The revised International Prognostic Scoring System (IPSS-R) improved the accuracy of outcome prediction in patients with myelodysplastic syndrome (MDS). Its major limitation, however, is the large heterogeneity within the intermediate-risk (int-risk) category (~20% of all patients). Recent evidence indicates variable outcomes of int-risk patients and need for additional prognostic factors to refine prognosis. Partly because the original IPSS is still used for azacitidine (AZA) approval, there have been no studies concerned with the effects of AZA specifically in IPSS-R int-risk MDS and, therefore, questions about benefits of AZA in these patients remain.
We aimed to establish simple, reproducible and widely applicable features that could help clinicians to identify int-risk IPSS-R patients at risk for poor outcomes at an early stage, and clarify whether AZA is effective as treatment. Our study population involving 468 patients from the database of the Hellenic National Registry of Myelodysplastic and Hypoplastic Syndromes is one of the world’s largest cohorts of int-risk MDS.
We paid particular emphasis on statistical methods. Data analysis was planned carefully on the basis of an existing hypothesis. Τhe statistical analysis plan was crucial in ensuring that solid, univariate and multivariate regression models were generated.
Similar to a previous analysis of the original IPSS-R cohort, we noted that an IPSS-R cutoff of 3.5 could divide int-risk patients into lower- and higher-risk groups (Fig. 1). Thus, int-risk patients may be considered as lower-risk if their score is 3.5 versus higher-risk if their score is >3.5. And, importantly, we showed for the first time that further refinement of the IPSS-R 3.5 subgroup was possible by means of the Endothelial Activation and Stress Index (EASIX), which separated populations of significantly different overall survival (OS). Additionally, we noted that the potential risk factors of older age (>70 years), circulating blasts (≥1%), IPSS-R >3.5, and high EASIX (>0.179 on the log2 scale) were the strongest predictors for OS and could identify patients with a poor prognosis at an early stage.
We have also analyzed the data in relation to AZA use. It emerges that an improvement in OS associated with AZA was evident only in patients achieving complete remission (CR). The CR rate, however, in these patients was modest (16.3%) and, therefore, AZA conferred no obvious survival benefit beyond standard care (Fig. 2). Mere improvement in the patient’s condition or hematologic status was not sufficient to increase OS, highlighting the major influence of CR on OS in the IPSS-R int-risk group.
The data that are reported here regarding risk of progression, azacitidine use, and specific risk factors may allow clinicians to communicate more accurate prognostic information to patients and choose the type of monitoring and treatment that is needed for each patient. We believe that OS is a reasonable primary endpoint for clinical studies in this challenging patient population. It is hoped that in the field of int-risk MDS, our findings will serve as the basis for further work, discussion in the MDS community, and improved clinical practice.