Correlation Between Bone Marrow Dysplasia and Genomic Profile in De Novo Acute Myeloid Leukemia (AML): A Study By the ALFA Group

Introduction: AML with multilineage dysplasia (MLD) are included in the WHO subset of "AML with myelodysplasia-related changes" (AML-MRC), together with AML arising from previous MDS or AML with MDS-related cytogenetic abnormalities. In the WHO classification, MLD is defined by dysplasia in at least 50% of the cells in at least two bone marrow (BM) myeloid cell lines. On the other hand, some genetically defined AML subgroups are specifically associated with morphologic changes, but close correlations do not exist for most of these entities. We searched for correlations between BM dysplasia and molecular aberrations in de novo AML patients included in 2 ALFA clinical trials

Methods: BM cytomorphology was retrospectively reassessed in 192 patients with de novo AML (excluding CBF-AML), aged 18 to 70 enrolled in ALFA-0702 (n=123) and ALFA-0701 (n=69) clinical trials in 5 centers. 4 distinct morphologists performed the analysis from BM smears. Dysmegakaryopoiesis (DM), dyserythropoiesis (DE) and dysgranulopoiesis (DG) were quantified (respectively on 30, 200 and 200 cells) using 22 criteria designed by GFHC, which allow better evaluation of cytoplasmic and nuclear dysplasia in all BM lineages. Dysplasia was also evaluated using WHO criteria. NPM1, FLT3, MLL, CEBPA, IDH1, IDH2, WT1, DNMT3A, RUNX1, TET2 and ASXL1 gene mutations and EVI1 gene overexpression were detected by standard methods, as previously published (Renneville et al. Oncotarget 2014).

Results: In the 192 patients analyzed, the incidence of molecular abnormalities was: MLL-PTD 5% (8/155), NPM1 31% (52/170), FLT3-TKD 9% (15/171), FLT3-ITD 19% (34/171), CEBPA double mutated (CEBPA-dm)11% (17/152), EVI1 overexpression 11% (17/152), IDH1 R132 9% (14/146), IDH2 R140 6% (10/159), IDH2 R172 2% (2/92), RUNX1 8% (6/67), DNMT3A 26% (11/43), TET2 12% (5/43) and ASXL1 7% (4/62). DG, DE and DM was evaluable in 59%, 83% and 85% of the patients, respectively. WHO-MLD was identified in 43/192 (22%) patients, and was not significantly associated with any genetic marker, even in AML with normal karyotype (Table 1). On the other hand, when using GFHC criteria, we observed in NPM1 mutated patients a higher % of bi-tri or multi nucleated megakaryocytes (25% vs 10%, p=0.03), of cytoplasmic DG (74% vs 58%, p=0.03); and more dysplasia in other cell lines including eosinophils, basophils, mastocytes, monocytes (p=0.008). In CEBPA-dm patients, lower % of global DG (21% vs 54%, p=0.04) was seen. In EVI1 overexpressing patients, we found a higher % of global DM, of micromegacaryocytes and of hypolobulated megacaryocytes (80% vs 31%, p=0.01; 18% vs 2%, p=0.01 and 19% vs 6%, p=0.001 respectively). In DNMT3A mutated patients, we observed a lower % of bi-tri or multi nucleated megakaryocytes (2% vs 28%, p=0.01) and a higher % of nuclear and cytoplasmic DG (21% vs 2%, p=0.005 and 1.2% vs 0%, p=0.03, respectively). In TET2 mutated patients, we observed less defects in nuclear segmentation and a higher % of abnormal chromatin condensation in granulocytes (1% vs 9%, p=0.02 and 6% vs 0%, p=0.008, respectively).

Conclusion: Presence of WHO-MLD was not significantly correlated with any genetic subgroup. The 22 BM dysplasia parameters designed by the GFHC were evaluable in a majority of patients, and allowed us to find some specific cytomorphologic features in de novo AML with NPM1, CEBPA-DM, DNMT3A, TET2 mutation, or EVI1 overexpression. Those findings suggest that the definition of MLD may be refined by using more in depth quantification of dysplasia, especially with GFCH parameters. This study will be expanded with the inclusion of whole exome sequencing data (ongoing).