Precision and strategic targeting of novel mutation-specific vulnerabilities in acute myeloid leukemia: the semi-centennial of 7 + 3

Abstract

The year 2023 marks the semi-centennial of the introduction of classic ‘7 + 3’ chemotherapy for acute myeloid leukemia (AML) in 1973. It also marks the decennial of the first comprehensive sequencing efforts from The Cancer Genome Atlas (TCGA), which revealed that dozens of unique genes are recurrently mutated in AML genomes. Although more than 30 distinct genes have been implicated in AML pathogenesis, the current therapeutic armamentarium that is commercially available only targets FLT3 and IDH1/2 mutations, with olutasidenib as the most recent addition. This focused review spotlights management approaches that exploit the exquisite molecular dependencies of specific subsets of AML, with an emphasis on emerging therapies in the pipeline, including agents targeting TP53-mutant cells. We summarize precision and strategic targeting of AML based on leveraging functional dependencies and explore how mechanisms involving critical gene products can inform rational therapeutic design in 2024.

Keywords:

• Acute myeloid leukemia
• targeted therapies
• mutational testing
• TP53
• metabolic vulnerabilities

1. Introduction

The mutational landscape of acute myeloid leukemia (AML) has been recently characterized by multiple groups and has led to significant advances in therapeutics [1,2]. In 2008, the first AML genome was sequenced and set the precedence for future studies [3]. To date, AML remains a treatable disease for some patients but a deadly disease for others. The heterogeneity in prognosis for AML is largely reflective of genomic features and various driver mutations, rather than traditional characteristics such as peripheral blood counts, blast count, age, demographics, and/or morphologic assessment [1,4]. Fifty years have elapsed since the introduction of classic ‘7 + 3’ induction chemotherapy in 1973, and remission rates for ‘7 + 3’ range from 60–80% in younger adults and 30–50% among patients above age 60 [5]. The field has seen significant advances in therapeutics in the recent 5 years based on genomically targeted therapies. Current Food & Drug Administration (FDA)-approved therapies that target mutation-specific vulnerabilities include midostaurin, gilteritinib, enasidenib, ivosidenib, and olutasidenib [6–11]. However, comprehensive genomic profiling has revealed that dozens of mutations (dispersed among several functional clusters) are involved in AML pathobiology, and most of these mutations are not targetable by currently available therapeutics (Figure 1). Numerous clinical trials in various phases have been performed with variable efficacy results, but FDA approvals have been limited by lack of solid Phase 3 data.

Figure 1. Functional genomic clusters in AML, organized alphabetically by known recurrently mutated genes. Width of chords in the circos plot do not correspond to frequency of aberrations.

2. Key principles of molecular oncology involving AML therapeutics

Key principles that govern the ability of novel agents to target mutation-harboring cells include synthetic lethality, oncogene addiction, and stemness. Pre-clinical data has exploited such dependencies and led to translational efforts for various investigational agents.

Synthetic lethality is a well-established paradigm in which cell death occurs with disruption of 2 different genes or gene products but not disruption of 1 of those genes or gene products [12,13]. Concurrent inactivation of both candidate genes affects cell viability, and the therapeutic power of this concept has been harnessed in patients who have heterozygosity of one gene. The classical example of synthetic lethality in clinical oncology involves breast cancer, in which BRCA-mutant breast cancer cells are particularly susceptible to PARP inhibitors. BRCA-mutant cells harbor defective double-stranded break repair and thus rely on base excision repair pathways to repair single-stranded breaks such that the cell can survive. Inhibition of PARP in these cells results in cell death since PARP is involved in base excision repair.

Oncogene addiction refers to the concept whereby a genomic lesion drives cancer cell proliferation and becomes critical for the viability of the tumor [14]. In many cases, mutations involved in genes encoding cell signaling confer a hyperproliferative phenotype, which leads to expansion of the cancer cell population. A classic example is activating EFGR mutations in lung cancer.

