Clozapine-related neutropenia and agranulocytosis in Korea: 2025 update for rethinking the role of monitoring system

Abstract

Background:

In February 2025, the FDA ended the mandatory Risk Evaluation and Mitigation Strategy requirements for clozapine. Korea continues mandatory hematologic testing under the Clozapine Patient Monitoring System (CPMS), yet no nationwide updates have been reported since 2006.

Objectives:

This study aimed to provide contemporary real-world evidence on the incidence, timing, and clinical outcomes of clozapine-associated neutropenia and agranulocytosis in Korea, and to evaluate adherence to monitoring regulations, to inform potential revision of the national CPMS.

Design:

A retrospective, multicenter cohort study was conducted using anonymized electronic medical records from two Korean tertiary hospitals between 2000 and 2023.

Methods:

Patients with schizophrenia spectrum disorders who had newly initiated clozapine were included. Absolute neutrophil count (ANC) data were used to identify neutropenia (ANC < 1500/μL), moderate neutropenia (<1000/μL), and agranulocytosis (<500/μL). Chart review determined causality for agranulocytosis. Monitoring adherence was assessed based on the proportion of ANC tests conducted within recommended intervals.

Results:

Among 2417 newly initiated patients (from 3364 total users), neutropenia occurred in 9.7%. Of these, 47.4% occurred within 18 weeks, and none who continued treatment progressed to agranulocytosis. Patients with neutropenia had lower baseline ANC than those without. Moderate neutropenia occurred in 1.7% and showed similar clinical patterns, most events being transient. Ten cases of agranulocytosis were identified, of which four (0.17%) were clozapine-related. All appeared abruptly within 5 weeks of initiation without preceding neutropenia or low baseline ANC, and recovered after discontinuation. Monitoring adherence was suboptimal: during the first 18 weeks, adherence to ⩽10-day interval was 73.7%, while 8.7% patients had no ANC tests; adherence after 1 year fell below 50%.

Conclusion:

Neutropenia is relatively common but usually low-risk, whereas clozapine-induced agranulocytosis is rare, abrupt, and confined to the early period. These findings suggest reconsideration of current hematological monitoring regulations regarding clozapine.

Introduction

Clozapine remains the most effective antipsychotic option for the treatment of schizophrenia and is uniquely indicated for reducing suicidal behavior.^1–4 Clozapine was first released in Europe in the 1970s as Leponex^® (Sandoz, Basel, Switzerland) but was briefly withdrawn from the market after reports of 18 cases of agranulocytosis, nine of which were fatal. ⁵ However, a subsequent double-blind, randomized trial demonstrated its superior efficacy, leading to its reintroduction into clinical practice. ⁶ Clozapine was approved in Europe in 1989 and by the United States Food and Drug Administration (FDA) in 1990 (Clozaril^®, Novartis, East Hanover, NJ, USA), under strict monitoring protocols aimed at early detection of potentially life-threatening agranulocytosis. The Clozapine Risk Evaluation and Mitigation Strategy (REMS) in the United States was the most stringent of these measures, mandating rigidly scheduled absolute neutrophil count (ANC) monitoring and centralized reporting before dispensing the medication. Despite its established efficacy, clozapine remains globally underutilized and often initiated later than clinically indicated, primarily due to concerns about hematologic risks, particularly neutropenia and agranulocytosis, and the burden of mandatory blood monitoring protocols.^7–9

In the United States, a pivotal regulatory shift occurred on February 24, 2025, when the U.S. FDA announced the elimination of the Clozapine REMS program. ¹⁰ As of that date, pharmacies are no longer required to verify laboratory test results or REMS authorization before dispensing clozapine, and prescribers are no longer obligated to report ANC values to the REMS registry. Although ANC monitoring is still recommended according to the prescribing information, the FDA concluded that REMS is no longer necessary to ensure the safety of clozapine treatment. This decision was based on the determination that labeling alone is sufficient to mitigate the risk of severe neutropenia associated with clozapine and that the REMS program had imposed unnecessary barriers to access. ¹¹ Elimination of the REMS program does not equate to removal of ANC monitoring recommendations in the United States; the clozapine prescribing information continues to recommend regular ANC monitoring (weekly for the first 6 months, every 2 weeks from months 6–12, and monthly thereafter). In contrast, regulatory approaches in Europe have evolved more substantively, with the European Clozapine Task Force and regulatory authorities advocating for meaningful revisions to long-term blood monitoring requirements. ¹²

This regulatory change reflects a broader reconsideration of how best to balance patient safety with treatment accessibility. Although the risk of clozapine-induced agranulocytosis is well-documented, its incidence is rare (less than 1%)^13,14 and most often occurs within the first 18 weeks of treatment. ¹⁴ In recent years, there have been increasing calls to relax rigid hematological monitoring requirements for clozapine, as such regulations have been recognized as a major barrier to the timely initiation and broader uptake of clozapine in clinical practice. Recent studies have questioned whether mandatory long-term monitoring remains proportionate to the low and temporally concentrated risk of clozapine-associated agranulocytosis. ¹⁵ Internationally, approaches to clozapine hematological monitoring differ substantially, with wide variation in both monitoring frequency and discontinuation thresholds across jurisdictions.^16,17 This international heterogeneity provides important context for understanding and re-evaluating country-specific monitoring systems.

