Acute Myeloid Leukemia: The Challenge of Capturing Disease Variety

Fifty years ago when the American Society of Hematology was founded, the diagnosis of acute myeloid leukemia (AML) was based solely on microscopic evaluation. There was no generally accepted treatment. In those days three classes of experimental agents were being introduced into treatment according a new concept of combination chemotherapy: corticosteroids, folic acid antagonists (methotrexate) and purine antagonists (6-mercaptopurine). Patients showing a response to treatment had somewhat better survival than nonresponders but there were no long-term survivors.^1,2 Since then more active antileukemic agents have been introduced into modern combination chemotherapy and brought cure to a significant fraction of patients. Initially AML was considered one monolithic disease for which “one size fits all” chemotherapy was applied. This is no longer true. AML is a genetically pleiomorphic disease. In 1958 the same cytotoxic treatment was applied to what was morphologically considered to be AML or acute lymphoblastic leukemia (ALL). In 2008 therapeutic and diagnostic developments have shifted the paradigm of AML therapeutics and created a perspective of personalized therapeutics. This review will highlight recent developments in AML that impact on diagnosis, prognosis assessment and treatment choice (Table 1 ).

Cytogenetic analysis has become part of the standard diagnostic approach of a patient suspected of AML. This allows the identification of (cyto)genetic entities that deserve targeted treatments (examples: acute promyelocytic leukemia and BCR-ABL–positive leukemia). It also allows the distinction of specific subtypes of disease with widely different prognosis (illustrating example: AML t(8;21)(q22;q22) with favorable risk versus AML abn3q26 with adverse risk) that deserve risk-guided treatment strategies. Numerous genetic abnormalities that escape cytogenetic detection (e.g., gene mutations, gene expression abnormalities) have more recently been discovered. These have allowed the further dissection of AML into molecular subtypes with distinctive prognosis (Table 2 ). Clinically favorable genotypes of AML, for example, involve mutations in the genes of the transcription factor CEBPA (CCAAT enhancer binding factor alpha)^3,–5 or nucleophosmin-1 (NPM1),^6,–13 whereas unfavorable genotypes may include those with partial tandem duplications of the MLL gene (MLL-PTD),^14,–17 internal tandem duplications of the gene of fms-like tyrosine kinase 3 (FLT3-ITDs)^18,–24 and mutations in Wilms’ tumor 1 gene (WT1).^25,26 These findings emerge from a field of intense scientific activity. There is little question that more genetic markers with clinical value will be discovered. These genetic events perturb diverse cellular pathways and functions, and they often confer a profound impact upon the clinical phenotype of the disease and treatment response.

Besides gene mutations aberrant levels of expression of particular genes may also be informative as regards prognosis. Examples of aberrant gene expression with adverse prognostic significance include the overexpression of Ecotropic Viral Integration-1 (EVI-1),^27,28,Brain and Leukemia Cytoplasmic Gene (BAALC),^29,30,meningioma-1 gene (MN1)³¹ and Ets-related gene (ERG)^32,33 (Table 2 ).

