Modification of the ELN Classification 2022 Refines Risk Assessment in MDS/AML Patients

Background: The International Consensus Classification (ICC) introduced MDS/AML as a novel myeloid disease entity defined by 10-19% blasts in the absence of AML-defining recurrent genetic abnormalities. MDS/AML patients should be eligible for AML-like treatment approaches in clinical trials. It has recently been shown that the AML-based risk classification according to European Leukemia Net (ELN) 2022 criteria fails in MDS/AML while an MDS-based risk classification according to Molecular International Prognostic Scoring System (IPSS-M) was applicable (Huber S et al. EHA 2023). Our aim was to refine the ELN 2022 criteria based on genome sequencing data to better reflect the outcome in this overlap category of MDS and AML.

Methods: Non-therapy-related cases of 403 MDS/AML and 686 AML patients classified according to ICC were included. Bone marrow samples were analyzed by cytomorphology, cytogenetics and targeted NGS panel sequencing (median coverage 1500x). Overall survival (OS) was assessed in all AML patients and in a subcohort of 137 MDS/AML patients (median follow-ups: 5.6 and 10.2 years, respectively) in whom also whole genome sequencing (median coverage 100x) was performed.

Results: MDS/AML patients had a median age of 74 years (female/male: 153/250). They were subclassified according to ICC as MDS/AML with mutated TP53 (14%), with myelodysplasia-related (MR) gene mutations (67%) or MR cytogenetic abnormalities (5%), or not otherwise specified (14%). Cytogenetic aberrations were detected in 173/403 (43%) cases including del(5q) (18%), -7/del(7q) (13%), and complex karyotypes (16%). Molecular aberrations were detected in 376/403 (93%) cases with a median number of 3 mutations. Most frequent gene mutations were ASXL1 (40%), TET2 (32%), SRSF2 (32%), and RUNX1 (29%); TP53 was mutated in (14%) (details in Figure 1A). Grouping of the MDS/AML cohort according to the ELN 2022 guidelines showed: no MDS/AML patient fulfilled criteria for the favorable ELN risk group, only 14% were classified as intermediate risk, and the vast majority (86%) as adverse risk. The survival of MDS/AML patients substantially differed from a bona fide AML cohort, in particular for adverse risk (Figure 1B left).

To more adequately risk stratify MDS/AML patients, we aimed to modify current ELN criteria. Notably, in multivariate Cox regression analysis in addition to -7/del(7q) (hazard ratio/HR: 5.5) and mutated RUNX1 (HR: 2.2), also mutations in EZH2 (HR: 2.2) and in SF3B1 (HR: 0.3) remained significant prognostic markers for OS (all p<0.05). Following this, we defined three hierarchically defined risk groups for MDS/AML patients i) adverse (criteria: complex karyotype, -7/del(7q), mutation in TP53, RUNX1, EZH2), ii) intermediate (criteria: del(5q), mutation in ASXL1, SRSF2, U2AF1, ZRSR2, BCOR or STAG2) and iii) favorable (criteria: sole SF3B1 mutation or none of the abnormalities defined above). Overall, 149 MDS/AML patients (37%) were re-grouped compared to ELN 2022 categories resulting in 49% adverse, 35% intermediate and 16% favorable cases. Survival analysis revealed strong prognostic separations of the three new MDS/AML risk categories with median OS of 1.3 years for adverse, 3.1 years for intermediate and 8.1 years for favorable (overall p<0.001; Figure 1B right). The general survival prediction within MDS/AML was improved in the modified version compared to the original ELN 2022 (corrected Akaike information criterion using Cox models: 876 for modified vs. 902 for ELN 2022). Of note, the outcome of adverse risk MDS/AML according to the modified ELN was comparable to the OS of adverse risk AML according to the original ELN (median OS: 1.3 vs 0.7 years; p=0.304).

Conclusion: ForMDS/AML patients, AML-based risk classification according to ELN 2022 guidelines is not applicable as the vast majority of cases are categorized into the adverse risk group based on mutations in splicing factor genes and other secondary-type mutations. Modification of the ELN risk categories reflects the partly MDS-based biology of the disease and substantially improves the prognostic stratification. Most importantly, patients with adverse outcome potentially benefitting from inclusion into clinical trials for novel substances are identified. The validity of the modified risk classification for MDS/AML patients undergoing specific treatment regimens remains to be confirmed in forthcoming studies.