Keywords

FormalPara Key Points

  • While acute myeloid leukemia is identifiable by morphologic assessment alone, characterization of the underlying genetic abnormalities is needed for definitive subclassification in most cases.

  • Current standard of care includes evaluating for selected gene sequence abnormalities (e.g., FLT3, NPM1, CEBPA, KIT, and others), in addition to traditional chromosome analysis and FISH studies.

  • Karyotype still represents the single most important prognostic factor in predicting remission rates, relapse risks, and overall survival outcomes in acute myeloid leukemia.

  • 40–50% of patients with de novo acute myeloid leukemia have a normal karyotype, and molecular profiling is quickly helping to better stratify this cohort with heterogeneous outcomes.

FormalPara Key Online Resources

Introduction

Acute myeloid leukemia (AML) represents a heterogeneous group of disease, but all subtypes are characterized by clonal proliferations of immature myeloid hematopoietic precursor cells. The first widely accepted subclassification system for AML, developed by the French-American-British (FAB) working group in 1972, was based solely on morphologic findings. This classification system was unfortunately found to lack clinical utility as the proposed disease subtypes were largely unable to provide meaningful prognostic stratification. While the terminology from the FAB classification system still persists in present-day medical vernacular, this system is considered obsolete.

Janet Rowley described the t(8;21)(q22;q22.1) translocation in 1973; this was the first recurrent genetic aberrancy reported in association with AML. As the cytogenetic profile for this disease was slowly elucidated over subsequent decades, a diagnostic paradigm shift occurred in the 2001 third edition of the World Health Organization (WHO) classification with inclusion of genetic abnormalities into the diagnostic algorithms for AML diagnosis. The importance of underlying cytogenetic aberrancies was recognized, as was secondary-AML arising from lower-grade myeloid neoplasms, prior cytotoxic therapies for unrelated malignancy, or disease arising in a background of multilineage dysplasia. These categories were expanded and refined further in 2008 and 2016. The 2016 WHO classification system is the most current AML classification system, and the use of prior less-specific terminology is discouraged [1].

The ability to risk-stratify cases of primary-AML was somewhat limited in first iteration of the WHO classification system. Recognized genetic defects were limited to chromosomal translocations at the time, and conventional cytogenetic testing modalities fail to detect aberrancies in a significant subset of cases (a disease subgroup often referred to as “normal karyotype AML”). The 2008 WHO revision broadened the scope of genetics in AML diagnosis, accepting that multiple types of genetic lesions could cooperate to create a leukemic process. More recent molecular sequencing studies have further characterized the genetic landscape of AML and have helped to close the knowledge gap. It is now understood that numerous cooperating mutations occur in AML [2]. While molecular profiling analysis is initially focused on normal karyotype AML, somatic sequence mutations appear to demonstrate prognostic importance across other genetic AML subtypes, and sequencing analysis appears to be indicated in all cases of AML [2, 3]. At present, most cases of primary/de novo AML can be genetically categorized, and several specific subtypes of AML can be diagnosed on the basis of underlying genetics without regard to blast cell count. The subgroup of AML, not otherwise specified, which has no distinct clinical, immunophenotypic, or genetic features is expected to continue to shrink as knowledge of AML pathogenesis accumulates (Fig. 14.1).

Fig. 14.1
figure 1

General breakdown of AML subtypes by 2016 WHO classification

Full size image

A Standard Genetic Workup

The specimen for evaluation (peripheral blood or bone marrow) should be obtained before initiation of any definitive therapy. At present, a standard workup for newly diagnosed AML should include:

  • Complete karyotype and/or FISH analysis for subtype defining aberrancies

  • NPM1, CEBPA, RUNX1, and FLT3 somatic sequence mutation analysis

  • IDH1/2 mutation analysis for potential targeted therapy in relapsed/refractory disease

  • KIT mutation analysis in all cases with t(8;21)(q22;q22.1) RUNX1-RUNX1T1 and AML with inv(16)(p13.1q22) or t(16;16)(p13.1;q22) CBFB-MYH11

Although detection of recurrent cytogenetic aberrancies generally provides the most significant prognostic information at diagnosis, nearly half of all adult AML cases will show no detectable abnormalities by karyotype. Molecular genetic analysis is quickly filling the knowledge gap. Many other gene mutations are also known to have prognostic significance or relevance for clinical trials in AML and may be readily evaluable by targeted next-generation sequencing gene mutation panels (a selection of these are found in Table 14.1).

