Introduction

Top Summary Introduction Procedures Results Discussion References

Clozapine (Fig. 1) is an atypical antipsychotic agent that is more effective than standard neuroleptic drugs in the treatment of refractory schizophrenia. Unfortunately, the use of clozapine has been limited because it causes agranulocytosis in almost 1% of patients treated with the drug (Safferman et al., 1992; Alvir et al., 1993; Atkin et al., 1996). The mechanism of clozapine-induced agranulocytosis is unknown. A possible mechanism involving a toxic serum factor with characteristics of an antibody (Pisciotta et al., 1992) has been proposed; however, there are limitations to this hypothesis (Liu and Uetrecht, 1995).

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Chemical structures of olanzapine (top) and clozapine (bottom).

Fischer et al. (1992) reported indirect evidence for the formation of a free radical after oxidation of clozapine by horseradish peroxidase and MPO. In addition, they identified two GSH conjugates formed from incubations of clozapine with horseradish peroxidase. It is unclear how the clozapine free radical would lead to formation of a GSH conjugate because it would be more likely to abstract a hydrogen atom from GSH, producing a glutathionyl radical and regenerating parent compound. More recently, we identified a reactive metabolite of clozapine that covalently binds to activated human neutrophils (Liu and Uetrecht, 1995), and others have demonstrated that it binds to bone marrow cells (Maggs et al., 1995). Formation of this reactive metabolite could lead to drug-induced agranulocytosis, either by covalently modifying critical cellular proteins, leading to cell death, or perhaps by acting as a hapten and eliciting an immune response. Although the evidence for an antibody-mediated immune response against clozapine is weak, it is conceivable that hapten formation could trigger a cell-mediated immune response (Liu and Uetrecht, 1995). Furthermore, recent work from the laboratory of Park has shown that exposure of human neutrophils to the clozapine reactive metabolite in vitro results in toxicity (Williams et al., 1998).

Regardless of the mechanism of clozapine-induced agranulocytosis, the seriousness of this problem has lead to attempts to find new therapeutic agents that share the unique pharmacological activity of clozapine without causing drug-induced agranulocytosis. One such agent is olanzapine, which is a thiobenzodiazepine derivative (Fig. 1). Despite the structural similarity between clozapine and olanzapine, no cases of agranulocytosis have been reported with the clinical use of olanzapine (Fulton and Goa, 1997). In fact, olanzapine has been used safely to treat patients who have had clozapine-induced agranulocytosis (Fulton and Goa, 1997). In this study, we evaluated the ability of olanzapine to form a reactive intermediate in a reaction analogous to that occurring with clozapine. In addition, we investigated the toxicity of clozapine and olanzapine reactive metabolites toward human leukocytes and the promyelocytic HL-60 cell line.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. Clozapine was provided by Sandoz Canada (now Novartis, Dorval, Quebec, Canada). Hydrogen peroxide (H₂O₂) was purchased from ACP Chemicals (Montreal, Quebec, Canada). Sodium hypochlorite (NaOCl) was purchased from Aldrich Chemical (Milwaukee, WI). Ketoprofen and dapsone were purchased from Sigma Chemical (Oakville, Ontario, Canada). MPO was obtained from Cortex Biochemical (San Leandrow, CA). One unit of MPO activity is defined as the amount of enzyme that would decompose 1 µmol of H₂O₂/min at 25° and pH 6. HBSS was obtained from Media Services, University of Toronto.

Preparation of olanzapine. Olanzapine was isolated from tablets (7 10-mg tablets, Zyprexa; Eli Lilly, Indianapolis, IN). The tablets were crushed and extracted with methanol (10 ml). The methanolic solution was centrifuged (500 × g, 5 min) to remove insoluble material. This was repeated three times, the methanolic solutions were combined, and the solvent was removed under reduced pressure. The residue was dissolved in water and applied to an LC-18 solid-phase extraction column (Supelco; Supelco Inc., Bellefonte, PA). The solid-phase extraction column was washed sequentially with water and 5% methanol, and the olanzapine was eluted with methanol. The drug was shown to be >99% pure by HPLC.