Finally, stemness considerations factor into the curative potential for targeted therapeutics. Leukemia stem cells were first described by the laboratory of Dr. John Dick in the 1990s, and these seminal studied paved the way for translational efforts to targets such cells [15,16]. Leukemia-initiating cells can reconstitute malignant hematopoiesis and fuel the production of mature blasts, so elimination of the stem cell compartment is critical in order to achieve durable and deep treatment responses [15–17]. Mechanisms of critical gene products and stem cell circuits can inform selection of therapeutic targets. Such targets may be the key to cell survival, and therapeutics against these targets may eliminate AML cells.

3. Mutation-specific vulnerabilities

3.1. Cohesin-Mutant AML

3.1.1. Functional biology

Genes encoding the cohesin complex include STAG1/2, SMC1A, RAD21, and SMC3. Cohesin proteins are involved in sister chromatid separation and DNA damage response [18]. Cohesin complex members are recurrently mutated in approximately 6–13% of AML genomes [4]. These mutations are thought to occur early in leukemogenesis, similar to NPM1 mutations [18]. The laboratory of Dr. Ravindra Majeti showed that, in primitive myeloid cell populations, mutations in the cohesin complex dominantly reinforce stem cell circuits and prevent differentiation, alluding to the role of cohesin members in leukemic stem cell biology [18]. STAG1 and STAG2 disruption specifically leads to enhanced hematopoietic stem cell (HSC) self-renewal, and loss of function of these proteins impairs differentiation [19]. Of note, STAG2 and SMC1A are located on the X chromosome, so males with these mutations have no intact wild-type gene in the malignant clone [19,20].

3.1.2. Precision targeting

Targeting of cohesin-mutant cells has been attempted through synthetic lethality. Synthetic lethality can be achieved in STAG2-mutant cell if STAG1 is disrupted: this abrogates sister chromatid adhesion [19,20]. The wild-type forms of both proteins redundantly ensure survival. Genetic dependency screens by the laboratory of Dr. Benjamin Ebert have identified a key role for STAG2 in DNA damage response [21]. In a murine model of STAG2-mutant AML arising from TET2-disrupted clonal hematopoiesis, the PARP inhibitor talazoparib selectively eliminated these cells. The basis for this was that DNA damage repair and DNA replication are functional dependencies of cohesin-mutant AML cells, and pharmacologic disruption of base excision repair results in cell death (Figure 2).

Figure 2. Summary of mutation-specific vulnerabilities and molecular dependencies in the experimental or pharmacologic pipeline for AML. Phase of experimentation for each therapeutic is included, ranging from pre-clinical through FDA approval.

Other therapies that have been studied in cohesin-mutant AML include the exportin-1 (XPO1) inhibitor selinexor as well as DOT1L inhibitors. Selinexor has been studied in a Phase 2 trial in relapsed/refractory (R/R) AML (including all comers and not just cohesin-mutant AML), but this was a negative trial [22]. Complete response rate in the cohesin-mutant subset was 33%. DOT1L, a histone methyltransferase, has been shown to be a therapeutic target in cohesin-mutant AML, and DOT1L inhibitors decrease H3K79 dimethylation and reverse the transcriptional changes induced by loss-of-function cohesin mutations [23].

3.2. Spliceosome-Mutant AML

3.2.1. Functional biology

Recurrent missense mutations in RNA splicing factors have been found in AML genomes. Specifically, SRSF2 and ZRSR2, as well as genes encoding SF3B complex members such as SF3B1 and U2AF1, are often mutated early in myeloid leukemogenesis. Cancer cells harboring such mutations rely on wild-type spliceosome activity, and further disruption of this complex pharmacologically may kill these cells. CRISPR screens have identified RNA-binding protein dependencies in human AML, including a key role for the RNA-binding motif protein 39 (RBM39) and the SF3B complex [24].

