Ann Hematol (1991) 62:16-21

Annals of

Hematology 9 Springer-Verlag 199t

Leading article A classification of acute leukaemia for the 1990s D. Calovsky, E. Matutes, V. Buccheri, V. Shelty, J. Hanslip, N. Yoshida, and R. Morilla Academic Department of Haematology and Cytogenetics Royal Marsden Hospital and Institute of Cancer Research Fulham Road, London, UK Received October 15, 1990/Accepted October 18, 1990

Summary. The need for reproducibility in the classification of acute leukaemia has made it necessary to incorporate information derived from new techniques which have become essential for the study of these disorders. In addition to classic morphology and cytochemistry (FAB proposals), it is necessary to add immunology and cytogenetics (MIC proposals), as well as to investigate further the biological and diagnostic significance of molecular events. As a result of these investigations a new group of leukaemias merit recognition as distinct entities. These include three types of A L L with specific chromosome abnormalities, namely, i) t (9; 22), ii) t (4; 11) and iii) t (1; 19) and four subtypes of AML, i) with minimal differentiation or AML-M0, ii) with basophilic precursors or M2Baso, iii) A M L ( M 4 / M 5 ) with t (8; 16) and iv) A M L with trilineage myelodysplasia. Biphenotypic acute leukaemia constitutes also a distinct entity with features of ALL and AML and represents a malignancy probably affecting multipotent stem cells. We propose an objective evaluation system for biphenotypic leukaemias based on a score in which the various lineage markers are graded according to their known specificity.

Introduction The classification of acute leukaemia (AL) should reflect as objectively as possible the biology and clinical features of the many disorders presenting as proliferations of haemopoietic precursors or blast cells. Advances in the fields of immunology (monoclonal antibodies), cytogenetics (chromosome abnormalities) and molecular genetOffprint requests to: D. Catovsky, Academic Haematology & Cytogenetics, The Royal Marsden Hospital, Fulham Road, London SW3 6J J, UK Abbreviations. AL, acute leukaemia; m, membrane; c, cytoplasmic; MPO, myeloperoxidase; rearr, rearrangement (by DNA analysis); IgH, immunoglobulin heavy chain gene; TCR, T-cell receptor gene

ics (gene rearrangements) have introduced a greater complexity which needs to be taken into consideration in order to improve both the diagnostic precision and the reproducibility of the classification [1]. The FAB (French, American, British) classification, introduced in the late 1970s, has been the basis for most studies to date. The morphological and cytochemical features proposed were refined ten years later [2] and new types, such as M7 or megakaryoblastic leukaemia [3] were recognized. The MIC proposals (morphology, immunology and cytogenetics) have been an important development which emerged from the knowledge that some ALs are probably better defined by chromosome translocations, as in the case of Ph-positive ALL and ALL with translocation t (4; 11) [4, 5], and that immunophenotyping was an essential component in the investigation of ALL [6]. The increasing availability of monoclonal antibodies (McAb) with specificity for leucocyte differentiation antigens has improved our ability to recognize not only the various types of T- and B-lineage derived leukaemias, but also AML [7]. However, findings with McAb and with DNA analysis for T-cell receptor and immunoglobulin genes have also introduced the concept of lineage 'promiscuity' [81 or 'infidelity' [91, and the recognition of a new category of A L described as either mixed-lineage [10] or biphenotypic [1]. The important role of ultrastructural cytochemistry for the identification of poorly differentiated blasts with different types of peroxidase activity, as seen in M7 [11] and in myeloblasts with small granules [12-15], has also been recognized. Therefore, it is apparent that improvements in diagnosis and classification will emerge from studies in which all the above techniques, including DNA analysis, are employed. From work in the last few years it appears that new types of AL with distinct biological features are being recognized and this information is beginning to be taken into account for therapeutic decisions, including the use of bone marrow transplantation. Within the remit of this report we will outline the areas in which these findings are clinically relevant and need to be included in a new classification system.

