Emerging Targeted Therapy for Specific Genomic Abnormalities in Acute Myeloid Leukemia

We describe recent updates of existing molecular-targeting agents and emerging novel gene-specific strategies. FLT3 and IDH inhibitors are being tested in combination with conventional chemotherapy for both medically fit patients and patients who are ineligible for intensive therapy. FLT3 inhibitors combined with non-cytotoxic agents, such as BCL-2 inhibitors, have potential therapeutic applicability. The menin-MLL complex pathway is an emerging therapeutic target. The pathway accounts for the leukemogenesis in AML with MLL-rearrangement, NPM1 mutation, and NUP98 fusion genes. Potent menin-MLL inhibitors have demonstrated promising anti-leukemic effects in preclinical studies. The downstream signaling molecule SYK represents an additional target. However, the TP53 mutation continues to remain a challenge. While the p53 stabilizer APR-246 in combination with azacitidine failed to show superiority compared to azacitidine monotherapy in a phase 3 trial, next-generation p53 stabilizers are now under development. Among a number of non-canonical approaches to TP53-mutated AML, the anti-CD47 antibody magrolimab in combination with azacitidine showed promising results in a phase 1b trial. Further, the efficacy was somewhat better in patients with the TP53 mutation. Although clinical evidence has not been accumulated sufficiently, targeting activating KIT mutations and RAS pathway-related molecules can be a future therapeutic strategy.


