We previously demonstrated that the presence of residual leukemic blasts (MRD) in patients in morphologic remission was predictive of subsequent leukemic relapse. More recently, we developed a four-color multidimensional flow cytometric method using standardized rather than patient specific panels, distinguishing the abnormal cells based on “difference from normal”. The feasibility of utilizing this method, which does not require a diagnostic specimen and is applicable even if the phenotype of the leukemic clone changes was tested in a blinded study as part of the recently completed COG AML pilot AAML03P1 that enrolled 341 patients. 302/341 patients in the study consented to participate in the biology portion of the study. 223 (74%) had evaluable specimens at the end of induction I, 191 (87%) of which achieved a morphologic CR at the end of induction I. Of the 191 patients in CR at the end of induction I, 48 (25%) had evidence of MRD. The level of MRD ranged from 0.02% to 3% with 70% of patients having an MRD level between 0.1% to 1%. Relapse risk was assessed in those with and without MRD. Those with evidence of MRD were at significantly higher risk of subsequent relapse, with a relapse-free survival (RFS) from end of induction I of 36% for those with MRD compared to 70% for those without MRD (p < 0.001). The corresponding overall survival at 2 years from induction I was 63% and 86% for those with and without MRD (p=0.003). 181 patients who were in morphologic CR at the end of induction II had evaluable samples for MRD analysis. MRD was detected in 36/181 patients (20%) and presence of MRD was associated with a RFS from end of induction II of 34% compared to that of 70% in those without MRD (p<0.001). Corresponding overall survival at 2 years from induction II was 56% and 83% for those with and without MRD (p=0.009). We determined whether MRD positive patients who clear their disease have an improved outcome. There were 16 patients with detectable MRD at the end of course I with undetectable MRD at the end of course II. Relative risk of relapse for these patients was 40% vs. 30% for those who were MRD negative at both time points. Data from end of induction I and Induction II were combined to maximize our ability to identify those at high risk of relapse. This study demonstrates that flow cytometric based MRD can be utilized in a multi-institutional setting to accurately identify those at high risk of relapse. Multivariate analysis of the prognostic significance of MRD in the context of other prognostic factors including cytogenetics (CBF, etc.) and molecular (FLT3/ITD) prognostic factors will be presented.
Disclosure: No relevant conflicts of interest to declare.