Reference bone marrows: outlining the MPR profile
The median age of RBM subjects was 53 years old (range 18 to 74), and 25 (52%) were female. MFC data analysis was performed in triplicates by three independent flow experts and no significant difference in MPR was found (p = 0.41). The mean MPR was 2.50 (± 2 standard deviations 1.25–3.75). The median MPR was 2.55 (range 1.27 to 3.80). MPR according to the origin of samples, i.e. healthy BM donors or patients with non-hematological malignancies, are shown in supplementary Fig. 1. For the rest of the study, a balanced MPR (bMPR) was set as ranging from 1 to 4. The MPR was considered unbalanced (ubMPR) if < 1 or > 4.
Patient characteristics
The whole cohort enrolled 206 patients with a median age of 62 years old (range 20–79), 42 of them (20.4%) being 70 or older. Ninety-six patients (46.6%) were female. Using the ELN 2010 risk stratification (30), 19 (9.2%) patients were classified as favorable risk, 153 (74.3%) as intermediate risk and 34 (16.5%) as adverse risk (8). The NPM1/FLT3-ITD status was obtained for 201 (97.6%) patients showing that 52 (25.9%) had a favorable prognosis (i.e. mutated NPM1 without FLT3-ITD or with FLT3-ITDlow).
Patients had required one (n = 182; 88.3%) or two cycles (n = 24; 11.7%) of intensive induction to reach CR/CRi. The median time between induction and evaluation was 35 days (range 21–92).
MPR at baseline
MFC data on BM samples were available for 195 (94.6%) patients at diagnosis. The 11 (5.4%) remaining had MFC data from PB. All MPR at baseline, evaluated on BM samples, were unbalanced (p < 0.0001) (Fig. 2). At baseline, ubMPR > 4 (n = 107 ; 54.9%) were more frequent in patients diagnosed as AML with minimal differentiation (AML0) (17/17, 100%) or AML with maturation (AML2) (31/54, 69%) whereas ubMPR < 1 (n = 88 ; 45.1%) were more recurrent in acute monoblastic and monocytic leukemia (AML5) (23/31; 74%) according to the morphological FAB classification (33) (Supplementary Fig. 2).
Post-induction MPR
Post-induction MPR were lower in patients in CR compared to CRi (median 2.90 vs 9.73, p < 0.0001) (Fig. 3A). For patients reaching CR, the proportion of post-induction bMPR was higher compared to those achieving CRi (61.9% vs 25.0%, p < 0.0001) (Fig. 3B). Patients with post-induction ubMPR needed more often 2 intensive induction cycles (n = 18) to reach CR/CRi than patients with bMPR (n = 6) (16.7% vs 6.1%, p = 0.018).
Post-induction MPR and baseline patient characteristics
Patients were divided in two groups according to their MPR status at the end of induction: balanced (n = 98) vs unbalanced (n = 108). Disease characteristics at baseline (Table 1), therapeutic lines and post-induction biology were then compared between these two groups.
There was no difference at baseline regarding ELN 2010 risk stratification (30) (p = 0.08), cytogenetic MRC 2010 stratification (34) (p = 0.10), ELN 2017 risk stratification (8) (p = 0.21) nor NPM1/FLT3-ITD status (p = 0.88) (Table 1).
Post-induction complete blood counts showed higher polymorphonuclear (mean 4.35x 109/L vs 2.8x109/L, p = 0.0002), monocyte (mean 0.90x109/L vs 0.64 x109/L, p = 0.0019) and platelet (mean 326 x109/L vs 175 x109/L, p < 0.0001) counts, together with less myelemia (mean 1.94% vs 5.41%, p < 0.0001) in patients with bMPR (Supplementary Fig. 3). Post-induction lymphocyte counts (mean 0.86x109/L vs 0.88x109/L, p = 0.74) and hemoglobin levels (mean 10.38 g/dL vs 10.14 g/dl, p = 0.10) were similar whatever the MPR group (Supplementary Fig. 3).
Post-induction MPR and clinical outcome
The median OS for patients with post-induction ubMPR was 36 months but was not reached for those with bMPR (HR, 2.96; 95%CI, 1.81–4.84; p < 0.0001, Fig. 4A). One year and 3-year OS were respectively 82.1% and 44.4% in patients with post-induction ubMPR vs 91.6% and 80.2% in patients with post-induction bMPR (p < 0.0001). In Cox model multivariate analysis, a status of ubMPR was significantly and independently associated with a worse OS (adjusted HR (aHR), 2.72; 95%CI, 1.42–5.20; p = 0.003, Table 2). Mutated NPM1 without FLT3-ITD or with FLT3-ITDlow (aHR, 0.36; 95%CI, 0.16–0.84; p = 0.017) and Allo-HSCT (aHR, 0.34; 95%IC, 0.18–0.65; p = 0.001) retained a significant positive impact on OS. ELN 2010 adverse risk held a statistically significant independent negative impact on OS (aHR, 2.99; 95%CI, 1.67–5.37; p = 0.0002).
The median RFS was 18 months and not reached for patients with post-induction ubMPR or bMPR (HR, 2.11; 95%CI, 1.39–3.18; p = 0.0004, Fig. 4B), respectively. One-year and 3-year RFS were respectively 57.7% and 38.7% in patients with post-induction ubMPR and 76.0% and 64.4% in those with bMPR (p = 0.0004). In multivariate analysis, post-induction ubMPR was significantly and independently associated with a shorter RFS (aHR, 2.27; 95%CI, 1.35–3.83; p = 0.002, Table 2). Both mutated NPM1 without FLT3-ITD or with FLT3-ITDlow (aHR, 0.30; 95%CI, 0.15–0.60; p = 0.0007) and Allo-HSCT (aHR, 0.44; 95%IC, 0.26–0.75; p = 0.003) still retained a significant positive impact on RFS. Patients classified as ELN 2010 adverse risk had a significantly worse RFS (aHR, 1.95; 95%CI, 1.16–3.29; p = 0.012).
The prognostic value of post-induction MPR within ELN 2010 subgroups showed worse outcomes for patients with ubMPR classified as intermediate 1 or 2 (n = 153 patients), for both OS (HR, 2.69; 95%CI, 1.45–4.98; p = 0.0018) and RFS (HR, 2.22; 95%CI, 1.33–3.70; p = 0.0021) (supplementary Fig. 4). Conversely, no difference was found for patients stratified in ELN 2010 adverse (OS, p = 0.0642; RFS, p = 0.1369) or favorable (OS, p = 0.49; RFS, p = 0.74) subgroups.