Educational Session 1 Chairman: W.E. Fibbe Haematologica 1999; 84:(EHA-4 educational book):1-3 Biology of normal and neoplastic progenitor cells

Emergence of the in the human embryo and foetus MANUELA TAVIAN,* FERNANDO CORTÉS,* PIERRE CHARBORD,° MARIE-CLAUDE LABASTIE,* BRUNO PÉAULT* *Institut d’Embryologie Cellulaire et Moléculaire, CNRS UPR 9064, Nogent-sur-Marne; °Laboratoire d’Étude de l’Hé- matopoïèse, Etablissement de Transfusion Sanguine de Franche-Comté, Besançon, France

he first haematopoietic cells are observed in es. In this setting, the recent identification in animal the third week of human development in the but also in human embryos of unique intraembryonic Textraembryonic yolk sac. Recent observations sites of haematopoietic stem cell emergence and pro- have indicated that intraembryonic liferation could be of particular interest. occurs first at one month when numerous clustered We shall briefly review here the successive steps of CD34+ Lin– haematopoietic cells have been identi- human haematopoietic development, emphasising fied in the ventral aspect of the aorta and vitelline the recent progresses made in our understanding of . These emerging progenitors express tran- the origin and identity of human embryonic and fetal scription factors and growth factor receptors known stem cells. to be acting at the earliest stages of haematopoiesis, and display high proliferative potential in culture. Primary haematopoiesis in the human Converging results obtained in animal embryos sug- embryo and foetus gest that haematopoietic stem cells derived from the As is the case in other mammals, human haema- para-aortic mesoderm – in which presumptive endo- topoiesis starts outside the embryo, in the yolk sac, thelium and -forming activity could be detect- then proceeds transiently in the before getting ed as early as 3 weeks in the human embryo by dif- stabilised until adult life in the marrow. Only T ferential expression of the CD34 and Flk-1/KDR genes are produced in the same tissue at – play an essential role in the foundation of definitive embryonic, foetal and postnatal stages. haematopoiesis. Aorta-associated CD34+ cells also The yolk sac represent a unique localised accumulation of primi- tive haematopoietic stem cells worthy of in-depth It is at about 18.5 days of development (early head molecular characterisation. Differential screening of process) that primitive haematopoietic cells appear a cDNA library has already revealed the expression of inside forming blood vessels in the intermediate novel genes in this population, one of which appears mesodermal cell layer of the human yolk sac wall. to be involved in the development of both haema- Studies on human haematopoietic cell emergence at these early stages are scarce, but our own observa- topoietic and endothelial cells. Active blood forma- 1,2 tion is observed in the liver and by the tions suggest that the sequence described in animal end of the first trimester. Inception of haemato- models also applies to the human yolk sac: meso- poiesis occurs earlier in the liver, where CD34+ cells derm-derived clusters of primitive haematopoietic are detected as early as 30 days, than in the marrow, stem cells – the – develop in close asso- where haematopoietic cells are not observed before ciation with the endothelium of emerging blood ves- week 11. sels, possibly from a common ancestor cell or hae- Current interest in early human blood cell ontoge- mangioblast. The coexpression of the CD34 surface ny may be partly related to the growing use of foetal molecule by haematopoietic precursor cells and stem cells for transplantation at postnatal stages, endothelial cells can be traced back to these initial stages, which may support the hypothesis of their and to emerging cell and/or gene therapies of the 1,2 3 blood system in utero, which justify a thorough char- common origin. Migliaccio et al. described several acterisation of embryonic and foetal human haema- generations of clonogenic progenitors in the human topoiesis. In addition, the prenatal haematopoietic yolk sac from 4.5 weeks of development, including system is characterised by an outstandingly high rate pluripotential (CFU-GEMM), granulomonocytic of progenitor cell expansion, migration and differ- (CFU-GM) and erythroblastic progenitors (BFU-E entiation and hence can be seen as a privileged mod- and CFU-E). The human yolk sac starts regressing at el to identify novel factors involved in these process- about 45-50 days post-ovulation and virtually all clonogenic progenitors have disappeared from that tissue by week six. Correspondence: Bruno Péault, Unité 506 INSERM, Bâtiment Lavoisier; The liver Groupe Hospitalier Paul Brousse, 12, avenue Paul Vaillant-Couturier, th 94807 Villejiuf Cedex, France. Tel. international +33-1-45595263 – The liver emerges during the 4 week of develop- Fax: international +33.1.45595268 – E-mail: [email protected] ment when the hepatic bud, an endodermal out- 2 M. Tavian et al. growth of the foregut, invades the adjacent mesoder- Concepts on the filiation of the stem cells that mal septum transversum. These two tissues con- found definitive haematopoiesis have changed in the tribute hepatocyte cords and vascular sinuses, respec- past few years with the demonstration that, in ani- tively. We have detected CD45+CD34– haematopoi- mals, these emerge inside the embryo, and not in the etic cells from day 23 of development in the liver yolk sac as previously believed. In mice and birds the anlage while the first CD34+ haematopoietic progen- original blood-forming territory develops intrinsically itors could be recognised on day 30.1 In vitro colony- in the para-aortic splanchnopleural mesoderm con- forming cells, i.e. BFU-E, CFU-GM and, slightly later, stituting the presumptive aorta-gonad-mesonephros CFU-E have been indeed detected at 4.5-5 weeks in (AGM) region of the embryo and contributes, at pre- the liver rudiment, where their frequency then increas- liver stages, multilineage haematopoietic progenitors es dramatically, paralleling their sharp decline in the and eventually long-term reconstituting true haema- yolk sac.3 At the end of the first trimester, and topoietic stem cells. The existence in the human onwards, more primitive progenitors – CFU-GEMM embryo of an equivalent site of haematopoietic stem and HPP-CFC – have also been detected in the liver. cell generation has been suggested by the identifica- Earlier studies, confirmed by more recent immuno- tion, at 4-6 weeks of development, of numerous clus- histochemical approaches, have documented the tered CD34+ haematopoietic cells on the ventral extensive erythro-myeloid haematopoiesis that takes endothelium of the aorta and vitelline artery.1,6 These place extravascularly in the human embryonic and cells express surface antigens that typify early blood foetal liver, and have stressed the prominence of ery- cell progenitors, being CD45+, CD34++, CD31+, thropoiesis therein (reviewed in ref. #4). Other CD43+, CD44+, CD164+, but display no CD38 or lin- myeloid cells present in the haematopoietic liver are eage-specific markers. In situ hybridisation on embryo , and rare . sections and screening of cDNA libraries prepared B- has been traced in the liver from from these sorted aorta-adherent progenitors have about 9 weeks of gestation by detection of surface also revealed that they express genes known to be IgM+ cells. associated with the early steps of haematopoietic development, such as Tal1/SCL, c-myb, GATA-2, The bone marrow GATA-3, flk-1/VGEFR2 and c-kit.7 When directly A cartilaginous presumptive skeleton is present in assayed in methylcellulose, human intraembryonic the 6-8-week human embryo. Bone rudiments are aorta-associated CD34+ cells exhibited negligible + then surrounded by a dense network of CD34 capil- clonogenic potential. In contrast, following a 4-10- + laries, by CD68 and by osteoblast pre- day co-culture on murine bone marrow stromal cells cursors which all invade the diaphyseal at (MS-5 cell line), they generated about six times more 8.5-9 weeks. Incoming macrophages rapidly digest progenitors, which yielded large multilineage colonies the cartilage, leaving only intact small islets of chon- in methylcellulose, than the liver rudiment.6 Of note, drocytes that soon become surrounded by osteo- the para-aortic splanchnopleura – but no other intra- blasts, from which ossification proceeds in a typical- embryonic tissue – exhibited dramatic haematopoietic ly endochondral manner. In-between ossifying tra- potential in culture as early as day 23 of development, beculae, large vascular sinuses develop leading to the i.e. several days before CD34+ stem cells can actually completion, at about 10 weeks, of bone marrow cav- be identified on the aortic wall. This result, as well as ities.5 Marrow haematopoiesis starts during the 11th provocative semi-thin section histology pictures, sug- week of development in specialised mesodermal gest that haematopoietic stem cells emerge from structures or primary logettes, constituted by a loose mesoderm in that territory, and not merely migrate network of mesenchymal cells supported by dense fib- there from another location.6 rillar material and surrounding a central artery, inside Differential screening of a cDNA library built from which CD15+ granulocytes appear first, closely fol- sorted embryonic aorta-associated CD34+ cells with lowed by erythroid cells. Haematopoiesis then devel- probes prepared from embryonic liver and foetal bone ops dramatically in rapidly enlarging logettes which by marrow CD34+ stem cells is in progress. Several dif- week 15 are densely packed with cells of the erythroid ferentially expressed genes have already been found, and granulocytic series.5 one of which encodes a serine-threonine kinase which, interestingly, is co-expressed in all developing endo- Haematopoietic stem cell emergence in thelial and haematopoietic stem cells.7 This argues early human development for the existence of haemangioblasts, i.e. common prog- As mentioned above, the emergence of the CD34 enitors for vascular and blood cells. cell surface antigen in ontogeny seems to be contem- We have reported that KDR, the human homologue porary with the earliest commitment of mesodermal of VEGFR2/Flk1, is strongly expressed in the human cells in the 3-week yolk sac to haematopoiesis and embryo by endothelial cells but barely detectable in vasculogenesis. CD34 expression is then consistently the first haematopoietic stem cells arising in the wall detected on the surface of haematogenous cells in the of the aorta.7 Conversely, the CD34 protein is detect- liver and bone marrow. ed from early stages at the surface of both cell types. Session #1 – Biology of normal and neoplastic progenitor cells 3

We took advantage of this differential expression pat- in living models of overexpression and inactivation tern to trace the emergence of putative human hae- such as mouse ES cells, transgenic mice and zebrafish. mangioblasts and their segregation into endothelial A better understanding of the microenvironment of and haematopoietic lineages. A population of KDR+ these different haematopoietic tissues is of theoreti- CD34– mesoderm cells emerges in early-somatic cal and practical interest in order to unravel similari- human embryos, by the beginning of the 4th week of ties with or differences from adult bone marrow stro- gestation. During blood vessel formation these KDR+ ma in terms of critical cells and mediators;8 progress CD34– cells gradually co-express increasing levels of in this domain should be obtained via the generation CD34. Simultaneously, in the yolk sac, the sole of immortalised stromal cell lines. haematopoietic tissue at that stage, most haemato- poietic progenitors exhibit a KDR– CD34+ phenotype. Remarkably, as development proceeds, a KDR+ CD34– References compartment persists in the splanchnopleura until 1. Tavian M, Hallais MF, Péault B. Emergence of just prior to the emergence of aorta-associated intraembryonic hematopoietic precursors in the pre- haematopoietic cell clusters. This cell compartment liver human embryo. Development 1999; 126:793- may include the putative haemangioblastic precursor 803. of human haematopoietic and endothelial lineages.8 2. Cortés F, Debacker C, Péault B, Labastie MC. Differ- ential expression of KDR/VEGFR-2 and CD34 defines distinct stages of endothelial and hematopoietic devel- Conclusions opment in early human embryos. Mech Dev 1999; in The localised accumulations of haematopoietic press. stem cells observed along intra-embryonic artery walls 3. Migliaccio G, Migliaccio AR, Petti S, et al. Human embryonic hemopoiesis. Kinetics of progenitors and and in the yolk sac in the third-fifth weeks of gestation precursors underlying the yolk sac->liver transition. J probably reflect phases of progenitor cell emergence Clin Invest 1986; 78:51-60. and amplification that do not occur any later in devel- 4. Péault B, Touraine JL, Charbord P. Haematopoietic opment. These unique, transient haematogenous ter- stem cell emergence and development in the human embryo and fetus; perspectives for blood cell therapies ritories are presently being actively studied: decipher- in utero. Semin Neonatol 1999; in press. ing the molecular control of mesoderm commitment 5. Charbord P, Tavian M, Humeau L and Péault B. - towards blood cell lineages is of prime interest for ly ontogeny of the human marrow from long : developmental biologists, while haematologists sus- an immunohistochemical study of hematopoiesis and its microenvironment. Blood 1996; 87:4109-19. pect that novel factors are to be identified at these 6. Tavian M, Coulombel L, Luton D, San Clemente H, early stages that could be used to manipulate the sur- Dieterlen-Lièvre F, Péault B. Aorta-associated CD34+ vival/renewal/proliferation of adult haematopoietic hematopoietic cells in the early human embryo. Blood cells. Experiments to that end are based on in vitro cell 1996; 87:67-72. 7. Labastie MC, Cortés F, Roméo PH, Dulac C, Péault B. or culture of the mouse or human yolk sac, Molecular identity of hematopoietic precursor cells para-aortic splanchnopleura and derived AGM tis- emerging in the human embryo. Blood 1998; 92: sues, the ability of which to drive the expansion and 3624-35. differentiation of co-cultured stem cells is tested in 8. Cortés F, Deschazeaux F, Uchida N, et al. HCA, an immunoglobulin-like adhesion molecule present on homospecific or xenogeneic combinations. Much the earliest human hematopoietic precursor cells, is emphasis is also being put on subtractive cloning of also expressed by stromal cell in blood-forming tis- novel genes whose function can be tested in vitro and sues. Blood 1999; 93:826-37. Educational Session 1 Chairman: W.E. Fibbe Haematologica 1999; 84:(EHA-4 educational book):4-7 Biology of normal and neoplastic progenitor cells

Characterisation and biology of normal human haematopoietic stem cells ROB E. PLOEMACHER Department of Haematology, Erasmus University, Rotterdam, The Netherlands

aematopoiesis is a life-long process respon- ate life-long blood cells of a multiplicity of lineages. sible for replenishing both haematopoietic In the laboratory, these properties are often project- Hprogenitor cells and mature blood cells from ed on individual HSC and the assays used are tuned a pool of pluripotent, long-term reconstituting to reveal one or more of their abilities. From physical haemopoietic stem cells (HSC). The daily turnover in sorting of various haemopoietic grafts it has become a normal adult of approximately 1012 blood cells is apparent that HSC assays should at least test for tightly regulated, involving, in part, a complex inter- multilineage cell production over extended periods action between soluble and membrane-bound stim- of time (i.e. months rather than weeks), in vitro but ulatory and inhibitory cytokines and their corre- preferentially in vivo. Preferentially such assays should sponding receptors. The molecular cloning of these be able to allow HSC frequency analysis as well as haemopoietic growth factors and their receptors has assessment of the proliferative ability of (individual) been instrumental in the partial delineation of the HSC subsets. Assays that allow such a read-out pathways that lead from a single HSC to the various include (a) analysis of long-term stroma-supported terminally differentiated cells in the haemopoietic haemopoiesis and generation of progenitors (long- system. term culture, LTC) and frequency analysis using The HSC compartment is distributed through the miniaturised LTC’s in limiting dilution (cobblestone separated bone marrow (BM) locations and HSC are area forming cell or CAFC assay; LTC initiating cell or known to traverse in low numbers via the peripheral LTC-IC assay);2,3 (b) assessment of human multi-lin- blood (PB) between these compartments. The HSC eage haemopoiesis in the bone marrow of immun- compartment is heterogeneous in various aspects, odeficient mice, e.g. in NOD/SCID mice (Scid repop- possibly as a result of different mitotic histories of ulating cell or SRC assay),4 or (c) in sheep, where the the individual HSC caused by stochastic mechanisms human cells are infused during the pre-immune foetal that regulate cycling activity in a predominantly qui- stage (human-foetal sheep model).5 The LTC, which escent HSC population. In analogy to the situation in is performed in flask cultures, gives an insight into the mouse, the human HSC compartment probably the capacity of a graft to produce progenitor cells represents a hierarchy of primitive cells on the basis while it does not reveal how many HSC and progen- of decreasing ability to generate new HSC, decreas- itors contribute to it. The latter issue is highly relevant ing pluripotentiality and proliferative potential, and when differences are anticipated in the progenitor increasing turn-over rate. In a transplant setting this cell generating capacity of HSC e.g. on the basis of heterogeneity may be reflected in the different time disease, chemotherapy or in vitro methods for physi- periods that individual HSC clones contribute to cal or chemical purging or selection. As frequency long-term reconstitution of a conditioned host, irre- analysis requires a limiting dilution set-up of the assay spectively of whether this reconstitution includes all it will be evident that animal assays are unfit to enu- haemopoietic lineages, and whether these lineages merate HSC on a routine basis, but nevertheless, can be detected at the same moment in time.1 probably provide the investigator with a more rele- vant estimation of the total graft ability than do in vit- HSC assays ro assays. Typical drawbacks of all these assays are HSC cannot be detected in bone marrow smears that they take a long time to complete and are often because they occur with low frequencies and have no tedious to quantify. unique features that allows their enumeration. Extensive studies using physically sorted murine BM Depending on the HSC assay used, HSC frequencies have allowed regression analysis on the applicability in normal BM are estimated to be between 1 in 104 of the murine CAFC assay as an in vitro equivalent of to 106 cells. The HSC compartment is typically char- assays for a series of HSC subsets in vivo. These stud- acterised by its ability to self-replicate and to gener- ies have indicated that, both in vivo and in vitro, early- developing, short-lived clones are initiated by the transient repopulating colony-forming cells Correspondence: Dr. Rob E. Ploemacher, Dept. of Haematology, Eras- (CFU-S day-12), whereas later developing and more mus University, P.O. Box 1738, 3000 DR, Rotterdam, The Netherlands. Phone: +31-10-408.7603; Fax: +31-10-436.2315; Email: ploemach- permanent clones are descendants of more primitive, [email protected]. long-term repopulating HSC, that induce stable chi- Session #1 – Biology of normal and neoplastic progenitor cells 5 maerism for more than a year. The CAFC assay, there- is that the markers that we use may be promiscuous, fore, seems fit to quantify HSC subsets with different or may not relate to function following manipulation repopulation potentials. of HSC. Ex vivo HSC manipulation protocols, e.g. for In order to facilitate more rapid enumeration of expansion of HSC from umbilical cord blood cells HSC and progenitor cells other methods have been (UCB), somatic gene transfer or tumour cell purging, and are being investigated, including colony forma- include either stimulation with a variety of cytokines, tion and phenotypic analysis. Although stem cells may or use of selection procedures, or both. It has been form (multi-lineage) colonies in vitro, they are greatly frequently observed that large numbers of CD34+ cells outnumbered by the more mature progenitor cells and progenitors can be generated ex vivo, however, which therefore renders colony formation unsuitable without a substantial maintenance or increase of HSC for HSC enumeration in a graft. Determination of the as tested by functional assays (e.g. SRC, LTC-IC, replating efficiency of individual colonies may give CAFC). We have recently found over 105-fold expan- more information on the proliferative ability of sion of the CD34+/ CD38– cells in 11 week cultures of colony-forming cells, but, again requires a large time human umbilical cord blood CD34+ cells, whereas expenditure. LTC-IC/CAFC were only modestly (1-100-fold) Data from many laboratory and clinical investiga- expanded and although SRC were initially expanded, tions indicate that almost all lymphohaemopoietic no SRC activity was detected at 11 weeks of culture. stem cells and all their progenitor cells are contained While the freshly uncultured CD34+/ CD38– cells were in approximately 1% of human BM mononuclear cells Lin–, it could be shown that most of the cultured and that express the surface marker CD34. Because stem resorted CD34+/CD38– cells, although still containing cells are a rare cell type in the CD34+ cell population, all CAFC activity, expressed lineage markers in that investigators have subdivided the CD34+ cell popula- fraction. These data show that one should be tion in order to enrich stem cells further. The CD34+/ extremely careful in extrapolating a function-pheno- CD38– cell subset comprises less than 10% of human type relationship as defined in fresh specimen to oth- CD34+ adult BM cells (equivalent to < 0.1% of mono- er circumstances. nuclear cells), lacks lineage (i.e. it is Lin–) antigens, contains cells with in vitro replating and long-term HSC quiescence, chemosensitivity and haemopoietic culture capacity, and is predicted to be radiosensitivity highly enriched for in vivo repopulating stem cells. BM- Virtually none of the stem cells from the mobilised derived CD34+/CD38– cells also contain SRC and gen- peripheral blood (MPB) and BM is cycling. Indeed, + erate long-term, multilineage human haemopoiesis in LTC-IC activity is higher in CD34 cells isolated in G0 the human-foetal sheep in vivo model, whereas trans- than in those residing in G1. HSC show fer of CD34+/ CD38+ cells generates only short-term longer persistence of CD34 expression in suspension human blood cells. culture than do G1 phase HSC, and maintain in vitro Another interesting glycoprotein antigen is AC133, haemopoiesis for longer periods. The deep quiescence which is selectively expressed on CD34bright HSC and of most HSC explains why they are refractive to a sin- progenitor cells derived from human foetal liver and gle treatment with cycle-specific drugs (e.g. 5-fluo- BM, and PB. AC133-selected cells, which also include rouracil, cytosine arabinoside, hydroxyurea, vincri- some CD34– cells, engraft successfully in a foetal stine), whereas repeated chemotherapy may lead to sheep transplantation model, and human cells har- HSC activation and their recruitment into a drug-sen- vested from chimaeric foetal sheep BM have been sitive state. In contrast, some alkylating drugs, e.g. shown to engraft secondary recipients successfully, busulphan, preferentially kill the most primitive CAFC providing evidence for the long-term repopulating (week 6) while sparing most of the transiently repop- potential of AC133+ cells. ulating CAFC (week 2).7 Multiple rounds of chemo- It should be noted that enumeration of HSC using therapy may, therefore, not only result in decreased phenotypic analysis may, or may not, generate simi- numbers of HSC in the BM and PB, but may also lead lar data as those from in vitro or in vivo functional to loss of the ability of the individual HSC to produce assays. Firstly, it would require great skill, superb progeny. probes and instrumentation to allow rare event detec- Ample evidence in the murine model has shown tion for enumeration of HSC in unseparated grafts that the most primitive, long-term repopulating HSC using the presently available technology. Thus, and CAFC-week 6 have the lowest sensitivity for ion- although we may find that the bulk of HSC activity is ising radiation (gamma, X-ray and fission neutrons) contained in a CD34+/CD38– fraction, not every and display unexpected high sub-lethal damage (SLD) CD34+/CD38– is necessarily a HSC. Moreover, some repair. In contrast, transiently repopulating HSC, HSC that do not meet these criteria may be over- CAFC-week 2 and cells that form spleen colonies in looked as was demonstrated by the recent finding irradiated recipients (CFU-S) are highly sensitive and that perhaps the most potent, long-term repopulat- lack SLD repair.8 ing HSC do not express the CD34+ antigen but are The regulated localisation, conservation, commit- Lin–CD34–.4,6 A second issue with phenotypic analysis ment and terminal differentiation of undifferentiated 6 R. E. Ploemacher

HSC is believed to occur in niches or local area net- ed secondary hosts long-term. Quiescent HSC, espe- works in BM stromal microenvironments. This results cially those in G0, have been demonstrated to show in preservation of the stem cell pool while permitting more effective seeding in the BM than do G1 or cycling controlled cell proliferation and differentiation. The HSC. What is more, using PKH26-labelled murine exact nature of such niches is only slowly emerging cells that were enriched for either transient or long- from many experiments. It is clear that complex inter- term in vivo repopulation suggestive evidence was actions between stromal cells and haemopoietic stem found that even at 48 hours after transplant long- and progenitor cells involve cell adhesion molecules, term repopulating cells are still quiescent in the bone and extracellular matrix molecules that may bind and marrow (BM). Short-term ex vivo cycle progression of present elaborated cytokines and chemokines. Stro- HSC, in the absence of cell division, appears to reduce mal cells display membrane bound cytokines (e.g. the seeding efficiency and long-term engraftment stem cell factor) and their receptors (e.g. c-kit), and capacity of both human and murine HSC.9 specific heparan sulphate proteoglycans containing From murine studies it appears that only about 1 of high 6-O-sulphation on the glucosamine residues. 4 infused HSC homes to the total BM of a condi- These interactions lead to specific docking of haemo- tioned recipient, although other studies have sug- poietic cells where they co-localise with regulatory gested a 100 percent seeding efficiency. The homing molecules in an as yet unsufficiently characterised con- of human CAFC in the NOD/SCID model is extreme- text. Specific niches might exist that induce conserva- ly low and less than 1 percent of infused human BM- tion and maintenance of primitive progenitors and derived HSC can be recovered 24 hrs after injection other niches that promote proliferation and differen- from the total recipient’s BM (Van Hennik et al., unpub- tiation, depending on the specific cytokines and matrix lished results). components present within. A wide variety of adhesion molecules and other lig- In vitro, primitive HSC require combined stimula- ands that mediate cell-to-matrix and cell-to-cell inter- tion by multiple cytokines for growth, but some actions have been implicated in HSC adherence to vas- cytokines selectively promote viability rather than cular endothelium in the BM and subsequent trans- growth when acting individually. These cytokines, e.g. migration of haemopoietic progenitor cells across it. Interleukin-3 (IL-3), the ligands for c-kit (KL) and flt3 Specific HSC homing is probably a complex multistep (FL), but especially Tpo, exert direct and selective via- process of rolling, crawling and nesting of the HSCs. bility-promoting effects on a small fraction of CD34+ There is evidence that a family of selectins (L, P and E) CD38–, but not CD34+CD38+, human bone marrow can mediate initial tethering, rolling and subsequent progenitor cells at the single-cell level. Tpo mRNA is adhesion of HSCs to endothelial cells. In fact, all of the expressed in many stromal cell cultures, while there is different classes of adhesion molecules appear to play variable expression of KL and FL. roles in anchoring HSCs within the BM or the promo- tion of differentiation. Intercellular adhesion molecule Mobilisation and homing of HSC (ICAM-1) in the Ig family, very late antigen 4 (VLA-4), Although in the steady state low HSC numbers can an integrin, L-selectin and CD44 are examples of such be detected in the PB, numerous agents, varying from important molecules. Another important considera- lipopolysachharides to specific cytokines, are able to tion is the functional activity of these adhesion recep- increase the numbers of HSC in the blood dramati- tors. The integrins can be activated by different cyto- cally. The probable underlying drive of the haemo- kines, including GM-CSF, IL-3 and KL. The chemokine poietic system is (a threatening) depletion of the stromal cell-derived factor-1 (SDF-1), too, was found body’s HSC reserve due to infection, blood loss or to be critical for bone marrow engraftment. SDF-1 treatment with antineoplastic agents. It is clinically binds to its receptor CXCR4, which is expressed on useful that repeated infusions of some cytokines (e.g. many cell types including some CD34+CD38– cells. G-CSF), in combination or not with chemotherapy, SDF-1 attracts CD34+CXCR4+ HSCs and its important mobilise large numbers of quiescent HSC into the role in homing is illustrated by the absence of haemo- blood that can be harvested by leukapheresis and poiesis in the bone marrow of mice that lack SDF-1 or used for autologous or allogeneic transplantation. do not express CXCR4. Recently, Peled et al.10 demon- This method is an advantage for the stem cell donors strated that KL and IL-6 induce CXCR4 expression on who would otherwise have to undergo anaesthesia human CD34+ cells. CXCR4 expression potentiates and repeated bone marrow punctures. migration to SDF-1 and engraftment in primary and Following their transplantation, the HSC have to secondary transplanted NOD/SCID mice. Moreover, find the BM niches that guarantee their life-long reg- anti CXCR4 antibody completely abrogated stem cell ulated preservation and outgrowth according to the engraftment in this model. body’s demands. Homing of transplanted HSCs in the recipient BM cavity is thus a critical step in the The use of HSC in clinics establishment of long term haemopoiesis after BMT, Primitive HSC and progenitors from BM, MPB and as only the cells that home to the marrow in a murine recently also UCB are targets for high-dose chemo- model are capable of reconstituting lethally irradiat- therapy or radiotherapy. With the discovery of the Session #1 – Biology of normal and neoplastic progenitor cells 7 currently known cytokines and chemokines and the Frequency analysis of human primitive haematopoi- improved definition of culture ingredients, liquid cul- etic stem cell subsets using a cobblestone area form- tures of unmanipulated or physically sorted human ing cell assay. 1994;8:1095-104. 4. Bhatia M, Bonnet D, Murdoch B, Gan OI, Dick JE. A BM, MPB, UBC and foetal liver have been and will be newly discovered class of human hematopoietic cells increasingly used in an attempt to expand repopulat- with SCID-repopulating activity. Nat Med 1998; 4: ing HSC, purge malignant cells and permit somatic 1038-45. gene therapy. Other potential clinical applications of 5. Civin CI, Almeida-Porada G, Lee MJ, Olweus J, Ter- these ex vivo graft manipulations include T-cell deple- stappen LW, Zanjani ED. Sustained, retransplantable, multilineage engraftment of highly purified adult tion for allogeneic HSC cell transplantation, adoptive human bone marrow stem cells in vivo. Blood 1996; immunotherapy via T-lymphocytes that are grown 88:4102-9. and educated in culture, and haemopoietic support 6. Zanjani ED, Almeida-Porada G, Livingston AG, Flake for haemopoietically compromised patients. Still AW, Ogawa M. Human bone marrow CD34- cells more extensive study is required in these areas to over- engraft in vivo and undergo multilineage expression come issues such as loss of repopulating stem cells that includes giving rise to CD34+ cells. Exp Hematol 1998; 26:353-60. due to manipulation, and inefficient gene transfer and 7. Down JD, Ploemacher RE. Transient and permanent expression in human HSC and their progeny. engraftment potential of murine hemopoietic stem cell subsets: differential effects of host conditioning with gamma radiation and cytotoxic drugs. Exp References Hematol 1993; 2:913-21. 8. Down JD, Boudewijn A, Van Os R, Thames HD, 1. Lemischka IR. What we have learned from retroviral Ploemacher RE. Variations in radiation dose survival marking of hematopoietic stem cells. In: Muller- among different bone marrow hemopoietic cell sub- Sieburg C, Torok-Storb B, Visser J, Storb R, eds. Cur- sets following fractionated irradiation. Blood 1995; rent topics in microbiology and . Hema- 86:122-7. topoietic stem cells, vol. 177. New York: Springer Ver- 9. Van der Loo JCM, Ploemacher RE. Marrow and spleen lag; 1992. p. 59-71. seeding efficiencies of all murine hematopoietic stem 2. Sutherland HJ, Lansdorp PM, Henkelman DH, Eaves cell subsets are decreased by preincubation with AC, Eaves CJ. Functional characterization of individ- hematopoietic growth factors. Blood 1995; 85:2598- ual human hematopoietic stem cells cultured at lim- 606. iting dilution on supportive marrow stromal layers. 10. Peled A, Petit I, Kollet O, et al. Dependence of human Proc Natl Acad Sci USA 1990; 87:3584-8. stem cell engraftment and repopulation of NOD/SCID 3. Breems DA, Blokland EAW, Neben S, Ploemacher RE. mice on CXCR4. Science 1999; 283:845-8. Educational Session 1 Chairman: W.E. Fibbe Haematologica 1999; 84:(EHA-4 educational book):8-10 Biology of normal and neoplastic progenitor cells

Gene regulation in normal and leukaemic progenitor/stem cells RUGGERO DE MARIA,°+ FRANCESCO GRIGNANI,+^ UGO TESTA,+ MAURO VALTIERI,°+ BENEDIKT L. ZIEGLER,* CESARE PESCHLE°+ °Kimmel Institute, T. Jefferson University, Philadelphia, PA, USA; + Dept. of Haematology and Oncology, Istituto Superiore di Sanità, Rome, Italy; * Dept. of Haematology and Oncology, University of Tübingen, Tübingen, Germany; ^ Istituto di Medicina Interna e Scienze Oncologiche, University of Perugia, Italy

New approaches for analysis of gene humans and flk-1 in mice) in murine embryonic regulation in normal and leukaemic haematoangiogenesis. Targeted gene disruption stud- haematolymphopoietic progenitor/ ies indicate that flk-1 is required for initiation of prim- stem cells (HPCs/HSCs) itive/definitive haematolymphopoiesis and vasculo- 5 While recent studies provided insight into gene reg- genesis: this suggests a role for flk1 in the generation ulation in normal and leukaemic HPCs/HSCs new of haemoangioblasts, i.e., hypothetical stem cells approaches and model systems may be required to bipotent for haematolymphopoietic and endothelial further our understanding in diverse key areas. Specif- lineages. ically, (i) HPCs have been stringently purified and We observed that in human post-natal haemato- extensively characterised, but isolation and pheno- poietic tissues [CB, BM, normal or mobilised periph- + typing of HSCs is still unsatisfactory; (ii) novel eral blood (PB, MPB)] CD34 cells comprise 0.1-0.5% methodology is required for analysis of gene regula- cells expressing KDR. Pluripotent HSCs repopulating tion at single HPC/HSC level; (iii) delineation of neg- the haematolymphopoietic tissues (assayed in both ative regulatory mechanisms for HPCs/HSCs is still NOD-SCID mice and foetal sheep xenografts) and unsatisfactory: particularly, the role of apoptotic putative HSCs [assayed in 12-wk extended Dexter type mechanisms in normal haematopoiesis is uncertain; long-term culture (LTC) as LTC initiating cells (LTC- 6 7 (iv) finally, while leukaemogenetic models have been ICs) or cobblestone area forming cells (CAFCs) ], are + + established in immortalised cell lines and transgenic restricted to and highly enriched in the CD34 KDR animals, an in vitro system for leukaemic transforma- subset. Conversely, oligo-unipotent HPCs with no tion of normal HPCs/HSCs is still unavailable. This self-renewal capacity are restricted to and highly puri- + – report reviews recent advances addressing these fied in the CD34 KDR cell fraction. Based on limit- aspects. ing dilution analysis, the HSC frequency in CD34+KDR+ cells is 22% in BM by NOD-SCID mice The KDR receptor allows isolation and assay and 25-42% in BM, PB, CB by 12-wk LTC assay. characterisation of normal HSCs The latter enrichment values rises to 53-63% in LTC The haematolymphopoietic hierarchy is defined by supplemented with VEGF, thus suggesting a func- functional assays. Pluripotent HSCs, endowed with tional role for the VEGF/KDR system in HSCs: the extensive self-renewal capacity, are assayed in vivo on purification index rises yet further to > 95% in the + + the basis of their capacity to repopulate the haema- CD34 KDR cell subfraction resistant to prolonged tolymphopoietic system,1 i.e., to xenograft irradiated GF starvation in culture. NOD-SCID mice2 and pre-immune sheep foetuses.3 These results indicate that KDR is a novel function- The major hurdle in HSC studies has been the lack al marker defining pluripotent repopulating HSCs and of a specific positive marker, comparable to CD34 for distinguishing them from oligo-unipotent HPCs: these early haematopoietic precursors:4 this hampered findings pave the way to characterisation and func- purification, characterisation and utilisation of this tional manipulation of HSCs/HSC subsets, as well as extremely rare cell population. Indeed, HSCs are usu- innovative approaches for HSC clinical utilisation. ally enriched in the CD34+/CD38– fraction, which comprises only 0.1-0.2% repopulating HSCs in adult Unicellular-unilineage erythropoietic bone marrow (BM) and cord blood (CB).1 cultures: molecular analysis of While HSC identification is still elusive, recent regulatory gene expression at sibling observations have suggested a role for vascular endo- cell level thelial growth factor receptor 2 (VEGFR2, KDR in In vitro studies on haematopoietic control mecha- nisms have been hampered by the heterogeneity of the analysed cell populations, i.e., lack of lineage speci- Correspondence: C. Peschle, M.D., Kimmel Cancer Institute, T. Jeffer- ficity and developmental stage homogeneity of prog- son University, B.L.S.B., Room # 902, 233 South 10th Street, Philadel- phia, PA 19107-5541, USA. Tel. international +1-215-5031763; fax. enitor/precursor cells growing in culture. We devel- international +1-215-9231098 – e-mail: [email protected]. oped unicellular culture systems for unilineage differ- Session #1 – Biology of normal and neoplastic progenitor cells 9 entiation of purified HPCs followed by daughter cell purified HPCs undergoing unilineage erythroid differ- analysis at cellular and molecular levels.8 In the culture entiation9 showed that Fas is rapidly upregulated in system reported here (i) the GF stimulus induces CB early erythroblasts and expressed at high levels through HPCs to proliferate and differentiate/mature exclu- terminal maturation. However, Fas crosslinking was sively along the erythroid lineage;9 (ii) this erythropoi- effective only in less mature erythroblasts, particular- etic wave is characterised by < 4% apoptotic cells; (iii) ly at basophilic level, where it induced apoptosis asymmetric divisions are virtually absent, i.e., non- antagonised by high levels of Epo. In contrast, FasL responsive HPCs with no erythropoietic potential are was selectively induced in late differentiating Fas- forced into apoptosis; (iv) the system is cell division insensitive erythroblasts, mostly at the orthochromat- controlled, i.e., the number of divisions performed by ic stage. FasL is functional in mature erythroblasts, as each cell is monitored. Single-cell reverse transcriptase it was able to kill Fas-sensitive lymphoblast targets in (RT)-PCR analysis was applied to this culture system to a Fas-dependent manner. Importantly, FasL-bearing investigate gene expression of diverse receptors, mark- mature erythroblasts displayed a Fas-based cytotoxi- ers of differentiation and transcription factors (EKLF, city against immature erythroblasts which was abro- GATA-1, GATA-2, NF-E2, PU.1, SCL/Tal1) at discrete gated by high levels of Epo. stages of erythropoietic development. Freshly isolated These findings suggest the existence of a negative CD34+ cells expressed CD34, c-kit, PU.1 and GATA-2 regulatory feed-back between mature and immature but did not express CD36, erythropoietin receptor erythroid cells, whereby the former cell population (EpoR), SCL/Tal1, EKLF, NF-E2 or GATA-1 and glyo- might exert a cytotoxic effect on the latter one in the phorin A (GPA). In early to intermediate stages of ery- erythroblastic island. Hypothetically, this negative throid differentiation, we monitored the induction of feedback operates at low Epo levels to moderate the CD36, Tal1, EKLF, NF-E2 and GATA-1, which preced- erythropoietic rate; however, it is gradually inhibited at ed expression of EpoR. In late stages of erythroid mat- increasing Epo concentrations coupled with enhanced uration, GPA was upregulated, while CD34, c-kit, PU.1 erythrocyte production. Thus, the interaction of Fas and GATA-2 were barely or not detected. and FasL may represent a major apoptotic mechanism In addition, competitive single-cell RT-PCR was for , contributing to the regulation of used to assay CD34 mRNA transcripts in sibling homeostasis. CD34+CD38– cells differentiating in unilineage ery- throid cultures: this analysis allowed quantification of A novel in vitro leukaemogenic model: the gradual downmodulation of CD34 mRNA from ␣ HPCs through their differentiating erythroid progeny. PML-RAR expression in normal HPCs It is concluded that this novel culture system, cou- dictates the leukaemic phenotype pled with controlled single-cell RT-PCR analysis, may While the role of fusion proteins in acute myeloid eliminate the ambiguities intrinsic to molecular stud- leukaemia (AML) is well recognised, the leukaemic tar- ies on heterogeneous populations of haematopoietic get cell and the cellular mechanisms generating the progenitors/precursors growing in culture, particular- AML phenotype are largely unknown. To address this ly in the initial stages of development. issue we have established a novel in vitro leukae- mogenic model: highly purified human HPCs/HSCs Apoptotic role of fas/fas ligand system are transduced with retroviral vectors carrying cDNAs in the regulation of erythropoiesis of the fusion protein and the green fluorescent protein, In the BM, erythropoiesis occurs in discrete anatom- purified to homogeneity12 and induced into multi- or ic units, the erythroblastic islands, consisting of one or unilineage differentiation by specific GF combina- two macrophages surrounded by one or more rings of tions.9 ␣ erythroblasts at different maturation stages. The inner Expression of PML/RAR fusion protein in human erythroblastic layers contain immature cells, whereas HPCs/HSCs dictates the acute promyelocytic the more mature cells are at the periphery of the leukaemia (APL) phenotype, largely via previously island.10 This spatial association of mature and imma- unreported effects: (i) rapid induction of HPC/HSC ture erythroblasts may play an important role in ery- differentiation to the promyelocytic stage: this is fol- thropoiesis as homocellular cell-cell interaction seems lowed by maturation arrest, which is abolished by to be required for erythroid cell growth and matura- retinoic acid; (ii) reprogramming of HPC commitment tion.11 We speculated that maturating erythroblasts to preferential granulopoietic differentiation, irre- might deliver negative signals to neighbouring cells, as spectively of the HGF stimulus: transduction of single a consequence of a decreased requirement for ery- sibling HPCs formally demonstrated this effect; (iii) throid cell production. We therefore studied the pos- HPC protection from apoptosis induced by HGF sible involvement of Fas and Fas ligand (FasL) in the deprivation. A PML/RAR␣ mutated in the co-repres- regulation of erythropoiesis. sor N-CoR/histone deacetylase binding region13 lost Immunohistochemistry of normal BM specimens these biological effects, showing that PML/RAR␣ revealed that several immature erythroblasts undergo alters the early haematopoietic programme via N- apoptosis in vivo. Analysis of BM erythroblasts and CoR-dependent target gene repression mechanisms. 10 R. De Maria et al.

These observations identify the cellular mechanism chi G, Peschle C. Pure human hematopoietic progen- underlying development of the APL phenotype, show- itors: permissive action of basic fibroblast growth fac- tor. Science 1990; 249:1561-4. ing that the fusion protein directly dictates the specif- 5. Shalaby F, Ho J, Stanford WL, et al. A requirement for ic lineage and differentiation stage of leukaemic cells. Flk1 in primitive and definitive hematopoiesis and vas- Altogether, this model system allows us to analyse culogenesis. Cell 1997; 89:981-90. the proliferation/differentiation potential of trans- 6. Hao QL, Thiemann FT, Petersen D, Smogorzewska duced primary HPCs/HSCs starting from the earliest EM, Crooks GM. Extended long-term culture reveals a highly quiescent and primitive human hematopoietic phases after oncogene transfer through differentia- progenitor population. Blood 1996; 88:3306-13. tion/maturation along each haematopoietic pathway. 7. van der Loo JC, Ploemacher RE. Marrow- and spleen- The investigative potential of this approach is shown seeding efficiencies of all murine hematopoietic stem by the fact that it allowed reproduction, in primary cell subsets are decreased by preincubation with ␣ hematopoietic growth factors. Blood 1995; 85:2598- HPC culture, of the PML/RAR effects detected so far 606. in cell lines (i.e., maturation block, protection from 8. Ziegler BL, Müller R, Valtieri M, et al. Unicellular-uni- apoptosis, sensitisation to RA differentiative stimu- lineage erythropoietic culture: molecular analysis of lus),14 while unveiling novel biological actions of the regulatory gene expression at sibling cell level. Blood (in press). fusion protein (i.e., HPC reprogramming and rapid 9. Ziegler B, Testa U, Condorelli GL, Valtieri M, Peschle differentiation to the promyelocytic stage). We believe C. Unilineage hematopoietic progenitor differentia- that this model system will have a wide impact, in that tion in bulk and single cell cultures. Stem Cells 1998; it represents a novel experimental tool for studies 16:51-73. 10. Bernard J. The erythroblastic island: past and future. aimed at recapitulating in vitro the genetic events lead- Blood Cells 1991; 17:5-14. ing to haematopoietic neoplasias. 11. Hanspal M. Importance of cell-cell interactions in reg- ulation of erythropoiesis. Curr. Opin. Hematol. 4:142, 1997. References 12. Grignani F, Kinsella T, Mencarelli A, et al. High effi- ciency gene transfer and selection of human hemato- poietic progenitor cells with a hybrid EBV/retroviral 1. Ogawa M. Differentiation and proliferation of hema- vector expressing the green fluorescence protein. Can- topoietic stem cells. Blood 1993; 81:2844-53. cer Res 1998; 58:14-9. 2. Bhatia M, Wang JCY, Kapp U, Bonnet D, Dick JE. 13. Grignani F, De Matteis S, Nervi C, et al. Fusion pro- Purification of primitive human hematopoietic cells teins of the retinoic acid receptor-␣ recruit histone capable of repopulating immune-deficient mice. Proc deacetylase in promyelocytic leukaemia. Nature (Lond) Natl Acad Sci USA 1997; 94:5320-5. 1998; 391:815-8. 3. Zanjani ED, Almeida-Porada G, Livingston AG, Flake 14. Grignani F Jr, Ferrucci PF, Testa U, et al. The acute AW, Ogawa M. Human bone marrow CD34- cells promyelocytic leukemia specific PML/RAR␣ protein engraft in vivo and undergo multilineage expression inhibits differentiation and promotes survival of U937 that includes giving rise to CD34+ cells. Exp Hematol myeloid precursors. Cell 1993; 74:423-431. 1998; 26:353-60. 15. Rubnitz JE, Look AT. Molecular basis of leukemogen- 4. Gabbianelli M, Sargiacomo M, Pelosi E, Testa U, Isac- esis. Curr Opin Hematol 1998; 5:264-70. Educational Session 2 Chairman: F.K Stevenson Haematologica 1999; 84:(EHA-4 educational book):11-13 Biotherapy strategies in haematological malignancies

DNA vaccines against haematological malignancies FREDA K. STEVENSON, CATHERINE A. KING, MYFANWY B. SPELLERBERG, DELIN ZHU, JASON RICE, SURINDER SAHOTA, ANDREW THOMPSETT, JIRI RADL, TERRY J. HAMBLIN Molecular Immunology Group, Tenovus Laboratory, Southampton University Hospitals Trust, Southampton, UK

accination as a treatment for cancer is an attrac- attack, and we need to tailor our vaccines with this tive option, particularly in the setting of a low in mind. Vlevel of residual disease. However, the task of We have investigated idiotypic determinants, which activating a defeated to recognise and are expressed by the immunoglobulin of neoplastic B destroy persistent tumour cells is formidable. Haema- lymphocytes as defined clonal markers, with the pri- tological tumours are a major challenge since those vate idiotypic determinants being tumour-specific. cells have survived the full power of the immune sys- They arise from the processes of genetic recombina- tem, possibly by inducing tolerance. Three develop- tion and somatic mutation which occur during nor- ments, all arising from molecular biological technol- mal B-cell maturation, and are largely preserved fol- ogy, may allow us to overcome the anticipated obsta- lowing neoplastic transformation. Idiotypic Ig can be cles. First, there has been a large expansion in our a surface protein, as in most lymphomas, or a secret- knowledge of candidate tumour antigens; second, we ed protein, as in . Development of have a greater understanding of immune mecha- vaccines which can specifically suppress both these nisms; third, vaccine vehicles are being developed for categories of B-cell tumour would have relevance for delivery of tumour antigens via pathways able to acti- other tumour antigens found in these cellular sites. vate the most effective immune attack. Idiotypic protein vaccines have been found to induce Our focus is on tumours of B lymphocytes, which protective immunity against B-cell lymphoma in sev- include a broad range of diseases, with various cur- eral mouse models, largely mediated by anti-idiotyp- rent treatment options. Low grade tumours are ic antibody,1 and a small clinical trial in patients with responsive to chemotherapy, but are likely to relapse low grade lymphoma is showing promising results.2 and are usually incurable. Even tumours such as dif- However, idiotypic protein is difficult to prepare on fuse large cell lymphoma, which may be cured by an individual patient basis, and antibody is not like- standard chemotherapy, prove lethal in 60% of cas- ly to be useful if there is a significant level of secreted es, and plasma cell tumours such as multiple myelo- protein. Both these problems could be solved by ma have a poor outlook with current treatment. turning to DNA vaccines which are simple to con- However, patients with B-cell tumours can often struct, and which are known to activate achieve remission, in some cases with transplant sup- responses. port, offering an opportunity to intervene with an immunotherapeutic approach. For vaccines it is of DNA vaccines course necessary that remission allows recovery of DNA plasmid vaccines consist of a backbone of immune capacity. bacterial DNA containing a cDNA sequence encod- A widening range of potential tumour antigens is ing the potential antigen. Transcription is usually dri- being identified on B-cell tumours, including viral ven by the powerful CMV promoter, and stimulation antigens, mutated proto-oncogene products, onco- of the immune system occurs due to unmethylated foetal antigens, sequences arising from chromoson- CpG dinucleotide repeats within the backbone al translocation events, mucins and idiotypes. Can- sequence.3 Vaccination with DNA containing genes didate antigens can be placed into various categories depending on how they are expressed by the tumour from pathogenic organisms has been shown to be effective in inducing all arms of the immune response, cell. One group of antigens is expressed as glycopro- 4 teins at the cell surface; a second group is presented and in protecting against infection. Injection is usu- as peptides bound to MHC class I or II molecules; ally in a muscle or skin site, and a gene gun has been and a third group consists of secreted antigens. Each used to deliver DNA to intradermal sites. There is evi- category of tumour antigen will require an appropri- dence for direct transfection of antigen-presenting ate immune effector mechanism for successful cells (APCs) in skin, and possibly in muscle. Howev- er, delivery of antigen from a muscle site is probably mainly by secretion and uptake by APCs, or by cross- Correspondence: Prof. F.K. Stevenson, Molecular Immunology Group, priming. Clinical trials of DNA vaccines against infec- Tenovus Laboratory, Southampton University Hospitals Trust, Southampton, S016 6YD, UK. Tel. international +44-1703-796923 – tious diseases are in progress, with immune respons- Fax. international +44-1703-701416 – E-mail: [email protected] es being induced.5 12 F. K. Stevenson et al.

DNA vaccines against B-cell tumours ic fragment (fragment C[FrC]) activated anti-idiotyp- To construct a DNA vector for idiotypic vaccina- ic responses against a range of human scFv sequences tion, it is necessary to identify the VH and VL genes in mice, where scFv alone had failed.9 Importantly, used to encode tumour Ig, and this can be achieved the scFv appeared to fold effectively in the scFv-FrC 9 by PCR/cloning methods.6 We have chosen to assem- fusion protein. Promotion of immunity probably ble genes as single chain Fv (scFv), but it is also pos- results from both increased recognition by APCs, and sible to construct vectors to express whole Ig, con- by a massive increase in T-cell help. This help is pro- taining either human or mouse constant region vided by T cells specific for FrC and, since gene fusion is an absolute requirement,10 this acts via a classical sequences.7 It soon became clear that DNA vaccines hapten-carrier mechanism. containing scFv alone, or as homologous Ig, could To assess the generality of the approach for other not activate a significant anti-idiotypic immune tumour antigens, carcinoembryonic antigen (CEA), 8 response. The presence of human constant region we replaced the scFv with a different tumour-associ- sequences improved the antibody response to ated antigen, also expressed at the cell surface. A sim- attached mouse V-region sequences,7 and inclusion of ilar promotional effect of fusion was seen with this cytokine sequences or peptides could also increase antigen, with CEA sequence alone inducing poor anti- responses. However, we obtained a dramatic body responses, whereas the CEA-FrC fusion gene improvement in anti-idiotypic responses by fusing a induced very high levels of anti-CEA antibodies. This gene encoding a fragment derived from tetanus toxin suggests that the same fusion format may be useful to the 3’ end of the scFv. Attachment of this non-tox- for many surface-expressed tumour antigens.

Figure 1. Induction of protec- tive immunity against B-cell tumours by DNAscFv-FrC vac- cines. Mice were vaccinated with DNAscFv-FrC or DNAscFv derived from either: A, a mouse lymphoma (A31); or B, a mouse myeloma (5T33), at days 0, 21 and 42. Anti-idio- typic antibody levels against the tumour Ig were measured, and mice were then chal- lenged with tumour. In each case, anti-idiotypic antibodies and protection were induced by the DNAscFv-FrC fusion vaccine but not significantly by the DNAscFv alone. Session #2 – Biotherapy strategies in haematological malignancies 13

DNA vaccines induce protective rational design for raising both antibody and T-cell immunity against tumour mediated attack against lymphoma and myeloma, A DNA vaccine containing scFv-FrC fusion genes which express idiotypic antigen at the cell surface or derived from a mouse lymphoma, A31, was injected as a secreted protein respectively. Intriguingly, pre- into syngeneic mice and was found to induce anti- liminary data indicate that the fusion gene approach idiotypic antibody. Importantly, this vaccine protect- promotes antibody responses against a different cell ed mice against challenge with a malignant lym- surface tumour antigen, CEA. Strategies for using phoma10 (Figure 1). It appeared, therefore, that this DNA vaccines to induce attack on processed peptides design was effective as a vaccine against a cell surface bound to MHC class I molecules are also being devel- tumour antigen. We then tested the same vaccine oped. We hope and anticipate that all categories of design against a mouse myeloma, 5T33. This myelo- tumour antigen may be susceptible to this powerful ma is one of a series which show characteristics sim- new technology. The critical clinical requirement, ilar to human myeloma, including osteolytic lesions. however, will be to treat the presenting tumour with The neoplastic plasma cells are completely surface Ig- maintenance or restoration of immune capacity. We negative, but secrete monoclonal Ig. Interestingly, the await results of the preliminary clinical trials with great scFv-FrC fusion gene also induced protective immu- interest. nity against this tumour10 (Figure 1). Although the vaccine-induced high levels of anti-idiotypic antibody, it was unlikely that antibody could mediate protec- References tion. This was confirmed by the fact that vaccination with idiotypic protein/CFA, which induced compara- 1. George AJT, Tutt AL, Stevenson FK. Anti-idiotypic ble levels of anti-idiotypic antibody, completely failed mechanisms involved in suppression of a mouse 10 lymphoma, BCL. J Immunol 1987; 138:628-34. to protect against tumour. We were unable to define 2. Hsu F, Caspar CB, Czerwinski D. Tumor-specific idio- a motif in the V-gene sequences suitable for binding type vaccines in the treatment of patients with to MHC class I, and it is therefore likely, but not yet relapsed low-grade non-Hodgkin’s lymphoma. Blood proven, that CD4+ T cells are involved in protection. 1997; 89:3129-35. 3. Sato Y, Roman M, Tighe H, et al. Immunostimulato- Clinical trials of scFv-FrC DNA vaccine ry DNA sequences necessary for effective intradermal gene immunization. Science 1996; 273:352-4. A phase 1 clinical trial of idiotypic DNA vaccines 4. Ulmer JB, Donnelly JJ, Parker SE, et al. Heterologous containing only scFv has been carried out in 7 patients protection against influenza by injection of DNA with end-stage low grade follicular centre lymphoma encoding a viral protein. Science 1993; 259:1745-9. (FCL) largely to assess any toxicity. No ill effects were 5. Wang R, Doolan DL, Le TP, et al. Induction of anti- gen-specific cytoxic T lymphocytes in humans by a observed, and we have now been allowed to proceed malaria DNA vaccine. Science 1998; 282:476-80. with a second trial using DNA scFv-FrC fusion genes. 6. Hawkins RE, Zhu D, Ovecka M, et al. Idiotypic vacci- The patients have FCL in first remission. We shall nation against human B-cell lymphoma. Rescue of investigate escalating doses of DNA, and will mea- variable region gene sequences from biopsy material sure antibodies against both tumour-derived scFv, for assembly as single-chain Fv personal vaccines. Blood 1994; 83:3279-88. and against FrC. This will provide information on the 7. Syrenglas AD, Chen TT, Levy R. DNA immunization ability of patients to respond to both a tumour anti- induces protective immunity against B cell lymphoma. gen and an exogenous pathogen-derived antigen, Nat Med 1996; 2:1038-41. both delivered by DNA. 8. Stevenson FK, Zhu D, King CA, Ashworth LJ, Kumar S, Hawkins RE. Idiotypic DNA vaccines against B-cell Summary lymphoma. Immunol Rev 1995; 145:211-28. 9. Spellerberg MB, Zhu D, Thompsett A, King CA, Ham- DNA vaccines against cancer have to activate an blin TJ, Stevenson FK. DNA vaccines against lym- inadequate or damaged immune system in order to phoma. Promotion of anti-idiotypic antibody respons- attack residual cancer cells. Although the potential es induced by single-chain Fv genes by fusion to problem of tolerance may be overcome by transplan- tetanus toxin fragment C. J Immunol 1997; 159:1885- tation, provision of high levels of T-cell help is likely to 92. 10. King CA, Spellerberg MB, Zhu D, et al. DNA vaccines be an important factor in stimulating effective with single chain Fv fused to fragment C of tetanus immune pathways. The fusion gene approach toxin induce protective immunity against lymphoma appears to provide the required help, and offers a and myeloma. Nat Med 1998; 4:1281-6. Educational Session 2 Chairman: F.K Stevenson Haematologica 1999; 84:(EHA-4 educational book):14-18 Biotherapy strategies in haematological malignancies

Monoclonal antibodies in the treatment of non-Hodgkin’s lymphoma patients BERTRAND COIFFIER Hospices Civils de Lyon and Université Claude Bernard, Lyon, France

uring the last ten years, significant progress tope attached to the antibody as well as the antigen. has been made in the treatment of non- Despite the simplicity of the concept, the choice of DHodgkin’s lymphoma (NHL) patients. The the antigen and the antibody is difficult. The selection refinement of the classification of the lymphomas of a suitable antigen was the first step for these and the incorporation of prognostic indices in deci- treatments. The antigen must not be shared by criti- sion making1 have allowed identification of groups cal tissues such as haematopoietic stem cells, must of patients who may be cured and others that need not be associated with much toxicity if all target cells further research to find the correct therapeutic are eliminated, and only be present on lymphoma options. Therapeutic improvements came from the cells. Unfortunately, specific antigens for B or T lym- introduction of G-CSF allowing higher dose of cura- phoma cells are unknown and all antigens are shared tive drugs with less risk of severe infections, the devel- with the normal B or T cell counterparts. The target opment of high-dose therapy with autotransplanta- antigen must be present either on all lymphoma cells tion, and the addition of interferon to multidrug or on the self-renewing clonogenic population of lym- chemotherapy regimens. Through these options, phoma cells. This target antigen must preferentially around half of the patients with de novo NHL may have a critical role in the homeostasis of these lym- expect to be alive ten years later. However, the diffi- phoma cells and be necessary for the cell survival. culty of overcoming tumour cell resistance to stan- Lymphoma cells should not be able to escape the dard drugs, the toxicity of the newly developed regi- antibody effect through the development of antigen mens, and the ageing population of patients chal- variants, antigen-negative clones, the shedding of the lenge us to develop less toxic but more effective treat- antigen in the extracellular compartment, or the ments. modulation of the antigen on the cell surface. If The idea of developing monoclonal antibodies MoAb-toxin conjugates need to be internalised for (MoAb) against cancer cells, particularly lymphoma the toxin to access the critical cellular processes, cells, appeared more than 20 years ago with the unconjugated MoAbs must remain on the cell surface description of the different antigens found on cell to allow the Fc portion of the antibody to activate membranes. Since the first attempt reported in immunologic mechanisms. Antigen density and 1980,2 the exponential increase in the progress in MoAb binding affinity may influence the cytotoxic molecular biology and protein engineering has efficacy for unconjugated antibodies but radioim- recently culminated in the approval, by drug agen- munoconjugates emit particles with enough energy to cies, of rituximab and, in a near future, of radio- kill adjacent cells, cells with a low antigen density or labelled antibodies, for in vivo treatment of lymphoma non-antigen-bearing cells. However, they may kill vital patients. Although monoclonal antibodies were used normal cells and increase the toxicity of the treat- for a long time to purge haematopoietic stem cells ex ment. vivo before autotransplant or for T-cell depletion of The MoAb must reach tumour cells in every parts of allogeneic cells and they are currently being devel- the organism and at all sites of the disease. This may oped against non-lymphoma cancer cells, this review be a problem in large tumours that are poorly vascu- will only focus on the future roles for MoAb in the larised. The presence of circulating antigens may be a treatment of lymphoma patients. problem leading to rapid clearance of the MoAb. The Three main approaches have been used in the MoAb must not be eliminated through immunologic development of MoAb therapy. Unconjugated anti- mechanisms because of its own difference from the bodies mediate cell death through different mecha- host. When xenophobic Ab are used, rapid appear- nisms related to the antigen and the antibodies. Con- ance of human anti-mouse antibodies (HAMA) may jugated antibodies act through a toxin or a radioiso- alter the pharmacokinetics of the MoAb, particularly during the re-treatment phases, and then further decrease their activity. Genetic engineering has Correspondence: Prof. B. Coiffier, Service d’Hématologie, CH Lyon-Sud, allowed the humanisation of antibodies and the cre- 69495 Pierre-Bénite Cedex, France. Tel. international +33 478 861194 – Fax: international +33 478 866566 – e.mail: coiffier@hema- ation of chimaeric proteins with a small antigen-bind- tologie.univ-lyon1.fr ing mouse part and a large human constant Fc region. Session #2 – Biotherapy strategies in haematological malignancies 15

These chimaeric MoAbs have substantially lower such as CD20, and humanised antibodies. Rituximab immunogenicity and, thus, prolonged half-life. They (MabThera®) is the most extensively studied uncon- also have an improved ability to mediate complement- jugated MoAb to date. This chimeric antibody con- dependent cytotoxicity (CDC) and antibody-depen- sists of the murine variable regions from the 2B8 dent cell-mediated cytotoxicity (ADCC), which MoAb grafted onto a human IgG1 constant region. In increases their potency. a phase I trial, the dose limiting toxicity was not The MoAb may kill the lymphoma cells through a reached which attests to a low side-effect profile of variety of different mechanisms. Radioimmunocon- the drug.5,6 Most phase II trials used the dose of 375 jugates or immunotoxins kill cells through the emis- mg/m2 once a week for 4 weeks.7,8 These trials accrued sion of particles or the internalisation of the toxin. predominantly patients with indolent follicular lym- Unconjugated MoAb may trigger CDC or ADCC. phoma (FL), refractory to standard chemotherapy They may have direct cytotoxic effects on tumour regimens. In more than 200 patients, the response cells,3 either on blocking the binding of an endoge- rate was around 50%, with 6% complete responses, nous ligand, which deprives the cell of a critical sur- and responses were observed in different subgroups vival signal, or by mimicking it, which triggers growth with adverse prognostic features, such as previous arrest. These functions may potentiate the effect of autotransplant or bulky tumour. Many patients had chemotherapy. no detectable residual tumour cells in blood or bone A large variety of antigens may potentially be cho- marrow, even as detected by PCR analysis for the sen as targets. While the early trials focused on Ig idio- t(14;18) translocation (molecular remission). The type, the CD20 antigen is probably the ideal target for median time to response was about 2 months, with B-cell lymphomas. It is not expressed on stem cells or many patients showing progressive responses for sev- precursor B-cells but is found on normal mature B- eral months. This may be correlated with the long cells and malignant B-cells, with the exception of plas- half-life of the antibody, some patients having resid- ma cells and myeloma cells. It is usually present on all ual levels detectable 6 months after the last infusion. cells of the tumour clone. It is expressed in high den- Median time to progression in responding patients sity on all B-cell lymphoma but not chronic lympho- was longer than 12 months. Interestingly, patients cytic leukaemia cells. This antigen is stable in the who progressed after a first response could be re- membrane of B-cell, does not have any known vari- treated and 50% of them responded. Several succes- ant, is not shed, and does not modulate or internalise sive responses were observed in some patients. in response to antibody binding. While its biologic Because of these results drug agencies approved the function is not fully known, it appears to function as indication of rituximab for relapsing FL patients. a calcium channel and it either forms a membrane Most adverse events were minor and associated with pore or controls a pore that is involved in calcium the first infusion. They consisted primarily of fever, transfer during the cell cycle. The majority of the mol- chills, mild nausea, mild fatigue, or malaise. Rarely, ecule is within the membrane or inside the cell and patients developed more serious reactions, including there is a small loop of 40 amino acids outside the hypotension, bronchospasm, or sensation of throat cell. All anti-CD20 appear to bind to the same section swelling. These symptoms were usually managed by of this external loop except L26, which binds to an temporarily slowing or stopping the antibody infusion. intracellular epitope of the molecule. The most serious reactions were observed in patients with peripheral blood involvement or large tumours. Unconjugated MoAbs These reactions could be prevented by stopping the Unconjugated MoAbs constitute the simplest appli- cation of targeted MoAb treatment. Table 1 lists the different antigens that have been used. The first trials Table 1. Different antigens chosen for unconjugated MoAb used MoAbs directed against the idiotype of the sur- therapy in lymphoproliferative diseases. face Ig of lymphoma cells which certainly represents a unique tumour-specific antigen. Most of these trials Antigens Monoclonal antibodies Humanised were conducted by Levy at Stanford.4 In different trials, anti-idiotype antibodies produced responses in 50% to CD4 cMT412 Chimaeric 70% of the patients. Although the median duration of CD5 T101 No these responses was only 6 months, some patients with CD10 J5 No complete response had long remissions. However, CD19 CLB-CD19 No CD20 1F5 No patients relapsed with idiotype-negative cells. The pres- CD20 IDEC-C2B8 (rituximab) Chimaeric ence of circulating shed idiotype and the formation of CD21 OKB7 No HAMA further limited the efficacy of unconjugated CD25 Anti-TAC No MoAbs. This, associated with the constraint of making CD52 Campath-1M No (rat) anti-idiotype specific for each patient, precluded fur- CD52 Campath-1H Chimaeric ther development of this therapy. HLA-DR LYM-1 No Subsequently, investigators used pan-B antigens, Ig idiotype anti-idiotype No 16 B. Coiffier infusion once mild adverse reactions occur. Myelo- Studies are only beginning or have been short lived suppression was rare. As expected, the normal B-lym- with other MoAb and no or few further developments phocytes rapidly declined and recovered over 6 to 9 are expected with them. months. However, there was no increase in infection rate and no occurrence of opportunistic infections, Immunotoxins probably because Ig levels and T-cells remained nor- An alternative approach to increase the activity of mal. HAMA was observed in less than 1% of the MoAb is the development of an immunotoxin, a con- patients and HACA (anti-chimaeric antibody) was not struct conjugating the antibody to cytotoxic plant or observed. bacterial toxic proteins. The commonly used toxins, Subsequently, rituximab was used in patients with ricin and diphtheria toxin, are highly potent natural more aggressive B-cell lymphoma, mantle cell lym- products that disrupt protein synthesis. Unlike uncon- phoma (MCL) and diffuse large B-cell lymphoma jugated MoAb, immunotoxins must be internalised (DLCL). In a phase II trial, patients with relapsing dis- after antigen binding to allow the toxin access to the ease showed a 32% response rate.9 In another phase cytosol. Although the conjugation to MoAb confers II trial, mantle cell lymphoma patients, in first line ther- some target specificity, the toxin continues to medi- apy or relapsing, showed a 40% and 30% response ate non-specific toxicity to normal tissues. Deglyco- rate, respectively. Most of the responses in these stud- sylated ricin A-chain has been used to eliminate such ies were incomplete, with less than 10% being com- non-specific toxicity. plete responses, and median time-to-progression was The vast majority of the immunotoxin trials have less than 12 months. These results indicate that ritux- been phase I studies designed to determine the max- imab has an anti-tumour activity in nearly all B-cell imum tolerated dose (Table 2). These trials have lymphomas. Currently, it is being tested in chronic shown that therapeutic serum levels may be achieved lymphocytic leukaemia, post-transplant lymphopro- with tolerable toxicity. A relatively uniform toxicity has liferative diseases, and multiple myeloma. Although been observed with vascular leak syndrome, hepato- plasma cells do not express the CD20 antigen, there toxicity, and myalgia. A strong immunologic response is some indication that the clonogenic precursors may. against the construct or the toxin was observed and Rituximab has been used in conjunction with re-treatment was not feasible in most patients. The chemotherapy in FL and DLCL patients, mostly with different trials have shown a low response rate of 10% the CHOP regimen.10 In a small study of 40 patients to 25% partial responses without durable efficacy. The a response rate of 95% was reached with 55% com- future of this therapy will depend on decreasing toxi- plete responses. At time of publication, 74% of the city, decreasing immune response against the con- responding patients had not progressed with a medi- struct, and on increasing the anti-tumour activity. an follow-up of 29 months. Seven of the 8 patients with bcl-2-rearrangement converted to PCR negativi- Radiolabelled antibodies ty after completion of the treatment. The adverse These consist of a radionucleide, usually 131iodine events were not different from those expected from (131I) or 90yttrium (90Y) emitting –particles, coupled to the CHOP regimen or rituximab treatment. No spe- a MoAb (Table 3). These compounds may selective- cific toxicity was observed with the combination of ly deliver ionizing radiation to tumour cells. These these drugs. Similar results were described in DLCL agents seem to possess several advantages over oth- patients. Currently, large co-operative group ran- er antibody constructs: they do not rely on recruit- domised trials are addressing the use of rituximab in ment of patients’ immune effector mechanisms and conjunction with multidrug chemotherapy as first line the –particles are capable of killing cells from a dis- treatment or as maintenance after chemotherapy. tance of several cell diameters permitting the killing of Other trials are testing in vivo purging before the har- antigen-negative tumour cells. Most studies of radio- vest of peripheral stem cells for autotransplant. Until these results are complete, the definitive place of rit- uximab cannot be established. Another humanised MoAb, CAMPATH-1H, has Table 2. Antigens, antibodies, and toxins used for immuno- toxin therapy. undergone evaluation in different indolent B-cell pro- liferations. CAMPATH-1 MoAb is directed against CD52, an antigen expressed by B- and T-lymphocytes, Antigens Antibodies Toxins granulocytes, and monocytes but not by stem cells. A series of trials documented responses in CLL and T-cell CD5 anti-H65-RTA blocked ricin prolymphocytic leukaemia.11 Adverse events consisted CD19 IgG-HD37-dgA deglycosylated ricin A-chain of infusion-related reactions, grade 4 , CD19 anti-B4-bR blocked ricin thrombocytopenia, or lymphopenia, which led to pro- CD22 RFB4-dgA deglycosylated ricin A-chain found immunosuppression and multiple opportunis- CD25 DAB389 IL-2 truncated diphtheria toxin tic infections. Complementary studies are in progress CD25 DAB486 IL-2 truncated diphtheria toxin to define a safer schedule of administration. CD30 BerH2-saporin saporin Session #2 – Biotherapy strategies in haematological malignancies 17

