|Home | About | Journals | Submit | Contact Us | Français|
We review recent advances in the biologic understanding and treatment of childhood acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML), identify therapeutically challenging subgroups, and suggest future directions of research.
A review of English literature on childhood acute leukemias from the past 5 years was performed.
Contemporary treatments have resulted in 5-year event-free survival rates of approximately 80% for childhood ALL and almost 60% for pediatric AML. The advent of high-resolution genome-wide analyses has provided new insights into leukemogenesis and identified many novel subtypes of leukemia. Virtually all ALL and the vast majority of AML cases can be classified according to specific genetic abnormalities. Cooperative mutations involved in cell differentiation, cell cycle regulation, tumor suppression, drug responsiveness, and apoptosis have also been identified in many cases. The development of new formulations of existing drugs, molecularly targeted therapy, and immunotherapies promises to further advance the cure rates and improve quality of life of patients.
The application of new high-throughput sequencing techniques to define the complete DNA sequence of leukemia and host normal cells and the development of new agents targeted to leukemogenic pathways promise to further improve outcome in the coming decade.
The treatment outcome for children with acute myeloid leukemia (AML) and especially acute lymphoblastic leukemia (ALL) has improved substantially with the use of risk-directed treatment and improved supportive care. The 5-year event-free survival rates for ALL now range between 76% and 86% in children receiving protocol-based therapy in the developed countries (Table 1),1–14and those for AML range between 49% and 63% in some of the more successful clinical trials (Table 2).15–30 The improved treatment has diminished or eliminated the impact of many conventional prognostic factors in ALL, such as male sex and black race.6,10 Thus current ALL trials have focused on improving not only the outcome of a few subtypes that remain refractory to treatment (eg, infant ALL with MLL rearrangement, hypodiploid ALL, and poor early responders), but also the quality of life of the patients. By contrast, with the exception of core-binding factor (CBF) leukemias [t(8;21)(AML1-ETO) and inv(16)(CBFβ-MYH11)] and AML in Down syndrome, most of the AML cases continue to pose a therapeutic challenge.
Childhood acute leukemias have long been the best characterized malignancies from a genetic viewpoint. Despite remarkable progress in cataloging the molecular lesions, our understanding of how such changes cooperate to produce overt leukemia or to induce drug resistance is still rudimentary. Recent genome-wide studies have begun to enlighten our understanding of leukemogenesis and prognosis and, in some instances, to stimulate the development of target therapy. Because of the space constraints, we review here only some of the most recent advances in the biology and treatment of childhood leukemias.
Standard genetic analyses can detect primary genetic abnormalities in more than 75% of the ALL cases, but they cannot identify the full repertoire of the genetic alterations.31 The advent of high-resolution genome-wide analyses of gene expression, DNA copy number alterations (CNA) and loss of heterozygosity, epigenetic changes, and whole-genome sequencing have led to the detection of many novel genetic abnormalities; to date, virtually all patients with ALL can be classified according to specific genetic abnormality (Fig 1A). These studies also provided new insights into the complex interactions of multiple genetic alterations in leukemogenesis and response to therapy.34
Primary somatic genetic abnormalities have important prognostic and therapeutic implications and play a critical role in leukemogenesis (Table 3).10,31,34–50 However, experimental models have established that cooperative mutations are necessary to induce leukemia and contribute to the development of drug resistance. Although high-throughput analyses of global gene expression have identified distinct subgroups and may in the future be used to stratify patients, these profiles do not distinguish pathways that are “drivers” of leukemogenesis from “passengers.” Using single-nucleotide polymorphism (SNP) array, Mullighan et al34 identified an average of six CNAs per case of childhood ALL. In general, deletions outnumbered amplifications by a ratio of 2:1. The lesions target genes regulating lymphoid differentiation, tumor suppression, cell cycle, apoptosis, signaling, microRNAs, and drug responsiveness. There were substantial differences in the frequency of CNAs among various subtypes. MLL-rearranged cases had less than one CNA per case, suggesting that MLL is a potent oncogene that requires very few cooperating lesions to induce leukemia transformation, whereas ETV6-RUNX1 (formerly known as TEL-AML1) and BCR-ABL1 leukemias demonstrated more than six lesions per case.34 These findings are compatible with the known behavior of these leukemias. MLL-rearranged leukemias often present during infancy and have a concordance rate close to 100% in identical twins, indicating in utero development and transplacental metastasis from one fetus to the other.31 By contrast, ETV6-RUNX1 leukemias present after infancy and have a concordance rate of only 10% in identical twins, and this gene fusion can be found in as many as 1% of normal newborn babies, a frequency 100 times higher than the prevalence of this subtype of leukemia, suggesting that additional postnatal mutations are necessary for malignant transformation.31