EVI-1–positive AML represents a genotype with a particularly adverse prognosis as has been demonstrated by Delwel and coworkers.^27,28 For quite some time the AML subtype with 3q26 chromosomal lesions has been recognized as belonging to the most aggressive forms of human AML.^27,34 The oncogene ecotropic viral integration site-1 (EVI-1) is involved in the cytogenetic translocations of 3q26.^27,35 AML with abn3q26 represent only about 1% of AML.³⁴ However, there are another 5% to 10% of AML that overexpress the oncogene EVI-1 (EVI-1 positive) in the absence of the abn3q26 anomaly.²⁷ In the EVI-1–positive subset CR rates are low, relapse rates are high and survival is low. Besides the common EVI1-1D splice form, four other splice variants of EVI1 were recently identified: EVI1-1A, -1B, -1C, and -3L5, which mainly differ in their 5′ untranslated regions. In addition, AML cells may also express MDS1/EVI1 (ME), an EVI1 fusion variant generated through intergenic splicing with MDS1, a gene with a currently unknown function that is located ~140 kb upstream of EVI1.²⁸ A recent study has addressed the prevalence and the prognostic value of each of the splice variants of EVI1 as well as those of the intergenic splice form MDS1/EVI1 (ME) by quantitative RT-PCR.²⁸ Among 534 AML patients 41 EVI1⁺ cases were identified. High EVI1 expression levels predicted a significantly reduced event-free survival (EFS) (3% vs 29%) and overall survival (OS) (13% vs 39% at 5 years) (P < .001). Multivariate analysis revealed that high EVI1 levels are an independent predictor of inferior EFS (HR = 1.9, P = .002) and DFS (HR = 2.1, P = .006). Seventeen of the EVI1⁺ leukemias were ME-negative, i.e., EVI-1^pos/ME^neg. Several of the latter EVI-1^po2/ME^neg AMLs with a dissociated high EVI1 expression but low MDS1/ EVI1 expression had cryptic cytogenetic 3q26 aberrations.²⁸ These abn3q26 abnormalities apparently may quite easily escape cytogenetic identification but they can be properly documented by FISH. The AML with the EVI-1^pos/ME^neg genotype represent one of the most unfavorable forms of AML. For the time being it remains elusive why AML with high EVI-1 expression represent such a bad prognostic subtype of AML that does not respond to currently available treatments. Exploring the mechanisms of transformation and treatment resistance by the EVI1 protein remains a major challenge of current research. EVI1 encodes a nuclear protein, which interacts with several proteins important in transcriptional control, e.g., CtBp1, HDAC, SMAD3, P/CAF and GATA1. Investigations into how these interactions take place may provide the necessary insights for developing tools to therapeutically target EVI1⁺ leukemia cells.

Particular cytogenetic abnormalities have been implicated to confer a poor prognosis in AML. These include −5, −7, del(5q), del(7q), abn3q, abn11q23, abn17p, t(6;9) t(9;22). However these abnormalities not infrequently are seen together or in association with other abnormalities. The negative prognostic value of coexisting multiple clonal cyto-genetic abnormalities, often referred to as “complex karyotypes,” is widely accepted. Commonly applied definitions of AML with complex cytogenetics are based on the summing up score of any type of abnormality. In a recent study in adults up to 60 years of age,³⁶ the value of each of the different sorts of cytogenetic abnormalities as determinants of poor prognosis has been assessed, such as loss of one or more chromosomes, extra copy of one or more chromosomes and structural cytogenetic abnormalities. The presence of autosomal chromosomal monosomies strongly predicted for an adverse prognosis. While the negative prognostic impact of autosomal monosomies in AML had been described for monosomies of chromosomes 5 and 7, it was demonstrated in multiple comparisons that there is nothing special about loss of chromosomes -5 or -7 as regards prognostic significance. Any type of autosomal monosomy in AML is associated, in fact, with a dismal outcome.³⁶ On the other hand, in direct comparisons extra copies of 1 or more chromosomes did not appear to exert an additive negative effect on prognosis in AML. Trisomies, tetrasomies or the presence of ring or marker chromosomes and structural chromosomal aberrations appeared to contribute minimal prognostic impact when they were considered in relationship to monosomal abnormalities. The negative prognostic influence of multiple, i.e., 2 or more, autosomal mono-somies or 1 autosomal monosomy in combination with at least 1 structural chromosomal abnormality in AML was profound (4-year OS: 4%). These findings have led to the proposal of the novel monosomal karyotype index for prognostically highly unfavorable AML with multiple cytogenetic abnormalities.^36,The “monosomal karyotype” index defines AML with a highly adverse prognosis as a karyotype with at least two autosomal monosomies or one single autosomal monosomy in the presence of one or more structural cytogenetic abnormalities. In a head-to-head comparison with the traditional complex karyotype, the simplified “monosomal karyotype” index is significantly superior. This is true into two ways: On one hand, AMLs with a complex karyotype but not fulfilling the criteria of monosomal karyotype have a better prognosis that is similar to that of cytogenetically abnormal AML in general. On the other hand, AMLs without a complex karyotype but with a monosomal karyotype have an extremely poor outcome. The better predictability of very unfavorable risk AML by the monosomal karyotype in comparison to “complex karyotypes” holds up regardless whether complexity is defined by ≥ 3 or ≥ 5 clonal cytogenetic abnormalities.³⁶ 

