|
|
| |||||||||||||||||||||||||||||||
Background Clozapine is an atypical antipsychotic agent that is more effective than standard neuroleptic drugs in the treatment of patients with refractory schizophrenia. Unlike classic neuroleptic agents, clozapine is not associated with the development of acute extrapyramidal symptoms or tardive dyskinesia. The main factor limiting its use is the risk of potentially fatal agranulocytosis, estimated to occur in 1 to 2 percent of treated patients. After clozapine was approved by the Food and Drug Administration, it became available for marketing in the United States in February 1990 only as part of a special surveillance system (the Clozaril Patient Management System, or CPMS), in which a weekly white-cell count was required for the patient to receive a supply of the drug.
Methods We evaluated the CPMS data for February 1990 through April 1991 by survival analysis to determine the incidence of agranulocytosis and the effects of potential risk factors such as age and sex. Data were available for 11,555 patients who received clozapine during the period after marketing began.
Results Agranulocytosis developed in 73 patients, resulting in death from infectious complications in 2 patients. Episodes of agranulocytosis occurred in 61 patients within three months after they began treatment. The cumulative incidence of this side effect was 0.80 percent (95 percent confidence interval, 0.61 to 0.99) at 1 year and 0.91 percent (95 percent confidence interval, 0.62 to 1.20) at 1 1/2 years. The risk of agranulocytosis increased with age and was higher among women.
Conclusions The occurrence of agranulocytosis is a substantial hazard of the administration of clozapine, but this hazard can be reduced by monitoring the white-cell count. The increasing risk of agranulocytosis with age and the reduced incidence after the first six months of treatment provide additional guidelines for the prescription and monitoring of clozapine treatment in the future.
Despite the promise of clozapine, its use has been limited by its propensity to cause agranulocytosis. Reports from Europe in the 1970s described rates of agranulocytosis among patients given clozapine (0.05 to 2 percent) that were higher than the rates among patients given standard psychotropic drugs10,11,12,13,14,15,16. Moreover, a number of patients died11,13. Clozapine-induced agranulocytosis selectively affects precursors of polymorphonuclear leukocytes in the bone marrow and can be reversed without hematologic sequelae if treatment is stopped promptly17,18. The pathophysiologic mechanism is unknown, but there is evidence that the reaction is immunologically mediated19. Previous estimates of the rate of clozapine-induced agranulocytosis among American patients, based on relatively small numbers who received clozapine on an investigational basis, ranged between 1 and 2 percent17. The generalizability of these estimates is uncertain, since they were based on stringently selected patients treated in research studies at a few tertiary care institutions.
Clozapine was approved by the Food and Drug Administration in October 1989 and was first marketed in the United States in February 1990. It was distributed solely by one firm (Caremark Homecare, an affiliate of Baxter Healthcare, Lincolnshire, Ill.) in conjunction with weekly measurement of the white-cell count under the Clozaril Patient Management System (CPMS) until this arrangement for distribution was discontinued on May 31, 1991. The data collected through the CPMS provided an opportunity to estimate the incidence of clozapine-induced agranulocytosis and to examine potential risk factors for it in the cohort of patients who received clozapine after commercial distribution of the drug began in the United States.
