Letter to the FDA Proposing Major Changes in the US Clozapine Package Insert Supported by Clozapine Experts Worldwide. Part II

Abstract

Purpose/Background

This is the second part of a 2-part article that proposes improving the United States (US) clozapine package insert. Part II focuses on fatal outcomes and the 5 boxed warnings, 4 specifically for clozapine: severe neutropenia, seizure, orthostatic hypotension and myocarditis, and 1 for all antipsychotics (elderly with dementia).

Methods

US reports to the World Health Organization's global pharmacovigilance database were analyzed from clozapine's introduction to January 15, 2023.

Findings/Results

The US was the top reporter worldwide for clozapine with 56,003 reports and 9587 associated fatal outcomes. The 4 clozapine boxed warnings were associated with 534 fatal outcomes (218 with severe neutropenia, 131 with seizures, 125 with orthostasis, 36 with myocarditis, 24 with cardiomyopathy, and 0 with mitral valve prolapse). With no boxed warnings, pneumonia was associated with 674 fatal outcomes and increased white blood cell count (a sign of infection) with 596 fatal outcomes. After considering overlaps, pneumonia and increases in white blood cell count explained 900 fatalities, or 9.4% of 9587 fatal outcomes. The Food and Drug Administration continues to focus on severe neutropenia which was associated with only 218 or 2.3% of fatal outcomes, whereas 97.7% of fatal outcomes reported in US clozapine-treated patients had another cause.

Implications/Conclusions

To help prevent future deaths in clozapine-treated patients, the clozapine package insert should focus on fatal outcomes during infections. Part II offers detailed solutions regarding current boxed warnings and lack of a warning for pneumonia and other infections. The Supplementary Material includes letters of support from 124 non-US clozapine experts from 44 countries/regions who support Parts I and II.

PHARMACOVIGILANCE

Randomized clinical trials (RCTs) conducted during the premarketing period provide some information about drugs. These RCTs are carried out under controlled conditions, exclude the most complicated patients, and usually last for a few weeks or a few months; thus, they may not identify rare but potentially lethal adverse drug reactions (ADRs). When new drugs are approved by the Food and Drug Administration (FDA) they have been studied in only 3000–5000 patients. Therefore, ADRs that occur in fewer than 1 in 5000 patients would likely go undetected.¹ For postmarketing surveillance then, the national drug agencies, including the FDA, consider pharmacovigilance a critical component of medication safety to detect, assess, understand, and prevent ADRs. Pharmacovigilance helps ensure that the benefits of medications outweigh their risks. Effective pharmacovigilance systems enable the early identification of safety signals, facilitating timely interventions to mitigate ADRs² and have led to withdrawal of several drugs from the US market.³

Furthermore, the FDA uses the voluntary spontaneous pharmacovigilance reports from the United States (US) to consider the need of boxed warnings to bring attention to the more serious and potentially lethal ADRs. Supplementary Box S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A948, summarizes the characteristics of the boxed warnings (previously called black box warnings).⁴

HISTORY OF CLOZAPINE PHARMACOVIGILANCE

Clozapine has a very peculiar history heavily influenced by pharmacovigilance.^5,6 Clozapine underwent a few RCTs in the 1970s in German-speaking countries, including a total of 2200 patients and was marketed in 1972 in Switzerland and Austria.⁷ In 1973, the clozapine marketer hired Honigfeld to start the US studies required by the FDA at the time.⁷ Then in 1975, clozapine was marketed in Finland and immediately after its marketing, a pharmacovigilance report described 8 deaths associated with agranulocytosis.⁸ This pharmacovigilance report served to halt the US studies. The Finnish deaths changed the perspective on the risk-benefits of clozapine as agranulocytosis cases described before 1975 but were not considered more concerning than those associated with chlorpromazine and other phenothiazines.⁹ A decade later after a complex negotiation,^10,11 the FDA required an RCT focused on treatment-resistant schizophrenia (TRS) to approve clozapine in the US for that indication. This RCT was published in 1988¹² and the FDA approved clozapine for TRS in 1989.⁷ Later in 1999, a pharmacovigilance report from Australia associated clozapine with myocarditis.¹³ This led first to a review of the US pharmacovigilance data by the FDA¹⁴ and then a modification of the US package insert in 2002.⁵

BOXED WARNING IN THE US CLOZAPINE PACKAGE INSERT

Following these pharmacovigilance studies, the FDA required that the US clozapine package insert include the following 5 boxed warnings concerning: 1) severe neutropenia, 2) seizure, 3) orthostatic hypotension, 4) myocarditis, and 5) one for the elderly with dementia (Table 1). Not all countries use the same boxed warnings in their clozapine package insert. Three other English-speaking countries, Australia, Canada, and the United Kingdom (UK) have sophisticated pharmacovigilance programs used to modify their packager inserts. They are the other leading reporters of clozapine ADRs in the world.¹⁵ These 4 countries have different regulations and provided different practical results regarding boxed warnings for clozapine. The Canadian clozapine package insert appears to be heavily influenced by the US and also has 5 boxed warnings. The Australian clozapine package insert has 2 boxed warnings. The UK clozapine package insert has a unique boxed warning which has a different purpose as it lists the indications together with agranulocytosis and myocarditis/myocardiopathy. Other Western European countries have national health systems similar to the UK, but after correcting for population and use, they report only 1%–10% of the UK's fatal outcomes in clozapine-treated patients.¹⁶ This difference is a further indication of the UK's excellence in clozapine safety.^15–17

TABLE 1.

Misalignments of 9587 Fatal Outcomes Reported by the FDA to VigiBase and Only 534* Fatal Outcomes From 5 CLO Boxed Warnings

	Year†	Fatal Outcomes		FDA Classification‡	Prevention
		1990–1999	Until 01/15/23
			Boxed warnings
Severe neutropenia	1989	48 (5% of 929)	218 (2% of 9587)	SeriousPreventable: WBCRestricting	WBC monitoring for 18 weeks REMS may ↑ TRS mortality due to CLO underuse in US
Seizures	1989	27 (3% of 929)	131 (<2% of 9587)	SeriousPreventable: Hx§	Preventable by careful dosing, attention to DDI and TDM
Orthostasis and	1992	46 (5% of 929)	125 (<2% of 9587)	SeriousPreventable: slow escalation	Personalized titration
CR arrest				SeriousPreventable: caution BZD	Avoid BZD during titration
Myocarditis	2002	0 (0% of 929)	36 (<1% of 9587)	Serious	CRP monitoring and personalized titration
Cardiomyopathy	2013	13 (<2% of 929)	24 (<1% of 9587)	Serious	Extremely rare
Mitral valve incompetency	2016	0 (0% of 929)	0 (0% of 9587)	Unclear why it was added	Not needed in boxed warning
Elderly with dementia	2005			CLO not approved	Rarely used in dementia Aspiration among nongeriatric
			No boxed warnings
Broad pneumonia		112 (12% of 929)	674 (7% of 9587)	No boxed warning	Education on relevance Dose reduction during infection and TDM
↑ WBC: infection sign		46 (5% of 929)	596 (6% of 9587)	No boxed warning	Education on relevance Dose reduction during infection and TDM

*The total of 534 fatal outcomes assumes that there were not overlaps between the fatal outcomes associated with the 5 boxed warnings. It is possible that some minor overlap exists and that the number is slightly lower.

†Year when the boxed warning was added to the CLO package insert.

‡Supplementary Box S1, http://links.lww.com/JCP/A948, provides the FDA definition for the various types of boxed warnings.

§At the time of approval, the package insert recommended caution in “patients having a history of seizures or other predisposing factors.” In the current version, it is more specific, “in patients having a history of seizures or other predisposing factors for seizures (CNS pathology, medications that lower seizure threshold, alcohol abuse).”

BZD, benzodiazepine; CR, cardiorespiratory arrest; DDI: drug-drug interactions; Hx: history of seizures.

The complications associated with constipation, usually labeled clozapine-associated gastrointestinal hypomotility (CIGH),¹⁸ led the FDA to strengthen this warning in 2020.¹⁹ In the section called Warnings and Precautions a new subsection was added titled “5.8 Gastrointestinal Hypomotility with Severe Complications.” This section describes the increased risk of adding anticholinergics and recommends considering “prophylactic laxatives in high-risk patients.”

