A. Introduction

The study of 'membrane markers' in human leukaemia has now been in progress for a decade. Starting from the initial observation ofL. Borella and colleagues at St. Judes on the sub-types of ALL [I] a wealth of data has accumulated particularly over the past few years with the introduction of monoclonal antibodies. Now is perhaps a good time to appraise the impact of these efforts and the implications for future research on leukaemia. As Seligmann, Kersey, myself and others have emphasised on many occasions, the single most fruitful product of this activity has been the appreciation of how the cellular heterogeneity of lymphoid leukaemia and lymphoma mirrors stages of normal differentiation. This clearly arises as a consequence of three salient features of haemopoietic malignancy: the restricted or clonal origin [2], the imposition of maturation arrest, and the broad conservation or fidelity of a qualitatively normal phenotype [3], The immunological and enzymatic definition of leukaemic cell phenotypes in relation to their normal counterparts has direct relevance to clinical problems of differential diagnosis, patient monitoring and variable prognosis [4]. Immunological features of ALL subgroups for example are linked to known prognostic features ( e.g. high white cell count in T-ALL) and not surprisingly, therefore, show a strong correlation with the outcome of chemotherapy [ I, 4-6]. Combinations of markers ( e.g. cell surface antigens and nuclear terminal transferase [7]) offer the possibility of monitoring leukaemia and detecting residual, minimal or re-emerging extramedullary disease (i.e. CNS or testis). The application of a panel of monoclonal antibodies has been routinely applied in my own laboratory for a national immunodiagnostic service over a number of years, It is difficult to determine precisely how useful such a service is; however, I estimate that the phenotypic data are essential in something like 15% of cases and are useful or supporting in many more (perhaps the majority). All of this is clear and undisputed; I would rather emphasise the broader and more substantial impact which I believe these studies should have. Firstly. they provide a rational, biological framework for attempts to improve the efficacy of therapy either by more selective or 'tailored' allocation of particular regimes to defined leukaemic subgroups or by exploiting the biological information to design new or more radical strategies, e.g. monoclonal antibody elimination of leukaemic cells, selective enzyme inhibition. Secondly, they provide an essential framework for pursuing the molecular basis of haemopoietic malignancy. Since cellular oncogenes (or their viral homologues) are probably limited in number and have some important function in regulating normal differentiation and/or proliferation, it is of some importance to search for these genes and the expression and function of their products in the context of particular leukaemic subtypes and their normal counterparts; this is indeed already happening (see papers by F. Wong-Staal and M. A. Lane in this volume). Some of the above points can be emphasised with reference to the biology of ALL.

B. Heterogeneity and Origins of ALL

Acute lymphoblastic leukaemia can be dissected in a number of subgroups with exclusive, composite phenotypes, which correlate with prognosis [4]. More recently, the use of monoclonal antibodies and immunoglobulin gene probes and the study of maturation induction in vitro has further elucidated the nature of ALL cells. It is now clear that ALL consists of two broad

Table 1. Biological features of two ALL subtypes

a Serologically defined cell surface antigens or intracellular enzymes terminal deoxynucleotidyl transferase and hexosaminidase isoenzyme I (plus charge variants of other acid lysosomal hydrolases. [42])
b Ig genes (e.g. V. D, J, u heavy chain) re-arranged from germ line configuration [41]

