Day-21 bone marrow morphology in patients with FLT3-mutated AML incorrectly designates persistent leukaemia

Background: In the registration trial for midostaurin (RATIFY) in intensively-treated FLT3-mutated acute myeloid leukaemia (FLT3-AML), day-21 bone marrow biopsy (D21-BMB) was protocolled. Those with persistent leukaemia (PL), as defined by a morphologic blast count ≥ 5% in a cellular (> 20%) marrow received a second induction (∼25% of participants).1 However, a similar D14-BMB early response assessment strategy is known to poorly predict overall induction outcomes.2 Our study aimed to assess the predictive value of D21-BMB on induction outcomes of our local FLT3-AML population treated intensively plus midostaurin. Methods: Between August 2018 and March 2022, all patients with newly diagnosed FLT3-AML at Monash treated with 7+3 (idarubicin) plus midostaurin who had D21-BMB (± 24H) were included in the analysis. To improve objectivity, D21-BMBs were morphologically assessed by two independent haematopathologists blinded to the original diagnostic reports. The predictive value of D21-BMBs was assessed against final induction remission status; determined from results of D21 or D28-BMB (where available), whichever was done later, in conjunction with peripheral blood count recovery and blast clearance. The RATIFY definition of PL was strictly used, while the 2022 ELN response criteria3 was used to designate complete remission (CR), CR with incomplete haematologic recovery (CRi), morphologic leukaemia-free state (MLFS) and non-evaluable marrows. This study was approved by the Monash Health Ethics Committee (RES-21-0000015Q-72713). Results: 15/144 newly diagnosed AML during the study period were FLT3-mutated and treated with 7+3+midostaurin: 9/15 had D21-BMB assessments (7/9 had a repeat D28-BMB) while 6/15 did not (1 induction death, 5 done at later time-points). Amongst the D21 original reports, 3/9 (33%) were classified as having PL, 4/9 CRi or MLFS and 2/9 non-evaluable due to marrow aplasia. Independent, duplicated, blinded review of D21-BMBs of the 3 patients originally assigned to have PL yielded an overall concordance of 89% (100%, 100%, 67%). At D28, all 7 patients, including the 3 original PL cases (who were not prescribed re-induction, based on review of ancillary flow cytometry and molecular results) attained CR as their final induction outcome. With the 2 patients with only D21-BMBs achieving subsequent count recovery and blast clearance, all 9 patients achieved CR at the end of 7+3+midostaurin induction.