Abstract
Background. Acute Myeloid Leukemia (AML) is a rare and genetically heterogeneous disease that constitutes 15 to 20% of childhood leukemia. Despite major treatment improvement over the past decades pediatric AML remains a challenging disease with poor outcome compared to acute lymphoid leukemia (ALL). About 50% of these patients relapse after standard intensive chemotherapy. Molecular analysis pointed out the prognostic impact of gene mutation such as FLT3-ITD, NPM1 or CEBPA; and new categories of regulators like epigenetic modifiers. More recently mutational profiling studies revealed distinct molecular subgroups with prognostic significant and stratification in adult AML. Nevertheless cytogenetic and mutational profiles are quite different between adult and pediatric AML. Extensive genomic studies have not been reported to date in pediatric AML. In this context it is of importance to identify additional genetic or molecular abnormalities to better understand leukemogenesis and also to predict outcome and serve as novel therapeutic targets.
Methods. We performed a mutational analysis on diagnostic samples from patients enrolled in the French National Multicenter ELAM02 trial. 438 patients with de novo AML (except AML3) were enrolled between march 2005 and December 2011 (median age: 8,22yrs [0-18.61]; median WBC: 15.4G/l [0.4-575]; cytogenetic subgroups: CBF-AML[n=97], NK-AML [n=109], MLL-AML[n=95], MRC2 other[n=77], MRC3 [n=55], failure [n=5]). Diagnostic samples were prospectively collected and 386 of the 438 patients (88%) were studied by next-generation sequencing (Miseq, Illumina with haloplex librairy and ion Proton, thermofischer with ampliseq librairy) including 36 genes frequently reported in myeloid malignancy. Two different technologies of next generation sequencing (NGS) were used, allowing direct validation. FLT3-ITD was detected and quantified by Genescan analysis.
Results. We identified 579 driver mutations involving 36 genes or regions in 386 patients (mean 1.5 per case), with at least 1 driver mutation in 291 patients (75%) and 2 or more driver mutations in 44% of samples. The number of mutation identified at diagnosis in cytogenetic subgroup is significantly lower in MLL-AML (0.44 mutation/patient; p<10-4). Mutations involving genes from the tyrosine kinase pathways (i.e RAS, FLT3, KIT, PTPN11, JAK2, MPL, CBL) were the most frequent and represent 56.3% of all aberrations. Among them N-RAS was detected in 26.4% of all cases, followed by FLT3-ITD, KIT and K-RAS in 14.8%, 12.4% and 12.2% respectively. We identified 64 driver mutations in the group of transcription factors (CEBPA, RUNX1, GATA, ETV6), 60 in the combined group of chromatin modifier (ASXL1, EZH2, BCOR) and DNA methylation (DNMT3A, IDH, TET2), 59 in the group of tumor suppressor genes (WT1, PHF6, TP53) 36 mutations in NPM1 gene, and few mutations in cohesion and spliceosome sub-groups. Identified mutations are indicated in the figure according cytogenetic subgroups. Among the 438 patients, 398 (91%) were in complete remission (CR) after two courses (induction and first consolidation), the 5-year overall survival (OS) is 71.5% [65-78] and the 5-year leukemia free survival (LFS) is 56.6% [49.7-63.5]. In univariate analysis, we found that FLT3-ITD, mutations in RUNX1, WT1 and PHF6 were associated with reduced LFS (p=0.0003 for FLT3-ITD, p=0.01 for RUNX1, p=0.02 for WT1 and p=0.025 for PHF6) and reduced OS (p=0.0003 for FLT3-ITD, p=0.0003 for RUNX1, p=0.015 for WT1 and p=0.04 for PHF6). Mutations in NPM1 is associated with an improved 5-yr LFS (p=0.014) and 5-yr OS (p=0.005). Multivariate analysis revealed that FLT3-ITD, RUNX1 and PHF6 were independently associated with an adverse outcome and NPM1 with an improved outcome.
Conclusions. We performed an extensive mutational study in de novo pediatric AML enrolled in the ELAM02 trial. We described the genomic landscape of 386 patients and showed the frequency of different mutations according cytogenetics. Interestingly we found mutations in genes involved in constitutional or pre-leukemic disease such as PTPN11, RUNX1, MPL or ETV6. We found that FLT3-ITD, RUNX1 and PHF6 mutations predict poor outcome although NPM1 mutations predict a better outcome. Mutational profiling reveals useful information for risk stratification and therapeutic decisions.
Figure Figure.
Disclosures
Baruchel: Amgen: Consultancy.
Defining driver mutations in the genomic landscape of breast cancer
