J Clin Pathol: first published as 10.1136/jcp.40.6.663 on 1 June 1987. Downloaded from J Clin Pathol 1987;40:663-669

Cytochemical profile of megakaryoblastic leukaemia: a study with cytochemical methods, monoclonal antibodies, and ultrastructural cytochemistry

MARIA S POMBO DE OLIVEIRA, CAROLYN GREGORY, ESTELA MATUTES, A PARREIRA, D CATOVSKY From the Leukaemia Unit, Royal Postgraduate Medical School, London

SUMMARY A cytochemical study using: Sudan black B; a-naphthyl acetate (ANAE) staining; estimation of cx-naphthyl butyrate (ANBE) esterase activity; acid phosphatase activity; and 5'nucleotidase activity was carried out in 15 cases of megakaryoblastic leukaemia. These included cases of M7 acute myeloid leukaemia and blast crises of chronic granulocytic leukaemia. The megakaryoblastic nature of the blasts was first established using two monoclonal antibodies against glycoproteins, and by estimating the platelet/peroxidase reaction at ultrastructural level. Our findings suggest that megakaryoblasts have a typical cytochemical profile comprising positive ANAE staining and acid phosphatase activity with a predominant localisation in the Golgi zone or and negative weak ANBE activity. A similar positive cytochemical pattem was also found in five copyright. cases of erythroleukaemia (M6). The specificity of the 5'nucleotidase activity for megakaryoblasts was not confirmed. In most cases of megakaryoblastic leukaemia there was no 5'nucleotidase activity only two cases showed positive reactions-reactions were positive in several cases of myeloblastic and lymphoblastic leukaemia. We suggest that cytochemical methods may be useful in diagnosing M6 and M7 acute leukaemia because less than 40% ofleukaemic cells react with specific monoclonal antibodies. http://jcp.bmj.com/

Megakaryocytes and their precursors influence the these cells has not previously been studied in detail pathogenesis of such diseases as acute megakaryo- because methods for their identification have only blastic leukaemia(M7), blast transformation of recently been developed. chronic granulocytic leukaemia, acute myelofibrosis The aim of this study was to see if megakaryoblasts and myelodysplastic syndromes. In 1973 Breton- have a characteristic cytochemical profile at light on September 28, 2021 by guest. Protected Gorius et al used ultrastructural cytochemistry to microscopical examination using five cytochemical measure platelet peroxidase activity with distinct tests. We studied cases that were shown to be predom- characteristics from myeloperoxidase in early mega- inantly or exclusively megakaryoblastic on mea- karyoblasts and suggested platelet peroxidase as a surement of platelet peroxidase activity and when marker for megakaryoblastic leukaemia.' - The several monoclonal antibodies, including two against development of membrane markers subsequently led platelet glycoproteins, were used. to the discovery of specific monoclonal antibodies against platelet glycoproteins, and these are now used Material and methods to diagnose megakaryoblastic proliferations.46 Megakaryocytic malignancy is difficult to recognise Specimens of peripheral or bone marrow, or by light microscopical examination because the mor- both, were taken for analysis from 15 adult patients phological characteristics of the blast cells are not with acute leukaemia, median 46 years (range 26 to 77 clearly defined and resemble either lymphoblasts or years). These included eight cases of recently diag- undifferentiated cells. The cytochemical profile of nosed acute megakaryoblastic leukaemia (M7 accord- ing to the FAB criteria)7 and seven cases of chronic Accepted for publication 21 January 1987 granulocytic leukaemia in blast crisis. They were 663 J Clin Pathol: first published as 10.1136/jcp.40.6.663 on 1 June 1987. Downloaded from

