Determination of Minimal Residual Disease in Patients with Acute Myeloid Leukemia by Multiparameter Flow Cytometry: Prognostic Impact at Different Checkpoints.

Abstract Guiding antileukemic treatment in patients with acute myeloid leukemia (AML) is increasingly based on levels of minimal residual disease (MRD) which can be quantified with high sensitivity by multiparameter flow cytometry (MFC). The optimum checkpoint for determination of MRD during the course of therapy, however, has not yet been determined. We applied MFC using a comprehensive panel of antibodies to identify leukemia-associated aberrant immunophenotypes (LAIPs) at diagnosis and to quantify MRD by individually selected antibody combinations. The prognostic impact of MRD levels was assessed in comparison to cytogenetics and age. Patients received double induction, consolidation, and maintenance therapies and underwent allogeneic stem cell transplantation if they were younger than 60 years and had a matched related donor. In 286 patients with newly diagnosed and untreated AML MFC-based assessment for the presence of LAIP has been performed. The median percentage of LAIP-positive bone marrow cells at diagnosis was 16.04% (range, 2.54%–76.14%). All individual LAIPs were applied to 26 normal bone marrow samples to estimate sensitivity based on the median percentages of LAIP-positive normal bone marrow cells which ranged from 0.00% to 1.01% (median, 0.02%). A total of 550 follow-up samples has been analyzed in these patients at different checkpoints (CP1, up to day 21 after start of therapy, n=85; CP2, day 22–60, n=122; CP3, day 61–120, n=158; CP4, day 121–365, n=137; CP5, after day 365, n=48). In order to adjust for differences in the percentages of LAIP-positive bone marrow cells at diagnosis the logarithmic difference (LD) between diagnosis and follow-up was calculated for each follow-up sample. The median LDs at the respective checkpoints were: CP1, 2.02; CP2, 2.29; CP3, 2.39; CP4, 2.53; and CP5, 2.81. Separation of patients according to the respective median LDs resulted in differences in event-free survival (EFS; CP1: 21.1 vs. 9.1 months, p=0.0711; CP2: 14.2 vs. 9.3 months, p=0.0095; CP3: 30.9 vs. 13.5 months, p=0.0055; CP4: median not reached vs. 14.1 months, p<0.0001; CP5: median not reached vs. 22.5 months, p=0.0001) and overall survival (OS; CP3: median not reached vs. 21.6 months, p=0.0332; CP4: 90% vs. 53% at 2 years, p=0.0058). Cox analysis using the LDs at the different checkpoints as continuous variables confirmed the prognostic impact on EFS (CP2, p=0.002; CP3, p=0.0003; CP4, p<0.0001; CP5, p<0.0001) and revealed an impact also on OS (CP3, p=0.003; CP4, p=0.001; CP5, p=0.029). Cox regression analysis taking into consideration cytogenetics and age as covariates proved the independent prognostic impact of LD at checkpoints 2 to 5 on both EFS and OS with the exception of LD at checkpoint 2 and OS. In fact, LD at checkpoint 5 was the only parameter independently related to EFS and OS. These data suggest that quantification of MRD by MFC in AML results in powerful and independent prognostic parameters. In particular during the first year of treatment MRD levels provide important prognostic information. Clincal trials should use MRD-based stratification in order to assess the efficacy of early treatment intensification in high-risk AML patients.