Table 14.1 Genes with recurrent somatic sequence mutations in AML
Full size table

AML with Recurrent Genetic Aberrancies

This category of acute myeloid leukemia includes entities that are defined by both by balanced chromosomal rearrangements and by specific gene sequence mutations . The 2016 WHO classification system recognizes eight subtypes defining balanced chromosomal gene fusions and three subtypes related to specific somatic gene sequence mutations, each with distinctive clinicopathologic features and prognostic associations. Many other balanced gene rearrangements are known to recur in AML [4], but these are very rare and are not currently recognized to represent distinct diagnostic entities.

The diagnosis of AML typically requires demonstrating a myeloblast population that represents at least 20% of the peripheral blood or bone marrow cellularity. However, the WHO permits assigning an AML diagnosis without regard to blast count for three entities, based on the strength of associated underlying cytogenetic aberrancies. These entities are the two core binding factor AMLs associated with t(8;21) and inv(16)/t(16;16) and acute promyelocytic leukemia with PML-RARA fusion. The minimum threshold of 20% myeloblasts is still required for an AML diagnosis with the remaining recurrent genetic aberrancies.

Core Binding Factor AML (Tables 14.2 and 14.3)

AML with t(8;21)(q22;q22.1) results in the fusion of RUNX1 (also known as core binding factor-α) and RUNX1T1, often presenting with large myeloblasts that have abundant basophilic cytoplasm, azurophilic granules, few large pseudo-Chédiak-Higashi granules, and perinuclear hoffs. AML with inv(16)(p13q22) or t(16;16)(p13;q22) CBFB-MYH11 disrupts the beta subunit of core binding factor, often presenting with myelomonocytic blasts and abnormal background eosinophils, usually with large basophilic colored granules. These translocations disrupt the function of core binding factor, a crucial heterodimeric transcription factor that helps control stem cell development and normal hematopoiesis. Together these represent about 12–15% of acute myeloid leukemia cases in adults and are commonly referred to as the core binding factor (CBF) leukemias .

Table 14.2 General features of AML with t(8;21)(q22;q22.1)
Full size table
Table 14.3 General features of AML with inv(16)(p13q22) or t(16;16)(p13;q22)
Full size table

Most of these cases will also carry other cytogenetic aberrancies. Presence of secondary cytogenetic aberrations or complex karyotypes do not appear to affect clinical outcomes for patients with t(8;21) AML [5]. In AML with inv(16) or t(16;16), trisomy 8 is associated with a worse prognosis, and trisomy 22 has been associated with an improved prognosis [6]. Somatic sequence mutations in KIT exons 8 and 17 are associated with a worse prognosis [1], and patients may benefit from hematopoietic stem cell transplant at first remission. Sequence mutations in genes activating tyrosine kinase signaling are frequent in both subtypes of CBF-AML; genes involving the RTK/RAS signaling pathways are affected in nearly 30% of cases and may suggest shorter event-free survival [7]. Genes involved in chromatin modification of the cohesin complex are seen at high frequencies in t(8;21) AML (42% and 18%, respectively), but are generally absent in inv(16)/t(16;16) AML [8]. Similarly ASXL2 mutations are seen in 20–25% of patients with t(8;21) AML, but are uncommon in inv(16)/t(16;16) disease [9]. RT-PCR targeted against fusion transcripts have been used for minimal residual disease (MRD) assessment in CBF-AML which appears to allow for identification of patients at high risk of relapse [10, 11]. MRD monitoring early after transplant may be more predictive of relapse risk than presence of KIT mutations [12].

Acute Promyelocytic Leukemia (APL) (Table 14.4)

APL presents with a predominance of abnormal promyelocytes and arises in the setting of fusion of the PML (a nuclear regulatory factor) and RARA (retinoic acid receptor alpha) genes. This fusion protein acts as a constitutive transcriptional repressor of RARα target-genes, but this repression may be alleviated by pharmacologic doses of tretinoin [13]. The leukemic blasts are highly sensitive to differentiating agents, tretinoin (also referred to as ATRA, all-trans-retinoic acid) and arsenic trioxide [14], as well as to anthracycline-based chemotherapy. APL is classically associated with the t(15;17)(q24.1;q21.2) translocation, but may arise from cryptic or variant PML-RARA fusions.

Table 14.4 General features of APL with PML-RARA
Full size table

Three breakpoint cluster regions (bcr) are described in the PML gene; fusions involving bcr1 and bcr2 are of similar size and are together referred to as long (L) isoform, and those involving bcr3 result in a short (S) isoform [15]. Hypergranular/typical APL represents ~70% of all cases and is often associated with the long isoform. The short isoform is more common in the microgranular (also called hypogranular) variant APL. Both variants are associated with a high risk of disseminated intravascular coagulation, increased fibrinolysis, and significant coagulopathy associated with early death [16].