3.2.2. Precision targeting

Two major avenues of precision targeting of spliceosome-mutant AML include small molecules that target core RNA splicing catalytic reactions and small molecules that inhibit RBM39. E7107 was studied as the first-in-class spliceosome complex modulator in clinical trials [25]. The SF3B complex modulator H3B-8800 has been shown to preferentially eliminate spliceosome-mutant cells. H3B-8800 is an oral selective inhibitor of the spliceosome complex that interferes with branch point adenosine recognition and has demonstrated efficacy in vivo [26]. In a Phase 1 trial of H3B-8800 involving 38 patients with AML, as well as other myeloid malignancies, H3B-8800 increased red blood cell transfusion independence, though no objective complete or partial responses were observed [27]. Importantly, patients were not required to have mutations in genes encoding spliceosome components for enrollment, although most (66%) did have a mutation in spliceosome machinery.

RNA binding proteins are essential for RNA splicing, and CRISPR screens performed by the laboratories of Dr. Omar Abdel-Wahab and Dr. Iannis Aifantis have revealed RNA-binding proteins that are upregulated in AML cells [23]. Such genome-wide screens have shown that RBM39 is a critical regulator for AML cell survival. Molecular inhibition of RBM39 results in synthetic lethality in spliceosome-mutant AML in vivo [23]. Translational efforts remains to be conducted.

In addition to these inhibitors, very recent data on the mechanisms of anemia in SF3B1-mutant myeloid neoplasm have uncovered a role for vitamin B5 and succinyl-coA [28]. Mis-splicing of coenzyme A synthase by SF3B1-mutant cells affects heme biosynthesis and depletes succinyl-CoA. Exogenous supplementation with vitamin B5, the substrate of CoA synthase, can rescue hematopoiesis [28]. This may pave the way for future clinical trials.

3.3. NPM1-Mutant AML

3.3.1. Functional biology

Nucleophosmin 1 (NPM1) is a nuclear-to-cytoplasmic shuttle protein that primarily resides in the nucleolus and interacts with XPO1. Wild-type NPM1 is a nucleolar stress sensor and functions in duplication of centrosomes, assembly of histones, and biogenesis of ribosomes [29]. It plays a role in maintenance of genomic integrity. From 2013 TCGA data, NPM1 was among the most common recurrently mutated genes in AML [4]. Mutant NPM1 maintains the leukemic state via HOX gene expression [29]. Patients with AML harboring the NPM1 mutation often have bone marrow hypercellularity with low or absent CD34 expression and also have a relatively favorable outcomes with cytotoxic chemotherapy if NPM1 is the sole mutation.

3.3.2. Precision targeting

The appeal of targeting NPM1 resides in the fact that NPM1 is a multi-domain protein that interacts with druggable targets. Disruption of NPM1 can promote unfolding, and NPM1-mutant AML cells may be susceptible to inhibition of chromatin regulators, as aberrant HOX gene expression is a hallmark of NPM1-mutant cells. CRISPR screens have shown a key dependency of NPM1-mutant AML cells on the menin-MLL interaction. Histone modifiers have been shown to be pharmacologic targets in NPM1-mutant AML. Inhibition of the menin-binding site in the menin interaction domain of MLL1 and inhibition of DOT1L have been shown to decrease HOX and FLT3 expression in NPM1-mutant AML [30]. The combination of menin-MLL1 and DOT1L inhibition results in anti-leukemic effect [30]. Other menin inhibitors under investigation include KO-539, SNDX-5613, JNJ-75276617, DS-1594b and BMF-219 [28–30].

Given that NPM1 is a nucleolar stress sensor, efforts to exploit this mechanism have been undertaken. Actinomycin D is an inhibitor of RNA polymerase I and ribosomal biogenesis that can induce nucleolar stress in NPM1-mutant cells. In a Phase 2 clinical trial, actinomycin D was shown to induce complete responses in NPM1-mutant AML [31].