In Philadelphia chromosome–positive ALL with constitutively active BCR-ABL1 tyrosine kinase, IKZF1 (encoding the lymphoid transcription factor IKAROS) is deleted in approximately 80% of the cases.51 Interestingly, there is a high-risk subgroup of BCR-ABL1–negative ALL that is characterized by IKZF1 deletion and has a genetic profile similar to that of cases with BCR-ABL1 fusion.46,47 In search of activated tyrosine kinase signaling in this leukemic subtype, Mullighan et al48 identified activating mutations in the Janus kinases (JAK1, JAK2, and JAK3) in approximately 10% of high-risk BCR-ABL1–negative cases. The presence of JAK mutations was associated with alteration of IKZF1 and deletion of CDKN2A/B. Recent studies showed CRLF2 over-expression (predominantly resulting from P2RY8-CRLF2 fusion or IGH@-CRLF2 rearrangement) in 6% to 7% of B-cell precursor ALL cases and strikingly in 50% to 60% of patients with Down syndrome–associated ALL, all of whom lacked recurring translocations commonly associated with non–Down syndrome ALL.40–42,52 CRLF2 alteration was associated with activating mutations of JAK1 or JAK2; these two genetic lesions together resulted in constitutive Jak-Stat activation and the growth of cytokine-dependent mouse B-progenitor cell lines in the absence of exogenous cytokine, indicating that they are cooperative mutations in leukemogenesis.40 Importantly, CRLF2 alteration was associated with Hispanic/Latino ethnicity and a poor treatment outcome.40,41
On the basis of gene expression profiling, T-cell ALL cases can be classified into several distinct genetic subgroups that correspond to specific T-cell development stages: HOX11L2, LYL1 plus LMO2, TAL1 plus LMO1 or LMO2, HOX11, and MLL-ENL (Table 3).43 Although HOX11L2 generally confers a poor outcome, HOX11 and MLL-ENL are associated with a favorable outcome.43,53 Using SNP and other genome-wide platforms, many novel genomic alterations have recently been identified, including focal deletions leading to dysregulated expression of TAL1 and LMO2, deletion and mutation of PTEN, mutations of NOTCH1 and FBXW7, deletions of RB1, duplications of MYB, deletions of RB1, and fusion of SET or ABL1 to NUP214.34 Hence T-cell ALL is also a heterogeneous disease. Thus far, mutations of NOTCH1 and FBXW7 (observed in 50% of T-ALL) have generally been associated with a favorable prognosis, and NUP214-ABL1 fusion has been associated with responsiveness to tyrosine kinase inhibition.43,44,49
To explore the genetic basis of relapse, genome-wide studies were conducted using matched diagnosis and relapse samples from the same patients.54,55 Although 90% of the cases exhibited differential CNAs (gaining or losing genetic lesions) from diagnosis to relapse, most relapse samples are clonally related to diagnosis samples, and backtracking studies showed that the relapse clones were often present as minor populations at diagnosis, suggesting that they were selected during treatment. Notably, many of the genetic alterations that emerge in the predominant clone at relapse involve genes that have been implicated in treatment resistance (eg, CDKN2A/B, IKZF1).54,55 Interestingly, some cases had focal deletion of MSH6, reduced expression of which was associated with resistance to mercaptopurine and prednisone, providing another plausible mechanism of drug resistance at relapse.54 Parallel gene expression studies, which have identified a proliferative gene signature that emerges at relapse and consistent upregulation of genes such as survivin, provide attractive targets for novel therapeutic intervention.56
Although candidate gene approaches have implicated inherited polymorphisms of several genes in leukemogenesis, the findings have not been consistent. Two recent genome-wide studies on patients of European ancestry failed to confirm these previously reported gene associations but independently identified germline polymorphisms of the IKZF1 gene and ARID5B gene to be associated with an increased risk of childhood ALL.57,58 The risk alleles of ARID5B, a gene that belongs to a family of transcription factors important in embryonic development, cell type–specific gene expression, and cell growth regulation, were specifically enriched in patients with hyperdiploid ALL and were also associated with greater methotrexate polyglutamate accumulation.57,58 Thus the same genetic variation of ARID5B that predisposes to the development of hyperdiploid ALL may also underlie the superior response of this subtype of ALL to chemotherapy. A subsequent study was performed in patients of African ancestry, showing that ARID5B germline polymorphisms were also associated with the risk of developing hyperdiploid ALL in black patients.59 The lower frequency of this risk allele in the control populations of African ancestry compared with that of European ancestry might also partly explain the lower incidence of hyperdiploid B-cell precursor ALL in the black population.