Secondary cytogenetic abnormalities are found in about 40% of cases. FLT3 mutations are found in 30–40% of cases, and FLT3-ITD is associated with a higher WBC count, microgranular morphology, and involvement of the bcr3 breakpoint [17]. Variant RARA translocations also occur with gene partners other than PML. Described variant fusion partners include ZBTB16 at 11q23.2, NUMA1 at 11q13.4, NPM1 at 5q35.1, and STAT5B at 17q11.2 [18]. Such cases should be diagnosed as “APL with a variant RARA translocation.” The ZBTB16-RARA and STAT5B-RARA translocations demonstrate resistance to ATRA differentiation therapy [19].

Minimal residual disease (MRD) monitoring for PML-RARA transcripts by PCR is currently the best predictor of relapse-free survival [20]. Detection of PML-RARA by RT-PCR in the immediate post-treatment period does not impact the clinical outcome, as abnormal promyelocytes may persist for several weeks after initiating therapy. However, detection of fusion transcripts after achieving complete remission is strongly predictive of relapse, and early pre-emptive therapy may prevent overt clinical relapse [20, 21].

AML with t(9;11)(p21.3;q23.3), KMT2A-MLLT3 (Table 14.5)

This subtype accounts for about 2% of adult AML but represents 9–12% of pediatric cases. The leukemic blasts often show monocytic or myelomonocytic differentiation, and patients may present with disseminated intravascular coagulation, myeloid sarcoma, or soft tissue infiltration. KMT2A encodes a histone methyltransferase which participates in chromatin remodeling. While fusions involving KMT2A are seen in 5–10% of all AML, the WHO classification for this category is limited specifically to t(9;11)(p21.3;q23.3) [1].

Table 14.5 General features of AML with t(9;11)(p21.3;q23.3)
Full size table

Over 130 different translocations involving KMT2A have been described, including greater than 90 different gene fusion partners and at least 6 translocations with no obvious gene fusions [22]. Translocations with MLLT3 are the most common of these (~30% of cases) and appear to define a more distinct pathologic entity [1, 22]. AML with other balanced translocations of 11q23.3 are classified as AML, not otherwise specified, though the translocation should also be stated in the diagnostic line (except in cases which meet criteria for therapy-related AML or AML with myelodysplasia-related changes).

Secondary cytogenetic aberrancies and complex karyotypes may be seen in t(9;11) AML, but do not appear to affect clinical outcomes for these patients [5]. Somatic sequence mutations of NRAS or KRAS are seen in 30–40% of cases, but the incidence of FLT3 mutations is low compared to other AML subtypes [23]. Overexpression of MECOM (previously known as EVI1) has been reported in about 40% of cases, and some reports suggest that t(9;11) AML positive for overexpression are biologically distinct from MECOM-negative cases [24]. Patients with de novo AML and t(9;11)(p21.3;q23.3) are at intermediate prognostic-risk but appear to have a relatively better survival than patients with other translocations at 11q23, who generally experience more adverse clinical outcomes [25]. Overexpression of MECOM in KMT2A-rearranged AML is associated with a very poor prognosis [24, 26].

AML with t(6;9)(p23;q34.1), DEK-NUP214 (Table 14.6)

This uncommon subtype accounts for 0.7–1.8% of all cases and often presents with basophilia (≥2% basophils), cytopenias, and multilineage dysplasia [1]. Despite the presence of multilineage dysplasia, the t(6;9) takes precedence over the less specific diagnosis of AML with myelodysplasia-related changes (AML-MRC). NUP214 encodes the CAN nucleoporin. Fusion with the DEK oncogene results in abnormal transcription factor activity, most likely due to altered nuclear transport due to binding of soluble transport factors [27].

Table 14.6 General features of AML with t(6;9)(p23;q34.1)
Full size table

The DEK-NUP214 fusion is the sole cytogenetic abnormality identified in nearly 90% of cases [28]. AML with t(6;9) has poor survival rates with conventional chemotherapy, and patients may benefit from allogeneic hematopoietic stem cell transplantation. Although the WHO requires ≥20% myeloblasts to diagnose this entity, this threshold requirement is controversial. FLT3-ITD mutation is found in 70–80% of cases, but the poor prognosis of this AML subtype is independent of FLT3 mutation status [29, 30].

