Relapsed AML Patients with Mutational Shifts Harbor An Oligoclonal Primitive Cell Compartment at Presentation; Evidence for Clonal Selection towards Relapse

Abstract 751

Previously we observed instability of type I/II mutations between diagnosis and relapse in a large portion (38%) of relapsed AML patients and this was associated with outcome (Bachas et al, Blood, 2010). Such instabilities may be explained by clonal selection during disease progression from oligoclonal initial AML populations. This hypothesis was tested in 6 AML patients with recurrent disease and mutational shifts between diagnosis and relapse. We determined the mutation profiles of mature and of immature subfractions of AML cells in initial diagnosis samples and related these to the corresponding relapse samples. Small amounts of cells (<50 cells) from mature and immature AML subfractions were sorted, followed by direct PCR assays for the detection of the frequently instable FMS like tyrosine kinase 3 (FLT3) internal tandem duplication (ITD), Wilms Tumor 1 (WT1) and RAS mutations and the relatively stable Nucleophosmin 1 (NPM1) mutation. AML blasts were cell sorted based on CD45^dim expression and primitive cell populations were defined by CD34/CD38 co-expression and refined based on the expression of CD133 and CD117. A lineage infidelity (Lin^inf) marker was used for designation of aberrant cells (Van Rhenen et al, Leukemia, 2007). Lymphocytes were sorted based on scatter properties and CD45^bright expression and were regarded as wild type controls of primary patient material in mutation analyses. Immunophenotypes, cytogenetics and minimal residual disease (MRD) status were routinely assessed in our laboratory. Our results are indicative of an oligoclonal immature AML cell compartment at presentation: the immature CD45^dimCD34⁺CD38^dim/- compartments of initial samples were more heterogeneous according to their mutation profiles, when compared to the bulk of mature CD45^dimCD34⁺ blast cells. Mutation profiles differed for at least one molecular marker among immature subfractions of the initial samples within 5 out of 6 patients. In the remaining patient, the mutation profiles were identical among sorted fractions of the initial sample. When comparing these data with the mutation profile of the bulk of AML cells of the relapse samples, in the latter patient and 2 out of the 5 above referred heterogeneous patients, mutational shifts could not be retraced to any subfraction of the initial sample. Strikingly, in the 3 other patients we found evidence for the clonal selection of a minor immature subclone of the initial sample, with a specific mutation profile that was predominantly present at relapse. In two of these patients we detected FLT3/ITD mutations of specific lengths, which were predominantly present at relapse, in a CD34⁺CD38⁻Lin^inf or CD34⁺CD38^dimLin^inf subfraction of the initial sample. This ITD was undetectable in the bulk of AML cells by standard diagnostics. Similarly, in the other patient with clonal selection, the CD34⁺CD38⁻Lin^inf compartment of the initial sample was enriched for a WT1 exon 7 mutation that predominated at relapse, whereas no WT1 mutations were detected in the bulk of AML cells at presentation. In concordance with these observations, immunophenotype and MRD data also suggested a clonal selection in these patients. In the 3 patients that did show a mutational shift, but lacked direct indications for clonal selection, a de novo mutation may have been induced during therapy. Alternatively, the use of additional immunophenotypical and molecular markers may be required for these samples to identify the relevant subclones. The key finding of our study is that we could detect in the immature fraction of initial samples of 3 AML patients a rare and otherwise undetectable subclone with a distinctive mutational profile that was predominant in the bulk of AML cells at relapse. This has several clinical implications: risk group stratification based on the presence of mutations that are associated with prognosis can be refined by genetic characterization of oligoclonal immature AML subpopulations that may play a role in the development of relapse. In addition, early and accurate detection of these subclones will be crucial in the prevention of relapse by accurate timing and application of targeted treatment. In conclusion, our results underline the concept of oligoclonality in AML and the involvement of clonal selection of immature chemoresistant subclones in the development of relapse. This study was supported by the Dutch Cancer Society (VU2005-3666).