Nucleophosmin 1: from its pathogenic role to a tantalizing therapeutic target in acute myeloid leukemia

ABSTRACT Nucleophosmin 1 (NPM1, also known as B23) is a multifunctional protein involved in a variety of cellular processes, including ribosomal maturation, centrosome replication, maintenance of genomic stability, cell cycle control, and apoptosis. NPM1 is the most commonly mutated gene in adult acute myeloid leukemia (AML) and is present in approximately 40% of all AML cases. The underlying mechanisms of mutant NPM1 (NPM1mut) in leukemogenesis remain unclear. This review summarizes the structure and physiological function of NPM1, mechanisms underlying the pathogenesis of NPM1-mutated AML, and the potential role of NPM1 as a therapeutic target. It is reported that dysfunctional NPM1 might cause AML pathogenesis via its role as a protein chaperone, inhibiting differentiation of leukemia stem cells and regulation of non-coding RNAs. Besides conventional chemotherapies, NPM1 is a promising therapeutic target against AML that warrants further investigation. NPM1-based therapeutic strategies include inducing nucleolar relocalisation of NPM1 mutants, interfering with NPM1 oligomerization, and NPM1 as an immune response target.


Introduction
AML is a malignancy characterized by clonal proliferation originating from primitive hematopoietic stem or progenitor cells. According to the 2016 WHO classifications, AML is classified into 6 categories: AML with recurrent genetic abnormalities, AML with myelodysplasia related changes, therapy-related myeloid neoplasms, AML, not otherwise specified (NOS), myeloid sarcoma, and myeloid proliferations associated with Down syndrome [1,2]. As a distinct subtype of AML with recurrent genetic abnormalities, AML with NPM1 mut accounts for about 40% of AML and 60% of cytogenetically normal AML (CN-AML) [3]. The frequency of NPM1 mut in the Asian population is relatively lower (about 20%) compared to that in the European and American populations [4]. AML patients with NPM1 mut show unique biological and clinical features; for example, NPM1 mut is more common in subtypes M1 (42%), M4 (57%), M5a (49%), and M5b (70%), leukemic cells are often marked by a cup-like nuclear invagination. Low expression or loss of CD34 expression was observed in more than 95% of NPM1 mut AML [5,6]. NPM1 mut is closely associated with the pathogenesis and prognosis of AML [7], NPM1 mut predicts higher response rates to induction therapies and longer overall survival (OS) in AML [8][9][10][11][12], especially in those without a high burden of FLT3-ITD (FLT3-ITD/FLT3 wt > 0.5), DNMT3A mutations and IDH mutations [13][14][15]. This review focuses on the mechanisms underlying the pathogenesis of NPM1 mut AML and the potential role of NPM1 as a therapeutic target.

Structure and function of NPM1
NPM1 is a multifunctional protein comprising 294 amino acid (AA) residues, it is mainly located in the nucleoli and is capable of shuttling between the nucleus and cytoplasm. It is encoded by the NPM1 gene located on chromosome 5q35 [16]. NPM1 plays critical roles in ribosomal maturation, centrosome replication, maintenance of genomic stability, cell cycle control, and apoptosis. Inactivation of NPM1 in mice causes defects that lead to death in mid-pregnancy [17]. In humans, NPM1 inactivation in adult hematopoietic stem cells (HSC) results in bone marrow failure [18]. NPM1 exerts a chaperone-like activity via its N-terminal domain (NTD), thereby protecting the structure of multiple proteins to maintain their enzymatic activities during thermal denaturation [19]. Two canonical nuclear export signals (NES) in the NTD of NPM1 ( Figure 1) mediate its nuclear export by interacting with nuclear export protein 1 (XPO1) [20] and are crucial for NPM1 function. In addition, NPM1 binds histones via the acidic domains (AD) of mimic DNA/RNA, which is rich in negatively charged aspartic acid and glutamic acid residues (acting as histone chaperones) and is further involved in nucleosome formation and chromatin remodeling [21,22]. The carboxy-terminal domain (CTD) of NPM1 is rich in hydrophobic aromatic amino acids (F268, Y271, F276, W288, and W290). Nucleolar localization signals (NoLS) located on the third helix (H3) of a special 'three-helix bundle' structure map of NPM1 in the nucleoli. Two tryptophan residues (W288 and W290) inside NoLS are essential in allowing NPM1 to maintain the 'three-helix bundle' structure [20]. Between AD and CTD, NPM1 contains an alkaline fragment with more than 50 AA residues that binds to DNA/RNA. In addition, this region contains a nuclear localization sequence (NLS) and another NLS located between the two ADs. The two NLS act as importin α/β recognition sites, which mediate the nuclear localization of NPM1 ( Figure 1) [22].

