Persistent minimal residual disease (MRD) is an independent prognostic marker in acute myeloid leukemia (AML) that has become increasingly important for risk stratification and treatment planning. MRD can be evaluated by multi-parameter flow cytometry (MFC) and molecular methods including PCR-based methods detecting leukemia-specific fusion transcripts and targeted deep sequencing of genes recurrently mutated in AML. However, while MFC is highly operator-dependent, PCR-based methods are only available for a minority of AML patients. In addition, targeted deep sequencing approaches are hampered to distinguish between mutations responsible for preleukemic clonal hematopoiesis and true leukemia-specific mutations.
In this study we aimed for developing a highly sensitive and broadly applicable method for MRD detection in AML by combining MFC-based leukemic cell enrichment followed by mutational analysis.
To identify markers suitable for leukemic cell enrichment, AML samples from 150 newly diagnosed patients were analyzed for 24 surface markers using MFC with a strictly standardized protocol. In 25 cases, samples were also available at relapse. To assess the normal hematopoietic stem and progenitor cell (HSPC) compartments normal bone marrow (BM) samples were obtained from 12 lymphoma patients without any evidence of disease in the BM. For validation of our method, BM samples were prospectively collected from a total of 42 patients in complete remission (CR) after induction chemotherapy. Mutational analysis of sorted leukemic cell enriched samples was done using targeted deep sequencing of 39 recurrently mutated genes using an Ion Torrent sequencing platform. In some NPM1-mutated samples analysis was performed using a mutation-specific digital droplet PCR assay.
A combination of antibodies against CLL-1, TIM-3, CD123 and CD117 was identified to perform best for leukemic cell enrichment by enabling staining of >90% of AML cells in 137 of 150 diagnostic AML samples (91.3%) and in 25 of 25 (100%) relapse samples. In dilution experiments using NPM1-mutated samples and normal BM, leukemic cell enrichment by these markers followed by mutational analysis showed a sensitivity of 10–5 for residual disease detection. In contrast, NPM1-mutations were not detected in cells negative for the marker combination in dilutions higher than 1:100. Using this marker combination for cell sorting allowed a 30- to 250-fold cell enrichment in prospectively collected BM samples of 42 patients in complete remission. In 39 samples DNA quality of sorted cells was sufficient for sequencing. Twenty-one samples tested MRD positive whereas 18 were negative. With a median follow-up of 372 days 71% of MRD positive (15/21) and 28% (5/18) of MRD negative patients relapsed (p = 0.007). Accordingly, median relapse free survival was significantly longer in MRD negative patients (497 vs. 242 days, p = 0.0035). Of note, in 8 out of 10 MRD positive remission samples, in which we also analyzed marker negative cells, mutations were exclusively found in marker positive cells emphasizing the importance of MFC-based cell enrichment prior to sequencing.
MFC-based leukemic cell enrichment using antibodies against CLL-1, TIM-3, CD123 and CD117 followed by mutational analysis is feasible for MRD detection with high sensitivity and informative on relapse risk in AML patients. Standardization of this method as well as its comparability to other approaches of MRD detection need to be evaluated in a multi-center clinical trial.