In South Korea, the Clozapine Patient Monitoring Service (CPMS) was introduced to implement a structured system for regular hematologic testing. According to current Korean CPMS guidelines, blood tests, including ANC and white blood cell (WBC) count, must be conducted weekly for the first 18 weeks of clozapine treatment and monthly after 18 weeks. Under this system, patients are categorized into green, amber, and red zones based on their WBC and ANC results, with specific clinical actions mandated for each range. Entry into the red zone requires discontinuation of clozapine; this threshold is defined as ANC < 1500/μL during the first 18 weeks of treatment, and ANC < 1000/μL thereafter. ¹⁸ A detailed summary of CPMS thresholds and corresponding management recommendations is provided in Supplemental Table 1. ¹⁸

To date, only one large-scale study, published in 2006, has analyzed Korean CPMS data. ¹⁹ This study reviewed hematologic outcomes in more than 6700 patients treated with clozapine over an 11-year period. The findings showed that more than half of agranulocytosis and neutropenia cases occurred within the first 18 weeks of treatment, with no new cases reported after 9 years of continuous use. The authors concluded that long-term monitoring may not be necessary beyond the ninth year of treatment. Despite these findings, the CPMS protocol has remained unchanged for nearly two decades, and no updated epidemiological data have been made available in Korea. Given the recent regulatory shift in the United States and the absence of contemporary Korean data, the necessity of ongoing intensive monitoring in chronic clozapine users warrants re-examination.

In this study, we conducted a large-scale retrospective cohort analysis using data from two major tertiary hospitals in Korea to assess the incidence, timing, and clinical relevance of hematologic abnormalities in clozapine-treated patients with schizophrenia. Specifically, we evaluated the frequency and temporal distribution of neutropenia and agranulocytosis, as well as adherence to monitoring protocols, thereby raising important questions about the current CPMS system in long-term clinical practice. Our goal was to provide updated epidemiological and clinical evidence on hematologic side effects among Korean patients with schizophrenia, thereby informing future revisions of monitoring guidelines in alignment with current international standards.

Methods

Study design

This retrospective cohort study utilized anonymized clinical data from two tertiary hospitals in Korea: Seoul National University Hospital (SNUH; Site A) and Asan Medical Center (AMC; Site B). The study was approved by the Institutional Review Boards (IRB) of Seoul National University Hospital (IRB No. 2003-240-1115) and Asan Medical Center (IRB No. 2022-1132), with a waiver of informed consent due to the use of de-identified data. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for observational studies. ²⁰

Patient selection and data collection

Patients who received clozapine prescriptions from January 2000 to January 2023 were identified and screened from the electronic medical records. Patients with a primary diagnosis of schizophrenia spectrum disorder were included, based on the International Classification of Diseases, 10th Revision (ICD-10): schizophrenia (F20), schizoaffective disorders (F25), delusional disorder (F22), acute and transient psychotic disorders (F23), and unspecified psychotic disorders (F29). Patients were required to have received clozapine for at least 7 consecutive days during the study period, which was confirmed using electronic prescription records documenting dispensing dates and prescribed daily doses.

For each patient, demographic and clinical data were collected: sex, age, clozapine prescription information (dates and doses), and laboratory results, including complete blood count (CBC) with ANC, and clozapine serum concentration.

Patients were classified as newly initiated on clozapine if they had no prior clozapine prescription in the medical record, and the first dose of clozapine was ⩽25 mg. The newly initiated group was used for analyses of the incidence of neutropenia and agranulocytosis, as well as monitoring adherence patterns.

Hematologic outcomes and group definitions

Baseline hematologic parameters, including WBC count and ANC, were evaluated. Baseline laboratory values were defined as those obtained within 30 days before clozapine initiation. Neutropenia and agranulocytosis occurrences were identified from laboratory data. Neutropenia severity categories adopted in this study are as follows: mild neutropenia, ANC 1000–1499/µL; moderate neutropenia, ANC 500–999/µL; and severe neutropenia (agranulocytosis), ANC < 500/µL. ²¹ Agranulocytosis was analyzed separately and underwent a detailed chart review by clinical experts to confirm a causal relationship with clozapine use. If an alternative cause was identified, the possible cause was documented, and the timing and outcomes of all agranulocytosis events were analyzed.

Patients were classified according to whether they experienced at least one episode of neutropenia (defined as ANC < 1500/µL) during clozapine treatment, and baseline hematologic characteristics were compared between the neutropenia and non-neutropenia groups. As a secondary analysis, a stricter threshold was applied to identify moderate neutropenia (defined as ANC < 1000/µL), and similar comparisons were conducted to assess whether consistent patterns were observed. Each analysis was conducted independently, with control groups defined according to the absence of the specific hematologic outcome under investigation. Agranulocytosis cases were excluded from these analyses.

The concentration-to-dose (C/D) ratio was included as an index of individual clozapine metabolism, which has been reported to be associated with susceptibility to inflammatory adverse effects during clozapine treatment.^22,23 Serum clozapine concentrations were not measured as part of a standardized routine protocol but were obtained based on individual clinician judgment, and were therefore available only for a subset of patients. For those with available concentration data, the C/D ratio was calculated by dividing the plasma clozapine concentration (ng/mL) by the corresponding prescribed daily dose (mg).^24,25 If multiple measurements were available for an individual patient, median values were used to minimize the influence of outliers. Because serum concentrations were obtained based on clinical indication rather than a standardized schedule, steady-state conditions were not uniformly ensured.

