Molecular characterization of acute myeloid leukemia

Acute myeloid leukemia (AML) is a genetically heterogeneous clonal disorder characterized by the accumulation of acquired somatic genetic alterations in hematopoietic progenitor cells that alter normal mechanisms of self-renewal, proliferation and differentiation.1 Non-random clonal chromosome aberrations (i.e., balanced translocations, inversions, deletions, monosomies, and trisomies) are detectable in the leukemic blasts of approximately 55% of adults with AML. These chromosome changes have contributed to the classification of the disease and in the past they have been recognized as the most important prognostic factor for achievement of complete remission, risk of relapse, and overall survival.2,3 In recent years, a number of gene mutations as well as deregulated expression of genes have been identified, illustrating the enormous heterogeneity of cytogenetically defined AML subsets, in particular in the large subset of AML exhibiting a normal karyotype (Table 1).4,5 The characterization of the gene mutations has provided insights into the mechanisms of leukemogenesis. From a clinical point of view there are two important aspects. First, some of these gene mutations have emerged as important prognostic and predictive markers. Second, novel therapies are now being developed that target these molecular changes. Based on these findings, it is now recommended to include molecular genetic diagnosis of several of these markers in the initial work-up of a patient with newly diagnosed AML. The importance of a comprehensive genetic diagnostic work-up is highlighted in a study by Lo-Coco et al. from the Gruppo Italiano Malattie Ematologiche dell’Adulto (GIMEMA) which appears in this issue of the journal.

Multi-step leukemogenesis

There are several lines of evidence showing that different genetic changes cooperate in leukemogenesis.1,6 First, data from murine models suggest that a single mutation is not sufficient to cause AML. For example, the RUNX-RUNX1T1 and CBFB-MYH11 fusion genes, resulting from t(8;21) and inv(16)t(16;16) respectively, impair myeloid differentiation in these models but they do not cause an overt leukemic phenotype. Second, rare germline mutations have been described in RUNX1 and CEBPA that predispose affected individuals to the development of AML. Constitutional heterozygous loss-of-function mutations in the transcription factor RUNX1 have been associated with familial platelet disorder with propensity to AML. In these individuals, overt leukemia likely develops upon the somatic acquisition of further mutations in hematopoietic progenitor cells. Finally, evidence comes from human disease, since in the majority of AML cases more than one genetic change can be detected (Figures 1 and 2).

There are different types of mutations which appear to fall into broadly defined complementation groups and cooperate in leukemogenesis.6 One group (class I) comprises mutations which activate signal transduction pathways resulting in enhanced proliferation and/or survival of leukemic progenitor cells such as mutations leading to activation of the receptor tyrosine kinase FLT3 or the RAS signaling pathway. The second complementation group (class II) comprises mutations that affect transcription factors or components of the transcriptional co-activation complex, resulting in impaired differentiation and/or aberrant acquisition of self renewal properties by hematopoietic progenitors. Prominent examples are the recurring gene fusions resulting from t(8;21), inv(16)/t(16;16), t(15;17), as well as mutations in CEBPA, MLL, and possibly also NPM1.

Molecular alterations in cytogenetically normal acute myeloid leukemia

In the last decade, somatically acquired mutations have been identified in several genes in cytogenetically normal (CN) AML: the nucleophosmin 1 (NPM1) gene, the fms-related tyrosine kinase 3 (FLT3) gene, the CCAAT/enhancer binding protein alpha (CEBPA) gene, the myeloid-lymphoid or mixed-lineage leukemia (MLL) gene, the neuroblastoma RAS viral oncogene homolog (NRAS) gene, the Wilms tumor 1 (WT1) gene, and the runt-related transcription factor 1 (RUNX1) gene.4,5 These gene mutations are most prevalent in CN-AML; however, they also occur in AML with abnormal karyotypes (Figure 1).

Table 1.Biological and clinical features associated with gene mutations in acute myeloid leukemia.

