Somatic mutations in the nucleophosmin (NPM1) gene occur in approximately 30% of de novo acute myeloid leukemias (AMLs) and are relatively enriched in normal karyotype AMLs. Earlier World Health Organization (WHO) classification schema recognized NPM1-mutated AMLs as a unique subtype of AML, while the latest WHO and International Consensus Classification (ICC) now consider NPM1 mutations as AML-defining, albeit at different blast count thresholds. NPM1 mutational load correlates closely with disease status, particularly in the post-therapy setting, and therefore high sensitivity-based methods for detection of the mutant allele have proven useful for minimal/measurable residual disease (MRD) monitoring. MRD status has been conventionally measured by either multiparameter flow cytometry (MFC) and/or molecular diagnostic techniques, although recent data suggest that MFC data may be potentially more challenging to interpret in this AML subtype. Of note, MRD status does not predict patient outcome in all cases, and therefore a deeper understanding of the biological significance of MRD may be required. Recent studies have confirmed that NPM1-mutated cells rely on overexpression of HOX/MEIS1, which is dependent on the presence of the aberrant cytoplasmic localization of mutant NPM1 protein (NPM1c); this biology may explain the promising response to novel agents, including menin inhibitors and second-generation XPO1 inhibitors. In this review, these and other recent developments around NPM1-mutated AML, in addition to open questions warranting further investigation, will be discussed.

© 2023 The Author(s). Published by S. Karger AG, Basel

Introduction

Nucleophosmin (NPM1) is a ubiquitously expressed nuclear and nucleolar phosphoprotein, which transits between those subcellular compartments and the cytoplasm to execute its normal functions [1]. Among its many roles in normal cell biology, NPM1’s regulation of genome stability, tumor suppressor proteins (e.g., p53, ARF), and the DNA damage response are some of the most well studied [2–7]. However, many of these processes have only been studied in vitro and therefore still warrant further investigation using in vivo model systems.

NPM1 has well-documented roles in neoplastic disease, including both solid tumor [8–17] and hematological malignancies [18]. While wild-type protein overexpression has been identified in several solid tumors, hematological malignancies have been variably linked to NPM1 deletions [19, 20], translocations [21–26], and somatic mutations [27], the latter two of which are unified by the aberrant cytoplasmic dislocation of residual wild-type NPM1 and/or its mutant counterpart. Approximately 30% of all acute myeloid leukemias (AMLs), and ∼60% of cases lacking karyotypic abnormalities (normal karyotype [NK]-AML), harbor pathogenic somatic mutations at the C-terminus of NPM1, resulting in a mutant protein product (NPM1c) that has long been thought to underlie the pathogenesis of NPM1-mutated AML [27, 28]. Some of the mechanisms by which NPM1c drives leukemogenesis have only been identified over the last several years [29–32], although our understanding of its likely numerous pathogenic functions has remained largely incomplete; very recent data will be summarized later in this review. Clinically, AML with mutated NPM1 has typically been considered to have a favorable outcome profile, in the absence of other high-risk features such as co-mutation in genes including DNA methyltransferase 3A (DNMT3A) and/or FMS-like tyrosine kinase 3 internal tandem duplications (FLT3-ITDs) [33–36]. However, some recent studies have suggested that adverse risk is specifically associated with the triple mutation pattern rather than, e.g., the more frequently encountered co-mutation in NPM1 and DNMT3A alone [37].

Over the last decade, NPM1 mutations in AML have progressed from recognition as defining of a unique disease biology to their current status as AML-defining at lower blast count thresholds than would otherwise be required for an AML diagnosis [18, 38, 39]. Clinical investigation of NPM1-mutated AML has been additionally intense given the role of mutant allele tracking for assessment of minimal/measurable residual disease (MRD), a typical hallmark of higher risk disease. Nevertheless, there remain open questions regarding the implication of detectable MRD in individual patients, and therefore, additional work may be required to better understand the biological significance of this finding. In this review, the evolving clinicopathologic features of NPM1-mutated AML, and some of the remaining open questions related to disease pathogenesis, will be discussed.