Introduction
Acute myeloid leukemia (AML) is an aggressive hematologic malignancy often characterized by specific genomic alterations [1]. The standard treatment strategy for AML is largely consisted of intensive chemotherapy with or without hematopoietic stem cell transplantation. However, long-term survival can be achieved in only up to three-quarters of patients, even in the favorable risk group [2]. Molecular-targeted therapy has had a significant impact on clinical practice, especially for patients with specific genomic abnormalities. The fms-like tyrosine kinase 3 (FLT3) mutation, for example, is known as one of the major adverse prognostic factors in AML. Further, potent FLT3 inhibitors have improved clinical outcomes as a part of salvage/alternative therapy as well as in combination with intensive chemotherapy in patients with FLT3-mutated AML. For patients with iso-citrate dehydrogenase (IDH)-1 or -2 mutation, specific IDH inhibitors ivosidenib and enasidenib are generally well-tolerated and expects complete remission (CR) rates of 30-40% as monotherapy [3,4]. Recently, menin-related leukemogenesis especially in AML with mixed lineage leukemia 1 (MLL)-rearrangement have gathered attention and several preclinical studies have evaluated its specific inhibitors. In addition, anti-tumor protein p53 (TP53), KIT, and RAS strategies have developed both in hematologic malignancies and solid tumors. A number of genomic abnormalities have been identified as potential targets and some have shown promising data in preclinical/early-phase studies. Here, we discuss emerging novel therapeutic approaches for AML, especially those targeting specific genomic abnormalities. Cortes, et al. [16] Blood 2019 FLT3-mutated AML (ITD only) R/R Quizartinib III CR/CRi 48% (118/245) 18.5 mo. [10.8-28.8] StdCTx: standard chemotherapy, CRi: CR with incomplete hematologic recovery, PR: partial response. ITD: internal tandem duplication, TKD: mutation of tyrosine kinase domain. ND: newly-diagnosed, R/R: relapsed or refractory to previous therapy. FLT3 inhibitors may also have an important role in maintenance therapy after allogeneic hematopoietic transplantation (allo-HSCT), and several clinical studies have evaluated the efficacy of sorafenib or midostaurin. A systemic review of 7 such studies including six-hundred-eighty FLT3-mutated AML patients revealed that FLT3 inhibitor maintenance therapy significantly reduced the risk of post-transplant relapse by 65% (HR 0.35, 95% CI 0.23-0.51) and improved survival rates (HR 0.48, 95% CI 0.36-0.64) [20]. The result of a phase 3 trial evaluating the efficacy of gilteritinib as post-transplant maintenance therapy is anticipated [21].
Combinations with low-intensity chemotherapy have also been evaluated. Retrospective and early-phase studies have suggested the modest efficacy of FLT3 inhibitors plus azacitidine or low-dose cytarabine with response rates of one-fifth to one-quarter (Table 2) [22][23][24]. However, the randomized phase 3 LACEWING trial, which compared the combination of gilteritinib plus azacitidine with azacitidine monotherapy in patients with ND FLT3-mutated AML or myelodysplastic syndrome (MDS) who were ineligible for intensive chemotherapy, did not meet the primary endpoint (overall survival) [25]. A combination strategy with non-cytotoxic agents has also been developed. A patientderived xenograft (PDX) mouse model experiment demonstrated that quizartinib in combination with the selective B-cell leukemia/lymphoma 2 (BCL-2) inhibitor venetoclax significantly prolonged survival of FLT3-mutant PDX mice compared with those receiving monotherapy [26]. This study also suggested that FLT3-ITD inhibition led to increased dependance on BCL-2 through indirect attenuation of B-cell lymphoma-extra large (BCL-XL) and myeloid-cell leukemia 1 (MCL-1) which are involved in anti-apoptotic mechanisms. Arsenic trioxide (ATO), one of the key drugs in treating acute promyelocytic leukemia and which induces differentiation of leukemic cells by enhancing RARA gene function, in combination with the FLT3 inhibitors showed a synergistic anti-leukemic effect in vivo along with reduced expression (or facilitated poly-ubiquitination) of FLT3 proteins [27]. A PDX model study suggested that inflammatory genes were upregulated in quizartinib-resistant FLT3-mutant leukemia, and co-administration of dexamethasone with quizartinib showed synergistic cell death in vivo [28]. Isocitrate dehydrogenase (IDH) 1 and IDH2 proteins are house-keeping enzymes involved in the tricarboxylic acid cycle in mitochondria. Mutant IDH1 and IDH2 confer an aberrant enzymatic activity, which converts alpha-ketoglutarate (alpha-KG) to the oncometabolite 2-hydroxyglutarate (2-HG) [29], resulting in epigenetic alterations that prevent hematopoietic differentiation [30,31]. IDH mutations impaired the histone demethylation that is required for cell differentiation through producing 2-HG [31]. Indeed, IDH1/2-mutated leukemic cells displayed global DNA hypermethylation and attenuated function of the alpha-KG-dependent enzyme Tet methylcytidine dioxygenase 2 (TET2) [30], which partially explains why IDH mutations and TET2 mutation are mutually exclusive in AML.
Both IDH1 and IDH2 gene mutations are found in approximately 20% of AML [6] as well as 12% of myelodysplastic syndrome (MDS) especially in high-risk cases [32]. IDH1/IDH2 mutations are also commonly found in high-grade gliomas and IDH1 inhibition prolonged disease control with favorable safety profile in a phase 1 clinical study [33]. The specific IDH1 inhibitor ivosidenib demonstrated a 21.6% CR rate in patients with R/R IDH1-mutated AML in a phase 1 study [34]. Ivosidenib also showed a 42.4% CR plus CR with partial hematologic recovery (CRh) rates as monotherapy [3] and a 60.9% CR rate in combination with azacitidine in patients with newly-diagnosed IDH1-mutated AML who were ineligible for standard therapy [35]. Similarly, the IDH2 inhibitor enasidenib showed a 40.3% overall response with a median response duration of 5.8 months in patients with R/R IDH2-mutated AML (Table 3) [4]. Ivosidenib plus azacitidine therapy for ND AML/MDS is now under evaluation in the phase 3 AGILE study [36]. Combinations with intensive chemotherapy have also been evaluated. A phase 1 clinical study revealed that ivosidenib or enasidenib in combination with standard induction chemotherapy and consolidation therapy showed 73-93% CR/CRh rates with 58-89% minimal residual disease (MRD)-negative rates in patients with IDH1or IDH2-mutated newly diagnosed AML [37].