In patients 60 years old and younger the monosomal karyotype enables (in addition to normal cytogenetics and core binding factor leukemias) the prognostic classification of two new aggregates of cytogenetically abnormal AML: abnormal cytogenetics (4-year OS: 26% ± 2) and the highly adverse risk monosomal karyotype categories (4-year OS: 4% ± 1).³⁶ 

The great variation of genomic aberrations in AML have opened a search for genome wide analysis that provides a snap shot of the complex genetic signature of a particular leukemia. Gene expression profiling (GEP), microRNA analysis, methylation profiling and whole genome sequence analysis offer powerful technologies that create opportunities for characterizing an individual’s leukemia in a genome-wide manner in diverse terms (gene expression, microRNA expression, gene mutations and epigenetics).

GEP is a technology in which DNA microarrays containing cDNAs or oligonucleotide probes are used to simultaneously measure levels of thousands of mRNA transcripts. A straightforward way of using GEP is to compare expression profiles among cases of AML and to examine similarities and differences. Thus, particular cytogenetic AML subtypes—e.g., AMLs with t(8;21), t(15;17) and inv(16)—can each be recognized by highly distinctive GEP profiles among a mixed series of AML.^37,-43 Likewise, AML genotypes with mutations in CCAAT/enhancer binding protein alpha (CEBPA), and mutations in nucleophosmin 1 (NPM1) show distinctive gene expression signatures.^6,42 The apparent fact that the genetic abnormality places such a prominent stamp on the transcriptome of the leukemia has important implications. It implies that a particular expression profile that becomes apparent following gene expression profiling may reflect an underlying genetic abnormality for that particular subset of AML. Thus, a notable gene expression pattern may offer the key to the disclosure of new genotypes. Evidence indicating that a distinctive gene expression AML subtype may guide the way to the discovery of biologically and clinically relevant genotypes is highlighted by the example of the recent identification of a unique form of AML characterized by epigenetic CEBPA silencing and an immature myeloid/T-lymphoid biphenotype.⁴⁴ In fact, these acute leukemias with silenced CEBPA were found to carry expression profiles similar to AML cases with CEBPA mutations, but such mutations were not present. Subsequent experiments elucidated that in the novel subgroup CEBPA expression was silenced, often through promoter hypermethylation. Mouse modeling subsequently demonstrated that knock down of CEBPA expression evokes the expression of T-cell genes in immature hematopoietic progenitor cells, thus furnishing a mechanistic insight into why these leukemias carry both myeloid and T-lymphoid features. In this example, the analysis of human AML GEP data in combination with pathway analysis and the use of informative mouse models uncovered the pathobiology of a subgroup of leukemia that is defined by a myeloid/T-lymphoid phenotype, CEBPA silencing and, in fact, also frequent NOTCH1 mutations.⁴⁴ 

Touw and colleagues have generated a series of myeloid leukemias in mice through retroviral insertion mutagenesis to search for critical suppressor genes involved in leukemogenesis. A direct comparison between the murine integration sites with a clinical GEP AML data set revealed that indeed mouse leukemia genes are frequently deregulated in human AML.⁴⁵ The comparisons of human AML GEP data with high-throughput murine data from dedicated experimental mouse models provide opportunities for identifying candidate suppressor genes (or oncogenes) of AML. Techniques such as chromatin immuno-precipitation on DNA microarray chips (ChIP-chip), ChIP-sequencing and RNA interference libraries may also be used in combination with the analysis of gene expression profiling and uncover target genes.

