P448: prognostic significance of measurable residual disease detection by flow cytometry in autologous stem cell apheresis products of patients with aml

Background: Measurable residual disease (MRD) detected by multiparameter flow cytometry (MFC) in bone marrow (BM) after two cycles of high-dose induction chemotherapy has strong prognostic value and predicts relapse when applied to patients with acute myeloid leukemia (AML) in complete remission (CR). However, still around 30% of MRD-negative patients relapse which is partly caused by limited sensitivity. Measuring MRD in autologous stem cell apheresis products (ASCAP), may increase sensitivity due to the enrichment of progenitor cells by granulocyte colony-stimulating factor (G-CSF) stimulation. Studies evaluating the applicability of MRD measured in ASCAP enriched with progenitor cells and possibly leukemic cells have been reported for acute lymphocytic leukemia and multiple myeloma (MM), but large studies are lacking for AML. Aims: In this study we aim to assess the prognostic significance regarding event-free survival (EFS) and overall survival (OS) for MRD measured using MFC-MRD in ASCAP samples in AML patients. Methods: ASCAP-MRD was measured in 229 de novo AML patients in CR following two cycles of standard induction chemotherapy in various HOVON-SAKK multicenter prospective phase III clinical trials with or without the addition of an experimental agent. Paired BM specimen was acquired for MRD in 185 patients. In addition, we analysed the ASCAP of healthy stem cell donors (n=10) and MM patients (n=4) to investigate the role of G-CSF stimulation on the potential induction of aberrant phenotypic markers on healthy progenitor cells, that could mimic a leukemia-associated phenotype (LAIP). For all samples, we used the same MFC-MRD assay as for regular BM-MRD samples. In order to quantify the prognostic value of measuring MRD in ASCAP, we performed the R MaxStat package based on LogRank values to determine the optimal cut-off. To demonstrate if MRD in ASCAP has additional prognostic value independent from MRD in BM, we have assessed the concordance and correlation between both sample types and associations with EFS and OS using Cox proportional hazard model. Results: In healthy donors and non-leukemic controls a LAIP was found in 3 cases. These markers (CD22+, CD13+/CD33- and CD14+/HLA-DR-) were thereby related to G-CSF exposure and excluded from further analysis in the ASCAP-MRD AML cases. A valid ASCAP-MRD result could be measured in 223 patients. The optimal cut-off for EFS was 0.1% of total WBC, which is similar to BM-MRD. In this cohort, 29/223 (13%) patients were ASCAP-MRD positive. The presence of MRD in ASCAP was associated with worse 3-year EFS (41% vs 58%, p=0.0025; Figure 1A) and inferior 5-year OS (49% vs 67%, p=0.0047; Figure 1B). A paired BM-MRD result was available for 185 (83%) patients. BM-MRD positive patients had worse outcome compared to MRD-negative patients in terms of EFS (3-year 35% vs 60%; p=0.004) and OS (5-year 53% vs 64%; p=0.022), respectively. ASCAP- and BM-MRD results were concordant for 93% (172/185) of patients, in which 83% (154/185) were MRD-negative and 10% (18/185) were MRD-positive. The non-concordant group of 13 patients consisted of 5 patients who were ASCAP MRD-pos and BM MRD-neg. When the four groups were compared, patients who were MRD-positive in both specimen had the lowest EFS (p=0.029) and OS (p=0.035). Summary/Conclusion: MFC-MRD detected in ASCAP is significantly associated with a lower EFS and OS. High concordance was found between paired ASCAP-MRD and BM-MRD. While BM has better discriminating power compared to ASCAP and therefore remains to be the primary choice for MRD assessment, data indicates that MRD assessment in ASCAP can be an option when BM is not available.Keywords: Autologous hematopoietic stem cell transplantation, Granulocyte colony-stimulating factor (G-CSF), Minimal residual disease (MRD), Acute myeloid leukemia