Top Banner
Leukemia Research 27 (2003) 95–120 Millennium review The myelodysplastic syndrome(s): a perspective and review highlighting current controversies David P. Steensma , Ayalew Tefferi Division of Hematology, Department of Internal Medicine, Mayo Clinic, Rochester, MN 55905, USA Received 21 May 2002; accepted 30 May 2002 Abstract The myelodysplastic syndrome (MDS) includes a diverse group of clonal and potentially malignant bone marrow disorders characterized by ineffective and inadequate hematopoiesis. The presumed source of MDS is a genetically injured early marrow progenitor cell or pluripotential hematopoietic stem cell. The blood dyscrasias that fall under the broad diagnostic rubric of MDS appear to be quite heterogeneous, which has made it very difficult to construct a coherent, universally applicable MDS classification scheme. A recent re-classification proposal sponsored by the World Health Organization (WHO) has engendered considerable controversy. Although the precise incidence of MDS is uncertain, it has become clear that MDS is at least as common as acute myelogenous leukemia (AML). There is considerable overlap between these two conditions, and the former often segues into the latter; indeed, the distinction between AML and MDS can be murky, and some have argued that the current definitions are arbitrary. Despite the discovery of several tantalizing pathophysiological clues, the basic biology of MDS is incompletely understood. Treatment at present is generally frustrating and ineffective, and except for the small subset of patients who exhibit mild marrow dysfunction and low-risk cytogenetic lesions, the overall prognosis remains rather grim. In this narrative review, we highlight recent developments and controversies within the context of current knowledge about this mysterious and fascinating cluster of bone marrow failure states. © 2002 Elsevier Science Ltd. All rights reserved. Keywords: Disease classification; FAB; IPSS; Myelodysplastic syndrome; Acute myelogenous leukemia; WHO 1. Introduction Almost every journal article reporting research on the myelodysplastic syndrome (MDS) or reviewing these disor- ders includes an introductory statement similar to the fol- lowing. The myelodysplastic syndrome(s) include(s) a heteroge- neous group of clonal bone marrow disorders character- ized by ineffective hematopoiesis and a variable risk of transformation to acute myelogenous leukemia. This broad definition, refined and supported by ample re- ports over several decades, is a statement with which almost all contemporary MDS investigators can wholeheartedly agree. But when it comes time to debate more specific details, this harmony quickly dissolves. This is because a number of important questions about MDS currently lack the definitive answers that all investigators crave: answers grounded in unambiguous data from rigorously reviewed Corresponding author. Tel.: +1-507-284-2479; fax: +1-507-266-4972. E-mail address: [email protected] (D.P. Steensma). scientific reports. As is the case in many other areas of medicine and science, in the absence of convincing evi- dence to answer the tough questions, strong and contrary opinions can flourish and vigorous discussion results. How should MDS be defined—what minimal criteria must individual cases meet in order to be labeled “MDS”? What are the most useful ways to classify the various subgroups of patients? There are many reports describing potential patho- physiological clues; which avenues of biological exploration are most likely to yield results? Of more immediate inter- est to suffering patients: what are the best therapeutic ap- proaches to the various subtypes of MDS, and which patient and disease features allow the most accurate prognostication of future events? Complex questions like these have no easy solutions, and many dedicated investigators across the globe are working diligently to try to burn away the fog that enshrouds MDS. The authors of this review certainly do not presume to have the answers to any of these challenging questions. Instead, we share in the excitement of those who are fortunate enough to be pursuing their research quarry with increasingly so- phisticated laboratory techniques in this present era of rapid 0145-2126/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved. PII:S0145-2126(02)00098-X
26

The myelodysplastic syndrome(s): a perspective and review ......syndrome behaved so aggressively and evolved so rapidly that there was never a real distinction from overt leukemia

Feb 09, 2021

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
  • Leukemia Research 27 (2003) 95–120

    Millennium review

    The myelodysplastic syndrome(s): a perspective and reviewhighlighting current controversies

    David P. Steensma∗, Ayalew TefferiDivision of Hematology, Department of Internal Medicine, Mayo Clinic, Rochester, MN 55905, USA

    Received 21 May 2002; accepted 30 May 2002

    Abstract

    The myelodysplastic syndrome (MDS) includes a diverse group of clonal and potentially malignant bone marrow disorders characterizedby ineffective and inadequate hematopoiesis. The presumed source of MDS is a genetically injured early marrow progenitor cell orpluripotential hematopoietic stem cell. The blood dyscrasias that fall under the broad diagnostic rubric of MDS appear to be quiteheterogeneous, which has made it very difficult to construct a coherent, universally applicable MDS classification scheme. A recentre-classification proposal sponsored by the World Health Organization (WHO) has engendered considerable controversy.

    Although the precise incidence of MDS is uncertain, it has become clear that MDS is at least as common as acute myelogenous leukemia(AML). There is considerable overlap between these two conditions, and the former often segues into the latter; indeed, the distinctionbetween AML and MDS can be murky, and some have argued that the current definitions are arbitrary. Despite the discovery of severaltantalizing pathophysiological clues, the basic biology of MDS is incompletely understood. Treatment at present is generally frustratingand ineffective, and except for the small subset of patients who exhibit mild marrow dysfunction and low-risk cytogenetic lesions, theoverall prognosis remains rather grim. In this narrative review, we highlight recent developments and controversies within the context ofcurrent knowledge about this mysterious and fascinating cluster of bone marrow failure states.© 2002 Elsevier Science Ltd. All rights reserved.

    Keywords:Disease classification; FAB; IPSS; Myelodysplastic syndrome; Acute myelogenous leukemia; WHO

    1. Introduction

    Almost every journal article reporting research on themyelodysplastic syndrome (MDS) or reviewing these disor-ders includes an introductory statement similar to the fol-lowing.

    The myelodysplastic syndrome(s) include(s) a heteroge-neous group of clonal bone marrow disorders character-ized by ineffective hematopoiesis and a variable risk oftransformation to acute myelogenous leukemia.

    This broad definition, refined and supported by ample re-ports over several decades, is a statement with which almostall contemporary MDS investigators can wholeheartedlyagree. But when it comes time to debate more specificdetails, this harmony quickly dissolves. This is because anumber of important questions about MDS currently lackthe definitive answers that all investigators crave: answersgrounded in unambiguous data from rigorously reviewed

    ∗ Corresponding author. Tel.:+1-507-284-2479; fax:+1-507-266-4972.E-mail address:[email protected] (D.P. Steensma).

    scientific reports. As is the case in many other areas ofmedicine and science, in the absence of convincing evi-dence to answer the tough questions, strong and contraryopinions can flourish and vigorous discussion results.

    How should MDS be defined—what minimal criteria mustindividual cases meet in order to be labeled “MDS”? Whatare the most useful ways to classify the various subgroups ofpatients? There are many reports describing potential patho-physiological clues; which avenues of biological explorationare most likely to yield results? Of more immediate inter-est to suffering patients: what are the best therapeutic ap-proaches to the various subtypes of MDS, and which patientand disease features allow the most accurate prognosticationof future events?

    Complex questions like these have no easy solutions, andmany dedicated investigators across the globe are workingdiligently to try to burn away the fog that enshrouds MDS.The authors of this review certainly do not presume to havethe answers to any of these challenging questions. Instead,we share in the excitement of those who are fortunate enoughto be pursuing their research quarry with increasingly so-phisticated laboratory techniques in this present era of rapid

    0145-2126/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved.PII: S0145-2126(02)00098-X

  • 96 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    biomedical progress. We confidently look forward to devel-opments in the near future that will lead to more certaindiagnoses and better treatments.

    This article is not meant to be comprehensive overviewof MDS. More than 4800 manuscripts about “myelodys-plastic syndromes” have been published since the NationalLibrary of Medicine began indexing this term in 1986, and1151 articles on “preleukemia” appeared between 1977 and1985. It would take a large volume indeed to do justiceto each critical topic and each major area of investiga-tion. In lieu of being all-inclusive, we hope to highlighthere some of the most active, controversial, or interestingareas.

    2. Terminology: the power of language

    Challenges with respect to MDS begin at the mostfundamental level: the language used to define and de-scribe the condition. The importance of accurate and un-ambiguous disease terminology extends far beyond thedevelopment of communication tools to aid researchersand clinicians. Writing in the neoplasia–nosology traditionestablished by literary critic Susan Sontag, the late Profes-sor Suzanne Fleischman, an exceptionally articulate MDSpatient who was a scholar of French and Romance Philol-ogy at the University of California at Berkeley, pointedout that the language of medicine also colors the waypatients think of themselves and their suffering and canaffect the perceptions of their physicians[1–3]. For ex-ample, “preleukemia”, an older term for MDS that is no

    Table 1Some previous terms for the myelodysplastic syndromes, with key references (modified from[370,371])

    Anemia pseudo-aplastica 1907 Luzzatto[372]Refractory anemia 1938 Rhoades and Barker[373]Odoleucosis 1942 Chevallier[374]Preleukemic anemia 1949 Hamilton-Paterson[375]Preleukemia 1953 Block et al.[376]Chronic refractory anemia with sideroblasts 1956 Bjorkman[377]Refractory normoblastic anemia 1959 Dacie et al.[378]Smoldering acute leukemia 1963 Rheingold et al.[379]Subacute myeloid leukemia 1960s First use uncertainChronic erythremic myelosis 1969 Dameshek[380]Refractory anemia with partial myeloblastosis 1969 Dreyfus et al.[381]Refractory anemia with excess myeloblasts 1970 Dreyfus et al.[382]Subacute myelomonocytic leukemia 1972 Zittoun et al.[383]Refractory megaloblastic anemia 1972 Lehrer et al.[384]Refractory macrocytic anemia 1970s First use uncertainPreleukemic syndrome 1973 Saarni and Linman[385]Chronic myelomonocytic leukemia 1974 Miescher and Farquet[386]Hypoplastic acute myelogenous leukemia 1975 Beard et al.[387]Hemopoietic dysplasia 1978 Linman and Bagby[138]Dysmyelopoietic syndrome 1980 Streuli et al.[388]Myelodysplastic syndromes 1982 Bennett et al.[78]

    The most appropriate name for what are now known as the myelodysplastic syndromes was a major topic at a 1975 symposium in Paris; a transcriptof the debate was published in a special 1976 issue of the journalBlood Cells[5]. Terms discussed included preleukemia, preleukemic states, myeloiddysplasia, myeloid dysplastic disorders, myelodysplastic syndrome(s), hematopoietic/hemopoietic dysplasia, stem cell dysplasia, and stem cell disease.

    longer widely used[4], cast a darker specter than the cur-rently favored terms because it included the emotionallycharged word “leukemia”. “Preleukemia” also fell into dis-favor because some patients died of complications of thecondition even though the dreaded full-blown leukemianever appeared, while in other instances the “preleukemic”syndrome behaved so aggressively and evolved so rapidlythat there was never a real distinction from overt leukemiaanyway.