Monitoring adherence

To assess adherence to hematologic monitoring recommendations, adherence was evaluated based on the Korean CPMS schedule, which recommends weekly ANC monitoring during the first 18 weeks of clozapine treatment and monthly monitoring thereafter. Because CPMS does not explicitly specify allowable grace periods between tests, we applied the grace-period definitions described in the FDA REMS advisory document to operationalize adherence assessment (⩽10 days instead of a strict 7-day interval during the first 18 weeks, and ⩽33 days instead of a strict 28-day interval thereafter). ²⁶ Adherence was quantified as the proportion of ANC tests meeting these interval criteria during the observation period, averaged across all patients.

Statistical analyses

Descriptive statistics were calculated for demographic and laboratory data and are presented as means ± standard deviations (SD) or median with interquartile range (IQR) for continuous variables and frequencies with percentages for categorical variables. Comparisons of baseline hematologic parameters between patients with and without neutropenia events were performed using independent samples t-tests. Two-tailed p-values < 0.005 were considered statistically significant. Where applicable, 95% confidence intervals (CIs) were calculated to quantify the precision of estimates. All statistical analyses were performed using R version 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Hematologic monitoring results

A total of 3364 patients diagnosed with schizophrenia spectrum disorders and treated with clozapine at two tertiary hospitals (Site A: 2685; Site B: 679) between January 2000 and January 2023 were included in the study. The average duration of clozapine use was 2191.8 days (approximately 6 years).

Across all patients, 191,383 WBC tests and 189,435 ANC tests were conducted. 181,656 ANC tests were performed during clozapine treatment, corresponding to an average of 54 tests per patient. Neutropenia (ANC < 1500/µL) occurred in 304 patients (9%). Moderate neutropenia (ANC < 1000/µL ) occurred in 56 patients (1.7%). Agranulocytosis (ANC < 500/µL) occurred in 10 patients (0.30%), of whom 4 patients (0.12%) were confirmed as clozapine-related events via chart review. Overall frequencies and site-specific data are summarized in Table 1.

Table 1.

Hematologic data and incidence of neutropenia and agranulocytosis among all patients.

Variable	Total	Site A	Site B
Number of patients, n	3364	2685	679
Sex, male (%)	1752 (51.2%)	1443 (53.7%)	309 (45.4%)
Average duration of clozapine use, days (years)	2191.8 days (6 years)	2311.9 days (6.3 years)	1717 days (4.7 years)
Number of ANC test results, n (per patient)	181,656 (54)	157,610 (58.7)	23,914 (35.22)
Neutropenia, n (%)	304 (9%)	248 (9.2%)	56 (8.2%)
Moderate neutropenia, n (%)	56 (1.7%)	46 (1.7%)	10 (1.5%)
Agranulocytosis, n (%)	10 (0.30%)	7 (0.26%)	3 (0.44%)
Agranulocytosis, clozapine-attributed,* n (%)	4 (0.12%)	4 (0.15%)	0 (0%)

Hematologic characteristics and frequency of neutropenia and agranulocytosis among patients treated with clozapine at two sites. The total number of patients, duration of clozapine use, number of ANC measurements, and the number of patients who showed neutropenia and agranulocytosis events are summarized. Neutropenia was defined as an ANC < 1500/μL (moderate neutropenia as an ANC < 1000/μL) and agranulocytosis was defined as an ANC < 500/μL. Values are presented as the number (percentage) unless otherwise indicated.

^*Clozapine-attributed cases of agranulocytosis confirmed via chart review.

ANC, absolute neutrophil count.

Neutropenia incidence among newly initiated clozapine users and baseline comparison of hematologic characteristics between neutropenia versus non-neutropenia groups

Among the 3364 patients, the incidence of hematologic side effects was investigated in the subgroup of patients who initiated clozapine after January 2000 (Figure 1). A total of 2417 patients with the prescription and ANC data from the clozapine initiation period were included for further analysis.

Figure 1.

Patient selection and inclusion criteria.

Flowchart illustrating the process of identifying and including patients for the study. Patients were screened based on clozapine prescription records and schizophrenia spectrum disorder diagnoses.

Among 2417 newly initiated patients, neutropenia was documented in 234 patients (9.7%), totaling 743 incidents. The average number of repeated neutropenia episodes per patient was 3.18 ± 6.94 (range 1–86). The time to first neutropenia showed a markedly right-skewed distribution, with a median of 156 days (IQR, 35–879 days); 47.4% (n = 111/234) of first episodes occurred within 18 weeks, whereas 39.3% (n = 123/234) occurred after 1 year of treatment (Table 2).

Table 2.

Summary of neutropenia and moderate neutropenia occurrence among new clozapine users (N = 2417).

