Nucleophosmin (NPM1) Mutations in Adult and Childhood Acute Myeloid Leukemia: Towards Definition of a New Leukemia Entity

Abstract

Nucleophosmin (NPM) is a ubiquitously expressed chaperone protein that shuttles rapidly between the nucleous and cytoplasm, but predominantly resides in the nucleous. It plays key roles in ribosome biogenesis, centrosome duplication, genomic stability, cell cycle progression and apoptosis. Somatic mutations in exon 12 of the NPM gene (NPM1) are the most frequent genetic abnormality in adult AML, found in approximately 35% of all cases and up to 60% of patients with normal karyotype AML. In children, NPM1 mutations are far less frequent, occurring in 8–10% of all AML cases, and in approximately 25% of those with a normal karyotype. NPM1 mutations lead to aberrant localization of the NPM protein into the cytoplasm, thus the designation, NPMc+ AML. NPMc+ AML is seen predominantly in patients with a normal karyotype and is essentially mutually exclusive of recurrent chromosomal translocations. Patients with NPM1 mutations are twice as likely as those who lack an NPM1 mutation to also have a FMS-like tyrosine kinase (Flt3) internal tandem duplication (ITD) mutation. NPMc+ AML is also characterized by a unique gene expression signature and microRNA signature. NPMc+ AML has important prognostic significance, as NPMc+ AML, in the absence of a coexisting Flt3-ITD mutation, is associated with a favorable outcome. NPM1 mutations have also shown great stability during disease evolution, and therefore represent a possible marker for minimal residual disease detection. Given its distinctive biologic and clinical features and its clear clinical relevance, NPMc+ AML is included as a provisional entity in the 2008 WHO classifications. There is still much to be learned about this genetic alteration, including its exact role in leukemogenesis, how it interacts with other mutations, and why it confers a more favorable prognosis. Further, it represents a potential therapeutic target warranting research aimed at identifying novel small molecules with activity in NPMc+ AML.

Introduction

Acute myeloid leukemia (AML) is a clinically and genetically heterogeneous disease that accounts for 15–20% of childhood leukemia and approximately 35% of adult leukemia. Currently, cytogenetic analysis at diagnosis allows for risk-stratification of AML into favorable, adverse, and intermediate risk¹. Treatment protocols are to some extent, risk-adapted in an attempt to improve survival and decrease treatment-related toxicity. Unfortunately, prognostic implications have not been reliably established for AML in the intermediate risk category, a group which includes 60%–70% of patients¹. In recent years, molecular analysis has identified novel markers with prognostic relevance in this diverse group. For example, AML with internal tandem duplication (ITD) in the fms-like tyrosine kinase-3 gene (FLT3) carries a poor prognosis^2,3, conversely, cases with a mutation in the transcription factor CCAAT/enhancer-binding protein-α (CEBPA) have a more favorable prognosis^4,5. Mutations in exon 12 of the nucleus-cytoplasm shuttling protein, nucleophosmin (NPM) have, since being identified in 2005, been established as a genetic alteration with important clinical and prognostic implications both in adult and childhood AML^6,7. We will herein review the molecular, biologic, clinical, and prognostic features of NPM1 mutations in adult and pediatric AML.

Nucleophosmin characteristics and functions

Nucleophosmin (NPM), also called nucleolar protein B23, numatrin, or NO38, is an abundant phosphoprotein that is ubiquitously expressed and highly conserved. The nucleophosmin gene (NPM1) is located on chromosome 5q35 and contains 12 exons⁸. The encoded protein is localized primarily in the nucleolus, but shuttles rapidly between the nucleus and cytoplasm⁹.

NPM has been shown to play an important role in many basic cellular processes. It has molecular chaperone activities including inhibition of protein aggregation, protection of enzymes against activity loss during thermal denaturation and promotion of renaturation of chemically denatured proteins¹⁰. It plays a key role in ribosome biogenesis through its shuttling properties and chaperone capabilities, which ensure proper transport of components from the nucleus to cytoplasm and prevents protein aggregation during ribosome assembly. Further, NPM mediates nuclear export of ribosomal protein L5/5S rRNA subunit complex¹¹. Other properties that implicate a role for NPM in the biogenesis of ribosomes include its intrinsic RNAse activity¹², ability to bind nucleic acids¹³, and capacity to process pre-RNA molecules¹⁴. NPM also functions as a histone chaperone that is capable of histone assembly, nucleosome assembly and increasing acetylation-dependent transcritption^15,16. Further, NPM has been implicated in the mitotic inhibition of GCN5 (general control of amino-acid synthesis 5)-mediated histone acetylation and transactivation which may be necessary to prevent premature histone acetylation before the onset of mitotic transcriptional reactivation ¹⁷. Thus, NPM appears to be important in regulating protein synthesis, cell growth, and proliferation.

