Abstract 237▪▪This icon denotes a clinically relevant abstractMutations in the nucleophosmin 1 (NPM1) gene represent one of the most frequent gene mutations in acute myeloid leukemia (AML), in particular in cytogenetically normal (CN)-AML. NPM1 mutations (NPM1mut) are considered as an early genetic event in the pathogenesis of AML. To address the role of clonal evolution from diagnosis to relapse in NPM1mut AML, we applied high-resolution genome-wide single nucleotide polymorphism (SNP) array analysis using the Affymetrix 6.0 platform to detect copy number alterations (CNAs) and uniparental disomies (UPDs) in paired samples from 42 patients. In addition, we determined NPM1 and FLT3 [internal tandem duplication (ITD) and tyrosine kinase domain (TKD)] mutation status in all samples. Blood or bone marrow samples obtained at complete morphologic remission were available for all patients to exclude germline copy number variations.At diagnosis, 29 cases (69%) had a normal karyotype by cytogenetics and no CNAs and UPDs by SNP analysis. In the 13 remaining cases, we found a total of 10 CNAs in 7 cases (19%), and 6 UPDs in 6 cases (14%): deletions of 9q21 (size range 0.9 to 17 Mb) were detected in 5 cases and were the only recurrent CNA; the only recurrent UPD affected the long arm of chromosome 13 in 4 cases, all resulting in homozygous FLT3-ITD mutations with FLT3-ITD/wildtype ratios >1; heterozygous FLT3-ITD and –TKD mutations were detected in 9 and 7 patients, respectively.At the time of relapse, the number of CNAs increased (34 CNAs in 16 cases, 38%) while the frequency of UPDs remained unchanged (6 UPDs in 6 cases, 14%). Of note, in 6 patients (14%) the NPM1 mutation was no longer detectable at the time of relapse; SNP analysis showed completely distinct CNAs/UPDs in 4 of these patients; 3 of these 4 cases had a small gain at 11q23 corresponding to MLL partial tandem duplications as confirmed by PCR. These findings suggest that these 4 cases were therapy-related AMLs (t-AML) rather than relapsed AML. The median interval from diagnosis to relapse/tAML in these 4 cases was 65 months compared with 9 months for the relapsed cases still having the NPM1 mutation. In the two remaining cases, genetic alterations were neither present at diagnosis nor at relapse. Analysis of other gene mutations (eg, IDH1 and 2, DNMT3A, ASXL1, p53) is currently under way to further elucidate the clonal origin of these cases.Of the 36 NPM1mut positive relapse samples, 15 maintained a “normal karyotype”, and 2 showed the CNAs already present at diagnosis; 19 relapse samples (53%) displayed clonal evolution with acquiring new (n=15) and/or loosing single aberrations (n=4): Acquired recurrent alterations comprised deletions of tumor suppressor genes [ETV6 (n=2), TP53 (n=2), NF1 (n=2), WT1 (n=2)], most of which are uncommon in de novo NPM1mut AML. All 6 UPDs detected in relapse samples affected 13q, of which 3 were already present at diagnosis. One patient with initial heterozygous FLT3-ITD mutation developed a homozygous state by acquiring UPD13q at relapse. Two cases with wild-type FLT3 at diagnosis acquired UPD13q at relapse. Of note, one UPD13q was not present in the corresponding relapse sample anymore.In conclusion, almost half (45%) of NPM1mut AML showed evolution to a more aberrant karyotype at relapse, including acquisition of high-risk genetic changes that may account for the adverse prognosis of relapsed patients. Conversely, other alterations such as UPD13q or del(9q) detected at diagnosis were not always present in relapse samples, implying that relapse had evolved from a more ancestral clone. In addition, our data suggest that in a proportion of cases t-AML rather than relapse had developed. Further analysis, such as gene mutation studies of paired diagnosis/ relapse samples, will provide more detailed information on clonal evolution events in the pathogenesis of NPM1mut AML. Disclosures:No relevant conflicts of interest to declare.