Cancer in Children: Clinical Management, 5th Edition

Chapter 13. Acute myeloid leukaemia

Brenda Gibson

Geoff Shenton


Leukaemia is the most common malignancy in childhood, accounting for 30 per cent of cancer registration in children under the age of 15 years in the USA and Europe. While cancer is the most common cause of death (after accidents) in children, leukaemia, despite its high cure rates, remains the most common cause of cancer-related mortality.


Acute myeloid leukaemia (AML) accounts for 20 per cent of leukaemias in childhood and is the seventh most common childhood malignancy. There are 70 new cases of AML per year in the UK. Data from the National Cancer Institute reported an overall incidence of AML in children of 7.6 per million for the period 1975–1995, with considerable age-related variation. The incidence of AML was found to be highest during the first 2 years of life (12 per million), decreasing to less than 4 per million at around 9 years of age, before increasing again throughout the teenage years to reach an incidence of approximately 8 per million by early adult life. There was no reported significant difference in the sex-adjusted incidence and, unlike acute lymphoblastic leukaemia (ALL), no difference between Black and White ethnic groups, although the incidence was higher in American Hispanic children. Data from the International Association of Cancer Registries reported a marked ethnic variation with the highest rates (14.4 per million) in the Maori population of New Zealand and one of the lowest incidences in the South American countries of Brazil (3.5 per million) and Colombia (1.6/ million). There was also marked geographical variation in subtype, with acute promyelocytic leukaemia being three times more common in Latin than non-Latin countries.

Risk factors for AML

A number of risk factors for the development of AML have been identified. The association is well established for some, such as previous exposure to chemotherapeutic agents, ionizing radiation, and congenital syndromes, but is less so for others.

Down syndrome (trisomy 21) carries a >10-fold increased risk of AML. Neonates with Down syndrome frequently develop a transient clonal megakaryoblastic myeloproliferative disorder that resembles AML and is termed transient abnormal myelopoiesis (TAM).This usually resolves within the first 3 months of life. However, 30 per cent of those affected will go on to develop AML by the age of 3 years. The median age at presentation for AML in Down syndrome is 23 months. Compared with non-Down-syndrome children with AML, Down syndrome patients tend to have lower leucocyte counts, preceding myelodysplasia, megakaryoblastic or undifferentiated AML, and a more favourable outcome. Approximately 10 per cent of all paediatric AML cases occur in children with Down syndrome, and patients with mosaicism are at risk of leukaemia in the affected bone marrow cells. The mechanism of this increased risk has yet to be elucidated, although GATA1, a gene involved in regulating survival and maturation of erythroid and megakaryocytic progenitor cells, has been implicated. There is an association with mutations in the GATA1 gene and megakaryoblastic leukaemia in Down syndrome patients, although the mechanism by which trisomy 21 provides a genetic or molecular environment that promotes leukaemia is unclear.

Fanconi's anaemia and Bloom's syndrome are characterized by increased chromosomal fragility and as a consequence of this are associated with an increased risk of AML. In children with severe congenital neutropenia (Kostman's syndrome), the infection-related mortality has been reduced with the prophylactic use of recombinant granulocyte colony-stimulating factor (rGCSF), whilst an increase in the incidence of AML has been noted in the children who have survived. Patients are reported with mutations in the GCSF receptor, leading to a truncated protein with defective internalization and sustained activation on stimulation with GCSF. This results in an abnormally high proliferative rate in myeloid precursors and a terminal differentiation block.

Neurofibromatosis type I is an autosomal dominant disorder with mutations of the NF1 gene on chromosome 17. The product of this gene plays a key role in the Ras signalling pathway. Mutations of NF1 lead to increased levels of activated Ras. Patients have a 20-fold increased risk of developing AML due to mutation of the remaining NF1 locus.

Other factors, such as maternal alcohol consumption during pregnancy and parental exposure to pesticides and benzene, have been implicated in leukaemogenesis. Further epidemiologic studies should better clarify the contribution of these risk factors to the development of AML.

Presenting features

Children with AML present with signs and symptoms of both failure of normal haematopoiesis and infiltration of extramedullary sites due to expansion of the leukaemic clone. These include pallor, fever, bone pain, anorexia, fatigue, mucosal bleeding, and bruising. Patients are commonly neutropenic at presentation and at increased risk of bacterial infection. Prodromal symptoms may be present from days to weeks before presentation.

Extramedullary infiltration with hepatosplenomegaly and lymphadenopathy is common. Chloromas, collections of leukaemic cells so called because of their greenish appearance, may occur in the skin (leukaemia cutis), soft tissues, gingivae, orbit, or at other sites. Organ infiltration is most common in monocytic/monoblastic leukaemias (FAB type M4 and M5). Chest radiographs may show abnormalities due to leukaemic infiltrates or infection. Central nervous system (CNS) involvement at presentation has been reported in 5–25 per cent of cases. Infants and older children with M4/M5 leukaemia are at particular risk. Chloromas of the nervous system may present as headaches, focal neurologic deficits, cranial nerve palsies, and seizures. The risk of leucostasis involving the lung or brain is greatest when the peripheral blast count is very high (> 200 109/liter). DIC or hyperfibrinolysis with haemorrhagic complications may be seen in all subtypes of AML, but most commonly in acute promyelocytic leukaemia (FAB type M3) and the M4/M5 subtypes.