Analytical studies. HPLC was performed using a Shimadzu system (LC-600 pump; SPD-6A UV spectrophotometer set at 254 nm, and a C-R6A integrator; Shimadzu, Kyoto, Japan). An Ultracarb ODS 30 column (2 × 100 mm, 5 µm; Phenomenex, Torrance, CA) equipped with a 2 × 30 mm guard column was used for the chromatography. The mobile phase consisted of water/acetonitrile/acetic acid (79:20:1, v/v/v) containing 2 mM ammonium acetate unless stated otherwise and was degassed before use.

Oxidation of olanzapine by hypochlorous acid. A Hewlett Packard diode-array spectrophotometer (HP 8452A; Hewlett Packard, Palo Alto, CA) was used to determine the rate of oxidation of olanzapine by hypochlorite. Scanning of the reaction mixture was initiated immediately after the addition of NaOCl (40 µl, 5 mM aqueous solution) to a solution of the drug [2 ml, 100 µM in phosphate buffer (0.1 M, pH 6.0)] with rapid stirring. A Hi-Tech stopped-flow spectrophotometer (Stopped-Flow SHU; Hi-Tech Scientific., Salisbury, UK; dead time, 2 msec) was used to obtain accurate kinetic data on the oxidation of olanzapine and clozapine by HOCl. Concentrations of drug and NaOCl were 500 and 50 µM, respectively. Reactions were performed in PBS (137 mM sodium chloride, 8 mM disodium hydrogen phosphate, 1.5 mM potassium dihydrogen phosphate, and 2.7 mM potassium chloride, pH 6.0) or in phosphate buffer (0.1 M, pH 6.0), and the reactive intermediates were monitored at 540 nm (olanzapine) or 460 nm (clozapine).

Trapping of the olanzapine reactive intermediate with glutathione and NAC. NaOCl (250 µl, 1 M aqueous solution) was added to olanzapine (5 ml, 20 mM in an aqueous solution of 60% ethanol with the pH adjusted with 20 µl of acetic acid) with rapid stirring. The solution immediately became dark red, and GSH or NAC (2 ml, 0.2 M aqueous solution) was quickly added to the solution. The reaction products were analyzed by LC/MS. The mobile phase consisted of water/acetonitrile/acetic acid (79:20:1, v/v/v) containing 2 mM ammonium acetate.

Purification of NAC adducts of the olanzapine reactive intermediate. After reaction of the olanzapine reactive intermediate with NAC (as described above), the sample was concentrated under reduced pressure. The pH was adjusted to 9.0 with NaHCO₃, and the solution was extracted with ethyl acetate (three times 15 ml). The aqueous layer was adjusted to pH 6.0 with 1 N HCl and applied to an LC-18 solid-phase extraction column (Supelco; Supelco Inc.). The solid-phase extraction column was washed sequentially with water and 5% methanol (v/v), and then the NAC conjugates were eluted with methanol. The methanolic solution was separated by normal-phase thin layer chromatography (100% methanol mobile phase), resulting in five bands. Two bands (retention time, 0.2 and 0.4) contained the olanzapine-NAC conjugates. The olanzapine-NAC conjugates were purified further by preparative HPLC (Ultracarb 5 ODS 30, 150 × 10 mm; Phenomenex, Torrance, Ca) using a mobile phase of water/acetonitrile/acetic acid (84:15:1, v/v/v) at a flow rate of 4 ml/min.

Oxidation of olanzapine by human MPO. Olanzapine (100 µM, in 1 ml PBS, pH 7.4) was incubated with MPO (1 unit) in the presence of H₂O₂ (100 µM). After incubation for 30 min at 37°, the reaction was stopped by cooling the sample on ice, and the mixture was analyzed by HPLC or LC/MS. GSH (1 mM), NAC (1 mM), or N-acetyllysine (1 mM) was included in some incubations. Control experiments involved the omission of olanzapine, H₂O₂, or MPO from the reaction mixture.

Human leukocyte isolation. Neutrophils and mononuclear cells were isolated from venous blood of healthy volunteers by differential centrifugation over Ficoll-Paque as described in detail previously (Liu and Uetrecht, 1995). Peripheral blood mononuclear cells (2 × 10⁶ cells/ml) were resuspended in RPMI 1640 (Media Services, University of Toronto) containing 10% heat-inactivated fetal bovine serum, 4 mM glutamine, 60 µg/ml penicillin, and 100 µg/ml streptomycin and then aliquoted onto 12-well plates (1 ml/well; Corning Glassworks, Corning, NY). The plates were incubated for 2 hr at 37° in an incubator (5% CO₂) to isolate monocytes (adherent cells) from T and B lymphocytes (nonadherent).