Table 3. Radioimmunoconjugates used in the treatment of dose of the antibody followed by a trace-imaging dose lymphomas. labelled with 131I. One or two weeks later, the patient received an additional dose of unlabelled antibody Antigens Antibodies Isotopes Dose in mCi followed by a therapeutic dose of 131I calculated to produce 75 cGy to the whole body. A 60% response Ig anti-idiotype 90Y 10 - 55 rate was observed, with 27% CR, and a median dura- 131 HLA-DR LYM-1 I 25 - 1050 tion of CR longer than 1 year. In de novo patients, a CD5 T101 131I or 90Y 25 - 150 CD20 B1 131I 30 - 850 higher response rate was reported. Longer follow-up CD20 1F5 131I 600 in more patients is necessary to determine whether CD20 C2B8 90Y 10 - 60 this approach will result in long-term disease control. CD21 OKB7 131I 90 - 200 In phase I-II trials, 90Y-labelled 2B8 antibody admin- CD22 LL2 131I 15 - 350 istered after a rituximab infusion yielded a 82% CD25 Anti-TAC 90Y 5 - 70 response rate. Larger numbers of patients are needed CD37 MB1 131I 25 - 650 Ferritin Anti-ferritin 131I or 90Y 20 - 100 to determine this antibody’s place. Even greater response rates have been reported in studies that used myeloablative doses of 131I-labelled anti-CD20 antibody. Patients received a therapeutic labelled MoAb require careful dosimetry measure- infusion of the antibody, were isolated, and 10-12 ments before the administration of the therapeutic days later were given their cryopreserved stem cells.14 doses to ensure that the radiation doses delivered to Toxicity of this approach included severe infections tumour sites exceed the doses to normal tissues. A and cardiomyopathy. However, 70% to 80% of the large tumour burden, particularly in the spleen, can CR patients remained disease-free 18 months later. interfere with the distribution of radiation. To min- Longer follow-up and randomised study will deter- imise the doses distributed to normal tissues, a first mine whether this approach confers survival advan- infusion of non-radiolabelled (cold) MoAb is often tage over conventional high-dose regimens. administered to the patients. Although 131I and 90Y both emit ␤-particles, 90Y emits higher energy parti- Conclusions cles, which have a deeper tissue penetration, and 131I In conclusion, impressive responses have been doc- also emits –radiation and has a longer half-life. These umented with MoAb therapy, even if their role in the last characteristics may present safety concerns limit- treatment of lymphoma remains to be determined. ing its use to large centres with strict radiation isola- Although the unconjugated MoAb, and particularly tion. Both compounds require an onsite radiophar- rituximab, have showed activity as single agents, they macy and dosimetry calculations that may also limit may have a greater role in combination with chemo- their application. therapy or as maintenance in responding patients. Radiolabelled MoAb therapy is associated with Their ability to purge blood and bone marrow to an myelosuppression, toxicity not found with cold undetectable level of lymphoma cells may allow re- MoAb, although a considerable inter-patient vari- infusion of minimally contaminated haematopoietic ability has been observed. With low dose radiation stem cell harvests after high dose therapy. The com- therapy neutropenia and thrombocytopenia are bination of these MoAb with a radionucleide may observed 3 to 4 weeks after the infusion and may per- result in greater efficacy but a greater toxicity, partic- sist for 16 weeks but with higher doses these haema- ularly myelosuppression. The definitive use of these tological effects are more rapid, profound, and pro- MoAb cannot be recommended before the results of longed. Extensive bone marrow involvement can lead current prospective studies are available. to a greater binding and a larger radiation dose deliv- ered to the normal haematopoietic cells. In addition the same adverse effects as those which are observed References with cold MoAb treatment exist and the use of iodine 1. Coiffier B. Non-Hodgkin’s lymphomas. In: Cavalli F, conjugates can cause hypothyroidism. Murine MoAb Hansen HH, Kaye SB, eds. Textbook of medical oncol- therapy is associated with the appearance of HAMA ogy. London: Martin Dunitz Ltd; 1997. p. 265-87. that may limit the possibility of re-treatment. 2. Nadler LM, Stashenko P, Hardy R, et al. Serotherapy The trials with non-myeloablative doses have used of a patient with a monoclonal antibody directed 131 against a human lymphoma-associated antigen. Can- I targeted to HLA-DR, CD20, CD21, or CD22 anti- cer Res 1980; 40:3147-54. gens and have achieved responses in 5% to 80% of 3. Shan D, Ledbetter JA, Press OW. Apoptosis of malig- treated patients. Preliminary studies with the 131I- nant human B cells by ligation of CD20 with mono- labelled Lym1 antibody in refractory patients yielded clonal antibodies. Blood 1998; 91:1644-52. 12 131 4. Brown SL, Miller RA, Levy R. Antiidiotype antibody a 50% response rate. The I-labelled anti-CD20 therapy of B-cell lymphoma. Semin Oncol 1989; 16: antibody (Bexxar®) produced durable CR in patients 199-10. with recurrent FL.13 Patients received an unlabelled 5. Maloney DG, Liles TM, Czerwinski DK, et al. Phase I 18 B. Coiffier

clinical trial using escalating single-dose infusion of 10. Czuczman MS, Grillo-Lopez AJ, White CA, et al. Treat- chimeric anti-CD20 monoclonal antibody (IDEC- ment of patients with low-grade B-cell lymphoma with C2B8) in patients with recurrent B-cell lymphoma. the combination of chimeric anti-CD20 monoclonal Blood 1994; 84:2457-66. antibody and CHOP chemotherapy. J Clin Oncol 6. Maloney D, Grillo-López A, Bodkin D, et al. IDEC- 1999; 17:268-76. C2B8: Results of a phase I multiple-dose trial in 11. Osterborg A, Dyer MJS, Bunjes D, et al. Phase II mul- patients with relapsed non-Hodgkin’s lymphoma. J ticenter study of human CD52 antibody in previously Clin Oncol 1997; 15:3266-72. treated chronic lymphocytic leukemia. J Clin Oncol 7. Maloney D, Grillo-López A, White C, et al. IDEC-C2B8 1997; 15:1567-74. (rituximab) anti-CD20 monoclonal antibody therapy 12. DeNardo GL, DeNardo SJ, Goldstein DS, et al. Maxi- in patients with relapsed low-grade non-Hodgkin’s mum-tolerated dose, toxicity, and efficacy of I-131- lymphoma. Blood 1997; 90:2188-95. 8. McLaughlin P, Grillo-Lopez AJ, Link BK, et al. Ritux- Lym-1 antibody for fractionated radioimmunothera- imab chimeric anti-CD20 monoclonal antibody ther- py of non-Hodgkin’s lymphoma. J Clin Oncol 1998; apy for relapsed indolent lymphoma: half of patients 16:3246-56. respond to a four-dose treatment program. J Clin 13. Kaminski MS, Zasadny KR, Francis IR, et al. Iodine- Oncol 1998; 16:2825-33. 131. Anti-B1 radioimmunotherapy for B-cell lym- 9. Coiffier B, Haioun C, Ketterer N, et al. Rituximab phoma. J Clin Oncol 1996; 14:1974-81. (anti-CD20 monoclonal antibody) for the treatment 14. Press OW, Eary JF, Appelbaum FR, et al. Phase II trial of patients with relapsing or refractory aggressive lym- of I-131-B1 (anti-CD20) antibody therapy with autol- phoma. A multicenter phase II study. Blood 1998; 92: ogous stem cell transplantation for relapsed B cell lym- 1927-32. phomas. Lancet 1995; 346: 336-40. Educational Session 2 Chairman: F.K Stevenson Haematologica 1999; 84:(EHA-4 educational book):19-22 Biotherapy strategies in haematological malignancies

Antisense oligonucleotides for haematological malignancies FINBARR E. COTTER LRF Molecular Haematology Unit, Institute of Child Health, London UK

he growth, differentiation, appearance and to improved ASO molecules and the prospect of function of all living cells is dictated by pro- effective molecular therapy. Tteins. The code for each protein is stored in the form of double helix DNA within the nucleus of every Antisense oligonucleotide design cell. Expression of such a gene is performed by tran- Optimising choice of target scribing the base sequence of the DNA into single The ability of naturally occurring RNA to anneal is stranded messenger RNA (mRNA) which is able to move from the nucleus to the cytoplasmic compart- a crucial process for living cells. However, the ability ment of the cell were it is translated into a protein. to design artificial antisense oligonucleotides has been Many disease processes are characterised by inap- hampered by the frequent inability of the ASO to propriate or excessive expression of normal or chi- anneal successfully to the mRNA. Efficient RNA-ASO maeric proteins as a consequence of altered DNA annealing involves the interaction between highly transcription. Current drugs predominantly function structured RNA elements. Empirical trial of a large by altering protein expression within the target cell, selection of oligonucleotides for an mRNA sequence however, the process is far from efficient and often has tended to be the method for effective ASO selec- lacks specificity. It is an attractive proposition to use tion, with many ineffective molecules being discarded. novel approaches to block transcription or transla- A lack of understanding of the structure of RNA and tion of individual genes in order to lower disease- the rules governing the annealing properties underlie causing proteins. An emerging and powerful this failure. Two approaches have been effective in approach is the use of antisense oligonucleotides improving ASO design. The first is an empirical (ASO) consisting of short sequences of synthetic sin- approach, essentially generating a large number of gle stranded DNA or RNA (around 12 to 20 bases) synthetic oligonucleotides to cover all the possible complementary to aberrantly expressed genes. ASO selections and hybridising these against the target were reported to inhibit gene expression as far back mRNA. This may be performed by generating gridded as 1978 with the inhibition of the Rous sarcoma arrays of ASO logically covering all computations for virus. Understanding of their kinetics and mode of a mRNA onto a glass plate and then hybridising the action at this time was, however, poor. Their design labelled message against the grid. The ASO that can permits Watson-Crick base pairing with a particular access the RNA structure and anneal gives a positive mRNA, blocking its translation either sterically or by signal. Identification of a good ASO molecule is read- the action of ribonuclease H (RNAse H) enzyme to ily provided but the technique is limited to examining cleave the ASO-mRNA duplex. Some antisense oligo- a maximum of 400 bases of RNA for each grid. In nucleotide modifications inhibit gene expression by most cases this is quite adequate. A greater length of the formation of triplex DNA, inserting a stretch of gene specific RNA can be examined by hybridising the nucleotides into the major groove of the nuclear labelled RNA against random or semirandom libraries double helix which forms side-to-side hydrogen (pools of single stranded ASO) of oligonucleotides, bonds (Hoogsteen base pairs) with one of the the latter having been shown to be more effective. The strands. The resultant triple helix is unable to tran- library approach is more prone to some false negative scribe RNA. While triplex formation has the aesthet- results but does not require the polymer technology ic advantage of blocking protein production at source required to attach the oligonucleotides to the glass for the triplex technology is considerably more complex the gridded arrays. Both are extremely effective at and less advanced. Antisense oligonucleotides have identifying accessible RNA sites for ASO. The second a number of inherent problems. These include nucle- ASO selection approach is based on the computer ase degradation, cellular delivery, effective sequence supported structural design of RNA. The secondary specificity, formulation and in vivo pharmacokinet- structure of the target RNA may be predicted by the ics. Greater understanding of all these areas has lead programme mfold 2.0 and the structural parameters recorded. Favourable ASO structures are searched for by examining for a maximal number of external bases Correspondence: Finbarr E. Cotter, LRF Molecular Haematology Unit, and components and a minimised loop degree. Sim- Institute of Child Health, 30 Guilford Street, London WC1N 1EH, UK. Tel. international +44-171-8138191 – Fax. international +44-171- plistically this means an ASO that complements an 8138100 – E-mail. [email protected] mRNA region with a weak secondary structure will 20 F. E. Cotter effectively compete with the RNA structure, displace Cellular internalisation the mRNA-mRNA double stranded binding and Overall cellular efficiency of ASO uptake is poor in anneal to exert its antisense effect. Such a theoretical- their present form, being in the region of 1% of total ly based selection approach has improved ASO iden- dose. Oligonucleotides are polyanionic in nature and tification considerably although again some erroneous thus easy passage through the cell membrane cannot predictions still occur. It is becoming clear that some be predicted. However, an 80kDa cell surface protein mRNA sites are inaccessible to ASO due to their struc- with features of the CD4 family of genes is present on ture. These may include chimaeric juctional points the surface of most cells and is capable of binding to from chromosomal translocations (as found in some ASO with cellular internalisation occurring in a sat- ) limiting their use against tumour specific urable, size-dependent, manner that is compatible mRNAs. The site of optimal ASO sequences remains with receptor-mediated endocytosis. It has been equivocal on occasions, however, some logical themes shown that the major part of ASO internalisation are evolving. occurs via fluid phase endocytosis or pinocytosis. The cell type affects ASO internalisation. B lymphocytes Protection against nuclease degradation take up ASO more readily than T cells. Mitogens may Antisense technology has in the past suffered from also increase uptake. It is likely that internalisation is poor quality control for ASO manufacture and before primarily dependent on the product of concentration modifications are considered, an emphasis on good and time. Strategies to improve ASO uptake may quality manufacture is essential. Breakdown of improve efficacy. In vitro cationic lipids have been par- unmodified ASO (PO) by circulating enzymes (nucle- ticularly effective, but their application to animal in ases) leads to rapid removal of the effective agent vivo ASO delivery shows an altered tissue distribution, before it has a chance to enter the cell. Chemically predominantly to the and liver, which may limit modifying the oligonucleotide backbone will confer their application. The possibility of conjugating the resistance to nuclease degradation and considerably ASO to a polymeric drug delivery system has a greater extend its active half life from minutes to days. The potential for improved oligonucleotide delivery in vivo first generation of ASO (providing a negative charge and may also permit differential tissue targeting to to the molecule) were modified by changing the oxy- give greater organ specificity. Optimisation of an gen backbone to a sulphur atom and were termed intracellular delivery system remains an area for con- phosphorothioates (PS). An alternative substitution siderably more research if antisense therapy is to be with the methyl group (creating methylphosphonates improved further. [MP]) is as effective for nuclease protection, is rela- tively non-toxic to cells, but is poorly water soluble Mechanism of action and difficult to manufacture, making it relatively There are a number of mechanisms of ASO effect. impractical for in vivo application. PS may cause some Direct competition with aminoacyl tRNA for sites on non-specific inhibition of cell growth, and being the mRNA molecule with subsequent blockade of strongly protein bound do have some undesirable ribosomal reading represents one, while the forma- effects in vivo. However, they are readily water solu- tion of the mRNA-DNA duplex with ionic ASO (i.e. PS ble, easily manufactured on a large scale, readily inter- or Gapmers) and subsequent activation of the enzyme nalised by most cells and have the advantage of act- RNAse H leading to duplex breakdown is another. It ing as good templates for RNAse H action when is apparent that several mechanisms are responsible annealed with the mRNA target. This has led to the for ASO mediated inhibition of gene expression and design of some second generation ASO which com- it is important to demonstrate a decrease in the bine a central core of seven PS molecules (to retain amount of protein produced by the gene targeted. water solubility and RNAse H activity) with outer MP molecules, to reduce PS toxicities. These molecules Pharmacokinetics of in vivo therapy have been called Gapmers due to the central PS gap. There have been several recent reports of the in vivo A number of additional chemical substitutions have efficacy of antisense molecules targeting oncogenes in been formulated to improve nuclease protection. haematological malignancy using the severe com- These include morpholino, phosphoramidate, guani- bined immunodeficient (SCID) mouse as a model. dine, methoxyethoxy and peptide modifications. There is some knowledge about the pharmacokinet- Probably the most interesting is the peptide modifi- ics of PS oligonucleotides when injected intravenous- cation, to form peptide nucleic acid (PNA) oligonu- ly (IV), intraperitoneally (IP), or subcutaneously (SC) cleotides with a neutral charge and the ability to com- The kinetics are similar with all three routes of admin- bine directly with chromosomal DNA to form a triplex istration. PS molecules are strongly protein bound. blocking transcription of the RNA. Their main draw- Although 30% of the dose is excreted in the urine with- back is the lack of intracellular uptake. Effective PNA in 24hrs, there are detectable levels in most tissues ASO will require some form of cellular delivery system for up to 48hrs, with only 15-50% degradation. Plas- if they are to become useful molecules. There is much ma clearance by both routes is biphasic with an ini- scope still to improve the formulation of ASO. tial half life of 15-25 minutes and a second half life of Session #2 – Biotherapy strategies in haematological malignancies 21

20 to 40 hours. The current human studies have associated with overexpression of a specific gene or administered the ASO by either a two hour infusion the presence of a foreign infectious agent. (IV) or by continuous SC infusion over a two week ASO for haematological malignancies period. It appears that both of these routes of admin- In the field of cancer, chromosomal translocations istration lead to a maximum tolerated dose of phos- often lead to deregulation of proto-oncogenes, or to phorothioate oligonucleotide dose in the region of 6 the formation of fusion genes, making them potential mg per kilogram per day. Above this dose reversible targets for manipulation by ASO in an attempt to falls in the count (thrombocytopenia) are downregulate oncogene expression and, it is to be observed and require therapy to be stopped. Platelet hoped, to have an anti-tumour effect. The systemic recovery is readily reversible. The pharmacokinetics of use of ASO also has its attractions where it might be the second and third generations of ASO are not yet hoped to have a specific anti-tumour effect while clarified. avoiding the many non-specific toxicities caused by How ASO should be assessed chemotherapeutic substances. In B-cell lymphomas the t(14;18) and t(8;14) translocation involving the Antisense research must be interpreted with ade- Bcl-2 and MYC genes respectively have been targeted quate reference to control parameters. These include for antisense therapy. The currently most advanced essential controls such as sense control (the comple- human Phase I/II study has been in B-cell lymphomas mentary sequence to the antisense sequence), non- with high Bcl-2 expression. The Bcl-2 oncogene has sense or scrambled control and mismatched control been implicated in the oncogenicity of a wide variety (the same oligomer but for one or two mismatches in of malignancies. Bcl-2 protein directly prolongs cellu- the central section). Most published antisense exper- lar survival by blocking chemotherapy-induced pro- iments have utilised a minimum of two controls and grammed cell death (apoptosis) making it an attrac- this should be regarded as a basic requirement when tive target for antisense therapy. Antisense, control designing antisense work. sense and nonsense oligonucleotides to the open Antisense experiments are targeted towards a par- reading frame of Bcl-2 were evaluated using a lym- ticular oncogene and should therefore be able to phoma cell line with high Bcl-2 expression and gave downregulate the translation of its mRNA, producing specific downregulation of Bcl-2 protein with subse- a decrease in its protein production. If this phenom- quent induction of apoptosis. An in vivo lymphoma enon cannot be demonstrated then the suspicion of model was set up with the same cell line. Anti-tumour non-specific effects must be raised. effect in vivo was demonstrated with a two week infu- Non-specific effects due to chemical composi- sion of Bcl-2 antisense (G3139, Genta USA) at 100 µg tion of oligonucleotides daily achieving a plasma level of approximately 0.1 Oligonucleotides are polyanions and can have phys- µM. Currently a human Phase I trial of this Bcl-2 anti- iological roles such as those of naturally occurring sense is being completed and shows it to be well tol- polyanions including and dermatan sulphate. erated with promising evidence of efficacy. It is envis- Base composition of the oligonucleotide such as the aged that the ASO may be used in a clinical setting to presence of four contiguous guanosine residues can overcome Bcl-2 mediated drug resistance by combin- result in it having an antiproliferative role which may ing Bcl-2 ASO to prime the patient before chemother- mimic an antisense effect if not adequately controlled apy, followed by the conventional chemotherapy. for. Other non-specific effects induced by oligomers, Such an experimental approach has been taken using with palindromic sequences of six or more bases, G3139 and chemotherapy in a SCID mouse include induction of ␣- and ␥-interferon production, melanoma model and is now being translated into a which may have effects on enhancing natural killer human trial for both melanomas and lymphomas. cell activity in in vivo tumour models such as in the Significant tumour regression is observed. Alterna- SCID mouse. These examples represent a fraction of tively it may prove to be of benefit as a form of main- the number of potential interactions between tenance therapy being given over a longer period. oligonucleotides and cellular proteins, and may give Caution will, however, be needed to ensure that long an explanation for some of the non-sequence-specif- term toxicities are not induced. Similar Phase I stud- ic effects seen with control oligomers. ies are now underway for other solid tumours using a c-Raf isoform antisense as an inhibitor of MAP kinas- es (ISIS 5132, ISIS pharmaceuticals, Carlsbad CA) Application of antisense oligonucleotides ␣ for human disease and Protein Kinase C- (ISIS 3521). There are currently a number of first generation ASO for other fields of medicine phosphorothioate ASO that are being investigated in The potential for ASO in the fields of medicine is the human setting. These include the fields of cancer vast. The field of myocardial infarction and cardiac (lymphoma), cardiac disease (coronary grafting), vein grafting have both been investigated with chronic inflammatory conditions (bowel disease) and favourable results. Antisense to cdk2 kinase mRNA infectious disorders (CMV retinitis). All conditions are prevents arterial intimal thickening after vein grafting 22 F. E. Cotter and may be a useful addition to cardiac surgery to go before they reach the clinic. Oligonucleotides are prevent coronary artery restenosis and could be use- currently the most advanced in the human therapeu- ful in preventing progression of atherosclerosis. In the tic setting. Total replacement of conventional thera- field of inflammatory bowel disease an ICAM-1 anti- py is not going to occur but we will be able to use sense (a protein mediating inflammation) has entered gene modification to enhance current pharmacolog- a Phase I study (ISIS 2302) and has been reported to ical agents and reduce their toxicity. Combinations have significantly reduced the use of steroids in the of antisense molecules may further enhance their effi- ASO treated group. It is envisaged that other systemic cacy and alterations of the antisense chemistry may inflammatory conditions such as rheumatoid arthri- improve their tolerance and usefulness. This is a whole tis or systemic auto immune diseases could benefit new area of research and as such requires much eval- from such an ASO approach. Recently the downreg- uation if it is to be applied otpimally. Although in vit- ulation of Bcl-xL an anti-apoptosis gene by ASO in ro efficacy may be established, it is highly desirable to the intima of vascular lesions has resulted in regres- test the hypothesis in in vivo animal models. With care sion, suggesting that targeting apoptosis by ASO may novel therapies based on the biology of the malignant play a major role as a novel therapy for vascular dis- cell may be determined on a scientific basis and may ease, a major cause of death in the Western world. help improve the treatment of patients with disease. The other area of considerable interest and one in which there are currently very few effective pharma- ceutical agents is that of antiviral therapy. Initial References attempts to eradicate HIV have not shown much suc- cess, possibly due to a failure to identify the correct 1. Cotter FE. Antisense therapy for B cell lymphomas. In: gene to target. However, an ASO to the mRNA for AC Wotherspoon, ed. Cancer Surveys. Lymphoma. the IE2 gene of cytomegalovirus (CMV) (ISIS 2922) Cold Spring Harbor: Cold Spring Harbor Laboratory administered directly into the eye for CMV infection Press; 1997. p. 311-25. 2. Cotter FE, Johnson P, Hall P, et al. Antisense oligonu- of the retina, which causes blindness particularly in cleotides suppress B-cell lymphoma growth in a SCID- immunocompromised patients, has been successful. hu mouse model. Oncogene 1994; 9: 3049-57. This compound is now completing the licencing pro- 3. Crooke ST. Proof of mechanism of antisense drugs. cedure for clinical use as the first ASO to come onto Antisense & Nucleic Acid Drug Development 1996; 6: the commercial market. This is a major advance for 145-7. 4. Dean N, McKay R, Miraglia L, et al. Inhibition of antiviral therapy and illustrates the versatility of ASO. growth of human tumor cell lines in nude mice by an Similar approaches could also be taken for bacterial antisense oligonucleotide inhibitor of protein kinase infections. C-alpha expression. Cancer Res 1996; 56:3499-507. 5. Milner N, Mir KU, Southern EM. Selecting effective Future perspectives antisense reagents on combinatorial oligonucleotide arrays. Nature Biotechnol 1997; 15:537-41. The potential ability of ASO to downregulate the 6. Monia BP, Johnston JF, Geiger T, Muller M, Fabbro D. expression of disease-causing genes, with minimal tox- Antitumor activity of a phosphorothioate antisense icity has been demonstrated and is now opening up oligodeoxynucleotide targeted against C-raf kinase. a whole new approach to the management of some Nat Med 1996; 2: 668-75. human diseases. In many respects antisense therapy 7. Raynaud FI, Orr RM, Goddard PM, et al. Pharmaco- kinetics of G3139 a phosphorothioate oligodeoxy- has been the forerunner of gene therapy as many of nucleotide antisense to bcl- following intravenous us understand it and have permitted us to show that administration or continuous subcutaneous infusion some genes truly play a significant role in human dis- to mice. J Pharmacol Exper Therap 1997; 281:420-7. ease. ASO give us the ability to switch off inappropri- 8. Reed J, Cuddy M, Haldar S, et al. BCL2-mediated ately expressed genes and modify the target cells’ bio- tumorigenicity of a human T-lymphoid cell line: syn- ergy with MYC and inhibition by BCL2 antisense. Proc logical behaviour. Vector-mediated gene therapy will Natl Acad Sci USA 1990; 87:3660-4. in the future allow replacement of genes that are lack- 9. Webb A, Cunningham D, Cotter F, et al. Bcl-2 anti- ing in the cell, that is to say switch on genes. ASOs rep- sense therapy in patients with non-Hodgkin’s lym- resent the first generation of a new class of pharma- phoma. Lancet 1997; 349:1137-41. cological agents based on a biological understand- 10. Ziegler A, Luedke GH, Fabbro D, Altmann K-H, Sta- hel W, Zangemeister-Wittke U. Induction of apopto- ing of the diseased cell. Other synthetically derived sis in small-cell lung cancer cells by an antisense molecules such as ribozymes and DNAzymes are also oligodeoxynucleotide targeting the Bcl-2 coding capable of silencing genes but still have a long way to sequence. J Natl Cancer Inst 1997; 88:1027-36. Educational Session 3 Haematologica 1999; 84:(EHA-4 educational book):23-25 Chairman: A. Lanzavecchia Dendritic cells and chemokines

Dendritic cell maturation and generation of immune responses ANTONIO LANZAVECCHIA Basel Institute for Immunology, Basel, Switzerland

Dendritic cell maturation also express high levels of the mannose receptor, a Dendritic cells (DC) represent a system of profes- pattern-recognition molecule that allows efficient sional antigen presenting cells (APC). Their function uptake of mannosylated and fucosylated antigens, as is to capture incoming antigens and present them in well as the Fcg receptor, CD32, that allows capture of secondary lymphoid organs to naive T cells in order immune complexes. Immature DC are consequently to generate immune responses.1 DC precursors con- extremely efficient antigen presenting cells (APC) for stitutively migrate at low rate from blood into non- soluble antigens. They can present tetanus toxoid inflamed tissues, but the rate of migration can be (TT), which is taken up by fluid phase at concentra- dramatically increased if there is an inflammation. tions of ~10–10 M, and are therefore as efficient as TT- Here the DC, which are in their so-called immature specific B cells that take up this antigen via specific state, encounter pathogens and undergo a complex mIg. The efficiency of presentation can be further change in their properties that is collectively called boosted (~100 fold) by targeting the antigen to CD32 maturation. or to the mannose receptor. This can be achieved by Evidence for DC maturation has been gained ini- complexing the antigen with antibodies or by man- tially from in vivo observations where immature DC nosylation. have been shown to migrate from peripheral tissues to secondary lymphoid organs while acquiring T cell Stimuli that induce dendritic cell stimulatory capacity.2 The molecular details and the maturation kinetics of the maturation process as well as the sig- The availability of cultured immature DC has been nals that trigger it have been defined using an in vitro instrumental in the identification of stimuli that induce system of -derived DC.3 Human peripheral DC maturation. These include inflammatory cytokines blood monocytes cultured in GM-CSF and IL-4 devel- (TNF-␣ and IL-1) and bacterial and viral products op without dividing into immature DC that can (LPS and double-stranded RNA).3-5 These stimuli are respond to various stimuli by undergoing the matu- likely be met by DC in peripheral tissues and are ration process. This process involves many aspects of important to the initiation of the maturation process. cell physiology, such as changes in the rate of endo- Once maturing DC have reached the secondary lym- cytosis, MHC biosynthesis, expression of co-stimula- phoid organs, they interact with activated T cells that tory molecules and cytokines, as well as molecules can deliver further stimulatory signals via CD40L and involved in adhesion and migration (Figure 1). We TRANCE, resulting in induction of cytokine produc- will discuss the properties of immature DC, the sig- tion and increased survival.6,7 DC maturation can be nals that induce maturation, and the consequence dissected into several distinct sub-programmes that of maturation for the generation of the immune involve changes in endocytosis, MHC biosynthesis, response. expression of co-stimulatory molecules, cytokine pro- duction and migration (see Figure 1). Mechanisms of antigen capture by immature DC Antigen presentation on MHC class II Immature DC have a high and constitutive level of molecules macropinocytosis that allows them to take up large In DC the maturation stimuli optimise antigen pre- volumes of fluid and concentrate the macrosolutes in sentation on class II molecules.8 In immature DC, the endocytic compartment.4 This capacity is depen- newly synthesised class II molecules are loaded effi- dent on the developmentally-regulated expression of ciently with antigenic peptides and transported to aquaporins and amiloride-sensitive sodium channels, the cell surface. From the cell surface, class II mole- which endow these cells with a very high capacity to cules are continuously internalised and recycled back transport water and ions across the membranes lead- to the surface. Thus, in immature DC, antigenic pep- ing to concentration of macrosolutes. Immature DC tides can be loaded on both newly synthesised and recycling molecules. The class II recycling compart- ment is particularly prominent in these cells and allows rapid loading of T cell epitopes that are dif- Correspondence: Antonio Lanzavecchia, Basel Institute for Immunolo- gy, Grenzacherstrasse 487, 4005 Basel, Switzerland; Fax. international ferent from those loaded into newly synthesised mol- +41.61.6051222 – E-mail: [email protected] ecules. Maturation induced by LPS or TNF-␣ results 24 A. Lanzavecchia

Antigen presentation on class I molecules A particular challenge for an APC is to present a cytopathic virus. To do this the cell must be suscepti- ble to infection in order to produce viral protein, but at the same time, it must be able to resist the cyto- pathic effect of the virus. In some cases, for instance infection with influenza virus, DC are able to achieve this difficult task.5 Immature DC are susceptible to infection with influenza virus and synthesise large amounts of viral proteins. But, soon after infection, they become resistant to the cytopathic effect of the virus. The dsRNA of the influenza virus stimulates DC maturation, which results in upregulation of synthe- sis of MHC class I molecules necessary for efficient presentation of viral antigens. As part of the matura- tion programme, DC produce type I IFN which upreg- ulates MxA, a protein that instates resistance to the cytopathic effect of the virus. By this autocrine mech- anism of stimulation, DC rapidly resist the cytopath- ic effect of the virus and efficiently present viral anti- gens by continuously loading antigenic peptides on newly synthesised class I molecules. Unlike that which Figure 1. Generation, maturation and activation of dendrit- is observed for class II molecules, the biosynthesis of ic cells. Human peripheral blood monocytes cultured with class I molecules is sustained for days while the half- GM-CSF and IL-4 develop into immature DC characterised life does not change significantly, remaining relative- by high endocytic activity, but low T cell stimulatory capac- ity. Inflammatory cytokines and pathogens rapidly convert ly short (~10-20 hours). This half-life may reflect an these cells into mature DC. DC maturation involves the intrinsically greater instability of class I as compared downregulation of endocytosis and the upregulation of to class II molecules in living cells. MHC, co-stimulatory and adhesion molecules. Stimulation by T cells via CD40L may further upregulate co-stimulatory The difference in stability between class I and class and adhesion molecules and selectively trigger IL-12 pro- II molecules makes good sense if we consider the dif- duction. The whole process from monocytes to immature, ferent functions of these two systems of peptide pre- mature and activated DC is highly efficient (up to 80-90% of initial cell input) and does not involve cell proliferation. sentation. Class II molecules should present antigens which are transiently encountered in the surrounding environment of a DC present in a peripheral tissue, so it is important that DC make a maximum effort to load antigenic peptides over a short period of exposure in an increase of ~3-4 fold in the rate of class II syn- to the antigen and then retain them as stable com- thesis, which is sustained for at least 24 hours, while plexes. On the other hand, class I molecules should at subsequent time points the synthesis is shut off. present endogenously-synthesised antigens, so that it The transient boost of class II biosynthesis provides is instrumental that these complexes be continuously the maturing DC with a large number of molecules generated inside the cell to allow expression of all pos- that can be loaded with antigenic peptides. At the sible viral proteins for the time that the cell remains same time the maturation process leads to a pro- infected. gressive decrease of endocytosis which results in a dra- Immature DC are also particularly efficient in cap- matic increase in the life expectancy of class II mole- turing apoptotic or necrotic cells. In this case, the cules. Indeed, in immature DC, class II molecules have antigens present in the phagocytosed cells can be a relatively short half-life of ~10 hours, which shifts to processed and presented on both class II and class I more than a 100 hours in mature DC. As a conse- molecules. This mechanism (defined as crosspriming) quence of these co-ordinated changes in class II may be important for presentation of tumour anti- biosynthesis and stability, DC, soon after induction of gens and, in general, of antigens which are carried maturation, assemble a large number of peptide exclusively by non-professional APC.9,10 MHC complexes that are retained in a stable form for long periods of time in the absence of further class II T cell stimulation and polarisation synthesis. This mechanism optimises the presentation Immature DC are very poor stimulators for naive T of infectious antigens, i.e., antigens, derived from cells since they do not express co-stimulatory mole- pathogens, which stimulate DC directly (for instance cules. On the other hand, all maturation stimuli via LPS), or indirectly (for instance via production of induce marked upregulation of B7.1 and B7.2 and TNF-␣). consequently mature DC acquire a marked capacity to Session #3 – Dendritic cells and chemokines 25

Table 1. The nature of the maturation stimuli determines stimulatory capacity of DC in order to generate the cytokine production and consequently T cell polarisation. appropriate type of effector response. As suggested earlier the mechanism of T cell-T cell help via APC acti- Maturation MIP1b IL-12 IL-6 TNF␣ INF␣ Polarisation vation may be exploited therapeutically by providing stimuli on the DC, in addition to the CTL epitope, a recall TNF + - + ND - Th2 >Th1 antigen such as TT, which could be recognised by LPS +++ + + + + Th1 memory T cells. This will allow stimulation of DC by dsRNA ++ + + + + Th1 T helper cells via CD40L-CD40 interaction to occur at CD40L + +++ + + - Th1 the appropriate sites in the lymph nodes, thus increas- ing the stimulatory capacity of the DC and inducing IL- 12 production at sites where it can most effectively stimulate a CTL response. prime naive T cells. Although most aspects of DC mat- uration (such as the downregulation of endocytosis, References increased MHC biosynthesis and increased co-stimu- lation) are induced to a comparable extent by all mat- 1. Banchereau J, Steinman RM. Dendritic cells and the uration stimuli, for cytokine production the nature of control of immunity. Nature (Lond) 1998; 392: 245- 52. the stimulus makes a difference (see Table 1). In gen- 2. Stossel H, Koch F, Kampgen E, et al. Disappearance eral, pathogens or helper T cells provide a much more of certain acidic organelles (endosomes and Langer- powerful maturation stimulus for cytokine production hans cell granules) accompanies loss of antigen pro- than TNF-␣ or IL-1, and push DC to a higher T cell cessing capacity upon culture of epidermal Langer- stimulatory state. For instance, IL-12 production is hans cells. J Exp Med 1990; 172:1471-82. 3. Sallusto F, Cella M, Danieli C, Lanzavecchia A. Den- triggered selectively by CD40L and/or LPS, but not by dritic cells use macropinocytosis and the mannose TNF-␣. The strong stimulation by CD40L explains the receptor to concentrate macromolecules in the major requirement for T cell help in the induction of cytotoxic histocompatibility complex class II compartment: responses to cellular antigens such as some minor his- downregulation by cytokines and bacterial products. tocompatibility antigens. On the other hand, the J Exp Med 1995; 182:389-400. 4. Sallusto F, Lanzavecchia A. Efficient presentation of strong stimulation induced by pathogens explains the soluble antigen by cultured human dendritic cells is lack of requirement for help in the generation of a CTL maintained by / colony-stim- response to influenza virus. Altogether, it appears that ulating factor plus interleukin 4 and downregulated while all mature DC have comparable capacity to trig- by tumor necrosis factor alpha. J Exp Med 1994; 179: 1109-18. ger T cell proliferation, they differ remarkably in their 5. Cella M, Salio M, Sakakibara Y, Langen H, Julkunen I, capacity to polarise T cells towards Th1 or Th2, Lanzavecchia A. Maturation, activation, and protec- depending on the type of cytokines produced. tion of dendritic cells induced by double-stranded RNA. J Exp Med 1999; 189: 821-9. Therapeutic approaches based on DC 6. Cella M, Scheidegger D, Palmer-Lehmann K, et al. Lig- ation of CD40 on dendritic cells triggers production of maturation high levels of interleukin-12 and enhances T cell stim- The better understanding of the mechanism of DC ulatory capacity: T-T help via APC activation. J Exp maturation is going to have an impact on several ther- Med 1996; 184:747-52. apeutic areas. First, antigens can be targeted more effi- 7. Wong BR, Josien R, Lee SY, et al. TRANCE (tumor necrosis factor [TNF]-related activation-induced ciently to dendritic cells by mannosylation, opsonisa- cytokine), a new TNF family member predominantly tion and special formulation (with apoptotic bodies or expressed in T cells, is a dendritic cell-specific survival exosomes being particularly interesting options). factor. J Exp Med 1997; 186:2075-80. These mechanisms may also be used to eliminate DC 8. Cella M, Engering A, Pinet V, Pieters J, Lanzavecchia A. Inflammatory stimuli induce accumulation of MHC when required. Second, adjuvants to be used for vac- class II complexes on dendritic cells. Nature (Lond) cination may be designed in a more rational way to 1997; 388:782-7. poise DC in such a way as to generate Th1 or Th2 9. Inaba K, Turley S, Yamaide F, et al. Efficient presenta- polarised responses. Finally, DC grown in tissue cul- tion of phagocytosed cellular fragments on the major ture may be used for immunisation after pulsing with histocompatibility complex class II products of den- dritic cells. J Exp Med 1998; 188: 2163-73. antigens or peptides, or better, after transfection with 10. Albert ML, Sauter B, Bhardwaj N. Dendritic cells appropriate vectors. This approach may take full acquire antigen from apoptotic cells and induce class advantage of the possibility of modulating the T cell I- restricted CTLs. Nature (Lond) 1998; 392: 86-91. Educational Session 3 Haematologica 1999; 84:(EHA-4 educational book):26-27 Chairman: A. Lanzavecchia Dendritic cells and chemokines

Cytotoxic T priming versus cytotoxic T lymphocyte tolerance induction: a delicate balancing act involving dendritic cells CORNELIS J.M. MELIEF, STEPHEN SCHOENBERGER, RENE TOES, RIENK OFFRINGA Leiden University Medical Center, Dept. Immunohaematolgy & Bloodbank, Leiden, The Netherlands

ost solid tissue cells including tumours cific CTL were ultimately deleted from the periphery.6 express MHC class I-molecules, but lack co- This deletion only required antigen recognition on a stimulatory molecules important for appro- BM-derived APC. Thus, cross-presentation of antigens M 1 priate cytotoxic T lymphocyte (CTL) activation. can also lead to tolerisation of CTL. These studies pro- Therefore, optimal CTL induction requires process- vide a mechanism by which potentially auto-reactive ing and presentation of peripheral antigens by pro- CTL can be deleted from the periphery when the anti- fessional antigen presenting cells (APC). This cross- gen recognised by these CTL is expressed outside the presentation of antigens provides the immune sys- recirculation pathway of naive T cells, but contrasts tem with a mechanism by which it can detect and with the activation of OVA-specific CTL by cross-prim- respond to antigens expressed in non-lymphoid tis- ing APC that occurs when OVA-expressing cells are sues. Antigen-specific CTL responses can be induced used for immunisation.7 The discrepancy between regardless of the haplotype of the immunising cell.2 these observations can be explained by lack (toler- For induction of many MHC class I-restricted ance) or presence (priming) of antigen-specific CD4+ tumour-specific immune responses, cross-presenta- Th cells. Indeed in the OVA-transgenic mouse system, tion of antigens that have been captured by profes- injection of T cell receptor transgenic OVA-specific sional APC plays a dominant role.3 Cross-presenta- CD4+ Th cells prevents the deletion of OVA-specific tion of antigens by professional APC is the pivotal CTL and favours the induction of autoimmunity.8 mechanism of CTL-priming even when cells are These observations thus indicate that provision of help transfected with the co-stimulatory molecule B7, or by CD4+ Th cells is an important factor in the preven- when plasmid DNA, inoculated in muscle tissue, is tion of peripheral CTL-tolerance as well as for the used as the immunogen.3 A role for bone marrow induction of CTL-immunity. (BM) derived cells [most likely dendritic cells (DC) The nature of this help was unknown until recent- able to present MHC class I restricted antigens in a ly. Th cells must recognise antigen on the same APC TAP-dependent fashion] appears to be important in that presents the CTL-epitope.7 This clarifies the the induction of tumour-specific CTL responses by requirement for epitope linkage between Th cell- and cross-priming.3 CTL epitopes important for induction of CTL- These observations thus indicate that exogenous responses, and was previously explained by a prox- antigen-uptake by professional APC, that process and imity requirement for the efficient delivery by Th cells present these antigens to CTL precursors, is the dom- of soluble factors, such as IL-2. Alternatively cognate inant mechanism by which specific CTL responses are interactions between Th cell and APC convert the primed with the exception of CTL responses induced APC to a state that is capable of priming naive CTL, by directly infected APC. Viruses and other micro- as now reported by 3 groups, including ours.4,9-11 This organisms can also directly intrude into professional model could explain our observation that the inabil- APC with or without concomitant induction of the ity of female B6.bm12 mice (which harbour a mutat- APC activation programme required for CTL induc- ed MHC class II molecule resulting in a minimal H- tion.4 Cross-presentation of antigens by professional Y-specific helper response) to reject male skin-grafts APC is important for efficient CTL-priming, but the is overcome by immunisation with (in vitro) activated same mechanism might also be important for toleri- DC.12 Importantly, by communicating with the cross- sation of autoreactive CTL. In studies with transgenic priming APC, a Th cell can assist in CTL priming with- mice expressing a membrane-bound form of ovalbu- out requiring the simultaneous interaction of a (rare) min (OVA) in , BM-derived APC cross- antigen-specific CTL precursor, the presence of which presented OVA to adoptively-transferred OVA-specif- cannot be noticed by the Th cell (murine CTL are ic CTL in lymph nodes draining the site of OVA expres- class II-negative) in the immediate vicinity of the same sion.5 After an initial phase of proliferation, OVA-spe- APC. As a result, the activity of only a few antigen- specific Th cells can in this way be amplified, since one Th cell can activate a number of APC which can Correspondence: Cornelis J.M. Melief, Leiden University Medical Cen- ter, Dept. Immunohaematolgy & Bloodbank, Albinusdreef 2, 2333 ZA then prime an array of antigen-specific CTL. Leiden, The Netherlands. We were alerted to this scheme of cellular interac- Session #3 – Dendritic cells and chemokines 27 tions by studying the role of CD40-CD40Ligand inter- through CD40. Female MHC class II knock-out mice action in the delivery of help for CTL-priming. This are not able to mount an H-Y-response when inject- interaction is important in the activation of profes- ed with male DC, but do respond when CD40-mod- sional APC. CD40 is a member of the tumour necro- ulated DC are used for vaccination.4 These results sis factor gene family and is expressed on several cell show that the Th cell and CTL do not need to meet types, including DC, B cells, macrophages, endothe- simultaneously at the surface of the antigen-present- lial cells and proximal tubular epithelial cells in the ing DC to enable CTL-priming. Moreover these results .13 CD40Ligand (CD40L) is expressed shortly indicate that DC that have received an activation sig- following TCR triggering on CD4+ T cells. We demon- nal through CD40-CD40L interactions can prime strated that CD40 signalling can replace CD4+ T cells naive CTL, whereas unmodulated DC that have not in the priming of helper dependent CD8+ CTL been in contact with antigen-specific, CD40L-express- responses.10 Vaccination of B6 mice (H-2b) with com- ing Th cells are unable to do so. pletely allogeneic tumour cells of BALB/c origin (H- 2d), transformed by the human adenovirus type 5 ear- ly region 1 (Ad5E1), leads to the induction of a strong References H-2Db restricted CTL-response directed against an 1. Allison J. CD28-B7 interactions in T cell activation. AdE1B-encoded CTL-epitope. Since these tumour Curr Opin Immunol 1994; 6:414-9. cells themselves cannot present the E1-derived CTL 2. Bevan M. Cross-priming for a secondary cytotoxic epitope to E1B-specific CTL (they lack H-2Db), the response to minor H antigens with H-2 congenic cells CTL must be primed by non-tumour cells that have which do not cross-react in the cytotoxic assay. J Exp processed and presented the E1-peptide to E1-spe- Med 1976; 143:1283-8. 3. Toes R, Schoenberger S, Voort van der E, Offringa R, cific CTL (cross-priming). In vivo depletion studies Melief C. CD40-CD40L interactions and their role in revealed that cross-priming of E1B-specific CTL is cytotoxic T lymphocyte priming and antitumor immu- strictly Th cell dependent, as mice depleted for CD4+ nity. Semin Immunol, in press. Th cells prior to immunisation with BALB/c Ad5E1 4. Ridge J, Rosa Di F, Matzinger P. A conditioned den- 10 dritic cell can be a temporal bridge between a CD4+ cells no longer mount an E1B-specific CTL response. T-helper and a T-killer cell. Nature (Lond) 1998; 393: Administration of a CD40 activating monoclonal 474-7. antibody (mAb) to either CD4-depleted or B6 I-Ab 5. Kurts C, Heath W, Carbone F, Allison J, Miller J, Kosa- knock-out mice, lacking functional MHC class II- ka H. Constitutive class I-restricted exogenous pre- restricted CD4+ T cells in the periphery, resulted in sentation of self-antigens in vivo. J Exp Med 1996; 184:923-30. efficient restoration of E1B-specific CTL responses. 6. Kurts C, Kosaka H, Carbone F, Miller J, Heath W. Moreover, blockade of CD40L by in vivo administra- Class I-restricted cross-presentation of exogenous self- tion of a mAb that blocks CD40L on the CD4+ T cells, antigens leads to deletion of autoreactive CD8+ T resulted in a profound inhibition of CTL-priming that cells. J Exp Med 1997; 186:239-45. 7. Bennet SRM, Carbone F, Karamalis F, Miller J, Heath was overcome by CD40 triggering. B cells are neither W. Induction of a CD8+ cytotoxic T lymphocyte required as APC for cross-priming of Ad5E1B-specif- response by cross-priming requires cognate CD4+ T ic CTL nor essential for the CD40-mediated restora- cell help. J Exp Med 1997; 186:65-70. tion of cross-priming in the absence of CD4+ cells.10 8. Kurts C, Carbone F, Barnden M, et al. CD4+ T cell help impairs CD8+ T cell deletion induced by cross presen- Similarly CD40- or CD40L-deficient mice are unable tation of self-antigens and favors autoimmunity. J Exp to generate OVA-specific CTL-responses after vacci- Med 1997; 186:2057-62. nation with OVA-expressing cells, showing that 9. Lanzavecchia A. Licence to kill. Nature (Lond) 1998; expression of both CD40 and CD40L is indeed 393:413-4. required for cross-priming of CTL.11 Moreover, mice 10. Schoenberger SP, Toes R, Van der Voort E, Offringa R, + Melief C. T help for cytotoxic T lymphocytes is medi- devoid of CD4 Th cells, are not able to mount OVA- ated by CD40-CD40L interactions. Nature (Lond) specific CTL after vaccination with OVA expressing 1998; 393:480-3. cells, unless immunisation is performed together with 11. Bennet SRM, Carbone F, Karamalis F, Flavell R, Miller injection of a mAb that triggers CD40.11 Taken J, Heath W. Help for cytotoxic T cell responses is medi- ated by CD40 signalling. Nature (Lond) 1998; 393: together, these results show that CTL can be primed 478-80. in the absence of Th-derived cytokines, and demon- 12. Boog CJP, Kast W, Timmers H, Boes J, De Waal L, strate the crucial importance of CD40-CD40L inter- Melief C. Abolition of specific immune response defect action. The contribution of CD4+ Th cells involves the by immunization with dendritic cells. Nature (Lond) activation of professional (non B-cell) APC via CD40- 1985; 318:59-62. 13. Grewal IS, Flavell R. A central role of CD40Ligand in CD40L interactions. Indeed, help for priming of H-Y the regulation of CD4+ T cell responses. Immunol specific CTL can be bypassed by activation of DC Today 1996; 17:410-4. Educational Session 3 Haematologica 1999; 84:(EHA-4 educational book):28-31 Chairman: A. Lanzavecchia Dendritic cells and chemokines

The role of chemokines and chemokine receptors in T cell priming and Th1/Th2-mediated responses FEDERICA SALLUSTO Basel Institute for Immunology, Basel, Switzerland

Chemokines and chemokine receptors ples of how changes in chemokine receptor expres- The immune system is made of mobile cells and its sion may be instrumental in inducing selective migra- function is dependent on the capacity of these cells tion of cells during the immune response.3 to migrate to the right place at the right time. T cell priming occurs in lymphoid tissues and requires Chemokines in antigen presentation encounters between naive T cells and antigen-loaded The induction of T cell responses requires migra- dendritic cells (DC) that have captured antigens at tion of DC from tissues, where they sample antigens, peripheral sites. In contrast, effector responses such to the T cell areas of lymph nodes, where they stimu- as delayed type hypersensitivity (DTH) or allergic late naive T cells. Immature DC and monocytes reactions occur in peripheral tissues following inter- express receptors for inflammatory chemokines 4,5 and action of Th1 or Th2 cells with effector leukocytes. consequently are attracted to inflamed tissues where Leukocyte migration is controlled at the level of the cognate ligands are produced. Here the DC will expression of selectins, chemokine receptors and inte- take up antigens and will be stimulated to mature by grins that co-operate in an ordered fashion in the pathogens or inflammatory cytokines. Maturing DC process of extravasation and migration within the tis- are themselves an extremely abundant source of sues. Recent evidence indicates that chemokines and inflammatory chemokines5 and therefore contribute chemokine receptors provide a flexible code to deter- to attracting immature DC as well as effector leuko- mine the selective migration of T cells, DC, and oth- cytes. The high level of inflammatory chemokines pro- er leukocytes involved in the immune response. duced leads to a prompt downregulation of the cog- Chemokines can be operationally divided into two nate receptors on maturing DC. At the same time, categories. The first is represented by those which are receptors for constitutive chemokines (especially produced constitutively in bone marrow, and CCR7) are upregulated at the transcriptional level.4,5 secondary lymphoid organs and which are denoted CCR7 allows maturing DC to enter the afferent lymph constitutive or immune as they control leukocyte traf- attracted by SLC which is bound on lymphatic fic under physiological conditions (Table 1). The sec- endothelial cells, and subsequently to localise in the ond is represented by inflammatory chemokines, that T cell areas of lymph nodes, following a gradient of are induced or upregulated by inflammatory stimuli ELC, another CCR7 ligand which is produced by res- and are involved in the recruitment of effector leuko- ident mature DC. Thus, in their life-cycle DC first cytes at site of tissue injury (Table 2).1,2 While there express receptors for inflammatory chemokines, then may certainly be exceptions to this rule, it represents produce inflammatory chemokines, then upregulate a useful paradigm to approach the complexity of receptors for constitutive chemokines, and finally, chemokine regulation. after a further time lag, start synthesising constitutive Chemokine receptors can be specific for a single chemokines. This time-dependent expression of recep- chemokine or can recognise several. The expression of tors and ligands allows DC to tightly regulate their these receptors is highly specific for a particular cell migratory capacity in an autonomous fashion, which type and determines the cell’s capacity to migrate to is consistent with their role as sentinels of the immune particular districts where the cognate ligands are pro- system (Figure 1). duced (Tables 1 and 2). An emerging principle is that in DC, B and T lymphocytes chemokine receptor Chemokines in T cell priming expression is developmentally regulated. In addition, Naive T cells continuously recirculate from the upon activation, chemokine receptor expression is blood into secondary lymphoid organs such as lymph rapidly modified by transcriptional or post-transla- nodes, where they scan DC for antigens and re-enter tional mechanisms, resulting in the acquisition of the blood stream through the efferent lymph. Naive novel migratory capacity. We will give a few exam- T cells express L-selectin and CCR7 which allow them to interact with endothelial venules (where the lig- ands for these receptors, CD34 and SLC are Correspondence: Federica Sallusto, Basel Institute for Immunology, Grenzacherstrasse 487, CH-4005 Basel, Switzerland. Fax. international expressed on endothelial cells). Following extravasa- +41-61-6051222 – E-mail: [email protected]. tion the cells enter the T cell areas, where ELC is pro- Session #3 – Dendritic cells and chemokines 29

Table 1. CCR7 and are therefore attracted to the T cell areas where ELC is produced. Conversely, some T cells that Constitutive/lymphoid chemokines are activated by encounter with antigen on DC upreg- Receptors MDC TARC ELC SLC SDF-1BCA-1 Expression ulate CXCR5 and thus migrate towards the B cell areas, guided by BCA-1. In this way, T and B cells CCR4 + + mDC (subset?), move out of their own areas and migrate towards Ba, Th2, T (subset) each other, meeting at a T and B boundary where they CCR7 + + + mDC, B, naive T, can interact in an antigen-specific fashion (Figure 1). act. T CCR8 + act. Th2 Chemokines in Th1/Th2-mediated CXCR4 + Mo, act. MF, mDC, B, T naive responses CXCR5 + B, T (subset) Following priming in the , T cells acquire effector function, i.e. the capacity to produce Abbreviations: Mo, monocytes: MF, macrophages; mDC, mature DC, act., cytokines following antigenic stimulation. Th1 cells activated; T, T lymphocytes; B, B lymphocytes; MDC, macrophage derived produce IFN␥ and are involved in responses against chemokine; TARC, thymus and activation regulated chemokine; ELC, EBI1 ligand chemokine; SLC, secondary lymphoid tissue chemokine; SDF-1, stro- intracellular pathogens and DTH reactions, while Th2 mal cell derived factor 1; BCA-1, B cell attracting chemokine 1. cells produce IL-4 and IL-5 and are involved in responses against extracellular parasites and generate allergic reactions. Together with the acquisition of the duced by resident mature DC (Figure 1). The impor- cytokine producing capacity, effector T cells acquire tance of CCR7 and its ligands in orchestrating the T- new homing potential. They lose CCR7 and acquire DC encounter is exemplified by the fact that mice receptors for inflammatory chemokines depending on lacking SLC or CCR7 are incapable of mounting pri- the type of polarisation (Figure 2). CCR3, a receptor 6 for eotaxin, which is also expressed by and mary T cell responses (and M. Lipp, personal communi- 8,9 cation). , is selectively expressed by Th2 cells. The expression of CCR3 allows these three cell types to co- Chemokines in T-B collaboration localise together at sites of allergic reactions. This may play a pathogenetic role because the Th2-derived Naive B cells also home to the lymph node where cytokines IL-4 and IL-5 function as activation and sur- they localise to the B cell areas. Resting B cells express vival factors for basophils and eosinophils. CXCR3, a CXCR5 which is essential for their localisation to the receptor for Mig and IP-10, two chemokines induced B cell follicles7 where the cognate ligand BCA-1 is pro- by IFN␥, is expressed at higher levels on Th1 than on duced by as yet undefined cell types. Activation of B Th2. This receptor may be important to localise Th1 lymphocytes to antibody production requires the cells to sites of DTH reactions. CCR5 is preferentially, encounter of two extremely rare cells, the antigen-spe- but not exclusively expressed on Th1 and its expression cific T and B cells. It is therefore not surprising that is upregulated by IL-2. CCR4 is expressed on Th2 cells this interaction needs to be tightly regulated. Follow- but also on non-polarised T cells that are incapable of ing stimulation with antigen, the B cells upregulate producing IL-4. Finally, CCR1 and CCR2 are

Table 2.

Inflammatory chemokines Receptors MIP-1a MIP-1b RNT MCP-1 MCP-2 MCP-4 Eot LARC I-309 IL-8 GRO IP-10 Expression MCP-3 Eot-2 GCP-2 ENA Mig ITAC

CCR1 + + + Mo, MF, iDC, Eo, Ba, Th1, N CCR2 + + + Mo, MF, iDC, Ba, T CCR3 + + + + Eo, Ba, Th2 CCR5 + + + Mo, MF, iDC, Th1 CCR6 + iDC (subset), B, T CCR8 + act. Th2 CXCR1 + iDC, N, Ba, T (subset) CXCR2 ++ N CXCR3 + B (subset), Th1

Abbreviations: Mo, monocytes; MF, macrophages; iDC, immature DC; Eo, eosinophils; Ba, basophils; N, ; T, T lymphocytes; B, B lymphocytes; MIP-1, macrophage inflammatory protein 1; RNT, RANTES, regulated on activation of normal T cell expressed and secreted; MCP, monocyte chemotactic protein; Eot, eotaxin; LARC, liver and activation regulated chemokine; GCP-2, granulocyte chemotactic protein 2; GRO, growth related oncogene; ENA, epithelial cell derived attractant; IP-10, interferon inducible protein 10; Mig, monokine induced by IFN-␥, I-TAC, interferon-inducible T cell alpha chemoattractant. 30 F. Sallusto

Figure 1. Selective expression of chemokines and chemokine receptors orchestrates the afferent and effector phases of the immune response. The critical migratory steps and cellular interactions in the immune response are highlighted, together with some of the molecules involved. (A) monocytes and immature DC are recruited by inflammatory chemokines to sites of antigen challenge, (B) pathogens induce DC-maturation resulting in a switch in chemokine receptor expression, (C) DC enter lymphatic vessels and are drained to the lymph node where they home to the T cell areas, (D) naive T and B cells home to the T and B cell areas of the lymph node, (E) T cells interact with DCs and are activated, (F) effector T helper cells migrate to B cell areas where they stimulate antigen- specific B cells, (G) and (H) Th1 or Th2 together with effector cells migrate to periph- eral sites of DTH or allergic reactions.

Figure 2. Developmental and activation induced control of chemokine receptor expres- sion in T cells. Session #3 – Dendritic cells and chemokines 31 expressed on both Th1 and Th2 cells.10,11 Interesting- Biol 1997; 62:634-44. ly the expression of chemokine receptors on effector 3. Sallusto F, Lanzavecchia A, Mackay CR. Chemokines T cells can be rapidly shifted by activation signals and chemokine receptors in T-cell priming and (Figure 2). Following TCR triggering effector T cells Th1/Th2-mediated responses. Immunol Today 1998; 19:568-74. downregulate receptors for inflammatory chemokines 4. Dieu M-C, Vanbervliet B, Vicari A, et al. Selective and transiently upregulate CCR7 as well as other recruitment of immature and mature dendritic cells receptors for constitutive chemokines. This rapid by distinct chemokines expressed in different anatom- switch occurs at the transcriptional level and results ic sites. J Exp Med 1998; 188:373-86. in novel homing capacities. The transient CCR7 5. Sallusto F, Schaerli P, Loetscher P, et al. Rapid and upregulation may allow T cells that have been acti- coordinated switch in chemokine receptor expression during dendritic cell maturation. Eur J Immunol 1998; vated in the tissues to migrate back to the lymph 28: 2760-9. nodes via the afferent lymphatics, or alternatively, to 6. Gunn MD, Kyuwa S, Tam C, et al. Mice lacking expres- localise within the chronically inflamed tissues to sion of secondary lymphoid organ chemokine have areas where ELC and SLC are produced. defects in lymphocyte homing and dendritic cell local- ization. J Exp Med 1999; 189:451-60. Conclusions 7. Forster R, Mattis AE, Kremmer E, Wolf E, Brem G, Lipp M. A putative chemokine receptor, BLR1, directs We have just started to appreciate the complexity of B cell migration to defined lymphoid organs and spe- the chemokine system and its regulation. This system cific anatomic compartments of the spleen. Cell 1996; provides a very flexible mechanism for regulating 87:1037-47. migration of different cells as well as their encoun- 8. Sallusto F, Mackay CR, Lanzavecchia A. Selective ters. Chemokine receptors can be useful markers of expression of the eotaxin receptor CCR3 by human T cell populations and their selective blockade may rep- helper 2 cells. Science 1997; 277:2005-7. resent a promising therapeutic strategy to interfere 9. Gerber B, Zanni MP, Uguccioni M, et al. Functional expression of the eotaxin receptor CCR3 in T lympho- with the afferent or efferent limb of the immune cytes co-localising with eosinophils. Curr Biol 1997; 7: response. 836-43. 10. Sallusto F, Lenig D, Mackay CR, Lanzavecchia A. Flex- ible programs of chemokine receptor expression on References human polarized T helper 1 and 2 lymphocytes. J Exp Med 1998; 187:875-83. 1. Baggiolini M. Chemokines and leukocyte traffic. 11. Bonecchi R, Bianchi G, Bordignon PP, et al. Differen- Nature (Lond) 1998; 392:565-8. tial expression of chemokine receptors and chemo- 2. Yoshie O, Imai T, Nomiyama H. Novel lymphocyte- tactic responsiveness of type 1 T helper cells (Th1s) specific CC chemokines and their receptors. J Leukoc and Th2s. J Exp Med 1998; 187:129-34. Educational Session 4 Haematologica 1999; 84:(EHA-4 educational book):32-35 Chairman: M. Greaves Acquired disorders of haemostasis

Autoimmune thrombophilic syndromes MIKE GREAVES Department of Medicine and Therapeutics, University of Aberdeen, Scotland

he role of autoantibodies in thrombosis has plete correction of the clotting time on mixing with been acknowledged for many years, but the normal plasma and confirmation of phospholipid- Trange of thrombotic disorders in which an dependence of the antibody. In contrast, ACL is usu- autoimmune pathogenesis is suspected continues to ally detected by enzyme-linked immunosorbent assay, expand (Table 1). In the 1960s the disorder which in which the anionic phospholipid cardiolipin is used subsequently became known as the antiphospho- to coat wells on plastic microtitre plates. In APS, lipid syndrome was first described. Paradoxically, assays for LA and ACL are complementary. although this condition has been subjected to exten- The antiphospholipid syndrome may be defined as sive investigation, the pathogenetic mechanisms the occurrence of thrombosis and/or recurrent mis- causing thrombosis are less well defined than those carriage in association with laboratory evidence of in more recently described autoimmune thrombotic persistent antiphospholipid antibody, either LA or disorders, such as heparin-induced thrombocytope- ACL. Thrombosis may affect or . Stroke nia with thrombosis. due to cerebral infarction is a prominent manifesta- tion, but occlusion of visceral and peripheral arteries Antiphospholipid antibodies and may also occur. In veins, limb deep vein thrombosis thrombosis is most common, but involvement of visceral and Antiphospholipid antibodies are associated with intracerebral veins, and pulmonary embolism also arterial and venous thrombosis, recurrent pregnancy occur. Thrombocytopenia is an occasional feature. loss and thrombocytopenia. Although the antibodies Recurrent miscarriage is a notable manifestation of have not yet been conclusively shown to be causal in the syndrome, often occurring in women with no pri- thrombosis and miscarriage, they are useful labora- or thrombotic history. tory markers for the antiphospholipid syndrome Where APS occurs in a subject with systemic lupus (APS). The identification of the syndrome is clinical- erythematosus, or less commonly some other disor- ly important because of the risk of recurrent throm- der, such as systemic sclerosis, rheumatoid arthritis bosis and the need for antithrombotic therapy in or Behçet’s syndrome, APS is regarded as secondary. many cases. Diagnosis and treatment of APS are sig- In primary APS there is no evidence of other underly- nificant challenges, however, due to the protean clin- ing disease. ical manifestations and associations, limitations of The nature of antiphospholipid antibodies currently available laboratory tests for antiphospho- lipid antibodies and the lack of clear evidence-based Why should antibodies apparently reactive with guidance on optimal management. some phospholipids be associated with a marked thrombotic tendency? A significant advance towards Antiphospholipid syndrome being able to answer this question was the recogni- Autoantibodies with apparent specificity for nega- tion of the dependence of antibody binding upon a tively charged phospholipids have long been recog- plasma protein, apolipoprotein H or ␤2 glycoprotein nised. The terms lupus anticoagulant (LA) and anti- I (␤2 GPI). It is noteworthy that this protein binds to cardiolipin (ACL) have been used to describe these anionic phospholipids and also possesses weak anti- antibodies. LA is an immunoglobulin which acts as a coagulant properties, principally through inhibition inhibitor but which does not recognise a of the contact phase of coagulation and of platelet specific coagulation factor. It slows the rate of throm- activity. The precise roles of ␤2 GPI bin generation, and therefore clot formation in vitro, and phospholipid in antibody binding have been dis- through interference in the interactions which require puted. ␤2 GPI may enhance antibody binding to phospholipid. It is therefore detected in coagulation phospholipid, although antibody binding to ␤2 GPI assays. The criteria for LA positivity are the prolon- immobilised on microtitre plates without the require- gation of a phospholipid dependent coagulation test, ment for phospholipid has been noted. One possi- with evidence of an inhibitor demonstrated by incom- bility is that ␤2 GPI bound to a surface, for example cell membrane phospholipid or the plastic of an assay plate, undergoes a conformational change and Correspondence: Mike Greaves. Department of Medicine and Thera- certain ’antiphospholipid’ antibodies bind to peutics, University of Aberdeen, AB25 2ZD, Scotland. Tel. international +1224-681818 ext. 53016 – Fax. international exposed neoepitopes. An alternative hypothesis is +1224-699884 – E-mail [email protected] that antibody binding to ␤2 GPI immobilised on a Session #4 – Acquired disorders of haemostasis 33