Laboratory features

Blood film and blood count

Children commonly present with pancytopenia [low haemoglobin (Hb), white cell count (WCC), and platelet count] and blasts (myeloblasts, monoblasts, or megakaryoblasts) on a peripheral smear. There is usually a normochromic normocytic anaemia and thrombocytopenia, but in a minority the platelet count may be normal or even high. The peripheral blood cells may show dysplastic features and Auer rods may be present in the myeloid blasts. Pancytopenia without circulating immature cells is common.

Bone marrow appearances

The diagnosis of AML is made by examining the bone marrow, and subtype classification is performed by histochemical staining, flow cytometry, and cytogenetics. It can be difficult or impossible to aspirate bone marrow because of fibrosis or a tightly packed marrow. In this situation of a ‘dry tap’, a trephine biopsy is necessary for diagnostic purposes and touch preparations of the biopsy specimen may be helpful.

Morphologic assessment is performed on a Romanowsky stained smear. Myeloblasts are larger than lymphoblasts and have more cytoplasm (which often contains a variable number of primary granules). The nucleus may have a number of distinct nucleoli. Auer rods (coalescences of primary granules) are pathognomic of malignant myeloid blasts (Fig. 13.1). Erythroblasts may show grossly abnormal morphology with multinucleate forms, nuclear fragments, and megaloblastic change. Megakaryoblasts are highly polymorphic, ranging from small round cells resembling lymphoblasts to larger more undifferentiated cells which may have cytoplasmic budding.


Cytochemical staining provides information on lineage involvement and maturation, and complements the information obtained from the Romanowsky stain and immunophenotyping.

Sudan Black and myeloperoxidase

Myeloperoxidase (MPO) stains the enzyme which is found in the peroxisomes of neutrophilic, eosinophilic, and monocytic cells. The Sudan Black stain non-enzymatically identifies the phospholipid in the membranes of primary and secondary granules. The strength of the reaction generally parallels that of MPO. Monoblasts, when positive, tend to show a fine scattered granular pattern whereas myeloblasts have more localized granular positivity. A positive reaction for Sudan Black is defined as >3 per cent staining.


The carboxylic esterases identify granulocytic and monocytic lineage. Chloroacetate esterase is specific for granulocytes and mast cells. Reactivity is due to the presence of several different enzymes in the secondary granules. Myeloblasts are generally negative, with positivity appearing as the cells become more differentiated. Non-specific esterase (e.g.α-napthyl butyrate esterase) stains monoblasts, monocytes, and megakaryocytes. Both chloroacetate esterase and non-specific esterase staining can be performed on the same slide (combined esterase).


Panels of monoclonal antibodies detect antigenic determinants on normal and leukaemic myeloid cells. The British Committee for Standards in Haematology (BCSH) recommends a standardized panel of monoclonal antibodies. This includes B-lymphoid (CD79a, CD22, CD19, CD10), T-lymphoid (CD3, CD2), myeloid (CD117, anti MPO, CD13), and non-specific (TdT). If the lineage is not clearly established, a second-line panel of CD33, CD41, CD42, CD61, glycophorin A, and CD7 should be used to clarify lineage. Additional useful markers include HLA-DR and the typically monocytic lineage expressed CD14, CD11, and antilysozyme. The BCSH guideline highlights the importance of interpreting the immunophenotype in combination with and not in isolation from the cytochemistry. (Table 13.1).

Fig. 13.1 (a) Myeloid blasts (FAB type M1); (b) acute promyelocytic leukaemia showing granular blasts and a faggot cell.

Table 13.1. Panel of monoclonal antibodies for the classification of acute leukaemias

Haemopoetic precursors

CD34, HLA-DR, TdT, CD45


CD 13, CD15, CD33, CD117, MPO


CD11b, CD11c, CD14, CD36, lysozyme


Glycophorin A


CD41, CD42, CD61


CD19, CD20, CD22, CD79a


CD2, CD3, CD5, CD7

Immunophenotyping should distinguish between lymphoid and myeloid blast cells in the majority of cases. Difficulty can arise with lymphoblastic leukaemias which express myeloid antigens and with true stem cell leukaemias which show no lineage commitment. Biphenotypic leukaemia can be identified by the coexpression of myeloid and T- or B-lymphoid markers, and diagnosis is dependent on an internationally agreed scoring system.

Classification of AML

For the past 30 years the classification of AML has been based on the French–American–British (FAB) criteria. These attempted to subclassify AML on the basis of the morphologic characteristics and cytochemical staining patterns of bone marrow blast cells. Eight subtypes of AML (M0–M7) are recognized, and the classification has been repeatedly revised to incorporate diagnostic advances, including immunophenotyping and improved cytogenetics. It is now accepted that not only do specific chromosomal rearrangements characterize subtypes, but that it is the genetic change that determines the biology of the leukaemia and the clinical outcome for the patient. In response to these developments the World Health Organization (WHO) has produced a new classification for AML incorporating morphologic, immunophenotypic, genetic, and clinical features. This aims first to define leukaemias that are biologically homogeneous and secondly to be of clinical relevance. AML is classified into four major categories:

·  AML with recurrent genetic abnormalities

·  AML with multilineage dysplasia

·  AML, therapy related

·  AML, not otherwise categorized.