17

Acute lymphoblastie leukaemia With the exception of L3-ALL (Burkitt type) which identifies by morphology cases with a mature B-cell phenotype that bear the translocation t (8; 14), it is nowadays essential to apply a battery of McAbs and chromosome analysis for a meaningful classification of A L L [4]. The known types of early B (or null)-ALL, common ALL, pre-B-ALL (with cytoplasmic u chain) and T-cell A L L can be further enriched by groups defined largely by karyotype, e.g. t (9; 22) or Ph positive ALL; t (4; 11) [15] and t (1; 19), the latter defining a subgroup within pre-BA L L with poor prognosis [16, 17]. Similarly, ALLs with hyperdiploid karyotype ( > 50 chromosomes) are seen exclusively within B-lineage ALL, more specifically common-ALL, and are strongly associated with a favourable outcome [18]. Although important in itself, the identification of prognostic groups should not be the only aim of a classification as advances in treatment may nullify any differences. If the disease features are defined by objective laboratory data this could lay the foundations for further progress in understanding the etiopathogenesis and molecular mechanisms involved in AL.

Acute myeloid leukaemia The classification of AML, from M 1 to M7, is still largely based on morphology and cytochemistry [1-3]. Studies with McAb have confirmatory value in many cases as they may correlate closely with differentiation features, e.g. the lack of expression of class II M H C antigens in promyelocytes of M 3 and the frequent expression of the CD 14 antigen in leukaemias with a monocytic component (M4 and M5). The biological importance of some of the FAB defined subtypes is supported by their strong correlation with chromosome abnormalities [5]: t (8;21) in M2; (15; 17) in M3 and M3-variant; inv(16) in M4 with bone marrow eosinophilia (M4Eo) and t (9; 11) in M 5 a. The contribution of immunology to the diagnosis of AML is particularly valuable in the poorly differentiated forms, myeloblastic (M0 and M 1), monoblastic (M 5 a), megakaryoblastic (M 7) and erythroleukaemia (M 6) with immature erythroid cells. McAbs such as CD13 and CD 33 are positive in the majority of AML cases, platelet glycoproteins can be detected in M7 blasts by McAb of the CD41, CD42 and CD61 clusters, and glycophorin A can be demonstrated on the cells of M6. There are four subtypes which have yet to find a place in the classification of AML: M0, M2Baso, A M L with trilineage myelodysplasia and monocytic leukaemia (M4/M5) with t (8; 16) and erythrophagocytosis.

cept for the enzyme deoxynucleotidyl terminal transferase (TdT) which is expressed in half the cases [20]. The key finding to define these cases as AML-M0 is their reactivity with McAb against myeloid associated antigens, usually CD 13 and CD 33. This group represents an early form of AML which can be distinguished from ALL (positive lymphoid markers) and other types of A M L (positive cytochemistry). Confirmation of the myeloblastic nature of AML-M0 can be obtained by means of ultrastructural cytochemistry with the demonstration of small myeloperoxidase (MPO) positive granules [12-14]. Two methodological considerations are important for the diagnosis of immature forms of AML, especially M0. Firstly, the knowledge that the myeloid antigen CD 13 is expressed in the cytoplasm (c) of myeloblasts earlier than on the membrane (m) [21]; thus, CD 13 should be tested by immunocytochemistry whenever the results of flow cytometry on cell suspensions are negative. This feature of CD 13 is similar to the findings with CD3 and CD22 which have cytoplasmic expression in T- and B-lymphoblasts, respectively [6, 22], before these antigens are expressed on the membrane. Secondly, recent studies with antibodies against MPO have shown that these reagents have greater sensitivity than the light microscopy cytochemical tests for MPO and Sudan black B [13,23,24]. Our experience in 140 cases of AL (90 AML, 50 ALL) using a McAb anti-MPO, which detects the proenzyme form and the alpha-chain of the MPO molecule, has confirmed the sensitivity and specificity of this reagent when used by immunocytochemistry with the APAAP method. Furthermore, in some cases of AML-M0, the positivity with anti-MPO correlates with evidence of MPO activity as seen at electron microscopy level. Our experience also suggests that the ultrastructural demonstration of MPO is slightly more sensitive than the McAb anti-MPO for the detection of this enzyme in very early myeloid cells.