NPM also is important in the maintenance of genomic stability. NPM regulates centrosome duplication as it associates with unduplicated centrosomes, inhibiting duplication. NPM dissociates from the centrosome upon CDK2-cyclinE mediated phosphorylation on thronine 199, triggering centrosome duplication¹⁸. NPM reassociates with centrosomes during mitosis after phosphorylation on serine 4 by PLK1¹⁹ and NEK2A²⁰. NPM inactivation causes unrestricted centrosome duplication and genomic instability²¹, with increased risk of cellular transformation. Thus, NPM acts as a licensing system for centrosome duplication ensuring the coordination of centrosome and DNA duplication as well as restricting centrosome duplication to occur once and only once within a single cell cycle. NPM may also help to maintain genomic stability through participation in DNA repair. NPM is mobilized to the nucleoplasm after double-strand DNA breakage where it binds to chromatin in a DNA-damage-dependent manner, implicating NPM in DNA repair and/or damage response²².

Further, NPM plays a key role in controlling cell cycle proliferation and apoptosis via its interactions with tumor suppressors p53 and ARF protein and their partners²³. NPM is crucial for the stabilization and activation of p53 in response to cellular stress²⁴. Human MDM2 (HMDM2) is a nucleoplasmic and nucleolar protein that controls the level of p53 by acting as an E3 ubiquitin ligase initiating p53 proteasomal degredation²⁵. Cell stress leads to nucleoplasmic localization of NPM where it interacts with and inhibits HMDM2 leading to p53 stabilization and activation²⁶. Further, p53 activation is controlled by GADD45α, a pro-apoptotic protein that is activated by genotoxic stress. Interaction of p53 and GADD45 leads to cell-cycle arrest at G2-M after cellular stress such as ionizing and UV irradiation²⁷. NPM directly binds to and regulates the cellular localization of GADD45 facilitating its nucleolar localization and interaction with p53²⁸. Thus, NPM plays a key role in potentiating p53-dependent cell-cycle arrest.

Alternate reading frame (ARF) protein (p19^Arf in mice, p14^Arf in humans) is a nucleolar protein that is involved in triggering cell-cycle arrest and apoptotic programs in response to oncogenic stress. It inhibits HMDM2 by relocalizing it to the nucleolus, which leads to the stabilization and activation of p53^29,30. NPM associates with ARF and protects it from degradation^31–33, thereby facilitating its activation. In addition to these p-53 dependent pathways, ARF also has p53-independent cell-cycle regulation capabilities. ARF is able to antagonize cell proliferation by inhibiting ribosome biogenesis³⁴. Perhaps the interaction of ARF and NPM in the nucleolus facilitates contact between ARF and the ribosomal machinery³¹. Also, ARF has been shown to inhibit NPM shuttling leading to the inhibition of cell proliferation³⁵.

NPM1 and human malignancies

NPM has been found to be overexpressed in tumors of various histological origins including gastric³⁶, colon³⁷, ovarian³⁸, and prostate³⁹. Further, the NPM1 gene is one of the most frequent targets of chromosomal translocations in hematopoietic malignancies⁴⁰. NPM1 is translocated with t(2;5) in anaplastic large-cell lymphoma creating the chimeric gene encoding for the fusion protein NPM-ALK⁴¹. Other translocation partners include t(3;5) in myelodysplasia/AML resulting in the fusion protein NPM-MLF1⁴², and t(5;17) in rare cases of acute promyelocytic leukemia resulting in an NPM-RAR fusion protein⁴³. The role of the NPM moiety of these chimeric products is not well characterized, but it is generally believed that it does not contribute to the transforming potential, but instead provides a dimerization interface for the oligomerization and the oncogenic conversion of the various partners⁴⁴. In addition, NPM1 is mapped to a region of chromosome 5 that is frequently deleted in therapy-related MDS⁴⁵.

In 2005, Falini et al reported that NPM1 was mutated at exon 12 resulting in aberrant cytoplasmic localization of NPM (NPMc+) in the leukemic blasts of approximately 35% of adult cases of AML, making NPM1 one of the most frequently mutated genes in AML⁶. Such exon 12 mutations are specific to AML as other neoplasms investigated have only shown nucleus-restricted NPM^6,46,47. NPMc+ has only been sporadically detected in chronic myeloproliferative disorders and myelodysplastic syndrome.^48–51 Upon careful review of the chronic myeloproliferative cases however, all were chronic myelomonocytic leukemias and most progressed to overt AML within one year, suggesting they in fact represent M4 or M5 AMLs with marked monocytic differentiation. Also, multilineage involvement and dysplastic features are frequent findings in NPMc+ AML⁵² making a distinction between NPMc+ AML and “NPM1-mutated MDS” difficult. Further, NPMc+ is closely associated with de novo AML, as AML secondary to myeloproliferative disorders/myelodysplasia and therapy-related AML rarely have NPM1 exon 12 mutations^6,53. This evidence indicates that NPM1 mutations and subsequent cytoplasmic dislocation of NPM is an event restricted to AML.

NPM1 mutations in AML

Immunohistochemical detection of cytoplasmic NPM in AML is predictive of NPM1 mutations⁵⁴. NPM1 mutations are characteristically heterozygous and result in frameshift mutations in exon 12⁷ (except two cases involving the splicing donor site of exon 9 and exon 11^55,56). There are currently 55 described mutations of NPM1 exon 12 in AML that result in similar alterations at the C-terminus of the mutant proteins.