Targeting Anti-Apoptotic BCL-2
The B-cell leukemia/lymphoma 2 (BCL-2) family regulates the mitochondrial apoptotic pathway by controlling mitochondrial outer membrane permeabilization (MOMP) and the release of cytochrome c [38]. The family consists of proapoptotic proteins (e.g., BCL-2 homology 3 (BH3)-only, BCL-2-associated X protein (BAX), and BCL-2 homologous antagonist/killer (BAK)) and anti-apoptotic proteins (e.g., BCL-2, BCL-XL, and MCL-1) [39]. BCL-2 proteins on the mitochondrial outer membrane surface ordinarily capture cytoplasmic BH3-only proteins and inhibit BAX/BAK, which are activated by BH3-only proteins, then initiate MOMP to release cytochrome c. BCL-2 inhibitors competitively bind to BCL-2 proteins to release BH3-only proteins and disinhibit BAX/BAK, which eventually initiates the apoptotic cascade of tumor cells ( Figure 1) [40]. Chan and colleagues demonstrated that the oncometabolite 2-HG inhibited the activity of cytochrome c oxidase, a key cation channel involved in the electron transport system of mitochondria, and then lowered the threshold of MOMP, resulting in dependence on BCL-2 activity ( Figure 1). In this study, ex vivo and PDX model experiments showed that the BCL-2 inhibitor ABT-199 (venetoclax) significantly suppressed the proliferation of IDH1or IDH2-mutated leukemic cells [41].
In the VIALE-A study, a phase 1b study, which proved the clinical efficacy of venetoclax in combination with azacitidine for previously untreated AML, the subgroup analysis suggested that IDH1 and IDH2 mutations were independently associated with a favorable prognosis (hazard ratio, 0.34) [42]. Notably, venetoclax is available for any type of AML regardless of genomic status, IDH1 and IDH2 mutations might help to predict responses to the agent. Given that FLT3 inhibition facilitates dependance on anti-apoptotic BCL-2, as mentioned above, the combination of FLT3 inhibitors and venetoclax can be a future strategy for AML [26]. inactivates key proapoptotic molecules such as BCL-2 homology 3 (BH3)-only proteins, BCL-2-associated X protein (BAX), and BCL-2 homologous antagonist/killer (BAK). BCL-2 inhibitors allow activation of these molecules to initiate the apoptotic cascade. The oncometabolite 2-hydroxyglutarate (2-HG), which is converted from alpha-ketoglutarate by mutant IDH, inhibits the activity of cytochrome c oxidase and results in decreased threshold of mitochondrial outer membrane permeabilization (MOMP), which eventually leads to dependance on BCL-2.
In the VIALE-A study, a phase 1b study, which proved the clinical efficacy of venetoclax in combination with azacitidine for previously untreated AML, the subgroup analysis suggested that IDH1 and IDH2 mutations were independently associated with a favorable prognosis (hazard ratio, 0.34) [42]. Notably, venetoclax is available for any type of AML regardless of genomic status, IDH1 and IDH2 mutations might help to predict responses to the agent. Given that FLT3 inhibition facilitates dependance on anti-apoptotic BCL-2, as mentioned above, the combination of FLT3 inhibitors and venetoclax can be a future strategy for AML [26].

Targeting MLL-Rearrangement
The mixed lineage leukemia 1 (MLL, also known as KMT2A) gene encoding a histone methyl-transferase is a proto-oncogene involved in a variety of chromosomal translocations [43]. Rearrangements of the MLL gene (MLL-r.) are found in approximately 5-10% of AML, are particularly prevalent in infant leukemias, and are associated with poor prognosis and resistance to chemotherapy [1,44]. Recent studies have suggested that oncogenic MLL fusion proteins interact with the chromatin-associated complex, including disruptor of telomeric silencing 1-like (DOT1L), a histone H3K79 methyltransferase, and the product of the MEN1 tumor suppressor gene (menin), which are required for the initiation of MLLmediated leukemogenesis ( Figure 2) [45,46]. Menin classically functions as a tumor suppressor for the endocrine lineage. However, menin also plays an essential role as a transcriptional cofactor for MLL oncoproteins [46]. Menin links MLL proteins with lens epithelium-derived growth factor (LEDGF), an epigenetic reader recognizing H3K36 histone marks, on cancer-associated target genes to upregulate the transcription [47]. In addition, oncogenic MLL fusion proteins are also linked with the histone H3K79 methyltransferase inactivates key proapoptotic molecules such as BCL-2 homology 3 (BH3)-only proteins, BCL-2associated X protein (BAX), and BCL-2 homologous antagonist/killer (BAK). BCL-2 inhibitors allow activation of these molecules to initiate the apoptotic cascade. The oncometabolite 2-hydroxyglutarate (2-HG), which is converted from alpha-ketoglutarate by mutant IDH, inhibits the activity of cytochrome c oxidase and results in decreased threshold of mitochondrial outer membrane permeabilization (MOMP), which eventually leads to dependance on BCL-2.

