Is it important to decipher the heterogeneity of “normal karyotype AML”?

Abstract

Almost half of adult acute myelogenous leukemia (AML) is normal cytogenetically, and this subgroup shows a remarkable heterogeneity of genetic mutations at the molecular level and an intermediate response to therapy. The finding of recurrent cytogenetic abnormalities has influenced, in a primary way, the understanding and treatment of leukemias. Yet “normal karyotype AML” lacks such obvious abnormalities, but has a variety of prognostically important genetic abnormalities. Thus, the presence of a FLT3-ITD (internal tandem duplication), MLL-PTD (partial tandem duplication), or the increased expression of ERG or EVI1 mRNAs confer a poor prognosis, and an increased risk of relapse. In contrast, the presence of cytoplasmic nucleophosmin or C/EBPA mutations is associated with lower relapse rates and improved survival. Although resistance to treatment is associated with specific mutations, it has also been suggested that the degree to which the leukemia resembles a stem cell in its functional properties provides greater protection from the effects of treatment. Although usually all of the circulating leukemia cells are cleared following treatment, a small residual population of leukemic cells in the bone marrow persists, making this disease hard to eradicate. Increased understanding of the biological consequences of at least some of these mutations in “normal karyotype AML” is leading to more targeted approaches to develop more effective treatments for this disease.

INTRODUCTION

The evolution of cytogenetics in the latter half of the 20th century has had a tremendous impact on the understanding and treatment of acute myelogenous leukemia (AML). The first cytogenetic abnormalities were identified in the 1960s and the first leukemia cell lines were established during the same period.^1–4 The isolation of reverse transcriptase,^5–7 and the development of DNA sequencing,⁸ polymerase chain reaction technology,⁹ and more recently, fluorescence in situ hybridization (FISH)¹⁰ have provided both profound insight and powerful tools for further research. These techniques have already elucidated more than a dozen mutations relevant to our detailed understanding of AML, such as KIT, FLT3 and NRAS mutations.¹¹

We are now in the process of trying to understand the biological and clinical implications of this genetic or molecular heterogeneity. The French-American-British (FAB) classification, proposed by Bennett et al in 1976¹¹ (Table 1), is based on morphology and cytochemistry. A few morphologically identifiable acute leukemias are associated with specific cytogenetic abnormalities (eg t(15;17) in acute promyelocytic leukemia; AML, M3), but there is no characteristic morphologic abnormality for the 40% to 45% of AML patients that have a normal karyotype. The prognosis of individuals with normal karyotype AML is intermediate, with a 5-year survival rate reported from 24% to 42%.^12–16 Clearly, the interplay between cytogenetics and the biological and clinical expression of the disease is more complex than it appears on the surface.

Table 1.

The FAB classification of AML

FAB subclassification of AML
M0	Undifferentiated	?
M1	Early Myeloid	?
M2	Late Myeloid	t(8;21)
M3	Promyelocytic	t(15;17)
M4	Myelomonocytic	Inv (16)
M5	Monocytic	t(9;11)
M6	Erythroleukemia	?
M7	Megakaryocytic	t(1;22), +21

Approximately 45% of AML patients have a normal karyotype. For intermediate risk patients, the 5-year-survival rate is 24% to 42%.

The heterogeneity of AML at the molecular level can be demonstrated biologically, providing insight into prognosis. As an example, ERG appears to function as an oncogene^17,18; it has been associated with prostate cancer,^{19–21 22} leukemia in Down’s syndrome/trisomy 21,^23,24 and Ewing’s sarcoma.^25–27 In AML, the increased expression of ERG is associated with a poor prognosis.²⁸

