Clozapine Re-Challenge and Initiation Despite Neutropenia and Outcomes from 14 Patients

Objective: Clozapine remains the most effective intervention for treatment resistant schizophrenia. But its use is limited because of the risk of neutropenia and agranulocytosis. In the absence of effective alternatives its use after blood dyscrasias is possible. Methods: A case series of 14 male patients with treatment resistant schizophrenia treated with clozapine despite previous episodes of neutropenia between 2006 and 2015 is presented. Data were collected during 2015 and 2019. Using routinely collected clinical data we describe the patient characteristics, causes of neutropenia, the strategies used and outcomes. Results: Previous neutropaenia was due to varying aetiologies: benign ethnic neutropaenia; clozapine induced; other medication; autoimmune. A range of risk mitigation strategies were used to prevent neutropaenia on clozapine rechallenge: G-CSF; lithium; watch and wait. There were no serious adverse events associated with clozapine or risk mitigation treatments. At follow up during 2019, half of the original 14 patients had improved sufficiently to transfer to conditions of lesser security. Conclusion: With careful planning and monitoring, taking into account what is known about neutrophil biology, clozapine can be successfully reinitiated following CIN or in the face of preexisting neutropenia resulting in significant benefits for patients.


Background
Clozapine remains the most effective intervention for treatment resistant schizophrenia (TRS) (1) but remains a third line treatment due to the risk of clozapine induced neutropenia (CIN, ANC(absolute neutrophil count) <1.5x10 9 /L) and agranulocytosois (CIA, ANC <0.5x10 9 /L) with rates of 2.7% and 0.7-0.8% respectively. The majority of the abnormalities occur within the first 18 weeks of treatment initiation and almost all within the first year (2) after which the risks become similar to those posed by other antipsychotics. Clozapine may only be used within licence if there is no previous history of neutropenia and with blood monitoring. This risk mitigation regime has been highly effective, with deaths from blood dyscrasias being orders of magnitude lower than those associated with cardiac and gastrointestinal side effects. The MHRA records eight deaths from clozapine induced blood dyscrasias to date as opposed to hundreds from cardiac and gastrointestinal adverse drug reactions (3). In 2006 we were faced with patients who, having had good responses to clozapine, had deteriorated severely when this had been discontinued due to blood dyscrasias and who had failed to respond to multiple alternative regimes; we therefore began to explore both clozapine rechallenge following CIN / CIA and then initiation in the face of pre-existing neutropenia. We present an overview of neutrophil biology relevant to clozapine rechallenge, summarise some of the recent literature relating to rechallenge, the use of lithium and cytokines as support and report the outcomes from 14 cases. In an accompanying article we describe the decision making and practical approaches used (REF).

Mechanisms of CIN and CIA
The mechanisms of CIN and CIA remain elusive and may be distinct (4). Whilst clozapine itself is not directly toxic to neutrophils its metabolites may be (5). Genetic linkage studies suggest both immune and anion transporter gene related mechanisms (6-9) but are not sufficiently predictive to be clinically useful and have predominantly examined European populations, so limiting applicability in other populations (10). CIN and CIA remain idiosyncratic, unpredictable type B adverse drug reactions (ADR). The hypothesis of an immune mediated mechanism is supported by the characteristic timing of both initial adverse effects and the tendency to have faster and more severe problems on rechallenge (11).