Mutations in NPM1

In 2005, Falini and colleagues made the important discovery that in a substantial proportion of AML cases, there is abnormal cytoplasmic localization of the NPM1 protein.7 This mislocalization is caused by mutations in exon 12 of the gene which result in loss of tryptophan residues normally required for NPM1 binding to the nucleoli and in the generation of an additional nuclear export signal motif at the C-terminus. NPM1 mutations are the most frequent genetic alteration in adult AML. They are found in 45–64% of CN-AML. NPM1 mutations occur less frequently in pediatric AML (2–8%).4,5,8

NPM1 mutations have been associated with several presenting clinical and biological features (Table 1). In univariate analysis, data on the prognostic impact of NPM1 mutations have been somewhat controversial with some studies showing a significant effect on complete remission (CR) rate (as in the study by Lo-Coco et al. in this issue of the journal), relapse-free survival (RFS), and event-free survival (EFS), while other studies did not reveal significant differences in these parameters. NPM1 mutations cooperate with other gene mutations in leukemogenesis (Figure 2). Approximately 40% of patients with NPM1 mutations also carry FLT3 internal tandem duplications (FLT3-ITD), and several studies unanimously showed that the genotype mutant NPM1 without FLT3-ITD represents a favorable prognostic marker.4,5,8 Figure 3 shows just such favorable results from the largest series of patients published by the German-Austrian AML Study Group (AMLSG).8 The data from this study also suggested that patients with the genotype mutant NPM1 without FLT3-ITD may be exempted from allogeneic hematopoietic stem cell transplantation during the first complete remission, since the outcome following allogeneic transplantation was not superior to that after conventional consolidation chemotherapy.

Mutations in FLT3

In AML, somatic mutations resulting in constitutive activation of FLT3 have been identified in two functional domains of the receptor, the juxtamembrane (JM) domain and the tyrosine kinase domain (TKD). Physiologically, the JM domain is essential for kinase autoinhibition. Disruption of this domain by ITDs of various sizes and insertion sites is detectable in 28–34% of CN-AML, whereas JM point mutations are rare.4,8 Mutations in the TKD mostly affect the activation loop in the carboxy-terminal lobe. These point mutations, small insertions, or deletions mainly involve codons 835 and 836 in 11%–14% of CN-AML.4,8 Point mutations or insertions located at other codons in the TKD are rare. Based on in vitro and in vivo studies there are similarities but also important differences in signal transduction properties between FLT3-ITDs and FLT3 TKD mutations which may explain differences in clinical phenotypes.

From a clinical perspective, FLT3 mutations are of major relevance because of their prognostic impact and because constitutively active FLT3 is an attractive target for molecular therapy. Numerous studies have shown that CN-AMLs harboring FLT3-ITD have a significantly inferior outcome compared to patients without FLT3-ITD. More recent data provide evidence that outcome is also related to the level of the mutant allele, and not just its mere presence. First evidence came from a study by Whitman et al. (2001) showing that only patients with FLT3-ITD and loss of wild-type FLT3, but not patients with heterozygous FLT3-ITD had inferior outcome.9 Thiede et al. (2002) used DNA fragment analysis to quantify the relative level of the mutant allele and were able to define a cut-off value that distinguished between prognostic subgroups.10 The prognostic relevance of FLT3 TKD mutations remains controversial.4,5 While a previous meta-analysis on more than 1,000 unselected AML cases (including 84 cases with FLT3 TKD mutations) showed a negative prognostic impact, another study even suggested a positive impact of TKD mutations on survival.

At present, several FLT3 inhibitors, including midostaurin (PKC412), lestaurtinib (CEP-701), sorafenib (BAY 43-9006), and sunitinib (SU11248), are under clinical investigation. Initial studies have shown that, when used as single agents, these compounds have limited activity in AML with mutant FLT3. However, the data look more promising when these inhibitors are combined with conventional chemotherapy. Results from phase III clinical trials, which are about to start, will answer the question whether inhibition of FLT3 will be a successful targeted approach for these patients.