Pathogenic NPM1 Mutations in Myeloid NeoplasiaA Brief Overview

There are three major NPM1 transcript variants, the longest of which (transcript 1, i.e., NPM1) comprises 11 coding exons and produces a 294 amino acid protein. In a search for NPM1-related aberrations in myeloid neoplasms (MNs), Falini and colleagues were the first to identify abnormal cytoplasmic NPM1 expression in a large proportion of AML samples using chromogenic immunohistochemistry (IHC); using polymerase chain reaction (PCR) studied, they link the aberrant cytoplasmic protein expression to recurrent 4 base pair insertion mutations (indels) at exon 11 of the C-terminus of NPM1[27]. Notably, while the canonical lesion is a 4 base pair indel, variant insertions at exon 11, insertions at other regions of the gene such as exon 5, and rare gene fusions, have also been described in AML cases [40]. Falini and others have also demonstrated frequent associations between NPM1 mutations, myelomonocytic phenotypes, NKs, and co-occurring mutations in methylation pathway genes [18, 27, 28, 33, 41].

Wild-type NPM1 contains several domains, among which are an N-terminal oligo/heterodimerization region, centrally located DNA binding domains, one nuclear export signal (NES) sequence, and two nuclear/nucleolar localization sequences. C-terminal insertion mutations in NPM1 replace one of the critical nucleolar localization signals with an additional NES. Thus, the unifying feature of pathogenic somatic mutations in NPM1 is aberrant cytoplasmic dislocation of the mutant protein, which can also sequester residual wild-type NPM1 into the cytoplasm through heterodimerization, thereby producing a haploinsufficient state [27, 42, 43]. It has been hypothesized that leukemogenesis results from the combinatorial effect of losing wild-type NPM1 in its normal nuclear/nucleolar position, and potentially neomorphic functions of NPM1c, potential mechanisms that have been under active investigation.

Diagnosis of NPM1-mutated AML, by definition, has required the detection of a pathogenic somatic mutation in a patient otherwise meeting conventional criteria for an AML diagnosis (i.e., >20% blasts in the peripheral blood or bone marrow) [38, 39]. However, there are many characteristic clinicopathologic features of this leukemia subtype that can assist in preparing a differential diagnosis prior to generation of the required molecular data: frequent associations with monocytic or myelomonocytic cytologic features and phenotype, the presence of “cup-like” blasts, absence of CD34 and/or human leukocyte antigen (HLA)-DR expression, and a NK or at least a lack of complex cytogenetic lesions [41] (Fig. 1). IHC, as described in the original report by Falini and colleagues, can also be a tremendously useful and cost-effective tool, especially in resource-poor settings and for the detection of extramedullary disease (i.e., myeloid sarcoma); while N-terminally directed antibodies can be used to detect aberrant cytoplasmic staining, mutant protein-specific antibodies have been developed [27, 41, 44–47].

Fig. 1.

A representative case of NPM1-mutated AML. a Peripheral blood smear preparation showing frequent abnormal large blasts (∼93% of nucleated cells by manual differential) with high N:C, irregular to occasionally folded nuclei, open chromatin pattern, and occasional “cup-like” nuclear invagination (inset) [Wright-Giemsa, ×1,000]. b Multiparameter flow cytometry (MFC) performed on the peripheral blood detected a predominant population in the conventional CD45+ (dim)/low side scatter (SSC) blast gate (∼88% of viable cells), which is predominantly positive for CD117 with only dim/partial expression of CD34. Color scheme for gated populations: mature lymphocytes (green), monocytes (orange), granulocytes (aqua), “blast” gate (red). Axes are on log scale, and major hashmarks denote 102–105. c Histomorphologic evaluation of the corresponding Bouin-fixed paraffin-embedded trephine biopsy shows a nearly ∼100% cellular marrow, markedly hypercellular for the patient’s age (H&E, ×40). d Higher magnification reveals cellular composition by a predominant population of medium to large sized atypical blasts (H&E, ×600). e The abnormal blasts exhibit cytoplasmic positivity for mutant NPM1 protein by chromogenic immunohistochemistry (Fast red, antibody clone: PA1-46356 [Thermo Fisher], ×600).

