The classification of acute myeloid leukemia (AML) has evolved to the most recent World Health Organization (WHO) schema, which integrates genetic, morphologic, and prognostic data into a single system. However, this system was devised using adult data and how this system applies to a pediatric cohort is unknown. Performing a retrospective chart review, we examined our single-center experience with AML in 115 children and classified their leukemia using the WHO 2008 schema. We examined patient samples for mutations of FLT3, NPM1, and CEBPA. Overall survival was calculated within categories. In our pediatric population, most cases of AML had recurrent genetic abnormalities of favorable prognosis. More than 10% of patients in our series were categorized as AML, with myelodysplasia-related changes, an entity not well-described in pediatric patients. In addition, a large proportion of patients were categorized with secondary, therapy-related AML. To our knowledge, this is the first application of the WHO 2008 classification to a pediatric cohort. In comparison to adult studies, AML in the pediatric population shows a distinct distribution within the WHO 2008 classification.
Materials and Methods
The recent explosion of genetic data in acute myeloid leukemia (AML) translates to improved understanding of AML pathogenesis, informing treatment and prognosis. The classification of myeloid neoplasms has evolved from the French-American-British Classification (FAB), which relied on morphologic features and cytochemical staining, to the 2001 World Health Organization (WHO) classification integrating cytogenetic abnormalities of prognostic value. Revisions are necessary as the pace of discovery quickens. The 2008 WHO classification of AML combines morphologic, cytogenetic, and molecular data into a system relevant for prognosis.1–5 The changes encompassed in the 2008 WHO classification were reviewed by Heerema-McKenney and Arber.6
The 2008 WHO classification system, with a few exceptions, was based primarily on data from the adult AML experience. The epidemiology, significance, and usefulness of the full classification in the pediatric population has yet to be demonstrated. Because these changes significantly affect the way pediatric AML is diagnosed and treated, we undertook a review of our experience at a single institution. We applied the 2008 WHO classification to pediatric AML cases to examine the distribution within the classification schema and the associated outcomes. We present this reclassification of more than 100 consecutive cases and the comparison of survival outcomes among categories.
The Stanford University institutional review board approved a retrospective analysis of patients at the Lucile Packard Children’s Hospital (Palo Alto, CA). Patients were identified using the International Statistical Classification of Diseases and Related Health Problems, 9th Revision, diagnosis codes for AML. Patients were considered eligible for review if they were younger than 25 years at the time of diagnosis of primary malignancy, they received care at Lucile Packard Children’s Hospital, and their diagnostic bone marrow findings were available for review. Clinical data regarding treatment course and outcomes was obtained by review of the electronic medical record via Stanford Clinical Research Integrated Data Environment (STRIDE), a central resource accessing the electronic medical records of our cohort and was reviewed by one of us (K.L.D.).7
We identified 115 newly diagnosed, consecutive pediatric patients with AML treated between 1993 and 2010 at Lucile Packard Children’s Hospital. These cases were originally diagnosed using the FAB classification (years 1993–2001), 2000 WHO classification (2001–2008), and the 2008 WHO classification (2008–2010). The 2008 WHO classification was applied with review of the original diagnostic bone marrow aspirate, blood smear, trephine biopsy, flow cytometry, and cytogenetic studies, when available. Unstained bone marrow aspirates were used for molecular analysis of NPM1, CEBPA, and FLT3 when available.
DNA samples were isolated from archived unstained bone marrow aspirate smears, as previously described.8 The FLT3-ITD and FLT3-D835 mutations were detected using the Invivoscribe (San Diego, CA) polymerase chain reaction (PCR)–based kit according to the manufacturer’s instructions. The allelic ratio of FLT3-ITD cases was calculated as previously described.9NPM1 exon 12 insertion mutations were detected using multiplex PCR followed by restriction enzyme detection and capillary electrophoresis, as previously described.10–12 The entire coding region of CEBPA was amplified and sequenced using PCR as previously described.13
NPM1, CEBPA, FLT3 Mutational Analysis
The primary outcome measure was overall survival defined as the time from diagnosis to death from any cause. Subjects were censored at the date of death or last follow-up. Kaplan-Meier curves for overall survival were performed using GraphPad Prism version 5.0d for Mac (GraphPad Software, San Diego, CA). Curves were compared using the log-rank test. All P values are two-sided.
