Distinct Genetic Lesions Drive Leukemogenesis in Secondary Acute Myeloid Leukemia,

Abstract 3559

Secondary acute myeloid leukemia (sAML) evolves from different types of chronic myeloproliferative neoplasms (MPN) and myelodysplastic syndromes (MDS). However, acute myeloid leukemia frequently arises de novo (dnAML) without a previous hematological phenotype. Since the genetic basis of sAML is poorly understood our objective was to define genetic aberration profiles of sAML patients and compare the distribution of these aberrations to the ones found in dnAML.

We studied a total of 195 patients of which 122 were diagnosed with dnAML and 73 with either post-MDS (n=27) or post-MPN (n=46) sAML. In order to obtain high-resolution karyotypes we genotyped DNA samples using Affymetrix SNP 6.0 arrays enabling analysis of 1.8 million data points per genome. In addition to copy number data, loss of heterozygosity associated with acquired uniparental disomy (UPD) was evaluated. Furthermore, the cytogenetic data was complemented with mutational analyses of commonly mutated genes in myeloid malignancies: IDH1, IDH2, NPM1, CBL, FLT3 and TP53.

With the arrays used we could not detect any cytogenetic lesion in 32% of dnAML and 18% of sAML. Furthermore, dnAML showed karyotypes with lower complexity compared to sAML. We observed 31 recurrent cytogenetic aberrations (>4 events). Of these 12 showed significant bias towards sAML. The most prominent were 9pUPD (P=0.0001), 1q gain (P=0.0023), del7q (P=0.0039), 11qUPD (P=0.0045) and del4q (P=0.006), targeting known genes JAK2, MDM4, CUX1, CBL and TET2, respectively. In our series gains of 3q involving EVI1 were exclusively found in sAML (P=0.018) and the 3q amplicon contained EVI1 and 10 other genes. The most common cytogenetic aberration observed overall was del5q (34 events) in our patient cohort, significantly associated with sAML (P=0.017).

As frequent 1q gains targeting MDM4 in sAML previously implicated the p53 pathway in sAML leukemogenesis, we performed exon sequencing of TP53 in the entire patient cohort. The results clearly showed that TP53 is significantly more mutated in sAML compared to dnAML (P=0.0052), with 16.4% of sAML patients carrying the mutation, comparing to only 4% of dnAML. We also confirmed the previously described co-occurrence of del5q with TP53 mutations (P=0.0031). Another gene showing significantly higher frequency of mutations in sAML is CBL (P=0.0035), found mutated in 10.3% of sAML and 0.8% of dnAML, respectively. On the other hand mutations in FLT3 and NPM1 were more common in dnAMLs (P=0.0090 and P=0.0001, respectively). FLT3 was found mutated in 22.6% of dnAML and 7.6% in sAML. Mutations of NPM1 were present in 24.4% of dnAML and in only 3% of sAML. Mutations of IDH1 and IDH2 were found to be present at similar frequencies in both AML types. Despite the genetic and phenotypic diversity of MPN and MDS in the chronic phase of the diseases we observed no significant genetic differences in the leukemic phases of these disorders within our sAML cohort.

Of all the aberrations, deletions were most common representing 55% of all lesions (361 total events). When we mapped all the detected deletion events we obtained a high-resolution deletion map of AML. A number of known leukemia-associated tumor suppressor genes were found within the common deleted regions such as TET2, CUX1, IKZF1, FOXP1, ETV6, NF1 and DNMT3A. We also identified a number of new tumor suppressor candidates distinct or common for both AML groups, such as ASXL2, BAI3 and others.

The higher cytogenetic complexity of sAML in contrast to dnAML and the differences of aberration frequencies between the two AML types are likely attributed to a longer disease duration and clonal evolution in sAML. Furthermore, clinical management of the chronic disease phase might influence the cancer genome evolution in sAML whereas such influences are absent in dnAML. Despite these differences, certain defects appear to be universally contributing to leukemogenesis such as IDH1/2 mutations. Our data indicate that leukemogenesis in MPN and MDS is not predominantly driven by lesions typical for dnAML. This study also provides a number of new potential markers for genetic stratification of patients with acute myeloid leukemia.