Neutrophil Biology
Neutrophils primarily protect the host from bacterial infections. They are produced in the bone marrow differentiating from pluripotent stem cells over 10-14 days. Mature neutrophils are stored in the bone marrow before being released into the blood where they circulate prior to leaving the circulation randomly into tissues with a half-disappearance time of about six hours. The cells then survive for a further 24 to 48 hours. The bone marrow neutrophil reserve is 10-20 times the number of circulating cells and can be mobilised in acute infection or inflammation. Bone marrow neutrophil production is regulated by cytokines. Recombinant pharmaceutical products are available which are used for the treatment of neutropenia in various clinical settings. Neutrophil counts in blood samples vary regardless of their absolute numbers (12,13) with diurnal variation peaking in the afternoon/evening in response to variation in Granulocycte Colony Stimulating Factor (G-CSF), one of the cytokines stimulating neutrophil differentiation and release (14,15). External influences such as exercise and smoking also lead to an increase in circulating neutrophils (16,17).
The normal neutrophil count is defined mathematically rather than pathologically. The lower limit for neutrophils counts is set at 1.5 x 10 9 /L, two standard deviations below the normal population mean; this is well above the level associated with a significant risk of infection (less than 0.5 x 10 9 /L) which is classified as severe neutropenia at which guidelines recommend action for the general or oncology patient (18). Neutropenia is either caused by reduced/ineffective production in the bone marrow or by accelerated cell destruction, sequestration or egress from the circulation and can be inherited or acquired (19). In the haematology literature agranulocytosis is defined as a complete absence of neutrophils but the term is often used to indicate severe neutropenia ANC<0.5x10 9 /L(20).
These processes are important when thinking about the use of clozapine. Clozapine blood testing may be optimised by afternoon blood sampling, ideally after exercise, and this alone has been demonstrated to be sufficient to improve outcomes (21). The Clozaril Patient Monitoring Service (CPMS) categorise neutrophil counts according to a colourcoded system (22). Table 1 If a patient's neutrophil count is less than 1.5 x 10 9 /L (red alert) clozapine treatment must be stopped immediately. A follow-up blood sample should be taken the 2 days following the date of the red alert sample; if either of these follow-up blood counts is in the red alert range then the red alert is confirmed and the patient must not restart clozapine. This provides a significant margin to neutrophil levels below which infection is likely and so in some cases flexibility regarding lower acceptable limits might be considered, especially in cases with low pre-clozapine neutrophil counts.

Benign Ethnic Neutropenia
Asymptomatic or benign reductions in neutrophils are observed in individuals of all ethnic backgrounds. Benign Ethnic Neutropenia (BEN), neutropenia without immune dysfunction or increased liability to infection is due not to a reduced absolute number of neutrophils but increased margination (23). It has been described in various non-Caucasian populations. In studies of Black US populations, clinically significant proportions (4.5%) show neutrophil counts less than 1.5 x 10 9 /L or even less than 1.0 x 10 9 /L (0.57%) (16).
The Duffy Null polymorphism which protects against some types of malaria is predicative of BEN (24) and can be determined simply by a request an extended blood type. It has recently been shown to be associated with the development of clozapine induced neutropenia and withdrawal in clozapine treated UK patients of African genetic ancestry (9) to the extent that there may be the potential to use this readily available genetic investigation in combination with other clinical information, the presence of a fluctuating neutropenia in the absence of infection or other pathology in an affected group, to adjust neutrophil thresholds (25). Since 2002 clozapine monitoring services in the UK have reference ranges 0.5 x 10 9 /L lower for patients with haematologically confirmed BEN (see Table 1) (26). Therefore some patients who received clozapine prior to 2002 will have had clozapine discontinued on account of 'red' results that can, in retrospect, be re-evaluated.
Afro-Caribbean patients are less likely that their peers to be prescribed clozapine (27) (28).

Clozapine Re-challenge after CIN and CIA
Given the limitations of case reports from populations treated with clozapine and then rechallenged after blood dyscrasia, the available evidence is useful but conflicting. In 2006, the CPMS published the data of all available rechallenges in the UK and Ireland. Of 53 patients presented, 20 patients (38%) experienced a further blood dyscrasia. The second blood dyscrasia was more severe in 17 of these patients (85%), in 12 patients (60%) longer and in 17 patients (85%) it occurred more quickly. Nine (45%) of the 20 cases of dyscrasias were agranulocytosis. Yet 55% of all cases (29 patients) did sufficiently well to continue to receive clozapine. There were no deaths. No clear risk factors for repeat blood dyscrasias were identified. Only one CIA patient was rechallenged and that individual did not go on to develop a further dyscrasia. None of the 20 failed re-challenge patients went on to a third attempt (29). More recently, data from the results of a nationwide pharmacovigilance study in Argentina presented the outcomes of 19 clozapine rechallenge patients; almost 70% of patients did not experience a second dyscrasia and of those that did, the dyscrasia occurred sooner during treatment but unwanted effects were less severe and of shorter duration (30). Evidence from case reports show positive outcomes in more than 60% of clozapine rechallenges post neutropenia (128 positive outcomes from 203 reported cases); fewer attempts are described following agranulocytosis and with less clinical success (3 positive outcomes from 17 reported cases) (31). There is a lack of data relating to the retitration regimes used and the outcome experienced. This is potentially important if immune mechanisms are responsible for the adverse events.