    The term “myelodysplastic syndrome(s)” emerged in themid 1970s from a long list of potential candidate descriptors(Table 1) [5] and has become sanctified through internationalcurrency. The ideal terminology remains elusive; althoughthe term MDS appears to have staying power, it may bemisleading in several respects[6].

    2.1. “Myelo-” may be misleading. . .

    Although the prefixmyelo- accurately designates the siteof origin of MDS in the bone marrow, the term has severalmeanings, andmyelo- can imply narrow restriction of a dis-order to non-lymphoid cells, i.e. those of erythroid, granu-locyte, megakaryocyte, and monocyte/macrophage lineage.Yet in some cases of MDS, lymphoid cells can also be shownto be part of the aberrant clone[7–12], albeit infrequently[13–16], and rarely, cases of MDS will transform to lym-phoblastic leukemias[17–20]. These phenomena reflect thefact that the cell of origin in MDS can be a very early multi-potential hematopoietic progenitor and perhaps even the truemarrow incunabulum, the omnipotent undifferentiated stemcell [21]. Admittedly, this terminological quibble is minor in

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 97

    comparison to the other problems described in the followingsections.

    2.2. “Dysplasia”: deceptive?

    The pathological termdysplasiausually refers to congen-ital, developmental disorganization of cells; in the contextof evolving neoplasia,dysplasiaencompasses architecturaldisruption accompanied by cellular pleomorphism and isclassically restricted to epithelial tissues[22]. The term“myelodysplasia” in the former sense is actually used insome medical circles to refer to the congenital neural tubedefects, confounding computerized searches of the medicalliterature. Hematopoietic MDS, which arises in a mesoder-mal tissue (marrow), usually represents a well-establishedneoplastic clone and not “dysplasia” in the strictest senseof the word[23].

    When MDS was first proposed as a diagnostic term inthe 1970s, the word “dysplasia” was evolving to a broaderusage that included virtually any pre-cancerous lesion; atthat time MDS was considered to be a preliminary stageto acute leukemia[5,24]. Although it was widely suspectedin the 1970s that MDS/preleukemia was a clonal disorderon the basis of the characteristic chromosomal abnormali-ties, several years passed before more definitive proof of theclonality of MDS arrived in the form of studies of X-linkedgene and gene product polymorphisms[10,25–27]. More re-cently, MDS has been shown to share some biological fea-tures with clonal “dysplasias” in other body sites, such as thedysplasias affecting the uterine cervix and gastroesophagealmucosa, including increased proliferation rates, increasedapoptosis, altered telomere dynamics, alterations in levelsof cell cycle and apoptotic regulatory proteins, alterations inmicroenvironmental cytokine levels, and a tendency to un-dergo genetic devolution[28]. However, MDS differs fromthese other dysplasias in several critical respects, such as thelack of a proven microbiologic origin and the extreme rarityof spontaneous regression[28,29].

    Some investigators have argued that the chronic myeloiddisorders are really “myeloneoplasias”—i.e. they do notrepresent preliminary stages of true clonal disorders butare already fully developed clonal, malignant disordersthat tend to have an indolent nature initially but undergoclonal evolution, analogous to the well-known Vogelsteinmodel for the progression of colon cancer[30,31]. Thesame considerations apply to the myeloproliferative dis-orders, which are also (almost always) clonal entities andshare some overlapping morphological and clinical fea-tures with MDS [32,33]. In fact, at one time, the term“myelodysplasia” was proposed as a broad categorical la-bel for all of the chronic myeloid disorders, and couldhave been the heir to an older, non-specific eponym onceused indiscriminately for marrow disorders chiefly af-fecting the erythroid elements, “DiGuglielmo syndrome”[34,380]. However, this terminology did not catch on,and the word myelodysplasia was eventually applied to

    the restricted set of disorders that are the subject of thisreview.

    The dysplasia versus neoplasia distinction has importantpractical consequences. Patients diagnosed with MDS oftenask their physicians, “Do I have a form of cancer?”— aquestion especially relevant for patients who carry one of theincreasingly common cancer-specific health insurance poli-cies[35]. Savvy patients with MDS may also ask whether ahematologist or an oncologist should direct their care. It canbe unsatisfying to explain to such patients that their diseaseis felt to reside along a shadowy frontier between malignantand benign disease. It can be equally challenging to persuaderesearch funding agencies dedicated to curing cancer thattheir money would also be well spent by bankrolling MDSinvestigations. Improving the terminology of the chronicmyeloid disorders might mitigate some of these problems.

    2.3. “Syndrome” is suspect: should it be superseded?

    The nebulous wordsyndromereflects the incomplete un-derstanding of MDS when the disorder began to be morewidely recognized in the early 1970s[36,37]. Although thepathogenesis and natural history of MDS have become morecompletely understood in the last 30 years, a great deal ofwork remains. Still, there has been enough progress thatsome have argued that the time may now be ripe for MDSto be considered a set of “diseases” with common biologicaland clinical parameters[37,38].

    From classical Greece until the 20th century, the term“syndrome” referred exclusively to a cluster of three or moresigns and symptoms often seen together, without referenceto etiology. During the 20th century, the term “syndrome”underwent devolution and is now used to describe virtuallyany characteristic pattern or bizarre occurrence, includingsome in areas of life far removed from medicine (e.g. Su-permom syndrome, Clinton syndrome)[36]. Medical librar-ians have pointed out that MDS represents an unusual useof the term syndrome as an all-encompassing label for acluster of possibly related pathologic conditions[36]. Per-haps, only the so-called 5q− syndrome[39] fits the former,stricter syndrome definition, because it is always associatedwith hypolobated micromegakaryocytes and isolated chro-mosome 5 deletions, while the responsible etiologic agent—presumably a tumor suppressor gene—remains obscure.

    In truth, some conditions that currently reside under thebanner of MDS have little apparent resemblance to one an-other. Pure sideroblastic anemia, for example, behaves quitedifferently from the typical case of chemotherapy-relatedrefractory anemia with excess blasts and a complex kary-otype [40,41]. Given this diversity, is the general categoryof MDS worth keeping at all? At least for the moment, itseems reasonable to do so, although disease “lumpers” and“splitters” may differ intelligently on this point. AlthoughMDS represents a cluster of associations without the impri-matur of a consistent genetic lesion, these conditions retainseveral hallmarks of “real” disease entities: pathologists

  • 98 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    can generally recognize the subtypes of MDS and theirdistinction has clear clinical relevance[42].

    While each patient is a unique individual, and the ad-vent of proteomic and genomic profiling threaten to smearthe several “primary colors” of MDS currently recognizedby marrow morphologists into a rainbow of intermediatehues, we believe general terms for these diseases will stillbe needed. There is power in group identity; most patientswant to know the name of their enemy and know that theyhave a disorder that their doctor recognizes and has seenbefore, even if their specific case has peculiar or idiosyn-cratic features[43]. Despite the pressure to fit patients intoa meaningful diagnostic category, the plurality of the blooddyscrasias that comprise MDS certainly should not be for-gotten, and clinicians need to be prepared for unpredictabil-ity. In light of this diversity, if the “syndrome” designationis to be maintained for MDS, perhaps the plural termsyn-dromesis preferable.

    3. Difficult diagnostic dilemmas

    All that glitters is gold, and not all bone marrow conditionswith pathologic features similar to MDS represent a truemonoclonal or oligoclonal neoplastic condition. The mini-mal criteria for a diagnosis of MDS are unclear, and severalversions have been proposed[37,44–49]. Some bone marrowmorphologists are uncomfortable making the diagnosis ofMDS in the absence of dysplasia in at least two cell lineages(erythroid, granulocytic, or megakaryocytic) since isolatederythroid dysplasia has so many potential etiologies. Therecent finding of mildly dysplastic hematopoiesis in a largepercentage of normal subjects[48,50] argues in favor of ahigher threshold for the morphologic diagnosis of MDS if anaccompanying MDS-associated cytogenetic abnormality islacking. Complicating matters is the recent finding of mon-oclonal hematopoiesis in some apparently normal elderlywomen[51], which may represent the myeloid counterpartof monoclonal gammopathy of undetermined significance(MGUS), an age-related clonal process. Whether myeloidclonality of undetermined significance has a risk for pro-gression similar to MGUS[52] has yet to be determined.

    Clinicians must be scrupulous to exclude MDS mimicssuch as nutritional deficiency (particularly Vitamin B12 andfolate [53]), toxin exposure (e.g. myelotoxic drugs, alcohol[54], lead, and arsenic[55]), and infection (HIV[56] andparvovirus B19[57,58]) in order to avoid a misdiagnosisand the potential for inappropriate administration of toxictherapy[59]. These disorders may look dysplastic, but theyare not clonal and they do not appear to have associatedgenetic abnormalities.

    One these non-clonal disorders are ruled out, however,the diagnostic work is not done. Distinguishing MDS fromsimilar clonal hematopoietic conditions can present a seri-ous challenge, as the frontiers between subsets of chronicmyeloid disorders are nebulous and there may be consid-

    erable overlap. Hybrid myeloproliferative–myelodysplasticsyndromes, for example, are not uncommon[32,60–62]. Di-agnosing MDS in the presence of a hypocellular marrow canbe particularly difficult[63], as cases of aplastic anemia thatare otherwise unremarkable may have detectable cytogeneticlesions, and the etiology of both disorders may be similar[64–66]. Marrow and peripheral blood cells in MDS canhave a paroxysmal nocturnal hemoglobinopathy (PNH) phe-notype with absence of glycosyl-phosphatidylinositol (GPI)anchored proteins[67–69]. MDS can also be seen in as-sociation with T-cell large granular lymphocyte disorders(T-LGL) [70], but the simple presence of a T-cell receptorgene rearrangement does not define a T-LGL, as such generearrangements are not lineage specific[71].

    Often, only the evolution of a particular patient’s disorderover time can allow clearer distinction between these severaloverlapping entities. The National Comprehensive CancerNetwork (NCCN) recognized this, and NCCN guidelinessuggest that if a case lacks classic features, several monthsof observation should pass before a diagnosis of MDS isassigned[72].