Incident neutropenia (ANC < 1500)	Value
Occurrence, n	743 events, in 234 patients
Incidence, %	9.7%
Number of repeated neutropenia per patient, mean ± SD (min–max)	3.18 ± 6.94 (1–86)
Onset of first neutropenia, days, median (IQR)	156 (35–879)
<18 weeks (%)	47.4%
⩾1 year (%)	39.3%
Incident moderate neutropenia (ANC < 1000)
Occurrence, n	105 events, in 42 patients
Incidence, %	1.7%
Number of repeated neutropenia per patient, mean ± SD (min–max)	2.50 ± 6.39 (1–42)
Onset of first neutropenia, days, median (IQR)	355 (38–1486)
<18 weeks (%)	33.3%
⩾1 year (%)	54.8%

Occurrence and characteristics of neutropenia (ANC < 1500) and moderate neutropenia (ANC < 1000) events among patients newly initiated on clozapine. Data include total number of neutropenia events, number of patients, repeated episodes per patient (mean ± SD), time to first neutropenia episode from clozapine initiation (median [IQR]), and distribution of first episodes occurring within the first 18 weeks and after 1 year of treatment. The onset of first neutropenia was presented as median with interquartile range due to skewed distribution.

ANC, absolute neutrophil count; IQR, interquartile range; SD, standard deviation.

Within 18 weeks, 21 patients discontinued clozapine in compliance with CPMS regulations while the other 90 patients continued clozapine after neutropenia development. Among the patients who continued clozapine after neutropenia detection within 18 weeks, none were followed by agranulocytosis.

Moderate neutropenia (ANC < 1000/μL) occurred in 42 patients (1.7%), totaling 105 episodes (Table 2). The mean number of repeated episodes per patient was 2.50 ± 6.39 (range 1–42). The time to first moderate neutropenia was also highly right-skewed, with a median of 355 days (IQR, 38–1486 days); 33.3% of first episodes occurred within the first 18 weeks, whereas 54.8% occurred after 1 year of treatment. Clozapine was discontinued in five patients within this subgroup, while the others continued treatment with more frequent monitoring. Among the patients who continued clozapine after moderate neutropenia, none were followed by agranulocytosis.

Baseline hematologic parameters were compared between patients with at least one neutropenia episode (neutropenia group, n = 234) and those without (non-neutropenia group, n = 2183). The neutropenia group had significantly lower baseline WBC counts (5528.3 ± 1487.3 vs 7013.4 ± 2056.3, t = 12.729, p < 0.001) and ANC values (2971.8 ± 1251.6 vs 4108.5 ± 1743.2, t = 11.521, p < 0.001) Similarly, patients who developed moderate neutropenia (ANC < 1000/μL) had significantly lower baseline WBC counts (5761.8 ± 1977.9 vs 6869.5 ± 2050.2, t = -3.297, p = 0.001) and ANC values (3269.0 ± 1751.9 vs 3995.2 ± 1729.0, t = -2.593, p = 0.010) compared with those who did not (Table 3). Clozapine C/D ratio did not differ significantly between the neutropenia group (t = 1.289, p = 0.198) and moderate neutropenia group (t = 0.038, p = 0.969) compared with their respective controls (Table 3).

Table 3.

Comparison of baseline hematologic characteristics between groups with neutropenia event (neutropenia group) and those without (non-neutropenia group).

Variable	Control group (N = 2183)			Neutropenia group (ANC < 1500) (N = 234)			t	p-value
	N	Mean	SD	N	Mean	SD
Baseline WBC	1590	7013.371	2056.317	202	5528.317	1487.316	12.729	<0.001*
Baseline ANC	1565	4108.498	1743.242	201	2971.761	1251.596	11.521	<0.001*
C/D ratio	914	1.603	1.050	118	1.473	0.821	1.289	0.198
Variable	Control group (N = 2375)			Moderate neutropenia group (ANC < 1000) (N = 42)			t	p-value
	N	Mean	SD	N	Mean	SD
Baseline WBC	1754	6869.458	2050.213	38	5761.842	1977.928	−3.297	0.001*
Baseline ANC	1727	3995.155	1728.964	39	3269.0	1751.861	−2.593	0.010*
C/D ratio	1012	1.588	1.033	20	1.596	0.606	0.038	0.969

Baseline hematologic parameters (hemoglobin, white blood cell count, absolute neutrophil count, eosinophil count, platelet count) and clozapine concentration-to-dose ratio were compared between patients who experienced a neutropenia (ANC < 1500) or moderate neutropenia (ANC < 1000) episode and those who did not. Values are presented as mean ± SD. Statistical significance was assessed using independent samples t-tests.

An asterisk (*) indicates significance at the < 0.05. level.

ANC, absolute neutrophil count; C/D ratio, concentration-to-dose ratio; Hb, hemoglobin; PLT, platelet; SD, standard deviation; WBC, white blood cell count.

Agranulocytosis incidence and the clinical features of agranulocytosis cases

Among 2417 patients, a total of 10 (0.41%) experienced agranulocytosis during clozapine treatment. Following a detailed chart review, four cases (incidence = 0.17%) were adjudicated as primarily clozapine-related, while the remaining cases had plausible alternative causes. All clozapine-attributed cases of agranulocytosis occurred at Site A, while no cases were identified at Site B. Table 4 summarizes the clinical characteristics of all 10 agranulocytosis cases and the chart-review adjudication, including plausible causes, clozapine dose at the time of onset, days from initiation to event, and whether clozapine was discontinued. In the remaining six cases, possible contributing factors included chemotherapy, etanercept (a TNF-α inhibitor), and folk remedies. Although these cases were considered more likely attributable to alternative causes other than clozapine, a contributory role of clozapine could not be fully excluded, particularly in the presence of concurrent myelosuppressive exposures.

