Current Oncology Reports

, Volume 14, Issue 5, pp 359–368

Acute Myeloid Leukemia with Normal Cytogenetics

Authors

  • Raya Mawad
    • Clinical Research DivisionFred Hutchinson Cancer Research Center
    • Department of Medicine, Division of Medical OncologyUniversity of Washington
    • Clinical Research DivisionFred Hutchinson Cancer Research Center
    • Department of Medicine, Division of Medical OncologyUniversity of Washington
Leukemia (A Aguayo, Section Editor)

DOI: 10.1007/s11912-012-0252-x

Cite this article as:
Mawad, R. & Estey, E.H. Curr Oncol Rep (2012) 14: 359. doi:10.1007/s11912-012-0252-x

Abstract

Acute myeloid leukemia (AML) is proving to be a heterogeneous disease process that is driven by various genetic mutations and aberrant protein expression. As our population ages, the incidence of AML is likely to increase, with approximately a third of adult cases categorized with normal cytogenetics. Advances in technology are now allowing us to explore the genetic expression and protein transcription patterns of AML, providing more information that must find its place in the prognosis and the therapeutic algorithm of this disease. As we learn more, we hope to further categorize patients with normal karyotype AML into discrete risk categories that will help in treatment decision making and further elucidate the necessity for hematopoietic cell transplantation. However, at this time, many of the identified mutations and expression patterns are still experimental, requiring further analysis to determine their exact role in AML.

Keywords

Acute myeloid leukemiaAMLNormal cytogeneticsNormal karyotypeIntermediate risk

Introduction

Since acute myeloid leukemia (AML) is a disease of the aging, the incidence will likely increase along with our aging population. Advances in our ability to analyze AML at the molecular level have increased our appreciation that it is a heterogeneous disease associated with various mutations. For many years, the probability of response to standard therapy (anthracycline and ara-C) was based on cytogenetic analysis in which 20 metaphases are typically analyzed [1]. Thus, inversion 16 and t (8;21), which affect the core binding transcription factor (CBF) were associated with a better response (40 %–80 % long-term remission rate). In contrast, AML characterized by monosomies of chromosomes 5 and/or 7 or deletion of their long arms (particularly when associated with a monosomal karyotype), t (6;9), inv 3 or t (3;3) have long-term remission rates <30 %. These low remission rates are seen even when standard induction therapy is followed by allogeneic transplant (HCT), placing these genetic mutations in the adverse risk group. The normal cytogenetic group accounts for approximately 35 % of patients, and these were initially placed in the intermediate risk category. Their prognoses with standard therapy, however, were more varied than patients in the favorable or adverse groups. This variability made it more difficult to recommend standard therapy vs a clinical trial, the fundamental management question in AML. However, increasing knowledge of various molecular mutations allows for further sub-classification of normal cytogenetics AML, with the hope that this improves prognostication and affects therapeutic decision-making. The main questions arising with identification of new mutations involve whether these changes are considered a driving force of the disease process, and whether they can be targeted by pharmacologic intervention.