MicroRNAs are small non-coding RNAs of 19-25 nucleotides that play a role in transcriptional or post-transcriptional regulation of genes involved in numerous biological processes, including differentiation and proliferation. Three recent studies have found characteristic microRNA expression profiles in cytogenetically and molecularly defined subtypes of AML.^46,-48 It has been demonstrated that AML can be classified according their variable and distinctive microRNA expression profiles.⁴⁸ Analysis (significance analysis of microarrays [SAM]) revealed individual differentially expressed microRNAs that characterize subtypes of AML, e.g., AML with t(8;21), inv(16), t(15;17) and AML with NPM1 and CEBPA mutations. Subsequently these subtypes of AML can be defined and classified according limited sets of selected microRNAs.

For instance, in the AML genotype with NPM1-mutations (NPM1^mut) which is present in approximately 35% of patients with AML, the microRNAs -10a, -10b, -196a, -196b are highly expressed.⁴⁸ AML with NPM1 mutations are not only characterized by typical microRNA patterns but also by specific mRNA gene expression signatures as had previously been shown.^6,49 The mRNA signatures are characterized by overexpression of homeobox genes (HOXA, HOXB families and TALE genes). Interestingly, the microRNAs -10a, -10b, -196a, -196b that are upregulated in AML with mutant NPM1, are all located within the ge-nomic cluster of HOX genes. Thus, these observations appear consistent with an aberrant regulatory network involving NPM1, HOX genes and microRNAs, which might be engaged in the arrest of cellular differentiation of hematopoietic progenitors and development of AML with mutant NPM1.

Marcucci et al⁵⁰ have reported data to suggest that microRNA profiling can have prognostic value in cytogenetically normal AML. The investigators derived a set of 12 microRNAs that can distinguish two subgroups of cytogenetically normal AML with EFS probabilities of 11% (bad risk) and 36% (intermediate risk). The profile was developed for the subset of AML with the genotypes FLT3-ITD^pos and FLT3-ITD^neg/NPM1^wt, which account for two-thirds of all cases of cytogenetically normal AML. The microRNA profile appeared independent of the prognostic effect of FLT3-ITD abnormalities. The survival estimates in this study, which were based on relatively small numbers of cases, require confirmation. These recent studies underscore the importance of microRNAs in the pathogenesis of AML and their potential usefulness in future diagnostics.

Changes in expression levels of critical genes may be due to small DNA amplifications or deletions undetected by conventional cytogenetics. Small chromosomal aberrations can be examined with single nucleotide polymorphism (SNP) arrays and array-based comparative genomic hybridization (CGH). A study of 60 patients with AML with complex karyotypes using array-CGH (bacterial artificial and P1-derived artificial chromosome clones) identified several recurrent lesions.⁵¹ It is not yet clear what the overall frequency and distribution of these genomic alterations in AML is. Copy-number neutral loss of heterozygosity through segmental uniparental disomy in AML is a relatively common and notable finding in AML.^52,–54 This phenomenon can point to the identification of homozygous mutations or deletions of leukemia-related genes, including FLT3 and CEBPA.^54,55 

Gene expression changes may be caused by aberrant methylation or mutations in yet-unknown genes. Whole genome sequencing is emerging as a method to address this issue on a global scale.⁵⁶ A challenge in this regard will be distinguishing functionally relevant mutations from the abundant so-called passenger mutations—unimportant genetic changes caused by genomic instability of cancer cells—that will be picked up at the same time. Studies on tyrosine kinase abnormalities have underscored the need for validation of the biological effects of novel mutations identified by high-throughput nucleotide sequencing.^57,58 