Table 1. Autoimmune thrombophilic conditions. has been demonstrated. Although some antien- dothelial cell reactivity appears to be ␤2 GPI depen- Antiphospholipid syndrome dent other antiendothelial antibodies in APS are dis- Heparin-induced thrombocytopenia and thrombosis tinct from antiphospholipid antibodies. In summary, although anti-␤2 GPI is commonly a Sporadic thrombotic thrombocytopenic purpura marker for APS, there appears to be considerable anti- body heterogeneity, with reactivity against a range of Due to autoantibody to von Willebrand factor proteins which bind to phospholipid as well as anti- Associated with antibody gens expressed on cell membranes. The pathogenesis of thrombosis in APS The paradoxical association between a marked pro- surface is not dependent upon expression of neoepi- thrombotic state and the presence of autoantibodies topes but is facilitated by concentration of the protein with in vitro anticoagulant effects has not yet been ful- on the surface, and therefore clustering of antigenic ly explained. Subjects with APS have evidence of per- sites, allowing bivalent binding of what are essential- sistent coagulation activation. Vascular occlusion is ly low affinity antibodies. Antibody binding may inter- due to thromboembolism, rather than vasculitis. fere with essential phospholipid-dependent steps in Some arterial events may also be due to embolisation coagulation, especially assembly of the and from sterile vegetations on cardiac valves. Although prothrombinase complexes on anionic phospholipid. there are many candidate prothrombotic mechanisms The result is a prolongation of the time to clot for- related to the properties of the antibodies listed mation which is reversible in the presence of excess above, the precise pathogenetic roles of the hetero- phospholipid. geneous array of autoantibodies which characterise Proteins other than ␤2 GPI are implicated in APS have not been clarified. Additional proposed antiphospholipid reactivity. Some antiphospholipid mechanisms include inhibition of fibrinolysis, pro- antibodies bind to immobilised or phospholipid motion of platelet activation and induction of tissue bound prothrombin. Usually the prothrombin con- factor on endothelial cells and monocytes. Further- centration in plasma is normal, consistent with the more, it may be that the antiendothelial antibodies concept that autoantibody only reacts with surface which frequently coexist with antiphospholipid anti- bound prothromin. The LA phenomenon can there- bodies may be pathogenic, as those found in systemic fore be due to antibodies reactive with prothrombin sclerosis have been shown to induce apoptosis of cul- or with ␤2 GPI. Anti-␤2 GPI positive sera are also fre- tured human umbilical vein endothelial cells. The quently positive in anticardiolipin assays, whereas LA pathogenesis of thrombosis in APS is almost certain- positive samples which are negative for ACL general- ly multifactorial. ly do not exhibit antibody binding to ␤2 GPI. Fur- A thrombotic pathogenesis of pregnancy failure in thermore, some antiphospholipid antibodies appear APS is suspected. Decidual vasculopathy and placen- to bind to phospholipid directly. This is a particular tal infarction have been observed but these are not feature of antiphospholipid antibodies associated apparent in all cases. Autoantibodies reactive with with syphilis and some other infections. trophoblast have been reported and some may cause The concept of heterogeneity amongst autoanti- displacement of annexin V from trophoblast and bodies in APS is strengthened by the demonstration accelerate coagulation. A non-thrombotic pathogen- esis, through inhibition of trophoblast proliferation of antibody reactivity with still other plasma proteins ␤ involved in haemostasis. Thus, inhibition of activated by anti- 2 GPI, has also been proposed. and its cofactor protein S by antiphospho- Although thrombosis and miscarriage are likely to lipid containing sera has been noted and antiphos- have an autoimmune basis in many cases of APS, in pholipid antibodies may induce resistance to the anti- other instances antiphospholipid antibodies could coagulant effect of activated protein C in vitro. Auto- represent an epiphenomenon, perhaps arising through antibodies to thrombomodulin and to phospholipid exposure of neoepitopes on proteins bound to cell bound protein C and protein S have also been report- membrane anionic phospholipids exposed following ed. Other autoantibodies impair the inhibition of cell injury from some other cause. by antithrombin III and still others react Problems in diagnosis with the phospholipid binding protein annexin V. Clearly APS is a heterogeneous condition, both in Cellular reactivity of autoantibodies is also a feature relation to its clinical manifestations and to the asso- in APS. Whilst ␤2 GPI dependent binding of antiphos- ciated range of autoantibodies. Whilst the diagnosis pholipid antibodies to has been demon- of APS can often be made with confidence because of strated, other autoantibodies react with the major the typical constellation of clinical symptoms and platelet membrane glycoproteins and are distinct signs and informative laboratory data, these may not from antiphospholipid antibodies and similar to be conclusive. Significant diagnostic and prognostic those responsible for autoimmune thrombocytope- difficulties arise because of the occurrence of anti- nia. Additionally, antiendothelial antibody reactivity phospholipid antibodies secondary to infections, in 34 M. Greaves relation to medications, and in some apparently bility of iatrogenic complications of treatment. Oral healthy subjects. The association with infection was anticoagulant therapy carries an inevitable risk of seri- first recognised in syphilis and was used to aid diag- ous haemorrhage. Although data from retrospective nosis. Antiphospholipid antibodies also occur in studies of highly selected patients suggests a possible infection with HIV and hepatitis C and in some other need for more intensive therapy, until prospective data infections, including cytomegalovirus. Such antibod- become available, such as from the ongoing WAPS Tri- ies may be transient and are generally not associated al, a target INR of 2.5 is a reasonable aim in most with thrombosis. Exceptions occur, however, for patients. example in the syndrome of purpura fulminans which Concerns have been expressed over the validity of occasionally complicates varicella infection in chil- the INR in oral anticoagulant dosing control, when LA dren in which the presence of LA has been linked to is present. The inhibitor occasionally causes prolon- protein S deficiency and extensive thrombosis. gation of the prothrombin time, and therefore the Drug-induced antiphospholipid antibodies may INR, which may thus not reflect the true degree of also cause diagnostic confusion. anticoagulation. This phenomenon appears to be Because of the heterogeneity of autoantibodies in reagent dependent and can usually be circumvented APS, a comprehensive approach to laboratory inves- by careful selection of the thromboplastin and instru- tigation is essential. This should include coagulation ment used in the prothrombin time test. based tests for LA as well as solid phase assays for In relation to the management of recurrent miscar- ACL, as both tests are positive in only around 50% of riage in APS, results of a controlled clinical trial indicate cases of undoubted APS. Unfortunately there are better outcome in women treated with low dose twice major limitations to the laboratory methods avail- daily unfractionated heparin with aspirin than in those able. Detection of LA relies upon prolongation of the given aspirin alone. Use of high doses of corticosteroids clotting times in phospholipid dependent coagula- in pregnancy is associated with generally unacceptable tion assays but there are pitfalls relating to the per- maternal morbidity and there are considerable doubts formance of these tests and their interpretation. For about the efficacy of these drugs in APS. example the reagents employed in coagulation assays vary considerably in their sensitivity. Furthermore, het- Heparin-induced thrombocytopenia with erogeneity of antiphospholipid antibodies leads to thrombosis variable levels of positivity in the different coagula- Thrombocytopenia is a common consequence of tion assays, antiprothrombin antibodies behaving dif- heparin therapy. The more common Type I heparin- ferently from anti-␤2 GPI antibodies. induced thrombocytopenia is early in onset, trivial in There are also problems in relation to tests for ACL. degree and clinically benign. This contrasts with Type Despite the establishment and distribution of cali- II (heparin-induced thrombocytopenia with throm- bration materials and the general adoption of units bosis-HITT), which occurs later (typically after at least for the expression of quantitative data, consistency 5 days of heparin exposure), is often severe in degree between laboratories is poor. Commercial reagents and is frequently associated with extension of, or new, and kits are popular but vary widely in their content arterial or venous thrombosis. It is an occasional and of ␤2 GPI in buffers and wells and in the recom- potentially life-threatening complication of treatment mended normal ranges. The level of agreement with heparin in prophylactic or therapeutic dosage. between these assays is unsatisfactory. This is clini- Pathogenesis of HITT cally important as the titre of IgG ACL appears to be prognostically informative and there is doubt over the The target antigen for HITT autoantibody is a com- relevance of low titre antibodies. Newer assays, par- plex between heparin and platelet factor 4 (PF4). This ticularly for ␤2 GPI, are under evaluation. Anti-␤2 GPI glycoprotein is released from platelet alpha granules may be more strongly associated with a history of during activation and complexes form with heparin, thrombosis than is ACL, but prospective data are some on the platelet surface. When an IgG autoanti- required to clarify the role of this assay. body has developed, it binds to the complex and induces platelet activation through interaction with Management of thrombosis and miscarriage the platelet Fc-␥RII receptor. When this receptor is There are few prospective clinical trial data on which tightly occupied stimulus response coupling occurs, to base treatment decisions. Therapeutic regimens with thromboxane synthesis and platelet release. This should be guided by the results of recent observation- leads to platelet consumption in thrombus, with ves- al studies, which support an association between sel occlusion and thrombocytopenia. Endothelial cell antiphospholipid antibodies and thrombosis, particu- activation may also be involved, perhaps through larly thrombosis recurrence. Attention should also be expression of after interaction between paid to the particular clinical manifestation and its autoantibody and heparin-glycosaminoglycan com- severity, the strength of the laboratory evidence for the plexes on the endothelial cell plasma membrane. presence of antiphospholipid antibodies, the coexis- HITT autoantibodies are predominantly IgG2 and tence of other, modifiable risk factors, and the possi- their ability to bind to platelet Fc-gammaRII is deter- Session #4 – Acquired disorders of haemostasis 35 mined in part by a polymorphism in the gene for the of such an autoimmune thrombotic disease is receptor (resulting in an Arg-His substitution at posi- unclear, but the clinical parallels with APS and patho- tion 131). genetic similarities to HITT are striking. Diagnosis and management In a further subject unexplained arterial thrombo- sis was associated with the presence of an autoanti- The most important diagnostic manoeuvre is the body against the thrombin anion-binding exosite. The maintenance of a high index of clinical suspicion for the condition. Available tests for the autoantibody are antibody caused prolongation of the plasma throm- either incompletely sensitive (platelet aggregation and bin time and inhibited platelet aggregation by throm- release assays) or not totally specific (immunoassays bin, but also inhibited thrombin-induced synthesis of for heparin-PF4) but assist in clinical assessment. In prostaglandin I2 by cultured endothelial cells and HITT, heparin therapy must be immediately with- blocked thrombin-thrombomodulin activation of drawn. Where anticoagulant therapy is necessary, the protein C. These last two phenomena suggest that choice usually lies between continuation of coumadin, this patient represents a further example of an auto- if oral anticoagulation has been established, and the immune thrombophilic syndrome. It is likely that oth- heparinoid orgaran. Cross-reactivity between HITT ers will be recognised in the future and that an under- autoantibody and orgaran is relatively uncommon. standing of the mechanisms involved will lead to more Where absence of cross-reactivity with a low molecu- rational therapeutic strategies. lar weight heparin preparation can be demonstrated this provides an alternative therapeutic option. More recently lepirudin (a recombinant form of hirudin) has References been successfully and safely employed as anticoagu- 1. Amiral J, Bridey F, Dreyfus M, et al. Platelet factor 4 lant therapy for thrombosis in subjects with HITT, as complexed to heparin is the target for antibodies gen- well as to cover cardiac surgery and haemodialysis in erated in heparin-induced thrombocytopenia. such individuals. It is a significant addition to the ther- Thromb Haemostas 1992; 68: 95-6. apeutic armamentarium in HITT. 2. Arnaud E, Lafay M, Gaussem P, et al. An autoanti- body directed against human thrombin anion-binding exosite in a patient with arterial thrombosis: Effects on Thrombotic thrombocytopenic purpura platelets, endothelial cells, and protein C. Blood 1994; (TTP) 84:1843-50. It has very recently been demonstrated that spo- 3. Costa JM, Fiessinger JN, Capron L, Aiach M. Partial radic TTP is an autoimmune disorder in which an IgG characterisation of an autoantibody recognizing the secondary binding site(s) of thrombin in a patient with autoantibody to plasma von Willebrand factor cleav- recurrent spontaneous arterial thrombosis. Thromb ing protease develops. The enzyme, a metallopro- Haemostas 1992; 67:193-9. tease, is responsible for the normal processing of 4. Furlan M, Robles R, Galbusera M, et al. von Wille- secreted vWF through cleavage of the peptide bond brand factor-cleaving protease in thrombotic throm- between tyrosine 842 and methionine 843 in bocytopenic purpura and the hemolytic-uremic syn- drome. N Engl J Med 1998; 339:1578-84. monomeric subunits, thereby degrading large vWF 5. Galli M, Comfurius P, Massen C, et al. Anticardiolipin multimers. Presence of the autoantibody appears to antibodies directed not to cardiolipin but to a plasma allow the large multimers to circulate and to bind to protein cofactor. Lancet 1990; 355:1544-7. platelet glycoprotein receptors under high shear con- 6. Greaves M. Antiphospholipid antibodies and throm- ditions and hence to induce platelet aggregation. This bosis. Lancet 1999; in press. 7. Hoylaerts MF, Thys C, Arnout J, Vermylen J. Recurrent accounts for the predominant clinical features- arterial thrombosis linked to autiommune antibodies thrombocytopenia and organ dysfunction due to enhancing von Willebrand factor binding to platelets microvascular occlusion by platelet thrombi. and inducing Fc-gammaRII receptor-mediated platelet The reason for the transient immune dysregulation activation. Blood 1998; 91:2810-7. in sporadic TTP is unknown, but the above patho- 8. McNeil HP, Simpson RJ, Chesterman CN, Krilis SA. Antiphospholipid antibodies are directed against a genetic mechanism provides an explanation for the complex antigen that includes a lipid-binding inhibitor effectiveness of high volume plasma exchange thera- of coagulation: beta2-glycoprotein I (apolipoprotein py (through removal of large multimers and autoan- H). Proc Natl Acad Sci USA 1990; 87:4120-4. tibody and provision of protease), and for the occa- 9. Schulman S, Svenungsson E, Granqvist S. Anticardio- sional response to other treatments such as immuno- lipin antibodies predict early recurrence of venous thromboembolism and death among patients with adsorption, immunosuppression and . venous thromboembolism following anticoagulant therapy. Am J Med 1998, 104:332-3. Other autoimmune thrombophilic states 10. Tsai H-M, Lian EC-Y. Antibodies to von Willebrand Recurrent arterial thrombosis and late foetal loss factor-cleaving protease in acute thrombotic throm- have been attributed to low affinity autoantibodies bocytopenic purpura. N Engl J Med 1998; 339:1585- 94. to vWF which enhance its binding to platelets and 11. Vermylen J, Hoylaerts MF, Arnout J. Antibody-medi- lead to platelet activation through antibody interac- ated thrombosis. Thromb Haemostas 1997; 78:420- tion with the platelet Fc-␥RII receptor. The prevalence 6. Educational Session 4 Haematologica 1999; 84:(EHA-4 educational book):36-39 Chairman: M. Greaves Acquired disorders of haemostasis

The pathogenesis and management of essential thrombocythaemia ANTHONY R. GREEN University of Cambridge, Department of Haematology, MRC Centre, Cambridge, UK

he myeloproliferative disorders (MPD) are selective advantage for those stem cells in which one thought to represent clonal disorders which parental X chromosome is active, relative to the Tresult from transformation of a pluripotent remaining stem cells in which the other parental X stem cell. Essential thrombocythaemia (ET) has been chromosome is active. However, whatever the mech- regarded as a member of the MPD for nearly fifty anism, the presence of a clonal pattern in granulo- years since Dameshek’s perceptive paper. However, cytes with polyclonal T cells is clearly not a specific there is a growing awareness that the term ET prob- diagnostic marker for the MPD in elderly women. ably includes a heterogeneous group of disorders Despite these limitations some useful information which differ in their pathogenesis and may require can be gained from clonality studies. In particular, different approaches to their management. two recent papers have demonstrated that approxi- Pathogenesis mately a third of patients with ET have a polyclonal pattern in both granulocytes and T cells.4,5 Although Clonality studies it is not possible to exclude the existence of a sub- Much of our current understanding of the patho- population of clonally derived granulocytes, these genesis of the MPD is based on X chromosome inac- data suggest the existence of pathogenetically dis- tivation studies. Earlier reports on rare glucose-6- tinct subsets of patients with ET. It has even been phosphate dehydrogenase (G6PD) heterozygotes proposed that patients with a polyclonal pattern in demonstrated that three patients with ET exhibited both granulocytes and T cells may be at lower risk of skewed G6PD expression in red cells, granulocytes thrombotic complications, although this suggestion and platelets, but not in skin. These results have been was based on very small numbers of patients and taken to indicate the presence of a clonal disorder requires confirmation. resulting from transformation of a haemopoietic stem cell. In support of this model several studies Cytogenetics subsequently analysed methylation of PGK, HPRT Karyotypic abnormalities are rare in ET and no con- and M27b loci and demonstrated skewing in unfrac- sistent chromosomal changes have been observed. tionated blood cells in a majority of patients with Rare reports of consistent chromosome changes have various MPDs. not been substantiated. However, this relatively simple picture was compli- Haemopoietic progenitors cated by two issues. Firstly, constitutional skewing (Lyonisation) can occur in normal females. In order Erythropoietin-independent BFU-E have proved a to exclude Lyonisation T cells have increasingly been useful diagnostic test in polycythaemia vera (PV). Sim- used as a control. The majority of patients with an ilar erythroid colonies are found in ET, although in a MPD have been found to exhibit a skewed pattern in smaller proportion of patients.6 Interestingly it has granulocytes but a polyclonal pattern in T cells or been suggested that many ET patients with erythro- mononuclear cells. This pattern was thought to indi- poietin-independent BFU-E subsequently develop PV. cate the presence of a clonal MPD. These results underline the similarities between ET However, the second complication then reared its and PV and suggest that erythroid colony assays may head - namely the existence of acquired skewing.1 be a useful way of distinguishing the two diseases. Two groups found that a clonal pattern in granulo- The growth of colonies in the cytes with a polyclonal pattern in T cells was present absence of exogenous sources of thrombopoietin has in approximately 25% of elderly women.2,3 Several also been investigated. In serum-containing cultures, mechanisms could be responsible for this phenome- factor-independent megakaryocyte colonies could be non.4 Perhaps the most likely explanation invokes X- found in the majority of patients with ET and also in linked allelic differences which give rise to a subtle a substantial proportion of patients with PV, but not in normal controls or in individuals with reactive Correspondence: A.R. Green, University of Cambridge, Department of thrombocytosis.6 However, this assay is difficult to Haematology, MRC Centre, Hills Road, Cambridge CB2 2QH. Tel. inter- national +44-1223-336835 – Fax: international +44-1223-762670 – standardise and is not widely available as a diagnos- E-mail: [email protected] tic tool. Session #4 – Acquired disorders of haemostasis 37

Molecular studies including ET and PV. Secondary thrombocytosis is Haemopoietic progenitors from patients with PV characterised by a normal megakaryocyte compart- display an abnormal sensitivity to, or a reduced ment responding to an extrinsic stimulus. Inherited requirement for, several different growth factors. This thrombopoietin mutations therefore fall into the cat- observation has focused attention on growth factor egory of secondary congenital thrombocytosis. Sec- signalling pathways. In the context of ET this reason- ondary acquired thrombocytosis will include the large ing has been reinforced by the description of two fam- number of pathological conditions which are accom- ilies with inherited thrombocytosis associated with panied by increased platelet formation. Finally, idio- mutations in the thrombopoietin gene.7,8 These muta- pathic thrombocytosis is a new term analogous to tions resulted in an increased production of throm- idiopathic erythrocytosis. What is currently called ET bopoietin and an increased serum thrombopoietin would include patients with idiopathic thrombocyto- level. However, no mutation of the thrombopoietin sis. What we need, but currently lack, are good posi- gene has been reported in patients with sporadic ET. tive diagnostic criteria for ET. Reduced expression of the thrombopoietin receptor (c-mpl) at both RNA and protein levels has been Management reported in the platelets of ET patients.9 By contrast, Many patients with ET remain well with minimal Moliterno et al.10 demonstrated reduced expression in treatment. Vascular events provide the main source of patients with PV and myelofibrosis but not in four morbidity and mortality. The dilemma is that vascu- patients with ET. The reasons for this discrepancy are lar complications, when they arise, can be cata- not clear, but may involve differences in diagnostic strophic. criteria or may reflect pathogenetic heterogeneity. No Haemorrhage is not common and usually only mutations of the c-mpl gene have been found in ET occurs in the presence of very high platelet counts patients. In the absence of any evidence for mutation (greater than 1000ϫ109/L). The major complication of the genes for either thrombopoietin or its receptor is thrombosis, both arterial and venous. Three main in patients with ET, attention is now shifting to intra- factors predispose to thrombosis: increasing age, pre- cellular signalling pathways. vious thrombosis and degree of thrombocytosis.11 In addition to its vascular complications, ET can also Classification and diagnosis transform into acute myeloid leukaemia (AML), There is an increasing realisation that a number of myelodysplastic syndromes or myelofibrosis. These disease entities may be lumped together under the transformations can all occur in patients who have heading of ET. It may therefore be timely to reconsider not received any cytotoxic therapy, but their frequen- how best to classify patients with thrombocytosis cy appears to be influenced by the choice of therapy (Table 1). (see below). When considering the best management Thrombocytosis is defined as a platelet count for patients with ET it may, therefore, be useful to ϫ 9 greater than 400 10 /L. Primary thrombocytosis stratify them into groups with high, intermediate or results from an intrinsic abnormality in megakary- low risks of vascular events (Table 2). ocyte formation. No example of a primary congenital thrombocytosis has yet been described but it seems High risk patients likely that inherited mutation of the c-mpl receptor or One of the few clear messages concerning the man- other signalling component may emerge before long. agement of patients with ET is that it is important to Primary acquired thrombocytosis can result from a lower the platelet count in patients at high risk of vas- number of clonal haematological malignancies cular events. In 1995 Cortelazzo et al.12 reported a ran-

Table 2. Proposed risk categories for patients with ET. Table 1. Proposed classification of the thrombocytoses.

High risk - patients with any of the following: Primary thrombocytosis Age ≥ 60 years Congenital Platelets > 1000ϫ109/L Acquired, e.g. essential thrombocythaemia, polycythaemia vera, History of ischaemia, thrombosis or embolic events chronic myeloid leukaemia, myelodysplastic syndromes Presence of hypertension or diabetes

Secondary thrombocytosis Intermediate risk Congenital, e.g. autonomous high thrombopoietin production Age 40-60 years Acquired, e.g. haemorrhage, inflammation, tumour Absence of high risk features

Idiopathic thrombocytosis Low risk Age < 40 Adapted from Pearson, Diagnosis and classification of erythrocytoses and Absence of high risk features thrombocytoses. Baillière’s Clinical Haematology. In press. 38 A. R. Green domised trial comparing hydroxyurea with no platelet tion with other such cytotoxic agents, and followed lowering drug in high risk patients with ET. Vascular up for eight years. The frequency of acute leukaemia events occurred in 3.6% of patients in the hydroxyurea or myelodysplastic syndrome was 3.5% in patients arm compared with 24% of patients in the control receiving hydroxyurea alone and 14% in patients arm over a median follow up of 27 months. Hydroxy- receiving hydroxyurea in combination with other cyto- urea remains the therapy of choice for most high risk toxic agents. In the absence of any therapy with cyto- patients. Busulphan is effective but is perceived as toxic drugs, the background frequency of AML/MDS having more side effects and as an alkylating agent, it in patients with ET is unknown. is also more likely to be leukaemogenic. Interferon is also effective, but is limited by cost and also by toxi- What about the use of hydroxyurea in PV? A recent city, especially in elderly patients. report compared hydroxyurea with pipobroman in the Anagrelide provides the most exciting potential treatment of patients with PV.15 The risk of leukaemia alternative to hydroxyurea.13 It acts to lower the in patients receiving hydroxyurea was approximately platelet count by a novel and poorly understood 5% at ten years and therefore comparable to the 3.5% mechanism. It is not a cytotoxic drug and is, therefore, observed in patients with ET.14 The risk of leukaemia unlikely to be leukaemogenic, but instead promotes was reported as increasing to 10% over thirteen years megakaryocyte differentiation. Most strikingly it does but these figures are very unreliable since only a small not suppress the white cell count or haemoglobin. minority of patients remained under observation at However, it is a positive inotrope drug and a vasodila- thirteen years. Unlike ET, there are some data on the tor, thus explaining its major side effects of palpita- background incidence of leukaemia in patients not tions, headaches and fluid retention. Moreover, ana- receiving cytotoxic drugs. A 1.5% incidence of grelide does not reduce megakaryocyte numbers and leukaemia was reported in patients entered into the the significance of this for the subsequent develop- phlebotomy-only limb of the PVSG-01 study. Howev- ment of myelofibrosis is unknown. The results of er, this figure is misleading for two reasons. First, it was prospective randomised trials comparing hydroxyurea not based on an intention to treat basis. Patients requir- and anagrelide, such as the ongoing MRC Primary ing cytoreductive therapy were censored from analysis Thrombocythaemia trial, should be very informative. and the figure of 1.5% therefore relates to a highly selected group of patients. Secondly, the number of Low risk patients patients managed by phlebotomy alone diminished There is considerable evidence that young patients rapidly over time to 10% by the tenth year. Thus the with ET have a particularly low risk of vascular events. median follow up time is significantly shorter than the In the absence of any high risk features, such as pre- other limbs of the PVSG-01 trial. vious thrombosis or a very high platelet count, it therefore seems reasonable to treat such patients with The jury therefore remains out. Existing data do not aspirin alone (75 mg/day). However, it is worth demonstrate that hydroxyurea is leukaemogenic when emphasising that the natural history of ET in young used alone to treat patients with ET or other MPD. patients is obscure, and the incidence of late compli- cations including myelofibrosis remains unknown. Intermediate risk patients References The management of this group of patients provokes 1. Busque L, Mio R, Mattioli J, et al. Nonrandom X-inac- the greatest controversy. Lowering the platelet count tivation patterns in normal females: Lyonization ratios exposes patients to the potential side effects of the vary with age. Blood 1996; 88:59-65. various platelet-lowering drugs. However, not lower- 2. Champion KM, Gilbert JGR, Asimakopoulos FA., Hin- ing the platelet count is likely to increase the risk of shelwood S, Green AR. Clonal haemopoiesis in nor- potentially catastrophic vascular events. Randomised mal elderly women: implications for the myeloprolif- erative disorders and myelodysplastic syndromes. Br J data will be needed to provide rational guidance. Haematol 1997; 97:920-6. Is hydroxyurea leukaemogenic? 3. Gale RE, Fielding AK, Harrison CN, Linch, DC. Acquired skewing of X-chromosome inactivation pat- This issue is much discussed but the facts are rela- terns in myeloid cells of the elderly suggests stochastic tively simple. There is currently no convincing evidence clonal loss with age. Br J Haematol 1997; 98:512-9. that hydroxyurea increases the incidence of leukaemia 4. El-Kassar N, Hetet G, Briere J, Grandchamp B. Clon- when it is used alone to treat ET. Reports of the ality analysis of hematopoiesis in essential thrombo- cythaemia: advantages of studying T lymphocytes and leukaemogenicity of hydroxyurea are based on anec- platelets. Blood 1997; 89:128-34. dotal case reports or on retrospective and therefore 5. Harrison CN, Gale RE, Machin SJ, Linch DC. A large systematically biased studies. Retrospectively com- proportion of patients with a diagnosis of essential paring treated and untreated patients is misleading thrombocythemia do not have a clonal disorder and may be at lower risk of thrombotic complications. since the two groups are biologically different. Blood 1999; 93: 417-24. In a large recently reported study,14 251 patients 6. Juvonen E, Ikkala E, Oksanen K, Ruutu T. Megakaryo- were treated with hydroxyurea alone or in combina- cyte and erythroid colony formation in essential Session #4 – Acquired disorders of haemostasis 39

thrombocythaemia and reductive thrombocytosis: tive study of the GIMMC group in two thousand diagnostic value and correlation to complications. Br patients [abstract]. Blood 1997; 90 (suppl 1): 348a. J Haematol 1993; 83:192-7. 12. Cortelazzo S, Finazzi G, Ruggeri M, et al. Hydroxyurea 7. Kondo T, Okabe M, Sanada M, et al. Identification of for patients with essential thrombocythemia and a TPO gene mutation in familial essential thrombo- high risk of thrombosis. N Engl J Med 1995; 332: cythemia. Blood 1997; 92:1091-6. 1132-6. 8. Wiestner A, Schlemper RJ, van der Maas APC, Skoda 13. Tefferi A, Silverstein MN, Hoagland HC. Anagrelide RC. An activating splice donor mutation in the throm- as a new platelet lowering agent in essential throm- bopoietin gene causes hereditary thrombocythaemia. bocythemia: mechanism of action, efficacy, toxicity, Nat Genet 1998; 18: 49-52. current indications. Semin Thromb Hemost 1997; 23: 9. Horikawa Y, Matsumura I, Hashimoto K, et al. Markedly reduced expression of platelet c-mpl recep- 379-83. tor in essential thrombocythaemia. Blood 1997; 90: 14. Sterkers Y, Preudhomme C, Lai JL, et al. Acute myeloid 4031-8. leukemia and myelodysplastic syndromes following 10. Moliterno AR, Hankins WD, Spivak JL. Impaired essential thrombocythemia treated with hydroxyurea: expression of the thrombopoietin receptor by platelets high proportion of cases with 17p deletion. Blood from patients with polycythemia vera. N Engl J Med 1998; 91: 616-22. 1998; 338:572-80. 15. Najean Y, Rain JD. Treatment of polycythemia vera: 11. Gugliotta L, Marchioli R, Fiacchini M, et al. Epidemi- the use of hydroxyurea and pipobroman in 292 ological, diagnostic, therapeutic and prognostic patients under the age of 65 years. Blood 1997; 90: aspects of essential thrombocythemia in a retrospec- 3370-7. Educational Session 4 Haematologica 1999; 84:(EHA-4 educational book):40-43 Chairman: M. Greaves Acquired disorders of haemostasis

Hepatic veno-occlusive disease following peripheral blood stem cell or bone marrow transplantation CHRISTOPH SALAT, ERNST HOLLER, ERHARD HILLER, RUDOLF PIHUSCH, HANS-JOCHEM KOLB Medical Department III, Klinikum Grosshadern, Ludwig-Maximilians-Universität München, Munich, Germany

epatic veno-occlusive disease (VOD) was first was introduced. It was classified as mild (no medica- described after ingestion of bush tea in tion required, complete resolution) in 12%, moderate HJamaica. Toxic pyrrolizidine alkaloids led to (requirement of sodium restriction, diuretics, anal- obstruction of small intrahepatic venules. gesics, complete resolution of liver injury) in 26% or The most relevant condition associated with VOD severe (adverse effects from liver damage, symptoms today is high dose radiochemotherapy followed by and laboratory abnormalites without resolution bone marrow or peripheral blood stem cell trans- before day 100 or death, whichever occurred first) in plantation. According to our present knowledge, the 15% of the patients. The corresponding day 100 fatal- toxicity of the myeloablative therapy in addition to ity rates were 9%, 23% and 98%, respectively. risk factors and secondary events is responsible for In another earlier study the Baltimore group this complication. For transplant patients VOD rep- defined VOD as follows: hyperbilirubinaemia equal resents the third leading cause of death and the most to or greater than 2 mg/dL and at least two of the fol- severe regimen-related toxicity. lowing three criteria: hepatomegaly, ascites, or weight Due to an increasing number of patients treated gain of at least 5% before day 21.2 In this study a with high dose chemotherapy and stem cell trans- VOD rate of 22% and a fatality rate of 45% was plantation the incidence of VOD may rise in the reported. future. On the other hand, the modifications of pre- A remarkably lower rate was recently reported by transplant conditioning regimens (mini transplants) in Carreras et al. in an EBMT survey including 1,652 the allogeneic setting may contribute to a reduction patients.3 In this largest evaluation to date, VOD was of VOD in this population of patients. defined as the occurrence of at least two of the fol- lowing criteria: hyperbilirubinaemia >2mg/dL, ascites Definition and incidence of VOD or weight gain (>5% of baseline body weight), and Two definitions of VOD are used: one proposed by painful hepatomegaly before day 21. A rate of 8.9% McDonald et al. from the Seattle group and the oth- after allogeneic and 3.1% after autologous trans- er by Jones et al. from Baltimore. plants was reported. Whereas in the EBMT trial only The first study was performed retrospectively by the 29% of the patients received allogeneic transplants, Seattle group in 1984 and described an incidence of the vast majority of patients in the Seattle trial1 were 22% with fatal outcome in 45%. The definition transplanted from allogeneic donors. included two of these four criteria: hyperbilirubi- When considering several risk factors, in the Euro- naemia >1.6 mg/dL, painful hepatomegaly and pean study the incidence of VOD for allogeneic trans- ascites or weight gain before day 30. plant recipients with a Karnofsky performance score In a more recent prospective analysis by the same <90%, who received high dose chemotherapy, was group the definition was modified and included two 25%. This rate is comparable to the incidence found of the following events: hyperbilirubinaemia >2 in the Seattle and Baltimore analyses. mg/dL, hepatomegaly or right upper quadrant pain of These differences in the VOD rate point to the liver origin, and sudden weight gain exceeding 2% of importance of the definition used, the composition baseline body weight due to fluid retention until day of the population studied in terms of autologous or 20 post-transplant.1 Considering these criteria the allogeneic transplants, and pre-existing risk factors. incidence of VOD increased to 54% and the mortali- Pathophysiology ty rate for patients with VOD before day 100 amount- ed to 39%. The pathophysiology of VOD is incompletely In this study a widely used classification of VOD understood. Endothelial cell injury due to the toxici- ty of the conditioning regimen, however, is believed to be the first step in the pathogenesis. The observa- ␣ Correspondence: Christoph Salat, MD, Medical Department III - Haema- tion, that TNF is elevated in patients with VOD, that tology/Oncology; Transplantation Unit, Klinikum Grosshadern der Lud- TNF␣ exerts procoagulant and hypofibrinolytic wig-Maximilians-Universität München, Marchioninistr. 15, 81377 Munich, Germany. Tel. international +49-89-70951 – Fax. international effects and that plasminogen activator inhibitor 1 +49-89-70954242 – E-mail: [email protected] (PAI-1), the main inhibitor of the fibrinolytic system, Session #4 – Acquired disorders of haemostasis 41 is remarkably elevated in VOD patients led us to pro- of the early post-transplant phase, manifestations at pose a two step model of the pathogenesis of VOD.4 later stages can occur. The initial conditioning-induced endothelial cell dam- Differential diagnosis includes other toxic liver age is followed by a second step resulting in Kupffer injury, mostly due to the conditioning regimen, cyclo- cell activation and cytokine release, which induces the sporine, antibiotics, methotrexate or other medica- release of PAI-1 from endothelial cells thereby pro- tion in the early phase after transplantation. Hepatic moting hypofibrinolysis. This second event may be graft-versus-host disease (GvHD) is another impor- the translocation of bacteria through the damaged tant differential diagnosis. It is rarely associated with gastrointestinal mucosa leading to endotoxinaemia. ascites. In the majority of cases GvHD of the liver man- Furthermore, the possible relevance of a depletion ifests later than VOD and is often associated with skin of glutathione, which can be provoked by radio- or gut symptoms. Nevertheless, there is a notable over- chemotherapy and may promote hepatocyte or lap between VOD and hepatic GvHD symptoms and endothelial cell damage, is presently being investigat- signs. ed. Deleve et al. have demonstrated that radiochemo- Fungal and viral infections or cholangitis in the therapy-induced alterations in glutathione levels lead- course of sepsis must be considered, although liver ing to increased vulnerability of the endothelium are damage in hepatitis B and C is mediated by a func- linked to endothelial cell damage. tioning immune system. In addition, cardiac, renal and pancreatic diseases may cause abnormal liver Risk factors for the development of VOD function and/or ascites and must be distinguished from VOD. McDonald1 identified elevated pre-transplant trans- aminase levels, high dose chemotherapy and persisting Making the diagnosis – histological, fever during conditioning as independent indicators of severe VOD. Amphotericin was associated with an laboratory and technical investigations increased risk of severe VOD and vancomycin and acy- VOD can not be adequately diagnosed by clinical clovir with an increased risk of a fatal outcome. Jones2 criteria alone, but is most reliably confirmed by biop- confirmed a pre-existing enzyme elevation (aspartate sy. Characteristic histological findings have been 7 aminotransferase, AST) as a risk factor by multivariate reported by Shulman. Severe VOD is associated with analysis, whereas the conditioning regimen and type of the histological findings of venous occlusion, zone 3 graft did not influence the VOD incidence. hepatocyte necrosis, sinusoidal fibrosis and eccentric In 1998, Hägglund5 described pre-transplant nor- luminal narrowing or phlebosclerosis. Deposition of ethisterone treatment, bilirubin >26 µmol/L before fibrin and von Willebrand factor and fibrous vessel BMT, one HLA-mismatch, previous abdominal irra- obliteration were also reported earlier by the same author. diation and conditioning with busulphan as risk fac- Since VOD is a complication occurring during the tors for VOD in a multivariate analysis. early post-transplant phase, patients suffer from In the large EBMT survey, Carreras et al.3described severe thrombocytopenia and impaired coagulation elevated levels of AST before transplant, allogeneic at the time of symptoms. Therefore, percutaneous BMT, high dose cytoreductive therapy, Karnofsky per- biopsy is associated with a high risk of severe or even formance score <90% and prior abdominal irradia- fatal bleeding complications, which has to be weighed tion as independent variables associated with an 3 against the potential benefit of having a biopsy sam- increased risk of VOD. In this analysis VOD was clas- ple for histological analysis. Additionally, due to a sified as being mild, moderate and severe in 8%, patchy pattern of histologic VOD alterations and the 64.4% and 27.6%, respectively. coincidence of hepatic VOD and GvHD, biopsy may Patients with pre-existing hepatitis are at a high risk yield false negative results. The risk of severe bleeding of developing VOD. To predict the course of VOD complications can be reduced by using transvenous 6 after conditioning, Bearman et al. developed a cal- techniques, e.g. transjugular biopsy. culation model which was found to be highly specif- Other proposals for diagnosing VOD have been ic and moderately sensitive. made. Several ultrasound studies have been pub- lished. Gray scale and Doppler ultrasound criteria, Clinical features and differential including hepatosplenomegaly, gall bladder wall diagnosis thickening, ascites, paraumbilical vein recanalisation Patients typically present with an increase of biliru- and modulation of intrahepatic vessel diameters and bin, hepatomegaly, right upper quadrant pain, ascites flow characteristics have been published by several and otherwise unexplained weight gain. Whereas sud- groups. A sensitivity and specificity of 83% and 87%, den weight gain and hepatomegaly are early signs respectively, at the most accurate cut-off level of the occurring as soon as day 0, hyperbilirubinemia (medi- total score was described by Lassau. An enhancement an day 6), peripheral oedema (day 7) and ascites (day of the wedge pressure in the hepatic vein (>9 mm Hg) 12) follow at later stages with remarkable interindi- has been demonstrated to be a highly specific para- vidual differences.6 Although VOD is a complication meter for VOD. 42 C. Salat et al.

Several laboratory parameters can be helpful in the which was associated with severe capillary leak syn- case of patients with hyperbilirubinaemia and sus- drome, was successfully treated with a combination of pected VOD. High levels of plasminogen activator rtPA and C1-esterase inhibitor concentrate. inhibitor 1 (PAI-1), the main inhibitor of the fibri- Whereas fast resolution of symptoms occurred in nolytic system, was found to be very specific and sen- some patients after thrombolytic therapy with rtPA, sitive for diagnosing VOD, whereas normal levels were severe bleeding complications were observed in oth- found in patients with GvHD and other causes of ers. The analysis of the 42 patients treated in Seattle hyperbilirubinaemia.4 These results have been con- – the largest analysis so far - yielded disappointing firmed by prospective weekly PAI-1 screening of trans- results.9 Symptoms improved in 29% of the patients, plant recipients in our centre (unpublished data) and but severe bleeding episodes were observed in ten point to a possible involvement of hypofibrinolysis in patients. Requirement for mechanical ventilation, the pathogenesis of VOD. supplemental oxygen and dialysis before thrombolyt- Protein C, the natural anticoagulant, is reduced in ic therapy were indicators of no response. These data most symptomatic patients and has been described are in accordance with the experience at other centres, to be indicative of VOD even before conditioning.8 showing that patients with multiorgan failure are not Nevertheless, it is unclear at the present time whether candidates for thrombolytic therapy. low protein C levels are a cause or symptom of VOD. Several shunting techniques including transjugular Procollagen III peptide levels have also been report- intrahepatic portosystemic shunts (TIPS), have been ed to be raised in VOD patients. A reduction of proposed for the management of severe VOD and, in antithrombin and protein S as well as increases of von isolated cases, liver transplants have been performed. Willebrand factor, F VIII, fibrinogen, tissue factor or Nevertheless, none of these therapeutic procedures F VII have been described in smaller surveys. can be regarded as standard therapy. A typical phenomenon in VOD patients is a high Recently, Richardson et al.10 reported on their expe- requirement for platelet transfusions and persistent rience with defibrotide, a polydeoxyribonucleotide, thrombocytopenia after engraftment. The patho- known to enhance tPA and thrombomodulin and physiology is not understood but the low platelet lower PAI-1 expression in endothelial cell cultures. count is not due to diminished hepatic thrombopoi- Nineteen patients with severe VOD were treated. The etin synthesis according to data recently published by drug was well tolerated and in 8 patients of this high Oh et al. risk group a reversal of hyperbilirubinaemia and oth- er VOD symptoms was achieved, 6 patients survived Prophylaxis beyond day 100. No severe haemorrhage occurred. Recommendations for the prophylaxis of VOD are These data point to the need for further investigation contradictory. Heparin, either in unfractionated or of defibrotide, a drug which holds promise of improv- low molecular weight form, is used in some centres. ing VOD therapy in the future. Attal et al. showed, in a randomised trial with 161 patients, a significant reduction of VOD in the group treated with unfractionated heparin (2.5 vs. 13.7%, Acknowledgements p<0.01) but the study contained only a small pro- We gratefully acknowledge the skilful technical assistance of portion of patients (11%) at high risk of VOD. In the Mrs. B. Reinhardt and the excellent co-operation of the staff EBMT survey3 and other large investigations the use of the transplant units L21 and M21. of heparin in patients with or without risk factors did not influence the incidence or mortality of VOD. Larger randomised trials with precisely defined inclu- References sion criteria are warranted to elucidate the role of A more extensive list of all references can be obtained upon heparin in the prophylaxis of VOD definitively. request from the corresponding author. Prostaglandin E1 and ursodeoxycholic acid have been used in some studies and beneficial effects have 1. McDonald GB, Hinds MS, Fisher LD, et al. Veno- occlusive disease of the liver and multiorgan failure been reported, but the surveys have been too small to after bone marrow transplantation: A cohort study of draw definite conclusions. 355 patients. Ann Intern Med 1993; 118:255-67. Pentoxyphylline, which inhibits TNF␣, has shown 2. Jones RJ, Lee KSK, Beschorner WE, et al. Veno-occlu- positive effects in smaller non-randomised trials, but sive disease of the liver following bone marrow trans- plantation. Transplantation 1987; 44:778-83. these results were not confirmed in randomised trials. 3. Carreras E, Bertz H, Arcese W, et al. Incidence and outcome of hepatic veno-occlusive disease after blood Therapy and marrow transplantation: a prospective cohort Thrombolytic therapy with recombinant tissue plas- study of the European Group for Blood and Marrow minogen activator (rtPA) was successfully used in sin- Transplantation. Blood 1998; 92:3599-604. 4. Salat C, Holler E, Kolb HJ, et al. PAI-1 confirms the gle patients and small surveys. In another report VOD diagnosis of hepatic veno-occlusive disease in patients was resolved in two patients who received rtPA in con- with hyperbilirubinemia after bone marrow trans- junction with antithrombin. In one case report VOD, plantation. Blood 1997; 89:2184-8. Session #4 – Acquired disorders of haemostasis 43

5. Hägglund H, Remberger M, Klaesson S, Lönnqvist B, urally occurring anticoagulants and bone marrow Ljungman P, Ringden O. Norethisterone treatment, a transplantation: Plasma protein C predicts the devel- major risk-factor for veno-occlusive disease in the liv- opment of veno-occclusive disease of the liver. Blood er after allogeneic bone marrow transplantation. 1993; 81:3458-62. Blood 1998; 92:4568-72. 9. Bearman SI, Lee JL, Baron AE, McDonald G. Treat- 6. Bearman SI. The syndrome of hepatic venoocclusive ment of hepatic venoocclusive disease with recombi- disease after marrow transplantation. Blood 1995; nant human tissue plasminogen activator and heparin 85:3005-19. in 42 marrow transplant patients. Blood 1997; 89: 7. Shulman HM, Fisher LB, Schoch HG, Henne KW, 1501-6. McDonald GB. Veno-occlusive disease of the liver after 10. Richardson P, Elias A, Krishnan A, et al. Treatment of marrow transplantation: Histological correlates of severe veno-occlusive disease with defibrotide: com- clinical signs and symptoms. Hepatology 1994; 19: passionate use results in response without significant 1171-8. toxicity in a high risk population. Blood 1998; 92: 8. Faioni EM, Krachmalnikoff A, Bearman SI, et al. Nat- 737-47. Educational Session 5 Haematologica 1999; 84:(EHA-4 educational book):44-45 Chairman: A. Bacigalupo Unrelated donor transplants

Donor selection-matching for mismatches ANDREA L. PAY, MARTHA PEREZ-RODRIGUEZ, J. ALEJANDRO MADRIGAL The Anthony Nolan Research Institute, The Royal Free and University College Medical School, London, UK

election for voluntary unrelated donors (VUD) sity. For HLA class I alleles the highest frequencies of for unrelated bone marrow transplantation substitutions occur within exons 2 and 3, which S(BMT) generally relies on matching for HLA-A, B encode the a1 and a2 domains of the HLA molecules and DR antigens. Initial studies have shown that and for practical purposes 97% of HLA class I alleles recipients of transplants from unrelated donors can be identified by sequence based analysis of exons matched for these HLA antigens have a lower risk of 2 and 3. In contrast, for HLA class II alleles, the high- acute graft-versus-host disease (GvHD) than recipi- est frequency of substitutions occurs mainly in exon ents of marrow from mismatched VUD, but this risk 2 of the b chain. is still much higher than that observed for HLA-iden- Molecular typing methods include: restriction frag- tical sibling transplants.1 Improvement of overall sur- ment length polymorphism analysis (RFLP), vival was not apparent in HLA-A, B, DR matched sequence-specific primer amplification (SSP), hybridi- unrelated recipients when compared with mis- sation with sequence-specific oligonucleotide probes matched related recipients. Thus the effect of HLA (SSO), heteroduplex analysis, single strand confor- matching on survival after unrelated transplantation mation polymorphism (SSCP), and direct nucleotide has been controversial. It is now becoming clear that sequencing. The techniques most commonly used are multiple factors could be associated with these results SSP and SSO.5 SSP utilises group and/or allele specific such as a) the presence of unidentified HLA-A, B and sequences for PCR primer design and SSO typing is DR disparities due to the low resolution HLA typing based on the identification of allele specific sequences techniques used, b) incompatibility for other HLA loci using oligoprobes, which either alone or in certain such as HLA-C, DQ and DP, c) mismatched minor combinations allow allele identification. Heterodu- transplantation antigens and d) non-HLA factors. plex analysis and SSCP allow comparison of the con- At present, typing of most unrelated individuals for formation of DNA molecules, and these are used as class I antigens is still achieved with serology. In addi- supplementary methods by tissue typing laboratories tion to serological typing, cellular assays, e.g. the for allelic subtyping. Direct sequencing in principle MLC reaction2 have served for a secondary definition allows the most accurate typing of HLA alleles. of class II molecular diversity. During the past 10 We have recently described a novel high resolution years, over 800 new HLA alleles have been recognised DNA based typing technique which offers a level of by DNA sequencing analysis and many more proba- resolution similar to direct sequencing techniques, bly remain to be identified. With the discovery of without the problem of heterozygous ambiguities.6 these new alleles, it became obvious that serological The method, now known as Reference Strand mediated specificities comprise multiple undetected molecular Conformation Analysis (RSCA), analyses the conforma- subtypes. For example, there are currently over 100 tion of HLA DNA duplexes. Chimaeric duplexes are recognised HLA-A alleles and of these only a fraction formed between a locus-specific fluorescently labelled can be typed by routine serology. Improvements in reference strand (FLR) and the locus specific alleles DNA-based methods for the detection of the many from the sample to be typed. The duplexes formed HLA alleles have provided the opportunity to investi- are resolved by polyacrylamide gel electrophoresis gate the relationship between HLA disparity and (PAGE) in an automated DNA sequencing instru- transplant complications.3,4 ment with a laser based fluorimetric detection sys- Tissue typing techniques are limited by the exten- tem.6 Only duplexes formed with labelled reference sive polymorphism of HLA genes. A pairwise com- sense strands are observed, i.e. two bands for a parison of the nucleotide sequences of known HLA homozygous sample (labelled reference homoduplex alleles indicates that genetic recombination has and labelled reference/allele duplex) and three bands played a key role in the generation of HLA diversity. for a heterozygous sample (labelled reference homo- Inter-allelic conversion or double recombination is duplex and two labelled reference/allele duplexes). the principal mechanism which generates HLA diver- This method is simple and easy to use, in contrast to the other DNA based methods described previously, Correspondence: Prof. J.A. Madrigal, MD, PhD, FRCP, MRCPath, The which require a large number of group and/or allele Anthony Nolan Research Institute, The Royal Free Hospital, Pond specific probes or primer mixes for HLA class I alleles Street, Hampstead, London, NW2 2QG, UK. Tel. international +44-171- 284-8315; Fax: international +44-284-8331 – E-mail: and thus cannot achieve such high level of resolu- [email protected] tion. Session #5 – Unrelated donor transplants 45

As these incompatibilities may be invisible in routine from all causes9 and class II mismatching is still seen matching techniques, cellular assays have been devel- to be important with regard to GvHD.10 In all cases oped in an attempt to confirm patient/donor identi- where a multiple mismatch with class I and class II ty. Limiting dilution analysis has proved to be a sen- alleles was seen the survival rate was low.10 sitive tool for the detection and investigation of T lym- In unrelated bone marrow transplantation, the ide- phocytes of defined specificity. The cytotoxic T lym- al situation would be to match donors and patients phocyte and helper T lymphocyte precursor (CTLp to the highest resolution possible for all loci. Howev- and HTLp) assays use limiting dilution analysis to er, due to the polymorphism seen in HLA this is most- quantify the frequency of donor cytotoxic T and ly impossible, especially with loci such as HLA-DP helper T cell precursors capable of responding to mis- where there is a very low level of linkage disequilibri- matched HLA antigens present on the patient’s cells.7 um. Therefore, the route to take must be to choose High CTLp frequencies correlate with class I mis- the donor who is most acceptable from those matches usually undetected by conventional typing, matched to the allelic level, for HLA and other poly- whereas HTLp appears capable of detecting class II morphic genes which can affect transplant outcome. differences.8 This decision will only be possible when a definitive How much do we have to match? Many studies hierarchy of factors has been established with respect have attempted to evaluate the clinical contribution to transplant outcome and many scientists are cur- of 0, 1, 2 and multiple mismatches,4,9 but by using rently working towards this goal. high resolution DNA typing it is becoming clear that once a mismatch is detected for one locus, there is a high probability of an associated mismatch for anoth- References er locus. This is understandable given the strong link- age disequilibria of HLA. We have observed a high 1. Madrigal JA, Scott I, Arguello R, Szydlo R, Little A-M, Goldman JM. Factors influencing the outcome of number of associated hidden mismatches between bone marrow transplants using unrelated donors. HLA-C and either HLA-A or B mismatches and there Immunol Rev 1997; 157: 153-66. seems to be a clear additive effect of additional mis- 2. Dupont B, Braun DW, EJY, Carpenter CB. HLA-D by matches in relation to GvHD and TRM. As HLA types cellular typing. In: Terasaki PI, ed. Histocompatibility testing. Los Angeles: UCLA, 1980. p. 229. are increasingly defined to higher degrees of resolu- 3. Fleischhauer K, Kernan N, O’Reilly R, Dupont B, Yang tion, so the probability of finding a completely S. Bone marrow-allograft rejection by T lymphocytes matched VUD is reduced. Until the true levels of HLA recognizing a single amino acid difference in HLA-B44. identity of donor/recipient pairs are resolved by high N Engl J Med 1990; 323:1818-22. resolution DNA typing, an understanding of HLA 4. Scott I, O’Shea J, Bunce M, et al. Molecular typing shows a high level of hla class I incompatibility in sero- matching and mismatching will be imprecise and logically well matched donor/patient pairs: implica- incompatibility will remain underestimated. tions for unrelated bone marrow donor s selection. Recent publications have shown these undetected Blood 1998; 92:4864-71. mismatches to be important with regard to the out- 5. Jordan F, McWhinnie AJ, Turner S, et al. Comparison 4,9,10 of HLA-DRB1 typing by DNA-RFLP, PCR-SSO and come of unrelated BMT. Previous results had indi- PCR-SSP methods and their application in providing cated that class II mismatching was more important matched unrelated donors for bone marrow trans- than matching for HLA class I, however, it is now clear plantation. Tissue Ant 1995; 45:103-10. that class I is just as important as class II, and that 6. Arguello R, Little A-M, Pay AL, et al. Double strand class II mismatches probably reflected mismatches at conformation analysis. Nat Gen 1998; 18: 192-4. 9 7. Kaminski E, Hows J, Man S, et al. Prediction of graft class I that were undetected. HLA-C mismatches are versus host disease by frequency analysis of cytotoxic now thought to be more important than ever before, T cells after unrelated donor bone marrow transplan- with results showing a C locus mismatch to a risk fac- tation. Transplantation 1989; 48: 608-13. tor for GvHD, and somewhat suprisingly, an HLA-C 8. O’Shea J, Madrigal JA, Davey N, et al. Measurement of cytotoxic T lymphocyte precursor frequencies match to be related to leukaemic relapse. This is reveals cryptic HLA class I mismatches in the context thought to be due to a graft-versus-leukaemia of unrelated donor bone marrow transplantation. response mediated by Killer Inhibitory Receptors (KIRs) Transplantation 1997; 64:1353-73. present on natural killer cells or T cells.11 However, 9. Sasazuki T, Juji T, Morishima Y, et al. Effect of match- KIR genotyping of HLA-C mismatched pairs has ing of class I HLA alleles on clinical outcome after transplantation of haemopietic stem cells from an shown that a the repertoire of HLA-C ligands pos- unrelated donor. N Engl J Med 1998; 339:1177-85. sessed by an individual does not directly correspond 10. Petersdorf EW, Gooley TA, Anasetti C, et al. Optimis- to the KIRs expressed, therefore KIRs may be present ing outcome after unrelated marrow transplantation which not only recognise the HLA-C alleles of that by comprehensive matching of HLA Class I and II alle- les in the donor and recipient. Blood 1998; 92:3515- individual but others as well (A.L.Pay, unpublished data). 20. A mismatch for HLA-A at the allelic level has been 11. Moretta A, Moretta L. HLA Class I specific inhibitory shown to be a risk factor for acute GvHD and death receptors. Curr Opin Immunol 1997; 9:694-701. Educational Session 5 Haematologica 1999; 84:(EHA-4 educational book):46-49 Chairman: A. Bacigalupo Unrelated donor transplants

Cytomegalovirus infection following stem cell transplantation HERMANN EINSELE, HOLGER HEBART Medizinische Klinik und Poliklinik, University Hospital, Tübingen, Germany

ytomegalovirus (CMV) infection remains a lineage cells are the source of latent CMV, which major cause of morbidity and mortality after might reactivate upon allogeneic stimulation. Laten- Callogeneic stem cell transplantation. Due to cy-associated CMV transcripts were found in CD33+ the broad application of antiviral prophylaxis and subpopulations co-expressing myeloid and dendritic pre-emptive therapy, a decrease in early onset and a cell markers. Co-culture of latently infected CD33+ subsequent increase in late onset CMV disease has cells with permissive human fibroblasts showed reac- been observed. Transplantation from unrelated tivation of infectious virus after 3-4 weeks of culture. donors or related donors mismatched for one or CMV has been reported to cause myelosuppres- more HLA class I or II alleles is clearly associated with sion after transplant. A significant reduction in the a prolonged and more profound deficiency of CD3+, number of haematopoietic progenitor cells was CD4+ and CD8+ T cell-populations when compared reported, and an association of particular CMV geno- with recipients of a BMT from a related donor. Thus, types with death due to myelosuppression described. these patients are at increased risk of early and late CMV has evolved diverse mechanisms for evading onset CMV disease, especially when receiving T-cell host cellular immunity: CTL lysis is inhibited by mul- depleted grafts. Sensitive diagnostic assays using tiple glycoproteins which inhibit MHC class I expres- nucleic acid amplification and hybridisation tech- sion, and US6 which inhibits TAP-peptide-complex niques have been commercialised and will allow stan- translocation, the CMV encoded class I homologue dardisation of CMV diagnosis in antiviral drug trials. UL18 blocks NK cell lysis, and CMV also inhibits Quantification of the viral load will be increasingly MHC class II expression via compromise of the considered for initiation and, in patients with persis- JAK/STAT pathway. tence of high viral titres despite antiviral therapy, The predominance of CMV disease in patients with screening for antiviral drug resistance. Apart from impaired cellular immunity indicates the pivotal role ganciclovir, clinical data are emerging that foscarnet of cell-mediated immunosurveillance. Patients lack- can be given safely even after allogeneic SCT. Addi- ing CMV-specific CD8+ and CD4+ T cells are at an tional drugs such as lobucavir and cidofovir have increased risk of developing CMV disease upon CMV been used for specific indications. Furthermore, post-transplant reactivation.1 immunotherapeutic approaches are being increas- Transplantation from unrelated donors or related ingly evaluated for the control of CMV infection fol- donors mismatched for one or more HLA class I or II lowing allogeneic stem cell transplantation. alleles is clearly associated with a prolonged and + + + Pathophysiology of cytomegalovirus more profound deficiency of CD3 , CD4 and CD8 T cell-populations when compared to a BMT from a infection related donor. Thus, these patients are at increased After primary infection, cytomegalovirus (CMV) risk of early and late onset CMV disease, especially establishes latency in the host characterised by the when receiving T-cell depleted grafts. Consequently, persistence of viral genomes without production of fatal opportunistic infections after successful engraft- infectious virus. Despite the fact that latent CMV is ment occur in 12-28% of unrelated transplant recip- obviously transmitted through transfusion of blood ients compared to in only 4-15% after HLA-matched products, bone marrow grafts and solid organs, cells sibling BMT. harbouring latent virus were not identified until recently. Reactivation of CMV in long-term cultures of Risk factors and manifestations of allogeneically stimulated adherent monocyte-derived CMV disease macrophages was demonstrated, with infectious virus detected 26-61 days after allogeneic stimulation. The In spite of recent developments in diagnosis and cells harbouring the virus were found to be CD14 and treatment, CMV infection remains one of the most CD83 positive, thus providing evidence that myeloid important opportunistic infections in recipients of an allogeneic stem cell transplant (SCT). The incidence of CMV infection increases with severity and dura- Correspondence: Hermann Einsele, M.D. Med. Klinik u. Poliklinik, tion of immunosuppression and approaches 70% in Abt.II, Otfried-Müller Str.10, D-72076 Tübingen, Germany. Tel. interna- tional +49-7071-2982808 – Fax. international +49-7071-293179 – allogeneic SCT recipients being either CMV-seropos- E-mail: [email protected] itive and/or receiving a transplant from a CMV-sero- Session #5 – Unrelated donor transplants 47 positive donor. CMV disease is associated with a high symptoms, and were highest in CMV seropositive mortality (70%) in this patient population, even when recipients transplanted from seropositive donors.4 A treated with antiviral chemotherapy (ganciclovir, fos- high viral load early after allogeneic SCT as detected carnet) plus CMV hyperimmunoglobulin.2 by a PCR plate assay from plasma might be an addi- The development of sensitive screening assays that tional predictor of late onset CMV disease. allow early pre-emptive initiation and monitoring of To increase the value of nucleic acid amplification antiviral therapy, as well as antiviral chemoprophy- techniques for positively predicting the occurrence of laxis have helped to reduce early post-transplant CMV disease, materials other than whole blood or CMV-associated mortality significantly. However, buffy coat, such as plasma or serum, were used. A patients receiving a transplant from a matched unre- study showed plasma PCR to be less sensitive than lated donor and T-cell depleted grafts are still at a PCR on whole blood and antigenaemia. In a limited high risk of developing early onset CMV disease. T-cell number of patients the results of a commercially avail- depletion by 2-3 logs as achieved by CD34+ selection able PCR assay on plasma samples were comparable using immunoabsorption techniques is associated to those of a pp65 antigenaemia assay in a cohort of with an increased risk of CMV reactivation, but not immunocompromised patients including recipients CMV disease, at least in patients receiving a stem cell of an allogeneic SCT. graft from an HLA-identical sibling donor. To improve the positive predictive value of PCR The epidemiology of CMV with the introduction of assays further, reverse transcriptase (RT) PCR assays pre-emptive and prophylactic antiviral strategies has for detection of immediate-early and late viral mRNAs changed markedly, with increasing reports of CMV dis- have been evaluated in a limited number of studies. ease occurring late (>100 days post-transplant), but However, since thse techniques are more time-con- less frequently during the first 100 days post-trans- suming and less sensitive, despite having a higher pos- plant. In a minority of CMV-seropositive patients, CMV itive predictive value than amplification of DNA, fur- disease occurs prior to engraftment and, probably due ther studies are warranted before these assays can be to the frequent presence of co-pathogens, has been recommended for routine screening of clinical sam- associated with a very high mortality. ples. Recently, an assay based on the amplification of Thus, late onset CMV disease occurring at least 100 late viral pp67 mRNA was commercialised and a first days post-transplant has now become the leading clinical study has demonstrated its potential for ear- CMV-related complication after allogeneic SCT. In two ly diagnosis of CMV infection and monitoring of studies, late onset CMV disease was associated with antiviral therapy. chronic graft-versus-host disease and long-term ganci- In conclusion, major progress has been achieved clovir application. Clinical manifestations of CMV concerning quantification of the viral load in the infection that had been only rarely observed early post- blood, which might help to improve monitoring of transplant, e.g. CMV retinitis, have been described in antiviral therapy. The availability of commercialised patients developing late onset CMV disease. sensitive assays will allow standardisation of CMV diagnosis in upcoming multicentre antiviral drug trials. Diagnosis of CMV infection Qualitative PCR and antigenaemia assays have been Prophylaxis of CMV infection used in recent years to screen patients undergoing To prevent the acquisition of exogenous virus in allogeneic SCT. Due to the low incidence of CMV dis- seronegative patients receiving a transplant from a ease in autograft recipients, monitoring by sensitive seronegative donor is of major importance. Primary assays should be limited to patients after allogeneic infection can be prevented efficiently by transfusion of SCT and possibly to autograft recipients at very high blood products from CMV-seronegative donors. risk e.g. CMV-seropositive patients receiving mye- Alternatively, leukocyte-depleted blood products have loablative conditioning regimens and manipulated been shown to be safe, if a leukocyte depletion of at grafts (4-hydroperoxycyclophosphamide purged or least three logs can be provided.5 heavily T-cell depleted autografts).1,3 However, even Antiviral prophylaxis to suppress reactivation of after allogeneic SCT, the positive predictive value of CMV if either patient and/or donor is seropositive is a culture, PCR and antigenaemia assays is low. As very efficient approach to prevent CMV infection and quantitative assessment of PCR reactions by the use disease. High-dose intravenous acyclovir at a dosage of internal standards and PCR-ELISA assays has of 500 mg/m2 three times a day until day +30 slightly become available, several studies have addressed the reduced the probability of and delayed the onset of question of whether there is a correlation between CMV infection. Most importantly, survival could be viral load in the blood and subsequent development significantly improved by high dose intravenous com- of CMV disease after SCT. In a longitudinal analysis pared to low dose oral acyclovir. Although prolonga- of 110 patients following SCT, the peak viral load in tion of acyclovir prophylaxis beyond day +30 did not patients with CMV disease was significantly higher further reduce the risk of developing CMV infection, than that in patients with asymptomatic CMV infec- survival seemed to be further improved.6 Valaciclovir, tion, peak values preceded the onset of CMV-related a valine ester of acyclovir that is rapidly metabolised 48 H. Einsele et al. to produce high acyclovir levels, applied prophylacti- oped CMV disease before day 100 compared to 2.7% cally was found to reduce CMV disease in PCR posi- in the ganciclovir group. Low-grade antigenaemia tive patients with advanced AIDS. progressed to CMV disease only in patients with Prophylaxis with intravenous ganciclovir in allo- grades III and IV acute graft-versus-host disease. The geneic BMT recipients at risk, given from the time of incidence of CMV disease until day 180, CMV-relat- engraftment until day +100 post-transplant, has been ed death, transplant survival, and neutropenia were assessed in 2 studies. In both studies, a significant not significantly different between the 2 groups at reduction of the incidence of CMV infection and in day 180 after transplantation. Ganciclovir at engraft- one study also of CMV disease was demonstrated. ment was associated with a higher rate of early inva- Moreover, in both studies survival of patients receiv- sive fungal infection and late CMV disease resulting ing ganciclovir was comparable to the survival of in a similar survival rate in both groups. patients in the high-dose acyclovir trial. However, gan- Foscarnet (PFA), a pyrophosphate analogue with ciclovir prophylaxis was associated with significant in vitro activity against all known human herpes virus- toxicity, specifically therapy-related neutropenia and es, does not cause myelosuppression but is associat- secondary bacterial infections. Hyperbilirubinaemia ed with significant nephrotoxicity. In earlier studies, > 6 mg/dL before day 20, low marrow cellularity at foscarnet was found to prevent CMV disease in recip- day 21, and a serum creatinine > 2 mg/dL after day ients of an allogeneic SCT, and using adequate pre- 20, but not CMV infection or disease, were reported hydration even a combination of ganciclovir and fos- as major risk factors for ganciclovir-related neutrope- carnet could be safely administered to high risk nia. Oral ganciclovir is currently being evaluated as a patients.9 For CMV prophylaxis following allogeneic new antiviral modality for CMV prophylaxis. SCT, foscarnet (60 mg/kg/day) was administered with low toxicity to patients who could not receive Antiviral therapy for documented CMV ganciclovir for delayed engraftment or ganciclovir- infection (pre-emptive therapy) induced neutropenia. However, in 15% of patients As the outcome of patients treated for overt CMV CMV was detected while they were receiving prophy- disease is poor, pre-emptive or early antiviral therapy laxis and 2 (5%) died of CMV disease. The results of has been introduced as a therapeutic strategy. In two the phase III EBMT protocol comparing pre-emptive large studies early treatment with ganciclovir based therapy with ganciclovir versus foscarnet based on on a positive virus culture assay, was evaluated. Gan- antigenaemia or PCR assay results will further eluci- ciclovir was either administered at the time of first date the role of foscarnet in treating patients under- CMV excretion from blood, urine or throat washing going allogeneic stem cell transplantation with doc- samples or from a bronchoalveolar lavage sample tak- umented CMV infection.10-12 Lobucavir, a desoxygua- en at day 35 post-transplant. In both studies, pre- nine nucleoside analogue with broad-spectrum antivi- emptive antiviral therapy reduced CMV disease and ral activity, has been shown to be active against CMV. transplant-related mortality. However, 12-13% of As lobucavir phosphorylation can occur in the patients presented with CMV disease before or coin- absence of viral phosphorylases, lobucavir may be cident with CMV excretion, leading to a 10% CMV- another alternative drug to treat ganciclovir-resistant related mortality. strains of CMV. Thus, more sensitive techniques were used to Cidofovir has been shown to be an effective and, detect CMV infection prior to the onset of CMV dis- due to its long half-life allowing a weekly administra- ease. CMV was detected up to 20 days earlier by PCR tion, convenient systemic treatment for CMV retinitis. than by the culture technique. Comparison of PCR Promising results with the administration of this drug and culture-based pre-emptive antiviral therapy also in marrow transplant recipients were reported at demonstrated that PCR screening allowed antiviral the EBMT-Congress in Hamburg. Antiviral resistance therapy to be started significantly earlier than culture has been reported more frequently, and prolonged results.7 Additionally, stopping and withholding administration of antiviral compounds has been antiviral therapy was found to be safe in the PCR- shown to predispose for selection of resistant viral negative patients. The incidences of CMV disease and strains. Thus, molecular screening for mutations in CMV-related mortality were decreased, and the dura- the UL97 and DNA polymerase gene might be indi- tion of antiviral therapy significantly shorter in the cated in patients with persistence of high viral load in PCR-monitored group leading to a reduced incidence the blood despite antiviral therapy. Moreover, new of ganciclovir-related side effects such as neutrope- antiviral compounds are needed to increase the ther- nia and non-viral infections. Overall survivals at day apeutic armamentarium against CMV. 100 and 180 post-transplant were significantly improved in the PCR-monitored group. Boeckh et al.8 Immunotherapy of CMV infection and compared antigenaemia guided antiviral therapy and disease ganciclovir prophylaxis in 226 CMV-seropositive allo- Transplantation from unrelated donors or related geneic marrow transplant recipients. Fourteen per- donors mismatched for one or more HLA class I or II cent in the antigenaemia-ganciclovir group devel- alleles is clearly associated with a prolonged and more Session #5 – Unrelated donor transplants 49 profound deficiency of CD3+, CD4+ and CD8+ T cell- populations in recipients when compared to a trans- References plant from a related donor. Thus, these patients are 1. Ljungman P, Biron P, Bosi A, et al. Cytomegalovirus at increased risk of early and late onset CMV disease, interstitial pneumonia in autologous bone marrow especially when receiving T-cell depleted grafts. Con- transplant recipients. Infectious Disease Working Party sequently, fatal opportunistic infections after suc- of the European Group for Bone Marrow Transplanta- cessful engraftment occur in 12-28% of unrelated tion. Bone Marrow Transplant 1994; 13:209-12. 2. Ljungman P, Engelhard D, Link H, et al. Treatment of transplant recipients compared with in only 4-15% interstitial pneumonitis due to cytomegalovirus with after HLA-matched sibling BMT. ganciclovir and intravenous immune globulin: experi- Adoptive immunotherapy with small numbers of ence of European Bone Marrow Transplant Group. unirradiated donor leukocytes was recently found to Clin Infect Dis 1992; 14:831-5. be associated with rapid restoration of CD3+, CD4+ 3. Hebart H, Schroeder A, Loeffler J. CMV-monitoring + by PCR of patients undergoing autologous bone mar- and CD8 T-cell numbers, antigen-specific T-cell row and peripheral blood progenitor cell transplan- responses, and resolution of CMV- and EBV-associ- tation. J Infect Dis 1997; 172:939-43. ated disease after unrelated T-cell depleted BMT. 4. Gor D, Sabin D, Prentice HG, et al. Longitudinal The prophylactic transfer of CMV-specific CD8+ T- analysis in cytomegalovirus load in bone marrow cell clones early post-transplant has been shown to transplant patients: relationship between peak virus provide protection against CMV disease. However, load, donor/recipient serostatus, GvHD and disease. Bone Marrow Transplant 1998; 21:597-605. long-term in vivo persistence of transferred CTLs 5. Bowden RA, Slichter SJ, Sayers M, et al. A comparison required the development of an endogenous CMV- of filtered leukocyte-reduced and cytomegalovirus specific T helper response. A phase II study is currently (CMV) seronegative blood products for the preven- evaluating co-administration of CD4+CMV-specific tion of transfusion-associated CMV infection after Th clones and CD8+ CMV specific CTL clones. marrow transplant. Blood 1995; 86:3599-603. 6. Prentice HG, Gluckman E, Powles RL, et al. Impact of Alternatively, peptide vaccines are a possible long-term acyclovir on cytomegalovirus infection and approach. Diamond et al. mapped an HLA*0201 survival after allogeneic bone marrow transplantation. restricted nonamer peptide out of the immuno- European Acyclovir for CMV Prophylaxis Study Group. dominant pp65 protein.13 This peptide was used to Lancet 1994; 343:749-53. amplify a memory CTL response in CMV-seropositive 7. Einsele H, Ehninger G, Hebart H, et al. Polymerase chain reaction monitoring reduces the incidence of individuals. Lipid modification of the amino terminus cytomegalovirus disease and the duration and side of the nonamer peptide resulted in its ability to simu- effects of antiviral therapy after bone marrow trans- late peptide-specific primary T-cell responses without plantation. Blood 1995; 86:2815-20. the use of adjuvant in a transgenic mouse model. 8. Boeckh M, Gooley TA, Myerson D, et al. Cytomegalo- virus pp65 antigenemia-guided early treatment with ganciclovir versus ganciclovir at engraftment after allo- Conclusions geneic marrow transplantation: a randomized dou- Pre-emptive as well as prophylactic antiviral treat- ble-blind study. Blood 1996; 88: 4063-71 ment strategies have helped to reduce CMV-associat- 9. Bacigalupo A, Bregante S, Tedone E, et al. Combined ed mortality significantly in recipients of an allogene- Foscarnet-Ganciclovir treatment for cytomegalovirus ic BMT. Pre-emptive antiviral therapy, compared to infections after allogeneic hemopoietic stem cell trans- plantation. Bone Marrow Transplant 1996; 62:376- antiviral prophylaxis, has the advantage of being able 80. to be used in patients stratified according to individ- 10. Ljungman P, Aschan J, Lewensohn-Fuchs I, et al. ual risk factors (active CMV infection, viral load) and Results of different strategies for reducing cytomegalo- thus helps to reduce the number of patients treated virus-associated mortality in allogeneic stem cell trans- and also the duration of antiviral therapy, which plant recipients. Transplantation 1998; 66:1330-4. 11. Reusser P, Attenhofer R, Hebart H, Helg C, Chapuis might have important implications concerning side B, Einsele H. Cytomegalovirus-specific T-cell immuni- effects and emergence of antiviral resistance. In the ty in recipients of autologous peripheral stem cell or future, immunotherapeutic strategies, e. g. donor vac- bone marrow transplants. Blood 1997; 89:3873-9. cination and transfer of donor-derived CMV-specific 12. Walter EA, Greenberg PD, Gilbert MJ, et al. Reconsti- T-cells to patients in a prophylactic or therapeutic set- tution of cellular immunity against cytomegalovirus in recipients of allogeneic bone marrow by transfer of T- ting will be more broadly applied and evaluated for cell clones from the donor. N Engl J Med 1995; 333: feasibility and efficacy following allogeneic stem cell 1038-44. transplantation. 13. Diamond DJ, York Y, Sun J-Y, Wright CL, Forman SJ. Development of a candidate HLA A*0201 restricted Acknowledgments peptide-based vaccine against human cytomegalo- Supported by SFB 510, project B3. virus infection. Blood 1997, 90:1751-67. Educational Session 5 Haematologica 1999; 84:(EHA-4 educational book):50-52 Chairman: A. Bacigalupo Unrelated donor transplants

