See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/234823631

Mechanism of clozapine-induced agranulocytosis. Current status of research

and implications for drug development

Article  in  CNS Drugs · February 1997

DOI: 10.2165/00023210-199707020-00005 · Source: PubMed

CITATIONS READS

77 5,133

2 authors:

Munir Pirmohamed Brian Kevin Park

University of Liverpool University of Liverpool
836 PUBLICATIONS   30,229 CITATIONS    1,023 PUBLICATIONS   34,224 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

UKRMP Safety and Efficacy of Regenerative Medicine Therapies View project

M420 deletion View project

All content following this page was uploaded by Munir Pirmohamed on 21 May 2014.

The user has requested enhancement of the downloaded file.

ADVERSE EFFECTS eNS Drugs 1997 Feb: 7 (2) 139-158
1172-7047/97/0002-0139/$10.00/0

© Adis Internotionalumlted. All rights reserved.

Mechanism of Clozapine-Induced
Agranulocytosis
Current Status of Research and Implications for
Drug Development
Munir Pirmohamed and Kevin Park
Department of Pharmacology and Therapeutics, The University of Liverpool, Liverpool, England

Contents
Summary ..................... . 139
1. General Overview of Adverse Drug Reactions . 141
1.1 Types of Adverse Drug Reactions . . . . . . 141
1.2 Mechanisms of Idiosyncratic Adverse Drug Reactions 141
2. Characteristics of Clozapine-Induced Agranulocytosis. 143
2.1 Syndromes.. 143
2.2 Epidemiology 143
2.3 Risk Factors . 144
2.4 Predictors .. 144
3. Other Haematological Abnormalities Induced by Clozapine 144
4. Mechanisms of Clozapine-Induced Agranulocytosis . . . . . 145
4.1 Site of Toxicity - Central or Peripheral? . . . . . . . . . . 145
4.2 Is Toxicity Due to Clozapine Itself and Is It Related to the
Pharmacology of the Drug? . . . . . . . . . . . . . . . . 146
4.3 Is Toxicity Due to Stable or Chemically Reactive Metabolites of Clozapine? . 147
4.4 Is Toxicity Direct or Immune Mediated? . . . . . . . . 148
4.5 What Determines Individual Susceptibility? . . . . . . 149
5. Comparison with Amodiaquine-Induced Agranulocytosis 152
6. Areas for Further Research . . . . . . . . 152
7. Prospects for Future Drug Development 154
8. Conclusion . . . . . . . . . . . . . . . . . 155

Summary Clozapine is an atypical antipsychotic agent that has several advantages over
conventional anti psychotics, not least of which is its superior efficacy. However,
the high risk of agranulocytosis (0.8% of patients) associated with c10zapine
therapy has resulted in restricted indications for its use.
The mechanism of c1ozapine-induced agranulocytosis is not clear. The target
cells affected are the myeloid precursors, although the mature neutrophil may
also be targeted simultaneously. There is no convincing evidence of direct toxicity
of the parent compound or its stable metabolites (demethyl-c1ozapine and c1ozap-
ine N-oxide). Clozapine is also metabolised by liver microsomes, peripheral
blood neutrophils and their bone marrow precursors to a chemically reactive
intermediate that has been postulated to be a nitrenium ion. This toxic metabolite
140 PirnlOhamed & Park

has been shown to covalently bind to neutrophil proteins, suggesting that it may
be involved in the pathogenesis of the toxicity. However, it is not clear how
toxicity is mediated. The nitrenium ion may bind to essential cellular proteins
and disrupt neutrophil function or, alternatively, it may act as a hapten and initiate
an immune reaction resulting in immune-mediated destruction of the neutrophil.
Indirect evidence exists to support both mechanisms, although clear direct evi-
dence is still lacking. The role of cytokines and apoptosis in the pathogenesis of
the agranulocytosis is unclear.
The reason why only approximately I % of individuals who are treated with
clozapine are affected by agranulocytosis has not been elucidated. Evidence ex-
ists to implicate both the major histocompatibility complex antigens and heat
shock protein variants in determining individual susceptibility, although more
patients of different ethnic backgrounds need to be studied.
The ultimate aim of research into clozapine-induced agranulocytosis should
be to either prospectively predict which individuals are going to develop
agranulocytosis and/or to develop analogues that retain efficacy but are not toxic.
The former is complicated by the fact that predisposition may be multifactorial,
and thus prediction may require multiple tests that may be of statistical but not
absolute validity. The latter depends on identifying the mechanism of toxicity and
the chemical characteristics of clozapine that are responsible for the toxicity. This
knowledge may allow rational design of new analogues that do not cause
agranulocytosis.

Clozapine, a dibenzodiazepine compound, is an In order to prevent fatalities from clozapine-in-

atypical antipsychotic that has a unique history (ta-
ble 1). In the mid 1970s, it was either withdrawn or
its use restricted in most countries as a result of the
high risk of agranulocytosis associated with its use.
By the late 1980s, largely because of the unique
pharmacological profile of the drug, the situation
had changed - its use had greatly increased, but
was accompanied by careful patient selection
and/or haematological monitoring.
Clozapine shows superior efficacy when com-
pared with conventional anti psychotics such as
chlorpromazine,1 I I and is now indicated primarily
in patients who have schizophrenia that is resistant
to conventional agents. Additional major advan-
tages of clozapine over conventional antipsychot-
ics are that it:
• has a lower incidence of extrapyramidal adverse
effects including tardive dyskinesia;121
• has a beneficial effect on the negative symptoms
of schizophrenia;13 1
• reduces suicidality.13 1
These advantages, however, are mitigated by its
propensity to cause agranulocytosis.1 41
duced agranulocytosis, in most countries, all pa-
tients being treated with clozapine are required to
have haematological monitoring. ISI The extent of
this monitoring differs in different countries. For
example, in the UK, white blood cell counts are
monitored weekly for the first 18 weeks of treat-
ment, fortnightly from 18 to 52 weeks, and
monthly thereafter. In the US, in contrast, the moni-
toring requirements are more stringent: patients are
monitored weekly for the entire duration of treat-
ment.
Since its reintroduction, a great deal of research
has been undertaken in an attempt to elucidate the
mechanism(s) of clozapine-induced agranulocyto-
sis. The purpose of this review is to critically eval-
uate what has been learnt from this research, what
still needs to be determined and how this may help
in future drug development. Clearly, the ultimate
aim of this research is to improve the benefit-risk
ratio associated with clozapine therapy, either by
developing a congener that has similar pharmaco-
logical properties but does not cause agranulocyto-

~) Adis International Limited. All rights reserved eNS Drugs 1997 Feb: 7 (2)
Mechanism of Clozapine-Induced Agranulocytosis
Table I. The regulatory and clinical history of clozapine
Year Event
1960 Developed by Hunziker and colleagues[11
1972 Marketed in Europe
1975 8 deaths from agranulocytosis (of 16 cases)
reported from Finland
1976 Clozapine either withdrawn or use restricted
in most countries
Mid-1980s Clozapine trials started as a result of
pressure from psychiatrists in the absence
of other compounds effective in
treatment-resistant schizophrenia
1989 Clozapine approved by the US Food and
Drug Administration and the UK Committee
on Safety of Medicines
1990 Reintroduced for patients with
schizophrenia that is resistant to
conventional therapy. Weekly monitoring of
white blood cell count mandatory (Clozaril@
Patient Monitoring Scheme)
sis, or to prospectively identify at-risk patients and
thus avoid the toxicity.
1. General Overview of
Adverse Drug Reactions
Before discussing cIozapine-induced agranulo-
cytosis, a brief overview of the different types of
adverse drug reactions is given together with a dis-
cussion of the mechanisms of idiosyncratic drug
reactions.
1.1 Types of Adverse Drug Reactions
In general, adverse drug reactions can be di-
vided into 2 types, A and B.l 6,71 Type A reactions
are an exaggeration of the normal pharmacological
effects of the drug and thus can be predicted from
the known pharmacology of the drug. These reac-
tions are usually dose-dependent and can be pre-
vented by dose reduction. They are common, usu-
ally not severe and are detected early during the
development of the drug. Agranulocytosis and
bone marrow depression occurring with cytotoxic
chemotherapeutic agents usually fall into this cat-
egory.
Type B reactions are also known as idiosyn-
cratic reactions. They cannot be predicted from the
© Adis International Umited. All rights reserved.
141
pharmacology of the drug, are dose-independent,
and tend to be influenced by host factors (which
may be genetic or environmental). They are less
common than type A reactions, but tend to be more
severe and occasionally result in fatalities. Be-
cause they are uncommon, they are usually not de-
tected during the early phase of drug development.
From the evidence available, cIozapine-induced
agranulocytosis falls into this category.
1.2 Mechanisms of Idiosyncratic
Adverse Drug Reactions
Many different mechanisms have been postu-
lated as being responsible for the idiosyncratic
toxicity caused by drugs, and these have been re-
viewed elsewhere ,171
The parent compound, or more usually its me-
tabolite(s), have often been implicated in the mech-
anism of drug-induced white blood cell toxicity. Of
particular importance with regard to the metabo-
lism of the drug is the formation of chemically
reactive intermediates, a process termed 'bioacti-
vation'17-1I1 (fig. 1). These are unstable, highly re-
active species that can bind covalently to cellular
macromolecules and cause toxicity by 2 mecha-
nisms (fig. 1),17.9 1 First, they may interfere with
essential functions of the cell (i.e. direct cell toxi-
city) and cause cell death. The typical example of
this is hepatic necrosis resulting from paracetamol
(acetaminophen) overdosage.l12-151 The mecha-
nism by which cell death occurs is not fully under-
stood. Secondly, the reactive species may act as a
hapten and initiate an immune reaction (i.e. indi-
rect or immune-mediated toxicity)p,9,16 1The im-
mune reaction may be directed against: (i) the drug
antigen; (ii) a neoantigen created by the interaction
of the metabolite with cellular protein; (iii) an auto-
antigen; or (iv) towards more than one of these
antigens. I 171 The immune reaction may be charac-
terised by antibody production (humoral immu-
nity), drug-specific T lymphocytes (cellular immu-
nity) or a combination of the two.
In the majority of individuals, bioactivation is
counter-balanced by the detoxification mecha-
nisms that are present in most cell types. lg ,9 1
eNS Drugs 1997 Feb: 7 (2)
142 Pirmo/wmed & Park