Despite the extremely heterogeneous nature of AML, the various subtypes seem to share some common pathways leading to leukemogenesis, and the hierarchical nature of the disease is generally well established.60 Several lines of evidence have led to a model that AML arises from the cooperation between two classes of genetic alterations that regulate self-renewal and differentiation.61 For example, mutations or epigenetic alterations involving CBF or MLL genes, termed type II alterations, play key roles in modifying the ability of precursor cells to differentiate, but are usually insufficient by themselves to generate AML. In support of this concept is the finding that the AML1-ETO fusion transcript resulting from the t(8;21) translocation can be detected in neonatal blood from teenagers with AML characterized by this translocation.62 Such fusion transcripts have also been identified to persist in the bone marrow of some patients in long-term morphologic remission.63 When a cooperative mutation or type I mutation (such as FLT/ITD or c-KIT point mutations) occurs, a proliferative signal is provided, leading to overt AML. That type II mutations such as AML-ETO can be detected far more frequently in neonatal samples than the incidence of leukemia further supports the concept of cooperating mutations in leukemogenesis.64 Genome-wide studies have identified additional genetic or epigenetic changes that lead to type I and II mutations. Currently, more than 90% of pediatric AML cases are identified to have at least one known genomic alteration (Fig 1B).
Similar to ALL, many of the chromosomal abnormalities in AML result in alteration of the function of transcription factors critical for normal hematopoiesis and have diagnostic and therapeutic implications (as discussed in Risk-Adapted Treatment of AML). The t(8;21)(q22;q22) and the inv(16)(p13q22) involving CBFα and β subunits respectively account for about a quarter of all cases. These translocations impair CBF transcriptional activation potential. Less common translocations include the t(15;17), 11q23/MLL rearrangements and the t(1;22). The t(15;17) generated PML-RARA fusion protein is resistant to physiologic concentrations of retinoic acid but sensitive to pharmacologic levels of all-trans-retinoic acid (ATRA), which destabilize the repressor complex, allowing for the expression of genes permissive for differentiation and making this subtype the first successfully treated acute leukemia by molecular targeting. The incidences of chromosomal rearrangements involving the MLL gene are age-dependent, being highest in children younger than 2 years and subsequently decreasing to less than 5% in adults.65 The t(1;22) translocation results in the fusion of the OTT gene on chromosome 1 to the Megakaryocytic Acute Leukemia gene on chromosome 22. The translocation is associated closely if not exclusively with acute megakaryoblastic leukemia (AMKL) and has been detected in up to one third of such childhood cases. Monosomy 7, monosomy 5, or 5q deletions are present in only approximately 2% to 4% of cases compared with more than 10% in adults.
Recent genome-wide microarray studies and direct-sequencing approaches have revealed important sub-karyotypic abnormalities. One important finding has been the high percentage of acquired uniparental disomy, resulting in homozygosity of chromosomal regions that occurs after the acquisition of a mutation in one allele.66,67 Examples of this have included FLT3-ITD and CEBPA mutations.68,69
Biallelic mutations in CEBPA, a key leucine zipper containing transcription factor regulating differentiation of several cell types, including myeloid precursors, have been identified in approximately 4% of the cases.70 The end result of such biallelic mutations is often that of a null phenotype or loss of function.71
WT1, originally described as a key oncogene in Wilms tumor, functions in non-nephrogenic tissues, including hematopoietic stem cells. Increased expression of WT1 has been used as a surrogate marker for minimal residual disease (MRD) in AML. Inactivating WT1 mutations were reported in 8% to 12% of patients with AML, and their prognostic significance remains uncertain and is likely treatment-dependent.
GATA1 is critical in regulating the differentiation of hematopoiesis, particularly for the erythroid and megakaryocyte lineages. Mutations leading to truncated forms of GATA1 are primarily observed in the AMKL or the megakaryoblastic transient myeloproliferative disease that occur in children with Down syndrome.72 Although not sufficient by themselves in causing leukemia, these GATA1 mutations have been established as a first genetic hit in these disorders and have even been detected prenatally.
