Background
Prognostic model development and external validation study. 2,957 patients with MDS were profiled for somatic mutations across 152 genes at diagnosis (development cohort), with external validation in 754 Japanese MDS patients. Endpoints included leukemia-free survival (LFS), leukemic transformation, and overall survival. Developed to improve on the Revised IPSS (IPSS-R), which uses only hematologic parameters and cytogenetics without molecular data.
Interventions and follow up
Design: Integration of clinical, cytogenetic, and somatic mutation data (31 genes) into the IPSS-M model to predict outcomes in primary MDS, compared against IPSS-R
Primary endpoint: Discrimination (concordance) for OS and LFS; reclassification rate vs IPSS-R
Median follow up: NR (retrospective)
Primary endpoint: Discrimination (concordance) for OS and LFS; reclassification rate vs IPSS-R
Median follow up: NR (retrospective)
Results
Reclassification: ~46% of patients reclassified vs IPSS-R, predominantly upgraded to higher risk
Risk categories: Six-category continuous model (Very Low / Low / Moderate Low / Moderate High / High / Very High)
Discrimination: IPSS-M outperformed IPSS-R for OS, LFS, and AML transformation
Validation: Improved prognostic discrimination confirmed in the external Japanese cohort
Risk categories: Six-category continuous model (Very Low / Low / Moderate Low / Moderate High / High / Very High)
Discrimination: IPSS-M outperformed IPSS-R for OS, LFS, and AML transformation
Validation: Improved prognostic discrimination confirmed in the external Japanese cohort
Adverse events
Applicability: Not applicable — observational prognostic model development/validation study.
Safety data: No treatment exposure; no adverse event reporting.
Safety data: No treatment exposure; no adverse event reporting.
Conclusions
The IPSS-M substantially improves MDS risk stratification over the IPSS-R by incorporating somatic mutation data, restratifying ~46% of patients to different risk categories. It provides individualized continuous risk scores and 6 clinically actionable risk groups with stronger prognostic discrimination for LFS, AML transformation, and OS. An open-access web calculator facilitates clinical use.
Key Limitations
Retrospective model derived largely from treated cohorts, so prognostic estimates may not fully reflect treatment-naïve natural history. Requires next-generation sequencing of 31 genes, which is not universally available and adds cost/turnaround. Validation cohort was geographically restricted (Japanese); broader prospective validation across diverse populations and therapy-related MDS is needed. Does not directly guide therapy selection or predict treatment response. Therapy-related and pediatric MDS underrepresented.
Clinical Context
IPSS-M is increasingly adopted as the molecular standard for MDS prognostication, complementing the WHO 2022 and ICC 2022 classifications which incorporate molecular and cytogenetic categories. It informs allogeneic transplant timing and intensity-of-therapy decisions in higher-risk MDS. ELN and international MDS guidance endorse molecular risk stratification; an open-access online calculator supports routine clinical application.