A M L with basophil precursors (M2Baso) Electron microscopy can also help to demonstrate a rare type of AML in which the blasts showed the characteristic granules of basophils. The nature of these cells can be suspected by light microscopy morphology because of the presence of few but distinct basophilic granules in their cytoplasm. Drs. G. Flandrin and M-T. Daniel have observed similar cases at the H6pital Saint-Louis in Paris, and we have jointly agreed to designate such cases as M 2 Baso, rather than AML-M 8, which was another alternative. Blasts with basophilic and/or mast cell granular structures are not rare in cases of chronic granulocytic leukaemia in blast crisis [11], but the designation of M2Baso should be reserved for Ph-negative cases.

Early myelobtastic leukaemia: AML-MO A M L with trilineage myelodysplasia A distinct group of cases, 3~ of all A M L in our experience, have blasts with immature morphology (often resembling A L L L2) and negative cytochemical tests which are insufficient for them to be classified as A M L by FAB criteria [2]. B and T lymphoid markers are negative, ex-

An important minority of primary AML (between 10-15% of cases) show marked myelodysplastic features involving erythroid, granulocytic and megakaryocytic cells [25]. Trilineage myelodysplasia could be seen in all

18 FAB subtypes of A M L except M3; it is more c o m m o n in M 6 and M7 and rare in M1. It is not clear whether such cases should be classified separately but it is likely that they represent cases of myelodysplastic syndrome (MDS) presenting as AML. In a number of these patients a unique type o f relapse, with MDS and without excess of blasts, has been documented [26]. In addition, the response rate of A M L with trilineage myelodysplasia is lower than in A M L without these features [24]. It is noteworthy that the proportion of such cases corresponds to those A M L with karyotypic abnormalities characteristic of MDS, e.g. - 5, 5 q - , - 7, 7 q - and which are also associated with a low remission rate [26]. As there is no available data to establish whether both types of AML, defined by morphology and karyotype, correspond to the same disease, further studies are necessary.

A M L ( M 4 / M 5 variant) with translocation t (8; 16) This rare but distinct form of A M L [5, 28] is characterized by monocytic features (described as M 5 in two thirds of cases and as M 4 in one third), erythrophagocytosis, dual positivity for granulocytic and monocytic enzymes and a bleeding diathesis with coagulation features suggestive of fibrinolysis and not of DIC as seen in M3. The chromosome abnormality t (8; 16) (pl 1;p13) is a consistent finding in these patients who may be o f any age group, from infants to adults. This type of A M L constitutes 0.6% of cases and is associated with a relatively low remission rate with few long term survivors. Central nervous system disease may be present at diagnosis and isolated meningeal relapses may occur.

Biphenotypic acute leukaemia The coexistence of blast cells from different lineages has been recognized for many years in the blast crisis of chronic granulocytic leukaemia. It is now apparent that a 'mixed' blast cell population or, more commonly, the coexpression of myeloid and B or T lymphoid antigens on the same cells, hence the term biphenotypic, is a feature of some cases of AL [10, 13-15]. The consensus is also emerging that if these cases are genuine and properly defined by excluding findings with non-specific markers and the spurious binding to Fc receptors [1,8], they probably represent a distinct type of A L with unique biology and prognosis [291. The high degree of correlation with certain chromosome abnormalities such as t (9; 22) and abnormalities of 11q23, including t (4; 11) [30], suggests that biphenotypic expression is more c o m m o n in leukaemias affecting haemopoietic precursor cells with potential for differentiation in several directions [31, 32]. Because of the availability of relatively specific McAb, cytogenetics, ultrastructural cytochemistry and D N A analysis, it is now possible to study such cases in detail and describe their clinical and pathologic correlations. The well documented demonstration of the phenomenon of 'phenotypic switch' is also highly relevant to this type of a c u t e leukaemia [31,331.

Within the context of biphenotypic A L a number of scenarios are possible:

1) Typical A L L or AML, defined by conventional criteria, in which blasts may express one aberrant or ectopic antigen (of the myeloid or lymphoid lineage, respectively). These are slightly more c o m m o n within cases of adult A L L [32] than childhood A L L [341 and probably represent minimal phenotypic deviations [35], without significant prognostic implications, at least in childhood A L L [341. 2) Cases which may be typical or appear undifferentiated by conventional morphological and cytochemical criteria, but in which there is unequivocal expression of two or more markers of different lineages. This group fits our definition of biphenotypic leukaemia (Table 1) and does not include cases of AML-MO, that express myeloid but no lymphoid antigens, which can be misdiagnosed as A L L if defined only on the basis of morphology and light microscopy cytochemistry [29]. In rare cases the blasts express lymphoid antigens but the myeloid features are only demonstrated by ultrastructural M P O or by antiM P O [13]. Because of the high ranking of M P O as a specific myeloid marker, such cases also qualify as biphenotypic (Table 1). D N A analysis usually demonstrates the rearrangement of functional genes (IgH and/or TCR) correlating with the expression of lymphoid antigens [10, 14, 15, 32, 36]. 3) Cases with similar features as above, but in which morphology, cytochemistry and/or markers suggest the existence of two separate blast cell populations. In our experience, these cases are rare and they often show some markers that overlap in the two cell populations. It is likely that these are also examples of biphenotypic A L but showing a greater degree of maturation in the cells involved.

Table 1. Scoring system for biphenotypic ALa

Points

B-lineage

T-lineage

Myeloid lineage

2

cCD 22 c/~ chain

cCD 3

MPO (any method including anti-MPO) b

1

CD10 CD 19 CD 24

CD2 CD5 TCR rearr. (fl or 6 chain)

CD33 CD 13 m/c CD 14 m/c AML morphology or cytochemistry (other than MPO) c

0.5

TdT IgH rearr.

TdT CD 7

CD 11b CD 11c CD15

Two or more points from 2 separate lineages are required to classify a case as biphenotypic AL. Most biphenotypic AL have a mixture of myeloid and lymphoid (B or T) markers; exceptionally, cases with coexpression of B and T markers have been reported u Cases suspected of being myeloid but with negative MPO by light microscopy cytochemistry or anti-MPO by McAb need analysis by electron microscopy c Diffuse/strong alpha naphthyl esterase (ANAE) sensitive to sodium fluoride and/or positive Sudan black B

a

19 4) Typical cases of A L L or A M L but in which, following complete remission, a relapse occurs with a complete lineage switch [31]. It appears that the nature of the treatment, anti-myeloid or antiqymphoid, may have a bearing on the nature of the subsequent relapse [10, 31]. A phenotypic switch is seen more often in cases which do not present as biphenotypic but as straight A L L or AML. Work in our laboratory with short term cultures of blast cells suggests that in a minority of cases a phenotypic switch can also be induced in vitro, from myeloid to lymphoid or lymphoid to myeloid (V. Buccheri, unpublished observation). It is not yet known whether the in vitro findings may predict a switch in vivo because, in the latter, the differentiation drive may be influenced by the type of therapy. The importance of D N A analysis to confirm that the malignant cells have the same genotype (by showing identical I g H or TCR rearrangement), regardless of the phenotypic expression, has been shown unequivocally in a few cases studied in both phases of the disease [32, 33, 371.

Scoring system to define biphenotypic AL A criteria for scoring the features necessary to classifiy cases of A L as biphenotypic has been suggested by Mirro et al. [31]. We propose herewith (Table 1) a modified system based on the different weight which needs to be given to the diagnostic tests according to their generally accepted specificity. This also presupposes, yet without proof, that cases with minimal deviation and thus not qualifying as biphenotypic by the criteria outlined in Table 1, have a different biological significance. The proposal is intended to facilitate the gathering of information and to restrict the criteria of biphenotypic to cases in which two or more differentiation genes are switched in a particular direction. Double staining procedures by flow cytometry [31] and electron microscopy for the demonstration of

Table 2. Application of the scoring system in a series of acute leukaemiaa Number of cases

Overall \\ ~:h low scores (1-1.5) Biphenotypic scores (2-5)

Diagnosis b AML

ALL

Total

98 9

68 5

166 14

13c (13%)

2 (3%)

15 (9%)

a Patients of all ages diagnosed over a period of 18 months. Information on gene rearrangements not yet completed, thus not used to calculate the scores b Based on conventional criteria: morphology and cytochemistry and positive markers (myeloid in AML and lymphoid in ALL) c Only 3 cases had score 2 and the phenotype TdT+, CD2+, CD7+; if CD2 is given a score of 0.5 rather than 1, these cases would not qualify as biphenotypic. CD2 was the most frequent 'aberrant' marker in AML, accounting for half of the 22 cases with scores 1-5