Table 4.

Summary of chart review from agranulocytosis cases.

Patient no.	Sex	Age	Onset (day)	ANC	Clozapine dose (mg/day)	Possible cause	Clozapine continued
1 a	M	56	26	126	37.5	Clozapine	No
2	M	33	26	120	50	Clozapine	No
3	F	40	35	492	200	Clozapine	No
4 a	F	35	35	23	62.5	Clozapine	No
5	F	44	1287	420	300	Chemotherapy	Yes
6	M	46	1799	368	225	Etanercept	No
7	F	44	3484	414	200	Chemotherapy	Yes
8	M	52	5212	395	212.5	Folk remedy b	Yes
9	F	19	5421	480	50	Chemotherapy	No
10	F	19	7191	230	200	Chemotherapy	Yes

Summary of chart review for the 10 patients who developed agranulocytosis during clozapine treatment, including demographic information, days from clozapine initiation to agranulocytosis event, ANC at onset, clozapine dosage at the time of event, suspected causes (including clozapine and other contributing factors), and whether clozapine therapy was continued following the event. In all cases, no mortality was found.

Bold values indicate cases adjudicated as clozapine-related agranulocytosis.

^aGranulocyte-Colony Stimulating Factor (G-CSF; Filgrastim) was administered to promote ANC recovery.

^bFolk remedy refers to non-prescribed traditional herbal preparations or fermented products sometimes used in Korean folk medicine. Specific ingredients could not be reliably identified from the medical records, and causality could not be definitively established. These products may contain unregulated components with potential hematologic effects.

ANC, absolute neutrophil count.

The four clozapine-attributed cases involved two female and two male patients, with agranulocytosis developing between days 26 and 35 after clozapine initiation. Clozapine was discontinued in all four cases, and two patients received granulocyte colony-stimulating factor (G-CSF) treatment. The mean baseline ANC among these four patients was 4073/µL (SD = 1117.5). The six cases attributed to other causes showed a wide range of onset days (1287 to 7191 days after clozapine initiation). In four of these six cases, clozapine was continued, and ANC levels recovered spontaneously. No mortality occurred in any of the 10 cases.

Adherence to hematologic monitoring

Adherence to hematologic monitoring recommendations was evaluated based on the proportion of ANC tests conducted within the recommended intervals (⩽10 days during the first 18 weeks, ⩽33 days thereafter, as described in the Methods section), averaged across all patients. Table 5 summarizes the average adherence rates across four time periods following clozapine initiation.

Table 5.

Monthly number of ANC tests by duration of clozapine use (n = 2417).

Duration of clozapine from initiation	N	Average adherence, %
First 18 weeks a	2332	73.7
18 weeks–6 months b	2006	75.4
6 month–12 month b	1935	58.3
12 month–24 month b	1746	43.2

The analysis included patients who continued to use clozapine until the end point of the specified period in the clozapine initiation cohort.

^aAdherence(gaps between ANC tests ⩽ 10 days), %.

^bAdherence (gaps between ANC tests ⩽ 33 days), %.

ANC, absolute neutrophil count.

Among all 2417 patients who initiated clozapine, 73.1% had at least one ANC test within 30 days prior to treatment initiation (baseline ANC). During the first 18 weeks of treatment, 8.7% of patients did not undergo any ANC testing, and the average adherence rate (⩽10-day interval) was 73.7%. Between 18 weeks and 6 months, adherence to the ⩽33-day interval was 75.4%. Thereafter, adherence declined to 58.3% during months 6–12 and to 43.2% during months 12–24 of treatment.

These findings indicate that monitoring adherence was relatively high during the early phase but progressively declined with longer treatment duration. Notably, no clozapine-related agranulocytosis occurred beyond the first 5 weeks of treatment, even among patients with reduced monitoring adherence.

Discussion

This study examined the real-world incidence and clinical significance of hematologic adverse events in clozapine-treated patients with schizophrenia using large-scale, longitudinal data from two tertiary hospitals in Korea. Among 3364 patients treated over a 23-year period, neutropenia (ANC < 1500/μL) occurred in 304 patients (9.0%), and moderate neutropenia (ANC < 1000/μL) occurred in 56 patients (1.7%). Agranulocytosis (ANC < 500/μL) was identified in 10 patients, but only four cases (0.12%) were adjudicated as primarily attributable to clozapine after expert chart review. All clozapine-attributed agranulocytosis events occurred abruptly between days 26 and 35 after initiation, with no cases observed beyond the first 5 weeks, and no mortality was observed. Monitoring adherence was suboptimal, particularly over time: during the first 18 weeks, only 73.7% met the recommended ⩽10-day ANC testing interval, and 8.7% had no testing at all, while adherence to the ⩽33-day interval fell below 50% after 1 year.