This FDA decision was a wise decision because 2 recent long-term studies from Canada²⁰ and Finland²¹ described CIGH as the second-leading clozapine ADR in burden (Table 2). Most clinicians would think that severe neutropenia would be the top ADR in burden, but they may be surprised to know that pneumonia was the leader in burden in both studies (Table 2).

TABLE 2.

Two Recent Studies on the Burden of Clozapine ADRs

	Canada20	Finland21
Method	Electronic medical records	Biobank data
Duration	Up to 10 years	Up to 25 years Median 12.7 years
Location	3 provinces	Whole country
Patients	45,616 adults 176,691 mental health relapses 11,405 rehospitalization due to ADRs	2659 on CLO
Outcome compared	CLO 14% lower relapse rate aHR = 0.86 (CI 0.83–0.90) CLO higher ADR risk HR = 1.34 (CI 1.18–1.54)	OLA used as control but they could not control for TRS ARP estimated CLO contribution to each ADR
Pneumonia*	1st cause of burden among studied 8319 incidents and 4521 hospitalizations	1st cause of burden† 29.5% cumulative incidence and ARP = 0.40
CIGH	2nd cause of burden among studied 2502 incidents and 1621 hospitalizations	2nd cause of burden: ileus 5.3% cumulative incidence and ARP = 0.53 constipation 22.7% cumulative incidence and ARP = 0.46
Neutropenia	3rd cause of burden among studied 284 incidents and 223 hospitalizations	2.4% cumulative incidence and ARP = 0.65
Cardiomyopathy	4th cause of burden among studied 228 incidents and 158 hospitalizations	7 cases
Myocarditis	5th cause of burden among studied 72 incidents and 64 hospitalizations	<5 cases of myocarditis 35 cases of all CLO-induced inflammations‡
Fever	NR	9.2% cumulative incidence and ARP = 0.63
All infections other than pneumonia	NR	Acute URI 17.6% cumulative incidence and ARP = 0.58 Other acute lower RI 10.4% cumulative incidence and ARP = 0.37 Intestinal infections 10.0% cumulative incidence and ARP = 0.34 Unspecified bacterial infections 8.5% cumulative incidence and ARP = 0.43
Seizures	NR	Antiepileptic drug use 46.3% cumulative incidence and ARP = 0.62 Epilepsy 8.1% cumulative incidence Drug-induced seizures 2.2% cumulative incidence and ARP = 0.88
Tachycardia	NR	Beta-blocker use 60.5% cumulative incidence and ARP = 0.50
Type 2 diabetes	NR	33.4% cumulative incidence and ARP = 0.15

*Pneumonia rehospitalization by diagnosis in adult Canadian patients per 1000 person-months, estimated using the Royston-Parmar model (vs other drugs) indicated that CLO's effects on pneumonia may vary per psychiatric diagnosis with the highest at 1.51 CI 0.83–3.03 vs 0.97, CI 0.80–1.18, in bipolar disorder, 1.37, CI 0.84–2.26 vs 1.01, CI 0.74–1.37, in schizoaffective disorder, and 1.26, CI 0.95–1.68, vs 0.83, CI 0.67–1.03, in schizophrenia.

†They described 408 patients with at least 1 case of pneumonia. It is unclear how many of these patients overlap with other patients including 419 with influenza or pneumonia, 166 with bacterial pneumonia, 26 with aspiration pneumonia, 290 with other respiratory infections, 201 with upper respiratory infections, 151 with other lower respiratory infections, 137 with bronchitis, 133 with bacterial infection, 131 with intestinal infections, 115 with pyelonephritis, and 69 with influenza.

‡They also described 13 cases of colitis, 10 of interstitial nephritis, 8 of pancreatitis and <5 of pericarditis. There were 35 cases of clozapine-induced inflammation, assuming no overlap among these cases.

aHR, adjusted hazard ratio; ARP, attributable risk proportion (the proportion of risk among clozapine users that can be directly attributed to CLO use); HR, hazard ratio; NR, not reported; OLA, olanzapine; RI, respiratory infections; URI, upper respiratory infections.

A BRIEF DESCRIPTION OF THE COMPLEXITY OF CHANGING THE US CLOZAPINE PACKAGE INSERT

The US clozapine package insert is not a scientific or a clinical document; it is essentially a very complex and long bureaucratic document. Its complexity is better understood when compared to the UK package insert. The number of words in the current version in November 2024 was 13,429 (vs 8862). The quantity of numbered sections was 17 (vs 8). The quantity of numbered subsections was 51 (vs 18). The quantity of tables was 16 (vs 4). Furthermore, the FDA uses a very cumbersome and bureaucratic process to make changes. A sentence modification requires modifying it from the prior version using the track changes system and a comment with a reference. For example, the word “myocarditis” appeared 23 times (vs 19) so to modify how myocarditis as described in the US package insert, up to 23 sentences need to be modified and depending on the section where “myocarditis” appears, different references on the topic of myocarditis may be needed. This type of detail and complexity required for modifying multiple sentences of the US clozapine package insert is beyond the limits of a scientific article. To make this article suitable for publication, we have developed very comprehensive Supplementary Material, http://links.lww.com/JCP/A949, to facilitate using this article in a potential negotiation with the FDA. One Supplementary Box, http://links.lww.com/JCP/A948, was developed for each major clozapine ADR to have a practical way to quote that Supplementary Box, http://links.lww.com/JCP/A948, with any proposed change to be considered by the FDA. In 2 cases (pneumonia and myocardial infarct) an additional Supplementary Box, http://links.lww.com/JCP/A948, had to be added to explain to skeptical readers whether the ADR was a clozapine ADR and/or was associated with the TRS present in most clozapine-treated patients.

THE DIFFERENCE BETWEEN EARLY VERSUS LATE CHANGES IN THE US CLOZAPINE PACKAGE INSERT

During the first 10 years of US marketing from 1989 to 1999, there was virtually no competition in the US for the marketer of clozapine, which had a large budget to further study clozapine ADRs if the company decided to do so. According to Meltzer,²² the company stopped marketing clozapine in 2003; therefore, it can be assumed that it lost interest in doing more studies and spending more money on them as 2003 was approaching. In summary, based on FDA practice, it is reasonable to propose that the first 10 years of accumulated US pharmacovigilance data should have been used by the FDA and the marketer to improve the US clozapine package insert to contribute to its safety in prescribing.

After a drug becomes generic there are no economic incentives to do further studies²³ and the companies producing generic drugs receive limited financial compensation for selling generic compounds and have no budget for more drug studies. Currently, there are 8 companies selling generic forms of clozapine in the US and managing the current hematological database system, which is controlled under a specific FDA regulation called risk evaluation and mitigation strategy (REMS). As far as we understand, each company contributes data for REMS, but only has access to their own pharmacovigilance data, and not the data from the other companies. The FDA has access to all the US data and sends it to the international database sponsored by the World Health Organization (WHO), called VigiBase.

This article utilizes clozapine reports sent by the FDA to VigiBase and a comprehensive review of the literature on clozapine-related ADRs grouped under the 5 existing boxed warnings compared to other unlisted ADRs. This is Part II of a comprehensive review article in which Part I focused on updates in basic pharmacology.⁶

METHODS

VigiBase Search

VigiBase is the WHO global pharmacovigilance database of reported potential ADRs of medicinal products. It receives data from all national drug or pharmacovigilance agencies including the FDA. The US VigiBase clozapine data described in this article are essentially the same as the FDA data, except that there can be a delay in the getting new FDA data into VigiBase.

This present study utilized VigiBase clozapine data for an observational, retrospective analysis conducted through January 15, 2023. Supplementary Box S2 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^15,24–30 provides details on the study design, statistics, confounders, underreporting, limitations and overlapping labels. The study follows the principles of the Helsinki Declaration. It is a retrospective review of deidentified patient data worldwide that does not require the signed consent of the individual patient, according to the ethics of the institutional review board of the university of the authors who accessed the VigiBase data.

Statistical Methods

A Bayesian confidence propagation neural network provides VigiBase with a statistical indicator called the information component (IC), which is interpreted using the confidence intervals (CIs). The IC is a sophisticated measure of disproportionality, indicating an increase or decrease in a specific ADR when compared to all drug ADRs in VigiBase (see details in Supplementary Box S2, Supplemental Digital Content 1, http://links.lww.com/JCP/A948).