Targeting MLL-Rearrangement
The mixed lineage leukemia 1 (MLL, also known as KMT2A) gene encoding a histone methyl-transferase is a proto-oncogene involved in a variety of chromosomal translocations [43]. Rearrangements of the MLL gene (MLL-r.) are found in approximately 5-10% of AML, are particularly prevalent in infant leukemias, and are associated with poor prognosis and resistance to chemotherapy [1,44]. Recent studies have suggested that oncogenic MLL fusion proteins interact with the chromatin-associated complex, including disruptor of telomeric silencing 1-like (DOT1L), a histone H3K79 methyltransferase, and the product of the MEN1 tumor suppressor gene (menin), which are required for the initiation of MLL-mediated leukemogenesis ( Figure 2) [45,46]. Menin classically functions as a tumor suppressor for the endocrine lineage. However, menin also plays an essential role as a transcriptional cofactor for MLL oncoproteins [46]. Menin links MLL proteins with lens epithelium-derived growth factor (LEDGF), an epigenetic reader recognizing H3K36 histone marks, on cancer-associated target genes to upregulate the transcription [47]. In addition, oncogenic MLL fusion proteins are also linked with the histone H3K79 methyltransferase DOT1L via its fusion partners, such as AF9 [48]. Indeed, preclinical studies have suggested that leukemic cells with MLL-r. were dependent on DOT1L activity [49][50][51], which results in upregulation of specific genes essential for hematopoietic proliferation, such as homeobox A9 (HOXA9) and myeloid ecotropic viral insertion site 1 (MEIS1) [52,53]. Although some previous studies suggested potential therapeutic activity of DOT1L inhibitors on AML with MLL-r. [54,55], it turned out to be far from clinically useful [56]. demonstrated uniformly suppressed expression of MLL-target genes, especially MEIS1, in the bone marrow of PDX mice which were treated with the menin-MLL inhibitor. Another menin-MLL inhibitor, MI-3454, also showed complete remission or regression of leukemia in a PDX model accompanied by downregulation of key leukemogenic genes such as MEIS1 [60]. Indeed, the agent was equally effective for MLL-r. and NPM1-mutated leukemia.  Still, several preclinical studies using leukemic cell lines and xenograft mice showed on-target antileukemic effects of menin-MLL inhibitors [57][58][59][60][61]. Krivtsov and colleagues demonstrated a dramatic reduction in leukemic burden when MLL-r. PDX mice were treated with the orally bioavailable menin-MLL inhibitor VTP50469 [61]. This study also demonstrated uniformly suppressed expression of MLL-target genes, especially MEIS1, in the bone marrow of PDX mice which were treated with the menin-MLL inhibitor. Another menin-MLL inhibitor, MI-3454, also showed complete remission or regression of leukemia in a PDX model accompanied by downregulation of key leukemogenic genes such as MEIS1 [60]. Indeed, the agent was equally effective for MLL-r. and NPM1mutated leukemia.

Targeting NPM1 Mutations
Nucleophosmin1 (NPM1) mutations are found in approximately 30% of AML patients and often co-exist with DNMT3A mutations [6,62]. Although mutant NPM1 with absent or low-allelic-ratio FLT3-ITD is known as a favorable prognostic factor of AML [1], NPM1 and FLT3 mutations are not uncommonly found simultaneously [6]. Normal NPM1 protein is a key chaperon protein in the nucleus that maintains genomic stability [63]. Mutant NPM1 protein loses the ability to be transported into the nucleus and consequently accumulates in the cytoplasm [64]. MLL proteins form a macromolecular complex involving menin that maintains expression of HOX genes, resulting in expansion of progenitor cells [65][66][67]. Recent studies have suggested that the interaction of menin and wild-type MLL plays a pivotal role in AML with NPM1 mutation by upregulating leukemogenic genes, such as HOXA, HOXB and MEIS1, similar to the action of MLL-fusion protein ( Figure 2) [62,68,69]. Aberrant demethylation of H3K79, the primary target region of DOT1L, and subsequent upregulation of HOXA9 and PBX3 genes have also been reported [70]. Highly potent menin-MLL inhibitors have been preclinically developed targeting NPM1-mutated leukemia, including MI-3454 mentioned above [60,71]. Interestingly, an in vivo study demonstrated that combined menin-MLL and FLT3 inhibition showed a synergistic anti-leukemic effect on NPM1-mutated and FLT3-mutated AML [72].