Activating mutations of several cytokine receptors have been shown to result in altered signal transduction and increased leukemia cell proliferation, survival, and chemotherapeutic resistance. FLT3-ITD, which involves a duplication of the internal juxtamembrane domain leading to constitutive receptor activation, is rare in infants but is present in 5% to 10% of 5- to 10-year-old patients, 20% of young adult patients, and more than 35% of patients older than 55 years of age with AML.73 An increased FLT3-ITD mutant to wild-type allelic ratio portends a poor prognosis.74,75 Children's Oncology Group (COG) trials are using a mutant to wild-type allelic ratio of ≥ 0.4 as a criterion for allogeneic hematopoietic stem-cell transplantation (HSCT) in first remission. Point mutations in the activating loop of the receptor (FLT3/ALM) also result in constitutively activated FLT3 but do not confer a poor prognosis.73 This finding may be in part due to differences in signaling pathway activation by the different forms of FLT3. FLT3/ALM occurs in only 6% to 8% of children with AML and is mutually exclusive of FLT3-ITD.73
Activating mutations of the c-KIT receptor also result in cytokine-independent proliferation, survival, and drug resistance. Such mutations preferentially activate STAT3 signaling and appear to be clustered with CBF abnormalities, occurring in up to 25% of CBF AML cases but in only ≤ 5% of overall childhood AML.74–77 In contrast to initial reports,74,75 recent studies failed to demonstrate c-KIT mutations to have adverse prognostic impact.78 Mutations involving FMS and PDGFR, class III type of tyrosine kinases, have been reported in adults but not in children with hematologic malignancies.
N-RAS has been reported to be mutated in approximately 10% of children and up to 30% of adults with AML.79 These activating mutations most frequently involve single-base changes of codons 12, 13, and 61, resulting in inhibition of RAS-GTPase. No clinical significance has been associated with mutations in RAS in pediatric AML, and RAS-directed therapies have not been successful thus far.80 However, the recent application of synthetic lethal screening has started to reveal some alternative pathways that could be therapeutically targeted.81
NPM1 mutations are associated with a favorable prognosis. They appear more frequently in AML with normal karyotypes and up to 20% of childhood and 50% of adult cases.82,83 NPM1 mutations seem to be present in leukemia-initiating cell populations and are usually maintained at relapse, suggesting they may represent a primary oncogenic event.84,85
The initial whole-genome sequencing of an adult with M1 AML and normal karyotype disclosed 10 somatic mutations and two leukemia-specific SNPs.86 Two of the mutations involved FLT3 and NPM1, whereas the other eight genes involved nonsense and missense mutations that had not previously been described in AML.86 These novel mutations were not identified in 200 additional AML samples, raising the question of whether they were merely passenger mutations. Analysis of the second case of cytogenetically normal AML revealed 12 mutations in annotated protein-coding genes or regulatory RNAs and 52 mutations in noncoding regions of the genome.87 Four genes, NPM1, NRAS, IDH1, and a conserved region on chromosome 10, were found in at least one additional sample of 180 adult AMLs. Although the IDH1 mutation was observed in 16% of 80 cytogenetically normal adult AML samples,87 it has not been found in childhood AML.88 Independent analyses of IDH1 and IDH2 mutations showed that they were more frequently associated with AML having a normal karyotype, but had no prognostic or therapeutic impact.89–92
Although RNA and microRNA expression signatures can accurately discriminate cases with specific chromosomal translocations and identify novel subsets of AML,93 two independent pediatric studies failed to identify an expression pattern with consistent prognostic significance.94,95 Some of the differences observed in gene expression studies are also beginning to be understood in terms of epigenetic regulation of chromatin function. The finding of a similar expression profile between an AML subset with the methylation and silencing of the CEBPA promoter and another characterized by CEBPA mutations illustrates the critical role that epigenetic changes play in AML.94 Of further interest, the distinctive myeloid/T-lymphoid characteristics of this subtype have been linked to the finding that decreased CEBPA expression leads to increased expression of T-lineage genes in hematopoietic precursors.94,95 On a genome-wide scale, altered CpG methylation profiles have been associated with distinct subsets of adult AML.96 A 15-gene methylation classifier has been reported to have prognostic significance.97
Several inherited syndromes associated with an increased incidence of AML have been instrumental in demonstrating the importance of key molecular pathways in the development of somatically acquired AML.98,99 Transient myeloproliferative disease and AMKL in patients with Down syndrome are characterized by specific mutations in the GATA1 gene, which have been documented even during prenatal development.72 Fanconi anemia (mutations in DNA repair genes), dyskeratosis congenita (X-linked mutations in dyskerin or autosomal-recessive forms with mutations in genes involved in telomere maintenance and RNA processing), Schwachman-Diamond's syndrome (SBDS gene involved in ribosome processing), and Kostmann's syndrome (defects in neutrophil elastase ELA2) are all associated with an increased incidence of AML. Mutations in the neurofibromin gene, a RAS-inactivating GTPase, are the cause of neurofibromatosis type I, along with its increased incidence of both juvenile myelomonocytic leukemia and AML. Similarly, mutations of the PTPN11 gene, which encodes the SHP-2 tyrosine phosphatase, cause Noonan's syndrome, which is also associated with the development of juvenile myelomonocytic leukemia and, less commonly, AML. This group of mutations are all linked to the activation of the RAS pathway. Familial platelet disorder with a propensity to develop AML, as well as congenital amegakaryocytic thrombocytopenia, are associated with an increased incidence of AML and caused by mutations in CFFA2 and the thrombopoietin receptor, c-mpl, respectively. Inherited mutations in the CEBPA have also been linked to familial AML.