Improved results in marrow transplantation from unrelated donors ANDREA BACIGALUPO,* TERESA LAMPARELLI,* MARIO BARBANTI,° NICOLETTA SACCHI,° GIOVANNI BATTISTA FERRARA,# SANDRA POZZI,* STEFANIA BREGANTE,* NICOLA MORDINI,* VITO VITALE,@ MARIA TERESA VAN LINT* FOR THE GENOA BMT GROUP *Divisione Ematologia II, Ospedale San Martino, Genova; °Italian Bone Marrow Transplant Registry (IBMDR), Ospedale Galliera, #Servizio Immunogenetica, @Servizio Radioterapia, Istituto Tumori, Genoa, Italy

Increasing numbers of unrelated BMT UD-BMT and HLA disparity Bone marrow transplantation (BMT) from unrelat- Major improvement has been achieved by molecu- ed donors (UD) is being increasingly used both in lar typing for HLA class II antigens: in an EBMT review Europe and in the United States: the European Group of 366 patients, 210 were matched for DRB1. Their for Blood and Marrow Transplantation (EBMT) reported TRM was 49%; 31 were DRB1 mismatched and their 953 UD-BMT in 1996 and an increase in the propor- TRM was 79% (p=0.00002).5 More recently HLA class tion of UD-BMT: 1% in 1983, 8% in 1990 (n=181), I antigens are being matched by molecular typing and 14% in 1993 (n=456), 22% in 1996 (n=953) and 24% it is now possible to assess the influence of mis- in 1997 (n=1150).1 In this Unit in Genoa the propor- matching at class I or class II:6 tion has increased from 4% in 1992 to 22% in 1996 • risk of aGvHD is highest with class II mismatching; and to 32% in 1998 (unpublished data). The Nation- • mismatching at one single allele for class I or class al Marrow Donor Program Data (NMDP) reports 82 II does not affect survival; transplants in 1987, 650 in 1993 and 1248 in 1997. • mortality is increased for mismatching at more than We can calculate that in 1998 there were approxi- one class I allele and by simultaneous mismatching mately 3000 UD-BMT world-wide, and that the num- for class I and class II alleles. ber is steadily increasing. In a recent analysis of the Italian Group for Bone Mar- row Transplantation (GITMO) TRM was predicted by UD-BMT, immunodeficiency and HLA matching : event free survival was 55% for DRB1 transplant-related mortality matched pairs and 15% for DRB1 mismatched UD-BMT are associated with high rate of mortali- patients (p=0.0006).7 ty (TRM), higher than that seen in transplants from At present a minimum requirement for an UD-BMT HLA identical sibling grafts (SIB-BMT).2 This is in adults would be class I identity in serology (includ- thought to be due to greater disparity between donor ing split antigens) and class II identity in molecular biol- and recipient when compared to genetically matched ogy (DRB1). A more stringent requirement would be donors. Greater disparity results in profound and identity at the molecular level for class I and class II. prolonged immunodeficiency with or without acute The report from the Seattle group of a comparable graft-versus-host disease (GvHD). In our Unit the mortality in the presence of one single antigen mis- median level of CD4 counts on day 50 post-BMT is match should be confirmed by other groups. 133/µL for SIB-BMT and 36/µL for UD-BMT An additional approach is matching for ancestral (p<0.0001); the CD4 count on day 100 is 114 vs haplotypes: a pilot study is ongoing within the GIT- 49/µL (p<0.01) (unpublished data). This results in a MO for UD-BMT in thalassaemia, with encouraging greater risk of opportunistic infections both early and results (La Nasa, personal communication). However, this late after transplantation.3 Cytomegalovirus (CMV) implies knowledge of ancestral haplotypes and a geo- infections are of particular concern and CMV-specif- graphical environment which donor and recipients ic T cell proliferation is significantly less likely to occur are likely to share (Sardinia is one example). in recipients of UD-BMT, who are therefore at greater The relevance of HLA-C mismatching has also been risk of developing CMV-disease.4 studied: an increased risk of graft failure has been Acute GvHD at day 100 is reported to be present reported in HLA-C mismatched, otherwise matched in 52% vs 42% (p=0.009) of UD-BMT and SIB-BMT pairs.6 These data need to be confirmed: one should recipients, respectively, and survival for patients with remember that approximately 50% of HLA-A, B, DRB1 early disease is 40% for UD-BMT vs 64% for SIB-BMT matched pairs have disparities at the HLA-C locus. (p=0.0001).2 In our Unit the risk of aGvHD grades III- IV is 25% in UD-BMT vs 5% in SIB-BMT (p<0.00001). In vitro functional tests for donor selection The mixed lymphocyte reaction (MLR) is no longer Correspondence: Dr A. Bacigalupo, Divisione Ematologia 2, Ospedale considered a reliable test for identifying unrelated San Martino, 16132 Genova, Italy. donors, and the test is not required to have access to Session #5 – Unrelated donor transplants 51 donors. Cytotoxic T cell precursors (CTL-P) have been • CyA 2 mg/kg i.v. is given day –1 day +20; then orally; extensively studied in some centres, and a high fre- • methotrexate (MTX) is given as a short course (day quency is known to be associated with mismatching +1, +3, +6, +11); at class I. Helper T cell precursors (HTL-P) have also • early treatment of acute GvHD; been studied, but conclusive results on their predictive • patients are treated at first signs of acute GvHD with value in GvHD is lacking. The same can be said of the prednisolone 2 mg/kg: non-responders are ran- mixed skin-lymphocyte reaction. domised within day +5 to receive 5 mg/kg of pred- nisolone or 5 mg/kg of prednisolone+ ATG low dose How transplant related mortality can (1.25 mg/kg on alternate days ϫ 5 doses); be reduced • aggressive CMV prophylaxis and treatment; TRM remains high in UD-BMT and the encouraging • all patients receive foscarnet 30 mg/kg ϫ 2 day –7 results achieved in chronic myeloid leukaemia (CML) day +30; then 90 mg/kg/day day +31 day +100 (5 patients grafted within one year from diagnosis (74%)8 days/week); do not necessarily apply to all patients, especially • pre-emptive treatment of CMV-antigenaemia is giv- those with advanced disease, with a long interval from en by adding ganciclovir (10 mg/kg/day ϫ 15 days) diagnosis to BMT and the elderly. If we want to reduce to foscarnet. transplant mortality we need to: 1. improve immunologic recovery; Results of the transplant programme in 2. improve haemopoietic recovery; Genoa 3. reduce GvHD. We have transplanted 60 patients from DRB1 To achieve these results, the transplant can be mod- matched unrelated donors in this programme. Twelve ified in several ways, and some areas of current inves- of the patients had acute leukaemia and 48 had tigation are outlined: chronic myeloid leukaemia (CML); 34 patients were • selection of donor and patient; in early phase. Their median age was 29 years (range • modified conditioning regimens; 18-49). The median interval between diagnosis and • manipulation of the graft; BMT was 1,248 days (range 262-3,064). • high cell dose; • low lymphocyte content; Actuarial transplant related mortality • expansion of marrow stromal cells; The overall TRM is 28% (Figure 1), being 20% for • pre and post-BMT GvHD prophylaxis; patients in early phase of the disease (n=34). There is • early treatment of acute GvHD; an effect of age (TRM 20% vs TRM 35%; p=0.2) at a • aggressive CMV prophylaxis and treatment. cut-off age of 35, and an effect of diagnosis-BMT interval (TRM 20% vs TRM 27%; p=0.5). UD transplant programme in Genoa Survival We have developed a transplant regimen which can The actuarial 5 year survival is 69% for all patients be summarised as follows: (n=60). For patients with CML (n=48), including 27 selection of donor and patient • in first chronic phase and 21 in accelerated phase HLA-A, HLA-B matched by serology and molecular • CML, the actuarial survival is 75%. biology sequence-specific oligonucleotide probes (SSOP, sequencing); Quality of graft function • DRB1 matched: high resolution molecular biology We analysed the quality of graft function as repre- (SSOP, sequencing); sented by platelet counts (ϫ109/L) on day +30, +50, • DRB3, 4, 5 matched (not prerequisite); +100: for SIB transplants platelet counts (ϫ109/L) • DQA, DQB, DPA, DPB : mismatch accepted (espe- were 122 (range 34-385), 113 (48-283), 99 (31-291); cially for advanced phase); • modified conditioning regimens; 100 HLA A, B, DRB1; DRB3 = MUD BMT: 60 pts • cyclosporin A from day –7 (CyA-7); 90 • CY 120mg/kg +TBI 9-9 - 12 Gy; 80 [CY= cyclophosphamide; TBI= fractionated total body irra- 70 ϫ ϫ ϫ diation (2 Gy 2 3 for 38 patients and 3.3 Gy 3 for 22 60 patients]; 50 • manipulation of the graft; 40 • in this programme the graft is given unmanipulated, % of patients 30 but care is taken to have the highest possible cell 20 dose; 10 • pre and post-BMT GvHD prophylaxis; 0 • rabbit ATG 4 days is being compared to no ATG in 0 6 12 18 24 30 36 a prospective ongoing randomised trial in adult Months from BMT patients, with the collaboration of the Italian Coop- erative BMT Group (GITMO); Figure 1. 52 A. Bacigalupo et al. for UD-BMT they were 38 (2-257), 50 (10-160), 45 Acknowledgements (10-247). The difference (Rank Sum test) was very This work was supported by the Associazione Ricerca significant on day +30 (p=0.001), on day +50 Trapianto Midollo Osseo (A.RI.T.M.O.), Genoa and Associ- (p<0.0001) and on day +100 (p=0.02) suggesting azione Italiana Ricerca contro il Cancro (A.I.R.C.), Milan, that graft function after UD-BMT is significantly poor- Italy. The valuable work of Marina Daneri as transplant co- er than after SIB-BMT. ordinator is acknowledged, together with the work of the nurs- Graft-versus-host disease ing staff of the transplant Unit. The work of International Reg- istries for donor search, together with Bone Marrow Donors Acute GvHD (aGvHD) occurred at a median inter- World Wide (BMDW) and European Marrow Donor Infor- val from BMT of 15 days (range 7-45) (and day 14 in mation System (EMDIS) has been the basis for these results. a control group of SIB-BMT). aGvHD was scored as grade 0-I, grade II, grade III-IV in 50%, 36% and 12% of SIB-BMTs, respectively, vs 25%, 43%, 31% in UD- References BMTs (p<0.0001). Chronic GvHD was scored as extensive in 23% of SIB-BMT and 36% of UD-BMT 1. Gratwohl A, Passweg J, Baldomero H, Hermans J, for (p=0.005). the European Group for Blood and Marrow Trans- plantation (EBMT). Blood and marrow transplanta- CMV infections tion activity in Europe in 1996. Bone Marrow Trans- The actuarial risk of CMV antigenaemia at 1 year plant 1998; 22:227-40. was 61% for SIB-BMT with high dose acyclovir pro- 2. Hows J, Bradley BA, Gore S, Downie T, Howard M, Gluckman E. Prospective evaluation of unrelated donor phylaxis and 69% for UD-BMT with foscarnet pro- bone marrow transplantation. The International Mar- phylaxis. row Unrelated Search and Transplant (IMUST) Study. Bone Marrow Transplant 1993; 4: 371-80. Relapse 3. Ochs L, Shu XO, Miller J, et al. Late infections after allo- The actuarial risk of relapse was 15%. geneic bone marrow transplantations: comparison of incidence in related and unrelated donor transplant Conclusions recipient: Blood 1995; 86:3979-86. 4. Krause H, Hebart H, Jahm G, Müller CA, Einsele H. If we compare SIB-BMT with UD-BMT, the latter Screening for CMV-specific T cell proliferation to iden- exhibit: tify patients at risk of developing late onset CMV dis- • slower haemopoietic recovery; ease. Bone Marrow Transplant 1997; 19:1111-6. • slower immunologic recovery; 5. Devergie A, Apperley JF, Labopin M, et al. for the higher incidences of acute and chronic GvHD; Chronic Leukaemia Working Party of the European • Group for Blood and Marrow Transplantation: Euro- • higher rate of transplant mortality. pean results of matched unrelated donor bone marrow transplantation for chronic myeloid leukaemia. Impact The 2,942 patients reported to the EBMT had an of HLA class II matching. Bone Marrow Transplant overall 5 year actuarial TRM of 45%; it was 33% for 1997; 20:11-9. 1,506 HLA identical siblings and 41% for 150 unre- 6. Petersdorf EW, Gooley TA, Anasetti C, et al. Optimis- lated transplants.9 Similar results were reported by ing outcome after unrelated marrow transplantation the National Marrow Donor Program (NMDP):10 survival by comprehensive matching of HLA class I and II alle- les in the donor and recipient. Blood 1998; 92:3515- for chronic phase CML patients receiving MUD grafts 20. was 37% and 19% for patients with more advanced 7. Dini G, Lamparelli T, Rondelli R, et al. Unrelated donor phase CML: an update on a larger number of patients marrow transplantation for chronic myelogenous suggests a survival of 52%.11 The Seattle team recent- leukaemia. Br J Haematol 1998; 102:544-52. ly reported an 80% survival for 30 1st chronic phase 8. Hansen JA, Gooley TA, Martin PJ, et al. Bone marrow transplants from unrelated donors for patients with CML patients grafted from a matched unrelated chronic myeloid leukaemia. N Engl J Med 1998; 338: donor within 1 year from diagnosis, although results 962-8. were less favourable for patients grafted later on in the 9. Gratwohl A, Hermans J, for the Working Party Chron- course of their disease.12 ic Leukaemia of the European Group for Blood and Marrow Transplantation (EBMT). Allogeneic bone TRM has decreased over the past decades by a fac- marrow transplantation for chronic myeloid leukemia. tor of 2, but still remains a major concern for physi- Bone Marrow Transplant; 17(suppl 3):S7-9. cians advising a patient whether he should or should 10. Kernan NA, Bartsch G, Ash RC, et al. Analysis of 462 not undergo an allogeneic BMT. Improved survival unrelated marrow transplants facilitated by the Nation- with the use of alternative haematologic therapy is al Marrow Donor Program (NMDP) for treatment of acquired and congenital disorders of the lymphohe- also a reality. matopoietic system and congenital metabolic disor- We believe that several combined strategies can be ders. N Engl J Med 1993; 328:592-602. applied to improve the outcome of UD-BMT: selec- 11. Anasetti C, Howe C, Petersdorf EW, Martin PJ, Hansen tion of donor/recipient pairs seems to be crucial, and JA. Marrow transplants from HLA matched unrelated lots of work is going on in that direction in laborato- donors: an NMDP update and the Seattle experience. Bone Marrow Transplant 1994; 13:693-5. ries. Clinical protocols with the use of less intensive 12. Clift R, Storb R. Marrow transplantation in CML: the conditioning, high cell dose, improved management Seattle experience. Bone Marrow Transplant 1996; of GvHD and CMV infections are also being explored. 17(suppl. 3):1-3. Educational Session 6 Haematologica 1999; 84:(EHA-4 educational book):53-56 Chairman: A.V. Hoffbrand Folic acid and thrombosis

Homocysteine, folate enzymes and neural tube defects ANNE M. MOLLOY,* DONALD G. WEIR,* JOHN M. SCOTT° *Departments of Clinical Medicine and °Biochemistry, Trinity College Dublin, Ireland

mong the most remarkable nutritional break- Inadequate folate or vitamin B12 to overcome throughs of the past decade was the conclu- a metabolic deficiency Asive discovery that periconceptional supple- A number of studies have examined folate and vit- mentation with the vitamin folic acid could prevent amin B12 status in relation to NTDs. These include the up to 70% of neural tube defects (NTDs). This find- evaluation of (a) levels in maternal blood and/or in ing, the result of major double-blind randomised amniotic fluid during NTD affected pregnancies, (b) clinical trials carried out in the UK and Hungary, was levels in the blood of non-pregnant women who had the culmination of more than ten years of research previously had an NTD affected pregnancy and (c) and speculation.1,2 Seven years have elapsed since levels in blood taken from NTD affected (spina bifi- then and in spite of an explosion of research in the da) patients and their families. The general consen- area, no clear-cut folate related mechanism has sus is that NTD affected pregnancies tend to be asso- emerged and there is still widespread debate and ciated with lower blood folate and possibly also low- uncertainty on the choice of strategy to adopt in er vitamin B12 levels, particularly during the first order to substantially reduce the incidence of NTDs. trimester.4 Several reports have also demonstrated This article reviews recent work in the area from low amniotic vitamin B12 or transcobalamin levels, the following two perspectives: 1. what is known indicating abnormal metabolism in the foetal com- about the underlying folate responsive mechanism partment. and 2) what folic acid/folate intake is required to pre- Increased homocysteine levels vent the occurrence of NTDs. For a number of reasons, much of the interest in The underlying folate responsive the mechanism of NTDs has involved the role that mechanism homocysteine might play. Firstly, while homocysteine is an essential intermediate in folate dependent path- The first question is whether folic acid acts by pre- ways (Figure 1), it is toxic both in vivo and in vitro. Intri- venting dietary deficiency or by treating a metabolic cate control mechanisms within the cell are designed block. Current opinion is that it is unlikely that NTDs to maintain low levels. Abnormally high homocys- result from maternal folate deficiency. Several stud- teine levels within the embryo could have a primary ies have confirmed that NTD affected mothers do effect on the neurulation process via a toxic effect on not have clinically low blood folate status although membrane function, protein structure, etc. Alterna- their blood folate concentrations tend to be lower tively, high intracellular homocysteine could act in a than mothers with non-affected pregnancies, which secondary manner by causing the level of S-adeno- is consistent with a folate related aetiology.3 More- sylhomocysteine within the cell to rise. Since S-adeno- over, folate malabsorption has not been demon- sylhomocysteine is a potent inhibitor of almost all strated in NTD affected mothers. Ongoing research, methyltransferase reactions, this would have the therefore, has focused on the developing embryo and effect of inhibiting the production of essential methy- the possibility of genetic variants in folate dependent lated products such as methylated proteins, DNA, enzymes giving rise to an inadequate supply of essen- etc (Figure 1). Finally, an alteration of homocysteine tial metabolites or excessive levels of toxic products. metabolism associated with NTDs may merely point Figure 1 illustrates pathways of folate metabolism in to an undefined malfunction in a folate related event. mammalian cells, which may be important during Apart from congenital deficiencies in specific homo- the neurulation process in embryonic cells. There are cysteine metabolising enzymes, the most common numerous ways in which these pathways may be dis- cause of plasma homocysteine accumulation is an turbed which could potentially result in abnormal inadequate supply of folate or vitamin B12. closure of the neural tube. To date, the evidence that elevated homocysteine is an important factor in the aetiology of NTDs is quite weak. Mildly elevated plasma homocysteine concen- trations have been detected in maternal blood during NTD affected pregnancies even after adjusting for Correspondence: Dr. Anne Molloy, Department of Biochemistry, Trinity 5 College Dublin, Dublin 2, Ireland. Tel: International +353-1-6081616 folate and vitamin B12 status. Amniotic fluid homo- – Fax: International +353-1-6772400 – E-mail: [email protected] cysteine is also moderately raised in NTD pregnan- 54 A. M. Molloy et al.

Figure 1. Pathways of folate metabolism.

cies.6 Some animal models have been used to examine results in higher than normal homocysteine levels. the teratogenicity of homocysteine, but they give con- Thus, an attractive hypothesis for the underlying aeti- flicting results and are difficult to interpret. Neverthe- ology of NTDs has been either a direct or indirect less, there has been particular interest in determining involvement of this enzyme. The methionine synthase the influence of the three enzymes involved in the gene has recently been cloned and some mutational metabolism of homocysteine. These are methionine analyses carried out. One reasonably prevalent muta- synthase, cystathionine-␤-synthase (CBS) and 5,10 tion (D919G) was examined in 56 NTD cases, 69 case methylenetetrahydrofolate reductase (MTHFR). mothers and 364 controls but there was no evidence Methionine synthase of an association between this amino acid change Methionine synthase occupies a central role in and NTDs. While the result is disappointing, it is pos- folate metabolism (Figure 1). It is an essential inter- sible that other mutations in the methionine synthase mediate in the incorporation of plasma folate into gene may exist in populations with a high NTD preva- the cell because 5-methyltetrahydrofolate, the circu- lence. Very recently, the cDNA was cloned and lating form of the vitamin, must be demethylated via mapped for methionine synthase reductase, an methionine synthase before it can be polyglutamated enzyme which provides the reducing system necessary and retained by the cell. In addition, the reaction is to maintain a properly functioning methionine syn- coupled with the recycling of homocysteine to thase. This enzyme provides an exciting new point of methionine, thus maintaining the supply of methyl interest for mutational analysis in relation to NTDs. groups through S-adenosylmethionine for the essen- Cystathionine-␤-synthase (CBS) tial methylation of DNA, proteins, etc. When cells are This vitamin B6 dependent enzyme is the first step folate replete these methyl groups are supplied to on the irreversible catabolic pathway for homocys- folate cofactors from serine or glycine and converted teine (Figure 1). Inborn errors of CBS cause massive to 5-methyltetrahydrofolate through MTHFR. The elevations of plasma homocysteine resulting in a vari- activity of the enzyme methionine synthase is the only ety of functional and clinical abnormalities. The fre- event known to be influenced by both folate and vit- quency of two relatively common mutations in the amin B12 status. Reduced activity of this enzyme also CBS gene was examined in NTD cases and controls in Session #6 – Folic acid and thrombosis 55 the Irish population. Neither the severely dysfunc- adopt in order to provide women of reproductive age tional G307S allele nor the 68 bp insertion/I278T with sufficient folic acid to prevent NTDs. In the ran- allele was present at increased frequency in the NTD domised controlled trials, doses of 4 mg folic acid per cases suggesting that this enzyme does not play a sig- day decreased the recurrence and 800 µg per day nificant role in the pathogenesis of NTDs. decreased the occurrence of NTDs. However, in earli- er non-randomised studies, doses of approximately 5,10 methylenetetrahydrofolate reductase 400 µg per day were effective. Thus, while recommen- The enzyme 5,10-methylenetetrahydrofolate reduc- dations for the prevention of NTD recurrence were tase (MTHFR) is also in a pivotal position in relation straightforward (periconceptional supplementation to folate and homocysteine metabolism. It irreversibly with 4 or 5 mg folic acid per day) because these moth- converts 5,10-methylenetetrahydrofolate to 5-methyl- ers were already in a specific high-risk group, the tetrahydrofolate, thereby functioning as a control approach to adopt for the rest of the population was point wherein one-carbon units are channelled away more contentious. At present, Department of Health from the DNA synthesis cycle and into the methyla- advice in many countries is that women who are plan- tion cycle (Figure 1). The genetic profile of this enzyme ning a pregnancy should increase their folate con- has been more thoroughly explored than the other sumption by 400 µg per day. However, campaigns to two homocysteine metabolising enzymes. Between 5% increase public awareness on the benefits of folic acid and 15% of Caucasian populations are homozygous have been largely ignored. To make matters worse, an for a thermolabile variant of the enzyme (C677T) increased intake of 400 µg of food folate per day is which confers reduced enzyme activity and mildly much less effective in changing folate status than tak- increased plasma homocysteine concentrations when ing a supplement of 400 µg folic acid. After much folate status is low. Homozygosity for this variant has debate and concern that high folic acid consumption now been clearly shown to be a risk factor for spina may mask the symptoms of pernicious anaemia in the bifida, albeit accounting for no more than 12% of the elderly, the Food and Drug Administration (FDA) in the US population attributable risk.7 Analysis of the data sug- adopted a population strategy to increase the folic gests that case TT genotype (rather than maternal or acid intake of all inhabitants from 1st January 1998 by paternal) is of paramount importance in determin- mandating that cereal grain be fortified with 1.4 mg ing risk, although it is not known how the variant of folic acid per kg of grain. This strategy is expected functions to confer this risk in the developing embryo. to add approximately 100 µg per day of folic acid to Mothers who are homozygotes for the thermolabile the average diet, which is unlikely to mask pernicious (TT) variant are also at increased risk of giving birth anaemia but may be too low to prevent many NTDs. to an NTD affected infant. However, in the general A major difficulty has been the lack of information on population homozygosity for this variant is associat- the minimum amount of folic acid needed to prevent ed with reduced red cell folate status, a known risk NTDs. Daly et al.9 addressed this question by examin- factor for NTDs. Thus, it seems likely that the enzyme ing the relationship between NTD risk and early preg- may act merely as a cause of low folate status in these nancy maternal red cell folate levels. They found an women.8 A second variant (A1298C), recently dis- eight-fold higher risk among women whose levels were covered in the MTHFR gene, was not more frequent less than 150 µg/L compared with those above 400 in NTD cases. Nevertheless, it is possible that there are µg/L and concluded that that if the average red cell other mutations or combinations of mutations in this folate of the population could be increased such that gene which contribute to folate responsive NTD risk. all pregnant women had values above 400 µg/L, then the risk of NTDs could be reduced by as much as 60%. Other enzyme defects In a recent randomised trial Daly et al.10 showed that In recent years a number of genetic mutations have doses of 400 µg/day would place all women over the been described which cause NTDs in the mouse. Only protective red cell folate concentration of 400 µg/L a few of these can be prevented by treating the preg- but at the expense of unnecessarily high exposure for nant dam with folic acid, however, these models may some people. Delivery of 200 µg/day would also pre- be important aids in elucidating the mechanisms vent many NTDs and 100 µg/day would be beneficial. involved in neural tube formation. Because NTDs are They called for immediate fortification at this low lev- sporadic even within families with high recurrence el; however, many experts believe that their conclu- rates it is probable that the underlying aetiology is a sions are over-optimistic and that fortification at high- combination of genetic and environmental factors er levels is the only long-term solution. present at the time of neural tube closure.

What intake of folate is needed References to prevent NTDs? 1. MRC Vitamin Study Research Group. Prevention of The results of the MRC and Hungarian trials pre- neural tube defects: Results of the Medical Research sented Public Health decision makers with a difficult Council Vitamin Study. Lancet 1991; 338: 131-7. dilemma – to decide on the most effective strategy to 2. Czeizel AE, Dudas I. Prevention of the first occurrence 56 A. M. Molloy et al.

of neural-tube defects by periconceptional vitamin bile” variant of methylenetetrahydrofolate reductase supplementation. N Engl J Med 1992; 327: 1832-5. and neural tube defects: an evaluation of genetic risk 3. Kirke PN, Molloy AM, Daly LE, Burke H, Weir DG, and the relative importance of the genotypes of the Scott JM. Maternal plasma folate and vitamin B12 are embryo and the mother. Am J Hum Genet 1999; (in independent risk factors for neural tube defects. Q J press). Med 1993; 86:703-8. 8. Molloy AM, Mills JL, Kirke JN, et al. Low blood folates 4. Wald NJ, Hackshaw AK, Stone R, Sourial NA. Blood in NTD pregnancies are only partly explained by ther- folic acid and vitamin B12 in relation to neural tube molabile 5,10-methylenetetrahydrofolate reductase: defects. Br J Obstet Gynecol 1996; 103:319-24. 5. Mills JL, McPartlin JM, Kirke PN, et al. Homocysteine low folate status alone may be the critical factor. Am metabolism in pregnancies complicated by neural J Med Genet 1998; 78:155-9. tube defects. Lancet 1995; 345:149-51. 9. Daly LE, Kirke PN, Molloy AM, Weir DG, Scott JM. 6. Steegers-Theunissen RPM, Boers GH, Blom HJ, et al. Folate levels and neural tube defects: Implications for Neural tube defects and elevated homocysteine levels prevention. JAMA 1995; 274: 1698-702. in amniotic fluid. Am J Obstet Gynecol 1995; 172: 10. Daly S, Mills JL, Molloy AM, et al. Minimum effective 1436-41. dose of folic acid for food fortification to prevent neur- 7. Shields DC, Kirke PN, Mills JL, et al. The “thermola- al-tube defects. Lancet 1998; 350: 1666-9. Educational Session 6 Haematologica 1999; 84:(EHA-4 educational book):57-60 Chairman: A.V. Hoffbrand Folic acid and thrombosis

Homocysteine and ischaemic disease MALCOLM R. LAW, NICHOLAS J. WALD Department of Environmental and Preventive Medicine, Wolfson Institute of Preventive Medicine, St Bartholomew’s and The Royal London School of Medicine and Dentistry, London, UK

link between increasing serum homocysteine Serum homocysteine concentration follows a concentration and increasing risk of athero- Gaussian distribution skewed to the right and Asclerotic disease was first shown 30 years ago,1 increases with age. In a cohort of men recruited in and a large body of evidence has since been accu- London (the BUPA study), the average concentration mulated on the subject. Serum homocysteine can be was 12 µmol/L (at an average age of 53 years),1 sim- lowered by folic acid supplementation,2-4 so if the ilar to the Western average at this age,2 and the 5th, relationship between homocysteine and cardiovas- 25th, 50th, 75th and 95th centiles were 7.5, 9.5, 11.1, cular disease is causal, folic acid supplementation 13.3 and 17.8 µmol/L respectively. In non-Western would lower cardiovascular mortality. Randomised communities levels tend to be lower. controlled trials of folic acid supplementation and ischaemic heart disease are underway,4 but the Epidemiological evidence on results are not expected for some time. In this review homocysteine and ischaemic heart we examine the evidence on homocysteine and cir- disease culatory diseases, ischaemic heart disease in partic- The majority of the epidemiological studies are of ular, assessing the likelihood that the relationship is retrospective design; homocysteine was measured in one of cause and effect and quantifying the reduc- subjects with known ischaemic heart disease (gener- tion in mortality to be expected through folic acid ally survivors of myocardial infarction) and in control supplementation. Our review draws on our previous subjects, adjusting (or matching) for age, sex and oth- 1 paper on homocysteine and heart disease, and also er confounding factors. Boushey and her colleagues draws on its citation list: to comply with the limita- reviewed 15 such studies in 1995 (ref. #10 cited in1). tion of 10 references here we identify papers cited in These studies show a strong and continuous rela- our previous paper by their reference numbers there. tionship between homocysteine and ischaemic heart disease. An analysis of eight retrospective studies Serum homocysteine in Western yielded the summary estimate that the odds ratio of populations ischaemic heart disease for a 5 µmol/L increase in Homocysteine is an intermediate compound gen- serum homocysteine was 1.84 (95% confidence inter- erated from the demethylation of methionine. It is val 1.52-2.23)1 – that is, an excess risk of 84% (52- metabolised in two ways. The first is the transsul- 123%) (in the remaining retrospective studies the rela- phuration pathway, in which cystathionine b-synthase tion could not be quantified in this way because the irreversibly binds homocysteine to serine to form cys- necessary data were not published or because homo- tathionine. The second is the remethylation cycle, cysteine was measured only after methionine load- which involves two major enzymes, 5,10-methylene- ing). No substantive source of systematic error in this tetrahydrofolate reductase (MTHFR) and methionine estimate from the retrospective studies can be identi- synthase. Specific genetic defects leading to deficien- fied. Retrospective studies are generally considered cies in the activities of all three enzymes have been less rigorous than prospective or cohort studies, but identified and these are discussed below. The activi- the major source of bias affecting retrospective stud- ties of the three enzymes, and hence the blood con- ies in general, recall bias, does not apply because a centration of homocysteine, also have environmental biochemical measurement was made to quantify the determinants, and are dependent on three of the B vit- exposure, and the retrospective studies were con- amins - folate, B6 and B12. Most of the evidence is on ducted sufficiently long after the ischaemic event to folic acid, which has been shown to have an impor- avoid the known acute change in serum homocysteine tant effect in lowering homocysteine.2,3 after myocardial infarction. The evidence from the cohort studies is less con- sistent. The results of eight published cohort studies Correspondence: MR Law, Wolfson Institute of Preventive Medicine, St are summarised in Table 1 (these studies were of nest- Bartholomew’s and The Royal London School of Medicine and Den- ed case-control design: blood was collected from healthy tistry, Charterhouse Square, London EC1M 6BQ, UK. Tel. international +44-171-9826268 – Fax: international +44-171-9826270 – E-mail: subjects at the outset and serum frozen, and at the [email protected] end of follow-up homocysteine was measured in 58 M. R. Law et al.

Table 1. Odds ratio of ischaemic heart disease (IHD) events for a 5 µmol/L increase in serum homocysteine levels: results from eight prospective studies (nested case-control analysis) of persons without disease at study entry, adjusted (or matched) for age and sex.

Estimated Odds ratio No of subjects average age at (95% confidence Study Sex with IHD without IHD IHD event, yrs interval)

Tromso Study (Arnesen et al.;* Norway) 90% men 122 478 53 1.54 (1.21-1.95) BUPA Study (Wald et al.;1 Britain)‡ men 229 1129 58 1.41 (1.20-1.65) British Regional Heart Study (Whincup et al.7) men 110 118 54 1.38 (1.02-1.86) Rotterdam (Bots et al.8 the Netherlands) 66% men 104 533 75 1.34 (1.10-1.69) US Physicians (Stampfer et al.*) men 271 271 62 1.29 (1.01-1.67) Atherosclerosis Risk in Communities men 174 395 59 1.00 (0.93-1.07)° (ARIC) Study (Folsom et al.9 USA) women 58 132 59 1.13 (1.01-1.29)° North Karelia Study (Alftham et al.;* Finland) men and women 191 269 61 1.00 (0.77-1.29) MRFIT (Evans et al.;* USA) men 230 474 61 0.95 (0.79-1.12)° *The Tromso, US Physicians, North Korelia and MRFIT studies are cited as refs. #12-15 respectively in our earlier review1; °Calculated from published data; ‡IHD death only (other studies recorded non-fatal myocardial infarction in addition). serum from subjects who developed ischaemic heart disease with raised serum homocysteine concentra- disease and matched controls). In all the studies the tion, and the presence of a cross-country association result could be expressed as an odds ratio associated (refs #31, 32 and 11 cited in1). with a 5 µmol/L increase in serum homocysteine. The In quantifying the association in the cohort studies, results from the eight studies vary considerably, and we take the result of our own study (BUPA) as repre- the variation is greater than would be expected sentative of the positive studies because it is large, its through chance alone. Five studies yield estimates of results were reported in some detail, and its result - an a 30% to 50% increased risk for a 5 µmol/L homo- odds ratio of 1.41 (95% confidence interval 1.20- cysteine increase, but three suggest no increase. In 1.65; p = 0.001), or 1.33 (1.22-1.59) after adjust- some of the studies the confidence intervals are wide, ment for other heart disease risk factors1 – is close to but the lower confidence intervals in two positive stud- the median of the positive cohort studies (Table 1). As ies (Tromso and BUPA) exceed the upper confidence with the retrospective studies, the true association interval in two negative studies [ARIC (men) and would be somewhat greater because of the effect of MRFIT] - see Table 1. The reason for the variation, regression dilution bias (the dilution of the effect of a with some studies negative and others positive, is risk factor in a cohort when based on single mea- uncertain. The subjects in the negative studies tended surements that fluctuate in an individual over time). to be older and relative risk is likely to decline with The bias could be allowed for using data from a study 1 age, but age alone does not totally account for the recording homocysteine measurements on two or variation. The heterogeneity between study results is more occasions in the same individuals, but at present too large for an overall average to be taken; this would no such data are published. be statistically inappropriate. The retrospective studies and the positive cohort The relationship between homocysteine and stroke studies show a dose-response relationship between follows the same pattern. Retrospective studies show serum homocysteine and ischaemic heart disease that a strong association but the cohort studies vary, and is continuous across the range of homocysteine levels individual studies tend to be consistent in showing in Western populations. Table 2 shows the estimates either an association or the lack of an association with both stroke and heart disease (refs. #10, 14, 16, 20 from the eight retrospective studies combined and & 21 cited in 1). from the BUPA prospective study of the risk of IHD In addition to these eight cohorts recruited from according to homocysteine level, with homocysteine the general population, there are three additional levels divided into four categories (<10, 11-20, 21-30, published cohort studies, one of progression of and 31-60 µmol/L). The continuous dose-response peripheral arterial disease, one of mortality in patients relationship provides strong evidence against the view with known coronary artery disease, and one of that only greatly elevated levels of homocysteine thrombotic events in patients with systemic lupus ery- increase the risk of ischaemic heart disease. thematosus, and these all demonstrated an associa- tion between homocysteine and occlusive vascular Genetic evidence on homocysteine and disease (refs. #18, 19 and 21 cited in1). Additional ischaemic heart disease confirmatory observational evidence is provided by There are autosomal-recessive inborn errors of the association of ultrasound-detected carotid artery metabolism affecting each of the three major enzymes Session #6 – Folic acid and thrombosis 59

Table 2. Odds ratio of ischaemic heart disease according to Combining the epidemiological and level of serum homocysteine: results from eight retrospec- tive studies and the prospective BUPA study. (Data from genetic evidence Wald et al.1). Table 3 shows estimates from three types of study on the magnitude of the relationship between homo- Serum Odds ratio (95% confidence interval) cysteine and ischaemic heart disease. The estimate homocysteine BUPA Retrospective from the retrospective studies is greater than that (µmol/L) study studies from the cohort study; this may be partly attributable to the fact that the subjects in the retrospective stud- <10 1.0 (0.5-1.9) 1.0 (0.9-1.2) ies were about seven years younger on average and 11-20 1.9 (1.0-3.5) 2.5 (2.2-2.9) relative risk decreases with age.1 The magnitude of the 21-30 2.2 (0.9-5.7) 5.3 (3.3-8.4) association in both the retrospective studies and the 31-60 5.7 (1.1-28.5) 6.1 (2.7-13.7) cohort study would be a little lower after adjustment for the confounding effects of other heart disease risk factors (the BUPA result changing from an odds ratio of 1.41 to 1.33), but this is offset by the fact that the Table 3. Estimates from three sources of the increase in inability to adjust the results for the regression dilu- risk of ischaemic heart disease for a 5 µmol/L increase in serum homocysteine. tion bias (because of the unavailability of published data on serum homocysteine measured on two or Odds ratio more occasions in the same subjects) has led to (95% confidence interval) underestimation. The estimate taken from the stud- ies of subjects homozygous for the thermolabile Retrospective studies1 1.8 (1.5-2.2) MTHFR defect8 is applicable to a similar age group Cohort study* 1.4 (1.2-1.7) (about 60) as the cohort study estimate, and is simi- lar in magnitude to this estimate. This quantitative Genetic evidence (subjects with agreement between the cohort study and the MTHFR thermolabile MTHFR)8 1.4 (1.0-2.0) genetic studies is remarkable and offers compelling *estimate from the BUPA cohort.1 evidence for homocysteine being directly responsible for the excess heart disease mortality associated with higher levels. At the age of 60, a 5 µmol/L increase in serum homocysteine is therefore likely to increase the risk of ischaemic heart disease by approximately 40%. in homocysteine metabolism,1 and in each defect homozygotes have very high serum homocysteine lev- Conclusions and implications for els (about 10-50 times higher than the general pop- prevention ulation) and very high risk of premature cardiovascu- The cohort studies taken together favour an asso- lar disease. Heterozygotes for the three disorders have ciation between homocysteine and heart disease serum homocysteine levels about three times the pop- although the negative results of three studies cannot ulation average and high risk of cardiovascular dis- satisfactorily be explained. The retrospective studies ease. The only biochemical change that is common to indicate a strong relationship, and it is not possible to these inborn errors of metabolism is a high homo- identify any likely source of error in these studies. The cysteine level; no other metabolite is consistently high three cohort studies on patients with existing circula- or low in all three. Given that circulatory disease is tory disease, and the additional observational evi- also common to all three genetic disorders, homo- dence, all show associations between homocysteine cysteine is almost certainly the cause of the disease and occlusive vascular disease. There is also epidemi- and not merely a marker of some other cause. ological evidence for a direct relationship between Another genetic defect affecting about 10% of the blood folate concentration and risk of cardiovascular population (being homozygous for a thermolabile disease. The genetic evidence for a cause and effect form of MTHFR8) offers a useful natural experiment relationship, discussed above, is strong. Results of to test the homocysteine hypothesis. This variant animal and in vitro experimental studies show that leads to moderate increases in homocysteine (2.8 increases in blood homocysteine levels increase the µmol/L on average in subjects homozygous for the extent of vascular and platelet damage (refs. #52-55 mutation compared to those without the mutation in cited in1). Taken together, the epidemiological, genet- a meta-analysis of eight studies8) and a moderate ic, and experimental evidence make a compelling case increase in risk of ischaemic heart disease (the sum- for a causal relationship between homocysteine and mary odds ratio was 1.22, 95% confidence interval ischaemic heart disease. The genetic and epidemio- 1.01-1.478). A significantly elevated risk was also logical evidence both indicate that the dose-response shown in studies identifying the abnormal phenotype relationship is continuous across the range of serum rather than the genotype.1 homocysteine levels in Western populations. 60 M. R. Law et al.

Serum homocysteine can be reduced by increasing consumption of folic acid.2-4 Folate is best given as References folic acid (either as supplements or as food fortifica- tion) because the bioavailability is greater than that of 1. Wald NJ, Watt HC, Law MR, et al. Homocysteine and 4 the conjugated folates that occur naturally in food. ischaemic heart disease. Arch Intern Med 1998; 158: The homocysteine-lowering effect appears to plateau 862-7. 2 at folic acid intakes of 1 mg/day or less, but there are 2. Homocysteine Lowering Trialists’ Collaboration. Low- few trial data on intakes of 0.4 mg/day or less (an ering blood homocysteine with folic acid based sup- appropriate level for food fortification). A supplement plements: meta-analysis of randomised trials. BMJ of 0.4 mg/day has been shown by Ward and col- 1998; 316: 894-8. leagues to reduce average homocysteine levels in mid- 3. Malinow MR, Duell PB, Hess DL, et al. Reduction of dle-aged subjects by 1.9 µmol/L (ref. #61 cited in1), plasma homocyst(e)ine levels by breakfast cereal for- and a study of breakfast cereal fortification at this lev- tified with folic acid in patients with coronary heart el showed a similar reduction in homocysteine.3 The cohort study result indicates that this reduction in disease. N Engl J Med 1998; 338:1009-15. homocysteine is equivalent to a reduction in mortali- 4. Omenn GS, Beresford SAA, Motulsky AG. Preventing ty from ischaemic heart disease of 10% (95% confi- coronary heart disease. B vitamins and homocysteine. dence interval 4-16%), and adjustment for the regres- Circulation 1998; 97:421-4. sion dilution bias would increase this estimate. There 5. Whincup PH, Mendall MA, Perry IJ, Strachan DP. is a need to confirm the size of the effect of this level Hyperhomocysteinaemia, Helicobacter pylori and of folic acid supplementation on serum homocysteine coronary heart disease. Heart 1997; 78: 524. levels, and to determine whether there is a homocys- 6. Bots ML, Launer LJ, Lindemans J, et al. Homocysteine teine threshold below which folic acid ceases to reduce and short-term risk of myocardial infarction and serum homocysteine concentration further. This could stroke in the elderly. Arch Intern Med 1999; 159: 38- be accomplished by a relatively small and short-term 44. randomised study of folic acid supplementation. 7. Folsom AR, Nieto FJ, McGovern PG, et al. Prospective The knowledge that increasing folic acid intake has study of coronary heart disease incidence in relation to an important effect in preventing neural tube defects fasting total homocysteine, related genetic polymor- justifies fortifying staple food with folic acid. The phisms, and B vitamins. Circulation 1998; 98: 204-10. expected benefit that this would have in reducing the 8. Kluijtmans LAJ, Kastelein JJP, Lindemans J, et al. Ther- risk of heart disease is an important associated ben- molabile methylenetetrahydrofolate reductase in coro- efit affecting everyone in the community. nary artery disease. Circulation 1997; 96: 2573-7. Educational Session 6 Haematologica 1999; 84:(EHA-4 educational book):61-63 Chairman: A.V. Hoffbrand Folic acid and thrombosis

Homocysteine, platelet function and thrombosis GIOVANNI DI MINNO, ANTONIO COPPOLA, FRANCESCO PAOLO MANCINI, MAURIZIO MARGAGLIONE Depts of Clinical and Experimental Medicine and Biochemistry and Biotechnology, "Federico II" University, School of Medicine, Naples, and Unit of Thrombosis and Athrosclerosis I.R.C.C.S. "Casa Sollievo della Sofferenza" S. Giovanni Rotondo, Italy

t was almost 30 years ago that hyperhomocys- after the addition of homocysteine to plasma. How- teinaemia was first associated with arteriosclero- ever, subsequent studies did not confirm these obser- Isis and since then a growing body of evidence has vations. Increased platelet aggregation after homo- established its independent contribution to cardio- cysteine exposure has been documented. This con- vascular disease. It is now clear that 15-40% of cept has been challenged as well. Discrepancies also patients with coronary, cerebral or peripheral arter- exist about the effect of homocysteine thiolactone ial disease have increased plasma levels of homocys- (HTL), the cyclic oxidation product of homocysteine, teine, and that, regardless of the severity and/or the on platelet function. Early studies suggested that HTL genetic/nutritional background leading to hyperho- had only a very small effect on platelet aggregation. mocysteinemia, subjects with elevated (>20 µM/L) However, at variance with the inactive salt (hydro- plasma levels of homocysteine have a tendency to chloride form), the free base of HTL fosters platelet arterial and venous thrombotic events. When severe- aggregation. On the other hand, in synergism with ly increased in plasma (>100 µM/L), homocysteine other methyltransferase inhibitors, HTL inhibits can leak into the urine causing homocystinuria. In platelet aggregation. A study that sheds some light on this severe form of hyperhomocysteinaemia, prema- such discrepancies is that of Stamler et al. Their exper- ture arteriosclerosis, and arterial and venous throm- iments indicate that homocysteine does not cause bosis are common findings. Biochemical measure- platelet aggregation per se. This is consistent with the ments of urinary metabolites and clinical trials with antiplatelet effects of other low molecular weight thi- aspirin, indicate that enhanced biosynthesis of ols with physical and chemical characteristics similar thromboxane A2 (TXA2) by platelets is a major con- to cysteine and glutathione. Instead, homocysteine tributor to the risk of thrombosis associated with can increase platelet adhesion to endothelial cells several risk factors. In subjects with homocystinuria, (EC) as a consequence of its toxic effect on the we have reported an abnormally high in vivo TXA2 endothelium itself. EC produces endothelium-derived biosynthesis, as reflected by the excretion of its major relaxing factor (EDRF) which reacts with homocys- enzymatic urinary metabolites. The possibility of a teine to form S nitroso-homocysteine (SNOHO). The platelet origin for the abnormally high TXA2, was sug- latter is a strong antiplatelet agent with a 15-min half- gested by the results of studies with 50 mg/day of life (for comparison, EDRF half life is about 5-30 sec). aspirin. In the following paragraphs, we shall review Therefore in normal conditions, the toxicity of homo- the effects of homocysteine on platelets. cysteine is abolished by the formation of SNOHO. Activation of the coagulation process and/or However when homocysteine levels saturate the avail- endothelial dysfunction are known to trigger specif- able amounts of EDRF, unmodified homocysteine ic effects on platelet biochemistry that in turn becomes available. This causes endothelial injury, enhance the biosynthesis of TXA2. A hypercoagulable with a consequent reduction of the EDRF produc- syndrome and endothelial dysfunction are triggered tion, followed by reduced formation of SNOHO and by homocysteine. Thus, we shall also review data con- in turn, of the antiplatelet potential. cerning the effects of high levels of homocysteine on These data suggests a role for an oxidant stress in blood coagulation/fibrinolysis and endothelial cells. the risk of thrombosis related to elevated levels of homocysteine. In keeping with this, we have reported Homocysteine and platelet function > 2 SD increase from the control mean of the urinary In homocystinuric patients, platelet survival has excretion of 11-dehydro-thromboxane B2 and of 2,3- been reported abnormally low. This was in line with dinor-thromboxane B2 (TXB2), major enzymatic deriv- the results obtained in non-human primates. In keep- atives of TXA2 in 11 homocystinuric patients. The ing with this, increased platelet stickiness was demon- abnormally high excretion of this valuable index of in strated in the blood of homocystinuric patients and vivo platelet activation, was independent of the pres- ence of major cardiovascular risk factors. On the oth- er hand, the fact that two metabolites resulting from Correspondence: Giovanni Di Minno, Clinica Medica, Istituto di Medici- na Clinica e Sperimentale, Policlinico "Federico II", via S. Pansini 5, two independent pathways were both increased in a 80131, Naples, Italy. Tel. and Fax. international +39-081-7462060. similar manner strongly argued for these changes to 62 G. Di Minno et al.

be a reflection of increased biosynthesis of TXA2. Both system that suggest a hypercoagulable state in this the profound suppression by low-dose aspirin and the setting. Reduced levels of antithrombin (AT), of fac- slow recovery after drug withdrawal were compatible tor VII, and protein C have been reported. In vitro stud- with a platelet source of the abnormally high TXA2 ies provided a biochemical background for a hyper- biosynthesis. To understand the mechanisms of the coagulable syndrome in hyperhomocysteinaemia. enhanced TXA2 biosynthesis in homozygous cysta- activity and prothrombin activation have thionine ␤-synthase deficiency (CBSD), 500 mg of the been shown to be increased by the addition of 0.5-10 antioxidant drug probucol was given to seven patients mM homocysteine to cultured bovine aortic endothe- for 3 weeks. In vitro probucol prevents the oxidative lial cells. While 8 hrs were required to detect an modification of low-density lipoproteins. This treat- increase in factor V activity in the presence of 10 mM ment resulted in a 40-60% drop in TX metabolite homocysteine, 24 to 30 hrs were needed in the pres- excretion, which did not correlate with a reduction in ence of 0.1 or 0.5 mM homocysteine. These effects blood cholesterol levels. Since oxidation of lipopro- appeared to depend on the effect of homocysteine teins – which can induce platelet TXA2 formation – is on the natural anticoagulant protein C. A direct effect facilitated by homocysteine, inhibition of TXA2 pro- of homocysteine on protein C activation was subse- duction by probucol is consistent with the possibility quently shown. Incubation of bovine or human that oxidized lipoproteins contribute to an increased umbilical vein cultured endothelial cells (HUVEC) arachidonic acid metabolism in platelets of patients with 7.5 to 10 mM homocysteine for 6 to 9 hrs pro- with CBSD. It is worth stressing that lipid peroxidation duced a 90% inhibition of protein C activation. This can be initiated not only by hydrogen peroxide, but effect could be partially explained by a competitive also by superoxide and hydroxyl radicals, which can be inhibition by homocysteine of the thrombomodulin- generated during oxidation of thiols. thrombin interaction. Hayashi et al. gained addition- al insight into the mechanism by which homocysteine Endothelial cells impairs thrombomodulin activity in HUVEC. They The data with endothelial cells suggest a major role found a time- and dose-dependent inhibitory effect of for some free radical species in the endothelial injury homocysteine on thrombomodulin cofactor activity. mediated by homocysteine. As for platelet activation, Thrombomodulin activity, measured as protein C the possibility therefore exists that an oxidant stress activation, was reduced to 5 or 10% of the baseline may be involved in the thrombogenic potential of values after incubation with 10 mM homocysteine homocysteine. The possibility that hydrogen perox- when the activity was determined on the cell surface ide is responsible for the cellular damage induced by or in whole cell extracts respectively. Thus, their data homocysteine has been analysed in detail by Starke- suggest that the effect of homocysteine on thrombo- baum and Harlan. Copper-catalysed auto-oxidation modulin activity is due to a reduction of the native of cysteine in alkaline media leads to the reduction of thrombomodulin, which is followed by a compen- oxygen and the generation of hydrogen peroxide. In satory increase in the expression of the thrombo- view of this, Starkebaum and Harlan showed that, in modulin gene and of the total thrombomodulin lev- a cell free system, increasing concentrations of copper el. Finally, in a cell-free system, they demonstrated (1-50 µM), increased homocysteine oxidation in a that homocysteine is able to inhibit the binding of dose-dependent fashion. The addition of catalase to thrombomodulin to thrombin and concluded that the system reduced oxygen consumption by nearly this is caused by a decreased binding capacity of the one half, thus suggesting that H2O2 was formed dur- reduced thrombomodulin. Also tissue factor (TF), a ing the reaction. However, H2O2 did not seem to central protein of the extrinsic pathway of the coagu- accumulate in the presence of homocysteine, sug- lation, has been indicated as another possible target gesting that homocysteine itself can scavenge H2O2. for the thrombogenic action of homocysteine. Incu- Interestingly, while at concentrations of 0.05-5 µM, bation of HUVEC with 10 mM homocysteine for 8 2+ Cu increased the rate of H2O2 formation, at con- hours increased TF activity by six-fold. A clear dose centrations above 5 µM, H2O2 formation was dependency of this effect (0.1-10 mM homocysteine) reduced. The effect of the higher concentrations may was demonstrated. Homocysteine induced TF activi- ++ be the result of Cu catalysed reduction of H2O2 to ty was inhibited by N-ethylmaleimide in HUVEC, thus water. The relationship between copper, homocys- indicating that the sulphur group was instrumental teine and endothelial injury was documented by the in the observed phenomenon. Finally, the ability of observation that a dose-dependent lysis of cultured homocysteine to induce TF mRNa, measured by a bovine aortic endothelial cells could only be observed quantitative polymerase chain reaction technique, when homocysteine (up to 5 mM) was added in the revealed an almost 4-fold increase in the TF mRNA, presence of copper (2 µM). when comparing HUVEC and fibroblasts after 3 hrs incubation with 10 mM homocysteine. Coagulation/fibrinolysis The effect of homocysteine on AT has been explored Ex vivo data from patients with homocystinuria have with emphasis on the interaction between AT and shown a variety of abnormalities of the coagulation heparin-like glycosaminoglycans in porcine aortic Session #6 – Folic acid and thrombosis 63 endothelial cells. The data showed that the maximal raises the possibility that anticoagulation should be AT binding capacity to heparin sulphate was reduced taken into consideration in preventing thrombosis in to 30% of normal after a 24-hour incubation with 1 these patients. On the other hand, since a homocys- mM homocysteine. This effect was dependent on teine-mediated oxidant stress may trigger platelet acti- sulphydryl groups and appeared to involve the gener- vation, the latter leading to a hypercoagulable state, ation of hydrogen peroxide, being prevented by cata- the question is whether lowering plasma homocys- lase, but not by superoxide dismutase. teine will per se correct the the tendency to thrombo- The interference of homocysteine with the fibri- sis of these patients. The latter strategy could be nolytic system has been addressed by Hajjar. She strengthened by the addition of antioxidants. Whether demonstrated a 65% decrease in cellular binding sites this is the case, will need to be established in large, for tissue plasminogen activator (t-PA), following prospective studies in appropriate experimental and treatment of cultured HUVEC with 1.5 to 7.5 mM clinical settings. homocysteine. The author also provided evidence that this was due to a reduction of the binding sites for t- PA on the 40 kDa receptor protein. Interestingly, the References receptor capacity to bind plasminogen was not altered, thus suggesting that the receptor had been 1. Boushey CJ, Beresford SAA, Omenn GS, Motulsky AG. A quantitative assessment of plasma homocysteine as altered only in the specific domain responsible for the a risk factor for vascular disease. JAMA 1995; 274: binding of t-PA, and that the COOH-terminal 1049-57. domain, which binds plasminogen, remained unmod- 2. Di Minno G, Davì G, Margaglione M, et al. Abnor- ified. Along the same line, Harpel et al., also focused mally high thromboxane biosynthesis in homozygous on the potential modulation of fibrinolysis by homo- homocystinuria. Evidence for platelet involvement and probucol-sensitive mechanism. J Clin Invest 1993; 92: cysteine. They studied the interaction of plasmin- 1400-6. modified fibrin and lipoprotein(a), Lp(a). Because of 3. Fryer RH, Wilson BD, Gubler DB, Fitzgerald LA, its homology to kringle IV of plasminogen, Lp(a) inter- Rodgers GM. Homocysteine, a risk factor for prema- feres with fibrinolysis by competing with plasmino- ture vascular disease and thrombosis, induces tissue gen binding sites. Harpel et al. demonstrated that factor activity in endothelial cells. Arterioscler Thromb 1993; 13:1327-33. homocysteine can enhance the binding of Lp(a) to 4. Hajjar KA. Homocysteine-induced modulation of tis- fibrin, especially to plasmin-treated fibrin. This bind- sue plasminogen activator binding to its endothelial ing was inhibited by ⑀-aminocaproic acid, thus indi- cell membrane receptor. J Clin Invest 1993; 91:2873- cating lysine binding site specificity, and was also 9. increased by cysteine, glutathione and N-acetylcys- 5. Hayashi T, Honda G, Suzuki K. An atherogenic stim- ulus homocysteine inhibits cofactor activity of throm- teine. Using gel electrophoresis and immunoblotting, bomodulin and enhances thrombomodulin expres- the authors observed changes in the mobility of the sion in human umbilical vein endothelial cells. Blood apo(a) moiety after exposure to homocysteine, and 1992; 79: 2930-6. therefore concluded that homocysteine could be 6. McCully KS. Vascular pathology of homocysteinemia: implications for the pathogenesis of arteriosclerosis. responsible for altering the structure of apo(a), pos- Am J Pathol 1969; 56: 111-28. sibly exposing additional binding sites for the fibrin 7. Mancini FP, Di Minno G. Hyperhomocysteinemia and surface. As a consequence, the thrombotic potential thrombosis: a difficult link. Nutr Met Cardiovasc Dis of Lp(a) would be increased by homocysteine. 1996; 6:168-77. 8. Nishinaga M, Ozawa T, Shimada K. Homocysteine, a Perspectives thrombogenic agent, suppresses anticoagulant heparan sulphate expression in cultured porcine aor- High levels of thrombin in the circulation and a sus- tic endothelial cells. J Clin Invest 1993; 92:1381-6. tained tendency to thrombosis occur in hypercoagu- 9. Stamler JS, Osborne JA, Jaraki M, et al. Adverse vas- lable states. In addition to its role in the conversion of cular effects of homocysteine are modulated by fibrinogen to fibrin, thrombin is a potent inducer of endothelium-derived relaxing factor and related oxides of nitrogen. J Clin Invest 1993; 91:308-18. platelet activation and TXA2 biosynthesis. The data 10. Starkebaum G, Harlan JM. Endothelial cell injury due presented above suggests a hypercoagulable syndrome to copper-catalyzed hydrogen peroxide generation in patients with severe hyperhomocysteinaemia. This from homocysteine. J Clin Invest 1986; 77: 1370-6. Educational Session 7 Chairman: G. Dighiero Haematologica 1999; 84:(EHA-4 educational book):64-66 Transgenic models of haematological diseases

Transgenic models of T-cell prolymphocytic leukaemia MARC-HENRI STERN From Unité INSERM U462, Institut Universitaire d’Hématologie, Hôpital Saint Louis, Paris, France

-cell prolymphocytic leukaemia (T-PLL) is a Elderly CD2-p13 transgenic mice develop rare form of mature T-cell leukaemia. It occurs T-cell prolymphocytic leukaemia in two populations, in elderly (mean age: 69 Although no immunologic abnormality or tumours T 1 years) and in young patients suffering from ataxia were detected in the first year of life of the transgenic telangiectasia (AT).2 This leukaemia is characterised mice, the survey of the animal cohort was maintained by an aggressive syndrome with lymphocytosis, lym- for a long period of time, keeping in mind that the fol- phadenopathy, hepatosplenomegaly and skin low-up of AT patients has shown that leukaemias lesions. Prolymphocytes are typically cells with a high with MTCP1 or TCL1 rearrangements have a long nucleocytoplasmic ratio, a basophilic cytoplasm and indolent pre-clinical course. It was not until 15 devoid of granules, and a single prominent nucleo- months that the first mice suffering from leukaemic lus. The immunophenotype is that of mature T lym- syndromes were detected. Because the spontaneous phocytes, but leukaemic cells bearing both CD4 and death of the transgenic animals could impair the CD8 are not infrequent.1 Interestingly, a characterisation of the leukaemia, the cohort was preleukaemic stage of T-PLL has been identified in sacrificed at 18 to 20 months of age, and compared AT patients.3 Recurrent chromosomal aberrations to non-transgenic age-matched siblings. associated with T-PLL involve on one hand the A typical leukaemia was characterised by lymphoid TCRA/D or TCRB gene, and on the other, the Xq28 cells with an irregular nucleus, condensed chromatin, or 14q32.1 regions.1 Their molecular characterisa- a unique and prominent nucleolus, and a basophilic tion led to the identification of the MTCP1 and TCL1 cytoplasm devoid of granules. Spleen and liver were genes, respectively.4-6 MTCP1 is organised into seven consistently invaded, even in the absence of organo- exons spread over approximately 10 kb. Alternative megaly. The immunophenotype of the leukaemic cells splicing generates two types of transcripts encoding was CD3+CD8+CD4–CD25–B220– in all cases but one two entirely different proteins, p8MTCP1 and p13MTCP1.4 which was CD4+CD8–. The rarity of CD4+ T-cell Both products were upregulated in leukaemic cells leukaemia was one of the rare differences between the bearing a t(X;14) translocation, none showed simi- human and murine T-PLLs and could not be larities with known oncogenes, and in vitro transfor- explained by a lower level of expression of p13MTCP1 in mation of a fibroblastic cell line was negative. Direct the CD4+ subset. Some leukaemias were also charac- evidence of the oncogenic activity of MTCP1 was terised by smaller cells, as has also been described in thus required. some human T-PLL. It should be noted that the analysis of a cohort of Characterisation of CD2-p13 transgenic old mice (18-20 months) revealed a large number of mice tumours not linked to the transgene (arising in trans- Transgenic mice were generated in which expres- genic and control mice). Solid tumours were easily dis- sion of p13MTCP1 was controlled by the CD2 regulatory carded from the analysis, but coincidental haemato- sequences (CD2-p13 mice). This construct was cho- logical diseases needed to be identified by precise and sen because it confers a level of expression indepen- complete diagnosis. Tissue samples from each trans- dent of the insertion site of the transgene. Three genic and control mouse were reviewed by a patholo- founders were obtained and bred. These transgenic gist, blood smears were reviewed by a cytologist, lines expressed p13MTCP1 in thymuses and in at were analysed by FACS with B and T cell different levels, in correlation with the number of markers and clonal rearrangements of the TCRb and transgene copies in the line. No expression was IgH were searched for by Southern blotting. Murine T- detected in the non-lymphoid organs. PLLs were unambiguously demonstrated in the three transgenic lines and never in the control mice, where- as up to 10 percent of various B cell and non-lym- phoid malignancies were found in transgenic and con- Correspondence: Marc-Henri Stern, MD, PhD, Unité INSERM U462, trol animals. The incidence of T-PLL was 100, 50 and Centre Hayem, Hôpital Saint Louis, 75475 Paris Cedex 10, France. Tel. international +33-1-53722215 – Fax. international +33-1- 21 percent in the three lines, and correlated with the 53722217 – E-mail. [email protected] level of expression of the transgene (see Table 1). Session #7 – Transgenic models of haematological diseases 65

Table 1. T-PLL incidence in aged CD2-p13 mice. very long incubation period. It is remarkable that the two animal models of T-PLL are so superimposable. Expression Florid Smouldering Small cell T-PLL free The use of transgenic animals clearly demonstrated of the transgene T-PLL T-PLL variants mice the role of a new oncogene family, MTCP1/TCL1, in the malignant transformation of mature T-cell lym- High 2/12 9/12 1/12 0/12 (17%) (75%) (8%) (0%) phocytes. Moderate 8/22 3/22 0/22 11/22 Interests in generating a murine (36%) (14%) (0%) (50%) model for T-PLL Low 0/14 1/14 2/14 11/14 The blood disorder arising in CD2-p13 transgenic (0%) (7%) (14%) (79%) mice shares most of the clinical and biological fea- Control 0/29 0/29 0/29 29/29 tures of the human T-PLL. Its late onset also mimics (0%) (0%)9 (0%) (100%) the human T-PLL, which generally arises in the elder- ly. The only differences are the more restricted immu- nophenotype of the leukaemic cells and the absence of detectable skin lesions in mice. As it is not the gen- eral rule that a human disease is mimicked so closely Natural history of the disease in transgenic mice, we can infer that the oncogenic Systematic analysis of the transgenic mice gave the activity of MTCP1 is very specific for a precise stage of opportunity to study the disease before it was clinically the differentiation of the T cell lineage. Given the par- or biologically symptomatic. Although no longitudinal allel between human and murine diseases, we can follow-up was performed, a plausible reconstitution of hypothesise that the initial (pre-diagnosis) stage of T- the malignant invasion could be proposed. No immu- PLL is similar to that observed in mice. It is thus pos- nological abnormality was detected before the emer- sible that a smouldering invasion of spleen and liver gence of a clonal population. A proliferative advantage precedes for years or decades the emergence of lym- of the polyclonal T-cell population is deduced from phocytosis. Diagnosis would be made only in indi- their susceptibility to develop leukaemia but needs to viduals who survived long enough to allow the dis- be defined. Clonal populations in the spleen were ease to progress until its florid stage. found in animals without macroscopic abnormality The latency period before emergence of the malig- or lymphocytosis. However, lymphoid secondary nancies, longer than in most murine models of onco- organs were disorganised and nodular invasion of the genesis, is in agreement with the necessity of addi- liver was constant. Even in animals with very large tional genetic events in the development of the T-PLL. spleens and , malignant populations were con- To date, studies of the human disease have demon- fined to these organs, bone marrow was found normal strated the importance of the inactivation of the ATM when analysed, and lymphocytosis was inconstant at gene8,9 and of the duplication of the long arm of chro- this stage. It was only in the animals with the most mosome 8.1 The role of these genetic abnormalities florid diseases that malignant cells were found in most could be tested by generating double mutant animals. organs including bone marrow, and that lymphocy- The striking similarity of the human and murine T- tosis could reach 2,000 G/L. PLLs makes CD2-p13 mice a valuable model to inves- tigate the biological functions of the family of onco- MTCP1 is an oncogene proteins formed by p13MTCP1 and p14TCL1, to identify The presence of T-cell leukaemia in the three trans- the secondary events necessary for the malignant phe- genic lines, and in none of the non-transgenic sib- notype, and to test therapeutics. lings, demonstrated that p13MTCP1 is an oncoprotein. In data not shown here, the alternative product of Acknowledgements MTCP1, p8MTCP1, was also expressed in transgenic ani- The author wishes to thank all the participants of this work. mals using a similar construct. No phenotype was This project was supported by the INSERM and the Ligue demonstrated in these transgenic animals. Thus, over- Nationale Contre le Cancer, France. expression of p8MTCP1 in leukaemias is probably coin- cidental whereas expression of p13MTCP1 is a causal factor of leukaemogenesis. Similar experiments were References performed by Virgilio et al. for the TCL1 gene.7 TCL1 1. Matutes E, Brito-Bapapulle V, Swansbury J, et al. Clin- has four exons and encodes a protein product of 14 ical and laboratory features of 78 cases of T-prolym- kD, p14TCL1, sharing significant homologies with phocytic leukaemia. Blood 1991; 78:3269-74. p13MTCP1. Both p14TCL1 and p13MTCP1 proteins have a 2. Taylor AMR, Metcalfe JA, Thick J, Mak YF. Leukaemia closely related ␤-barrel structure. Animals transgenic and lymphoma in ataxia telangiectasia. Blood 1996; 87:423-8. for TCL1 were generated using another T-cell specific 3. Stern MH, Theodorou I, Aurias A, et al. T-cell non- promoter, lck. CD8+ T-cell leukaemias similar to those malignant clonal proliferation in ataxia telangiecta- arising in the CD2-p13 animals were detected after a sia: a cytological, immunological, and molecular char- 66 M-H. Stern

acterisation. Blood 1989; 73:1285-90. 7. Virgilio L, Lazzeri C, Bichi R, et al. Deregulated expres- 4. Stern MH, Soulier J, Rosenzwajg M, et al. MTCP-1: a sion of TCL1 causes T cell leukemia in mice. Proc Natl novel gene on the human chromosome Xq28 translo- Acad Sci USA 1998; 95:3885-9. ␣ ␦ cated to the T cell receptor / locus in mature T cell 8. Stilgenbauer S, Schaffner C, Litterst A, et al. Biallelic proliferations. Oncogene 1993; 8:2475-83. mutations in the ATM gene in T-prolymphocytic 5. Fisch P, Forster A, Sherrington PD, Dyer MJ, Rabbitts leukaemia. Nat Med 1997; 3:1155-9. TH. The chromosomal translocation t(X;14) (q28;q11) in T-cell pro-lymphocytic leukaemia breaks within one 9. Vorechovsky I, Luo L, Dyer MJ, et al. Clustering of mis- gene and activates another. Oncogene 1993; 8:3271- sense mutations in the ataxia-telangiectasia gene in a 6. sporadic T-cell leukaemia. Nat Genet 1997; 17:96-9. 6. Virgilio L, Narducci MG, Isobe M, et al. Identification 10. Gritti C, Dastot H, Soulier J, et al. Transgenic mice for of the TCL1 gene involved in T-cell malignancies. Proc MTCP1 develop T-cell prolymphocytic leukaemia. Natl Acad Sci USA 1994; 91:12530-4. Blood 1998; 92:368-73. Educational Session 7 Chairman: G. Dighiero Haematologica 1999; 84:(EHA-4 educational book):67-69 Transgenic models of haematological diseases

A murine transgenic model of human cold agglutinin disease SÉVERINE HAVOUIS,* GÉRARD DUMAS*, PATRICK AVÉ,° OTTO PRITSCH,* MICHEL HUERRE,° GUILLAUME DIGHIERO,* CHRISTINE POURCEL* *Unité d’Immuno-hématologie et d’Immunopathologie et °Unité d’Histopathologie, Institut Pasteur, Paris, France

n 1900, the group working with Metchnikoff sug- bodies have been demonstrated under normal con- gested the concept of autoimmunisation by ditions; 3) the presence of normal autoreactive B cells Idemonstrating the presence of autoantibodies in can also be demonstrated; 4) autoantibodies have normal conditions; which was opposed to the con- been induced from normal B lymphocytes upon mito- cept of horror autotoxicus raised by Ehrlich. Nearly at genic stimulation. Since they were able to produce the same time, Landsteiner described the rules gov- autoantibodies, the precursor B-cells producing these erning blood compatibility. He showed that a subject autoantibodies should exist.1 with group A blood will never be able to produce anti- In collaboration with B. Guilbert and S. Avrameas, A antibodies. These results provided strong support to we demonstrated the existence of high levels of nat- Ehrlich’s idea. The influence of the Ehrlich’s ideas was ural autoantibodies (NAA) in normal conditions and so strong that the experiments from Metchnikoff’s that the B cell repertoire secretes these so-called NAA. school were forgotten, and when Donath and Land- Overall, these results indicate that in normal serum, steiner described for the first time an autoantibody, a substantial proportion of circulating Igs are indeed the biphasic haemagglutinin responsible for paroxys- NAA and, that the precursors of this autoreactive tic a frigore haemoglobinuria, they failed to call it repertoire account for a substantial part of normal B autoantibody. In 1949 Burnet and Fenner proposed cells. As these autoantibodies express recurrent the clonal deletion theory. This theory was strongly idiotopes, express V genes frequently in germinal con- influenced by the experiments carried out by Owen figuration and predominate early in life, they are the with dizygotic calves sharing a single . These expression of the germinal repertoire. Interestingly, two calves were mixing their red blood cells but, this repertoire has been found in species phylogenet- despite the fact that they were expressing different ically distant from mammals such as fish and batra- blood groups, they were unable to produce allo-anti- chians. They are autoantibodies since they bind bodies against the blood groups from the other calf. autoantigens. However, they are not self-specific, The interpretation of these experiments and the exper- since nobody has been able to demonstrate this type iments from Medawar’s group, demonstrating that of autoantibodies against very critical self-antigens injection of new born mice with spleen or marrow such as the A and B red blood cells groups. On the cells from an unrelated strain of mice enables these contrary, these autoantibodies bind public epitopes animals to accept skin-grafts from the second strain, shared by all individuals belonging to a given species led both groups to conclude that during embryonic and even antigens that are well conserved during evo- life the immune system learns to tolerate self antigens. lution. However, this is also the characteristic of path- The clonal deletion theory was a magnificent theory ogenic autoantibodies observed in autoimmune dis- explaining tolerance and autoimmunity in a simple eases. For instance anti-red blood cell autoantibod- way. Autoimmunity, according to this theory, will only ies recognise public antigens, anti-DNA from SLE arise in the case of somatic mutations, which is an patients recognise human, rat and murine DNA, anti- unusual phenomenon. AChR autoantibodies recognise human and even fish However, in 1956 Witebsky and Rose were able, for receptors (for reviews see refs. #1-4). the first time, to induce an experimental autoimmune Thus, we have evolved from Ehrlich’s horror auto- disease mediated by autoantibodies: autoimmune toxicus notion, to Burnet’s forbidden clones hypothesis, thyroiditis. They succeeded in inducing this disease to reach, now, the view that autoimmunity is a nor- by injecting thyroglobulin in the presence of Freund mal physiological phenomenon. But, how can we rec- adjuvant. During the last several years, considerable oncile the experiments from Metchnikoff with the data have accumulated raising doubts concerning the experiments from Ehrlich and Landsteiner, since these clonal deletion theory as a general explanation for tol- results have never been challenged. erance since: 1) autoimmune diseases can be induced Experiments with transgenic mice allow us to inte- by injecting organ extracts; 2) numerous autoanti- grate all this experimental evidence. Nemazee’s group has created double transgenic mice expressing H-2Kk and anti-H2k transgenes.5 This is a very critical situa- Correspondence: Professor Guillaume Dighiero, Institut Pasteur, 25, tion, since the transgene is recognising a determinant Rue du Docteur Roux, Paris 75015, France. Phone: international +33- 1-45688212 – Fax: international +33-1-45688951 – e-mail: of polymorphism, that is a real-self antigen. Accord- [email protected] ing to Burnet’s prediction, B cells expressing the anti- 68 S. Havouis et al.