Liver Neutrophil

CytotoxiCity Sensitisation

Necrosis!
apoptosis

Fig. 1. The postulated role of drug metabolism in the pathogenesis of agranulocytosis. Drugs may be metabolised by either the liver
or neutrophils (or their precursors) to chemically reactive intermediates, a process that is termed 'bioactivation'. Bioactivation usually
represents a minor metabolic pathway, the conversion of drugs to stable metabolites being by far the most common biotransformation.
If the chemically reactive metabolite is not bioinactivated, it may bind to neutrophil proteins and cause either direct toxicity resulting
in cell death (i.e. cytotoxicity) or act as a hapten and initiate immune-mediated toxicity (i.e. hypersensitivity). With compounds other
than clozapine, both mechanisms have been shown to result in agranulocytosis.

Clearly, those individuals who have an imbalance
between bioactivation and detoxification are most
likely to develop idiosyncratic toxicity. Of rele-
vance here is the site of metabolism of the drug.
Different tissues in the body will have different
complements of drug activation and drug detoxifi-
cation enzymes, and this may be one factor deter-
mining site-specific toxicity. I'll The liver is obvi-
ously the commonest site of metaboli sm in the
body.!lgl Thus, metabolism within the liver leading
to hepatic injury, as occurs in paracetamol overdos-
age, is easy to understand since metabolism and
toxicity are occurring within the same organ.
With regard to extra-hepatic toxicity, it is diffi-
cult to envisage how a chemically reactive interme-
diate (often with a half-life of less than I minute)
can be formed within the liver and then travel in
the circulation to its site of toxicity.!9 1 Therefore, it
is likely that metabolism and bioactivation also oc-
curs in extra-hepatic tissues (fig . I). With regard to
drugs causing agranulocytosis, there is evidence to
support metabolism within white blood cells as be-
ing responsible for the toxicity.IID,II ,19 1
An alternative mechanism may involve the he-
patic metabolism of the parent drug to a stable in-
termediate which then passes to its site of toxicity
where it undergoes bioactivation to the ultimate
toxic metabolite. Such a mechanism has been pos-
tulated for the haematological toxicity associated
with benzeneI2D-22I and chloramphenicol.123I

© Ads Internatio nal Limited . All rights reserved. e NS Drugs 1997 Fe b : 7 2)

(
Mechanism of Clozapine-Induced Agranulocytosis
Even when an imbalance exists between drug
bioactivation and detoxification, this does not nec-
essarily mean that the patient will go on to develop
toxicity.18 1 For example, the damage caused by a
chemically reactive metabolite may be nullified
by other cellular repair mechanisms, such as DNA
repair enzymes in the case of potentially carcino-
genic drugs. For drugs causing immune-mediated
toxicity, the formation of a drug antigen may not
result in an immune response as the patient may
not be able to mount such a response,l24 1 Even
when an immune response is mounted, a hyper-
sensitivity reaction does not necessarily ensue.
With penicillin, for example, there is little inter-
individual variation in the circulating levels of the
penicilloyl hapten,125 1yet over 60% of patients fail
to mount a serological response.[24 1Similarly, with
halothane it is thought that all individuals can gen-
erate the antigen but susceptibility to hepatitis is
dependent on immune responsiveness to the anti-
gen.[261
It can be seen from this overview that the mech-
anism(s) of idiosyncratic drug reactions are com-
plex and multifactorial, susceptibility being deter-
mined by individual variability at different levels.
In fact, with most drugs, multifactorial predisposi-
tion is likely to be the rule rather than the excep-
tion. 1271 This is also likely to be the case with
clozapine-induced agranulocytosis, as discussed in
section 4.
2. Characteristics of Clozapine-
Induced Agranulocytosis
In this section, the clinical characteristics of
clozapine-induced agranulocytosis, as well as the
effects of the drug on other haematological param-
eters, are reviewed. This information is relevant, as
it may throw light on the possible mechanisms of
the toxicity.
2.1 Syndromes
Clozapine-induced suppression of the granulo-
cyte series can result in 3 contiguous syn-
dromes: 12H1
© Adis International Lirnited. All rights reserved.
143
• leucopenia - a white blood cell count of less
than 3500 cells/mm 3 with granulocytes above
1500 cells/mm3;
• neutropenia - a granulocyte count of less than
1500 cells/mm 3 but more than 500 cells/mm3;
• agranulocytosis - a granulocyte count of less
than 500 cells/mm 3 .
2.2 Epidemiology
Agranulocytosis is obviously the most severe
form and has resulted in fatalities. An epidemio-
logical study in the US covering the period be-
tween February 1990 and April 1991 showed that
the incidence of agranulocytosis after treatment
with c10zapine for I year was 0.8%.[29,301 This con-
trasts with an incidence of 1.5 to 2.0% for c1oza-
pine-induced neutropenia.J3 11 The US study also
showed that in 24 of 73 patients who developed
agranulocytosis, the white cell count did not fall
below 3500 cells/mm 3 , and in 16 of these patients,
the count had remained above 3500 cells/mm 3
within 8 days before the occurrence of agranulocy-
tosis.[ 30I
In the UK, of 6316 patients who recei ved
c10zapine from January 1990 to July 1994, 2.9%
developed neutropenia and 0.8% developed agran-
ulocytosis.1 321 The risk of both agranulocytosis and
neutropenia decreased with time (fig. 2).
More recently, an analysis of 99 502 US pa-
tients who received c10zapine in the 5-year period
starting from 1990 revealed 2931 cases of leuco-
penia (2.95%), 382 cases of agranulocytosis
(0.38%) and 12 deaths (0.012%).1331 The rate of
agranulocytosis is less than half of that which
would have been predicted, suggesting that the
stringent criteria used for haematological monitor-
ing of clozapine therapy has reduced the incidence
of agranulocytosis by early detection of suscepti-
ble patients. The risk of death associated with the
occurrence of agranulocytosis is 3 to 4% in the
US;1 33 1 this again is a marked improvement on a
predicted death rate of 15% and the observed rate
of 50% in Finland in 1975 when the drug was first
introduced.J341
eNS Drugs 1997 Feb; 7 (2)
144
• Neutropenia
D Agranulocytosis
1st year 2nd year 3rd year 4th year
Year of use
No of
patients 6316 2858 1625 661
receiving
clozaplne
Fig. 2. The incidence of clozapine-induced agranulocytosis and
neutropenia in patients who received clozapine for 1, 2, 3or 4
years in the UK and Ireland between 1990 and 1994. The error
bars represent the upper 95% confidence interval. The risk of
agranulocytosis and neutropenia is highest during the first year
of use, with the risk decreasing thereafter. The lower part of the
figure shows the number of patients receiving clozapine and the
duration of use. Data kindly provided by Dr Karen Atkin (Sandoz
Pharmaceuticals, UK) [adapted from Atkin et a1. 132) I.
About 75 % of the cases of c1ozapine-induced
agranulocytosis occur within the first 24 weeks of
treatment, with 95 % occurring within 6 months.
Although the ri sk of agranulocytosis decreases
with time, it does not reach zero, with some cases
being reported after 2 years or more of continued
therapy. 12~ I
2.3 Risk Factors
Older age and female gender are considered to
be risk factors for c1ozapine-induced agranulocy-
tosis.1 301 Certain ethnic groups such as Scandina-
vians l341 and Ashkenazic Jews l3 )1 may be more sus-
<0 Adis International Limited . All rig hts reserve d .
Pirmolzamed & Park
ceptible. Conversely, it has been suggested that the
Chinese may be less susceptible, since no cases of
agranulocytosis were reported among 290 Chinese
patients treated with c1ozapine. i361 However, this is
based on small numbers of patients, and since there
is no mandatory monitoring system in China it is
difficult to confirm. Concomitant treatment with
other drugs known to induce agranulocytosis such
as carbamazepine may increase the risk of agranu-
locytosis.[37,38 1
It is important to note that c10zapine dosage and
the baseline white blood cell count do not act as
risk factors for agranulocytosis . 12~ ,30.3 21
2.4 Predictors
Recently, further analysis of a cohort of II 555
patients who received c10zapine between 1990 and
1991 has shown that the occurrence of a white
blood cell count spike of at least 15 % above the
previous measurement predicted the development
of agranulocytosis (risk ratio 3.02) within 75
days.l3 9J However, although this was a sensitive in-
dicator, it was not specific, since 79% of matched
controls who were receiving c10zapine had at least
I spike but did not develop agranulocytosis.
Neutropenia after use of c10zapine is associated
with little morbidity and has a short clinical course
of between 2 to 8 days .140 1 Agranulocytosis is
reversible provided that c10zapine is withdrawn,
but lasts for between 14 to 21 days.140 1 Recent-
ly, the use of granulocyte-macrophage colony-
stimulating factor (GM-CSF) or granulocyte
colony-stimulating factor has gained wide accep-
tance as a treatment for c1ozapine-induced agranu -
locytosis and can reduce the duration of agranulo-
cytosis from a mean of 16 to 8 days.140A) I
3. Other Haematological Abnormalities
Induced by Clozapine
Clozapine has been reported to cause other
haematological abnormalities. 12x1 However, these
are much less common and less serious than its
effects on myeloid cells.
Mild anaemia is sometimes noted in patients
taking c1ozapine, although it is not related to the
e NS Drugs 1997 Fe b ; 7 (2)
Mechanism of Clozapine-Induced Agranulocytosis
occurrence of agranulocytosis.l28,42 1 One case of
hypoplastic anaemia has also been reported in a
patient taking clozapine, but the patient was also
recei ving other drugs known to cause bone marrow
suppression, most notably carbamazepine. 1371
Chronic leucocytosis, lymphopenia and isolated
cases of thrombocytopenia have also been re-
ported. 128]
Clozapine has also been reported to cause iso-
lated eosinophilia. A recent prospective study in
118 patients showed the incidence of eosinophilia
to be 14%.[43 1 Women (23%) had a higher inci-
dence than men (7%), and the eosinophilia oc-
curred 3 to 5 weeks after the onset of treatment.
There was no relationship between the eosinophil
and neutrophil counts, and no patient developed
agranulocytosis in this study. A recent case report
has suggested that the magnitude of eosinophilia
rather thanjust its occurrence may be related to the
development of neutropenia.l 441 Clearly, this ob-
servation is only from I patient and thus requires
confirmation.
4. Mechanisms of Clozapine-Induced
Agranulocytosis
When considering the mechanism of agranulo-
cytosis induced by a drug, it is important to ask the
following questions:
• Is the drug acting centrally, i.e. in the bone mar-
row, or in the peripheral blood to cause deple-
tion of granulocytes?
• Is the neutrophil toxicity due to the parent drug,
and is it related to the pharmacological or chem-
ical characteristics of the drug?
• Is the neutrophil toxicity due to a stable meta-
bolite of the drug, or can the drug be
bioactivated to chemically reactive metabo-
lites?
• Is the agranulocytosis due to direct toxicity of
the drug/metabolite to neutrophils (or their pre-
cursors), or is it secondary to an immune reac-
tion?
• What is the individual susceptibility factor(s)
that determines that only a minority of patients
exposed to the drug will develop agranulocytosis?
© Adis International Limited. All rights reserved.
145
With regard to the agranulocytosis caused by
clozapine, we will consider each of these questions
in turn. The current state of knowledge from the
research that has been carried out will be
emphasised, and the deficiencies that still exist in
our understanding of the mechanism(s) of cloza-
pine-induced agranulocytosis will be highlighted.
4.1 Site of Toxicity - Central or Peripheral?
Before considering whether the toxicity of
clozapine is central or peripheral, it is important to
review the normal process of haematopoiesis.l 451
All blood cells are derived from a pluripotential
haematopoietic stem cell (fig. 3). By a process of
differentiation and commitment, this stem cell
gives rise to committed precursors that ultimately
result in the production of mature blood cells. In
the case of neutrophils, the process takes 4 to 7
days. Neutrophils have a mean transit time (i.e.
survival time) of about 9 hours, with the cells dying
by the process of apoptosis.[ 46 1
It is important to note that the transition from a
pluripotential stem cell to the mature cell is char-
acterised by differentiation and a loss in the ability
of the cell to divide and proliferate.[471 With regard
to neutrophils, a balance exists between their pro-
duction in the bone marrow and the loss of cells by
apoptosis. Thus, put simply, any factor that reduces
their production or increases apoptosis may lead to
a depletion of neutrophils. The whole process of
haematopoiesis is tightly regulated by a large num-
ber of cytokines.l 471
In patients with clozapine-induced agranulocy-
tosis, bone marrow examination shows an absence
of myeloid precursors, the presence of only occa-
sional promyelocytes and myeloblasts, and rela-
tive erythroid hyperplasia.l 42 1This suggests that an
early myeloid precursor in the bone marrow, rather
than the peripheral blood neutrophil, is the site of
toxicity in clozapine-induced agranulocytosis.
This picture has to be contrasted with that seen in
clozapine-induced neutropenia where there is evi-
dence of myeloid maturation in the bone marrow,
reflecting peripheral destruction of granulocytes)311
The factors leading to progression in severity from
eNS Drugs 1997 Feb; 7 (2)
146 Pirmohamed & Park