Wild type NPM (wtNPM) contains two Crm1(exportin 1)-dependent nuclear export signal (NES) motifs, one within residues 94–102 and one at the N-terminus within amino acids 42–61^11,57. Wild type NPM also contains a nucleolar localization signal (NLS) at its C-terminus, which drives NPM from the cytoplasm to the nucleoplasm, then into the nucleolus via its nucleolar-binding domains⁵⁸. In normal cells NPM is predominately localized to the nucleus as the NLS greatly predominates over the relatively weak NES motifs⁵⁹. The majority of NPM1 exon 12 mutations encode mutant proteins that have a novel NES motif inserted at the C-terminus and have disruption of the NLS due to mutations of tryptophan residues 288 and 290 (or 290 alone)^7,60,61. In transfected cells, both alterations were found to be crucial for aberrant cytoplasmic localization of NPMc+ mutants⁶⁰.

There also appears to be a correlation between the type of NES motif inserted in the mutant and mutations of tryptophans 288 and 290. The most common NES motif (L-xxx-V-xx-V-x-L) is present in 29/55 described mutants including the three most frequent mutants, “A”, “B”, and “D”⁷. Most other mutants carry a NES motif with another hydrophobic amino acid replacing the valine at the second position. Specifically, 8/55 mutants have the L-xxx-L-xx-V-x-L motif, 5/55 carry the L-xxx-M-xx-V-x-L motif, 4/55 have the L-xxx-F-xx-V-x-L motif and 2/55 display the L-xxx-C-xx-V-x-L motif. One variant carries a NES in which leucine replaces valine at the third position (L-xxx-V-xx-L-x-L)⁵³. The most common NES motif (L-xxx-V-xx-V-x-L) has been shown to require loss of both tryptophans to be efficiently transported out of the nucleus, whereas the variant NES motifs can be exported efficiently despite retention of tryptophan 288, suggesting they are functionally stronger than the common motif^59,60. This suggests there is mutational selective pressure towards efficient transport into the cytoplasm, implicating this delocalization as a critical event for leukemogenesis.

In adult NPMc+ AML, mutation “A” (tandem duplication of TCTG) accounts for approximately 80% of all NPMc+ cases^53,62–65, whereas in children mutation “A” accounts for 11.1–50% of all NPMc+ cases^66–70. Therefore, in children, the variant NES motifs tend to occur more frequently compared to adults^66–68. This suggests that there are significant differences in the molecular mechanisms in NPMc+ AML in children compared to adults.

NPM1 mutation and leukemogenesis

Though it appears that the cytoplasmic localization is critical for leukemic transformation, the exact role of NPMc+ in leukemogenesis is unclear. One possible mechanism involves the interaction of NPMc+ and the ARF tumor suppressor. NPMc+ binds to and delocalizes ARF to the cytoplasm, inhibiting it from its interaction with HMDM2 thus preventing p53 initiation^71,72. In addition to perturbing p53-dependent activities of ARF, NPMc+ also disrupts ARF's p53-independent tumor suppressive activities, likely via destabalization of the ARF protein. Wild type NPM is capable of inhibiting ARF turnover, greatly increasing it stability. NPMc+ is capable of delocalizing wtNPM to the cytoplasm possibly preventing this stabilization of ARF^54,72. However, perturbed ARF function appears to be insufficient to explain the oncogenic effect of NPM1 mutations⁷² indicating other factors are involved. This could include perturbations of other wt NPM functions. This possibility is supported by the finding that NPM1 loss significantly affects genomic stability resulting in increased susceptibility to oncogenic transformation. Heterozygous knock-out mice (NPM1^+/−) develop hematologic features similar to those seen in humans with MDS. In fact, many of these mice go on to develop hematopoietic malignancy with myeloid malignancy being the most common^21,73. However, in these mice, numerical and structural chromosomal abnormalities were invariably found⁷³, whereas NPMc+ AML is seen predominantly in patients with normal karyotype and lacking recurrent chromosomal abnormalities suggesting it is not simply a similar loss of genomic stability the leads to oncogenic transformation.

Bonetti et al have recently suggested another possible mechanism by which NPMc+ could contribute to leukemogenesis. NPM regulates the turnover of the c-Myc oncoprotein by interacting with the F-box protein Fbw7γ, which is a component of the E3 ligase complex involved in the ubiquitination and proteasome degradation of c-Myc. NPMc+ led to c-Myc stabilzation by binding to and delocalizing Fbw7γ to the cytoplasm where it was degraded⁷⁴. Therefore, it is possible that NPMc+ facilitates c-Myc oncogene-induced hyperproliferation. Such hyperproliferation normally activates ARF- and p53-dependent cell cycle arrest or apoptosis⁷⁵. As NPMc+ also leads to perturbed ARF and p53 activity, it could also attenuate this normal cell cycle checkpoint. Thereby the single genetic mutation of NPMc+ could act to both activate proliferation and attenuate the ARF- and p53-dependent checkpoint response likely resulting in accelerated leukemogenesis without the need for futher cooperating genetic abnormalities.