Toxicity of the reactive intermediates of olanzapine and clozapine to neutrophils. Three different protocols were used to produce reactive metabolites. First, reactive metabolite was produced chemically by allowing clozapine or olanzapine to react with HOCl. Drug was dissolved in PBS (100 µl, pH 7.4), and 0.9 molar equivalent of NaOCl (dissolved in PBS, pH 6.0, 100 µl) was added. This results in the formation of an orange (clozapine) or red (olanzapine) reactive metabolite. Immediately, neutrophils (4 × 10⁶ in 1 ml of HBSS) were added to the reactive metabolite solution. Assuming complete reaction between drug and NaOCl, the concentrations of reactive metabolite produced were 0.18, 1.8, or 18 µM. Given the rapid reaction between hypochlorite and drug (Liu and Uetrecht, 1995), the use of molar excess of drug ensures that the neutrophils are not exposed to NaOCl, which may be directly toxic to the cells. In the second protocol, 4 × 10⁶ neutrophils were incubated in 1 ml of HBSS containing drug and MPO (1 unit), and the reaction was started by the addition of 200 µl of PBS containing H₂O₂. In the final 1.2-ml volume, the concentration of drug was 0, 0.2, 2, or 20 µM, and the H₂O₂ concentration was 10 mM. In the third protocol, 4 × 10⁶ neutrophils were incubated in 1 ml of HBSS containing drug, and the reaction was started by the addition of 200 µl of HBSS containing H₂O₂. In the final 1.2-ml volume, the concentration of drug was 0, 0.2, 2, or 20 µM, and the H₂O₂ concentration ranged from 0 to 10 mM. After initiation of the reaction by one of the protocols outlined above, cells were incubated at 37° for 2 hr in a shaking water bath. At the end of this period, the cells were counted using a hemocytometer, and cell viability was assessed by trypan blue exclusion. In some experiments, the time course of clozapine toxicity was investigated. In these experiments, 100-µl aliquots were removed at 0, 15, 30, 60, 90, and 120 min. At each time point, cells were counted using a hemocytometer, and cell viability was assessed by trypan blue exclusion. In a series of additional experiments, neutrophils were incubated in the presence of clozapine (20 µM) and H₂O₂ in the presence of either 10 or 100 µM dapsone or ketoprofen.

Toxicity of the reactive intermediates of olanzapine and clozapine to HL-60 cells and monocytes. HL-60 (human promyelocytic) cells were grown in RPMI 1640 containing 10% heat-inactivated fetal bovine serum, 4 mM glutamine, 60 µg/ml penicillin, and 100 µg/ml streptomycin and passaged twice a week. Cell culture was carried out at 37° in a humidified atmosphere containing 5% CO₂. The number of cells was determined using a hemocytometer, and cell viability was determined by trypan blue exclusion. Initial viability of the cells used in these experiments was always >95%. For exposure to drug reactive metabolites, cells were washed three times in HBSS, resuspended to 2 × 10⁶ cells/ml, and then exposed to reactive metabolites as described above for neutrophils.

Statistical analysis. Statistical tests were performed using InStat 2.01 (GraphPAD Software, San Diego, CA). Unless otherwise stated statistical significance was tested by one-way analysis of variance with Student-Newman-Keuls post-test. Values of p <0.05 were considered statistically significant.