NPM1 mut in AML
3.1. NPM1 mut clone and AML pathogenesis NPM1 mut -AML was initially thought to be caused by haploinsufficiency of NPM1 [23]. However, the identification of NPM1 mut in CD14 + monocytes, CD66b + granulocytes, CD19 + B cells, CD3 -CD14 -CD16 + CD56 + NK cells as well as CD3 + T cells in AML patients harboring NPM1 mut , suggested that NPM1 mut AML originated from early-stage stem cells with the capacity to differentiate into any lineage [24]. Moreover, NPM1 mut was also found in leukemia stem cells (LSCs), and mice transplanted with CD34 + leukemic cells developed morphologically and immunohistochemically confirmed leukemia, predicting that NPM1 mut is an initiating factor in the pathogenesis of AML [5]. However, subsequent studies showed that a single NPM1 mutation only affected megakaryocyte development in mice and was not sufficient to cause AML [25]. In addition, 36.5% of patients with relapsed NPM1 mut -AML lost NPM1 mut and developed new cytogenetic and molecular alterations [26]. In recent years, it is hypothesized that different genetic events can cause leukemia transformation. Clinically, almost all NPM1 mut were found to co-occur with mutations in genes involved in DNA methylation regulation (such as DNMT3A, TET2, IDH1 and IDH2), RNA splicing (SRSF2 and SF3B1) or genes associated with adhesive protein complexes (RAD21, SMC1A, SMC3 and STAG2) [27]. Mutations associated with cell signaling pathways (FLT3, NRAS and PTPN11) also coexist with NPM1 mut in AML [27]. Furthermore, mice with DNMT3A-mutant clonal hematopoiesis (CH) progressed to myeloproliferative neoplasms (MPNs) after the introduction of NPM1 mut , and more aggressive MPN was observed with a longer latency between the two mutations. MPN uniformly progressed to AML in secondary recipient mice. These results indicate that NPM1 mut drove the evolution of DNMT3A-mutant CH to AML [28]. Above all, these data suggest that NPM1 mut might be a secondary molecular event in AML leukemogenesis.