Fig. 3. A schematic representation of the normal process of haematopoiesis. The shaded area represents the cells affected by
clozapine-induced agranulocytosis. Abbreviations: BFU-E = burst forming units-erythrocyte; CFU-Eo = colony forming units-eosino-
phils; CFU-G =colony forming units-granulocyte; CFU-G/M =colony forming units-granulocyte/macrophage; CFU-M =colony forming
units-macrophage; CFU-mega = colony forming units-megakaryocyte.

neutropenia to agranulocytosis (although in some
patients, as stated in section 2, there is no such
progression), are not known. Indeed, it seems
likely that the mechanism of toxicity in the major-
ity of patients with neutropenia who do not pro-
gress to agranulocytosis is distinct from that respon-
sible for the agranulocytosisl 311 (see section 4.3).
4.2 Is Toxicity Due to Clozapine Itself and Is It
Related to the Pharmacology of the Drug?
The mode of action of clozapine has not been
clearly elucidated. However, clozapine has been
shown to interact with dopamine (in particular 0 1
and 0 4 ), serotonin (5-hydroxytryptamine; 5-HT),
muscarinic anticholinergic, and adrenergic recep-
tors. 14X1 Given the absence of most of these recep-
tors on white blood cells, it can be safely assumed
that the agranulocytosis is not due to an extension
of the known pharmacological effects of the drug
(or its metabolites).
At therapeutic drug concentrations, c10zapine
has not been shown to be directly cytotoxic to neu-
trophils and does not interfere with the turnover of
bone marrow precursor cells. 142 ,49,SOI A possibility
to consider is whether c10zapine affects the func-
tion of the neutrophil and whether this in some way
contributes to the toxicity. Acharacteristic property
of drugs known to affect neutrophil function (for
example, antimalarials lSII ) is their ability to accu-
mulate within the neutrophil , often within Iyso-

© Adis International Limited . All rights rese rved . eNS Drugs 1997 Feb: 7 (2)
Mechanism of C1ozapine-Induced Agranulocytosis 147

somes (so-called 'lysosomotropic agents'152 1). Pre-
liminary data from our laboratory have shown that
c10zapine does not accumulate within neutrophils
to any great extent (unpublished observations).
Whether c10zapine affects the ability of the neutro-
phil to undergo a respiratory burst is unknown and
requires further study.
4.3 Is Toxicity Due to Stable or Chemically
Reactive Metabolites of Clozapine?
Clozapine is a highly lipophilic compound that
undergoes extensive metabolism (fig. 4), with only
2 to 5% of the drug being excreted unchanged.l 531
The major stable metabolites of the drug are
demethyl-c1ozapine and clozapine N-oxide, al-
though the serum concentrations of these metabo-
lites show a great deal of inter-individual variabil-
ity.154 1Demethylation of c10zapine is dependent on
the cytochrome P450 (CYP) isoform I A2,155, 56 1
while N-oxidation may be due to either the CYP
enzymes or flavin monooxygenases .1551 Initial re-
ports that c10zapine was metabolised by the poly-
morphic isoform CYP2D6,1 57 1 have largely been
discounted by in vivo l5H1 and in vitro l551 studies.
Theoretically, c10zapine toxicity may be due to
either the parent compound or one of its stable me-
tabolites. Given that such stable metabolites are of
low molecular weight and cannot form covalent
linkages with macromolecules, they would be un-
able to react irreversibly with cellular molecules
and induce cell death or act as haptens. Thus, their
toxicity would have to be assumed to be due to a

Hydroxylated
metabolites

o~
I .... CH J

N=C
N
J
~N
N=C I

0
~
Clv-( Clv-( 0 CI~N=UC
I
~+. \
~N~
I
~N~
I
0-- N
I
__
H H H

/ 1-
Clozapine N-oxide Clozapine

H
'

Demethyl·clozapine Glutathionyl-clozapine Nitrenium ion

Fig. 4. The metabolism of clozapine. In the liver, clozapine is metabolised to the stable metabolites, demethyl-clozapine, clozapine
N-oxide and hydroxylated metabolites. Clozapine can also undergo bioactivation in the liver and neutrophil to the same chemically
reactive metabolite, postulated to be a nitrenium ion. This electrophilic metabolite can react spontaneously with glutathione (GSH)
to form glutathione conjugates, the 2 most common being C6-glutathionyl-clozapine and C9-glutathionyl-clozapine (adapted from
Pirmohamed et al.(55) and Maggs et aI. 160I ).