664 Oliveira, Gregory, Matutes, Parreira, Catovsky Table 1 Summary ofresults in 15 cases ofleukaemia with megakaryoblasts CGL-BC, megakaryoblastic Case I Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Age/sex 31/F 48/M 38/M 74/F 34/F 45/M 37/M White blood count (109/l) 46 90 40 76 90 20 26 Blast (% in peripheral blood) 78 68 67 65 80 20 46 LM cytochemistry (% positive blast cells) Sudan black B 14+/- Negative 14+/- 8+/- Negative Negative Negative ANAE 60+ + 33+ 30+ 31+ + 100+ + 16+ + 46+ + ANBE 36 + /- 20 + /- Negative Negative 15 + /- Negative Negative Acidphosphatase 64++ 63++ 60++ 20++ 100++ 20+ 40++ 5'nucleotidase Negative Negative Negative Negative ND Negative Negative Platelet peroxidase (electron microscopy level) + + ND ND + + Negative ND ND Monoclonal antibodies (% positive blast cells) GRBI ND ND ND 30 26 61- 47 3C5 12 ND ND 15 ND ND 38 My9 51 21 Negative Negative 49 8 62 C17 Immunofluorescence 31 7 2 22 13 ND 20 Immunofluorescence 23 35 12 24 16 24 24 AN51 Immunofluorescence 14 1 ND 20 Negative ND ND Immunoperoxidase 5 30 20 ND Negative 5 23 copyright. ND = not done. Intensity of cytochemical reactions: +/-, weak; +, moderate; + +, strong. ANAE, alpha-naphthyl acetate esterase ANBE, alpha-naphthyl butyrate esterase. selected from a series of over 150 cases submitted to determine the pattern, strength, and percentage of our laboratory for marker studies of reactivity with positive blasts. http://jcp.bmj.com/ monoclonal antibodies against platelet glycoproteins: many were also tested for platelet peroxidase activity. CELLS MARKER ANALYSIS Mononuclear cells from peripheral blood bone mar- MORPHOLOGY AND CYTOCHEMISTRY row, or both, were isolated by Lymphoprep. An indi- Peripheral blood or bone marrow films, or both, and rect immunofluorescence technique was performed cytospin slides, were stained using May-Grunwald- on cell suspensions using monoclonal antibodies, fol- Giemsa stain, and Sudan black B stain, by Scheehan lowed by a second layer of a fluorescein iso- and Storey's technique,8 Alpha naphthyl acetate thiocyanate conjugated goat antimouse F(ab)2 serum on September 28, 2021 by guest. Protected esterase (ANAE) and ax-naphthyl butyrate esterase (Cappel Laboratories). The cells were washed in (ANBE), activity were estimated by the method of Li buffer resuspended in phosphate buffered saline and et al,9 using a-naphthyl acetate or a-naphthyl glycerol (50/50), mounted on slides, and examined butyrate as substrates and hexazotised pararosanilin under a Zeiss 14 fluorescence microscope. A control as coupler; acid phosphatase activity was estimated in which the first layer monoclonal antibody was by the Goldberg and Barka technique,'0 with naphtol replaced by non-immune mouse ascites was included AS-BI-phosphate as substrate and hexazotised para- in all samples. AB serum at a dilution of 2% was used rosanilin as coupler; 5'nucleotidase activity was mea- throughout to block non-specific binding to Fc recep- sured by the method of El Mohandes and Hayhoe," tors. The following mouse monoclonal antibodies in which slides were incubated for six hours at 37°C in were used: GRB1 (IgG2a), gift from Professor a mixture containing adenosine-5-monophosphate. Garrido-Torres, against class II MHC antigens (anti- A control slide was incubated in a reaction mixture Ia); 3C5 (IgG1; CD34), which identifies an antigen containing B glycerophosphate instead of adenosine present in early haematopoietic precursor cells'2; 5'monophosphate, as this substrate is split by alkaline My9 (IgG2; CD33), which is reactive with immature phosphatase, but not by 5'nucleotidase. myeloid and monocytic cells'3; B4 (CD19) against All slides were examined by light microscopy to early acute lymphoblastic leukaemia blast cells and B J Clin Pathol: first published as 10.1136/jcp.40.6.663 on 1 June 1987. Downloaded from Cytochemistry ofmegakaryoblasts 665

Megakaryoblastic leukaemia, M7 Case 8 Case 9 Case 10 Case 11 Case 12 Case 13 Case 14 Case 15 26/M 49/F 46/M 50/M 79/M 33/M 30/M 77/F 3-4 1 1 4-6 50 1-6 7-8 2-9 71-9 65 45 68 89 65 89 35 80 Negative 4 + /- Negative Negative Negative Negative Negative 7 + /- 50+/+ + 52+ + 75+ + 65+ + 45+ + 100+ + 63+ 55 Negative 15 + /- 25 + /- Negative Negative Negative ND ND 53+ 30+ 75+ 80+/- 28+ + 80+ + 60+ + 45+ + Negative Negative Negative Negative Negative 30+ + ND 18+ +

++ * + + + +/* ND + + ND + +/* ND 52 51 42 ND 42 36 68 48 + 5 42 48 ND 24 67 65 13 39 48 68 ND 14 ND 43 48 12 3 24 ND 14 ND 3 20 45 4 31 17 15 21 ND 25 ND ND 17 ND 10 ND Negative 20 23 ND 22 12 10 7 ND