Other factors contributing to the oncogenic effects of NPM1 mutations could include NPMc+ interactions with yet unknown partners. Further, it may be possible that other cooperating mutations may be present that contribute to leukemic transformation.

Cell of origin of NPMc+ AML

Research has been conducted in an attempt to identify the cell of origin of NPMc+ AML. Clonal NPM1 mutations affect different cell lineages in about 60% of NPMc+ AML⁵² and the leukemic mutant is absent in B and T lymphocytes of NPMc+ AML patients⁷⁶, suggesting either a common myeloid or an earlier progenitor without the ability to differentiate into lymphoid lineages is involved in NPMc+ AML. Also, NPMc+ AML is observed in a wide morphologic spectrum, explained by combinations and diverse ratios of NPMc+ leukemic clone-derived myeloid, monocytic, erythroid, and megakaryocyte cells as demonstrated immunohistologically⁵². This is a significant finding, as in the 2001 WHO AML classifications, NPMc+ was included in the category of AML not otherwise characterized. This category was subclassified based on FAB criteria, however with varying combinations and ratios of different lineages giving rise to a wide morphologic spectrum, NPMc+ AML is not characterizable by FAB criteria⁵².

Gene expression profiling of NPMc+ AML has shown an upregulation of several members of the homeodomain-containg family of transcription factors, including HOX and TALE genes^69,77,78. These genes are known to be important in stem-cell maintenance, further supporting the idea that NPMc+ AML may derive from a multipotent hematopoietic progenitor. MLL-rearranged leukemias also have been shown to have a gene expression signature with increased expression of multiple HOX genes⁷⁹, however different HOX genes are upregulated in NPMc+ AML compared to MLL-rearranged leukemia⁶⁹. In both NPMc+ AML and MLL-rearranged leukemia there is increased expression of HOXA4, A6, A7, A9 and B9 and the TALE genes MEIS1 and PBX3. However HOXB2, B3, B5, B6 and HOXD4 are overexpressed in NPMc+ AML but not MLL-rearranged leukemia⁶⁹, suggesting a different mechanism leading to HOX dysregulation. The mechanism by which NPM1 mutation leads to aberrant HOX expression is unclear. Perhaps NPM1 mutation directly influences HOX expression or possibly, NPM1 mutation leads to the arrested development of hematopoeitic precursors at a primitive stage at which HOX expression is elevated.

Recently, it has been suggested that microRNAs (miRNAs) may play a role in this up-regulation of such HOX genes. MiRNAs are 19–25 nucleotide noncoding RNAs that have been linked to the development of cancer^80–82. NPMc+ AML was found to have a unique miRNA signature that includes the up-regulation of miR-10a, miR-10b, miR-196a and miRNA-196b, several let-7 and miR-29 family members^83,84. Several miRNAs were also found to be down-regulated including miR-204 and miR-128a⁸³. While more study is needed, there is data that implicates an aberrant regulatory circuit of NPM1, HOX genes and miRNAs. This includes the fact that miRNAs 10a, 10b, 196a and 196b are all located within the genomic cluster of HOX genes⁸⁴. Further, miR-204 has been shown to target HOXA10 and MEIS1 suggesting that the HOX up-regulation seen in NPMc+ AML could be the result of loss of HOX regulation by miRNAs⁸³. Further, miRNA 196a directs translational inhibition of HOXB8 mRNA⁸⁵. If further study confirms these interactions, one could hypothesize that the interplay of miRNAs and HOX genes in NPMc+ AML could lead to the development of AML by causing an arrest of cellular differentiation of hematopoieic progenitors.

Clinical features of patients with NPMc+ AML

Several clinical features have been shown to be significantly associated with NPMc+ AML. Studies have shown the NPM1 mutations have an increased incidence with increasing age. The incidence of NPM1 mutations in AML is significantly higher in adult AML compared to pediatric AML. In adult studies including over 4,300 patients the overall frequency of NPM1 mutations was 31.4% (range 25.4 to 41%)^{6,53,62,64,86–88}. In studies including over 900 pediatric AML patients, the frequency of NPM1 mutations was 7.5% overall (range 0 to 12%)^{66–70,86,89}. Further, in adult studies, several authors have reported a higher median age in patients with NPM1 mutations^{6,62,64,87,88,90}. Pediatric studies have also shown a trend towards higher age and increased frequency of NPMc+ AML, with one study reaching statistical significance⁶⁹, but others not^66,67,70. However, in the two largest pediatric studies, there was a conspicuous absence of NPM1 mutations in children under 3 years (0/126) with a steady incidence of approximately 10% in children 3 years and above^67,68. After the age of 21 years, the incidence appears to increase steadily such that approximately 15% of patients 21–35 years of age, 40% of patients 35–60 years of age and 50% of patients 60 years of age and older harbor NPM1 mutations⁶⁴. This data suggests that the risk of acquiring NPM1 mutation in a myeloid stem/progenitor cell is cumulative. The striking absence of NPM1 mutation in the youngest children suggests that latency between the acquisition of NPM1 mutation and the acquisition of the cooperating mutation(s) required for leukemic transformation is on the order of years. Alternatively, myeloid stem/progenitor cells in young children may be relatively resistant to the acquisition of NPM1 mutations.