    Detailed genetic profiling aided by “gene chips” andrelated microarray technology promises eventual relieffor these difficult diagnostic dilemmas[73]. Eventually,MDS-specific DNA and mRNA transcript patterns may bedefined, which will set the diagnosis and classification ofthese marrow conditions on more solid ground, as genomicprofiling is already showing potential to do for other dis-eases[74,75]. MDS gene and protein expression patternswill become even more relevant when specific and effec-tive therapies are developed based upon them. Until thesepatterns are ferreted out and the critical genes are defined,the MDS counterparts of highly specific therapies such asSTI-571 (imatinib mesylate, GleevecTM) for chronic myel-ogenous leukemia[76] must remain only a distant dream.

    4. Disease classification: schemes and controversies

    The appropriate classification of the many varieties ofMDS has been a contentious topic for many years, and thedebate shows no signs of abating. The fundamental prob-lem seems to be that it is simply not possible to classifyincompletely understood disorders like MDS with absolutecertainty and to the complete satisfaction of all investiga-tors. Yet, paradoxically, further understanding of enigmaticdisorders cannot easily be achieved in the absence of theframework of a working classification. Each patient withMDS is unique and presents idiosyncratic clinical problems,yet if every patient were to be considered a “special case”because of their peculiar constellation of clinical and patho-logical features, the overall syndrome would suffer death bydeconstruction. Reproducible patterns of bone marrow be-havior are clearly seen; recognition of these can facilitatecommunication between investigators and allow forecastingof a particular patient’s disease course.

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 99

    Table 2The 1982 French–American–British (FAB) Cooperative Group classification of the myelodysplastic syndromes

    Subtype Myeloblasts inperipheralblood (%)

    Myeloblasts inbone marrow(%)

    Ringedsideroblasts(%)

    Absolutemonocytes inperipheral blood

    Auer rodspresent in bonemarrow?

    Refractory anemia (RA)

  • 100 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    Table 3The proposed World Health Organization (WHO) classification of neo-plastic diseases of the hematopoietic and lymphoid tissues[42]: categoriesrelevant to the myelodysplastic syndromes

    Myelodysplastic syndromesRefractory anemia

    With ringed sideroblasts (pure sideroblastic anemia)Without ringed sideroblasts

    Refractory cytopenia with multilineage dysplasia

    Refractory anemia with excess blastsWith 5–10% myeloblasts (RAEB-1)With 11–19% myeloblasts (RAEB-2)

    5q− syndromea

    Myelodysplastic syndrome, unclassifiable

    Myelodysplastic/myeloproliferative syndromesChronic myelomonocytic leukemiaa

    Atypical chronic myelogenous leukemiaJuvenile myelomonocytic leukemia

    Relevant acute myeloid leukemia (AML) categoriesAML with multilineage dysplasia

    With prior myelodysplastic syndromeWithout prior myelodysplastic syndrome

    AML and myelodysplastic-syndromes, therapy-relatedAlkylating agent-relatedEpipodophyllotoxin-relatedOther types

    a Like RAEB, CMML can also be subdivided based on myeloblastcount. The 5q− syndrome is narrowly defined to include only cases withde novo isolated del(5q) and the characteristic morphologic findings ofhypolobated megakaryocytes and less than 5% marrow myeloblasts[37].

    count of 19%. Based on these considerations, there werecalls to eliminate RAEB-T and to split the heterogeneousRAEB category into multiple subgroups.

    4.3. The World Health Organization (WHO) proposal

    In 1997, a working group of more than 100 cliniciansand pathologists met at Airlie House in Virginia under theauspices of the World Health Organization (WHO) to dis-cuss a new master classification of hematologic disorders[42]. Included in this classification was a proposal for there-classification of MDS (Table 3). In 2001, the final versionof the classification was published and incorporated into the10th edition of the WHO International Classification of Dis-eases (ICD-10), which was first used in 1994 and is the mostcurrent ICD classification[87,88].

    Clinicians and investigators with an interest in lym-phoma were already exposed to a forerunner of the newWHO classification in the form of the 1994 RevisedEuropean–American Lymphoma (REAL) classification[89]and were perhaps somewhat battle-weary after many decadesof contentious classification debate[90,91]. This group of-fered comparatively little resistance to the new WHO clas-sification[92]. Controversy over the subclassification of themyeloid disorders, in contrast, contributed to delays in thefinal publication of the classification, and several revisions

    were made between the initial and final WHO proposals.Another comparison between myeloid and lymphoprolif-erative disorders is revealing. Although MDS appears asclinically diverse as lymphoma, in the original WHO pro-posal MDS was represented by only six “pure” MDS sub-types and three myelodysplastic–myeloproliferative overlapdisorders; the lymphoproliferative disorder classificationcomprised more than 40 entities[42].

    4.4. Single lineage versus multilineage dysplasia: doesdegree matter?

    One major change between the WHO and the FAB clas-sifications is the recognition in the former that there is in-deed a difference between cases of MDS with morphologicdysplasia primarily restricted to one cell lineage (usuallyerythroid) and those with more widespread dysplasia. Asmentioned above, the FAB originally defined refractoryanemia and refractory anemia with ringed sideroblasts assyndromes with dysplasia largely restricted to the erythroidlineage, rendering cases with marked trilineage dysplasiadifficult to classify [93]. One key caveat is that mild dys-plasia restricted to the erythroid lineage is sometimes notclonal [50]. Stricter definitions for MDS such as thoserequiring at least bi-lineage dysplasia avoid inadvertentlyaffixing the label of MDS to cases that are not monoclonal,but also risk excluding genuine MDS cases with mini-mal dysplasia—a sacrifice of sensitivity at the expense ofspecificity for which there is no easy solution.

    Several reports have demonstrated that multilineage dys-plasia (e.g. “refractory cytopenia with multilineagedysplasia”) carries a worse prognosis than simple erythroiddysplasia[41,84,94], including cases with ringed siderob-lasts[40,95]. Whether this finding is independent of otherknown prognostic variables for MDS remains unclear[37].In addition, more severe dysplasia within a given lineagemay also portend a worse prognosis[96,97], although fur-ther studies are needed to support this claim and the samecaveat about independent prognostic value applies. A re-port by the Vienna group reviewing 431 MDS patients didnot validate the prognostic value of the new category ofRCMD, but a German group reviewing 1600 patients didsupport the WHO proposal on this point[41,98]. Several ofthe architects of the WHO classification pointed out that theVienna group used a different threshold (50% dysplasticcells versus 10% for other groups) to define whether alineage exhibited dysplastic features or not[97]. This dis-tinction may turn out to be quite important—dysplasia isunlikely to be a “yes or no” issue—but at present, thereare no studies that have published the effect of changesin the value of the “dysplasia differential” on prognosis.Most morphologists who diagnose RCMD appear to oper-ate more by a general gestalt (i.e. is there heavy dysplasia,occasional dysplasia, or none at all?) rather than actuallyenumerating the dysplastic and normal red cells, white cells,and megakaryocytes. More work is needed in this area.

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 101

    4.5. Chronic myelomonocytic leukemia (CMML): asyndrome with a new home

    The WHO classification removes CMML from MDSproper and puts it in a myeloproliferative–myelodysplasticoverlap category, along with two unusual disorders, atypicalchronic myelogenous leukemia and juvenile myelomono-cytic leukemia. This distinction seems reasonable, as CMMLwas always an awkward bedfellow in the FAB scheme, asmentioned above (fans of the musical group “The Beatles”might compare CMML’s uncomfortable presence along-side the FAB 4—RA, RARS, RAEB, and RAEB-T—to theawkward coupling of Yoko Ono with John, Paul, Ringo,and George).

    Although there is universal agreement on CMML’s het-erogeneity[79,82,99], it is unclear at present how one mightreproducibly split the disorder into distinct subtypes[81].Placement of CMML in a separate overlap category is a par-tial solution to this ambiguity, and dividing CMML into twocategories based on the blast count, as the WHO has donein their final proposal, may have more prognostic relevancethan attempts to divide CMML based on peripheral whitecount[97].

    But other overlap MDS-myeloproliferative conditionsdo not easily fit into one of the WHO categories[32,61],and there is currently no appropriate default category forsuch cases. Otherwise typical cases of MDS may have neu-trophilia, monocytosis, thrombocytosis, splenomegaly, orother myeloproliferative features, raising diagnostic angst:are such cases truly MDS with minor variation, or are theyactually different enough to require re-designation as aunique overlap syndrome?

    4.6. “Secondary” MDS/AML: “secondary” to what?

    The WHO classification includes a sub-category fortherapy-related MDS and AML. This category reflects thehistorical distinction between “secondary” and “primary”MDS; the former applies to patients who have previouslyreceived chemotherapy or radiotherapy for another disease.Confusing matters is the fact that the term “secondary”AML is used to describe both patients with prior genotoxictherapy exposures as well as patients with leukemia arisingout of a prior chronic myeloid disorder. Although pretreatedpatients generally do more poorly than those without sucha history[100–102], the prognosis appears to depend pri-marily on the cytogenetic profile (frequently abnormal and“high-risk” [83,103]) and not the history of treatment perse [21]. Whether or not such patients represent a distinctsubset of MDS deserving of a separate classification isquestionable.

    Someday it may become clear thatall MDS is “secon-dary”, albeit not always iatrogenic. Although findings havebeen somewhat inconsistent, there are already considerabledata supporting the fact that many cases of MDS and AMLmay result from toxin exposure in a susceptible host. Some

    of these susceptible persons may be those with polymor-phisms in genes encoding NADP(H) quinone oxidoreductase(NQO1) and glutathioneS-transferases (especially GSTT1and GSTM1), a group of enzymes involved in hydrocarbondetoxification[104–110]. Whether an MDS-inciting agentis encountered in the home or workplace or was instead dis-pensed by a pharmacist or radiotherapist may turn out to bean unimportant distinction.

    4.7. Distinguishing MDS from AML: scratchinglines in the shifting sand

    In some cases of AML, the diagnosis is obvious, whilein other cases making a clear distinction between MDSand AML can be extremely difficult. The FAB classifica-tion imposed an arbitrary threshold value of 30% marrowblasts to define AML; the WHO proposal lowers this thresh-old to 20% and thereby does away with the FAB categoryof “refractory anemia with excess blasts in transformation”(RAEB-T or RAEBIT). This proposed change has engen-dered strong criticism. Some have argued that RAEB-T isbiologically different from AML and should be retained asa diagnostic category[37,111], while others have empha-sized the similar prognosis for the two entities as well as theidentical response to treatment[85,86] and certain biologi-cal features which are indistinguishable[112]. Both propos-als suffer from the limitation that regardless of which blastcut-off for AML is accepted (30 or 20%), such numbers areof course fundamentally arbitrary[37]. The effect of blastpercentage on prognosis in MDS appears to be a continu-ous variable, as is true of most biological systems. Reflect-ing this, the final WHO proposal separates RAEB into twosub-categories depending on whether the marrow blast per-centage is 5–10% or 11–19%, a distinction that has beenshown to be prognostically important[37,83,113,114]. Thismay be the best that can be done given the limited precisionof bone marrow differential counts.

