Intended for healthcare professionals

Clinical Review State of the Art Review

Advances in acute myeloid leukemia

BMJ 2021; 375 doi: (Published 06 October 2021) Cite this as: BMJ 2021;375:n2026
  1. Laura F Newell, assistant professor,
  2. Rachel J Cook, associate professor
  1. Knight Cancer Institute, Hematology and Medical Oncology, Oregon Health & Science University, Portland, OR, USA
  1. Correspondence to: L F Newell newelll{at}


Acute myeloid leukemia (AML) is an uncommon but potentially catastrophic diagnosis with historically high mortality rates. The standard of care treatment remained unchanged for decades; however, recent discoveries of molecular drivers of leukemogenesis and disease progression have led to novel therapies for AML. Ongoing research and clinical trials are actively seeking to personalize therapy by identifying molecular targets, discovering patient specific and disease specific risk factors, and identifying effective combinations of modalities and drugs. This review focuses on important updates in diagnostic and disease classifications that reflect new understanding of the biology of AML, its mutational heterogeneity, some important genetic and environmental risk factors, and new treatment options including cytotoxic chemotherapy, novel targeted agents, and cellular therapies.


Acute myeloid leukemia (AML) is a heterogeneous hematologic malignancy characterized by the proliferation of myeloid blasts or progranulocytes that fail to undergo normal differentiation. Outcomes for patients with AML are often described as “dire” or “poor,” despite the fact that patients are sometimes cured and life can often be prolonged with treatment. Standard intensive chemotherapy for AML (combining an anthracycline and cytarabine) has been in use for more than 40 years. Recently, major advances in molecular and cell biology have improved our understanding of the pathophysiology of AML, expanded treatment options, and provided a deeper understanding of how and why individual patients may be at increased risk for development of leukemia. These advances provide personalized treatment options for many patients and may provide future opportunities to preempt the development of AML. We have not included a discussion of acute promyelocytic leukemia (APL), which has unique disease biology, diagnosis, treatment, and monitoring algorithms.


An estimated 21 450 new cases of AML occurred in the US in 2019 (1.2% of all new cancer diagnoses), with an increasing incidence with advanced age: 25.1% of new AML cases were among adults aged 65-74 years and 33.7% among those aged 75 and older. The five year survival for patients with AML is a dismal 28.3%, with an estimated 10 920 deaths from AML in 2019. As with the incidence, the percentage of deaths also increases with age, being highest in patients aged 75 or older (43.7%).1

Sources and selection criteria

For this review, we searched PubMed and Medline databases from 2010 to 2020. We also included studies written before 2010 that were highly cited or of historical significance, and we selected only articles from peer reviewed journals written in English. We used the search terms “acute myeloid leukemia”, “AML”, “AML and decision making”, “FLT3”, “genetic and AML”, “bone marrow transplant and AML”, “stem cell transplant and AML”, “CHIP and AML”, “CAR-T and AML”, and “treatment and AML”. We prioritized studies by their relevance to the search criteria, by study year and publication date, and by study design and number of patients included. We excluded case reports but included selected small case series.

We prioritized larger, more recent studies, as well as landmark and novel trials in the field, when available. As expected, in the section on emerging therapies, we found many early phase trials with low numbers of patients. We took into account the fact that the review process was inherently limited by the prevalence of the disease (many of the pivotal phase III registration studies enrolled only a few hundred patients, and more recent studies of novel agents often enroll <100). Additionally, patient outcomes are often censored at time of hematopoietic stem cell transplant (HSCT), leading to abbreviated time spent on the new therapies. Consequently, this review is not intended to be formally systematic, but our evaluation of combined datasets from trials allows for reinterpretation of risk stratification and deeper understanding of initial metrics used for overall prognostication and prediction of responses to specific therapies.23

Clinical presentation

The presenting signs and symptoms can vary from patients without symptoms identified, through routine screening blood work, to those with critical illness due to sepsis in the setting of severe bone marrow failure. Early manifestations of disease are non-specific and may be initially seen by care givers of disparate types (fig 1). Assessing a patient’s medical history, social history, and general fitness early in the evaluation is important, as these factors weigh heavily in the decisions about treatment. Recent studies have shown the prognostic value of fitness evaluations, particularly in older adults, and several reviews provide clear roadmaps for completing such evaluations.456

Fig 1
Fig 1

Clinical presentation of acute myeloid leukemia (AML). Patients with a new AML diagnosis may present for care across the healthcare system, with patients seeking care for varied manifestations of AML: dentists for gingival hyperplasia, dermatologists for new rashes, cardiologists for cardiac symptoms related to anemia, neurologists for new weakness or cranial nerve abnormality, or primary care providers for progressive fatigue, bruising, fevers, lymphadenopathy/organomegaly, palpable masses (myeloid sarcoma), and recurrent infections. Patients range from mildly to catastrophically ill but benefit from early identification


Formal diagnosis of AML can be made either from a peripheral blood sample or from a bone marrow biopsy, depending on the white blood count and the presence or absence of circulating blasts. Although some leukemic disorders can be diagnosed by the detection of pathognomonic chromosomal abnormalities—for example, APL with chromosome 15;17 translocation (t(15;17) and core binding factor leukemias (t(8;21), inv(16), or t(16;16)—ordinarily the diagnosis of AML requires the presence of an abnormal fraction of immature myeloid cells (either myeloblasts or promyelocytes). Histologically, myeloid leukemias require the presence of ≥20% blasts in either a blood or a bone marrow sample. A bone marrow assessment typically includes both an aspiration of the liquid portion of the marrow and a core biopsy. Immunophenotyping is done on the aspirate or, if the marrow is inaspirable, can be done on a core that has been crushed.

Flow cytometric analyses for certain cell surface and cytoplasmic markers confirm the diagnosis of AML, and the specific immunophenotype at diagnosis is useful after treatment to evaluate for persistence of disease. Cytogenetic analysis of metaphase chromosomes reveals the “karyotype,” the most prognostic element of the patient’s leukemia evaluations. The World Health Organization requires evaluations of AML to include clinical history (for example, a patient who has previously undergone treatment with cytotoxic chemotherapy or radiation will automatically be given a diagnosis of a “treatment related AML”), cytogenetics, molecular genetics, morphology, and immunophenotype.78 Molecular testing with next generation sequencing (NGS) has gained importance at diagnosis, for risk stratification, to identify targets for treatment, and as a baseline with which future assessments of disease burden will be compared to assess for depth of response.91011 All patients should be assessed for comorbidities, functional status, and social support, as well as for their goals and values.12

AML classification and risk stratification

Box 1 reviews classification schemes for AML.131415161719Box 2 provides details about AML disease monitoring and “day 14” mid-induction bone marrow assessment,2021222324252627 and box 3 describes the nuances of monitoring for minimal/measurable residual disease (MRD).28293031323334353637383940

Box 1

Classification schemes for acute myeloid leukemia (AML)

French-American-British (FAB) classification

  • The FAB classification scheme was first proposed in 1976.13 Subtypes of AML were defined on the basis of morphologic features and cytochemical methods, categorized as M0 through M7