AML with inv(3)(q21.3q26.2) or t(3;3)(q21.3;q26.2); GATA2, MECOM (Table 14.7)

This uncommon subtype accounts for 1–2% of AML and often presents with normal to increased platelet counts and multilineage dysplasia, typically with prominent uni- or bi-lobed dwarf megakaryocytes [1]. Despite the presence of multilineage dysplasia, the presence of inv(3) or t(3;3) takes diagnostic precedence over the less specific diagnosis of AML-MRC. This rearrangement pairs the oncogene MECOM with a GATA2 enhancer. No abnormal fusion transcript is generated by this gene rearrangement, but it contributes to leukemogenesis by both stimulating MECOM expression and causing GATA2 insufficiency [31, 32]. Inappropriate expression of MECOM (previously known as EVI1) is seen in a variety of AMLs, and high expression is a poor prognostic indicator independent of 3q26.2 translocations [33]. While other aberrancies involving chromosome 3q26.2 and variant fusion partners for MECOM have been described, these are currently excluded by WHO from the AML with recurrent genetic abnormalities disease category [1].

Table 14.7 General features of AML with inv(3)(q21.3q26.2) or t(3;3)(q21.3;q26.2)
Full size table

Secondary cytogenetic aberrancies are found in most cases of AML with inv(3) or t(3;3) and are of the variety that are typically associated with myelodysplasia. Monosomy 7 can be found in up to 66% cases, and chromosome 5q deletions and complex karyotypes are also commonly described [34]. Activating mutations in genes affecting the RAS/receptor tyrosine kinase signaling pathways are found in about 98% of cases, including NRAS, PTPN11, FLT3, KRAS, NF1, CBL, and KIT [35]. Other commonly mutated genes include GATA2, RUNX1, and SF3B1 [35, 36].

This subtype of AML is typically associated with an aggressive disease course, therapy resistance, and short survival. Although the WHO requires ≥20% myeloblasts to diagnose this entity, this threshold requirement is controversial as disease associated with inv(3) or t(3;3) and <20% blasts have an equally poor outlook, similar clinicopathologic features, and identical mutational patterns at the molecular genomic level [1]. A complex karyotype or concomitant monosomy 7 worsens the already adverse prognosis associated with this subtype [34].

AML (Megakaryoblastic) with t(1;22)(p13.3;q13.1), RBM15-MRTFA (Table 14.8)

This rare subtype with megakaryoblasts represents <1% of all AML cases. It presents almost exclusively in infants; 80% of diagnoses are made within the first year of life, and most occur within the first 6 months [1]. These children usually have marked hepatosplenomegaly, cytopenias, and a densely fibrotic marrow with bilateral periostitis or osteolytic lesions. The patient may also present with a soft tissue mass, mimicking other small round blue cell tumors. The translocation fuses RBM15, a RNA recognition motif-encoding gene, to MRTFA (previously known as MLK1), a protein with a DNA-binding motif involved in chromatin organization [37].

Table 14.8 General features of AML with t(1;22)(p13.3;q13.1)
Full size table

AML with t(1;22) represents only ~14% of the non-Down syndrome acute megakaryoblastic leukemias [38]. In most cases, the RBM15-MRTFA fusion is the sole cytogenetic aberrancy [1]. When compared to other de novo cases of non-Down syndrome acute megakaryoblastic leukemia, presence of the t(1;22) translocation appears to be associated with intermediate-risk disease and inferior event-free survival [38, 39].

AML with Mutated NPM1 (Table 14.9)

Mutations in NPM1 are among the most frequent acquired genetic abnormalities in AML, occurring in 2–8% of childhood cases, 27–35% of adult cases, and 45–64% of adult normal karyotype (NK) AML [1]. The NPM1 gene encodes nucleophosmin, a multifunctional chaperone protein which localizes to the nucleus, participates in the biogenesis of ribosomes, and helps regulate the ARF-TP53 tumor suppressor pathway [40, 41]. Mutations typically involve exon 12 and lead to a frameshift in the C-terminal protein region, with subsequent cytoplasmic displacement of the protein [42]. NPM1 mutations are also considered late aberrancies in leukemogenesis, following earlier somatic mutations in genes involved in epigenetic regulatory processes such as DNA methylation, histone modification, and chromatin looping [43, 44].

Table 14.9 General features of AML with mutated NPM1
Full size table

The leukemic blasts often have a monocytic or myelomonocytic phenotype.

NPM1 mutations are usually mutually exclusive of other recurrent AML-defining cytogenetic aberrancies and are typically associated with normal karyotypes [45]. A minority of cases (5–15%) will carry nonspecific chromosomal alterations such as +4, +8, −Y, del(9q), and +21; however these findings do not appear to alter the disease profile or survival outcomes, when compared to “normal-karyotype disease” [45]. Cytogenetic aberrancies typically associated with myelodysplasia are uncommon in this setting of NPM1-mutated AML [46], but morphologic dysplasia may be seen in up to a quarter of cases. However, AML-MRC-related cytogenetic abnormalities should take diagnostic precedence if detected. Other acquired sequence mutations are common, and commutated genes often include FLT3, TET2, DNMT3A, IDH1/2, and KRAS/NRAS and cohesin complex genes [47].