Open in new tab
Over a period of 17 years (1993–2010), we identified 115 patients treated for AML at Lucile Packard Children’s Hospital. Five patients were excluded because their diagnostic bone marrow results were not available for review. Demographic characteristics for the remaining patients included a median age of 9.5 years (range, 2 weeks-24 years), 53 were female, and 78 were white. Evaluation of age as a prognostic factor revealed a statistically significant survival disadvantage for those diagnosed before 1 year of age (P = .0109) or after 10 years of age (P = .0398) compared with those aged 1 to 10 years Figure 1A. All patients with primary AML were younger than 21 years at diagnosis. In 5 patients with therapy-related AML, the original cancer was diagnosed before the age of 21 years, but they developed the secondary AML between the ages of 21 and 25 years.
Open in new tab
Pathology and Cytogenetic/Molecular Rearrangements
The new category of AML with myelodysplasia-related changes (AML MRC) comprised 14% of cases. The average age was 11 years. Of these 16 cases, 11 were diagnosed as AML MRC because of a myelodysplasia-associated karyotype Table 2. In 3 cases the designation resulted from morphologic multilineage dysplasia alone. An additional 2 cases that did not meet cytogenetic criteria for the diagnosis had multilineage dysplasia and a history of preceding myelodysplastic syndrome. Two patients with AML MRC had mutated NPM1, 1 had multilineage dysplasia, and the other had a history of myelodysplastic syndrome. Twelve cases of AML MRC were analyzed for FLT3 mutations. A single FLT3-ITD mutation was identified with a high allelic ratio. As a group, the survival appears poor compared with core-binding factor AML and similar to AML with 11q23 abnormalities in childhood. We found no difference in survival between those who met criteria for AML MRC based on morphology compared with those who had defining cytogenetic changes (P = .3143).
A large proportion of our cases (10% each) was classified as tAML or DS-AML. The average age of tAML cases was 15 years and none of these had any of the recurrent genetic abnormality karyotypes. Two were of normal karyotype, 2 had MLL rearrangements, 3 had complex karyotypes, and 2 had monosomy 7. Multilineage dysplasia was common. One of the 8 cases tested for FLT3 had an internal tandem duplication mutation with a high allelic ratio. No case had NPM1 or CEBPA mutations. Seven patients suffered tAML after the initial diagnosis of acute lymphoblastic leukemia, 2 patients had an initial diagnosis of osteosarcoma, and 2 patients were initially treated for rhabdomyosarcoma. The average interval between diagnosis of the primary cancer and tAML was approximately 5 years, with a range of 2 to 11 years. Four of 11 patients were treated with stem cell transplantation and all received some combination of chemotherapy. Survival for this group is shown in Figure 2A. The average age for DS-AML cases was 1.7 years. Four of these patients had a documented history of transient abnormal myelopoiesis. All cases had megakaryoblastic differentiation, and multilineage dysplasia was commonly observed. None of the 7 cases tested had mutations of FLT3, NPM1, or CEBPA.
