Myelodysplastic neoplasms (MDS) are a group of clonal hematopoietic disorders characterized by dysplasia in one or more myeloid cell lines, ineffective hematopoiesis, and an increased risk of progression to acute myeloid leukemia (AML). The transition from MDS to AML is a pivotal event that significantly impacts patient outcomes. While the blast percentage in bone marrow is a well-established predictor of leukemic conversion, it often overshadows other potentially significant prognostic factors.

This study aims to identify independent predictive factors for leukemic conversion in patients with MDS, adjusting for the blast percentage in the bone marrow.

We conducted a retrospective cohort at a single reference center, involving patients diagnosed with MDS between 2004 and 2024. We included patients with available data on survival and leukemic conversion outcomes. Missing data for other variables were handled using the ’cart’ multiple imputation method, following confirmation of non-completely at random missingness through Little's test. We performed multivariate Cox proportional hazards (PH) regression to model leukemic conversion, adjusting for multiple relevant clinical parameters, including blast percentage, allowing for the evaluation of the independent effects of other variables. All statistical analyses were conducted using R software, utilizing the ’mice’ package for imputation and the ’survival’ package for survival analysis.

162 patients were included in this study. In the multivariable Cox PH regression, we adjusted for blast percentage, sex, age at diagnosis, cytogenetic prognostic, presence of dysplasia in specific lineages in bone marrow, complete hemogram parameters, iron-related parameters, and transfusion burden. Optimal cut-offs were defined based on clinical relevance. In our analysis, only blast percentage, male sex, transfusion burden, hemoglobin count lower than 10 g/dL and serum iron higher than 65 μg/dL were significant (p < 0.001, < 0.001, 0.03, 0.012, and 0.03, respectively). Other than blast percentage, the most discriminative variable was sex, with males having a HR of 10.46 (95% CI: 3.31-33.08) for AML conversion in comparison to females. Transfusion burden (HR = 6.16, 95% CI: 1.85 - 20.52) and hemoglobin higher than 10 ng/dL (HR = 4.31, 95% CI: 1.36 - 13.65) were also relevant predictors for leukemic conversion, while serum iron higher than 65 μg/dL (HR = 0.28, 95%CI: 0.09-0.86) was a protective factor.

Male sex has been associated with worse outcomes in MDS, such as a higher incidence of transformation to AML, which is partly due to a higher prevalence of high-risk somatic mutations in men. Transfusion burden has also been associated with AML conversion, as iron overload can contribute to leukemic transformation due to the accumulation of free iron radicals, which can cause DNA damage and promote leukemogenesis. However, the protective impact of high serum iron may relate to premature death, as these patients may be less susceptible to AML conversion. Similarly, although higher hemoglobin levels have been associated with better survival, its impact on leukemic transformation in our study may be due to death as a competing risk.

Our study identified important discriminative factors for AML transformation in MDS patients when adjusting for blast percentage. These factors will allow for a more refined risk stratification in this population.