Molecular Mutations

Flt3

Flt3 (Fms-like tyrosine kinase III) belongs to the class III receptor tyrosine kinase family, and was independently cloned by 2 groups in 1991 [2, 3]. It is expressed on normal hematopoietic progenitors, including CD34+ cells that have high levels of CD117 expression [4]. In vitro studies have shown that Flt3 mutations cause autophosphorylation of the receptor, leading to cell proliferation, and the inhibition of apoptosis [5, 6]. Mutations are seen in approximately 30 % of adult AML patients, causing constitutive activation of the Flt3 gene and formation of an abnormal FLT3 protein. The 2 most common mutations involve internal tandem duplications (ITD) affecting the juxtamembrane domain and mutations affecting the tyrosine kinase domain (TKD). Flt3-ITD mutations are associated with de novo AML with higher white count, LDH, and blast count at presentation [7]. Several studies have proven that the presence of Flt3-ITD mutations confers a poorer prognosis, related to increased relapse rates [79]. Specifically, a higher ratio of FLT3 protein derived from the ITD gene in relation to the normal gene predicts a higher relapse rate [9, 10]. The prognostic impact of Flt3-TKD mutations, however, remains controversial. Although 1 study suggested a favorable prognosis compared to ITD mutations [11], a subsequent analysis revealed poor outcomes with Flt3-TKD mutations [12]. It is possible that the prognosis of Flt3-TKD mutations depends on the presence or absence of other concurrent molecular abnormalities, as suggested by analysis of a cohort of 1720 patients by Bacher et al [13]. However, given its poor prognosis, Flt3 ITD + patients are generally referred for HCT, with the hopes of improving their overall survival and relapse rate. Pharmacologic targeting of Flt3+ mutations is currently being evaluated in various clinical trials. Previous evaluation of non-specific tyrosine kinase inhibitors (TKIs) such as midostaurin and sorafenib have shown significant clinical reductions in the leukemic burden [14]. However, this effect is short-lived, suggesting the development of resistance to these agents [15]. Current evaluation of more specific Flt3 inhibitors such as AC220 and PLX are underway in clinical trials. While AC220 produces a higher response rate than the older inhibitors, thus providing a “bridge to transplant,” it is not clear that the quality of responses is sufficient to predict a good outcome after HCT. Thus, AC220 is being combined with standard induction therapy, while there is interest in single agent AC220 in FLT 3 ITD positive patients in remission, for example after HCT.

NPM1

Nucleophosmin 1 (NPM1) is an intracellular chaperone protein involved in cellular proliferation and apoptosis. Exon 12 mutations in the NPM1 gene are seen in one-third of all adult AML cases and up to 50 %–60 % of adult AML with normal cytogenetics [16, 17]. NPM1 mutations often occur in the setting of other cytogenetic or molecular mutations, and are thought to confer a favorable prognosis associated with higher CR rates and lower relapse rates (even in elderly patients) [10, 1820]. NPM1+ mutated outcomes are improved to such a degree that NPM1+ AML in the absence of FLT3 mutation is now considered a favorable genetic subtype [1]. By analogy to other favorable subtypes, these patients are hypothesized to benefit from intensification of therapy, involving either daunorubicin 90 mg/m2 or high-dose ara-C. Allogeneic HCT is not recommended in patients who are NPM1+/FLT3– since they have a 5-year overall survival rate of approximately 55 % with standard chemotherapy, and no survival advantage was seen with HCT [21]. The role of all-trans retinoic acid (ATRA) in NPM1+/FLT3– patients is yet to be determined. Although a benefit was seen in older patients treated with ATRA and conventional chemotherapy [22], these results were note reproduced in a trial conducted by the British Medical Research Council [23]. NPM1 mutations have been thought to occur early in leukemogenesis, in comparison to Flt3 mutations, and are thus considered more fundamental to development of AML. Indeed, it appears that NPM mutation status is less likely to change at relapse than FLT3 ITD status, making it an attractive marker for the detection of minimal residual disease.

CEBPA

CCAAT Enhancer Binding Protein alpha (the protein encoded by CEBPA) is expressed in myeloid progenitors, and is thought to be important in myeloid differentiation [24]. Mutations in CEBPA can occur through either truncating mutations in the N-terminal region or in-frame mutations affecting the C-terminal leucine zipper [25]. Somatic CEBPA mutations are found in approximately 10 % of patients with normal karyotype AML, and they are uncommonly seen with NPM1 mutations [18]. Mutated CEBPA confers a favorable prognosis in various clinical trials, even when accounting for cytogenetics and FLT3 karyotype [26, 27]. However, this benefit is only seen with bi-allelic mutations, with the effect of a single CEBPA mutation similar to that of CEBPA wildtype [2830]. Biallelic CEBPA mutations confer a similar outcome to NPM1+/FLT3– AML, placing them in the favorable risk category by the World Health Organization (WHO). Therefore, recommended therapy involves cytarabine-containing induction with high dose cytarabine consolidation. Based on the favorable risk classification, it is assumed that allogeneic HCT would not benefit these patients, although the low incidence of this mutation would make such assessments difficult.