NPM1-mutant cells also have a dependency on XPO1 activity. Pharmacologic disruption of nuclear export of cytoplasmic NPM1 (NPM1c) via selinexor, a nuclear export inhibitor, results in downregulation of NPM1c and promotes differentiation of AML. This, in turn, has shown a survival benefit in murine models [29]. The second-generation XPO1 inhibitor eltanexor has been shown to improve survival in vivo via irreversible HOX gene downregulation in NPM1-mutant cells [32]. Translational efforts are underway.

Single-agent venetoclax has also shown efficacy for NPM1-mutant AML based on subgroup analysis from the CAVEAT trial [33]. The response rate for NPM1-mutant AML to single-agent venetoclax was 80%. Long-term data remains to be seen.

3.4. TP53-mutant AML

3.4.1. Functional biology

Recent studies of prognostication in myelodysplastic neoplasms (MDS) and AML have shown that TP53 aberrations constitute the highest risk molecular subgroup of myeloid neoplasms [17,34]. Specifically, a major recent effort that gained prognostic clarity on this genetically defined subgroup is the IPSS-M model for MDS [35]. As the guardian of the genome, p53 plays an integral role in genomic integrity, and disruption of TP53 is often linked with complex karyotype and treatment refractoriness [36].

3.4.2. Precision targeting

Early attempts at improving outcomes in TP53-mutant AML studied an extended course of decitabine. Welch et al. reported that the response rate for 10-day decitabine was 67% in patients with unfavorable-risk cytogenetics and 100% in patients with TP53 mutation [37]. Numerous efforts have been made in the following years to translate TP53 mutation-directed or p53 pathway-directed therapeutics into the clinic. These agents include eprenetapopt and nutlin analogs. However, the mutant p53 protein as a therapeutic target has been quite elusive in the clinical setting.

Eprenetapopt (or APR-246) is reactivator of p53 and works via covalent interactions with the core domain of p53. This agent induces apoptosis by rescuing the mutated form of p53 and restoring its wild-type function. However, the mechanism is not purely p53 reactivation, as activity is not abrogated by treatment of p53 knockout cells. In two back-to-back studies in the Journal of Clinical Oncology, Sallman et al. and Cluzeau et al. studied eprenetapopt [38,39]. The deployment of eprenetapopt and azacitidine in MDS showed a response rate of 73%, with a complete response rate of 50%. However, primary endpoints were not met in Phase 3 trials. A Phase 2 Groupe Francophone des Myélodysplasies (GFM) trial of eprenetapopt and azacitidine showed an overall response rate of 62% and complete response rate of 47% for patients with MDS and AML. Among responders, 73% achieved measurable residual disease (MRD)-negativity by next-generation sequencing of the TP53 mutation [39]. Eprenetapopt administered with azacitidine has also been recently studied in the post-transplant maintenance setting in AML in a Phase 2 trial [40]. Median relapse-free survival was 12.5 months, with a median overall survival of 20.6 months. Of note, Phase 3 trials have not been performed to date.

TP53 targeting was also studied in the setting of CD47 inhibition. CD47 was first reported as a leukemic stem cell marker in two back-to-back studies in Cell in 2009 at Stanford [41,42]. CD47 serves as an anti-phagocytic signal when interacting with its ligand SIRPα on macrophages. Magrolimab (hu5F9-G4) is an anti-CD47 antibody that blocks the macrophage checkpoint and thus facilitates phagocytosis. Magrolimab was investigated in a Phase 1B trial involving patients with TP53-mutant AML [43]. This was studied in combination with azacitidine, and the response rate within the TP53-mutant subgroup was 88%. Very recent Phase 1 data showed 40% complete response rate with median overall survival of 16.3 months for TP53-mutant MDS [44]. Overall response rate reported at ASCO 2022 for azacitidine plus magrolimab for TP53-mutant AML was 48.6% in intention-to-treat analysis. A Phase 3 trial of magrolimab in combination with azacitidine versus physician’s choice of venetoclax plus azacitidine or intensive chemotherapy in treatment-naive patients with TP53-mutant AML is in progress.

