Novel Antipsychotics and the Neuroleptic Malignant Syndrome: A Review and Critique

Cases of neuroleptic malignant syndrome related to clozapine or risperidone reported in English were collected (by means of a MEDLINE literature search and contacting pharmaceutical companies as well as authors when necessary) and analyzed against three sets of criteria that represent a continuum of formulations of neuroleptic malignant syndrome: most stringent (Caroff and Mann [5]), intermediary (DSM-IV), and least stringent (Levenson [6]). All criteria sets require adequate workup for exclusion of other etiologies; importantly, the extent of workup is unspecified.

Cases were categorized as high or low probability of being neuroleptic malignant syndrome according to criteria sets, agreement across these criteria, and documentation of adequate, exclusionary workup. For cases designated as having high probability of being neuroleptic malignant syndrome, potential predisposing, precipitating, or contributing factors (use of concomitant medications, rapid titration, age, use of depot neuroleptics, and previous neuroleptic malignant syndrome) were identified.

We identified 19 reports of clozapine-induced neuroleptic malignant syndrome and 13 reports of risperidone-induced neuroleptic malignant syndrome (table 1) with variability in ascribed diagnostic certainty. When we applied the criteria of Caroff and Mann (5), we identified four of the clozapine cases and five of the risperidone cases as being neuroleptic malignant syndrome; when we applied DSM-IV criteria, 10 of the clozapine cases and 10 of the risperidone cases were so identified; when we applied Levenson's criteria (6), 14 of the clozapine cases and 11 of the risperidone cases were so identified. All of the cases identified as being neuroleptic malignant syndrome when we applied Caroff and Mann guidelines were also identified by DSM-IV and Levenson's criteria. All but one (a risperidone case reported by Singer et al. [26]) of the cases identified by DSM-IV also met Levenson's criteria. Four clozapine cases (3 [patients 1 and 2], 9, 13) and two risperidone cases (21, 27) met Levenson's criteria alone. There was consensus for nine clozapine cases (3 [patients 3 and 4], 10, 12, 14, 16, 17 [patient 1], 18, 20) and six risperidone cases (22, 24, 25, 28, 30, 32); of these, five clozapine cases (3 [patients 3 and 4], 16, 18, 20) and one risperidone case (24) failed to fulfill any criteria.

Clozapine cases were designated as having low probability of being neuroleptic malignant syndrome if 1) presentations were not causally linked to clozapine because important differential diagnoses were not clearly excluded and concomitant medical illness (3 [patient 1], 18), and concomitant psychotropic medications (15) were present or 2) they did not fulfill neuroleptic malignant syndrome criteria but were suggestive of possible neurotoxicity of clozapine alone (9, 13, 16) or with polypharmacy (3 [patient 4], 8, 20). Risperidone cases were designated as having low probability if the presentations were attributable to infection (26), neurotoxicity from polypharmacy (31), benzodiazepine withdrawal (21, 31), serotonin syndrome (31), and risperidone-induced extrapyramidal side effects (24, 27).

Nine of 19 clozapine cases were designated as having a high probability of being neuroleptic malignant syndrome, with consensus across criteria sets for four cases (10, 12, 14, 17 [patient 1]). The remaining five cases were close in presentation to typical neuroleptic malignant syndrome, and this was the most parsimonious diagnosis (2 [patient 2], 7, 11, 17 [patient 2], 19). Diagnostic ambiguity remained for three patients (3 [patient 2], 12, 14) with infections of unclear temporal sequence. Eight of 13 risperidone cases were designated as having high probability of being neuroleptic malignant syndrome, with consensus for five cases (22, 25, 28, 30, 32). Most cases had both fever and rigidity, except two (3 [patient 1], 22) with minimal rigidity and two (7, 29) with fevers around 99°F.