We examined outcomes in our cohort after stratification into risk groups based on recent Children’s Oncology Group (COG) clinical stratification (eg, AML00531) as well as WHO 2008 groups. Favorable risk groups included those with core-binding factor AML, acute promyelocytic leukemia, NPM1 mutated/FLT3 wild type, CEBPA mutated, and DS-AML. Standard risk groups included AML NOS, AML with t(1;22), AML with t(9:11), and AML MRC. Unfavorable risk groups included AML with monosomy 7, tAML, and FLT3-ITD high allelic ratio. Patients with t(6;9) are not considered unfavorable in the current COG treatment stratification though the literature supports it as a poor prognostic factor. In our cohort, both patients with t(6;9) were classified as unfavorable because of an FLT3-ITD mutation with high allelic ratio.14,15 Stratification into risk groups resulted in 48 patients with favorable risk, 44 with standard risk, and 19 with unfavorable risk status Figure 3A. The overall survival of those with favorable characteristics was significantly different from those with intermediate or poor risk status (P < .0001) Figure 3B. However, we were unable to detect significant differences in overall survival for patients in the intermediate or high-risk categories Figure 3C, Figure 3D, Figure 3E. Overall survival at 10 years was 38% for intermediate risk patients and 26% for poor risk patients (P = .32) Figure 3B.
Treatment data are summarized in Table 3. All patients had remission induction therapy with a backbone of cytarabine and an anthracycline. Depending on the treatment protocol used, patients may have also received therapy with etoposide, 6-thioguanine, l-asparaginase, and steroids. Patients with acute promyelocytic leukemia were treated with all-trans retinoic acid and/or arsenic trioxide often with the same backbone of anthracycline and/or cytarabine. Twenty-five patients received a stem cell transplant as part of their therapy. Ten-year overall survival was 56% in the patients who underwent stem cell transplantation and 57% in those who were treated with chemotherapy alone (P = .9461, not significant) Figure 2B.
Examination of a single-center experience with pediatric AML reveals striking differences in the incidence of AML subtypes in the pediatric population compared with adult cohorts. To our knowledge, this is the first description of pediatric AML categorized using the 2008 WHO classification. In comparison with adult data, pediatric AML is characterized by recurrent genetic abnormalities. In our cohort, 48 patients (44%) would fall into this category under the WHO 2008 schema. In the adult cohort from our institution described by Weinberg et al,16 only 10% of patients were categorized as AML with recurrent genetic abnormalities. The overall survival outcomes for patients in the subcategory of AML with recurrent genetic abnormalities was favorable, as supported by the 71% of cases in this subgroup characterized as core-binding factor leukemia, acute promyelocytic leukemia, or AML with mutated NPM1 or CEBPA, all of which have been associated with improved prognosis.11,17–19 These cases comprise the majority of patients with favorable-risk AML and are responsible for the significantly improved survival of this group compared with those with intermediate or poor risk disease (P < .0001).
AML MRC is a new entity in the 2008 classification, expanding the prior category of AML with multilineage dysplasia. This category has been assumed to be rare in pediatrics.2 Using the 2008 classification criteria, AML MRC encompasses 14% of the cases in our cohort. Whether this is related to a higher-than-expected incidence in the pediatric population or to a referral pattern in our area cannot be ascertained in our study. Evaluation of the incidence of this condition in a larger cooperative group series is warranted. It is well-recognized that some of the associated cytogenetic abnormalities of AML MRC, specifically monosomy 7 and monosomy 5, are associated with a poor prognosis in both children and adults.20 Two-thirds of the patients with AML MRC in our series resulted from myelodysplasia-associated karyotype with or without dysplasia, whereas the remaining one-third had multilineage dysplasia without karyotypic changes.
In the adult population, AML MRC portends a poor prognosis.16,21–23 Less is known about the incidence of multilineage dysplasia and its risk in the pediatric population. Adachi et al24 examined the Japanese experience with trilineage dysplasia in 341 newly diagnosed pediatric patients with AML. They found an incidence of trilineage dysplasia of 2.6% and no association with FAB classification or risk group. Multilineage dysplasia is common in other categories such as AML with t(6;9), tAML, and DS-AML.