    One proposed compromise position is to redefine AMLin terms of a rate of progression rather than a strict blastpercentage; such an assessment would require several mea-surements over time, with relative stability consideredthe hallmark of MDS[37,38]. Cytogenetic features arealso important; MDS is most often characterized by dele-tions and (less commonly) gains of chromosomal material,while recurrent translocations are more common in AML[101,115]. Most hematologists would consider a marrowexhibiting a classic AML-associated genetic lesion such ast(8;21)(q22;q22), t(15;17)(q22;q11–12), inv(16)(p13q22),or an anomaly of 11q23 as diagnostic of AML regardless ofthe blast count, and in fact the WHO scheme does categorizesuch cases as AML[42]. An MDS phase is observed onlyrarely with such lesions[116,117]. Yet, there are certainlycases of AML with a high blast count which are character-ized more by ineffective hematopoiesis than by blast burden[38], and there have been few calls to reclassify such cases asMDS. In addition, in some cases of AML, the bone marrow

  • 102 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    demonstrates multi-lineage dysplasia but no preceding MDSwas recognized. Although the WHO includes an AMLsub-category for such cases, the simple presence of dysplasiain AML has not had consistent prognostic value[118–123].

    4.8. Unclassifiable MDS: a spacious prison for caseswith disorderly conduct?

    The inclusion of an “unclassifiable” MDS category inthe WHO classification has been criticized[37]. The WHOworking group recognized that cases are occasionally seenwhich do not easily fit into the FAB classification andwanted to have a category for such patients for epidemi-ologic purposes. However, the number of unclassifiablecases is likely to be very dependent on each individualhematologist or morphologist and the strictness of the stan-dard to which each case is held. Among other difficultcases[60], hypoplastic/hypocellular MDS[63,65], MDSwith fibrosis[124], “refractory thrombocytopenia” and “re-fractory neutropenia” without anemia[125], MDS in thepresence of a simultaneous untreated lymphoplasmacyticclone [126,127], MDS associated with a granulocytic sar-coma[128], clonal cytogenetic abnormalities characteristicof MDS without clear morphologic changes[129,130],MDS with an associated T-cell clonal gene rearrangement[70], “paraneoplastic” MDS (of uncertain clonality)[131],and miscellaneous MDS-myeloproliferative overlap cases[62] could all be considered unclassifiable under the WHOscheme. It is not clear whether these should be considereddiscrete entities, though, because most of them are rareenough that they have not clearly been shown to behavedifferently from more typical MDS cases in large series.The WHO classification has been validated in a large ret-rospective study in which unclassifiable patients apparentlyaccounted for a very small number of cases, but not allmorphologists are convinced[41,98].

    In addition to these criticisms, the pediatric MDS com-munity has pointed out that their needs have not beenwell-served by either the FAB or WHO schema[38,132].The WHO classification controversy highlighted the factthat a robust, evidence-based pediatric MDS classificationwas sorely needed, and a consensus system has recentlybeen proposed[133].

    4.9. Beyond the FAB 5 and the WHO: rolling towarda less rocky classification

    The ideal classification for MDS will be simple, re-producible, and useful for treatment and prognostication.Among other virtues, such a classification scheme will“lump” similar biologic entities and “split” disparate ones,will minimize arbitrary distinctions, and will be fluid andeasily alterable in the face of progressive enlightenmentby research reports. Although the current schemes appearto fall somewhat short of these goals, emerging data ongene expression patterns and other biologic parameters may

    clarify distinctions among subtypes of MDS and betweenAML and MDS and move the science of classificationforward [73,134].

    The WHO classification architects are to be commendedfor successfully building on the FAB scheme with whichclinicians and pathologists were familiar, including some cy-togenetic data, and refining diagnostic categories in reason-able ways. But as with all classification schemes, the WHOproposal should be considered a work in progress, and asmore evidence accumulates about the significance of spe-cific genetic lesions and clinical features, revisions will benecessary.

    5. Epidemiology: how trustworthy are the numbers?

    Several factors have made true the incidence and preva-lence of MDS difficult to ascertain. As one epidemiologistcomplained, “we are put off by the fact that MDS is aheterogeneous, vaguely defined group of conditions withseemingly ever-changing names”[135]. MDS terminologyhas been somewhat consistent only since the 1982 FABclassification, and as discussed earlier, many MDS casesremain unclassifiable, uncertain, or diagnostically prob-lematic. Compounding the difficulty is the fact that casesof MDS are not routinely reported to cancer registries[136,137], a legacy of the benign vs. malignant murkinessdetailed earlier. Standard disease classifications such as thestill widely used ICD-9 carry terminology that is not evenconsistent with the FAB classification, let alone the WHO[135,137]. Additionally, MDS can be confused with otherconditions with similar names. It is not uncommon to finddeath certificates, hospital summaries, and patient databaserecords in which the terms “myeloproliferative disorder”and “myelodysplastic syndrome” and “myeloid leukemia”have been used interchangeably. Such muddle confoundsregistry-based work. Further, most epidemiologic studies inMDS have been limited to data from small, regional reg-istries[137]; large national or international studies are rare.

    5.1. How common is MDS?

    In the 1970s, shortly after MDS was morphologicallydefined by the FAB, it was estimated that there would be ap-proximately 1500 new cases of MDS per year in the UnitedStates[138]. More recent estimates suggest that this figureis too small by at least a factor of 10. If the annual incidencein the USA were as high as the crude rate of 9.3–12.6 casesper 100,000 persons per year suggested by two English stud-ies [139,140], more than 30,000 American cases of MDSwould be diagnosed annually. In comparison, the overallage-adjusted incidence rate of acute myeloid leukemia inthe USA was estimated at 2.9 cases per 100,000 personsper year in 1998, a figure that has not changed significantlysince 1973[141]. Even if only the more conservative MDSincidence estimates are accepted—crude incidence figures

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 103

    which generally range between 2 and 4 cases per 100,000persons per year[136,142–144]—it seems probable thatMDS is at least as common as AML. The elderly are partic-ularly vulnerable, as annual MDS incidence rates in patientsover 70 years of age have ranged between 15 and 50 casesper 100,000 persons per year[145].

    5.2. Rumors of the coming plague: is the incidenceof MDS really increasing?

    Many hematologists from a variety of locations aroundthe world believe that the incidence of MDS is increasing[44,146–149]. It remains unclear whether this is actuallytrue or whether such observed trends are simply a matterof increasing recognition of the syndromes. Changes in de-mographics such as the overall aging of the population inindustrialized nations can also be misleading[145].

    Several investigators have found no evidence for an in-crease in the incidence of MDS over time[136,143]. Sinceslight macrocytosis, for example, is not easily recognizableon a peripheral blood smear and since a high red cell meancorpuscular volume (MCV) may be one of earliest signsof MDS [150], it is probable that routine use of automatedhematology counters has highlighted certain milder casesthat might previously have been overlooked[140]. Otherfactors such as the increasing use of remission-inducing orcurative therapies for other disorders that are associated witha subsequent risk of MDS[151,152]may also be making asmall contribution to changes in MDS epidemiology[137].A widely held perception that the general environment is be-coming more “toxic” has been suspected of fanning fears ofa looming epidemic of exposure-related diseases like MDSand asthma[153].

    What will it take to obtain reliable incidence data forMDS? Uniform syndrome definitions, clear and comprehen-sive record-keeping, and multinational collaboration over aprolonged period of time represent challenging but poten-tially achievable goals that will greatly clarify the epidemi-ology of MDS.

    6. The mysterious biology of MDS

    Among the chief challenges in studying MDS are thepaucity of adequate cell lines and the lack of an appropri-ate animal model of the disease. Most mechanistic stud-ies have been carried out on fresh human tissue, of whichthere is necessarily a limited supply. The current collec-tion of MDS-related cell lines was recently reviewed[154].Most of the 10 MDS-specific (i.e. not secondary-AML) celllines that have been described are not currently availablefrom major cell banks, and several of these are lymphoidlines (of less interest for myeloid mechanistic work) and arepoorly validated[154]. The myelomonocytoid P39/Tsuganecell line, an MDS/AML cell line that has been the object ofseveral mechanistic studies of myeloid differentiation, has

    recently been reported by the Japanese Collection of Re-search Bioresources (JCRB) Cell Bank to be cross-contami-nated by HL60 cells[155–158]. Additional well-validatedMDS cell lines are clearly needed.

    A detailed discussion of the mechanisms that initiate andsustain MDS is beyond the scope of this paper, and only afew insights can be mentioned here.

    6.1. MDS causation: the prime mover remains unknown

    Oncogenesis is thought to be a multi-step process in whichseveral critical genetic lesions accumulate, eventually re-sulting in overt cancer. The development and progression ofMDS from its earliest stages through more advanced diseaseand its eventual transformation to AML appear to mesh withthis concept[106,137]. Multiple genetic pathways can beinvolved, and sometimes several distinct clones are presentin the same patient[101,110,159–161].

    Loss of genetic material from chromosomes 5, 7, 13,17, and 20 or a sex chromosome and (less commonly) ge-netic gains such as trisomy 8 are well described in MDS[162]. Several hundred other MDS-associated karyotypeshave been described[163], some of which are rare but re-current, yet the critical genes lost or grained in most ofthese lesions remain a mystery[164]. The region of chromo-some 5 often deleted in MDS, for example, contains multiplehematopoietic growth factors, but attempts to define whichgenes are the sine qua non of the classic 5q− syndromehave thus far been unrevealing[165,166]. Large cytogeneticlesions detectable by conventional karyotypic analysis arelikely to be late developments in the pathogenesis of MDS.Several patients have been described in whom clonality (asassessed by analysis of various X-linked genes) precededthe development of an overt cytogenetic lesion by severalyears[167].

    Almost half of de novo MDS cases have normalmetaphase cytogenetic findings. Sorting out the relevantaltered gene pathways in such cases remains an active areaof investigation. “FISHing expeditions” with multiplex-fluorescent in situ hybridization (M-FISH) and spectral kary-otyping techniques in cytogenetically normal MDS caseshave been generally unrevealing, although these proceduresmay be useful in clarifying the karyotype in complex cases,which at present is not often clinically relevant[168–170].Panel FISH techniques, in which a group of FISH probesare used to search for cryptic expression of common MDSgenetic abnormalities, are also of limited value[171].