© Adis International Limited. All rights reserved. eNS Drugs 1997 Feb: 7 (2)
148
direct toxic action on cellular function by a revers-
ible interaction with an essential cellular enzyme,
cellular membrane, ion channel or receptor.
Gerson and co-workers fSOI have shown that
demethyl-clozapine was 4 to 10 times more toxic
towards haematopoietic progenitors than clozapine
or its other metabolites. However, the concentra-
tion required to produce toxicity in vitro was 3 to
6 times the normal serum concentration. The au-
thors postulated that certain patients may have an
intrinsic susceptibility to this metabolite, although
marrow taken from 4 patients with clozapine-in-
duced agranulocytosis was not more sensitive to
the metabolite than marrow from healthy donors.
Furthermore, demethyl-clozapine was as toxic to
different progenitor lineages despite the fact that
the major clinical toxicity only affects the granulo-
cyte precursors. Additionally, a recent study did not
find any relationship between the plasma concen-
trations of clozapine and demethyl-clozapine and
the occurrence of agranulocytosis in 5 patients.l S91
Taken together, these facts suggest that agranulo-
cytosis is unlikely to be due to a direct toxic effect
of either clozapine or its stable metabolites. In con-
trast, the neutropenia may be due to the parent
drug, as in vitro experiments have shown it to be
directly toxic to late myeloid maturation. 1311
Clozapine also undergoes biotransformation to
toxic, chemically reactive metabolites, which may
be involved in the pathogenesis of agranulocyto-
sis.l 5s ,60.63 1We have shown in vitro that in human
liver microsomes clozapine is converted to a reac-
tive metabolite that binds covalently to microsomal
protein. ISS I This biotransformation was catalysed
by several CYP isoforms, but not CYP I A2. The
major adduct formed in the presence of glutathione
was C-6 glutathionyl-clozapine (fig. 4). In vivo
studies in mice and rats have also demonstrated
bioactivation of clozapine to the same adduct. 1601
The functional effects of generating this chemi-
cally reactive intermediate have been shown us-
ing an in vitro assay, in which clozapine was co-
incubated with lymphocytes and human, mouse or
rat liver microsomes.lM,6SI It was found that
clozapine by itself was nontoxic, but was trans-
© Adis International Limited. All rights reserved.
Pirmohamed & Park
formed to a cytotoxic metabolite by the addition of
NADPH, a co-factor for CYP-mediated metabo-
lism. It is important to note, however, that the target
cell was the lymphocyte rather than the neutrophil.
Clozapine is also bioactivated to a chemically
reactive metabolite when it is incubated with
myeloperoxidase (the major enzyme present in
neutrophils),161] isolated neutrophilsl60,63I and their
bone marrow precursors. 160] In the cellular system,
bioactivation is only observed when the cells are
activated by phorbol myristate acetate. Impor-
tantly, the adduct formed by neutrophils in the pre-
sence of glutathione was the same as that observed
with liver microsomes.l 60] The toxic metabolite has
been tentatively identified as a nitrenium ion.l 631
Given that the toxic metabolite can be formed
in the vicinity of the target cell, i.e. the bone mar-
row precursor cells, toxicity may either be direct,
by irreversible interaction with cellular macromol-
ecules, or indirect, by acting as a hapten and initi-
ating an immune response. However, at present
there is no clear indication of the mechanism of
toxicity (see section 4.4). Although the bioactiva-
tion of clozapine by neutrophils has been demon-
strated in vitro under nonphysiological conditions,
it has not been possible to either demonstrate or
investigate the formation of the chemically reac-
tive metabolite in vivo in human white blood cells
because of the very short half-life (less than I min-
ute 1631 ) of the metabolite. It has been suggested that
concomitant infection that activates white blood
cells may act as a predisposing factor by causing
bioactivation of clozapine in vivo. 1621 However,
there is no epidemiological or mechanistic evi-
dence to support the hypothesis.
4.4 is Toxicity Direct or immune Mediated?
At present, there is no clear evidence of the
mechanism of clozapine-induced agranulocytosis.
Rechallenge with clozapine in patients with a his-
tory of either leucopenia or agranulocytosis re-
sulted in the recurrence of toxicity after a mean of
14 weeks, which contrasts with a mean time of 24
weeks to the occurrence of toxicity on primary ex-
posure.1 661 In general, this shorter time interval to
eNS Drugs 1997 Feb; 7 (2)
Mechanism of Clozapine-Induced Agranulocytosis
the recurrence of toxicity on re-exposure to the
drug is in keeping with an anamnestic response of
an immune-mediated mechanism. I171 Neverthe-
less, the average delay of 14 weeks that was ob-
served in these patients l661 is longer than that usu-
ally observed with immune-mediated reactions.
It is, however, difficult to draw conclusions re-
garding the mechanism of agranulocytosis from
this study l66 1because: (i) patients with both leuco-
penia and agranulocytosis, which may have differ-
ent mechanisms, were studied; and (ii) although
the mean time to the recurrence of toxicity was 14
weeks, the standard deviation was 34 weeks, with
I patient being a 2-year outlier, suggesting that the
patients studied were a clinically heterogeneous
group.
Pisciotta and colleagues l671 have postulated an
immune mechanism on the basis of complement-
mediated neutrophil toxicity of acute phase serum
from patients with clozapine-induced agranulo-
cytosis. The toxicity was not dependent on the pre-
sence of clozapine or its metabolites and was
attenuated by anti-immunoglobulin M (lgM) anti-
bodies. The development of colony forming units-
granulocyte (CFU-G) was also inhibited by the se-
rum. Also in favour of an immune mechanism is
the finding of antibodies towards myeloperoxidase
in the serum of patients with clozapine-induced
agranulocytosis.1 681 This enzyme is implicated in
the bioactivation of clozapine. However, this was
an inconsistent finding and the antibodies were
present in low titre.
The association of clozapine-induced agranulo-
cytosis with human leucocyte antigen (HLA) types
in certain popUlations also supports an immune
mechanism (see section 4.5).
The evidence for an immune mechanism, how-
ever, must be regarded as indicative rather than
conclusive. More direct evidence such as the pres-
ence of antidrug antibodies has not been forthcom-
ing. Although the discussion so far has focused on
humoral immunity, the role of cellular immunity in
the pathogenesis of clozapine-induced agranulo-
cytosis should be considered in future studies. In
this respect, it is important to note that there are T
© Adis International Limited. All rights reserved.
149
lymphocytes present in normal human bone mar-
row that are capable of suppressing phagocyte col-
ony formation. 1451
The evidence for a nonimmunogenic mecha-
nism is also indirect. The lack of anti neutrophil or
antimyeloid cell antibodies in patients with agran-
ulocytosis supports a nonimmunogenic mecha-
nism, although only small numbers of patients
have been studied.149.631 As discussed in section
4.3, demethyl-clozapine is more toxic to bone mar-
row cultures than the parent compound, but this is
only observed at supratherapeutic plasma concen-
trations. 1501 It is important to note, however, that
plasma concentration may not be an appropriate
measure, since some compounds (i.e. lysosomo-
tropic drugs) can selectively accumulate within
neutrophils.151.521 Whether this occurs with
demethyl-clozapine is unknown.
Perhaps the most compelling evidence gathered
thus far is the ease with which clozapine can be
bioactivated to a chemically reactive intermedi-
ate,1 60 I thought to be a nitrenium ion. 1631 This toxic
species has been shown to bind to neutrophil pro-
teins,1 63 1 and thus could cause toxicity either
through direct interaction with the membrane or
with cellular macromolecules, or by initiating an
immune mechanism. With regard to the latter
mechanism, it would be important to demonstrate
cell-surface haptenation.
Given that neutrophil numbers are governed by
a delicate balance between their production and
their death, any factor decreasing their production
or increasing their death rate will lead to neutro-
penia. Although this process has been examined
with regard to clozapine and its stable metabo-
lites,150,64,65I no such studies have thus far been per-
formed with regard to the reactive metabolite. It is
of particular importance to determine whether
clozapine or its metabolites induce neutrophil
apoptosis. In this respect, it is important to con-
sider the role of cytokines in clozapine-induced
agranulocytosis. Cytokines govern the process of
haematopoiesis and are also of importance in the
process of apoptosis.l 471 Recent studies have indi-
cated that clozapine causes a dose-dependent de-
eNS Drugs 1997 Feb; 7 (2)
150
crease in the release of GM-CSF from bone marrow
cultures,[69[ one of the most important lineage-
specific cytokines. How this contributes to the patho-
genesis of agranulocytosis is unclear, but may suggest
that haematopoietic stress in patients treated with
clozapine is not followed by adequate release ofGM-
CSF. It is known that GM-CSF inhibits neutrophil
apoptosisPO[ This finding also suggests that the use
of recombinant GM-CSF in patients with agranulo-
cytosis is a logical therapeutic manoeuvre.
4.5 What Determines Individual Susceptibility?
If other drugs causing idiosyncratic toxicity are
taken as examples, it can be seen that the factors
that may determine individual susceptibility are
multiple and complex. For example, with hydrala-
zine-induced lupus, HLA-DR4[71[ and slow acetyl-
ator phenotype[72 1have so far been found to be pre-
disposing factors, but there are likely to be others
that have not yet been identified. Susceptibility to
clozapine-induced agranulocytosis is likely to re-
side at various levels including: (i) bioactivation of
clozapine; (ii) its detoxification; and (iii) factors
responsible for inducing cell death and tissue in-
jury. The factors that have so far been investigated
as determinants of susceptibility to clozapine-
induced agranulocytosis are summarised below.
4.5. 1 Viral Infection
Clozapine bioactivation by neutrophils in vitro
has been shown by the use of nonphysiological ac-
tivators. It has been suggested that such bio-
activation could occur in vivo if the patients receiv-
ing clozapine developed an infection.[62[ There is
clinical and laboratory evidence suggesting that in-
fluenza may serve as a predisposing factor for ves-
narinone-induced agranulocytosis.[ 73 1 Large scale
prospective studies would be needed to determine
whether infection was a co-factor in clozapine-
induced agranulocytosis. Additionally, an explana-
tion would have to be put forward for why most
cases of agranulocytosis occur within 3 months of
starting the drug)301 However, it could be specu-
lated that the white blood cell count spike that may
be a predictor for the occurrence of agranulocytosis
(see section 2.4)1 39 1is the result of an infection.
© Adis International Limited. All rights reserved.
Pirmohamed & Park
4.5.2 Oxidative Stress
Given that free radicals may be responsible for
the agranulocytosis associated with clozapine (see
section 4.3),1611 the role of variation in antioxidant
levels as predisposing factors has received some
attention. Covalent binding can be reduced by
ascorbic acid (vitamin C), with the formation of
ascorbyl radicals.l 611 Thus, given that low levels of
ascorbate have been noted in patients with schizo-
phrenia,[741 it has been suggested that co-adminis-
tration of ascorbic acid with clozapine may protect
against agranulocytosis.f61 1To date, however, this
remains untested, and will require large scale pro-
spective studies.
More recently, it has been suggested that the pa-
tients with agranulocytosis may have a relative de-
ficiency of free radical scavenging enzyme activ-
ity,l751 Multiple regression analysis suggested that
plasma glutathione peroxidase and red blood cell
peroxidase activities, and selenium levels were key
parameters in distinguishing patients at risk of
agranulocytosis from those not at risk,l751 How-
ever, the use of these parameters was not able to
distinguish between the patients with agranulocy-
tosis and one group of controls, and thus cannot be
used to predict toxicity.
4.5.3 Genetic Factors
In order to identify genetic predisposing factors
for clozapine-induced agranUlocytosis, Yunis and
colleaguesl35.761 have concentrated on the major
histocompatibility complex (MHC) antigens.
In initial studies, they showed that the HLA-
B38, DR4, DQ3 and the HLA-DR2, DQ I haplo-
types were associated with the occurrence of
clozapine-induced agranulocytosis in Ashkenazic
Jews and non-Jewish patients, respectively,l35. 76 1
The haplotype HLA-B 16, variant B39, DR4 and
DQ3 was also found in a native American patient
who had developed agranulocytosis after treatment
with clozapine.l 77 [ Two other non-Jewish patients
with clozapine-induced agranulocytosis were
found to express HLA-B38, but not DR4 or
DQ3.1 781 However, a larger study in a European
popUlation failed to show any association with
MHC antigens,l79 1
eNS Drugs 1997 Feb; 7 (2)
Mechanism of Clozapine-Induced Agranulocytosis 151