Recent Developments

While the unique transcriptional features associated with NPM1 mutation have been well established [48], until recently the potential functions of NPM1c have remained largely unknown. Brunetti and colleagues were the first to demonstrate that the transcriptional program of NPM1-mutated cells, including features such as HOX gene upregulation, is entirely dependent on the presence of NPM1c [30]. These data led to hypothesized roles for NPM1c, including a neomorphic transcription factor-like function in driving the observed transcriptional program; nevertheless, definitive proof remained elusive.

In very recent parallel studies, the groups of Uckelmann and Wang have demonstrated that NPM1c directly binds to chromatin [49, 50]. Uckelmann and colleagues showed that NPM1c binds specific chromatin targets and works in collaboration with the histone methyltransferase KMT2A (MLL1) to drive expression of RNA polymerase II and the characteristic transcriptional program in NPM1-mutated cells, including HOX gene overexpression; importantly, targeted degradation of NPM1c abrogated its neomorphic activity, as evidenced by loss of RNA polymerase II expression and activating histone modifications at its chromatin targets [50]. Importantly, these data provide a mechanistic explanation for the observed preclinical efficacy of menin inhibition in NPM1-mutated AML (reviewed later) [51, 52]. Similarly, Wang and colleagues demonstrated that NPM1c binds to a subset of active gene promoters, including those associated with HOXA/B cluster genes and MEIS1, and is critical in establishing a transcriptional hub to drive the expression of the downstream genes; in addition, NPM1c was shown to maintain an active chromatin state through inhibition of histone deacetylases. These data provide complementary evidence of a neomorphic role for NPM1c and additional rationale for the development of targeted approaches to disrupt the leukemogenic transcription factor network orchestrated by NPM1c.

Classification of NPM1-Mutated MNs in 2022World Health Organization versus International Consensus Classification

NPM1 mutations are well-recognized as defining of a unique biological subtype of AML [18, 53]. Earlier studies concluded that NPM1 mutations are rare in chronic myeloid proliferations, such as myelodysplastic syndromes and myelodysplastic/myeloproliferative overlap states, based on detection in only ∼1–5% of cases [54–63]. Although rare, evaluation of such cases can be additionally complicated given that NPM1-mutation can also be associated with multilineage dysplasia [64]. However, until very recently, there remained much debate about whether or not NPM1 mutations could alone be used to render a diagnosis of AML in the absence of a blast count at or above the standard threshold of 20% in either the blood or bone marrow. More recent studies using larger cohorts of non-acute NPM1-mutated MNs (NPM1+ MNs) have independently demonstrated that NPM1-mutated disease has common biological features regardless of the blast percentage; moreover, clinical outcome data suggest that NPM1-mutated myeloid disease may uniformly benefit from AML-type therapy [65, 66]. Based largely on these data, the latest 2022 World Health Organization (WHO 5th Ed.) and International Consensus Classification (ICC) diagnostic schema have uniformly resolved the ongoing debate by lowering the blast count threshold required for a diagnosis of AML with mutated NPM1; however, the two consortia remain incompletely aligned and therefore the nuances in their diagnostic guidelines warrant focused discussion, particularly because either criteria applied to the same patient may result in a diagnosis of AML by one scheme but not the other [38, 39].

Per the WHO 5th Edition criteria, AML with NPM1 mutation can be diagnosed irrespective of the blast count, although the basis for the lack of a blast count threshold has not been provided in the preliminary guidelines [38]. Such loose criteria may create diagnostic dilemmas that will need to be overcome. For example, the lack of a single gold standard assay for detecting NPM1 mutation, and variable inter-assay sensitivity, has the potential to dramatically alter management for a patient depending on the institution where they are evaluated. Next-generation sequencing (NGS)-based assays, although now widely available and routinely used to profile treatment-naïve patients, can vary in their lower limit of variant allele detection for individual somatic mutations, in the range of ∼2–5%; as a result, the same patient evaluated in two different centers may receive an AML diagnosis in one, but not the other. Furthermore, it remains unclear whether patients harboring such low levels of NPM1 mutant allele burden will exhibit the same clinical behavior and/or whether AML-directed therapy would be as beneficial in patients with very low blast counts and/or NPM1 mutant allele burden. Prospective studies will be needed to resolve the management dilemmas that are likely to arise.

