Original Research Relation of Blood Counts Dur ing Clo zap ine Treat ment to Serum Concentrations of Clo zap ine and Nor-Clozapine L Kola Oyewumi, MB, BS, FRCPC 1, Zack Z Cer novsky, PhD 2, David J Free man, PhD 3, David L Streiner, PhD 4 Ob jec tive: To de ter mine the re la tion be tween se rum clo zap ine and nor- clozapine levels and blood cell counts during clo zap ine treat ment. Method: We un der took a pro spec tive lon gi tu di nal study of 37 con secu tive pa tients with a di ag no sis of schizo phre nia treated with clo zap ine. We ob tained in formed con sent and then de ter mined se rum con cen tra tions of clo zap ine and nor- clozapine weekly. Clo zap ine was ad min is tered daily in di vided doses given every 12 hours and ad justed ac cord ing to clini cal guide lines for its use. Sam ples for se rum con cen tra tions were taken at steady state, im me - di ately be fore the next morning s dose, for 4 to 8 weeks. Com plete blood counts (CBC), weight, and vital signs (that is, blood pres sure, pulse, and tem pera ture) were also moni tored weekly be fore the morn ing s dose of clo zap ine was administered. Re sults: Analy ses of vari ance showed no sig nifi cant changes over the 8- week treat ment course in the ob served mean white blood count (WBC), red blood count (RBC), neutro - phils, and lym pho cytes counts, or in the he mo glo bin and he ma to crit. Only a few weak cor - re la tions (r < 0.21) were found be tween these he ma to logi cal pa rame ters and the meas ures of se rum clo zap ine and nor- clozapine. Con clu sions: The mecha nism of clozapine- induced he ma to tox ic ity at the thera peu tic dos - age range is probably not by direct tox ic ity of clo zap ine or nor- clozapine to the blood cells or their pre cur sors. The for ma tion of the cy to toxic nitrenium com pound from clo zap ine by neu tro phils may be nec es sary. (Can J Psy chia try 2002;47:257 261) Clinical Implications The white blood count (WBC), red blood count (RBC), neu tro phils, and lym pho cytes counts show no sig nifi cant changes within 8 weeks of clo zap ine treatment at therapeutic dosages. There is no remarkable correlation between blood counts and se rum clozapine and norclozapine levels. The clo zap ine dosage is significantly and highly cor re lated with levels of se rum clozapi ne and nor-clozapine, but not with the se rum nor-clozapine clozapine ratio. Limi ta tions The sam ple size was mod er ate. Only data from the first 8 weeks of treatment were in cluded. Cross- cultural replicability is un known. Key Words: clo zap ine, nor- clozapine, agranu lo cy to sis W Can J Psychiatry, Vol 47, No 3, April 2002 257
The Canadian Journal of Psychiatry Original Research Clo zap ine, the pro to type atypi cal an tipsy chotic medi ca - tion, has proven ef fi cacy in cases of treatment- resistant schizo phre nia. How ever, its pro pen sity for agranu lo cy to sis is a major draw back (1). Re cent reports sug gest that clozapineinduced agranu lo cy to sis may be due to an im mune or a toxic pro cess or a com bi na tion of both (2,3). Ac cord ing to Veys and oth ers (4), clozapine- induced neu tro penia may be me di ated by the com bined tox ic ity of clo zap ine and its major me tabo - lite, N- desmethyl clo zap ine (nor- clozapine), against myeloid matu ra tion and mye loid mi totic pro cess. Ger son and oth ers (2,5) found that nor- clozapine was sev eral times more toxic than clo zap ine to he mo poie tic stem cells in cul ture. These find ings stimu lated our search to de ter mine whether thera peu - tic blood lev els of clo zap ine or nor- clozapine, or both, may be as so ci ated with ma jor changes in blood cell counts of pa tients treated with clo zap ine. Cen torrino and oth ers ob served a de crease by up to 60% to 73% in white cell and granu lo cyte counts in a group of 44 pa - tients started on clo zap ine, but they found no posi tive cor re la - tions be tween these dec re