The WHO classification recognizes factors which are important in predicting the biology of the leukaemic process and, where possible, allocates leukaemias to one of three categories based on specific criteria. Leukaemias which cannot be classified in this manner are grouped together in a fourth category which is based on a modified version of the FAB system (Table 13.2). A number of cytogenetic abnormalities are associated with molecular events which correlate strongly with both response to therapy and outcome, allowing risk stratification of patients with AML. Specific chromosomal abnormalities are associated with a favourable, intermediate, or unfavourable outcome. The translocations inv(16), t(8;21) and t(15;17) are associated with a favourable outcome, whilst 7/del(7q) and 5/del(5q) are seen in AML with myelodysplasia and are associated with an unfavourable outcome. Secondary-therapy-related AML occurring after cytotoxic chemotherapy is recognized as a distinct disease entity.

Table 13.2. The WHO classification of AML

Acute myeloid leukaemia with recurrent genetic abnormalities

·  Acute myeloid leukaemia with t(8;21)(q22;q22); (AML1/ETO)

·  Acute myeloid leukaemia with abnormal bone marrow eosinophils, inv(16)(p13q22) or
t(16;16)(p13;q22); (CBFβ/MYH11)

·  Acute promyelocytic leukaemia with t(15;17)(q22;q12); (PML/RARα) and variants

·  Acute myeloid leukaemia with 11q23 (MLL) abnormalities

Acute myeloid leukaemia with multilineage dysplasia

·  Following myelodysplastic syndrome or myelodysplastic/myeloproliferative disorder

·  Without antecedent myelodysplastic syndrome

Acute myeloid leukaemia and myelodysplastic syndromes, therapy related

·  Alkylating agent related

·  Topisomerase type II inhibitor related (may include some lymphoid)

·  Other types

Acute myeloid leukaemia not otherwise categorized

·  Acute myeloid leukaemia minimally differentiated (FAB M0)

·  Acute myeloid leukaemia without maturation (FAB M1)

·  Acute myeloid leukaemia with maturation (FAB M2)

·  Acute myelomonocytic leukaemia (FAB M4)

·  Acute monoblastic and monocytic leukaemia (FAB M5)

·  Acute erythroid leukaemia (FAB M6)

·  Acute megakaryoblastic leukaemia (FAB M7)

·  Acute basophilic leukaemia

·  Acute panmyelosis with myelofibrosis

·  Myeloid sarcoma

Reproduced from R.D. Brunning et al. (2001) In E.S. Jaffe et al. (eds) Pathology and Genetics of Tumours of
Haemopoetic and Lymphoid Tissues
WHO Classification of Tumours. Lyon: IARC Press.

Prior to the WHO classification the diagnosis of leukaemia required an infiltrate of 30 per cent blasts in the bone marrow, but the revised classification has lowered the requirement to 20 per cent. This followed the recognition that patients with 20–30 per cent blasts in their bone marrow have a similar outcome to those with >30 per cent. Therefore patients with myelodysplasia and refractory anaemia with excess blasts in transformation (RAEB-T) are now included within the diagnostic criteria for AML. It must be appreciated that any threshold is arbitrary, and that clinical and other laboratory features must be taken into account when making a diagnosis of AML.

AML with t(8;21)(q22;q22) (AML1-ETO)

Occurring in 8–16 per cent of childhood AML, t(8;21) is one of the most common structural abnormalities and is associated with FAB M2 morphology. The blasts are usually large and frequently contain Auer rods. The blasts express myeloid markers (CD13, CD33, MPO) and CD34 is characteristically present. Some patients show weak TdT positivity and coexpression of the lymphoid marker CD19 in a subset of blasts on immunophenotyping.

In t(8;21)(q22;q22), the AML1 gene (also known as the RUNX gene), encoding CBFα, a DNA binding protein, becomes fused with the ETO (eight-twenty-one) gene which encodes a zinc-finger transcription factor. The AML1–ETO transcript is consistently found in patients with t(8;21), and the disruption of the AML1 gene is reported to occur within a single intron.

AML with inv(16)(p13q22) or t(16;16)(p13q22); (CBFβ–MYH11)

AML with inv(16) occurs in 3–12 per cent of childhood AML. It tends to show monocytic and granulocytic differentiation with an abnormal eosinophil component (M4 with eosinophilia or M4eo). Cases of inv(16)(13q22) have been seen which lack the characteristic eosinophilia. Auer rods may be observed in the myeloblasts and at least 3 per cent show MPO positivity. In addition to myeloid antigens (CD33, CD13, and MPO), the immunophenotype often shows markers of monocytic differentiation (CD14, CD11b, CD11c, CD36, and lysozyme).

The core binding factor β gene is located at 16q22 and encodes half of a heterodimeric transcription factor which is known to bind the enhancers of various leukaemia viruses. The translocation results in the fusion of this gene with the smooth muscle myosin heavy chain (MYH11) at 16p13. Occasionally there may be characteristic morphology but without karyotypic evidence of a chromosomal abnormality, and in these cases CBFβ–MYH11 may be demonstrated by molecular studies.