H2k transgene were stringently downregulated sialylated motifs can be observed: anti-Pr, anti-Sia-1b through programmed cell death (PCD) and receptor (anti-Gd), anti-Sia-b (anti-Fl) anti-Sia-l (anti-Vo).10,11 editing.5 However, in the case of Ig transgenes with CA disease is a rare disease (approximate incidence autoantibody activity, which are not directed against of 5 cases per million population) predominantly critical self polymorphisms, the transgene is not delet- occurring in the elderly. A substantial proportion of cas- ed; it is simply downregulated or energised. In double es develop in patients with B-cell lymphomas, Walden- transgenic mice expressing HEL and anti-HEL trans- ström’s macroglobulinaemia or chronic lymphocytic genes, B-cells were downregulated.6 B cells were also leukaemia. The fact that CA disease can also evolve in downregulated through energy and receptor editing two steps, the first one limited to AIHA, which subse- in transgenic mice expressing pathogenic anti-DNA quently can evolve to a tumoural disease, makes this transgenes.7 disease a paradigmatic model of the relationship These experiments allow us to understand the between autoimmune and malignant diseases. Pioneer apparent discrepancy between Ehrlich, Landsteiner work demonstrated that CA shared recurrent idiotopes and Metchnikoff’s research. Indeed, a subject express- and this was confirmed by the demonstration that anti- ing the A or B group will never produce typical autoan- I/i exclusively express the VH4-34 gene, frequently asso- tibodies against these determinants and no autoim- ciated with members of the VkIII family.12 Since the vast mune haemolytic anaemias displaying autoantibod- majority of CA express a unique VH gene, therapeutic ies with such specificities have been reported. Howev- strategies based on the preparation of anti-idiotypic er, the production of autoantibodies against public vaccines directed against recurrent idiotopes may be antigens, such as the I group, is a common phenom- devised and should allow the manufacture of vaccine enon. Haemolytic anaemia autoantibodies are main- that could be used in most cases of CA. This is partic- ly directed against these public antigens. So, the B cell ularly important becuase there is no effective therapy for repertoire that is going to be directed against poly- this disease. morphic determinants will be probably submitted to One of the major obstacles limiting progress in man- very stringent negative selection, i.e. deletion or strin- agement strategies for this disease is the absence of an gent receptor editing mechanisms, whilst the reper- animal model. In a previous work, we studied one toire that is directed against public determinants is human CA displaying the rare anti-Sia-1b (Gd) speci- probably not deleted and is an important component ficity (CAGAS) at the molecular level.13 In contrast to of the normal immune repertoire. classical anti-Ii CAs, which do not bind to mouse red blood cells, CAGAS agglutinates murine erythrocytes Construction of a murine model for (MRBC) with better affinity than it binds human red human cold agglutinin disease blood cells (HRBC) and is able to haemolyse MRBC Autoimmune haemolytic anaemia (AIHA) induced in the presence of complement. These characteristics by cold agglutinins (CA) is due to autoantibodies usu- make CAGAS a suitable candidate for the construc- ally expressing the µk isotype, that recognise carbohy- tion of a transfectoma and a transgenic model of drate epitopes on human red blood cells (RBC) and autoimmune haemolytic anaemia. GAS cause haemagglutination at temperatures below 37°C We have introduced CA VH and VL domains into and, optimally, in the cold (i.e. 4°C).8-11 The serum of eukaryotic expression vectors and transfected them healthy adult individuals frequently exhibits low titres into the non-secreting mouse myeloma X63 cell line. of CA. Transient pathological increase of CA levels Clones expressing complete engineered pentameric may occur during Mycoplasma pneumoniae infection and IgMk CAGAS (eCAGAS) recapitulating the characteristics a persistent increase as the result of a monoclonal B- of serum CA (sCAGAS), could be obtained. The i.p. cell expansion, which can be idiopathic or associated injection of eCAGAS to normal BALB/c mice induced a to a well-defined B cell malignancy. In addition to typical haemolytic anaemia, as demonstrated by the AIHA, CA can induce clinical manifestations such as presence of spontaneous cold agglutination of RBC, acropathy (due to RBC agglutination in peripheral ves- induction of anaemia and significant reticulocytosis. sels, where the temperature upon exposure to cold can Of interest, conspicuous bilateral ear loss was remain below 30°C), which can range from local observed in one of these animals. In addition, i.p. blood stasis to gangrene, in the more exposed to cold injection of X63 transfected line into BALB/c nude areas (extremities, , nose, etc).9,10 The thermal mice induced ascites, typical haemolytic anaemia and amplitude of the CA, defined as the titre at the high- shortening of the mean RBC survival. These findings est temperature causing RBC agglutination,9 corre- validated the practical interest of constructing a trans- lates better than its titre at 4°C with the pathogenic genic mouse model expressing eCAGAS.14 potential of a CA. Transgenic mice for either the CA.GAS heavy or light Different specificities of CA have been discerned, the chain gene were then produced. most frequent being anti-I/i, which recognises devel- Expression of the human H chain alone resulted in opmentally regulated polyglycosyl ceramides epitopes, a block in B lymphocyte maturation at the pro-B stage, that constitute the precursor molecule of the ABO and did not induce allelic exclusion. Double-trans- antigen system. Less frequently, CA directed against genic (dTg) mice, obtained by mating the mice of the Session #7 – Transgenic models of haematological diseases 69 two monotransgenic lines, produced significant We hope that these different approaches will lead to amounts of serum CA.GAS, though the agglutination the establishment of a full-blown CA disease, as was titre was insufficient to induce AIHA. In these mice, co- observed in the transfectoma model. expression of the human H and L chains mostly released the pro-B cell block but the majority of mature cells expressing the human IgM in the bone References marrow were subsequently eliminated by deletion, or 1. Dighiero G. Natural autoantibodies, tolerance and submitted to receptor editing. Hybridomas secreting autoimmunity. Ann NY Acad Sci 1997; 815:182-92. CA.GAS could be derived from splenic lymphocytes of 2. Avrameas S. Natural autoantibodies. From horror these mice, indicating the presence of splenic B cells autotoxicus to gnothi seauton. Immunol Today 1991; displaying this specificity. Interestingly, peritoneal cav- 12: 154-9. 3. Dighiero G, Lymberi P, Guilbert B, Ternynck T, ity B cells, were enriched in B cells co-expressing the Avrameas S. Natural auto-antibodies constitute a sub- complete transgene. In addition, a large proportion stantial part of normal circulating immunoglobulins. of splenic B cells co-expressed the human H chain with Ann NY Acad Sci 1986; 475:135-45. a murine one, and probably secreted mixed IgM with 4. Bouvet JP, Dighiero G. From natural polyreactive a lower than expected agglutination capacity. autoantibodies to “à la carte” monoreactive antibod- ies to infectious agents: is it a small world after all? Indeed, this model demonstrates that CA.GAS is Infect Immunol 1998; 6:1-4. definitively pathogenic for mice, thus constituting the 5. Nemazee DA, Burki K. Clonal deletion of B lympho- first reported case of a CA pathogenic for both human cytes in a transgenic mouse bearing anti-MHC class I and mouse. When transgenic mice were constructed, antibody genes. Nature (Lond) 1989; 337:562-6. 6. Goodnow C, Brink R, Adams E. Breakdown of self-tol- it could be demonstrated that despite significant dele- erance in anergic B lymphocytes. Nature (Lond) 1991; tion and receptor editing, the transgenic clone secret- 352: 532-6. ing the CA still secreted significant amounts of the CA, 7. Erikson J, Radic MA, Camper SA, Hardy RR, Carmack although not sufficient to induce the disease. C, Weigert M. Expression of anti-DNA immunoglobu- lin transgenes in non-autoimmune mice. Nature (Lond) We are trying to induce the disease using the fol- 1991; 349:331-5. lowing strategies: 8. Williams RC, Kunkel HG, Capra JD. Antigenic speci- a) mating the double transgenic mice, secreting the ficities related to the cold agglutinin activity of gamma CA at levels insufficient to induce the disease, with M globulins. Science 1968; 161:379-84. 9. Pruzanski W, Shumak KH. Biologic activity of cold- Bcl-2 transgenic mice. The anti-apoptotic effect of reacting autoantibodies. N Engl J Med 1977; 297:538- the Bcl-2 transgene may increase the survival of B 42. cells expressing the transgene. In addition, as lym- 10. Roelcke D. Sialic acid-dependent red blood cell anti- phoma tumours occur in this type of mouse, a lym- gens. In: G. Garraty, ed. Immunobiology of transfusion phoma expressing the CA transgene may be medicine. New York: Marcel Dekker, Inc. 1993. p. 69- 95. obtained. In that case, the disease should be 11. Silverman GJ, Chen PP, Carson DA. Cold agglutinins: observed; specificity, idiotype and structural analysis. In: DA Car- b) since Mycoplasma has been reported to induce the son, PP Chen, TJ Kipps, eds. Idiotypes in Biology and disease in humans, through polyclonal stimulation Medicine. Basel: Karger, 1990. Chem Immunol 48:109- 25. of the normal B cell population producing CA, work 12. Pascual V, Victor K, Lelsz D, et al. Nucleotide sequence is in progress in our laboratory to infect the trans- analysis of the V region of two IgM cold agglutinins. Evi- genic animals with this micro-organism; dence that the VH4-21 gene segment is responsible for c) LPS has been reported to be able to rescue imma- the major cross-reactive idiotype. J Immunol 1991; 146: 4385-92. ture B cells in the bone marrow and allow them to 13. De la Fuente MA, Egile C, Pereira A, et al. Molecular migrate to the peripheral compartment. Since in characterization of a monoclonal IgMk (GAS) cold this transgenic model, a significant number of B agglutinin (CA) of anti-Sia-Ib (anti-Gd) specificity that cells are deleted in the bone marrow, it is reason- co-existed with a clonally related monoclonal IgG3k (GAS) lacking CA activity. Blood 1994; 83:1310-22. able to test whether LPS is able to rescue these cells 14. Dumas G, Pritsch O, Pereira A, Gallart T, Dighiero G. and to allow their migration to the peripheral com- A murine model of human cold agglutinin disease. Br partment. J Haematol 1997; 98:589-96. Educational Session 8 Haematologica 1999; 84:(EHA-4 educational book):70-71 Chairman: F. Mandelli Acute promyelocytic leukaemia

The APL-associated fusion proteins SAVERIO MINUCCI, MARIO CIOCE, MARCO MACCARANA, PIER GIUSEPPE PELICCI European Institute of Oncology, Department of Experimental Oncology, Milan, Italy

cute promyelocytic leukaemia (APL) has RAR␣-fusion proteins affect differentiation aroused interest well outside the haematolog- Despite clinical similarities, RA induces differentia- Aical field during the last ten years. Two fea- tion of leukaemic blasts and disease remission only in tures, both of which are unique to APL, have attract- PML/RAR␣ APLs whereas PLZF/RAR␣ APLs are ATRA ed the attention of various sectors of biomedical resistant.1 Recent experimental evidence, obtained research: i) the remission of the disease obtained with PML/RAR␣ and PLZF/RAR␣ model systems, has with retinoic acid (RA) treatment, whose mechanism shown that RAR␣-fusion proteins interfere directly of action consists of inducing the APL blasts to dif- with the programme of terminal differentiation and ferentiate; ii) the presence in the APL blasts of fusion are involved in both disease pathogenesis and proteins that involve one of the retinoic acid recep- response to RA: i) before the onset of leukaemias, the tors (RAR).1 RAR␣ is a member of the super-family PML/RAR␣ transgenic mice show a pre-leukaemic of nuclear hormone receptors that are involved in condition characterised by increased and poorly dif- development and differentiation.2 In the last 5 years ferentiated haematopoietic precursors in the bone- ␣ ␣ crucial insights have been gained into the issues con- marrow; ii) expression of PML/RAR or PLZF/RAR nected with the molecular basis of APL and RA- in haematopoietic precursor cell lines blocks differ- entiation induced by physiological stimuli;3,4 iii) RA treatment, raising further questions about the cellu- ␣ lar and molecular mechanisms underlying the induces differentiation of PML/RAR –expressing ␣ cells, but not of PLZF/RAR␣-, both in the patients leukaemogenic activities of RAR -fusion proteins ␣ and the physiological functions of the RAR␣-translo- and in transgenic mice; iv) expression of PML/RAR , but not PLZF/RAR␣, in resistant or poorly RA-sensi- cation partners. Present knowledge, however, allows tive cell lines restores a differentiative response to us to speculate that understanding of such mecha- RA.3,4 It appears, therefore, that PML/RAR␣ and nisms will not only be relevant to APL, but to the PLZF/RAR␣ block differentiation at physiological processes of leukaemogenesis and growth/differen- concentrations of RA, while only PML/RAR␣ favours tiation control in general. it at pharmacological doses. ␣ The RAR -fusion proteins are leukaemogenic RAR␣-fusion proteins regulate transcription APL is cytogenetically characterised by a reciprocal RA-induced degradation of PML/RAR␣, which translocation that constantly involves chromosome occurs through the activation of caspase and pro- ␣ 17, which breaks within the locus encoding for RAR . teosomal pathways, has been proposed as a critical Usually the chromosome partner is chromosome 15, mechanism accounting for the response of APL blasts with the break located within the PML locus, and less to this agent. However, the inhibition of the frequently, chromosomes 11 and 5 with breaks in the PML/RAR␣ proteolysis obtained by using various PLZF or NuMa and NPM loci, respectively. The hybrid caspase inhibitor peptides does not impair the RA genes so formed encode a PML/RAR␣, PLZF/RAR␣, effect on APL blast differentiation, indicating that NuMa/RAR␣ or NPM/RAR␣ fusion protein, all of PML/RAR␣ is actively involved in conferring ATRA which retain the same portion of RAR␣.1 The RAR␣- sensitivity. translocations are primary chromosome aberrations The potential of PML/RAR␣ and PLZF/RAR␣ to and are often the only cytogenetic anomaly in the neo- interfere with haematopoietic differentiation and to plastic metaphases. Experimental evidences for mediate RA sensitivity is thought to involve transcrip- leukaemogenic potential is, however, only available tional regulation of RA target genes. Retinoic acid for PML/RAR␣ and PLZF/RAR␣. Mice transgenic for receptors are transcription factors involved in the con- PML/RAR␣ and PLZF/RAR␣ manifest myeloid differ- trol of terminal myeloid differentiation. They behave entiative alterations with the phenotypic features of as ligand-dependent transcriptional regulators, promyelocytic leukaemia. repressing transcription in the absence of ligand and activating transcription in its presence. The different effects on transcription are carried out through recruitment of coregulators: unliganded receptors Correspondence: Pier Giuseppe Pelicci. European Institute of Oncolo- gy, Department of Experimental Oncology, 20141 Milan, Italy. tel: bind corepressors (NCoR and SMRT) that are found (0)2-57489831; fax: (0)2-57489851; e-mail: [email protected] within a complex containing histone deacetylase Session #8 – Acute promyelocytic leukaemia 71

(HDAC) activity, whereas liganded receptors recruit ments (nuclear bodies; NB) and this localisation is coactivators with histone acetylase activity (HATs). apparently crucial to their function. Notably, the NB Chromatin remodelling activities (such as the NURD structure is destroyed in APL cells and is restored by and hBRG1 complexes) have also shown to be ATRA treatment. ATRA induced re-assembly of the required for transcriptional regulation by retinoid PML-nuclear bodies has been recently shown to be a receptors and other members of the nuclear hormone direct consequence of the PML/RAR␣ degradation fol- receptor superfamily, suggesting a hierarchy of pro- lowing ATRA treatment. Thus, PML/RAR␣ may alter moter structure modifications in RA target genes car- the PML pathway by interfering with the function of ried out by multiple co-regulatory complexes.2,5 ␣ ␣ nuclear bodies. PLZF is another growth suppressor Both PML/RAR and PLZF/RAR retain the abili- which localises within nuclear bodies.9 It may, there- ty of RAR␣ to regulate transcription of RA-target ␣ ␣ fore, be that the nuclear bodies are involved in growth genes. PML/RAR and PLZF/RAR fusion proteins control and that their deregulation is a constant fea- recruit the N-CoR/HD complex through their RAR␣ ture of RAR␣-fusion proteins. moiety.6 PLZF/RAR␣ contains a second, ATRA-resis- tant, binding site in the PLZF N-terminal region. High ␣ doses of ATRA release HD activity from PML/RAR , References but not from PLZF/RAR␣. Mutation of the N-CoR ␣ binding site abrogates the ability of PML/RAR to 1. Grignani F, Fagioli M, Alcalay M, et al. Acute promye- block differentiation, whereas inhibition of HD activ- locytic leukemia: from genetics to treatment. Blood ity switches the transcriptional and biological effects 1994; 83:10-25. of PLZF/RAR␣ from inhibitor to activator of the RA- 2. Minucci S, Ozato K. Retinoid receptors in transcrip- signalling pathway. Therefore, recruitment of HD and tional regulation. Curr Opin Gen Dev 1996; 6: 567-74. regulation of RA-target genes are crucial to the trans- 3. Grignani F, Ferrucci PF, Testa U, et al. The acute promyelocytic leukaemia specific PML/RAR␣ fusion forming potential of APL-fusion proteins while the dif- protein inhibits differentiation and promotes survival ferent effects of ATRA on the stability of the of myeloid precursor cells. Cell 1993; 74: 423-31. PML/RAR␣– and PLZF/RAR␣–corepressor complex- 4. Ruthardt M, Testa U, Nervi C, et al. Opposite effects es determine the differential response of APLs to of the acute promyelocytic leukaemia PML/RAR␣ and ATRA. PLZF/RAR␣ fusion proteins on retinoic acid signalling. Mol Cell Biol 1997; 17:4859-69. RAR␣ fusion proteins affect survival 5 Grunstein M. Histone acetylation in chromatin struc- An additional biological activity which may con- ture and transcription. Nature (Lond) 1997; 389: 349- tribute to the leukaemogenic potential of RAR␣ fusion 52. 6. Grignani F, De Matteis S, Nervi C, et al. Fusion pro- proteins is their interference with cell survival. Expres- ␣ ␣ teins of the retinoic acid receptor- recruit histone sion of PML/RAR in haematopoietic cells inhibits deacetylase in promyelocytic leukemia. Nature (Lond) programmed cell death, while expression of the fusion 1998; 391;815-8. protein into non-haematopoietic cells induces apop- 7. Ferrucci PF, Grignani F, Pearson M, Fagioli M, Nico- tosis.7 These effects depend on the integrity of the letti I, Pelicci PG. Cell death induction by the acute ␣ incorporated PML sequences, and may be the conse- promyelocytic leukemia-specific PML/RAR fusion ␣ protein. Proc Natl Acad Sci USA 1997; 94:10901-6. quence of the interference of PML/RAR with the 8. Wang ZG, Delva L, Gaboli M, et al. Role of PML in cell growth-suppressive function of the wild-type PML pro- growth and the retinoic acid pathway. Science 1998; tein. Indeed, forced expression of PML in a variety of 279: 1547-51. cell lines induces growth arrest through mechanisms 9. Ruthardt M, Orleth A, Tomassoni L, et al. The acute which involve induction of apoptosis; targeted dis- promyelocytic leukemia specific PML and PLZF pro- ruption of the PML locus in mice increases the rate of teins localize to adjacent and functionally distinct spontaneous or induced carcinogenesis.8 However, nuclear bodies. Oncogene 1998; 16:1945-53. ␣ 10. Lin R, Nagy L, Inoue S, Shao W, Miller WH, Evans the molecular mechanisms through which PML/RAR RM. Role of the histone deacetylase complex in acute deregulates the PML intracellular pathways are not promyelocytic leukemia. Nature (Lond) 1998; 391; clear. PML localises within distinct nuclear compart- 811-4. Educational Session 8 Haematologica 1999; 84:(EHA-4 educational book):72-74 Chairman: F. Mandelli Acute promyelocytic leukaemia

Diagnosis, front line treatment and molecular monitoring of acute promyelocytic leukaemia FRANCESCO LO COCO, DANIELA DIVERIO, GIUSEPPE AVVISATI, FRANCO MANDELLI Department of Cellular Biotechnologies and Haematology, University “La Sapienza” of Rome, Italy

Diagnosis In cases with the typical hypergranular morpholo- A unique t(15;17) chromosome aberration result- gy, however, treatment initiation need not be post- ing in the PML/RAR␣ gene fusion and an exquisite poned pending the results of genetic studies. These sensitivity to the differentiating agent all-trans retinoic patients may in fact immediately receive specific ther- acid (ATRA) are the main distinguishing features of apy and ATRA could be subsequently withdrawn in acute promyelocytic leukaemia (APL). Given in com- those rare cases lacking the genetic abnormality bination with anthracycline-containing chemo- despite having typical morphology. If not available therapy, ATRA has been shown to provide the best on site, RT-PCR and karyotypic characterisation treatment results in this disease. Because the presence might be requested from specialised centres in order of the PML/RAR␣ hybrid in leukaemic cells is known to obtain further potentially relevant information to predict response to ATRA, and due to frequent life- (e.g. additional chromosome abnormalities, variant threatening haemorrhagic diatheses, rapid diagnosis translocations, PML/RAR␣ isoform type) and to at the genetic level is recommended for promptly ini- define the correct RT-PCR strategy to be used in the tiating APL-tailored therapy.1-4 individual patient for minimal residual disease Demonstration of the disease’s genetic hallmark (MRD) monitoring. can be carried out at chromosome, protein, DNA or RNA levels, using conventional karyotyping and/or Front line treatment fluorescent in situ hybridisation (FISH), anti-PML anti- Following the advent of ATRA, large trials using bodies, Southern blotting and reverse-transcriptase this agent in variable combinations with chemother- polymerase chain reaction (RT-PCR), respectively. apy (CHT) for newly diagnosed APL have been car- Each of these procedures has its own advantages and ried out in Europe (APL ’93, GIMEMA, and PETHE- pitfalls (Table 1). In particular, karyotyping may MA studies), the USA (New York Memorial Sloan- occasionally give false-negative results [i.e. absence of Kettering Cancer Center, US Intergroup and MD t(15;17) in cases later found to contain cryptic Anderson studies), the UK (MRC), China and Japan ␣ PML/RAR rearrangement] and like Southern blot- (JALSG). Overall, nearly 2,000 patients were includ- ting is time-consuming, requiring a few days for exe- ed in these trials.3 The advantage of ATRA inclusion cution. RT-PCR allows a rapid and highly sensitive in front line treatment was established in 1993 by the diagnosis, but it is prone to artefacts and technical- APL European group in a randomised study com- ly difficult if not performed in experienced laborato- paring ATRA followed by CHT (ATRA→CHT) with ries. Recently, immunohistochemical analysis of PML CHT alone.6 Subsequent studies were aimed in most staining with monoclonal antibodies has proven use- cases at establishing the optimal ATRA and CHT ful for specific diagnosis, which is established fol- combination. Other issues which were addressed lowing the identification of the so-called microspeckled were the importance of genetic diagnosis and the role PML protein distribution consequent to the translo- of maintenance therapy including or not ATRA. The cation.5 In light of its widespread availability for main results may be summarised as follows: rapid, specific and low cost diagnosis, this procedure 1. resistant leukaemia to regimens variably combin- might be recommended as a convenient tool for iden- ing ATRA and CHT is virtually absent in patients tification of the disease’s hallmark, particularly in with genetically confirmed diagnosis. Failure to centres not equipped or trained for more sophisti- cated analyses. Apart from confirming morphologi- achieve haematological complete remission (CR, cally typical APLs, this assay would clarify diagnosis reported in 7-20% of cases) is mainly due to early in cases with uncertain cytological features and its death from haemorrhage or infectious complica- application could be extended to all acute myeloid tions; leukaemias (AMLs). 2. the simultaneous ATRA+CHT combination pro- vides the best results in terms of CR and disease- free survival (DFS), and seems more effective in Correspondence: Francesco Lo Coco, M.D. Department of Cellular diminishing the occurrence of overt ATRA syn- Biotechnologies and Haematology, University La Sapienza, via Benevento 6, 00161 Rome, Italy. Phone: international +39.06.85795530 – Fax: drome. This latter is also successfully counteract- international +39.06.44241984 – e-mail: [email protected] ed, however, by strict adherence to the recom- Session #8 – Acute promyelocytic leukaemia 73

Table 1. Technical approaches for genetic diagnosis in APL. CHT vs. ATRA+CHT vs. observation. The final analysis of these studies should be available soon. ATRA therapy is given intermittently in these two Technique Main advantages Main pitfalls trials (i.e. for 15 days every 3 months), since con- tinuous administration is known to result in sig- Karyotype Additional genetic Frequent poor quality information (lesions metaphases and cryptic nificant toxicity and more frequent development of other than t15;17) translocations (false resistance; negative) 5. the prognostic outcome of elderly APL patients Southern blot Specific Time consuming; has dramatically improved after the advent of hybridisation with > 1 ATRA. In fact, complete remission and event-free probe is often required survival rates of >80% and >50% have been report- RT-PCR Highly sensitive Frequent poor RNA yield ed in recent trials for patients aged >60. and amplification artefacts Molecular assessment of response and Immunohistochemistry Rapid, simple Provides no information MRD monitoring (PML-staining pattern) and low-cost on the PML breakpoint Besides refining diagnosis, PML/RAR␣ RT-PCR type studies offer the additional advantage of identifying patients with distinct transcript isoforms (bcr1 or long type, bcr2 or variable type and bcr3 or short type) and, particularly, of providing a sensitive tool to mendations established by the Memorial Sloan- assess response to treatment. Due to technical diffi- Kettering group,7 i.e. rapid institution of corticos- culties in distinguishing long and variable forms, in teroids at the earliest manifestation of symptoms most reported studies these two types are included in related to the syndrome (respiratory distress, unex- a common group referred to as long transcript, and plained weight gain, fever and pleuro-pericardial this latter compared to the short type. Contrary to effusion). Death rates attributed to the ATRA syn- earlier reports indicating a worse outcome in patients drome have dropped to 1-3% in the most recent with the short form, recent trials have shown no sig- studies; 3. the benefit of cytarabine administered in addition nificant differences in DFS. It has to be emphasised, to anthracyclines during induction and/or consol- however, that the former analyses were done in idation is unclear. Studies in which cytarabine has patients treated with ATRA alone for remission induc- been omitted from induction (GIMEMA) or from tion, whereas the latter derive from trials in which both induction and consolidation (MD Anderson patients received CHT added to ATRA. and PETHEMA) have reported results compara- Several groups have analysed the response to treat- ble or even superior (at least in terms of reduced ment by RT-PCR tests performed at various times dur- toxicity) to those obtained in studies including this ing and after treatment. Using assays with sensitivity –4 agent. Whilst the effectiveness of anthracycline- levels of approximately 10 , up to 50% of patients in based monochemotherapy for remission induc- haematological remission after induction have ␣ tion of APL was established as long as 26 years detectable PML/RAR transcript in their marrow ago by the pioneering work of Jean Bernard,8 the cells. No correlations were found between PCR status omission of cytarabine from overall treatment (i.e. at the time of remission achievement and relapse risk including consolidation) is a recent attempt and in the GIMEMA and MRC studies. The vast majority long-term results of the MD Anderson and PETHE- (up to 95%) of patients tested after completion of MA groups are awaited to resolve this issue. How- consolidation were PCR-negative in the New York ever, no study has as yet compared different con- Memorial Sloan-Kettering, MRC, GIMEMA and solidation options including or not cytarabine in PETHEMA series.3 Interestingly, detection of residual phase III trials; disease at the end of the third CHT course (of 4 giv- 4. two large randomised studies (APL’93 and US en) predicted an increased relapse risk associated with Intergroup) have demonstrated a benefit conferred poorer survival in the MRC study.9 by ATRA-containing maintenance therapy. This is The clinical value of post-treatment molecular mon- also confirmed in the analysis of the GIMEMA itoring has been emphasised in a recently reported study which adopted the same randomisation prospective study of the GIMEMA group.10 This designed by the European APL ’93 study (unpub- showed that conversion from PCR-negativity to PCR- lished observations). The advantage of using CHT positivity after consolidation therapy is almost always maintenance with methotrexate and 6-mercap- associated with subsequent haematological relapse. topurine had been suggested in France and the Based on these results, patients in this trial who con- USA in the past decade; hence, both the APL ’93 vert to PCR-positivity in two successive marrow sam- and the GIMEMA groups included four randomi- ples are now defined as being in molecular relapse sation arms in their studies to compare ATRA vs. and given anticipated salvage treatment. 74 F. Lo Coco et al.

Conclusions and future perspectives References Despite considerable improvement in diagnosis and management, a sizeable proportion of APL patients 1. Chen S-J, Wang Z-Y, Chen Z. Acute promyelocytic who receive state-of-art front line treatment still die of leukemia: from clinic to molecular biology. Stem Cells early complications or following disease recurrence. 1995; 13:22-31. With regard to early death, a special effort should be 2. Fenaux P, Chomienne C, Degos L. Acute promyelo- cytic leukemia: Biology and treatment. Semin Oncol made to define risk categories better (hyperleukocy- 1997; 24:92-102. tosis, older age, severity of the coagulopathy, other 3. Lo Coco F, Nervi C, Avvisati G, Mandelli F. Acute unknown factors) in order to promptly reinforce ade- promyelocytic leukemia: a curable disease. Leukemia quate supportive care and to evaluate the feasibility 1998; 12: 1866-80. 4. Warrell RP Jr. Pathogenesis and management of acute of a tailored (less intensive?) initial treatment. promyelocytic leukemia. Annu Rev Med 1996; 47: At least 60% of newly diagnosed patients become 555-65. long-term survivors and are probably cured with 5. Falini B, Flenghi L, Fagioli M, et al. Immunocyto- ATRA+CHT. It is conceivable to hypothesise that a chemical diagnosis of acute promyelocytic leukemia relevant fraction of these patients are being overtreat- (M3) with the anti-PML monoclonal antibody PG- M3. Blood 1997; 90:4046-53. ed and that they might, therefore, be spared this risk. 6. Fenaux P, Le Deley MC, Castaigne S, et al. Effect of All- On the other hand, we are still unable to identify trans retinoic acid in newly diagnosed acute promye- those cases (approximately 20%) who will ultimately locytic leukemia. Blood 1993; 82:3241-9. relapse after initial therapy and are, therefore, in need 7. Frankel SR, Eardley A, Heller G, et al. All-trans retinoic acid for acute promyelocytic leukemia. Results of the of treatment intensification. The definition of risk cat- New York study. Ann Intern Med 1995; 120:278-86. egories at diagnosis for more appropriate treatment 8. Bernard J, Weil M, Boiron M, Jacquillat C, Flandrin G, stratification remains a big challenge for future clini- Gemon MF. Acute promyelocytic leukemia. Results of cal investigation in this disease. Finally, the place of treatment with daunorubicin. Blood 1973; 41:489- 96. novel drugs which have proven effective in relapse, 9. Burnett AK, Grimwade D, Solomon E, Wheatley K, such as arsenic trioxide, and the advantage of antici- Goldstone AH. Presenting white cell count and kinet- pating salvage treatment at the time of molecular ics of molecular remission predict prognosis in acute recurrence remain to be established. promyelocytic leukemia treated with all-trans retinoic acid: Results of the randomised MRC trial. Blood Acknowledgements 1999; in press. Supported by A.I.L. (Associazione Italiana contro le 10. Diverio D, Rossi V, Avvisati G, et al. Early detection of relapse by prospective RT-PCR analysis of the Leucemie), Italy. PML/RAR␣ fusion gene in patients with acute promye- locytic leukemia enrolled in the GIMEMA-AIEOP mul- ticenter “AIDA” trial. Blood 1998; 92:784-9. Educational Session 8 Haematologica 1999; 84:(EHA-4 educational book):75-77 Chairman: F. Mandelli Acute promyelocytic leukaemia

Arsenicals and inhibitors of histone deacetylase as anticancer therapy RAYMOND P. WARRELL, JR. Developmental Chemotherapy and Leukaemia Services, Memorial Sloan-Kettering Cancer Centre and the Cornell University Medical College, New York, NY, USA

everal new therapeutic approaches have recent- breadth of this activity, we then initiated a clinical ly been clinically tested in patients with study to evaluate the pharmacokinetics, safety, and Sadvanced cancer, including the use of arseni- potential efficacy of melarsoprol in patients with cals and various drugs that affect gene transcription relapsed leukaemia. Using the anti-trypanosomal via inhibition of histone deacetylase. Although our dose and schedule, patients received escalating intra- interest in these agents arose from our programme venous doses daily for 3 days, repeated weekly for 3 in acute promyelocytic leukaemia (APL), we expect weeks. Doses were 1 mg/kg on day 1, 2 mg/kg on that these agents will prove useful for a number of day 2, and 3.6 mg/kg on day 3 and on all days there- other cancers. The following pages summarise some after, up to a maximum daily dose of 200 mg. Eight aspects of our group’s clinical work in these areas. patients [5 AML, 1 APL, 1 CML, and 1 CLL] were treated. Mean peak plasma concentrations of the Arsenicals as cancer treatment parent drug were obtained immediately after injec- Like many others, we were intrigued by reports tion. These concentrations ranged from 1.2 µg/mL emanating from the People’s Republic of China in on day 1 to 2.4 µg/mL on day 3, were dose propor- 1 ␤ ≅ the early 1990s that intravenous arsenic had re- tional, and decayed with a t /2 15 minutes. A emerged as a cancer treatment. Arsenic has been minor clinical response (regression of splenomegaly used as a medicinal for many centuries, both in top- and lymphadenopathy) was observed in a patient ical and injectable forms (as escharotic agents) as with chronic lymphocytic leukaemia. One patient well as oral forms that were first described in the with t(11;17)-variant APL who had been refractory to 1700s. Fowler’s solution given orally was used in the chemotherapy and all-trans retinoic acid had slight 19th and early 20th centuries as a treatment to reduce improvements in both his leukocyte and platelet extreme leukocytosis in patients with chronic myelo- counts with the first course of treatment. He then cytic leukaemia (CML). In the West, however, this received a second 3-week course without further use dwindled after the widespread application of improvement and was removed from the study. radiotherapy in the 1930s and the advent of other Unfortunately, central (CNS) toxi- cytotoxic drugs after the 2nd World War. city proved limiting. Three patients experienced gen- The recent upsurge of interest stems from reports eralised grand mal seizures during the second week of from Harbin China that an intravenously injectable therapy. Other neurologic reactions included vertigo formulation induced remissions in patients with (2 patients) and paraesthesia (1 patient), both of APL.1 Since we perceived an enormous regulatory which resolved after discontinuing therapy. The hurdle to direct application of this therapy in the patient who received two complete treatment cycles United States, we first tested an organic arsenical, experienced moderately severe peripheral neuropa- melarsoprol, which was already formulated for thy, with bilateral lower extremity paraesthesia that caused moderate functional impairment, but which human use and was immediately available on an resolved approximately 5 weeks after the last dose. investigational basis because of its use for the treat- Other common reactions were nausea, vomiting, and ment of African trypanosomiasis due to T. bruceii. irritation at the injection site. We concluded that the antitrypanosomal dosing schedule of melarsoprol is Laboratory and clinical studies of melarsoprol associated with excessive CNS toxicity, and that ver- We evaluated both As2O3 and melarsoprol for pos- ification of the striking preclinical activity in patients sible antileukaemic activity in vitro. In these studies, with leukaemia requires development of an alterna- melarsoprol exhibited broad antileukaemic activity tive dosing schedule. against both myeloid and lymphoid cells. Given the Arsenic trioxide We and others have recently confirmed that arsenic Correspondence: Raymond P. Warrell, Jr. MD, Memorial Sloan-Ketter- trioxide induces complete remissions (CR) in a high ing Cancer Center, 1275 York Avenue, New York, NY 10021, USA. Tel. 2-5 international +1-212-6398168 – Fax. international +1-212-717-3215 proportion of patients with APL. In our initial pilot – e-mail: [email protected] study, 12 heavily pretreated patients were treated 76 R. P. Warrell, Jr. with arsenic trioxide at doses ranging from 0.06 to APL patients who receive all-trans RA for induction 0.2 mg/kg/day until leukaemic cells were eliminated develop one or more signs of the RA syndrome, a poten- from the marrow. Patients who achieved CR were eli- tially lethal complication. The RA syndrome is a loose- gible to receive up to 5 additional courses of therapy, ly defined constellation of problems that include fever, with each course given for a cumulative total of 25 weight gain, dyspnoea, pulmonary oedema, pleural days at a daily dose of 0.15 mg/kg/d every 3-6 weeks. effusions, musculoskeletal pain, or effusions. Marrow leukaemic cells were eliminated in 11 of 12 Signs/symptoms of the RA syndrome were observed in patients after a median treatment duration of 33 days 8 of the 23 patients (35%) treated with arsenic triox- (range, 12-39 days)(one patient who had an intracra- ide. Several patients were presumptively treated with nial haemorrhage on day 1 died on day 5). The medi- high-dose dexamethasone. All patients with this prob- an cumulative dose during induction was 360 mg lem have recovered, and interruption of arsenic ther- (range, 160 to 515 mg). CR by all criteria was apy was not required. These data show that even non- attained at a median time of 47 days (range, 24-83 terminal cytodifferentiation induced by arsenic trioxide d). Impressively, 8 of 11 patients tested converted RT- is associated with induction of leukocytosis and the PCR assays for PML/RAR-␣ to negative at or before RA syndrome in APL. These effects appear to be relat- the end of the second course, while 3 patients who ed to intrinsic biological responsiveness to differenti- remained positive relapsed early and appeared to ating agents. 5 have rapidly developed arsenic resistance. Several Pharmacokinetics of arsenic trioxide patients have received up to 6 cycles of arsenic triox- and studies in other cancers ide therapy without showing evidence of cumulative We have conducted several dose-ranging studies to toxicity. examine the pharmacokinetics and biological effects We subsequently initiated a confirmatory multicen- of arsenic trioxide in patients with other types of tre study in patients with relapsed or refractory APL haematologic cancers and solid tumours. Pharmaco- using the same treatment schedules at a fixed daily kinetic analysis of blood and urine for elemental dose of 0.15 mg/kg/d. To date, 28 patients have been arsenic content showed that arsenic is distributed in accrued, and 19 of 21 evaluable patients (90%) have both plasma and red blood cell fractions of whole achieved complete remission. A new study of arsenic blood. Parallel elimination curves suggest that these trioxide plus all-trans RA for induction therapy of new- 2 compartments are freely exchangeable. The mean ly diagnosed patients with APL will begin shortly. area under the plasma ϫ concentration curve (AUC) Adverse reactions to arsenic trioxide on day 1 after a dose of 0.15 mg/kg was ~ 400 Adverse effects due to arsenic trioxide have gener- ng.hr/mL. Approximately 20% of the administered ally been mild. These reactions have included skin dose was recovered in urine within the first 24 hrs. rash, light headache during the infusion, fatigue, mild The terminal half-life appears to be quite prolonged. hyperglycaemia, musculoskeletal pain, and EKG In an ongoing dose-ranging study in patients with changes (particularly QT prolongation). In the past, haematologic cancers other than APL, we have all-trans retinoic acid (RA) was shown to induce ter- employed a daily IV dosing schedule for a cumulative minal differentiation of APL cells in vivo, and leukocy- total of 25 days per treatment course, with addition- tosis was commonly observed during this process, al cycles administered every 3-5 weeks. Dose levels probably as a result of a transient increase in have ranged from 0.1 to 0.25 mg/kg/day. A separate leukaemic cell lifespan. Similarly, previous studies by study in patients with solid tumours is using a daily ϫ us and others have shown that arsenic treatment is 5 days schedule, with additional cycles in responding associated with induction of partial, non-terminal patients administered every 4 weeks. Doses in that immunophenotypic cytodifferentiation. study have ranged from 0.15 to 0.3 mg/kg/day. Over In clinical studies of APL patients treated with this range of doses, the drug has continued to be rel- arsenic trioxide at this centre, leukocytosis (defined atively well-tolerated. However, skin rash, diarrhoea, as a total peripheral blood leukocyte count ≥ 10ϫ109 QTc prolongation on EKG, and fatigue (but not leuko- cells/L) was observed in 13 of 23 patients (57%). The cytosis or RA syndrome) have emerged as increasing- median baseline leukocyte count for all treated ly prominent side-effects in these other diseases. patients (2.6ϫ109/L [range, 0.7-65.2]) was not sub- stantially different from the cohort of patients who Targeting of histone deacetylase as subsequently developed leukocytosis (3.5ϫ109/L anticancer therapies [range, 1.3-65.2). The median peak leukocyte count Acetylation of DNA-associated histones is linked to of the latter group was 60.3ϫ109/L (range, 17.1- gene-specific activation, whereas histone deacetyla- 247.0), which occurred at a median of 19 days from tion is associated with transcriptional repression.6 the start of arsenic therapy (range, 4-24 days). No Recent studies have shown that inhibitors of histone other cytotoxic therapy was administered, and the deacetylase (HDAC) can relieve transcriptional repres- leukocytosis resolved in all patients. sion caused by certain oncogenes.7-9 These inhibitors Distinct from leukocytosis, approximately 50% of include trichostatin A, trapoxin, depudecins, bicyclic Session #8 – Acute promyelocytic leukaemia 77 depsipeptides, hybrid polar compounds and various ja, and Peter Maslak for their contributions to these efforts. butyrates. Supported in part by CA-77136 and CA-09207 from the Since sodium phenylbutyrate was immediately National Cancer Institute, FD-R-001364 from the Orphan available for investigational use in human subjects, Products Development, Food and Drug Administration, EDT- we tested whether this drug, as well as the concept of 83025 from the American Cancer Society, and by grants from HDAC inhibition, could be effective in vitro and then the Lymphoma Foundation and PolaRx Biopharmaceuticals, Inc. clinically in patients with highly resistant APL. Having documented that sodium phenylbutyrate (PB) inhib- ited HDAC, we found that high concentrations of PB alone decreased cell growth but had no effect upon References differentiation of NB4 cells; however, the combina- tion with all-trans RA caused a substantial increase in 1. Sun HD, Ma L, Hu XC, Zhang TD. Treatment of acute differentiation. promyelocytic leukemia by Ailing-1 therapy with use We then treated a child who had relapsed after of syndrome differentiation of traditional Chinese medicine. Chim J Comb Trad Chin Med West Med numerous cytotoxic drugs, all-trans RA, allogeneic 1992; 12:170-1. bone marrow transplantation, and arsenic trioxide. 2. Chen G-Q, Zhu J, Shi X-G, et al. In vitro studies on cel- Blood and marrow mononuclear cells were assayed lular and molecular mechanisms of arsenic trioxide for histone hyperacetylation by immunohistochem- (As2O3) in the treatment of acute promyelocytic istry and Western blot analysis. Minimal residual dis- leukemia: As2O3 induces NB4 cell apoptosis with ease was assessed by RT-PCR for PML/RAR-␣. After downregulation of Bcl-2 expression and modulation reconfirming that the patient was clinically resistant of PML-RAR␣/Pml proteins. Blood 1996; 88:1052- to RA, PB was added to the treatment regimen. The 61. RA dose ranged from 45-90 mg/m2/d in 2 equally 3. Shen Z-X, Chen G-Q, Ni J-H, Li XS, Xiong SM, Qiu QY, divided doses; the PB dose ranged from 150-210 et al. Use of arsenic trioxide (As2O3) in the treatment mg/kg IV BID infused IV over 1 hr. By day 23, the of acute promyelocytic leukemia (APL): II. Clinical effi- patient’s marrow showed complete elimination of vis- cacy and pharmacokinetics in patients at relapse. Blood 1997; 89:3354-60. ible leukaemic cells, and she achieved a complete clin- 4. Zhu J, Koken MH, Quignon F, et al. Arsenic-induced ical and cytogenetic remission shortly thereafter. By PML targeting onto nuclear bodies: implications for the 2nd treatment course, the previously positive RT- the treatment of acute promyelocytic leukemia. Proc ␣ PCR assay for PML/RAR- had converted to nega- Natl Acad Sci USA 1997; 94:3978-83. tive. Peak PB levels in plasma were 2.5 mM; however, 5. Soignet SL, Maslak P, Wang Z-G, et al. Complete these levels had decayed to baseline levels by 4 hrs. remission after induction of non-terminal differentia- Immunofluorescence and Western blot analysis tion and apoptosis in acute promyelocytic leukemia showed that PB clinical treatment caused a time- by arsenic trioxide. N Engl J Med 1998; 339:1341-8. dependent increase in histone hyperacetylation in 6. Grunstein M. Histone acetylation in chromatin struc- blood and bone marrow mononuclear cells.10 ture and transcription. Nature (Lond) 1997; 389:349- These preliminary data suggest that although PB is 52. a relatively weak HDAC inhibitor, the drug can 7. Grignani F, De Matteis S, Nervi C, et al. Fusion proteins of the retinoic acid receptor-␣ recruit histone de- reversibly increase chromatin hyperacetylation in tar- acetylase in promyelocytic leukaemia. Nature (Lond) get cells, possibly restoring RA sensitivity in retinoid- 1998; 391:815-8. resistant APL patients. The concept of HDAC inhibi- 8. He L-Z, Guidez F, Tribioli C, et al. Distinct interactions tion, if it can be selectively induced, may prove useful of PML-RAR␣ with co-repressors determine differential in other neoplastic diseases associated with onco- responses to RA in APL. Nat Gen 1998; 18:126-35. genic repression of gene transcription due to recruit- 9. Lin RJ, Nagy L, Inoue S, Shao W, Miller WH Jr., Evans ment of histone deacetylases. This concept is currently RM. Role of the histone deacetylase complex in acute being tested in a broader study. promyelocytic leukaemia. Nature (Lond) 1998; 391: 811-4. Acknowledgements 10. Warrell RP Jr., He L-Z, Richon V, Calleja E, Pandolfi PP. Numerous colleagues have collaborated in studies by our Targeted transcription therapy of acute promyelocyt- group. I am especially indebted to Drs. Pier Paolo Pandolfi, ic leukemia using an inhibitor of histone deacetylase. Victoria Richon, Steven Soignet, Paul Marks, Elizabeth Calle- J Natl Cancer Inst 1998; 90:1621-5. Session #9 – Trials that have influenced clinical practice in the treatment of thrombosis 79 procedures surrounding the test need to be standard- Ventilation-perfusion scintigraphy ised and the criteria for a normal and an abnormal test Ventilation-perfusion scintigraphy is the most wide- result need to be defined, using for example receiver- ly used first line diagnostic test when pulmonary operator curve analysis. In addition the inter- and embolism is suspected. The criteria for the interpreta- intraobserver variability of the test should be assessed. tion of lung scans have been a matter of debate for Ideally, the development of the diagnostic test will pass many years. The most clinically applicable classifica- into the next phase only after all these aspects have tion is that of three categories. These are: 1) a normal been adequately investigated. Phase 2 would then con- perfusion scan on the basis of which pulmonary sist of studies in which the diagnostic accuracy of the embolism is excluded; 2) a high probability lung scan, new test is evaluated in a large group of consecutive defined as one or more defects of at least segmental patients with a suspicion of the disease for which the size with associated local normal ventilation, which test is intended. The outcomes of the new test must be strongly indicates the presence of emboli; and 3) the blindly and independently compared with the results of remaining lung scan results which are considered to be the gold standard test for the relevant disease. The sen- non-diagnostic.5 The interobserver variability of venti- sitivity, specificity and positive and negative predictive lation-perfusion scintigraphy using the various classi- values of the test can then be determined. In case of fications has been investigated in several studies and insufficient accuracy, one might need to return to has been found to be acceptable.4 phase 1 to re-evaluate the test procedures and the def- Numerous studies have compared lung scintigra- inition of normal and abnormal results. phy with pulmonary angiography. The overall posi- Only after the requirements of both phases 1 and 2 tive predictive value of a high-probability lung scan have been adequately met, can the development has been shown to be 88% (95% CI 84% - 91%).4 This process advance into phase 3 in which the role of the is generally considered to be enough evidence to new test in the diagnostic process is evaluated. In this accept the diagnosis of pulmonary embolism. How- phase of development management studies are per- ever, pulmonary embolism cannot be considered formed, in which therapeutic decisions are made proven or excluded on the basis of a non-diagnostic based on the result of the new diagnostic test. lung scan. Studies have shown that the prevalence of A new diagnostic test should only be implemented angiographically proven pulmonary embolism is in routine clinical use after all three phases of devel- approximately 25% in patients with such a scan opment have been properly performed with good result.4 Further diagnostic investigation is therefore results. warranted in these patients. Management studies using ventilation-perfusion Pulmonary angiography scintigraphy in the clinical work-up of patients sus- At present pulmonary angiography is regarded as pected of having pulmonary embolism have been per- the gold standard diagnostic test for pulmonary formed. In these studies a normal perfusion lung scan embolism. There are generally accepted standardised was used to reject the diagnosis and anticoagulant procedures for the performance of the test and crite- treatment was withheld. In a total of 693 patients ria for a normal and an abnormal test result have been with a normal perfusion lung scan in whom antico- formulated.2 Furthermore, the observer variability of agulants were withheld 0.3% (95% CI 0.2% - 0.4%) the test has been assessed and has been found to be suffered a thromboembolic event during a follow-up minimal.3 Assessment of the sensitivity and specificity period of at least 3 months.4 Hence, it is deemed safe is not possible since pulmonary angiography is itself to withhold anticoagulants in patients with a normal the reference method. The ability of pulmonary perfusion lung scan. angiography to confirm or exclude pulmonary embolism is, however, considered to be high. Several D-dimer studies in patients who underwent pulmonary angiog- In recent years the measurement of D-dimer, a raphy for the suspicion of acute pulmonary embolism degradation product of fibrin, has been suggested for have shown that it is safe to withhold anticoagulant use in the diagnostic work-up of patients with clini- treatment from those patients who have a normal test cally suspected pulmonary embolism. Absence of an result. In a total of 840 patients with suspected pul- elevated concentration of D-dimer has been suggest- monary embolism and a normal pulmonary angio- ed to have a high negative predictive value for venous gram, anti-coagulant therapy was withheld and thromboembolism. As a result, numerous D-dimer patients were followed-up for a minimum of 3 tests (quantitative and qualitative) have been devel- months. The overall incidence of recurrent pulmonary oped which are now widely commercially available. embolism was 1.9% (95% CI 1.4% - 3.2%) and the inci- Proper evaluation of the interassay and intra- and dence of fatal pulmonary embolism was 0.3% (95% interobserver variation has only been performed for CI 0.09% - 1.08%).4 On the basis of these studies it is the minority of the available D-dimer tests. The stud- deemed safe to consider pulmonary embolism exclud- ies which have been performed have shown that this ed in the case of a normal pulmonary angiography variation is rather high.6 Several D-dimer tests have and proven in the case of an abnormal result. been compared to a gold standard diagnostic method 80 B-J. Sanson et al. to determine the accuracy of the tests, albeit often in trates the fact that a normal spiral CT scan appears non-consecutive series of patients. The high sensitiv- to be inappropriate to refute the diagnosis of pul- ity and moderate specificity of the D-dimer assays monary embolism and that subsequent testing should seem to qualify the test as potentially useful for the be initiated to exclude the disease safely. exclusion of venous thromboembolism. However, the sensitivity of D-dimer tests varies widely among the Discussion studies, ranging from 61% to 100% and data are still The introduction of a new diagnostic test is usual- limited.7,8 The cut-off values of the D-dimer assays ly accompanied by a great amount of enthusiasm can be varied so as to increase the sensitivity leading, concerning the potential indications for its use. A new however, to a decrease in specificity, which could lim- test is often presented as an improvement in the diag- it the clinical usefulness. When a receiver operator nostic arsenal. This all too frequently results in the curve (ROC) analysis is performed, the various quan- wide acceptance and large-scale implementation of a titative D-dimer assays appear to be nearly identical.9 new diagnostic test directly following its introduction. The determination of optimal cut-off values for the This often occurs without adequate foundation on individual D-dimer assays is, therefore, essential. properly performed studies. However, after the initial Management studies have evaluated the use of D- enthusiasm, the limitations and the true value in clin- dimer integrated in a diagnostic strategy in combina- ical practice become clear. In the past this has often tion with other non-invasive tests, such as ventilation- resulted in the almost complete disappearance of new perfusion lung scanning and clinical decision rules.10 and initially promising tests from the diagnostic arma- In this setting, management with D-dimer assays mentarium. One may assume that with proper evalu- seems safe and cost-effective. However, one needs to ation these tests could have been found valuable with- interpret these data with great caution. The use of D- in the proper setting. As example of the problems dimer assays within a strategy combined with other associated with the introduction of new diagnostic diagnostic tests seems safe, while the exclusion of pul- methods we have assessed 4 tests currently available monary embolism on basis of a normal D-dimer test for the diagnosis of pulmonary embolism using the alone seems unjustified. suggested guidelines for the proper evaluation of diag- nostic tests. Spiral computed tomography Pulmonary angiography, presently considered the Spiral computed tomography (spiral CT) is a rela- gold standard method, has been evaluated quite tively new diagnostic technique for patients with sus- extensively throughout the various phases mentioned pected pulmonary embolism, first described in earlier. Both the role in the diagnostic work-up of a 1992.11 In one study the interobserver agreement was patient with suspected pulmonary embolism and the found to be relatively high (kappa 0.77).12 However, consequences of its result are known. Ventilation-per- this does not appear to be true for pulmonary emboli fusion scintigraphy has also been shown to be of use that are confined to the subsegmental arteries. In in the diagnosis of pulmonary embolism. With the addition, no accepted criteria for normal and abnor- performance of proper studies in comparison to pul- mal test results have been formulated. The first study monary angiography and numerous clinical follow- that compared spiral CT with angiography was per- up studies, the value and limitations of the test have formed in 42 patients and found a sensitivity of 100% become clear. Both pulmonary angiography and ven- and a specificity of 96%.11 There have been many sub- tilation-perfusion scintigraphy have therefore been sequent studies comparing spiral CT with angiogra- properly evaluated according to the suggested guide- phy or lung scintigraphy. The overall sensitivity and lines and are well-established diagnostic methods. specificity are 91% (95% CI 87-94) and 93% (95% CI When considering the evaluation of D-dimer assays, 90-96), respectively. However, the sensitivity varies it is apparent that all phases of development have from 54% to 100 % and the specificity from 67% to been traversed for some of the available assays. How- 100% in these studies.4 This can in part be explained ever, there are also some troubling aspects in the by differences in patient selection in the various stud- development of this test. Not all of the assays have ies, the higher sensitivities and specificities generally been properly evaluated in phase 1 and 2 studies, and being found in the studies with highly selected patient the results of the studies that have been performed groups. Furthermore, different criteria for the diag- should not be directly extrapolated to other D-dimer nosis of pulmonary embolism and various procedures assays. Furthermore, a wide spread in the sensitivity for the performance of the test were used. and specificity has been found between studies, fur- One management study implementing the spiral CT ther complicating the choice of an optimal cut-off in the diagnostic work-up of patients suspected of point. Optimal cut-off levels for each individual D- pulmonary embolism has been performed.13 In this dimer assay need to be determined. The observation study a recurrence rate of venous thromboembolism that the ROC curves of the various quantitative D- of 5.5% (95% CI 1.3 to 9.7) during a follow-up peri- dimers are nearly identical would seem to support od of 3 months was found in patients with a normal this. The management studies that have so far been spiral CT and a normal ultrasound. This finding illus- performed show promising results. Use of the D- Session #9 – Trials that have influenced clinical practice in the treatment of thrombosis 81 dimer within certain diagnostic strategies appears safe sary. The criteria that need to be met by consecutive and cost-effective. However, further studies are war- phases of development need to be generally accept- ranted before the widespread use of D-dimer assays ed and implemented. in routine clinical practice is justified. The use of a D- dimer assay as a single test for the exclusion of pul- monary embolism, a practice that is gaining ground, References is certainly not founded on adequate studies. When the studies evaluating spiral CT scanning are 1. PIOPED Investigators. Value of the ventilation/perfu- critically assessed, one must conclude that the phas- sion scan in acute pulmonary embolism. Results of the prospective investigation of pulmonary embolism es of development have been passed in a less orderly diagnosis (PIOPED). JAMA 1990; 263:2753-9. fashion. Firstly, consensus does not exist concerning 2. Stein PD, Athanasoulis C, Alavi A, et al. Complica- the optimal method of performance of the test. Fur- tions and validity of pulmonary angiography in acute thermore, accepted criteria by which the presence or pulmonary embolism. Circulation 1992; 85:462-8. absence of pulmonary embolism can be evaluated are 3. van Beek EJR, Bakker AJ, Reekers JA. Pulmonary embolism: interobserver agreement in the interpreta- still lacking. This translates to the wide variation tion of conventional angiographic and DSA images in found in the performance of the test when compared patients with nondiagnostic lung scan results. Radi- to pulmonary angiography and ventilation-perfusion ology 1996; 198:721-4. scanning. A more important cause of this variation is, 4. Beek EJR van, Ginsberg JS, Hirsh J, Büller HR. Diag- however, that these comparisons were not consis- nosis of pulmonary embolism. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, eds. and tently performed in representative groups of patients thrombosis, basic principles and clinical practice. 4th with suspected pulmonary embolism. This can be ed., 1999. In press. seen by the wide range of the prevalence of pulmonary 5. Becker DM, Philbrick JT, Bachhuber TL, Humphries embolism in the various studies (33-88%). The only JE. D-dimer testing and acute venous thromboem- management study which has so far been performed bolism. A shortcut to accurate diagnosis? Arch Intern Med 1996; 156:939-46. using spiral CT scanning in the management of 6. Hull RD, Hirsh J, Carter CJ, et al. Diagnostic value of patients with suspected pulmonary embolism by Fer- ventilation-perfusion lung scanning in patients with retti et al., has shown disappointing results (recur- suspected pulmonary embolism. Chest 1985; 88:819- rence rate 5.4%, upper 95% CI 9.7%).13 Further phase 28. 1 studies in which methodology and criteria are stan- 7. Janssen MC, Heebels AE, de Metz M, et al. Reliability of five rapid D-dimer assays compared to ELISA in the dardised, and phase 2 studies in which spiral CT scan- exclusion of deep venous thrombosis. Thromb ning is compared with pulmonary angiography or ven- Haemostas 1997; 77:262-6. tilation-perfusion scintigraphy are, therefore, needed 8. Turkstra F, Beek EJR van, ten Cate JW, Büller HR. Reli- before further management studies or implementa- able rapid blood test for the exclusion of venous tion in routine clinical practice should occur. thromboembolism in symptomatic outpatients. Thromb Haemostas 1996; 76:9-11. In conclusion, at present pulmonary angiography 9. de Monyé W, Sanson BJ, Büller HR, Pattynama PM, and ventilation-perfusion scintigraphy are the only Huisman MV. The value of rapid quantitative D-dimer properly evaluated diagnostic tests for pulmonary assays versus gold standard for suspected pulmonary embolism. Although there are many new develop- embolism and the influence of embolus size on accu- ments in this field, such as D-dimer assays and the spi- racy. Submitted for publication. 10. Perrier A, Desmarais S, Miron MJ, et al. Non-invasive ral CT scan which are certainly promising, further diagnosis of venous thromboembolism in outpatients. studies are needed to determine their real value and Lancet 1999; 353:190-5. safety in the diagnostic work-up of patients suspect- 11. Remy-Jardin M, Remy J, Wattinne L, Giraud F. Central ed of having a pulmonary embolism. pulmonary thromboembolism: diagnosis with spiral The diagnostic tests for pulmonary embolism exem- volumetric CT with the single-breath-hold technique – comparison with pulmonary angiography. Radiology plify the problems often surrounding the introduction 1992; 185:381-7. of new tests. Due to the initial enthusiasm, a new test 12. van Rossum AB, Pattynama PM, Ton ER, et al. Pul- is often implemented in clinical use before proper monary embolism: validation of spiral CT angiography evaluation has occurred. Studies are also often per- in 149 patients. Radiology 1996; 201:467-70. formed without adequate data from earlier phases of 13. Ferretti GR, Bosson JL, Buffaz PD, et al. Acute pul- monary embolism: role of helical CT in 164 patients development. We would suggest that guidelines for with intermediate probability at ventilation-perfusion the proper evaluation of diagnostic tests before their scintigraphy and normal results at duplex US of the implementation in routine clinical practice are neces- legs. Radiology 1997; 205: 453-8. Educational Session 8 Haematologica 1999; 84:(EHA-4 educational book):72-74 Chairman: F. Mandelli Acute promyelocytic leukaemia

Diagnosis, front line treatment and molecular monitoring of acute promyelocytic leukaemia FRANCESCO LO COCO, DANIELA DIVERIO, GIUSEPPE AVVISATI, FRANCO MANDELLI Department of Cellular Biotechnologies and Haematology, University “La Sapienza” of Rome, Italy

Diagnosis In cases with the typical hypergranular morpholo- A unique t(15;17) chromosome aberration result- gy, however, treatment initiation need not be post- ing in the PML/RAR␣ gene fusion and an exquisite poned pending the results of genetic studies. These sensitivity to the differentiating agent all-trans retinoic patients may in fact immediately receive specific ther- acid (ATRA) are the main distinguishing features of apy and ATRA could be subsequently withdrawn in acute promyelocytic leukaemia (APL). Given in com- those rare cases lacking the genetic abnormality bination with anthracycline-containing chemo- despite having typical morphology. If not available therapy, ATRA has been shown to provide the best on site, RT-PCR and karyotypic characterisation treatment results in this disease. Because the presence might be requested from specialised centres in order of the PML/RAR␣ hybrid in leukaemic cells is known to obtain further potentially relevant information to predict response to ATRA, and due to frequent life- (e.g. additional chromosome abnormalities, variant threatening haemorrhagic diatheses, rapid diagnosis translocations, PML/RAR␣ isoform type) and to at the genetic level is recommended for promptly ini- define the correct RT-PCR strategy to be used in the tiating APL-tailored therapy.1-4 individual patient for minimal residual disease Demonstration of the disease’s genetic hallmark (MRD) monitoring. can be carried out at chromosome, protein, DNA or RNA levels, using conventional karyotyping and/or Front line treatment fluorescent in situ hybridisation (FISH), anti-PML anti- Following the advent of ATRA, large trials using bodies, Southern blotting and reverse-transcriptase this agent in variable combinations with chemother- polymerase chain reaction (RT-PCR), respectively. apy (CHT) for newly diagnosed APL have been car- Each of these procedures has its own advantages and ried out in Europe (APL ’93, GIMEMA, and PETHE- pitfalls (Table 1). In particular, karyotyping may MA studies), the USA (New York Memorial Sloan- occasionally give false-negative results [i.e. absence of Kettering Cancer Center, US Intergroup and MD t(15;17) in cases later found to contain cryptic Anderson studies), the UK (MRC), China and Japan ␣ PML/RAR rearrangement] and like Southern blot- (JALSG). Overall, nearly 2,000 patients were includ- ting is time-consuming, requiring a few days for exe- ed in these trials.3 The advantage of ATRA inclusion cution. RT-PCR allows a rapid and highly sensitive in front line treatment was established in 1993 by the diagnosis, but it is prone to artefacts and technical- APL European group in a randomised study com- ly difficult if not performed in experienced laborato- paring ATRA followed by CHT (ATRA→CHT) with ries. Recently, immunohistochemical analysis of PML CHT alone.6 Subsequent studies were aimed in most staining with monoclonal antibodies has proven use- cases at establishing the optimal ATRA and CHT ful for specific diagnosis, which is established fol- combination. Other issues which were addressed lowing the identification of the so-called microspeckled were the importance of genetic diagnosis and the role PML protein distribution consequent to the translo- of maintenance therapy including or not ATRA. The cation.5 In light of its widespread availability for main results may be summarised as follows: rapid, specific and low cost diagnosis, this procedure 1. resistant leukaemia to regimens variably combin- might be recommended as a convenient tool for iden- ing ATRA and CHT is virtually absent in patients tification of the disease’s hallmark, particularly in with genetically confirmed diagnosis. Failure to centres not equipped or trained for more sophisti- cated analyses. Apart from confirming morphologi- achieve haematological complete remission (CR, cally typical APLs, this assay would clarify diagnosis reported in 7-20% of cases) is mainly due to early in cases with uncertain cytological features and its death from haemorrhage or infectious complica- application could be extended to all acute myeloid tions; leukaemias (AMLs). 2. the simultaneous ATRA+CHT combination pro- vides the best results in terms of CR and disease- free survival (DFS), and seems more effective in Correspondence: Francesco Lo Coco, M.D. Department of Cellular diminishing the occurrence of overt ATRA syn- Biotechnologies and Haematology, University La Sapienza, via Benevento 6, 00161 Rome, Italy. Phone: international +39.06.85795530 – Fax: drome. This latter is also successfully counteract- international +39.06.44241984 – e-mail: [email protected] ed, however, by strict adherence to the recom- Session #8 – Acute promyelocytic leukaemia 73

Table 1. Technical approaches for genetic diagnosis in APL. CHT vs. ATRA+CHT vs. observation. The final analysis of these studies should be available soon. ATRA therapy is given intermittently in these two Technique Main advantages Main pitfalls trials (i.e. for 15 days every 3 months), since con- tinuous administration is known to result in sig- Karyotype Additional genetic Frequent poor quality information (lesions metaphases and cryptic nificant toxicity and more frequent development of other than t15;17) translocations (false resistance; negative) 5. the prognostic outcome of elderly APL patients Southern blot Specific Time consuming; has dramatically improved after the advent of hybridisation with > 1 ATRA. In fact, complete remission and event-free probe is often required survival rates of >80% and >50% have been report- RT-PCR Highly sensitive Frequent poor RNA yield ed in recent trials for patients aged >60. and amplification artefacts Molecular assessment of response and Immunohistochemistry Rapid, simple Provides no information MRD monitoring (PML-staining pattern) and low-cost on the PML breakpoint Besides refining diagnosis, PML/RAR␣ RT-PCR type studies offer the additional advantage of identifying patients with distinct transcript isoforms (bcr1 or long type, bcr2 or variable type and bcr3 or short type) and, particularly, of providing a sensitive tool to mendations established by the Memorial Sloan- assess response to treatment. Due to technical diffi- Kettering group,7 i.e. rapid institution of corticos- culties in distinguishing long and variable forms, in teroids at the earliest manifestation of symptoms most reported studies these two types are included in related to the syndrome (respiratory distress, unex- a common group referred to as long transcript, and plained weight gain, fever and pleuro-pericardial this latter compared to the short type. Contrary to effusion). Death rates attributed to the ATRA syn- earlier reports indicating a worse outcome in patients drome have dropped to 1-3% in the most recent with the short form, recent trials have shown no sig- studies; 3. the benefit of cytarabine administered in addition nificant differences in DFS. It has to be emphasised, to anthracyclines during induction and/or consol- however, that the former analyses were done in idation is unclear. Studies in which cytarabine has patients treated with ATRA alone for remission induc- been omitted from induction (GIMEMA) or from tion, whereas the latter derive from trials in which both induction and consolidation (MD Anderson patients received CHT added to ATRA. and PETHEMA) have reported results compara- Several groups have analysed the response to treat- ble or even superior (at least in terms of reduced ment by RT-PCR tests performed at various times dur- toxicity) to those obtained in studies including this ing and after treatment. Using assays with sensitivity –4 agent. Whilst the effectiveness of anthracycline- levels of approximately 10 , up to 50% of patients in based monochemotherapy for remission induc- haematological remission after induction have ␣ tion of APL was established as long as 26 years detectable PML/RAR transcript in their marrow ago by the pioneering work of Jean Bernard,8 the cells. No correlations were found between PCR status omission of cytarabine from overall treatment (i.e. at the time of remission achievement and relapse risk including consolidation) is a recent attempt and in the GIMEMA and MRC studies. The vast majority long-term results of the MD Anderson and PETHE- (up to 95%) of patients tested after completion of MA groups are awaited to resolve this issue. How- consolidation were PCR-negative in the New York ever, no study has as yet compared different con- Memorial Sloan-Kettering, MRC, GIMEMA and solidation options including or not cytarabine in PETHEMA series.3 Interestingly, detection of residual phase III trials; disease at the end of the third CHT course (of 4 giv- 4. two large randomised studies (APL’93 and US en) predicted an increased relapse risk associated with Intergroup) have demonstrated a benefit conferred poorer survival in the MRC study.9 by ATRA-containing maintenance therapy. This is The clinical value of post-treatment molecular mon- also confirmed in the analysis of the GIMEMA itoring has been emphasised in a recently reported study which adopted the same randomisation prospective study of the GIMEMA group.10 This designed by the European APL ’93 study (unpub- showed that conversion from PCR-negativity to PCR- lished observations). The advantage of using CHT positivity after consolidation therapy is almost always maintenance with methotrexate and 6-mercap- associated with subsequent haematological relapse. topurine had been suggested in France and the Based on these results, patients in this trial who con- USA in the past decade; hence, both the APL ’93 vert to PCR-positivity in two successive marrow sam- and the GIMEMA groups included four randomi- ples are now defined as being in molecular relapse sation arms in their studies to compare ATRA vs. and given anticipated salvage treatment. 74 F. Lo Coco et al.

