ELN vs. IPSS-M in MDS/AML: Which Prognostic System Should Guide Clinical Decision-Making?

The recognition of myelodysplastic syndrome/acute myeloid leukemia (MDS/AML) as a distinct overlap entity by the International Consensus Classification (ICC) has introduced new challenges in prognostication and therapeutic decision-making. Defined by 10–19% blasts in the absence of AML-defining genetic abnormalities, MDS/AML occupies a biological continuum between myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) and is characterized by unique molecular features, patterns of clonal evolution, and clinical behavior that are not fully captured by prognostic systems originally developed for either disease. Consequently, the emergence of this entity has renewed interest in how existing classification and risk stratification frameworks should be applied to overlap disease. This review examines the biological basis of MDS/AML and critically evaluates the strengths and limitations of the Molecular International Prognostic Scoring System (IPSS-M) and the European LeukemiaNet (ELN) classifications. Available evidence supports IPSS-M as the preferred framework for baseline disease-specific prognostic assessment, with preserved prognostic discrimination in this entity. In contrast, direct application of ELN 2022 results in substantial adverse-risk compression and limited prognostic discrimination. Emerging evidence suggests that disease-specific recalibration and treatment-contextual frameworks, including ELN 2022 modified and ELN 2024 Less-Intensive, may provide additional information, although validation specifically in MDS/AML remains limited and largely retrospective. Collectively, current evidence supports a complementary rather than competitive approach. IPSS-M should provide the disease-specific prognostic foundation, whereas ELN-based assessment may add AML-oriented molecular and treatment-contextual information relevant to therapeutic planning. Neither framework should independently determine treatment intensity. Instead, optimal management should integrate disease-specific prognosis with patient fitness and transplant eligibility, molecular characteristics, treatment context, and dynamic response assessment. Emerging approaches incorporating clonal hierarchy, single-cell analysis, epigenomic and multi-omic profiling, measurable residual disease, and artificial intelligence (AI)- and machine learning (ML)-based predictive models may further refine prognostic assessment, improve prediction of treatment response, and facilitate increasingly individualized therapeutic strategies in MDS/AML, although their clinical application remains investigational.