Integration of genomic and clinical variables improves the prediction of myelodysplastic syndromes to acute myeloid leukaemia transformation and prognosis.

Myelodysplastic syndromes (MDS) are heterogeneous stem cell disorders with a 30%-40% risk of transformation to acute myeloid leukaemia (AML). This study characterised the genetic and clinicopathological features of 437 newly diagnosed MDS patients. A predictive model for AML transformation was developed, which identified the bone marrow blast percentage, KRAS, STAG2 and TP53 mutations as significant predictors of disease progression. Novel prognostic models for overall survival (OS) and progression-free survival (PFS) were established and validated. The OS model incorporated age, sex and mutations in TP53, NF1, SRSF2, CSF3R, ETV6, CEBPA, BCOR, TET2, JAK2 and SF3B1. The PFS model incorporated age, sex, bone marrow blast percentage and mutations in STAT3, TP53, NF1, CEBPA, ETV6, CALR, DNMT3A, IDH2, GATA2, BCOR, JAK2, TET2 and SF3B1. These models stratify patients into favourable, intermediate-1, intermediate-2 and adverse risk groups, demonstrating superior risk stratification compared to the Revised/Molecular International Prognostic Scoring System (IPSS-R/M). This work provides the first genomic characterisation of a diagnosed MDS cohort in China and establishes the first risk prediction model for MDS-to-AML transformation, alongside novel prognostic models for OS and PFS. These tools offer improved prognostic prediction and potential guidance for therapeutic strategies in Chinese patients with MDS.