After a remission has been attained the next therapeutic objective is to eradicate any residual leukemia (minimal residual disease) by applying additional chemotherapy, autologous stem cell transplantation or allogeneic stem cell transplantation. Allogeneic stem cell transplantation is the most effective antileukemic modality. But an allogeneic stem cell transplantation for whom? The reason to turn to allogeneic transplantation is that the immune-mediated graft-versus-leukemia effect of the transplanted cells reduces the risk of relapse considerably and improves relapse-free survival. The alloSCT advantage obviously has to be cautiously balanced against the increased risk of death and morbidity that is typically connected with alloSCT. Excess mortality (ranging between 10% and 40%) and morbidity due to transplant-related complications, such as infection and graft-versus-host disease, can abrogate all of the benefit of a reduced risk of relapse.⁵⁹ For this reason, allogeneic stem cell transplantation is usually avoided in a type of AML that has a pattern of cytogenetics with a relatively favorable prognosis, such as AML with the chromosomal translocations t(8;21) or inv(16)/t(16;16).^59,60 In the latter subtypes the risk of relapse is in the order of 35% to 40% or less. In a recent study two low-risk genotypically defined subsets within the large category of cytogenetically normal AML were investigated,⁶⁰ each with a risk of relapse of about 35%. The first genotype was defined by the presence of ‘favorable’ mutations in NPM1 and the absence of concurrent ‘unfavorable’ FLT3-internal tandem duplications (NPM1^mut /FLT3-ITD^neg).⁶¹ This genotype accounts for approximately 16% of all newly diagnosed patients younger than 60 years old. There was no demonstrable benefit from transplantation in patients with NPM1^mut/FLT3-ITD^neg AML. The second subset of AML with ‘favorable’ mutations in the transcription factor gene CEBPA (CEBPA^mut) could not be analyzed in this way because of a lack of statistical power due to a limited number of cases. The latter low-risk subtype CEBPA^mut accounts for 8% of all AML. Nevertheless, the available body of evidence suggests that these AMLs are unlikely to profit from an alloSCT. Thus, by and large patients with AML with t(8;21), AML with inv(16)/t(16;16), AML with NPM1^mut/ FLT3-ITD^neg and AML with CEBPA^mut are not considered for alloSCT.^59,61 By contrast, a transplant would be more attractive in a patient whose leukemia cells bear a cytogenetic or molecular abnormality that predicts a high risk of relapse after chemotherapy.⁵⁹ Obviously, in the near future these results will also need to be considered more specifically in the light of the extended scale of allogeneic stem cell transplantation strategies with respect to reduced-intensity conditioning regimens and in relationship to different transplant sources and donor types (matched unrelated and haploidentical donors, umbilical stem cell grafts).⁶² 

As discussed above, cytogenetics as well as various genomic markers (gene mutations, gene overexpression) may provide input for algorithms for remission induction and post-remission treatment decisions. At the same time it remains appropriate to realize that prognostic factors in fact remain a moving target and they are only relevant to therapies available at a given time. Algorithms that provide a basis for risk-adapted therapeutic choices may include cytogenetic factors, molecular markers as well as clinical parameters (e.g., age, attainment of an early or late complete remission) and hematological determinants (e.g., secondary AML, white blood cell count at diagnosis). The Dutch-Belgian Leukemia Cooperative Group (HOVON) and the Swiss Leukemia Cooperative Group (SAKK) have recently introduced a prognostic stratification score for newly diagnosed AML in adults less than 60 years of age that distinguishes four prognostic categories with different outcomes. The risk classification has been derived from (a) an analysis of the data of 1975 patients from the previous HOVON/ SAKK AML studies for patients up to 60 years of age, registered before January 1, 2004 and with successful cytogenetic analysis, and (b) an analysis of the data of a subset of 424 patients for whom marker information and microarray expression data were also available. The risk stratification system for prognosis is used as an algorithm for assigning patients in complete remission (CR) depending on their relapse risk to allogeneic stem cell transplantation (Table 3 ).

As soon as the definitive diagnosis of AML has been established, the treatment begins with a remission induction phase aimed at establishing a CR. The anthracyclines (daunorubicin, idarubicin) and cytarabine have remained the cornerstones of modern remission induction therapy for adult AML since their introduction. They induce CR in an average of 70% to 80% of adults aged less than 60 years enrolled in clinical trials and on average in 50% of patients older than 60. Continuous efforts are being made to improve the efficacy of remission induction treatment. Improved induction therapy should yield more and better quality CRs (i.e., CRs of longer duration). The responsiveness to therapy depends mainly on the genotype of the disease and age of the patient. Thus, for instance, leukemias with a monosomal karyotype or with EVI1 oncogene overexpression (see above) and patients of older age with AML respond very poorly to treatment. Is it possible to target particular genotypes of AML with specific drugs and advance the therapeutic possibilities for these subtypes?