*MPO positive in a small proportion of cells. copyright.

lymphocytes was tested in five cases; J5 (CD 10) (Cou- estimated with 3-3' diamino-benzidine tet- lter clone) against the common acute lymphoblastic rahydrochloride in phosphate buffered saline for 10 leukaemia antigen; antiglycophorin A (LICR minutes. The slides were counterstained with hae- LON/R10)'" or Gero (early erythroid cells; Gerbich matoxylin for one minute, dried, and mounted in http://jcp.bmj.com/ blood group) were used to recognise erythroid precur- DPX. sors; C17 (IgGI; CDw41) against platelet gly- coprotein Illa"5; and ANSI (IgG; CDw42), which TRANSMISSION ELECTRON MICROSCOPY recognises an antigen on glycoprotein lb,'6 were used Mononuclear cells were fixed for ultrastructural to show megakaryoblasts. Terminal deoxynucleotidyl analysis by transmission electron microscopy in 3% transferase (TdT) activity was estimated using an glutaraldehyde in 01 M phosphate buffered saline indirect immunofluorescence method described by (pH 7 4) at room temperature for 90 minutes. The on September 28, 2021 by guest. Protected Bollum.'7 myeloperoxidase reaction was estimated on cells pre- viously fixed in 3% glutaraldehyde in 0-1 M phos- phate buffered saline for 30 minutes at 4°C, by the IMMUNOCYTOCHEMISTRY Graham and Karnovsky method, and the platelet The reactivity of C17 and ANS1 were also examined peroxidase reaction on unfixed cells by Roel's using an indirect immunoperoxidase technique, method.2 Osmication, dehydration, and embedding slightly modified from that of Stein et al."8 Briefly, were done using standard methods. cells on Cytospin slides were fixed in acetone for 10 minutes and endogenous peroxidase blocked by Results immersing them in chloroform for 10 minutes, both at room temperature. After washing slides were incu- MORPHOLOGY bated with 2% complement inactivated AB serum, All the patients except case 6 had high percentages- and then with the first monoclonal antibody, for 30 between 45 and 89%-ofblast cells in their peripheral minutes. Two layers of peroxidase conjugated anti- blood (table 1), and often bone marrow could not be body were used (rabbit antimouse and swine anti- aspirated. The leukaemic cells were predominantly rabbit). After washing the peroxidase reaction was immature blast cells with some polymorphism, rang- J Clin Pathol: first published as 10.1136/jcp.40.6.663 on 1 June 1987. Downloaded from

666 Oliveira, Gregory, Matutes, Parreira, Catovskj,

Fig I Blast (cells showi'ing scanmt X cvtoplasm,fine1v reticulated chromatin, and one or two prominent nucleoli. Fig 2 Blast cell releasing (case 8). Fig 3 Small blast cell (case 9) showing high nucleocytoplasmic ratio, heavy nuclear chromatin, and Cv11toplasmic blebs.

Fig I copyright. http://jcp.bmj.com/

Fig 2 Fig 3 on September 28, 2021 by guest. Protected ing from small cells with a high nucleocytoplasmic case, occasional Auer Rods. That these blast cells ratio resembling lymphoblasts, to medium and large were myeloid was confirmed by Sudan black B stain- cells with agranular and moderately blue cytoplasm. ing on light microscopy and the myeloperoxidase The nuclei were round or oval, with a dense nuclear reaction on transmission electron microscopy. chromatin in the small blasts; the large cells had finely reticulated nuclear chromatin with regular outlines CYTOCHEMISTRY and between one and three prominent nucleoli (figs Table 1 summarises the results. The Sudan black B 1-3). In every case there was a small proportion of stain reaction was negative in 10 of 15 cases. Five cells with megakaryocytic differentiation, such as cases showed a minority of blast cells stained with cytoplasm budding, acidophilic granules, and platelet Sudan black B. The blast cells in all cases were posi- shedding. Large and bizarre platelets and micro- tive for ANAE the activities varying from moderate were seen in seven cases. Five cases, to strong. ANAE activity was seen as scattered gran- (1, 3, 4, 9 and 15), also showed between 5 and 14% of ules localised in the Golgi area (figs 4 and 5). The with fine granular cytoplasm, and in one ANAE activity decreased when sodium fluoride was J Clin Pathol: first published as 10.1136/jcp.40.6.663 on 1 June 1987. Downloaded from Cytochemistry of megakaryoblasts 667

F. I.O.-.7:7

P F:: .i..: s: .s 40F

Xi .,::: :W.