ORB1·0402
•
2l
00B1·0302
•
~
"'"
c.
.<::
(/)
00A1·0301
•
•
.~

'"
....,
.!.!
ORB1·11
N

"'"
"''""
.<::
«
(/)
00B1·0301
•
OPB1·0401
•
Haplotype 1
•

i1
ORB1·02
•
'c."
~
.<::
00B1·0502
•
•
(/)
.~

'"
7
00A1·0102

z"
•
0

Haplotype 2

I I I
0.01 0.1 10 100
Odds ratios

Fig. 5. The association of certain human leucocyte antigen (HLA) subtypes with clozapine-induced agranulocytosis in Ashkenazic
Jews and non-Jewish patients. The total numbers of patients studied were 24 cases of agranulocytosis and 54 controls in Ashkenazic
Jews, and 40 cases of agranulocytosis and 32 controls in non-Jewish patients. The odds ratios together with their corresponding
95% confidence intervals are shown for each allele or haplotype. An odds ratio of 1 indicates no effect of the HLA subtype on the
risk of agranulocytosis, while an odds ratio of below 1 indicates that the HLA allele may be protective and above 1 indicates that it
may represent a risk factor for agranulocytosis. The data are derived from Yunis et al. l801 and the 95% confidence intervals were
determined by the use of the Arcus statistical software package (Or lain Buchan, The University of Liverpool, UK). Haplotype 1:
ORB1*0402, ORB4*0101, 00B1*0302, 00A1*0301. Haplotype 2: ORB1*1601, ORB5*02, 00B1*0502, 00A1*0102.

In a further study involving a larger number of
patients, Yunis et al. ISOI confirmed the HLA associ-
ation of clozapine-induced agranulocytosis in both
Jewish and non-Jewish patients using molecular
techniques (fig. 5). The HLA alleles ORB I * II and
DQ B I *030 I were found to be protective factors
against agranulocytosis. It is important to note that
although the odds ratios for susceptibility ex-
ceeded 10 with some alleles, the 95% confidence
intervals are extremely wide (fig. 5), indicating
that larger numbers of patients need to be studied
to provide more definitive data. Furthermore, di-
rect evidence for HLA involvement in immune re-
actions against clozapine has not been found. ISOI It
is also difficult to understand why there are differ-
ent HLA associations between Jewish and non-
Jewish patients.
Given these uncertainties, the authors have pos-
tulated that the HLA alleles may be in linkage dis-
equilibrium with other genes [for example, heat

© Adis International Umited. All rights reserved. eNS Drugs 1997 Feb: 7 (2)
152
shock proteins (HSP) and tumour necrosis factor
(TNF)] in the MHC region. Subsequent studies
have shown that agranulocytosis is associated with
HSP-70-1 and HSP-70-2 variants in both Jewish
and non-Jewish patients. lSI I This is an interesting
finding that needs to be confirmed in larger num-
bers of patients from different ethnic backgrounds.
However, it is important to note that HSP-70 ex-
pression has been shown to decrease proliferation
and induce apoptosis in granulocyte precursors,
both of which may be important in clozapine-
induced agranuiocytosis.lS 21 Similarly, TNFa also
induces apoptosis,[S3 1 although its role in cloza-
pine-induced agranulocytosis needs to be defined.
5. Comparison with
Amodiaquine-Induced Agranulocytosis
Amodiaquine is a 4-aminoquinoline antimalar-
ial that was used for the prophylaxis and treatment
of malaria. IS41 However, following reports of
agranulocytosis and hepatic damage, it was with-
drawn from use as a prophylactic and is now only
used for treatment. Research into the mechanisms
of the idiosyncratic toxicities caused by amodia-
quine has been ongoing in our laboratory for a
number of years. Some of these findings are dis-
cussed in this section to compare the clinical and
mechanistic features of the agranulocytosis caused
by amodiaquine with those of that caused by
clozapine, and to provide possible avenues for fur-
ther research into clozapine-induced agranulocyto-
sis.
Like that caused by clozapine, amodiaquine-
induced agranulocytosis occurred within 3 months
of the onset of treatment and was characterised by
the absence of myelopoiesis in the bone marrow,
suggesting that the precursor cells rather than the
mature neutrophils were the target cells. ISSI Also
like clozapine, the drug has been shown to undergo
bioactivation by neutrophils l86 ,87 1and liver micro-
somesIXX,X~1 to a chemically reactive metabolite,
The toxic metabolite in this case is the quinoneim-
ine (fig. 6).190 1 As with the reactive metabolite of
clozapine, we have shown that quinoneimine
readily conjugates with glutathione;IXXI the gluta-
© Adis International Limited. All rights reserved.
Pirmohamed & Park
thione conjugates of both drugs have been shown
to undergo biliary excretion in in vivo studies in the
rat.[60,881
Importantly, amodiaquine has been shown to be
immunogenic in rats when given by oral, intra-
peritoneal or intramuscular routes.l 84 ,86 1 Anti-
amodiaquine antibodies have also been detected in
patients with agranulocytosis.1 911 These antibodies
recognised a synthetic drug antigen designed and
synthesised with a knowledge of the chemical
mechanisms of bioactivation of amodiaquine, pro-
viding evidence that the reactive metabolite rather
than the parent drug was responsible for the toxi-
city.[841
Interestingly, amodiaquine has also been shown
to be toxic in bone marrow cultures in vitro.l 921 This
may reflect its ability to undergo spontaneous auto-
oxidation to the quinoneimine metabolite.l 901
Given our knowledge of the mechanism of
amodiaquine-induced agranulocytosis, the aim of
our laboratory is to design an analogue of
amodiaquine that retains efficacy yet does not
cause toxicity. To this end, several analogues have
been synthesised, and preliminary studies have
shown that the therapeutic effects of amodiaquine
can be separated from its toxicity.187,~31 Clearly,
such an approach may hold promise in the devel-
opment of clozapine analogues (see section 7).
6. Areas for Further Research
From the evidence available, it can be surmised
that clozapine-induced agranulocytosis may be
mediated by the reactive metabolite of clozapine,
which has been postulated to be a nitrenium
ion.ISS.60.631 This toxic moiety affects bone marrow
precursor cells rather than (or as well as) the mature
peripheral neutrophils.l 421 However, as has already
been pointed out, many questions remain unan-
swered. Further research is need into the areas
summarised below. This is important not only to
prevent clozapine-induced agranulocytosis, but it
may also provide important insights in to the mech-
anisms of agranulocytosis associated with other
drugs. The issues that require clarification are:
eNS Drugs 1997 Feb; 7 (2)
Mechanism of Clozapine-Induced Agranulocytosis 153

y~~'-N/
'---- ~I
y '---
CH - N:
H
2

m
Am......'".

