Therapy-related acute myeloid leukemia (t-AML) and de novo AML (dnAML) are a group of aggressive diseases with poor outcome, in spite of recent advance in therapy. However, the alterations in different genes seems to be associated with this dismal prognosis. Recently, different groups have shown that t-AML in adult patients has a mutational profile distinct from dnAML. TP53 mutations were mostly described in t-AML. Next-generation sequencing (NGS) is a fast and accurate method used to demonstrate the presence of multiple alterations in several genes. The comparison of t-AML and dn-AML mutational profiles could be useful for improving our understanding of the molecular genetics events and in the outcome of each group as well.

Objectives: 1-To describe the incidence of pathogenic variants in t-AML and dn-AML using a customized NGS panel, 2-to study correlations of these mutations with classical genetic features in each group of AMLs, 3-to compare the mutational profile of both diseases, and 4-to evaluate the prognostic impact of these mutations.

Patients and Methods

A total of 136 available DNA samples from non-APL dn-AML and 11 t-AML pediatric patients were retrospectively analyzed. The samples were sequenced on a Miseq (Illumina) platform, using a customized amplicon-based NGS panel (Illumina) including 65 genes. A total of 172,360 bp of DNA target regions were covered with 1,207 amplicons with a mean coverage of 1,308. Sequences obtained were analyzed by in-house bioinformatics pipeline, with particular analysis for NPM1, CEBPA and FLT3 genes, using Pindel and ITDseek algorithms.


At least one pathogenic variant was detected in 100 out of 136 (74%) dn-AML analyzed cases, and 11 out of 11 (100%) t-AML patients. Overall, 317 mutations were found in 49 genes with a mean of 3.2 variants per patient for dn-AML, and 53 mutations in 26 genes with a mean of 4.8 variants per patient for t-AML. The most frequently mutated genes were: BCORL1 (5.4%, p=ns), FLT3 (5.0%, p=ns), KMT2A (5.0%, p=ns), BCOR (4.7%, p=0.0126), RB1 (4.4%, p=ns) for dn-AML, and KDM6A (15.1%, p<0.00001), BCOR (9.4%), CUX1 (9.4%, p=0.0020), BCORL1 (5.7%), FLT3 (5.7%) for t-AML. TP53 mutations were only detected in 5 dnAML cases. In the group of patients with dnAML, the distribution of classical genetic abnormalities was: KMT2A/11q23 (24%), RUNX1-RUNX1T1 (22%), Normal karyotype (19%), CBFB-MYH11 (9%). The most frequent mutations in normal karyotype were: FLT3 (38.1%, p=0.0019), CEBPA (27.3%, p=0.0001), NPM1 (28.6%, p<0.00001) and WT1 (23.8%, p=0.0001). In the group of t-AML, the genetic features were: KMT2A/11q23 (45%, p=ns) and Normal karyotype (9%, p=ns). CBFA2T2 mutation was statistically associated with the presence of RUNX1-RUNX1T1 in dn-AML (p=0.0002). No other significant association between variants and genetic groups was observed. Complete remission was achieved in 92% of cases in dn-AML and 72% in t-AML (median follow-up: 21 and 5 months respectively). Leukemia-free survival probability (LFSp) was 60(5)% [High-Risk: 55(6)%; Low-Risk: 71(10)%] for dn-AML and 13(11)% for t-AML (p=0.0008). In dn-AML group, LFSp(SE) of patients with at least one deleterious variant was 60(6)%, whereas those without any variant showed LFSp(SE): 58(10)(p=ns). In the normal karyotype group, the genotype NPM1/CEBPA-mutated plus FLT3/WT1-wild-type disclosed a LFSp(SE) of 86(13)% while remaining patients showed a LFSp(SE) of 42(14)% (p=ns).


We describe the incidence of mutations detected by NGS in a series of pediatric t-AML and dn-AML in our country. In spite of the limited number of patients, the mutation profile of t-AML seems to be different from the de novo counterpart. BCOR mutations were significantly associated with dn-AML cases, while KDM6A and CUX1 variants with t-AML patients. In contrast with adult reports, non TP53 mutations were observed in our group of t-AML patients. There was a significantly higher incidence of mutations in FLT3, CEBPA, NPM1 and WT1 within the normal karyotype group of dn-AML. The combined genotype NPM1/CEBPA-mutated plus FLT3/WT1-wild-type showed a trend to a better LFSp in dn-AML. The molecular genetic profile characterization of these heterogeneous diseases allows expansion of our understanding about the leukemogenic mechanisms involved.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.