Given the usefulness of measurable residual disease (MRD) in evaluating treatment efficacy and to predict relapse, MRD monitoring is becoming more frequent for the management of patients with acute myeloid leukemia (AML). Therefore, the accuracy of the methods is essential to obtain reliable results in order to incorporate MRD into therapeutic protocols. MRD AML by multiparametric flow cytometry (MFC) requires complete standardization of processes to ensure optimal results. In addition, clinical validation of the MRD assessment strategy is essential. This study aimed to retrospectively evaluate the clinical effectiveness of a previously analytically validated 10-color MFC protocol to detect AML MRD.

Databases from the flow cytometry laboratory and from the bone marrow transplantation service as well as information from the electronic medical records of AML patients were accessed. Inclusion criteria: patients with MRD evaluated by the validated 10-color MFC protocol, with at least four months of follow-up after the last MRD assessment or who died due to leukemia relapse before the end of this 4-months follow-up. The samples included should have at least 500,000 evaluable CD45+ cells. All levels of MRD positivity were considered for survival analysis. Eighty-two patients with non-promyelocytic AML with a median age of 38 (10-69) years were included in the cohort; 43 were female. Seventy-six patients underwent allogeneic hematopoietic stem cell transplantation (alloHSCT) and six were treated with intensive chemotherapy. The strategy to validate the effectiveness of the 10-color MFC protocol in detecting AML MRD was the concordance of MRD results with patients’ expected outcomes. Clinical sensitivity and specificity, positive and negative predictive values, and diagnostic concordance were calculated even for patients with MRD evaluations post-alloHSCT and after the second cycle of induction chemotherapy. Overall survival (OS) and relapse free survival (RFS) were estimated using the Kaplan-Meyer method and the cumulative incidence method was used to calculate the incidence of relapse (CIR). p-values ≤ 0.05 were considered statistically significant. The results showed high sensitivity (87%), specificity (97%), positive predictive value (78%), negative predictive value (98%) and diagnostic concordance (95%) of the 10-color MFC protocol. The two-year OS, RFS, and CIR of pre-HSCT patients (N = 68) according to MRD < or > 0.1% were 76.4% vs 51.5% (p = 0.117), 79.8% vs 5.3% (p = 0.006), and 53% vs 2% (p < 0.001), respectively. Regarding post-HSCT and after 2nd cycle of intensive chemotherapy patients (N = 74), OS and RFS were higher in patients with MRD < 0.1% vs > 0.1% (75.5% vs 50% p = 0.008%) and (78.4% vs 14.3% p < 0.006) respectively, with smaller CIR (3% vs 6% p < 0.001). In conclusion, objective parameters were used in this study to calculate the clinical sensitivity and specificity of the MRD method. However, the aspects such as the clonal evolution, the MRD kinetics and the therapeutic interventions have impact on sensitivity of MRD results and should be considered for the interpretation of AML relapse. Despite the small cohort and short follow-up time, the 10-color MFC protocol for AML MRD showed high correlation of levels of MRD with clinical outcome and proved to be useful for clinical decision-making.