Activating mutations in KIT (KIT^mut), the receptor of the hematopoietic growth factor SCF (stem cell factor) are predominantly seen in AML with core binding factor abnormalities (CBF) t8;21 and inv(16)/t(16;16). Some studies have suggested that AML with CBF and KIT^mut have a prognosis that is less than that of the other AMLs with CBF).^63,–65 The KIT gene encodes for a tyrosine kinase receptor that in case of the mutation is constitutively activated. Kinase inhibitors (e.g., dasatinib) that abrogate KIT signaling are currently in early clinical trial and await an evaluation as regards their possible therapeutic effect. Fms-like tyrosine kinase 3 or FLT3 are membrane kinase receptors expressed on hematopoietic progenitor cells that are stimulated following ligand binding. Mutations may disrupt the function and lead to the constitutive activation of the receptor. The most common kinase receptor mutations in AML are internal tandem duplications in fms-like tyrosine kinase 3 or FLT3 (apparent in approximately 25% of AML), which express a negative effect on prognosis (Table 2 ). The other class of so-called FLT3-TKD mutations (i.e., point mutations in the tyrosine kinase domain of the receptor) seen in 8% of cases has controversial prognostic significance. Various small molecule FLT3 kinase inhibitors or anti-FLT3 blocking antibodies are in Phase I–III clinical trial. These include sorafenib, lestaurtinib and staurosporin and various others.^66,–69 These agents have produced responses in AML with FLT3 gene mutations as well as in AML with wild-type FLT3. The KIT and FLT3 mutations are generally held to have been acquired as late events during leukemogenesis. The question as to whether the KIT and FLT3 inhibitors in combination with chemotherapy will add significant therapeutic benefit is currently addressed in various currently ongoing phase III trials. Although many of the new agents appear active in vitro and have generated much excitement, the clinical activity for the majority of them has as yet to be demonstrated.

Mechanisms determining chemoresistance may not only reside in the leukemic cells themselves but may also depend on an interaction between the leukemia and the external environment. The microenvironment in which the AML cells seed and develop may influence the pathobiology and the treatment response of the disease. AML infiltrations are associated with increased angiogenesis, as is apparent from the increased microvessel density. Angiogenesis in AML may be promoted by various provascular factors, including vascular endothelial growth factor (VEGF), and basic fibroblast growth factor. VEGF may be produced by the AML blasts, and VEGF-receptors (VEGF-R) may be expressed by the AML cells themselves as well as by endothelial cells, thus enabling an autocrine and paracrine stimulatory circuit of vascular growth in leukemic areas (prosurvival effects). VEGF-C production by stromal cells, via interaction with VEGF receptor-3, has been shown to protect AML cells from chemotherapy-induced apoptosis.⁷⁰ Considering these data one might postulate that targeting VEGF or VEGF-R could offer a therapeutic target in AML.⁷¹ Interruption of these circuits, e.g., by antiVEGF-R antibodies or appropriate kinase blockers (peptides) is currently in study as a therapeutic strategy of overcoming leukemic chemoresistance.⁷² Chemotherapy resistance of AML may also be provoked by adherence of the leukemic cells to the stromal environment. Leukemic cells settling in the niches of the microenvironment are relatively protected against the cytotoxic effects of chemotherapy. Adhesion is accomplished through leukemic progenitor cell–stromal cell interactions via specific receptors and adhesion molecules. One such interaction is known as the CXCR-4-SDF1α (stroma-derived factor 1α) axis. SDF1α is produced by marrow stroma and activates its cognate receptor CXCR4. CXCR4 has been implicated in chemotaxis, homing, survival/apoptosis of hematopoietic cells. AMLs with high CXCR-4 cell surface expression (e.g., on the AML progenitor population) and a presumed high tendency of stromal protection have been shown to have a poor prognosis.^73,74 CXCR-4 inhibitors that target the stromal interaction and release the leukemic cells from the microenvironment are currently in clinical trial. These inhibitors have both a mobilizing and cell-cycle activating effect upon the AML cells and may sensitize the AML for chemotherapeutic cell killing.^75,76 The effects of CXCR-4 inhibition remarkably resemble AML cell stimulation by granulocyte colony-stimulating factor (G-CSF) and granulocyte-macrophage CSF (GM-CSF). Following in vivo stimulation with G-CSF or GM-CSF leukemia cells are also mobilized to the blood and also become activated. The administration of G-CSF or GM-CSF concurrently with chemotherapy, commonly referred to as CSF priming, has also been shown to sensitize for cell killing by chemotherapy in vitro. A clinical benefit of CSF priming has been demonstrated in two large randomized studies where a reduced relapse probability and improved disease-free survival have been reported.^77,78 The positive effects on outcome were particularly apparent in intermediate-risk cytogenetic types of AML. Some other published studies, in particular in AML at older age, have failed to show enhanced therapeutic efficacy of CSF priming.⁷⁹ The reasons for these discrepant results for the time being has not been resolved but information of the genetic leukemia diversity has not been available in those studies. Interestingly, synergy between the activities of G-CSF stimulation and CXCR-4 inhibition has also been reported.⁸⁰ Future studies that make use of CXCR-4 inhibition and take the level of cellular CXCR-4 expression phenotype into account, and studies using the combination of G-CSF stimulation and CXCR-4 inhibition in combination with antileukemic chemotherapy appear warranted.