Underlying mechanisms of NPM1 mut in AML pathogenesis
NPM1 exon 12 mutations are frequent in AML patients and are classified as either type A (NPM1 mA ) and non-Type A mutations. NPM1 mA (tandem duplications of TCTG occur between 863-864 nucleotides resulting in AA chain extension) is the main type of mutation in adult AML patients (accounting for 75-80% of adult NPM1 mut -AML). Non-type A NPM1 mutations are called 'rare type' and are mainly caused by insertion of different bases at the end of exon 12. There are more than 40 mutations in total, which were named alphabetically in the order of detection (B-ZY) [29].
These mutations lead to frameshift mutations in the CTD of NPM1, thus inserting a new NES or leading to loss of W288 and W290 (or W290 alone) in the original NLS, resulting in ectopic localization of NPM1 in the cytoplasm of leukemia cells (also known as NPM1c+) [30]. Polypeptides harboring NPM1 mut have different kinetics and oligomerization levels and therefore have different abilities to form amyloid aggregates. Thus, like wild-type NPM1 (NPM1 wt ), NPM1 mut may have a modulated function by regulation of oligomerization [7]. Moreover, NPM1-CTD bound with highaffinity G-quadruplex DNA regions found in ribosomal DNA, while NPM1 variants completely lost this activity, which is crucial for its nucleolar localization [31]. This suggests that the abnormal structure of NPM1 mut might lead to its cytoplasmic localization. In addition, NPM1 wt could also abnormally localize to the cytoplasm by forming a heterodimer with NPM1 mut [32].
Compared with NPM1 wt -patients, NPM1 mut -AML patients were more likely to express NPM1 mut , suggesting that the expression preference of mutant transcripts may be related to the pathogenesis of AML [33]. NPM1 mut further induces AML through a variety of mechanisms. In unstressed cells, NPM1 disrupts the ARF-MDM2 interactions, thereby inhibiting p14ARF-dependent p53 activation. In response to cellular stress, NPM1 and ARF may relocate to the cytoplasm and competitively bind to MDM2. Similar to the above, the abnormal cytoplasmic localization of ARF was induced by NPM1 mut , resulting in its ubiquitination and degradation, thus impairing the function of the tumor suppressor ARF-MDM2-p53 axis [20].
Fbw7γ (a member of the F-Box protein family and an important component of the E3 ligase complex of the Myc oncoprotein) is degraded following its abnormal cytosolic delocalization by mutated NPM1, thus stabilizing Myc and further promoting the proliferation and survival of leukemia cells [29,[34][35][36]. NPM1 mut may also act as a 'molecular partner' to directly form a complex with multiple transcription factors (such as CTCF, MEF/ELF4 and NF-κB, etc.) to interfere with the transcription of downstream target genes (such as MDM2, etc.), thereby fine-regulating the pathogenesis of AML [37][38][39].
Another key event in the leukemogenesis of AML is the failure of LSCs to differentiate into mature hematopoietic cells. In embryonic stem cells, downregulation of NPM1 increased the transcription of key genes involved in mesodermal differentiation, suggesting its fundamental role in the maintenance of the stem cell phenotype [40]. In AML, NPM1 mut regulates a series of differentiation-inducing genes such as homobox (HOX), PBX homobox 3 (PBX3) and Meis homobox 1 (Meis) genes. XPO1 inhibition induced the relocalisation of NPM1 mut to the nucleus and degradation of NPM1 mut ; this led to the down-regulation of the expression of HOX and Meis gene, thereby further inducing differentiation of AML cells into mature cells and prolonging the survival of leukemia mice carrying NPM1 mut . The DOT1L inhibitor EPZ5676 promoted the apoptosis of leukemia cells harboring NPM1 mut by decreasing the expression of HOXA9 and PBX3. These results suggest that persistent cytoplasmic localization of NPM1 mut is necessary for the maintenance of leukemia stem cells [41][42][43]. However, it has been reported that co-overexpression of NPM1 mA with Meis1 or HOXA9 was not sufficient to induce AML [44], highlighting the necessity of prodromal molecular events before NPM1 mut acquisition. In a mouse model of AML in which mice were transplanted with NPM1 +/− hematopoietic stem/progenitor cells co-transduced with mutant genes (such as IDH2 R140Q ), conditional knockout of IDH2 R140Q resulted in the failure of LSC maintenance and implantation, and promoted the apoptosis of NPM1 mut cells. These results suggested that IDH2 mut is critical for the development and maintenance of AML-LSCs [45]. Both knock-in mice carrying double mutations (NPM1 mA/+ NRAS G12D/+ and NPM1 mA FLT3-ITD) had AML-like manifestations and shared common characteristics (overexpression of HOX gene, enhanced self-renewal ability, amplification of hematopoietic progenitor cells, and myeloid differentiation preference). In addition, NPM1 mutation has significant molecular synergies with both NRAS G12D and FLT3-ITD. Moreover, HOX is necessary for survival of both types of dual-mutated AML [46].
In addition, in recent years, aberrations in noncoding RNAs (e.g. overexpression of miR-21, miR-10a, and miR-10b) have been found to be closely related to the occurrence and development of NPM1 mut -AML [47]. Long non-coding RNA (lncRNA) HOXB-AS3 upregulated proliferation of NPM1 mut -AML cells in vitro and vivo, regulates ribosomal RNA transcription and de novo protein synthesis, which may serve as a compensation mechanism for adequate protein production in leukemia cells [48]. Overexpression of HOXBLINC (a HOXB gene-related lncRNA) in mice enhanced the self-renewal capacity of HSC and led to AML-like disease [49].

