Acute myeloid leukemia (AML) is an heterogeneous disease, with still a dismal prognosis, which requires bone marrow transplantation (BMT) as a most effective treatment in the majority of patients.1 The recent and continuously increasing identification of genetic mutations allows to define prognostic subgroups and potential targets for new treatments, aiming to improve the survival rates. The first attempt to perform the whole-genome sequencing of an adult AML case identified, in addition to known aberrancies, the mutation of the IDH1 gene as a novel acquired somatic mutation in AML,2 which could represent a new genetic entity in both pediatric and adult AML.
The IDH1 gene encodes for NADP+-dependent isocitrate dehydrogenase 1, an enzyme that catalyzes the oxidative decarboxylation of isocitrate to α-ketoglutarate. Acquired somatic mutations of the R132 residue of IDH1 have previously been described in gliomas,3 and then detected in adults with de novo AML, with a prevalence of 4.4%, rising to 16% in those with a normal karyotype.2, 4, 5 Although the role of IDH1 mutations in the pathogenesis of AML is still unclear, it has been suggested that an alteration in IDH1 expression may lead to a depletion of cellular α-ketoglutarate, activating the downstream oncogenic HIF-1α pathway.
In a recent paper,4 no IDH1 mutations were detected in a series of 257 children. IDH1 mutations were found in adult patients older than 34 years, frequently associated with a normal karyotype and with NPM1 and FLT3/ITD mutations. The low frequency of mutations did not allow to determine any significant difference in outcome or in the association with other known molecular markers of prognosis.
From 1 December 2002 to 31 December 2007, 205 childhood (non-acute promyelocytic leukemia) AML patients were enrolled into the multicentric AIEOP-LAM 2001/02 protocol. Among them, we analyzed the prevalence of IDH1 mutations in 165 consecutive patients, selected only on the basis of DNA availability.
The clinical and biological features of the analyzed and not analyzed population were not statistically different. The male to female proportion was 1.06, the median age at diagnosis was 11.3 (range 1.4–17.7) and the mean white blood cell count was 46 × 103 per mm3 (range 1–274). The analyzed patients represented different French American British (FAB) classes: 11 patients were M0; 33 were M1; 42 M2; 25 M4; 17 M4eo; 28 M5; 3 M6 and 6 patients were M7. Cytogenetic analysis was successful in 149 patients; the karyotype was normal in 42 patients (28%), whereas 107 children (72%) had aberrations; 23 patients(15%) had a complex karyotype with 3 or more structural or numerical abnormalities. Alterations of the core-binding factor as unique abnormality was detected in 20/165 patients (12%), therefore stratified as standard risk.
IDH1 gene mutations have been analyzed as previously reported.6 In this consecutive series of childhood AML cases, 4 out of 165 cases (2.4%) were positive for IDH1 mutations. The main biological and clinical features of the IDH1-mutated cases are listed in Table 1. All patients were male, the age at diagnosis ranged from 3 to 14 years, whereas the white blood cell count at diagnosis ranged from 8750 to 233 970 cells/μl. Three of them had FAB M1 and one M2; none of them had any localization in the central nervous system, whereas one had lymph nodes involvement. Two of the four children with the IDH1 mutation had a normal karyotype, whereas two carried different clonal translocation. Whereas one child carried the FLT3-ITD mutation at diagnosis, none had any other known mutation associated, such as FLT3-D835, CEBPA or NPM1. Based on cytogenetics, all of them were classified as high risk; complete remission was achieved in all cases and all but one received BMT after 5, 7 and 4.5 months from diagnosis, respectively. Two patients had a medullary relapse 15 and 33 months after BMT, respectively. All are alive after 26, 33, 33 and 20 months from BMT. All patients carried the R132H IDH1 mutation, in which an arginine was replaced by hystidine; this is one of the two most frequent IDH1 mutations described in literature.4, 7 The mutation was specific of the leukemia cells, being absent in the remission phase of all patients. Interestingly, the mutation was detected at the relapse stage of patient number 3, but not at the relapse of patient number 1 (Figure 1), suggesting that IDH1 mutations could represent a secondary lesion in the pathogenesis of leukemia. Although the phenotypic profile of the relapse is similar to the diagnostic one, the absence of a specific marker did not allow to formally discriminate between true relapse and secondary leukemia.
As 2/4 IDH1-mutated cases in the sequential screening had a normal karyotype, we extended the mutational screening to all Italian childhood cases diagnosed as AML with normal karyotype from 13 October 2000 to 15 April 2010, from whom we had DNA available. Out of the additional 97 cases with normal karyotype, only 1 (patient number 5) carried a IDH1 mutation (R132H). The clinical and biological characteristics of this additional patient are listed in Table 1.
In summary, in this study we show that, differently from what previously reported, although rare, IDH1 gene mutations can be detected also in pediatric AML, with an estimate prevalence of 2.4% (4/165) in the Italian series. The low prevalence does not allow any prediction on the outcome, although all patients are alive at different time after BMT, even in the presence of FLT3-ITD mutation (patient number 4). The clinical and biological characteristics of the mutated patients seemed not to be different from the overall childhood AML population, and similar to the adult IDH1-mutated cases. The R132 mutation is the only pediatric mutation detected so far. The extended series to a total number of 186 childhood AML with normal karyotype identified 3/186 mutations (1.6%) in this specific subgroup. Therefore, it does not seem that IDH1 mutations are more prevalent in normal karyotype, certainly are not exclusive of this subgroup. However, this further indicates that the so far defined ‘normal’ cytogenetic subgroup is any time reduced (toward the extinction?) as much as new abnormalities can be detected by improved technologies and new knowledge. It is interesting to recognize that all different genes affected by mutations in AML (FLT3, CEBPA, WT1, IDH1 or NPM1) are quite different at the functional level, and it is still not clear whether it is possible to define a common pathway to the disease. In addition, even considering that the IDH1 mutation could not be present at the relapse (like it happens for FLT3 mutations), it is questionable the role of these abnormalities, and whether those mutations in ‘normal cytogenetic’ subgroups could be sufficient for the clinical disease emergence or further events must be discovered in the complex and multistep pathogenesis.
This work was supported by Comitato ML Verga and Fondazione Tettamanti (Monza), Fondazione Città della Speranza (Padova), Associazione Italiana per la Ricerca sul Cancro (AIRC), Fondazione Cariplo, Ministero dell'Istruzione and Università e Ricerca (MIUR). We thank Vincenzo Rossi, Dr Sabrina Gelain and Dr Alessandra Beghin for their excellent technical support.