In March, 1958 Frei et al submitted one of the first comparative studies in the chemotherapy of malignant disease.¹ The study dealt with patients with AML and ALL. They concluded from their studies that, “survival time from onset of treatment was greater in patients who attained remissions than in those who did not . . . . The combination of antimetabolites used in this study could exercise its optimum effect only on patients with leukemic cells sensitive to both drugs.” In 1960 F.G.J. Hayhoe wrote about acute leukemia in his renowned monograph “Leukemia, Research and Clinical Practice”: “While comparisons . . . suggest that an increase in longevity of some few months is the most that has been achieved by the advances in treatment of the past 10 years, they are in a sense misleading, and are not helpful in determining prognosis in individual patients . . . .”² These observations and statements half a century ago already pointed to the need of a better understanding of patient variability and disease heterogeneity for treatment development in AML and drew attention to the necessity for the development of drugs that would overcome resistance to therapy. The occasion of the celebration of a half century of ASH may make us especially aware of the remarkable progress as regards the diagnosis, prognosis and treatment of AML that the last 50 years have witnessed. Nevertheless, there is no question that the challenges of therapeutics in AML today remain considerable. In 2008 most of our patients (primary refractory leukemias, recurrences after initial complete remission, leukemia at older age) continue to fail treatment. At the same time the intense scientific activity at multiple frontlines holds promise for a continuing positive trend in developmental therapeutics in AML.

Acknowledgments

This manuscript alludes to the results of studies from many investigators around the world. I apologize for not having been able to reference all individual publications because of restricted space. The research in Rotterdam has been the product of a team effort at Erasmus University Medical Center/Daniel den Hoed. I am grateful to my colleagues in Rotterdam for an inspiring cooperation and for their important contributions to the field. As regards this manuscript I owe special thanks to Ivo Touw, Ruud Delwel, Peter Valk, Marieke von Lindern, Pieter Sonneveld, Jan Cornelissen, Bas Wouters, Sanne Lugthart, Mojca Lavrencic-Jongen, Dimitri Breems, Wim van Putten, and the clinical collaborators in Holland, Belgium, and Switzerland in the leukemia studies of the Dutch-Belgian HOVON and Swiss SAKK cooperative trial groups. The research has been supported by The Queen Wilhelmina Foundation (Dutch Cancer Society) and ZonMw, the Netherlands Organisation for Health Research and Development.