Therapy of NPM1 mut -AML
Despite the high complete remission (CR) rate of NPM1 mut -AML without FLT3 mutations after the standard '7 + 3' IA regimen induction, all patients relapsed within 41 months. The CR rate and 3-year OS in patients who received fludarabine in combination with IA were significantly higher than those in the non-fludarabine group. These results suggested that AML patients with NPM1 mut may benefit from intensive chemotherapy [50,51]. However, these patients did not benefit from allogeneic hematopoietic stem cell transplantation (allo-HSCT) compared with patients with autologous hematopoietic stem cell transplantation (ASCT) consolidation [52,53]. This contrasts with the results of the AML2003 trial, in which patients in the allo-HSCT arm had significantly better 3-year relapse-free survival (RFS) and OS than those in the non-allograft group (ASCT or chemotherapy consolidation). Among patients with normal karyotype and FLT3 wt , the 3-year RFS of the allo-HSCT subgroup was still superior to that of the non-allo-HSCT group, but there was no significant difference in OS. This suggests that AML cases with NPM1 mut , especially those with FLT3-ITD mutation, may still benefit from allo-HSCT [54]. In addition, allo-HSCT is recommended as first-line consolidation therapy for NPM1 mut -AML patients with other dismal prognostic factors, such as DNMT3A R882 mutation. If no suitable donor is available, these patients may still benefit from ASCT [53,55]. But allo-HSCT was not observed to improve the outcomes of patients with co-occurrence of FLT3-TKD and NPM1 mutations [56]. In elderly or unfit patients, no significant difference was observed in OS between the NPM1 mut and NPM1 wt cohorts after azacytidine-containing induction therapy, suggesting that hypomethylating agents are unable to improve long-term survival when used in the first-line treatment of NPM1 mut -AML [57]. However, maintenance therapy with lenalidomide combined with azacytidine enhanced the function of cytotoxic T lymphocytes in AML patients with NPM1 mut in remission, which further enabled sustained remission. Therefore, azacytidine/lenalidomide as maintenance therapy for highrisk AML deserves further study [58].

NPM1 as a novel therapeutic target
Despite notable advances in the treatment of this frequent AML subtype in recent years, approximately 50% of AML patients with NPM1 mut treated with conventional regimens eventually died of disease progression [59]. In vitro, knockdown of NPM1 resulted in proliferation inhibition and increased the apoptosis of leukemia cells [60]. The stable expression of NPM1 variants in NPM1 mut -AML cells makes it a promising therapeutic target, and in recent years, an increasing number of therapies targeting NPM1 have been explored.

Nucleolar relocalisation of NPM1 mut
Persistent cytoplasmic localization of NPM1 mut is a key event in leukemia development [41], therefore, interfering with NPM1 mut mislocalisation is a feasible strategy for the treatment of NPM1-mutated AML. Both natural product avrainvillamide (AVA) and its fully synthetic AVA analog exerted potent anti-leukemia activities in vitro and vivo. And the NPM1-mutated cell line OCI-AML3 and primary AML cells were more sensitive to AVA than cells expressing NPM1 wt , mainly due to AVA induced NPM1 mut nuclear retention and enhanced proteasomal degradation of NPM1 mut and XPO1 [61]. Similarly, XPO1 inhibitor and EAPB0503 restored NPM1 nucleolar localization and promoted NPM1 mut degradation. Furthermore, XPO1 inhibition led to the down-regulation of HOX, thus inducing AML cell differentiation, promoting apoptosis, and reducing the leukemia burden in xenograft mice [41,62].

Nucleolar stress induction
Actinomycin D induces nucleolar oxidation, while NPM1 is S-glutathionylated at Cys275 under stress, further leading to its dissociation from rDNA/rRNA and translocation to the nucleoplasm, thus interfering with ribosomal biogenesis. TP53 mutations or 17p deletion do not frequently co-occur with NPM1 mutations in AML, thus, p53-mediated nucleolar stress still exists, and there is still a low-level expression of NPM1 wt in the nucleoli of NPM1 mut -cells. It is speculated that the nucleoli of NPM1 mut -AML cells may be sensitive to medicines that trigger nucleolar stress (such as actinomycin D). Falini et al. reported that a newly diagnosed 60-year-old AML patient with NPM1 mut and FLT3 wt who could not undergo intensive chemotherapy due to heart disease responded to Actinomycin D. Later, Guillaume et al. reported that 3/17 (18%) of relapsed/refractory (R/R) adult AML carrying NPM1 mutation achieved CR after 1 course of actinomycin [9,63]. In another phase II single-centre clinical trial, 10 adult patients with R/R NPM1 mutated AML were treated with actinomycin. All patients responded to the treatment, with 44% achieving CR/CRi, with good treatment tolerance. Compared with NPM1 wt cells, low-dose actinomycin displayed a more effective stress response in NPM1 mut -cells, suggesting that NPM1 mutated AML was more sensitive to nucleolar stress. Therefore, actinomycin is a potential treatment option for R/R NPM1 mutated AML and warrants further investigation [64].