Methods
Data Collection and Patients
Each patient taking clozapine (Clozaril, Sandoz, East Hanover, N.J.) was assigned a case administrator who was responsible for monitoring compliance with measurement of the white-cell count, maintaining regular contact with the treating physician, and maintaining the patient's records. The case administrator recorded the patient's age, sex, height, weight, and diagnosis during registration for treatment. Weekly visits were scheduled for monitoring white-cell counts and dispensing the drug. Only one week's supply of medication was dispensed to facilitate compliance with hematologic monitoring. The data collected during these visits included the amount of drug dispensed and the reasons for interrupting or discontinuing treatment. Pilot testing of the CPMS revealed a compliance rate of more than 99 percent for weekly measurement of the white-cell count20. Blood samples were delivered to a central laboratory for analysis (Roche Biomedical Laboratories, Raritan, N.J.); the coefficients of variation were 5, 3, and 2 percent for white-cell counts with expected values of 2000, 7000, and 18,000 cells per cubic millimeter, respectively. The hematologic test results were provided weekly to the treating physician. The following results were immediately communicated by the case administrator to the treating physician: a white-cell count below 3500 per cubic millimeter; a large absolute drop in the white-cell count, even if the count remained above 3500 per cubic millimeter; or three consecutive drops in the count. Treatment guidelines specified that a complete blood count with a differential count be determined twice weekly if the white-cell count dropped below 3500 per cubic millimeter. Clozapine treatment was interrupted if the white-cell count fell below 3000 per cubic millimeter or the absolute polymorphonuclear-leukocyte count fell below 1500 per cubic millimeter, and daily hematologic monitoring was instituted; treatment could be resumed if the white-cell count and the absolute polymorphonuclear-leukocyte count returned to levels above 3000 and 1500 per cubic millimeter, respectively. The guidelines stipulated that clozapine be permanently discontinued if leukopenia developed, indicated by a white-cell count below 2000 per cubic millimeter or an absolute polymorphonuclear-leukocyte count below 1000 per cubic millimeter. Weekly monitoring was continued for at least four weeks after clozapine was discontinued for any reason21. A master list was maintained (by Caremark) of all patients reported to have agranulocytosis and patients withdrawn from clozapine treatment because of leukopenia.
Treatment data from the distributor and hematologic data from the central laboratory on each visit were merged by the drug manufacturer according to the patient's social security number. If this identifier was unavailable, other identifiers were used, including the patient's name. The data were processed in biweekly batches of records. The proportion of unlinkable records was 8 percent for batches processed at the start of the period after marketing began, declining to 3 to 4 percent within a month and to 1 percent for batches processed at the end of 1990. The unlinked records did not contain any white-cell counts indicating agranulocytosis or leukopenia. The demographic, clinical, and hematologic data, without the patient's name, were further processed and analyzed at the Mental Health Clinical Research Center of Hillside Hospital, Long Island Jewish Medical Center.
Analyses were limited to patients taking clozapine during the period from February 5, 1990, through April 1991 who had been followed for at least three weeks. Patients without agranulocytosis were included in the study cohort only if blood samples were available for at least three time points. Patients who began taking clozapine before marketing but who were still taking it during the period after marketing began were included in the analyses. Patient accrual continued through the study period.
The analyses were performed on data from 11,555 patients, including 1353 who began treatment during the premarketing period; 5323 patients with less than three weeks of follow-up and 164 patients with only two hematologic records from the data base were excluded. Demographic and base-line clinical information was available for 11,382 of the 11,555 patients in the analyses. The numbers of patients followed for one, three, and six months were 11,033, 8608, and 5780, respectively. Follow-up covering 1 1/2 years was available for 898 patients who had begun treatment during the premarketing period. Clozapine treatment was stopped for reasons other than agranulocytosis or leukopenia in 1281 patients, including 1244 whose recorded white-cell count never fell below 3500 per cubic millimeter. The most common reasons for stopping treatment were a lack of response (34 percent), noncompliance (18 percent), and adverse reactions other than agranulocytosis or leukopenia (16 percent).
Patients with Agranulocytosis
Patients with agranulocytosis were identified from the hematologic records and the master list of patients reported to have this complication. An episode of agranulocytosis was indicated by a confirmed absolute polymorphonuclear-leukocyte count below 500 cells per cubic millimeter. Patients whose hematologic records indicated an episode of agranulocytosis were checked against those on the master list. Nine patients not included on the master list had single hematologic records that indicated agranulocytosis. The records of these nine patients were determined to contain errors: the white-cell counts determined in six patients immediately before and after the supposed episodes were clearly incompatible with agranulocytosis; the seventh patient was mistakenly thought to have agranulocytosis because of a clerical error (transposition of the digits of a social security number); and the eighth and ninth patients, in whom agranulocytosis was detected at base line, had never taken clozapine.