Acute promyelocytic leukaemia with t(15;17)(q22;q12) (PML–RARα) and variants

Acute promyelocytic leukaemia (APL) (FAB subtype M3) is characterized by an arrest of myeloid differentiation at the promyelocyte stage with abnormal proliferation of these cells. It accounts for approximately 4–11 per cent of childhood leukaemias, with significant geographical variations in incidence. There are two subtypes, one characterized by ‘typical’ hypergranular promyelocytes and the other by ‘variant type’ hypogranular promyelocytes. The latter subtype is seen in 10–30 per cent of cases of APL.

Patients with APL present with the usual signs and symptoms of pancytopenia. At presentation, APL is more often associated with a severe bleeding diathesis and DIC than other subtypes of AML. This potentially serious complication has been greatly reduced by the use of all-trans retinoic acid (ATRA). Correction of the coagulopathy does not completely abrogate the risk of CNS haemorrhage and early death. This complication is more commonly reported in paediatric patients with the M3v subtype. Hypergranular APL usually presents with a lower WCC than other AML subtypes, whilst in variant APL the WCC tends to be very high, with a rapid doubling time.

The diagnosis of APL has traditionally been based on morphology, immunohistochemistry, flow cytometry, and demonstration of the cytogenetic abnormality t(15;17). However, the widespread availability of RT-PCR allows demonstration of the fusion product and rapid confirmation of the diagnosis. Morphologically, classical APL cells are hypergranular promyelocytes containing numerous Auer rods. Characteristic cells containing bundles of Auer rods (‘faggot cells’) are present in many cases. The cells of APLv have distinct morphology with a paucity or absence of granules and a bi-lobed shape to the nucleus. However, even in APLv a small number of faggot cells or abnormal promyelocyes with clearly visible granules are usually present. By flow cytometry, APL cells are positive for CD33, CD13, and CD9 but negative for HLA-DR, CD34, CD7, CD11b, and CD14. Blasts in APL may also be positive for CD2.

In the majority of cases of APL the t(15;17) can be detected by standard cytogenetics or PCR for the gene transcript. The t(15,17) results in fusion of the retinoic acid receptorα (RARα)on 17q12 to the promyelocytic leukaemia (PML) gene on 15q22, which is a nuclear regulatory factor. In normal cellular development the PML-encoded protein is a component of an intranuclear structure which may play a part in mRNA processing. The chimeric PML–RARαprotein disrupts the formation of this structure, leading to a differentiation block. Treatment with ATRA, a retinoic acid compound, results in selective proteolytic degradation of the PML–RARα protein and the promotion of differentiation of APL cells. This leads to differentiation and apoptosis of the malignant cells, and improvement in the associated coagulopathy.

Variant translocations may also be seen in APL and include t(11;17)(q23;q21) in which RARα on chromosome 17 fuses with the promyelocytic leukaemia zinc-finger gene (PLZF)on chromosome 11, t(5;17)(q23q12) in which the nucleophosmin (NPM) gene on chromosome 5 fuses with RARα, and t(11;17)(q13;q21) in which the nuclear matrix associated (NuMA) gene fuses with RARα. The PLZF–RARα fusion is particularly important because patients with this translocation are resistant to ATRA.

AML with 11q23 (MLL) abnormalities

AML with 11q23 abnormalities is found in 6–11 per cent of childhood leukaemias. The incidence is higher in infants and in secondary-therapy-related AML which has developed after treatment with topoisomerase II inhibitors. Patients may present with DIC and tissue infiltration, the latter especially of skin or gingiva. Morphologically there is a strong association with monocytic and myelomonocytic leukaemias, particularly monoblastic leukaemia.

Molecular studies have shown that the MLL gene is a human homologue of the Drosophila trithorax gene, a developmental regulator, which is structurally altered in leukaemia-associated translocations. Although the MLL gene at 11q23 is involved in a number of translocations with at least 20 different partner chromosomes, the most common in childhood AML are t(9;11) (p21q23) and t(11;19)(q23p13.1)/t(11;19)(q23;13.3). Interestingly t(4,11) is not seen in infants with AML but is common in infant ALL. Molecular techniques have also shown that 11q23 rearrangements may be found more frequently than is seen by conventional cytogenetics alone.

AML with multilineage dysplasia

AML with multilineage dysplasia is an acute leukaemia with >20 per cent blasts in the blood or bone marrow and dysplasia in 50 per cent of the cells of at least two or more myeloid cell lines. It may occur de novo or following a myelodysplastic disorder. Patients often present with severe pancytopenia. The bone marrow is usually hypercellular with dysgranulopoiesis (hypogranular and/or abnormally segemented forms), dyserythropoiesis (abnormal nuclear outlines, multinucleate forms, and abnormal haemoglobinization) and dysmegakaryopoiesis (micromegakaryocytes, monolobular or disparate nuclei). Cytogenetic abnormalities, especially monosomy 7, are seen relatively frequently.

Therapy-related AMLs and myelodysplastic syndromes

AML and myelodysplasia can follow treatment with cytotoxic chemotherapy and/or radiation therapy and are categorized into two main types based on the implicated causative agent. If appropriate, they may be further classified by morphology or cytogenetics with the qualifying term ‘therapy related’.