In NK-AML, NPM1 mutation confers a favorable prognosis, similar to that of core binding factor AMLs [48]. A significant minority (~25%) of NPM1-mutated NK-AML may have multilineage dysplasia, but the finding does not impact the good prognosis associated with NPM1 mutation unless myelodysplasia-associated cytogenetic aberrancies are also detected [49]. About 40% of NPM1-mutated AML will have concurrent FLT3-ITD mutations, and this abnormality appears to negate the favorable prognostic effect [50]. The relative allelic ratio of FLT3-ITD appears to have prognostic significance in this setting, and NPM1-mutated patients with a low-allelic burden of FLT3-ITD (i.e., <0.5) seem to retain favorable outcomes [51]. Regardless, patients with NPM1 mutation appear to have a better prognosis than patients with FLT3-ITD and wild-type NPM1, especially in cases with a high FLT3-ITD allelic ratio (i.e., ≥0.5) [52, 53]. Concurrent mutations of NPM1, FLT3-ITD, and DNMT3A appear to have a particularly adverse impact on overall and event-free survival [54].

AML with Biallelic Mutations of CEBPA (Table 14.10)

Biallelic mutations of CEBPA may be seen in 4–9% of children with AML, at a lower frequency in adult disease, and are generally associated with a good prognosis similar to that seen in CBF-AML [1]. CEBPA encodes a protein called CCAAT enhancer-binding protein alpha, which serves multiple functions including as a hematopoiesis-associated transcription factor and also as a tumor suppressor gene. Biallelic gene mutation is required for diagnosis; the favorable prognostic association is linked to a specific gene expression profile that is not identified with single allele mutation [1, 55, 56]. Only sequence mutations of the CEBPA gene are taken into diagnostic consideration for this subtype, though there are many routes that can lead to CEBPA inactivation. This AML subtype does not have particularly distinctive morphologic features.

Table 14.10 General features of AML with biallelic mutations of CEBPA
Full size table

More than 70% of cases will be associated with a normal karyotype. Factors which might negatively impact the favorable prognostic risk include presence of cytogenetic aberrancies (i.e., an abnormal karyotype) and co-mutation with FLT3-ITD [57, 58], though this still requires additional clarification [1]. Concurrent GATA1 and WT1 mutations are relatively frequent in patients with biallelic CEBPA mutation, but FLT3-ITD, NPM1, ASXL1, and RUNX1 mutations are uncommon and seen more frequently in CEBPA monoallelic cases [59]. Cytogenetic aberrancies typically associated with myelodysplasia are uncommon in this setting of biallelic CEBPA mutation, but morphologic dysplasia may be seen in about a quarter of cases [1, 60]. The finding of dysplasia alone does not influence the prognosis, but AML-MRC-related cytogenetic abnormalities should take diagnostic precedence if detected [1, 60] (see Table 14.11).

Table 14.11 Cytogenetic abnormalities diagnostica for AML-MRC
Full size table

Germline mutation of CEBPA is also a described phenomenon and is well associated with predisposition to develop AML. Therefore, identification of biallelic CEBPA mutation in AML should prompt evaluation of possible germline inheritance , especially in patients presenting as children or young adults (see section “Myeloid Neoplasms with Germline Predisposition”).

Provisional 2016 WHO AML with Recurrent Genetic Abnormality Subtypes

BCR-ABL1 fusion and RUNX1 mutation define new provisional entities in the 2016 WHO AML classification system . AML with BCR-ABL1 is a de novo AML with no evidence of chronic myeloid leukemia, both prior to and after therapy. This is a rare subtype, accounting for <1% of all cases of AML [1, 61]. Most cases demonstrate the p210 fusion, though a minority of reported cases had p190 transcripts. Most cases have additional cytogenetic abnormalities such as loss of chromosome 7, gain of chromosome 8, or complex karyotypes [61,62,63]. AML with BCR-ABL1 is reported to be an aggressive disease with poor response to traditional AML therapy or tyrosine kinase inhibitor therapy alone.