subtypes, both of which originate in lymphocyte progenitors (Table I ); one is 'pre-T' or equivalent to thymic precursors of mature T cells: the other, more common, variant is 'pre-B' or equivalent to B-cell progenitors and precursors in bone marrow. Within these two categories subtypes can be defined which broadly reflect sequential stages of maturation within the 'early' compartments of these two distinct cell lineages [8-10). Detailed studies on the antigenic phenotypes of these leukaemias provide no evidence for qualitatively aberrant gene expression or for a progenitor cell shared by and exclusive to the T and B lineages. Thus, ALL cells do not express glycophorin [ II ] or other restricted non-lymphoid markers; neither do they show concurrent expression on single cells or within a single leukaemic clone of markers unique to T and B cells. The 'pre- T' and 'pre-B' categories are also consistent features and although individual markers may change in relapse [12] there is no shift between these two subtypes during malignant progression in individual patients [3]. Normal counterparts of the ALL subtypes with qualitatively similar phenotypes (excluding karyotype) can be found in bone marrow [9, 13] and thymus [8. 10], It is of some interest to note that whereas malignancies of lymphocyte precursors occur predominantly in children and young patients, malignancies of mature lymphoid cells (leukaemia, lymphoma, myeloma) are almost exclusively ad u]t diseases [ ] 3 a]; one interpretation of this correlation and the similarly striking age associations of other cancers ( e.g. neural tumours versus epithelial carcinomas) is that they are a reflection of cell populations (stem cells?) at risk through proliferative demand at various stages of early development or during prolonged function (and turnover) in adult life. The simplest interpretation of this descriptive data is therefore that ALL can originate in progenitor cells or either the T or B-cell lineage and invariably sufiers from the imposition of maturation arrest with the conservation of phenotype 'appropriate' for the particular stage of differentition in which the leukaemic cells become frozen or stabilised. Whilst I believe this general conclusion to be manifestly correct there are some relevant and important qualifications that should be ap preciated: I. The phenotypes observed are not identical for every leukaemic blast cell of an jndividual patient. Phenotypic categorisation reflects the dominant phenotype, but in practice some diversity can always be detected either with respect to quantity (e.g. antigen density) or in what appears to be quality. Figure 1 illustrates one such case, in which one-third of the leukaemic cells have a different but clearly related phenotype to the other two-thirds. The interpretation favoured for this intraclonal diversity is that it reflects in large part the variable stringency of maturation arrest, i.e. all cells do not appear to be stopped in their tracks at precisely the same developmental position. Superimposed upon this maturational control there is also some phenotypic diversity which is linked to cell proliferation, e.g. expression of the monoclonal antibody defined receptor for transferrin [ 14 ]. 2. Detailed scrutiny of ALL phenotypes in relation to their supposed normal counterparts suggests that they are probably not perfect replicas; an analogy with the minjmally deviated hepatomas of Potter [15] may be appropriatc. The "abnormalities" concern some apparent deletions, such as lack of expression of the E rosette "receptor" or TdT when the remainder of the composite phenotype dictatcs that they be present or what can best he described as asynchronies of gene expression, i.e. com binations of markers which are normally sequentially expressed in maturation, such as TdT and high-density HLA-ABC in T-ALL [ 10. 16], TdT and .u chains in pre-B ALL[17,40]. 3. Leukaemias with an identical (non-chromosomal) phenotype to ALL can arise in the pluripotential stem cell. As reviewed elsewhere [ 18] approximately one-third of Ph1-positive CGL in blast crisis have the common ALL or B-cell progenitor phenotype which includes monoclonal antibody defined antigens, selective enzyme expression and also re-arranged Ig genes (Mulgaard, Gould and Greaves, unpublished observations). Some adult patients can present with Ph1 ALL without a clinically evident chronic phase CGL but may after

Fig. I. Variable position of "maturation arrest" in cALL. Bone marrow Iymphoblasts were stained with various combinations of reagents to analyse phenotypic diversity, e.g. anti-DR, anti-cALL, anti µ or anti-Ig (chi/lamda ) in combination with TdT; anti-chi in combination with anti-DR or anticALL

therapy revert to CGL [18]. It is important to note that whereas B-cell progenitor ALL (e.g. common ALL) is curable with chemotherapy. blast crises manifest in this cellular compartment are not, although as expected they may achieve short-term remissions with steroids [ 19]. This sharp distinction provides an excellent example of the importance of target cell" biology for understanding clinical outcome and developing appropriate alternative therapeutic strategies (e.g. marrow transplants for Phl-positive leukaemia). 4. ALL of either B or T progenitor type may not be diagnosed haematologically as ALL. Thus the majority of those rare (~5%) acute leukaemias which haematologists consider to be acute undifferentiated leukaemia are usually identifiable as ALL subtypes or more rarely as immature myeloid cells [4, 20]. Paediatric cases diagnosed as non-Hodgkin lymphoma may also belong or at least be very closely related to the two major subtypes of ALL. Conversely, not all cases diagnosed as ALL may be bona fide ALL. Thus, B-ALL is probably a misnomer: this relatively mature B-cellleukaemia probably represents a rapidly disseminating lymphoma [4, 21]. Rare cases of newborn acute leukaemia diagnosed as ALL may in fact be 'cryptic' erythroleukaemias as assessed by studies with monoclonal antibodies including antiglycophorin [11,22]. 5. The maturation arrest imposed in ALL may be reversible. at least partially in vitro. Thus, some T -ALL cell lines can be induced by phorbol ester (TPA) to irreversibly modulate their composite phenotype from that of an immature or thymic variety to that of a mature T-cell subset [23, 24]. We and others have also been able to modulate the expression of TdT and cell surface antigen in B-cell progenitor ALL, although in our experience Ig synthesis cannot be induced in Ig- ALL despite the presence of re-arranged fA. chain genes. Our interpretation of this is that in leukaemia and in normal B-cell differentiation these recombinational genetic events are inefficient, with most clones failing to achieve a productive or functional re-arrangement. The observation that maturation arrest in ALL is reversible as demonstrated previously with other leukaemias (e.g. Friend virus erythroleukaemia and myeloid leukaemia in rodents. avian erythroleukaemia and in some human leukaemic cell lines, e.g. HL-60, K562) carries the important corollarv that maturation arrest, a central "lesion'; in acute leukaemia, is a regulatory defect which, although having a genetic, inheritable basis. is reversible in its phenotypic consequences. C. Is the Conservation of Phenotype Telling Us Anything Interesting About Leukaemic Cells? It could be argued that since malignancy involves rare genetic events, it is to be expected that these will not have catastrophic effects on a cell's pattern ofgene expression and that the broad fidelity of phenotype observed in ALL is (a) just what we would expect, and (b) boring and of no relevance or even downright misleading with respect to the central issue of what distinguishes a leukaemic cell from normal. Furthermore, it can always be that the 'critical' gene products in leukaemia arc not those which we rather arbitrarily elect to stuliy (so tar) and that a more appropriate screen would reveal distinct, qualitative and consistent differences between leukaemic cells and their normal counterparts. These are not unreasonable views and I am surprised that they are not made more often. I have favoured a different view initially because it was more interesting and subsequently because I believe it is supported by data. That is that the expression of qualitatively normal phenotype or pattern of gene expression is an integral and essential feature of most if not allleukaemias and other malignancies. Qualitative abnormalities (e.g. new or lost antigens, altered glycolipids, altered drug recognition) may occur and indeed have some selective advantage with malignant progression and treatment: however, they need not be considered as essential components of the malignant state. In the context of ALL, therefore, and as suggested some years ago [25, 26] a qualitatively normal lymphoid progenitor cell phenotype which is normally only transiently expressed on proliferating cells quite compatible with leukaemic