ments and drug dos age, se rum clo zap ine or nor- clozapine lev els, or their ra tio (6). In their study, the blood counts were ob tained be fore clo zap ine treat - ment, as well as af ter 2, 6, and 16 weeks of treat ment. How - ever, it is pos si ble that ma jor changes in he ma to logi cal pa rame ters oc cur al ready at the end of the first week of treat - ment. It is also pos si ble that the com plete se quence of bio - chemi cal cop ing, during which the or gan ism ini tially adjusts to clo zap ine and its me tabo lites, can be de ter mined only if weekly blood counts are ana lyzed during the first 7 to 8 treat - ment weeks. It is then that the clo zap ine dos age is typi cally in - creased, as rec om mended in the usual clini cal guide lines. Our study used weekly meas ures to evalu ate blood count changes dur ing the first 8 weeks of clo zap ine treat ment. Method Plasma clo zap ine and nor- clozapine lev els were de ter mined us ing a High Per form ance Liquid Chro ma tog ra phy (HPLC), as pre vi ously de scribed by our group (7). Thirty- seven adults ad mit ted to the clini cal evalua tion unit of a pro vin cial psy chi - at ric hos pi tal with a di ag no sis of treatment- refractory schizo - phre nia en rolled in the study af ter giv ing in formed con sent for clo zap ine treat ment and ve ne punc ture for re lated blood tests. The sam ple com prised 26 (70.3%) men and 11 (29.7%) women whose mean age was 35.2 years (SD 10.2, range 18 to 57). Twenty- nine (78.4%) were smok ers, and 7 (18.9%) had a his tory of re cent can na bis use. All pa tients com pleted the first 4 weeks of the study. How ever, only 30 of the 37 pa tients com - pleted 8 weeks. This sub group in cludes pa tients for whom some of the clo zap ine and nor- clozapine meas ures from week 4, 5, 6, and 7 were not avail able. Clo zap ine treatment was ini ti ated at a dosage of 25 to 50 mg daily in di vided doses given every 12 hours. If tol er ated and clini cally in di cated, the dosage was in creased by 25 mg every 12 hours 3 times a week until clini cal im prove ment or doselimiting side ef fects pre cluded any further in crease. Two blood sam ples of 5 ml each were ob tained weekly for com - plete blood count (CBC) and de ter mi na tion of plasma clozap - ine and nor- clozapine. Sam ples were col lected at trough con cen tra tion in the morning im me di ately be fore the next dose of clo zap ine was given. No dos age change was made within at least 3 days of blood work to en sure a steady state based on a clo zap ine half- life of 16 hours (8). Plasma concen - tra tions of clo zap ine and nor- clozapine were moni tored over the first 4 to 8 weeks of ther apy, and thereafter as needed to op - ti mize bene fit and monitor com pli ance. None of the 37 pa - tients re ceived any regu lar con comi tant medi ca tion during our study: all were given clo zap ine only. Lo raze pam 2 mg daily was used spar ingly as needed for agitation and ag gres - sive be hav iour. Al though the study de sign ini tially ap peared suit able for the repeated- measures analy sis of variance (re peated meas ures ANOVA), our pre limi nary tests of spheric ity de tected sig nifi - cant vio la tion of sta tis ti cal as sump tions for the re peated meas - ures ANOVA on some of the key vari ables. For this rea son, we assessed the weekly changes in he ma to logi cal pa rame ters us ing ANOVA only, which pro vides more con ser va tive es ti mates. Re sults Clo zap ine Dosage and Blood Con cen tra tion As shown in Table 1, there was a cor re spond ing in crease in se - rum clo zap ine and nor- clozapine as the clo zap ine dos age in - creased. The overall change in clo zap ine dosage and in its blood con cen tra tion in this data set is highly sig nifi cant (Pear - son cor re la tion: r = 0.76 for clo zap ine dos age, r = 0.60 for se - rum clo zap ine, and r = 0.64 for se rum nor- clozapine; P < 0.001). The changes in the ra tio of se rum