RUNX1 mutations are reported to occur in 4–16% of AML, but can also be found in numerous other myeloid neoplasms. The diagnosis of AML with mutated RUNX1 should not be made for cases that fulfill criteria for any of the other specific AML subtypes, including AML with recurrent genetic abnormalities, AML with myelodysplasia-related changes, and therapy-related myeloid neoplasms [1]. RUNX1 mutations found in the setting of myelodysplasia (MDS) frequently coincide with additional gene mutations including SRSF2, EZH2, STAG2, and ASXL1, and this profile appears similar in AML with mutated RUNX1 [64, 65]. Some studies have associated RUNX1 mutations with worse overall survival in AML. Germline mutation of RUNX1 is also described and is associated with an autosomal dominant thrombocytopenia and also increased risk for MDS/AML. When identified in AML, RUNX1 mutation should prompt evaluation of family history and possible consideration for germline sequence analysis (see section “Myeloid Neoplasms with Germline Predisposition”).

AML with Myelodysplasia-Related Changes

This diagnostic category represents 24–35% of AML and encompasses disease with ≥20% peripheral blood or bone marrow myeloblasts and (1) dysplasia in ≥50% of at least two cell lines, (2) a prior history of MDS or myelodysplastic/myeloproliferative neoplasm (MDS/MPN), or (3) underlying MDS-associated cytogenetic abnormalities [1]. Identifying an AML-associated recurrent cytogenetic aberrancy or history of cytotoxic/radiation therapy for unrelated disease would exclude this diagnostic category. The cytogenetic aberrancies associated with this category of AML are similar to those found in MDS and include complex karyotypes, unbalanced gains/losses of major chromosomal regions, and number of uncommon balanced translocations (Table 14.11). Some abnormalities that are common in MDS, such as trisomy 8, del(20q), and loss of chromosome Y, are not sufficiently specific in isolation to diagnose AML-MRC [1].

This category is generally associated with a poorer prognosis and lower rates of complete remission than other AML subtypes [66, 67]. There are generally no significant differences in survival between AMLs arising from myelodysplasia and de novo AMLs with multilineage dysplasia [67]. Some cases with a prior history of MDS, intermediate-risk cytogenetics , and relatively low blast counts (20–29%) may exhibit clinical behavior more similar to MDS [68], with response and survival benefit from hypomethylating agents. Cases with high-risk cytogenetics generally have no survival differences compared to AML cases with ≥30% blasts [68]. Of note, a significant minority of AML associated with NPM1 or biallelic CEBPA mutations will show multilineage dysplasia. In the absence of MDS-specific cytogenetic aberrancies, these cases retain a good prognostic outlook, with similar behavior to cases without multilineage dysplasia [49, 60].

The 2016 WHO classification system does not recognize any somatic gene sequence mutations as being diagnostically specific for the AML-MRC category. However, acquired variants in some genes have frequent association with secondary AMLs arising from antecedent myeloid malignancy. Mutations in SRSF2, SF3B1, U2AF1, ZRSR2, ASXL1, EZH2, BCOR, STAG2, RUNX1, and TP53 are common in AML-MRC and occur at a higher frequency than in other forms of AML, NOS [69, 70]. Presence of TP53 mutations is almost always associated with complex karyotypes and may suggest an even worse prognosis than other cases in this already poor prognostic group [69,70,71,72].

Therapy-Related Myeloid Neoplasms

Therapy-related myeloid neoplasms (t-MNs) arise as an uncommon late effect of chemotherapy and/or radiation therapy for an unrelated illness, usually another malignancy, solid organ transplant, or autoimmune disease. The morphologic presentation at diagnosis can be variable, and this category represents about 10–20% of all cases of AML, MDS, and MDS/MPN [1]. t-MNs are morphologically heterogeneous and can look like either MDS or AML, but the 2016 WHO classifies them collectively in a single category due to general behavioral similarities and extremely poor outcomes that are independent of blast counts [1]. The most common antecedent malignancies are breast, lung, and hematologic cancers, chiefly lymphomas and multiple myeloma [73, 74]. The leukemic blasts do not have diagnostically specific morphologic or immunophenotypic features.

The leukemic cells in t-MNs will demonstrate an abnormal karyotype in >90% of cases [75, 76]. Essentially all the balanced cytogenetic abnormalities associated with AML-MRC are also found in t-MNs; thus the clinical history is central to assigning a correct diagnosis. A positive history of cytotoxic therapy takes diagnostic precedence over morphologic dysplasia and MDS-associated cytogenetics. Two general subsets of t-MNs are clinically recognized, associated with either (1) alkylating agents and/or ionizing radiation therapy or (2) topoisomerase II inhibitor therapy (Table 14.12). However, as patients may undergo multiple therapeutic exposures, there can be overlap between the general archetypes [77]. The pathogenic effect of isolated limited-field radiation therapy is unclear, and the incidence of associated t-MNs associated with this form of therapy is uncertain [78].