M P O [15], combined with the immunogold method to document the cell expressing an 'aberrant' antigen [38], could contribute further to the study of such cases. We have tested the scoring system in 166 consecutive ALs studied in our laboratory although information on gene rearrangements was not yet available. The preliminary analysis is summarized in Table 2. In our experience, high scoring cases were found mainly a m o n g cases diagnosed as AML; within this group, 13~ of them could be considered biphenotypic. We are still concerned with the weight given to C D 2 in some American studies [15]. In fact, C D 2 was positive in 11 of 22 A M L with positive lymphoid markers (Table 2). If we downgrade C D 2 to score 0.5 (as TdT and CD7) rather than 1, this reduces the number of A M L s considered biphenotypic in Table 2 from 13 to 10. Interestingly, these three 'AMU cases had the same phenotype, CD 2 +, CD 7 +, TdT+. Information on TCR gene rearrangements will be forthcoming in these cases. However, in the recent study by Kantarjian et al. [15], three cases of 'mixed lineage' A L which expressed C D 2 and TdT did not show rearrangement of the TCR [3 chain gene, thus questioning their T-cell specificity. From the above it is apparent that there is insufficient data correlating i m m u n o p h e n o t y p e with gene rearrangements in biphenotypic cases and this is one area where new information is urgently needed. A report by Fontenay et al. [36] provides evidence that i.nappropriate rearrangement of the TCR 8 gene in A M L correlates with the expression of lymphoid antigens, chiefly CD 10. The use of a restricted definition and the need to correlate with chromosome abnormalities may be critical to provide meaningful prognostic information, as already suggested in some adult leukaemias [29, 32]. If biphenotypic acute leukaemia is defined on the lines outlined in Table 1, it may be possible to identify this disease more clearly as a distinct type of AL, and establish more conclusively whether it represents a stem cell leukaemia with potential for multilineage gene expression [8] or is a reflection of changes in the differentiation programme brought about by cytogenetic and/or molecular events [9, 31]. One clinical implication, extrapolating from findings in cases showing a phenotypic switch, is that therapy should cover the whole spectrum of anti-myeloid and anti-lymphoid agents. A classification of biphenotypic leukaemia should include, in its broadest sense, cases in which a phenotypic switch is documented in vivo and possibly also those in which it can be demonstrated in vitro. The precise diagnosis of such cases may thus allow a better selection of appropriate treatment modalities.

Acknowledgements. This work was supported in part by the Leukaemia Research Fund (VS), CAPES, Brasil (VB) and Locally Organized Research at The Royal Marsden Hospital (JH).

References 1. Bain B, Catovsky D (1990) Practical aspects of the classification of acute leukaemia. J Clin Pathol 43:882-887 2. Bennett JM, Catovsky D, Daniel M-T, Flandrin G, Galton DAG, Gralnick HR, Sultan C (1985) Proposed revised criteria for the classification of acute myeloid leukemia. Ann Intern Med 103: 620-625