    • M0-M5: include leukemia involving myeloid blasts with varying degree of maturation

    • M6: acute erythroid leukemia

    • M7: acute megakaryocytic leukemia

    • M3: represents the distinct subtype of acute promyelocytic leukemia (APL)

  • Using FAB criteria, the most common subtypes were M2 (25%) and M4 (20%) disease

  • Subsequent modifications to the FAB classification scheme were made to incorporate emerging genetic findings14

  • However, more recent classification schemes, based on molecular and prognostic scoring systems, have largely replaced use of the FAB scheme

World Health Organization classification

  • WHO introduced a new classification scheme for myeloid neoplasms in 2001 and revised it in 2008.1516 The WHO categorization reflected increasing emphasis on morphologic, genetic, and clinical factors

    • Separation of de novo AML, from secondary AML evolving from myelodysplastic syndrome, was an important distinction, as well as subtypes of therapy related disease

    • The initial classification required 30% myeloblasts for diagnosis of AML, but the revised WHO guidelines lowered the threshold for AML diagnosis to 20% blasts (notable exceptions being APL and core binding factor AML)

  • The most recent 2016 WHO revision updated the previously defined disease categories, incorporating new diagnostic markers and clinical impact of specific gene mutations.7 Many of these revisions involved prognostication derived from gene expression analyses and next generation sequencing studies

    • Categories include AML with recurrent genetic abnormalities, AML with myelodysplasia related changes, therapy related myeloid neoplasms, myeloid sarcoma, myeloid proliferations related to Down syndrome, and AML not otherwise specified

    • Additionally, a new category, the myeloid neoplasms with germline predisposition, was added to incorporate the subgroup of familial AML associated with germline mutations (eg, germline RUNX1, GATA2, CEBPA)

European Leukemia-Net (ELN) classification

  • The ELN categories were developed to correlate genetic abnormalities with clinical variables and prognostic impact17

  • Most recently updated in 2017,18 these guidelines provide classifications to assess disease risk or resistance. The categories are divided into favorable, intermediate, and adverse subtypes based on presence of cytogenetic abnormalities and mutations

  • The favorable category includes:

    • Core binding factor acute myeloid leukemia (AML), defined by the cytogenetic abnormalities t(8;21) and inv(16)

    • This subtype also reflects the favorable prognosis of AML with NPM1 or biallelic CEBPA mutations, regardless of coexisting gene mutations, as well as the similar favorable impact of NPM1 with low allelic ratio (<0.5) FLT3-ITD versus NPM1 without FLT3-ITD

  • The intermediate classification includes:

    • Those cytogenetic abnormalities not classified as favorable or adverse

    • Mutated NPM1 with high allelic ratio FLT3-ITD

    • Low allelic ratio FT3-ITD without mutated NPM1

    • t(9;11) abnormalities

  • The adverse risk subtype includes:

    • The high risk mutation TP53 and mutations in RUNX1 and ASXL1 (not coexisting with favorable risk subtypes)

    • Monosomal or complex karyotypes

    • Specific cytogenetic markers of high risk disease including t(6;9), t(9;22), inv(3), KMT2A rearrangement with t(v;11q23.3), and monosomy 5 or del(5q)

    • FLT3-ITD with high allelic ratio (>0.5) carries a poor prognosis and is included in the adverse risk group

National Comprehensive Cancer Network (NCCN) guidelines

  • The NCCN guidelines provide AML disease risk classifications of better risk, intermediate risk, and poor risk, based on genetic risk and aligned with evidence based treatment recommendations19

  • The NCCN subtypes are similar to ELN’s, but some differences exist in the classification schemes, including:

    • The distinction between wild type NPM1 with low (ELN intermediate) and high (ELN adverse) allelic ratio FLT3-ITD

    • The NCCN risk stratification does not delineate between allelic ratio with any FLT3-ITD within the poor risk subtype

Box 2

Disease monitoring and mid-induction bone marrow assessment (nadir marrow)

  • Disease response has historically been monitored at predetermined time points during treatment, although the timing of disease re-evaluation varies considerably between centers, clinical trials, treatment regimens, and providers, and re-evaluation is sometimes foregone completely, such as in regimens FLAG-Ida and CLAG-M

    • In patients undergoing intensive cytotoxic chemotherapy, a “nadir” bone marrow biopsy has often been performed around day 14 to determine whether residual leukemia is present that would warrant a “re-induction” course of either the same or a different chemotherapy regimen

  • Considerations for completion of mid-induction bone marrow and re-induction are:

    • Fitness of the patient at the time of the nadir marrow and ability to tolerate additional therapy

    • Likelihood of obtaining blast clearance based on cytogenetic and molecular risk stratification

  • The day 14-21 marrow biopsy remains a practice recommended by the NCCN, with determination of the need for additional therapy based on a combination of hypocellularity and blast clearance or as recommended by a clinical trial protocol

    • Clearance of blasts (<5% blasts in the marrow) at this early time point is prognostically favorable2021

    • For patients with a partial response (blasts 5-19%), whether additional chemotherapy at the day 14 time point is able to improve outcomes is unclear222324

    • In large cohorts of patients with acute myeloid leukemia on clinical trials, approximately 50% of patients with residual disease at day 14 undergo re-induction, suggesting that the bone marrow biopsy may be used as much for prognostication as for treatment decisions at day 14 and that strict adherence to guidelines risks overtreatment in this population25262728

Box 3

Monitoring for minimal/measurable residual disease (MRD)


No uniform approach exists for MRD detection in acute myeloid leukemia (AML)

  • Current methods rely on multi-parameter flow cytometry or molecular platforms (quantitative polymerase chain reaction (qPCR), next generation sequencing), and are estimated to cover approximately 60% of AML in patients <60 years of age28

Post-induction MRD

The 2018 consensus document from the European Leukemia-Net (ELN) MRD Working Party on minimal/measurable residual disease in AML provided recommendations for MRD assessment by flow cytometry10

  • A retrospective analysis of 245 adult patients with AML quantified MRD status by 10 color flow cytometry in marrow samples taken at time of bone marrow recovery after induction therapy and correlated with remission responses29

  • Frequency of MRD was lowest in patients who achieved complete remission (19%), whereas detection of MRD was significantly higher in patients with complete remission with incomplete platelet recovery (54.2%) and with complete remission with incomplete blood count recovery) (60.9%) (P<0.01)

Mutation specific monitoring

For patients harboring mutations targetable by current standardized qPCR assays, the PCR approach has been shown to have high sensitivity.10 Various mutations have also been shown to have differential impact, as described below:

  • PCR based MRD testing was shown to have an independent prognostic effect, using samples from patients with NPM1 mutations receiving treatment in the UK National Cancer Research Institute’s AML17 trial.30 Residual presence of NPM1 mutated transcripts after the second chemotherapy cycle was associated with increased risk of disease relapse as well as decreased overall survival. Most patients who relapsed had detectable NPM1 mutations at time of relapse