Notwithstanding methodological limitations, the following tentative conclusions appear justified: 1) Neuroleptic malignant syndrome can occur during clozapine or risperidone monotherapy. 2) A proportion of published cases of neuroleptic malignant syndrome appear to have implicated clozapine or risperidone inappropriately. 3) The diagnostic conundrum inherent to neuroleptic malignant syndrome is further complicated when atypical antipsychotics are used because side effects are likely to be misattributed to neuroleptic malignant syndrome or because diagnoses of concomitant infection or neurotoxic drug effects may be consistent with the clinical presentation. 4) Insufficient evidence exists currently to support the concept of “atypical” neuroleptic malignant syndrome with novel antipsychotics.

The potential for diagnostic confusion is considerable because of overlap between features of neuroleptic malignant syndrome and adverse effects of atypical antipsychotics. Approximately 3% of patients receiving clozapine develop benign hyperthermia, and approximately 25% have autonomic instability during initial titration of clozapine or risperidone therapy (33). Also, massive creatine phosphokinase elevations with atypical antipsychotics may occur in the absence of other features of neuroleptic malignant syndrome (34). These observations suggest that clinicians should be particularly judicious in ascribing a diagnosis of neuroleptic malignant syndrome (especially early or “incipient” forms) at the initiation of atypical antipsychotic therapy.

Some reports were most likely drug-drug interactions occurring during polypharmacy with lithium, benzodiazepines, antidepressants, or other neuroleptics (8, 15, 21, 31). In several cases (3 [patients 3 and 4], 8, 9, 16, 20, 21, 24, 27), neurotoxicity related to initiation or discontinuation of neuroleptics seemed a more apt explanation. Clarifying risk factors and mechanisms for such adverse reactions and distinguishing these from neuroleptic malignant syndrome is particularly important because treatment of neurotoxicity differs from that of neuroleptic malignant syndrome. Among the high-probability neuroleptic malignant syndrome cases, seven of the nine clozapine-treated patients had schizophrenic or schizoaffective diagnoses and four of the eight risperidone cases had schizoaffective or affective disorders. Concomitant diagnoses of mental retardation were also prominent. This is consistent with findings of high rates of comorbidity and affective disorder in neuroleptic malignant syndrome from typical neuroleptics (35). Three of the nine patients given clozapine (but none given risperidone) had a history of neuroleptic malignant syndrome from typical antipsychotics; there may be a selection bias for clozapine treatment in such patients. Although neuroleptic malignant syndrome patients may tolerate clozapine without recurrence, this clinical practice may over time accrue cases of clozapine-induced neuroleptic malignant syndrome. Four of the clozapine-treated and four of the risperidone-treated patients received concomitant psychotropics; three of the clozapine-treated patients were receiving decanoate preparations before clozapine treatment (with washout periods of 3 weeks to 11 months). Notably, substantial dopamine receptor blockade occurs even at 16 weeks after discontinuation of depot neuroleptics (36). Three of the four patients with clozapine-induced neuroleptic malignant syndrome were rechallenged with clozapine and one with risperidone without subsequent problems. Both patients with risperidone-induced neuroleptic malignant syndrome developed some signs and symptoms short of neuroleptic malignant syndrome when rechallenged with antipsychotics: one had creatine phosphokinase elevation (3293 IU/liter); another had rigidity, fever (99.1°F), and creatine phosphokinase elevation (220 IU/liter). It is important to evaluate carefully whether neuroleptic malignant syndrome can occur with other novel antipsychotics. This process will be advanced by reporting based on recognized diagnostic criteria and by meticulous attention to alternative medical diagnoses and possible neurotoxicities.

Received June 3, 1997; revision received March 9, 1998; accepted March 24, 1998. From the Department of Psychiatry, Case Western Reserve University and University Hospitals of Cleveland; and the Northcoast Behavioral Healthcare System, Cleveland. Address reprint requests to Dr. Buckley, Department of Psychiatry, Case Western Reserve University, 2040 Abington Rd., Cleveland, OH 44106; [email protected] (e-mail).