nor- clozapine to se rum clo - zap ine were not sig nifi cant (Pearson cor re la tion, r = 0.09, P = 0.186). Since the Pear son cor re la tion tech nique can de tect only lin ear trends, we cal cu lated an ANOVA to evalu ate the trends in the ra tio of se rum nor- clozapine to se rum clo zap ine over the 8 weeks. The re sults were non sig nifi cant (P = 0.662). The clo zap ine dos age was sig nifi cantly and highly cor re lated with the blood level of se rum clo zap ine (r = 0.64) and se rum nor- clozapine (r = 0.68) but not with the ra tio of se rum norclo zap ine to se rum clo zap ine (r = 0.09, P = 0.182). Changes in Blood Count Meas ures From Base line to Week 8 Ta ble 1 shows the av er age blood count meas ures at base line (week 0) and sub se quently at each of the 8 weeks. To evalu ate 258 W Can J Psychiatry, Vol 47, No 3, April 2002
Relation of Blood Counts During Clozapine Treatment to Serum Concentrations of Clozapine and Nor-Clozapine Table 1 Average levels of clozapine and of blood cell count Variables by weeks of treatment Weeks 0 1 2 3 4 5 6 7 8 n 37 34 36 36 35 33 30 29 28 Clozapine, average daily dose (mg): None 97.3 (55.2) 218.9 (85.1) 296.6 (103.8) 364.9 (126.9) 42 (157.3) 457.8 (177.9) 486.2 (182.7) 468.3 (179.3) Clozapine concentration (ng/ml): None 63.5 (46.0) 199.3 (149.1) 251.9 (178.0) 300.7 (200.2) 316.2 (189.4) 364.7 (195.6) 351.9 (176.2) 379.5 (156.5) Nor-clozapine concentration (ng/ml): None 38.7 (30.8) 118.1 (79.4) 156.8 (105.4) 202.9 (111.3) 220.7 (136.2) 244.6 (126.4) 246.2 (129.0) 249.6 (94.7) Ratio nor-clozapine clozapine (ng/ml): na 0.65 (0.31) 0.65 (0.27) 0.67 (0.21) 0.74 (0.23) 0.69 (0.13) 0.71 (0.18) 0.70 (0.18) 0.70 (0.16) Total WBC count (10 9 /L): 8.5 (2.8) 7.6 (1.8) 7.6 (2.1) 8.2 8.2 7.9 (2.3) 8.4 (2.9) 8.0 8.7 (2.5) Granulocytes (10 9 /L): 6.0 (2.5) (1.7) 5.4 (2.2) 5.8 (2.3) 5.7 () 5.3 (1.9) 5.9 (2.6) 5.6 (2.2) 6.1 (2.1) Lymphocytes (10 9 /L): (0.7) 2.1 1.7 1.9 (0.7) 1.9 2.1 RBC (10 12 /L): 4.9 4.9 (0.4) Platelets (10 9 /L): 253.6 (62.3) 250.9 (56.6) 246.9 (59.9) 283.8 (74.4) 300.5 (97.0) 276.5 (74.8) 274.4 (73.1) 268.7 (67.9) 273.8 (82.9) WBC = white blood cell; RBC = red blood cell. changes in blood pa rame ters from base line to week 8 of clo - zap ine treat ment, ANOVA was cal cu lated for the WBC, neu - tro phils, lym pho cytes, RBC, he mo glo bin, he ma to crits, and plate lets. None of the changes were sig nifi cant at the level of P < 0.01. Al though these ex plora tory analy ses in volved nu mer - ous cal cu la tions, some re search ers may fa vour a less strin gent cri te rion of sig nifi cance at P < 0.05. How ever, even with this cri te rion, the only sig nifi cant change de tected was with re - spect to plate lets. Tukey post hoc tests showed that the sig nifi - cant dif fer ence (P = 0.042) in the platelet count when on clo zap ine was be tween week 2 (mean count 246.9, SD 59.9) and week 4 (mean count 300.5, SD 97.0). The un der ly ing trend is weak (Pear son r = 0.32). Blood Count Meas ures With Clo zap ine Dos age and Serum Con cen tra tion as Co vari ates Since the in ter in di vid ual vari abil ity in the dos age of clo zap - ine, se rum clo zap ine, or nor- clozapine lev els within each time slot could obscure the nature of the changes in he ma to logi cal pa rame ters, we cal cu lated analy ses of co vari ance (AN CO - VAs) for the he ma to logi cal pa rame ters with clo zap ine vari - ables as co vari ates. None of these nu mer ous com pu ta tions led to the dis cov ery of sig nifi cant changes in blood counts (P > 0.01). The Mag ni tude of the Re la tion Be tween Clo zap ine and Blood Count We used Pearson cor re la tion co ef fi cients to evalu ate the mag - ni tude of the re la tion of vari ous meas ures of clo zap ine or norclozapine to he ma to logi cal pa rame ters (see Ta ble 2). The cor - re la tions are non sig nifi cant (P > 0.01), ex cept for the cor re la - tions of nor- clozapine con cen tra tion to lym pho