Nonspecific Deaths

The Supplementary Figures (Supplemental Digital Content 1, http://links.lww.com/JCP/A948) describe all US deaths in clozapine-treated patients over the years. A methodological clarification is needed: the top category associated with death in clozapine-treated patients both worldwide and in the US is the label “death” (Supplementary Table S1, http://links.lww.com/JCP/A948). In the US, “death” was used to label 5918 fatal outcomes or 61.7% of 9587 total fatal outcomes. Supplementary Figure S2, http://links.lww.com/JCP/A948, displays the distribution of such cases over time. Supplementary Table S1, http://links.lww.com/JCP/A948, shows that of 5918 of these fatal outcomes, 3733 had no other information and 2815 had another label besides “death.” After excluding 3733 fatal outcomes with no label other than death from the total of 9587 fatal outcome reports, 5854 were designated as specific fatal outcomes in the US.

Literature Search

Box S3, Supplemental Digital Content 1, http://links.lww.com/JCP/A948, provides a detailed description of how the literature search led to the selection of 270 references.³¹

RESULTS

US Clozapine Reports to VigiBase

The US Is the Top Clozapine Reporter in the World in Absolute Numbers

The US is the top reporter of clozapine ADRs with 56,003 reports, which accounts for 29% of 191,557 worldwide reports.¹⁵ Supplementary Table S2, provides the US ICs for the major clozapine ADRs; 5 ICs were significantly lower than the global ICs (Supplemental Digital Content 1, http://links.lww.com/JCP/A948). On the other hand, the US significantly overreports agranulocytosis (IC = 6.63, IC₀₂₅ = 6.57) versus the worldwide average (IC = 4.58, IC₀₂₅ = 4.54). Only 18 cases of clozapine-associated agranulocytosis were expected in the US based on all the worldwide drug reports on agranulocytosis but 1855 were reported.

The US was the leading reporter of clozapine-associated fatal outcomes with 9587 or 43% of the 22,596 worldwide total (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). The US had lower percentage rates of 20 fatal outcomes associated with other clozapine ADRs when compared with global averages, indicating the possible underreporting of clozapine ADRs among US ADRs (Supplementary Table S3, http://links.lww.com/JCP/A948). On the other hand, 4 fatal outcomes (death, myocardial infarct, completed suicide and neutrophilia) were higher than the US average, indicating relative overreporting.

US Reporting Style: Possible Greater Underreport Than the Other 3 Top Reporting Countries

Because of variations in population characteristics, clinical environments, reporting practices, and regulatory frameworks, it is anticipated that different countries will exhibit diverse patterns of clozapine ADR reporting. When compared with the 3 other top reporting countries (Australia, Canada, and the UK), the data needs to be corrected for clozapine use, population, and propensity to report ADRs in general. According to an international comparison,^32,33 the US has relatively low clozapine use and we estimated that the US prescribed 79% of what is prescribed per 100,000 people in the UK¹⁵ with even lower percentages of use when compared with other Western European countries. For example, the US may prescribe only around 29% as much clozapine as Finland, the top clozapine prescriber worldwide.³²

Supplementary Table S4 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^15,32,33 provides 3 parameters as comparisons with the UK, Australia, and Canada.¹⁵ Clozapine ADRs account for only 0.4% of all US drug reports to VigiBase versus 1.9%–3.5% in the other 3 countries. These values are not influenced by population size and suggest that when compared with these 3 other countries, the US may be underusing and/or relatively underreporting clozapine ADRs by a factor of 4.8–8.8 (red font numbers in Supplementary Table S4, http://links.lww.com/JCP/A948). If it is assumed that the UK has the correct number of clozapine fatal outcomes with 100%, the US only provided 37% of them (after adjusting for population size and clozapine use), whereas Canada (106%) and Australia (86%) are closer to the UK.

VigiBase US Fatal Outcomes and Boxed Warnings in the US Clozapine Package Insert

Severe Neutropenia Boxed Warning and Its 218 US Fatal Outcomes

Since clozapine's approval in 1989, the FDA boxed warning has been used to justify hematological monitoring and is currently used to justify restricted use of clozapine under REMS regulation. The VigiBase data indicates that worldwide hematological monitoring has been enormously successful in limiting the risk of clozapine-associated agranulocytosis. Between 1989 and 2019, there were 34,931 reports worldwide of all probable cases of clozapine-associated severe neutropenia, but these were associated with 550 fatal outcomes, indicating a low lethality rate of 1.6%.³⁴

In the US, “agranulocytosis” has been associated with only 64 fatal outcomes from among 1855 clozapine reports (3.5% lethality). Figure 1 describes all potential VigiBase labels that may represent fatal outcomes associated with severe neutropenia, which total 218 fatal outcomes (2.3% of 9587). That means 97.7% of clozapine fatal outcomes are not explained by severe neutropenia, according to US pharmacovigilance data.

FIGURE 1.

Clozapine-associated fatal outcomes including a broad definition of severe neutropenia. There were 64 fatal outcomes associated with agranulocytosis, 100 with neutropenia, 49 with leukopenia and 36 with granulocytopenia. The sum of these reports with fatal outcomes is 249 (64 + 100 + 49 + 36 = 249). There were only 218 individual fatal outcomes; this means that 31 (249–218 = 31) reports were eliminated as some fatal outcomes had more than 1 label.

Boxed Warning for Seizure and Its 138 US Fatal Outcomes

Supplementary Box S4 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^35–50 provides a brief review of the literature of clozapine-associated epileptic seizures. Seizures accounted for only 131 fatal outcomes from among 1752 seizure reports (7.5% relative lethality). Supplementary Figure S3, http://links.lww.com/JCP/A948, combines “seizures” and “generalized tonic-clonic seizures,” accounting for 138 fatal outcomes (7.7% relative lethality) and 1.8% of 9587 US fatal outcomes.

Orthostatic Hypotension Boxed Warning and 125 US Fatal Outcomes

A third boxed warning was added in 1992 for orthostatic hypotension with or without syncope with the rare possibility of collapse from respiratory and cardiac arrest during clozapine initiation, particularly in the presence of benzodiazepines. Later on, the word “bradycardia” was added. Supplementary Box S5 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^51–57 provides a brief review of the literature of this topic. “Circulatory collapse” led to 60 fatal outcomes. Supplementary Figure S4, http://links.lww.com/JCP/A948, combined “circulatory collapse,” “syncope,” “hypotension,” and “orthostatic hypotension,” with a total of 125 fatal outcomes (1.3% of 9587).

Myocarditis Boxed Warning and 36 US Fatal Outcomes

A fourth boxed warning was added in 2002 for myocarditis. Supplementary Figure S5, http://links.lww.com/JCP/A948, shows that myocarditis led to only 36 fatal outcomes among 311 reports (11.6% relative lethality). Supplementary Box S6, http://links.lww.com/JCP/A948,^{13,14,21,44,58–87} reviews the concept of clozapine-associated myocarditis in the context of clozapine-associated inflammation. Slower and personalized titration may decrease risk for clozapine-associated myocarditis. Denmark and the Netherlands have very low rates of clozapine-associated myocarditis and use very slow outpatient titrations in a large percentage of their clozapine-treated patients.⁶⁰ Very slow outpatient treatments in the US have been proposed by the EASE model.⁸⁸

In 2013, the US clozapine package insert's boxed warning for myocarditis incorporated cardiomyopathy. Clozapine-associated myocarditis almost always occurs during titration⁵⁹; clozapine-associated cardiomyopathy remains a poorly defined category.^89–92 There were 24 other US fatal outcomes associated with cardiomyopathy.

In 2016, mitral valve incompetency was added to the myocarditis boxed warning. However, this addition is perplexing because 1) the British, Canadian and Australian package inserts do not include it in their boxed warnings, 2) we could find no fatal outcomes among US clozapine-treated patients associated with “mitral valve incompetence” or “mitral valve disease,” and 3) a PubMed search on October 30, 2023, based on “clozapine and mitral valve” as search terms provided 1 case report from the Netherlands in a patient taking clozapine with lithium.⁹³

Boxed Warning for the Elderly With Dementia Who Use Clozapine

In 2005, the FDA reported an association between treatment with second-generation antipsychotics and deaths in patients with dementia.⁹⁴ In the clozapine package insert, the FDA noted that clozapine “is not approved for this condition.” Supplementary Box S7 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^94–100 provides a brief review of the content in the US package insert, which does not adequately represent findings in the research literature on this complex issue, including 2 recent articles exploring the risks and benefits of clozapine use in this population.