Alkylating agent related AML

This form of treatment-related AML, which usually occurs 5–6 years after exposure to alkylating agents or radiotherapy, arises as a direct result of the mutagenic effect of these agents. Often presenting initially as a myelodysplastic disorder with evidence of isolated or pancytopenias, the disorder progresses through a dysplastic phase with multilineage dysplasia to frank leukaemia. Initially the percentage of marrow blasts may be low (<5 per cent), although a minority of patients present with frank AML.

The marrow morphology reflects the underlying dysplasia with hypolobulated and hypogranular neutrophils and dysplastic erythroid cells. In a minority of cases Auer rods are present. The marrow is usually hypercellular, but hypocellularity and marked marrow fibrosis may be present. Immunophenotyping reflects the heterogeneity of the underlying morphology with blasts constituting only a subpopulation of marrow cells. The blasts are generally CD34 positive and express myeloid markers (CD13, CD33). Aberrant expression of CD56 and/or CD7 may be seen.

There is a high incidence of clonal cytogenetic abnormalities which are similar to those seen in MDS and AML with multilineage dysplasia. These are primarily unbalanced translocations involving chromosomes 5 or 7 and involve the loss of all or part of the long arm of the chromosome. Other chromosomes frequently involved include 1, 4, 12, 14, and 18. Complex chromosomal abnormalities are a common finding.

Topoisomerase II inhibitor related AML

Topoisomerase II induced AML tends to have a shorter latent period than alkylating agent induced AML, with the median presentation 3 years after exposure. Although the main causative agents are the epipodophyllotoxins etoposide and teniposide, anthracyclines such as doxorubicin and 4-epi-doxorubicin have been implicated. The presentation tends to be with frank AML, although a preceding myelodysplastic phase is sometimes seen.

Morphologically there is usually a significant monocytic component with morphology characteristic of FAB M4 or M5 types, although other FAB types have been reported. The predominant cytogenetic finding is a balanced translocation involving the MLL gene (11q23) and this is commonly t(9;11), t(11;19), or t(6;11). Acute lymphoblastic leukaemias have been reported following topisomerase II inhibitors and are usually associated with t(4;11)(q21q23).

Therapy in AML

The survival of children with AML has improved dramatically over the past 30 years and continues to do so (Fig. 13.2). This is in part due to the delivery of increasingly intensive chemotherapy, but equally to improvements in supportive care, including antibacterials, antifungals, growth factors, anti-emetics, and blood product support, which have allowed intensive chemotherapy to be given with less morbidity. Since the introduction of daunorubicin and cytosine arabinoside (Ara-C) in the mid 1970s, treatment has become increasingly intensive and evolved through clinical trials. This has included myeloablative therapy with bone marrow transplantation, and more recently the introduction of newer agents which target leukaemia cells semiselectively. The increase in our understanding of the intracellular pathways in both normal and leukaemic cells has led to the development of agents which can target specific enzymes involved in leukaemogenesis and cell proliferation. The role of immunotherapy is becoming more clearly understood. Advances in therapy have been mirrored by technologic improvements which allow risk stratification and the detection of minimal residual disease. Each new technology and treatment has to be evaluated by clinical trials before its benefit can be accepted.

The aim of treatment of AML is eradication of the malignant clone and restoration of normal haematopoiesis. Paediatric AML is a rare disease, and most children in the developed world are treated within clinical trials organized by their national group. National cooperative protocols differ in the choice of drugs, dosages, and scheduling, but generally share the same overall treatment strategy of remission induction and consolidation. The approach to CNS prophylaxis varies between cooperative groups, as does the use of maintenance therapy.

Fig. 13.2 Improvement in AML survival (Medical Research Council data).

Induction therapy

The aim of induction therapy is to induce a morphologic remission (<5 per cent blasts in bone marrow). However, it is acknowledged that induction treatment influences not only the remission rate but also the disease-free survival (DFS). Induction regimens vary but generally include an anthracycline and Ara-C. This combination chemotherapy is highly myelosuppressive and results in a prolonged period of pancytopenia. Complete remission (CR) rates of 80–90 per cent are now achievable for children with AML; 4–5 per cent of children have resistant disease and a similar number die from infection or haemorrhage before achieving remission.

The UK Medical Research Council (MRC) trials have serially tested the benefit of intensifying chemotherapy by increasing both the number of blocks of treatment and their intensity. Early trials identified DAT 3+10 (daunorubicin, Ara-C, and thioguanine) as superior to previous regimens for remission rates, time to achieve remission, and DFS. MRC AML 10 compared DAT with ADE (Ara-C, daunorubicin, and etoposide) and found no significant difference in the remission rate or DFS (CR rate 89 per cent versus 93 per cent, P = 0.3). Therefore the use of etoposide conferred no benefit in monoblastic leukaemia, although benefit had previously been reported. In its subsequent trials, MRC AML 12 and MRC AML 15, the MRC adopted ADE as the standard treatment for children against which other regimens should be tested. It is unlikely that any one regimen of this intensity will prove superior to another, but it is important that trials continue.

