiminostilbene, either in the liver or in other tissues (Lertratanangkoon and Horning, 1982). In this study, we were able to detect the glucuronide conjugate of the 2-hydroxyiminostilbene by LC/MS. Two of the hydroxyiminostilbenes had been detected in human urine during an earlier study in which they were characterized as their trimethylsilyl (TMS) derivatives after enzymatic hydrolysis of the urine sample with Glusulase (Lertratanangkoon and Horning, 1982). One of the TMS derivatives of the hydroxyiminostilbenes was observed to have the same GC/MS properties as the TMS derivative of 2-hydroxyiminostilbene (Lertratanangkoon and Horning, 1982). Although the amount of the metabolite was too small for us to identify the structure by NMR, the fact that we also observed a peak due to the iminoquinone in the same sample suggested that 2-hydroxyiminostilbene glucuronide conjugate must be excreted in the urine because other hydroxyiminostilbene isomers would not generate the iminoquinone intermediate upon oxidation. In addition, when the sample was hydrolyzed by -glucuronidase to liberate 2-hydroxyiminostilbene, a much larger peak due to the iminoquinone was observed instead of a peak due to 2-hydroxyiminostilbene (Fig. 8). This was not unexpected since we had shown that 2-hydroxyiminostilbene was readily oxidized to the iminoquinone by HOCl, H₂O₂ or even by air. The free 2-hydroxyiminostilbene released from enzymatic hydrolysis of the glucuronide conjugate was presumably oxidized to iminoquinone readily before it was detected by LC/MS. We also detected a urinary metabolite with the same molecular weight as the glucuronide conjugate of 4-methylthio-2-hydroxyiminostilbene. However, we did not obtain enough material to identify its structure.

In this work, we have also demonstrated that the iminoquinone reacted with sulfhydryl-containing nucleophiles, such as GSH and NAC, to form conjugates. The sulfur was substituted on the aromatic ring in the meta position to the hydroxy group. This result suggested that the iminoquinone is an reactive electrophile, and it may bind to macromolecules (i.e., proteins) in vivo to cause direct toxicity or act as a hapten to modulate the immune system. It could also generate reactive oxygen species by redox cycling.

In summary, we have demonstrated that the -glucuronide conjugate of 2-hydroxyiminostilbene was excreted in the urine of patients taking carbamazepine. We have also shown that 2-hydroxyiminostilbene was readily oxidized to a reactive iminoquinone intermediate that can be trapped by GSH and NAC. The ease of oxidation of 2-hydroxyiminostilbene makes the iminoquinone an attractive candidate for the reactive metabolite of carbamazepine responsible for idiosyncratic reactions in the bone marrow and skin. Specifically the multistep metabolic pathway could explain the relatively low covalent binding of carbamazepine in human hepatic microsomes and the 2-hydroxyiminostilbene could be readily oxidized in target organs, such as bone marrow and skin, by peroxidases or even by oxygen.