	Results

Top Summary Introduction Procedures Results Discussion References

Oxidation of olanzapine by hypochlorous acid. Olanzapine reacted with hypochlorous acid to produce an unstable intermediate with a _max of 540 nm (Fig. 2). Stopped-flow spectrophotometer measurements revealed that under the conditions used, the reactive intermediates of clozapine and olanzapine had formation rate constants (K, 1/sec) of 225 ± 0.7 and 213 ± 20 (mean ± standard deviation; three or four determinations), respectively. The olanzapine reactive intermediate had a half-life of 35.5 ± 0.7 sec. The disappearance of the clozapine reactive intermediate was best described by a biexponential equation. The two half-lives were 5.1 ± 0.7 and 45.0 ± 3.5 sec (mean ± standard deviation; three determinations). The longer half-life is equivalent to the value reported previously for the disappearance of the clozapine reactive intermediate (Liu and Uetrecht, 1995). In the absence of chloride anion, the disappearance of the clozapine reactive intermediate was described by a biexponential equation with half-lives of 5.06 and 93.52 sec.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Repetitive absorption spectra from the oxidation of olanzapine by HOCl. Cycle time, integration time, and total run time were 2, 0.5, and 30 sec, respectively. Drug and NaOCl concentrations were 100 µM in PBS, pH 6.0. Dashed line, UV absorbance spectrum of parent olanzapine. Dotted line, initial spectrum recorded after the addition of NaOCl.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Mass spectrum of the reactive intermediate in the reaction of olanzapine with HOCl obtained with a flow system coupled to a Sciex API III mass spectrometer.

Oxidation of olanzapine by the MPO/H₂O₂/Cl system. When olanzapine was incubated with MPO and H₂O₂ in the presence of chloride, three species with MH⁺ ions at m/z 329 were detected, which corresponds to the addition of oxygen to olanzapine. The three species had retention times of 1.5, 2.2, and 5.5 min. LC/MS/MS analysis revealed that the conjugate with a retention time of 1.5 min showed major peaks at m/z 329 (24%), 272 (loss of CH₂CHNHCH₃, 48%), 242 (40%), 229 (loss of methylpiperazine, 100%), 213 (52%), 188 (63%), and 84 [CH₂CHN(CH₃)CH₂CH₂⁺, 37%]. LC/MS/MS of the conjugate with a retention time of 2.2 min showed major fragments at m/z 329 (17%), 242 (35%), 229 (100%), 213 (60%), and 188 (37%). LC/MS/MS of the conjugate with a retention time of 5.5 min showed major fragments at m/z 329 (26%), 159 (19%), 146 (79%), 133 (100%), and 84 (38%). Inclusion of GSH in the incubation resulted in the formation of two olanzapine-GSH conjugates (with protonated molecular ions at m/z 618), which were detectable by LC/MS. The conjugates had retention times of 1.4 and 2.2 min. In the absence of MPO, no GSH conjugates were observed, suggesting that there is no direct reaction between olanzapine and H₂O₂ leading to the formation of a reactive intermediate. The corresponding NAC conjugates were observed when NAC was included in the incubation, but no olanzapine conjugates were detectable when N-acetyllysine was included in the incubation (data not shown).

Toxicity of olanzapine and clozapine reactive intermediates toward human leukocytes and HL-60 cells. The toxicity of clozapine toward human neutrophils varied with the method used to generate reactive metabolites (Fig. 4). Although no toxicity was observed when NaOCl was used (Fig. 4A), significant toxicity was observed at clozapine concentrations of 2 and 20 µM when H₂O₂ (10 mM) was included in the incubations (Fig. 4C). The toxicity was found to be dependent on both clozapine (Fig. 4C) and H₂O₂ concentration (Fig. 5). The addition of exogenous MPO (1 unit) to the reaction mixtures in the presence of H₂O₂ (10 mM) had little effect on the amount of toxicity observed, and under both conditions, the major effect of clozapine was to produce a decrease in the number of neutrophils with few blue-stained cells being observed. Time course studies revealed that toxicity was not detectable at 15 min after the initiation of reactive metabolite generation (Fig. 6). By 30 min, statistically significant toxicity was observed, and the number of viable neutrophils continued to decrease until the experiment was ended (120 min). No toxicity was observed with clozapine alone at concentrations up to 50 µM (data not shown). Although it seemed that there might be some toxicity at the highest concentration of olanzapine tested, this did not reach statistical significance.