In our cohort, neutropenia was relatively common but largely transient and clinically benign, occurring in 9.7% of patients newly initiated on clozapine, with nearly 40% of first episodes arising after 1 year of treatment. A recent large-scale study from Australia and New Zealand analyzed over 2.6 million ANC values in 26,630 clozapine-treated patients and reported that only 4.3% experienced minor neutropenia. ²⁷ In the study, the weekly incidence rate peaked at 9 weeks (0.218%) and declined to a stable average of 0.01% thereafter. ²⁷ As in other observational studies, neutropenia events in our cohort could not always be definitively attributed to clozapine, as categorical confirmation is often not possible in the absence of identifiable alternative causes, such as exposure to other medications known to induce neutropenia.

In our study, the neutropenia group had lower baseline ANC values, although mean levels remained well above the conventional neutropenia threshold. This group also exhibited more frequent and recurrent neutropenia episodes, accompanied by more frequent ANC monitoring. Importantly, no patients with mild neutropenia who continued clozapine subsequently developed severe neutropenia or agranulocytosis, suggesting that mild neutropenia may not necessarily warrant the same level of concern as more severe hematologic events. When a stricter definition of moderate neutropenia (ANC < 1000/μL) was applied, the incidence was 1.7%. Most episodes were transient and occurred after 1 year of treatment, and many patients continued clozapine with intensified monitoring rather than discontinuation. These findings suggest that even below the 1000/μL threshold, clozapine continuation may be clinically feasible in carefully monitored patients. Genetic variants associated with benign ethnic neutropenia (BEN), such as the Duffy-null (rs2814778 c-c) genotype, ²⁸ are known to be exceedingly rare in East Asian populations,^28,29 suggesting limited relevance to the lower baseline ANC observed in our cohort. Notably, despite prior reports that East Asian patients may be more susceptible to clozapine-related fever or pneumonia and tend to have higher clozapine C/D ratios, ²² we found no significant difference in C/D ratios between patients who developed neutropenia and those who did not. These findings suggest that inter-ethnic differences in clozapine metabolism are unlikely to explain the occurrence of neutropenia. Taken together, these findings suggest that lower baseline ANC may be associated with an increased likelihood of neutropenia during clozapine treatment, while not necessarily indicating an elevated risk of agranulocytosis. Baseline hematologic profiles may help contextualize neutropenia events and distinguish transient fluctuations from clinically meaningful toxicity, particularly in the early phase of treatment.

These observations also raise an important question regarding the attribution of neutropenia observed during clozapine treatment. Neutropenia is not uncommon in the general population, and population-based analyses have shown that mild neutropenia can be observed in otherwise healthy individuals in the general population. ³⁰ Clinical evaluations of apparent clozapine-associated neutropenia further suggest that a substantial proportion of cases may in fact be unrelated to clozapine exposure and instead attributable to alternative or background causes. ³¹ Episodes of neutropenia have also been reported with several non-clozapine antipsychotics, including olanzapine, risperidone, quetiapine, and aripiprazole, and comparative observational studies have not consistently demonstrated a higher incidence among clozapine-treated patients than among those receiving other antipsychotics. ³² Given that concomitant medications could not be fully controlled in this retrospective study, it is difficult to attribute all observed neutropenia events solely to clozapine exposure. Recent analyses have also suggested that many episodes of mild neutropenia detected during clozapine treatment may reflect physiological variability in neutrophil counts that becomes apparent through intensive hematologic monitoring. ³³ Longitudinal monitoring cohorts have shown that mild neutropenia may occur without progression to agranulocytosis. ²⁷ Taken together, these findings suggest that reductions in neutrophil counts observed during clozapine treatment may not uniformly reflect a direct drug-induced effect.

In our dataset, 10 agranulocytosis cases were identified in electronic ANC records, but chart review identified only 4 cases (0.12%) in which clozapine was considered the primary contributing factor; none of the cases led to death. The incidence of clozapine-related agranulocytosis in our cohort was lower than reports from Western countries. Alvir et al. ³⁴ reported a rate of 0.8% in the United States, and a similar incidence of 0.7% has been reported in Italy. ³⁵ By contrast, our findings were more comparable to the lower incidence of 0.21% observed in China. ³⁶ The lower incidence observed in our cohort may partly reflect stricter causality adjudication, highlighting the importance of distinguishing true clozapine-related agranulocytosis from coincidental severe neutropenia due to alternative causes.

All cases adjudicated as clozapine-related agranulocytosis occurred within the first 5 weeks of treatment. These findings align with prior studies indicating that the risk of clozapine-induced agranulocytosis is rare and primarily concentrated in the early phase of treatment.^14,32 For example, a review by Andersohn et al. ³⁷ reported that most cases develop within the first 2 months of treatment. More recent large-scale epidemiological studies have challenged the historical perception of agranulocytosis risk. A Finnish study of 61,769 patients treated between 1996 and 2017 reported a cumulative incidence of 1.37% (95% CI, 0.58–3.16) among clozapine users. ³⁸ The risk decreased steeply over time: adjusted odds ratio (aOR) of 36.01 for those treated <6 months, compared to 4.38 for those treated > 54 months. Only one of 3559 patients died due to clozapine-induced agranulocytosis. Similarly, the Australia/New Zealand study found that serious neutropenia requiring clozapine discontinuation occurred in 1.2% of patients, with most events within the first 18 weeks and becoming negligible after 2 years. ²⁷ All patients with clozapine-related agranulocytosis recovered their ANC after drug cessation, and two received G-CSF. No mortality occurred. Although agranulocytosis is a serious adverse event, these results indicate that, with early detection and prompt discontinuation, it is generally reversible. In contrast, several non–clozapine-related agranulocytosis cases were associated with concurrent myelosuppressive treatments, underscoring the need for vigilant monitoring when such agents are used.