Lithium and neutrophils
Lithium has varying effects on haematopoiesis. It enhances the production of G-CSF and directly stimulates the differentiation of stem cells into neutrophils resulting in increased absolute neutrophil numbers. It also protects neutrophils from the toxic effects of some drugs. In haematological practice it has shown some utility in treating cases idiopathic neutropenia and syndromes in which G-CSF production is reduced (32). The use of lithium to support clozapine rechallenge has been reviewed by both Manu (2012) and Boazak (2018) with high rates of success. However, Boazak noted that, in the few reported cases where lithium was withdrawn following an initially successful rechallenge, a second dyscrasia had followed. As an adjunct to clozapine rechallenge the familiarity psychiatrists have with the use of lithium, patient acceptability, low cost and the attitude of downstream services are all helpful (33)(34)(35). There are a number of disadvantages: lithium's effects on neutrophil production, although significant, are markedly less than the alternative of G-CSF; there is a risk of reversible neurotoxicity (36); and case reports of the use of lithium and clozapine include one in which there was fatal agranulocytosis (37) and another by a failure of subsequent G-CSF (38). There is also the concern of poor compliance increasing relapse of affective symptoms, whether or not this has been clearly demonstrated in patients with a specific diagnosis of schizophrenia or schizoaffective disorder opposed to bipolar disorder (39).

G-CSF and neutrophils
G-CSF is a cytokine glycoprotein that stimulates the differentiation, release and survival of neutrophils and other granulocytes. As recombinant products, they have been available for therapeutic use since the early 1990's and were originally used mainly to treat patients with chemotherapy-induced neutropenia for short durations, over time, the indications expanded to include severe chronic neutropenia and for the mobilisation of haematopoietic progenitor cells for stem cell transplantation (20,40,41). Filgrastim (G-CSF) is a WHO essential medicine (42) and is used to treat drug induced neutropenia in general (43). There are multiple case reports in relation to CIN and CIA (40,(44)(45)(46)(47)(48)(49)(50)(51)(52)(53)(54)(55)(56) reviewed by Lally et al (57,58); the general finding being that rechallenge supported by either regular G-CSF or as required (plus rescue) G-CSF was successful in over 70% of initial CIN cases. However, the reported clinical success was markedly reduced and with a risk of repeat agranulocytosis following CIA; in such cases, G-CSF can be safely used to speed recovery following CIA.
The use of G-CSF is not part of usual psychiatric practice. Common side effects of shortterm use include flu like symptoms, bone pain, headache, pyrexia and fatigue. Rare complications include splenic rupture, Acute Respiratory Distress Syndrome, glomerulonephritis, alveolar haemorrhage, serious allergic reactions, thrombocytopenia, cutaneous vasculitis and capillary leak syndrome. Long term safety data following short term G-CSF use is available from healthy volunteer donors of peripheral blood progenitor cell (PBPC); approximately 20,000 healthy donors receive G-CSF worldwide yearly (59).
Appropriate precautions relating to the longer term use of G-CSF include the management of bone pain with paracetamol as well as regular ultrasonography to monitor for splenomegaly (60). Long-term G-CSF administration predisposes to the development of osteopenia/osteoporosis (61). However, osteopenia is a recognised finding in this patient population and is often present prior to treatment initiation. Regular bone mineral density assessment in patients receiving long term G-CSF is recommended, as it will enable the detection of subtle bone changes and early treatment initiation that can prevent the development of symptomatic osteoporosis. Initial concerns that G-CSF may cause an increased leukaemia risk have been alleviated by long term follow up (62).

Patterns of Clozapine Response
Patients with TRS who do respond to clozapine frequently show this early in treatment, within days or weeks especially when adequate levels have been achieved (63)(64)(65); this is consistent with response to other antipsychotics when used first line (66). Persisting with a potentially hazardous treatment in the absence of significant improvement after an adequate trial is not recommended.