Targeting NUP98 Fusion
The nucleoporin 98 (NUP98) gene was originally identified as a component of the nuclear pore complex [73]. NUP98 is involved in a variety of balanced translocations and inversions as the fusion partner of dozens of genes such as HOXA9 and lysine-specific demethylase 5A (KDM5A), also known as JARID1A [74]. The majority of NUP98 fusions are accompanied by overexpression of HOXA9 [75,76], which is associated with poor prognosis in AML [77,78]. Recent studies have suggested that leukemic cells with NUP98 fusions are dependent on MLL protein in terms of recruiting the fusion proteins onto the HOXA locus to initiate leukemogenesis [79]. Given the synergy of MLL-r. and NPM1-mutated leukemia, Heikamp and colleagues demonstrated that the menin-MLL inhibitor VTP50469 prolonged survival of PDX mice with human leukemia following implantation with NUP98-HOXA9 and NUP98-JARID1A, along with suppressed pro-leukemic gene expression and upregulated differentiation markers [80].

Targeting SYK Signaling
Spleen tyrosine kinase (SYK) was originally identified as a signaling molecule downstream of the B cell antigen receptor. SYK also plays a key role in AML in terms of phosphorylating signal transducer and activator of transcription 5 (STAT5) [81] and cooperating with FLT3-ITD in maintaining leukemia [82], and is also associated with an unfavorable prognosis [83]. Mohr and colleagues revealed that Meis1 induced SYK signaling through multiple transcriptional events, including downregulation of microRNA(miR)-146a, which negatively regulates SYK expression in Hoxa9-overexpressing myeloid progenitor cells.
The study also showed that SYK overexpression enhanced Meis1 transcriptional patterns, resulting in dependence on SYK activity in Hoxa9/Meis1-overexpressing myeloid progenitors. Indeed, SYK inhibition prolonged survival of mice with Hoxa9/Meis1-driven leukemia [84]. In an international multicenter phase 1b/2 study, 34 previously untreated AML patients were treated with the SYK inhibitor entospletinib in combination with standard induction chemotherapy, resulting in a 56% CR rate with acceptable toxicity [85]. In this study, patients with HOXA9/MEIS1 overexpression showed significantly better OS (HR 0.32, 95% CI 0.100-0.997) than others (Table 4).

TP53 Mutations 2.4.1. p53 Stabilizers
Tumor protein p53 (TP53) is a major tumor suppressor inducing growth arrest or apoptosis. TP53-induced apoptosis is partially mediated by stimulation of BAX and repression of BCL-2 expression [86,87]. TP53 prevents activity of cyclin-dependent kinase (CDK) 7, a part of CDK-activating kinase (CAK), to stop cell cycle in response to DNA damage [88]. TP53 gene mutations are found in less than 10% of AML patients, is one of the most well-recognized adverse genomic factors, and is often associated with complex karyotype [1,6]. A number of in vitro and in vivo studies revealed that TP53-mutated leukemia gains enhanced self-renewal capacity and a competitive growth advantage, subsequently accumulating additional mutations such as DNMT3A, TET2, and ASXL1 [89]. APR-246 (eprenetapopt) is the first-in-class anti-tumor agent targeting TP53 mutation, which thermodynamically stabilizes p53 protein and shifts the equilibrium toward a functional conformation [90]. In a phase 1b/2 clinical study, the combination therapy of APR-246 and azacitidine produced a 71% overall response rate, including a 41% CR rate, in patients with TP53-mutated high-risk MDS or AML (Table 4) [91]. Despite this promising result, the combination therapy did not meet the primary endpoint (CR rate) in a phase 3 clinical trial for patients with TP53-mutated MDS, though the CR rate tended to be superior in the combination group than the azacitidine monotherapy group (33.3% vs. 22.4%) [92]. The next-generation p53 stabilizer APR-548 is now being evaluated in an early phase clinical trial [93].