.. _. .**

.>.. F ii ON ....

..s.

Fig 4 ANAE reaction in blasts. Strong cytochemical Fig 5 ANAE reaction (case 5). copyright. reac tion is seen mainly in paranuclear area or Golgi zone. added. On the other hand, the ANBE reaction was negative in nine of the 14 cases tested. The positive cases showed weak activity in a small proportion of blast cells, which were also sensitive to sodium http://jcp.bmj.com/ fluoride. Acid phosphatase activity was positive in all cases and the localisation pattern was similar to that of ANAE with strong positivity in the Golgi area and scattered granules throughout the cytoplasm (fig 6). The leukaemic cells were uniformly negative for 5'nucleotidase activity, except for two cases (13 and

15), where the enzyme reaction was strongly positive on September 28, 2021 by guest. Protected with a coarse brown granular pattern.

MARKER STUDIES The results obtained with several monoclonal anti- bodies in samples of peripheral blood or bone mar- row, or both, in the 15 cases with megakaryoblastic leukaemia are summarised in table 1. The blast cells reacted with GRB1 (anti-Ia) in all 10 cases in which it was tested. 3C5 was also consistently expressed in all the cases (8 to 15) of M7; in CGL-BC with mega- karyoblasts alone one of four cases tested was posi- I~~~~~~ tive for 3C5. Eight of 13 cases tested with My9 were positive, with seven showing more than 40% positive blast cells. B lymphoid markers (TdT, J5, and B4) were negative, as were both antierythroid monoclonal Fig 6 Acidphosphatase reaction in cells (caseS) showing antibodies. heavy reaction product in the Golgi zone. J Clin Pathol: first published as 10.1136/jcp.40.6.663 on 1 June 1987. Downloaded from

668 Oliveira, Gregory, Matutes, Parreira, Catovsky The reactivity of the two monoclonal antibodies light microscopical cytochemistry to the diagnosis of against platelet glycoprotein was variable, showing megakaryoblasts using five cytochemical tests. between 10 and 40% positive blast cells. There were Our findings show that megakaryoblasts are nega- no major differences in the proportion of cases reac- tive to Sudan black B staining and myeloperoxidase ting with one or other antibody, although the values estimation, are usually ANBE negative, and show a were slightly higher for C17 than for AN51 in three characteristic cytochemical pattern seen with the cases. In two cases (10 and 15) small percentages of reactions to ANAE and acid phosphatase, enzymes blast cells were weakly positive with C17, but their which are localised in and around the Golgi area. megalokarocytic nature was shown by the level of These results confirm our earlier observation in a ultrastructural platelet peroxidase activity. On the different series of patients and suggest that this cyto- other hand, one case (5) reacted positively to C17 in chemical profile may be typical for M7 and CGL-BC the absence of platelet peroxidase activity. C17 and with megakaryoblasts. The strong reactivity with AN51 were tested in most cases by both ANAE, but not with ANBE, using a different sub- immunofluorescence and immunoperoxidase, the lat- strate to show non-specific esterases, is characteristic ter mainly to confirm whether the reactivity was of megakaryoblasts.20 2 detected in blast cells, as opposed to large platelets. In this study (table 2) we also compared the cyto- The results were usually similar with the two tech- chemical reactivity of the five enzymes in other types niques, but immunoperoxidase seemed to be more of blast cells. Surprisingly, in cases of M6 (erythro- sensitive than immunofluorescence. leukaemia) we showed a cytochemical profile similar to that of M7 blast cells, including the reactivity with Discussion ANAE and acid phosphatase and the absence of reac- tion when tested for ANBE. Specific monoclonal antibodies against platelet The acid phosphatase reaction has been described glycoproteins46 151619 and the ultrastructural as characteristic of T cell acute lymphoblastic marker, platelet peroxidaset 2 4 20 21 have allowed us leukaemia23 with a strong reaction localised in the to establish objective criteria for the diagnosis of Golgi zone. This reaction, sometimes scattered over copyright. megakaryoblastic leukaemia.7 Although both the rest of the cytoplasm, was also seen in M6 and M7 morphological7 and cytochemical criteria6 22 have blasts, suggesting that in undifferentiated acute been proposed, we know of no systematic cyto- leukaemias which are Sudan black B and MPO chemical study in cases of megakaryoblastic leu- negative-a localised strong reaction to acid phos- kaemia diagnosed by immunological and phatase is not necessarily unique to T cell acute lym- ultrastructural studies. We have attempted to apply phoblastic leukaemia. Clearly, the reactivity with T http://jcp.bmj.com/