US Fatal Outcomes and ADRs Not Listed in US Boxed Warnings

This section describes clozapine-associated ADRs which are among the leading causes of fatal outcomes but do not have a corresponding boxed warning in the US package insert: 1) cardiac arrest, 2) sudden death, 3) pneumonia, 4) other infections, 5) myocardial infarct, 6) lung cancer, 7) pulmonary embolism, 8) suicide, and 9) CIGH.

Cardiac Arrest Is Associated With 415 US Fatal Outcomes

The fourth leading category associated with deaths in clozapine-treated patients around the world and in the US is labeled “cardiac arrest” (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). In the US, “cardiac arrest” explained 415 fatal outcomes (4.3% of 9587). Supplementary Figure S6, http://links.lww.com/JCP/A948, displays the temporal distribution throughout the years.

Sudden Death Is Associated With 110 US Fatal Outcomes

Worldwide, “sudden death” was the sixth leading label for clozapine-associated fatal outcomes but was much less prevalent in the US (Supplementary Figure S7, Supplemental Digital Content 1, http://links.lww.com/JCP/A948) with only 110 fatal outcomes (1.1% of 9587). Supplementary Box S8, http://links.lww.com/JCP/A948,^21,101–115 describes the limited available published information on the risk of sudden death secondary to arrythmia in clozapine-treated patients.

Pneumonia Is Associated With 674 US Fatal Outcomes

The second most prevalent category associated with deaths in clozapine-treated patients worldwide and third in the US was “pneumonia” (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). In the US, pneumonia accounted for 559 fatal outcomes from among 1438 pneumonia reports in clozapine-treated patients (Supplementary Figure S8, http://links.lww.com/JCP/A948) with a relative lethality of 39%. Moreover, Supplementary Table S1, http://links.lww.com/JCP/A948, explains that this label was first among specific causes after excluding nonspecific fatal outcomes. Fatal outcomes associated with pneumonia explained 9.5% of 5854 specific fatal outcomes among US clozapine-treated patients. Aspiration pneumonia led to 115 US fatal outcomes among 823 reports (Supplementary Figure S9, http://links.lww.com/JCP/A948) with a relative lethality of 14%. All the US fatal outcomes associated with respiratory aspiration were also included among the fatal outcomes of aspiration pneumonia so no additional data are presented. Figure 2 shows that broadly defined pneumonia, which combines pneumonia and aspiration pneumonia, represented 674 fatal outcomes; 7.0% of 9587 US clozapine fatal outcomes and 11.5% of 5854 US specific clozapine fatal outcomes.

FIGURE 2.

Clozapine-associated fatal outcomes including a broad definition of pneumonia. There were 558 fatal outcomes associated with “pneumonia” and 115 with “aspiration pneumonia.” The sum of these reports with fatal outcomes is 796 (558 + 115 = 796). There were only 674 individual fatal outcomes; this means that 122 (796–674 = 122) reports were eliminated as some fatal outcomes had more than 1 label.

Supplementary Box S9 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^{15,21,116–126} presents an updated review of the literature on the fatal outcomes associated with pneumonia in clozapine-treated patients. In 2659 Finnish patients followed on average for 12 years, 408 patients developed pneumonia at least once; the mortality rate was 6%.²⁰ In VigiBase reports with complete data, the average worldwide mortality rate was 45% in 2415 patients but ranged from 0% to 75% with greater mortality for male sex, longer duration of treatment and older age.¹²⁴ In the 618 US cases with complete data, lethality was 61% with an OR of 2.3 (CI 1.93–2.8), indicating a much higher risk of lethality in US pneumonia cases than in those of other reporting countries. When adjusted for other variables, the US OR increased from 2.3 to 2.9 (CI 2.2–3.6). The US tends to report more severe clozapine ADRs across all ADR categories,⁵ so currently it is not known whether or not this tendency contributes to this increased US lethality in pneumonia of clozapine-treated patients (Supplementary Box S9, http://links.lww.com/JCP/A948).

Supplementary Box S10 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^{19,87,96–98,124,127–130} notes that the contribution of clozapine versus TRS to pneumonia is not currently very well understood and may vary from aspiration pneumonia compared to infectious pneumonia with no signs of aspiration. It is reasonable to consider aspiration pneumonia an ADR associated with clozapine and other prescribed co-medications that may interfere with swallowing. To explore aspiration pneumonia in VigiBase clozapine-treated patients, a new category for all cases of pneumonia with signs of aspiration was developed. This new category included those labeled as aspiration pneumonia, or any pneumonia case with 1) aspiration, 2) salivary hypersecretion, 3) dysphagia, 4) vomiting, or 5) choking. Using this definition, 30% of the pneumonia cases in clozapine-treated patients had signs of aspiration, thus being a potential ADR due to clozapine and co-medications.¹²⁴

The antipsychotic studies of the national registries do not distinguish aspiration pneumonia and several antipsychotics have been associated with pneumonia in the Danish^131,132 and the Finnish^21,125 registries. A mirror-image study in the Danish registry found the annual pneumonia incidence was 1.23% for TRS in the first year of clozapine treatment, and 0.76% in patients with schizophrenia before taking any antipsychotic.^131,132 This design allowed estimations that TRS contributed to 70% (0.45/0.64) of the pneumonia risk while clozapine contributed to 30% (0.419/0.64) of the pneumonia risk.¹³² It is unclear how these estimations can be extrapolated to clozapine treatment all over the world, as the Finnish data suggests that pneumonia risk may increase with clozapine dose, duration of clozapine treatment and age.^21,125 Thus, in each country variations in these contributing variables may lead to different incidences of pneumonia in clozapine-treated patients. The most recent Finnish study²¹ estimated the risk of pneumonia in clozapine patients at years 1, 5 and 20 were 3.0%, 7.9% and 29.5% versus 0.7%, 2.9%, and 10.8% in olanzapine patients, but this study could not control for the effects of TRS, which is more likely to be present in clozapine-treated patients than in olanzapine-treated patients. The exposure impact number for clozapine was 30.7 (95% CI 17.2 to 138.5), this means that almost 31 patients need to be exposed to clozapine to develop 1 new case due to clozapine.²¹

In a meta-analysis, pneumonia was the major cause of natural mortality in patients with schizophrenia.¹³³ In an umbrella review of risk of life-threatening medical events associated with antipsychotics, pneumonia provided the most robust results.¹³⁴ Thus, many clozapine-treated patients accumulate pneumonia risk if they suffer from the worst forms of schizophrenia, have taken antipsychotics for many years and those risks are associated with clozapine. At an individual level, when a clozapine patient has survived an episode of pneumonia, it is important to consider whether or not it was an episode of aspiration pneumonia, likely to be secondary to clozapine and other medications. If aspiration pneumonia is present or suspected, clozapine dose reductions and/or discontinuation of other medications that may interfere with swallowing should be considered to avoid recurrences.¹²⁴ In 2659 Finnish schizophrenia clozapine-treated patients studied for up to 20 years, 408 of them developed pneumonia; 20% of these 408 patients had pneumonia twice and 14% had it 3 or more times.²¹

Increased White Blood Cell Count (a Possible Sign of Infection) Was Associated With 596 US Fatal Outcomes

Supplementary Figure S10 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948) shows that the combination of all forms of increased white blood cell (WBC) count were associated with 596 fatal outcomes (6.2% of 9587), which is more than the 218 fatal outcomes (2.2% of 9587) associated with all forms of severe neutropenia. These labels indicate that patients had a high WBC count at the time of death. In a clozapine-treated patient dying with leukocytosis 2 major diagnoses need to be considered: 1) leukemia and 2) infections. In 2659 Finnish schizophrenia clozapine-treated patients studied for up to 20 years, less than 5 developed leukemia while infections were very frequent. Besides the 408 patients with pneumonia, other infections included 290 other respiratory infections, 201 upper respiratory infections, 151 other lower respiratory infections, 137 bronchitis infections, 133 bacterial infections, 131 intestinal infections, 115 with pyelonephritis, and 69 with influenza.²¹ The number of overlapping infections in the same patients was not described. This Finnish data affirms that almost all VigiBase fatal outcomes in the context of leukocytosis should be associated with infections and, in rare instances, leukemia may explain some cases.