Two patients who had agranulocytosis according to the master list had no records in the hematologic data base. The dates of the start of treatment and agranulocytosis in these two patients were added to the data base. Seventeen patients on the master list had no hematologic records indicating this complication. Further investigation revealed that they had been hospitalized and their blood monitoring had been taken over by the hospitals. All but 4 of these 17 patients had recorded white-cell counts below 3500 per cubic millimeter before the date that agranulocytosis occurred. For 10 of these 17 patients, the white-cell counts indicating agranulocytosis and the dates of these counts were recovered from hospital laboratories; for the other 7 patients, the dates of agranulocytosis were estimated on the basis of the dates of hospitalization and ancillary hematologic data. In all, 73 patients were considered to have had agranulocytosis.
Prodrome to Agranulocytosis
The onset of the prodrome to agranulocytosis was determined by examining individual plots of white-cell counts and determining the date from which the count showed a continual decline until agranulocytosis developed, allowing for one possible increase in the count during that interval but only to a level that did not exceed the base-line count. The duration of the prodrome, defined as the time from onset to the date of agranulocytosis, was determined in 68 of the 73 patients. In one patient, an atypical and irregular pattern of white-cell counts precluded identification of the date of onset. In four patients, weekly data for the period immediately before agranulocytosis were sparse or missing.
Limitations of the Data
The CPMS data were not specifically designed to provide information about the incidence of agranulocytosis and the risk factors for this complication. Potentially interesting characteristics of the patients, such as their race or ethnic background, were not routinely recorded on the registration forms. The blood counts of patients who were hospitalized for agranulocytosis were often not included in the data base because their blood monitoring was taken over by the hospitals. However, the data did provide information necessary to estimate the incidence of agranulocytosis as well as the effects of several potential risk factors.
Statistical Analysis
The Lifetest and PHREG procedures of SAS/STAT software22,23 were used to carry out life-table and Cox proportional-hazards regression analyses. The regression analyses tested for the effects of age, sex, and base-line white-cell count on the risk of agranulocytosis. Survival analysis was particularly appropriate because the accrual of patients continued throughout the study period, resulting in unequal duration of follow-up24.
The effects of dosage were tested with the use of conditional logistic regression. Each patient with agranulocytosis was matched according to the date of the initial visit with 10 randomly selected controls (other patients in the data base who did not have agranulocytosis) for whom data were available through the date when agranulocytosis developed in the patient. Four patients had fewer than 10 controls. One patient, who made the initial visit on December 24, was matched with six controls who made their initial visits during the week surrounding this date. The PHREG procedure23 was used to carry out the conditional logistic regression. The analyses evaluated the effects of the maximal dose, defined as the highest dose received during clozapine treatment, the last recorded dose, and the cumulative dose. The collection of data on dosage in the controls was cut off on the date when agranulocytosis developed in the patient. Body weight was controlled for in the dosage analyses.
The association between the duration of the prodrome and age, sex, and the base-line white-cell count was also investigated.
Results
Of the 11,382 patients for whom demographic data were available, 7032 were men (62 percent) and 4350 were women (38 percent). Table 1 shows their clinical characteristics, maximal doses, and white-cell counts according to whether or not agranulocytosis developed. The recorded diagnosis was schizophrenia in 95 percent of the patients, affective psychosis in 2.2 percent, and organic psychosis in 1.0 percent.
|
|
|
|
|
|
The mean (±SD) duration of the prodrome of agranulocytosis was 29.3 ±12.6 days (median, 28). Thus, in half of the 68 patients for whom this information was available, the white-cell count started to decline at least four weeks before agranulocytosis occurred. In only six patients did the prodrome last two weeks or less. However, 16 patients, all of whom had declining white-cell counts in the period before agranulocytosis, had no white-cell counts below 3500 per cubic millimeter within eight days before the occurrence of agranulocytosis. Although the patients who eventually had agranulocytosis had consistent declines in their white-cell counts, their clozapine treatment was continued until their counts fell below 3000 per cubic millimeter.
Neither age nor sex was associated with the duration of the prodrome. The length of the prodrome correlated with the magnitude of the base-line white-cell count (r = 0.26; 95 percent confidence interval, 0.02 to 0.47) -- i.e., in patients whose count had further to decline, the period before their granulocyte count dropped below 500 per cubic millimeter was longer.