  • In a series of 430 patients with at least one mutation at diagnosis, MRD assessment in post-treatment marrows at time of morphologic complete remission detected persistent mutations in 51.4%.31 Mutations in DNMT3A, TET2, and ASXL1 were the most common (78.7%, 54.2%, and 51.6%, respectively) but did not correlate with an increased risk of relapse. By contrast, molecular MRD in mutations other than the above (eg, NRAS, PTPN11, KIT, KRAS) was associated with significantly greater risk of disease relapse (55.4% v 31.9%) and lower overall survival (41.9% v 66.1%) compared with no MRD

  • Whether specific residual mutations (eg, DNMT3A, TET2, and ASXL1) represent persistence of the antecedent clone of pre-malignant cells,32 reflective of clonal hematopoiesis,33 is unclear, as is whether they might contribute to relapse

Pre-transplant MRD

Studies have also shown a negative prognostic impact of MRD positivity before hematopoietic stem cell transplant (HSCT) on post-transplant survival and risk of relapse343536

  • In a retrospective analysis of 359 adult patients with AML undergoing myeloablative HSCT, three year overall survival and progression-free survival estimates were similar for patients with pre-transplant MRD positivity (26% and 12%) and those with pre-transplant active AML disease (23% and 13%) and inferior to those of patients entering transplant with MRD negative status (73% and 67%)

  • Relapse estimates were similar between patients with MRD positive and active disease (67% and 65%, respectively), compared with patients in MRD negative remission (22%)37

Monitoring for MRD in the era of personalized medicine

Questions remain about the optimal method or combination of methods for MRD assessment, sample source (peripheral blood, bone marrow), most impactful time point(s) for analysis, differential impact of MRD by various mutations and assays, and how best to integrate MRD assessment and status into treatment algorithms

  • All genetic testing for MRD assessment are not the same, and nor are all assays capable of detecting MRD38

  • Further standardization and validation of assays across centers or in centralized laboratories will be needed, as well as continued incorporation into response assessments in clinical trials

  • How best to target post-therapy MRD with mutation directed therapy is also an area of active ongoing investigation, with novel strategies recently reviewed3940


“Risk factors” for AML

Familial pre-disposition syndromes

That families exist in which multiple members (in both childhood and adulthood) are affected by a myeloid malignancy has been long recognized. In recent decades, specific genotypes associated with heritable hematologic malignancies have been defined. In the most recent WHO categorization,7 provisional recognition of this phenomenon includes multiple categories for myeloid neoplasms with germline predispositions including those that are isolated abnormalities (CEBPA, DDX41) or associated with additional abnormalities (either a platelet disorder or organ dysfunction), including RUNX1, ANKRD26, ETV6, GATA2, Li Fraumeni syndrome (TP53), and bone marrow failure syndromes such as Fanconi anemia and dyskeratosis congenita. Awareness and recognition of these conditions are increasing, which allows for both tailored treatment and genetic counseling.

Some affected families are identified by astute clinicians who note multiple affected family members or phenotypic abnormalities in the patient, whereas others are identified because of increasingly large NGS panels including myeloid predisposition genes. In the latter case, the presence of a suggestive family history may or may not be known. In a study of 360 patients with hematologic malignancies, NGS was completed using a panel including nine genes implicated in hereditary hematologic malignancies, resulting in the identification of 74 (21%) pathogenic variants of which six were ultimately confirmed to be germline.41 The increasing use of NGS that includes germline predisposition mutations allows for identification of at risk individuals. Confirmation of germline mutations should be completed on genetic material obtained by skin biopsy or buccal swab. A recent detailed review by the University of Chicago Hematopoietic Malignancies Cancer Risk Team describes clinical and laboratory features suggestive of a hereditary myeloid malignancy syndrome.42

Identifying these conditions to avoid treating the wrong disease (for example, presuming refractory immune thrombocytopenic purpura in a patient with an underlying platelet disorder predisposing to AML) is critically important, to avoid using affected family members with no symptoms as donors for HSCT and to recognize associated conditions for which the patient may need screening or modification of leukemia treatment (for example, patients with Fanconi anemia in whom cytotoxic chemotherapy should be minimized). Expert consensus recommendations are available for management of family members who have been identified as having the causal mutation but who have not developed the malignancy, which include consultation with a disease expert, genetic counseling, and screening for associated malignancies/conditions in addition to blood work and a baseline bone marrow biopsy to ensure that no hematologic malignancy has developed and as a baseline for future studies.43

Acquired risk factors

Environmental exposures

The adverse effect of environmental exposures on development of leukemia has been well described among the survivors of the 1945 Hiroshima and Nagasaki atomic bombs. Survivors were noted to have an increased risk for the development of leukemia (and other malignancies) following radiation exposure, with increasing risk dependent on age at exposure and dose received. For radiation associated AML, the risk persisted throughout the study follow-up period, out to 55 years after the bomb detonations.44 Additionally, hematologic malignancies including AML and myelodysplastic syndrome associated with the Chernobyl nuclear power plant accident have been described, particularly in the clean-up workers with high exposures to ionizing radiation.45

Secondary AML

Secondary AML refers to myeloid leukemia that develops after an antecedent hematologic disorder or newly diagnosed AML presenting with myelodysplasia related changes and represents approximately 20% of all AML.46 The WHO classification of AML with myelodysplasia related changes (AML-MRC) requires the presence of at least 50% dysplastic cells in at least two cell lineages, excluding cases with NPM1 or biallelic CEBPA mutations.7 Myelodysplastic syndrome and secondary AML often have overlapping cytogenetic and molecular abnormalities but are formally distinguished by the percentage of myeloblasts (<20% equals myelodysplastic syndrome, ≥20% blasts equals AML). Approximately 30% of patients with myelodysplastic syndrome develop secondary AML in the course of their disease. AML-MRC is associated with poor prognosis and lower complete remission rates after conventional induction chemotherapy.4748495051

Risk factors for the transformation from myelodysplastic syndrome to secondary AML include increased circulating peripheral blood blasts, complex cytogenetics, high risk molecular abnormalities, and certain epigenetic and DNA methylation changes.52 As opposed to de novo AML, secondary AML is characterized by multilineage dysplasia, complex karyotypes, and aberrant expression of multidrug resistance proteins and antiapoptotic proteins, as well as subclonal heterogeneity.53 The overall prognosis is poor, with low response rates to cytotoxic chemotherapy and a reported overall survival of six to 12 months.54

Disease progression to secondary AML can evolve stepwise, with the accumulation of somatic mutations and subclonal disease progression occurring over time (box 4).5556 The pathogenesis of secondary AML is complex and dynamically evolving, with the acquisition of mutations in dominant and subclonal populations that respond differentially to chemotherapy. This may affect responses to current therapies and contribute to disease relapse, as well as provide the potential for targeted therapeutic intervention for secondary AML and at even earlier stages to delay or even prevent disease progression.