Conclusions and future perspectives References Despite considerable improvement in diagnosis and management, a sizeable proportion of APL patients 1. Chen S-J, Wang Z-Y, Chen Z. Acute promyelocytic who receive state-of-art front line treatment still die of leukemia: from clinic to molecular biology. Stem Cells early complications or following disease recurrence. 1995; 13:22-31. With regard to early death, a special effort should be 2. Fenaux P, Chomienne C, Degos L. Acute promyelo- cytic leukemia: Biology and treatment. Semin Oncol made to define risk categories better (hyperleukocy- 1997; 24:92-102. tosis, older age, severity of the coagulopathy, other 3. Lo Coco F, Nervi C, Avvisati G, Mandelli F. Acute unknown factors) in order to promptly reinforce ade- promyelocytic leukemia: a curable disease. Leukemia quate supportive care and to evaluate the feasibility 1998; 12: 1866-80. 4. Warrell RP Jr. Pathogenesis and management of acute of a tailored (less intensive?) initial treatment. promyelocytic leukemia. Annu Rev Med 1996; 47: At least 60% of newly diagnosed patients become 555-65. long-term survivors and are probably cured with 5. Falini B, Flenghi L, Fagioli M, et al. Immunocyto- ATRA+CHT. It is conceivable to hypothesise that a chemical diagnosis of acute promyelocytic leukemia relevant fraction of these patients are being overtreat- (M3) with the anti-PML monoclonal antibody PG- M3. Blood 1997; 90:4046-53. ed and that they might, therefore, be spared this risk. 6. Fenaux P, Le Deley MC, Castaigne S, et al. Effect of All- On the other hand, we are still unable to identify trans retinoic acid in newly diagnosed acute promye- those cases (approximately 20%) who will ultimately locytic leukemia. Blood 1993; 82:3241-9. relapse after initial therapy and are, therefore, in need 7. Frankel SR, Eardley A, Heller G, et al. All-trans retinoic acid for acute promyelocytic leukemia. Results of the of treatment intensification. The definition of risk cat- New York study. Ann Intern Med 1995; 120:278-86. egories at diagnosis for more appropriate treatment 8. Bernard J, Weil M, Boiron M, Jacquillat C, Flandrin G, stratification remains a big challenge for future clini- Gemon MF. Acute promyelocytic leukemia. Results of cal investigation in this disease. Finally, the place of treatment with daunorubicin. Blood 1973; 41:489- 96. novel drugs which have proven effective in relapse, 9. Burnett AK, Grimwade D, Solomon E, Wheatley K, such as arsenic trioxide, and the advantage of antici- Goldstone AH. Presenting white cell count and kinet- pating salvage treatment at the time of molecular ics of molecular remission predict prognosis in acute recurrence remain to be established. promyelocytic leukemia treated with all-trans retinoic acid: Results of the randomised MRC trial. Blood Acknowledgements 1999; in press. Supported by A.I.L. (Associazione Italiana contro le 10. Diverio D, Rossi V, Avvisati G, et al. Early detection of relapse by prospective RT-PCR analysis of the Leucemie), Italy. PML/RAR␣ fusion gene in patients with acute promye- locytic leukemia enrolled in the GIMEMA-AIEOP mul- ticenter “AIDA” trial. Blood 1998; 92:784-9. Educational Session 8 Haematologica 1999; 84:(EHA-4 educational book):75-77 Chairman: F. Mandelli Acute promyelocytic leukaemia

Arsenicals and inhibitors of histone deacetylase as anticancer therapy RAYMOND P. WARRELL, JR. Developmental Chemotherapy and Leukaemia Services, Memorial Sloan-Kettering Cancer Centre and the Cornell University Medical College, New York, NY, USA

everal new therapeutic approaches have recent- breadth of this activity, we then initiated a clinical ly been clinically tested in patients with study to evaluate the pharmacokinetics, safety, and Sadvanced cancer, including the use of arseni- potential efficacy of melarsoprol in patients with cals and various drugs that affect gene transcription relapsed leukaemia. Using the anti-trypanosomal via inhibition of histone deacetylase. Although our dose and schedule, patients received escalating intra- interest in these agents arose from our programme venous doses daily for 3 days, repeated weekly for 3 in acute promyelocytic leukaemia (APL), we expect weeks. Doses were 1 mg/kg on day 1, 2 mg/kg on that these agents will prove useful for a number of day 2, and 3.6 mg/kg on day 3 and on all days there- other cancers. The following pages summarise some after, up to a maximum daily dose of 200 mg. Eight aspects of our group’s clinical work in these areas. patients [5 AML, 1 APL, 1 CML, and 1 CLL] were treated. Mean peak plasma concentrations of the Arsenicals as cancer treatment parent drug were obtained immediately after injec- Like many others, we were intrigued by reports tion. These concentrations ranged from 1.2 µg/mL emanating from the People’s Republic of China in on day 1 to 2.4 µg/mL on day 3, were dose propor- 1 ␤ ≅ the early 1990s that intravenous arsenic had re- tional, and decayed with a t /2 15 minutes. A emerged as a cancer treatment. Arsenic has been minor clinical response (regression of splenomegaly used as a medicinal for many centuries, both in top- and lymphadenopathy) was observed in a patient ical and injectable forms (as escharotic agents) as with chronic lymphocytic leukaemia. One patient well as oral forms that were first described in the with t(11;17)-variant APL who had been refractory to 1700s. Fowler’s solution given orally was used in the chemotherapy and all-trans retinoic acid had slight 19th and early 20th centuries as a treatment to reduce improvements in both his leukocyte and platelet extreme leukocytosis in patients with chronic myelo- counts with the first course of treatment. He then cytic leukaemia (CML). In the West, however, this received a second 3-week course without further use dwindled after the widespread application of improvement and was removed from the study. radiotherapy in the 1930s and the advent of other Unfortunately, central nervous system (CNS) toxi- cytotoxic drugs after the 2nd World War. city proved limiting. Three patients experienced gen- The recent upsurge of interest stems from reports eralised grand mal seizures during the second week of from Harbin China that an intravenously injectable therapy. Other neurologic reactions included vertigo formulation induced remissions in patients with (2 patients) and paraesthesia (1 patient), both of APL.1 Since we perceived an enormous regulatory which resolved after discontinuing therapy. The hurdle to direct application of this therapy in the patient who received two complete treatment cycles United States, we first tested an organic arsenical, experienced moderately severe peripheral neuropa- melarsoprol, which was already formulated for thy, with bilateral lower extremity paraesthesia that caused moderate functional impairment, but which human use and was immediately available on an resolved approximately 5 weeks after the last dose. investigational basis because of its use for the treat- Other common reactions were nausea, vomiting, and ment of African trypanosomiasis due to T. bruceii. irritation at the injection site. We concluded that the antitrypanosomal dosing schedule of melarsoprol is Laboratory and clinical studies of melarsoprol associated with excessive CNS toxicity, and that ver- We evaluated both As2O3 and melarsoprol for pos- ification of the striking preclinical activity in patients sible antileukaemic activity in vitro. In these studies, with leukaemia requires development of an alterna- melarsoprol exhibited broad antileukaemic activity tive dosing schedule. against both myeloid and lymphoid cells. Given the Arsenic trioxide We and others have recently confirmed that arsenic Correspondence: Raymond P. Warrell, Jr. MD, Memorial Sloan-Ketter- trioxide induces complete remissions (CR) in a high ing Cancer Center, 1275 York Avenue, New York, NY 10021, USA. Tel. 2-5 international +1-212-6398168 – Fax. international +1-212-717-3215 proportion of patients with APL. In our initial pilot – e-mail: [email protected] study, 12 heavily pretreated patients were treated 76 R. P. Warrell, Jr. with arsenic trioxide at doses ranging from 0.06 to APL patients who receive all-trans RA for induction 0.2 mg/kg/day until leukaemic cells were eliminated develop one or more signs of the RA syndrome, a poten- from the marrow. Patients who achieved CR were eli- tially lethal complication. The RA syndrome is a loose- gible to receive up to 5 additional courses of therapy, ly defined constellation of problems that include fever, with each course given for a cumulative total of 25 weight gain, dyspnoea, pulmonary oedema, pleural days at a daily dose of 0.15 mg/kg/d every 3-6 weeks. effusions, musculoskeletal pain, or joint effusions. Marrow leukaemic cells were eliminated in 11 of 12 Signs/symptoms of the RA syndrome were observed in patients after a median treatment duration of 33 days 8 of the 23 patients (35%) treated with arsenic triox- (range, 12-39 days)(one patient who had an intracra- ide. Several patients were presumptively treated with nial haemorrhage on day 1 died on day 5). The medi- high-dose dexamethasone. All patients with this prob- an cumulative dose during induction was 360 mg lem have recovered, and interruption of arsenic ther- (range, 160 to 515 mg). CR by all criteria was apy was not required. These data show that even non- attained at a median time of 47 days (range, 24-83 terminal cytodifferentiation induced by arsenic trioxide d). Impressively, 8 of 11 patients tested converted RT- is associated with induction of leukocytosis and the PCR assays for PML/RAR-␣ to negative at or before RA syndrome in APL. These effects appear to be relat- the end of the second course, while 3 patients who ed to intrinsic biological responsiveness to differenti- remained positive relapsed early and appeared to ating agents. 5 have rapidly developed arsenic resistance. Several Pharmacokinetics of arsenic trioxide patients have received up to 6 cycles of arsenic triox- and studies in other cancers ide therapy without showing evidence of cumulative We have conducted several dose-ranging studies to toxicity. examine the pharmacokinetics and biological effects We subsequently initiated a confirmatory multicen- of arsenic trioxide in patients with other types of tre study in patients with relapsed or refractory APL haematologic cancers and solid tumours. Pharmaco- using the same treatment schedules at a fixed daily kinetic analysis of blood and urine for elemental dose of 0.15 mg/kg/d. To date, 28 patients have been arsenic content showed that arsenic is distributed in accrued, and 19 of 21 evaluable patients (90%) have both plasma and red blood cell fractions of whole achieved complete remission. A new study of arsenic blood. Parallel elimination curves suggest that these trioxide plus all-trans RA for induction therapy of new- 2 compartments are freely exchangeable. The mean ly diagnosed patients with APL will begin shortly. area under the plasma ϫ concentration curve (AUC) Adverse reactions to arsenic trioxide on day 1 after a dose of 0.15 mg/kg was ~ 400 Adverse effects due to arsenic trioxide have gener- ng.hr/mL. Approximately 20% of the administered ally been mild. These reactions have included skin dose was recovered in urine within the first 24 hrs. rash, light headache during the infusion, fatigue, mild The terminal half-life appears to be quite prolonged. hyperglycaemia, musculoskeletal pain, and EKG In an ongoing dose-ranging study in patients with changes (particularly QT prolongation). In the past, haematologic cancers other than APL, we have all-trans retinoic acid (RA) was shown to induce ter- employed a daily IV dosing schedule for a cumulative minal differentiation of APL cells in vivo, and leukocy- total of 25 days per treatment course, with addition- tosis was commonly observed during this process, al cycles administered every 3-5 weeks. Dose levels probably as a result of a transient increase in have ranged from 0.1 to 0.25 mg/kg/day. A separate leukaemic cell lifespan. Similarly, previous studies by study in patients with solid tumours is using a daily ϫ us and others have shown that arsenic treatment is 5 days schedule, with additional cycles in responding associated with induction of partial, non-terminal patients administered every 4 weeks. Doses in that immunophenotypic cytodifferentiation. study have ranged from 0.15 to 0.3 mg/kg/day. Over In clinical studies of APL patients treated with this range of doses, the drug has continued to be rel- arsenic trioxide at this centre, leukocytosis (defined atively well-tolerated. However, skin rash, diarrhoea, as a total peripheral blood leukocyte count ≥ 10ϫ109 QTc prolongation on EKG, and fatigue (but not leuko- cells/L) was observed in 13 of 23 patients (57%). The cytosis or RA syndrome) have emerged as increasing- median baseline leukocyte count for all treated ly prominent side-effects in these other diseases. patients (2.6ϫ109/L [range, 0.7-65.2]) was not sub- stantially different from the cohort of patients who Targeting of histone deacetylase as subsequently developed leukocytosis (3.5ϫ109/L anticancer therapies [range, 1.3-65.2). The median peak leukocyte count Acetylation of DNA-associated histones is linked to of the latter group was 60.3ϫ109/L (range, 17.1- gene-specific activation, whereas histone deacetyla- 247.0), which occurred at a median of 19 days from tion is associated with transcriptional repression.6 the start of arsenic therapy (range, 4-24 days). No Recent studies have shown that inhibitors of histone other cytotoxic therapy was administered, and the deacetylase (HDAC) can relieve transcriptional repres- leukocytosis resolved in all patients. sion caused by certain oncogenes.7-9 These inhibitors Distinct from leukocytosis, approximately 50% of include trichostatin A, trapoxin, depudecins, bicyclic Session #8 – Acute promyelocytic leukaemia 77 depsipeptides, hybrid polar compounds and various ja, and Peter Maslak for their contributions to these efforts. butyrates. Supported in part by CA-77136 and CA-09207 from the Since sodium phenylbutyrate was immediately National Cancer Institute, FD-R-001364 from the Orphan available for investigational use in human subjects, Products Development, Food and Drug Administration, EDT- we tested whether this drug, as well as the concept of 83025 from the American Cancer Society, and by grants from HDAC inhibition, could be effective in vitro and then the Lymphoma Foundation and PolaRx Biopharmaceuticals, Inc. clinically in patients with highly resistant APL. Having documented that sodium phenylbutyrate (PB) inhib- ited HDAC, we found that high concentrations of PB alone decreased cell growth but had no effect upon References differentiation of NB4 cells; however, the combina- tion with all-trans RA caused a substantial increase in 1. Sun HD, Ma L, Hu XC, Zhang TD. Treatment of acute differentiation. promyelocytic leukemia by Ailing-1 therapy with use We then treated a child who had relapsed after of syndrome differentiation of traditional Chinese medicine. Chim J Comb Trad Chin Med West Med numerous cytotoxic drugs, all-trans RA, allogeneic 1992; 12:170-1. bone marrow transplantation, and arsenic trioxide. 2. Chen G-Q, Zhu J, Shi X-G, et al. In vitro studies on cel- Blood and marrow mononuclear cells were assayed lular and molecular mechanisms of arsenic trioxide for histone hyperacetylation by immunohistochem- (As2O3) in the treatment of acute promyelocytic istry and Western blot analysis. Minimal residual dis- leukemia: As2O3 induces NB4 cell apoptosis with ease was assessed by RT-PCR for PML/RAR-␣. After downregulation of Bcl-2 expression and modulation reconfirming that the patient was clinically resistant of PML-RAR␣/Pml proteins. Blood 1996; 88:1052- to RA, PB was added to the treatment regimen. The 61. RA dose ranged from 45-90 mg/m2/d in 2 equally 3. Shen Z-X, Chen G-Q, Ni J-H, Li XS, Xiong SM, Qiu QY, divided doses; the PB dose ranged from 150-210 et al. Use of arsenic trioxide (As2O3) in the treatment mg/kg IV BID infused IV over 1 hr. By day 23, the of acute promyelocytic leukemia (APL): II. Clinical effi- patient’s marrow showed complete elimination of vis- cacy and pharmacokinetics in patients at relapse. Blood 1997; 89:3354-60. ible leukaemic cells, and she achieved a complete clin- 4. Zhu J, Koken MH, Quignon F, et al. Arsenic-induced ical and cytogenetic remission shortly thereafter. By PML targeting onto nuclear bodies: implications for the 2nd treatment course, the previously positive RT- the treatment of acute promyelocytic leukemia. Proc ␣ PCR assay for PML/RAR- had converted to nega- Natl Acad Sci USA 1997; 94:3978-83. tive. Peak PB levels in plasma were 2.5 mM; however, 5. Soignet SL, Maslak P, Wang Z-G, et al. Complete these levels had decayed to baseline levels by 4 hrs. remission after induction of non-terminal differentia- Immunofluorescence and Western blot analysis tion and apoptosis in acute promyelocytic leukemia showed that PB clinical treatment caused a time- by arsenic trioxide. N Engl J Med 1998; 339:1341-8. dependent increase in histone hyperacetylation in 6. Grunstein M. Histone acetylation in chromatin struc- blood and bone marrow mononuclear cells.10 ture and transcription. Nature (Lond) 1997; 389:349- These preliminary data suggest that although PB is 52. a relatively weak HDAC inhibitor, the drug can 7. Grignani F, De Matteis S, Nervi C, et al. Fusion proteins of the retinoic acid receptor-␣ recruit histone de- reversibly increase chromatin hyperacetylation in tar- acetylase in promyelocytic leukaemia. Nature (Lond) get cells, possibly restoring RA sensitivity in retinoid- 1998; 391:815-8. resistant APL patients. The concept of HDAC inhibi- 8. He L-Z, Guidez F, Tribioli C, et al. Distinct interactions tion, if it can be selectively induced, may prove useful of PML-RAR␣ with co-repressors determine differential in other neoplastic diseases associated with onco- responses to RA in APL. Nat Gen 1998; 18:126-35. genic repression of gene transcription due to recruit- 9. Lin RJ, Nagy L, Inoue S, Shao W, Miller WH Jr., Evans ment of histone deacetylases. This concept is currently RM. Role of the histone deacetylase complex in acute being tested in a broader study. promyelocytic leukaemia. Nature (Lond) 1998; 391: 811-4. Acknowledgements 10. Warrell RP Jr., He L-Z, Richon V, Calleja E, Pandolfi PP. Numerous colleagues have collaborated in studies by our Targeted transcription therapy of acute promyelocyt- group. I am especially indebted to Drs. Pier Paolo Pandolfi, ic leukemia using an inhibitor of histone deacetylase. Victoria Richon, Steven Soignet, Paul Marks, Elizabeth Calle- J Natl Cancer Inst 1998; 90:1621-5. Session #9 – Trials that have influenced clinical practice in the treatment of thrombosis 79 procedures surrounding the test need to be standard- Ventilation-perfusion scintigraphy ised and the criteria for a normal and an abnormal test Ventilation-perfusion scintigraphy is the most wide- result need to be defined, using for example receiver- ly used first line diagnostic test when pulmonary operator curve analysis. In addition the inter- and embolism is suspected. The criteria for the interpreta- intraobserver variability of the test should be assessed. tion of lung scans have been a matter of debate for Ideally, the development of the diagnostic test will pass many years. The most clinically applicable classifica- into the next phase only after all these aspects have tion is that of three categories. These are: 1) a normal been adequately investigated. Phase 2 would then con- perfusion scan on the basis of which pulmonary sist of studies in which the diagnostic accuracy of the embolism is excluded; 2) a high probability lung scan, new test is evaluated in a large group of consecutive defined as one or more defects of at least segmental patients with a suspicion of the disease for which the size with associated local normal ventilation, which test is intended. The outcomes of the new test must be strongly indicates the presence of emboli; and 3) the blindly and independently compared with the results of remaining lung scan results which are considered to be the gold standard test for the relevant disease. The sen- non-diagnostic.5 The interobserver variability of venti- sitivity, specificity and positive and negative predictive lation-perfusion scintigraphy using the various classi- values of the test can then be determined. In case of fications has been investigated in several studies and insufficient accuracy, one might need to return to has been found to be acceptable.4 phase 1 to re-evaluate the test procedures and the def- Numerous studies have compared lung scintigra- inition of normal and abnormal results. phy with pulmonary angiography. The overall posi- Only after the requirements of both phases 1 and 2 tive predictive value of a high-probability lung scan have been adequately met, can the development has been shown to be 88% (95% CI 84% - 91%).4 This process advance into phase 3 in which the role of the is generally considered to be enough evidence to new test in the diagnostic process is evaluated. In this accept the diagnosis of pulmonary embolism. How- phase of development management studies are per- ever, pulmonary embolism cannot be considered formed, in which therapeutic decisions are made proven or excluded on the basis of a non-diagnostic based on the result of the new diagnostic test. lung scan. Studies have shown that the prevalence of A new diagnostic test should only be implemented angiographically proven pulmonary embolism is in routine clinical use after all three phases of devel- approximately 25% in patients with such a scan opment have been properly performed with good result.4 Further diagnostic investigation is therefore results. warranted in these patients. Management studies using ventilation-perfusion Pulmonary angiography scintigraphy in the clinical work-up of patients sus- At present pulmonary angiography is regarded as pected of having pulmonary embolism have been per- the gold standard diagnostic test for pulmonary formed. In these studies a normal perfusion lung scan embolism. There are generally accepted standardised was used to reject the diagnosis and anticoagulant procedures for the performance of the test and crite- treatment was withheld. In a total of 693 patients ria for a normal and an abnormal test result have been with a normal perfusion lung scan in whom antico- formulated.2 Furthermore, the observer variability of agulants were withheld 0.3% (95% CI 0.2% - 0.4%) the test has been assessed and has been found to be suffered a thromboembolic event during a follow-up minimal.3 Assessment of the sensitivity and specificity period of at least 3 months.4 Hence, it is deemed safe is not possible since pulmonary angiography is itself to withhold anticoagulants in patients with a normal the reference method. The ability of pulmonary perfusion lung scan. angiography to confirm or exclude pulmonary embolism is, however, considered to be high. Several D-dimer studies in patients who underwent pulmonary angiog- In recent years the measurement of D-dimer, a raphy for the suspicion of acute pulmonary embolism degradation product of fibrin, has been suggested for have shown that it is safe to withhold anticoagulant use in the diagnostic work-up of patients with clini- treatment from those patients who have a normal test cally suspected pulmonary embolism. Absence of an result. In a total of 840 patients with suspected pul- elevated concentration of D-dimer has been suggest- monary embolism and a normal pulmonary angio- ed to have a high negative predictive value for venous gram, anti-coagulant therapy was withheld and thromboembolism. As a result, numerous D-dimer patients were followed-up for a minimum of 3 tests (quantitative and qualitative) have been devel- months. The overall incidence of recurrent pulmonary oped which are now widely commercially available. embolism was 1.9% (95% CI 1.4% - 3.2%) and the inci- Proper evaluation of the interassay and intra- and dence of fatal pulmonary embolism was 0.3% (95% interobserver variation has only been performed for CI 0.09% - 1.08%).4 On the basis of these studies it is the minority of the available D-dimer tests. The stud- deemed safe to consider pulmonary embolism exclud- ies which have been performed have shown that this ed in the case of a normal pulmonary angiography variation is rather high.6 Several D-dimer tests have and proven in the case of an abnormal result. been compared to a gold standard diagnostic method 80 B-J. Sanson et al. to determine the accuracy of the tests, albeit often in trates the fact that a normal spiral CT scan appears non-consecutive series of patients. The high sensitiv- to be inappropriate to refute the diagnosis of pul- ity and moderate specificity of the D-dimer assays monary embolism and that subsequent testing should seem to qualify the test as potentially useful for the be initiated to exclude the disease safely. exclusion of venous thromboembolism. However, the sensitivity of D-dimer tests varies widely among the Discussion studies, ranging from 61% to 100% and data are still The introduction of a new diagnostic test is usual- limited.7,8 The cut-off values of the D-dimer assays ly accompanied by a great amount of enthusiasm can be varied so as to increase the sensitivity leading, concerning the potential indications for its use. A new however, to a decrease in specificity, which could lim- test is often presented as an improvement in the diag- it the clinical usefulness. When a receiver operator nostic arsenal. This all too frequently results in the curve (ROC) analysis is performed, the various quan- wide acceptance and large-scale implementation of a titative D-dimer assays appear to be nearly identical.9 new diagnostic test directly following its introduction. The determination of optimal cut-off values for the This often occurs without adequate foundation on individual D-dimer assays is, therefore, essential. properly performed studies. However, after the initial Management studies have evaluated the use of D- enthusiasm, the limitations and the true value in clin- dimer integrated in a diagnostic strategy in combina- ical practice become clear. In the past this has often tion with other non-invasive tests, such as ventilation- resulted in the almost complete disappearance of new perfusion lung scanning and clinical decision rules.10 and initially promising tests from the diagnostic arma- In this setting, management with D-dimer assays mentarium. One may assume that with proper evalu- seems safe and cost-effective. However, one needs to ation these tests could have been found valuable with- interpret these data with great caution. The use of D- in the proper setting. As example of the problems dimer assays within a strategy combined with other associated with the introduction of new diagnostic diagnostic tests seems safe, while the exclusion of pul- methods we have assessed 4 tests currently available monary embolism on basis of a normal D-dimer test for the diagnosis of pulmonary embolism using the alone seems unjustified. suggested guidelines for the proper evaluation of diag- nostic tests. Spiral computed tomography Pulmonary angiography, presently considered the Spiral computed tomography (spiral CT) is a rela- gold standard method, has been evaluated quite tively new diagnostic technique for patients with sus- extensively throughout the various phases mentioned pected pulmonary embolism, first described in earlier. Both the role in the diagnostic work-up of a 1992.11 In one study the interobserver agreement was patient with suspected pulmonary embolism and the found to be relatively high (kappa 0.77).12 However, consequences of its result are known. Ventilation-per- this does not appear to be true for pulmonary emboli fusion scintigraphy has also been shown to be of use that are confined to the subsegmental arteries. In in the diagnosis of pulmonary embolism. With the addition, no accepted criteria for normal and abnor- performance of proper studies in comparison to pul- mal test results have been formulated. The first study monary angiography and numerous clinical follow- that compared spiral CT with angiography was per- up studies, the value and limitations of the test have formed in 42 patients and found a sensitivity of 100% become clear. Both pulmonary angiography and ven- and a specificity of 96%.11 There have been many sub- tilation-perfusion scintigraphy have therefore been sequent studies comparing spiral CT with angiogra- properly evaluated according to the suggested guide- phy or lung scintigraphy. The overall sensitivity and lines and are well-established diagnostic methods. specificity are 91% (95% CI 87-94) and 93% (95% CI When considering the evaluation of D-dimer assays, 90-96), respectively. However, the sensitivity varies it is apparent that all phases of development have from 54% to 100 % and the specificity from 67% to been traversed for some of the available assays. How- 100% in these studies.4 This can in part be explained ever, there are also some troubling aspects in the by differences in patient selection in the various stud- development of this test. Not all of the assays have ies, the higher sensitivities and specificities generally been properly evaluated in phase 1 and 2 studies, and being found in the studies with highly selected patient the results of the studies that have been performed groups. Furthermore, different criteria for the diag- should not be directly extrapolated to other D-dimer nosis of pulmonary embolism and various procedures assays. Furthermore, a wide spread in the sensitivity for the performance of the test were used. and specificity has been found between studies, fur- One management study implementing the spiral CT ther complicating the choice of an optimal cut-off in the diagnostic work-up of patients suspected of point. Optimal cut-off levels for each individual D- pulmonary embolism has been performed.13 In this dimer assay need to be determined. The observation study a recurrence rate of venous thromboembolism that the ROC curves of the various quantitative D- of 5.5% (95% CI 1.3 to 9.7) during a follow-up peri- dimers are nearly identical would seem to support od of 3 months was found in patients with a normal this. The management studies that have so far been spiral CT and a normal ultrasound. This finding illus- performed show promising results. Use of the D- Session #9 – Trials that have influenced clinical practice in the treatment of thrombosis 81 dimer within certain diagnostic strategies appears safe sary. The criteria that need to be met by consecutive and cost-effective. However, further studies are war- phases of development need to be generally accept- ranted before the widespread use of D-dimer assays ed and implemented. in routine clinical practice is justified. The use of a D- dimer assay as a single test for the exclusion of pul- monary embolism, a practice that is gaining ground, References is certainly not founded on adequate studies. When the studies evaluating spiral CT scanning are 1. PIOPED Investigators. Value of the ventilation/perfu- critically assessed, one must conclude that the phas- sion scan in acute pulmonary embolism. Results of the prospective investigation of pulmonary embolism es of development have been passed in a less orderly diagnosis (PIOPED). JAMA 1990; 263:2753-9. fashion. Firstly, consensus does not exist concerning 2. Stein PD, Athanasoulis C, Alavi A, et al. Complica- the optimal method of performance of the test. Fur- tions and validity of pulmonary angiography in acute thermore, accepted criteria by which the presence or pulmonary embolism. Circulation 1992; 85:462-8. absence of pulmonary embolism can be evaluated are 3. van Beek EJR, Bakker AJ, Reekers JA. Pulmonary embolism: interobserver agreement in the interpreta- still lacking. This translates to the wide variation tion of conventional angiographic and DSA images in found in the performance of the test when compared patients with nondiagnostic lung scan results. Radi- to pulmonary angiography and ventilation-perfusion ology 1996; 198:721-4. scanning. A more important cause of this variation is, 4. Beek EJR van, Ginsberg JS, Hirsh J, Büller HR. Diag- however, that these comparisons were not consis- nosis of pulmonary embolism. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, eds. Hemostasis and tently performed in representative groups of patients thrombosis, basic principles and clinical practice. 4th with suspected pulmonary embolism. This can be ed., 1999. In press. seen by the wide range of the prevalence of pulmonary 5. Becker DM, Philbrick JT, Bachhuber TL, Humphries embolism in the various studies (33-88%). The only JE. D-dimer testing and acute venous thromboem- management study which has so far been performed bolism. A shortcut to accurate diagnosis? Arch Intern Med 1996; 156:939-46. using spiral CT scanning in the management of 6. Hull RD, Hirsh J, Carter CJ, et al. Diagnostic value of patients with suspected pulmonary embolism by Fer- ventilation-perfusion lung scanning in patients with retti et al., has shown disappointing results (recur- suspected pulmonary embolism. Chest 1985; 88:819- rence rate 5.4%, upper 95% CI 9.7%).13 Further phase 28. 1 studies in which methodology and criteria are stan- 7. Janssen MC, Heebels AE, de Metz M, et al. Reliability of five rapid D-dimer assays compared to ELISA in the dardised, and phase 2 studies in which spiral CT scan- exclusion of deep venous thrombosis. Thromb ning is compared with pulmonary angiography or ven- Haemostas 1997; 77:262-6. tilation-perfusion scintigraphy are, therefore, needed 8. Turkstra F, Beek EJR van, ten Cate JW, Büller HR. Reli- before further management studies or implementa- able rapid blood test for the exclusion of venous tion in routine clinical practice should occur. thromboembolism in symptomatic outpatients. Thromb Haemostas 1996; 76:9-11. In conclusion, at present pulmonary angiography 9. de Monyé W, Sanson BJ, Büller HR, Pattynama PM, and ventilation-perfusion scintigraphy are the only Huisman MV. The value of rapid quantitative D-dimer properly evaluated diagnostic tests for pulmonary assays versus gold standard for suspected pulmonary embolism. Although there are many new develop- embolism and the influence of embolus size on accu- ments in this field, such as D-dimer assays and the spi- racy. Submitted for publication. 10. Perrier A, Desmarais S, Miron MJ, et al. Non-invasive ral CT scan which are certainly promising, further diagnosis of venous thromboembolism in outpatients. studies are needed to determine their real value and Lancet 1999; 353:190-5. safety in the diagnostic work-up of patients suspect- 11. Remy-Jardin M, Remy J, Wattinne L, Giraud F. Central ed of having a pulmonary embolism. pulmonary thromboembolism: diagnosis with spiral The diagnostic tests for pulmonary embolism exem- volumetric CT with the single-breath-hold technique – comparison with pulmonary angiography. Radiology plify the problems often surrounding the introduction 1992; 185:381-7. of new tests. Due to the initial enthusiasm, a new test 12. van Rossum AB, Pattynama PM, Ton ER, et al. Pul- is often implemented in clinical use before proper monary embolism: validation of spiral CT angiography evaluation has occurred. Studies are also often per- in 149 patients. Radiology 1996; 201:467-70. formed without adequate data from earlier phases of 13. Ferretti GR, Bosson JL, Buffaz PD, et al. Acute pul- monary embolism: role of helical CT in 164 patients development. We would suggest that guidelines for with intermediate probability at ventilation-perfusion the proper evaluation of diagnostic tests before their scintigraphy and normal results at duplex US of the implementation in routine clinical practice are neces- legs. Radiology 1997; 205: 453-8. Educational Session 9 Chairman: J.W. Ten Cate Haematologica 1999; 84:(EHA-4 educational book):82-84 Trials that have influenced clinical practice in the treatment of thrombosis

Clinical significance of non clinical end-points in studies on the prevention of venous thromboembolism GIANCARLO AGNELLI Istituto di Medicina Interna e Medicina Vascolare, Università di Perugia, Perugia, Italy

The end-point(s) in studies on the disadvantages of clinical and non-clinical end-points prevention of venous thromboembolism will be discussed. The large majority of studies on the prevention of venous thromboembolism are carried out in patients Accuracy of venography as end-point undergoing surgery. Indeed, post-operative pul- measurement in clinical trials on monary embolism is the most common cause of pre- prevention of venous thromboembolism ventable death in high risk patients, such as those Contrast venography of the lower limbs is the undergoing major orthopaedic and cancer surgery. method of choice to measure the end-point in clini- The definition of the proper end-point to be used in cal trials on the prevention of DVT. Venography is clinical trials on the prevention of venous throm- not routinely performed in the large majority of the boembolism has been a matter of debate for a long hospitals. Indeed, the introduction and validation of time. Mortality is the ideal end-point for clinical tri- real time B-mode ultrasonography in patients with als on prevention of pulmonary embolism. However, clinical suspicion of DVT have reduced the use of post-operative mortality, although high in epidemio- venography in the diagnosis of this disease. In clini- logical terms, is relatively low in statistical terms mak- cal studies on prophylaxis of venous thromboem- ing the study sample size quite large. Furthermore, bolism the study centres are selected on the basis of the overall mortality during the peri-operative period the availability of venography. A standardised tech- is influenced by factors unrelated to venous throm- nique is usually adopted to improve the interpreta- boembolism. This background noise reduces the speci- tion of the results. However, a certain degree of inter- ficity of the study intervention and amplifies the sam- observer variability may be difficult to eliminate and ple size required to demonstrate the study hypothe- can confound the results of multicentre clinical tri- sis. An alternative end-point to death is death due to als.1 We determined the inter-observer agreement venous thromboembolism. Unfortunately, this end- between the local and central assessment of venogra- point has a limited diagnostic accuracy. phies in a multicentre trial comparing enoxaparin and The ideal end-point is both reliably measurable and placebo in the prevention of DVT after elective neu- clinically relevant. Unfortunately, accuracy and clini- rosurgery. The central and local adjudication panels cal relevance are not necessarily associated as far as were both blind with respect to the assigned treat- study end-points are concerned. This is indeed the ment. The central panel was unaware of the local case in clinical studies on the prevention of post- adjudication. Venographies were adjudicated as pos- operative venous thromboembolism. Studies target- itive, negative or inadequate for adjudication and ing accuracy use venography to measure the end- positive venographies as proximal or distal DVT. point. However, patients with venography-detected Inter-observer agreement was assessed according to post-operative deep vein thrombosis (DVT) are the Cohen’s inter-observer variability index (K index). almost invariably asymptomatic and the clinical rel- All 266 venographies (8 monolateral) were consid- evance of venography-detected DVT is still unclear. ered adequate for adjudication by both the central Studies targeting clinical relevance are based on clin- ically overt events. Unfortunately, the clinical diag- and local panels. A disagreement was found in 25 nosis of DVT and pulmonary embolism has a low cases, K index = 0.75. Fourteen venographies adjudi- accuracy, both in terms of sensitivity and specificity. cated as negative centrally were considered positive Furthermore, studies on the prevention of venous locally (3 were proximal DVT). Eleven venographies thromboembolism with clinical end-points require a adjudicated as positive centrally (1 was a proximal large number of patients because of the low number DVT) were considered negative locally. Enoxaparin of events. In this paper the relative advantages and was found to be more effective than placebo accord- ing to both the central and local adjudication: 16.9% versus 32.6% (relative risk, RR = 0.52. 95% CI 0.33- Correspondence: Giancarlo Agnelli, MD Istituto di Medicina Interna e 0.82) according to central adjudication; 18.5% ver- Medicina Vascolare Università di Perugia Via Enrico dal Pozzo Perugia, Italy Phone : +39-075-572 2905 Fax : +39-075-572 2011 E-mail: sus 33.3% (RR = 0.56; 95% CI 0.36-0.87) according [email protected] to local adjudication. It was concluded that there Session #9 – Trials that have influenced clinical practice in the treatment of thrombosis 83 was a good inter-observer agreement between the Limitations of clinically overt events as central and local adjudication in clinical study centres end-points in clinical trials on preven- frequently involved in clinical studies on the preven- tion of venous thromboembolism tion of DVT. Disagreement was mainly confined to Although objectively confirmed venous throm- distal DVT. The cost and work overloading of central boembolic events are the most appropriate outcome assessment of venographies in multicentre trials on for prevention studies, they present some limitations.7 DVT prevention seems not to be justified. Reduced accuracy and validity of clinically overt events, as study end-points, primarily affect the gen- Role of real time B-mode ultrasonography eralisation (external validity) of the study results. The Real time B-mode ultrasonography in asympto- internal validity of the comparison between different matic patients is insufficiently accurate.2-4 The limit- treatment interventions will not be compromised, as ed sensitivity of this test might be compensated for by long as the design of the clinical trial is methodolog- enrolling additional patients if the diagnostic tech- ically sound. nique is highly specific. This is the case for proximal The results of clinical outcome study may be con- DVT but not for distal DVT. High quality studies using founded by the non-antithrombotic pharmacological compression, duplex and colour Doppler ultrasound effect of the study drug. This is clearly illustrated by show sensitivities of 42-70% and positive predictive studies assessing the value of aspirin in the preven- values of 35-83% for proximal DVT. For calf DVT, the tion of venous thromboembolism in orthopaedic accuracy is undoubtedly lower. surgery. In the first studies objective testing was per- formed to confirm the diagnosis in symptomatic Limitations of non-clinical end-points in patients only. These studies were randomised but studies on the prevention of venous were not blind and adopted clinical diagnosis for the thromboembolism evaluation of the occurrence of DVT. The results of The value of non-clinical outcomes to measure the these trials indicated that aspirin was effective in the efficacy of antithrombotic agents in studies on the prophylaxis of DVT. However, since aspirin has a sig- prevention of venous thromboembolism is not clear. nificant analgesic effect and pain is one of the most Studies based on venography-detected asymptomatic frequent clinical symptoms of DVT, the efficacy of DVT overestimate the event rates compared with aspirin was probably overestimated. Indeed, aspirin studies based on clinical end-points. When compar- might have concealed pain without preventing the ing the efficacy of one agent versus another, the underlying DVT and the subsequent risk of pulmonary absolute benefit of the more efficacious agent may embolism. Subsequent trials, conducted by the same be overestimated if the event rate is high. The long- investigators and by others, adopted correct meth- term outcome of asymptomatic patients with untreat- ods of DVT adjudication and indicated that aspirin ed venography-detected DVT is unknown. Some has a modest efficacy, being much less effective than authors consider this event only a surrogate end-point a number of other available prophylactic methods. of mortality or morbidity due to venous thromboem- bolism. Other authors believe that venography- Major bleeding as an end-point in clinical detected DVT has an undoubted pathophysiological trials on prevention of venous relevance and is predictive of clinical outcome. thromboembolism The major side effect of antithrombotic therapy is Clinically overt events as end-points in bleeding. This unexpected event may cause consider- clinical trials on prevention of venous able patient morbidity and may even be fatal. Typi- thromboembolism cally, major bleeding has been considered to be a sec- Symptomatic outcomes are of greatest concern to ondary outcome of trials of antithrombotic therapy. patients and their physicians.5 Prevention of these However, several recent studies have considered outcomes is the primary goal of antithrombotic pro- symptomatic objectively documented venous throm- phylaxis.6 To minimise bias, randomised controlled boembolism and major bleeding as a composite out- clinical trials on prevention of post-operative venous come measure to evaluate the effectiveness of anti- thromboembolism are usually designed taking into thrombotic therapy. This approach may be adequate account the following essential requirements. Only to assess the benefits and risks of individual anti- patients developing symptoms presumed to be due to thrombotic agents. venous thromboembolism are considered for study end-points. The criteria for diagnosis of venous Conclusions thromboembolic events are based on validated objec- Clinically overt events confirmed by objective test- tive diagnostic testing. Investigators blind to the study ing are the most relevant outcomes to be used to intervention review suspected outcome events. Ideal- determine the rates of unfavourable events in studies ly the study is double blind, so that there is no refer- designed to set the standard of care. However, low ral bias whereby patients (or physicians) might be sensitivity and specificity hamper clinical diagnosis of have (or investigate) differently. DVT and pulmonary embolism. The gold standards 84 G. Agnelli for the diagnosis of asymptomatic DVT and pul- patients. Ann Intern Med 1992; 117:735-8. monary embolism are bilateral venography and pul- 3. Wells PS, Lensing AWA, Davidson BL, Prins MH, Hirsh monary angiography, respectively. Both these tests J. Accuracy of ultrasound for diagnosis of deep venous are invasive and difficult to interpret. The clinical val- thrombosis in asymptomatic patients after orthopedic ue of asymptomatic venography-detected DVT is still surgery. A meta-analysis. Ann Intern Med 1995; 122: 47-53. unknown. However, until it can be demonstrated that 4. Robinson KS, Anderson DR, Gross M, et al. Ultra- asymptomatic venography-detected DVT are entirely sonographic screening before hospital discharge for without deleterious consequences, it seems prudent deep venous thrombosis after arthroplasty: the post- to prevent them. The accuracy of non-invasive diag- arthroplasty screening study. A randomized controlled nostic techniques to assess study end-points has still trial. Ann Intern Med 1997; 127:439-45. to be adequately validated. 5. Fleming TR, DeMets DL. Surrogate end points in clin- ical trials: are we being misled? Ann Intern Med 1996; 125:605-13. 6. Clagett GP, Anderson FA, Heit J, Levine MN, Wheeler References HB. Prevention of venous thromboembolism. Chest 1995; 108:312S-334S. 1. Lensing AWA, Buller HR, Prandoni P, et al. Contrast venography, the gold standard for the diagnosis of 7. Leclerc JR, Gent M, Hirsh J, Geerts WH, Ginsberg JS. deep-vein thrombosis: improvement in observer The incidence of symptomatic venous thromboem- agreement. Thromb Haemostas 1992; 67:8-12. bolism during and after prophylaxis with enoxaparin. 2. Davidson BL, Elliott CG, Lensing AWA. Low accuracy A multi-institutional cohort study in patients who of colour Doppler ultrasound in the detection of prox- underwent hip or knee arthroplasty. Arch Intern Med imal leg vein thrombosis in asymptomatic high risk 1998; 158:873-9. Educational Session 9 Chairman: J.W. Ten Cate Haematologica 1999; 84:(EHA-4 educational book):85-87 Trials that have influenced clinical practice in the treatment of thrombosis

Randomised evaluation of treatment modalities in thromboembolic disorders JAN W. TEN CATE, HARRY R. BÜLLER Division of Vascular Medicine, Academic Medical Center, Amsterdam, The Netherlands

enous thromboembolism (VTE) is a serious Initial treatment UFH versus LMWH disorder and frequently heralds a fatal out- Rapid initiation of optimal anticoagulant treat- Vcome. The prevalence of VTE is approximately ment is an absolute requirement in patients with con- 2-3 per 1,000 inhabitants of the Western world per firmed DVT to prevent recurrent VTE at follow-up. A year. The clinical implications of deep venous throm- double-blind placebo-controlled study in the Nether- bosis (DVT) and pulmonary embolism (PE) will be lands3 clearly revealed the necessity for initial dose- discussed with a focus on complications, prevention adjusted heparin treatment. of these complications by rapid (initial) anticoagu- Twenty percent of patients who presented with lant treatment with unfractionated (UFH) or low- proximal DVT and received a placebo infusion, in molecular-weight heparin (LMWH), and by long- addition to concurrently commenced coumarin term treatment with coumarins. The various treat- treatment, had a recurrent VTE by three months’ fol- ment modalities for DVT and PE have changed con- low-up, while only 6.7% of those who received initial siderably over the last 10-20 years and will continue UFH treatment concurrent with coumarins did so. to do so in the next century. Major changes in the Once heparin treatment is started, careful labora- treatment of VTE have been facilitated by clinical tri- tory control of coagulation, using an activated par- als comparing entire treatment modalities rather tial thromboplastin test (APPT) to assess adequate than single components. heparinisation early in the course of therapy, is a mat- ter of some importance. The APPT is used to adjust Clinical implications of a first VTE event the dose of heparin at regular intervals, in order to The long-term clinical course of a first acute DVT maintain the clotting time prolongation within the has been revealed by carefully conducted, prospec- prescribed therapeutic range, i.e. an APPT that is tive, cohort studies in consecutive outpatients; mea- increased at least 1.5 times the normal range. Sub- surements included recurrent VTE, and death.1,2 therapeutic heparinisation within the first 24 hours is Potential risk factors for these outcomes were evalu- associated with a high rate of recurrent VTE at follow- ated, in particular the late discovery of malignancy. up (23.3%) versus 4.6% in patients receiving thera- The average cumulative incidence for recurrent VTE peutic doses of heparin.4 after the qualifying event was approximately 15% at These and other studies support the clinical signif- the first year and gradually increased to approxi- icance of immediate and optimal heparinisation in mately 25% at eight years follow-up. Of the initial patients with DVT. However, even in the best centres cohort of patients, approximately 15% died within some patients treated with intravenous heparin will the first year and about 70% survived the total of the receive sub-optimal therapy, reflecting that LMWH is eight years follow-up. Death was frequently due to the likely practical solution to this dilemma. The cancer which became clinically overt mainly within advantages of LMWH over UFH are longer plasma the first year of follow-up, this being an association half-life, an almost complete bio-availability follow- particularly found in patients having either idiopath- ing subcutaneous injection, as well as a reduced inter- ic or recurrent DVT. The elucidation of this associa- and intra-individual variability following subcuta- tion may have an impact on the design of cancer neous injection. These pharmacological characteris- screening and the detection of occult malignancy at tics allow for fixed dose treatment which only needs an early stage. The frequency of cancer observed in adjustment for specific bodyweight categories with- the first year was approximately 7.6-8%.2 Similarly, out requiring laboratory monitoring. incidence of fatal outcomes - also due to malignan- Initial in-hospital randomised studies comparing cy in most cases - was observed in patients with con- the efficacy and safety of fixed bodyweight-adjusted firmed PE. dose LMWH given subcutaneously against a UFH- adjusted dose given intravenously revealed promising results, i.e. at least a comparable efficacy, improved Correspondence: J.W. ten Cate, Academic Medical Center, Div. Vascu- safety and, interestingly, a reduced mortality at fol- lar Medicine (F4-211), Meibergdreef 9, 1105 AZ Amsterdam, Tel: international +31-20-5665976 – Fax: international +31-20-6968833 low-up. The results of published studies were – E-mail: [email protected] assessed by the Cochrane review group on peripher- 86 J. W. ten Cate et al. al vascular diseases;5 this report comprises all ran- Home treatment with LMWH versus in-hospital domised controlled clinical trials which were identi- treatment with UFH fied by electronic (Medline® and Embase®) or manu- Two major clinical studies have been reported, i.e. al (relevant journals) searches, and by communica- the Dutch Tasman Study6 and another study from tion with colleagues and relevant pharmaceutical Canada.7 Both studies – though utilising different companies. All studies were independently reviewed brands of LMWH – yielded similar results in terms of by three investigators; discordant results were solved efficacy, safety and mortality (Table 1). Moreover, by majority voting. Only those studies which directly both studies showed that out-of-hospital treatment compared fixed-dose LMWH with an adjusted dose with LMWH was, as expected, associated with a con- UFH for treatment of DVT and PE were included, and siderable reduction in days spent in the hospital and only if they were truly randomised, precluding prior hence an important reduction in costs, without a knowledge of next treatment allocation. Further inclu- detrimental effect on the quality of life.6 These data sion criteria were: an unconfounded comparison; an imply that out-of-hospital treatment of patients with objective confirmation of VTE before randomisation proximal DVT is a feasible option, but with the fol- by either contrast venography of compression ultra- lowing precautions: 1. patients should be referred to sonography for the diagnosis of DVT, and ventilation- expert centres to receive objective diagnosis of DVT; perfusion lung scanning and pulmonary angiography 2. patients should be analysed for risk factors for DVT; for the diagnosis of PE; a prospective long-term fol- and 3. community facilities for anticoagulant treat- low-up for objectively confirmed VTE recurrence and ment and control should be available. for mortality and an independent adjudication of pre- Long-term treatment with coumarins: shift in defined clinical end-points. duration of treatment? Studies that were excluded from the analysis were Until recently, the duration of oral coagulation fol- preliminary reports later duplicated in full reports; lowing a first or a recurrent episode of DVT has not dose-ranging studies which made use of doses of been submitted to well-designed clinical investigation. LMWH other than those currently in use; studies with Two studies from Sweden have now addressed this heparinoids. The following relevant clinical outcomes issue.8,9 were considered: In the first study8 direct comparison was made 1. symptomatic recurrent VTE at long-term follow- between six weeks’ versus six months’ treatment in up (> 3 months); 2. major haemorrhage during initial patients with a first thromboembolic event; patients treatment; 3. death due to any cause and in particu- with an inherited risk factor predisposing to VTE lar death associated with clinically overt malignancy. (antithrombin, protein C or S deficiency) were exclud- Of the 26 potentially eligible clinical trials, 13 sat- ed from the analysis. In the remaining patients a sig- isfied the inclusion criteria; these studies were pub- nificant reduction of recurrent VTE was achieved in lished over the years 1988-1997. Pooled analysis for the six months’ treatment group: OR for recurrence in all these clinical trials revealed a reduction of recurrent the six weeks group was 2.1 (95% CI 1.4-3.1) without VTE, at three to six months’ follow-up in favour of differences in mortality or major haemorrhage. By LMWH (OR 0.75; 95% CI 0.55-1.01; p=0.06). The analysis of subgroups, the OR in patients with tem- pooled efficacy analysis in patients with PE revealed porary risk factors for thrombosis was 1.9 (95% CI no such statistically significant difference (OR 0.91; 95% CI 0.42-1.97; p=0.02). Important and significant reduction in mortality was observed with LMWH treatment (OR 0.74; 95% CI 0.57-0.98; p=0.03). Mortality reduction was main- Table 1. Efficacy and safety outcomes in patients with prox- imal DVT treated with LMWH at home versus UFH in hospi- ly confined to patients with malignancies. tal This systematic review thus revealed that LMWH in a fixed dose given subcutaneously and only adjusted for bodyweight, compared to APPT-adjusted thera- Tasman study UFH LMWH peutic doses of UFH, was at least as effective and like- (n=198) (n=202) ly to be safer. These, and previous data, led to clini- cal studies investigating the feasibility and clinical effi- Recurrent VTE 17 (8.6%) 14 (6.9%) cacy and safety of home treatment with LMWHs. In Major bleeding* 4 (2.0%) 1 (0.5%) Mortality 16 (8.1%) 14 (6.9%) these studies a direct comparison was made between entire treatment modalities, e.g. in hospital using the established treatment module consisting of i.v. Canadian study UFH LMWH unfractionated heparin continuously by a syringe (n=253) (n=247) pump driven system, APTT-monitoring and dose- Recurrent VTE 17 (6.7%) 13 (5.3%) adjustment when necessary, versus ambulant treat- Major bleeding* 3 (1.2%) 5 (2.0%) ment (at home) with fixed-dose s.c. LMWH. Subcu- Mortality 17 (6.7%) 11 (4.5%) taneous injections were carried out by the patients, a close relative or a nurse. *During the initial treatment phase or within 48 hours thereafter. Session #9 – Trials that have influenced clinical practice in the treatment of thrombosis 87

0.8-4.5; p=0.24), while in patients with permanent risk factors, the OR was 2.3 (95% CI 1.5-3.6; References p<0.001). Hence, the risk reduction for recurrent VTE was observed more prominently in patients with per- 1. Prandoni P, Lensing AWA, Cogo A, et al. The long- manent risk factors receiving six months of treatment. term clinical outcome of acute deep venous throm- A similar VTE reduction was observed in patients bosis, Ann Intern Med 1996; 125:1-8. with a second episode of DVT or PE, who were receiv- 2. Prandoni P, Lensing AWA Büller HR, et al. Deep vein thrombosis and the incidence of subsequent sympto- ing continuing oral anticoagulant treatment (48 matic cancer. N Engl J Med 1992; 327: 1128-33. months’ follow-up) as compared to six months of 3. Brandjes DPM, Heijboer H, Büller HR, et al. Aceno- 9 treatment. As in the first study, patients with docu- coumarol and heparin compared with acenocoumarol mented congenital risk factors were excluded. alone in the initial treatment of proximal-vein throm- After four years’ follow-up, the rate of recurrence in bosis. N Engl J Med 1992; 327:1485-9. the six months group was 20.7% versus 2.6% in the 4. Hull RD, Raskob GE, Brant RF, et al. Relation between patients receiving continuing coumarin treatment, for the time to achieve the lower limit of the APTT thera- a relative risk of 8.0 (95% CI 2.5-25.9). As expected, peutic range and recurrent venous thromboembolism the risk of bleeding was lower in the six months’ treat- during heparin treatment for deep vein thrombosis. ment group, i.e. 0.3 (95% CI 0.1-1.1). Arch Intern Med 1997;157:2562-8. The following conclusions may be derived from 5. van den Belt AGM, Prins MH, Lensing AWA, Hirsh J. these and other data: The initial treatment of venous thromboembolism. 1.patients with a first VTE event and having only a Ph.D. Thesis; University of Amsterdam, July 1998. 6. Koopman MMW, Prandoni P, Piovella F, et al. Treat- temporary risk factor are better off with 4-12 weeks ment of venous thrombosis with intravenous unfrac- of treatment (calf vein thrombosis: 4-6 weeks, prox- tionated heparin administered in the hospital as com- imal DVT or PE: 12 weeks); pared with subcutaneous low-molecular weight 2.patients with DVT or PE and without a temporary heparin administered at home. N Engl J Med 1996; risk factor are better off with six months of treat- 334:682-7. ment; 7. Levine M, Gent M, Hirsh J, et al. A comparison of low- 3.patients with a second episode of VTE deserve molecular-weight heparin administered primarily at longer treatment. Duration should be carefully con- home with unfractionated heparin administered in the sidered, taking into full account the risk for recur- hospital for proximal deep-vein thrombosis. N Engl J rent VTE and the risk of major bleeding especially Med 1996; 334: 677-81. with older patients; 8. Schulman S, Rhedin AS, Lindmarker P, et al. A com- 4.continuing coumarin treatment for at least four parison of six weeks with six months of oral anticoag- years, e.g. the duration of follow-up in this study9 ulant therapy after a first episode of venous throm- boembolism. N Engl J Med 1995; 332: 1661-5. should be seriously considered at least on the basis 9. Schulman S, Granqvist S, Holmstrom M, et al. The of the data currently available. duration of oral anticoagulant therapy after a second The duration of anticoagulation in patients with episode of venous thromboembolism. N Engl J Med inherited thrombophilia (excluded in the Swedish 1997; 336: 393-8. studies) is still uncertain and has been challenged 10. van den Belt AGM, Sanson BJ, et al. Recurrence of recently.10 Currently, indefinite treatment is the prevail- venous thromboembolism in patients with familial ing option, though this is not based on solid evidence. thrombophilia. Arch Intern Med 1997; 157:227-32. Educational Session 10 Haematologica 1999; 84:(EHA-4 educational book):88-91 Chairman: A. Polliack Chronic lymphocytic leukaemia

Cytogenetics and molecular genetics of chronic lymphocytic leukaemia DAVID G. OSCIER Department of Haematology, Royal Bournemouth Hospital, Bournemouth, UK

dvances in the detection of genetic abnormal- or spleen) but are more likely to reflect either techni- ities in chronic lymphocytic leukaemia (CLL) cal differences or the inclusion of patients with relat- Ahave occurred entirely in conjunction with ed chronic lymphoproliferative disorders such as technical developments. Early cytogenetic studies mantle cell lymphoma or splenic lymphoma with vil- which used PHA as a mitogen concluded that the lous lymphoctyes. karyotype in CLL was normal. With the advent of Secondly, studies in which the same patients have polyclonal B cell mitogens trisomy 12 and subse- been investigated with a variety of techniques such as quently structural abnormalities of chromosome cytogenetic analysis, interphase and metaphase FISH 13q14 were found in between 10 and 20% of and Southern blotting or PCR have shown that the patients with CLL. Abnormalities of chromosomes complexity of abnormalities is frequently much 11q, 6q, 17p and 14q were found in fewer than 5% greater than might be anticipated if a single tech- of cases. Cytogenetic analysis is a global technique nique was performed. For example a patient with a but suffers from two main disadvantages. Firstly it is 13q14 deletion on cytogenetic analysis and apparent relatively insensitive in detecting small deletions and heterozygous loss of a 13q14 marker on Southern secondly the metaphases obtained using standard blotting may harbour subclones with homozygous mitogens such as TPA are frequently derived from T loss only detectable when FISH is also used. cells rather than the malignant B cell clone. A recent Genetic abnormalities study evaluated the use of different combinations of mitogens including TPA, TNF-␣, SAC and IL2 in cas- 13q14 es of CLL and showed that by optimising the mito- Structural abnormalities of 13q14 are found in genic combination for each individual case, the pro- approximately 20% of CLL patients by karyotyping, in portion of B cell mitoses rose from 7 to 57% com- 30-40% by Southern blotting and in 50-60% using pared to the proportion following TPA stimulation interphase FISH. Chromosome analysis shows that alone and the incidence of clonal aberrations also two thirds of patients have a deletion of 13q14-q22 rose from 50 to 79%. or 13q12-q14 while the remainder have a transloca- The use of fluorescent in situ hybridisation with cen- tion involving 13q14. The translocations are fre- tromeric, YAC, or cosmid probes applied to non quently complex involving two or more partner chro- dividing peripheral blood cells, without prior culture, mosomes and invariably result in the loss of genetic has shown a significantly high incidence of abnor- material from the q14 region. Several groups have malities that are detectable by routine chromosome identified a minimum region of genetic loss contain- analysis. In one study which used probes for chro- ing numerous previously unidentified transcripts, mosomes 3q, 6q, 8q, 11q, 12q, 13q and 14q, telomeric to the retinoblastoma gene. The smallest abnormalities were detected in 82% of cases. Addi- region identified is less than 10kb. This includes exons tional abnormalities such as amplification of chro- from at least two genes which show little homology mosome 4q were identified in 2 recent studies using to previously identified genes. However no mutations comparative genomic hybridisation. In a small study have been discovered in these genes in a large series using highly polymorphic microsatellite markers on of patients with CLL.2 purified leukaemic cells loss of heterozygosity was 13q12 found on chromosome 3p and 9p.1 One group has reported a high frequency of dele- There are two important caveats to the interpreta- tions at 13q12 involving the BRCA 2 gene. While this tion of cytogenetic and genetic data in CLL. Firstly would be anticipated in cases with deletions of q12 when results differ between studies, this may be par- to q14 many patients had loss at q12 and q14 with tially explainable by the use of samples from patients retention of intervening sequences. Other groups at different stages of their disease, or from different have been unable to duplicate these findings partic- anatomical sites (blood, lymph node, bone marrow ularly using Southern blotting and the significance of q12 loss in CLL remains uncertain. Trisomy 12 Correspondence: David G. Oscier, Department of Haematology, Royal Bournemouth Hospital, Bournemouth UK. Tel. international +44-1-202- The incidence of trisomy 12 varies between 10 and 704794 Fax. international +44-1-202-00248. 20% evaluated cytogenetically and 15 to 45% using Session #10 – Chronic lymphocytic leukaemia 89 interphase FISH. Case selection is an important factor In those studies in which a higher incidence is found in determining the incidence of this abnormality (see the lymphocyte morphology and immunophenotype below). Less frequently translocations involving 12p11- are atypical and such cases may represent the leukaemic 12 are found as well as translocations or duplications phase of mantle cell lymphoma. The t(14;19) is found of 12q12-14. Comparative genomic hybridisation in less than 0.5% of cases and involves the BCL3 gene (CGH) also reveals amplification of 12q and detailed which encodes an i-kB-like protein. genetic analysis on patients with structural abnormal- 6q ities of chromosome 12 will be important in defining the key genetic defects of importance in the pathogen- Structural abnormalities of 6q are found in less esis of CLL. than 5% of cases on chromosome analysis and in 7% using FISH. There has been less progress in charac- 11q23 terising the deletions involving 6q compared to those Although deletions and translocations of 11q23 are of 13q and 11q. In NHL two or three separate regions found in less than 5% of patients by chromosome of loss have been found. Preliminary data in CLL analysis, the incidence rises to 20% using interphase shows that deletions of q21 are more common than FISH. A critical region of loss of 2-3 MB has been distal deletions involving q27. identified containing the ATM gene. Recent studies have suggested that ATM loss may be important in Clinical and laboratory correlations CLL. In one study 8 patients with a cytogenetic abnor- Lymphocyte morphology mality of 11q were also shown to have an ATM muta- tion in the other allele. A second study found reduced One of the cardinal diagnostic features of CLL is or absent ATM protein in 8 of 20 patients with CLL. the presence of small lymphocytes with central round Six of the 8 cases had ATM mutations and in two they nuclei, condensed chromatin and minimal agranular were germline. There was no loss of heterozygosity at cytoplasm. Morphology is described as atypical if 11q23 in any case.3 more than 10% of cells have the features of prolym- phocytes or if more than 15% of cells have either a 17p13 cleaved nucleus or lymphoplasmacytoid features. The Although deletions and translocations involving this incidence of chromosome abnormalities is higher in region are infrequent in CLL (less than 5%) abnor- patients with atypical morphology and there is a malities of p53 are found in 10 to 20% of cases of strong correlation with trisomy 12 and p53 abnor- CLL. The incidence varies with the stage of the disease malities. (see below) and the techniques used to study p53 abnormalities. The most frequently used methods are Clinical features measurement of p53 protein expression using immu- CLL shows marked clinical heterogeneity not only in nocytochemistry or flow cytometry, identification of the rate of disease progression but also in the distri- p53 gene mutations, and detection of p53 loss using bution of the leukaemic population. Deletions of 11q FISH. Although there is a strong correlation between and 6q are associated with bulky lymphadenopathy5 p53 gene mutation, loss of the remaining allele and and a recent study has shown a correlation between overexpression of p53 protein this is not absolute.4 11q abnormalities and expression of adhesion mole- For example not all patients with a mutation show cules of the leukaemic cells. allelic loss and approximately 15% of mutations do not lead to p53 protein overexpression. VH gene status Immunoglobulin gene translocations The variable regions of immunoglobulin heavy and Translocations involving the immunoglobulin heavy light chains undergo somatic hypermutation during chain locus are both frequent and critical to the their passage through germinal centres. Early small pathogenesis of many chronic lymphoproliferative studies in CLL showed that most cases had unmutat- disorders but appear to be rare in CLL. The t(14;18) ed VH genes and it was hypothesised that CLL arose and the light chain variants t(2;18) and t(18;22) are from a naive B cell, possibly CD5 positive mantle cells. found in 1-2% of cases of CLL and juxtapose the Two large studies have now shown that VH genes in immunoglobulin gene loci with the BCL2 gene. The CLL are mutated in approximately 50% of cases.6 Pre- BCL2 breakpoints are usually 5/ and distinct from liminary data have shown a strong correlation between those found in follicle centre cell lymphomas. BCL2 trisomy 12 and unmutated VH genes and between protein expression is often higher than typically found abnormalities of 13q14 and VH gene mutations. A in CLL but patients with BCL2 rearrangements appear larger study is required to confirm these findings. to have no distinctive clinical features. The incidence of the t(11;14) in CLL is controver- Prognosis sial. In most studies the incidence is extremely low A large multicentre study has clearly shown an asso- and many would regard its presence as a diagnostic ciation between complex karyotypes and poor prog- exclusion criterion. nosis.7 Among patients with single chromosome 90 D. G. Oscier abnormalities, those with trisomy 12 had a shorter an abnormal karyotype is difficult since the original survival than those with structural abnormalities of normal karyotype may have derived from T cells rather chromosome 13 or those with a normal karyotype. than a mitotically inactive leukaemic clone. Sequen- The introduction of interphase FISH has enabled a tial FISH studies show the occasional acquisition of more sophisticated analysis of the prognostic signifi- trisomy 12 and RB1 loss during the course of the dis- cance of genetic abnormalities.8 The results of a ease and also the expansion of the trisomy 12 clone recent study are shown in the table, and emphasise particularly in patients with progressive disease.9 No the poor prognosis of patients with p53 abnormali- acquisition or expansion at the D13S25 locus was ties, which are frequently accompanied by drug resis- found suggesting that loss in this region is an early tance both to alkylating agents and purine analogues. event in CLL. In contrast p53 mutation or loss is much Another study has yielded similar results but also commoner in advanced disease and may be associ- showed that in patients with abnormalities of 13q14, ated with Richter’s transformation. survival is considerably better for patients with dele- The mechanisms responsible for genomic abnor- tions at D13S25 compared to patients with RB1 loss. malities and instability in CLL are poorly understood. In a multivariate analysis which included clinical Microsatellite instability appears to be rare in CLL. stage, lymphocyte morphology, and response to ther- Shortening of telomeres is a feature of advanced CLL apy both D13S25 loss and p53 deletions were inde- but it is not known whether this correlates with chro- pendent prognostic variables. mosome abnormalities.10 Both p53 mutations and ATM loss can be responsible for genetic instability. Genetic instability in CLL Since ATM mutations may be germline, they may be The predominant chromosomal abnormalities in linked to the intriguing clinical observation of antici- CLL involve either the loss or gain of genetic materi- pation in familial CLL in which the disease appears to al. Translocations are less frequent and also appear to arise approximately 20 years earlier in the children of result in genetic loss rather than the creation of a a parent with CLL. fusion gene or overexpression of an oncogene. There is data both in haematological malignancies Although genetic abnormalities are more common in and in solid tumours that the site of genetic abnor- advanced disease, 40-50% of patients with stage A malities such as chromosomal deletions and gene CLL have an abnormal karyotype. The combined use amplifications correlates with the presence of fragile of immunophenotyping and interphase FISH has sites. A recent study identified a polymorphic trinu- shown that genetic abnormalities such as trisomy 12 cleotide repeat sequence on chromosome 11q22-q23 and p53 loss only occur in a subclone of leukaemic within the minimal region commonly deleted in CLL cells. The initiating event (or events) in CLL remain and showed an association between the presence of unknown. the larger repeat and poor prognosis. Sequential chromosome analysis shows karyotypic evolution in only 10-15% of cases. However, inter- Conclusions pretation of an apparent transition from a normal to New techniques have uncovered a multitude of genetic abnormalities in CLL. Much further work is required to understand both the mechanisms where- by CLL cells acquire genetic defects and also their importance in the pathogenesis and clinical course of Table 1. Incidence and significance of genetic abnormalities in CLL (adapted from Dohner et al., ASH 1998). the disease.