Conversely, the ICC has established a blast count threshold of ≥10% in blood or bone marrow as a requirement to render a diagnosis of AML with mutated NPM1. Although the basis for this threshold is also not thoroughly described, it may have been selected based on the separate studies led by Patel and Montalban-Bravo, where the median bone marrow blast count in non-acute NPM1-mutated MNs was found to be approximately 10% in both cohorts. It is important to note that blast counting is inherently subjective, and therefore, an individual patient may be classified differently by different practitioners and/or laboratories (i.e., with respect to the 10% threshold), resulting in the same issues for downstream patient management that may be associated with variability in molecular methodology for mutation detection. Issues associated with inter-assay variability (e.g., limit of variant allele detection) for detection of the NPM1 mutation may be less problematic at this higher blast count threshold, as the NPM1 mutant allele has been noted to correlate with both the peripheral blood and bone marrow blast percentage in prior studies [66, 67]. However, the accrual of additional data will be required to determine whether or not patients in the ∼5–10% blast count range may also benefit from more intensive therapy. Overall, it is evident that a more precise definition or threshold (e.g., variant allele frequency) may be required in the future, to ensure greater consistency between laboratories and individual practitioners in establishing a diagnosis of NPM1-mutated AML.

Therapy-Related MNs: Where Does NPM1 Mutation Fit In?

Therapy-related MNs (t-MNs), per the revised 4th edition WHO criteria, represented a separate diagnostic category comprising any myeloid malignancies occurring in patients with history of prior exposure to chemotherapy and/or radiotherapy [18]. Therapy-related AMLs account for approximately 7–10% of AML cases [68]. This designation of t-MN is clinically impactful, because such patients are considered to have high-risk disease warranting aggressive therapy as a bridge to allogeneic hematopoietic stem cell transplantation (HSCT), if otherwise fit. Although characteristically associated with complex karyotypes and pathogenic somatic mutations in DNA repair pathway genes, including TP53 and PPM1D, it has been recognized that the revised 4th edition WHO-defined criteria for a t-MN diagnosis may on occasion lead to spurious misclassification of a de novo AML otherwise lacking high-risk biological features. Very recent work comparing NK AMLs, otherwise classified as de novo or therapy-related based on clinical history, has revealed that NK t-AMLs may have an “intermediate” clinical behavior between NK de novo AMLs and t-AMLs that have more classical high-risk genetic features [69]. There remains uncertainty regarding how best to manage patients with classically defined t-AML lacking canonical high-risk genetic characteristics (e.g., complex karyotypes, TP53/PPM1D mutations), given the genetic heterogeneity among such cases. Furthermore, the frequency of this challenging scenario is likely to increase in parallel with the increasing survival rates for many other cancer types. To address this confusion, both the 5th edition WHO and ICC schema have eliminated the prior t-MN category, favoring the use of “post-cytotoxic therapy” (WHO) or “therapy-related” (ICC) as qualifiers for AMLs and MNs that are otherwise defined using separate diagnostic criteria [38, 39].

Given the frequency of NPM1 mutations in de novo AML, and the unique clinicopathologic features of this AML subtype, it has remained critically important to determine the significance of NPM1 mutation occurring in the setting of prior therapy. In a recent study, Othman and colleagues compared the clinical and genetic features of therapy-related (as previously defined) NPM1-mutated AML, de novo NPM1-mutated AML, and therapy-related (as previously defined) AML with wild-type NPM1[70]. Unlike “conventional” t-AML lacking NPM1 mutation, NPM1-mutated AMLs, whether “therapy-related” or de novo, shared several clinicopathologic features: high frequency of NKs, DNMT3A and/or TET2 co-mutation, consistently wild-type TP53 and PPM1D, similar transcriptional profiles (e.g., upregulation of HOX, downregulation of CD34 and CD133), and perhaps most importantly, similar rates of relapse-free and overall survival. These data appear to conclusively demonstrate that NPM1 mutation is a hallmark of de novo AML, regardless of a history of therapy for an unrelated malignancy.