View larger version (37K): [in this window] [in a new window]   Fig. 4.   Toxicity of clozapine (open bars) or olanzapine (striped bars) reactive intermediates to human neutrophils. Drug concentration varied between 0 and 20 µM as indicated. Three methods were used to generate the reactive intermediates: (A) reactive metabolite was produced chemically by reacting drug with NaOCl; (B) neutrophils were incubated with drug and MPO (1 unit), and the reaction was started by the addition of H2O2 (10 mM); (C) neutrophils were incubated with drug, and the reaction was started by the addition of H2O2 (10 mM). Cells not receiving drug (0) were treated with (A) 20 µM NaOCl, (B) 1 unit of MPO and H2O2 (10 mM), or (C) H2O2 (10 mM). Cells then were incubated at 37° for 2 hr. Results are expressed as percentage of the number of viable cells present in untreated cell samples incubated at 37° for 2 hr. Values are mean ± standard error of 5 (olanzapine) or 6-13 (clozapine) individual experiments. *, p < 0.05; **, p < 0.001, Statistical differences compared with values in the absence of drug (0).

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Influence of H2O2 concentration on clozapine-induced toxicity toward neutrophils. Neutrophils were incubated at 37° for 2 hr with clozapine (CLOZ) (0 or 20 µM) and H2O2 (0.1-10 mM) as described in the text. Values are mean ± standard error of four individual experiments. The number of viable cells is expressed relative to the number of cells in control incubations (without clozapine or H2O2). *, p < 0.05, Statistically different (Student's t test) from cells incubated with the equivalent H2O2 concentration in the absence of clozapine.

View larger version (20K): [in this window] [in a new window]   Fig. 6.   Time course of clozapine toxicity toward human neutrophils. Neutrophils were incubated at 37° for various time periods with clozapine (CLOZ) (0 or 20 µM) and H2O2 (10 mM) as described in the text. Control cells were incubated in HBSS alone. Values are mean ± standard error of three individual experiments. The number of viable cells is expressed relative to the initial value (time 0) determined for each experiment. *, p < 0.05; **, p < 0.001, Statistically different from the viable cell number at time 0.

View larger version (42K): [in this window] [in a new window]   Fig. 7.   Toxicity of clozapine (open bars) and olanzapine (striped bars) toward human monocytes. Monocytes were incubated at 37° for 2 hr with drug concentrations of 0-20 µM and H2O2 (10 mM). Toxicity was assessed by trypan blue exclusion. Control cells were incubated in HBSS alone. Values are mean ± standard error of three or four individual experiments. *, p < 0.01, Statistically different from those of cells receiving H2O2 alone.

View larger version (20K): [in this window] [in a new window]   Fig. 8.   Effect of dapsone and ketoprofen on clozapine toxicity toward human neutrophils. Neutrophils were incubated at 37° for 90 min with 20 µM clozapine, 10 mM H2O2 (CLOZ), and dapsone (10 or 100 µM) or ketoprofen (10 or 100 µM) as indicated. H2O2, cells incubated with H2O2 in the absence of clozapine. Control, neutrophils incubated in the absence of both clozapine and H2O2. Values are mean ± standard error of four individual experiments. *, p < 0.01, Statistically different from those of cells incubated with clozapine and H2O2.

Toxicity of clozapine reactive intermediates toward neutrophils from patients who have had clozapine-induced agranulocytosis. When neutrophils from patients who had clozapine-induced agranulocytosis were incubated with HOCl-generated clozapine reactive intermediate at concentrations up to 20 µM, no toxicity was observed (data not shown). When neutrophils from these patients were incubated with clozapine in the presence of H₂O₂, the cells seemed to be more sensitive to the toxic effects than those from normal control subjects; however, the difference was small and did not allow clear differentiation from control cells (Fig. 9).

View larger version (13K): [in this window] [in a new window]   Fig. 9.   Neutrophils from healthy volunteers () (5-10) or patients who had had clozapine-induced agranulocytosis () (2) were incubated with clozapine (0-20 µM) and H2O2 (10 mM) for 2 hr at 37°. The results are expressed as the percentage of the number of viable cells present in untreated cell samples incubated at 37° for 2 hr.

	Discussion

Top Summary Introduction Procedures Results Discussion References

The results of these experiments demonstrate that like the structurally related antipsychotic drug clozapine, olanzapine is oxidized to a reactive intermediate by HOCl and by the MPO/H₂O₂/Cl system. It was possible to trap the reactive olanzapine metabolite using soft nucleophils, such as GSH and NAC, to form two olanzapine-nucleophile conjugates. However, the olanzapine reactive intermediate did not react with the harder nucleophil, N-acetyllysine. The two olanzapine-NAC conjugates had very similar mass spectra, suggesting that they have similar chemical structures. The observation that the same GSH conjugates were formed when olanzapine was incubated with activated neutrophils provides strong evidence that the reactive intermediate was a true metabolite.