Table 2 Cytochemical profile ofmegakaryoblasts compared with other types ofblast cells No of Leukaemia type cases ANAE ANBE Acidphosphatase S'nucleotidase Megakaryoblastic 15 Moderately to strong Negative in 10 Strong + in 2 of 3 Positive in 2 of 13

(M7 and CGL-BC) in 30-100% of blast cases; others cases; scattered cases in 18% and 30% on September 28, 2021 by guest. Protected cells; diffuse heavy weak positive granules localised of blast cells granules and localised in Golgi area in the Golgi area Monocytic (AML 10 Moderately to strong in Moderately to Not done Negative in 2 of 2 M5a and M5b) 60- 100% of blast cells; strong in 45-90% fine granular to diffuse of blast cells; pattern fine granular pattern Erythroleukaemia 5** Strong in 16-80% of blast Negative 3 of 3 strong + Not done (AML, M6) cells; large dot localised localised in Golgi in the Golgi area area Myeloblastic 10 6 negative; 4 with a small Negative in 4 of 4 Not done Strong in 80% blast cells (AML MI and percentage of weak (scattered granules in M2) positive blast cells 2 of 5 cases) Lymphoblastic (ALL 9 Negative Not done Not done Strong in 40-100% of and CGL-BC) blast cells in 6 cases ANAE and ANBE reaction was NaF sensitive in M5, M6 and M7. **all shown by morphological criteria and monoclonal antibody (anti-glycophorin and Gero). J Clin Pathol: first published as 10.1136/jcp.40.6.663 on 1 June 1987. Downloaded from Cytochemistry ofmegakaryoblasts 669 lineage specific monoclonal antibody has priority, and electron microscopic levels. Leuk Res 1984;8:769-81. 5 Vinci G, Tabilio A, Deschamps JF, et al. Immunological study of although some of them, like those of the CD7 group in vitro maturation of human megakaryocytes. Br J Haematol (for example 3A1), may also react with some early 1984;56:589-605. myeloid blast cells, including occasional M7 cases. 6 San Miguel JF, Tavares de Castro J, Matutes E, et al. Character- A positive 5'nucleotidase reaction in blast cells ization of blast cells in chronic granulocytic leukaemia in trans- formation, acute myelofibrosis and undifferentiated leukaemia. from a case of M7, and a negative one in other types Br J Haematol 1985;59:297-309. of acute leukaemia, was reported by El-Mohandes 7 Bennett JM, Catovsky D, Daniel M-T, et al. Criteria for the and Hayhoe.' We failed to confirm this, using the diagnosis of acute of lineage (M7). A same cytochemical method. Table 2 shows that only report of the French-American British Cooperative Group. Ann Intern Med 1985;103:460-2. two of the 13 megakaryoblastic cases that we tested 8 Sheehan HL, Storey GW. An improved method of staining leuco- showed distinct 5'nucleotidase positivity. Further- cyte granules with Sudan Black B. Journal of Pathology and more, a similar strong reactivity was seen in several Bacteriology 1947;59:336. acute lymphoblastic leukaemias, and some 9 Li CY, Lam KW, Yam LT. Esterases in human leukocytes. J Histochem Cytochem 1973;21:L-12. myeloblastic leukaemias (M 1 and M2). 10 Barka T, Anderson PJ. Histochemical methods for acid Our findings suggest that there is a distinct cyto- phosphatase using hexazonium pararosanilin as coupler. chemical profile for M6 and M7 blast cells, which J Histochem Cytochem 1962;10:741. could, in the absence of more specialised electron 11 El Mohandes E, Hayhoe FGJ. 5-nucleotidase activity of mega- karyoblasts in a case of acute megakaryoblastic leukaemia microscopic studies, support a diagnosis based on Br J Haematol 1983;53:523-6. immunological data. The value of the simple cyto- 12 Tindle RW, Nichols RAB, Chan L, Campana D, Catovsky D, chemical methods, which also showed distinct pat- Birnie GD. A novel monoclonal antibody BI-3C5 recognised terns in other types of acute myeloid leukaemia, like myeloblasts and non-B non-T lymphoblasts in acute leu- kaemias and CGL blast crises, and reacts with immature cells M5 (table 2) is emphasised by the fact that less than in normal bone marrow. Leuk Res 1985;9:1-9. 40% of the leukaemic cells react with the specific 13 Griffin JD, Mayer RJ, Weinstein HI, et al. Surface marker monoclonal antibody against platelet glycoprotein. analysis of acute myeloblastic leukemia: Identification of The remaining blast cells may, therefore, be of differentiation-associated phenotypes. Blood 1983;62:557-63. different cell lineage. Other workers' experience with 14 Edwards PAW. Monoclonal antibodies that bind to the human