Minimal/Measurable Residual Disease

Given its durability over the course of disease (i.e., presence at diagnosis, frequent “loss” in clinical remission, re-emergence at relapse), and association with inferior outcomes when detected at lower levels following initial therapy (MRD) [71, 72], NPM1 mutant allele transcript detection has become essential for disease monitoring and longitudinal patient management. The 2017 and updated 2022 European LeukemiaNet (ELN) guidelines for AML management recommend baseline assessment of NPM1 mutant transcript burden by reverse transcriptase quantitative PCR (RT-qPCR) followed by assessment at 3-month intervals over the course of 2 years [35, 36]. It is important to note the heterogeneity in molecular methods for MRD detection; while ELN guidelines have recommended the RT-qPCR assay, highly sensitive DNA-based approaches such as digital droplet PCR and NGS-based ultradeep sequencing [73, 74] also warrant consideration for laboratories pursuing MRD test implementation. Whether or not other assays, including multiparameter flow cytometry (MFC) or NGS, can be used as substitutes or provide complementary data remains an open question.

MFC: Does It Still Have a Role?

Several large studies have evaluated the immunophenotypic profile of NPM1-mutated AML with a focus on MFC [75–77]. Common leukemia-associated immunophenotypes (LAIPs) include decreased or absent expression of CD34, increased CD33, decreased or absent HLA-DR, increased CD123, and decreased or absent CD13. Cases with more conspicuous monocytic differentiation are often characterized by a variable proportion of leukemic cells with both conventional and abnormal antigen expressions. Conventional expression patterns include expression of CD13, CD14, CD16, CD33 (bright), CD34, CD117, HLA-DR, CD64, CD4, and CD15 (variable). Abnormal phenotypes can include aberrant expression of CD56, with absence of CD13, CD14, CD16, CD34, CD117, and/or HLA-DR. However, such LAIPs are not entirely specific and can be very challenging to detect at the low levels required to qualify for an MRD assessment, which is further complicated by inter-assay and inter-observer variability (i.e., diagnostic subjectivity). MFC-based MRD evaluation in NPM1-mutated AML can also be particularly difficult in cases with prominent monocytic differentiation at diagnosis, or in instances where the residual disease is predominantly monocytic (Fig. 2a, b). Nonetheless, the updated 2022 ELN guidelines include recommendations for a core panel of markers to perform MRD assessment by MFC, which include CD45, HLA-DR, CD34, CD117, CD13, CD33, CD7, CD56, and additional monocytic markers when necessary, such as CD4, CD11b, and CD64.

Fig. 2.

Longitudinal assessment pre- and post-induction therapy in a representative case of NPM1-mutated AML with monocytic differentiation. a Multiparameter flow cytometry (MFC) was performed on a peripheral blood specimen characterized by an exuberant monocytosis (∼86%) including immature forms and ∼8% circulating blasts by manual differential count. MFC shows a predominant abnormal population in the conventional CD45+ (moderate to bright)/intermediate side scatter (SSC) monocyte gate (∼91.6% of viable cells), which expresses several monocytic antigens (not shown) and aberrant partial CD56 positivity. Color scheme for gated populations: mature lymphocytes (green), monocytes (orange), granulocytes (aqua), “blast” gate (red). Axes are on log scale, and major hashmarks denote 102–105. b Multiparameter flow cytometry (MFC) was performed on a post-induction therapy bone marrow aspirate specimen with ∼13% monocytosis and ∼3% blasts by manual differential count. MFC shows a small but discrete abnormal population in the conventional CD45+ (moderate to bright)/intermediate side scatter (SSC) monocyte gate (∼17% of viable cells), which expresses several monocytic antigens (not shown) and aberrant CD56 positivity, suspicious for minimal/measurable residual disease. Color scheme for gated populations: mature lymphocytes (green), monocytes (orange), granulocytes (aqua), “blast” gate (red). Axes are on log scale, and major hashmarks denote 102–105. c, d Mutant NPM1 protein immunohistochemistry performed on the accompanying Bouin-fixed paraffin-embedded trephine biopsy (fast red, antibody clone: PA1-46356 [Thermo Fisher], ×1,000) highlights foci of clustered immature-appearing mononuclear cells (c) as well as conspicuous terminally differentiated megakaryocytes (d).