Several authors have reported a higher incidence of NPM1 mutations in adult female AML patients compared to males^53,63,65,88. This female preponderance has also been found in pediatric populations as well^67,68,70. However, other adult and pediatric studies have not found a significant difference in frequency of NPM1 mutations between sexes^{6,62,64,66,69,86,87,90}.

There also appears to be ethnic differences in the frequency of NPM1 mutations with Asian populations having significantly lower frequencies. In one pediatric study, there were no NPM1 mutations found in 33 Japanese children with NK AML⁸⁹. Another study of 47 Taiwanese children found the incidence of NPM1 mutations was only 2.1%⁸⁶. In contrast, the frequency of NPM1 mutations in children with AML from Europe and the United States ranges from 6.4 to 8.4%^66–69. This ethnic difference has also been seen in adults with studies of patients from China and Japan reporting incidences ranging from 14.3 to 28.2%^50,62,86,87 compared to approximately 35% of patients of European descent^6,64,88.

NPM1 mutations have been found in all FAB morphologic subtypes of AML. However, of studies that included patients with M3 AML, only one patient with M3 AML was found to have an NPM1 mutation⁶⁴, whereas all other studies found no cases of M3 AML with NPM1 mutations^{6,53,62,69,86,87}. In adult studies, NPM1 mutations occur most frequently in M4 and M5 AML^{6,53,62–65,88,90}. In children this distribution is somewhat different. In one study of 107 children with AML, NPM1 mutations were found in 1/21 patients with M1, 2/19 with M2, 3/13 with M4, 0/18 with M5 and 1/3 with M6⁶⁶. Other pediatric studies have also found an absence of NPM1 mutations in M5 AML^68–70. This morphologic discrepancy can possibly be explained by the fact that in children, most myelomonoblastic AML occurs in very young children with MLL rearrangements⁹¹, which are essentially mutually exclusive of NPM1 mutations.

Other clinical features that have been found to be associated with NPM1 mutations include a higher white blood cell count at presentation^{53,62–65,86–88,90} and higher blasts percentage at diagnosis^53,62,63,86. Some authors have found a significantly higher platelet count at the time of diagnosis in patients with NPM1 mutations^53,63,86. Lower expression of CD 34 and CD 133 have been consistently found in NPM1 mutant blasts.^{6,53,63–66,69,86,87} NPMc+ AML has also been associated with significantly higher incidence of extramedullary involvement, mostly accounted for by gingival hyperplasia and lymphadenopathy⁶³. This finding likely reflects the fact that this pattern of dissemination is most often seen in patients with monocytic AML (M4/M5), which are the FAB subtypes with the highest incidence of NPM1 mutations ^7,92.

Relationship of NPMc+ AML to karyotype, cytogenetic, and molecular abnormalities

NPM1 mutations have been shown to be significantly more common in patients with normal karyotype (NK) AML. As stated above, the frequency of NPM1 mutations in adult AML patients is significantly higher than in children with AML. However the difference in frequency between adults and children cannot be accounted for by the higher incidence of normal karyotype in adult AML. In adults with NK AML the frequency of NPM1 mutations in 10 studies ranged from 45.7 to 64%^{6,53,62–65,86–88,90} whereas in children NPM1 mutations were present in 0 to 26.9% with NK AML^{66–70,86,89}. In patients with karyotype abnormalities NPM1 mutations were present in 8.5 to 19% of adults^{6,53,62,64,86–88} and 0 to 4% of children^66–70,86. In addition to being associated with normal karyotype, NPM1 mutations have also been shown to be essentially mutually exclusive of recurrent genetic abnormalities ^{6,7,53,63–67,70,86,88,90,93}.

A focus of many studies has been the analysis of the distribution of NPM1 mutations within AML with different molecular mutations. This investigation is relevant, as according to the two-hit hypothesis of leukemogenesis, at least two different mutation types are necessary for leukemic transformation, type I which activate signal-transduction pathways and confer a proliferative advantage, and type II which affect transcription factors and cause a block in differentiation⁹⁴. Several studies have found the MLL-PTD mutations are exceedingly rare in patients with NPM1 mutations^6,53,63,65. Many studies have found no difference in frequency of mutations in the transcription factor CCAAT/enhancer binding protein-α (CEBPA) in patients with and without NPM1 mutations^63–65. Other studies have found that CEBPA mutations are significantly less frequent in patients with NPM1 mutations^86,90. No correlations was found between NPM1 mutations and the frequency of KIT mutations⁶⁵ or TP53 mutations⁶². Some studies have shown no significant association between NPM1 mutations and NRAS mutations^62,65,86, whereas one study found a negative correlation⁶⁴. Mutations in KRAS were found in one study to occur more frequently in patients with NPM1 mutations⁶⁴, however another study found no significant correlation⁸⁶.