Box 4

Evidence for stepwise progression to secondary acute myeloid leukemia (AML)

  • Whole genome sequencing studies of paired bone marrow samples at time of diagnosis of myelodysplastic syndrome, as well as time of secondary AML, found persistence of the antecedent founding clone and progression of subclone(s) at disease evolution to secondary AML55

    • Additionally, the number of secondary AML specific mutations was greater in patients with slower disease progression (defined as ≥20 months), compared with patients with more rapid development of secondary AML (defined as <6 months)

  • Mutational analysis comparing samples from patients with de novo AML, secondary AML, and therapy related AML by next generation sequencing was recently reported56

    • In secondary AML, there was >95% specificity for mutations occurring in eight genes (SRSF2, SF3B1, U2AF1, ZRSR2, ASXL1, EZH2, BCOR, STAG2), with an under-representation of mutations in NPM1 and rearrangements in MLL/11q23 and CBF

    • Persistence of disease mutations was found in the post-induction therapy marrow samples of 69% of patients achieving morphologic complete remission, including a subset with persistence of the founder clone, suggesting a differential sensitivity to chemotherapy and showing conceptually that pre-leukemic clones likely existed before the onset of overt leukemia as well


Therapy related AML

Therapy related AML develops in patients with previous chemotherapy (particularly alkylating agents and topoisomerase II inhibitors), radiation exposure, or both; it is associated with an extremely poor prognosis, with historically reported median overall survival of six to eight months and up to 14.6 months in a recent series from the Italian Network on secondary leukemias.57 The only curative treatment option is HSCT. Box 5 provides details of the incidence, classification, and risk of therapy related AML.585962636465 The pathogenesis of and possible genetic susceptibility for the development of therapy related AML is complex and involves the acquisition of somatic mutations driven by chemotherapy/radiation induced DNA damage in the hematopoietic stem cells, which can be influenced by inherited risk factors such as alterations in telomere shortening and single nucleotide polymorphisms in genes linked to drug metabolism and DNA repair (fig 2).6263

Box 5

Therapy related acute myeloid leukemia (AML)

Incidence and characteristics of therapy related AML

  • The incidence varies from <1% to 10%, depending on underlying disease and previous therapies, and represents <10% of all AML cases46

  • Characteristic features of therapy related AML include58:

    • Complex cytogenetics

    • Abnormalities of chromosomes 5 and/or 7 (usually partial or complete loss)

    • Translocations involving the MLL gene and chromosomes 11q23 or 21q22 (particularly after topoisomerase II inhibitors)

    • Chemotherapy refractory disease

    • Mutations in TP53 that are associated with adverse cytogenetics and particularly poor outcomes

    • Less frequently, therapy related AML can present with cytogenetic markers of traditionally “good risk” disease—t(8;21), inv(16), or t(15;17)

Classification, onset, and risk of therapy related AML

  • The WHO classification of myeloid neoplasms and acute leukemia categorizes therapy related myeloid neoplasms as a distinct entity, subdivided into therapy related myelodysplastic syndrome and therapy related AML7

  • In population based cancer registries, therapy related AML has been well described in patients with Hodgkin’s lymphoma, non-Hodgkin’s lymphoma, and multiple myeloma; after autologous hematopoietic stem cell transplantation; and after treatment for solid tumor malignancies including breast, ovarian, bone, gastrointestinal, and other cancers59

  • Mutational analyses have suggested that therapy related myelodysplastic syndrome and therapy related AML have distinct molecular profiles compared with de novo AML, with differences in the frequency of mutated genes such as TP53, epigenetic regulators, and spliceosome genes60

    • However, more recently, next generation sequencing showed that acquired mutations in therapy related AML and therapy related myelodysplastic syndrome were more similar to de novo AML with the same alterations than to the therapy related AML cohort as a whole56

    • Thus the classification of therapy related AML may not be distinct on the basis of mutational profile alone, but driven by the complexity of karyotypic abnormalities61

  • The timing of leukemia presentation varies according to previous exposures:

    • Therapy related AML after topoisomerase II inhibitors occurs earlier, often within 2-3 years after treatment

    • By contrast, therapy related AML after alkylating agents typically occurs later, more commonly within 5-7 years of therapy, and is often preceded by the development of therapy related myelodysplastic syndrome62

  • The risk of developing therapy related AML varies depending on the alkylating agent, with a dose-response effect based on cumulative exposure63

  • Therapy related AML has also been described after exposure to fludarabine and azathioprine and after solid organ transplantation6465

Fig 2
Fig 2

Pathogenesis of therapy related acute myeloid leukemia (t-AML).606166676869 Experimental evidence supports the notion that chemotherapy and/or radiation selects for the expansion of a mutant clone that is more resistant to DNA damage.6668 Normal hematopoietic cells ordinarily thrive in contact with a complex microenvironment that includes non-hematopoietic (stromal) cells and well differentiated (auxiliary) hematopoietic cells. The microenvironment normally supports stem cell function but, if challenged with genotoxic agents, can inhibit stem cell survival providing an opportunity for the selection of pre-leukemic founder mutations resistant to such agents.69 The stem cells with founder mutations marked with “?” are meant to emphasize that mutations can develop either before or as a direct result of DNA damage inflicted by the genotoxic stress. Specifically, in leukemic cells of patients with t-AML, mutations of TP53 were not only found in the post-chemotherapy leukemic cells but were also detectable at low concentrations years before initial chemotherapy.67 Finally, founder mutations alone are insufficient to “drive” the expansion of fully malignant AML clones but provide the right cellular context for the acquisition of additional mutations that subsequently drive the emergence of fully leukemic cells

Clonal hematopoiesis

Clonal hematopoiesis, defined as the expansion of a clone of blood cells with one or more somatic mutations,66 has been described in older adults and linked to specific acquired gene mutations.707172 Different nomenclature has been used to describe clonal hematopoiesis, including clonal hematopoiesis of indeterminate potential (CHIP), age related clonal hematopoiesis, and idiopathic or clonal cytopenias of undetermined significance. More recently, age related clonal hematopoiesis has been identified as a risk factor for the development of hematologic malignancies,737475 as well as for adverse cardiovascular outcomes such as coronary heart disease and ischemic stroke.76 Whole exome sequencing of peripheral blood DNA from large cohorts of adults without known malignancies was used to detect somatic mutations in genes that are epigenetic regulators of hematopoiesis and known potential drivers for hematologic disease. The frequency of clonal hematopoiesis increased with age, ranging from <1% in adults aged <50 years to >10% in adults >65 years of age.737475

Most mutations were found to occur in genes DNMT3A, TET2, and ASXL1, as well as JAK2 and PPM1D. As opposed to patients with myelodysplastic syndrome or AML, most of whom have mutations in multiple driver genes,77 most adults with clonal hematopoiesis had only one detectable mutation.7374 The presence of clonal hematopoiesis was identified as a risk factor for subsequent development of a hematologic malignancy and for mortality, and also as a risk for therapy related myeloid disease (box 6).737478798081

Box 6

Clonal hematopoiesis and myeloid disease

Clonal hematopoiesis as a risk factor for hematologic malignancy and mortality

  • In one study,73 the hazard ratio was:

    • 12.9 (95% confidence interval 5.8 to 28.7) for development of a hematologic malignancy

    • 1.4 (1.0 to 1.8) for risk of death

  • In another study,74 the hazard ratio was:

    • 11.1 (3.9 to 32.6) for risk of hematologic cancer

    • 1.4 (1.1 for 1.8) for all cause mortality

  • The risk of developing a hematologic malignancy among patients with clonal hematopoiesis was 0.5-1% per year depending on the variant allele fraction (proportion of mutated alleles), compared with <0.1% for those without clonal hematopoiesis, and these mutations persisted over time737478

Therapy related clonal hematopoiesis

  • Clonal hematopoiesis has also been described in patients with non-hematologic malignancies, such as therapy related clonal hematopoiesis

  • Use of the Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT) to sequence paired blood derived versus tumor derived DNA samples showed that79:

    • 25.1% of patients had at least one somatic mutation, with presumptive driver mutations in 4.5% of patients

    • TP53 and PPM1D mutations were associated with previous chemotherapy and radiation

    • Statistically significant associations were found between clonal hematopoiesis and older age, tobacco use, and previous radiation therapy but not chemotherapy

    • Patients with clonal hematopoiesis had an increased risk of developing a hematologic malignancy (18 month estimate of 1% (95% confidence interval 0.5% to 1.8%) versus 0.3% (0.1% to 0.5%) for those without clonal hematopoiesis).

    • Additionally, overall survival was inferior among patients with clonal hematopoiesis with a presumptive driver mutation, with the most common cause of death being disease progression of the non-hematologic malignancy

Clonal hematopoiesis and therapy related myelodysplastic syndrome and AML

  • Clonal hematopoiesis has also been associated with the development of therapy related AML and myelodysplastic syndrome

  • A nested case-control study determined the prevalence of clonal hematopoiesis in patients >70 years of age with chemotherapy treated cancer, who subsequently did or did not develop therapy related myelodysplastic syndrome or therapy related AML80

    • The prevalence of clonal hematopoiesis of indeterminate potential (CHIP) among the entire cohort was 33%, and was significantly higher in patients who developed therapy related myeloid disease (62%) compared with those who did not (27%)

    • Patients with CHIP were at greater risk of developing therapy related myelodysplastic syndrome/AML than those without CHIP (adjusted odds ratio 5.75, 95% confidence interval 1.52 to 25.09; P=0.013)

  • In another case-control study,81 clonal hematopoiesis with driver mutations was detected in 71% of patients who ultimately developed therapy related myeloid malignancies, compared with 31% in age matched controls.


Because a significant fraction of patients with CHIP have clones that do not expand or progress to myeloid malignancy, somatic clonal mutations clearly are not sufficient to function as “drivers” of leukemia. Whether the presence of such mutations alone would be actionable for leukemia prevention trials or whether pre-treatment CHIP status might affect therapeutic decisions for patients with non-hematologic malignancies remains to be determined. Additionally, although the current definition of CHIP specifies a somatic mutation frequency of greater than 2% in a defined set of genes linked with leukemogenesis, the cut-off for variant allele frequency used to define the presence of this somatic mutation will clearly evolve.6682

Treatment paradigms

The care of a patient with AML involves treatment of the primary malignancy and extensive supportive care measures (table 1).90919293949596

Table 1

Considerations for supportive care

View this table:

Cytotoxic chemotherapy

For decades, the standard intensive treatment for AML has been the combination of seven days of cytarabine with three days of an anthracycline (idarubicin, daunorubicin), referred to as the “7+3” regimen.979899100101 Rates of complete remission and cure vary according to age, ranging from 60-85% complete remission and 35-40% cure for adults ≤60 years of age to 40-60% complete remission and 5-15% cure for adults >60 years of age.99 Patients with mutations in genes defined as “secondary type” or TP53 mutations have been shown to have more chemoresistant disease requiring multiple cycles of induction therapy, compared with those with de novo or pan-AML genes.56Box 7 details induction chemotherapy regimens.1983848586102103104105106

Box 7

Induction chemotherapy

Conventional chemotherapy

  • Fludarabine, high dose cytarabine, and idarubicin (FLAG-ida) has been shown to be an effective regimen102

    • Superior rates of remission and reduced risk of relapse compared with a 10 day cytarabine plus daunorubicin regimen were shown, but also with increased toxicity and thus no improvement in overall survival

  • Other intensive cytotoxic regimens incorporating clofarabine have been investigated103104105

  • Cladribine, another nucleoside analog, has also been investigated in combination with daunorubicin and cytarabine (eg, the “DAC regimen”), with response rates of up to 70% in de novo acute myeloid leukemia (AML) and 40-50% in relapsed/refractory AML838486106

    • Induction regimens containing cladribine are included as category 1 recommendations in the National Comprehensive Cancer Network guidelines for AML patients <60 years, as well as in the salvage setting19


  • Recently, the US Food and Drug Administration and European Medicines Agency approved CPX-351 (Vyxeos), a liposomal dual drug encapsulation of cytarabine and daunorubicin at a fixed 5:1 molar ratio, for the treatment of adults with newly diagnosed therapy related AML or AML with myelodysplasia related changes, on the basis of a randomized phase III study in adults aged 60-75 years85

  • Patients were randomized to a 7+3 regimen versus CPX-351

    • Higher response rates were seen with CPX-351 than with conventional 7+3 chemotherapy (complete remission/complete remission with incomplete blood count recovery 47.7% v 33.3%; P=0.016)

    • Median overall survival was improved with CPX-351 versus 7+3 (9.56 v 5.95 months; hazard ratio 0.69; P=0.003)

    • No significant differences were seen in early mortality rates though day 30 and day 60


Consolidation therapy

Essentially all patients who achieve complete remission after induction ultimately relapse without post-induction consolidation therapy. Consolidation options include intensive chemotherapy and stem cell transplantation (discussed below). Details of consolidation chemotherapy are in box 8 (109-111).

Box 8

Consolidation chemotherapy

  • High doses of cytarabine (up to 3000 mg/m2), six doses per cycle, given for up to four cycles, are commonly used (dose reduced in older patients)

  • However, some data suggest that intermediate dose (2000 mg/m2) cytarabine may be an alternative regimen, particularly for patients with intermediate risk AML.19 Additionally, recent studies have failed to show a benefit for high dose cytarabine regimens in favorable risk AML with younger age (<65 years),878889101102 as is also discussed in the European Leukemia-Net (ELN) guideline18

  • Whether higher doses of cytarabine are more effective than intermediate dosing, what the optimal number of cycles of consolidation should be, what the impact of age is on dose selection, and whether adding other agents (including additional anthracyclines) is beneficial are not yet known18101