Whether or not MFC and molecular diagnostic methods, such as RT-qPCR, are equivalent for MRD assessment of NPM1-mutated AML has remained an area of ongoing investigation. A notable confounding factor for MFC-based MRD assessment in NPM1-mutated AML is the presence of persistent clonal hematopoiesis (CH)-associated mutations in the absence of detectable mutant NPM1. Using an NGS-based assay for molecular MRD evaluation, Jongen-Lavrencic and colleagues analyzed treatment-naïve and post-induction therapy specimens from a cohort of de novo AML patients and detected persistent mutations in 51.4%; notably, persistent “DTA” mutations (DNMT3A, TET2, ASXL1), commonly found in CH, were not associated with an increased rate of relapse [78]; however, upon exclusion of DTA-type mutations, mutational persistence correlated significantly with inferior clinical performance. A comparison of MFC-based MRD assessment with molecular detection of persistent non-DTA mutations revealed that using both approaches in combination could have additive value for patient prognostication and management.

In NPM1-mutated AML, co-occurring DNMT3A mutations are frequent [33], and studies using preclinical models have suggested that sequential acquisition of mutations in DNMT3A followed by NPM1 may represent a potent mechanism for transformation of CH to AML [79, 80]. In the context of MRD evaluation, one lingering question centers on the potential correlation between CH mutations and abnormal phenotypes as detected by MFC, including so-called difference from normal (DfN) patterns, in addition to LAIPs as discussed earlier. The possibility that persistent CH, following the eradication of an NPM1 mutation, may lead to spurious detection of MRD by MFC has warranted further investigation, since the relapse risk associated with persistent CH alone has been shown to be less significant. Loghavi and colleagues recently addressed this issue by comparing MFC- and NGS-based MRD assessments in a cohort of NPM1-mutated AML patients [81]. Interestingly, they found that approximately half of the patients with persistent CH-type mutations in the absence of mutant NPM1 had a small number (<1%) of aberrant CD34+ myeloblasts with a phenotype that differed from that seen on the pre-treatment blast population (referred to as a “pre-leukemic” [PL] phenotype); of note, PL phenotypes did not correlate with relapse-free survival. Inter-observer variability in the interpretation of PL phenotypes, e.g., resulting in erroneous classification as a difference from normal pattern indicative of MRD, may adversely alter the management for at least some patients who do not truly have a higher relapse risk. These data further raise the possibility that the challenges in interpretation of MFC-based MRD data in NPM1-mutated AML cases might outweigh the benefits of the assay, particularly in the absence of orthogonal assessment by a molecular method.

MRD Positivity: What Are We Actually Detecting?

The ELN has defined molecular persistence at low copy number (MP-LCN) in NPM1-mutated AML as an MRD transcript level <2% with a <1-log change between any 2 positive samples obtained following the end of treatment [36]. Notably, persistent low-level expression of mutant NPM1 may not be prognostic of relapse [82, 83]. Tiong and colleagues recently analyzed a cohort of patients characterized by molecular MRD positivity at end of treatment and found that 42% remained progression-free at 1 year; of these, 30% spontaneously achieved a complete molecular remission. At present, there is no reliable method to predict the trajectory of a patient with MP-LCN NPM1-mutated disease.

Cytotoxic therapy is most deleterious to rapidly dividing cells, and the molecular detection of persistent NPM1 mutation is notably agnostic to the cell type(s) harboring the mutant allele. Thus, at least in some patients with MP-LCN, the residual NPM1-mutated cells may include at least a subset that are incapable of driving relapse. Shortly after the initial discovery of NPM1 mutations in AML, Falini and colleagues also demonstrated that NPM1 mutations were ubiquitous in the myeloid compartment [44]. In our laboratory, we occasionally identify post-treatment MRD-positive samples by mutant NPM1 protein-specific IHC that are characterized by mutant protein expression in mononuclear elements, but also by conspicuous expression in terminally differentiated cells such as mature megakaryocytes (Fig. 2c, d). The relative contribution of these various cell types to the measurement of mutant NPM1 transcript warrants consideration, given the potential for non-leukemogenic NPM1-mutated cells to drive the detectable mutant transcript level above 2%, thereby potentially leading to misclassification of some cases that would otherwise be considered MP-LCN by ELN criteria. Recently, we compared MFC, RT-qPCR, and immunohistochemical methods for detection of residual NPM1-mutated disease and found that indeed, biopsy staining could provide complementary information regarding the cellular/architectural context of residual disease not revealed by “bulk” testing methods [84]. Multiparameter in situ imaging has also been demonstrated in preliminary studies to be helpful in localizing mutant NPM1 protein to specific cell types, providing potentially valuable information that cannot be obtained by MFC or molecular diagnostic techniques [85]; however, additional studies will likely be required to further evaluate the role of mutant NPM1 distribution within the myeloid compartment, particularly in the post-therapy setting.