Interfere with NPM1 oligomerization
NSC348884 is a small molecule inhibitor that inhibits the oligomerization of NPM1, further inducing the apoptosis of OCI-AML3 cells and primary NPM1 mut -AML cells sensitized by all-trans retinoic acid (ATRA). However, a recent study showed that NSC34884mediated cytotoxicity in AML cells by regulating cell adhesion signaling rather than inhibiting NPM1 oligomerization [65]. The 1-benzyl-2-methyl-3-indolemethyl-thiobarbituric acid analogs 7K and 7L were found to cause dose-dependent apoptosis in NPM1 mut harboring OCI-AML3 cells. Molecular docking studies showed that they exert their potent anti-leukemia activities mainly by binding to the pocket in the central channel of the NPM1 pentameric structure [66]. In irradiated cells, YTR107, a small molecule radiation sensitizer, suppressed the formation of the NPM1 pentamer by binding to NPM1, thus inhibiting SUMOylated NPM1 from associating with RAD51, preventing RAD51 foci formation and repair of DSBs. YTR107 can also acted synergistically with the PARP1/2 inhibitor ABT 888 to increase replication stress and radiation-induced cell lethality [67].
The multivalent pseudopeptide N6L, a synthetic ligand of cell surface nucleolin, sensitized AML cells to doxorubicin and cytarabine treatment. N6L bound NPM1-NTD with high affinity and inhibits its phosphorylation. In AML cells, N6L co-localized with NPM1 mut in the cytoplasm and interfered with its protein-protein associations, hence displaying its anti-leukemia activity [68].

Down regulation of NPM1 mut
Arsenic trioxide (ATO) and ATRA cause proteasomedependent NPM1 mut degradation, induce apoptosis of NPM1 mut -AML cell lines and primary AML cells, and sensitize leukemia cells to daunomycin-containing chemotherapy. However, prolonged survival was not observed in NPM1 mutated AML treated with ATRA [69,70]. Both epigallocatechin3-gallate (EGCG) and deguelin (a rotenoid isolated from several plant species) inhibited the expression of NPM1, exhibited proliferation inhibition and apoptosis induction of NPM1-altered AML cells in vitro. However, deguelin has no such effect in OCi-AML2 cells (harboring NPM1 wt ), suggesting that these molecules are expected to be used in AML cases with NPM1 mutations [71][72][73].

Therapeutic immune responses against the mutated NPM1
Somatic mutations in distinct cancers are a potential source of cancer-specific neoantigens, and NPM1 mut is expected to be an ideal target for personalized immunotherapy due to its specificity in leukemia. Specimens of AML patients were stimulated with two endogenous HLA class I peptides (LAVEEVSLR and AVEEVSLRK) carrying NPM1 mut originated from primary AML cells. ELISPOT assay revealed that interferon γ-secreting NPM1 mut -specific T lymphocytes were observed in 50.6% of peripheral blood and 42.5% of bone marrow samples. The minimal residual disease (MRD) level in patients was negatively correlated with anti-leukemia specific T cells [74]. Specific immunoresponsive CD8 cytotoxic T cells (CTLs) against NPM1 mutatants-derived epitopes may be involved in the clearance of MRD and is associated with a favorable prognosis and expected persistent MRD negative for this AML subtype [75,76]. These findings suggest that adoptive immunotherapy may be a promising treatment option for AML patients with NPM1 mut , especially for maintenance therapy. 4.6. Factors co-expressed with NPM1 as therapeutic target 4.6.1. CD33 CD33 expression was significantly higher in newly diagnosed NPM1 mut -AML patients than in NPM1 wt patients, which laid a foundation for the application of anti-CD33 antibody (GO) in AML carrying NPM1 mut [77]. The CR/CRi rate and cumulative recurrence rate (CIR) of AML patients treated with a GO-containing regimen as initial induction therapy were significantly lower than those treated with standard induction group without GO [78]. In the AMLSG 09-09 trial, the median NPM1 mut copy number of patients treated with GO were significant lower across all treatment cycles, resulting in a significantly greater proportion of patients achieving MRD negativity. Moreover, addition of GO to the treatment of NPM1 mut MRD-positive patients significantly reduced the NPM1 mut copy number after two treatment cycles, thus significantly reducing CIR [79].