erythrocyte-membrane glycoproteins glycophorin A and B and copyright. the platelet peroxidase reaction seen on transmission 3. Biochem Soc Trans 1980;8:334-5. electron microscopy4 19 20 suggests that most of the 15 Tetteroo PAT, Lansdorp PM, Leeksma OC, Von Dem Borne blasts are megakaryoblasts, despite the percentage of AEGKr. Monoclonal antibodies against human platelet glyco- protein Illa. Br J Haematol 1983;55:509-22. reactive cells with the monoclonal antibody being 16 McMichael JA, Parham P, Rust N, Brodsky FM. A monoclonal low. antibody that recognises an antigenic determinant shared by Our results partly confirm this. A uniform cyto- HLA A2 and B17. Hum Immunol 1980;1:121-9. 17 Bollum FJ. Terminal deoxynucleotidyl transferase as a hema- chemical pattern was often seen in most of the blast http://jcp.bmj.com/ topoietic cell marker. Blood 1979;54:1203-15. cells in the M7 cases studied, but this was not as clear 18 Stein H, Gerdes J, Schwab U, et al. Identification of Hodgkin and in the cases of blast crises in chronic granulocytic leu- Sternberg-Reed cells as a unique cell type derived from a newly kaemia, which have a mixed cell population. Cyto- detected small-cell population. Cancer 1982;30:445-59. chemistry tests may be able to differentiate between 19 Vainchenker W, Deschamps JF, Bastin JM, et al. Two mono- clonal antiplatelet antibodies as markers of human mega- them and may also indicate whether many of the karyocyte maturation: immunofluorescent staining and plate- leukaemic cells belong to the megakaryoblast or let peroxidase detection in megakaryocyte colonies and in in erythroid lineages. vivo cells from normal and leukemic patients. Blood 1982;59:514-21. on September 28, 2021 by guest. Protected 20 Polli N, O'Brien M, Tavares de Castro J, Matutes E, San Miguel E Matutes was supported by a grant from the JF, Catovsky D. Characterization of blast cells in chronic gran- Leukaemia Research Fund. ulocytic leukaemia in transformation, acute myelofibrosis and undifferentiated leukaemia. I. Ultrastructural morphology and cytochemistry. Br J Haematol 1985;59:277-96. References 21 Huang M-J, Li C-Y, Nichols WL, Young J-H, Katzmann JA. Acute leukemia with megakaryocytic differentiation: a study of 1 Breton-Gorius J, Daniel MT, Flandrin G, Kinet Denoel G. Fine 12 cases identified immunocytochemically. Blood 1984;64: structure and peroxidase activity of circulating micro- 427-39. megakaryoblasts and platelets in a case of acute myelofibrosis. 22 Den Ottolander GJ, Te Velde J, Prederoo P, et al. Mega- Br J Haematol 1973;25:331-9. karyoblastic leukaemia (acute myelofibrosis): a report of three 2 Breton-Gorius J, Reyes F, Duhamel G, Najman A, Gorin NC. cases. Br J Haematol 1979;42:9-20. Megakaryoblastic acute leukemia: identification by the ultra- 23 Catovsky D, Galetto J, Okos A, Miliani E, Galton DAG. Cyto- structural demonstration of platelet peroxidase. Blood chemical profile of B and T leukemic lymphocytes with special 1978;S1:45-60. reference to acute lymphoblastic leukemia. J Clin Pathol 3 Bain BJ, Catovsky D, O'Brien M, et al. Megakaryoblastic leuke- 1 974;27:767-7 1. mia presenting as acute myelofibrosis. A study of four cases with the platelet-peroxidase reaction. Blood 1981 ;58:206- 13. Requests for reprints to: Dr D Catovsky, MRC Leukaemia 4 Tabilio A, Vainchenker W, Van Haecke D, et al. Immunological Unit, Royal Postgraduate Medical School, Ducane Road, characterization of the leukemic megakaryocytic line at light London W12 OHS, England.