Chromosome Genetic Incidence Median surv. Clinical References abnormality abnormality (%) (years) features 1. Gartenhaus RB. Allelic loss determination in chron- all del or (+)13q14 unknown 54 ic lymphocytic leukemia by immunomagnetic bead sorting and microsatellite marker analysis. Oncogene single del or 36 >15 typical 1997; 14: 375-8. (t)13q14 morphology 2. Liu Y, Corocoran M, Rasool O, et al. Cloning of two del 11q23 ?ATM 17 6.6 bulky candidate tumor suppressor genes within a 10 kb disease region on chromosome 13q14, frequently deleted in chronic lymphocytic leukemia. Oncogene 1997; 15: +12 unknown 15 10.9 atypical 2463-73. morphology 3. Stankovic T, Weber P, Stewart G, et al. Inactivation del 17p13 p53 8 3.6 p53 mutations of ataxia telangiectasia mutated gene in B-cell chron- drug resistance ic lymphocytic leukaemia. Lancet: 1999; 353: 26-30. 4. Cordone I, Masi S, Mauro FR, et al. p53 Expression del 6q21 unknown 7 11 bulky in B-cell chronic lymphocytic leukaemia: a marker of disease disease progression and poor prognosis. Blood1998; Session #10 – Chronic lymphocytic leukaemia 91

91: 4342-9. leukemia. Second IWCCLL compilation of data on 5. Dohner H, Stilgenbauer S, Fischer K, Bentz M, Lichter 662 patients. Leuk Lymphoma 1991; (Suppl.)21-5. P. Cytogenetic and molecular cytogenetic analysis of 8. Dohner H, Stilgenbauer S, James MR, et al. 11q dele- B cell chronic lymphocytic leukemia: specific chro- tions identify a new subset of B-cell chronic lympho- mosome aberrations identify prognostic subgroups cytic leukaemia characterised by extensive nodal of patients and point to loci of candidate genes. involvement and inferior prognosis. Blood 1997; 89: Leukemia 1997: 11(Suppl. 2)S19-S24. 2516-22. 6. Fais F, Ghiotto F, Hashimoto S, et al. Chronic lym- 9. Garcia-Marco JA, Price CM, Catovsky D. Interphase phocytic leukemia B cells express restricted sets of cytogenetics in chronic lymphocytic leukemia. Can- mutated and unmutated antigen receptors. J Clin cer Genet Cytogenet 1997; 94: 52-8. Invest 1998: 102:1515-25. 10. Counter CM, Gupta J, Harley CB, Leber B, Bacchet- 7. Juliusson G,Oscier D, Cahrton G. Cytogenetic find- ti S. Telomerase activity in normal leukocytes and in ings and survival in B-cell chronic lymphocytic hematologic malignancies. Blood 1995; 85: 2315-20. Educational Session 10 Haematologica 1999; 84:(EHA-4 educational book):92-93 Chairman: A. Polliack Chronic lymphocytic leukaemia

Morphology, atypical chronic lymphocytic leukaemia and prognostic features affecting choice of therapy DANIEL CATOVSKY Academic Department of Haematology and Cytogenetics, The Royal Marsden Hospital and Institute of Cancer Research, London, UK

orphological analysis is as important for lymphocytic leukaemia (B-PLL) can, in turn, be dis- diagnosis in a patient with lymphocytosis tinguished from CLL/PL by the immunophenotype as the histopathological review of lymph which, in CLL/PL, resembles that of CLL and in B- M 6,7 nodes is for the diagnosis of lymphoma. As for lymph PLL is similar to the B-cell lymphomas. node biopsies, the immunophenotype of the circu- lating cells is an essential complement to morphol- Immunophenotype ogy. Of several thousand cases of lymphocytosis Nowadays it is relatively simple to employ a battery studied in our Department, chronic lymphocytic of monoclonal antibodies and identify a patient with leukaemia (CLL) remains the most common diag- lymphocytosis as having predominantly B or T cells. nosis. Only a few cases of CLL are diagnosed by As the B-cell disorders are the more common and the lymph node histology. ones which often present diagnostic problems, we A detailed description of the morphological char- have focused our work on developing a scoring sys- acteristics of the lymphocytes in blood films is impor- tem based on the distinct antigenic profile of CLL tant for the differential diagnosis between CLL and which is different from all other cases of monoclon- other disorders, in particular non-Hodgkin’s lym- al B-cell lymphocytosis. Five antibodies can now phomas (NHLs) presenting or evolving with circulat- define the CLL immunophenotype (Table 1), includ- ing lymphoma cells. The most common low/interme- ing a component of the B-cell antigen receptor, the diate grade NHLs which evolve with leukaemia CD79b molecule.7 According to an earlier proposal6 (defined as a lymphocyte count greater than 5ϫ109/L) we used the intensity of CD22 as one of the five mark- are follicular lymphoma (FL), mantle cell lymphoma ers. Table 2 depicts the relative incidence of the var- (MCL), splenic lymphoma with villous lymphocytes ious B-cell diseases and their score using CLL mark- (SLVL), also known as splenic marginal zone lym- ers. It is obvious from Table 2 that the combination phoma, and lymphoplasmacytic lymphoma (LPL). of immunophenotype and morphology represents a Another reason why morphology is important in robust and objective methodology for diagnosis of CLL is in order to define cases described broadly as this group of lymphoid malignancies. Minor depar- atypical or mixed cell which include those with more tures from the score in CLL are seen mainly in cases than 10% prolymphocytes or CLL/PL,1-3 others with of atypical morphology and chiefly in CLL/PL. Such a spectrum of small and large lymphocytes, and yet cases may occasionally show positivity with FMC7 or others in which cells with cleft nuclei and/or plas- CD79b or strong staining for membrane immuno- macytoid features are seen.1,3 Cases of atypical CLL globulin (SmIg). Preliminary data from the MRC may have a different clinical course and are often the CLL3 trial suggest that positivity with FMC7 is asso- ones which present problems of differential diagno- ciated with worse outcome than in FMC7 negative sis, in particular with MCL and occasionally with FL patients and that this feature is independent of stage or SLVL. Circulating villous lymphocytes are seen not and age. only in SLVL but also in hairy cell leukaemia (HCL) and the rare HCL-variant. Again here, membrane CLL with more than 10% prolymphocytes markers are useful to separate CLL from SLVL4 and the latter from HCL and its variant.5 Cases of T-cell (CLL/PL) leukaemia and lymphoma such as T-PLL, large gran- CLL/PL has been recognised since the 1980s as a ular lymphocyte (LGL) leukaemia, adult T-cell distinct form of the disease. It is not clear whether leukaemia/lymphoma (ATLL) and Sézary syndrome, patients may present as typical CLL and evolve to typically have lymphocytosis and can be distin- CLL/PL. In at least half of the cases the percentage of guished from CLL by a combination of membrane prolymphocyte-like cells (prominent nucleolus, larg- markers and peripheral blood cytology. B-cell pro- er size, basophilic cytoplasm) tends to increase at a variable rate. In a minority they may reach values greater than 50% and therefore resemble B-PLL, although they retain the CLL immunophenotype; Correspondence: Prof. D Catovsky, The Royal Marsden Hospital, Ful- ham Road, London SW3 6JJ, UK. Tel. international +44-171-8082880 minor departures are seen in 10-20% of cases (see – Fax. international +44-171-3516420 – E-mail. [email protected] above). Session #10 – Chronic lymphocytic leukaemia 93

Table 1. Diagnostic criteria for CLL based on a scoring immunology (FMC7+) and genetic changes (+12, system.* abnormal p53, del 11q23) are gradually emerging as important for prognosis in individual series. Whether Marker CLL (Score) B-cell disorders other (Score) and how this should modify the choice of therapy is than CLL not clear. However, the systematic use of such mark- SmIg° Weak (1) Moderate/strong (0) ers in prospective clinical trials will allow such obser- CD5 Positive (1) Negative (0) vations to be made. The greater availability of treat- CD23 Positive (1) Negative (0) ment options in CLL, including several nucleoside FMC7 Negative (1) Positive (0) analogues used alone or in combination and the CD79b# Negative (1) Moderate/strong (0) increasing use of high dose therapy programmes with *After Matutes et al., 1994;6 updated by Moreau et al., 1997;7 °SmIg stem cell transplantation for younger patients will intensity; the pattern is always monoclonal by light chain restriction (K or L stain) in B-cell disorders; #This reagent is now used instead of CD22 (10). provide the answers if a systematic integration of lab- oratory markers of prognosis, including morphology, into the analysis of clinical trials is achieved.

Table 2. Incidence and CLL score in B-cell leukaemias and lymphomas with lymphocytosis. References

1. Bennett JM, Catovsky D, Daniel M-T, et al. The Disease Incidence* Score° French-American-British (FAB) Cooperative Group. Proposals for the classification of chronic (mature) B CLL and T lymphoid leukaemias. J Clin Pathol 1989; 42: typical 43% 4-5 atypical; CLL/PL 5% 3-5 567-84. 2. Criel A, Verhoef G, Vlietinck R, et al. Further charac- B-PLL 1.5% 0-1 terization of morphologically defined typical and atyp- ical CLL: a clinical, immunophenotypic, cytogenetic HCL# and prognostic study on 390 cases. Br J Haematol typical 8% 0-1 1997; 97:383-91. variant 1% 0-1 3. Matutes E, Oscier D, Garcia-Marco J, et al. Trisomy 12 defines a group of CLL with atypical morphology. NHL Correlation between cytogenetic, clinical and labora- SLVL 9% 0-1 tory features in 544 patients. Br J Haematol 1996; MCL 5% 0-2 92:382-8. FL 8% 0-2 4. Matutes E, Morilla R, Owusu-Ankomah K, Houlihan Other (LPL, not specified) 11.5% 0-1 A, Catovsky D. The immunophenotype of splenic lym- *From a series of 1,685 cases of B and T cell leukaemia studied in our phoma with villous lymphocytes and its relevance to Department over a four-year period (1992-1995). The T-cell cases repre- the differential diagnosis with other B-cell disorders. sented 8% and included T-PLL (2%), T-NHL (1%), LGL leukaemia (4%) and Blood 1994; 83:1558-62. Sezary syndrome (1%); °CLL score (Table 1); *#Using a score for HCL with 5. Matutes E, Morilla R, Owusu-Ankomah K, Houlihan four additional markers - CD25,CD11c,CD103,HC2 - it is possible to distin- guish typical HCL from the other B-cell disorders (Matutes et al 1994).5 A, Meeus P, Catovsky D. The immunophenotype of hairy cell leukemia (HCL). Proposal for a scoring sys- tem to distinguish HCL from B-cell disorders with hairy or villous lymphocytes. Leuk Lymphoma 1994; 14:57-61. Our group and others have reported important 6. Matutes E, Owusu-Ankomah K, Morilla R, et al. The associations of CLL/PL with genetic changes, namely immunological profile of B-cell disorders and proposal trisomy 122,3 and p53 abnormalities (deletions, over- of a scoring system for the diagnosis of CLL. Leukemia expression, mutations).8 These changes are known to 1994; 8:1640-5. 7. Moreau EJ, Matutes E, A’Hern RP, et al. Improvement be associated with poor response to therapy, high of the chronic lymphocytic leukemia scoring system labelling index with the antibody Ki-67, and overall with the monoclonal antibody SN8 (CD79b). Am J poor prognosis.2,8,9 As both trisomy 12 and p53 Clin Pathol 1997; 108:378-82. abnormalities are genetic changes acquired during the 8. Lens D, Dyer MJS, Garcia-Marco JA, et al. p53 abnor- evolution of CLL, it is likely that the morphological malities in CLL are associated with excess of prolym- expression of CLL/PL is also a secondary event caused phocytes and poor prognosis. Br J Haematol 1997; 99:848-57. by, or closely related to, the above changes. 9. Catovsky D. The search for genetic clues in chronic lymphocytic leukemia. Hematol Cell Ther 1997; 39: Prognostic features of CLL and choice S5-S11. of therapy 10. Catovsky D, Fooks J, Richards S, for the MRC Work- Stage, age and response to therapy remain the ing Party in Leukaemia in Adults. Prognostic factors in 10 chronic lymphocytic leukaemia: the importance of strongest prognostic features of CLL. However, the age, sex and response to treatment in survival. Br J new findings, such as morphology (CLL/PL), Haematol 1989; 72: 141-9. Educational Session 10 Haematologica 1999; 84:(EHA-4 educational book):94-95 Chairman: A. Polliack Chronic lymphocytic leukaemia

Chronic lymphocytic leukaemia. Current strategy and new perspectives of treatment KANTI R. RAI Long Island Jewish Medical Center, New Hyde Park, New York, USA

What is our current practice in CLL? Choice of drugs Until recently, when fludarabine was demonstrat- Risk-adapted approach ed to be superior to chlorambucil, the latter drug was A review of our current method of practice in the the one more often used for initial treatment of CLL. management of patients with CLL reveals a risk- It had previously been shown that fludarabine is an adapted approach. All CLL patients, from a thera- effective agent as salvage therapy for patients who peutic decision-making point of view, can be divided had failed chlorambucil, or other alkylating agent into the following groups: 3 1 therapy. Keating et al. in 1991 were the first to report 1. the “better” prognosis group. These are the low risk that fludarabine is also an effective drug in the front- category (modified Rai criteria) or Binet’s stage A line therapy of CLL. Two large multi-institutional ran- patients. Until recently it was not clear whether domised trials, one in the U.S. and Canada4 and the therapeutic intervention, with chlorambucil as a second in Europe5 have now been reported. Both of single agent or in combination with prednisone, these studies prove the superiority of fludarabine as would be of benefit to stage A CLL patients. How- the initial drug in the treatment of CLL. The North ever, the French Cooperative Group on CLL recently American study4 compared it against chlorambucil published a long follow-up report on a large ran- and the European study compared fludarabine versus domised trial conducted between 1980 and 2 cyclophosphamide, doxorubicin and prednisone 1990. This study assigned 609 stage A patients to (CAP-regimen).5 Fludarabine induced significantly either no treatment or chlorambucil and subse- more frequent complete remissions (CR) and overall quently an additional 926 stage A patients to remissions (CR plus partial remissions, PR), and flu- either notreatment or chlorambucil plus pred- darabine-induced remissions were significantly more nisone. With a median follow-up of more than six durable than those achieved with chlorambucil or years, this study demonstrated that early inter- CAP. vention with chlorambucil with or without pred- nisone did not prolong survival in stage A CLL Because of the above-noted results, fludarabine is patients. Therefore, the previously held view of increasingly being used as the drug of choice for the deferring therapy until the disease progresses to initial treatment of CLL. It is especially recommend- stage B or C was convincingly validated; ed when it is desirable to quickly achieve a CR or as 2. intermediate prognosis group. These are Binet’s good a PR as is achievable without undue toxicity. stage B or the intermediate-risk category (modified Fludarabine should be the treatment of choice for a Rai criteria) patients.1 Initiation of therapeutic relatively young CLL patient who, after chemothera- intervention is indicated if these patients exhibit py, may wish to consider the option of autologous or disease-related symptoms or evidence of progres- allogeneic haematopoietic stem cell transplant ther- sively increasing disease-activity (rapid doubling apy. With fludarabine there seems to be a somewhat time of blood lymphocyte count, or rapid increase higher risk of development of autoimmune haemolyt- in the size of enlarged lymph nodes or spleen, etc). ic anaemia than is expected in CLL patients, and pro- The aim of therapy is to obtain as good a state of longed exposure to fludarabine may result in signifi- remission as possible without an unacceptable lev- cant degrees of immunosuppression. It is best to el of toxicities; choose between fludarabine and an alkylating agent 3. worse prognosis group.1 These are the high-risk cat- on a case-by-case basis, taking into consideration the egory (modified Rai criteria) or stage C (Binet) desired therapeutic end-points and potential toxicities patients who are started on treatment without the with any of these drugs. usual wait-and-watch phase. The aim of therapy is to decrease the extent of disease and increase the New perspectives of treatment blood levels of haemoglobin and platelets which are abnormally low prior to treatment. Need to increase the incidence of CRs The results from the two large multi-institutional randomised studies mentioned above indicate that Correspondence: Dr. Kanti R. Rai, Long Island Jewish Medical Center, with fludarabine as the front-line therapy, approxi- 270-05, 76th Ave., New Hyde Park, NY 11040, USA. E-mail: rai @lij.edu mately 25 to 30 percent of CLL patients achieve a CR Session #10 – Chronic lymphocytic leukaemia 95 and an additional 40 percent achieve a PR for an over- approaches and identification of new agents with all (CR+PR) response incidence of about 70%. How- potential to increase the incidence of complete remis- ever, the overall survival curves have shown no signif- sions significantly, have brought us to a point where icant improvement. In my view, only CRs can lead to new perspectives in treatment may soon lead us to a longer life expectancy and PRs do not make a major curative strategies in chronic lymphocytic leukaemia. contribution towards improving survival. The progress recorded in three other examples of haematological malignancies during the past few decades lends sup- Acknowledgements port to this view: most recently hairy cell leukaemia, Research grants support from: United Leukaemia Fund, and Hodgkin’s disease and childhood acute lympho- Inc, Wayne Goldsmith Leukaemia Foundation, Berlex Labo- cytic leukaemia about twenty years earlier, witnessed ratories. The author acknowledges excellent assistance of Dale a significant increase in survival and even cures as soon Janson, RPA-C. as the incidence of CR increased to about 80 or 90%. With fludarabine we have made significant progress, increasing the CR incidence from the 3% achievable References with chlorambucil to 27%. New perspectives of CLL 1. Rai KR, Patel DV. Chronic lymphocytic leukemia. In: therapy will be directed towards further increase of Hoffman R, Benz, Jr EJ, Shattil SJ, Furie B, Cohen HJ, this proportion – to 80 or 90% CRs, we hope. Silberstein LE, eds. : Basic Principles and Practice. 2nd ed. New York: Churchill Livingstone; New therapies 1995. p. 1308-22. The most attractive agents on the horizon are mon- 2. Dieghiero G, Maloum K, Desablens B, et al. Chlo- oclonal antibodies – Campath 1-H6 and Rituxan,7 rambucil in indolent chronic lymphocytic leukemia. N Engl J Med 1998; 338:1506-14. which are currently undergoing clinical trials in previ- 3. Keating MJ, Kantarjian H, O’Brien S, et al. Fludara- ously untreated patients with CLL. Radiolabelled anti- bine: A new agent with marked cytoreductive activity bodies such as Bexxar (I-131 labelled anti-CD20) are in untreated chronic lymphocytic leukemia. J Clin likely to enter clinical trials in the near future. It is Oncol 1991; 9:44-9. expected that fludarabine and a monoclonal antibody 4. Rai KR, Peterson B, Elias L, et al. A randomized com- used together, either simultaneously or sequentially, parison of fludarabine and chlorambucil for patients with previously untreated chronic lymphocytic will enable us to increase the incidence of CR signifi- leukemia: a CALGB, SWOG, CTG/NCI-C and ECOG cantly from the base line of 25-30%. Intergroup study. Blood 1996; 88:141a. 5. Johnson S, Smith AG, Loffler H, et al. Multicentre Newer drugs prospective randomized trial of fludarabine versus Interesting new drugs,8 which are currently in phase cyclophosphamide, doxorubicin and prednisone (CAP) for treatment of advanced stage chronic lym- I and phase II clinical trials, are listed below: phocytic leukemia. Lancet 1996; 347:1432-8. • protein kinase-C (PKC) inhibitor: UCN-01; 6. Osterborg A, Fassas AS, Anagnostopoulos AMJ, • cyclin-dependent kinase inhibitor: Flavopiridol; Catovsky D, Mellstedt H. Humanized CD52 mono- • new nucleotide analogues: Pro-drug for Ara-G clonal antibody Campath as first-line treatment in (Compound GW 506-U78). chronic lymphocytic leukaemia. Br J Haematol 1996; 93:151-3. 7. Jensen M, Winkler U, Manzke O, Diehl V, Engert. Although bone marrow and haematopoietic stem Rapid tumor lysis in a patient with B-cell chronic lym- cell transplantations have been done both in autolo- phocytic leukemia and lymphocytosis treated with an gous as well as allogeneic (HLA-matched sibling anti-CD20 monoclonal antibody (IDEC-C2B8, ritux- donor) settings, this method of treatment in CLL will imab). Ann Hematol 1998; 77:89-91. need controlled clinical trials before its true value can 8. Byrd JC, Rai KR, Sausville EA, Grever ME. Old and new therapies in chronic lymphocytic leukemia: Now is the be determined. A recent report suggesting benefits time for a reassessment of therapeutic goals. Semin with non-myeloablative chemotherapy, allogeneic Oncol 1998; 25:64-74. stem cell transplant followed 2 to 3 months later with 9. Khouri IF, Keating M, Korbling M, et al. Transplant- donor lymphocyte infusion, especially for its graft-ver- lite: induction of graft-versus-malignancy using flu- sus-leukaemia effect is worthy of further exploration.9 darabine-based nonablative therapy and allogeneic In addition to haematopoietic stem cell transplanta- blood progenitor-cell transplantation as treatment for lymphoid malignancies. J Clin Oncol 1998; 16:2817- tion, new approaches in the future will also focus on 24. gene therapy in CLL.10 10. Kipps TJ. Chronic lymphocytic leukemia. Curr Opin In summary, the currently applied risk-adapted Hematol 1998; 5: 244-53. Educational Session 10 Haematologica 1999; 84:(EHA-4 educational book):96-97 Chairman: A. Polliack Chronic lymphocytic leukaemia

Prognostic factors in chronic lymphocytic leukaemia JACQUES-LOUIS BINET Centre d’Ecologie Cellulaire, Hôpital de la Salpêtrière, Paris, France and French Cooperative Group on Chronic Lymphocytic Leukaemia and Centre Claude Bernard

he best definition of indolent chronic lympho- or progressive splenomegaly. cytic leukaemia (CLL) is still clinical staging but 5. Massive nodes or clusters (i.e. > 10 cm in longest Tbiological features will improve this prognostic diameter) or progressive lymphadenopathy. approach. 6. Progressive lymphocytosis with an increase of > 50% over a 2-month period, or an anticipated Clinical staging doubling time of less than 6 months. Before the staging systems, male sex, young age, 7. Marked hypogammaglobulinaemia or the devel- anaemia, thrombocytopenia, blood lymphocytosis, opment of a monoclonal protein in the absence of lymph nodes, splenomegaly and cytological features any of the above criteria for active disease is not were, in univariate analysis, prognostic factors in CLL. sufficient for protocol therapy. In 1975, Rai et al. described the first staging system with five stages and their respective survival times.1 Patients with CLL may present with a markedly ele- The number of clinical stages can be reduced to vated leukocyte count; however, the symptoms refer- three: low risk, intermediate risk and high risk. In able to leukocyte aggregates that develop in patients 1981, we presented a three stage system.2 Jaksic et al.3 with acute leukaemia rarely occur in patients with published an evaluation of total tumour mass (TTM) CLL. Therefore, the absolute lymphocyte count based on the number of peripheral blood lympho- should not be used as the sole indicator for treat- cytes, the diameter of the largest palpable lymph ment, but should be included as a part of the total node and enlargement of the spleen. Phillips, Run- clinical picture, which includes the lymphocyte dou- dles, Rozman and Baccarani described other clinical bling time (see earlier). staging systems. To give a better definition of indo- lent CLL, Montserrat defines smouldering CLL (Hb Biological staging ≥ ϫ 9 120 g/L and lymphocytes < 30 10 /L and we divide Biological features can be studied in four parts: ≤ ϫ 9 stage A into A’ with lymphocyte count of 30 10 /L classical, new features, cytogenetic abnormalities and and haemoglobin concentration of 120 g/L and A’’ immunoglobulin heavy chain genes. ϫ 9 with lymphocyte count of 30 10 /L and haemoglo- For Rozman and Montserrat, bone marrow histo- bin concentration of 120 g/L or both. In patients with logic pattern was, in 1984, the best single prognostic stage A’ disease survival is close to that of a sex- parameter in CLL.5 Retrospective lymphocyte doubling matched and age-matched normal population. In time is very useful.6 Some patterns of immu- 1996, the National Cancer Institute4 revised guide- nophenotype can be used. Serum soluble CD23 lev- lines for treatment of aggressive forms of disease: el is more sensitive than ␤2 microglobulin and LDH 1. a minimum of any one of the following disease in our experience.7 related symptoms must be present: a. weight loss ≥ 10% within the previous 6 months; The overall frequency of p53 protein positivity in CLL is 15% but it is a marker of disease progression b.extreme fatigue (ie. ECOG PS 2 or worse, can- 8 kip1 not work or unable to perform usual activities); and poor prognosis. Cell cycle inhibitor p27 is c. fever of greater than 100.5°F for ≥ 2 weeks with- strongly correlated with both lymphocyte and total out evidence of infection; tumour mass doubling. High p27 expression may be d.night sweats without evidence of infection. a valuable kinetic marker by providing instantaneous 9 2. Evidence of progressive marrow failure as mani- estimation of the disease doubling time. Cellular fested by the development of, or worsening of, expression and serum release of adhesion molecules, anaemia and/or thrombocytopenia. degree of abnormal angiogenesis, serum levels of IL- 3. Autoimmune anaemia and/or thrombocytopenia 8 and IL-6, and soluble forms of some molecules poorly responsive to corticosteroid therapy. belonging to the growth factor receptor super- 4. Massive (i.e. > 6 cm below the left costal margin) family should be better explored in prognostic stud- ies.10 Patients with 13q14 abnormalities characteristi- Correspondence: Jacques-Louis Binet, Centre d’Ecologie Cellulaire – Hôpital de la Salpêtrière, 47 boulevard de l’Hôpital, 75651 Paris cally have a benign disease that usually manifests as Cedex 13, France. an isolated, stable or slowly progressive disease. Session #10 – Chronic lymphocytic leukaemia 97

These patients survive as long as their age-matched 2. Binet J-L, Auquier A, Dighiero G, et al. A new prog- controls. In contrast, a significant association nostic classification of chronic lymphocytic leukemia between trisomy 12 and CLL with atypical morpho- derived from a multivariate survival analyses. Cancer logic and/or immunologic features, high proliferative 1981; 48:198-206. 3. Jaksic B, Vitale B. Total Tumor Mass Score (TTM): a activity, advanced disease and poor prognosis have new parameter in chronic lymphocytic leukaemia. Br been reported. Forty-three out of 214 (20%) patients J Haematol 1981; 49:405-13. exhibited 11q23 deletions, which affected the prog- 4. Cheson BD, Bennett JM, Grever M, et al. National nosis of patients under 55 years of age. Given the con- Cancer Institute sponsored working group guidelines flicting data concerning clinical risk factors in young for chronic lymphocytic leukemia: revised guidelines patients suffering from CLL, this biological finding for diagnosis and treatment. Blood 1996; 87:4990-7. may be of great relevance in selecting patients for 5. Rozman C, Montserrat E, Rodriguez-Fernandez J-M, et intensive treatment approaches.11 Döhner et al. stud- al. Bone marrow histologic pattern, the best single prognostic parameter in chronic lymphocytic ied 338 B-CLL patients by interphase cytogenetics leukemia: a multivariate survival B anlaysis of 329 cas- using a disease-specific set of diagnostic DNA probes. es. Blood 1984; 64:642-8. Patients with deletion 13q as a single cytogenetic 6. Montserrat E, Sanchez-Bisono J, Viñolas N, Reverter abnormality had the longest median survival (> 15 JC, Rozman C. Natural history of chronic lymphocyt- years), followed by those with 6q deletion and tri- ic leukemia: on the progression and prognosis of ear- somy 12 (median survival 11 and 10.9 years, respec- ly clinical stages. Nouv Rev Fr Hematol 1988; 30:359- tively). In contrast, 17p and 11q deletions were asso- 61. ciated with rapid disease-progression and shorter sur- 7. Sarfati M, Chevret S, Chastang C, et al. Prognostic 12 importance of serum levels of CD23 in chronic lym- vival (median times, 3.6 and 6.6 years, respectively). phocytic leukemia. Blood 1996; 88:4259-64. Configuration role of immunoglobulin heavy chain 8. Cordone I, Masi S, Mauro FR, et al. p53 expression in genes in prognostic features has been recently dis- B-cell chronic lymphocytic leukemia: a marker of dis- covered. Hamblin13 proved that, in CLL, germline con- ease progression and poor prognosis. Blood 1998; figuration of immunoglobulin heavy chain genes is 91:4342-9. associated with a more aggressive form of the disease. 9. Vrhovac R, Delmer A, Tang R, Marie J-P, Zittoun R, Patients whose tumours showed somatic mutation of Ajchenbaum-Cymbalista F. Prognostic significance of the cell cycle inhibitor p27kip1 in chronic B-cell lym- VH genes characteristically had stable stage A disease phocytic leukemia. Blood 1998; 91:4694-700. with typical morphology, a normal karyotype or 10. Molica S. It is time for a reassessment of prognostic abnormalities at 13q14. Damle14 identified the same features in B-cell chronic lymphocytic leukemia? two groups of B-CLL patients who differ in clinical Hematology and Cell Therapy 1999; in press. course and in response to therapy. Difference in CD38 11. Döhner H, Stilgenbauer S, James. 11q deletions iden- expressing cells between the mutated and unmutated tify a new subset of B-cell chronic lymphocytic group was highly statistically significant (p < 0.001). leukemia characterized by extensive nodal involve- ments and inferior prognosis. Blood 1995; 89:2516- 22. Therapeutical staging? 12. Döhner H, Stilgenbauer S, Krober A, et al. Chromo- The best prognostic factor in CLL is treatment some aberrations identify prognostic subgroups of B- response but can in vitro tests predict drug sensitivity? cell chromosome lymphocytic leukemia. Blood 1998; This question remains a matter of debate.15 92 (suppl. 1):429a. 13. Hamblin TJ, Davis Z, Oscier DG, Stevenson FK. In In conclusion, biological parameters and chemo- chronic lymphocytic germline configuration sensitivity data may be incorporated into clinical of immunoglobulin heavy chain genes is associated with a more aggressive form of the disease. Blood prognostic models thus leading to the formulation of 1998; 92 (suppl. 1):515a. clinico-biological and therapeutic systems for CLL. 14. Damle RN, Fais F, Wasil T, et al. B-CLL patients can be divided into two distinct clinical categories based on CD38 expression and Ig V gene mutation status. References Blood 1998; 92 (suppl. 1):431a. 15. Kitada S, Andersen J, Akar S, et al. Expression of apop- 1. Rai K, Sawitsky A, Cronkite EP, Chanana AD, Levy RN, tosis-regulating proteins in chronic lymphocytic Pasternack BS. Clinical staging of chronic lymphocyt- leukemia: correlations with in vitro and in vivo ic leukemia. Blood 1975; 46:219-34. chemoresponses. Blood 1998; 91:3379-89. Educational Session 11 Haematologica 1999; 84:(EHA-4 educational book):98-101 Chairman: A.M. Hagemeijer New cytogenetics

Metaphase FISH, microdissection, and multicolour FISH. Applications in haematology CRISTINA MECUCCI, DANIELA FALZETTI, ROBERTA LA STARZA Ematologia, University of Perugia, Italy

onventional cytogenetics has been greatly After over-night hybridisation at 37°C, post- enriched by the development of molecular hybridisation washes of the slides are performed Cgenetics. DNA probes labelled with haptens or under conditions in which stringency can be modu- fluorochromes and used in experiments of fluores- lated to remove the excess of probe and keep stable cence in situ hybridisation (FISH) on metaphases allow hybrids. Probes labelled with haptens must be us to go into molecular details of structural chromo- detected with fluorochrome conjugated molecules. somal rearrangements, namely translocations, dele- Moreover, to detect small sequences, the fluores- tions, and inversions.1 cence signal can be amplified using a second layer of We will review how metaphase FISH, chromosome antibodies. Chromosomes are counterstained with microdissection, and, more recently, multifish, sig- DAPI or propidium iodide and kept in antifade solu- nificantly improve our knowledge of molecular bases tion after dehydration in an ethanol series (70%-85%- of chromosome changes in malignant haematologi- 100%) and air dried. cal disorders. Microdissection Technical aspects Chromosome microdissection of specific chromo- somal regions is obtained under an inverted micro- Probes scope using glass microneedles controlled by a micro- In situ hybridization techniques are based on the manipulator. The dissected material is further ampli- use of probes complementary to specific DNA fied by PCR, namely DOP-PCR, in which partially regions of the human genome. Centromeric and degenerated oligonucleotides (DOP) are inserted painting probes give rough information for the between the 5’ and 3’ ends of the primer. In a second assignment of the material to specific chromosomes, round of PCR, biotin- or digoxigenin-labelled nucleo- while for identification of specific genes or genomic tides are introduced to obtain FISH probes. Further- segments, probes are used in which known sequences more, FISH with microdissected probes (microFISH) are inserted in vectors such as plasmids, cosmids, is conducted following standard methods. The PAC, BAC, and YAC. Vectors differ for the size of source of microdissected material is confirmed on insert they bear, from 20-30 kb till 1000 kb. Probes original abnormal metaphases, while hybridization are labelled with haptens or fluorochromes and on normal metaphases is necessary to identify the revealed by fluorescence microscopy. Apart from the chromosomes to which the amplified material SKY system, multicolour karyotyping is obtained by belongs. combinations of different chromosomes and adapt- ed filters. Investigations using probes with cDNA are MultiFISH in progress. This is an interesting technological improvement of karyotyping based on the differential and simulta- FISH neous painting of each pair of homologues chromo- Both interphase and metaphase FISH are largely some. It may help to solve major cytogenetic prob- used in haematological malignancies. FISH can be lems, such as detection of subtle translocations or performed with directly (Texas Red, Fluorescein, Lis- identification of complex structural rearrangements samine, Cy3, etc) or indirectly labelled (biotin or dig- in one single experiment. This is of particular interest oxigenin-conjugated) probes. Pretreatment with in malignancies in which cytogenetics and molecular RNAse and pepsin or proteinase K to remove RNA cytogenetic investigations are frequently limited by and protein is optional. Simultaneous denaturation lack of material. There are already a number of pro- of DNA target and probes can be done at high tem- posals from different companies in the market. perature on a hot-plate. Alternatively, they can be denaturated separately in formamide containing Multiple colour FISH uses conventional fluores- buffer. This step is not necessary for single strand cence microscopy and multiple exposures. Another DNA probes. system, the so-called SKY (spectral karyotyping), is based on Fourier spectroscopy and fluorescence is detected after a single exposure. Correspondence: Cristina Mecucci, MD PhD, Ematologia, Policlinico Monteluce; 06123 Perugia, Italy. Tel. international +39-075-5783808 A major limit to applications of this technology are - Fax. international +39-075-5783691 – E-mail: [email protected]. high costs of both equipment and probes. Further- Session #11 – New cytogenetics 99 more, at present, fine changes, such as microdele- eosinophilia. A series of such cases was described as tions, inversions, telomeric translocations, are not being associated with good prognosis. seen by multicolour FISH. In the so-called 17p syndrome in which a myelo- dysplastic syndrome is associated with a chromoso- Metaphase FISH and chromosomal mal change involving 17p13 and typical haemato- deletions logical stigmata, such as dysgranulopoiesis with pseu- Chromosomal deletions may be found in all types do Pelger-Huet anomaly and vacuolated neutrophils, of malignant blood diseases, although they are par- a biallelic involvement of p53 at 17p13 has been ticularly frequent in myelodysplastic syndromes. shown by molecular cytogenetics with loss of the sec- Important new insights into comprehension of the ond allele from the normal appearing chromosome. molecular side of such structural aberrations have In follicular lymphomas characterised by the typical been obtained by FISH investigations in abnormal t(14;18) translocation, FISH with YACs containing metaphases. two suppressor genes, i.e., PTEN/MMAC1 at 10q23.3 First of all FISH may be helpful in identifying dele- and MXI1 at 10q24-25, showed that deletion is a fre- tions hiding translocations. An historical example is quent secondary event and pointed to the existence represented by the t(6;11)(q27;q23) translocation, of a third suppressor gene in 10q involved in follicular one of the variants involving MLL gene at 11q23, lymphomas.8 which can be missed, despite careful cytogenetic Once a deletion has been established at cytogenet- analysis, and which was revealed by painting during ic or molecular level, it is relevant to know whether all a screening of cases classified as 11q- anomaly.2 affected cases lose the same material. Thus, identifi- Even more striking is the discovery of a t(12;21) cation of genomic sequences always being lost (so- (p13;q11) translocation in a subgroup of childhood called commonly deleted region, CDR), has been chosen pre-B ALL with low white cell count and a favourable by many laboratories as the privileged approach to course. This translocation cannot be seen by classical discover the critical suppressor gene(s) underlying cytogenetics because the small material exchanged deletions. Commonly deleted regions have been between chromosomes 12 and 21 has similar mor- restricted by chromosome walking with FISH probes phology at chromosome banding. Only FISH studies in the majority of typical deletions of malignant blood in cases with a 12p deletion revealed the underlying diseases, namely, del(5q) and del(7q) in MDS and translocation. Cloning of corresponding genes, i.e., AML; del(11q) in lymphoproliferative disorders; ETV6 at 12p13 and AML1 at 21q11, provided us with del(12p) and del(13q); in both lymphoid and myeloid specific probes to characterise such typical translo- malignancies; and del(20q) in MDS and myeloprolif- cation further.3 Interestingly, by using cosmids spe- erative disorders. cific for ETV6 gene, Raynaud et al.4 showed that the t(12,21) is accompanied by deletion of the second Metaphase FISH and translocations allele from the normal chromosome 12. Moreover Reciprocal translocations are mainly evidenced by deletion of the second allele only involves a portion of double colour FISH, using painting probes, or, spe- malignant cells, as expected for a secondary event in cific probes for the two partner genes, when known. the leukaemogenic process. Interestingly, in translocations resulting in a fusion More recently an isolated 5q deletion at karyotyp- gene, such as BCR-ABL from the t(9;22) of chronic ic level in childhood leukaemias has been proven to myeloid leukaemia, the fusion emerges as a result of mask a translocation between chromosome 5 and 11, overlapping of the two fluorochromes. This specific i.e., t(5;11)(q35;p15.5).5 reaction is useful not only to increase the specificity of In addition to correction of apparent deletions, a the analysis, but also to understand where a given most important and new piece of information com- fusion happens in reciprocal translocations, as well as ing from FISH studies arises from the demonstration in complex changes. of genomic deletions accompanying other structural FISH on metaphases is one of the most interesting rearrangements. Studies on balanced and unbalanced approaches to narrowing and cloning breakpoints of 12p translocations in both lymphoid and myeloid reciprocal novel translocations identified by conven- malignancies showed that cryptic deletions may hap- tional karyotyping. Breakpoint restriction of recipro- pen either in the 12p material adjacent to breakpoints cal translocations implies chromosome walking pos- or in more centromeric segments independently of sibly with contigs covering the genomic portion of the the breakpoints involved in the translocation.6 Simi- chromosomal sub-bands containing the breakpoints. larly, cryptic 13q deletion may accompany complex The aim is to obtain the splitting of a probe between translocations.7 In some cases the cryptic deletion the two chromosomes involved in a given transloca- may involve very short genomic segments, even only tion, meaning that the insert of the probe contains the part of single genes, such as ETV6 in the 12p. genomic sequences where the breakpoint falls. By this A cryptic deletion of the MRP gene on the 16p arm approach an impressive variation of partners has been may be found in a subgroup of leukaemic cases with identified to exchange material with BCL6 at 3q27 in the typical inv(16)(p13q22) and M4 leukaemia with B cell diffuse lymphomas; with ETV6 at 12p13 in both 100 C. Mecucci et al. lymphoid and myeloid malignancies; and with MLL at rearrange with FGFR3 and MMSET on 4p16.315 or 11q23 in acute myeloid leukaemias. Indeed the num- with the MUM1 gene on 6p25, respectively. Indeed ber of such so-called promiscuous genes is increasing SKY is not able to pick up the t(6;14) translocation due to extensive investigations of chromosomal break- which has been identified by molecular analysis of IgH points with FISH probes for specific genes. gene rearrangements. The 6;14 involves telomeric Metaphase FISH has also been shown to be useful9 genes escaping resolution of painting analysis which to detect minimal residual disease in leukaemia. is the basis of the SKY approach. 14q32 appears as Indeed the authors showed that cell culturing with the hot site of recombination, as expected in a B cell long exposure of cells to colcemid allows recruitment neoplasm. Partners have been recognised on 12q24, of a high number of metaphases that may be 20q11, 16q22, and 22q11. analysed by FISH with specific probes for transloca- Acknowledgements tions, even in poor material unusable for conven- Cristina Mecucci is partially supported by A.I.R.C., Italy. tional karyotyping.

Chromosome microdissection References Chromosome microdissection is an established tool to produce FISH probes for selected chromosomal 1. Falzetti D, La Starza R, Mecucci C. FISH, MicroFISH sites. Further refinements of such technology may also and multicolor FISH to identify rearrangements lead to the cloning of translocation breakpoints in undetected by classic cytogenetics. Rev Clin Exp malignancies. Microdissection is a direct way of iden- Hematol 1997; 4:3-23. tifying the nature of the amplified genomic material 2. Kobayashi H, Espinosa III R, Thirman MJ, et al. Do contained in double minute chromosomes and terminal deletions of 11q23 exist? Identification of homogeneously staining regions.10,11 Cytogenetic clas- undetected translocations with fluorescence in situ sifications of malignant blood diseases contain an hybridization. Genes Chromosom Cancer 1993; 7: 204-8. undefined group in which complex structural 3. Romana SP, Poirel H, Leconiat M, et al. High fre- rearrangements may be not identified with chromo- quency of t(12;21) in childhood B-lineage acute lym- some banding. Microdissection is extremely powerful phoblastic leukemia. Blood 1995; 86:4263-9. in such cases since the isolation of just one or two 4. Raynaud S, Cavé H, Baens M, et al. The 12;21 copies of a marker chromosome, followed by ampli- translocation involving TEL and deletion of the oth- fication and labelling by PCR, may be sufficient to er TEL allele: two frequently associated alterations generate a probe that, by reverse painting on normal found in childhood acute lymphoblastic leukemia. metaphases, may rapidly reveal the chromosomal Blood 1996; 87:2891-9. constitution of markers. Thus, cytogenetically unde- 5. Jaju RJ, Haas OA, Neat M, et al. A new recurrent fined rearrangements may contain typical deletions translocation t(5;11)(q35;p15.5) associated with del(5q) in childhood AML. Blood 1998; 92:311a. or translocations. In human B-cell lymphoma a com- 6. Wlodarska I, Marynen P, La Starza R, Mecucci C, mon deletion including 6q11 was shown after Van den Berghe H. The ETV6, CDKN1B and microdissection of different types of 6q changes.12 D12S178 loci are involved in a segment commonly deleted in various 12p aberrations in different hema- Multicolour karyotyping tological malignancies. Cytogenet Cell Genet 1996; Hidden chromosomal abnormalities were identi- 72:229-35. fied in selected cases of leukaemias and lymphomas 7. La Starza R, Wlodarska I, Aventin A, et al. Molecu- with multicolour karyotyping.13 lar delineation of 13q deletion boundaries in 20 A major step forward in the knowledge of cytoge- patients with myeloid malignancies. Blood 1998; 91:231-7. netics of multiple myeloma came from multicolour 8. Siebert R, Gesk S, Harder S, et al.Deletions in the spectral karyotyping. Cytogenetic studies in multiple long arm of chromosome 10 in lymphomas with myeloma were very difficult because of the lack of suf- t(14,18): A pathogenetic role of the tumor suppres- ficient metaphases from plasma cells. Some improve- sor genes PTEN/MMAC1 and MXI1? Blood 1998; ments were obtained over the last ten years by dedi- 92:4487-9. cated laboratories after the development of cultures 9. El-Rifai W, Ruutu T, Vettenranta K, et al. Follow-up with B-mitogens. Thus 13q- emerged as a recurrent of residual disease using metaphase-FISH in patients structural aberration. A t(11;14), apparently identical with acute lymphoblastic leukemia in remission. to the translocation occurring in mantle cell lym- Leukemia 1997; 11:633-8. phomas, was also observed. Hyperdiploidy emerged 10. Sen S, Sen P, Mulac-Jericevic B, Zhou H, Pirrotta V, Stass SA. Microdissected double-minute DNA as a non-random event. Moreover cryptic transloca- detects variable patterns of chromosomal localiza- tions, undetectable by conventional cytogenetics, tions and multiple abundantly expressed transcripts have already been shown by FISH investigations.14 in normal and leukemic cells. Genomics 1994; New genes have been cloned from the t (4,14) and the 19:542-51. t(6;14) in which immunoglobulin genes on 14q32 11. Guan X-Y, Horsman D, Zhang HE, Parsa NZ, Meltzer Session #11 – New cytogenetics 101

PS, Trent JM. Localization by chromosome microdis- ical malignancies detected by multicolor spectral section of a recurrent breakpoint region on chro- karyotyping. Nat Genet 1997; 15:406-10. mosome 6 in human B-cell lymphoma. Blood 1996; 14. Rao PH, Cigudosa LC, Ning Y, et al. Multicolor spec- 88:1418-22. tral karyotyping identifies new recurring breakpoints 12. Su YA, Trent JM, Guan X-Y, Meltzer PS. Direct isola- and translocations in multiple myeloma. Blood tion of genes encoded within a homogeneously 1998; 92:1743-8. staining region by chromosome microdissection. 15. Avet-Loiseau H, Li JY, Facon T, et al. High incidence of Proc Natl Acad Sci USA 1994; 91:254-68. translocations t(11;14)(q13;q32) and t(4;14) (p16; 13. Veldman T, Vignon C, Schrock E, Rowley JD, Ried T. q32) in patients with plasmacell malignancies. Cancer Hidden chromosome abnormalities in haematolog- Res 1998; 58: 640-5. Educational Session 11 Haematologica 1999; 84:(EHA-4 educational book):102-106 Chairman: A.M. Hagemeijer New cytogenetics

Interphase FISH in chronic lymphoproliferative disorders and comparative genomic hybridisation in the study of lymphomas MARTIN BENTZ, STEPHAN STILGENBAUER, PETER LICHTER,* HARTMUT DÖHNER Medizinische Klinik und Poliklinik V, Universität Heidelberg and *Abteilung "Organisation komplexer Genome", Deutsches Krebsforschungszentrum, Heidelberg, Germany

Interphase FISH in chronic lympho- Sets of specific DNA probes have been developed proliferative disorders for the interphase cytogenetic analysis of the most fre- quent numerical and structural aberrations in B-CLL.2 Methods of cytogenetic analysis in chronic Using such a probe set in more than 250 B-CLL cas- lymphoproliferative disorders es, we found higher incidences of certain aberrations Due to the low proliferative capacity of tumour than those found by banding analysis. In particular, cells, in chronic B-cell lymphocytic leukaemias (B- the percentage of cases without clonal aberrations CLL) conventional chromosome banding analysis is was only 20%. By interphase cytogenetic analysis, the difficult in this disease. Only in 40-50% of cases with most frequent aberrations were 13q14 deletions, this disease can clonal genetic aberrations be identi- which were present in 53% of the cases. 11q deletions fied using this method. In cases without chromoso- were identified in 19% of cases, followed by trisomies mal aberrations, mitotic cells often do not arise from 12 (15%), 6q21-deletions (9%) and 17p13-deletions the malignant cell clone, but rather are derived from (8%) (Table 1). Most likely, the striking differences non-leukaemic T-cells. This was demonstrated using between banding results and interphase cytogenetic a technique of sequential immunophenotyping and data are not due to patient selection but rather reflect karyotype analysis. different sensitivities of the methods used. Using FISH About ten years ago, the development of fluores- analysis, the real incidences of chromosomal aberra- cence in situ hybridisation (FISH) using specific DNA tions are revealed. probes greatly enhanced the potential to detect genet- Characterisation of specific aberrations using ic aberrations in tumour cells. FISH is based on the molecular cytogenetics specific base pairing of a genomic DNA probe to com- Recurring chromosomal aberrations pinpoint to plementary sequences in cells of the specimen. With regions of the genome containing genes of potential this technique, chromosomal aberrations in both pathogenetic relevance. In case of deletions, such metaphase cells and interphase nuclei can be detect- pathogenetically relevant genes are tumour suppres- ed (interphase cytogenetics). Numerical and structural sor genes. Many known tumour suppressor genes, aberrations resulting in a change of the copy number such as p53 and p16, have important roles in the con- in tumour cells are identified by an abnormal signal trol of cell proliferation and cell death. Thus, loss of number per cell. Chromosomal translocations can be function of these genes contributes to tumourigene- detected by an aberrant signal distribution within the sis by an uncontrolled proliferation or a loss of apop- 1 interphase cells. totic capabilities of the cells. In most cases, such a Specific chromosome aberrations in B-CLL loss of function requires the inactivation of both alle- In 1990 and 1991, comprehensive data from chro- les of a tumour suppressor gene. Typically, this is mosomal banding analyses collected within the first achieved by a deletion of one allele, and a mutation and second meeting of the International Working Party of the second allele. Interphase cytogenetics can be on Chromosomes in CLL (IWCCLL) were published. The used for further characterisation of sub-regions con- second report was based on data from 662 patients sistently involved in a specific recurring aberration. In from eleven institutions. In this series, clonal chro- the following, this approach is outlined for the char- mosomal aberrations were identified in 51% of cas- acterisation of the commonly deleted region on chro- es. The most frequent aberrations were trisomy 12 mosome arm 11q in B-CLL. (19% of available cases) and structural aberrations of Using chromosomal banding analysis and com- chromosomes 13 (10%), 14 (8%), 11 (8%), 6 (6%) parative genomic hybridisation in all B-CLL cases with and 17 (4%)(Table 1). an 11q-deletion, the smallest commonly deleted region extended from chromosomal bands 11q21 to 11q25. Based on these data, 17 representative DNA Correspondence: Dr. Martin Bentz, Medizinische Klinik und Poliklinik V, clones were selected from a contig map encompass- Hospitalstr. 3; 69115 Heidelberg, Germany. Tel. international +49- 6221-568001. Fax. international +49-6221-565813 – E-mail: mar- ing chromosomal bands 11q14.3-q23.3. With these [email protected] probes, 43 cases of B-cell neoplasms exhibiting an Session #11 – New cytogenetics 103

11q-deletion (40 B-CLLs and 3 mantle cell lym- ture: this aberration is associated with a younger age phomas) were analysed. Thus, for each case, the size at diagnosis, advanced Rai stages and with extensive and extent of the deletion was characterised. These lymphadenopathy. The negative prognostic effect of experiments resulted in a region of minimal overlap of 11q-deletions is dependent on the age at diagnosis: 2-3 Mbp, which was deleted in all 43 cases. This con- whereas in older patients, there was no significant dif- sensus region contains several genes including the ference in duration of survival, in the age group < 55 ataxia teleangiectasia mutated (ATM) gene. Recently, years at diagnosis, this difference in survival duration a similar 11q22-q33 deletion cluster was delineated in was highly significant (64 months versus 209 months; T-cell prolymphocytic leukaemia.3 In this latter dis- p<0.001) (Figure 1). The clinical implications of chro- ease, missense mutations or small intragenic deletions mosome aberrations in B-CLL are summarised in were identified in the remaining ATM allele of cases Table 2. 3,4 which had an 11q-deletion. This is strong indication Comparative genomic hybridisation in the study for a tumour suppressor function of ATM in T-cell pro- of lymphomas lymphocytic leukaemia. Recently, such biallelic inacti- In comparison to B-CLL and other leukaemias, in vations were also described for some cases of B-CLL. nodal NHL far less is known about the correlation of specific chromosome aberrations with clinical find- Clinical relevance of chromosomal aberrations ings. This is mainly due to the lack of appropriate detected by interphase cytogenetics tumour material: in most cases, lymphoma samples Based on chromosome banding studies, some are only obtained at the time of diagnosis, when a genetic aberrations were associated with shorter sur- is performed. Usually only paraf- vival times in univariate analysis. These included the presence of clonal aberrations (versus normal kary- otypes) and the presence of a trisomy 12. However, an Table 1. Frequency of specific chromosomal aberrations in abnormality of chromosome 17 was the only aberra- B-CLL as detected by banding analysis and interphase cyto- tion of independent prognostic value. genetics. Interphase cytogenetic analysis provides more reli- able information about the real incidence of specific Chromosomal aberration Chromosome Interphase aberrations, independent of the in vitro proliferative banding* cytogenetics# activity of the tumour cells. For three aberrations, an n% n % association with shortened duration of survival has Trisomy 12 112/604 19 46/310 15 been published. The respective studies are outlined Structural 13q aberrations 62/604 10 164/310 53 below. Structural 11q aberrations 49/604 8 59/310 19 Trisomy 12. Altogether at least ten studies reporting Structural 6q aberrations 36/604 6 28/310 9 interphase cytogenetic data for trisomy 12 have been Structural 17p aberrations 22/604 4 25/310 8 published. Only in one of these was an association *Banding data were obtained within the Second International Working Party with the clinical course described. In this series by on chromosomes in CLL (IWCCLL; Juliusson et al., 1991). #Interphase cyto- Escudier and co-workers, there was no significant dif- genetic data are from the Heidelberg interphase cytogenetic study (see ference between survival probabilities in cases with also Döhner et al., 1999). (n=41) and without (n=76) trisomy 12. If, however, banding data were included in the analysis, median survival in patients with trisomy 12 was significantly shorter than in patients with a normal karyotype (7.8 versus 14.4 years). p53-deletions. In an interphase cytogenetic study, monoallelic TP53-deletions were detected in 17 of 100 cases. Patients with this deletion had a significantly higher probability of shorter survival. In addition, the presence of a TP53-deletion was a strong predictor for non-response to purine analogues. In a multivari- ate analysis, TP53-deletion was the strongest prog- nostic factor for poor survival, followed by clinical parameters such as age, Rai stage and haemoglobin level.5 A study using SSCP analysis in 53 patients also showed the strong predictive value of TP53 gene muta- tions. 11q-deletions. Strong clinical implications were also Figure 1. Survival probabilities of 310 B-CLL cases with 6 (n=59) and without (n=251) the presence of an 11q-dele- identified for 11q-deletions. Patients with B-CLL and tion. Survival times are measured from the time of diagno- an 11q-deletion exhibit a characteristic clinical pic- sis. 104 M. Bentz et al.

Table 2. Clinical correlations of specific chromosomal aber- trisomies or DNA amplifications) or underrepresent- rations in B-CLL detected by interphase cytogenetic analy- ed (e.g. monosomies) in the test genome, can be sis. detected by a stronger or weaker staining of the respective target regions in the metaphases. Since sig- Chromosomal Authors, year Clinical characteristics Aberration nal inhomogeneities could also be caused by experi- mental parameters, an internal standard is introduced Trisomy 12 Matutes et al., 1996 atypical cell morphology, by co-hybridisation of normal genomic DNA. Such advanced stages control DNA is obtained ideally from normal cells of Escudier et al., 1993 atypical cell morphology, the patient or alternatively from a proband. Signal shorter survival times inhomogeneities of diagnostic relevance are identified Que et al., 1993 atypical cell morphology by comparison of the differentially visualised signal Criel et al., 1994 atypical cell morphology, advanced stages intensities of the test and control DNAs along the chromosomes. This comparison is performed using 6q deletions Stilgenbauer et al., 1998 no difference in survival digitised image analysis procedures which have been developed for the quantitative evaluation of CGH 11q deletions Döhner et al., 1997 extensive lymphadenopathy experiments. advanced stages shorter treatment-free Chromosomal gains and losses detected in NHL intervals shorter survival times Although by CGH no balanced chromosomal aber- rations (e.g. translocations or inversions) can be 17p aberrations Döhner et al., 1995 shorter survival times, no response to purine detected, this technique has considerable relevance in analogues lymphomas. In a compilation of cytogenetic data in El Rouby et al., 1993* shorter survival times, NHL, a pattern common to this group of tumours was resistance to therapy identified by Johansson and co-workers. The primary aberrations which define the subtype of NHL are bal- *In this study the TP53 gene was analysed for alterations by SSCP. anced aberrations. Examples are the t(14;18) (q32;q21) of follicular lymphomas or the t(11;14) fin-embedded material is stored by the pathologist, (q13;q32) of mantle cell lymphomas. In contrast, and neither fresh tissue nor frozen samples are avail- most of the secondary aberrations present in addition able for further studies. Although interphase cytoge- to the respective translocation seem to be chromoso- mal gains and losses. These additional aberrations netic analysis can be performed on paraffin-embedded may be of particular importance for tumour progres- tissue samples, in most published studies the use of sion and therefore may be associated with a specific this analysis has been restricted to the detection of clinical presentation. Since 1995, CGH studies focus- numerical aberrations. In addition, FISH using specif- ing on different types of NHL have been performed. ic DNA probes depends on the pre-knowledge of can- These include disease entities for which no (primary didate regions that are altered in a specific tumour. mediastinal B-cell lymphoma) or very few cytogenetic Furthermore, only very few chromosomal regions can data (e.g. aggressive gastrointestinal lymphomas, mar- be examined in a single experiment. In contrast, com- ginal zone lymphomas) were available. Even in groups parative genomic hybridisation (CGH) is a molecular- which were analysed in large banding studies, novel cytogenetic technique that allows comprehensive aberrations were identified by CGH (see e.g. Bentz et analysis of the entire genome. Using CGH, a genome al.,8 Table 3). can be screened for the presence of chromosomal gains (e.g. polyomies, duplications or amplifications) Identification of target genes for gains and and deletions of chromosome regions. In addition, losses of DNA the imbalanced regions are mapped within the The types of genes typically associated with chro- genome. Because this technique can be performed on mosomal imbalances are tumour suppressor genes (dele- paraffin-embedded samples, it is particularly attractive tions) and proto-oncogenes (gains). An approach for for the analysis of solid tumour specimens, e.g. epithe- further molecular cytogenetic analysis of regions con- lial tumours and lymphomas.7 taining possible tumour suppressor genes has been outlined above. Due to the limited spatial resolution Comparative genomic hybridisation - principle of CGH, this technique is not suitable for the fine map- of the method ping of deleted regions. However, as a genome screen- CGH is based on a modified in situ hybridisation. ing procedure, CGH has contributed to the identifi- Whole genomic DNA of the tumour of interest (test cation of new deletion regions, such as 11q23 dele- DNA) is hybridised as probe to well defined (normal) tions in CLL and mantle cell lymphomas. metaphase cells (reverse in situ hybridisation, reverse paint- In contrast, CGH has proven to be a very efficient ing). Hybridisation of genomic DNA results in a more method for the characterisation of gains of chromo- or less homogeneous staining of all chromosomes. somal sequences. Such aberrations are supposed to be Chromosomal regions that are overrepresented (e.g. associated with possible activation of proto-onco- Session #11 – New cytogenetics 105

Table 3. Comparative genomic hybridisation in various lymphoma entities.

Lymphoma type Authors, year number of patients Most frequent aberrations gains losses

Chronic B-cell leukaemias Bentz et al., 1995 28 8q, 12 17p, 11q, 6q, 13q Karhu et al., 1997 25 12 11q, 13q

Follicular lymphomas Bentz et al., 1996 28 X, 7, 12, 8, 18 6q Avet-Loiseau et al., 1997 34 18q, X, 7, 2, 6p, 8q 6q, 17p

Primary mediastinal B-cell lymphoma Joos et al., 1996 26 9p, 12q, Xq Diffuse large cell lymphoma Monni et al., 1996 32 X, 1q, 7, 3, 6p, 11. 12, 18 6q, X Marginal zone B-cell lymphoma Dierlamm et al., 1997 25 3, 18, X, 1q 17, 9 Aggressive GI-lymphomas Barth et al., 1998 31 11, 12, 1q, 3q 6q, 17p Mantle Cell lymphoma Monni et al., 1998 27 3q, 8q, 15q 13q, 1p, 6q, 9p, 11q

Studies focusing on amplifications amplified genes

Diffuse large cell lymphoma Houldsworth et al., 1996 1 (CGH-study)* REL (n=26 of 111 cases; Southern Blot) B-cell lymphomas Werner et al., 1997 108 N-MYC, BCL2, CCND2, GLI Diffuse large cell lymphomas Monni et al., 1997 26 BCL2 (31%, Southern Blot) Diffuse large cell lymphomas Rao et al., 1998 20 (CGH-study)* Southern-Blot (n=96): REL, MYC, BCL2, GLI, CDK4, MDM2

* In these studies, CGH was used for for screening a smaller number of cases for amplified chromosomal regions. In a second step, more cases were investigated by Southern Blot analysis using DNA probes for candidate genes mapping to the amplified chromosomal regions.

genes. Due to an increased gene dosage, the level of (DNA chips; Matrix-CGH). For matrix-CGH, whole expression of such genes can be up-regulated resulting genomic tumour DNA and a control DNA are co- in a profound deregulation of cell proliferation and hybridised to an array of DNA probes immobilised on differentiation. In the case of gene amplifications, a glass slide. In the near future this technique may which are known to play a key role in the pathogene- allow a simple and rapid overview of the specific genes sis of certain epithelial tumours, this increase of the and chromosomal loci involved in the pathogenesis gene dosage is particularly evident. CGH is a very sen- of a specific tumour. Furthermore, with further exten- sitive method for screening DNA amplifications. In sion of molecular cytogenetic analysis, it can be addition, the amplified sequences are mapped within expected that, as for acute leukaemias and B-CLL, so the genome by CGH analysis. This has resulted in sev- for nodal NHL, genetic risk factors will be identified eral new findings in NHL. that can be used for the development of risk adapted Before NHLs were analysed by CGH, gene amplifi- treatment strategies in this group of tumours. cations were considered rare events in NHL. Using Acknowledgements banding techniques, the chromosomal hallmarks of This research was supported by grants from the Tumorzen- such gene amplifications, i.e. homogeneously stain- trum Heidelberg/Mannheim (I/I.1), the Deutsche Forschungs- ing regions and double minute chromosomes, were gemeinschaft (Grant Be1454/5-2) and the Deutsche Kreb- reported in only 19 of more than 3,500 NHL cases. By shilfe. CGH, however, DNA amplifications were identified in 10 to 20% of the cases (Werner et al.9). In NHL, ampli- fied target genes were mainly identified based on a References candidate gene approach. Using such an approach, amplification of several genes was performed, alter- 1. Lichter P, Bentz M, Joos S. Detection of chromosomal ations of which were previously only rarely identified aberrations by means of molecular cytogenetics: Paint- (e.g. CCND2) or not at all (GLI and N-MYC) in NHL ing of chromosomes and chromosomal subregions and before. The CGH findings regarding gene amplifica- comparative genomic hybridisation. Methods Enzymol 1995; 254:334-59. tion are summarised in Table 3. 2. Döhner H, Stilgenbauer S, Döhner K, Bentz M, Lichter P. Chromosome aberrations in B-cell chronic lympho- Future prospects cytic leukaemia: reassessment based on molecular cyto- Currently, novel molecular cytogenetic methods are genetic analysis. J Mol Med 1999; 77:266-81. 3. Stilgenbauer S, Schaffner C, Litterst A, et al. Biallelic being developed that will further extend the applica- mutations of the ATM gene in T-prolymphocytic tion of molecular cytogenetics. Recently, most atten- leukaemia. Nat Med 1997; 3:1155-9. tion has been attracted by microarray technologies 4. Vorechovsky I, Luo L, Dyer MJS, et al. Clustering of mis- 106 M. Bentz et al.

sense mutations in the ataxia-teleangiectasia gene in a 8. Bentz M, Huck K, du Manoir S, et al. Comparative sporadic T-cell leukaemia. Nat Genet 1997; 17:96-9. genomic hybridisation in chronic B-cell leukaemias 5. Döhner H, Fischer K, Bentz M, et al. p53 gene deletion reveals a high incidence of chromosomal gains and predicts for poor survival and non-response to therapy with purine analogs in chronic B-cell leukaemias. Blood losses. Blood 1995; 85: 3610-8. 1995; 85:1580-9. 9. Werner CA, Döhner H, Joos S, et al. High-level DNA 6. Döhner H, Stilgenbauer S, James MR, et al. 11q dele- amplifications are common genetic aberrations in B- tions define a new subset of B-cell chronic lymphocyt- cell neoplasms. Am J Pathol 1997; 151:335-42. ic leukemia characterised by extensive nodal involve- 10. Monni O, Joenssuu H, Franssila K, Klefstrom J, Alitalo ment and inferior prognosis. Blood 1997; 89:2516-22. 7. Forozan F, Karhu R, Kononen J, Kallioniemi A, Kallion- K, Knuutila S. BCL2 overexpression associated with iemi OP. Genome screening by comparative genomic chromosomal amplification in diffuse large B-cell lym- hybridization. Trends Genet 1997; 13:405-9. phoma. Blood 1997; 90:1168-74. Educational Session 11 Haematologica 1999; 84:(EHA-4 educational book):107-109 Chairman: A.M. Hagemeijer New cytogenetics

Detection of recurrent translocations using real time PCR; assessment of the technique for diagnosis and detection of minimal residual disease JEAN GABERT Institut Paoli - Calmettes, Marseille, France

arge cytogenetic studies have clearly shown that evaluation; this fusion gene was discovered in 1995 certain chromosomal abnormalities and in par- and represents up to 25% of childhood B-lineage ALL. Lticular some recurrent balanced translocations, Quantitative analyses with competitive PCR on small represent independent prognostic factors for acute series of patients have suggested that it is important leukaemia patients: to determine the level of MRD precisely for assessing - t(9;22) (q34;q11), t(1;19) (q23;p13) and 11q23 the clinical relevance of molecular MRD.4 However, translocations are poor prognostic factors for such competitive techniques are not applicable in acute lymphoblastic leukaemias (ALL); large prospective studies because they are very labo- - for acute myeloid leukaemias (AML), inv(16) (p13; rious, time consuming and too complicated to be q22), t(15;17) (q22;q21) and t(8;21) (q22;q22) standardised. All these qualitative and quantitative identify a good risk group (65% disease free survival PCR techniques are end-point PCR methods. (DFS) at 5 years) while 11q23 abnormalities are cri- A new technique has recently emerged: real time teria for the poor prognostic group (less than 10% quantitative PCR (RQ-PCR) that allows precise quan- DFS at 5 years). tification of the target to be amplified which should revolutionise the field of MRD studies. At the molecular level, these translocations corre- spond to the formation of fusion genes that can be Principles of real time quantitative PCR detected by reverse transcriptase polymerase chain Two methods have been described to detect and reaction (RT-PCR) techniques. This molecular screen- quantify specific PCR products in real time,5 while ing has some advantages: 1) the analyses can be done the PCR reaction proceeds: on a few cells from either blood or bone marrow with- a. during the extension phase, an internal fluorogenic out the need of cell culture so the vast majority of the probe called the Taqman probe, labelled with two patients can be screened and 2) between 10 to 20% dyes: the reporter in 5’ and the quencher in 3’, is of cryptic translocations (patients with a normal degraded by the 5’-3’ exonuclease activity of the karyotype but bearing the molecular abnormality) Taq polymerase, resulting in emission of a fluo- have been identified for each translocation . rescent signal that accumulates during the reac- Besides these diagnostic advantages, PCR analyses tion (Figure 1a).6 This reaction is measured by the allow, because of their high sensitivity, detection of ABI 7 700 or more recently by ABI 5 700 ABI prism residual leukaemia cells bearing these fusion genes machines from Perkin Elmer; during the course of therapy, the so-called minimal b. during the annealing step of the PCR reaction, the residual disease (MRD).1 The clinical value of quali- emission of fluorescence by two hybridisation tative RT-PCR analyses for MRD molecular studies probes is obtained by energy transfer when the two with chromosomal aberrations as targets has been probes are contiguous (Figure 1b). This reaction is clearly established for chronic myeloid leukaemia measured by the Light Cycler System from Roche. after allogeneic transplant and for acute promyelo- cytic leukaemia,2 but their clinical relevance in other In both cases, the fluorescence signal increases with diseases remains unclear. Although a role in MDR each amplification cycle. During the exponential evaluation has been suggested for CBF␤ MYH11 phase of the PCR reaction the signal is proportional [inv(16)],3 contradictory results have been reported to the starting amount of the target present in the for AML1 ETO, the molecular equivalent of t(8;21) tested sample. The value measured is the Cycle for AML and for IgH-Bcl2, the molecular equivalent Threshold, i.e. the number of cycles necessary to of t(14,18) for follicular lymphomas, and finally it is reach a target-specific fluorescent signal 10 standard still too early to assess the role of TEL AML1 for MRD deviations above the base line of fluorescence. When RT-PCR targets are quanitified, the amount of fusion-gene target must be normalised against a Correspondence: Jean Gabert, Laboratoire d’Hématologie, Institut ubiquitous gene, expressed in all haematopoietic Paoli-Calmettes, 232 Bd. Sainte Marguerite, 13009 Marseille, France. Tel. international +33-491-223585 - Fax. international +33-491- cells, in order to assess the quality of the sample 223544 – E-mail: [email protected] analysed. 108 J. Gabert

A. During the extension phase: B. During the annealing phase:

Figure 1. Real time quantitative PCR: principles of detection.