    Two recent microarray studies of gene expression inMDS have shown increased expression of a gene encod-ing the delta-like (dlk) protein in low-risk MDS patientscompared with high-risk MDS, AML, CML, and normalcontrols[73,134]. The function of this gene is uncertain, butthere is new evidence that it may have a role in the cellulargrowth and differentiation programs, including differentia-tion of hematopoietic cells. In mice, forced overexpressionof dlk was a negative regulator of adipocyte differentiation,

  • 104 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    and a soluble dlk-IgG Fc chimeric protein was shown tocompletely inhibit formation of lineage-marker negative(Lin−) bone marrow cell colonies by colony stimulatingfactors in the presence of stem cell factor (SCF)[172,173].These microarray experiments represent a proof of concept,and it can be expected that global genomic work will yieldother interesting candidate genes whose pathophysiologicalrelevance in MDS can then be studied in more detail.

    Several MDS investigators have probed specific genesknown to be mutated in other forms of cancer andpre-cancerous lesions. Mutations ofras, fms, andp53 haveall been described in MDS, although the exact prevalenceremains uncertain[174–177]. In contrast, other genes of-ten mutated in AML such asAML1 do not appear to beabnormal in the majority of patients with MDS[178,179].

    The root cause of the underlying genetic lesions in MDSis often not known, although various epidemiologic asso-ciations support the importance of exposure to a genotoxicagent in a susceptible host. Patients treated with alkylatingagents, epipodophyllotoxins, and ionizing radiation are atincreased risk for MDS. MDS is more common in men,in persons with agricultural or industrial occupations orregular exposure to petroleum products, in smokers, and inpersons who use hair dyes[142,180–184]. MDS has alsobeen described after exposure to atomic radiation. A numberof cases have appeared in the Chernobyl nuclear accidentdecontamination workers[185], and although MDS was notyet recognized in the years following the atomic bomb blastsin Hiroshima and Nagasaki, re-review of the marrow speci-mens from those patients who developed AML revealed dys-plastic changes[186]. Atomic bomb survivors were recentlyfound to have a dose-dependent increase in the risk of MDSeven many years after exposure, lending yet more supportfor a multi-step pathogenesis of MDS[187]. The “naturalexperiments” of familial MDS[188–190]and MDS arisingin patients with congenital deletion of a tumor suppressorgene (e.g. neurofibromatosis type 1[191]) or a DNA repairdefect (e.g. Fanconi syndrome[192] and Bloom syndrome[193,194]) underscore the importance of host suscepti-bility.

    6.2. Too much apoptosis?

    In the late 1980s, apoptosis-associated morphologicchanges such as chromatin condensation and cytoplasmicblebbing were observed in hematopoietic progenitor cellsin bone marrow from patients with MDS[195]. Thesefindings suggested a resolution to one of the paradoxes ofMDS: the presence of peripheral blood cytopenias despitea typically hypercellular bone marrow[196,197]. Numer-ous subsequent studies have supported the hypothesis thatexcessive programmed cell death is a contributing factor tothe ineffective hematopoiesis in MDS. The marrow failureis usually not a result of decreased progenitor synthesis, asmarrow kinetic studies have determined increased prolifer-ation [198].

    Various techniques can be used to measure typical apop-totic changes in MDS marrow, including in situ end-labeling(ISEL) and nick-end labeling (TUNEL) of DNA strandbreaks[199–204], detection of phosphatidylserine migra-tion to the outer portion of the cellular phospholipid bilayerusing the annexin V binding protein[205,206], and quantifi-cation of sub-diploid (sub-G1 phase) DNA[207]. Althoughthere continues to be significant disagreement among in-vestigators regarding the degree and extent of apoptosis inMDS marrow and the culpability of stromal cells, severalclear trends have emerged.

    Multiple studies have demonstrated that there is an in-crease in the apoptotic index in the marrow of patients withearly MDS compared with normal controls. However, theapoptotic index represents a numerator (the number of apop-totic cells detected) over a denominator (the total number ofcells counted), and changes in the apoptotic index can re-sult from changes in either number. If the denominator wereCD34-positive cells, for instance, as has been the case inmost studies measuring the apoptotic index, alterations inthe pool of early hematopoietic progenitor cells might givemisleading results. Apoptotic phenomena are also time de-pendent, and not all studies have described the freshness ofstudied samples[208].

    The increased apoptotic index in MDS is associated withan increased proliferative fraction (as measured by Ki-67monoclonal antibody staining[206] or bromodeoxyuri-dine/iododeoxyuridine labeling indices[198,199,202,209])and signal antonymy[209]. In contrast, apoptotic indicesappear to be decreased in late MDS (refractory anemiawith excess blasts and refractory anemia with excess blastsin transformation) and in AML arising from a pre-existingMDS when compared with early MDS. The percentage ofapoptotic cells in early MDS also appears to decrease sig-nificantly after treatment with erythropoietin and/or G-CSF,suggesting one possible mechanism for the salutary increasein peripheral blood counts in patients treated with theseagents[201].

    Several investigators (including groups at Stanford Uni-versity, Rush Medical College, the University of Arizona,King’s in London, and others) have studied the mechanisticchanges contributing to excessive apoptosis in MDS, andmuch progress is being made. There is evidence for involve-ment of members of the Bcl-2 family, for example. In earlyMDS, the pro-apoptotic members of the Bcl-2 family (Bad,Bax) are overexpressed relative to the anti-apoptotic mem-bers (Bcl-2), but this ratio drops in late MDS and secondaryAML [206,210]. The ratio of c-myc(pro-apoptotic) to Bcl-2(anti-apoptotic) is elevated in cases of early MDS comparedto normal controls[207]. Vascular endothelial growth factor(VEGF) appears to be an important cytokine for leukemiccell proliferation and contributes to the morphologic phe-nomenon of abnormal localization of immature myeloidprecursors within the marrow microenvironment[211]. Theexpression of tumor necrosis factor (TNF), a pro-apoptoticcytokine, also appears to be increased in MDS[212–214],

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 105

    and this elevation correlates with increased activity of akey biochemical effector of apoptosis, caspase 3[215]. Inaddition, increased expression of another death-inducingprotein, Fas/CD95, has been noted in MDS and correlateswith ineffective erythropoiesis[216,217]. The degree of ex-pression of this receptor’s partner, Fas ligand, in MDS hasbeen correlated with FAB subtype, degree of anemia, andoverall survival[218]. In contrast, in one study granulocytesand CD34+ cells from MDS patients were resistant to Fas,TNF-�, and interferon-induced apoptosis[219]. Not to beoutdone, several of the receptors for the other known mem-ber of the death ligand/receptor family, TRAIL, have alsobeen shown to be overexpressed in MDS marrow comparedto normal controls, and treatment with exogenous TRAILinhibits myeloid progenitor proliferation[220,221].

    The underlying genetic lesions contributing to thesechanges remain obscure. It is also unclear whether increasedapoptosis is a desperate cellular reaction to a rapidly prolif-erating and genetically disturbed clone, or whether excessiveapoptosis is instead an integral part of the pathophysiologyof the syndrome.

    6.3. The enigma of ringed sideroblasts

    Ringed sideroblasts are abnormal erythroid precursors inwhich iron-stuffed mitochondria encircle the nucleus; thisiron is unavailable for incorporation into heme, resulting inanemia. The finding of ringed sideroblasts can be associ-ated with a number of congenital syndromes that cause de-fects in heme synthesis—usually via decreased activity of5-aminolevulinate (5-ALA) synthase, an X-linked enzymewhich has been found to be mutated in a subset of patientswith sideroblastic anemia[222]. Treatment with pyridoxine,a precursor for the cofactor of 5-ALA synthase, can improvethe anemia in some cases of MDS with ringed sideroblasts.Several polyclonal acquired ringed sideroblastic states alsoexist, such as lead intoxication and alcohol abuse, whichmay present a diagnostic challenge[223].

    Monoclonal acquired sideroblastic anemia is considereda myelodysplastic syndrome[78]. As mentioned above,“pure” acquired sideroblastic anemia carries a much morebenign prognosis than sideroblastic anemia associated withmulti-lineage dysplasia; the latter is also much more fre-quently associated with clonal cytogenetic anomalies[40].

    The genetic lesions responsible for most cases of acquiredsideroblastic anemia remain uncertain; mitochondrial DNAlesions may play an important role[224–226]. In one smallstudy, substitutional, deletional, and insertional mutationsin cytochromec oxidase subunit genes were detected in 13of 20 MDS patients but only 2 of 10 normal individuals; thesignificance is uncertain at present but mitochondrial cis-ternae morphologic abnormalities have also been observed[226,227]. Autosomal defects too small to be detected byconventional karyotypic analysis are also possible contrib-utors to sideroblastic anemia. The recent cloning of ABC7,an X-linked gene coding for an iron transporter which local-

    izes to the mitochondrial membrane, is of particular interestbecause the Xq13 locus where this gene is located has oc-casionally been associated with acquired myeloid disordersincluding some cases with ringed sideroblasts[228–230].

    7. Treatments: too few and often too futile

    Effective treatments for MDS are limited, but the longlitany of potential treatments to which a few patients will re-spond also makes therapeutic nihilism untenable. It is oftenstated that the only potentially curative treatment for mostpatients is also the riskiest, allogeneic stem cell transplanta-tion, but in fact a few patients (especially younger patientswith a normal karyotype) will achieve a prolonged poly-clonal remission after high-dose chemotherapy even withouttransplantation and may turn out to be “cured”[231]. Still,given the older age of most patients, gentle, supportive careremains the treatment standard[232].

    A number of conventional and experimental treatmentsfor MDS are briefly reviewed below. Although the list ofpotential treatments is long, very few are effective in a largenumber of patients. A major challenge lies in predictingwhich patient is most likely to respond to which treatment.Since the detailed mechanism of action for many agents isunknown, some of the agents listed later in one categorymay actually have activity that would make them equally athome in a different category (e.g. the interferons, which havegrowth, differentiation, cytotoxic, and immunomodulatoryeffects—or thalidomide, which has many poorly understoodbiological effects).

    7.1. The importance of response criteria: whatconstitutes success?

    For chronic disorders such as MDS, treatment success isalmost never a simple question of “yes” or “no”. Treatedpatients often exhibit gradations of clinical change in multi-ple organ systems—some insignificant and others of majorimportance, some salutary and others detrimental. Definingarbitrary criteria for therapeutic success or failure that allinvestigators can agree upon in the face of this biologicalcontinuum represents a major problem. When clinical tri-als in MDS are submitted to medical journals, disagreementover response criteria may be a cause of manuscript rejec-tion [233].

