Abstract
Charles M. Beasley, Lilly Research Laboratories, Indiana, Indianapolis:
We would like to bring the readers’ attention to several points relevant to the recent letter by Professor Tim Lambert titled ‘Olanzapine after clozapine: the rare case of prolongation of granulocytopenia’ [1]. Lambert appears to accept the case observations of Flynn et al. [2] as demonstrating a causal relationship between olanzapine therapy and prolongation of neutropenia associated with clozapine therapy, when olanzapine is initiated immediately upon discontinuation of clozapine. He refers to the cases of Flynn et al. [2] as follows: ‘The finding that olanzapine may further suppress and also prolong agranulocytosis in clozapine discontinuers suggests that extreme caution should be applied when introducing olanzapine in these situations’.
We would suggest that Lambert [1] and Flynn et al. [2] should have discussed the possibility that the observations of Flynn etal. [2] were chance findings and not indicative of any pathophysiologic process being caused by olanzapine, especially in light of the literature not cited in either letter describing the temporal course of resolution of neutropenia associated with clozapine. Flynn etal. [2] described three patients treated with olanzapine following development of neutropenia associated with clozapine treatment, instituted immediately following discontinuation of clozapine. The neutropenia in these three patients required between 17 and 24 days (mean = 21 days) to resolve. These three patients were compared with 12 consecutive patients who experienced neutropenia associated with clozapine, clozapine was discontinued, and olanzapine was not instituted. For these 12 patients, time to recovery ranged from 1 to 9days (mean = 3 days).
Obviously, this case series was not the result of a blinded treatment comparison. More importantly, receiving olanzapine or not was not a matter of random assignment. Therefore, these observations cannot be considered a study and the group of 12 patients must be considered a historical control group. Use of historical controls, rather than random assignment controls, introduces the possibility of multiple unrecognised biases in comparisons. Other groups developing neutropenia on clozapine with their time to recovery have been published in the context of describing recovery in patients treated with granulocyte colony stimulating factor (G-CSF) compared to those not receiving such therapy. The reported groups not receiving G-CSF could have served equally as well as controls but they were not cited by Flynn et al. [2] or Lambert [1]. They are in Table 1.
Neutropenia: time to recovery in days
Based on the data in Table 1, Flynn etal.'s [2] observation with respect to clozapine would appear to be atypical (mean = 3 days) and the time taken to recovery among olanzapine-treated patients (mean = 21 days), although longer than that observed by Chengappa et al. [3], Gerson et al. [4] and Atkin et al. [5], is within the range of observed variance with clozapine alone.
Lambert suggests that these observations may reflect a risk of initiating any antipsychotic with discontinuation of clozapine or reflect an increased risk of initiating olanzapine with discontinuation of clozapine due to the pharmacologic similarities between olanzapine or clozapine. Although the precise mechanism of human neutropenia associated with clozapine is unknown, recent studies have not demonstrated the in vitro toxicity with olanzapine shown by clozapine [6, 7].
This leads us to our major concern. Given the above caveats and data, we believe the introductory paragraph of Lambert's letter [1]:
‘This letter is to alert physicians to a potentially major haematological problem with the atypical neuroleptic olanzapine.'
to be unduly alarming in tone. It is important to consider case reports as hypothesis generators and maintain careful scrutiny of safety with all medications. However, we must also be balanced and place our data in their objective context.
Finally, the published letter contains two typographical errors. Reference 6, the letter by Flynn et al. [2] published in the Journal of Clinical Psychopharmacology which Lambert's letter recounts, is cited as being published in 1977. It was actually published in 1997. Lambert summarises the letter of Flynn et al. [2] as follows: ‘The mean time to recovery in patients treated with olanzapine (immediately following development of neutropenia associated with clozapine) was 24.3 days while those not treated with clozapine was 3.4 days (range = 1–9)’. The reference to ‘…not treated with clozapine…’ should have read ‘…not treated with olanzapine…’.