Patient Demographics:
Fourteen subjects were identified; four had transferred to lesser security by the first census date and a further three by the second. All were male with a primary diagnosis of schizophrenia (ICD-10 F20). The indication for clozapine in each case was treatment resistance. All had been extremely unwell for many years having been in high security for an average 5.1 years prior to the interventions described. All but two patients were being nursed in seclusion or long term segregation (67) at the time of the studied clozapine trial for an average of 486 days; for two patients this was in excess of 2,000 days. These restrictions had been used on account of persistent and at times extreme violence. Three patients had originally been transferred to high security in the context of deteriorations secondary to clozapine discontinuation due to neutropenia.

Previous treatments:
All had failed to adequately respond to multiple non-clozapine antipsychotics (min 3, max 15, average 7.8) alone and as polypharmacy; 12 patients had received these drugs at doses above the recommended UK maxima (TABLE 2). Ten had been prescribed clozapine for an average of 394 days (min 35, max 1785) before this had been stopped. Seven patients had shown improvement with clozapine; there was no evidence to suggest the remaining three patients had improved on clozapine although they had taken clozapine on average for 77 days only (min 35, max 150). During previous clozapine treatment a change to an alternative antipsychotic treatment was necessary following a period of neutropenia (N=8) or non-compliance (N=2) In all cases, a clinical decision was made to trial a period of treatment with clozapine due to the failure of all other antipsychotic treatment regimes and the severity of the patients' psychoses. The risk of possible treatment complications, in particular a further episode of neutropenia, was considered against likely benefits of clozapine treatment. In addition, the anticipated degree of difficulty in maintaining patient and staff safety in the event of a blood dyscrasia resulting in clozapine discontinuation, rebound psychosis and extreme behavioural disturbance was considered.

Aetiology of neutropenia:
All had previous neutropenia (ANC less than 1.5 x 10 9 /L due to various aetiologies). Five had previous CIN one of whom had also had agranulocytosis following treatment with another antipsychotic. Two patients had neutropenia associated with other medications (sodium valproate and risperidone) prior to the use of clozapine and one developed neutropenia related to the use of lamotrigine during clozapine therapy (duration of 1 day with a neutrophil nadir of 0.6 x 10 9 /L). BEN was present in five cases: two of these patients had pre-clozapine neutrophil counts that were frequently below the BEN adjusted neutrophil cut offs; two had previously been exposed to clozapine prior to BEN having been recognized and clozapine had been stopped when neutrophils fell below the non-BEN levels; and one patient had not had a previous diagnosis of BEN. Another patient had an autoimmune neutropenia. Unfortunately, the data were incomplete as a small number of the adverse events occurred a number of years previously in other healthcare settings; where available, previous clozapine related neutropenic episodes had duration of 7 days on average (min 2, max 21) with an average nadir of 1.0 x 10 9 /L (min 0.7, max 1.3) mostly on the second neutropenic day (n=4) (see Table 3). There was one previous episode of agranulocytosis with associated spontaneous infection in the patient who had had both CIN and agranulocytosis secondary to another antipsychotic.

Risk Mitigation Strategies:
The relationship between the type of neutropenia and risk mitigation strategy used is set out in TABLE 4. A watch and wait approach was taken with the patients who had previous non-clozapine related drug induced neutropenia. For one patient who developed neutropenia during established clozapine treatment when lamotrigine was introduced, two doses of G-CSF were used, lamotrigine was withdrawn and clozapine continued. The patient with autoimmune neutropenia was given clozapine using the BEN reference ranges with provision for G-CSF if required. For the cases of established CIN, prophylactic or rescue G-CSF was used. A variety of approaches including lithium, rescue and prophylactic G-CSF were used taking into account the challenges of the patients with BEN and low neutrophils.
However, as Von Moltke observed, no plan survives first contact with the enemy (68) and so the initial strategies did not all succeed. One patient prescribed lithium did not tolerate this due to sedation and so the alternative of rescue G-CSF was made available. Another patient with BEN and low pre-clozapine neutrophils had a series of neutropenic episodes when clozapine was introduced as planned; these were initially treated with rescue G-CSF but after neutrophils reached a nadir of 0.4x10 9 L and there was then a good response to clozapine and G-CSF, both were continued regularly. One CIN case required prophylactic G-CSF following repeated episodes of neutropenia with rescue G-CSF. A further CIN case was switched from prophylactic G-CSF to rescue G-CSF without any further neutropaenia.