NH ------------------.. NH

CI N
~"-'-dl-
c,AA N)

(
"" ~tochrome P450

•....
~Iivated white blood cells

o ;-

Cytochrome P450 l)~'-N,----

~ CIA)l)
~
N

Amodiaquine
quinoneimine

~ 0"

~
/
Protein C H2 - N

m
CI
~

~ I
: NH

h-
N
I --,,~","
conjugate

Antigenlimmunogen
formation

Agranulocytosis

Fig. 6. The role of metabolism and bioactivation in the pathogenesis of agranulocytosis induced by the antimalarial amodiaquine.
The figure shows the proposed mechanism for immunotoxicity via the formation of the unstable intermediate amodiaquine quinoneim-
ine (adapted from Park et aI. 1841).

© Adis International Limited. All rights reserved. eNS Drugs 1997 Feb: 7 (2)
154 Pirmohamed & Park

Remoxipride Sui pi ride

C(y~
N Y C H2 CH2 N

F
Olanzapine Risperidone

Ziprasidone lIoperidone
F

Fig. 7. The structures of some of the new antipsychotic drugs that have either recently been introduced into clinical practice or are
entering the last stages of development prior to marketing. Remoxipride has now been withdrawn because of the risk of aplastic
anaemia.

• Does bioactivation of clozapine by white blood
cells occur in vivo, and if so, what is responsible
for neutrophil activation?
• Does clozapine or one of its metabolites specif-
ically accumulate in neutrophils or their precur-
sors, leading to a high local concentration that
may mediate toxicity?
• What is the mechanism of toxicity? If it is im-
mune mediated, are there any drug (metabolite)-
specific antibodies or T lymphocytes in affected
patients?
• If the mechanism of toxicity is not immune-
mediated, how is agranulocytosis mediated?
What is the role of apoptosis, and what role do
cytokines play?
• What are the individual susceptibility factors?
Is the HLA system involved or is a closely
linked gene on chromosome 6 involved?
• What is the role of polymorphisms in genes cod-
ing for activation and detoxification enzymes?
7. Prospects for Future
Drug Development
The ultimate aim of research into clozapine-
induced agranulocytosis should be to either iden-

