Myelodysplastic syndromes, from French-American-British to World Health Organization: comparison of classifications on 431 unselected patients from a single institution

Abstract

In 1999 a working group of the World Health Organization (WHO) published a revised classification for myelodysplastic syndromes (MDS): RA, RARS, refractory cytopenia with multilineage dysplasia (RC+Dys), RAEB I and II, del (5q) syndrome, and MDS unclassifiable. Chronic myelomonocytic leukemia (CMML) and RAEB-t were excluded. Standard French-American-British (FAB) and new WHO classifications have been compared in a series of patients (n = 431) from a single center, analyzing morphologic, clinical, and cytogenetic data. According to the WHO findings, dysgranulocytopoiesis or dysmegakaryocytopoiesis only were found in 26% of patients with less than 5% medullary blasts. These patients are thus unclassified and should remain in the subgroups RA and RARS. Splitting of heterogeneous RAEB into 2 subgroups according to blast count was supported by a trend to a statistically significant difference in the single-center study population. Patients with CMML whose white blood cell counts are above 13 000/microL may be excluded from the MDS classification, as warranted by WHO, but a redistribution of patients with dysplastic CMML according to medullary blast count leads to more heterogeneity in other WHO subgroups. Although the natural courses of RAEB-T and acute myeloid leukemia (AML) with dysplasia are different, comparable median survival durations after treatment in patients with RAEB-T and AML were in favor of the proposed 20% medullary blast threshold for AML. The homogeneity of subgroups was studied by evaluating prognostic scores. A significant shift into lower IPSS risk groups was evident in the new classification. These data cannot provide evidence for the new WHO proposal, which should not be adopted for routine clinical use at present. Some of its aspects can provide a starting point for further studies involving refined cytogenetics and clinical results.