Discussion
The incidence of agranulocytosis was 0.80 percent at 1 year and 0.91 percent at 1 1/2 years. The majority of cases of agranulocytosis (61 of 73 patients) occurred within the three months after the start of treatment, with the risk peaking in the third month. Only three cases occurred after 6 months, and one case after 1 1/2 years. The prodromal period leading to agranulocytosis was long in the majority of cases. However, agranulocytosis developed in 16 patients within eight days of white-cell counts above 3500 per cubic millimeter, the threshold level for beginning semiweekly monitoring, and above the threshold level for interrupting clozapine treatment (3000 per cubic millimeter), despite consistent declines in the counts. These results suggest that the frequency of hematologic monitoring could be reduced after the initial six months of treatment, provided that clozapine treatment may be interrupted at a higher threshold level for the white-cell count in patients whose counts show consistent declines.
The risk of agranulocytosis increased with age, although it was somewhat higher among patients below the age of 21 years than among those 21 to 40 years old. Although the higher risk among women was reduced after adjustment for age, the adjusted risk ratio for women (1.60; 95 percent confidence interval, 0.99 to 2.58) was still substantial. These results are consistent with previous reports associating female sex and increasing age with an increased risk of drug-induced neutropenia and agranulocytosis26,27.
The risk of clozapine-induced agranulocytosis has been reported to be increased among patients of Ashkenazi Jewish descent with the haplotype HLA-B38, DR4,DQw328. Unfortunately, information about the race or ethnic background of the patients in our study was not available in our data base.
Two of the 73 patients who had agranulocytosis according to our data base died25. In addition, five patients who began taking clozapine after April 1991 have since died of disorders related to agranulocytosis (two women 59 and 72 years old and three men 40, 49, and 60 years old) (Krassner M [Sandoz]: personal communication). All five patients died within three months after they began clozapine treatment. In all other patients with agranulocytosis, this complication has proved reversible, with no persisting hematologic sequelae17,29. Given the increased risk of agranulocytosis and the limited data on the efficacy, safety, and optimal dose range of clozapine in elderly patients, clinicians should be particularly careful in selecting and treating elderly patients until the risks and benefits of clozapine in this age group are better known.
Since the discontinuation of the CPMS system, treating physicians have had primary responsibility for hematologic monitoring. Experience suggests that it may not be possible to prevent the occurrence of agranulocytosis, but the purpose of hematologic monitoring is to prevent complications and deaths due to this disorder. Our report suggests that the hematologic guidelines required of clinicians by the maker of clozapine (Sandoz) can be modified so that treatment with this drug can be managed in a less burdensome yet safe manner. Currently, in several European countries, including the United Kingdom, Germany, and Finland, the frequency of monitoring is reduced after the initial 18-week period of maximal risk, to testing every 2 weeks in the United Kingdom and at least monthly in Germany and Finland (Jager P [Sandoz]: personal communication). Future modifications might include reducing the frequency of white-cell counts to testing every two weeks after six months of treatment. However, clinicians must pay attention to patterns of decline in the counts in addition to absolute threshold levels. In addition, new methods of treating agranulocytosis that stimulate granulopoiesis (the use of granulocyte and granulocyte-macrophage colony-stimulating factors30,31,32) should be evaluated to determine whether they can reduce the complications associated with clozapine-induced agranulocytosis.
Supported by a Research Scientist Development Award (MH-00537) to Dr. Lieberman from the National Institute of Mental Health and the Mental Health Clinical Research Center of Hillside Hospital (MH-41960).
We are indebted to Freddy Howard (Hillside Hospital), Regina Ofiera (Caremark Homecare), and Gil Honigfeld and Peter Jager (Sandoz) for their comments and suggestions.
Source Information
From the Department of Psychiatry, Hillside Hospital, Division of Long Island Jewish Medical Center, Glen Oaks, N.Y. (J.M.J.A., J.A.L., A.Z.S.); the Department of Psychiatry, Albert Einstein College of Medicine, Bronx, N.Y. (J.M.J.A., J.A.L., A.Z.S.); and Sandoz Pharmaceuticals, East Hanover, N.J. (J.L.S., J.A.S.).
Address reprint requests to Dr. Alvir at Hillside Hospital, Long Island Jewish Medical Ctr., 75-59 263rd St., Glen Oaks, NY 11004.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