Response to treatment of AML is definitely associated with cytogenetic subtypes. For example, good prognosis is associated with t(15;17), inv(16), t(16;16) and t(8;21); intermediate prognosis is associated with normal karyotype, -Y; and poor prognosis is associated with monosomy 7 or 5, complex cytogenetics, most 11q23 abnormalities and loss of 17p.²⁹ However, these classifications are not fully predictive of response. One has to ask: what underlies these prognoses? For the last few years, our laboratory and others have focused on defining the stem cell “gene signature” and on examining how the presence of stem-cell-like properties of the leukemia cell can serve as a potential determinant of response to treatment. The question remains as to whether we are curing core binding factor (CBF) leukemias [inv(16), t(16;16), t(8,21)] because they are caused by CBF mutations or because the cellular origins of these leukemias are not stem-cell-like? A similar argument can be made for acute promyelocytic leukemia, which is usually CD34-. Morphologically, the M2, M3 and M4eo AMLs are more differentiated than other AMLs, and the target cell in these leukemias is presumably different than that of the FAB M0 and M1 leukemias. This line of reasoning (that poor prognosis AML is stem-cell-like and good prognosis AML is more progenitor-cell-like) does not hold absolutely, because there are well differentiated leukemias, erythroleukemia (M6) and megakaryocytic leukemia (M7), that are extremely difficult to treat with currently available therapies. Even so, several questions remain: does some fundamental aspect of stem cells determine the success of our treatments, and could that critical aspect be their quiescent nature? Are quiescent cells really the reason why we are not curing more patients with this disease? In solid tumors, the presence of quiescent cells may be more important than whether the cancer arose in a stem cell. One in every 10,000 cells in a solid tumor is said to be a stem cell.^30,31 However, chemotherapy for most solid tumors does not eliminate all but the 1 stem cell in 10⁴ cancer cells. Factors other than the presence of cancer stem cells clearly affect our ability to achieve complete remissions in patients with solid tumors.

REGULATION OF HEMATOPOETIC STEM CELL QUIESCENCE

We now know that stem cells reside in niches within the endosteal surface of bone, where they are protected from oxidative damage. From there, stem cells can move towards the vascular niche, enter the circulation, and later return to their niche. Several mechanisms keep the stem cells quiescent as long as they are in their niche. Intrinsic cell cycle regulators like p57 and p21 modulate the quiescence of stem cells, but there are also signals that are received by hematopoietic stem cells residing within the stem cell niche that lead to their quiescence. For example, CD56 is a known marker for poor prognosis in almost all AMLs.^32–34 A possible biologic explanation for this is that CD56 may be important for the adherence of some of the leukemic stem cells to the niche, so that the cells can receive pro-survival signals that keep them alive. Perhaps certain leukemias, which begin in the earliest stem cells, and others, which occur in slightly more committed cells but have acquired the ability to behave like a stem cell, are able to evade chemotherapy by residing in the highly protective stem cell niche.

Our laboratory has spent the last decade studying myeloid ELF-1-like factor (MEF), an ETS family member that, like AML1B, can strongly transactivate several promoters. This led us to examine whether MEF interacts functionally or physically with AML1 proteins.³⁵ MEF is located on Xq26, where it binds both AML1 and PML. It has now been shown to be involved in the t(X;21) translocation, where it fuses with ERG.³⁶ MEF null murine hematopoietic stem cells showed a marked increase in quiescence. This was seen in vivo, using bromodeoxyuradine (BrdU) uptake studies. Normally, over a 72 hour period, about 60% of the lineage -, Sca-¹⁺, c-kit⁺, (LSK) cells incorporate BrdU, whereas MEF-null LSK cells incorporate about one-third as much as is BrdU (22%) indicating that they divide at a much slower rate. Similarly, G0/G1 analyses demonstrated that about twice as many MEF null LSK cells are in G0, compared to wild-type LSK cells.³⁷

This enhanced quiescence is associated with chemotherapy resistance. Thus, dosing wild type mice with 5-fluorouracil (5-FU), results in severe neutropenia, with survival of only 2 of 6 wild type mice. However, MEF knockout mice given the same dose of 5-FU have less neutropenia and all 6 survive 5-FU treatment. Bone marrow examination on day 8 after a high dose of 5-FU demonstrates an empty bone marrow in wild-type, “chemosensitive normal” mice, and a relatively hypercellular bone marrow in the MEF-null mouse. Few stem cells are observed in the marrow of the “chemosensitive normal mouse,” whereas, in the chemoresistant MEF-null mouse, the stem cell niche appears relatively full of hematopoietic stem cells. This suggests that the pathways controlled by MEF (and probably other cell cycle regulators) maintain the leukemic stem cells in the niche, keeping them quiescent and protecting them from chemotherapeutic agents. Activation of these pathways may partially explain the reason why we are not particularly effective in treating most cancers.

