Flow Cytometry Quantification of Leukemic Stem Cells Is Associated with Risk Stratification and May Be Useful for Minimal Residual Disease in Acute Myeloid Leukemia

Abstract 1470

Investigation of minimal residual disease (MRD) after remission induction (RI) therapy provides important information for risk assessment in patients with acute myeloid leukemia (AML). The presence of immature and chemotherapy-resistant leukemic stem cells (LSC) within the bulk of AML blasts at diagnosis and in post-induction bone marrow (BM) may lead to relapses. Nevertheless, whether the frequency of cells with LSC characteristics and their clearance after induction is correlated with prognosis has not been established. Using four-color multiparametric flow cytometry (MFC), BM quantification of leukemic associated phenotypes (LAPs) and LSCs was performed at diagnosis and after the first RI from 37 AML patients, excluding acute promyelocytic leukemia, with a median age of 48 years and a male/female ratio of 0.95. All patients received the conventional daunorrubicin and cytarabin (3 + 7) RI chemotherapy. Complete remission (CR) was defined as BM blast count inferior to 5%. Thirty-three patients were classified according to the European Leukemia Net recommendation and 9/32 (28.1%) patients were allocated in favorable, 14/32 (43.8%) in intermediate and 9/32 (28.1%) in poor cytogenetic/molecular risk group. FLT3-ITD mutation was detected in 9/33 (27.3%) and 5/28 (17.9%) carried NPM1 mutation. MRD identification was performed on 20 erythrocyte-lysed whole BM samples after staining with a panel of directly conjugated monoclonal antibodies. Five patients were excluded from this analysis because a LAP could not be identified. Blasts gating was performed considering the low expression of CD45 and sideward scatter (SSC) and CD34 expression (these latter, when more than 20% of blasts were detected at diagnosis). Within this population, the following LAPs were investigated: CD15/CD117, HLA-DR/CD13, HLA-DR/CD33, CD2/CD56, CD19/CD11b, CD42a/CD33, CD64/CD11c or CD14/CD11c. LSCs were selected by the same CD45^dim × SSC gating strategy and defined as CD34⁺/CD38⁻/CD123⁺. After staining procedures, at least one hundred thousand events were acquired in a FACScalibur flow cytometer and analysis was performed using the Cell Quest software. LAP and LSC quantification, at diagnosis and at days 21 to 30 of RI, was analyzed as a categorical variable defined as lower or higher than the median and was compared to the following variables: age (< or > 60 years old); WBC count (< or > 30 × 10³); cytogenetic/molecular risk; and morphological CR. The comparison between LAP and LSC quantification at both time points was also assessed. Comparison of categorical variables was performed using Fisher's exact test or Yates' corrected chi-square for two or more variables, respectively. Statistical analyses were performed using SPSS 13.0 software and P < 0.05 was considered to be significant. LSC quantification at diagnosis was found at varying frequencies across different cytogenetic/molecular risk groups, being higher at the poor risk group (P = 0.041). Of note, 100% of the poor risk patients had high levels of LSC at diagnosis. In addition, the presence of FLT3-ITD mutation was associated with higher amounts of the LSC population at diagnosis (P = 0.043). The most frequent detected LAP was CD45^dim/CD34⁺/HLA-DR⁺/CD13⁺. Low or high expression of the LAP was not correlated with the prognostic variables at diagnosis. CR rate was 83.33% and was not different in the groups with high or low levels of LSC at diagnosis. However, LAP and LSC quantification after RI were found to be correlated (P = 0.018), suggesting that the LSC subpopulation can be useful for MRD monitoring at this early treatment time-point. Therefore, LSC quantification by MFC at diagnosis can identify patients at high risk of relapse and offers the opportunity to study the stem cell compartment after chemotherapy. These findings are particularly important for the intermediate normal karyotype risk group patients, who frequently do not have specific molecular targets for MRD monitoring.