The Children's Cancer Group (CCG) in the USAtested the benefit of increasing the intensity of drug scheduling in induction chemotherapy and compared different modalities of post-induction treatment for outcome. The CCG-2891 (1989–1995) trial compared ‘standard-timing’ induction with a more intensive approach in a randomized manner. Patients in both arms of the study received the standard five-drug regimen DCTER (dexamethasone, Ara-C, 6-thioguanine, etoposide, and daunorubicin). Patients in the intensive arm received a second cycle of therapy after a 6-day rest interval regardless of bone marrow response. Patients randomized to standard timing received a second cycle of treatment depending on their bone marrow findings on day 14 and either preceded immediately to a second cycle or waited for count recovery. There was a significant improvement in outcome for those patients who received intensive-timing induction therapy, regardless of the post-remission therapy they received.

AML BFM-93 compared two different anthracyclines, daunorubicin and idarubicin, in induction. Patients who received idarubicin had significantly better blast cell clearance in the bone marrow on day 15 (17 per cent with >5 per cent blasts versus 31 per cent, P = 0.01), but this did not translate into an improved 5-year event-free survival (EFS) or DFS.

The evidence from these and other clinical trials clearly demonstrates the benefit of intensive induction chemotherapy. This has been achieved by choice of drugs, dosage, and scheduling. Induction chemotherapy influences not only the remission rate but also the relapse rate and the DFS. The use of increasingly intensive induction regimens carries the risk of increased treatment-related mortality which must be carefully monitored.

Consolidation therapy

Long- term remission can only be achieved with intensive consolidation chemotherapy. The optimal consolidation therapy is unknown. The benefit of regimens which employ high-dose Ara-C is agreed, but the role of allogeneic and autologous bone marrow transplant (BMT) remains controversial.

Patients entered into the MRC AML-10 trial (1988–1995) received four courses of very intensive chemotherapy. Children with a histocompatible sibling donor were eligible for an allogeneic BMT and those without were randomized to receive a fifth course of high-dose therapy with autologous bone marrow rescue or to stop treatment. Following such intensive chemotherapy, neither matched allografts nor auto-BMT improved overall survival, although both types of transplant were associated with a reduced relapse risk. The benefit from the reduced relapse rate was negated by a higher procedure-related mortality after allogeneic BMT and a better survival from relapse for previously non-autografted patients.

Analysis of data from this trial enabled three prognostic groups to be identified (good, standard, and poor) based on cytogenetics and response to the first block of treatment. Poorrisk patients (20 per cent) were those with >15 per cent blasts in the bone marrow performed after course 1 or adverse cytogenetic abnormalities [5, 7, del(5q), abn(3q), complex karyotypes]; good risk patients (28 per cent) were those with favourable cytogenetics [t(8,21), t(15,17), inv 16] irrespective of their marrow status after course 1, and all others were standard-risk patients. Five-year overall survival was 82 per cent, 60 per cent, and 22 per cent in the good-, standard-, and poor-risk groups, respectively. Because of these findings, allogeneic BMT was restricted to standard- and poor-risk patients in the subsequent MRC AML 12 trial. ABMT was replaced by an extra block of chemotherapy because of its treatment-related mortality and lack of overall survival benefit.

In the CCG-2891 study, following induction therapy, patients with a matched family donor (MFD) were allocated to allogeneic BMT. All other patients were randomized between autologous bone marrow transplantation or non-marrow ablative chemotherapy which included high-dose Ara-C. There was no significant difference in overall survival between those patients who received chemotherapy or an autologous bone marrow transplant (48 per cent versus 53 per cent, P = 0.31). Allogeneic MFD transplant was associated with an improved overall and DFS at 8 years (60 per cent and 55 per cent p 0.02) despite the increase in therapy related mortality, and this was largely due to a reduction in the relapse rate.

In BFM-93 the BFM group investigated the benefit of increasing intensity of consolidation therapy by adding an additional course of high-dose Ara-C with mitozantrone (HAM) for their high-risk patients. The addition of HAM in the high-risk group significantly improved the EFS (51 per cent versus 41 per cent, P= 0.01).

The optimal consolidation therapy for children in first CR of AML remains controversial. CCG, BFM, and MRC have shown no benefit for ABMT when compared with intensive chemotherapy. Results for allogeneic BMT in the first CR vary with CCG reporting benefit, and MRC and BFM reporting no benefit. This is in part due to the excellent results achieved with chemotherapy alone, and the benefit for allogeneic BMT maybe less apparent in studies which use very intensive chemotherapy. Whilst allogeneic BMT may reduce the risk of relapse, the results with chemotherapy alone are very encouraging and benefit has to be balanced against transplant-related mortality and long-term sequelae. In current MRC trials allogeneic BMT in the first CR is restricted to patients with unfavourable cytogenetics or a slow and suboptimal response to chemotherapy.

The optimal consolidation chemotherapy is unknown. MRC AML 12 showed no benefit for a fifth block of chemotherapy after four intensive courses, suggesting that there is a ceiling for benefit from chemotherapy.