Table 2. Structure, genetics and function of ALL-associated membrane proteins identified by monoclonal antibodies

a Mean value: multiple spots observed with variable positions reflecting allelic polymorphism
b Katz, Povey and Greaves, unpublished observations
c Natural, physiological function unknown

cell behaviour and only requires that the genetic change provoking clonal selection effectively, uncouples proliferation from maturation. This view accords with recent molecular studies which reveal the central role of normal genes (c-onc) or their inserted viral (v-onc) homologues which may facilitate clonal advantage via amplification or excessive promotion ([27] and various papers in this volume). There is no evidence to date that qualitatively altered gene products are involved (An important example of such an alteration has however recently been reported [43] ) Much emphasis therefore rests on quantitative aspects of c-onc expression. Even this phenotypic distinction between leukaemic and normal cells could be small or perhaps only evident in the time frame, i.e. equivalent normal cells may express similar levels of c-onc gene products but only transiently.

D. Epilogue

Several of the ALL-associated membrane antigens have now been biochemically characterised and their control mapped to particular chromosomes (Table 2). Whether any of these proteins has any important regulatory role in differentiation or are even c-onc gene products is at present unknown. One of these structures does have a definite function. The monclonal antibody OKT9 identifies the transferrin receptor [14]; this observation has enabled rapid progress to be made in the biochemica! studies of this receptor [28] and also facilitated the mapping of controlling (presumably structural) genes to chromosome 3 [29]. We have also suggested that the transferrin receptor may serve as a common 'target' structure on malignant and normal cells for so-called natural killer (NK) cells [30]. There are still many gaps in our understanding of lymphoid malignancy and of normal lymphopoiesis. Compared with myelopoiesis for example (see paper by Metcalf in this volume) we have little insight into soluble regulators of early lymphocyte development. Despite these limitations lymphoid malignancy in humans provides, I believe, an excellent example of a disease whose molecular, cellular and clinical complexity can be best understood in relationship to normal cellular differentiation.