Seminal studies at MD Anderson Cancer Center in the pre-genomic era showed a role for the MDM2 antagonists called nutlins in AML [45]. Nutlin analogs are small molecule MDM2 inhibitors that prevent degradation of p53 and thereby facilitate cell cycle arrest and apoptosis [45]. In a Phase 1 study, the MDM2 inhibitor RG7112 showed single-agent activity in patients with AML with wild-type TP53 [46]. These early studies led the path for Phase 3 trials of nutlins in R/R AML [47]. In the Phase 3 MIRROS trial, idasanutlin was studied in combination with cytarabine for R/R AML [47]. The overall response rate was 38.8% and the complete response rate was 20.3%. However, the primary endpoint (overall survival) in the wild-type TP53 intention-to-treat population was not met (median, 8.3 vs. 9.1 months with idasanutlin-cytarabine vs. placebo-cytarabine). Very recent Phase 1B data on venetoclax plus idasanutlin in R/R AML has shown a composite complete remission rate of 20% in patients with TP53 mutation [48].

3.5. IDH1-Mutant AML

3.5.1. Functional biology

The first description of IDH1/2 mutations in AML was in 2009: IDH1 mutations were found at a frequency of 6–10% [49]. Mutant IDH1 results in production of the oncometabolite 2-hydroxyglutarate, which leads to a differentiation block via inhibition of the hydroxylation and demethylation activity of TET2. IDH1-mutant AML cells are dependent on activity of BCL-2 as well as acetyl CoA carboxylase 1 (ACC1) [13,50].

3.5.2. Precision targeting

Ivosidenib was first studied in the R/R setting for IDH1-mutant AML and showed an overall response rate of 41.6% and complete remission rate of 21.3%, and this led to its FDA approval [9]. The very recent Phase 3 AGILE trial studied azacitidine plus ivosidenib compared to azacitidine alone [51]. Event-free survival at 12 months was 37% vs. 12%, respectively. The median overall survival was 24 months vs. 7.9 months, respectively. The safety profile was favorable in the azacitidine plus ivosidenib. The AGILE trial paved the way toward lower-intensity and better tolerated precision therapeutics for patients who are ineligible for induction chemotherapy.

Olutasidenib (FT-2102), a potent small molecule inhibitor of mutant IDH1, was studied very recently in the Phase 1/2 multicenter trial of IDH1-mutant AML or high-risk MDS. For R/R AML, the overall response rate was 41% [10]. Concurrent data published in 2023 showed a composite complete response rate of 35%, with overall response rate of 48% [10,52]. The median overall survival was 11.6 months, and 34% of patients achieved transfusion independence. Based on these results, the FDA approved olutasidenib for patients with relapsed or refractory AML with IDH1 mutations.

Very recent synthetic lethal experiments involving IDH1-mutant AML showed a role for pharmacologic inhibition of BCL-2 as well as ACC1 [13,50]. For this reason, venetoclax-based therapy works well in IDH1/2-mutant AML. The allosteric ACC1 inhibitor 5-tetradecyloxy-2-furoic acid (TOFA) can reduce the viability of IDH1-mutant cells. The novel selective ACC1 inhibitor also has activity against venetoclax-resistant, IDH1-mutant AML [50]. This mutation-specific vulnerability merits investigation in the clinical setting.

3.6. IDH2-Mutant AML

3.6.1. Functional biology

Wild-type IDH2 plays an essential role in AML metabolomics, similar to IDH1. Studies from the CALGB cohort in 2010 showed an IDH2 mutation frequency as high as 19% [53]. TCGA data showed a frequency of 20% for both IDH1/2 mutations [4].