In VigiBase infections are reported with hundreds of labels. Recently, Chrétien et al¹³⁵ comprehensively studied infections in clozapine-treated patients in the worldwide data from VigiBase and found that infections were distributed across 196 labels. The number of US fatal outcomes associated with infections is hard to estimate due to these multiple labels. Most of the infection labels differ from pneumonia, as described by Chrétien et al,¹³⁵ and did not reach the top 20 US clozapine ADRs. Supplementary Table S3 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948) lists only 2, sepsis and lower respiratory tract infections, respectively, with 161 and 5 US fatal outcomes.

Among clozapine-treated patients, it is not currently known how much clozapine versus TRS contributes to the increased risk of infections. Studies from the Danish registry have associated increased risk and severity of infections with severe mental illnesses¹³⁶ and antipsychotics.¹³⁷ A recent meta-analysis associated severe mental illnesses and increased risk of infections.¹³⁸

Myocardial Infarct Was Associated With 574 US Fatal Outcomes in Clozapine-Treated Patients

Reviews of clozapine ADRs^139–143 do not include myocardial infarct. Supplementary Table S2, Supplemental Digital Content 1, http://links.lww.com/JCP/A948, indicates that the US IC for myocardial infarct is 0.45, which is much lower than other definitive clozapine ADRs and does not indicate that clozapine is disproportionally associated with myocardial infarcts in the US.

Myocardial infarct was the second highest category associated with deaths in clozapine-treated patients in the US, but was third worldwide (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). In the US, “myocardial infarct” (Supplementary Figure S11, http://links.lww.com/JCP/A948) explained 574 fatal outcomes (6.0% of 9587).

Supplementary Box S11 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^144–155 explains that the data from US studies and the Denmark registry does not associate clozapine with increased cardiovascular deaths in schizophrenia, whereas the data from the Finnish registry indicates that clozapine may be protective when compared to patients with schizophrenia who are not taking antipsychotics. However, it is not known whether the Finnish data applies to the US, which has a much different health system and much lower clozapine use.

Supplementary Box S12 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^{21,134,156–169} deals with an even more complicated topic: whether a myocardial infarction in a clozapine-treated patient can be considered a clozapine ADR or not. In a short-term RCT, clozapine appeared to be the worst of the second-generation antipsychotics regarding metabolic ADRs. However, a clozapine-treated patient suffering from a myocardial infarct has other risk factors besides clozapine and they may include schizophrenia and a history of long-term antipsychotic use. In conclusion, current evidence suggests that TRS may be a major contributor to MI in clozapine-treated patients. TRS is associated with other nonspecific risk factors for MI such as obesity, metabolic ADRs from all prior treatments, and smoking.

Lung Cancer Was Associated With 182 US Fatal Outcomes in Clozapine-Treated Patients

The ninth highest category associated with deaths in clozapine-treated patients in the US and the world was “malignant lung neoplasm.” In the US (Supplementary Figure S12, Supplemental Digital Content 1, http://links.lww.com/JCP/A948), “malignant lung neoplasm” explained 182 fatal outcomes (2% of 9587).

Schizophrenia is strongly associated with increased and heavy tobacco use.^157–159 Moreover, the most severe cases of schizophrenia, overrepresented in TRS, may be even more strongly associated with heavy smoking.¹⁵⁸ Thus, the fatal outcomes due to lung cancer in patients taking clozapine may well be explained because TRS is associated with smoking, a major cause of lung cancer. In a comprehensive study that favors this interpretation, Shi et al¹⁷⁰ found no significant association between schizophrenia and lung cancer in nonsmokers.

Pulmonary Embolism Was Associated With 157 US Fatal Outcomes in Clozapine-Treated Patients

In a Swedish pharmacovigilance study, Hägg et al¹⁷¹ first described the association of clozapine with venous thromboembolism. Pulmonary embolism was the 10th leading category associated with deaths in clozapine-treated patients (Supplementary Table S1, http://links.lww.com/JCP/A948). In the US, “pulmonary embolism” explained 155 fatal outcomes. Supplementary Figure S13, http://links.lww.com/JCP/A948, presents the broadly defined pulmonary embolism with 157 fatal outcomes (1.6% of 9587).

The association between clozapine and fatal outcome with pulmonary embolism is probably complex, as described in Supplementary Box S13 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^21,172–179 and reflects the following: 1) effects of TRS and the frequent co-medications present in clozapine -treated patients; 2) the association of clozapine with nonspecific factors associated with venous thromboembolism, such as obesity and other combined clozapine ADRs that may increase pulmonary embolism risk; and 3) possibly some specific clozapine effects increasing coagulation factors.

Suicide Was Associated With 250 US Fatal Outcomes in Clozapine-Treated Patients

The sixth leading category associated with deaths in US clozapine-treated patients and the fifth in the world was “completed suicide.” In the US (Supplementary Figure S14, Supplemental Digital Content 1, http://links.lww.com/JCP/A948), “completed suicide” accounted for 250 fatal outcomes (2.6% of 9587), but many of these were in the context of polypharmacy.

Schizophrenia is associated with mortality as a result of suicide.¹³³ Supplementary Box S14 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^{113,145,149–151,180–205} reviewed the literature of antisuicidal effects of clozapine including: 1) an RCT leading to the antisuicide indication by the FDA, 2) other clozapine studies, 3) studies indicating clozapine's antisuicidal effects may be specific and not present in other antipsychotics, 4) the possible mechanism of the antisuicidal effect of clozapine, and 5) the need for further studies in children and adolescents.

CIGH Was Associated With 97 US Fatal Outcomes in Clozapine-Treated Patients

Supplementary Figure S15 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948) illustrates that the combination of all forms of CIGH was associated with 97 fatal outcomes (1.0% of 9587). Supplementary Box S15, http://links.lww.com/JCP/A948,^{15,21,34,206–213} presents a brief summary of CIGH including information on the following: 1) the label “constipation” in VigiBase, which does not reflect CIGH well, 2) the fatal outcomes in pharmacovigilance studies, 3) the fatal outcomes in the recent long-term Finnish study, and 4) systematic reviews.

DISCUSSION

Limited Alignment Between Boxed Warnings and US Fatal Outcomes

Supplementary Box S1 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948) provides the criteria which the FDA uses to justify a boxed warning for a serious ADR: 1) risk-benefit is impacted, 2) can be prevented, and/or 3) restricted use is needed. Table 1 discusses how the 5 boxed warnings can be classified based on the details provided in the clozapine package insert.

The Scandinavian^{145,149–151,181–183} and Taiwanese registries¹⁸⁴ indicate that clozapine may decrease TRS mortality. Extrapolating this finding to the US with very low clozapine use (29% of that in Finland) may suggest that the lives of some patients with TRS are lost because of clozapine underuse in the US.

The next 2 subsections critically assess the imbalance between boxed warnings and fatal outcomes in US clozapine-treated patients during the first 10 years, when clozapine pharmacovigilance was definitively controlled by the marketer and the FDA.

The First 10 Years of Clozapine Use in the US and Its Fatal Outcomes in VigiBase

It is not known what the FDA and the marketer did to monitor clozapine ADRs between 1989 and 1999, nor of their view regarding clozapine-associated fatal outcomes in the US. Moore et al,²¹⁴ a research group interested in ADRs, reviewed what the FDA considered to be serious ADRs between 1998 and 2005. Clozapine was considered the third most lethal drug with 3277 deaths after oxycodone with 5548 deaths and fentanyl with 3455 deaths. However, Moore et al²¹⁴ provided no data on the causes of death in clozapine-treated patients.