Menin-MLL1
Overexpression of HOX genes is observed in almost all NPM1 mut -AML patients. The histone modification enzymes MLL1 and DOT1L control the expression of HOX and FLT3 in NPM1-mutated AML, and the survival of NPM1 mut -AML cells is particularly dependent on the Menin binding site in the MLL1 structure. In the NPM1mutated AML mouse model, small-molecule inhibitors of Menin-MLL1 protein interaction (e.g. MI-3454) exhibit potent anti-leukemia effects. Blockage of both MLL1 and DOT1L led to significant inhibition of HOX and FLT3 expression, leukemia cell differentiation, and stronger anti-leukemia activity. In addition, MI-3454 was well tolerated and did not impair normal hematopoietic function in mice [80,81]. Menin inhibitor mainly targets CD34 + CD38 + cells, while Venetoclax, a Bcl2 inhibitor, mainly targets CD34 + CD38 − cells. The combination of these two drugs may be effective at clearing hematopoietic stem/progenitor cells. Moreover, inhibition of Menin decreases the expression of a number of anti-apoptotic proteins and increases the expression of pro-apoptotic proteins. Therefore, in the AML PDX model carrying NPM1 mut FLT3 mut , treatment with a Menin inhibitor in combination with Venetoclax further improves mouse survival. The combination was more effective against primary AML cells harboring NPM1 mut FLT3 mut in vitro. On this basis, a triple combination regimen with FLT3 inhibitors may further enhance the efficacy [82]. Therefore, Menin-MLL1 is a promising therapeutic target for NPM1-mutated AML. AML is often preceded by a premalignant state (clonal hematopoiesis or myelodysplastic syndrome), characterized by significant myeloid progenitor cell proliferation and self-renewal for a relatively long period. VTP-50469, a small molecule targeting Menin-MLL1, which reverses this selfrenewal, might serve as a promising prophylactic treatment in the premalignant stage [83]. 4.6.3. Interleukin-3 receptor α chain (IL-3Rα, also known as CD123) CD123 is overexpressed in NPM1 mut -AML cells, especially in LSC and patients with FLT3 mut . Tagraxofusp (SL401) formed by IL-3 fusion with diphtheria toxin has been approved for the treatment of patients with blastocytic plasmacytoid dendritic cell tumor (BPDCN). SL401 has strong cytotoxic effect on CD123 + AML and MDS-derived primary cells, and has shown efficacy in R/R AML patients. However, since significant cytotoxic activity against normal hematopoietic progenitor cells has been observed, adverse events might be the major limiting factor of its application [84]. In December 2018, MB-102 (CD123 CAR-T) developed by Mustang Bio Inc. received the Orphan drug designation for the treatment of BPDCN. Subsequent studies have shown that MB-102 can be used as a novel immunotherapy agent for patients with R/R AML [59,85], and several clinical trials are currently underway. Therefore, CD123 as a valuable therapeutic target is worth further evaluation.