References

I Sen L, Borella L ( 1975) Clinical im portance of Iymphoblasts with T markcrs in childhood acute leukemia. N Engl J Med 292: 828-832
2. Fialkow PJ, Denman AM, Singer J, Jacobson RJ, Lowenthal MN (1978) Human myeloproliferative disorders. clonal origin in pluripotential stem cells. In. Clarkson B, Marks PA, Till JE (eds) Differentiation of normal and neoplastic hemopoietic cells. Cold Spring Harbor, New York, pp 131-144
3. Greaves MF (1982) 'Target' cells, cellular phenotypes and lineage fidelity in human leukaemia. J Cell Physiol Suppl I: 113-126
4. Greaves MF (1981) Analysis of the clinical and biological significance of lymphoid phenotypes in acute leukemia. Cancer Res 41:4752-4766
5. Dow L W, Borella L, Scn L, Aur RJA, George SL, Mauer AM, Simone JV (1977) Initial prognostic factors and Iymphoblasterythrocyte rosette formation in 109 children with acute lymphoblastic leukemia. Blood 50.671-682
6. Greaves MF, JanO.5sy G, Pcto J, Kay H ( 1981) Immunologically dcfined subclasscs of acutc lymphoblastic leukaemia in children. their relation5hip to prcsentation features and progno5is. Br J Haematol 48179-197
7. Janossy G, Bollum FJ, Bradstock KF, AshIcy J (1980) Cellular phcnotypes of normal and leukemic hemopoietic cclls determined by analysis with 5elected antibody combi nations. Blood 56. 430-441
8. Reinherz EL, Kung PC, Goldstein G, Levey RH, Schlossman SF ( 1980) Discrete stages or human intrathymic differentiation. analysis or normal thymocytes and leukemic Iymphoblasts of T lineage. Proc Natl Acad Sci USA 771588-1592
9. Greaves MF, Delia D, Robin,5on J, Sutherland R, Newman R (1981) Exploitation or monoclonal antibodies: A 'Who's who' or haemopoietic malignancy. Blood Cells 7:257-280
10. Greaves, MF, Rao J, Hariri G, Verbi W, Catovsky D, Kung p, Goldstein G (1981) Phenotypic heterogeneity and cellular origins of T -cell malignancies. Leukcmia Res 5:281-299
II. Greaves MF (1981) Monoclonal antibodies as probcs ror leukaemic heterogeneity and haemopoietic differentiation. In: Knapp W. (ed.) Leukemia markers. Academic, New York, pp 19-32
12. Greaves MF, Paxton A, Janossy G, Pain C, Johnson S, Lister TA (1980) Acute lymphoblastic leukaemia associated antigen III. Alterations in expression during treatment and in relapse. Leukemia Res 4.1-14
13. Greaves MF, Robinson JB, Delia D, Ritz J, Schlossman S, SieffC, Goldstein G, Kung P, Bollum r, Edwards P ( 1981) Comparative antigenic phenotypes or normal and leukemic hcmopoietic precursor cells analysed with a 'library' of monoclonal antibodies In. Neth R, Gallo RC, Graf T, Mannwcilcr K, Winklcr K (eds) Modern trends in human leukemia 4. Springer, Berlin Heidelberg New York, pp 296-304 (Haematology and blood transfusion, vol 26)
13 a. Greaves MF (to be publishcd) Subtypes of acute lymphoblastic leukaemia: implications for the pathogenesis and epidemiology of leukaemia. In: Magrath I, Ramot B (eds) The influence of the environment on leukaemia and lymphoma subtypes. Natl Cancer Inst Monogr
14. Sutherland R, Delia D, Schneider C, Newman R, Kemshead J, Greaves MF (1981) Ubiquitous, cell surface glycoprotein on tumour cells is proliferation-associated receptor for transferrin. Proc Natl Acad Sci USA 78:4515-4519
15. Potter VR ( 1978) Phenotypic diversity in experimental hepatomas: the concept of partially blocked ontogeny. Br J Cancer 38.1-23
16. Bradstock Kf', Janossy G, Bollum FJ, Milstein C (1980) Anomalous gene expression in human thymic acute lymphoblastic leukaemia (Thy-ALL). Nature 284.455-457
17. Greaves MF, Verbi W, Vogler L, Cooper M, Ellis R, Ganeshaguru K, Hoffbrand V, Janossy G, Bo!lum FJ (1979) Antigenic and enzymatic phenotypes of the pre-B subclass of acute lymphoblastic leukaemia. Leukemia Res 3.353-362
18. Greaves MF ( 1982) 'Target' cells, differentiation and clonal evolution in chronic granulocytic leukaemia: A 'model' for understanding the biology of malignancy. In: Shaw MT (ed) Chronic granulocytic leukaemia. Praeger, New York, pp 15-47
19. Greaves MF (1981) Biology of acute lymphoblastic leukaemia. 16th Annual Guest Lecture: Leukaemia Research Fund Publ