In the first report of NPMc+ AML, Falini et al found a high frequency of fms-like tyrosine kinase-3 (FLT3) internal tandem duplications (ITD) in patients with NPMc+ AML⁶. This finding has been confirmed by both adult studies^{53,62–65,86,88} and studies of pediatric patients^67,68,70,86. In adults and children, approximately 40% of patients with NPMc+ AML also have FLT3/ITD mutations compared to approximately 15–25% in patients with wt NPM^{6,53,63–65,70,86,88,90}. It is possible that these mutations are secondary events from a primary process that predisposes a myeloid stem/progenitor cell to errors in DNA replication. Supporting this possibility is the fact that many of the NPM1 mutants, including the most common mutant “A”, are a 4 base pair ITD, the same type of mutation that occurs in FLT3 to cause the FLT3/ITD.

One study found a significantly higher NPM1 mutant/wild type ratio than the FLT3-ITD/wild type ratio, suggesting that NPM1 mutations occurred prior to FLT3/ITD mutations in cases with both⁵³. To further support this hypothesis, cases with several FLT3-mutant bands (found in 15–20% of patients with FLT3/ITD mutations) were evaluated and found to have only a single NPM1 mutation, which suggest the FLT3/ITD mutations evolved from a single NPM1 mutated clone⁵³.

Some studies have found a positive correlations between NPM1 mutations and point mutations in the activation loop of the kinase domain of FLT3 (FLT3-PM)^{53,62,63,65,95}. One recent large study of adult AML patients found that FLT3-PM were significantly overrepresented in patients with NPM1 mutations, as 8.8% of patients with NPM1 mutations also had FLT3-PM compared to 4.8% in the entire population⁹⁵. Other studies have found no correlation^6,64.

Recently, PTPN11 mutation in adult AML has been found to be closely associated with NPM1 mutation. Six out of 13 patients with PTPN11 mutations had concurrent NPM1 mutations. In the same study, PTPN11 and FLT3/ITD were mutually exclusive⁹⁶.

Overall, these findings are compatible with the two-hit theory of leukemogenesis. NPM1 mutations likely represent a type II mutation as they occur with type I mutations such as PTPN11 and FLT3/ITD and are mutually exclusive of other type II mutations.

Prognostic significance of NPM1 mutations in AML

Currently, cytogenetic analysis at AML diagnosis, to a certain extent, allows for risk stratification into favorable, adverse, and intermediate risk groups¹. Given that the majority of NPMc+AML cases have normal karyotype and lack recurrent cytogenetic abnormalities, they are included in the large heterogenous group of AML patients in the intermediate risk group. Studies of both adults and children have suggested that NPM1 mutations may be a novel molecular marker with prognostic relevance.

In patients with NK AML, NPM1 mutation has been associated with higher rate of complete remission (CR) compared to NK AML with wt NPM^{6,53,62,65,88}. There is data to suggests that NPMc+ AML has an increased sensitivity to chemotherapeutic agents secondary to its interaction with nuclear factor κB (NK-κB)⁹⁷. NK-κB is a transcription factor that contributes to the maintenance and survival of malignant clones and impaired response to chemotherapy⁹⁸. NPMc+ has been shown to interact with and sequester NK-κB in the cytoplasm leading to its inactivation and reduced DNA binding⁹⁷.

Many studies have found a significant association between FLT3/ITD mutational status and NPM1 status in terms of CR rate, event-free survival (EFS), disease-free survival (DFS), and overall survival (OS). Numerous adult studies have demonstrated that the group of NK AML patients with NPM1 mutation lacking FLT3/ITD mutation (NPMc+/ITD−) have the best CR rate compared to those with wtNPM and those with FLT3/ITD mutations^53,63,88. For example, in a study of 1217 adults with AML, for patients with NPMc+/ITD− the CR rate was 94% compared to 78% in patients lacking both mutations and 87% in patients carrying both mutations (p<0.00001)⁸⁸. Further, in adults, a significantly better EFS and DFS has been found in NPMc+/ITD− patients^53,64,88. Relapse-free survival (RFS) in adults with NK AML is also better in the NPMc+/ITD− group compared to other NK AML patients^53,63,65,88. Importantly, in patients with NK AML with NPMc+/ITD− status, the OS and RFS were significantly better compared to other groups, and the availability of a matched sibling hematopoeitic stem cell transplant (HSCT) had no effect on OS or RFS⁶³, indicating that perhaps these patients should be exempt from this treatment modality in first-line therapy. A large study of adult patients found that EFS was superior in the group of patients with NPM1 mutations and FLT3-PM, and this cooperative positive effect was even stronger in patients with NK de novo AML lacking FLT3/ITD mutation⁹⁵.

In pediatric studies, CR rate has not been shown to be significantly different for patients with or without NPM1 mutations^66–68. However, this can perhaps be accounted for by the small numbers of patients with NPM1 mutations and the relatively high rate of CR overall in children compared to adults with AML. In our study of pediatric AML patients, there was a trend towards better 5 year EFS in children with NPM1 mutations compared to those with wt NPM (50% and 33% respectively), however this did not reach statistical significance⁶⁷. Further, when considering FLT3/ITD status, those patients with NPMc+/ITD− status achieved the best 5 year EFS and OS (69±13% and 77±12% respectively) compared to 35±3% and 51±4% in patients negative for both, 25±15% and 25±15% in patient with both mutations, and 21±7% and 34±7% in patient with FLT3/ITD mutations and wt NPM⁶⁷. Hollink et al found similar results in their study of 297 pediatric patients with AML with patients with NPM1 mutations doing significantly better than those with wt NPM in terms of 5 year OS and EFS. However, they found that FLT3/ITD status had no significant impact on survival in patients with mutations and was only prognostically significant in patients with wt NPM1⁶⁸. These results must be interpreted with caution however as there were only 25 total patients with NPMc+ AML and only 10 patients with both a NPM1 mutation and FLT3/ITD mutation. All these pediatric survival data however are important, as the 5 year EFS and OS of those patients in the NPMc+/ITD− group are comparable to patients with the favorable cytogenetic abnormalities of inv (16) and t(8;21)^67,68, suggesting that this group of patients could be identifiable as a favorable risk group.