Bcl-2
Bcl-2 is found to be down-regulated by NPM1 mut in AML cell lines [86], however, in AML cases with NPM1 mut or IDH2 mut , treatment with the Bcl2 inhibitor venetoclax in combination with hypomethylated agents (HMA) or low-dose Ara-C results in a high response rate and durable molecular response, suggesting that patients with NPM1 mut may benefit from venetoclax-containing chemotherapy [87]. Venetoclax in combination with low-dose Ara-C or azacytidine induced durable molecular remission without transplantation in all AML patients with persistent NPM1 mut -MRD. After 1-2 cycles of venetoclax, MRD turned negative in 85.7% (6/7) of patients with molecular relapses [88]. Cell line OCI-AML3 is highly sensitive to ATO and moderately sensitive to venetoclax. However, ATO combined with venetoclax significantly inhibited cell proliferation and induced apoptosis without affecting NPM1 expression, suggesting that venetoclax and ATO have synergistic anti-leukemia effects. Subsequently, two R/R AML patients with NPM1 mut were found to be sensitive to venetoclax in combination with oral arsenic [89]. DiNardo et al reported that venetoclax plus azacitidine combination provided higher and moredurable response rates and an improved OS in previously untreated patients who were ineligible for standard induction therapy. Besides, in the cases with NPM1 mutations, venetoclax containing regimen was associated with a significantly higher incidence of composite CR rate (CR + CRi) than the control regimen (66.7% vs 23.5%, P = 0.012) [90]. Similar results were observed in another study, significant improvement in OS was seen in NPM1 mut -AML patients age >65 years treated with HMA+ venetoclax vs HMA or intensive chemotherapy. On account of the impressive results compared with traditional regimens, lower-intensity venetoclax combination therapies are largely considered an improved standard of care for older patients with NPM1 mut -AML [91]. 4.6.5. Other NPM1-related therapeutic targets Pediatric lymphoma driven by the NPM1-ALK fusion gene has a response rate to ALK inhibitors such as crizotinib of 54%-90%, and these can thus can be used as an alternative to standard chemotherapy with lower toxicity and side effects. However, a proportion of patients progress within merely 3 months after initial of treatment [92]. The FLT3 inhibitor Gilteritinib inhibited the phosphorylation of the NPM1-ALK fusion kinase and hence blocked the downstream signaling pathway, further inducing G0-G1 cell cycle arrest, apoptosis, decreased c-Myc expression and downregulated CD30 expression. Therefore, Gilteritinib can be used to treat NPM1-ALK positive anaplastic large cell lymphoma [93]. In addition, there are many other molecular events related to NPM1, such as autophagy activation and co-expression of IDH1/2, which are expected to provide more targets for new targeted therapies for NPM1-mutated AML in the future [94,95].