The Philadelphia chromosome–positive ALL had historically been associated with a dismal outcome, even with allogeneic HSCT, especially in older patients who presented with a high leukocyte count or had a slow early response to initial therapy. In a recent COG study, the addition of continuous exposure of imatinib into an intensive chemotherapy regimen has yielded a 3-year event-free survival of 80%, more than twice that of the historical controls.37 Patients who were treated with intensive chemotherapy plus imatinib fared as least as well as those historical controls who underwent matched-related or matched-unrelated transplantation. Continuous exposure of imatinib seemed to abrogate the prognostic impact of age, leukocyte count, high level of MRD at the end of induction, and even induction failure. Additional follow-up is needed to determine whether the treatment improved the cure rate rather than merely prolonging the disease-free survival. Meanwhile, many pediatric oncologists have reserved transplantation for therapy after relapse in children with Philadelphia chromosome–positive ALL. Future studies will need to address whether the aggressive backbone therapy that was used in the COG study can be safely omitted or reduced in combination with a tyrosine kinase inhibitor and whether the new generation of tyrosine kinase inhibitors (eg, dasatinib, nilotinib) can further improve outcome. Because of its dual targeting of ABL and SRC, more potent suppression of the BCR-ABL1 signaling, and efficacy to most imatinib-resistant BCR-ABL1 mutants (except for T3151), as well as good tolerability in adult clinical trials,101 dasatinib is now being tested in children.
With the use of intensive treatment including asparaginase and dexamethasone, T-cell cases fared as well as B-cell precursor cases in some studies.6,10 Although high-dose methotrexate at 5 g/m2 has been suggested to improve outcome in T-cell ALL, methotrexate was given parenterally at only 30 mg/m2 in the DFCI 95-01 study, which featured intensive asparaginase and doxorubicin for consolidation therapy and has perhaps the best treatment result for this subtype of ALL, with a 5-year event-free survival rate of 85%.102 In the current COG study for T-cell ALL, the relative efficacy and safety of high-dose methotrexate at 5 g/m2 with leucovorin rescue are being compared with escalating medium doses of methotrexate without leucovorin rescue plus PEG-asparaginase. In the same study, patients are also randomly assigned to receive or not receive nelarabine (a prodrug of 9-β-D-arabinofuranosylguanine), which has considerable antileukemic effects for T-cell ALL in phase II studies.103
Early T-cell precursor ALL, a subset of T-cell ALL characterized by a gene expression profile similar to that of normal early thymic precursor cells with multilineage differentiation potential and a distinct immunophenotype (CD1a-negative, CD8-negative, CD5-weak, and the expression of stem-cell or myeloid markers), was recently identified.45 These cases have extremely poor outcome, despite the use of transplantation. Because of the significantly improved outcome of patients with T-cell ALL who were treated with high-dose dexamethasone (10 mg/m2 per day) during remission induction in the recently completed Associazione Italiana Ematologia ed Oncologia Pediatrica Berlin-Frankfurt-Muenster AIEOP-BFM-2000 study, albeit with increased toxicity,104 this dose of dexamethasone is being tested during remission induction for this subset of patients in the current St Jude Total Therapy Study XVI.
Even with intensive therapy including high doses of cytarabine and methotrexate, treatment outcome for infants with ALL remains poor, and for those with MLL gene rearrangement, the event-free survival rates range between 30% and 40% only.100,105,106 Although several small studies have suggested that allogeneic HSCT improves outcome, results from large group studies failed to show any benefit of this treatment modality.105,106 The Interfant-06 Study Group is conducting a randomized trial to test whether the addition of two early intensification courses typically used for AML might improve outcomes for infants with medium-risk or high-risk ALL, including those with MLL rearrangement. The role of allogeneic HSCT in first remission is being investigated in very high-risk cases, as defined by MLL rearrangement, age less than 6 months, and WBCs more than 300 ×109/L. The current COG trial randomly assigns MLL-rearranged infant cases to receive the FLT3 inhibitor lestaurtinib, which has been shown to have schedule-dependent synergistic effects with chemotherapy for MLL-rearranged leukemia in preclinical studies.107 Recent findings suggested that aberrant DNA methylation occurs in the majority of infant ALL case with MLL rearrangement,108,109 raising the possibility of using DNA methyltransferase inhibitors in these patients.