Using a flow system, the molecular ion of the reactive olanzapine intermediate was observed to be m/z 311, which is 1 mass unit less than that of the parent compound. It is probable that the only hydrogen that could be lost to form an intermediate of this mass and react with NAC or GSH to form the observed conjugates is the N10 hydrogen (Fig. 1). Because the reactive species observed in the mass spectrometer has a positive charge, it is formally a nitrenium ion; however, the positive charge must be highly delocalized. The formation of the reactive metabolite from olanzapine described in this report is entirely analogous to the formation of a clozapine reactive metabolite we reported previously (Liu and Uetrecht, 1995).

Although the use of clozapine has been associated with a relatively high incidence of agranulocytosis (0.8%), no cases of agranulocytosis have been reported after the administration of olanzapine. Because the two compounds produced similar reactive metabolites, we wanted to determine whether exposure of human neutrophils and monocytes and the promyelocytic HL-60 cell line to the reactive metabolites of clozapine and olanzapine resulted in toxicity. Three different methods were used to produce the reactive metabolite. The toxicity of clozapine and olanzapine reactive metabolites toward target cells was found to vary both with the drug used and with the method of reactive metabolite generation. With clozapine, no toxicity was observed when NaOCl was added to drug to produce the reactive intermediate that was then added to the cells (Fig. 4A). However, when neutrophils were incubated with clozapine and H₂O₂ (in the presence or absence of exogenous myeloperoxidase), significant toxicity toward human neutrophils was observed (Fig. 4). No toxicity was observed in the absence of H₂O₂. The most likely explanation of this finding is that the H₂O₂ enters the cell and is used as a cofactor by intracellular enzymes (presumably MPO) to generate the cytotoxic clozapine reactive metabolite. It was not possible to detect oxidation of clozapine after mixture of the drug with H₂O₂, suggesting there is no direct chemical reaction between the two substances (Liu and Uetrecht, 1995). Cytotoxicity also was observed when HL-60 cells and monocytes were exposed to clozapine and H₂O₂ (in the presence or absence of exogenously added myeloperoxidase). Because both HL-60 cells (Koeffler et al., 1985; Murao et al., 1988) and monocytes (Winterbourn, 1989) are known to express MPO, these results are consistent with the MPO-catalyzed conversion of clozapine to a cytotoxic metabolite. Further support for this hypothesis comes from the data showing that dapsone can inhibit the toxicity produced by clozapine (Fig. 8). Previous studies have shown that dapsone is both an inhibitor (Kettle and Winterbourn, 1991) and a substrate for human MPO (Uetrecht et al., 1993). Although it is not possible to discount the involvement of other pharmacological effects of dapsone in the inhibition of clozapine induced toxicity, it is of interest that ketoprofen, an inhibitor of cyclooxygenase, was unable to prevent the clozapine-induced toxicity.

We cannot be certain why only the H₂O₂-based methods of reactive metabolite production should cause toxicity, but some explanations can be proposed. One possibility is that the toxicity may be related to the length of time to which the cells are exposed to drug reactive metabolite. In the HOCl method, exposure of the cells to reactive intermediate occurs for only a short initial period of time. In the H₂O₂ methods, the reactive metabolite presumably is generated over a longer time period; hence, the duration of exposure of the cells to reactive metabolite is longer. A second possible explanation is that when cells are exposed to HOCl-generated reactive metabolite, the reactive metabolite is not able to reach critical sites inside the cells, but when H₂O₂ is added, the reactive metabolite can be generated intracellularly. Support for this hypothesis comes from studies showing that extracellular GSH can inhibit covalent binding of HOCl-generated clozapine reactive intermediate to neutrophils but does not inhibit covalent binding of H₂O₂-generated clozapine reactive metabolite (Gardner et al., 1998). These two hypotheses are not mutually exclusive.