Table 14.12 Major subtypes of therapy-related myeloid neoplasms
Full size table

The more common subtype arises after alkylating agent and/or radiation therapy (~70% of patients). There is usually latency period of 5–10 years, an MDS-like phase with dyspoiesis and cytopenias and rapid progression to overt AML with multilineage dysplasia. These cases are associated with unbalanced chromosomal losses (often involving chromosomes 5 and/or 7), complex karyotypes, and mutations or loss of TP53. Loss of 5q is often seen with additional chromosomal abnormalities in a complex karyotype, and up to 80% of patients with del(5q) will also have mutations or loss of TP53 [1].

The second subtype arises after topoisomerase II inhibitor therapy, but may also be seen with radiation therapy alone. The latency period is shorter (1–5 years), patients usually do not have an MDS-phase, and overt leukemia is found on presentation. Balanced translocations are more frequent in this subgroup, often involving KMT2A at 11q23.3 or RUNX1 at 21q22.1. Category-specific balanced chromosomal rearrangements have been described, such as the t(15;17) PML-RARA fusion associated with APL or the inv(16) CBFB-MYH11 fusion associated with CBF-AML. The clinical behavior of these cases is still unresolved; some groups have reported comparable outcomes to de novo disease, while others have indicated worse overall and event-free survival [79, 80].

In general, the prognosis of this disease category is exceptionally poor with overall 5-year survival rates that are often reported at <10%. Cases with abnormalities of chromosome 5 and/or 7, TP53 mutations, or complex karyotypes have a median survival time of <1 year regardless of presentation as overt t-AML or as t-MDS [1]. Somatic sequence mutations are frequently reported in the TET2, PTPN11, IDH1/2, NRAS, and FLT3 genes, but the clinical significance of these findings is still undetermined [81, 82].

Myeloid Neoplasms with Germline Predisposition and AML in Children

Myeloid Neoplasms with Germline Predisposition

A number of germline abnormalities have been linked with an inherited predisposition toward myeloid malignancies, but only a few are specifically predisposing to AML. These are rare disorders which represent <1% of AMLs, but the relative frequency of subtypes within this diagnostic category has not been well established [1]. Patients present more frequently in childhood, though few subtypes with late-onset have been described, and recognition is important for the screening of family members. These disorders are quite rare but the few better-characterized entities fall into three groups within the 2016 WHO system, as summarized below:

  • Myeloid neoplasms with germline predisposition without a pre-existing disorder or organ dysfunction

    • Acute myeloid leukemia with germline CEBPA mutation

    • Myeloid neoplasms with germline DDX41 mutation

  • Myeloid neoplasms with germline predisposition and pre-existing platelet disorders

    • Myeloid neoplasms with germline RUNX1 mutation

    • Myeloid neoplasms with ANKRD26 mutation

    • Myeloid neoplasms with ETV6 mutation

  • Myeloid neoplasms with germline predisposition and other organ dysfunction

    • Myeloid neoplasms with germline GATA2 mutation

    • Myeloid neoplasms associated with bone marrow failure syndromes

    • Myeloid neoplasms associated with telomere biology disorders

    • Juvenile myelomonocytic leukemia associated with neurofibromatosis, Noonan syndrome, or Noonan syndrome-like disorders

    • Myeloid neoplasms associated with Down syndrome

AML may be seen associated with any of the germline predisposition entities, but a clinical picture dominated by either MDS or AML with no other significant organ dysfunction is primarily seen with the first group, including CEBPA and DDX41 mutations. Disorders associated with germline DDX41 mutation appear to have a longer latency period, with a median age of 62 years at malignancy onset [1]. An increased risk for lymphoid malignancies is also reported for the entities associated with DDX41, RUNX1, ANKRD26, and ETV6 mutations and also with Down syndrome [1].

Transient Abnormal Myelopoiesis and Myeloid Leukemia Associated with Down Syndrome

Persons with Down syndrome have a 10- to 100-fold increased risk of developing acute leukemia than unaffected persons. About 70% of these cases have a megakaryoblastic phenotype, which is rare in non-Down syndrome associated AML [1]. Additionally, a significant minority of infants with Down syndrome may also present with a temporary clonal myeloid proliferation whose features can mimic and even meet criteria for AML. This unusual condition is referred to as transient abnormal myelopoiesis (TAM) associated with Down syndrome . The blasts found in the vast majority of TAM also exhibit a megakaryoblastic immunophenotype [1]. The unique clinical characteristics of both these myeloid proliferations were recognized by the WHO, resulting in a separate categorization in 2008, which has persisted into the 2016 update.

