Immunophenotypic profiling of leukemic stem cells to track FLT3-ITD positive chemo-resistant clones in AML

Introduction Acute Myeloid Leukemia (AML) is characterized by heterogeneous genetic abnormalities, immunophenotypes and clinical outcomes. Although available treatments induce complete remission (CR) in ~80% of AML patients, some of them will eventually relapse, due to the emergence of resistant clones. Persistence of leukemia stem cells (LSCs) in AML patients achieving CR after chemotherapy, is known to drive disease recurrence and worsen outcome. Therefore, the identification and characterization of LSCs, that generally reside within the CD34+ /CD38- cell fraction and confer resistance to therapy, represents an important challenge. The Internal Tandem Duplication of FLT3 gene (also known as FLT3-ITD mutation), present in approximately 25% of AML patients, has been shown to occur at the LSCs level, and may be a primary event in leukemogenesis. It is also well known that FLT3-ITD mutations represent an independent predictor of poor prognosis in AML, being associated with an increased risk of relapse. According to these data, a recent publication from our group demonstrated that chemo-resistant FLT3-ITD clones, present at AML relapse, were already identifiable at subclonal level at the time of diagnosis. Presence of these subclones actually directly correlates with the detection of CD34/CD123/CD99/CD25 positive cells by multiparameter flow cytometry (MFC). In this line, it has been recently demonstrated that expression of CD99 allows for separation of LSCs from functionally normal hematopoietic stem cells in AML. Based on the preliminary identification of the MFC fingerprint associated with FLT3-ITD mutated cells, our study aimed at better characterizing the phenotypic and genetic profiles of FLT3-ITD positive AML LSC. These cells may sustain resistance mechanisms driving disease relapse, and may be ideal treatment targets. Patients and Methods Following previous results and to better identify patients at risk of relapse, we investigated the FLT3-ITD molecular status at the DNA level in a cohort of 150 patients diagnosed with AML between January 2017 and December 2018. Samples were analysed at at the time of diagnosis, of relapse, and in selected cases during follow-up, using standard procedures. Based on FLT3-ITD V molecular status and samples availability, a sequential gating strategy was carried out in 12 AML samples at diagnosis, to sort the CD34/CD123/CD99/CD25+ (CD38+ and CD38-) LSCs-enriched fraction, the CD34+ stem cell subset (CD123/CD99/CD25-), and T-lymphocytes. Cells were purified using high-speed cell sorting, in collaboration with the IRCCS Santa Lucia Foundation. In order to better characterize FLT3-ITD positive LSCs at the genome level, targeted sequencing was performed on 32 AML samples collected from the 12 AML patients at diagnosis and relapse, using the Oncomine™ Myeloid Research Assay panel on the Ion Torrent™ S5 sequencer. This panel contains targeted multibiomarkers to simultaneously detect variants across 40 key genes, 29 driver genes and a broad fusion panel to cover most of the genetic changes associated with AML. Results Enrichment of the FLT3-ITD positive population, characterized by a significantly higher FLT3-ITD mutation load, was observed within the CD34+ compartment of CD123/CD99/CD25+ cells, as compared to MNC (p=0.006). Conversely, the lymphoid/myeloid precursors showed low or absent CD123/CD99/CD25 expression (p=0.002). In one patient with two different FLT3- ITD mutated clones at diagnosis, the LSC population defined by CD34/CD123/CD99/CD25+ expression was homozygous for one ITD mutated clone, likely originated by loss of heterozygosity (LOH) of the mutated allele. Furthemore, based on several studies indicating that LSCs are particularly enriched in the CD34+/CD38- cell fraction, we investigated whether the FLT3-ITD mutation was enriched in the CD34/CD123/CD99/CD25+/CD38- LSCs compartment. In one AML patient, these cells represented the dominant FLT3-ITD mutated population, as compared to the CD34+/CD38+ counterpart. Furtermore, to trace the clonal evolution of LSCs carrying the FLT3-ITD mutation, we sorted different BM-cell population at diagnosis and during follow-up from one AML patient who relapsed 8 months after the initial diagnosis. A progressive increase of the FLT3-ITD allele burden was detected in the CD34/CD123/CD99/CD25+ subset, as compared to the CD34+ cells lacking CD123/CD99/CD25 expression, and to the total MNC population. In this patient, a high FLT3-ITD mutant allele burden in in the sorted LSC was detected already at time of complete morphologic remission, and 2 months VI before haematological relapse. These data confirm that CD34/CD123/CD99/CD25+ LSC may represent the FLT3-ITD reservoir, driver of disease relapse. Finally, based on recent global genome studies carried out by the TCGA network revealing that de novo AML arises from combinations of several recurrent driver gene mutations, we perfomed targeted NGS mutation analysis on the total leukemic cell population (12 pts at diagnosis and 3 pts at relapse) and on several purified leukemia cell fractions in 7 AML patients. We observed a median of 3.5 and 3 mutations per patient at diagnosis and relapse, respectively, with 3 mutations in common in paired diagnostic and relapse samples. In addition to FLT3-ITD mutations, the most frequently mutated genes were NRAS (40% of cases), RUNX1, TET2 (30%), DNMT3A, BCOR, IDH2, NPM1, FLT3-TKD, and KRAS (20%). Mutations in DNMT3A, NPM1, IDH2, RUNX1 and TET2 genes were stable during the disease course. On the contrary, NRAS, FLT3-ITD and EZH2 mutations were lost at the time of relapse in 2 pts, confirming that leukemias may evolve by a process of clonal expansion and/or selection. Comparing the gene mutation load in the diagnostic vs relapse samples, we found an almost costant increase in the variant allele frequency (VAF) for FLT3-ITD, FLT3- TKD, NRAS and RUNX1 (from 0.47%, 2.7%, 3.3%, 3%, and 34% at diagnosis, vs 16%, 9.5%, 39% and 4.2% at relapse). Conversely, in one patient, VAF of DNMT3A and TET2 was similar at diagnosis (41% and 91%) and relapse (49% and 99%), indicating that they may represent early founder mutations, stable during disease course. On the contrary, mutations affecting other myeloid genes such as FLT3 can be gained or lost during disease progression. Aiming at shedding light on the molecular heterogeneity and subclonal structure of AML genetic, we then compared the mutational profiles of MNCs to that of CD34/CD123/CD25/CD99+ highly purified cell population and/or to the CD34+/ CD123/CD25/CD99- counterpart, available for 7 AML patients (N=11 samples). Targeted NGS analysis revealed two different clonal evolution models. In 7 of 11 cases, the same mutation pattern was present both in MNCs and LSC, suggesting that the LSC represent the same genetic background of the bulk leukemic cells. We conclude that in AML, assessment of the genetic profile at diagnosis in a powerful indicator for the presence of somatic mutations, which remain stable VII also at relapse in most cases, with the exception of FLT3-ITD mutations, which may expand at relapse. In this line, muttional analysis of MNC is informative in most cases and is representative of LSC profiles. FLT3-ITD mutate subclones are on the contrary entriched in CD34/CD123/CD99/CD25+/CD38-/+ LSCs