Mutations in CEBPA

Two major types of heterozygous CEBPA mutations have been identified in sporadic and familial AML. Nonsense mutations affecting the N-terminal region of the molecule prevent expression of the full-length CEBPA protein, thereby up-regulating the formation of a truncated isoform with dominant-negative properties; and in-frame mutations in the C-terminal basic region-leucine zipper domain resulting in CEBPA proteins with decreased DNA-binding or dimerization activity. N- and C-terminal mutations often occur simultaneously. Interestingly, by using gene expression profiling, a subgroup of AML could be defined that exhibits a transcriptional signature which resembles that of AMLs with CEBPA mutations, while lacking such mutations.11 In most, but not all, of these AMLs, the CEBPA gene was silenced by promoter hypermethylation. In addition, this subset of AML showed a strong association with putatively activating mutations in the NOTCH1 gene.

Figure 1.Major cytogenetic subgroups of acute myeloid leukemia (AML) (excluding acute promyelocytic leukemia) and associated gene mutations. In the subgroup various, NPM1 mutations are frequently found in AML with 9q deletion and trisomy 8, CEBPA mutations in AML with 9q deletion, MLL mutations in AML with trisomy 11, and RUNX1 mutations in AML with trisomy 13 and trisomy 21. Frequencies of the cytogenetic subgroups are derived from 2,654 cytogenetically characterized adult (≥18 years) patients with de novo or secondary AML entered on five AMLSG treatment trials.

Figure 2.Distribution of class I gene mutations among 259 cytogenetically normal acute myeloid leukemia (AML) with NPM1 mutation. In about 28% of cases, NPM1 mutation is the only detectable genetic change, whereas in the majority of cases additional mutations are found in genes such as FLT3, NRAS and WT1. A small number of AML with NPM1 mutations have an additional hypothetical class II mutation, e.g., NPM1 and concurrent CEBPA mutation.

Several studies have shown that CN-AMLs with CEBPA mutations are associated with favorable outcome (Figure 3).4,8 The survival data reported are very similar to those of CN-AML patients with mutant NPM1 without FLT3-ITD. It remains an open question whether the presence of a FLT3-ITD impacts on prognosis in patients with mutant CEBPA.

Mutations in MLL

Partial tandem duplications (PTD) of the MLL are detectable in 5%–11% of patients with CN-AML.4,8 At the molecular level, MLL-PTDs most commonly involve duplication of a genomic region spanning exons 5 through 11 and insertion of the duplicated segment into intron 4 of the gene. In contrast to MLL chimeric fusion proteins, MLL-PTDs retain all functional protein domains. MLL-PTDs have been found to cooperate with silencing of the MLL wild-type allele, which appears to be mediated through epigenetic mechanisms. Whitman et al. (2005) showed that the combination of decitabine, a DNA methyltransferase inhibitor, and depsipeptide, a histone deacetylase inhibitor, can reactivate transcription of the wild-type allele in MLL-PTD-positive blasts.12 The induction of MLL wild-type expression was associated with enhanced cell death of the blasts. These data may provide a rationale for evaluating DNA methyltransferase and/or histone decetylase inhibitors in AML with MLL-PTD.

Clinically, MLL-PTD has been associated with shorter CR duration and worse RFS and EFS; however, in these studies MLL-PTDs had no effect on OS.

Mutations in NRAS

NRAS mutations are found in 9–14% of younger adults with CN-AML.4,8 Mutations are almost exclusively located at codons 12, 13 and 61, resulting in loss of intrinsic GTPase activity and constitutive activation of the RAS protein. None of the larger studies has found an impact on prognosis, neither in the CN-AML subgroup nor in AML with other intermediate-risk karyotypes. Nevertheless, these mutations may represent a target for molecular therapy.

Mutations in WT1

Mutations in the WT1 gene in AML were first reported by King-Underwood and Pritchard-Jones in 1998.13 In a recent study of 70 patients with CN-AML by Summers et al. (2007),14 WT1 mutations were detected in 10% of cases. Mutations consisted of insertions or deletions that mainly clustered in exons 7 and 9. Preliminary data resulting from these two small studies on heterogeneous patient populations suggest that WT1 mutations may be associated with induction failure. However, the prognostic impact of WT1 mutations needs to be evaluated in larger patient cohorts and within the context of other molecular markers.