Older adolescents 15 to 21 years of age have had an inferior outcome as compared with that of children and younger adolescents with ALL, partly due to an increased incidence of Philadelphia chromosome– positive and T-cell ALL, lower incidence of ETV6–RUNX1 fusion and hyperdiploidy, and poorer compliance.31,110–112 The relatively poor outcome of older adolescents 16 to 21 years of age with ALL has prompted comparisons of outcome of patients in this age group treated in pediatric versus adult clinical trials in North America and Western Europe. Consistently, pediatric trials have yielded significantly better outcome than adult trials.112–114 This finding, in all likelihood, reflects differences in the more intensive use of nonmyelosuppressive agents such as dexamethasone, vincristine, and asparaginase; the incorporation of high-dose methotrexate; and early and frequent administration of intrathecal therapy in pediatric regimens.
The importance of these treatment components is supported by several other pediatric trials. In the Children's Cancer Group 1961 trial, which included 262 adolescents 16 to 21 years of age treated from 1996 to 2002,112 the 5-year event-free survival rate was 71.5%. This randomized trial demonstrated that augmented rather than standard postinduction intensification therapy with additional doses of vincristine and PEG-asparaginase in lieu of mercaptopurine during interim maintenance therapy and increased doses of intravenous methotrexate without leucovorin rescue improved outcome for patients with a rapid early response. The Dana-Farber Cancer Institute (DFCI) protocols (1991 to 2000)6 and St Jude Total Study XV (2000 to 2007),115 which yielded excellent results for adolescents, likewise featured early intensification therapy with vincristine and asparaginase. Although some contemporary adult ALL trials have recommended allogeneic transplantation as the best treatment option for good-risk adult ALL patients in first remission,116 the excellent results obtained in these pediatric trials do not support the routine use of transplantation in older adolescents with ALL.
There are three other important observations from the Children's Cancer Group 1961 study.112 First, there was no statistical difference in treatment outcome between adolescents with a rapid response who were randomly assigned to receive either one or two courses of postinduction intensification therapy. Second, there was an increased risk of toxic deaths among older adolescents, with 18% of the first events being nonrelapse deaths that occurred during induction or in first remission. Third, there was an increased risk of osteonecrosis. Thus it will be of interest to test whether prednisone pulses, which are commonly given during maintenance therapy, can be omitted in adolescents and young adults with ALL and rapid early response.
The excellent results recently achieved in older adolescents with ALL have prompted many centers and cooperative groups to start treating young adults with pediatric-like regimens. A recent report showed 6-year event-free survival rates of 60% for the 35 adolescents 15 to 18 years old and 63% for the 46 young adults 19 to 30 years old.117 The DFCI Consortium attained a 2-year rate of 72.5% among 75 adults 18 to 50 years of age.118 Investigators at Princess Margaret Hospital in Toronto have also adapted the DFCI treatment regimen and achieved a 3-year relapse-free survival rate of 77% among 64 adults with BCR-ABL–negative ALL who did not undergo transplantation.119 To this end, a US adult intergroup study is testing a pediatric-like regimen in adolescents and young adults up to 39 years of age.
Hypodiploidy (< 44 chromosomes), t(17;19)(q22;p13.3) [TCF3-HLF], remission induction failure, and the presence of MRD more than 1% at the end of remission induction are also associated with dismal outcome and collectively occur in approximately 5% of children with ALL.31 Although allogeneic transplantation is frequently used to treat these patients, there is no strong evidence to document the efficacy of this approach. Among many novel therapeutics under investigation (Table 4),101 killer-cell immunoglobulin-like receptor–mismatch natural cell therapy120,121 and immunotherapy with a T-cell–engaging CD19-/CD3-bispecific antibody construct (blinatumomab) seem to be particularly promising.122
Two recent studies showed that with effective chemotherapy, prophylactic cranial irradiation can be safely omitted altogether in the treatment of childhood ALL,10,123 showing isolated CNS relapse rates of 2.7% and 2.6%. The complete omission of prophylactic cranial irradiation in these studies allowed the clear identification of risk factors for CNS relapse, which were any CNS involvement at diagnosis, t(1;19)[TCF3-PBX1], and T-cell ALL.10,123 In the current St Jude Total Therapy Study XVI, triple intrathecal therapy is further intensified for these patients with twice weekly administration in the first 2 weeks of remission induction.