Advantages 4. prices of the machines and reagents are still high, These RQ-PCR methods have a very large dynamic although a cost-effectiveness analysis was per- detection range over five orders of magnitude, there- formed suggesting that they are in the range of by eliminating the need to perform serial dilutions of those of current PCR techniques; details will be pre- follow-up samples. Quantitative data are available in sented during the meeting. a brief period of time since post-PCR processing is not necessary, providing a high through-put tech- Current studies nique. Furthermore, as the assay is performed in an These all show the feasibility of the technique. enclosed tube the risk of cross-contamination, a Studies with this new technology concerning three major problem during PCR MRD studies, is greatly recurrent translocations used as MRD targets have decreased. Finally, standardisation is at last possible been published so far: with this new technology. 1. chronic myeloid leukaemia patients with t(9;22) or Philadelphia chromosome.7 The sensitivity of the Limitations assay was described: at least 10 copies of a plasmid These are mainly due to the novelty of the technique diluted in water were detectable with a dynamic and should be resolved in the near future: range of six orders of magnitude. The BCR ABL 1. current dilution cell lines in limiting dilution assays quantification was normalised by assessment of the or modified artificial constructs in competitive amplification of a ubiquitous gene, the PBGD. assays used to assess the sensitivity of the end point Cytogenetic analysis was negative below a nor- PCR analyses are no longer suitable for RQ-PCR. malised BCR ABL dose of 3x10-2; New standards need to be set and a unit value to 2. acute myeloid patients with t(8 ;21).8 Again using express data has yet to be defined and agreed on; cDNA dilutions in water, the threefold level of sen- 2. for the normalisation with a ubiquitous gene two sitivity of the assay was 5 molecules. Serial samples problems remain: first, the definition of the good from six patients at diagnosis and during the course control gene(s) (GAPDH, ABL, PBGD, etc.) which of therapy, were analysed. Normalisation was done needs to be expressed stably under various condi- with the ␤ actin gene. Each patient showed 103 or tions and to be expressed within the same range as more copies of AML1 ETO fusion transcript at diag- the specific target; second, the target and the con- nosis and each showed a 2 to 4-log decrease dur- trol are currently measured in separate tubes in the ing successful induction therapy; various reported studies, although the technology 3. in follicular lymphoma patients with t(14;18) allows the use of two different dyes and ideally for (q32;q21).9 Here the amplification was done from the specific target and the internal control to be genomic DNA with a sensitivity estimated at 5 pg of measured in the same tube; DNA, equivalent to 0.6 to 0.8 genomes per reac- 3. rules for designing primers and probes and for per- tion. The authors compared this technique with the forming the assay under the best conditions are classical PCR analyses (one round PCR + Southern still to be learned empirically; blotting); the overall concordance was 98%. Session #11 – New cytogenetics 109

Future developments ease during the course of therapy allowing precise An European network has started a standardisation quantification with a high through-put. This technique and quality control programme for RQ PCR analyses is, however, still in its infancy. Retrospective studies of fusion transcripts for leukaemia patients with the will indicate the clinical significance of such molecular support of the European Commission DG V “Europe data which current molecular MRD studies already Against Cancer”. This network involves more than 20 suggest are important. Guidelines need to be estab- expert laboratories from 11 European countries: Aus- lished in order to use this new technique efficiently for tria, Belgium, Denmark, Germany, Great Britain, Italy, the benefit of the largest number of patients in the Netherlands, Portugal, Spain and Sweden. The aims coming years. are to develop a common platform for the main recur- rent fusion genes, to assess the reproducibility between laboratories and to perform a quality control pro- References gramme. This new technology should allow much better stan- 1. Campana D, Pui C-H. Detection of minimal residual disease in acute leukaemia: methodologic advances dardisation than that possible with classical PCR and clinical significance. Blood 1995; 85:1416-34. analyses allowing comparison of results inter-labora- 2. Diverio D, Rossi V, Avvisati G, et al. Early detection of tory and ultimately between therapeutic trials. Studies relapse by prospective reverse transcriptase-poly- on large series of patients with a long follow up are merase chain reaction analysis of the PML/RAR-␣ fusion gene in patients with acute promyelocytic mandatory to assess the usefulness of real time quan- leukemia enrolled in the GIMEMA-AIEOP multicenter titative PCR in a clinical setting. It is then reasonable "AIDA" trial. GIMEMA-AIEOP Multicenter "AIDA" to predict that molecular analyses generated by real Trial. Blood 1998; 92:784-9. time quantitative PCR will be used for monitoring 3. Costello R, Sainty D, Blaise D, et al. Prognosis value patients bearing a marker at diagnosis in large thera- of residual disease monitoring by polymerase chain reaction in patients with CBF beta/MYH11-positive peutic trials. Obviously the interest and applicability of acute myeloblastic leukaemia [letter]. Blood 1997; 89: this new technology will be adapted according to the 2222-3. disease and therapy: 4. Lion T, Henn T, Gaiger A, Kalhs P, Gadner H. Early a. for CML patients, MRD after allogeneic transplant detection of relapse after bone marrow transplanta- in particular, will be definitively assessed using this tion in patients with chronic myelogenous leukaemia. Lancet 1993; 341:275-6. new technology. Furthermore, even the very well 5. Heid CA, Stevens J, Livak KJ, Williams PM. Real time established cytogenetic follow-up of the large quantitative PCR. Genome Res 1996; 6:986-94. majority of patients under interferon therapy may 6. Holland PM, Abramson RD, Watson R, Gelfand DH. be challenged by this new technology; Detection of specific polymerase chain reaction prod- uct by utilizing the 5'----3’ exonuclease activity of Ther- b. for AML patients, the challenge is to identify, with- mus aquaticus DNA polymerase. Proc Natl Acad Sci in the good cytogenetic sub-group, patients who USA 1991; 88:7276-80. will eventually relapse under conventional therapy. 7. Mensink E, van de Locht A, Schattenberg A, et al. The best potential application of this technology Quantitation of minimal residual disease in Philadel- probably lies in this area; phia chromosome positive chronic myeloid leukaemia patients using real-time quantitative RT-PCR. Br J c. for ALL patients, immunoglobulin and T cell recep- Haematol 1998; 102:768-74. tor gene rearrangements have been shown to be 8. Marcucci G, Livak KJ, Bi W, Strout MP, Bloomfield suitable targets for MRD studies in two large ret- CD, Caligiuri MA. Detection of minimal residual dis- rospectives studies.10,11 They have the advantage, ease in patients with AML1/ETO- associated acute compared to fusion genes, of being available for myeloid leukemia using a novel quantitative reverse transcription polymerase chain reaction assay. almost every patient, but much simpler techniques Leukemia 1998; 12:1482-9. need to be developed for prospective studies and 9. Luthra R, McBride JA, Cabanillas F, Sarris A. Novel 5’ clinical application. Real time quantitative PCR is exonuclease-based real-time PCR assay for the detec- one option;12 tion of t(14;18)(q32;q21) in patients with follicular d. finally, this technology appears as a very useful tool lymphoma. Am J Pathol 1998; 153:63-8. 10. Cave H, van der Werff ten Bosch J, Suciu S, et al. Clin- to monitor, quickly and objectively, the efficiency of ical significance of minimal residual disease in child- innovative therapies such as donor lymphocyte hood acute lymphoblastic leukaemia. European infusion or purging of bone marrow or selected Organisation for Research and Treatment of Cancer- peripheral blood stem cells. Childhood Leukemia Cooperative Group [see com- ments]. N Engl J Med 1998; 339:591-8. 11. Van Dongen JJ, Seriu T, Panzer-Grumayer ER, et al. Conclusions Prognostic value of minimal residual disease in acute Currently, at diagnosis, the RQ PCR technique seems lymphoblastic leukaemia in childhood. Lancet 1998; to have a limited application in acute leukaemias or 352:1731-8. lymphomas because of the heterogeneity of some 12. Pongers-Willemse MJ, Verhagen OJ, Tibbe GJ, et al. Real-time quantitative PCR for the detection of mini- fusion genes examined such as MLL AF4 [t(4;11)] or mal residual disease in acute lymphoblastic leukemia CBF␤ MYH11 [inv(16)]. In contrast, it appears to have using junctional region specific TaqMan probes. great potential for monitoring minimal residual dis- Leukemia 1998; 12:2006-14. Educational Session 12 Haematologica 1999; 84:(EHA-4 educational book):110-114 Chairman: V. Vicente Transfusion medicine

Neonatal alloimmune thrombocytopenia MICHAEL F. M URPHY, RACHEL MANLEY, DAVID ROBERTS National Blood Service, and the Department of Haematology, John Radcliffe Hospital, Oxford, UK

eonatal alloimmune thrombocytopenia Pathophysiology of FMAIT (NAIT) is the commonest cause of severe Platelet alloantigens neonatal thrombocytopenia. The condition N There are five well characterised biallelic platelet is the platelet equivalent of haemolytic disease of the alloantigen systems, and a number of rare private newborn (HDN). Foetomaternal incompatibility for alloantigens have also been described;1 most were a foetal platelet alloantigen inherited from the father first discovered during the investigation of cases of and absent in the mother may cause maternal FMAIT. Platelet antigens are known to be expressed alloimmunisation, and foetal and neonatal throm- from 16 weeks’ gestation, and placental transfer of bocytopenia may result from placental transfer of IgG antibodies can occur from 14 weeks, so foetal IgG antibodies. Most cases are diagnosed after birth, thrombocytopenia can occur very early in pregnancy. hence the terminology neonatal alloimmune thrombo- Knowledge of the genetic basis of platelet-specific cytopenia. However, the condition develops in utero, antigens makes it possible to carry out molecular and the foetus may be severely affected, and this is genotyping on whatever material is available, for sometimes emphasised through the use of an alter- example platelet typing using foetal DNA from amni- native term for the condition, such as foetomaternal otic fluid or chorion villous samples. alloimmune thrombocytopenia (FMAIT). Considerable progress has been made in laborato- Platelet alloantibodies implicated in FMAIT ry aspects of platelet immunology since FMAIT was Series of cases of FMAIT with detectable platelet- first recognised, allowing more precise diagnosis of specific alloantibodies have shown that the com- the condition. There have been advances in foetal monest antibody in Caucasian women is anti-HPA-1a in 78-89% of cases, followed by anti-HPA-5b in 6-15% and transfusion medicine resulting in improvements 2 in its management, particularly the antenatal man- and the remainder due to other specificities. In Japan, agement of women with a previous history of preg- anti-HPA-5b and anti-HPA-4b are the most frequent antibodies; the HPA-1a/1b polymorphism is very rare nancies affected by FMAIT. However, current ante- in Japan and anti-HPA-1a antibodies have never been natal management strategies all require foetal blood found. FMAIT due to anti-HPA-5b has been reported sampling (FBS), which is invasive. Improved non- to cause less severe thrombocytopenia than that due invasive methods for identifying high risk pregnan- to anti-HPA-1a, but there are few data on the severi- cies are needed to allow antenatal intervention to be ty of FMAIT due to other HPA antibodies. directed at those foetuses at greatest risk of severe HLA antibodies are frequently found in pregnant thrombocytopenia and haemorrhage. The question women, but are not thought to cause FMAIT. This is of antenatal screening for FMAIT has been raised, because antibodies against foetal HLA antigens are but ideally the relationship between the presence of absorbed by placental HLA antigens; only HLA anti- antibodies and clinical disease should first be more bodies against HLA antigens not expressed by the closely defined so that those pregnancies in alloim- foetus pass into the foetal circulation. munised women which might benefit most from the antenatal intervention can be predicted. A better Incidence of FMAIT understanding of the immune response to platelet The incidence of FMAIT is between 1 in 1,000 to 1 antigens and the mechanisms involved in foetal in 5,000 live births.3-6 This is surprisingly low consid- platelet destruction may allow the design of novel ering that 2% of women are HPA-1a negative and approaches to the prevention and treatment of that 98% of men are HPA-1a positive. However, only FMAIT in the future. about 10% of HPA-1a negative women develop anti- HPA-1a.4-6 The low incidence of alloimmunisation to HPA-1a in HPA-1a negative women is because the antibody Correspondence: Dr.M.F.Murphy, National Blood Service, The John response to HPA-1a is strongly associated with a cer- Radcliffe Hospital, Oxford, UK OX3 9DU. Tel. international +44-1865- 447902 – Fax. international +44-1865-447953 – E-mail. mike.mur- tain HLA class II type. This association was initially [email protected] reported with HLA-DR3, which was present in 71- Session #11 – New cytogenetics 111

platelet alloantibody detection in quality assessment exercises, particularly in antibody quantification. Alter- 300x109/L natively, parameters of the functional activity of HPA antibodies measured by antibody-dependent cell- 30x109/L mediated cytotoxicity or chemiluminescence assays could be used as markers of the severity of FMAIT, as they have been in HDN, but a close correlation with disease severity has not yet been established. Gestational age (weeks) Post-natal age (days) Diagnosis of FMAIT CS Clinical aspects Figure 1. Seventh pregnancy of a patient who had had five The diagnosis of FMAIT is usually made on clinical miscarriages. The last of these was shown to have hydrops grounds following the observation of bleeding, and hydrocephalus and a platelet count of only 17ϫ109/L. although occasionally the diagnosis is suspected fol- The serological findings supported a diagnosis of FMAIT due to anti-HPA-1a. The foetal platelet count was <10ϫ109/L at lowing the finding of a low platelet count in the 25 weeks’ gestation in the sixth pregnancy, and a cord absence of bleeding. The usual presentation is that of haematoma developed during FBS resulting in foetal death. an otherwise well neonate who is observed to have In the seventh pregnancy, prednisolone 20 mg/day and intra- venous immunoglobulin 1 g/week were administered to the widespread purpura at the time of delivery or in the mother from 16 weeks until delivery. The figure shows pre- first 48 hours after birth. The mother is healthy, and and post-transfusion platelet counts following serial FBS and her first child is affected in 50% of cases. Intracranial platelet transfusions (↑). The foetal platelet count was haemorrhage (ICH), which is the major cause of mor- <10ϫ109/L at 26 weeks. The aim was to maintain the foetal platelet count above 30ϫ109/L by raising the immediate tality and long-term morbidity occurs in 15-20% of post-transfusion platelet count to above 300ϫ109/L after cases. In one study of 127 cases, death due to severe each transfusion. The foetal platelet count fell below haemorrhage occurred in 7% and there were neuro- 10ϫ109/L on one occasion when there were problems in preparing the foetal platelet concentrate and the dose of logical sequelae in 21%. Although there is a serious platelets was inadequate. risk of severe haemorrhage at the time of delivery, CS = Caesarean section. nearly 50% of ICHs occur in utero, usually between 30 and 35 weeks of gestation, but sometimes even before 20 weeks. There may be more unusual presentations such as isolated foetal hydrocephalus, unexplained 95% of women developing anti-HPA-1a. It was later foetal anaemia, or recurrent miscarriages. found that all alloimmunised women were positive for The first step in the diagnosis of FMAIT is confir- HLA DRw52, and all responders tested by restriction mation of isolated thrombocytopenia, followed by fragment length polymorphism had the DRw52a allele exclusion of other causes of neonatal thrombocy- at the DRB3 locus. Further studies using polymerase topenia, such as infection, disseminated intravascu- chain reaction with sequence-specific probes showed lar coagulation, maternal autoimmune thrombocy- that two cases previously typed as HLADRw52a topenia, and conditions causing impaired neonatal (HLADR3*0101) were in fact HLA-DRw52c megakaryocytopoiesis. (HLADR3*0301), and one additional case out of sev- Laboratory investigations en new ones was found to be HLADRw53 A diagnosis of suspected FMAIT on clinical grounds (HLADR4*0101).7 This means that FMAIT is not requires laboratory confirmation. Testing is aimed at always associated with HLADRw52a, and that preg- detection of a maternal platelet-specific alloantibody nancies in HPA-1a negative women without this HLA against the father’s platelets, which are a more con- type cannot be excluded from the group at risk of venient source of platelets expressing the relevant anti- FMAIT. There is no similar association between HPA- gen than the baby’s own platelets, and determination 1b alloimmunisation and HLADRw52a, but a signifi- of the platelet antigen types of the mother and father. cant association has been reported between The investigation is best carried out in a reference HLADRw6 and alloimmunisation to HPA-5b. laboratory, which will be able to carry out the neces- The factors influencing the development of foetal sary tests speedily and advise on clinical management. thrombocytopenia and its severity in pregnant women Testing usually comprises the combination of a sen- with HPA antibodies are poorly understood, and there sitive assay using intact platelets, such as the platelet is no reliable or precise correlation with the titre or immunofluorescence test (PIFT), with a glycoprotein- isotype of the IgG antibodies. One study found a sig- specific antigen capture method, such as the mono- nificant correlation between severe thrombocytope- clonal antibody-specific immobilisation of platelet nia and a third trimester titre of anti-HPA-1a ≥ 1:32,6 antigens (MAIPA) assay, which allows detection of but the usefulness of antibody titres as a measure to weak antibodies and mixtures of antibodies. Molec- predict clinical severity is doubtful because there is ular methods, using PCR, are now commonly used considerable variability between laboratories in for platelet genotyping. 112 M. F. Murphy et al.

Table 1. FMAIT due to HPA-la alloimmunisation during pregnancy.

Study Number of pregnant HPA-la-negative HPA-la-positive Maternal Overall incidence women studied foetus (if tested) anti-HPA-la of FMAIT

Mueller–Eckhardt et al. (1985) 1,211 26 (2.2%) 23/25 2/23 (8.7%) 2 (0.16%) Blanchette et al. (1990) 5,000 81 (1.6%) 25/29 3/50 (6%) 1 (0.02%) Burrows and Kelton (1993) 15,471 – – – 10 (0.06%) Doughty et al. (1995) 3,473 74 (2.2%) – 2/22 (9.1%) 3* (0.09%) Williamson et al. (1998) 24,417 618 (2.5%) – 46/387 (12%) 22 (0.09%)

*including one twin pregnancy.

Investigation is more complicated if it is not possi- magnetic resonance scan should be performed to ble to demonstrate parental incompatibility for detect clinically silent ICH. platelet-specific alloantigens with detection of the cor- The transfusion of platelet concentrates from ran- responding maternal antibody. Maternal platelet-spe- dom donors is unlikely to be effective. There has only cific antibodies have been reported to be undetectable been one formal study of the use of intravenous in up to 20% of cases, but this is almost certainly less immunoglobulin (IVIgG) for the neonatal manage- with improved methods for detecting antibodies, and ment of FMAIT; the response rate was 75% and the it may be possible to demonstrate parental incom- increase in platelet count was delayed for 24-48 hours patibility for HPA-1a or another HPA in this situation. during which time the infant remained at risk of ICH. Another cause of difficulty in laboratory investigation Antenatal management is if a rare private antigen is involved; the mother has The recurrence of FMAIT in subsequent pregnancies an antibody against the father’s platelets, but it reacts is very high (>85%); the risk obviously depending on with platelets from very few or no normal donors and whether the partner’s platelet genotype is homozy- its specificity is not obvious. gous (HPA-1a/1a) when the rate of recurrence is Management essentially 100%, or heterozygous (HPA-1a/1b) when the rate of recurrence is 50%. If a previous sibling has Unexpected neonatal thrombocytopenia had an ICH, the risk of antenatal ICH is high in a sub- There is as yet no routine antenatal screening for sequent pregnancy, strongly suggesting that antena- FMAIT, and so most cases will present without warn- tal intervention is indicated. ing. The need for immediate treatment depends on If the previous infant was thrombocytopenic, but the presence of bleeding and the severity of the throm- did not have a major haemorrhage, the risk of ante- bocytopenia. Any delays or difficulties in the labora- natal haemorrhage is more difficult to assess. There tory confirmation of the diagnosis of FMAIT should are no non-invasive tests for the prediction of which not prevent treatment of an infant with bleeding or foetuses are at greatest risk of haemorrhage. Levels of severe thrombocytopenia. A recent survey identified maternal platelet antibodies have not been found to some of the problems in the delivery of prompt treat- be reliably predictive of the severity of thrombocy- ment, including a lack of awareness of the potential topenia, and cerebral ultrasound scans will only seriousness of the condition amongst clinical staff, become abnormal after ICH has already occurred. and in the availability of compatible platelets.8 The only method available at the present time for The best way of raising the platelet count rapidly is assessing the foetal platelet count is foetal blood sam- to transfuse compatible platelets. A strategy for pro- pling (FBS), which allows the diagnosis and severity of viding compatible platelets before the results of sero- FMAIT in utero to be made with certainty.9,10 In addi- logical testing are available is to transfuse platelets tion, this technique provides a means for transfusing from donors who are either HPA-1a negative, or HPA- compatible platelets to severely affected foetuses. The 1a and HPA-5b negative. Some transfusion centres procedure usually only takes a few minutes, and a few ensure that HPA-1a negative, or HPA-1a and HPA-5b- millilitres of pure foetal blood are obtained. Foetal negative platelet concentrates are always available. platelet counts are reliable and it is unusual to have Alternatively, compatible platelets can be provided by problems because of contamination of the sample plateletpheresis of the mother. Often only one trans- with maternal blood or amniotic fluid; the normal fusion of compatible platelets is sufficient, as the range is similar to adults and there is no significant thrombocytopenia is self-limiting. The clinical condi- variation with gestational age. If the father is het- tion of the infant should be carefully monitored and erozygous for the relevant HPA, foetal platelet typing the platelet count measured at least once daily until can be performed using a foetal blood sample it is obvious that it has reached a ’safe’ level and is obtained from 18 weeks’ gestation, or using chorion- increasing spontaneously. If there has been severe ic villous or amniotic fluid samples obtained earlier in thrombocytopenia, a cerebral ultrasound or nuclear pregnancy. Session #11 – New cytogenetics 113

The main risks of FBS are severe haemorrhage from A number of studies have shown the value of platelet the cord puncture site and obstruction of the cord transfusions given by cordocentesis in raising the circulation due to a haematoma. The main problems platelet count, but the platelet count is raised for only are most likely with the initial FBS at around 20-22 a few days. A single pre-delivery transfusion may pro- weeks’ gestation when the cord is small and the tect against bleeding at the time of delivery, but the platelet count may be very low. FBS for FMAIT should foetus remains at risk of spontaneous ICH earlier in be performed at referral centres, where the risk is pregnancy. Weekly in utero platelet transfusions have about 1%. In the authors’ experience of over 200 FBS been shown to be effective in preventing ICH in severe for FMAIT, there have been two foetal deaths, one cases of FMAIT.12 An example of the use of serial foetal due to bleeding and one due to a cord haematoma. platelet transfusions for FMAIT is shown in Figure 1. Severe bleeding requiring foetal red cell transfusions If the foetal platelet count is raised to 300-500ϫ109/L occurred in two other pregnancies, but there was after each transfusion, it is usually no lower than eventually a successful outcome in these pregnancies. 30ϫ109/L one week later; this is thought to be a safe Another group has reported five foetal deaths due to level for the foetal platelet count, and one week to be exsanguination.11 Because of the risk of bleeding, it an acceptable interval between transfusions in high has become routine practice to transfuse platelets to risk pregnancies where serial transfusions are used. the foetus following FBS for suspected FMAIT. More information is needed about the frequency The aim of antenatal management when there has and time of occurrence of ICH to determine the opti- been a previously affected pregnancy is to minimise mal timing of the initial FBS. As already discussed, the risk of severe haemorrhage such as ICH, but the there is no reliable non-invasive method for assessing optimal management remains controversial. The ther- the severity of thrombocytopenia, and because of this apeutic options being explored include maternal uncertainty FBS is usually carried out at 20-22 weeks’ administration of IVIgG and foetal platelet transfu- gestation. If the foetal platelet count is less than sions; early Caesarean section alone is not considered 50ϫ109/L, antenatal treatment is required to min- to be effective in preventing antenatal or perinatal imise spontaneous antenatal ICH because the foetal haemorrhage. For both approaches to antenatal platelet count will continue to fall in untreated foe- management, FBS is required for initial assessment tuses. If it is greater than 50ϫ109/L, and the foetus is of the foetal platelet count, usually at 20-22 weeks’ confirmed to be positive for the relevant HPA, FBS gestation, and for the monitoring of the effectiveness may need to be repeated at regular intervals, possibly of treatment. every 4 weeks, to detect a significant fall in the foetal Maternal administration of IVIgG has been report- platelet count indicating the need for antenatal treat- ed to be successful with no instances of ICH and ment. Antenatal treatment appears to have improved most, but not all infants, having a platelet count of the outcome of severely affected cases of FMAIT, but greater than 30ϫ109/L at the end of pregnancy.9 The there is little information on the long-term develop- addition of steroids does not add to the effect of ment of the children who have been treated in utero. IVIgG. However, ICH has been found during maternal The main issues in the preparation of platelet con- treatment with IVIgG, and other studies have report- centrates for foetal transfusions are compatibility with ed this treatment to be ineffective. A group of Euro- maternal antibodies, avoidance of transfusion-trans- pean centres treating 27 pregnancies found success mitted infection and transfusion-associated graft-ver- with the use of maternal IVIgG in only 4/11 (36%) sus-host disease, and adequate dosage without vol- cases, with steroids in 1/10 (10%), and steroids and ume overload. Concentration of the platelet concen- IVIgG in 1/6 (16%); the treatment was considered to trate to a platelet count of 2,500-4,000ϫ109/L is be successful if there was no ICH, and if there was a essential to achieve satisfactory post-transfusion significant increase in the foetal platelet count and it platelet counts without an unacceptably high trans- was above 30ϫ109/L at the end of therapy.10 fusion volume. It is difficult to understand why there is such a dif- ference in the success of maternal administration of Antenatal screening IVIgG between North America and Europe. Relevant Progress in the antenatal management of FMAIT in factors may include the methods used for assessing women who have had previously affected pregnancies the success of treatment and the dose, timing and draws attention to the issue of how to identify and type of IVIgG used. The selection of cases may also be manage the first affected infant. The diagnosis of important, and there is some evidence that thera- FMAIT in first affected infants is usually only made peutic failures occur more frequently in severe cases. after birth, by which time serious haemorrhage might It has been suggested that the direct injection of IgG have already occurred. Improvements in methods for to the foetus by cordocentesis may be more effective large scale platelet antigen typing and platelet anti- than its indirect administration to the mother. How- body detection have raised the possibility of antena- ever, this method of administering IgG has all the risks tal screening for pregnancies at risk of FMAIT along- of FBS discussed above, and was not found to be side established programmes of screening for effective by another group. haemolytic disease of the newborn. Recent studies 114 M. F. Murphy et al. have demonstrated the feasibility of such programmes tion of immunological and genetic tests. in determining the frequency of HPA-1a alloimmuni- Modulation of the disease process by molecular sation and its clinical sequelae.5,6 Typing for manipulation of the B or T cell and/or effector mech- HLADR3*0101 following initial identification of HPA- anisms may be possible. In principle, the currently 1a negative pregnant women would focus on women available detailed knowledge of the molecular com- most likely to develop anti-HPA-1a, but its positive ponents of the immune response would permit atten- predictive value has been estimated to be only 35%.6 uation of the maternal B cells making HPA-1a anti- Important obstacles to the introduction of antena- body or the maternal T cells that provide help to the tal screening are the lack of reliable non-invasive meth- B cells. Finally, it may be possible to make designer’ ods for identifying which foetuses of pregnant women antibodies that inhibit the binding of harmful mater- with anti-HPA-1a are severely affected by FMAIT, and nal HPA-1a antibodies to foetal platelets. While it is uncertainty about the optimal management of FMAIT true that none of these mechanisms of molecular when there is no previous history of affected pregnan- manipulation is yet established in clinical practice the cies. FBS has significant risks, which have to be bal- definition of restricted mechanisms to generate a dele- anced against those of a conservative policy with no terious HPA-1a antibody response in FMAIT suggest antenatal intervention but prompt identification and that modern molecular treatments may succeed here treatment of affected cases after delivery, and only a before more immunologically heterogeneous diseases large trial will answer this question. are tackled. Future directions The current antenatal management of women References affected by FMAIT, relying on FBS, is invasive and car- ries significant risks to the foetus. These dangers, com- 1. Santoso S, Kiefel V. Human platelet-specific alloanti- gens: update. Vox Sang 1998; 74 (suppl.2):249-53. bined with the inability to predict the severity of clin- 2. Mueller-Eckhardt C, Kiefel V, Grubert A, et al. 348 cas- ical disease have restricted the development of safe es of suspected neonatal alloimmune thrombocyto- and effective strategies to screen all pregnant women penia. Lancet 1989; 1:363-6. for FMAIT and to treat all affected cases. The aim of 3. Mueller-Eckhardt C, Mueller-Eckhardt G, Willen-Ohff H, et al. Immunogenicity of and immune response to current research is, therefore, to develop methods for the human platelet antigen Zwa is strongly associat- the non-invasive diagnosis of FMAIT, reliable predic- ed with HLA-B8 and DR3. Tissue Antigens 1985; 26: tion of disease severity in affected infants and non- 71-6. invasive, effective treatment of FMAIT. 4. Burrows RF, Kelton JG. Foetal thrombocytopenia and Work towards achieving these ambitious goals is its relation to maternal thrombocytopenia. N Engl J Med 1993; 329: 1463-6. helped considerably by the detailed understanding we 5. Doughty HA, Murphy MF, Metcalfe P, Waters AH. now have of the molecular basis of the disease. In the Antenatal screening for foetal alloimmune thrombo- majority of cases of FMAIT, a strong antibody cytopenia: the results of a pilot study. Br J Haematol response to the HPA-1a platelet antigen is only gen- 1995; 90:321-5. 6. Williamson LM, Hackett G, Rennie J, et al. The natur- erated in HPA-1a negative women who carry the al history of fetomaternal alloimmunisation to the HLADR3*0101 allotype. Furthermore, this associa- platelet-specific antigen HPA-1a as determined by tion permits the identification of the fragments of the antenatal screening. Blood 1998; 92:2280-7. HPA-1a antigen that bind to the MHC Class II mole- 7. L’Abbe D, Tremblay L, Goldman M, Decary F, Char- cule to prime helper T cells. trand P. Alloimmunization to platelet antigen HPA- 1a (Zwa): association with HLA-DRw52a is not 100%. This knowledge and other modern molecular meth- Transfus Med 1992; 2:251. ods will be crucial in several areas. First, non-invasive 8. Murphy MF, Verjee S, Greaves M. Inadequacies in the foetal genotyping in the first trimester of pregnancy, postnatal management of fetomaternal alloimmune using the foetal DNA circulating in the maternal cir- thrombocytopenia. Br J Haematol 1999; in press. 9. Bussel JB, Zabusky MR, Berkowitz RL, McFarland JG. culation, has been possible for the Rhesus D blood Foetal alloimmune thrombocytopenia. N Engl J Med group antigen. Similar non-invasive foetal genotyping 1997; 337: 22-6. for platelet antigens would spare foetal blood sam- 10. Kaplan C, Murphy MF, Kroll H, Waters AH. Fetoma- pling of pregnant women with circulating anti HPA- ternal alloimmune thrombocytopenia: antenatal ther- 1a antibodies and a partner who is heterozygous at apy with IVIgG and steroids - more questions than answers. Br J Haematol 1998; 10: 62-5. this locus (HPA-1a/1b). 11. Paidas MJ, Berkowitz RL, Lynch L, et al. Alloimmune Prediction of the severity of disease in affected cas- thrombocytopenia: foetal and neonatal losses related es would be of considerable value by allowing reliable to cordocentesis. Am J Obstet Gynecol 1995; 172: targeting of foetal platelet transfusion. Elucidation of 475-9. 12. Murphy MF, Waters AH, Doughty HA, et al. Antena- the maternal foetal factors, both acquired and innate, tal management of foetal alloimmune thrombocy- that determine disease severity will require co-opera- topenia: report of 15 affected pregnancies. Transfus tive large scale longditudinal studies and the applica- Med 1994; 4:281-92. Educational Session 12 Haematologica 1999; 84:(EHA-4 educational book):115-119 Chairman: V. Vicente Transfusion medicine

Potential and current development of platelet products JOSÉ RIVERA, VICENTE VICENTE Unit of Clinical Oncology and Haematology, University General Hospital, Regional Transfusion Center, Murcia, Spain

latelets, the smallest cellular components of mild agitation to facilitate gas exchange through the blood, are critically involved at each step of the plastic container. However, these storage conditions haemostatic response, from the initial sealing are by no means optimal, and transfusion of conven- P 1 of damaged endothelium to supporting coagulation tional 22ºC liquid stored PCs has several drawbacks. reactions and finally, retraction of the fibrin clot First, the transmission of an infectious agent is a which enhances fibrinolysis and wound healing. major concern for the transfusion of any blood prod- Consequently, if the platelet concentration is uct, and only recently have approaches been devel- decreased and/or the platelet function is abnormal, oped to attempt inactivation of infectious pathogens the risk of haemorrhage is increased. – viruses, bacteria, and protozoa – in PCs. Particu- Ever since Duke’s report in 1910 of the first use of larly, septic reactions associated with transfusion of platelet-containing fresh whole blood to treat three bacterial contaminated PCs are a recognised and still bleeding thrombocytopenic patients, platelet trans- unsolved problem of relevant dimensions (approxi- fusions have gained value in medicine and nowadays mately 1 in 1,500 transfusions), with potentially fatal they are essential for the management of patients consequences for recipients. This risk is higher for with primary thrombocytopenia, or for support of PCs than for other blood components stored under those treated with to intensive chemo/radiothera- refrigeration, such as red cells, since storage at room peutic regimens associated with prolonged periods of temperature facilitates the propagation of a poten- bone marrow aplasia. Unfortunately, the supply of tially present bacterial load. In fact, the increased risk platelets for clinical use is hampered by an increasing of bacterial growth associated with the 22ºC storage demand and our limited capacity to preserve this temperature is the primary reason for the current 5- blood component under in vitro conditions. Thus, day shelf-life of PCs. platelet shortage occurs frequently and, consequent- In addition, alloimmunisation against histocom- ly, the development of platelet products with longer patibility antigens occurs in many patients receiving shelf-life and/or of platelet substitutes is a major goal multiple transfusions of random donor platelets. in modern transfusion medicine. Alloimmunised patients who, in recent experience, This overview will briefly discuss the approaches may reach as many as 25-35% of newly diagnosed that have been taken over the past few years to devel- subjects with acute myeloid leukaemia, become op alternative products to the current 22ºC liquid- refractory and can be extremely difficult to treat with stored platelet concentrates (PCs). platelet transfusions. Histocompatible donors are often not available, and as many as 50% of appar- Drawbacks of conventional 22°C liquid ently histocompatible platelet transfusions adminis- stored PCs tered to alloimmunised patients do not achieve good Conventional liquid PCs consisting of platelets with post-transfusion recovery. Thus, the cost and diffi- a distinct number of contaminating white cells, re- culty of platelet transfusion therapy in these alloim- suspended in autologous plasma, are routinely pre- munised patients are high, and often they remain at pared by fractionation of whole blood units by either risk of haemorrhagic morbidity and mortality. Fortu- the platelet-rich-plasma (PRP) or the buffy coat meth- nately, there is now substantial evidence suggesting ods, or directly harvested from blood circulation by that the leukocytes contaminating platelet prepara- apheresis. Although each procedure has advantages tions are the primary stimulus for alloimmunisation. and disadvantages, there is a tendency in Western Thus, the use of leukodepleted PCs is becoming a rou- countries to increasing use of single donor PCs tine in transfusion medicine that could drastically obtained by apheresis. Currently, all types of liquid reduce the incidence of alloimmunisation in the near PCs are stored in highly permeable plastic bags at a future. Additional benefits of leukodepletion of PCs controlled temperature of 22±2ºC, and with constant would be the lower incidence and severity of febrile non-haemolytic transfusion reactions mediated by bioreactive substances released by passenger leuko- Correspondence: Prof. Dr. Vicente Vicente García, Centro Regional de cytes during storage, and the decreased risks of Hemodonación, C/Ronda de Garay s/n, 30003-Murcia, Spain. Tel. inter- national +34-968-341990 – Fax. international +34-968-261914 – E- immunomodulatory effects and of graft-versus-host mail: [email protected] disease. 116 J. Rivera et al.

Finally, it is well established that during storage of clinical trials in volunteers and/or in patients. In addi- PCs at 22ºC serial deleterious changes in the platelet tion, this task would be hampered by 1) the lack of properties occur, collectively referred to as the platelet consensus regarding the in vitro tests that best reflect storage lesion.2 Thus we, and other groups, have the viability and function of platelets; 2) the contro- demonstrated that storage promotes platelet shape versy about what are valid species for pre-clinical stud- change, surface expression of activation proteins, loss ies regarding generalisation of the data obtained to of GP Ib, and impaired response to agonists. Whether humans, and about the correct way to induce the these in vitro observed changes are reversible upon animals’ thrombocytopenia; and 3) the lack of vali- transfusion or significantly affect the in vivo viability dated end points or surrogate measurements for and function of platelets remains unclear, but the demonstration of clinical benefit to the patient pop- awareness of the platelet storage lesion is an addi- ulation included in trials. tional reason for the current 5-day restriction of con- ventional 22ºC liquid-stored PCs. It seems clear that Snapshot of advances in platelet the short shelf-life of PCs under current storage meth- products and substitutes ods results in complex inventory management, loss of The research performed in this field over the past units due to outdating, difficulty in providing platelets few years can be classified in five general areas. far from the production place, and frequent platelet shortages. 1. Storage of liquid PCs under refrigeration For all the above reasons, much effort has been Early attempts to store platelets were performed, as made in academic and commercial settings to for whole blood or red cells, at 4ºC. However, refrig- improve the current platelet storage methods, and to erated storage was soon abandoned due to the find- develop novel platelet substitutes. ing that cold temperatures significantly and irreversibly affect platelets. Advances in understanding the cold- The challenge of designing and evaluating induced platelet responses, a process termed cold acti- novel platelet products vation, have shown similarities with physiological ago- As mentioned above, platelets have many complex nist-induced platelet activation (shape change, calci- functions (adhesion and spreading on injured vessel, um mobilisation, actin filament assembly, aggrega- support for coagulant activity, modulation of fibri- tion, etc.). This knowledge, and awareness of the risk nolysis, and others). Therefore, an important ques- of bacteraemia associated with transfusion of room tion in the quest for platelet substitutes is, what func- temperature-stored PCs, have renewed the interest for tion(s) of platelets is needed to be mimicked by the development of platelet cold storage.3-5 platelet product or substitute? For example, products Several physical methods (increased atmospheric retaining the platelet procoagulant activity may be pressures, temperature cycling), and biochemical sufficient to reduce the risk of venous haemorrhage. strategies (cytoskeletal stabilisers, antifreeze glyco- By contrast, maintenance of haemostasis in the arte- proteins, signal transduction inhibitors), for preven- rial circulation, where a high shear stress exists, might tion of cold-induced platelet activation have been require most, if not all, of the platelet functions. Thus, explored. The microtubule stabiliser agent taxol was we may speculate that in the future it would be pos- used in early attempts to prevent platelet changes sible to tailor and produce specific platelet products induced by chilling, but discoid shape and physio- for the management of different clinical situations logical responses were not fully preserved in treated (dilutional thrombocytopenia, immune thrombocy- platelets. More recently, platelets cooled to 4ºC in topenic purpura, chemotherapy-induced thrombocy- the presence of the drug cytochalasin, an inhibitor of topenia, or a dysfunctional platelet disease). actin filament assembly, in combination with a cyto- In the short term, however, we can only hope that plasmic calcium chelator (Quin 2) have been shown alternative platelet products have a few desirable to remain discoid and responsive to glass and throm- properties. Some of these properties relate to practi- bin activation. Whether this treatment preserves oth- cal aspects such as, need of no specialised storage er platelet properties is not known. conditions, prolonged shelf-life compared to that of Another interesting approach to the cold storage current PCs, and minimal manipulation before trans- of platelets is the use of antifreeze glycoproteins iso- fusion. Others relate to safety and clinical effects on lated from polar fish. It has been shown that these recipients, such as being sterile or suitable for sterili- proteins prevent, in a dose-dependent manner, the sation, haemostatically effective during long intervals, cold-induced platelet shape change and surface and having no major thrombogenic or immunogenic expression of activation proteins (LAMP, CD63), and effects. Appropriate demonstration that proposed preserve the thrombin responsiveness of cold-stored platelet products accomplish these requirements of platelets. The underlying mechanism of these actions safety and efficacy would be a difficult task, because is unclear, but some evidence suggests that antifreeze it would need: 1) the performance of in vitro assays glycoproteins could protect membrane phospholipids valuable for testing platelet function; 2) the carrying from a phase transition change during platelet chill- out of pre-clinical studies in animal models; and 3) ing, thus preventing damage to the membrane. Session #11 – Transfusion medicine 117

We, and others, have recently evaluated the storage have observed that platelets frozen with this solution at 4ºC of PCs supplemented with a combination of display in vitro properties similar to those cells cryo- signal transduction inhibitors.6,7 This cocktail, named preserved with the standard 6% concentration of ThromboSol, is made up of agents that enhance the DMSO. Moreover, platelets cryopreserved in this man- platelet cAMP/cGMP concentrations (adenosine, ner retain their haemostatic function in rabbits, and it sodium nitroprusside and dipyridamole), amiloride, has recently been shown that their recovery and sur- ticlopidine, and quinacrine. In these studies, Throm- vival when re-infused to autologous human volunteers boSol protected platelets from morphological changes is superior to that of 6% frozen platelets, and compa- and spontaneous aggregation during cold storage. rable to that of platelets stored at 22ºC for five days.8 Moreover, treated platelets stored-refrigerated for 9 3.Preservation of platelets in a freeze-dried days displayed responses to agonists (ADP, collagen, state thrombin) comparable to those of platelets stored at Freeze-dried platelets are seen as a potentially 22ºC for 5 days. The protective effect of this additive durable platelet product, lightweight, and easier to solution was related to the ability of its components transport and store than liquid or cryopreserved PCs. to sustain high levels of cAMP and to inhibit TxA2 pro- Early attempts in the 50’s to use lyophilised platelets duction during the entire storage period at 4ºC. An in animal models and in humans were unsuccessful, additional benefit of cold storage of PCs treated with and this technology was abandoned for many years. ThromboSol was impaired production of cytokines However, the strategy has emerged with great vigour (IL6, IL8) compared to that found in PCs convention- thanks to the work of Read et al.9 These authors devel- ally stored at 22ºC. oped a lyophilisation procedure in which platelets are 2. Cryopreservation of platelets first stabilised with 1.8% paraformaldehyde for 1 hour, As for many cell types, freezing has long been con- washed, re-suspended in 5% albumin, and then freeze- sidered an alternative for platelet preservation, with dried for 24 hours at –20 to –40°C. When re-hydrat- the major advantage of extending the storage period ed, the platelets in the lyophilised product appear to from a few days to years. Unfortunately, current meth- have normal morphology, surface expression of recep- ods for platelet cryopreservation require expertise, are tors, adhere appropriately to thrombogenic surfaces, labour-intensive, and involve the use of cryoprotec- and have procoagulant activity. Studies in vivo in rats, tant agents, potentially harmful for recipients, and rabbits, and dogs have shown that lyophilised thus requiring wash-out before infusion. Thus, frozen platelets correct the bleeding time. In addition, the storage of platelets is not routinely considered and is platelet treatment with 1.8% paraformaldehyde is viru- scarcely used. cidally and bactericidally effective, providing a 5-7 log 4,5 Although it has been shown that cryopreserved reduction in infectious material (including HIV). platelets display several metabolic and functional Other recent works have also shown the presence of changes, they are claimed to exert haemostatic prop- GP receptors Ib and IIIa in lyophilised platelets, and erties when infused in vivo. In fact, several studies have have demonstrated that these freeze-dried cells retain proven the usefulness of frozen platelets in the pro- their ability to interact with exposed subendothelium phylaxis of bleeding in different clinical settings (car- in a perfusion model. The safety and efficacy of these diopulmonary bypass, onco-haematological patients lyophilised platelet products in humans remain to be undergoing high-dose chemotherapy and/or haema- elucidated in clinical trials. topoietic progenitor cell transplantation).4,5 4. Use of platelet fragments or microparticles So far, the most widely used cryoprotectant agent Activation of platelets is known to induce shedding for platelet freezing is dimethyl sulphoxide (DMSO) of platelet membranes or microvesicles with procoag- (at 5 to 10%). Other substances sporadically being ulant properties that support the haemostatic func- employed are glycerol based solutions, hydroxyethyl- tion of intact platelets. This microparticle formation starch, trephalose, or propane-1,2-diol. Since DMSO also occurs during conventional storage of PCs, and has well recognised adverse effects, a reduction of its increases with platelet chilling. Recently, human concentration in platelet freezing or substitution by infusible microvesicles have been developed and stud- less toxic cryoprotectant regimens is desirable to allow ied.4,5 The manufacturing process involves repeated direct infusion of frozen-thawed platelets. In this freezing/thawing of platelets, high speed centrifuga- regard, PCs frozen with a solution containing 20% tion to isolate particulate material, wet heat viral inac- polyvinyl-pyrrolidone, 10% mannitol, 5% glycerol, and tivation, and lyophilisation. The resulting product has a mixture of salts, appear to function well when a phospholipid content similar to that of platelets, infused to thrombocytopenic rabbits without post- and retains varying quantities of platelet membrane thaw washing of the cryoprotectants. Also, a reduced receptors and procoagulant activity. It appears to have DMSO concentration (2%) combined with a modified reduced class I HLA expression, and has a shelf life of ThromboSol mixture (amiloride, sodium nitroprusside three years. These platelet microparticles were first and adenosine) is being tested as an alternative cryo- found to shorten the bleeding time with no evidence preserving solution. Other authors and our group, of toxicity in a rabbit model of busulfan-induced 118 J. Rivera et al. thrombocytopenia. In phase I studies, human volun- uct, namely Plateletsome, has been suggested. This teers tolerated the treatment well with platelet mem- product consists of a deoxycholate extract of platelet branes, which were not associated with adverse membranes, including major glycoprotein receptors, changes in biochemical or coagulation indices. In incorporated into unilamellar lipid vesicles. Platelet- addition, no signs of immunogenicity (antibodies somes have no in vitro effect on platelet aggregation, reactive to normal platelets or lymphocytotoxic HLA but decrease the tail bleeding time in thrombocy- antibodies) were observed in fifteen normal subjects topenic rats by 67%. No evidence of intravascular injected twice with microparticles for up to two coagulation is observed upon intravenous infusion of months after the first exposure. There is also a phase Plateletsomes to rabbits, and no pathologic thrombi II clinical study of microparticles in thrombocytopenic were detected on post-mortem examination of treated patients with non-life threatening active mucosal rats. Procoagulant liposomes have also been explored. bleeding. Sixty-five per cent of patients infused with In one study, infusion of a combination of phos- microparticles achieved improvement or cessation of phatidylcholine: phosphatidylserine (80:20) (PC/PS) bleeding. More importantly, 58% of patients refracto- vesicles and factor Xa normalised the bleeding time in ry to normal platelets did respond to microparticle haemophiliac dogs. However, this preparation was treatment. No patients experienced serious adverse toxic in dogs and baboons, inducing a decrease in fac- events attributable to microparticle infusion. These tor V and VIII as well as in the platelet count and are promising data, but routine bleeding prophylaxis haematocrit.4,5 Recently, Galán et al.12 described lipo- with platelet microparticles awaits definitive confir- somal preparations that under in vitro conditions of mation of safety and efficacy in prospective ran- flow, fulfil the procoagulant function of platelets. By domised controlled trials. using the Baumgartner perfusion system, these 5.Synthetic or semi-synthetic platelet authors demonstrated that addition of vesicles made substitutes of PC, phosphatidylinositol (PI), phosphatidylethanol- amine (PE)/PC (1:1), and (PS/PC) (3:1) to platelet There have been several investigative approaches to and white cell depleted blood, significantly increased preparing synthetic or semi-synthetic products that fibrin formation on exposed subendothelium, and could sustain some of the properties of functional increased post-perfusion levels of F1+2, suggesting platelets.4,5 thrombin generation. These in vitro results further sup- Agam and Livne10 explored the haemostatic prop- port the value of phospholipid preparations as poten- erties of red cells coated with fibrinogen. They found tial platelet substitutes for treatment of thrombocy- that these coated erythrocytes enhance platelet aggre- topenic patients. However, issues such as toxicity, gation induced by agonists, and shorten by 4-fold the optimal lipid composition, concentration, size, and tail bleeding times in thrombocytopenic rats. Dr. Bar- stability of these procoagulant liposomes need to be ry Coller and co-workers11 have developed and evalu- determined. ated thromboerythrocytes, which are red cells that are covalently coupled with RGD-containing peptides. Concluding remarks This Arg-Gly-Asp (RGD) sequence is present in The technologies to produce many kinds of platelet fibrinogen and others ligands that bind to the GP products are in place, and should be encouraged in IIb/IIIa receptor. Thromboerythrocytes bind to order to avoid shortages of this blood component, platelets adhered to collagen under static or low shear and the problems associated with transfusion of con- conditions. They also co-aggregate with platelets stim- ventional PCs. The major goal now is to develop ulated with ADP. However, discrepant results have appropriate ways to assess the safety and efficacy of been found concerning the capacity of thromboery- these products before they are used in the clinical set- throcytes to reduce prolonged bleeding time in ani- ting. mal models. More artificial products are thrombospheres and liposome-based agents. The former are spheres of References cross-linked human albumin with fibrinogen covalent- ly bound to the surface. They are small (1.2 µm), thus 1. Norfolk DR, Ancliffe PJ, Contreras M, et al. Consen- circulate freely through small vessels with appropriate sus conference of platelet transfusion, Royal College haemorrheologic characteristics. These microspheres of Physicians in Edinburgh, 27-28 November 1997. do not spontaneously clump, but co-aggregate with Synopsis of background papers. Br J Haematol 1997; 10: 609-17. platelets in the presence of a platelet agonist. When 2. Seghatchian J, Krailadsiri P. The platelet storage infused as a single bolus to thrombocytopenic rabbits, lesion. Transfus Med Rev 1997; 11:130-44. thrombospheres shorten ear bleeding time and reduce 3. Vostal JG, Mondoro TH. Liquid cold storage of 51Cr-blood loss, while they do not seem to have appar- platelets: A revitalised possible alternative for limiting bacterial contamination of platelet products. Transfus ent thrombogenicity. Their safety and efficacy in Med Rev 1997; 11:286-95. humans have not yet been investigated. 4. Alving BM, Reid TJ, Fratantoni JC, Finlayson JS. Frozen The haemostatic efficacy of a liposome-like prod- platelets and platelet substitutes in transfusion med- Session #11 – Transfusion medicine 119

icine. Transfusion 1997; 37: 866-76. hemostatic and structural properties of rehydrated 5. Lee DH, Blajchman MA. Novel platelet products and lyophilised platelets. Potential for long-term storage of substitutes. Transfus Med Rev 1998; 12: 175-87. dried platelets for transfusion. Proc Natl Acad Sci USA 6. Connor J, Currie LM, Allan H, Livesey SA. Recovery of 1995; 92: 397-401. in vitro functional activity of platelet concentrates 10. Agam G, Livne AA. Erythrocytes with covalently bound stored at 4ºC and treated with second-messenger fibrinogen as cellular replacement for the treatment of effectors. Transfusion 1996; 36:691-8. thrombocytopenia. Eur J Clin Invest 1992; 22:105-12. 7. Rivera J, Lozano ML, Corral J, et al. Quality assess- 11. Coller BS, Springer KT, Beer JH, et al. Thromboery- ment of platelet concentrates supplemented with sec- ond-messenger effectors. Transfusion 1999; 39:135- throcytes: In vitro studies of a potential autologous, 43. semi-artificial alternative to platelet transfusions. J Clin 8. Currie LM, Lichtiger B, Livesey SA, Connor J. In vivo cir- Invest 1992; 898:546-55. culatory parameters of human platelets cryopreserved 12. Galán AM, Hernández MR, Bozzo J, et al. Prepara- with ThromboSol [abstract]. Blood 1998; 92 (suppl tions of synthetic phospholipids promote procoagu- 1): 546a. lant activity on damaged vessels: studies under flow 9. Read MS, Reddick RL, Bode AP, et al. Preservation of conditions. Transfusion 1998; 38:1004-10. Educational Session 12 Haematologica 1999; 84:(EHA-4 educational book):120-123 Chairman: V. Vicente Transfusion medicine

Current status of non-transfusional haemostatic agents MARCO CATTANEO, PIER MANNUCCIO MANNUCCI Angelo Bianchi Bonomi Haemophilia and Thrombosis Centre, Department of Internal Medicine, IRCCS Ospedale Maggiore and University of Milan, Italy

on-transfusional haemostatic agents are indi- Urinary tract bleeding cated in the treatment of bleeding resulting After prostatectomy, the urine, which is rich in plas- Nfrom multiple or unknown defects of minogen activators, dissolves clots in the prostatic haemostasis. These drugs may also be indicated in cavity, resulting in haematuria and sometimes patients who refuse blood transfusion or in those anaemia. In clinical trials EACA or tranexamic acid who undergo surgical procedures or suffer from reduced blood loss by approximately 50 percent, as mucosal lesions associated with large blood losses compared with placebo. The drugs are contraindi- necessitating many transfusions of donor blood. cated in patients with upper urinary tract bleeding Non-transfusional haemostatic drugs of proven clin- because of the risk of retention of clots in the ical efficacy include antifibrinolytic amino acids and the bladder. Recommended dosage is: tranex- (epsilon-aminocaproic acid and tranexamic acid), amic acid, 10 to 15 mg per kilogram every 8 hours aprotinin, desmopressin and conjugated oestrogens. intravenously, starting immediately after surgery, fol- lowed by 20 mg per kilogram orally every 8 hours dai- Antifibrinolytic amino acids ly until macroscopic haematuria stops; EACA, 50 to Two synthetic derivatives of the amino acid lysine, 60 mg per kilogram intravenously 6 times daily fol- epsilon-aminocaproic acid (EACA) and tranexamic lowed by oral administration of the same dose. acid, bind reversibly to plasminogen thereby blocking its binding to fibrin and its activation to plasmin. Oral bleeding in congenital and acquired EACA and tranexamic acid (which is about 10 times coagulation disorders more potent and has a longer half-life) are haemo- Antifibrinolytic drugs are useful adjuvants for the statically effective even when bleeding is not associ- control of bleeding after dental extractions in patients ated with laboratory signs of hyperfibrinolysis. Since with haemophilia, reducing the amount of clotting- both drugs enter the extravascular space and accu- factor replacement therapy needed. Mouthwashes mulate in tissues, the basis for their efficacy is thought containing tranexamic acid (1 g every 6 hours) are to be inhibition of tissue fibrinolysis and consequent effective for prevention of oral bleeding in haemophil- clot stabilisation. iacs and in patients who need dental extractions while receiving long-term oral anticoagulant thera- Primary menorrhagia py. Extractions can be performed without stopping or Tranexamic acid reduces menstrual bleeding by 40 reducing the anticoagulant, which may increase risk to 50 percent. It is recommended only when organic of thrombosis in patients with atrial fibrillation or lesions in the have been excluded. artificial heart valves. Recommended oral doses: in Recommended oral doses are 10 to 15 mg per kilo- adults are of 50 to 60 mg EACA per kilogram every 4 gram of body weight every 8 hours, from the onset hours or 20 to 25 mg tranexamic acid per kilogram until the arrest of menstrual bleeding. every 8 hours until healing of the sockets is complete. Gastrointestinal bleeding Bleeding in patients with thrombocytopenia A meta-analysis, based on results from 1,267 EACA may be efficacious to stop mucosal (nasal, patients with peptic ulcers, mucosal erosions or oth- uterine, gastrointestinal) bleeding and bleeding asso- er causes of bleeding demonstrated that use of ciated with dental extractions in patients with throm- antifibrinolytic drugs is associated with a 20 to 30 bocytopenia. percent reduction of re-bleeding, 30 to 40 percent reduction in the need for surgery, and 40 percent Bleeding after thrombolytic treatment reduction in mortality. Despite these results tranex- Antifibrinolytic drugs are of potential efficacy in amic acid is not widely used to treat patients with controlling bleeding complicating thrombolytic ther- upper digestive tract bleeding because of the effica- apy. There is, however, little evidence that they are cy of other medical and endoscopic treatments. useful when bleeding complications develop during or shortly after thrombolytic treatment. Surgical blood loss Correspondence: M. Cattaneo, MD, via Pace 9, 20122 Milan, Italy. Phone: international +39-02-5503 5426 – Fax: international +02- Cardiac surgery is the prototype operation war- 5516093 – e-mail: [email protected] ranting adoption of blood saving measures. The Session #11 – Transfusion medicine 121 results of clinical trials involving at least 1,000 patients Aprotinin treated with tranexamic acid or EACA consistently Aprotinin, a polypeptide with a molecular weight of demonstrated that either drug reduced blood loss by 6,512 daltons, inhibits several serine proteases 30 to 40 percent, as compared with placebo. Howev- (trypsin, chymotrypsin, plasmin and tissue and plas- er, the amount of blood products given was either not ma kallikrein) through the formation of reversible measured or not reduced by antifibrinolytic drugs, and enzyme-inhibitor complexes. By inhibiting kallikrein, none of the studies was of sufficient size to determine aprotinin indirectly inhibits the formation of activat- whether the incidence of serious side-effects was ed factor XII. Hence, aprotinin inhibits the initiation increased by the treatment. A recent meta-analysis, of fibrinolysis induced by the contact of blood with a however, showed that EACA is as effective as apro- foreign surface. The enzymatic activity of the com- tinin in reducing blood loss and transfusion require- pound is expressed in kallikrein inactivator units ments and had no effects on risks of post-operative (KIU). Plasma concentrations of 125 KIU per milli- myocardial infarction or overall mortality. liter are necessary to inhibit plasmin and concentra- Recommended dosage is: a bolus intravenous dose tions of 300 to 500 KIU per milliliter are needed to of 150 mg EACA per kilogram given before the oper- inhibit kallikrein. ation, followed by an infusion of 15 mg per kilogram Surgical blood loss per hour during the operation; 10 mg tranexamic acid The broad antiproteolytic action of aprotinin per kilogram intravenously before the operation, fol- prompted its use to reduce blood loss in patients lowed by 1 mg per kilogram per hour during the oper- undergoing cardiac surgery, during which there is ation. increased plasma proteolysis. Several double-blind Although there are demonstrations that antifibri- studies demonstrated that aprotinin is effective in nolytic drugs reduce blood loss and transfusion reducing blood loss and transfusion requirements in requirements in patients undergoing knee replace- patients at high bleeding risk, such as those under- ment surgery, their use should be considered only for going repeat surgery or cardiac transplantation, tak- patients in whom large blood losses are predicted, ing acetylsalicylic acid, or suffering from endocarditis, such as those undergoing double joint replacements and also in those undergoing more common opera- and re-operations. tions at lower bleeding risk, such as valve replacement Patients undergoing orthotopic liver transplanta- and coronary artery bypass grafting. tion lose large amounts of blood, due in part to pre- Aprotinin is effective only when infused intra- existing coagulopathy and intraoperative fibrinolysis. venously. The first clinical studies of aprotinin in car- In a clinical trial of 45 patients given high-dose tranex- diac surgery used a loading dose of 2 million KIU fol- amic acid (20 to 30 mg per kilogram) or placebo dur- lowed by a continuous infusion of 500,000 KIU per ing surgery, treated patients had about 50 percent hour during surgery, with 2 million KIU added to the less intraoperative blood loss and lesser transfusion priming solution. Lower-dose regimens have been sub- requirements. These preliminary results need confir- sequently proposed, which, with few exceptions, are as mation. effective haemostatically as the full-dose aprotinin reg- Side effects and contraindications imen. Aprotinin is less effective when given post-oper- The drugs are contraindicated in patients with sub- atively than when given prophylactically. arachnoid bleeding, because they induce vasospasm Results of non-randomised, uncontrolled trials sug- and ischaemic stroke. The side effects are dose-depen- gest that aprotinin may be effective in reducing blood dent and usually involve the loss and transfusion requirements in patients under- (nausea, vomiting, abdominal pain and diarrhoea). going orthotopic liver transplantation. However, a The main risk of these drugs is thrombotic complica- small randomised study of 20 patients given aprotinin tions, through inhibition of fibrinolysis, a naturally- showed that the drug was not effective as compared occurring defence mechanism against thrombus for- with placebo. mation. There are at least ten case reports of forma- Side effects tion of thrombi in abnormal amounts or at abnormal Aprotinin is an heterologous polypeptide, and locations associated with the use of these drugs. On therefore it could cause hypersensitivity reactions, par- the other hand, no strikingly increased risk of throm- ticularly after repeated exposure. Theoretically, apro- bosis has emerged when the drugs were used during tinin could cause venous and arterial thrombosis and operations often complicated by venous and arterial therefore occlusion of coronary bypass and other vas- thromboembolism, such as cardiac surgery and knee cular grafts. However, in controlled studies, aprotinin replacement. However, these studies were not did not lead to an increased rate of mortality, acute designed to evaluate thrombotic complications and myocardial infarction or early occlusion of saphenous usually were too small to detect differences in rela- vein or internal mammary artery grafts after coronary tively low-incidence outcomes such as stroke, myocar- bypass grafting, or to an increased risk of venous dial infarction or coronary bypass graft occlusion. thromboembolism after hip replacement. 122 M. Cattaneo et al.

Desmopressin Hence the drug is a possible prophylactic treatment Plasma concentrations of factor VIII, the clotting for patients who need invasive diagnostic procedures factor deficient or defective in haemophilia A, and and have a prolonged bleeding time. von Willebrand factor, the adhesive protein deficient Surgical blood loss or defective in von Willebrand disease, can be In a single study of 70 patients undergoing com- increased for a short time by 1-deamino-8-D-arginine plex cardiac surgery, desmopressin given at the time vasopressin (desmopressin), an analogue of arginine of chest closure reduced blood loss and transfusion vasopressin. These effects, mimicking replacement requirements by about 30 percent. However, subse- therapy with blood products, are the rationale for the quent studies carried out in patients undergoing sim- use of desmopressin in the treatment of patients with pler cardiac operations, showed that there were no these congenital bleeding disorders. Subsequently, significant differences between desmopressin and desmopressin has also been used in patients with oth- placebo. In a meta-analysis of 17 clinical trials which er congenital and acquired bleeding disorders. In included 1,171 patients desmopressin significantly these, the effect of desmopressin is mediated by reduced post-operative blood loss by 9 percent, a val- mechanisms that are both dependent and indepen- ue of little clinical importance. However, a sub-analy- dent of the attainment of high plasma concentrations sis of the data revealed that desmopressin reduced of von Willebrand factor with ultralarge multimers, blood loss by more than 30% under conditions caus- which support platelet aggregation and adhesion to ing excessive blood loss. the vascular subendothelium more than multimers of normal size. Other mediators of increased haemosta- Side effects sis might be high plasma concentrations of factor VIII, Frequent side effects include mild facial flushing a rate accelerating factor in the process of fibrin for- and headache. Because of its potent antidiuretic mation. effect desmopressin can cause water retention and Congenital bleeding disorders hyponatraemia. In patients given more than one dose, plasma sodium and body weight should be measured Desmopressin is the treatment of choice for patients daily and excessive administration of fluids avoided. with mild haemophilia A or type 1 von Willebrand dis- Arterial thrombosis (sometimes fatal stroke or ease who bleed spontaneously or who are to undergo myocardial infarction) has occurred in a few patients any operation. Plasma concentrations of factor VIII after treatment. In patients at high risk for thrombo- and von Willebrand factor increase approximately 2 to sis (such as those undergoing coronary artery bypass 4 times, with a peak 30 to 60 minutes after intra- grafting) there was no excess in thrombotic compli- venous infusion and 60 to 90 minutes after subcuta- cations in those given desmopressin. neous and intranasal administration. These doses can be repeated as necessary at intervals of 12 to 24 hours, Prophylactic value of haemostatic but tachyphylaxis may occur after 3 to 4 doses. Effects on the bleeding time. Desmopressin shortens drugs in cardiac surgery the prolonged bleeding time in most patients with EACA, tranexamic acid, desmopressin and apro- type 1 von Willebrand disease and also in some tinin have been evaluated in patients undergoing car- patients with congenital defects of platelet function. diac surgery. Based on direct comparison studies and However, the bleeding time in patients with type 3 or a meta-analysis all four reduce operative blood loss. type 2 von Willebrand disease is usually not short- The order of efficacy in terms of reduction in blood ened. The optimal intravenous or subcutaneous dos- loss (greatest to least) is aprotinin, tranexamic acid, es of desmopressin are 0.3 µg per kg and the optimal EACA and desmopressin; the order of cost at the intranasal dose is 300 µg in adults and 150 µg in chil- most commonly recommended doses is the same. In dren. terms of reduction of blood transfusion requirements, the most important efficacy criterion, the results Acquired bleeding disorders favour aprotinin. In terms of safety, firm data from Desmopressin has also been used in patients with clinical trials demonstrating no increased frequency of uraemia, who have complex abnormalities of graft occlusion are available for aprotinin only. haemostasis reflected in part by a prolonged bleeding The cumulative evidence leads to the choice of time. In a group of these patients given an intravenous aprotinin, but it should be reserved for those patients infusion of desmopressin, the prolonged bleeding who are likely to need transfusion of donor blood. time became normal for 4 to 6 hours in about 75 per- They are those undergoing re-operation, those with cent. Desmopressin given before invasive procedures pre-existing haemostatic defects or taking antiplatelet (biopsies and major surgery) seems to prevent bleed- drugs and those with sepsis. ing, but controlled studies are lacking. In spite of the fact that patients with cirrhosis have Conjugated oestrogens high plasma concentrations of factor VIII and von Conjugated oestrogens shorten the prolonged Willebrand factor, they have a prolonged bleeding bleeding times and improve or stop bleeding in time that is shortened by intravenous desmopressin. patients with uraemia. The mechanism whereby con- Session #11 – Transfusion medicine 123 jugated oestrogens affect the bleeding time in these bleeding in patients with other haemorrhagic disor- patients is unknown. Recommended dosage in ders, including congenital defects of platelet function, patients with uraemia is a single i.v. daily infusion of chronic liver disease and haemostatic defects induced 0.6 mg per kilogram repeated daily for 4 to 5 days by the therapeutic use of antithrombotic drugs such which shortens the bleeding time by approximately 50 as aspirin and ticlopidine, but there is still no well percent for at least two weeks. A daily oral dose of 50 conducted clinical trial that demonstrates efficacy. In mg shortened the bleeding time after an average of 7 cardiac surgery, antifibrinolytic drugs (lysine deriva- days of treatment. tives and aprotinin) are more effective than desmo- The clinical value of conjugated oestrogens in pressin, aprotinin being preferable because its effica- patients with uraemia rests on data indicating that cy and safety have been more extensively evaluated. the bleeding tendency of these patients is directly related to degree of prolongation of the bleeding time. The chief advantage of conjugated oestrogens over References desmopressin is the longer duration of the effect on 1. Bonnar J, Sheppard BL. Treatment of menorrhagia the bleeding time (10 to 15 days vs 6 to 8 hours). during menstruation: randomised controlled trial of Hence, conjugated estrogens should be used when ethamsylate, mefenamic acid and tranexamic acid. Br long-lasting haemostatic competence is required, Med J 1996; 313: 579-82. such as during elective surgical procedures or recur- 2. Cattaneo M, Harris AS, Stromberg U, Mannucci PM. The effect of desmopressin on reducing blood loss in rent episodes of gastrointestinal or nose bleeding. On cardiac surgery. A meta-analysis of double-blind, the other hand, desmopressin should be given when placebo-controlled trials. Thromb Haemost 1995; 74: an immediate effect on haemostasis is required (for 1064-70. instance, to stop acute bleeding or to prevent bleed- 3. Cattaneo M. Review of clinical experience of desmo- pressin in patients with congenital and acquired bleed- ing at the time of emergency surgery). The two prod- ing disorders. Eur J Anaesthesiol 1997;14 (Suppl. 14): ucts can be given concurrently, exploiting the different 10-8. timing of their maximal effects. 4. Fremes SE, Wong BI, Lee E, et al. Meta-analysis of pro- In patients with chronic renal insufficiency, recom- phylactic drug treatment in the prevention of postop- binant erythropoietin causes a dose-dependent rise erative bleeding. Ann Thorac Surg 1994; 58: 1580-8. 5. Mannucci PM. Desmopressin (DDAVP) in the treat- in the haematocrit and eliminates the need for blood ment of bleeding disorders: the first 20 years. Blood transfusions. The progressive increase in the haema- 1997; 90:2515-21. tocrit is paralleled by a pronounced shortening of the 6. Mannucci PM. Hemostatic drugs. N Engl J Med 1998; bleeding time and improvement of platelet adhesion. 339:245-53. 7. Moia M, Mannucci PM, Vizzotto L, Casati S, Cattaneo Since most patients with chronic renal insufficiency M, Ponticelli C. Improvement in the haemostatic are now regularly treated with erythropoietin, there is defect of uraemia after treatment with recombinant less need for short-acting haemostatic drugs such as human erythropoietin. Lancet 1987; 2:1227-9. desmopressin and conjugated oestrogens. 8. Munoz JJ, Birkmeyer NJO, Birkmeyer JD, O’Connor GT, Dacey LJ. Is ⑀-aminocaproic acid as effective as Conjugated oestrogens are well tolerated and side aprotinin in reducing bleeding with cardiac surgery? effects are negligible or absent. Circulation 1999; 99:81-9. 9. Royston D, Bidstrup BP, Taylor KM, Sapsford RN. Conclusions Effect of aprotinin on need for blood transfusion after repeat open heart surgery. Lancet 1987; 2: 1289-91. The antifibrinolytic drugs EACA and tranexamic 10. Sindet-Pedersen S, Ramstrom G, Bernvil S, Blomback acid are useful in patients with a broad range of M. Hemostatic effect of tranexamic acid mouthwash haemorrhagic conditions, particularly when there is in anticoagulant treated patients undergoing oral excessive bleeding from mucosal sites. Desmopressin surgery. N Engl J Med 1989; 320:840-3. is the treatment of choice for patients with mild 11. Verstraete M. Haemostatic drugs. In: Bloom AL, Forbes CD, Thomas DP, Tuddenham EGD, eds. haemophilia and type 1 von Willebrand disease. It Haemostasis and thrombosis. 3rd ed. London: has also been used successfully to treat or prevent Churchill Livingstone, 1994. p. 1057-75.