Mutations in RUNX1 (AML1)

RUNX1 encodes a transcription factor that is involved in regulation of normal hematopoietic differentiation through dimerization with the CBFB transcription factor. Notably, both RUNX1 and CBFB are involved in chromosomal translocations associated with AML. Mutations in RUNX1 have recently been shown to occur in approximately 10% of CN-AML.15 At present, no data are available regarding the clinical relevance of RUNX1 mutations in CN-AML.

Figure 3.Kaplan–Meier survival estimates, according to genotype, in younger adult patients with cytogenetically normal AML. Data are shown for relapse-free survival (Panel A) and overall survival (Panel B). Other genotypes is defined as the FLT3-ITD genotype and the triple-negative genotype consisting of wild-type NPM1 and CEBPA without FLT3-ITD. Tick marks represent patients whose data were censored at the last time they were known to be alive and in complete remission (Panel A) or whose data were censored at the last time they were known to be alive (Panel B) (Reprinted by permission of Schlenk RF, Döhner K, et al., N Engl J Med 2008;358:1909-18).

Molecular alterations in acute myeloid leukemia with balanced translocations/inversions

As has been shown in murine models of leukemogenesis, at least some of the recurrent gene fusions that are generated by balanced translocations and inversions are not sufficient to induce leukemia, suggesting a requirement for secondary genetic lesions. Consistent with these models, additional genetic lesions are found in many human AML with balanced translocations or inversions. For example, these lesions may include secondary chromosome abnormalities such as trisomy 22 in AML with inv(16)/t(16;16), deletion 9q in AML with t(8;21), or monosomy 7 in AML with inv(3)/t(3;3). The mechanisms underlying the formation of these secondary chromosome abnormalities remain elusive. Furthermore, several gene mutations have been identified in cases of AML with balanced translocations or inversions.

Gene mutations in core-binding-factor acute myeloid leukemia

CBF leukemias defined by inv(16)(p13.1q22)/t(16;16) (p13.1;q22) or by t(8;21)(q22;q22) account for about 10–15% of AML. At the molecular level, inv(16)/t(16;16) result in the generation of a CBFB-MYH11 fusion protein, whereas t(8;21) creates a RUNX1-RUNX1T1 fusion product. CBF AML are associated with favorable outcome; nevertheless, there is substantial heterogeneity with approximately 30–40% of younger adults relapsing. Over recent years, mutations have been identified in several genes in CBF AML, i.e. the KIT gene, the NRAS and KRAS genes, and the FLT3 gene.

KIT exon 8 mutations located in the extracellular portion of the receptor or mutations in the activation loop at codon 816 (exon 17) are detectable in about 30% of CBF AML. A number of studies have evaluated the prognostic significance of KIT mutations in CBF AML.5,16,17 In t(8;21) AML, mutant KIT, and in some studies specifically mutations at codon 816, have been associated with inferior outcome.16 In inv(16)/t(16;16), a study by the Cancer and Leukemia Group B (CALGB) on a larger patient cohort showed that KIT mutations are associated with a higher cumulative incidence of relapse; this difference was mainly due to the effect of KIT exon 17 mutations.17 In multivariable analysis, KIT mutation was an adverse prognostic factor for OS. These results await independent confirmation in large patient cohorts that have received uniform treatment. The prognostic significance of the other gene mutations is less well defined.

Irrespective of their prognostic significance, mutant KIT alleles represent a potential target for molecular therapies. Compounds such as the tyrosine kinase inhibitor midostaurin or the dual SRC/ABL kinase inhibitor dasatinib (BMS-354825) have significant inhibitory efficacy against mutant and wild-type KIT. The efficacy of tyrosine kinase inhibitors combined with conventional chemotherapy needs to be tested in future clinical trials of CBF AML.