DNMT3A

DNMT3A encodes for a methyl-transferase, and DNMT3A mutations can lead to hypermethylation of certain tumor suppressor genes associated with the development of AML [31]. Massive genome sequencing of AML patients with normal cytogenetics revealed DNMT3A mutations in 22 % of adult AML patients and 33 % of patients with normal cytogenetics AML [32]. Further analysis of 489 AML patients by Thol et al showed that DNMT3A mutations were associated with statistically significant decrease in overall survival (OS) of all patients, including patients with normal cytogenetics. Sub-analysis revealed that this decrease in OS and relapse free survival (RFS) was most significant in patients with normal cytogenetics in the high risk NPM1–/FLt3+ group [33]. Possible confounders to keep in mind regarding this analysis involve the inclusion of patients with secondary AML and the use of allogeneic HCT in many of the patients analyzed. Additional groups have confirmed findings that DNMT3A mutations occur frequently in normal cytogenetics AML and are often associated with NPM1 and FLT3 mutations. However, several multivariate analyses have shown DNMT3A mutations to be an independent predictor of worse overall survival for AML patients [3436]. Marcucci et al found a particular mutation, R882-DNMT3A to occur in 62 % of patients with the mutated gene, suggesting a possible target for future pharmacologic intervention [37]. Additionally, the potential benefit of hypomethylating agents and higher dosed daunorubicin in patients with DNMT3A mutations has been suggested and likely warrants further investigation [38, 39].

MLL-Partial Tandem Duplications

MLL partial tandem duplications (PTD) occur in approximately 5 %–11 % of AML patients with normal cytogenetics [40, 41]. The mutation involves the duplication of a specific genomic region (exon 5 through 11) that is then inserted into intron 4, causing protein elongation. Interestingly, this mutation does not affect the MLL protein functional domains. However, it is thought to cause silencing of the MLL wild-type allele in AML blasts through epigenetic mechanisms [42]. MLL-PTDs are associated with statistically significant decreases in remission duration, with associated decreases in relapse free survival (RFS) [40, 41]. The MLL-PTD positive patients had a much shorter duration of remission compared to their MLL-PTD negative counterparts, 7.75 months vs 19 months (P < 0.001) [40]. Most recent mutational analysis of 18 genes in 398 AML patients identified MLL-PTD as a poor-risk molecular feature that is associated with significantly reduced overall survival in normal-cytogenetics AML (P = 0.009) [39]. In this evaluation, the MLL-PTD prognostic effect is similar to that of adverse cytogenetic mutations. However, the use of high dose anthracycline-based therapy was shown to improve 3-year overall survival from 25 % to 44 %. Other potential therapeutic strategies involve the use of hypomethylating agents or histone deacetylase (HDAC) inhibitors. The possibility of reversing the epigenetic silencing of wild-type MLL with decitabine and depsipeptide has shown promising results in 1 preclinical study [42].

IDH1 and 2

Isocitrate dehydrogenase genes 1 or 2 (IDH1/IDH2) mutations have been detected in up to 15 %–33 % of AML patients with normal cytogenetics and are often seen with concurrent NPM1 or Flt3 mutations [43, 44]. They encode metabolic enzymes which, when dysfunctional, generate 2-hydroxylglutarate, thought to be an oncogenic metabolite [45]. However, their prognostic significance and role as therapeutic targets remains controversial. Some studies have found a negative prognostic association with IDH1/IDH2 mutations [43, 44, 46], while others have noted a favorable or unaffected outcome in patients [47, 48]. Most recently, mutational analysis of 18 genes in 398 patients younger than age 60 with AML revealed R140Q IDH2 mutations to be associated with improved outcomes, with a 3-year overall survival of 66 % (P = 0.01). This same study revealed that NPM1 mutations in the setting of concurrent IDH1/2 mutations conferred a 3 year overall survival of 89 % compared with 31 % with mutated NPM1 and wildtype IDH1/2 (P < 0.001) [39]. Further studies are necessary to decipher the exact prognostic role of IDH mutations and their potential as therapeutic targets.