cyte count and of the nor- clozapine clozapine con cen tra tion ra tio to RBC and to plate lets. None the less, all these are very weak (r < 0.21). These cor re la tion co ef fi cients sug gested that a higher nor- clozapine con cen tra tion was as so ci ated with lower lym - pho cyte counts and that a high ra tio of nor- clozapine to clo - zap ine was as so ci ated with lower plate let counts and higher RBC. When we applied the Bon fer roni cor rec tion, how ever, only the cor re la tions of RBC and plate lets to the norclozapine clozapine ra tio re mained sig nifi cant. These re sults are con sis tent with re cent findings by Cen torrino and oth ers (6), in which no posi tive relation be tween clo zap ine meas ures and he ma to logi cal pa rame ters was found. Al though we cal cu lated these cor re la tions, we are also aware of their sta tis ti cal short com ings. When in ter pret ing these co - ef fi cients, it is im por tant to con sider that the dosage of 480 mg at week 7 (reached af ter the medi ca tion had gradu ally been W Can J Psychiatry, Vol 47, No 3, April 2002 259
The Canadian Journal of Psychiatry Original Research Table 2 Correlations between clozapine variables and blood cell counts Clozapine variables Blood cell counts White blood countl Neutrophils Lymphocytes Red blood count Platelet Average daily dose (mg) 0.03 0.01 0.07 0.07 0.11 Clozapine concentration (ng/ml) 0.08 0.04 0.15 0.01 0.07 Nor-clozapine concentration (ng/ml) 0.07 0.02 0.16 a 0.06 0.02 Nor - clozapine Ratio: Clozapine 0.04 0.00 0.12 0.20 a 0.20 a a Sig nifi cant at P < 0.005 (l- tailed) in tro duced with a mod er ate weekly in crease) had probably a very dif fer ent im pact on the in tra in di vid ual bio chemi cal pro - cesses than would the same dos age have had at the first week. The cor re la tional ap proach fails to correct for this. How ever, when we cal cu lated the cor re la tions of clo zap ine vari ables to WBC, RBC, lym pho cytes, neu tro phils, and plate lets sepa - rately for each of the 8 treatment weeks to avoid this con tami - na tion, the cor re la tions still re mained un re mark able and rather er ratic. None of these co ef fi cients re mained sig nifi cant af ter we ap plied the Bon fer roni cor rec tion. Weight Gain in the First 8 Weeks of Clo zap ine Treat ment We re corded weight meas ures be fore treat ment and at each of the 8 weeks of treat ment. There was no sta tis ti cally sig nifi cant dif fer ence among these measures over the dif fer ent time points (ANOVA, P > 0.05). Dis cus sion We ob served no sig nifi cant changes in WBC, RBC, lym pho - cytes, and neu tro phils measures dur ing the 8- week course of clo zap ine treat ment. A mi nor in crease in the platelet counts was noted from week 2 to week 4. The trend is only weak, and its rep li ca bil ity and clini cal sig nifi cance need to be ex plored in fu ture stud ies. Of in ter est are also the cor re la tions of levels of se rum clo zap ine to he ma to logi cal pa rame ters. Only er ratic cor re la tional re la tions were ob served be tween se rum clo zap - ine meas ures and blood counts in our study, and this re mained true even when we con trolled for the ef fect of pre vi ous expo - sure to clo zap ine by re cal cu lat ing the cor re la tion matrix sepa - rately for each of the 8 weeks of treat ment. Clozapine- induced agranu lo cy to sis has a de layed on set. Ap - proxi mately 76% of cases oc cur be tween the fourth and eight - eenth week of ther apy, with peak in ci dence at the tenth week (2,9). This 8- week study al lowed us to study pa tients in both the first 4 weeks of mini mal risk and the next 4 weeks of in - creased risk. Leu co penia was ob served in only 1 pa tient dur - ing the 8 weeks of treat ment, the low est value being 3.4 x 10 9 /L. Pe riph eral blood cells are de rived from pre cur sors in the bone mar row. The WBC and granu lo cytes have a rela tively shorter matu ra tion pe riod of 4 to 