Monitoring minimal residual disease in AML with NPM1 mutations

NPM1 mutation has shown great stability during disease evolution^62,68,86,90 implicating it as potentially useful marker for MRD detection. Using quantitative polymerase chain reaction (PCR) assays for NPM1 mutations, two studies have shown that the number of NPM1 mutated copies closely correlate with clinical status and can predict impending hematologic relapse^99,100. Therefore, quantitative PCR assays for NPM1 mutations appear to be able to monitor and quantify for MRD and may have prognostic impact in AML patients with NPM1 mutations.

Future Directions

NPMc+ AML has been well characterized in terms of its clinical features and prognostic relevance. However, there is still much to be learned about this frequent genetic alteration. While there is strong evidence to suggest cytoplasmic localization of NPM is critical for leukemic transformation⁵⁹−⁶¹, the exact role NPMc+ plays is unknown. We know that NPMc+ interferes with ARF and p53^72,101 function and causes an increase in the oncoprotein cMyc⁷⁴, which likely contribute to leukemic transformation however, given NPM's diverse functional repertoire other yet unidentified proteins may also be involved. More research is needed to clarify how mutated NPM1 promotes leukemogenesis, likely through the identification of putative partners.

NPM1 mutations and Flt3 ITD mutations frequently occur together, therefore one might speculate that the two cooperate to cause leukemic transformation. However, a mechanistic link has yet to be established. Research should focus on developing an experimental model to investigate the interaction between these two frequently coexisting mutations. Evidence exists to suggest that PTPN11 mutations might also be closely associated with NPM1 mutation and therefore possibly synergistically important⁹⁶, however more investigation is necessary. Research should also continue to search for other possible cooperating mutations, the identification of which might help further unravel the question of how NPMc+ contributes to leukemic transformation.

Given the frequency with which NPM1 is mutated in AML, research should be aimed at devising a molecularly targeted therapeutic approach, which could dramatically improve patient outcome. NPMc+ is an alluring target for molecularly directed therapy. With small molecule inhibitors, it may be possible to kill leukemia cells with NPM1 mutations by specifically targeting NPMc+ while leaving wtNPM undisturbed. Also, given the frequent coexistence of NPMc+ and Flt3 ITD, it may be possible to utilize existing small molecule inhibitors of Flt3 to revert NPMc+/Flt3 ITD patients into the more favorable prognostic category of patients with NPMc+ alone. Further, it is unclear why NPMc+ AML is more sensitive to conventional chemotherapy and confers a more favorable prognosis. Identifying possible mechanisms that result in improved outcome could point toward potential targets to augment chemosensitivity in AML and perhaps other malignancies.

In the WHO classification, entities are diseases that are clearly defined, clinically distinctive and non-overlapping. Most of the diseases described are distinct entities, however some are not as clearly defined and are listed as provisional entities. In the current WHO classifications, NPMc+ has been recognized as a provisional entity secondary to its relatively recent description¹⁰². More work is necessary to confirm NPMc+ AML as a true disease entity with distinctive clinical and biological features.

Table 1.