Other Genetic Mutations

Wilms Tumor 1 gene (WT1) encodes a transcription regulatory protein involved in cellular replication and maturation. WT1 mutations are seen in approximately 10 % of AML patients with normal cytogenetics [49, 50]. The prognostic significance of WT1 mutations has not been clearly elucidated, since some studies have shown inferior outcomes and chemotherapy resistance [5052], while others have not [49, 53]. Runt-related transcription factor 1(RUNX1) encodes a transcription involved in hematopoietic differentiation, and its mutations are associated with the M0 FAB classification [54]. RUNX1 mutations are seen in 10 % of normal cytogenetics AML, and they are associated with resistance to chemotherapy, as well as lower RFS and OS [55, 56]. NRAS mutations result in constitutive RAS activation and are seen in up to 13 % of normal cytogenetics AML [21]. However, no clinically significant effect has been seen thus far with NRAS mutations [57]. Other genetic mutations in TET2, TP53, and ASXL1 are seen in various myeloid neoplasms, however, their clinical significance is still being evaluated [39].

Genetic Overexpression

Certain genes have been shown to be up-regulated in AML patients, with data supporting their independent prognostic significance in patients with normal cytogenetics. The brain and acute leukemia cytoplasmic gene (BAALC) is normally expressed in hematopoietic precursors and neuro-ectodermally derived tissues, but not in mononuclear cells or mature marrow components. It encodes a protein that has no known homologes [58]. Its overexpression has been associated with chemoresistance, and is an independent prognostic factor that decreases CR rates, DFS, and OS in patients with normal cytogenetics (also in older patients) [5964]. BAALC overexpression has recently been associated with concurrent overexpression of microRNA miR-3151, suggesting a relationship between the two and significantly worse outcomes with overexpression of both [65]. In several studies examining genetic expression profiles, overexpression of ERG (an ETS related gene) was also noted to be independently associated with worse outcomes, especially in NPM1+/FLt3- patients [61, 62, 66, 67]. Similarly, increased MN1 gene expression has been associated with chemoresistance, higher relapse rates, and decreased OS in AML patients with normal cytogenetics [60, 6870].

Gene Expression, Epigenetic, and Proteomic Profiling

Advances in DNA micro-array technology have led to more rapid gene expression profiling (GEP). Initial published results confirmed the ability of unsupervised genetic cluster analysis to generate specific molecular signatures that were associated with AML [71]. Several of these clusters were driven by genetic mutations with known prognostic impact on AML. However, new clusters were seen in samples with normal cytogenetics, raising the possibility that GEP would provide independent prognostic information in AML with no detectable karyotypic abnormality [71, 72]. Bullinger et al used gene expression profiling to separate patients with normal cytogenetics into two distinct clusters that had statistically significant differences in overall survival (P = 0.046) [72]. The group with the worse survival was characterized by high expression of DNMT3A/B, GATA2, and NOTCH1 expression. This group also notably had more Flt3+ mutations and was more often categorized with FAB M1 or M2 disease by morphology. However, multivariate analysis revealed this gene-expression outcome prediction tool to have independent prognostic significance (OR 8.8, 95%CI 2.6–29.3, P < 0.001) [72]. A later CALGB study validated this prediction model in a larger group of cytogenetically normal patients with longer follow-up time [73]. The observation that Flt3 negative had decreased survival associated with a certain genetic cluster supports the notion that GEP may provide additional prognostic information in patients with normal cytogenetics. Metzeler et al used GEP in a test cohort of 163 normal-cytogenetics AML patients to create a continuous risk score [74]. This score was then validated in 2 different cohorts as part of two different clinical trials. A higher gene-expression risk score was associated with statistically significantly decreased RFS and OS in both validation cohorts. A multivariate analysis adjusting for age, Flt3-ITD status, and NPM1 mutation status confirmed the gene-expression risk score as an independent predictor of OS (P = 0.37) [74]. Importantly, the interaction of this score with other mutations such as CEBPA, MLL-PTD, and Flt3-TKD was not assessed. Since gene expression and translation results in protein formation, it seems logical and natural to examine the relation between prognosis and protein profiling (proteomics) [75]. Kornblau et al assayed 256 samples from 256 newly diagnosed AML patients for 51 proteins involved in signal-transduction pathways (STPs), cell cycle regulation, and apoptosis [76]. Specific proteomic expression patterns were seen in different FAB subtypes, as well as Flt3-ITD samples and those with –5 or –7 karyotype. Seven signature protein groups were identified that correlated to a certain extent with known cytogenetic risk categories. However, these groups were thought to provide prognostic information, in addition to cytogenetics, which affected CR, relapse rates, and OS. Similarly, epigenetic profiling of gene methylation status can provide insight into the protein expression patterns implicated in leukemogenesis with normal cytogenetics. Figueroa et al examined the methylation profiles of 344 AML patients, resulting in data that segregated people into 16 distinct groups. Among the patients with normal cytogenetics, CEBPA mutations had a distinct methylation profile, and patients with NPM1 mutations were further categorized into 4 subgroups [77]. It is clear that sophisticated screening methods are emerging that can provide us with various additional information beyond simple cytogenetics. The key is determining how to use this information to improve risk stratification, treatment decisions, and therapy options.