    Recently, an International Working Group (IWG) labor-ing under the auspices of the National Cancer Institute inBethesda, MD proposed a set of standardized response cri-teria for clinical trials in MDS[234]. Although such stan-dardized criteria are desperately needed, the IWG proposalhas been criticized by some. Several of these response crite-ria have been perceived as clinically irrelevant or difficult toapply consistently[233,235]. Using the IWG criteria, for ex-ample, independent groups reviewing raw data from clinicaltrials may come to different conclusions about the number

  • 106 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    of patients responding to treatment[233]. Less meaningfulhematologic changes could qualify as a treatment responseunder the current IWG criteria, while in other situations anovert cure would not technically qualify as a complete re-sponse[235]. For now, the IWG criteria may provide a usefulcommunication tool for reporting clinical trials, but minormodifications might help investigators use them with moreconfidence.

    7.2. Flogging a recalcitrant marrow: the role ofgrowth factors in MDS

    Recombinant human hematopoietic growth factors canbe an effective palliative tool in MDS, but they do notseem to prolong survival. The detailed studies of theScandinavian group have shown that erythropoietin (EPO)ameliorates anemia in about 20% of MDS patients witha serum EPO level of less than 200 U/l, but only rarelyworks in patients with RARS or those with higher endoge-nous EPO levels[236]. Very high doses of rHuEPO (upto 240,000 units per week) have been tolerated well byMDS patients, but using very high doses does not seem toimprove the response rate over the more typical dose of40,000–60,000 units per week[237]. In some patients whodo not respond to EPO, the addition of low doses of gran-ulocyte colony stimulating factor (G-CSF) (0.3–3.0 mcg/kgper day) or granulocyte-macrophage colony stimulating fac-tor (GM-CSF) may help recruit erythroid progenitors andthereby improve anemia as well as neutropenia[238–242].Both G-CSF and GM-CSF can increase neutrophil countand reduce infections in MDS[243–245], but no survivaladvantage has been demonstrated and there is still concernabout a small risk of accelerating transformation to acuteleukemia[21]. Results from trials of modified, longer-actinggrowth factors in MDS (e.g. darbepoeitin and pegylatedfilgrastim) have not yet been reported.

    Options are more limited for growth factor-based treat-ment of thrombocytopenia. Recombinant interleukin-11is a megakaryocyte growth factor recently FDA approvedfor the prevention of severe thrombocytopenia and todecrease platelet transfusion needs after myelosuppres-sive chemotherapy. It appears to have mild efficacy inMDS [246]. Side effects such as atrial arrhythmias andfluid retention are common and troublesome with thisagent in standard doses (50 mcg/kg per day), but maybe seen less often when very low doses (10 mcg/kg perday) are used[246]. Thrombopoietin (TPO) dynamics inMDS are complex[247,248], and no data are availableyet on the therapeutic use of recombinant TPO in thiscondition.

    Other agents with growth factor activity that have un-dergone or are undergoing clinical evaluation in MDS in-clude interleukin-6, interferon alpha and gamma (which alsohave differentiation properties in vitro), and interleukin-3.Each has had rather limited efficacy and substantial toxicity[249–260].

    Several growth factors have been tried in combinationwith other biologic agents such as amifostine, with varyingbut generally unimpressive results[261–264]. The use of allof the hematopoietic growth factors is currently limited bytheir high cost and the need for parenteral administration.

    7.3. The peril and promise of high dose chemotherapy

    High-risk MDS has very little distinguishing it fromAML, and therefore, AML-like therapy is a reasonableconsideration for patients with aggressive MDS such asthose with a high blast count. Although elderly patients tol-erate AML-type chemotherapy poorly, the median age forMDS patients is not markedly different from that for AML.Studies of patients with high-risk MDS reveal a 40–50%remission rate with high dose AML induction therapy, butpatients in trials of such regimens are usually a select groupyounger and healthier than the typical patient with aggres-sive MDS. Almost all patients achieving a remission viasuch therapy will relapse promptly[85,265,266]. In onestudy, only 5% of patients receiving high-dose therapy werealive at 3 years[231]. Growth factor support following ag-gressive therapy is tolerated but of uncertain benefit[267].

    Several high dose regimens such as the combination ofmitoxantrone and intermediate-dose cytosine arabinoside(ARA-C) have shown excessive toxicity with little redeem-ing characteristics[268]. In an attempt to move beyondthese limitations, the MD Anderson group is currently pi-oneering several combination programs containing neweragents such as topotecan, a topoisomerase inhibitor thathas been demonstrated to be useful in a number of solidtumors, and fludarabine, a nucleoside analog designed totreat lymphoproliferative disorders[231,269–272]. Thesedrugs appear to be somewhat toxic but effective in MDSwhen used in combination with ARA-C and anthracy-clines, and the optimal dosing and schedule has yet to bedetermined.

    7.4. Low dose chemotherapy: gentleness rebuffed

    For older and sicker patients with MDS in whom high dosecytotoxic therapy is deemed to be too dangerous, it wouldbe beneficial to have a useful low dose chemotherapeuticregimen that might assist in palliation. Unfortunately, nolow dose program has been particularly successful, althoughoccasional patients will have salutary results. In CMML,etoposide has been used palliatively with conflicting resultsthat may be schedule dependent[273–275]. Low dose oralmelphalan may also have brief palliative benefit in MDS[276], but all patients described in the original encouragingtrial have since relapsed and several developed a new 17pdeletion[277]. Low dose ARA-C looked promising whenfirst attempted in the 1980s, but an intergroup trial showedlittle efficacy and made this agent appear less exciting asmonotherapy[278]. Today, there is little enthusiasm for lowdose, non-specific cytotoxic agents.

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 107

    7.5. Differentiation therapy has yet to make a difference

    One of the hallmarks of MDS is that the neoplastic cloneexhibits a maturation block, a fixed stage of differentiationbeyond which the abnormal cells apparently cannot progress.Since the 1980s, multiple attempts have been made to in-duce maturation of these “stalled” cells. The resoundingsuccess with differentiation therapy with all-trans-retinoicacid in acute promyelocytic leukemia (AML M3) greatlyencouraged investigators with such aspirations[279]. Thusfar, there has not been a resounding success; the litany ofunsuccessful or marginally promising MDS trials with dif-ferentiation agents was recently reviewed[263].

    In addition to the interferons described earlier, otherdifferentiating agents that have been tried in MDS in-clude hexamethylene bisacetamide and its derivatives (thepolar–planar compounds[280,281]), homoharringtonine (aplant-derived alkaloid used in traditional Chinese medicine[282]), 5-azacytidine and 5-aza-2′-deoxycytidine (nucleo-side analogs which may also be cytotoxic and can modifygene methylation status), butyrates such as butyric acidand sodium phenylbutyrate (histone deacetylase inhibitorswhich may alleviate histone deacetylase-mediated tran-scriptional repression[283]), amifostine (an inorganic thio-phosphate originally designed to be a radioprotective agentbut subsequently found to stimulate hematopoiesis), hemearginate (a promoter of heme biosynthesis[284–286]),bryostatin (a macrocyclic lactone isolated from a micro-scopic sea creature[287]), retinoids such as Vitamin A andits analogs (stimulants of normal erythroid and myeloidprogenitor proliferation[262,264,288–292]), and VitaminD and its analogs (which bind to transcription factors thatcan induce terminal monocytic differentiation). With a fewexceptions, these agents have shown little real benefit, orhave been excessively toxic, or have induced only transientimprovements in hematopoiesis[263]. Amifostine, a rela-tively well-tolerated agent, showed considerable promise inan early trial, but subsequent studies have engendered pes-simism about the drug’s use in unselected cohorts of MDSpatients[293–295].

    Advocates of “biological”, differentiation-style therapyrecently received a boost when the Cancer and LeukemiaGroup B (CALGB) reported the results of a 191-patientrandomized trial of azacytidine versus supportive care inMDS [296]. In this study, statistically significant benefits inquality of life (i.e. less fatigue and dyspnea with improvedmood and overall performance status), rate of leukemictransformation and overall survival were observed whenazacytidine was administered subcutaneously at a dailydose of 75 mg/m2 for 7 days every 4 weeks, especiallyamong those patients who completed at least four cyclesof the drug[296]. Whether these results will translate intobroad usefulness in general clinical practice and whether5-aza-2′-deoxycitabine (decitabine) will prove even moreefficacious than its demethylating cousin remains to beseen[297]. Despite the many disappointments, the concept

    of differentiation therapy is so appealing that considerablework continues in this area, and one can only hope for moresuccesses in the near future.

    7.6. Anti-apoptosis therapy: suicide prevention

    Since excessive apoptosis seems to contribute to ineffec-tive hematopoiesis in MDS, it seems logical that prevent-ing programmed cell death might improve peripheral bloodcounts. Whether detrimental cell death can actually be pre-vented without inadvertently immortalizing a frankly neo-plastic clone remains to be seen. Since tumor necrosis factor(TNF) can induce apoptosis, and since some studies havesuggested that MDS marrow has higher than normal TNFlevels, anti-TNF therapy with etanercept (a soluble TNF-�receptor) and infliximab (a chimeric anti-TNF monoclonalantibody) is currently being attempted and has shown someinitial promise[298,299]. In one small study, ciprofloxacinand pentoxifylline down-regulated TNF expression but therewas no hematologic effect[300]. Combination therapy withamifostine, pentoxifylline, ciprofloxacin, and dexametha-sone is currently in vogue, but the trial introducing this regi-men appears somewhat less encouraging when patients whohad a neutrophil increment after receiving dexamethasoneare excluded from the tally of responders (see later for a dis-cussion of the effects of corticosteroids on neutrophil count)[301]. Inhibitors of the caspases, a family of enzymes thatrepresent the final common pathway in effecting apoptosis,have shown inconsistent results in improving hematopoiesisin vitro and have not yet been tried in vivo[302–304].Anti-apoptosis therapy has many obstacles to overcome be-fore it becomes reality.

    7.7. Immunologic manipulation: suppressing theoppressors

    Some patients with MDS may respond to immunesuppression directed at potentially auto-reactive T-cells.Because of the success of immune modulating therapyin aplastic anemia, patients with hypoplastic MDS havebeen assumed to be particularly good candidates for suchtherapy[21,305]. Agents including antithymocyte globulinand antilymphocyte globulin have shown responses in ap-proximately 10–20% of patients, but because these agentsare derived from horses, goats, or other animals, serumsickness and other toxicities can be problematic[306,307].Cyclosporin has also been used with some success[308,309]. To the best of our knowledge, there are no dataon the use of other immunosuppressive agents such astacrolimus (FK506) or rapamycin in MDS.