Gene mutations in acute myeloid leukemia with t(6;9)(p23;q34)

The translocation t(6;9) is found in about 0.5%–1.0% of adult patients with AML, and the presence of this abnormality has been associated with poor treatment outcome. In a recent international meta-analysis, 55 patients (AML, n=50; MDS, n=5) were studied for activating FLT3 mutations.18 FLT3-ITDs were found in 42 of the 55 (76.4%) patients. Notably, compared to patients with wild-type FLT3, patients with FLT3-ITD had a significantly lower probability of achieving a CR (75% vs. 36%, p=0.042). In addition, patients with FLT3-ITD had significantly inferior DFS and OS. These data provide evidence that the poor prognosis of patients with t(6;9) may not only be due to the DEK-NUP214 gene fusion resulting from this chromosomal translocation, but also to the presence of activating FLT3 mutations. These data provide a rationale for the evaluation of FLT3 inhibitors in this subtype of AML.

Molecular alterations in acute myeloid leukemia with inv(3)(q21q26.2)/t(3;3)(q21;q26.2)

inv(3)/t(3;3) are detected in about 1–2% of all AML and result in fusion of the ecotropic virus integration-1 (EVI1) gene or the related MDS-EVI1 gene at 3q26.2 with the ribophorin 1 gene (RPN1) at 3q21. These rearrangements lead to increased EVI1 gene expression. inv(3)/t(3;3) have been consistently associated with a dismal outcome. Monosomy 7 and 5q deletions are recurrent secondary chromosome aberrations, and NRAS gene mutations have been found in about 25% of cases with inv(3)/t(3;3). Notably, deregulated expression of EVI1 has previously also been identified in a subset of AML without 3q rearrangements. In a recent study of 534 untreated adult patients with de novo AML, deregulated expression of various EVI1 splice variants and of the related MDS1/EVI1 (ME) gene was found in 7.8% of the cases, and high EVI1 expression was an independent predictor of poor survival.19 Some of these cases lacked cytogenetically visible 3q rearrangements, but using fluorescence in situ hybridization, additional cryptic 3q26 breakpoints were identified. In addition, there was a correlation between EVI1/ME overexpression and the presence of 11q23 chromosome rearrangements, suggesting a role for MLL fusion proteins in the regulation of EVI1 and ME expression. The authors argued that quantitative EVI1/ME expression analysis should be added to the molecular diagnostic procedures in AML.

Molecular alterations in acute myeloid leukemia with various cytogenetic abnormalities

Most of the gene mutations described for CN-AML are not restricted to this cytogenetic subset of AML but occur with varying frequencies in AML with abnormal karyotypes.

In an AMLSG-CALGB intergroup study,20 AMLs with deletion 9q occurring in the context of a non-complex karyotype were shown to be associated with either CEBPA or NPM1 mutations; the prevalence of CEBPA and NPM1 mutations was 43% and 37% respectively. Notably, CEBPA and NPM1 mutations occurred mutually exclusive, a finding that supports the hypothesis that both CEBPA and NPM1 mutations act as class II mutations which cooperate with a class I mutation affecting an as yet unknown gene on 9q. NPM1 mutations have also been found in up to 20% of cases with trisomy 8. Large patient numbers need to be studied to evaluate the exact prevalence of NPM1 mutations in this cytogenetic subgroup. Such studies may eventually help clarify the enormous clinical heterogeneity seen among patients whose leukemic cells exhibit trisomy 8 as a single abnormality. Finally, in a recent study, RUNX1 gene mutations have been shown to be associated with specific cytogenetic abnormalities.15 Interestingly, in this study 14 out of 14 cases with trisomy 13, 5 out of 9 (56%) cases with trisomy 21, and 10 out of 32 (31%) cases with monosomy 7 exhibited RUNX1 mutations.

Molecular alterations in acute myeloid leukemia with complex karyotype

AML with complex karyotype is commonly defined by the presence of at least 3 acquired chromosome aberrations in the absence of t(8;21), inv(16)/t(16;16), and t(15;17). Complex karyotype AML constitutes 10–12% of all AML, with its incidence increasing with age. So far, little is known about the molecular pathogenesis of these leukemias.21 Notably, complex karyotype AML is characterized by the rare occurrence of balanced chromosomal rearrangements, and also by a low frequency of mutations in NPM1, FLT3, CEBPA, RAS or KIT, which are frequently mutated in other types of AML, suggesting deregulation of other molecular pathways in these cases. In AML with complex karyotype, there is a predominance of chromosomal imbalances, with chromosomal losses occurring more frequently than chromosomal gains.