7 days and mean tran sit time in pe - riph eral blood of 8 to 9 hours, com pared with tran sit times of 120 days for RBC and 7 to10 days for plate lets. The finding that, over the 8 weeks of study, lym pho cytes and RBC counts re mained relatively un changed and platelets in creased be - tween weeks 2 and 4 may sug gest a dif fer en tial ef fect of clo - zap ine on the bone mar row mul ti line age cells. The sites of these tran sient clozapine- induced changes in pe riph eral blood cells are proba bly 1) sup pres sion of the colony- forming unit (CFU)-GM stem cells, which are mye loid pre cur sors, and 2) stimu la tion of the CFU- Meg and Mega kar yo cyte pre cur sors. The ef fect on CFU-E and erythroid pre cur sors is un clear from this 8- week study be cause of the long RBC tran sit time (120 days). Iden ti fied risk fac tors (2,9) for clozapine- induced agranu lo - cy to sis in clude ex po sure to clo zap ine treat ment (es pe cially weeks 4 to 18 of ex po sure), in creas ing age, fe male sex, and Hu man Leu co cyte An ti gens (HLA)-B38, -DR4, -DQW3 in Ash ke nazi Jews. Ger son and Melt zer (2) suggested that a toxic me tabo lite of clo zap ine may be re spon si ble for clozapine- induced agranu lo cy to sis. They also rec om mended pro spec tive studies to evalu ate whether high me tabo lite con - cen tra tions in sen si tive in di vidu als are re spon si ble for the agranu lo cy to sis. Our finding could not con firm any di rect tox - ic ity of clo zap ine or its me tabo lites at thera peu tic dosages. In vi tro stud ies by Ger son and oth ers (2,5) found that the mean con cen tra tion of clo zap ine that in hib ited growth of the mye - loid CFU (CFU- GM) by 50% was 32 mg/l, and that of norclozapine was 3.2 mg/l. These con cen tra tions are many times higher than those in typi cal clini cal se rum sam ples. For ex am - ple, in our sam ple the high est mean nor- clozapine con cen tra - tion, at week 8, was 249.6 ng/ml (SD 94.7). Al though we used a pro spec tive lon gi tu di nal ap proach, our find ings cor robo - rated a re cent re port that used a cross-sec tional ap proach (6). Both studies found no relation be tween WBC and granu lo cyte counts and any of the clo zap ine vari ables stud ied. These 260 W Can J Psychiatry, Vol 47, No 3, April 2002
Relation of Blood Counts During Clozapine Treatment to Serum Concentrations of Clozapine and Nor-Clozapine find ings sug gest that un der nor mal clini cal con di tions clo zap - ine and nor- clozapine are not di rectly re spon si ble for clozapine- induced agranu lo cy to sis. How ever, as Uetrecht has pro posed, re ac tive me tabo lites such as the nitrenium ion formed by neu tro phil or its pre cur sor could be re spon si ble for clozapine- induced agranu lo cy to sis, either by di rect tox ic ity or through an immune- mediated mecha nism (10). Clini cally, we have ob served that when CBCs show leu co - penia in some of our pa tients, the ra tio of nor- clozapine to clo - zap ine is usu ally above 0.7. The sig nifi cance of this is un clear at this time, but we sug gest that fu ture studies on the re la tion of clo zap ine con cen tra tion and that of its me tabo lites to he ma - to logi cal pa rame ters use a lon gi tu di nal ap proach span ning the first 6 months of clo zap ine treat ment, which is the es ti mated pe riod of maxi mum risk for clozapine- induced agranu lo cy to sis. References 1. Lie ber man JA, Kane JM, Johns CA. Clo zap ine: Guide lines for Clini cal Man age - ment. J Clin Psy chia try 1989;50:9 13. 2. Ger son SL, Melt zer H. Mecha nisms of clozapine- induced agranu lo cy to sis. Drug Safety 1992;7 (Suppl.1):17 25. 3. Pis ci otta AV, Kon ings SA, Cie semier LL, Cronkite CE, Lie ber man JA. On the pos si ble mecha nisms and pre dict abil ity of clozapine- induced agranu lo cy to sis. Drug Safety 1992;7 (Suppl.1):33 44. 