International Prognostic Scoring System and Other Prognostic Systems for Myelodysplastic Syndromes

To the Editor:

Although the French-American-British (FAB) classification1 has enabled the natural history of myelodysplastic syndromes (MDS) to be clarified, it has limitations as a prognostic parameter owing to the great variability in survival within each FAB subtype, especially refractory anemia (RA) and RA with ring sideroblasts (RARS). To obtain a better prognostic characterization of individual MDS patients, different scoring systems have been proposed,2-7 but imprecision persists with these analyses. In a recent issue of Blood, Greenberg et al8 described an International Prognostic Scoring System (IPSS) for MDS that was generated during an International MDS Risk Analysis Workshop. This system takes into account the number of cytopenias, bone marrow blast percentage, and refined cytogenetic characterization. In the investigators' experience, it provides an improved method for predicting survival and acute leukemia evolution with respect to the FAB, Sanz, and Morel categorization methods.

This publication prompted us to apply the new scoring system to MDS patients observed in our institution over the last 10 years and to compare its predictive value with that of the FAB classification and five previously published scoring systems.

The records of patients observed between 1986 and 1995 were examined: 143 cases of primary MDS not receiving intensive chemotherapy were found and the scoring systems of Mufti et al,2 Sanz et al,3 Goasguen et al,4 Morra et al,5 and Morel et al,7 in addition to the IPSS, were applied on the basis of data at diagnosis. According to the FAB classification, 52 patients had RA, 14 had RARS, 37 had RA with excess of blast cells (RAEB), 32 had RAEB in transformation (RAEB-t), and 8 had chronic myelomonocytic leukemia (CMML). Application of the Mufti, Sanz, Goasguen, and Morra scoring systems was possible in all cases, whereas IPSS and the Morel scoring systems were applied to 98 subjects only, because karyotype at diagnosis was not available for 45 subjects. This situation derived from a lack of evaluable mitoses in 15 cases and from the fact that cytogenetic investigation was not performed at diagnosis in 30 other patients. This latter phenomenon is not surprising, because chromosomal studies require specialized laboratories and are not always readily available everywhere; for example, in our institution, the cytogenetic laboratory is not operative on weekends and for 1 month during the summer.

Freedom from leukemia evolution and survival time were calculated using Kaplan-Meier curves.

Table 1 compares the effectiveness of the prognostic systems in the 95 patients with cytogenetic characterization at diagnosis; similar results were obtained when the scoring systems not requiring chromosomal mapping were applied to all 143 patients. Our results show that all prognostic systems gave similar results in the identification of patients with very short survival time and a high propension to acute leukemia transformation, whereas IPSS worked better in the categorization of subjects with intermediate risk. However, the IPSS was not satisfactory in the identification of patients with very long survival, in which the scoring systems of Morel, Sanz, and Goasguen and the FAB classification gave better results. The FAB classification and the Sanz and Goasguen systems offered the additional advantage in our study of MDS cases of allowing prognostic categorization of all patients, because they do not require cytogenetic studies.

In conclusion, our experience confirms that the IPSS is effective in separating MDS patients into four groups with different life expectancies and tendencies to leukemia evolution. However, it was inferior to some of the older prognostic systems and to the FAB classification in the identification of very long survivors. Therefore, although the IPSS represents progress in the prognostic classification of intermediate-risk patients, it is not fully satisfactory and further investigation is required to devise better staging systems for proper risk-based therapeutic choices.