20 3. Bennett JM, Catovsky D, Daniel M-T, Flandrin G, Galton DAG, Gralnick HR, Sultan C (1985) Criteria for the diagnosis of acute leukaemia of megakaryocyte lineage (M7). Ann Intern Med 103:460-462 4. First MIC Cooperative Study Group (1986) Morphologic, immunologic and cytogenetic (MIC) working classification of acute lymphoblastic leukemias. Cancer Genet Cytogenet 23: 189-197 5. Second MIC Cooperative Study Group (1988) Morphologic, immunologic and cytogenetic (MIC) working classification of the acute myeloid leukemias. Cancer Genet Cytogenet 30: 1-15 6. Janossy G, Coustan-Smith E, Campana D (1989) The reliability of cytoplasmic CD3 and CD22 antigen expression in the immunodiagnosis of acute leukemia: a study of 500 cases. Leukemia 3:170-181 7. Greaves MF, Janossy G, Peto J, Kay H (1981) Immunologically defined subclasses of acute lymphoblastic leukaemia in children: their relationship to presentation features and prognosis. Br J Haematol 48:179-197 8. Greaves MF, Chart LC, Fnrley AJW, Watt SM, Molgaard HV (1986) Lineage promiscuity in hemopoietic differentiation and leukemia. Blood 67:1-11 9. Smith LJ, Curtis JE, Messner HA, Senn JS, Furthmayr H, McCulloch EA (1983) Lineage infidelity in acute leukemia. Blood 61:1138-1145 10. Mirro J, Zipf TF, Pui C-H, Kitchingman G, Williams D, Melvin S, Murphy SB, Stass S (1985) Acute mixed lineage leukemia: clinicopathologic correlations and prognostic significance. Blood 66:1115-1123 11. Breton-Gorius J, Gourdin MF, Reyes F (1981) Ultrastructure of the leukemic cell. In: Catovsky D (ed) The leukemic cell, Churchill Livingstone, Edinburgh 1981, pp 87-128 12. Marie JP, Perrot JY, Boucheix C, Zittoun J, Martyre MC, Kayibanda M, Rosenfeld C, Mishal Z, Zittoun R (1982) Determination of ultrastructural peroxidases and immunologic membrane markers in the diagnosis of acute leukemias. Blood 59:270-276 13. Vainchenker W, Villevat JL, Tabilio E, Matamis H, Karanakis G, Guichard J, Henri A, Vernant JP, Rochant H, BretonGorius J (1988) Immunophenotype of leukemic blasts with small peroxidase-positive granules detected by electron microscopy. Leukemia 2:274-281 14. Matutes E, Pombo de Oliveira M, Foroni L, Morilla R, Catovsky D (1988) The role of ultrastructural cytochemistry and monoclonal antibodies in clarifying the nature of undifferentiated cells in acute leukaemia. Br J Haematol 69:205-211 15. Kantarjian HM, Hirsh-Ginsberg C, Yee G, Huh Y, Freireich E J, Stass S (1990) Mixed-lineage leukemia revisited: acute lymphocytic leukemia with myeloperoxidase-positive blasts by electron microscopy. Blood 76:808-813 16. Leglise M-C, Riviere D, Briere J (1990) Acute leukemia with a translocation t(4;ll) (q21;q23): a distinct clinicopathological entity. Leukemia Lymphoma 2:353-368 17. Christ W, Boyett J, Jackson J, Vietti T, Borowitz M, Chauvenet A, Winick N, Ragab A, Mahoney D, Head D, Iyer R, Wagner H, Pulla J (1989) Prognostic importance of the pre-B-cell immunophenotype and other presenting features in B-lineage childhood acute lymphoblastic leukaemia: a pediatric oncology group study. Blood 74:1252-1259 18. Christ WM, Carroll AJ, Pui C-H (1990) Pre-B-cell acute lymphoblastic leukemia in childhood. Leukemia Lymphoma 3: 1-6 19. Bloomfield CD, Secker-Walker LM, Goldman AI, Van Den Berghe H, de la Chapelle A, Ruutu T, Alimena G, Garson OM, Golomb HM, Rowley JD, Kaneko Y, Whang-Peng J, Prigogina E, Philip P, Sandberg AA, Lawler SD, Mitelman F (1989) Sixyear follow-up of the clinical significance of karyotype in acute lymphoblastic leukemia. Cancer Genet Cytogenet 40:171-185 20. Parreira A, Pombo de Oliveira MS, Matutes E, Foroni L, Morilla R, Catovksy D (1988) Terminal deoxynucleotidyl transferase positive acute myeloid leukaemia: an association with immature myeloblastic leukaemia. Br J Haematol 69:219-224