MicroRNA Expression Patterns

Micro RNAs are small sized non-coding RNAs that bind to complementary sections of messenger RNA (mRNA) causing down-regulation or silencing of the translational product. Deregulation of micoRNAs has been implicated in various malignancies by affecting steps in cell proliferation, differentiation, and apoptosis. Abnormal expression of microRNAs has also been documented in AML, and specific patterns are seen with known cytogenetic mutations as well as FAB classifications [78, 79]. MicroRNA expression patterns have also been associated with various molecular mutations. For example, FLT3-ITD is seen with upregulation of miR-155 [79]. NPM1 expression is associated with a gene expression signature related to HOX gene overexpression, which correlates with increased MiR-10a, MiR-10b, and MiR-196a-I expression [80, 81]. Additionally, CEBPA mutations are related to high expression of miR-181 subtypes, which are associated with the erythroid differentiation and may explain the partial erythroid differentiation seen in AML with CEBPA mutations [82]. This association correlates with the improved outcomes seen with both CEBPA mutated AML and AML with higher miR-181 expression [83]. An evaluation of GEP and miRNA expression in high-risk AML patients with normal cytogenetics (FLt3-ITD+, NPM1 WT, or both) discovered 12 probes (5 of these involved the miR-181 family) that were associated with decreased event-free survival (EFS) (P < 0.04 in multivariate analysis accounting for FLt3-ITD status) [84]. It is difficult to tell if miRNA expression provides additional prognostic information above molecular and cytogenetic classification. However, information regarding miRNA expression in AML patients may provide insight into driving pathways or protein expression profiles that may hopefully lead to therapeutic innovation.

Post-Induction Factors

Although pretreatment factors such as cytogenetics/molecular mutations, age, and white count are considered important prognostic variables, disease response to therapy is also predictive of outcomes. The rate of blast clearance after induction has been shown to correlate with relapse rates and overall survival of AML [85, 86]. Although the rate of count recovery was previously thought to be simply an effect of chemotherapy, Estey et al have shown that slower time to achieving CR was an independent predictor of disease-free survival [87]. Similarly, Yanada et al’s analysis of 891 AML patients in first CR showed that the extent of count recovery at CR was an independent predictor of RFS [88]. These findings suggest that the timing and extent of normal count recovery are likely reflective of any residual disease present, despite achievement of a morphologic CR. Recent technologic advances have improved our ability to detect the presence of residual leukemia, down to the genetic and molecular levels. This has led to re-categorization of disease response to therapy, moving well past the simple attainment of a morphologic CR [89]. The presence of persistent flow cytometry, genetic, or molecular abnormalities detected in the setting of morphologic CR is considered minimal residual disease (MRD). Analysis of the role of MRD has shown it to be associated with increased relapse rates and decreased OS, even after hematopoietic cell transplantation (HCT) [9092]. However, the exact method, timing, cut-off values, and target mutations for MRD detection are still being evaluated [91]. An integration of karyotype, FLt3 status and presence of MRD was tested and validated in 143 AML patients by Buccisano et al, creating a low-risk group, which had much better RFS and OS rates compared with the high-risk group [93]. Integration of additional molecular risk factors and more specific MRD criteria will likely improve the ability to prognosticate and guide therapeutic decisions in patients with cytogenetically normal AML.