MUTATIONS AND PROGNOSIS: C/EBP ALPHA AND FLT3, CYTOPLASMIC NUCLEOPLASMIN, AND BAALC

Mutations that are associated with improved prognosis have been identified. For example, mutations resulting in either dominant negative forms of (C/EBP alpha) or to haplo insufficiency of CCAAT/enhancer-binding protein (C/EBP alpha) function are associated with better than average prognosis in AML.^14,38–41 While the complete response rates to chemotherapy are not different, fewer patients relapse, and overall survival is better. In addition to C/EBP-alpha mutations, there are a variety of FLT3 mutations detectable in 30% to 40% of AML patients.⁴⁰ FLT3 can phosphorylate and inactivate C/EBP alpha (leading to a block in differentiation),^42,43 and FLT3 mutations are associated with worse outcome in AML.^44–46 FLT3 mutations combined with partial tandem duplication (PTD) of the mixed lineage leukemia gene (MLL) are rare but convey an even worse prognosis.^47–49

Why should C/EBP-alpha mutations confer a better than average prognosis in normal karyotype AML? Work on a C/EBP-alpha knockout mouse model, published by Zhang and Tenen et al⁵⁰ demonstrated profound neutropenia that was unresponsive to G-CSF treatment, and the C/EBP-alpha null cells appear to be blocked (ie a maturation arrest) at the promyelocyte stage. Perhaps leukemias that have C/EBP-alpha mutations are arrested at a late stage in myeloid differentiation, and the better-than-average prognosis is due to the type of cell that harbors this mutation or to effects of the loss of C/EBP-alpha function arresting the cell at this stage of differentiation. Either way, the cell itself would be more susceptible to chemotherapy.

In 2005, Falini and colleagues with the GIMEMA Acute Leukemia Working Party published their work demonstrating a high frequency of nucleophosmin (NPM) mutations in patients with normal-karyotype AML.⁵¹ Almost 40% of these patients have NPM1 mutations, which lead to frame shifts that cause expression of a nuclear export signal, resulting in mislocalization of NPM to the cytoplasm. Using immunostaining techniques and monoclonal antibodies against anaplastic lymphoma kinase (ALK) and NPM, these researchers identified a large subgroup of patients with normal karyotype AML and cytoplasmic NPM. These mutations have been associated with female sex, higher white blood count (WBC), increased blast percentage, and low or absent CD34 expression. This suggests that cytoplasmic NPM may also be found in a cell that is at a later stage of differentiation than the more difficult-to-treat leukemias, hence its better prognosis. Among the patients with NPMc, who did not also have a FLT3 mutation, prognosis was generally more favorable; that is, they were more likely to respond to induction therapy and stay in remission.

The brain and acute leukemia cytoplasmic (BAALC) gene encodes a protein with still unidentified homology to any known proteins or functional domains. It is expressed in neuroectoderm-derived tissues and hematopoietic cells. When BAALC is overexpressed in AML with normal cytogenetics, it confers a poor prognosis, especially in the presence of FLT3 mutations. In a study reported by Baldus et al,⁵² pretreatment blood samples were used to measure BAALC expression in 307 adults with AML and normal cytogenetics. Patients were divided into low- and high- BAALC expression groups. This study showed that high BAALC expression was associated with a higher incidence of relapse and a lower overall survival. BAALC is over-expressed in AML, ALL, and in CML in blast crisis, but not in CLL or CML in chronic phase. As yet, we have no clear explanation for this other than the possibility that BAALC expression is characteristic of stem cells with increased potentiality, ie, a cell in an earlier stage. With respect to FLT3 mutations and BAALC expression, 4 subgroups were identified: BAALC low/FLT3 low (n = 125), BAALC high/FLT3 low (n = 110), BAALC low/FLT3 high (n = 12), and BAALC high/FLT3 high (n = 21). Patients with low-risk FLT3 mutation and low BAALC expression showed the best outcome, whereas those with high-risk FLT3 and high BAALC expression had the worst outcome (Figure 1).⁵²

Figure 1.