Over recent years, molecular cytogenetic and array-based techniques have contributed to an improved characterization of these genetic changes.21,22,23 Deletions have been mapped, in order of decreasing frequency, to 5q, 17p, 7q, 18q, 16q, 17q, 12p, 20q, 18p, and 3p, and genomic gains to 8q, 11q, 21q, 22q, 1p, 9p, and 13q. Efforts to identify the target genes of these recurrent genomic losses have been complicated by the fact that these tumors are mostly characterized by large genomic deletions. The TP53 gene is the likely target for losses occurring on 17p; in fact, TP53 gene mutations have been identified in about two-thirds of AML with complex karyotype. Loss of p53 protein function by TP53 mutation is known to result in genetic instability and TP53 inactivation by gene loss or gene mutation may therefore represent one molecular pathway leading to complex genetic changes. Another difficulty in the analysis of chromosomal deletions is that of identifying genes that contribute to tumorigenesis by haploinsufficiency, i.e. the inactivation of a single allele. The value of functional genetic techniques in this regard is illustrated by a recent study in which down-regulation of candidate genes by RNAi was used to identify RPS14 on 5q32 as a causal gene for the 5qsyndrome, a subtype of the myelodysplastic syndromes.24 It is conceivable that other genes located within critical regions of recurrent genomic losses may contribute to leukemogenesis through haploinsufficiency.

The detection of recurrent genomic amplifications by modern techniques, such as array-based comparative genomic hybridization, has led to the identification of genes that are potentially involved in the pathogenesis of cytogenetically complex AML as well as other cytogenetic AML subtypes.21,22 Genes that have been shown to be frequently amplified include members of the ETS gene family, specifically ERG and ETS2 on chromosome 21q, and ETS1 and FLI1 on chromosome band 11q23.3–q24.1. Other recurrent amplifications (candidate genes within amplicons) have been mapped to 8q24 (MYC, TRIB1), 9p24 (JAK2), 11q13.5 (GAB2), 11q23.3 (DDX6, MLL), and 13q12 (CDX2, FLT3, FOXOA1). The frequent finding of chromosomal amplification suggests that oncogene activation through genomic gain is more common in this type of AML than previously thought.

Conclusions

Important discoveries have been made over recent years that have contributed to a better understanding of the molecular pathogenesis and to a refinement of the classification of AML. Furthermore, mutations in genes such as NPM1, FLT3 or CEBPA have been shown to provide important prognostic information, and it is now recommended to include mutation analysis of these genes in the initial diagnostic work-up of a patient with AML, at least in the context of a clinical trial.

Nevertheless, we have only just begun to unravel the enormous genetic diversity of AML. The constant improvements in genomics technology will lead to the identification of additional genetic changes and novel mechanisms of leukemogenesis. The introduction of genome-wide single nucleotide polymorphism based mapping arrays has led to the identification of a novel mechanism of leukemogenesis involved in AML, partial uniparental disomy, that has been shown to occur in up to 20% of AML cases at the time of diagnosis. In addition, high-throughput DNA sequence analysis in large numbers of primary patient samples will become possible at an affordable cost, which may ultimately result in the development of comprehensive, disease and allele-specific gene mutation profiling strategies. Finally, innovative functional genetic approaches, such as large-scale RNA interference screens, have great potential for the identification of novel cancer genes. It is hoped that the data resulting from these studies will also eventually lead to the development of successful molecular targeted therapies.

Acknowledgments

Supported in part by grants 01GI9981 [Network of Competence Acute and Chronic Leukemias], 01KG0605 [IPD-Meta-Analysis: A model-based hierarchical prognostic system for adult patients with acute myeloid leukemia (AML)] from the Bundesministerium für Bildung und Forschung (BMBF), Germany, the Deutsche José Carreras Leukämie-Stiftung (DJCLS R 06/06v), and the Else Kröner-Fresenius-Stiftung (P38/05//A49/05//F03). We thank Dr. Stefan Fröhling, Dr. Richard F. Schlenk, Dr. Verena Gaidzik and Dr. Sabine Kayser for their critical reading of the manuscript.