Table 1 shows that, until 1999, all forms of agranulocytosis accounted for only 5% of the 929 deaths in clozapine-treated patients. The number of deaths associated with agranulocytosis was only 48, somewhat higher than the 27 associated with seizures or the 46 associated with all forms of orthostasis. It is not known why the 112 fatal outcomes associated with pneumonia received no attention. Similarly, deaths in patients with high values of WBC were 46, close to that with agranulocytosis. These patients probably had other infections than pneumonia; combining them with pneumonia there were 158 fatal outcomes if no overlap is assumed.

US Fatal Outcomes in Clozapine-Treated Patients Until 2023 in VigiBase

Table 1 shows the fatal outcomes associated with 5 boxed warnings until 2023, which range in each box from 0 to 218. The total number of US fatal outcome deaths associated with all 5 boxed warnings was 534 if there is not any overlap among categories (218 severe neutropenia, 131 seizure, 125 orthostasis and 60 broadly defined myocarditis). On the other hand, pneumonia by itself had 674 deaths, which was more than the combination of the 4 boxed warnings specific for clozapine. Increase of WBC, a sign of infection, has not had a boxed warning but was associated with 596 fatalities, which was more than all 4 boxed warnings combined specific for clozapine.

In 2023, by adding only the broad definition of pneumonia (674 fatal outcomes) and of high WBC values (596), and after excluding overlaps, there were 900 fatal outcomes (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). Many other fatal outcomes associated with infections (probably divided among as many as 196 labels¹³⁵) are not included in the 900 fatal outcomes. Even with this underestimation, the US pharmacovigilance data suggests that, for preventing future deaths in US clozapine-treated patients, the US clozapine package insert should address the prevention of fatal outcomes during infections. Thus, it seems plausible that the introduction in 1999 of a boxed warning for risk of fatalities during infections might have saved hundreds of lives of US clozapine-treated patients.

Problems in the Current US Clozapine Package Insert Regarding Boxed Warnings

Box 1^{5,6,67,77,81–87,215–220} summarizes the major problems with the current US clozapine package insert. Compared to the UK package insert, it took the FDA 10 additional years to develop a benign ethnic neutropenia (BEN) pathway (2005 vs 2015).²¹⁵ Moreover, in 2020 on its webpage the UK drug agency recommended clozapine therapeutic drug monitoring (TDM) including during infections, but TDM has not been recommended by the FDA.^216–218 In retrospect, the FDA focus in 1989 on agranulocytosis appears reasonable, but it is unclear why in 2025 the focus continues to be on agranulocytosis when 97.7% of fatal outcomes in the US pharmacovigilance database had some other cause. The lack of attention paid to the 7.0% of fatal outcomes associated with pneumonia and another 6.2% possibly associated with other infections appears to be a major oversight. The recently added paragraph (see Supplementary Box S10, Supplemental Digital Content 1, http://links.lww.com/JCP/A948) on pneumonia to the clozapine package insert in May 2023¹⁹ was not based on the US pharmacovigilance data but quoted published case reports. On the other hand, it did not use published articles to provide instructions to clinicians on how to prevent or deal with pneumonia in clozapine-treated patients, as proposed in Box 2.

BOX 1.

Major problems with the US CLO package insert regarding fatal outcomes.

1. Compared with the UK's CLO package insert: delayed additions

• The UK introduced a BEN pathway in 2005. The FDA took another 10 years, but provided a more reasonable BEN pathway.215 The FDA dropped the WBC requirements in favor of ANC-only monitoring and lowered the threshold of the ANC for treatment interruption from 1500/mm3 to 1000/mm.3 • Since 2004, a German expert group216 has strongly recommended CLO TDM and provided a therapeutic reference range recently endorsed by US experts.217 The UK drug agency took 16 years to recommend CLO TDM on its webpage including during infections. Unfortunately, the FDA has not recommended it yet despite approving a point-of-care device for CLO TDM.218

2. It is unclear why severe neutropenia continues to be the FDA's focus among CLO ADRs

• Agranulocytosis is overreported in the US, but all severe neutropenia labels explained 249 deaths or 2.2% of 9587 US VigiBase fatal outcomes. In Finland, CLO ↓ natural/suicide deaths in TRS. If this can be extrapolated to the US, the restricted use of CLO in the US (<1/3 of Finland's use) can be associated with more deaths than with the agranulocytosis associated with CLO use. • In 1990, hematological monitoring helped introduce CLO in the US. Now, most experts agree (Supplementary Box S16, http://links.lww.com/JCP/A948) on the need for weekly monitoring only for 18 weeks, recommended in 1977 by the CLO marketer after the Finnish deaths.219 The REMS system should be eliminated.

3. The lack of attention to fatal outcomes during pneumonia and infections in CLO patients

• In 1978, it was recognized that infections led to theophylline intoxications (a CYP1A2 substrate). The US theophylline package insert reflects that, but the CLO (another CYP1A2 substrate) package insert does not.5,6 • Based on US pharmacovigilance data, broad pneumonia is the leading specific CLO-associated cause of death (674, or 7.0% of 9587 US fatal outcomes) while other infections may contribute to CLO fatalities (596, or 6.2% of 9587 US fatal outcomes). • To estimate all CLO fatal outcomes during infections requires adding 196 VigiBase labels. • It is unclear why, after 10 years of monitoring, the FDA and CLO's marketer did not compare the 48 fatal outcomes associated with severe neutropenia with the 112 associated with pneumonia and the 46 with increases in WBC count, presumably associated with infections. • Until 2023, there were 218 US fatal outcomes associated with severe neutropenia versus 674 with pneumonia and 596 with increases in WBC count. In 2023, there were at least 900 nonoverlapping US fatal outcomes associated with these 2 categories (Supplementary Table S1, http://links.lww.com/JCP/A948). Many more nonoverlapping fatal outcomes associated with infections in US clozapine-treated patients are expected if all 196 VigBase labels for infections are combined to provide an estimation.

4. Slower personalized titration may be needed to limit the risk of CLO-associated inflammations

• CRP monitoring during titration was first promoted to prevent myocarditis in Australia. Other forms of CLO-induced inflammation can ↑ CRP and are associated with rapid titrations.67 A systematic review endorsed CRP monitoring for 4–8 weeks during CLO titration.81 • A 2022 international titration guideline recommended baseline and weekly CRP monitoring for at least 4 weeks and slow personalized titration considering: 1) ancestry group, 2) sex-smoking subgroup, and 3) presence or absence of CLO PM status.77 • Slower personalized titrations are being endorsed by recent studies from Australia,85 Korea,86,87 Japan,82–84 and the US.220

5. There is no easy mechanism for funding US prospective CLO titration studies

• Nevertheless, CRP and personized titration are simple inexpensive options that any CLO prescriber can implement without waiting for changes in the US package insert.

ADR, adverse drug reaction; ANC, absolute neutrophil count; BEN, benign ethnic neutropenia; CLO, clozapine; CRP, C-reactive protein; CYP1A2, cytochrome P450 1A2; FDA, Food and Drug Administration; PM, poor metabolizer; REMS, Risk Evaluation and Mitigation Strategy; TDM, therapeutic drug monitoring; TRS, treatment-resistant schizophrenia; UK, United Kingdom; US, United States; WBC, white blood cell.

BOX 2.

Proposals for improving the boxed warnings in the US CLO package insert.

1. Proposal for the boxed warning regarding severe neutropenia

• Severe neutropenia accounts for only around 2% (218/8578) of US deaths; thus, hematological monitoring has been successful in preventing the high number of deaths expected by the FDA that led to the cessation of US clozapine studies in the 1970s. • Based on the risk after 18 weeks being no greater than with other antipsychotics, the Sandoz researchers, Amsler et al219 recommended monitoring weekly WBC only for 18 weeks of CLO treatment. • Supplementary Box S16, http://links.lww.com/JCP/A948, describes recent recommendations by 6 groups of CLO experts from throughout the world. 126,221–230 There are disagreements about the need for and duration of monthly monitoring after 18 weeks, but there is general agreement among experts that it may be time to limit the weekly hematological monitoring to 18 weeks. • The REMS database needs to be eliminated. Countries in continental Europe do not have more fatal outcomes with CLO-induced agranulocytosis and do not use centralized databases.222

2. Proposal for the boxed warning regarding seizures

• To avoid CLO-induced seizures, Table 1 recommends paying attention to dosing and DDI and, more importantly, to TDM because it provides the best way to personalize dosing. • It is concerning that the US CLO package insert does not mention clozapine TDM when the UK drug agency recommended it in August of 2020.130

3. Proposal for the boxed warning regarding orthostasis

• Benzodiazepines should not be co-prescribed during titration to avoid circulatory collapse. • The risk of orthostatic hypotension can best be prevented by measuring orthostatic changes after first doses and by using slow, personalized titrations based on: 1) ancestry group; 2) sex-smoking subgroups, and 3) absence/presence of clozapine PM status.77

4. Proposal for the boxed warning regarding myocarditis

• The warning on mitral valve incompetence should be eliminated. There are no fatal outcomes associated with mitral valve incompetence in CLO-treated patients in the literature or VigiBase. • Myocarditis may best be prevented by CRP monitoring and by using slow, personalized titrations based on consideration of ancestry, sex-smoking subgroup, and PM status. • Cardiomyopathy, a poorly understood ADR, should not be grouped with myocarditis.

