A DNA methylation–based risk score improves prognostic stratification in myelodysplastic neoplasms

Abstract

Myelodysplastic neoplasms (MDS) are clonal hematopoietic disorders characterized by ineffective hematopoiesis, cytopenias, and variable risk of progression to acute myeloid leukemia (AML). Although mutational data are already implemented into the most recent MDS risk models, (i.e. IPSS-M), clinical heterogeneity continues to complicate decision making. Aberrant DNA methylation is considered an important contributor to malignant transformation in MDS. We here provide evidence that methylation adds additional prognostic information with respect to AML transformation-free survival (AML-TFS). Genome-wide methylation data from 68 bone marrow samples were used to develop a model predicting AML-TFS, and findings were validated in an independent cohort ( n = 172). Furthermore, gene expression differences between risk groups were assessed. A 55-CpG methylation signature stratifies MDS patients into distinct risk groups and predicts transformation in a manner that appears largely independent from established clinical, cytogenetic, and mutational parameters. A CpG located in the promoter of SPAG6 , identified as the most important contributor to the 55-CpG methylation score, showed hypomethylation associated with inferior AML-TFS. Higher-risk methylation profiles, such as SPAG6 , a cancer–testis antigen (CTA), are associated with the coordinated upregulation of other CTAs and developmentally restricted genes, suggesting possible reactivation of early developmental and stemness-related transcriptional programs. Notably, the methylation signature showed preliminary concordance in a limited set of paired peripheral blood/bone marrow samples, suggesting potential as a minimally invasive prognostic tool, although this requires confirmation in larger sample sets. In summary, a compact DNA methylation signature shows promising results as independent predictor of AML-TFS in MDS and warrants prospective validation in larger patient cohorts.

The authors' abstract, as published at the source. Clinical Epigenetics, 2026 · DOI ↗