LAUSR

There are unsolved questions in the field of minimal/ measurable residual disease (MRD) detection by multiparameter flow cytometry (MFC) in acute myeloblastic leukemia (AML). One of them, the value of peripheral blood (PB) testing, is raised repeatedly [1]. A Pubmed search on the topic is, however, exceedingly frustrating and mostly retrieves publications based on the morphological assessment of blast cells in PB [2–4]. The 2007 study by Maurillo et al. [5] was the first to propose pertinent MFC data. In this work, 50 and 48 patients, respectively, were studied in MFC for matched samples of PB and bone marrow (BM) after induction and consolidation of a chemotherapy regimen. At that time, all post-induction PB and BM samples contained detectable MRD. However, the 10 patients who had no detectable PB MRD after consolidation had a significantly better outcome, highlighting the potential value of PB MRD. In 2009, very early detection of peripheral blasts in MFC during the first days of induction was also shown to be of high prognostic value [6], a result confirmed by Yu et al. in 2015 [7]. The second most pertinent work of MRD during and after chemotherapy appeared in Leukemia in 2016, from the Dutch group HOVON [8]. The relevance of MRD assessment in PB on survival was confirmed there on a larger series of 76 evaluable patients. Among them, at a threshold of 0.04% established by ROC curves, respectively, 9/55 and 2/29 patients had detectable PB MRD after induction and consolidation. Their outcome was significantly poorer than that of MRD-negative patients. A good correlation was also noted with BM MRD. In the course of a multicenter research program on AML MRD that was supported by the French Institute of Cancer (Inca), peripheral blood (PB) sampling was recommended for an ancillary investigation, besides that of classical bone marrow (BM) assessment. Results of the latter have been published recently [9], detailing the types of patients and treatments received. This work demonstrated the robustness of a novel analysis algorithm using a patient-tailored protocol established with MFC data of the diagnosis and a fixed set of monoclonal antibodies. Being then directly applied to the same patient’s follow-up samples, this protocol allowed to detect MRD down to a level of 5 × 10. Patients who never had an MRD above this threshold during follow-up fared significantly better than those having at least one “positive” follow-up sample. Here we report the application of the same strategy for the 96 patients of this cohort who also benefited from PB sampling and who are representative of the whole 256