5. Proposal for the boxed warning regarding the elderly with dementia

• The package insert needs to acknowledge that respiratory aspiration can occur in CLO-treated nongeriatric patients and in the absence of overdose,98 this risk may be better addressed in the context of the continuum of aspiration – aspiration pneumonia.

6. Lack of boxed warnings for pneumonia and other infections

• If the FDA assumed that a boxed warning was needed for 5 CLO ADRs, it is not clear why pneumonia, the most significant cause of death in US CLO-treated patients, was ignored. • The UK drug agency in August 2020 at least indicated the need for TDM during infections.130 • The recent FDA change of the US CLO package insert in April of 2023 was insufficient and confusing for clinicians (see Supplementary Box S10, http://links.lww.com/JCP/A948). • Until better information is available, 4 recommendations are important in minimizing the risk of pneumonia and other infections: ○ using slow and personalized titration schedules, ○ using the lowest therapeutic CLO dose possible in each patient, and ○ educating patients/family to contact the psychiatrist as soon as severe infection develops ○ and educating primary care providers in the management of infections in CLO-treated patients.

ADR, adverse drug reaction; CLO, clozapine; CRP, C-reactive protein; FDA, Food and Drug Administration; PM, poor metabolizer; REMS, risk evaluation and mitigation strategy; TDM, therapeutic drug monitoring; UK, United Kingdom; US, United States; WBC, white blood cell.

Using C-reactive protein (CRP) monitoring during titration^77,81 and using personalized titrations⁷⁷ that are much slower than that recommended by the US package insert may be a simple, inexpensive option for limiting the risk of clozapine-associated inflammation and pneumonia. Recent studies, ^81–87,220 though limited methodologically, support the wisdom of that approach. Ideally, prospective titration studies are required, but there is no easy mechanism to pay for them now that clozapine is a generic drug marketed in the US by 8 manufacturers with limited resources.

Box 2^{77,98,126,219,221–231} proposes detailed improvements to the US package insert by reviewing the boxed warnings for 1) severe neutropenia, 2) seizures, 3) orthostasis, 4) myocarditis, and 5) the elderly with dementia, and considering the lack of a boxed warning for pneumonia and other infections.

In 1977, Sandoz owned clozapine and its experts²¹⁹ proposed limiting the WBC monitoring to the first 18 weeks of clozapine treatment. Supplementary Box S16 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^{126,221–231} provides an updated review of the literature from 6 groups of clozapine experts from around the world proposing simplifications in WBC monitoring during clozapine treatment.

The data from the Scandinavian^{145,149–151,181–183} and Taiwanese¹⁸⁴ registries indicates that clozapine saves the lives of patients with TRS in those countries. The US prescription rates of clozapine are much smaller than in those countries that use clozapine safely with no centralized database such as the REMS system. The REMS needs to be eliminated to eliminate burdens caused to patients, prescribers and pharmacies. Worldwide experts agree that weekly monitoring should be limited to the first 18 weeks (Supplementary Box S16, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). Long-term follow-up data from Finland²¹ and, more importantly, from Canada,²⁰ which is more like the US, currently indicates that focusing on hematological monitoring is wrong as the main burdens and fatal outcomes are not associated with severe neutropenia, but first with pneumonia and secondly with CIGH.

Relevance of Changes in the Package Insert Beyond US Patients With TRS

Part I⁶ and Part II of this report propose major changes in the US clozapine package insert aimed at increasing safety for patients taking clozapine in the US and elsewhere. We also hope to contribute to increasing patient access to clozapine as a uniquely effective antipsychotic agent.

Relevance for Other US Patients

The FDA is responsible for deciding on approved indications for treatment, although US prescribers, based on their clinical judgment, can use clozapine for nonapproved indications. US clozapine experts co-authoring this article as well as other literature reports support the use of clozapine as a safe and effective drug for some of the most severely ill psychiatric patients who do not respond to any other pharmacological or nonpharmacological intervention. Thus, it is possible that clozapine can save lives beyond those with TRS or with suicide risk in schizophrenia. Supplementary Box S17 (Supplemental Digital Content 1, http://links.lww.com/JCP/A948)^{57,192–194,197,232–254} summarizes the available literature on off-label use of clozapine. They include the following: 1) aggression in patients with schizophrenia psychoses, 2) substance use disorders in patients with schizophrenia psychoses, 3) catatonia, 4) treatment-resistant bipolar disorder, 5) psychotic depression, 6) delusional disorder, 7) borderline personality disorder at high risk of suicide, 8) polydipsia associated with severe mental illness, 9) severe hetero- or auto-aggressive behavior in patients with intellectual disability, 10) psychosis in patients with Parkinson's disease, and 11) children and adolescents with schizophrenia.

Relevance for Non-US Patients

In many areas of the world, clozapine pharmacovigilance is severely underdeveloped including countries in Continental Western Europe,¹⁶ Eastern Europe,²⁵⁵ Latin America,²⁵⁶ Asia, and Africa.²⁵⁷ Clozapine is used relatively frequently in China,²⁵⁸ but only 6050 reports had been sent to VigiBase from China with no fatal outcomes associated with the 9 leading clozapine ADRs as of January 15, 2023.¹⁵ Clozapine is also frequently used in Russia,²⁵⁹ but with no clozapine reports yet sent to VigiBase.²⁵⁵ Clozapine use in Middle Eastern countries is low and variable,²⁶⁰ with fewer than 1000 clozapine reports with 3 fatal outcomes in VigiBase by January 15, 2023. Use of clozapine in Sub-Saharan Africa is very limited, as of January 15, 2023¹⁵; we found only 190 reports with 5 fatal outcomes from the country of South Africa.

Other regulatory agencies tend to follow FDA decisions.²⁶¹ Moreover, the FDA led the resurrection of clozapine use by requiring an RCT in TRS prior to its marketing in the US.^10,11 We now propose major revisions in the US clozapine package insert to increase safety in the US and to encourage similar changes in clozapine package inserts worldwide. Such improvements have the potential to save the lives of thousands of clozapine-treated patients. In VigiBase identified cases of pneumonia alone account for more than 2000 fatal outcomes, approximately 9% of the fatal outcomes associated with clozapine in VigiBase (footnote c in Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). These deaths are the tip of the iceberg as many more deaths of clozapine-treated patients associated with pneumonia have not been reported to or recognized as such in VigiBase; some may be reported under the unspecific label, “death.” By improving the safety of clozapine and simplifying WBC monitoring without compromising safety, access to potentially life-saving clozapine treatment is a likely and welcome result.

Description of the FDA Clozapine Meeting on November 19, 2024

The Meeting

On November 19, 2024, the FDA called a clozapine meeting of the Drug Safety and Risk Management Advisory Committee and the Psychopharmacologic Drugs Advisory Committee focused on the reevaluation of the REMS. The meeting lasted 8.5 hours including the lunch break.²⁶² There were 15 advisors and 14 of them voted in favor of eliminating the need to send reports to the centralized WBC database associated with the REMS. The same 14 of 15 advisors also voted in favor of eliminating the certification of psychiatrists and pharmacists who prescribe clozapine in the US.²⁶³

The FDA officials revealed some information that has never been published: a Risk Management Analyst with the FDA indicated that the agency estimates that 25%–35% of patients taking clozapine may not be participating in the REMS.^262,263 It is perplexing that in a litigious country such as the US, approximately one-third of clozapine-treated patients are managed by clinicians and pharmacies willing to take the legal risk of prescribing clozapine outside of the REMS. This may be an indication that these clinicians and pharmacies are absolutely convinced that clozapine is needed in those patients; therefore, risk such high legal liability in prescribing clozapine outside of the REMS, but appear unwilling to comply with the very cumbersome bureaucratic regulations of the REMS.