3.6.2. Precision targeting

Part of the pre-clinical rationale for targeting IDH2-mutant AML came from the Majeti laboratory, in which synthetic lethality screens showed a dependency of IDH2-mutant AML on BCL-2, thereby conferring sensitivity to the BCL-2 inhibitor venetoclax [13]. Enasidenib, the small molecule inhibitor of the mutant form of IDH2, was studied in a Phase 1/2 trial in R/R AML. Overall response rate was 40% [8]. Median overall survival was 9.3 months with enasidenib. In the very recent randomized Phase 3 trial published in 2023, enasidenib was shown to confer clinically meaningful morphologic and hematologic responses in patients with IDH2-mutant R/R AML after 2 or 3 prior AML-directed therapies [54]. There was improvement in median event-free survival compared to conventional care (4.9 months vs. 2.6 months) [54]. However, the primary study endpoint was not met; median overall survival was not significantly different for enasidenib vs. conventional care (6.5 months vs. 6.2 months). This analysis was confounded by early dropout and subsequent AML-directed therapies. Enasidenib remains the only commercially available mutant IDH2 inhibitor to date, but other inhibitors are under investigation. On-target drug resistance to mutant IDH1/2 inhibitors also remains an area of investigation.

3.7. FLT3-Mutant AML

3.7.1. Functional biology

FLT3-ITD and FLT3-TKD mutations are late-arising genomic aberrations in AML. FLT3 mutations confer increased proliferative potential upon AML cells, as these mutations can result in autonomous cell activation independent of the FLT3 ligand binding to the FLT3 receptor. Recurrent FLT3 mutations were among the first described mutations in AML genomes, with an estimated frequency of 28% according to TCGA data [4]. FLT3 aberrations typically occur late in leukemogenesis.

3.7.2. Precision targeting

Oncogene addiction plays a role in FLT3-mutant AML, as the FLT3 mutation serves a driver of cell proliferation in AML. Early attempts at targeting the FLT3 mutation began in the 2010s, when the Phase 2 SORAML trial showed a 12-month improvement in the event-free survival with sorafenib compared to placebo [55]. However, toxicity with sorafenib was quite high, given the non-selectivity of sorafenib. There are currently two commercially available FLT3 inhibitors: the multikinase inhibitors midostaurin and gilteritinib. Midostaurin was studied in a Phase 3 RATIFY trial (CALGB 10603), which showed that the combination of 7 + 3 induction chemotherapy plus midostaurin led to improved overall survival compared to 7 + 3 alone [6]. Gilteritinib was studied in the ADMIRAL trial in the R/R setting and showed a composite complete response rate of 34%, compared to 15.3% in the chemotherapy arm [7]. In the very recent Phase 3 LACEWING trial, Wang et al. reported improved composite complete response rates with gilteritinib plus azacitidine compared to azacitidine alone (58.1% vs. 26.5%) for newly diagnosed FLT3-mutant AML in patients ineligible for intensive chemotherapy [56]. In the post-transplant setting, the Phase 3 BMT CTN 1506 trial has completed accrual and will report on relapse-free survival for gilteritinib maintenance. Data will be presented at the European Hematology Association (EHA) conference in 2023 [56].

Other FLT3-targeted therapies (commercially unavailable) that have been studied in clinical trials include quizartinib and crenolanib [57]. Quizartinib was studied in the multi-center Phase 3 QuANTUM-R trial and showed improved median overall survival compared to conventional salvage chemotherapy (6.2 months vs. 4.7 months) for R/R AML [57]. Crenolanib has pan-selective FLT3 activity, and Phase 2 trial data has been reported [11]. A Phase 3 trial randomizing patients with newly diagnosed FLT3-mutant AML to crenolanib versus midostaurin with daunorubicin and cytarabine is ongoing. FLT3 BiTEs have been studied in the recent years: these can induce T-cell dependent cellular toxicity in vitro and in vivo against FLT3-mutant cells [58]. Importantly, the FLT3-directed BiTEs do not necessarily require presence of the FLT3 mutation: high expression of wild-type FLT3 may be sufficient for effective tumor lysis by T cells. On-target resistance to several of these FLT3 inhibitors merits further research.