Targeting CD47
The transmembrane protein CD47, also known as the "don't-eat-me signal", is the ligand for signal regulatory protein alpha (SIRPα) on macrophages and dendritic cells and results in inhibition of phagocytosis [94]. Increased expression of CD47 on AML stem cells has been shown to be associated with poor prognosis [95]. The anti-CD47 antibody magrolimab in combination with azacitidine showed a 57% CR/CRh rate in patients with treatment-naïve AML (65% had TP53 mutation) who were ineligible for intensive therapy in a phase 1b study. Although the immune escape mechanism through CD47 would not be specific for TP53-mutant AML, the CR/CRh rates were somewhat higher (67%) in TP53-mutated AML in this study (Table 4) [96].

KIT Inhibitors
KIT protein, also known as CD117, is expressed in 70% of AML as well as normal hematopoietic progenitor cells. KIT mutations are seen in less than 10% of all subsets of AML and in approximately 30% of the core-binding factor (CBF) AML [97][98][99]. KIT mutations are associated with worse prognosis in CBF-AML [2,100]. KIT protein is a cell-surface receptor for the cytokine SCF (Kit ligand) and plays an essential role in anti-apoptosis, proliferation, and hematopoiesis via the Ras-Erk pathway, the PI3K/AKT pathway, and the JAK/STAT pathway [101,102]. Activating mutations of KIT gene, commonly on the exon 8 and 17, are frequently found in gastrointestinal stromal tumor (GIST) and systemic mastocytosis, as well as CBF-AML. A number of tyrosine kinase inhibitors (TKIs) targeting KIT such as imatinib, sunitinib sorafenib, regorafenib, dasatinib, nilotinib, ponatinib, and midostaurin are suggested to be a potential therapeutic agent for KIT-mutated tumors [103]. A part or all of imatinib, sunitinib, regorafenib and avapritinib are involved in the standard treatment for GIST and systemic mastocytosis. While some KIT A-loop mutations (e.g., D816 alterations) render resistance to imatinib and sunitinib in patients with GIST, some agents (e.g., avapritinib and ponatinib) remain effective for such resistant cases [104,105]. Although the clinical role of KIT inhibitors in treatment of AML has not been established, there are a few clinical reports suggesting an anti-leukemic effect of KIT inhibitor (e.g., dasatinib) for KIT-mutated AML [106,107].
Kampa-Schittenhelm and colleagues reported 77-years-old patient with KIT D816Vmutated CBF-AML which relapsed after the first-line decitabine monotherapy [106]. Upon careful consideration and informed consent, he received a KIT inhibitor dasatinib as salvage chemotherapy. Peripheral blasts started to reduce on day 15 and, instead, neutrophils/monocytes increased in a few days, suggesting the release of differentiation blockade. The patient's mononuclear cell sample showed dose-dependent cytoreduction ex vivo responding to dasatinib and vanishment of phosphorylated KIT proteins on Western immunoblotting after dasatinib administered. Although the patient eventually shifted to best supportive care because of inacceptable tolerability of dasatinib, this report provided proof-of-concept that KIT inhibition has anti-leukemic effect on KIT-mutated CBF-AML.
Recently, a heat shock protein (HSP)-90 inhibitor pimitespib (TAS-116) showed significant improvement of PFS in patients with GIST resistant to imatinib, sunitinib, and regorafenib in the phase 3 CHAPTER-GIST-301 trial [108]. HSP-90 is a molecular chaperone that engages a variety of clients including KIT by interacting with co-chaperone proteins [109]. Thus, HSP-90 inhibition results in reduced functional stability of KIT then attenuates it's signaling. Preclinical studies have suggested that inhibition of HPS-90 led to tumor shrinkage in human tumor xenograft mouse model as well as depletion of multiple HSP-90 clients [110]. Katayama and colleagues reported that HSP-90 inhibition restored the sensitivity to FLT3 inhibitors in AML cell-lines with FLT3-ITD plus FLT3-TKD (e.g., N676K, F691L, D835V, and Y824C) which render resistance to FLT3 inhibition [111]. HSP90 inhibitors in combination with other TKIs can be a future strategy in AML treatment.