The WHO 2008 classification has 2 new provisional categories of recurrent genetic abnormalities: mutations in NPM1 and CEBPA. Nucleophosmin, a protein largely present in the nucleolus of cells is involved in multiple cellular processes and is frequently mutated in AML. The mutation alters the nuclear localization signal and, in mutated cells, the protein is found in the cytoplasmic compartment.25,26 In adult patients with AML, this mutation often is found in conjunction with FLT3 ITD, and it does not appear to trump the poor prognosis conferred by FLT3 mutation. In children, in the absence of FLT3 ITD, the NPM1 carries a favorable prognosis.27 The incidence of NPM1 mutations in childhood AML has been reported by many groups as approximately 10%. Our results are in agreement with this; 10 patients (9%) of our cohort carried a mutation in NPM1. Of those, 7 were positive for NPM1 mutations alone and 3 had an activating mutation of FLT3 as well. In pediatric patients, as in adults, NPM1 mutations are largely found in the group of patients with cytogenetically normal AML; in this cohort, 70% of patients harboring a mutation in NPM1 also had cytogenetically normal AML. Interestingly in our cohort, only those patients who also carried a mutation in FLT3 had an abnormal karyotype.
Likewise, mutations in CCAAT/enhancer–binding protein α (CEBPA), a transcription factor involved in the regulation of proliferation and terminal differentiation in granulocytes are implicated in the pathogenesis of AML. These mutations are present in 5% to 10% of adult AML and carry a favorable prognosis in this group, especially when both alleles are mutated.17 The incidence of CEBPA mutations in the pediatric population has been reported to be between 4% and 8%.18,28,29 In our cohort, we identified 3 patients (2.7%) with mutations in CEBPA. Two of these patients had normal karyotype. Two of the 3 patients had biallelic mutations of CEBPA. Two of these 3 patients died, 1 of Streptococcus viridans sepsis and the other of disease. The patient who died of disease had only 1 mutated allele.
Open in new tab
A, Overall survival of patients with treatment-related acute myeloid leukemia (tAML). B, Overall survival based on therapy received—chemotherapy alone vs chemotherapy and stem cell transplantation (P = .9461).
This study specifically highlights 2 important and prognostically different subgroups of AML in the pediatric population. The first is the high incidence of Down syndrome–related leukemia. In the current cohort, 10% of cases were DS-AML. This is not surprising given that AML in this population typically occurs in patients between 1 and 5 years of age and thus is enriched in the pediatric population. In agreement with previous observations, patients in our cohort with DS-AML had excellent outcomes, with overall survival of 90% at 10 years, consistent with other reports of event-free survival of more than 80% in this population.30
Secondly, tAML was observed in 10% of our cohort. Initial diagnoses included both solid tumors and leukemia. It is possible that this is an underestimation of the true prevalence of tAML in our group because patients treated as children for their primary malignancy may have been diagnosed and treated for secondary AML outside our institution and thus not captured in this analysis. This observation is particularly poignant because of the abysmal 5-year survival rate of 13% in this subgroup of patients. The issue of secondary AML is of particular importance in the pediatric population. Though treatment for pediatric cancer has drastically improved and 80% of patients survive 5 years, the late effects observed in these patients can be significant. Secondary myeloid leukemia exemplifies the challenges that remain to improve outcomes.31 This group presents an opportunity to investigate the mechanisms of leukemogenesis as well as the genetic factors that lead to secondary AML. A recent analysis of the Surveillance, Epidemiology, and End Result database examined the incidence of secondary hematologic malignancies in survivors of childhood cancer.31 This report found that 111 of 34,867 patients had documented secondary hematologic malignancy, and of these, 49% were tAML. This group was characterized by the shortest median latency period (36 mo) from primary malignancy diagnosis to diagnosis of tAML and had the worst outcome of secondary hematologic malignancies, with 5-year survival of just 18%. The literature suggests that the risk of secondary leukemia plateaus 10 years after primary diagnosis. A recent report by Nottage et al32 describes a 3.5-fold increased risk of subsequent leukemia for survivors of childhood cancer over the general population more than 15 years after the primary diagnosis. Thus, our results highlight the significance of this particular subtype of AML and a need to improve outcomes for this group of patients.