Background: Aside from 3 small case series (total n=17) commenting on hematological malignancies in clozapine treated patients, the only study to examine this issue was a 2021 analysis of the World Health Organization association pharmacovigilance database (VigiBase) that noted an increased risk of leukemia and lymphoma cases among clozapine treated patients compared to those on structurally similar antipsychotics (olanzapine, quetiapine and loxapine). However, there were significant concerns about the strength of this association given that this database is dependent on spontaneous, anonymized reports from various countries, and thus subject to significant reporting biases. As noted on the VigiBase website: “The VigiBase system is dependent on national centres for the timeliness, completeness, and quality of reports and relies to a large extent on the goodwill of the participating pharmacovigilance centres.”
Given these concerns about reporting bias, investigators performed a case-control analysis of hematological malignancy rates for clozapine compared to non-clozapine antipsychotics using national registry data from Finland. The results of this analysis by Tiihonen et al. appeared online in Lancet Psychiatry on March 22, 2022. The study cohort included all adults admitted for inpatient care with a diagnosis of schizophrenia from 1972-2014, with the analysis of drug exposure confined to the years 2000-20117 to ensure that all patients had a least 5 years of antipsychotic usage data since the creation of the Finnish National Prescription Register in 1995. The diagnosis was based on a range of ICD-10 codes covering lymphoma, leukemias, polycythemia vera, and hemophagocytic lymphohistiocytosis (C81–C96, D45–47, D76, ICD-O-3 behaving as malignant). All cases were matched 1:10 with controls on the basis of demographic factors, duration and dosing of antipsychotic exposure, and prior medical conditions of relevance.
Results: Of the 61,889 individuals in the base analysis, only 375 cases were found, with the following diagnosis distribution: lymphoma 81.3%, leukemia 11.2%, myeloma 5.9%, unspecified 1.6%. During the mean follow-up of 12.3 ± 6.5 years the rate was 61 cases per 100,000 person-years for those treated with clozapine, compared to 41 cases per 100,000 person-years for other antipsychotics (mean follow-up over 12.9 ± 7.2 years). The absolute difference in hematological malignancy rates was thus very small between clozapine and other long-term antipsychotics (difference = 21 per 100,000 person-years of medication use). It is also worth noting that among the 176 cases who died due to a hematological malignancy the mortality rate was 35.9% among those who used clozapine, but 51.1% in the non-clozapine group.
Comment: This analysis removed the reporting bias inherent in the prior pharmacovigilance database study and found a risk difference of 21 cases per 100,000 person-years. One theory for this result is that the small difference is due to surveillance bias, since all clozapine treated patients have routine ANCs. It would also be interesting to see if this finding is replicated from databases of more diverse populations. From the risk difference, the number needed to harm (NNH) is 5000; moreover, given the significantly higher mortality rate in the nonclozapine group from hematological malignancy, the evidence is that clozapine patients, perhaps due to routine surveillance, may have these malignancies captured earlier. Other analyses have noted that patients treated with clozapine have longer life expectancies from natural and unnatural causes providing convincing evidence that clozapine reduces mortality. In conclusion, given the positive benefits of clozapine for treatment resistant schizophrenia, there is no need to change our practices as clozapine saves lives.
Tiihonen J, Tanskanen A, Bell JS, et al: Long-term treatment with clozapine and other antipsychotic drugs and the risk of hematological malignancies in people with schizophrenia: a nationwide case-control and cohort study in Finland. The Lancet Psychiatry 2022; 9(5):353-362.