CNS therapy

Intrathecal chemotherapy with cytosine methotrexate and hydrocortisone has been shown to be effective in preventing CNS relapse for patients with no CNS disease at presentation who are treated on protocols which include high-dose Ara-C. The CCG, Pediatric Oncology Group (POG), and MRC trials which did not include cranial irradiation report very low isolated and combined CNS relapse rates. The BFM group report benefit from CNS irradiation due to a reduction in bone marrow relapse risk. Cranial irradiation, with its associated neurotoxicity, is only justified for patients with CNS disease at presentation.

The role of maintenance therapy in AML

Several studies have investigated whether there is benefit for maintenance therapy in AML. The MRC and CCG-213 studies found no benefit for post-remission maintenance therapy. The LAME 89/91 study failed to show a significant difference in 5-year DFS between those who did and did not receive maintenance treatment, whilst the overall survival was better in the nonmaintenance group. This was because of a higher salvage rate following relapse, believed to be due to drug resistance and subsequent treatment failure in those exposed to maintenance therapy. However, the BFM group continue to use maintenance therapy although duration has been reduced from 2 to 1.5 years.

Acute promyelocytic leukaemia

A priority in the treatment of APL is the management of the associated coagulopathy, which can cause significant haemorrhagic problems. It is important to be aware of the potential risk of haemorrhage associated with procedures such as insertion of a Hickman line. Support with platelets, fresh frozen plasma, and cryoprecipitate is commonly required. Various approaches have been employed including maintaining the platelet count >50 109/liter and the fibrinogen >0.75–1.0 g/l until the coagulopathy resolves. There is no evidence to support the use of low-dose heparin. The coagulopathy improves with treatment of the leukaemia and the use of ATRA which induces differentiation of the malignant clone.

Studies of RT-PCR for PML-RARA transcripts have demonstrated prognostic value in the detection of minimal residual disease during clinical remission in APL. Serial negative PCR after chemotherapy is associated with prolonged remission, whereas patients who remain or revert to PCR positivity after consolidation are likely to relapse. Ongoing studies are assessing whether intervention prior to overt relapse will improve the clinical outcome.

Paediatric patients with relapsed APL usually achieve a second remission with further chemotherapy. The bone marrow is the usual site of relapse, but isolated CNS relapses have been reported. Patients can develop resistance to ATRA therapy, and some of these patients may respond to arsenic trioxide which has been reported to give CR rates of 80–100 per cent. Adult studies suggest that if a second remission is achieved, matched sibling allogeneic BMTor autologous transplant should be considered.

Relapsed and refractory AML

Despite improvements in therapy, about 30 per cent of children with AML still relapse. Children treated on MRC AML 10 who relapsed within 1 year of achieving their first CR had a very poor outlook (overall survival 15 per cent of those treated). Patients with a first CR >1 year fared better [CR 92 per cent; 3-year EFS 60 per cent]. High dose Ara-C alone, or in combination with other agents (anthracycline ± fludarabine) are commonly used to achieve a second remission. The majority of these patients proceed to allogeneic BMT with family or alternative donors. Allogeneic BMT may improve outcome for patients who relapse late. Patients who relapse after allogeneic BMT may benefit from augmentation of the graft versus leukaemia (GVL) effect. This may be achieved by stopping or reducing immunosuppression and by the use of donor lymphocyte infusions, although the latter remains unconfirmed.

Approximately 5 per cent of patients still fail to achieve remission with standard induction chemotherapy. Possible treatment strategies for this group of patients include alternative intensive chemotherapy and the use of novel treatments such as monoclonal antibody therapy. If a remission is achieved, allogeneic transplant from a matched family or unrelated donor is generally recommended, although no definite advantage of this approach has been demonstrated. Failure to achieve remission with salvage therapy is associated with a very poor prognosis, and transplantation in this setting carries a very high procedure-related mortality and little hope of cure.

Infant AML

Infants <12 months of age usually have acute monoblastic or myelomonoblastic AML characterized by hyperleucocytosis and extramedullary involvement. Many patients have 11q23 translocations involving the MLL gene. Infant AML has a superior outcome to infant ALL and ~90 per cent can achieve CR with intensive chemotherapy. Studies from the USA and Japan have reported a 3-year EFS of 61–72 per cent, suggesting that many infants with AML can be cured with intensive chemotherapy alone. Matched related donor transplant may be associated with a slightly better EFS but the risk of late effects must be considered.

Minimal residual disease

Relapse remains the main cause of treatment failure in AML and studies suggest that monitoring minimal residual disease (MRD) may be of benefit in identifying those patients at high risk of relapse. Complete remission is classically defined as the presence of <5 per cent blasts in the bone marrow as assessed by traditional light microscopy. At diagnosis, leukaemic tumour burden may be 1012leukaemic cells, and even at the time of morphologic remission there may still be as many as 1010 residual leukaemic cells. Whilst risk stratified treatment based on MRD is common practice in ALL, MRD is less commonly measured in AML and treatment is uniform regardless of the tumour load.

There are currently four main methods for the detection of minimal residual disease (Table 13.3). These are conventional cytogenetics, fluorescence in situ hybridization (FISH), multiparameter flow cytometry, and nucleic-acid-based amplification (PCR).