Targeted therapy/novel agents

The development of lower intensity treatments has increased the therapeutic options for all patients, but in particular may offer additional treatments that are more palatable to providers and to patients who are older or unfit or wish to avoid prolonged hospital admission, given the lack of mucosal toxicity and barrier disruption traditionally associated with cytotoxic chemotherapies. Comprehension of complex treatment algorithms and previously unfathomable medical complications may be difficult for patients grappling with a new life threatening diagnosis and can contribute to discordance between patients’ and providers’ perceptions of the risks and potential benefits of AML directed therapy; this can make “informed” consent difficult.107 The importance and difficulty of shared decision making in patients with AML has been well described, and the challenges are evident in population studies showing the high percentage of patients who historically have received no treatment despite data suggesting that most would benefit from some treatment.12108 The intensity of treatment requiring time spent in clinics and hospitals in addition to up-front risks of treatment related morbidity and mortality has been, historically, a barrier to treatment for AML.109 A recent review of therapies in development for targeting pathways in AML with both novel therapies and combination therapies shows the extraordinary progress and opportunity in novel treatments.110Table 2 lists several emerging and novel therapies for AML.111112

Table 2

Emerging therapies and selected clinical trials for acute myeloid leukemia

View this table:

Hypomethylating agents and low dose chemotherapy

Hypomethylating agents (azacitidine and decitabine) are administered in the clinic as an intravenous or subcutaneous (azacitidine only) injection and are appropriate for all subtypes of AML. Hypomethylating agents enhance and inhibit expression of sets of genes in cells by influencing the methylome. Many possible mechanisms of action have been proposed for the drugs, and no clear cytogenetic or molecular predictors of response are available.125

TP53 mutated AML is known to have particularly poor outcomes and is seen more commonly in older adults, patients with complex karyotypes, and patients with treatment related AML.126127 A phase II study found that a 10 day course of decitabine improved overall survival of patients with TP53 mutated AML enough to resemble that of intermediate risk AML,128 and a five day course was sufficient in patients with newly diagnosed AML.129 The ideal hypomethylating agent regimen has not been established in the setting of TP53 mutated AML. Some data from retrospective studies support withholding traditional cytotoxic chemotherapies if the TP53 mutation status is known in advance of the time at which definitive chemotherapy must be started.130 Although the likelihood of obtaining complete remission is lower when using the hypomethylating agents, the survival benefit of being in complete remission was just as potent in patients who obtained complete remission with azacitidine as those who obtained complete remission with a “7+3” based regimen.131132 Azacitidine has been studied in a phase III randomized trial and showed both hematologic response rates and complete remission rates that provide benefit above conventional care and low dose cytarabine (LDAC).133 The relative benefits in terms of quality of life of complete remission versus complete remission with incomplete blood count recovery versus hematologic improvement remain understudied. LDAC given subcutaneously either once or twice daily for 10 days is also low intensity chemotherapy that does not require admission to hospital for administration and has been used in combination with novel agents.134135

B cell lymphoma-2 inhibition

The B cell lymphoma-2 (BCL-2) family of proteins is involved in multiple regulatory processes of apoptosis. BCL-2 inhibition was identified as a drug target for the purposes of re-enabling activation of apoptosis in malignant cells. Venetoclax is an oral BCL-2 inhibitor first developed in chronic lymphocytic leukemia (CLL), a disease in which BCL-2 is usually significantly overexpressed leading to impaired apoptosis.136 Venetoclax proved effective in heavily pre-treated patients with CLL, occasionally leading to tumor lysis syndrome.137 Although less BCL-2 overexpression is seen in AML than in CLL, pre-clinical data suggested efficacy and the initial phase II trial of single agent venetoclax for treatment of relapsed/refractory AML showed a marginal overall response rate of 19%.138139 Subsequent trials studied combinations of venetoclax with LDAC or hypomethylating agents, with improved results exceeding those of historic controls.140 The trial VIALE-A randomized older adult patients with newly diagnosed AML who were not candidates for induction chemotherapy to either single agent azacitidine or azacitidine plus venetoclax. Results for the combination arm showed higher complete remission and overall response rates (36.7% and 66.4%) compared with single agent azacitidine (17.9% and 28.3%).141

Initial real world data using a hypomethylating agent in combination with venetoclax continue to support favorable response rates, though as expected they are lower than those reported in clinical trials.142 A parallel trial of venetoclax with LDAC showed a complete remission rate of 26% and overall response rate of 54% in previously untreated older adults.134 Additional details on molecular sub-populations of patients with AML who are most likely to benefit from venetoclax are emerging, showing high response rates and durable remissions in patients with NPM1 or IDH2 mutations.143 Although uncommon, tumor lysis syndrome can develop in patients starting therapy with a regimen containing venetoclax, particularly in those with white blood cell counts greater than 25×109/L. Recommended prophylaxis, monitoring, and management of tumor lysis syndrome have been well described in recent publications, in addition to practical recommendations on management of drug interactions with commonly used “azole” antifungals.144145 The role of venetoclax in the post-transplant setting is being studied.

Myeloid cell leukemia-1 (MCL-1) is another member of the anti-apoptotic family of BCL-2 proteins for which inhibitory drugs are being developed for treatment of AML.146 MCL-1 is highly expressed in patients with untreated AML, and several MCL-1 inhibitors are in development, some of which are being used in combination with venetoclax. Additional therapies affecting the MCL-1 apoptotic pathway include cyclin dependent kinase inhibitors, many of which are in varying stages of development and clinical trials.146

FLT3 inhibition

The increasing recognition of genomic heterogeneity in AML has led to the development of therapies that directly target the mutations. Targeted therapies are increasingly available for specific driver mutations, although the number of patients for whom these drugs (FLT3 and isocitrate dehydrogenase inhibitors) are in routine clinical use remains a minority. Multiple novel agents target the FLT3 pathway, both FLT3-ITD mutations within the juxtamembrane domain and FLT3-TKD mutations in the tyrosine kinase domain. FLT3-ITD mutations are associated with a poor prognosis because of the increased frequency of relapse after a first complete remission has been obtained. FLT3-ITD mutations are seen in approximately 25% of patients with newly diagnosed AML and should be tested for at diagnosis.18 Testing should be quantitative and include the variant allele frequency or allelic ratio, as these are prognostic, and additional co-mutations such as NPM1 that further modify risk stratification.147148 Midostaurin and gilteritinib are now approved by the US Food and Drug Administration (FDA) for up-front and relapsed/refractory FLT3-ITD AML respectively.149150 First generation FLT3 inhibitors including sorafenib and midostaurin are yielding to next generation FLT3 inhibitors including crenolanib, quizartinib, and gilteritinib in ongoing clinical trials designed to establish the benefits of early and prolonged FLT3 inhibition for patients with newly diagnosed AML and in relapsed disease. Several recent reviews have summarized the evolving FLT3 landscape and identified important unanswered questions.151152153 Additional tyrosine kinase inhibitors are being studied for targets other than FLT3 and are in various stages of development, as recently described.154

Isocitrate dehydrogenase inhibition

The isocitrate dehydrogenase inhibitors ivosidenib and enasidenib target mutations in the enzymes isocitrate dehydrogenase 1 and 2 respectively that are involved in the Krebs cycle. Mutations in the isocitrate dehydrogenase enzymes cause an accumulation of oncometabolites that lead myeloid blasts to linger in an immature state.8 The addition of the isocitrate dehydrogenase inhibitors blocks this process and allows for resumption of more normal differentiation, sometimes leading to the complication of “differentiation syndrome,” in which the rapid maturation of many immature white blood cells leads to dyspnea, fever, pulmonary infiltrates, and hypoxia.155156 Isocitrate dehydrogenase 1 mutations are present in approximately 7-14% of patients with AML and isocitrate dehydrogenase 2 mutations in 8-19%.157 Both drugs were initially tested as single agents in the relapsed/refractory setting to demonstrate efficacy sufficient to warrant FDA approval. Ivosidenib is also approved for use as initial therapy and is now in phase III combination studies with azacitidine for newly diagnosed disease. The availability of novel agents with potential to improve outcomes emphasizes the importance of initial diagnostic testing to identify patient specific mutations that affect not only prognosis but also choice of therapy in both the up-front and relapsed/refractory setting.