The Role of Transplant

Patients with normal cytogenetics AML have been traditionally categorized in the intermediate risk category. Thus, there has been controversy over whether they should receive HCT in first CR vs high dose consolidation. Studies have shown that allogeneic HCT can improve outcomes in adverse risk AML and has no significant impact on survival for favorable risk disease, when compared with consolidation with chemotherapy or autologous transplant [9496]. The benefit of allogeneic HCT has been less clear in patients traditionally categorized with “intermediate risk” disease. However, molecular analysis allows close to half of patients with normal karyotype to be placed into the “favorable” or “adverse” risk group. Included in the former are patients who are NPM1+/FLT3ITD– or who have a double mutation of CEBPA, and in the latter are patients who are FLT3ITD+, particularly if they have a high allelic ratio. By analogy then it would seem reasonable to defer HCT in CR1 in patients who are NPM1+/FLT3 ITD– or who are CEBPA double mutated. Data to support this approach come from Schlenk et al’s analysis of 872 patients on 4 different clinical trials [21]. Patients with NPM1, Flt3, CEBPA, MLL, and NRAS gene mutations were evaluated, with the benefit of transplant limited to patients with Flt3-ITD+ or wild type NPM1 and CEBPA without Flt3-ITD mutations. This finding helps one decide on treatment options for patients with a normal karyotype (Fig. 1). However, the previously mentioned additional genetic and molecular mutations have not been taken into account when evaluating, which patients with normal cytogenetics should receive HCT. In addition, post-induction factors, such as blast clearance kinetics and the presence of MRD, will likely be integrated in the treatment algorithm of this increasingly heterogeneous group of patients. It is clear that with the various new prognostic factors identified through techniques such as GEP and proteomic profiling, a prospective evaluation of allogeneic HCT in these patients is warranted. However, the increasing use of alternative donor sources will make it difficult to use conventional door vs no-donor analyses in future transplant trials.
https://static-content.springer.com/image/art%3A10.1007%2Fs11912-012-0252-x/MediaObjects/11912_2012_252_Fig1_HTML.gif
Fig. 1

Treatment algorithm for AML with normal cytogenetics. This includes the basic molecular testing that is now available at most academic institutions. The role of other mutations, such as DNMT3A, MLL-PTD, TET2. and BAALC is still being evaluated and is considered experimental at this point

Conclusion

There is little doubt that normal cytogenetic AML will be shown to be increasingly heterogeneous in the setting of such technologic advancement. Although more mutations and expression patterns are being identified, the prognostic and therapeutic implication of many of these discoveries remains to be seen. In the future, different treatments may be used in normal karyotype patients with specific molecular mutations or genetic expression profiles, FLT3 inhibitors being an example. Although clearly desirable, this development will likely call into question current methodology for phase 3 trials, the optimal means for assessing the values of such therapies. As the number of distinct molecular groups increases it will be impossible to enroll sufficient patients of a given type to maintain the conventional false positive (5 %) and false negative rates (20 %) without looking for much larger differences between control and new therapies than is currently the case. This challenge may prompt us to redefine trial design for patients with normal karyotype AML, in order to discover potential therapeutic benefits for this increasingly heterogeneous group.

Disclosure

No potential conflicts of interest relevant to this article were reported.

Copyright information

© Springer Science+Business Media, LLC 2012