Table 13.3. Methodologies used in MRD detection


percentage of
suitable cases






5 × 10-2





1-5 × 10-2

Specific marker
Monitors multiple events




1 × 10-2

Specific marker
Can be used in interphase

Relatively insensitive

flow cytometry


10-2 - 10-4

Relatively sensitive

False positive
Low frequency of normal
cell with leukaemic
Phenotype switch


Specific fusion
Other molecular

10-4 - 10-6

Highly sensitive
Allows standardization

False positive due to
Reduced sensitivity due
to degraded RNA
Low mRNA expression
Inefficient reverse

Reproduced from J.Yin and K. Tobal (1999) Br J Haematol 106, 578–90.

Emerging therapies

It is likely that there is a ceiling to the benefit that can be achieved from intensive chemotherapy. In addition, the role of BMT may be limited. New agents and technologies are desperately needed. Alternative approaches include monoclonal antibodies such as gemtuzimab (anti CD33 antibody) which targets an epitope highly expressed on many myeloid blasts. Tyrosine kinase inhibitors, which are directed at the fusion products of abnormal genes and drugs and modulate proliferation and apoptosis, are under assessment. Microarray technology and gene profiling will improve understanding of leukaemogenesis and its effect on intracellular pathways. The benefits of these developments require careful assessment in well-designed clinical trials.


There has been considerable improvement in the survival in AML over the past 30 years, largely because of the high rate of enrollment in clinical trials. With intensive induction chemotherapy 90 per cent of children achieve remission, and with intensive consolidation >60 per cent are long-term survivors. Significant treatment-related mortality, despite improvements in supportive care, suggests that there may be a limit to therapy intensification. Improved understanding of the molecular pathways in AML will allow the development of targeted cytotoxic therapy. It is hoped that combining this with better characterization of prognostic groups will allow optimization of therapy in the future.

Key references and suggested reading

Arceci RJ (2002). Progress and controversies in the treatment of pediatric acute myelogenous leukemia. Curr Opin Hematol 9, 353–60.

Bain BJ, Barnett D, Linch D, et al. (2002). Revised guidelines on immunophenotyping in acute leukaemias and chronic lymphoproliferative disorders. Clin Lab Haematol 24, 1–13.

Brunning RD, Matutes E, Harris NL, et al. (2001). Acute myeloid leukaemia. In: Jaffe ES, Harris NL, Stein H, Vardiman JW (eds) Pathology and Genetics of Tumours of Haemopoetic and Lymphoid TissuesWorld Health Organization Classification of Tumours. Lyon: IARC Press.

Clark JJ, Smith FO, Arceci RJ (2003). Update in childhood acute myeloid leukemia: recent developments in the molecular basis of disease and novel therapies. Curr Opin Hematol 10, 31–9.

Creutzig U, Ritter J, Zimmerman D, et al. (2001). Improved treatment results in high-risk pediatric acute leukemia patients after intensification with high-dose cytarabine and mitoxantrone: Results of Study Acute Myeloid Leukaemia–Berlin–Frankfurt–Munster 93. J Clin Oncol 19, 2705–13.

Gregory J, Feusner J (2003). Acute promyelocytic leukemia in children. Best Pract Res Clin Haematol 16, 483–94.

Ishii E, Kawasaki H, Isoyama K, et al. (2003). Recent advances in the treatment of infant acute myeloid leukemia. Leuk Lymphoma 44, 741–8.

Smith MA, Ries LAG, Gurney JG, et al. (1999). Leukemia. In: Ries LAG, Smith MA, Gurney JG, et al. (eds) Cancer Incidence and Survival Among Children and Adolescents: United States SEER Program 1975–1995. NIH Publication 99–4649. Bethesda, MD: National Cancer Institute, 17–34.

Swirsky DM, Richards SJ (2001). Laboratory diagnosis of acute myeloid leukaemia. Best Pract Res Clin Haematol 14, 1–17.

Stevens RF, Hann IM, Wheatley K, Gray RG (1998). Marked improvements in outcome with chemotherapy alone in paediatric acute myeloid leukaemia: Results of the United Kingdom Medical Research Council's 10th AML trial. Br J Haematol 101, 130–40.

Webb DKH, Wheatley K, Harrison G, et al. (1999). Outcome for children with relapsed acute myeloid leukaemia following initial therapy in the Medical Research Council (MRC) AML 10 trial. Leukemia13, 25–31.

Wheatley K, Creutzig U, Chen AR, et al. (2002). Current controversies. Which patients with acute myeloid leukaemia should receive a bone marrow transplantation? Br J Haematol 118, 365–84.

Wheatley K, Burnett AK, Goldstone AH, et al. (1999). A simple, robust, validated and highly predictive index for the determination of risk-directed therapy in acute myeloid leukaemia derived from the MRC AML 10 trial. United Kingdom Medical Research Council's Adult and Childhood Leukaemia Working Parties. Br J Haematol 107, 69–79.

Woods WG, Neudorf S, Gold S, et al. (2001). A comparison of allogeneic bone marrow transplantation, autologous bone marrow transplantation, and aggressive chemotherapy in children with acute myeloid leukemia in remission: a report from the Children's Cancer Group. Blood 97(1) : 56–52.

Yin J, Tobal K. (1999). Detection of minimal residual disease in acute myeloid leukaemia: methodologies, clinical and biological significance. Br J Haematol 106, 578–590.

If you find an error or have any questions, please email us at Thank you!