Hedgehog pathway inhibition

Glasdegib is an additional new agent in the treatment of AML that inhibits the hedgehog pathway and smoothened protein. A 2:1 randomized trial in 132 patients who were not candidates for intensive chemotherapy compared glasdegib plus LDAC versus LDAC alone. It found that the combination achieved a 17% remission rate compared with 2.3% in the single agent LDAC arm (P< 0.05).135 Ongoing trials are examining its role in the up-front setting in combination with induction chemotherapy.

TP53 pathway modulation

Mutations in the TP53 tumor suppressor gene occur in about 10% of new AML diagnoses and are more common in patients with secondary AML.158 Survival is markedly worse for patients with TP53 mutations and is commonly seen in patients with complex karyotypes.159 The presence of a TP53 mutation is inherently problematic for patients being treated with cytotoxic chemotherapy, as mutated p53 causes dysregulation of the apoptotic pathway which decreases the efficacy of DNA damaging chemotherapy agents.130 New agents targeting the TP53 pathway are in development, including APR-246, which restores wild type p53 function and has been advanced to a phase III clinical trial after the phase Ib/II study of patients with myelodysplastic syndrome and oligoblastic AML showed promising results.160

Cellular therapies and immunotherapy for AML

Allogeneic hematopoietic stem cell transplant

When the goal of treatment for AML is cure, HSCT is often a consideration for post-remission therapy in patients with intermediate or adverse risk disease for whom relapse is most likely. The efficacy of HSCT relies variably on the anti-leukemic effect of the “conditioning” (pre-transplant chemotherapy and radiation) and the graft versus leukemia effect mediated primarily through donor T cells. Candidates for HSCT may be identified shortly after diagnosis of AML on the basis of risk stratification and the patient’s characteristics or later when leukemia persists in an MRD positive state or with overt recurrence or refractory disease.10

Eligibility to proceed with a transplant is based on a favorable response to induction chemotherapy, patient’s fitness, and availability of a stem cell donor. HSCT has higher up-front treatment related mortality than chemotherapy based treatment, but for selected patient populations it can provide improved long term survival. Decision making about the risks and benefits of transplant are complicated, although clear indications of disease states that most strongly require transplant for potential cure exist.161 With increasing “alternative” donor options (donors who are less well HLA matched or from conventional HLA matched donors), transplants are becoming more available to patients. Important strides have been made in recent years in identifying options for patients who previously would not have had an identified stem cell donor, with the ready availability of umbilical cord blood units and successful transplantation of haplo-identical stem cells.162 With improvements in prophylaxis for graft versus host disease and the use of post-transplant cyclophosphamide, further improvement in outcomes with mismatched transplants may be possible in the future.163

Relapse of AML post-transplant remains the major obstacle for patients who undergo HSCT and is mediated by both persistence of disease and immune evasion. Recent studies have elegantly described mechanisms of relapse via down-regulation of HLA class II, effectively allowing the leukemia to “hide” from the donor immune system.164165 T cell exhaustion (both CD4+ and CD8+) is an additional mechanism of relapse, for which donor lymphocyte infusions have shown efficacy.166 Ongoing studies are examining ways in which the post-transplant immune system and leukemia cells can be modulated to re-engage an effective immune response.

Maintenance therapy after transplant is also an area of active study. The randomized trial of post-transplant gilteritinib maintenance (US CTN 1506) for patients with FLT3 ITD positive disease recently met accrual, and results are awaited. Studies of post-transplant hypomethylating agents and targeted therapies including venetoclax and immunotherapies are all ongoing with the goal of mitigating the risk of relapse. Some studies have shown that post-transplant azacitidine was safe and may prolong event-free and overall survival,167 but others have not observed such an effect.168 Additional studies are needed to identify which patients are candidates for which therapies and on what schedules in the post-transplant setting.

Chimeric antigen receptor T cell (CAR-T) therapy

The field of cellular therapy has been revolutionized by the development of CAR-T therapies, in which autologous or allogeneic T cells are isolated and genetically modified to express chimeric antigen receptors targeting specific cell surface antigens. The re-programmed T cells possess effector T cell function, combined with the ability of antibodies to recognize targeted antigens, in a major histocompatibility complex (MHC) independent manner.169170 One method of immune evasion used by tumor cells is the loss of MHC class I expression, leading to avoidance by the tumor of T cell surveillance and recognition171; thus the ability of CAR-T cells to function in an MHC independent manner is important in efforts to improve the anti-tumor effect.170 The first approvals by the FDA were for the anti-CD19 CAR-T product tisagenlecleucel (Kymriah) for relapsed or refractory pediatric B lineage acute lymphoblastic leukemia and for relapsed or refractory diffuse large B cell lymphoma (DLBCL) in adults and axicabtagene ciloleucel (Yescarta) for the treatment of relapsed or refractory DLBCL and for relapsed or refractory primary mediastinal B cell lymphoma.121172

However, despite this success in B lineage acute lymphoblastic leukemia and DLBCL, the development of CAR-T therapy for AML has been more challenging, partly because of the lack of a suitable target antigen. Given that AML target antigens such as CD33 and CD123 are often expressed not just on leukemia blasts but also on normal hematopoietic stem and progenitor cells,173174175 concern exists that eradication of all such antigen expressing cells would lead to ablation of the myeloid cell pool and marrow failure.176 Efforts to circumvent this hematologic toxicity include the identification of novel target antigens such as CLL-1, Lewis Y, FLT3, and CD44v6; compound CAR-T targeting both CD33/CD123 or CD123/CLL1, which could bypass the problem of loss of target antigen seen on leukemia cells at the time of relapse; and the incorporation of HSCT as a stem cell rescue or in combination with CAR-T and a genetically modified allograft.122123172177178179180181

Nineteen CAR-T studies are actively recruiting for AML, as well as six additional trials that are either not yet recruiting or of unknown status, as recently reviewed.172122 The FDA has granted orphan drug designation to the CD123 targeted CAR-T product MB-102 (, NCT02159495),182 as well as to the PRGN-3006 UltraCAR-T (NCT03927261),183 for patients with AML.