Similar to ALL, risk-adapted therapy has also become an important approach in childhood AML. Many traditional prognostic factors have been replaced by cytogenetic and molecular features, as well as flow cytometric assessment of MRD (Table 5).
With the introduction of intensively dosed and/or timed chemotherapy regimens, AML characterized by alterations of core-binding and transcription factors have emerged as a favorable prognostic group. Both t(8;21) and inv(16) AML have an approximately 80% overall survival rate when treated with three to four courses of intensive chemotherapy.15–28 Although patients with t(8;21) AML may have a higher relapse rate than those with inv(16), they are still able to be cured with allogeneic transplantation after attaining a second remission.127
Similar to adults, children with acute promyelocytic leukemia (APL) have an excellent overall survival in the 75% to 85% range, with some reports as high as 90%.124,128 This leukemia is particularly sensitive to ATRA and arsenic trioxide. Children with APL also benefit from maintenance therapy with ATRA and antimetabolites.129,130 Allogeneic transplantation is not recommended for these children in first remission. Survival after relapse is approximately 70% after reinduction, usually with arsenic trioxide, and then autologous or allogeneic transplantation.131,132
MLL-rearranged AML represents a diverse group with quite variable outcomes.65 In an international study, survival was 100%, 63%, 27%, and 22% for patients with the t(1;11), the t(9;11), the t(4;11), and the t(6;11), respectively.65 In part because of the variability in outcomes, small numbers of patients, and lack of definitive data demonstrating the superiority of HSCT, most cooperative trials do not include patients with MLL-rearranged AML in the high-risk group requiring allogeneic HSCT in first remission.133 Once associated with a poor prognosis,134 AMKL with the t(1;22) has improved outcome with contemporary treatment, and HSCT may not be indicated in these patients.26,135
FLT3-ITD mutations, particularly when associated with a high mutant to wild-type allelic ratio, are associated with an overall survival rate of usually less than 30%.136 Although controversial, there are some indications that transplantation might improve outcome of children with this type of AML.137 Some clinical trials are testing the efficacy of various FLT3 inhibitors with different degrees of specificity plus allogeneic transplantation in these children. In the next COG trial for newly diagnosed AML, patients with a high FLT3-ITD to wild-type allelic ratio will be offered an allogeneic HSCT in first remission. Of note, patients with point mutations of FLT3 (eg, FLT3/ALM) do not have poor outcome and should be treated with chemotherapy only.
The c-KIT mutations, most commonly observed in CBF AML, has been reported to portend a poor prognosis in some studies, but not in a recent COG study.138 Similarly, RAS mutations have not been definitively associated with a poor prognosis. Thus patients with these mutations are usually treated with chemotherapy only.
NPM mutations convey both increased chemosensitivity and improved outcome.83 However, whether or not the presence of NPM mutations can partially abrogate the adverse affect of coexpressed FLT3-ITD is controversial.82 CEBPA mutations are often associated with normal karyotype AML and improved overall survival.70 Thus HSCT in first remission is not recommended for pediatric patients with isolated NPM or CEBPA mutations. WT1 mutations had an independent, poor prognostic impact in one study,125 but not in another (except in patients with coexpression of FLT3-ITD).126 Most clinical trials do not recommend allogeneic transplantation based on isolated WT1 mutations.
Other than a Berlin-Frankfurt-Munster study, the detection of MRD has been associated with adverse prognostic significance.26,139 COG and St Jude trials are using this factor to stratify patients for risk-directed treatment.
In COG, the combination of cytogenetic, molecular, and MRD information is being used to stratify patients into two groups for risk-directed therapy. Low-risk AML includes patients with mutations involving CBF, CEBPA, and NPM and those with no MRD at the end of induction therapy. This group represents approximately 73% of patients, with a predicted survival close to 75%. The high-risk group (the remaining 27% of patients with survival < 35%) includes patients with adverse cytogenetic abnormalities (monosomy 7, del(5q)–, −5), high FLT3-ITD to wild-type allelic ratio, or MRD at the end of induction and will be offered HSCT in first remission with the most suitable donor.
Molecularly targeted drugs can be tested in both low- and high-risk groups with the intention of reducing toxic therapy in the former group while improving survival of the latter group (Table 4). For example, the use of bortezomib to augment chemotherapy effects on AML stem cells will be randomized in the next COG phase III trial. In addition, FLT3-ITD inhibitors (ie, sorafenib) will be tested in the high-risk group with a high FLT3-ITD to wild-type allelic ratio.