A Potential Role for the Bone Marrow Microenvironment

The clinical course of patients characterized by MP-LCN may also be shaped by NPM1 mutant cell-extrinsic features of the bone marrow environment. Ferraro and colleagues recently compared patients with NK AML who were characterized by either short (relapse within 2 years; SFR) or long (>5 years; LFR) first remissions [86]; of note, the core genetic variables comprising ELN criteria for prognostication were insufficient to explain the outcomes of many SFR patients. Interestingly, their single-cell RNA sequencing studies revealed that SFR AML cells differentially expressed many genes associated with immune suppression, upregulation of MHC class II genes, and a microenvironmental profile characterized by fewer CD4+ Th1 cells, which exhibited an exhaustion signature; furthermore, the exhaustion could be attributed to the presence of AML cells and ameliorated by AML cell eradication or targeted inhibition of the MHC class II:LAG3 pathway. These features were not observed in most LFR cases. Of note, NPM1 mutation was detected in 54.8% and 85.7% of SFR and LFR cases, respectively.

It has been previously established that mutant NPM1 peptides can be presented by HLA molecules and recognized by T cells [87–90]. However, the potential interactions between NPM1-mutated cells and their microenvironmental co-constituents in human samples cannot be easily studied using aspirated cells, where spatial information is lost. These interactions are potentially even more important to study in the post-therapy setting, where the differences between SFR and LFR cases could be explained by critical interactions between rare residual leukemic cells and nearby immune effectors. Anecdotal evidence from our diagnostic laboratory, based on mutant NPM1 IHC applied to post-treatment bone marrow biopsy tissues, has revealed that residual mutant cells can exist in either tight clusters or as singly dispersed elements (Fig. 3); however, the potential significance of these distribution patterns remains unknown. Case 1 illustrates an additional benefit of NPM1c-specific IHC; in this case, the aspirated marrow material was insufficient for flow cytometry or PCR-based MRD assays, and thus, detection of residual mutant protein-expressing cells by IHC performed on the trephine biopsy provided critical evidence of MRD. In this overall context, additional multiparametric in situ imaging studies may also enable dissection of the leukemic “niche” for residual NPM1-mutated disease and provide additional insight into the prediction of SFR and LFR states.

Fig. 3.

Two representative examples of mutant NPM1 protein-specific immunohistochemistry in bone marrow trephine biopsy specimens from NPM1-mutated AML patients obtained following induction chemotherapy. a Case 1: NPM1-mutated cells in clusters (left) and singly dispersed (right). The aspirated marrow material was insufficient for reliable flow cytometry- or PCR-based MRD assessment in this case. b Case 2: NPM1-mutated cells in clusters (left) and singly dispersed (right). Using a conventional NGS-based assay (non-deep sequencing), the type A mutant NPM1 allele was detected at a variant allele frequency of 2.1% (data not shown). Chromogenic immunohistochemistry performed on Bouin-fixed paraffin-embedded tissues, using anti-mutant NPM1 protein-specific antibody (fast red, antibody clone: PA1-46356 [Thermo Fisher], ×1,000).

Novel Therapeutic Strategies

The established approaches to management of NPM1-mutated AML patients include intensive chemotherapy for fit adults (≤60 years), FLT3 mutation-directed agents when applicable, and HSCT for patients with higher risk disease [35, 36, 91, 92]. Patients >60 years, regardless of fitness status, have been shown in recent studies to benefit from treatment with regimens that include the BH3 mimetic venetoclax [93–95], while hypomethylating agent treatment alone may lack efficacy [96]; in older patients who can tolerate HSCT, reduced intensity conditioning regimens may be preferable. In addition to standard chemotherapeutic approaches, several experimental strategies to treat NPM1-mutated AML have emerged over the last several years, which are either undergoing preclinical optimization (e.g., XPO1 inhibition with Eltanexor, treatment with dactinomycin) or preliminary testing in early-phase clinical trials (e.g., menin inhibition). A comprehensive discussion of these novel strategies falls outside of the scope of this review, and therefore, only two select approaches will be briefly discussed.