Natural killer cell therapy

Natural killer cells are effector lymphocytes that play a role in the innate immune response and possess direct cytotoxic and anti-tumor effects without the requirement for antigen specificity.123 Natural killer cell therapies can be derived from both autologous and allogeneic sources, including haplo-identical and in combination with HSCT,184 and can be used in the allogeneic setting without evidence of graft versus host disease.185 Efforts are under way to optimize natural killer cell sources (for example, haplo-identical peripheral blood, umbilical cord blood, pluripotent stem cells, and CAR expressing natural killer cells)186187188 and in vivo natural killer cell expansion and persistence, including lymphodepletion chemotherapy before natural killer cell infusion and co-administration of cytokines such as interleukin 2 and interleukin 15.185189190

A recently published first in human study assessed lymphodepleting chemotherapy followed by haplo-identical natural killer cell infusion with recombinant human interleukin 15, to potentiate in vivo natural killer cell activation.189 Recombinant human interleukin 15 was administered either intravenously or subcutaneously, in escalating doses. Assessment of response by day 42 showed a complete remission/complete remission with incomplete blood count recovery rate of 35% in both cohorts (32% for intravenous recombinant human interleukin 15, 40% for subcutaneous recombinant human interleukin 15), with median durations of response of 107 days (intravenous) and 278 days (subcutaneous), and one year overall survival of 19% (intravenous) and 21% (subcutaneous).

More than a dozen actively recruiting trials of natural killer cell and CAR-natural killer cell therapies for AML are registered on Active studies include natural killer therapies including haplo-identical and third party off-the-shelf products, as consolidation therapy or for relapsed/refractory disease, in tandem with haplo-identical or cord blood transplantation, as well as in combination with other agents (for example, recombinant human interleukin 2, immune checkpoint inhibitors).191 Strategies to enhance natural killer cell activity, as well as natural killer cell based cellular therapies under commercial development, have been recently reviewed.124

Antibody based therapy

High expression of CD33 on AML blasts has also been exploited for antibody based therapy.173 Gemtuzumab ozogamicin (Mylotarg), an immunoconjugate combining anti-CD33 antibody with the toxin calicheamicin, was initially granted accelerated approval on the basis of a phase II study with a reported 30% response rate192193; subsequently, gemtuzumab ozogamicin was discontinued given safety concerns and disappointing efficacy results.113114115 Recent studies showed high efficacy in APL and AML and efficacy when given in fractionated doses in combination with chemotherapy.116117118192 The US FDA, followed by the European Medicines Agency, reapproved the use of gemtuzumab ozogamicin combined with daunorubicin and cytarabine for patients with newly diagnosed CD33 positive AML in 2017.117119120194195 The FDA has also approved gemtuzumab ozogamicin for use as a single agent in relapsed or refractory AML and in children.

Additional antibody based therapies include unconjugated monoclonal antibodies that can enhance antibody dependent, cell mediated cytotoxicity mediated by natural killer cells; multivalent antibodies (including bi-specific and tri-specific) designed to link tumor associated antigens with T cell or natural killer effector cells; and radioimmunoconjugates with a radionuclide attached to a monoclonal antibody, as recently summarized.196197


European, American, and WHO guidelines for the diagnosis and management of AML are referenced throughout this review. The European Leukemia Net (ELN), European Society for Medical Oncology (ESMO), and National Comprehensive Cancer Network (NCCN) all provide detailed guidelines for the clinical management of patients with AML, starting with initial investigation, through diagnosis and management, to end of life care, and are important resources for the practicing clinician.1819198 Because the guidelines cover a vast number of patient specific situations, they do not permit direct comparison. The ELN risk stratification algorithm, which may need updating in the era of new therapies that may differentially affect prognosis, is still used by both ESMO and NCCN in their more recent iterations, with differences as noted earlier in the review.199 The NCCN guidelines are developed by members from US Cancer Centers that are designated “comprehensive cancer centers” on the basis of recent FDA approvals, presentations at scientific meetings and congresses, third party submissions, PubMed literature review, clinical data review, and institutional review. The recommendations are rated according to categories of evidence and consensus. Concerns have been raised about recommendations for newer oncologic therapies in the absence of robust randomized clinical trial data.200 More recently, several focused guidelines have been completed, identifying a specific sub-group of patients for whom recommendations might differ.201202 The rapidity of changing knowledge about AML is likely to require incremental updates of guidelines with specific diagnostic, risk stratification, and management recommendations rather than complete revisions in the future.

Emerging treatments

Emerging treatments are incorporated throughout the text in the relevant sections, as well as in table 2.


AML remains a complicated and difficult to treat disease, but major developments in our understanding of the molecular pathogenesis have led to improvements in diagnosis, monitoring, and treatment. Targeted therapies, immunotherapies, and novel agents are increasing the number of therapeutic choices for patients and providers. Studies on the impact of factors such as cost, quality of life, time spent in treatment, and patients’ preferences will now be more relevant given the increasing number of decisions that must be made. Pre-leukemic conditions and genetic predisposition states are also increasingly recognized at a genomic level and raise the possibility of ultimately considering studies of leukemia prevention in the future. Providers should be prepared to refer patients and family members for appropriate testing and genetic counseling as needed. Ongoing cooperative efforts in basic science, translational studies, and clinical trials will continue to advance improvements for patients with AML.

Glossary of abbreviations

  • AML—acute myeloid leukemia

  • APL—acute promyelocytic leukemia

  • BCL-2—B cell lymphoma-2

  • CAR-T—chimeric antigen receptor T cell

  • CHIP—clonal hematopoiesis of indeterminate potential

  • CLL—chronic lymphocytic leukemia

  • DLBCL—diffuse large B cell lymphoma

  • ELN—European Leukemia-Net

  • FAB—French-American-British

  • HSCT—hematopoietic stem cell transplant

  • LDAC—low dose cytarabine

  • MCL-1—myeloid cell leukemia-1

  • MHC—major histocompatibility complex

  • MRC—myelodysplasia related changes

  • MRD—minimal/measurable residual disease

  • NCCN—National Comprehensive Cancer Network

  • NGS—next generation sequencing

  • qPCR—quantitative polymerase chain reaction

Research questions

  • In what clinical situations should treatment of acute myeloid leukemia (AML) be delayed to obtain additional prognostic information?

  • How has risk stratification changed on the basis of availability of novel therapies?

  • What is the most practical, replicable tool to measure residual AML after treatment?

  • Can patients at the highest risk for therapy related AML be predicted before deciding therapy for their first malignancy, so that additional consideration could be given to choice of treatment?

    • Would the risks and benefits of chemotherapy and/or radiation be reconsidered in light of an increased risk of therapy related AML?

  • Do targetable mechanisms exist that might be used as therapies to prevent drivers of leukemogenesis in patients at risk (eg, post-chemotherapy/radiation, clonal hematopoiesis of indeterminate potential)?

  • How should quality of life and patient reported outcomes influence treatment choices?


We gratefully acknowledge Grover Bagby for his review of the manuscript, tables, and figures.


  • Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors

  • Contributors: LFN and RJC developed the outline for this manuscript, did the literature review, and wrote all sections of the manuscript.

  • Funding: LFN is supported by the National Institute of Child Health and Human Development (1K23 HD091369-01).

  • Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.

  • Provenance and peer review: Commissioned; externally peer reviewed.

  • Patient involvement: No patients were involved in the creation of this article.


View Abstract