Outcome based on BAALC expression and FLT3 mutation status. The probability of outcome based low- or high-risk FLT3 mutation status and BAALC expression was measured in this cohort of patients. Those with FLT3 low/BAALC low (n-125) had the best outcome, followed by low FLT3/high BAALC (n=110), then high FLT3/low BAALC (n=12), and those with high FLT3/high BAALC (n=21) had the worst outcome.⁵²

Molecular epidemiology is important, but what can it tell the clinician and the patient about therapeutic options? Perhaps poor prognosis mutations portend a poor prognosis despite varying the intensity of treatment. However, the study of BAALC expression suggested that allogenic stem cell transplantation consolidation therapy may be associated with improved long-term outcome for patients with high BAALC expression. Indeed, it is our hope that intensifying the treatment of certain subtypes of normal karyotype AML can lead to improved outcome. However, we not only need to identify these subgroups but also to sufficiently understand the biology involved to target it meaningfully. MLL-PTD is an interesting example of this subtype. Several groups have recently published studies linking MLL-PTD to worse outcomes.^53–55 The MLL protein has histone methyltransferase activity, which may be modulated or targeted in the future.⁵⁶

NEW THERAPIES

Currently, the most promising avenues of research, which take advantage of our increasing understanding of the biologic consequences of mutations in malignant myeloid cells, have focused on inhibiting enzymatic activities, especially tyrosine kinases. However, epigenetic based therapies are showing some success, and combination therapies are being widely explored. Gore and colleagues at Johns Hopkins and our group at MSKCC have reported encouraging phase 1 results using a combination of DNA methyltransferase inhibition followed by histone deacetylase (HDAC) inhibition.^57,58 These studies are beginning to address the hypothesis that reversal of aberrant gene silencing can improve remission rates in some patients with normal karyotype AML.⁵⁷ In vitro work on histone demethylase inhibitors is exploring these same pathways with the idea that the right sequence of inhibitors may be able to restore gene expression, leading to a more differentiated phenotype and better disease control.⁵⁹ Kinase inhibitors active against FLT3, the mTOR pathway, AKT, and PI3K, among others, are all under investigation.^60–64 HDAC inhibitors may overcome the consequences of haploinsufficiency for C/EBP alpha.^65,66 Other agents may affect hematopoietic stem cell quiescence, but, so far, our attempts to activate quiescent cells to resume cycling using agents such as GM-CSF or G-CSF have been largely unimpressive.^67–69

Immunotherapy in AML goes back to studies of Bacillus Calmette-Guérin (BCG) as an adjuvant to chemotherapy.⁷⁰ However, this “shotgun” approach failed over time to show clear survival benefit.⁷¹ More recently, defining how leukemic cells present antigens, and learning how to maximize that presentation, has led to various approaches intended to allow the individual’s immune system to engage and eliminate leukemic cells.^72–74 Additionally, blockade of the export and import of RNA or protein species from the nucleus appears to be a critical factor in translocations in leukemias, through the function of the nuclear pore complex proteins, NUP98, NUP96, and NUP214. Restoring the normal functions of these nuclear pore proteins may be another path to modulating the response to treatment in AML.^75,76

SUMMARY AND CONCLUSIONS

Our therapies generally fail to kill all of the malignant cells, and a small population of cells that is unaffected by therapy is responsible for persistence or recurrence of the disease. There is a need to understand both cancer cell quiescence and the cancer initiating cell, which in some cases may be a stem cell. Although much is known about the cytogenetic and genetic mutations in AML and the prognosis related to these mutations, little is known about how to better treat patients with poor prognosis, other than with allogeneic stem cell transplantation. Given that as many as 25% of AML patients with normal cytogenetics do not have any of the mutations discussed above, a better understanding of the molecular mechanisms underlying these subtypes of the disease will hopefully lead to the discovery of more effective ways to target them.