Targeting RAS Pathway-Related Genes
RAS proteins have GTPase activity responding to growth factor receptor activation and play an essential role in cell proliferation [112,113]. RAS genes (e.g., KRAS, NRAS, and HRAS) are one of most popular proto-oncogenes and frequently mutated in human cancer, affecting the mitogen-activated protein kinase (MAPK), phosphatidylinositol-3-kinase (PI3K), and Ras-like (Ral) small GTPase (RalGEF) signaling pathways. Activating mutations of NRAS and KRAS are found in approximately 15-25% of patients with AML [6]. RAS genes commonly co-mutated with RAS-regulating genes (e.g., PTPN11 and NF1) and/or signaling receptor genes that rely on RAS-involving pathways (e.g., FLT3 and KIT) [6,114]. Recent studies of cancer clonal evolution have suggested that additional RAS mutations are associated with resistance to FLT3 inhibitors in FLT3-mutated AML. McMahon and colleagues sequentially analyzed clinical samples from patients with FLT3-mutated AML which progressed on gilteritinib treatment [115]. Among forty-one participants of this study, fourteen patients (34%) acquired new NRAS and/or KRAS mutations after progression on FLT3 inhibition. Almost complete substation of RAS-mutated subclones for the original RAS-wild FLT3-mutated clone was observed in a few cases. In addition, leukemic cell lines with both FLT3 and RAS mutations showed resistance to gilteritinib monotherapy, which was canceled by co-administration of a MEK inhibitor trametinib.
Preclinical experiments have shown that inhibition of both MAPK and PI3K pathway did not cause significant leukemic cell death but led to static effects instead [116,117]. Similarly, early phase clinical trials evaluating MEK or AKT inhibitor in patients with relapsed/refractory AML demonstrated only transient or little anti-leukemic effects [118,119]. Whereas, Pomeroy and colleagues found that NRAS-driven AML cell line relapsed after genetic suppression on NRAS (NRI-AML) was devoid of MAPK and PI3K signaling but dependent on the cyclin-dependent kinase (CDK) 5-mediated activation of RalGEF signaling. In in vivo experiments using PDX mouse model, a CDK inhibitor dinaciclib which inhibits CDK5-mediated RalGEF signaling induced leukemic cell apoptosis and prevented evolution of NRI-AML, suggesting that RalGEF signaling is involved in escaping mechanism of RAS-mutated AML and can be a potential therapeutic target.
The KRAS G12C -specific inhibitor sotorasib demonstrated 7-32% of overall response rates in advanced solid tumors (mainly non-small cell lung cancer and colorectal cancer) with KRAS G12C mutation in a phase 1 study [120] and now available in some countries. The efficacy of KRAS inhibition have not been proved in hematologic malignancies neither clinically nor preclinically. However, given that activating RAS mutations are associated with worse prognosis and drug resistance in AML, KRAS inhibition might be included in future possibility.

Conclusions and Perspective
Molecular-targeted therapies, especially for specific genomic abnormalities, have had significant impacts on AML treatment. FLT3 inhibitors and IDH inhibitors have proved its efficacy in clinical trials and now available in practice. In addition, menin-MLL inhibitors have shown anti-leukemic effects in preclinical studies of AML with menin-related genomic alterations such as MLL rearrangements, NPM1 mutations, and NUP98 fusion genes. SYK inhibitors can be another strategy for AML with menin-related genomic alterations. For AML with TP53 mutation, p53 stabilizers and anti-CD47 antibodies can be a candidate for gene-specific therapeutic agents. Although clinical evidence has not been sufficient yet, anti-KIT strategy such as HSP-90 inhibitors and RAS-pathway interference might be a future approach for AML treatment. Mutations of FLT3, MLL, and TP53 are considered to be unfavorable prognostic factors so far. However, the development of targeted therapies with or without conventional therapy may improve or perhaps reverse the current situation. Aberrant oncogenic mechanisms depending on specific genomic abnormalities can be an ideal therapeutic target, though preventing or overcoming treatment resistance remains a challenge. Determining genomic features and following case-specific molecular-targeted strategy should be future therapeutic strategy in AML. Results from current developments and ongoing trials are anticipated.