XPO1 Inhibition

Nuclear export of wild-type NPM1, and its mutant counterpart NPM1c, relies on binding to the nuclear pore protein exportin-1 (XPO1/CRM1) [43]. Based on the rationale that relocalization of NPM1c to the nucleus may abrogate the mutant cell phenotype, XPO1 inhibition was initially applied to NPM1-mutated AML cells in vitro and proven to have the desired effect on transcription and leukemic cell differentiation [29, 30]. It is important to note that other proteins, such as p53 and p21 [97], also contain NES sequences, and therefore, the mechanism of XPO1 inhibition is likely to be complex. The selective XPO1 inhibitor Selinexor was initially tested in AML; however, associated neurotoxicity limited the dosing frequency, resulting in limited antileukemic efficacy due to a lack of sustained XPO1 inhibition [98]. The second-generation XPO1 inhibitor, Eltanexor, appears to have a better toxicity profile as a result of lower penetration of the blood-brain barrier [99]. Piangiani and colleagues recently demonstrated using preclinical models that Eltanexor could induce prolonged XPO1 inhibition, resulting in irreversible HOX gene downregulation, AML blast differentiation, and prolonged survival of leukemic mice [100]. Although Eltanexor is currently in early-phase trials for a wider variety of hematologic malignancy indications, these recent data will hopefully lay the foundation for focused testing in patients with NPM1-mutated AML.

Menin Inhibition

As discussed in several sections of this review, a hallmark feature of NPM1-mutated cells is the overexpression of HOX gene network members and their cofactor MEIS1, which is directly fostered by the neomorphic role of NPM1c. As part of this neomorphic function, NPM1c complexes with other proteins including MLL1, which itself associates with its cofactor menin. The menin-MLL interaction has been shown to induce HOX overexpression in other leukemia subtypes, including those with MLL rearrangement. Thus, even prior to the most recent studies described above that have elucidated the specific functions of NPM1c, it has been of interest to disrupt the menin-MLL1 network as a therapeutic strategy in NPM1-mutated AML. Two novel menin-MLL1 inhibitors (MI3454 and VTP-50469) have demonstrated promising efficacy in preclinical models [51, 52], paving the way for ongoing early-phase clinical trials in patients with relapsed/refractory NPM1-mutated AML (NCT05453903, NCT04067336, NCT04065399).

Concluding Remarks

The original discovery of NPM1 mutations in AML was made nearly 20 years ago; given the frequency of this genetic lesion in AML, it has been possible for many different groups to study large cohorts of patients, as a result of which our understanding of the clinicopathologic and biological features of the disease has advanced tremendously. However, studies within the last several years, including two within the past year alone, have finally elucidated the critical mechanisms underlying the function of NPM1c in promoting leukemogenesis. Our changing view of NPM1-mutated disease is in many ways evidenced by the fact that the blast count threshold required for an AML diagnosis has been lowered in both of the new WHO and ICC classification schema. Recent data have not only clarified our understanding of the biology of NPM1-mutated AML, but also provided compelling bases for therapeutic intervention that will hopefully be free of the common toxicities associated with conventional chemotherapy. Although some critical questions have been answered, others nevertheless remain: which assay will become the gold standard for longitudinal monitoring of NPM1-mutated AML patients? How should patients characterized by low-level MRD positivity post-therapy be managed? Are there tumor cell-extrinsic components of the bone marrow environment that may be targetable, either alone or in combination with leukemia cell-directed therapies? At least some of these lingering questions are likely to form the basis of ongoing investigations over the coming years.

Conflict of Interest Statement

The author acts as a subject matter expert/consultant for various clients via Guidepoint, LLC. and GLG, LLC. intermediaries.

Funding Sources

None relevant to the preparation of this manuscript.

Author Contributions

Sanjay S. Patel conceived of the topic for the review, prepared all figures, and wrote the manuscript.

This article is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC). Usage and distribution for commercial purposes requires written permission.
Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug.
Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.