    7.8. The disappointing and the untested: miscellaneousstandard and novel therapies

    There are several novel agents and therapeutic agents ofhistorical importance in MDS that do not clearly fit into one

  • 108 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    of the other categories in this section. Since occasional pa-tients with sideroblastic anemia will respond to pyridoxine(Vitamin B6), a brief trial of this agent can be worthwhilein patients who appear to have RARS. Supplementation ofMDS patients with other vitamins is not usually worthwhile,including folate supplementation, as this group generally hasadequate folate stores[310,311]. Androgen therapy is gener-ally ineffective in MDS[312–314]. Danazol, a synthetic an-drogen, improved thrombocytopenia in occasional patientsin some studies but had no effect in other trials[315–317].

    Corticosteroids reliably increase the peripheral blood neu-trophil count, but this phenomenon is due to not to increasedneutrophil synthesis but rather to neutrophil release fromblood vessel walls and egress of a storage pool from bonemarrow[318]. Such increases are not associated with a de-creased risk of infection; trials of corticosteroids in MDShave been ineffective and associated with increased infec-tions [319].

    Iron chelation therapy with deferoxamine has been advo-cated for red cell transfusion-dependent patients, in orderto prevent iron overload and secondary hemochromato-sis [320]. Such therapy is cumbersome, however, sincedeferoxamine requires prolonged parenteral infusion bya pump. Sadly, most transfusion-dependent patients withMDS will not live long enough to be concerned about ironoverload, making such treatment perhaps not worthwhile.The use of chelation therapy may be justifiable especiallyin low-risk patients such as those with classic 5q− syn-drome. More recent data suggesting that iron overload maycontribute to ineffective hematopoiesis argues in support ofa wider role for chelation therapy[320]. Oral iron chelatorsare greatly needed, but progress has been slow. Deferiponeis a moderately effective oral iron chelator that is currentlyavailable, but there is controversy about its ability to preventiron-related hepatotoxicity[321].

    Novel agents currently undergoing clinical trials in MDSinclude farnesyltransferase inhibitors (which interfere withprocessing of theras oncogene, mutated in 10–40% ofMDS cases [175,177,322]—one farnesyltransferase in-hibitor has shown activity in relapsed AML[323]), gem-tuzamab ozogamicin (a monoclonal antibody against theCD33 epitope conjugated to a toxin; the CD33 epitope ispresent on early myeloid progenitors, and this agent hasbeen approved for acute myelogenous leukemia), variousputative anti-angiogenesis agents including matrix metal-loproteinase inhibitors, arsenic trioxide (an ingredient ina traditional Chinese remedy that has shown efficacy inacute promyelocytic leukemia[324,325]), inhibitors of theMDR-1 “drug-resistance” glycoprotein, troxicitabine (anovel nucleoside analog[326]), leucovorin (folinic acid,used because the dihydrofolate reductase gene maps to thelong arm of chromosome 5 and may be deleted in 5q−syndrome[327]), STI571 (a targeted tyrosine kinase in-hibitor custom-made for chronic myelogenous leukemia[76,328]), flavopiridol (a cyclin-dependent kinase in-hibitor), and others. Some clinical trials of novel agents

    designed for refractory AML are also open to high-riskMDS patients, expanding their options for clinical trialenrollment.

    7.9. Thalidomide: a ray of hope from a dark remedy?

    Thalidomide, the tragic teratogen that in 1961 catapultedthe Food and Drug Administration (FDA) into the modernera, was recently approved by the FDA for the rare clini-cal situation of leprosy-associated erythema nodosum[329].The drug is now enjoying extensive “off-label” use in hema-tologic diseases use because of its proven success in refrac-tory multiple myeloma[330], and it is currently being testedin a wide range of other disorders. Its biologic mechanismremains mysterious, as it has both immunomodulatory andanti-angiogenic actions, and dozens of associated cytokinechanges have been described[329,331].

    Several reports have described responses to thalidomidein MDS [332–335]. Many institutions are currently studyingthe drug alone and in combination with other agents, whichshould more clearly define its role in MDS[301,336]. Unfor-tunately, the drug can be difficult to tolerate, especially forthe older age group typical of MDS, because of its commonside effects such as drowsiness, rash, postural hypotension,peripheral neuropathy, bradycardia, constipation, and erec-tile dysfunction. Thalidomide embryopathy still occurs inleprosy-plagued areas of South America[337], and carefulsafeguards have been introduced in North America to limitthe possibility of the drug falling into the hands of anyonewho might become pregnant[336].

    7.10. Flavors of stem cell transplantation: auto, allo,mini, cord

    The role of stem cell transplantation in MDS has recentlybeen reviewed[338,339]. Overall, approximately 30–40%of MDS patients can be cured with allogeneic transplan-tation, but many of those will suffer scars such as chronicgraft-versus-host disease, which can be so disabling thatsome “successfully” transplanted patients have committedsuicide [340]. The best outlook with transplant is for pa-tients with early MDS who receive stem cells from fullyHLA matched donors; in this subgroup more than 75% willbe long-term disease free survivors. The major limitationto allogeneic transplantation is the older age of MDS pa-tients, whose median age is approximately 65 years[144].Older patients tolerate allogeneic transplantation poorly(although some success was recently achieved in a cohortof 55–66-year-old MDS patients, with a 3-year disease freesurvival of 33–53%, depending on disease subtype[341])and have a higher risk of disease relapse after the proce-dure [342]. Registry data have shown that patients whoreceive stem cells from an unrelated donor have a 2-yeardisease-free survival of only 29% and a treatment-relatedmortality of 54%[343]. No superior marrow ablation reg-imen has been defined[344], and the appropriate timing

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 109

    of transplant in MDS remains unclear[345]. Some havesuggested that patients transplanted early in their diseasecourse may do better long-term[346].

    Autologous stem cell transplantation for MDS at firstblush would seem a futile proposition—why transplantmarrow that potentially contains corrupt stem cells? Thesuccess of this endeavor hinges on the ability to selec-tively harvest and transplant polyclonal stem cells that arenot part of the neoplastic process, which can be accom-plished in some cases[347,348]. Some patients with MDSretain significant polyclonal hematopoiesis[349,350], andothers who receive high-dose chemotherapy can achievea remission with restoration of polyclonal hematopoiesis[351,352]. Stem cell collection is occasionally successfulin such patients[353]. The European Group for Blood andMarrow Transplantation has led the way in this area[346].Although autologous transplant was well tolerated in thesmall trials that have been reported[354,355], it is stilltoo early to know just what role autologous transplantationwill have in MDS, and the results of randomized trials areawaited.

    Novel transplantation methods include non-myeloablative(“mini-allogeneic”) transplantation, in which an attempt ismade to harness the immunologic activity of donor cellsto purge the recipient’s malignant clone and re-establishhealthy hematopoiesis[356,357]. This treatment modality atpresent is attempted primarily in patients felt to be too old ortoo sick to proceed with standard allogeneic transplantation,or in those whose disease has returned after conventionalallogeneic transplant.

    Allogeneic transplantation using umbilical cord blood of-fers distinct advantages such as a decrease in graft-versus-host disease[358]. Although cord blood transplantationcan be successful in adults[359], fetal blood often doesnot contain enough stem cells to re-establish hematopoiesisin a large adult. Efforts to expand the stem cell pop-ulation in cord blood in order to allow more frequenttransplantation options for larger individuals are ongoing[360].

    Table 4International Prognostic Scoring System for myelodysplastic syndromes[83]

    Score

    0 0.5 1 1.5 2

    Marrow myeloblast percentage

  • 110 D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120

    9. Conclusion

    Much has been accomplished in the last 10 years, butmany challenges remain in understanding the myelodysplas-tic syndromes. As the biology of these disparate disordersbecomes better understood, more appropriate classifica-tion schemes and more accurate prognostic indices will beachieved. Hopefully, improved treatments will also becomeavailable to brighten the outlook for patients with what atpresent can only be described as dismal diseases. Clinicaltrials of carefully selected novel therapeutic agents deservewidespread and enthusiastic support. Even as new agents be-come available, excellent supportive care by conscientiousphysicians will remain the best that medicine has to offer.

    References

    [1] Sontag S. Illness as metaphor. New York: Farrar, Straus & Giroux,1988.

    [2] Fleischman S. I am. . . , I have . . . , I suffer from . . . : a linguistrelfects on the language of illness and disease. J Med Human1999;20:3–32.

    [3] Wood ML. Naming the illness: the power of words. Fam Med1991;23:534–8.

    [4] Bagby Jr GC. The concept of preleukemia: clinical and laboratorystudies [review] [123 references]. Crit Rev Oncol Hematol1986;4:203–20.

    [5] Anonymous. Is preleukemic states an adequate designation? BloodCells 1976;2:347–51.

    [6] Lichtman MA. Language and the clonal myeloid diseases. Blood2002;99:725–6.

    [7] Tefferi A, Thibodeau SN, Solberg Jr LA. Clonal studies inthe myelodysplastic syndrome using X-linked restriction fragmentlength polymorphisms. Blood 1990;75:1770–3.

    [8] Culligan DJ, Cachia P, Whittaker J, Jacobs A, Padua RA. Clonallymphocytes are detectable in only some cases of MDS. Br JHaematol 1992;81:346–52.

    [9] White NJ, Nacheva E, Asimakopoulos FA, Bloxham D, Paul B,Green AR. Deletion of chromosome 20q in myelodysplasia canoccur in a multipotent precursor of both myeloid cells and B cells.Blood 1994;83:2809–16.

    [10] Janssen JW, Buschle M, Layton M, Drexler HG, Lyons J, van denBerghe H, et al. Clonal analysis of myelodysplastic syndromes:evidence of multipotent stem cell origin. Blood 1989;73:248–54.

    [11] Lawrence HJ, Broudy VC, Magenis RE, Olson S, Tomar D, BartonS, et al. Cytogenetic evidence for involvement of B lymphocytes inacquired idiopathic sideroblastic anemias. Blood 1987;70:1003–5.

    [12] Nilsson L, Astrand-Grundstrom I, Arvidsson I, Johansson B,Hokland P, Kjeldsen L, et al. Involvement of the lympho-myeloidstem cells in different cytogenetic subgroups of myelodysplasticsyndromes. Leuk Res 2001;25:S2.

    [13] Kibbelaar RE, van Kamp H, Dreef EJ, de Groot-SwingsG, Kluin-Nelemans JC, Beverstock GC, et al. Combinedimmunophenotyping and DNA in situ hybridization to study lineageinvolvement in patients with myelodysplastic syndromes. Blood1992;79:1823–8.

    [14] Saitoh K, Miura I, Takahashi N, Miura AB. Fluorescence in situhybridization of progenitor cells obtained by fluorescence-activatedcell sorting for the detection of cells affected by chromosomeabnormality trisomy 8 in patients with myelodysplastic syndromes.Blood 1998;92:2886–92.