© Adis International limited. All rights reserved. eNS Drugs 1997 Feb; 7 (2)
Mechanism of Clozapine-Induced Agranulocytosis
tify susceptible individuals and/or to develop a
clozapine analogue that retains efficacy but does
not cause toxicity. With the former option, it is
likely that predisposition is multifactorial, and thus
routine testing of patients before starting clozapine
may be practically and economically unfeasible.
The latter option is hampered by the fact that the
mode of action of clozapine in schizophrenia is
unclear,1 48 1 making it difficult to evaluate the po-
tential therapeutic efficacy of new analogues.
Given the problems associated with conven-
tional antipsychotics, many new agents have been
developed that have advantages over conventional
compounds such as chlorpromazine (fig. 7). How-
ever, clinical experience with the new drugs is lim-
ited and further comparative efficacy studies with
clozapine need to be performed.l94-961 It is also im-
portant to note that new compounds are not neces-
sarily safer than the older compounds that are
available. This is clearly illustrated by the example
of remoxipride, which following its introduction
was shown to cause aplastic anaemia.197-991 It also
emphasises the general point that better testing sys-
tems need to be developed so that idiosyncratic
reactions can be detected at an early stage in the
development of a drug. Perhaps with the accessi-
bility of human haematological tissue, it may be
possible to develop such tests to prevent haemato-
logical toxicity.
The ultimate solution to clozapine-induced
agranulocytosis would be to define unequivocally
the chemical species (drug, stable metabolite or
chemically reactive metabolite) responsible for the
toxicity. Subsequent synthesis of clozapine ana-
logues could then be followed by systematic anal-
ysis in in vitro test systems designed to: (i) inves-
tigate their potential to accumulate within the
white blood cell and affect its function, and
whether they are bioactivated by the cells; and (ii)
investigate their pharmacological interaction with
various receptors in comparison with clozapine. In
this way, it may be possible to design out the toxi-
city of clozapine, but retain the pharmacological
activity.
<0 Adis International Limited. All rights reserved.
155
If indeed bioactivation of clozapine to a nitren-
ium ion is responsible for agranulocytosis, then
fluorine substitution of the aromatic groups, which
would lower the oxidation potential of the mole-
cule, may prevent bioactivation without loss of
pharmacological activity.lIOO] Various analogues of
clozapine have been synthesised in the past but
were either ineffective or caused adverse ef-
fects,IIOl] and thus were abandoned. Clearly, the
synthesis of these compounds was performed with-
out a knowledge of the chemistry of bioactivation.
With the studies that have been performed in the
past few years, it may be possible to develop new
analogues that do not cause agranulocytosis.
8. Conclusion
There is no doubt that clozapine is a highly ef-
fective agent and that it causes agranulocytosis in
about 0.8% of individualspo. 32 1The Clozaril® Pa-
tient Monitoring Scheme has proved to be highly
successful in the early detection of patients devel-
oping white blood cell toxicity, and has prevented
many fatalities. 1331 However, by virtue of its ability
to induce agranulocytosis, clozapine is restricted
to the patients most severely affected by schizo-
phrenia. The elucidation of the mechanisms oftox-
icity may ultimately allow the development of
strategies to prevent agranulocytosis, and thus al-
low the use of the drug in other patients with
schizophrenia and possibly other neurological dis-
orders.l 102 ,I031
Thus, research into clozapine-induced agranu-
locytosis, which has been ongoing for the past 5
years, needs to continue and address the questions
outlined in this review. Importantly, this may also
provide clues to the general mechanisms of drug-
induced agranulocytosis, one of the most severe
forms of idiosyncratic toxicity.
Acknowledgements
The authors wish to thank Sandoz Pharmaceuticals for
their support. BKP is a Wellcome Principal Fellow.
eNS Drugs 1997 Feb; 7 (2)
156
References
I. Kane J, Honigfeld G, Singer J, et al. C10zapine for the treat-
ment-resistant schizophrenic: a double-blind comparison
with chlorpromazine. Arch Gen Psychiatry 1988; 45: 789-96
2. Lieberman JA, Safferman AZ. Clinical profile of c1ozapine: ad-
verse reactions and agranulocytosis. Psychiatr Q 1992; 63:
51-70
3. Lindstrom LH. A retrospective study on the long-term effi-
cacy of c10zapine in 96 schizophrenic and schizoaffective pa-
tients during a J3-year period. Psychopharmacology 1989;
99: S84-6
4. Kane JM. Drug therapy - schizophrenia. N Engl J Med 1996;
334: 34-41
5. Baldessarini RJ, Frankenburg FR. Clozapine. A novel antipsy-
chotic agent. N Engl J Med 1991; 324: 746-54
6. Rawlins MD, Thompson JW. Pathogenesis of adverse drug re-
actions. In: Davies DM, editor. Textbook of adverse drug re-
actions. Oxford: Oxford University Press, 1977: 44-60
7. Park BK, Pirmohamed M, Kitteringham NR. Idiosyncratic drug
reactions: a mechanistic evaluation of risk factors. Br J Clin
Pharmacol 1992; 34: 377-95
8. Pirmohamed M, Kitteringham NR, Park BK. The role of active
metabolites in drug toxicity. Drug Saf 1994; II: 114-44
9. Park BK, Pirmohamed M, Kitteringham NR. The role of cyto-
chrome P450 enzymes in hepatic and extrahepatic human
drug toxicity. Pharmacol Ther 1995; 68: 385-424
10. Uetrecht JP. Mechanism of hypersensitivity reactions: proposed
involvement of reactive metabolites generated by activated
leukocytes. Trends Pharmacol Sci 1989; 10: 463-7
II. Uetrecht JP. The role of leukocyte-generated reactive metabo-
lites in the pathogenesis of idiosyncratic drug reactions. Drug
Metab Rev 1992; 24: 299-366
12. Mitchell JR, Jollow DJ, Potter WZ, et al. Acetaminophen-in-
duced hepatic necrosis. I. Role of drug metabolism. J Phar-
macol ExpTher 1973; 187: 185-94
13. Jollow DJ, Mitchell JR, Potter WZ, et al. Acetaminophen-in-
duced hepatic necrosis. II. Role of covalent binding ill vivo. J
Pharmacol Exp Ther 1973; 187: 195-202
14. Potter WZ, Davis DC, Mitchell JR, et al. Acetaminophen-
induced hepatic necrosis. III. Cytochrome P-450-mediated
covalent binding in vitro. J Pharmacol Exp Ther 1973; 187:
203-JO
15. Mitchell JR, Jollow DJ, Potter WZ, et al. Acetaminophen-in-
duced hepatic necrosis. IV. Protective role of glutathione. J
Pharmacol Exp Ther 1973; 187: 211-7
16. Park BK, Coleman JW, Kitteringham NR. Drug disposition and
drug hypersensitivity. Biochem Pharmacol 1987; 36: 581-90
17. Pohl LR, Satoh H, Christ DD, etal. Immunologic and metabolic
basis of drug hypersensitivities. Ann Rev Pharmacol 1988;
28: 367-87
18. Pessayre D, Larrey D. Acute and chronic drug-induced hepati-
tis. Baillieres C1in Gastroenterol 1988; 2: 385-423
19. UetrechtJ. Drug metabolism by leukocytes and its role in drug-
induced lupus and other idiosyncratic drug reactions. Crit Rev
Toxicol 1990; 20: 213-35
20. Eastmond DA, Smith MT, Irons RD. An interaction of benzene
metabolites reproduces the myelotoxicity observed with ben-
zene exposure. Toxicol Appl Pharmacol 1987; 91: 85-95
21. Bhat RV, Suhrahmanyam VV, Sadler A, et al. Bioactivation of
catechol in rat and human bone marrow cells. Toxicol Appl
Pharmacol 1988; 94: 297-304
22. Brodfuehrer JI, Chapman DE, Wilke TJ, et al. Comparative
studies of the ill vitro metabolism and covalent binding of
© Adis International Limited. All rights reserved.
Pirmohamed & Park
14C-benzene by liver slices and microsomal fraction of
mouse, rat and human. Drug Metab Dispos 1990; 18: 20-7
23. Yunis AA. Chloramphenicol toxicity? 25 years of research. Am
J Med 1989; 87: 44-8
24. Adkinson NF, Wheeler B. Risk factors for IgE dependent reac-
tions to penicillin. In: Kerr JW, Ganderton MA, editors. XI
International Congress of Allergology and Clinical Immunol-
ogy. London: MacMillan Press Ltd, 1983: 55-9
25. Lafaye P, Lapresie C. Fixation of penicilloyl groups to albumin
and appearance of anti-penicilloyl antibodies in penicillin-
treated patients. J C1in Invest 1988; 82: 7-12
26. Gut J, Christen U, Huwyler J. Mechanisms of halothane toxic-
ity: novel insights. Pharmacol Ther 1993; 58: 133-55
27. Pirmohamed M, Kitteringham NR, Park BK. Idiosyncratic re-
actions to antidepressants: a review of possible mechanisms
and predisposing factors. Pharmacol Ther 1992; 53: J05-25
28. Mendelowitz AJ, Gerson SL, Alvir JMJ, et al. Clozapine-in-
duced agranulocytosis - risk-factors, monitoring and manage-
ment. CNS Drugs 1995; 4: 412-21
29. Alvir JMJ, Lieberman JA. A re-evaluation of the clinical char-
acteristics of c1ozapine-induced agranulocytosis in light of the
United States experience. J Clin Psychopharmacol 1994; 14:
87-9
30. Alvir JMJ, Lieberman JA, Safferman AZ, et al. Clozapine-in-
duced agranulocytosis. Incidence and risk factors in the
United States. N Engl J Med 1993; 329: 162-7
31. Gerson SL. C10zapine - deciphering the risks. N Engl J Med
1993; 329: 204-5
32. Atkin K, Kendall F, Gould D, et al. The incidence of neutropenia
and agranulocytosis in patients treated with c10zapine in the
UK and Ireland. Br J Psychiatry 1996; 169: 483-8
33. Honigfeld G. Effects of the c10zapine national registry system
on incidence of deaths related to agranulocytosis. Psychiatr
Serv 1996; 47: 52-6
34. Idanpaan-Heikkila J, Alhava E, Olkinuora M, et al. Clozapine
and agranulocytosis [Ietterl. Lancet 1975; 2: 611
35. Lieberman lA, Yunis l, Egea E, et al. HLA-B38, DR4, DQw3
and c1ozapine-induced agranulocytosis in Jewish patients
with schizophrenia. Arch Gen Psychiatry 1990; 47: 945-8
36. Potter WZ, Ko GN, Zhang LD, et al. Clozapine in China: a
review and preview of USIPRC collaboration. Psychophar-
macology 1989; 99: S87-91
37. Gerson SL, Lieberman JA, Friedenberg WR, et al. Polypharm-
acy in fatal c1ozapine-associated agranulocytosis [Ietterl.
Lancet 1991;338: 262
38. Junghan U, Albers M, Woggon B. Increased risk ofhematolog-
ical side-effects in psychiatric-patients treated with c10zapine
and carbamazepine [Ietterl. Pharmacopsychiatry 1993; 26:
262
39. Alvir JMJ, Lieberman JA, Safferman AZ. Do white-cell count
spikes predict agranulocytosis in c10zapine recipients. Psy-
chopharmacol Bull 1995; 31: 311-4
40. Gerson SL. G-CSF and the management of c1ozapine-induced
agranulocytosis. J Clin Psychiatry 1994; 55: 139-42
41. Lamberti JS, Bellnier TJ, Schwarzkopf SB, et al. Filgrastim
treatment of 3 patients with c1ozapine-induced agranulocyto-
sis. J Clin Psychiatry 1995; 56: 256-9