21. Pombo de Oliveira MS, Matutes E, Rani S, Morilla R, Catovsky D (1988) Early expression of MCS2 (CD13) in the cytoplasm of blast cells from acute myeloid leukaemia. Acta Haematol 80: 61-64 22. Rani S, de Oliveira MSP, Catovsky D (1988) Different expression of CD3 and CD22 in leukemic cells according to whether tested in suspension or fixed on slides. Hematol Pathol 2: 73 -78 23. Van der Schoot CE, von dem Borne AEGK, Tetteroo PAT (1987) Characterization of myeloid leukemias by monoclonal antibodies, with an emphasis on antibodies against myeloperoxidase. Acta Haematol 78 [Suppl 1]: 32-40 24. Van der Schoot CE, Daams GM, Pinkster J, Vet R, von dem Borne AEGK (1990) Monoclonal antibodies against myeloperoxidase are valuable immunological reagents for the diagnosis of acute myeloid leukaemia. Br J Haematol 74:173-178 25. Brito-Babapulle F, Catovsky D, Galton DAG (1987) Clinical and laboratory features of de novo acute myeloid leukaemia with trilineage myelodysplasia. Br J Haematol 66:445-450 26. Brito-Babapulle F, Catovsky D, Galton DAG (1988) Myelodysplastic relapse of de novo acute myeloid leukaemia with trilineage myelodysplasia. Br J Haematol 68:411-415 27. Samuels BL, Larson RA, Le Beau MM, Daly KM, Bitter MA, Vardiman JW, Barker CM, Rowley JD, Golomb HM (1988) Specific chromosomal abnormalities in acute nonlymphocytic leukemia correlate with drug susceptibility in vivo. Leukemia 2: 79-83 28. Tanzer J, Brizard A, Guilhot F, Benz-Lemoine E, Dreyfus B, Lessard M, Herchkovitch C, Mossafa H, Davi F, Huret J, Le Bezu M (1988) La leucemie aigue a translocation (8;16) Nouv Rev Fr Hematol 30:83-87 29. Sobol RE, Mick R, Royston I, Davey FR, Ellison RR, Newman R, Cuttner J, Griffin JD, Collins H, Nelson DA, Bloomfield CD (1987) Clinical importance of myeloid antigen expression in acute adult lymphoblastic leukemia. N Engl J Med 316: 1111-11t7 30. Chen SJ, Flandrin G, Daniel M-T, Valensi F, Baranger L, Grausz D, Bernheim A, Chert Z, Sigaux F, Berger R (1988) Philadelphia-positive acute leukemia: lineage promiscuity and inconsistently rearranged breakpoint cluster region. Leukemia 2:261-273 31. Mirro J, Kitchingman GR (1989) The morphology, cytochemistry, molecular characteristics and clinical significance of acute mixed-lineage leukaemia. In: Scott CS (ed) Leukaemia Cytochemistry: principles and practice. Ellis Horwood, Chichester, pp 155-179 32. Childs CC, Hirsch-Ginsberg C, Walters RS, Andersson BS, Reuben J, Trujillo JM, Cork A, Stass SA, Freireich E J, Zipf TF (1989) Myeloid surface antigen-positive acute lymphoblastic leukemia (My+ ALL): immunophenotypic, ultrastructural, cytogenetic and molecular characteristics. Leukemia 3: 777-783 33. Nosaka T, Ohno H, Doi S, Fukuhara S, Miwa H, Kita K, Shirakawa S, Honjo T, Hatanaka M (1988) Phenotypic conversion of T lymphoblastic lymphoma to acute biphenotypic leukemia composed of lymphoblasts and myeloblasts. J Clin Invest 81: 1824-1828 34. Pui C-H, Behm FG, Singh B, Rivera GK, Schell M J, Roberts WM, Crist WM, Mirro J (1990) Myeloid-associated antigen expression lacks prognostic value in childhood acute lymphoblastic leukemia treated with intensive multiagent chemotherapy. Blood 75:198-202 35. Del Vecchio L, Schiavone EM, Ferrara E Pace E, Lo Pardo C, Pacetti M, Russo M, Cirillo D, Vacca C (1989) Immunodiagnosis of acute leukemia displaying ectopic antigens: proposal for a classification of promiscuous phenotypes. Am J Hematol 31: 173-180 36. Fontenay M, Flandrin G, Baurman H, Loisea P, Valensi F, Daniel M-T, Sigaux F (1990) T cell receptor 8 gene rearrangements occur predominantly in immature myeloid leukemias exhibiting lineage promiscuity. Leukemia 4:100-105

21 37. Scott CS, Vulliamy T, Catovsky D, Matutes E, Norfolk DR (1989) DNA genotypic conservation during phenotypic switch from T-cell acute lymphoblastic leukaemia to acute myeloblastic leukaemia. Leukaemia Lymphoma 1:21-28 38. Robinson D, Tavares de Castro J, Polli N, O'Brien M, Catovsky

D (1984) Simultaneous demonstration of membrane antigens and cytochemistry at ultrastructural level: a study with the immunogold method, acid phosphatase and myeloperoxidase. Br J Haematol 56:617-631