Trisomy 21 itself causes perturbation of fetal hematopoiesis with abnormal production in the liver, increases in the number of megakaryocyte-erythroid progenitors, and increases in the hematopoietic stem cell compartment [83]. These abnormalities are congenital and precede the acquisition of disease-associated somatic mutations [84]. Essentially all cases of Down syndrome-associated TAM and AML will acquire a subsequent mutation of GATA1, a hematopoietic transcription factor that regulates normal megakaryocyte and erythrocyte differentiation [83, 85, 86]. More than 95% of the pathologically significant variants are in exon 2 with the remainder in exon 3, with resultant N-terminal protein truncation [87]. Additionally, up to 25–30% of all neonates with Down syndrome may be found to carry these mutations, though the reason for the high frequency in this setting is unclear [88].

However, GATA1 mutations are insufficient in isolation to cause myeloid leukemia associated with Down syndrome; 80–90% of patients with TAM will show spontaneous regression of the process within the first 3 months of life [1, 89, 90]. Patients with TAM who do progress to acute leukemia usually do so within the first 5 years of life, and acquisition of additional oncogenic mutations can usually be demonstrated. Trisomy 8 is common in this setting (13–44% of cases), but monosomy 7 is very rare [1, 91]. Whole genome or exome sequencing studies at progression to acute leukemia have shown about 50% of cases acquire mutations in cohesin complex genes (RAD21, SMC1A, SMC3, and STAG2), 45% will involve epigenetic regulators such as EZH2 and KANSL1, and 20% will involve the transcription factor CTCF [89, 90]. Other signaling pathways such as JAK kinases, MPL, and RAS pathway genes (NRAS, KRAS, CBL, PTPN11, and NF1) were implicated in a smaller subset of cases [89, 90]. However, no specific genetic abnormalities can consistently predict transformation of TAM to acute leukemia at present.

Childhood AML

AML accounts for only 20% of pediatric acute leukemias, but is overtaking acute lymphoblastic leukemia as the leading cause of childhood leukemia-related mortality [92]. Both adult and childhood AML have a low overall mutation burden compared to other human cancers, with a broad spectrum of recurrently impacted but relatively infrequently affected genes [92]. However, the landscape of structural and sequence-related genomic aberrancies in pediatric AML shows significant differences from the adult cohort.

While there is some overlap of recurrent cytogenetic abnormalities seen in adult and childhood AML, the general types of balanced and unbalanced chromosomal abnormalities are different. Structural variants are disproportionately prevalent in younger patients, with a variety of uncommon recurrent balanced translocations and inversions beyond the specifically named entities in WHO classification system. A selection of these rare balanced rearrangements with higher prevalence in pediatric AML may be found in Table 14.13. Rearrangements involving KMT2A are the most common, seen in ~10–20% of children but in nearly half of affected infants [92, 93]. Similar to adult patients, AML associated with t(8;21), inv(16), and t(15;17) are associated with superior outcomes, while complex karyotypes and monosomy 7 are associated with poor outcomes [92,93,94]. Monosomal karyotypes have also been described as an indicator of poor outcome [95, 96]. Recurrent focal deletions are other characteristic findings in pediatric AML. Copy number loss are more common in children, and the ZEB2, MBNL1, and ELF1 genes are often affected; ZEB2 and MBNL1 co-deletion is a relatively frequent finding, and half of these are found accompanying KMT2A-MLLT3 fusions [92]. KMT2A fusions were also commonly associated with RAS-related mutations (KRAS, NRAS, PTPN1, or NF1), and a subset of KMT2A fusions also showed recurrent mutation in post-transcriptional splicing genes (i.e., SETD2, U2AF1, and DICER1) as the sole additional abnormality [92].

Table 14.13 Chromosomal translocations with a higher prevalence in pediatric AML
Full size table

Mutations of WT1 appear to be mutually exclusive with those in ASXL1 and EZH2, but WT1 or EZH2 variants are seen in about one-quarter of pediatric AML cases and may represent early clonal or near-clonal origin [92]. Widespread gene silencing by aberrant promoter methylation is enriched in younger patients with WT1 mutations, and mutations of WT1, ASXL1, or EZH2 are associated with induction failure [92]. Other recurrently mutated genes in pediatric AML include variants in GATA2, CBL, MYC-ITD, NRAS, and KRAS. NRAS and WT1 are mutated more often in younger patients than adults; conversely, mutations in DNMT3A, IDH1/2, RUNX1, NPM1, and TP53, which are common in adults, are seen more often in older patients [92]. Given the ongoing discovery clarifying the genetics of pediatric AML, more robust classification systems for diagnosis and treatment of childhood AML will likely be forthcoming.