Application of the International Working Group MDS Molecular Taxonomy Classes to a Single Institution Cohort Open Access

1. The IWG MDS Molecular Taxonomy can be applied in real-world hematopathology practice, and molecular classes are largely stable over time.

2. In a cohort receiving modern therapies, several MDS molecular classes had distinct clinicopathologic features and outcomes.

Hematopathology classification increasingly relies on genetics, but most myelodysplastic syndromes/neoplasms (MDS) remain classified based on morphologic criteria such as blasts and dysplasia. Recently, the MDS International Working Group (IWG) developed an MDS molecular taxonomy using a large international cohort. Here, we applied this taxonomy to a single-institution cohort of 392 patients with MDS and oligomonocytic CMML (oCMML) receiving modern standard of care. Cases were also classified per International Consensus Classification (ICC) and WHO 5th edition (WHO5). All patients were assigned to a molecular class; TP53-CK (29%), IDH-STAG2 (14%), SF3B1 (9%), and No-event (6%) were the most frequent. Clinical presentation varied across classes, with oCMML being frequent in bi-TET2 (42%, p<0.001), EZH2-ASXL1 (50%, p=0.005), and CCUS-like (33%, p=0.011). Molecular classifications were stable; only 18% changed upon follow-up. Molecular classes showed significant differences in overall survival (OS, p<0.0001). On multivariable analyses (MVAs) that included stem cell transplant, longer OS was seen with SF3B1 (HR 0.266 p=0.0003), del5q (HR 0.363, p=0.015), and biTET2 (HR 0.479, p=0.0495); shorter OS was seen with TP53-CK (HR 2.766, p=0.0002) and -7/SETBP1 (HR 2.465, p=0.015). In MVAs for WHO5 and ICC entities, oCMML2 was the only morphologically-defined entity that affected OS (WHO5: HR 3.460, p=0.0009; ICC: HR 2.644, p=0.018). For predicting OS, IWG molecular classification had a Harrell’s C-index of 0.75, compared to WHO5 and ICC Harrell’s C-indices of 0.74 and 0.73, respectively. Altogether, these results show that the IWG molecular taxonomy can be applied in routine practice, with several classes exhibiting distinct clinical features and outcomes.