Medication and dose:
At the first census date one patient experienced two amber results before a single red result within the first month of treatment reaching a maximum dose of 200mg per day before clozapine was withdrawn. He had reportedly shown some clinical improvement. The remaining patients were able to tolerate clozapine and at the first data collection point the most recent average dose of clozapine (N=13) was 354mg/day (min 225mg, max 450mg). Hematological advice on the use of G-CSF took into account the patient's body weight, the severity of the previous neutropenia and the effect of clozapine. Where regular G-CSF was used doses varied between 15 and 60 million units per week. One patient was prescribed lithium both for his mental disorder and to stimulate neutrophil counts and so this was at 800mg/day. A lower dose of lithium at 400mg/day was used for another patient when only a neutrophil effect was needed.

Progress:
The treating clinicians considered that all patients had showed a clinical improvement on clozapine. By the first census point all except one remained on the drug and four had been transferred to lesser security. At the second data collection point, of the ten patients remaining in 2015, a further three patients had discontinued clozapine treatment: one patient had died of an unrelated cause; one had refused clozapine for unspecified reasons; and clozapine had been discontinued in another patient due to suspected poor clozapine absorption in relation to an inflammatory bowel condition. A further three patients had remained on clozapine and improved sufficiently to transfer to less secure services. The remaining three patients showed a variable progress but remained on clozapine; one patient had progressed to the extent that very prolonged seclusion had ended.
In summary, by the second census date, half of the original 14 patients had improved sufficiently to transfer to lesser security; one patient was transferred directly to a low secure unit and the others to medium secure facilities. The transfer process was slow,

Unwanted effects of treatment:
Four had anomalous blood results. It was suspected that one related to inadvertent morning blood sampling, late in treatment, so the logistics of blood taking were revisited but provision was made for rescue G-CSF. One patient with BEN who had no prior clozapine experience rapidly developed repeated low neutrophil counts of less than 1.0 x 10 9 /L with one brief episode at 0.4 x 10 9 /L. Each episode responded to G-CSF given first intermittently and then regularly. One patient whose initial clozapine had been discontinued after neutropenia at over 800 days of treatment had six episodes of low neutrophils with a nadir of 0.4 x 10 9 /L; these episodes, within the first two months of clozapine re-challenge, responded to intermittent and then regular G-CSF. Another patient had two amber results followed by a single red result (neutrophil count 1.3 x 10 9 /L) within the first month of clozapine treatment; clozapine was stopped by the treating team. No patient developed infection or required transfer to general hospital. Of those prescribed G-CSF (rescue or prophylactic), one patient was noted to have splenomegaly on ultrasound imaging; this continues to be monitored by hematology colleagues; no other intervention has been required. No patient has complained of bone pain while on G-CSF. Lithium was discontinued in one case due to sedation.