42. Gerson SL, Meltzer H. Mechanisms of c1ozapine-induced
agranulocytosis. Drug Saf 1992; 7 Suppl. I: 17-25
43. Banov MD, Tohen M, Friedberg J. High-risk of eosinophilia in
women treated with cJozapine. J Clin Psychiatry 1993; 54:
466-9
eNS Drugs 1997 Feb; 7 (2)
Mechanism of Clozapine-Induced Agranulocytosis
44. Galletly C, Wilson D, McEwen S. Eosinophilia associated with
decreasing neutrophil count in a c1ozapine-treated patient. J
Clin Psychiatry 1996; 57: 40-1
45. Nathan DG. Stem cells and haematopoiesis. In: Weatherall DJ,
Ledingham JGG, Warrell DA, editors. Oxford textbook of
medicine. Oxford: Oxford University Press, 1987: 19.7-19.12
46. Terui Y, Furukawa Y, Sakoe K, et al. Expression of differentia-
tion-related phenotypes and apoptosis are independently reg-
ulated during myeloid cell differentiation. J Biochem 1995;
117: 77-84
47. Metcalf D. The molecular control of cell division, differentia-
tion commitment and maturation in haemopoietic cells. Na-
ture 1989; 339: 27-30
48. Wagstaff AJ, Bryson HM. C1ozapine: a review of its pharma-
cological properties and therapeutic use in patients with
schizophrenia who are unresponsive to or intolerant of clas-
sical antipsychotic agents. CNS Drugs 1995; 4: 370-400
49. Veys PA, Wilkes S, Shah S, et al. Clinical experience of c1ozap-
ine-induced neutropenia in the UK. Laboratory investigation
using liquid culture systems and immunofluorocytometry.
Drug Saf 1992; 7 Suppl. I: 26-32
50. Gerson SL, Arce C, Meltzer HY. N-desmethylclozapine - a
c10zapine metabolite that suppresses hematopoiesis. Br J
Haematol 1994; 86: 555-61
51. Labro MT, Babin-Chevaye C. Effects of amodiaquine, chloro-
quine, and mefloquine on human polymorphonuclear neutro-
phil function ill vitro. Antimicrob Agents Chemother 1988;
32: 1124-30
52. de Duave C, de Barsy T, Poole B, et al. Lysosomotropic agents.
Biochem Pharmacol 1974; 23: 2495-531
53. Jann MW, Grimsley SR, Gray EC, et al. Pharmacokinetics and
pharmacodynamics of c1ozapine. Clin Pharmacokinet 1993;
24: 161-76
54. Lin SK, Chang WH, Chung MC, et al. Disposition of c10zapine
and desmethylclozapine in schizophrenic-patients. J C1in
Pharmacol 1994; 34: 318-24
55. Pirmohamed M, Williams D, Madden S, et al. Metabolism and
bioactivation of c10zapine by human liver ill vitro. J Phar-
macol Exp Ther 1995; 272: 984-90
56. Jerling M, Lindstrom L, Bondesson U, et al. Fluvoxamine in-
hibition and carbamazepine induction of the metabolism of
c10zapine - evidence from a therapeutic drug-monitoring ser-
vice. Ther Drug Monit 1994; 16: 368-74
57. Fischer V, Vogels B, Maure G, et al. The antipsychotic c10zapine
is metabolised by the polymorphic human microsomal and
recombinant cytochrome P450 2D6. J Pharmacol Exp Ther
1992; 260: 1355-60
58. Dahl M-L, Llerena A, Bondesson U, et al. Disposition of
c10zapine in man: lack of association with debrisoquine and
S-mephenytoin hydroxylation polymorphisms. Br J Clin
Pharmacol 1994; 37: 71-4
59. Hasegawa M, Cola PA, Meltzer HY. Plasma c10zapine and
desmethylclozapine levels in c1ozapine-induced agranulocy-
tosis. Neuropsychopharmacology 1994; II: 45-7
60. Maggs JL, Williams D, Pirmohamed M, et al. The metabolic
formation of reactive intermediates from c1ozapine: a drug
associated with agranUlocytosis in man. J Pharmacol Exp
Ther 1995; 275: 1463-75
61. Fischer V, Haar JA, Greiner L, et al. Possible role offree radical
formation in c10zapine (Clozaril)-induced agranulocytosis.
Mol Pharmacol 1991; 40: 846-53
62. Uetrecht JP. Metabolism of c10zapine by neutrophils. Possible
implications for c1ozapine-induced agranulocytosis. Drug Saf
1992; 7 Suppl. I: 51-6
© Adis International Limited. All rights reserved.
157
63. Liu ZC, Uetrecht JP. Clozapine is oxidized by activated human
neutrophils to a reactive nitrenium ion that irreversibly binds
to cells. J Pharmacol Exp Ther 1995; 275: 1476-83
64. Pirmohamed M, Templeton E, Breckenridge AM, et al. III vitro
bioactivation of c10zapine to a chemically reactive metabolite
by human and mouse liver microsomes [abstract\. Br J Clin
Pharmacol 1992; 34: 168P
65. Tschen AC, Rieder MJ, Qyewumi K, et al. In vitro toxicity of
c10zapine metabolites [abstract\. Clin Pharmacol Ther 1996;
59:PlI12
66. Safferman AZ, Lieberman JA, Alvir JM, et al. Rechallenge in
c1ozapine-induced agranulocytosis [Ietter\. Lancet 1992; 339:
1296-7
67. Pisciotta AV, Konings SA, Ciesemier LL, et al. Cytotoxic activ-
ity in serum of patients with c1ozapine-induced agranulocy-
tosis. J Lab Clin Med 1992; 119: 254-66
68. Jaunkalns R, Shear NH, Sokol uk B, et al. Antimyeloperoxidase
antibodies and adverse reactions to c10zapine [Ietter\. Lancet
1992; 339: 1611-2
69. Sperner-Unterweger B, Gaggl S, Fleischhacker WW, et al. Ef-
fects of c10zapine on hematopoiesis and the cytokine system.
BioI Psychiatry 1993; 34: 536-43
70. Brach MA, de Vos S, Gruss HJ, et al. Prolongation of survival
of human polymorphonuclear neutrophils by granulocyte-
macrophage colony-stimulating-factor is caused by inhibi-
tion of programmed cell death. Blood 1992; 80: 2920-5
71. Batchelor JR, Welsh KI, Tinoco RM, et al. Hydralazine-induced
systemic lupus erythematosus: influence of HLA-DR and sex
on susceptibility. Lancet 1980; I: 1\07-9
72. Timbrell JA, Facchini V, Harland SJ, et al. Hydralazine-induced
lupus: is there a toxic metabolic pathway? Eur J Clin Phar-
macol 1984; 27: 555-9
73. Uetrecht JP, Zahid N, Whitfield D. Metabolism of vesnarinone
by activated neutrophils: implications for vesnarinone-in-
duced agranulocytosis. J Pharmacol Exp Ther 1994; 270:
865-72
74. Suboticanec K, Folnegovic-Smalc V, Korbar M, et al. Vitamin
C status in chronic schizophrenia. BioI Psychiatry 1990; 28:
959-66
75. Linday LA, Pippenger CE, Howard A, et al. Free-radical
scavenging enzyme-activity and related trace-metals in
c1ozapine-induced agranulocytosis - a pilot-study. J Clin
Psychopharmacol 1995; 15: 353-60
76. Yunis JJ, Lieberman J, Yunis EJ. Major histocompatibility com-
plex associations with c1ozapine-induced agranulocytosis.
The USA experience. Drug Saf 1992; 7 Suppl. I: 7-9
77. Pfister GM, Hanson DR, Roerig JL, et al. Clozapine-induced
agranulocytosis in a Native American: HLA typing and fur-
ther support for an immune-mediated mechanism. J C1in Psy-
chiatry 1992; 53: 242-4
78. Joseph G, Nguyen V, Smith 10. HLA-B38 and c1ozapine-in-
duced agranulocytosis [Ietter\. Ann Intern Med 1992; 116:
605
79. Claas FH, Abbott PA, Witvliet MD, et al. No direct clinical
relevance of the human leucocyte antigen (HLA) system in
c1ozapine-induced agranulocytosis. Drug Saf 1992; 7 Suppl.
I: 3-6
80. Yunis JJ, Corzo D, Salazar M, et al. Hla associations in c1ozap-
ine-induced agranulocytosis. Blood 1995; 86: 1177-83
81. Corzo D, Yunis JJ, Salazar M, et al. The major histocompatibil-
ity complex region marked by hsp70-1 and hsp70-2 variants
is associated with c1ozapine-induced agranulocytosis in 2 dif-
ferent ethnic-groups. Blood 1995; 86: 3835-40
eNS Drugs 1997 Feb; 7 (2)
158
82. Filippovich I, Sorokina N, Khanna KK, et a!. Butyrate-induced
apoptosis in lymphoid cells preceded by transient over-
expression of HSP70 mRNA. Biochem Biophys Res Com-
mun 1994; 257: 198-204
83. Takeda Y, Watanabe H, Yonehara S, et a!. Rapid acceleration of
neutrophil apoptosis by tumor-necrosis-factor-alpha. Int Im-
munol 1993; 5: 691-4
84. Park BK, Pirmohamed M, Tingle MD, et a!. Bioactivation and
bioinactivation of drugs and drug metabolites: relevance to
adverse drug reactions. Toxicol In Vitro 1994; 8: 613-21
85. Neftel KA, Woodtly W, Schmid M, et a!. Amodiaquine induced
agranulocytosis and liver damage. BMJ 1986; 292: 721-3
86. Clarke JB, Maggs JL, Kitteringham NR, et a!. Immunogenicity
of amodiaquine in the rat. Int Arch Allergy Appl Immunol
1990; 91: 335-42
87. Tingle MD, Jewell H, Maggs JL, et a!. The bioactivation of
amodiaquine by human polymorphonuclear leukocytes i/1
!'ilro - chemical mechanisms and the effects of fluorine sub-
stitution. Biochem Pharmacol 1995; 50: 1113-9
88. Harrison AC, Kitteringham NR, Clarke JB, et al. The mecha-
nism of bioactivation and antigen formation of amodiaquine
in the rat. Biochem Pharmacol 1992; 43: 1421-30
89. Jewell H, Maggs JL, Harrison AC, et a!. Role of hepatic meta-
bolism in the bioactivation and detoxication of amodiaquine.
Xenobiotica 1995; 25: 199-217
90. Maggs JL, Tingle MD, Kitteringham NR, et al. Drug-protein
conjugates. XIV. Mechanism of formation of protein-arylat-
ing intermediates from amodiaquine, a myelotoxin and hepa-
totoxin in man. Biochem Pharmacol 1988; 37: 3(J3-11
91. Clarke JB, Neftel K, Kitteringham NR, et al. Detection of anti-
drug IgG antibodies in patients with adverse drug reactions
to amodiaquine. Int Arch Allergy Appl Immunol 1991; 95:
369-75
92. Lind DE, Levi JA, Vincent Pc. Amodiaquine-induced agranu-
locytosis: toxic effect of amodiaquine in bone marrow cul-
tures i/1 vilm. BMJ 1973; I: 458-60
© Adis International limited. All rights reseNed.
View publication stats
Pirmohamed & Park
93. Ruscoe JE, Jewell H, Maggs JL, et al. The effect of chemical
substitution on the metabolic-activation, metabolic detoxica-
tion, and pharmacological activity of amodiaquine in the
mouse. J Pharmacol Exp Ther 1995; 273: 393-404
94. Kane JM. Newer antipsychotic drugs. A review of their phar-
macology and therapeutic potential. Drugs 1993; 46: 585-93
95. Gerlach J, Peacock L. New anti psychotics: the present status.
Int Clin Psychopharmacol 1995; 10 Suppl. 3: 39-48
96. Livingston MG. Risperidone. Lancet 1994; 343: 457-60
97. Kerwin R. Adverse drug reaction reporting and new antipsy-
chotics. Lancet 1993; 342: 1440
98. Laidlaw ST, Snowden JA, Brown MJ. Aplastic anaemia and
remoxipride [letter]. Lancet 1993; 342: 1245
99. Philpott NJ, Marsh JC, Gordon-Smith EC, et a!. Aplastic anae-
mia and remoxipride. Lancet 1993; 342: 1244-5
100. Park BK, Kitteringham NR. Effects of fluorine substitution on
drug metabolism - pharmacological and toxicological impli-
cations. Drug Metab Rev 1994; 26: 605-43
10 I. Davis DA, de Paulis T, Janowsky A, et al. ChI oro-substituted,
sterically hindered 5, II-dicarbo analogues of clozapine as
potent chiral antipsychotic agents. J Med Chern 1990; 33:
809-14
102. Rabey JM, Treves TA, Neufeld MY, et a!. Low-dose clozapine
in the treatment of levodopa-induced mental disturbances in
Parkinson's disease. Neurology 1995; 45: 432-4
103. Sajatovic M, Ramirez L. Clozapine therapy in patients with
neurologic illness. Int J Psychiatry Med 1995; 25: 331-44
Correspondence and reprints: Dr M. Pirmohamed, Depart-
ment of Pharmacology and Therapeutics, The University of
Liverpool, Liverpool, L69 3BX, UK.
Email: munirp@liverpooLac.uk
eNS Drugs 1997 Feb: 7 (2)