    [15] van Kamp H, Fibbe WE, Jansen RP, van der Keur M, de GraaffE, Willemze R, et al. Clonal involvement of granulocytes and

    monocytes, but not of T and B lymphocytes and natural killer cellsin patients with myelodysplasia: analysis by X-linked restrictionfragment length polymorphisms and polymerase chain reaction ofthe phosphoglycerate kinase gene. Blood 1992;80:1774–80.

    [16] Kroef MJ, Fibbe WE, Mout R, Jansen RP, Haak HL, Wessels JW, etal. Myeloid but not lymphoid cells carry the 5q deletion: polymerasechain reaction analysis of loss of heterozygosity using mini-repeatsequences on highly purified cell fractions. Blood 1993;81:1849–54.

    [17] Escudier SM, Albitar M, Robertson LE, Andreeff M, PierceS, Kantarjian HM. Acute lymphoblastic leukemia followingpreleukemic syndromes in adults. Leukemia 1996;10:473–7.

    [18] Bonati A, Delia D, Starcich R. Progression of a myelodysplasticsyndrome to pre-B acute lymphoblastic leukaemia with unusualphenotype. Br J Haematol 1986;64:487–91.

    [19] Ascensao JL, Kay NE, Wright JJ, Arthur D, Finkel B, Rydell R,et al. Lymphoblastic transformation of myelodysplastic syndrome.Am J Hematol 1986;22:431–4.

    [20] Inoshita T. Acute lymphoblastic leukemia following myelodysplasticsyndrome. Am J Clin Pathol 1985;84:233–7.

    [21] Heaney ML, Golde DW. Myelodysplasia. N Engl J Med1999;340:1649–60.

    [22] Cotran RS, Kumar V, Collins T. Robbins pathologic basis of disease.6th ed. Philadelphia: Saunders, 1999.

    [23] Kouides PA, Bennett JM. Morphology and classification ofmyelodysplastic syndromes. Hematol Oncol Clin North Am1992;6:485–99.

    [24] Koeffler HP, Golde DW. Human preleukemia. Ann Intern Med1980;93:347–53.

    [25] Fialkow PJ. Clonal origin of human tumors. Biochim Biophys Acta1976;458:283–321.

    [26] Nowell PC. Tumors as clonal proliferation. Virch Arch B CellPathol 1978;29:145–50.

    [27] Raskind WH, Tirumali N, Jacobson R, Singer J, Fialkow PJ.Evidence for a multistep pathogenesis of a myelodysplasticsyndrome. Blood 1984;63:1318–23.

    [28] Raza A. Consilience across evolving dysplasias affecting myeloid,cervical, esophageal, gastric and liver cells: common themes andemerging patterns. Leuk Res 2000;24:63–72.

    [29] Raza A. Myelodysplastic syndromes may have an infectiousetiology. J Toxicol Environ Health A 2000;61:387–90.

    [30] Lichtman MA. Myelodysplasia or myeloneoplasia: thoughts onthe nosology of clonal myeloid diseases. Blood Cells Mol Dis2000;26:572–81.

    [31] Fearon ER, Vogelstein B. A genetic model for colorectaltumorigenesis. Cell 1990;61:759–67.

    [32] Dickstein JI, Vardiman JW. Issues in the pathology and diagnosisof the chronic myeloproliferative disorders and the myelodysplasticsyndromes. Am J Clin Pathol 1993;99:513–25.

    [33] Harrison CN, Gale RE, Machin SJ, Linch DC. A large proportionof patients with a diagnosis of essential thrombocythemia do nothave a clonal disorder and may be at lower risk of thromboticcomplications. Blood 1999;93:417–24.

    [34] Michiels JJ. The myeloproliferative disorders. An historicalappraisal and personal experiences. Leuk Lymphoma 1996;22(Suppl 1):1–14.

    [35] Bennett CL, Weinberg PD, Lieberman JJ. Cancer insurance policiesin Japan and the United States. West J Med 1998;168:17–22.

    [36] Jablonski S. Syndrome—a changing concept. Bull Med Libr Assoc1992;80:323–7.

    [37] Greenberg P, Anderson J, de Witte T, Estey E, Fenaux P, GuptaP, et al. Problematic WHO re-classification of myelodysplasticsyndromes. Members of the International MDS Study Group. J ClinOncol 2000;18:3447–52.

    [38] Head D, Bennett J, Greenberg P, Hamblin T, Hasle H, NiemeyerC, et al. Classification of myelodysplastic syndrome: FAB versusWHO and pediatric MDS. Leuk Res 2001;25:S1.

    [39] Tefferi A, Mathew P, Noel P. The 5q− syndrome: a scientific andclinical update. Leuk Lymphoma 1994;14:375–8.

  • D.P. Steensma, A. Tefferi / Leukemia Research 27 (2003) 95–120 111

    [40] Germing U, Gattermann N, Aivado M, Hildebrandt B, Aul C.Two types of acquired idiopathic sideroblastic anaemia (AISA): atime-tested distinction. Br J Haematol 2000;108:724–8.

    [41] Germing U, Gattermann N, Strupp C, Aivado M, Aul C.Validation of the WHO proposals for a new classification ofprimary myelodysplastic syndromes: a retrospective analysis of1600 patients. Leuk Res 2000;24:983–92.

    [42] Harris NL, Jaffe ES, Diebold J, Flandrin G, Muller-Hermelink HK,Vardiman J. World Health Organization classification of neoplasticdiseases of the hematopoietic and lymphoid tissues: report ofthe Clinical Advisory Committee meeting, Airlie House, Virginia,November 1997. J Clin Oncol 1999;17:3835–49.

    [43] Cassell EJ. The nature of suffering: and the goals of medicine.Oxford: Oxford University Press, 1994.

    [44] Tricot GJ. Minimal diagnostic criteria for the myelodysplasticsyndrome in clinical practice. Leuk Res 1992;16:5–6.

    [45] Ost A, Reizenstein P. Minimal diagnostic criteria for themyelodysplastic syndrome. Leuk Res 1992;16:9–11.

    [46] Culligan DJ, Jacobs A. Minimal diagnostic criteria for themyelodysplastic syndrome. Leuk Res 1992;16:4–5.

    [47] Farhi DC. Bone marrow dysplasia. A continuing diagnosticchallenge. Am J Clin Pathol 1992;98:473–5.

    [48] Ramos F, Fernandez-Ferrero S, Suarez D, Barbon M, RodriguezJA, Gil S, et al. Myelodysplastic syndrome: a search for minimaldiagnostic criteria. Leuk Res 1999;23:283–90.

    [49] Gardais J. Dyshaemopoiesis in adults: a practical classification fordiagnosis and management. Leuk Res 2000;24:641–51.

    [50] Bain BJ. The bone marrow aspirate of healthy subjects. Br JHaematol 1996;94:206–9.

    [51] Champion KM, Gilbert JG, Asimakopoulos FA, HinshelwoodS, Green AR. Clonal haemopoiesis in normal elderly women:implications for the myeloproliferative disorders and myelody-splastic syndromes. Br J Haematol 1997;97:920–6.

    [52] Kyle RA, Therneau TM, Rajkumar SV, Offord JR, LarsonDR, Plevak MF, et al. A long-term study of prognosis inmonoclonal gammopathy of undetermined significance. N Engl JMed 2002;346:564–9.

    [53] Imbert M, Scoazec JY, Mary JY, Jouzult H, Rochant H, SultanC. Adult patients presenting with pancytopenia: a reappraisal ofunderlying pathology and diagnostic procedures in 213 cases.Hematol Pathol 1989;3:159–67.

    [54] Casagrande G, Michot F. Alcohol-induced bone marrow damage:status before and after a 4-week period of abstinence from alcoholwith or without disulfiram. A randomized bone marrow study inalcohol-dependent individuals. Blut 1989;59:231–6.

    [55] Rezuke WN, Anderson C, Pastuszak WT, Conway SR, Firshein SI.Arsenic intoxication presenting as a myelodysplastic syndrome: acase report. Am J Hematol 1991;36:291–3.

    [56] Karcher DS, Frost AR. The bone marrow in humanimmunodeficiency virus (HIV)-related disease. Morphology andclinical correlation. Am J Clin Pathol 1991;95:63–71.

    [57] Baurmann H, Schwarz TF, Oertel J, Serke S, Roggendorf M, HuhnD. Acute parvovirus B19 infection mimicking myelodysplasticsyndrome of the bone marrow. Ann Hematol 1992;64:43–5.

    [58] Hasle H, Heegaard E, Kerndrup G, Jensen IM, Peterslund NA,Hornsleth A. Parvovirus B19 infection infrequently involved inchildren and adults with myelodysplastic syndrome. Leuk Res1996;20:81–3.

    [59] Hofmann WK, Ottmann OG, Ganser A, Hoelzer D. Myelodysplasticsyndromes: clinical features. Semin Hematol 1996;33:177–85.

    [60] Kampmeier P, Anastasi J, Vardiman JW. Issues in the pathologyof the myelodysplastic syndromes. Hematol Oncol Clin North Am1992;6:501–22.

    [61] Bain BJ. The relationship between the myelodysplasticsyndromes and the myeloproliferative disorders. Leuk Lymphoma1999;34:443–9.

    [62] Neuwirtova R, Mocikova K, Musilova J, Jelinek J, Havlicek F,Michalova K, et al. Mixed myelodysplastic and myeloproliferativesyndromes. Leuk Res 1996;20:717–26.

    [63] Tuzuner N, Cox C, Rowe JM, Watrous D, Bennett JM. Hypocellularmyelodysplastic syndromes (MDS): new proposals. Br J Haematol1995;91:612–7.

    [64] Barrett J. Myelodysplastic syndrome and aplastic anemia—diagnostic and conceptual uncertainties. Leuk Res 2000;24:595–6.

    [65] Barrett J, Saunthararajah Y, Molldrem J. Myelodysplastic syndromeand aplastic anemia: distinct entities or diseases linked by a commonpathophysiology. Semin Hematol 2000;37:15–29.

    [66] Appelbaum FR, Barrall J, Storb R, Ramberg R, Doney K, Sale GE,et al. Clonal cytogenetic abnormalities in patients with otherwisetypical aplastic anemia. Exp Hematol 1987;15:1134–9.

    [67] Graham DL, Gastineau DA. Paroxysmal nocturnal hemoglobinuriaas a marker for clonal myelopathy. Am J Med 1992;93:671–4.

    [68] Cermak J, Marinov Y, Jelinek J. CD59 and CD55 deficiency inpatients with primary myelodysplastic syndrome: an analysis of 35patients. Leuk Res 2001;25:S11–2.

    [69] Okada M, Okamoto T, Takatsuka H, Wada H, Fujimori Y,Kakihsita E. Relationship between immunological abnormalities andparoxysmal nocturnal hemoglobinuria