20. Greaves MF, Bell R, Amess J, Lister T A (to be published) What is 'undifferentiated' leukaemia?
21. Magrath IT, Ziegler JL (1980) Bone marrow involvement in Burkitt-s lymphoma and its relationship to acute B-cell leukemia. Leukemia Res 4.33-60
22. Greaves MF, Sieff C, Edwards P (1983) Monoclonal anti-glycophorin as a probe for erythroleukaemia5. Blood (in press)
23. Nagasawa K, Mak TW(1980) Phorbol esters induce differentiation in human malignant T Iymphoblasts. Proc Natl Acad Sci USA 77.2964-2968 24. Delia D, Greaves M, Newman R, Sutherland R, Minowada J, Kung P, Goldstein G (1982) Modulation of T leukaemic cell phenotype with phorbol ester. lnt J Cancer 29.23-31
25. Greaves MF, Janossy G (1978) Patterns of gene expression and the cellular origins of human leukaemia. Biochim Biophys Acta 516:193-230
26. Greaves MF (1979) Tumour markers, phenotypes and maturation arrest in malignancy. A cell selection hypothesis. In Boelsma E, Riimke P (eds) Tumour markers. Elsevier, Amsterdam, pp 201-211
27. Varmus H (1982) Recent evidence for oncogenesis by insertion mutagenesis and gene activation. In. Greaves M (ed) Leukaemia cell differentiation. Cancer Surveys, vol 2. ICRF, London, pp 309-320
28. Schneider C, Sutherland R, Newman R, Greaves M (1982) Structural features of the cell surface receptor for transferrin that is recognised by the monoclonal antibody OKT9.J Bioi Chem 251:8516-8522
29. Goodfellow PN, Banting G, Sutherland R, Greaves M, Solomon E, Povey S (1982) Expression of the human transferrin receptor is controlled bye gene on chromosome 3. assignment using the species specificity of a monoclonal antibody. Somatic Cell Genet 8.197-206
30. Vodinclich L, Sutherland DR, Schneider C, Newman R, Greaves MF (1983) The receptor for tran5ferrin may be a 'target' structure for natural killer cells. Proc Natl Acad Sci (in press)
31. Ritz J, Pesando JM, Notis-McConarty J, Lazarus H, Schlossman SF ( 1980) A mono clonal antibody to human acute lym phoblastic leukaemia antigen. Nature 283.583-585 32. Sutherland R, Smart J, Niaudet P, Greaves MF (1978) Acute lymphoblastic associated antigen. 11. Isolation and partial characterization. Leukemia Res 2: 115-126
33. Newman RA, Sutherland R, Greaves Mf' ( 1981) The biochemical characterization of a cell surface antigen associated with acute lymphoblastic leukemia and lymphocyte precursors. J ImmunoI126:2024-2030
34. Kersey JH, LeBien TW, Abramson CS, Newman R, Sutherland R, Greaves M (1981) p24. a human leukemia-assocjated and Iymphohemopoietic progenitor cell surf'ace structure identified with monoclonal antibody J Exp Med 153.726- 731
35. Newman RA, Sutherland DR, LeBien TW, Kersey JH, Greaves Mf' (1982) Biochemical characteriZiltion of a leukaemia-associated antigen (p24) defined by the monoclonal antibody BA-2. Biochim Biophys Acta 701:318-327
36. Newman R, Greaves MF (1982) Characterisation of HLA-DR on leukaemic cells. Clin Exp ImmunoI50.41-50
37. Terhorst C, Van Agthovan A, LeCJair K, Snow P, Reinherz E, Schlossman S ( 1981 ) Biochemical studies of the human thymocyte cell-surf'ace anti~ens T6, T9 and TIO. Cell 23.771-780 v
38. Verbi W, Greaves MF, Schneider C, Koubek K, Jano55y G, Stein H, Kung P, Goldstein G (1982) Monoclonal antibodies OKTIO and OKTIIA have pan T reactivity and b1ock sheep erythrocyte 'receptors'. Eur J Immuno11281-86
39. Brodsky FM, Parham P, Barnstable CJ, Crumpton M, Bodmer WF (1979) Hybrid myeloma monoclonal antibodies against MHC products. Immunol Rev 47.3-61
40. Vogler LB, Crist WM, Bockman DE, Pearl ER, Lawton AR, Cooper MD (1978) Pre-B cellleukemia. a new phenotype of childhood lymphoblastic leukemia. N Engl J Med 298.872-878
41. Korsmeyer SJ. Hieter PA. Ravetch. JV. Poplack DG. Waldmann TA. Leder P (1981) Developmental hierarchy of immunoglobulin gene rearrangements in human leukemic pre-B cells. Proc Natl Acad Sci USA 78.7096-7100
42. Dewji N. Rap5on N. Greaves M, Ellis R (1981) Isoenzyme profiles of lysosomal hy drolases in Ieukaemic cells. Leukemia Res 5.19-27
43. Reddy EP, Reynolds RK, Santos E, Barbacjd M (1982) A point mutation is responsible for the acquisition of transforming properties by the T24 human bladder carcinoma oncogene. Nature 300.149-152