Abstract
Abstract 760
Currently, the most important prognostic factors for acute myeloid leukemia (AML) include molecular aberrancies and karyotype of the leukemic blasts. Although these factors have showed to be of utmost importance in upfront risk stratification in current treatment schedules, the treatment outcome of patients within as such defined risk groups is still quite heterogeneous. Therefore, there is an unmet need for therapy-dependent prognostic factors which can be implemented into risk-adapted treatment strategies. Minimal residual disease (MRD) frequency is such a parameter. MRD cells are considered responsible for the outgrowth of AML after treatment, leading to a relapse in 30–40% of the patients in complete remission (CR). In this study, we are the first to report prospective multicenter data on the prognostic impact of MRD frequency in adult AML. In our retrospective study (N.Feller et al. Leukemia 2004), we explored which cut-off points for percentage of MRD cells would define MRD positive (levels above cut-off, MRD+) patients with a relatively poor prognosis, from MRD negative (levels below cut-off, MRD-) patients who showed a longer overall and relapse-free survival (OS and RFS). In search for the most optimal cut-off level which can be used for clinical purpose in risk stratification-directed therapy, we used these cut-offs to evaluate the prognostic value in the current prospective setting. Diagnosis and follow-up samples were collected of 462 patients treated uniformly according to the HOVON/SAKK42a protocol (www.hovon.nl) and MRD frequency was assessed blindly without knowledge of clinical course. MRD detection was accomplished by immunophenotyping by flow cytometry (FCM) through aberrant expression of markers on AML blasts. Together with the expression of normal immature cell markers and/or myeloid lineage markers, this offers a leukemia associated phenotype (LAP). Each LAP was individually designed for each patient in diagnosis bone marrow (BM) or peripheral blood. Subsequently, BM samples obtained during follow-up were analysed for the presence of LAP-positive cells. MRD frequency was expressed as a percentage of leukocytes. The median MRD frequencies of patients in clinical CR after first induction cycle (n=164), second induction cycle (n=182) and consolidation (n=121) were 0.040%, 0.022% and 0.020%, respectively. The cut-off levels for MRD frequency as defined retrospectively were all significant in the identification of patients with adverse (MRD+) and favourable (MRD-) OS and RFS, respectively. After the first cycle, the most significant cut-off was 0.8%, leading to 17 MRD+ patients who showed a median RFS of only 8.6 months, while 147 MRD- patients had a median RFS of >47 months (p=0.003,A). The relative risk of relapse (RR) was 2.9 (95% c.i. 1.4–6.0, p=0.004). After the second induction cycle, a cut-off level of 0.06% was most significant. Above this cut-off, 49 patients showed a median RFS of 7 months, while 133 MRD- patients showed a RFS of more than 47 months (p<0.00001, fig B). The RR was 3.2 (95% c.i. 2.0–5.0, p<0.00001). After consolidation therapy, 11 MRD+ patients with extremely poor prognosis were identified (median RFS 7.3 months vs. >47 months for 110 MRD- patients, p<0.00001, fig C), with a RR of 10.6 (95% c.i. 4.9–22.8, p<0.00001). Multivariate analysis was performed with conventional prognostic factors for AML: cytogenetic risk groups and time to achieve CR. After every cycle of therapy, MRD frequency was an independent prognostic factor for RFS after all cycles (1st cycle: p=0.010, 2nd cycle and consolidation p<0.00001) and for OS after 1st (p=0.023) and 2nd induction cycle (p=0.010). In this prospective multicenter study, already after first induction cycle, MRD detection by FCM was an independent significant factor in the identification of poor prognostic patients. In future treatment studies, risk stratification, e.g. for allogeneic stem cell transplantation, should not only be based on risk estimation determined at diagnosis, but also on MRD frequency as a therapy-dependent prognostic factor. This work was supported by Netherlands Cancer Foundation KWF.
No relevant conflicts of interest to declare.
This icon denotes an abstract that is clinically relevant.
This feature is available to Subscribers Only
Sign In or Create an Account Close Modal