In view of superior prognosis of Down syndrome patients with AML,140 current trials are testing strategies to reduce overall drug exposure, especially cardiotoxic drugs, because of special concerns in this population.141
Since the demonstration that ATRA significantly improved the outcome of patients with APL, these patients have been treated separately from other patients with AML, resulting in 5-year overall survival rates as high as 87%.128 A major, adverse prognostic factor is presenting WBC more than 10 × 109/L, associated with an event-free survival of approximately 60%.128 Approximately 3% of patients died during induction from hemorrhagic complications, accounting for half of the induction failures. The microgranular variant (M3v), a bcr3 PML breakpoint, and the presence of FLT3-ITD had also been associated with a poor prognosis,142 partly due to increased presenting WBCs. In the current pediatric studies, patients are considered low risk or high risk on the basis of WBCs ≤ or more than 10 × 109/L, respectively.
Although intensive anthracycline treatment (cumulative doses ranging from 400 to 750 mg/m2) has been attributed to improve outcome,130 one of the key issues for pediatric patients has been to reduce exposure to these cardiotoxic drugs. The efficacy of arsenic in relapsed and newly diagnosed patients has made it an attractive alternative to anthracyclines.143–145 Thus the current COG trial replaces a course of anthracycline-containing chemotherapy with arsenic, reducing anthracycline exposure to 355 mg/m2 of daunorubicin equivalents for standard-risk patients with negative MRD and to 455 mg/m2 for high-risk patients and standard-risk patients with positive MRD after the third treatment course. This trial also includes high-dose cytarabine based on the survival advantage observed in a European trial.145 Maintenance therapy with ATRA plus antimetabolites will be given to all patients for approximately 2 years. This trial is similar to the European trial, with one primary difference being the introduction of arsenic during consolidation. Future studies may include other targeted agents, such as anti-CD33 monoclonal antibody therapy.146,147
Because spontaneous remissions have been reported more frequently in the neonate age group, it is commonly recommended to initially observe such patients and, if cytoreduction is necessary, to perform an exchange transfusion. However, AML-directed therapy with dosing adjustments is usually required. Outcome in this age group is worse than that in older children, despite similar therapy, as a result of treatment toxicities and resistant disease.148 Novel approaches to directly target MLL149 and bcl-2150 are being developed (Table 4). In contrast to neonates, infants more than 1 month old seem to do as well as older children when treated with current, intensive regimens.15–30 The role of transplantation in infants is controversial because it has not been shown to definitely improve outcome and has significant adverse sequelae.
In conclusion, cure rates for childhood leukemias have improved largely through more intensive use of conventional cytotoxic agents evaluated in the context of large, randomized clinical trials. Clinical factors, genetic features of the leukemia, and initial response to therapy are now used in concert to personalize treatment for all patients. Advances in high-throughput genomics have led to the discovery of additional recurrent somatic lesions in the leukemic cell that offers not only additional prognostic information, but also opportunities for the application of novel targeted treatment. Finally, the recognition of host factors that are associated with the risk of leukemic transformation and the response to therapy will likely lead to more sophisticated treatment strategies in the near future.151,152
We thank many colleagues for their discussions and work on childhood leukemia. Our apologies go to colleagues whose work we were not able to reference because of space limitations.
Supported by the National Cancer Institute (Grant No. CA21765), CureSearch, and American Lebanese Syrian Associated Charities. R.J.A. is supported in part through an endowed King Fahd Professorship in Pediatric Oncology.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “U” are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
Employment or Leadership Position: None Consultant or Advisory Role: Ching-Hon Pui, Genzyme (C) Stock Ownership: None Honoraria: Ching-Hon Pui, Enzon Pharmaceuticals, sanofi-aventis Research Funding: None Expert Testimony: None Other Remuneration: None
Conception and design: Ching-Hon Pui, Robert J. Arceci
Financial support: Ching-Hon Pui
Administrative support: Ching-Hon Pui
Provision of study materials or patients: Ching-Hon Pui, Robert J. Arceci
Collection and assembly of data: Ching-Hon Pui, Robert J. Arceci
Data analysis and interpretation: Ching-Hon Pui, Robert J. Arceci
Manuscript writing: Ching-Hon Pui, William L. Carroll, Soheil Meshinchi, Robert J. Arceci
Final approval of manuscript: Ching-Hon Pui, William L. Carroll, Soheil Meshinchi, Robert J. Arceci