Discussion
We have described one of the largest series of the use of clozapine following neutropenia from any cause and we believe our length of follow-up exceeds any previously reported.
Our high secure forensic patient group presented particular problems, with extreme disturbance, highly uncooperative behavior, lengthy periods of treatment resistant psychosis, the prolonged use of isolation and lengths of stay perhaps more meaningfully measured in years than weeks. Yet even in this group the unique efficacy of clozapine was evident, with life changing effects for many of the men we describe. Some were transformed, from individuals completely incapacitated by psychosis with life threatening harm both to themselves and others and needing the most restrictive interventions imaginable, to men able to participate freely in all aspects of hospital life and who have now been discharged, not only to less secure hospitals but also now to community settings. One man went on to give evidence about Mental Health services at the House of Lords. This response to clozapine is, of course, the usual pattern, even in the most difficult forensic populations (69,70). As well as transforming patients' lives these interventions have resulted in dramatic reductions in length of stay, expense, and the use of seclusion, segregation, isolation or whatever other argot is used to describe solitary confinement; an issue of both national and international concern (71, 72).
Despite progress, understanding of the aetiology of life-threatening clozapine blood dyscrasias remains insufficient. There are many causes of neutropenia; its definition is not uncontroversial and a low neutrophil count is not necessarily indicative of significant pathology. The existing neutrophil thresholds for the continuation of clozapine are extremely effective at safeguarding patients, but may well be so safe that many who would benefit are not eligible, particularly those of African ancestry. Since at least 1993 the psychiatric literature relating to clozapine has defined agranulocytosis as a neutrophil count of less than 0.5x10 9 /L; (73) this has been convention since. But at the time of the first Finnish cases, argranulocytosis was described as the presence of very few or no granulocytes in the peripheral blood (74) and our haematology colleagues subdivide neutropenia into mild (1.5-1.0x10 9 /L) moderate (1.0-0.5x10 9 /L) and severe (0.5-0.2x10 9 /L) with or without sepsis. Agranulocytosis is strictly the complete absence of neutrophils and it is at counts of less than 0.2x10 9 /L at which infection is most likely (20). Absolute neutrophil counts have rarely been set out in the reported clozapine rechallenge cases and it is possible that a significant number have included subjects with either benign mild neutropenia coincidental to clozapine use or have taken a broader definition of agranulocytosis than that in the original Finnish cases. In 2015 the US FDA guidelines for interruption of clozapine treatment changed. White blood cell count (WBC) monitoring is no longer required and once clozapine is established the threshold for an absolute neutrophil count requiring treatment interruption was lowered from 1.5x10 9 /L to 1.0x10 9 /L and to below 0.5x10 9 /L in cases of BEN (75)(76)(77).
We took an arguably aggressive approach to supporting neutrophil counts using significant amounts of G-CSF. Our decision-making was in the light of the existing UK neutrophil thresholds and with patients for whom cessation of clozapine would result in adverse outcomes: at best a failure to progress; at worst life threatening violence to themselves or others. We were particularly mindful of the difficulties that would be faced by all should we need to care for an acutely psychotic man, experiencing a clozapine rebound psychosis with severe neutropenia or agranulocytotosis requiring treatment in reverse barrier isolation on a haematology ward; being sure that this scenario would be unmanageable we took what we considered to be the necessary steps to avoid it. As with other reported series of clozapine rechallenge other less interventional approaches may have been successful. In particular a simple revision of the neutrophil thresholds on an off-license basis. Better still would be wholesale change to align UK and US practice (78).
We are unclear why the psychiatric literature has used ANC's of below 0.5x10 9 /L as a single cut off to define agranulocytosis. Similarly there has been no differentiation between the different degrees of neutropenia as used elsewhere. The clozapine literature could be improved by taking that approach, particularly when attempting to improve our understanding of the aetiologies of these difficult adverse events. As described by Legge we hope that a wider understanding of the role of the Duffy-null genotype in BEN in general and in low neutrophil counts in patients taking clozapine in particular (and potentially other psychotropic drugs with a risk of neutropenia) will translate in to an easily obtainable pharmacogenetic test (9). Duffy typing can be performed following a routine laboratory request for a blood group. Prevention is better than cure. Given that a significant proportion of cases of CIN and CIA likely have an immunological mechanism it may well be that slower rates of titration, both on initial use and in the event of rechallenge may result in reductions in adverse events (79). This approach has been effective for rechallenge of lamotrigine and carbamazepine following adverse serious side effects (80)(81)(82).
We have described several approaches to allow rechallenge of clozapine despite previous neutropaenia. We are pleased to report that there were no serious adverse events but instead marked improvements in the quality of life for our patients. Careful mitigation of further neutropaenia with a flexible approach is needed to ensure the patient is offered the best opportunity for recovery; it may be necessary to change mitigtaion strategy at short notice to encourage compliance and minimize risk of life-threatening blood dyscrasia. We advocate close working with haematology colleagues when prescribing clozapine to patients with previous neutropaenia. Our case series demonstrates that rechallenge with clozapine following neutropenia should not be dismissed without full consideration of the potential benefits.