4. Veys PA, Wilkes S, Shah S, Noyelle R, Hoffbrand AV. Clini cal ex pe ri ence of clo zap ine- in duced neu tro penia in the UK: labo ra tory in ves ti ga tion us ing liq uid cul ture sys tems and immuno- fenorocytometry. Drug Safety 1992;7 (Suppl. 1):26 32 5. Ger son SL, Arce C, Melt zer HY. N- desmethyl clo zap ine: A clo zap ine me tabo lite that sup presses Hae mo poie sis. Br J He ma tol 1994;86:555 61. 6. Cen torrino F, Bald es sarini RJ, Flood JG, Kando JC, Frank en burg FR. Re la tion of leu co cyte counts dur ing clo zap ine treat ment to se rum con cen tra tion of clo zap ine and me tabo lites. Am J Psy chia try 1995;152:610 2. 7. Free man DJ, Li MC, Oyewumi LK. Solid- phase ex trac tion and HPLC analy sis of clo zap ine and nor- clozapine in hu man plasma. Ther Drug Monit 1996;18:668 92. 8. Choc MG, Lehr RG, Hsuan F, Honig feld G, Smith HT. Mul ti ple dose phar ma - coki net ics of clo zap ine in pa tients. Phar ma ceu tic Res 1987;4:402 5. 9. Krupp P, Bar nes P. Clozapine- associated agranu lo cy to sis: risk and ae ti ol ogy. Br J Psy chia try 1992;160 (Suppl 17):38 40. 10. Uetrecht JP. Me tabo lism of clo zap ine by neu tro phils. Pos si ble im pli ca tions for clozapine- induced agranu lo cy to sis. Drug Safety 1992;7 (Suppl 1):51 6. Manuscript received April 2001, revised, and accepted February 2002. 1 Professor of Psychiatry, Queen s University, Kingston, Ontario. 2 Professor of Psychiatry, University of Western Ontario, London, On tario. 3 Associate Professor, Faculty of Medicine, University of Western Ontario, London, Ontario. 4 Director, Kunin-Lunefeld Applied Research Unit, Toronto, Ontario. Address for correspondence: Dr LK Oyewumi, Hotel Dieu Hospital, 166 Brock St, Kingston, ON K7L 5G2 e-mail: oyewumik@post.queensu.ca Résumé: Relation entre les numérations globulaires durant le traitement à la clozapine et les concentrations sériques de clozapine et de nor-clozapine Ob jec tif : Dé ter mi ner la re la tion en tre les ni veaux sé ri ques de clo zap ine et de nor- clozapine, et les numé ra tions globu laires du rant le traite ment à la clo zap ine. Méth ode : Nous avons en tre pris une étude lon gi tu di nale pro spec tive de 37 pa tients consé cu tifs pré - sen tant un di ag nos tic de schizo phré nie et traités à la clo zap ine. Nous avons ob tenu leur con sen te ment éclairé, puis avons dé ter miné chaque se maine les con cen tra tions sé ri ques de clo zap ine et de norclozapine. La clo zap ine était ad min is trée chaque jour en doses frac tionnées à toutes les 12 heures et était ra justée selon les lig nes di rec tri ces clin iques qui en ré gis sent l u tili sa tion. Les échan til lons des con cen tra tions sé ri ques étaient pré levés à un état sta tion naire, im médi ate ment avant la dose du matin, pen dant 4 à 8 se maines. Les numé ra tions globu laires com plètes (NGC), le poids et les sig nes vitaux (ten sion arté ri elle, pouls et tempé ra ture) étaient aussi vé ri fiés chaque se maine avant l ad min is tra tion de la dose du matin de clo zap ine. Résul tats : Les analy ses de variance n ont in diqué au cun change ment sig ni fi ca tif, au cours du traite - ment de 8 se maines, dans la moy enne ob servée de la numé ra tion des glob ules blancs (NGB), de la numé ra tion des glob ules rouges (NGR), des polynu cléaires neu tro philes et de la numé ra tion lym pho - cy taire, ou dans l hémo globine et l héma to crite. Seules quelques fai bles corré la tions (r < 0,21) ont été con statées en tre ces paramètres héma tolo giques et les me sures sé ri ques de clo zap ine et de norclozapine. Con clu sions : Le mé can isme d héma to tox ic ité in duite par la clo zap ine à l é chelle de dosage thé ra - peu tique n est prob able ment pas par in toxi ca tion di recte des glob ules san guins ou de leurs pré curseurs par la clo zap ine ou la nor- clozapine. La for ma tion d un com posé ni tré nium cy to toxique par les polynu cléaires neu tro philes à par tir de la clo zap ine peut être né ces saire. W Can J Psychiatry, Vol 47, No 3, April 2002 261