Comprehensive List of Reported NPM1 mutations

Mutation	DNA sequence of exon 12 of NPMl gene											Protein
WT	gatctct		g		gcag		t		ggagg		aagtctctttaagaaaatag	DLWQWRKSL
A	gatctct		g	TCTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCLAVEEVSLRK
B	gatctct		g	CATG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCMAVEEVSLRK
C	gatctct		g	CGTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCVAVEEVSLRK
D	gatctct		g	CCTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCLAVEEVSLRK
E	gatctct		g		gcag		t	CTCTTGCCC			aagtctctttaagaaaatag	DLWQSLAQVSLRK
F	gatctct		g		gcag		t	CCCTGGAGA			aagtctctttaagaaaatag	DLWQSLEKVSLRK
E*	gatctct		g		gcag		t	CCCTCGCCC			aagtctctttaagaaaatag	DLWQSLAQVSLRK
G*	gatctct		g		gcag		t	GCTTCGCC			aagtctctttaagaaaatag	DLWQCFAQVSLRK
H*	gatctct		g		gcag		t	GTTTTTCAA			aagtctctttaagaaaatag	DLWQCFSKVSLRK
J	gatctct		g		gcag		t	CTCTTTCTA			aagtctctttaagaaaatag	DLWQSLSKVSLRK
L	gatctct	CCCG	g		gcag		t				aagtctctttaagaaaatag	DLSRAVEEVSLRK
K	gatctct		g		gcag		t	CCCTTTCCA			aagtctctttaagaaaatag	DLWQSLSKVSLRK
M	gatctct		g	TAGC	gcag		t		ggagg		aagtctctttaagaaaatag	DLCTAVEEVSLRK
N	gatctct		g	CCAC	gcag		t		ggagg		tctctttaagaaaatag	DLCHAVEELSLRK
O	gatctct		g		gcag	CGTTTCC			agg		aagtctctttaagaaaatag	DLWQRFQEVSLRK
P	gatctct		g	TACCTTCC			t		ggagg		aagtctctttaagaaaatag	DLCTFLEEVSLRK
Q	gatctct		g		gcag	AGGA	t		ggagg		aagtctctttaagaaaatag	DLWQRMEEVSLRK
Gm	gatctct		g	CAGG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCRAVEEVSLRK
Km	gatctct		g	CCGG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCRGVEEVSLRK
Lm	gatctct		g	CCGCGG	ag		t		ggagg		aagtctctttaagaaaatag	DLCQAVEEVSLRK
Nm	gatctct		g	CCAG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCLAVEEVSLRK
Om	gatctct		g	TTTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCRAVEEVSLRK
Qm	gatctct		g	TCGG	gcag		t		ggagg		tctctttaagaaaatag	DLWQSMEEVSLRK
1	gatctct		g		gcag	TCCA	t		ggagg		aagtctctttaagaaaatag	DLCHAVEEVSLRK
3	gatctct		g	TCAT	gcag		t		ggagg		aagtctctttaagaaaatag	DLCLAVEEVSLRK
4	gatctct		g	CTTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLWQDFLNRLLFKKIV
6	gatctct		g		gca	AGATTTCTTAATTC					gtctctttaagaaaatag	DLCLAVEEVSLRK
7	gatct	ATGC			gcag		t		ggagg		aagtctctttaagaaaatag	DLCLAVEEVSLRK
12	gatctct		g	CGCC	gcag		t		ggagg		aagtctctttaagaaaatag	DLCAAVEEVSLRK
13	gatctct		g	TAAG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCKAVEEVSLRK
10	gatctct		g		gcag		tg	CTGCTCCC			aagtctctttaagaaaatag	DLWQCCSQVSLRK
14	gatctct		g		gcag		t	TATTTTCCC			aagtctctttaagaaaatag	DLWQCCSQVSLRK
G†	gatctct		g	TTTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCLAVEEVSLRK
H†	gatctct		g	CTTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCLAVEEVSLRK
I†	gatctct		g	TAAG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCKAVEEVSLRK
J†	gatctct		g	TATG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCMAVEEVSLRK
I*	gatctct		g	CAGA	gcag		t		ggagg		aagtctctttaagaaaatag	DLCRAVEEVSLRK
S	gatctct		g	CAAA	gcag		t		ggagg		aagtctctttaagaaaatag	DLCKAVEEVSLRK
R	gatctct		g		gcag		t	CTTTCTCCC			aagtctctttaagaaaatag	DLWQSFSQVSLRK
DD2	gatctct		g	CTGG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCWAVEEVSLRK
DD4	gatctct		g	TGTG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCVAVEEVSLRK
DD5	gatctct		g	TCAG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCQAVEEVSLRK
DD10	gatctct		g		gcag	AGAA	t		ggagg		aagtctctttaagaaaatag	DLWQRMEEVSLRK
DD11	gatctct		g		gcag	AGAC	t		ggagg		aagtctctttaagaaaatag	DLWQRLEEVSLRK
DD12	gatctct		g		gcag	CGCT	t		ggagg		aagtctctttaagaaaatag	DLWQRLEEVSLRK
DD13	gatctct		g		gcag	GGGGTGGGGAATC					tctctttaagaaaatag	DLWQGVGNLSLRK
JH1	gatctct		g	TACG	gcag		t		ggagg		aagtctctttaagaaaatag	DLCTAVEEVSLRK
JH2	gatctct		g		gcag		C		gga	TGGCCg	aagtctctttaagaaaatag	DLWQRMAEVSLRK
*JH3	gatctct		g	CGGA	gcag		t		ggagg		aagtctctttaagaaaatag	DLWGAVGEVSLRK
JH4	gatctct		g		gcag		C		gga	TTCCgg	aagtctctttaagaaaatag	DLWQRIPEVSLRK
JH5	gatctct		g		gcag	CGTT	C		ggagg		aagtctctttaagaaaatag	DLWQRSEEVSLRK
*JH6	gatctct		g		gcag		t		gga	TGGAgg	aagtctctttaagaaaatag	DLWQWMEEVSLRK
*JH7	gatTtTt		g		gcag		G		ggagg		aagtctctttaagaaaatag	DFWQGRKFF
Hollink et al1	gatctct		g		gc	TCCGATTTGC			gg		aagtctctttaagaaaatag	DLWLRFAEVSLRK
Hollink et al2	gatctct		g		gcag		t	ATCTGGGGGCCC			tctctttaagaaaatag	DLWQYLGALSLRK