Acute myeloid leukemia (AML) is generally regarded as a stem cell disease, known as leukemic initiating cells (LIC), which initiate the disease and contribute to relapses. Although the phenotype of these cells remains unclear in most patients, they are enriched within CD34+CD38- compartment. In core binding factor (CBF) AML, the cytogenetic abnormablities are also existed in LIC. The aim of this study was to determine the prognostic power of minimal residual disease measured by fluorescence in situ hybridization (FISH) in flow sorted CD34+CD38- cells (FISH+CD34+CD38- population) at different period during the therapy. Thirty-six patients under 65 years of age with de novo CBF AML and treated with CHAML 2010 protocol were retrospectively included in this study. FISH efficiently identified the LICs (FISH+CD34+CD38-) in the CD34+CD38- population. The last follow-up was March 31, 2013, and the median follow-up was 336 days (range: 74-814 days). 33 patients with complete remission (CR) were eligible for the study, and 23 patients (23/33, 69.7%) with t (8;21) or AML1/ETO, and the remaining (10/33, 30.3%) with inv(16)/t(16;16) or CBFβ/MYH11. Flow-cytometry based FISH (F-FISH) procedure was performed at diagnosis, before every cycle of consolidation therapy, and every 3 months during follow-up. The FISH+ percentage at diagnosis constituting an average of 2.1% (range: 0.01%-27.5%) of the blast cells and 64.6% (range: 14%-87.8%) of the CD34+CD38- cells. Before the consolidation, FISH+CD34+CD38- population was detected in 13/33 (39.4%) patients. At this checkpoint, we have found the existence of FISH+CD34+CD38- population had prognostic value for the end points relapse free survival (RFS, 12% versus 68%, P=.008), and retained prognostic significance for RFS in multivariate analysis. Furthermore, the detection of FISH+CD34+CD38- before consolidation was found to be significantly associated with decreased OS. (11% versus 75%, P=.0005) Minimal residual disease (MRD) detected with F-FISH had a prognostic value at an earlier checkpoint when compared with flow cytometry and RT-PCR. Meanwhile, the concordance of flow cytomety, RT-PCR and F-FISH was investigated in the same patient cohort. 14 (70%) of 20 samples with detectable fusion transcripts by PCR did not have detectable leukemic cells by F-FISH. Therefore, the concordance for PCR and F-FISH was 63.7%. The concordance of FC and F-FISH was 64.3%: in 40 samples MRD was detected by both methods and in 61 samples MRD was ruled out by a negative result with the tests. With further analysis, the discrepancies among MRD detected with different MRD monitoring approaches before consolidation and after the first consolidation therapy contribute to 84% of the disconcordance. In summary, the detection of FISH+CD34+CD38- cells before consolidation therapy was significantly correlated with long-term survival in de novo CBF AML patients. F-FISH might be easily adopted as MRD monitor approach in clinical practice to identify patients at risk of treatment failure from the early stage during therapy. Disclosures:No relevant conflicts of interest to declare.