An important consideration in interpreting hematologic safety data for clozapine is the distinction between clozapine-induced neutropenia (CIN) and clozapine-induced agranulocytosis (CIA), which are often treated as part of the same continuum but may reflect fundamentally distinct pathophysiological processes.^39,40 It has also been suggested that many neutropenia events observed during clozapine treatment may reflect surveillance bias rather than a direct drug effect, as intensive hematologic monitoring increases detection of otherwise clinically silent neutrophil fluctuations. ⁴¹ Clinically, CIN is relatively more common, tends to emerge gradually, and is often reversible.^42,43 Rechallenge is frequently feasible after successful resolution, and onset timing is highly variable across individuals.^43,44 In contrast, CIA is rare but typically abrupt in onset, more severe, and often considered a contraindication for rechallenge due to its potentially fatal course.^34,39 CIA also tends to occur early in the course of treatment, prompting current monitoring protocols to emphasize intensive surveillance during the initial months. ³⁴ Notably, these patterns were also observed in our results. In our cohort, neutropenia occurred throughout treatment, including late-onset cases, and was not followed by agranulocytosis in any of these patients. Neutropenia was more frequent in those with lower baseline ANC, suggesting that some cases may reflect a benign predisposition rather than drug toxicity. Conversely, CIA occurred abruptly, early in treatment, and in patients without low baseline ANC. These findings support the notion that CIN and CIA are pathophysiologically distinct.

The mechanisms underlying CIN and CIA are not yet fully understood. Proposed explanations include direct toxicity from reactive clozapine metabolites and oxidative stress that damages neutrophils, as well as the involvement of an immune-mediated process.^45,46 Recent genetic studies suggest that CIA may be more strongly associated with immune-mediated pathways. ⁴⁷ In particular, certain HLA haplotypes such as HLA-DQB1 and HLA-B have been implicated in genome-wide association studies, indicating a potential T cell–driven or antibody-mediated destruction of granulocytes.^42,48 Yet, many genetic studies have relied on broad case definitions that assume all neutropenia or agranulocytosis during clozapine treatment to be clozapine-related, potentially obscuring distinct biological mechanisms. More refined phenotypic definitions, particularly focused on adjudicated early-onset agranulocytosis, may be needed to clarify true genetic susceptibility.

While it remains unclear whether CIN and CIA represent two ends of a single continuum or fundamentally distinct phenomena, this possible divergence is clinically relevant. Recent work has highlighted that clozapine-related agranulocytosis may show a distinct early temporal pattern and that pattern-based approaches may help differentiate true clozapine-related dyscrasia from coincidental neutropenia.^41,49 Current clozapine ANC monitoring protocols are intended to enable early detection of agranulocytosis and prevent serious clinical consequences. Therefore, if these two conditions arise from different mechanisms and carry different levels of risk over time, tailoring the intensity and duration of monitoring may become an important consideration. In our study, all identified agranulocytosis cases occurred within the first 5 weeks of treatment, and no new cases were observed during later treatment phases—even as adherence to monitoring declined over time. These observations reinforce the importance of intensive monitoring during the early high-risk period, while also suggesting that the occurrence of later-onset neutropenia does not necessarily indicate imminent progression to agranulocytosis or mandate immediate clozapine discontinuation. Together with prior literature indicating a declining risk of agranulocytosis after prolonged treatment, our results support ongoing efforts to reevaluate long-term monitoring requirements in low-risk patients, such as patients with stable long-term clozapine use, consistently normal baseline ANC values, absence of prior severe hematologic events, and no exposure to concurrent myelosuppressive medications.

Under the U.S. REMS guidelines—prior to its termination in February 2025—clozapine discontinuation was required if the ANC dropped below 1000/μL, or below 500/μL for patients with documented BEN. In Korea, the CPMS mandates clozapine discontinuation when ANC falls below 1500/μL during the first 18 weeks of treatment, and below 1000/μL after 18 weeks. However, our real-world data suggest that strict adherence to these cutoffs may not always be clinically warranted. Before 18 weeks, a substantial number of patients experienced ANC < 1500/μL but continued clozapine, and none of these cases progressed to agranulocytosis. Although current CPMS recommendations specify clozapine discontinuation in cases of early neutropenia, we observed that some clinicians opted to continue or reduce clozapine with close hematologic follow-up, reflecting real-world clinical decisions. While this practice is not fully aligned with CPMS recommendations, it appears to be relatively common and reflects a risk–benefit consideration aimed at avoiding premature clozapine discontinuation. Likewise, after 18 weeks, several patients showed ANC values < 1000/μL; clozapine was discontinued in only five of these cases, while the remainder continued treatment with closer monitoring, again without subsequent agranulocytosis. These observations indicate that rigid enforcement of CPMS thresholds can lead to premature discontinuation in patients who might otherwise safely remain on therapy, highlighting the potential value of adopting more flexible, risk-stratified monitoring strategies.