Pressure was applied by multiple professional and advocacy organizations, including The Angry Moms, who contacted politicians and other national agencies, leading to the scheduling of the FDA meeting. A founding member of The Angry Moms explained that 1) the group submitted to the FDA 4000 signatures calling for the end of the REMS and 2) those most harmed by the clozapine REMS are the millions of patients who have never taken a single dose of clozapine, as the use of clozapine in the US is very limited. Many prescribers and pharmacies do not want to participate in the cumbersome REMS program, and those that do often misinterpret the program requirements. The chair of the panel of advisors said the public testimony was “very moving” and that families and patients had described “the intensity of suffering that people go through prior to getting to clozapine.”²⁶³

The FDA can follow the recommendation of the Advisory Committees to eliminate the REMS or not, but at the meeting the FDA officials made clear that even if the REMS program is eliminated, they do not plan to change the US package insert by reducing the requirements for hematological monitoring. Moreover, an FDA official presented a slide noting that the package insert will not be modified (at the 4:14:47 timestamp).²⁶²

The FDA briefly presented 2 US studies about neutropenia in clozapine-treated patients. These studies were funded by the FDA but have never been published in peer-reviewed journals, their authors were not revealed, and the studies were considered inconclusive. A British research group led by Taylor has published a series of studies^224–228 indicating that most cases of neutropenia in clozapine-treated patients are not life-threatening and may not even be caused by clozapine. In a study during the period 2007–2020 including approximately 3500 clozapine-treated patients, Taylor et al²²⁴ identified 23 patients (0.66% of 3500) with episodes recorded as agranulocytosis. Of these, 9 (0.26% of 3500) met predefined criteria and were considered episodes of life-threatening agranulocytosis. The life-threatening episodes exhibited a distinct pattern of continuous and rapid neutrophil count decline to zero or near zero that led to what has been called the Taylor criteria for diagnosing clozapine-induced neutropenia (at least 2 neutrophil counts <0.5 × 10⁹/L, life-threatening as indicated by medical treatment, and where no other cause was apparent).²²⁵ During the FDA meeting,²⁶² one of the advisory experts from the panel asked about the Taylor criteria (at the 4:19:02 timestamp). Finally, after 3 minutes (4:22:52), an FDA psychiatrist acknowledged that they had not looked at this study. After lunch (at the 7:17:40 mark) an FDA epidemiologist presented a biased critique of this study by calling it a case series of 23 cases and defining it as “hypothesis generating.” The advisory expert who asked the question corrected the FDA epidemiologist, explaining that the study started with 3500 patients initiated on clozapine (at the 7:18:30 mark).

The FDA First Draft Article for the Meeting Is No Longer Available

During the FDA meeting,²⁶² the FDA officials insisted several times that the risk of clozapine-induced agranulocytosis persists as long as the patient is taking clozapine, which is not consistent with the opinion of US and non-US clozapine experts (Box S16, Supplemental Digital Content 1, http://links.lww.com/JCP/A948). More importantly, this opinion of the FDA experts is not consistent with the information that the FDA experts had before the meeting. On November 17, the FDA released a 91-page “FDA Briefing Document”²⁶⁴ that described 2 studies on agranulocytosis that were called the “Veterans Affairs Study” and “the Brigham and Women's Hospital and Harvard Medical School Studies.” This “FDA Briefing Document”²⁶⁴ was no longer available on the day of the meeting but there were 2 final drafts presented by the 8 pharmaceutical companies²⁶⁵ totaling 93 pages plus a 10-page addendum presenting the “Veterans Affairs Study.”²⁶⁶

As the “FDA Briefing Document”²⁶⁴ is no longer available but reflects these 2 unpublished studies, we provide a summary. The Veterans Affairs Study included a cohort of 6488 patients followed on clozapine for up to 23 years. They identified only 32 (0.35%) cases of severe neutropenia and 2 (0.03%) deaths were likely attributable to clozapine-induced neutropenia. These numbers described here were not described during the meeting or in the addendum available at the meeting on November 19.²⁶⁵

The Harvard study matched 16,873 clozapine and olanzapine initiations.²⁶⁴ The tables of that document provide results on neutropenia that were unexpected by the FDA, but known worldwide by clozapine experts, that neutropenia is frequent in these patients before starting clozapine. In the 30 days before starting clozapine there were 136 (0.81%) cases of neutropenia and 85 (0.50%) cases before olanzapine. In the days −31 to −180 before starting clozapine, there were 94 (0.56%) cases of neutropenia and 83 (0.49%) cases of neutropenia before olanzapine. These numbers clearly indicate that neutropenia is present in these patients with ranges 0.49–0.81% in the absence of clozapine. In clozapine-treated patients 0.49–0.81% ranges of incidence of neutropenia should be expected as not caused by clozapine. Furthermore, these results suggest that Taylor et al²²⁴ are right and that many of the cases of neutropenia seen in clozapine-treated patients are probably not caused by clozapine. As far as we know, these ranges of the incidence of neutropenia in the absence of clozapine from the Harvard study were not presented to the public on November 19.

An important statement from this FDA Briefing document,²⁶⁴ which is no longer available on the FDA website, reveals that the FDA currently knows that 97% of deaths in clozapine-treated patients are not explained by agranulocytosis. The FDA studied clozapine ADRs found in the worldwide data that is described as the FDA Adverse Event Reporting System. They stated on page 40 of this document,²⁶⁴ “Additionally, we conducted a high-level overview of 11,610 FDA Adverse Event Reporting System worldwide reports coded with the serious outcome of death associated with clozapine from January 2015 through September 2023. Nearly 9 years of reports revealed that neutropenia was described in a low proportion of these fatal reports. In US-only reports of fatalities, only 3% mention neutropenia. Nearly half of the domestic fatal reports were not coded with any adverse events other than death, which precludes assessment of the cause of death. Upon review of the domestic fatal reports describing neutropenia, most were in the context of concurrent malignancy or infection.”

CONCLUSIONS

According to international comparisons,³² use of clozapine in the US is less frequent than in other counties. There is need to educate US psychiatry residents²⁶⁷ and other US mental health professionals on these advances in clozapine reviewed here and to encourage greater use of clozapine. Changes in the US package insert based on US pharmacovigilance data reviewed here are recommended to promote wider and safer use of clozapine in the US.²⁶⁸ Other countries will also benefit from changes in their package inserts and promoting wider and safer clozapine use.²⁶⁹ Physicians around the world need to be informed in particular that systemic inflammation during pneumonia and other infections manifests with CRP elevations and releases of cytokines that can double or triple serum/plasma clozapine concentrations. The combination of pneumonia and a clozapine intoxication may explain the high lethality of pneumonia in clozapine-treated patients, calling for early diagnosis. We recommend halving the dose of clozapine while CRP is elevated during pneumonia or other severe infections.^{122–124,270} Factors associated with TRS may contribute to risk of pneumonia in clozapine-treated patients, but we recommend the following: 1) using slower titrations and the lowest possible clozapine dosage (including dose decreases as patients reach geriatric age)¹²⁴ and 2) educating patients and family to contact the prescriber as soon as signs of severe infection develop.^124,270

This is Part II of an article by 40 authors proposing changes in the US package insert and these proposed changes are also supported by 124 other clozapine experts outside of the US from 45 countries/regions. Part II focuses on the pharmacovigilance sections while Part I⁶ focused on the sections devoted to basic pharmacology. The list of 124 non-US clozapine experts and their 44 respective countries/regions who support Parts I and II of this article is contained in Supplementary Table S5, Supplemental Digital Content 1, http://links.lww.com/JCP/A948; their letters are provided in Supplemental Digital Content 2, http://links.lww.com/JCP/A949.

DATA AVAILABILITY STATEMENT

VigiBase does not allow distribution of their data, but this article's Supplementary Material at http://links.lww.com/JCP/A949 provides more details on the reported findings.