3.8. RARA-aberrant AML

3.8.1. Functional biology

Acute promyelocytic leukemia, which is a specific subtype of AML, is characterized by rearrangement of the retinoic acid receptor alpha (RARA) locus, most commonly due to translocation of chromosomes 15 and 17. This aberration leads to a differentiation block in myeloid cells. In addition to rearrangements of the RARA locus, RARA can also be overexpressed in about 30% of AML cells [55]. Enhancer landscape analysis of AML genomes has shown that the RARA gene locus harbors a strong superenhancer which may be targetable [59].

3.8.2. Precision targeting

The original hallmark of targeted therapy in AML is the treatment of acute promyelocytic leukemia with all-trans retinoic acid (ATRA) and arsenic trioxide (ATO). The APL0406 trial showed two-year event-free survival as high as 97% [60]. Such superb clinical outcomes for RARA-aberrant AML have rendered the ATRA/ATO combination as one of the most successful targeted therapies for AML.

Seminal data at Stanford University showed that the presence of a RARA superenhancer in myeloid neoplasms confers sensitivity to the selective RARA agonist tamibarotene [59]. This pre-clinical data set the precedence for the SELECT-AML-1 and SELECT-MDS-1 trials where tamibarotene was administered in combination with venetoclax and azacitidine or azacitidine, respectively. Initial results from the Phase 2 open-label multi-center SELECT-AML-1 trial showed a composite complete response rate of 100% in response-evaluable newly diagnosed patients with RARA-positive AML. This trial is still active recruiting (NCT04905407), and the field will await additional efficacy data.

3.9. MLL-rearranged AML

3.9.1. Functional biology

Recurrent translocations involving the KMT2A (MLL) locus on chromosome 11q23 are found in approximately 10% of all cases of AML and impart high risk disease behavior [61]. MLL has nearly 60 different fusion partners that have been described to date, and such fusions are often leukemogenic, especially in murine models [61]. Translocations involving the MLL locus create a chimeric protein that recruits DOT1L, a histone methyltransferase, and HOX-A family members to sites of transcription. Co-occurring mutations in the RAS pathway are common in MLL-rearranged AML and can confer a proliferative phenotype. One of the largest cohorts to date showed uniformly poor outcomes for various subgroups of MLL-rearranged AML [61].

3.9.2. Precision targeting

The critical dependency of MLL-rearranged leukemia cells on the MLL fusion allows for targeting of this mutation-specific vulnerability. The most well-described targets are DOT1L and menin [62]. DOT1L and menin are members of the MLL-AF9 multi-protein complex, and inhibition can lead to degradation of MLL-AF9 [62,63]. The small molecule aminonucleoside DOT1L inhibitor EPZ-5676 showed high selectively for DOT1L over other methyltranferases and resulted in potent leukemia cell killing with complete tumor regression [62,63]. The menin-MLL inhibitor VTP50469 can displace menin and MLL from chromatin and can reduce leukemia cell burden in murine models [62,63]. In the Phase 1 AUGMENT 101 trial, SNDX-5613 (revumenib) has also shown anti-leukemic activity in R/R MLL-rearranged AML [64]. Its on-target effect involves disruption of menin-MLL1 interaction. Small molecule inhibitors have also been developed against ENL, a fusion partner for MLL. TDI-11055 is an ENL inhibitor that displaces ENL from chromatin and can hinder disease progression. In a Phase 1 study, the DOT1L inhibitor pinometostat has shown clinically meaningful responses in patients with AML. The field awaits long-term data. Other menin inhibitors under investigation include KO-539, JNJ-75276617, DS-1594b and BMF-219. Such epigenetic targeting forms the basis for precision therapeutics for MLL-rearranged AML.

4. Challenges and solutions to precision therapeutics

The current standard of care for many patients with AML is the combination of azacitidine plus venetoclax from the VIALE-A trial, which showed a clear overall survival beefit across multiple subgroups (DiNardo et al. NEJM 2020). There is an emerging role for targeted BCL2 inhibition in combination with existing therapies as well as with investigational therapies such as magrolimab. Only a handful of targeted therapeutics are commercially available as of 2023 for the treatment of AML with specific genomic aberrations. However, the mutational spectrum spans more than 100 genes and nearly a dozen functional groups. The lack of readily available therapeutics against the known recurrent gene mutations suggests a major unmet need in this area.