Efforts to revise rigid ANC monitoring thresholds have gained momentum internationally, reflecting a broader shift toward more proportionate, risk-based approaches to clozapine safety. In the UK and Europe, Bastiampillai and colleagues ⁵⁰ proposed lowering the BEN-related ANC cutoff to 500/μL (with the current UK cutoff for BEN set at 1000/μL), highlighting concerns that overly conservative thresholds may unnecessarily restrict access to clozapine, particularly among ethnically diverse populations. These discussions underscore increasing recognition that uniform ANC cutoffs may not adequately reflect biological variability or true clinical risk. In the United States, the FDA’s 2025 decision to eliminate the mandatory ANC reporting requirement under the REMS program represented a regulatory shift focused on reducing administrative and access-related barriers. The agency concluded that such reporting was no longer essential for risk mitigation and that overly rigid policies may in fact hinder access to clozapine for patients with treatment-resistant schizophrenia.^11,51 In parallel, an international Delphi consensus panel led by Siskind et al. updated global recommendations for ANC and adverse drug reaction (ADR) monitoring during clozapine. ⁵² Drawing on input from over 90 international experts, the panel recommended lowering the ANC cessation threshold to 1.0 × 10⁹ cells/L (and to 0.5 × 10⁹ cells/L for individuals with Duffy antigen receptor for chemokines [DARC]-null phenotypes) and discontinuing routine ANC monitoring after 2 years of treatment, based on robust epidemiologic data demonstrating that the risk of clozapine-induced agranulocytosis becomes negligible beyond the early treatment period. While these recommendations align closely with current U.S. prescribing thresholds, they are notable in that they arise from a global, evidence-based clinical consensus rather than from regulatory policy considerations. Taken together, these developments suggest that recent U.S. regulatory changes are not isolated or exceptional, but are broadly concordant with evolving international expert opinion regarding the low, temporally concentrated risk of CIA. This convergence of regulatory pragmatism and clinical consensus provides an important context for reconsidering long-standing national monitoring systems, such as Korea’s CPMS, which continue to apply relatively conservative ANC thresholds and prolonged mandatory monitoring without recent empirical re-evaluation.

Adherence to the recommended ANC monitoring schedule was suboptimal, particularly after the early phase. During the first 18 weeks, only 73.7% of patients met the recommended ⩽10-day ANC monitoring interval, and 8.7% had no ANC testing at all. Given that intensive monitoring during this period is mandatory under CPMS, this adherence rate is notably low and underscores substantial real-world challenges in sustaining frequent blood testing. After 1 year, adherence to the ⩽33-day interval fell below 50%. Nevertheless, no clozapine-attributable agranulocytosis occurred beyond the early phase, which may suggest that the clinical yield of prolonged routine monitoring is limited in clinically stable patients.

Our findings align with the evolving international view that long-term intensive ANC monitoring may offer limited clinical value. All agranulocytosis cases occurred within the first 5 weeks of treatment. No additional cases were observed beyond this period, even though adherence to monthly ANC monitoring declined steadily over time. Neutropenia was more common, showed broader variability in onset, and was not followed by agranulocytosis in any case. These findings indicate that the risk of agranulocytosis is both rare and temporally concentrated. Given the declining monitoring adherence in long-term users and the absence of late-onset cases, these data reinforce the need to reevaluate Korea’s CPMS protocol. Aligning national guidelines with emerging international standards may reduce unnecessary testing, prevent unwarranted discontinuations, and improve access to clozapine for patients with treatment-resistant schizophrenia.

Limitations

This study has several limitations that should be acknowledged. First, the retrospective design limits causal inference and is subject to incomplete follow-up. Loss to follow-up was not systematically captured, which may have influenced estimates of monitoring adherence, particularly in later treatment phases. In addition, hematologic events observed during clozapine treatment cannot always be definitively attributed to clozapine itself, especially neutropenia, which may arise from intercurrent medical conditions or concomitant medications. Furthermore, the reasons for clozapine discontinuation could not always be clearly determined from the medical records, as treatment cessation may have been influenced by clinical judgment, patient preference, or non-hematologic adverse events rather than hematologic abnormalities alone. Second, serum clozapine concentrations were not obtained through a standardized monitoring protocol and were available only for a subset of patients, with steady-state conditions not uniformly ensured. This limits the interpretation of concentration-related measures, including the C/D ratio. Third, generalizability may be limited, as this study was conducted within the Korean CPMS framework in an East Asian population and at tertiary referral centers with relatively robust monitoring resources. Finally, although major hematologic outcomes were systematically captured, residual confounding by underlying medical conditions, concurrent non-psychiatric medications, and antipsychotic polypharmacy could not be fully addressed.

Conclusion

Our findings indicate that the risk of CIA is both rare and temporally concentrated in the early phase of treatment, whereas later-onset neutropenia is common but largely benign and not predictive of progression to agranulocytosis. These findings provide timely evidence to inform potential revisions of Korea’s CPMS system and highlight the need for more individualized, evidence-based approaches to hematologic monitoring in clozapine therapy. Future research should focus on prospective, longitudinal studies to better characterize the clinical significance of neutropenia during clozapine treatment. Studies incorporating standardized therapeutic drug monitoring, baseline hematologic profiling, and careful assessment of comorbid conditions and concomitant medications will be important for refining risk stratification and optimizing monitoring strategies.