Although incubation of clozapine and H₂O₂ with neutrophils produced toxicity, no statistically significant toxicity was observed when olanzapine and H₂O₂ (or HOCl) were incubated with human neutrophils, monocytes, or HL-60 cells. Thus, although both olanzapine and clozapine produce reactive nitrenium ions, these metabolites seem to have differing abilities to cause cytotoxicity. This difference was surprising because the two nitrenium ions seem to have similar chemical reactivities. In particular, both clozapine and olanzapine reactive metabolites preferentially bind to S-nucleophils, with little reactivity toward harder nucleophils such as N-acetyllysine being observed. However, although the reaction to simple nucleophiles in vitro was similar, the covalent binding of the two drugs to neutrophil proteins under the same conditions that led to clozapine toxicity in these studies was significantly different (Gardner et al., 1998). This difference in covalent binding of reactive metabolite to cellular protein may explain the differences in toxicity.

The observation that the reactive metabolite of clozapine is toxic to neutrophils, as well as HL-60 cells, and that of olanzapine is not under the same conditions could be taken as evidence that clozapine-induced agranulocytosis is due to the cytotoxicity of the reactive metabolite. In addition, the observation that the clozapine metabolite seems to be more toxic to neutrophils from patients with a history of clozapine-induced agranulocytosis supports this hypothesis. However, if direct cytotoxicity were the mechanism, it would be somewhat difficult to explain the idiosyncratic nature of clozapine-induced agranulocytosis and the long delay between starting the drug and the onset of toxicity. Furthermore, we have been unable to induce neutropenia in animals despite treatments with high doses of clozapine and additional treatments to increase the formation of the reactive metabolite in bone marrow (Uetrecht JP, unpublished observations). Phenytoin, carbamazepine, and sulfamethoxazole are examples of other drugs in which the reactive metabolite of the drug seems to be more toxic to cells from patients with a history of a severe idiosyncratic reaction to that drug, but the adverse reaction does not seem to be a simple cytotoxic reaction (Shear and Spielberg, 1988; Rieder et al., 1989). It may be that cytotoxicity plays a role in such reactions by leading to an increase in the phagocytosis of drug-modified cells, which may in turn help to stimulate an immune reaction leading to a severe idiosyncratic reaction. However, there is no strong evidence that clozapine-induced agranulocytosis is immune mediated, and it is difficult to explain the long delay (6 weeks in several cases) in the onset of clozapine-induced agranulocytosis on reexposure of patients with a previous history of clozapine-induced agranulocytosis if the mechanism were a simple immune reaction against drug-modified cells (Safferman et al., 1992).

This in vitro test system showed an interesting difference in the ability of clozapine and olanzapine to induce toxicity toward human neutrophils, but the methods used have some limitations. Induction of toxicity required exposure of neutrophils to clozapine in the presence of high concentrations of H₂O₂. The neutrophils were relatively resistant to short term exposure to H₂O₂, but a variable amount of toxicity was observed (loss of up to 25% of cells present in control incubations) in the absence of clozapine (Fig. 9). This background toxicity means that the method is not very sensitive toward a small amount of drug-induced toxicity. In monocytes, which have lower levels of MPO (Klebanoff, 1990), statistically significant toxicity was observed in the absence of clozapine; in monocyte-depleted lymphocytes, it was not possible to observe toxic effects of clozapine because ~50% of cells died after exposure to H₂O₂ (10 mM) in the absence of clozapine (data not shown). This value is similar to a published LC₅₀ value for H₂O₂ toward human lymphocytes (O'Donnell-Tormey et al., 1985).

In summary, we demonstrated that like clozapine, olanzapine is oxidized to a reactive nitrenium ion by HOCl, the major oxidant produced in activated neutrophils, and by an MPO/H₂O₂/Cl system. However, the olanzapine reactive metabolite has a lower propensity to cause toxicity toward human neutrophils, monocytes, and HL-60 cells than the reactive clozapine nitrenium ion in an in vitro cytotoxicity assay. The lower toxic potential of the olanzapine reactive metabolite in conjunction with the lower therapeutic plasma concentrations of olanzapine (Aravagiri et al., 1997) compared with clozapine (Weigmann and Hiemke, 1992) (0.1 µM versus 2 µM) may help to explain why compared with clozapine, olanzapine is not associated with agranulocytosis in humans. Furthermore, understanding the mechanism of toxicity of clozapine in this model system may provide new insights into the mechanism of clozapine-induced agranulocytosis, which currently is undefined.