One barrier is the significant heterogeneity even within a given mutational subset of AML. Not all AML cells of a given mutational subclass are created equally: the functional diversity within each subclass often leads to different clinical trajectories for individual patients. The co-occurrence of other mutations within an AML mutational subset can be highly variable. Mutational epistasis occurs in myeloid neoplasms, and such gene-gene interactions might hinder the efficacy of a targeted therapy against a certain mutation-specific vulnerability. Gene-gene interactions must be considered when developing novel therapeutic approaches.

Another barrier is the presence of a malignant stem cell compartment within each patient’s AML. Complex clonal architecture involving self-renewing cells is a feature of AML and is an obstacle to effective therapy. Genomic instability can lead to the development of resistant clones. Efforts at targeting certain mutations may thus not be effective at eliminating the stem cell fraction of a given patient’s AML. There is more robust evidence that DNMT3A, TET2, and TP53 mutations arise within the self-renewing compartment, but other mutations including FLT3 are often late-arising. Although there is high clinical value in targeting the late-arising proliferative subclones, as noted by the success of numerous FLT3 inhibitors, responses may be transient if stem-like clones persist. Thus, even though targeting mutation-specific vulnerabilities may eliminate the bulk of leukemic cells at a given time point, leukemic stem cells can eventually reconstitute malignant hematopoiesis.

Another significant barrier is the acquisition of resistance and development of bypass pathways with targeted therapies. Unlike cytotoxic chemotherapeutics which destroy cells en bulk, targeted therapies eliminate only the cells with a specific genotype or phenotype. Resistance pathways might emerge, leading to repopulation of the blood compartment by these cells. Stem cell transplantation still remains the major curative intent therapy, but transplantation often requires a patient to be subjected to high-dose chemotherapeutics.

In addition, aging associated with high-risk disease poses a barrier to the efficacy of targeted agents. Our aging population is more likely to skew the epidemiology of AML in favor of a higher incidence of adverse-risk AML compared to favorable-risk AML, as genomic disruptions often accumulate with aging. In such cases, targeted therapies may be less effective at eliminating cells. For example, one may presume that AML with antecedent MDS is a difficult-to-treat leukemia in an octogenarian, given high-risk disease biology and limited functional reserve to tolerate therapy. However, it is exactly these patients with the most to gain from the discovery and validation of potentially well-tolerated targeted therapeutics.

A handful of solutions may help enhance target validation and drug development in AML. Synthetic lethal screens in AML may facilitate target validation through identification of critical dependencies and thus vulnerabilities for specific genomically defined subsets. Combinatorial strategies may translate into improved efficacy. In an effort to address the intra-patient heterogeneity, one solution is to perform single-cell genomic assessments. Resolution of each patient’s mutational state at the single-cell level may help minimize off-target effects and facilitate a precision approach to care. Clinical trial data from cooperative group studies are currently available for various targeted therapies, and it is worthwhile to perform subgroup analyses for these trials with respect to treatment response of certain mutational subsets. Biomarker assessment from cooperative group trials may help refine the therapeutic impact of targeted agents. Such efforts will help each and every patient gain access to a personalized therapeutic landscape for such a heterogeneous disease.

Ethical approval

There was no testing or experimentation involving animals, patients, or other subjects.

Disclosure statement

SAP serves on the Acute Myeloid Leukemia advisory board for Bristol Myers Squibb and served on the Multiple Myeloma advisory board for Pfizer.

Data availability statement

There was no new data generated.

Additional information

Funding

SAP receives research funding from the UMass Center for Clinical and Translational Science (CCTS) Pilot Project Program grant [NIH/NCATS Grant UL1TR001453].