NPM1 mut as a MRD monitoring marker
NPM1 mutations are considered as an ideal marker for MRD monitoring in AML since they are stable over the course of disease including relapse in most cases, and can also express in the whole leukemic populations. The NPM1 mut -MRD of AML in first CR (CR1) was positively correlated with their baseline NPM1 mut VAF, suggesting that NPM1 mut -MRD can be used as a good prognostic factor to guide post-remission treatment [96]. Both achievement of MRD negativity and deep reduction in NPM1 mut transcripts (>4-log) at multiple time points after conventional chemotherapy (especially after double induction therapy) predict low risk of relapse and high survival rates, even in FLT3-ITD mutated AML patients, independent of the allelic ratio [97,98]. After induction therapy, the cumulative 2-year CIR of patients with relative expression of NPM1 mut higher than 0.01 was 77.8%, and the 2-year CIR of the remaining cases was 26.4% [96,99]. Among 152 NPM1 mut -AML with evaluable MRD assessment in CR1 enrolled in ALFA-0702, patients who did not achieve a 4-log reduction of NPM1 mut in peripheral blood after induction therapy had a higher CIR [100]. However, Ing S. Tiong et al documented that patients with NPM1 mut -MRD positivity after completing intensive chemotherapy had a variable course. One hundred NPM1 mut -AML who received ≥2 cycles of intensive chemotherapy without allo-HSCT in CR1 were included if bone marrow was NPM1 mut MRD positive at the end of treatment (EOT). After a median of 23.5-month follow-up, 42% remained free of progression at 1 year, either spontaneously achieving complete molecular remission (30%) or retaining lowlevel transcript in the bone marrow for ≥ 12 months (12%) [101].
As ELN recommended, using a dynamic risk assessment approach including MRD monitoring, allo-HSCT is clearly indicated when the estimated leukemia relapse risk is above 35-40% [100]. Cornellissen JJ et al reported that subjects who were MRD-test-positive pretransplant had a greatest benefit from a transplant than those who were MRD-test-negative pretransplant [102]. However, pretransplant MRD positive was also found to be associated with increased overall mortality and recurrence rates [103,104]. Dillon et al showed that AML patients who had high NPM1 mut transcripts before allo-transplant had a dismal 2-year survivals than those without or with low NPM1 mut transcripts (13%, 83% and 63%, respective) [105]. And, the prognostic MRD threshold is still controversial to date. Michal Karas et al reported that higher NPM1 mut -MRD level with cut-off 0.1% in bone marrow before allo-HSCT was independent poor prognostic factor for OS [103,104].
Moreover, post-transplant MRD was a strong independent predictor of both relapse risk and 3-year OS (20% ± 17.9% versus 88.6% ± 7.8% for MRD-positive and MRD-negative patients, respectively) in adult NPM1-mutated AML patients [104]. Platzbecker et al documented that in AML or advanced MDS cases who became MRD-positive in CR after either conventional chemotherapy only or consecutive allo-HSCT, pre-emptive 5-azacitidine resulted in overall response rates of 71% and 48% in patients with and without allo-HSCT, respectively. After a median follow-up of 13 months, OS rate was 76%, suggesting that MRDguided therapy may prevent or delay morphologic relapse [106]. However, the pre-emptive strategies are not currently standardized, which need to be further studied.
According to ELN recommendations, serial MRD monitoring at multiple timepoints was necessary. During active treatment phase, MRD should be measured at least at diagnosis, after two cycles of therapy and at the EOT. During follow-up period, MRD assessments is recommended every three months for the first two years. Thereafter, timing of MRD monitoring should be personalized according to relapse risk [107]. Molecular NPM1 mut MRD evaluation includes RNA-based (RT-qPCR) and DNA-based (Q-PCR, ddPCR and deep sequencing) levels. DNA-based methods show strong correlation (95% agreement). Although more sensitive, RT-qPCR failed to detect leukemic MRD in 10% of samples with detectable NPM1 mut DNA, suggesting that DNA-based method may act as a complementary for RT-qPCR [108]. Besides the standard molecular methods, flow cytometry and immunohistochemical (IHC) methods are also good choices for MRD detection. Flow cytometry is able to reliably detect NPM1 mut populations of at least 10% by using the mean fluorescence intensity. The results were consistent with conventional PCR, and the advantage is the rapid determination of NPM1 status in newly diagnosed patients [109]. Anita Chopra et al reported that ASO-PCR detected NPM1 mut in 21 (60%) of the 35 AML patients, and IHC detected NPM1 mut in 19 patients. 13/35 patients were negative for both tests, and one IHC positive patient was not detected by ASO-PCR. Compared with ASO-PCR, the sensitivity and specificity of IHC were 90% and 93%, respectively. Advantages of IHC include the ectopic expression of cytoplasmic NPM1 mut could be determined, and IHC is critical for diagnosis of AML cases presented with myeloid sarcoma [110].
However, 13.5% (14/104) cases relapsed with NPM1 wt -AML based on analysis of paired bone marrow samples from NPM1 mut -AML patients [111]. Moreover, the predictive value of NPM1 mut MRD was influenced by DNMT3A mutation status in elderly AML patients. DNMT3A status exerted no impact on the probability of having a > 4-log MRD reduction after induction. But post-induction NPM1 mut MRD reduction was not predictive of OS and LFS in DNMT3A mut patients, predicting that the presence of DNMT3A mutation seems to abrogate the predictive value of NPM1 mut good molecular responses. Therefore, as a monitoring marker for MRD in AML patients, NPM1 mut should be analyzed together with other markers [98].

Conclusion
In conclusion, the structural complexity and functional diversity of NPM1 can lead to several molecular abnormalities, including the formation of fusion genes and gene mutations that play an important role in the occurrence and development of AML. Further elucidation of the underlying mechanism of NPM1 in AML is expected to provide a better theoretical basis for finer clinical prognostic stratification and targeted therapy with new medications.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Ethical approval
This article does not contain any studies with human participants or animals performed by any of the authors.