A Retrospective Validation Study Comparing the IPSS-M and IPSS-R Risk Stratification Scores in Primary and Secondary Myelodysplastic Syndrome (MDS)

Introduction:

Myelodysplastic syndrome (MDS) is a stem cell disorder characterized by bone marrow stem cell dysfunction, cytopenia(s), and increased risk of acute myeloid leukemia (AML) transformation. Treatment of MDS varies based on a patient's estimated risk of transformation to AML and the presence of molecular and cytogenetic changes known to confer poor clinical outcomes. Significant effort over the last few decades has been undertaken to develop prognostic models to risk stratify patients, resulting in the international prognostic scoring system (IPSS), revised version (IPSS-R), and WHO classification-based prognostic scoring system (WPSS). Recently, the molecular international prognostic scoring system (IPSS-M) was proposed to consider clinical indices, laboratory and pathology parameters, cytogenetics, and mutational data in the progression of MDS. Currently, this model has only been validated within a Japanese database, wherein 42% of patients were re-stratified by IPSS-M in comparison to IPSS-R. To date, these results have yet to be validated in a cohort of patients in the U.S.

Methods:

We conducted a single-center retrospective validation cohort study of adults with a known diagnosis of MDS referred to Oregon Health & Science University from 2019 through 2021. The study was approved by the Institutional Review Board prior to initiation (OHSU IRB number- STUDY00023980). Patients were identified using relevant diagnostic codes for MDS and those who were <18 years of age, did not have a biopsy-confirmed diagnosis of MDS, did not have next-generation sequencing performed on diagnostic tissue, or were lost to follow-up shortly after diagnosis were excluded. Patients with treatment-related MDS were included.

The primary outcomes, measured from MDS diagnosis and determined through May 2022, were overall survival (OS) and leukemia-free survival (LFS). Predictors of interest were IPSS-R, age-adjusted IPSS-R, WPSS, and IPSS-M, each with categories of "very low", "low", "intermediate" (called "moderate" for IPSS-M), "high", and "very high" risk. OS and LFS in specific prognostic risk categories/groups were estimated by the Kaplan-Meier method and compared using the log-rank test. Median follow-up was estimated by the reverse Kaplan-Meier method. Cox regression models of OS and LFS were fit to estimate the relative risk of AML transformation or death with hazard ratios (HRs) and the prognostic accuracy with concordance indices (C-index).

Results:

One hundred and forty-nine patients, diagnosed between 2011 and 2021, met inclusion criteria and were analyzed: 136 diagnosed with primary MDS and 13 with secondary/treatment-related MDS (Table 1). After IPSS-M re-stratification, 32% of patients were up-staged (with 24 moving into the "very high" group) and 12% down-staged (with 3 patients exiting the "very high" group) from their IPSS-R risk category. Median follow-up time was 2.8 years and 54 patients died (36%) during the study period. Among the 32 patients (21%) whose disease transformed to AML, median time from MDS diagnosis to transformation was 7 months. When considering death as a competing risk, estimated cumulative incidence of AML transformation was 14% at 1 year and 19% at 3 years post-diagnosis.

IPSS-R groups were significantly segregated by LFS (log-rank p=6.3e-05; "very high" vs. "low" HR=4.08, p=3.3e-04; Figure 1, Table 1) and by OS (log-rank p=4.9e-06; "very high" vs. "low" HR=5.38, p=1.0e-04; Fig 1, Table 1). IPSS-M was also highly significantly associated with LFS (log-rank p=1.2e-07; "very high" vs. "low" HR=5.64, p=4.4e-06; Fig 1, Table 1) and OS (log-rank p=4.6e-08; "very high" vs. "low" HR=7.30, p=6.1e-06; Fig 1, Table 1).

IPSS-M had better discrimination than IPSS-R for LFS (C-index 0.72 [95% CI: 0.67 - 0.78] vs. 0.68 [95% CI: 0.62 - 0.75]) and for OS (C-index 0.74 [95% CI: 0.68 - 0.81] vs. 0.70 [95% CI: 0.63 - 0.78]). This discrimination advantage was also observed for LFS (0.73 vs. 0.70) and OS (0.76 vs. 0.73) when taking into account a patient's transfusion dependence at MDS diagnosis.

Discussion & Conclusion:

This single-center study is the first U.S. cohort validation of the IPSS-M, finding that this mutation-incorporating prognostic index has greater discriminative potential than IPSS-R in predicting AML transformation and any-cause mortality. Further multicenter prospective studies are needed to confirm the utility of the IPSS-M model.

No relevant conflicts of interest to declare.