Assessing Prognosis by Magnetic Resonance Imaging in Patients With Myelodysplastic Syndromes

abstract & Commentary

Synopsis: Myelodysplastic syndromes (MDS) are a heterogeneous group of bone marrow disorders that have a full spectrum of clinical aggressiveness. A classification system has been developed that provides valuable prognostic information. Magnetic resonance imaging (MRI) of the femur in patients with MDS revealed five different patterns (fatty, faint, nodular, scattered, or uniform) and these correlated well with clinical outcomes, including the development of leukemia and survival. Scattered or uniform MRI patterns indicated advanced disease, with an increased number of myeloblasts and greater cytogenetic abnormalities.

Source: Tokagi S, et al. J Clin Oncol 1999;17:277-383.

Myelodysplastic syndromes are a group of bone marrow disorders characterized by ineffective hematopoiesis. Although there is substantial heterogeneity, hematopathologists attempt to characterize the observed bone marrow morphology using the French-American-British (FAB) classification system.1 Patients with refractory anemia (RA) or refractory anemia with ring sideroblasts (RARS) typically have a favorable prognosis, while those with refractory anemia and excess myeloblasts (RAEB) or RAEB in transformation (RAEB-T) have an unfavorable prognosis, often progressing to acute leukemia. A large number of factors other than FAB classification can be used to assess prognosis including age, sex, hemoglobin concentration, WBC, neutrophil count, platelet count, percentage of blasts in the blood or marrow, cytogenetic changes, and DNA ploidy.2,3 In this report, an additional prognostic measure is considered.

Magnetic resonance imaging (MRI) of the femur (femoral marrow) was undertaken in 42 consecutive patients with newly diagnosed MDS and results were correlated with patterns of disease progression and survival. Five patterns of femoral marrow were described. These were: 1) "fatty," showing a low-signal intensity marrow that appeared dark on short T1 inversion recovery technique (STIR) but bright on T1-weighted spin echo (SE) images; 2) "faint," showing an abnormally faint signal intensity in the fatty-signal-density marrow; 3) "nodular," showing low signal marrow density with small nodules of a higher intensity on STIR images; 4) "scattered," showing multiple scattered areas of higher signal intensity in the low signal intensity marrow on STIR images; and, 5) "uniform," a uniform high-signal intensity marrow as compared with the low-signal intensity muscles on STIR images. Serial MRI scans revealed that the pattern changed from fatty, faint or nodular to scattered or uniform as the disease progressed.

In this series of 42 patients, leukemia developed in 13. In all 13, the MRI pattern was either scattered or uniform. In all, there were 13 patients (31%) with fatty, faint or nodular patterns and these patients all had either RA or RARS. None of these patients developed leukemia, and survival was excellent. In contrast, of the 29 (69%) who had femoral marrow with scattered or uniform MRI patterns, 11 had RAEB-T, six had RAEB, three had CMML, six had RA and three had RARS. Patients with scattered or uniform MRI patterns were more likely to have unfavorable cytogenetics (7 of 29 vs 0 of 13), leukemia (13 of 29 vs 0 of 13) and death (22 of 29 vs 2 of 13). The nonleukemic deaths in this group were primarily from hemorrhage or infection.

The overall survival of the 29 patients with scattered or uniform MRI patterns was significantly shorter than survival of the 13 patients with fatty, faint, or nodular MRI patterns (7-year survival 10.7% vs 73%, respectively, P < 0.01). Tokagi and colleagues have suggested that magnetic resonance images of the femoral marrow can provide valuable information for assessing the prognosis and determining the most appropriate management of patients with MDS.


MDS are a heterogeneous group of disorders which result in cytopenias and their well-characterized clinical consequences. Approximately one-third of patients with MDS progress to a disorder that resembles acute myelogenous leukemia. However, leukemia that has developed in patients with antecedent MDS is more difficult to treat and less frequently goes into complete remission when compared to de novo ANLL. The FAB classification system has provided useful prognostic information based upon bone marrow characteristics. Thus, the presence of dysplastic features, the involvement of more than one cell lineage, and the presence of increased numbers of myeloblasts characterize the more malignant variants of MDS (RAEB, RAEB-T).1 Other features, such as the presence of karyotypic abnormalitites and patient age, also seem to be important.3 The International Prognostic Scoring System3 involves assessment of the percent of marrow myeloblasts, the number of cytopenias, and a grading of the cytogenetic abnormalities: normal karyotype, Y chromosome deletion, del(5q) alone, and del (20q) alone are good, the presence of three or more abnormalities and any lesions involving chromosome 7 are bad, and any other abnormalities are intermediate in prognosis. Based upon these characteristics, four prognostic groups were determined. Thirty-one percent of patients fall into the low-risk category; median survival of this group is 5.7 years. Thirty-nine percent of patients fall into the first intermediate risk category; median survival is 3.5 years. Twenty-two percent of patients fall into the second intermediate risk category; median survival is 1.2 years. Eight percent of patients are in the high-risk category; median survival is 0.4 years.

Now there is another tool to assist in the assessment of risk for MDS patients. MRI scans of the femur may become routine testing for such patients. Patients whose femoral marrow demonstrated a scattered or uniform MRI pattern had a significantly more advanced disease (P < 0.0001) and a significantly higher percentage of blasts in the bone marrow (P < 0.0013) than patients whose femoral marrow had a faint, fatty, or nodular pattern. When the International Prognostic Scoring System3 was applied to the same patients, those with the fatty, faint, or nodular pattern fell into either low or intermediate-1 risk, whereas those with a scattered or uniform pattern fell into either intermediate-2 or high risk, indicative of a much greater likelihood for developing leukemia.

It is interesting that the femur turns out to be the skeletal site of choice for this determination. Femoral marrow is typically fatty and has little hematopoietic tissue. Other sites, such as the spine, are rich in "red marrow" and more quickly become involved in leukemic processes.4 It is suggested that changes in the femoral marrow occur more gradually and lead to MR images of greater variety. Studies that evaluated the vertebral marrow of MDS patients have shown no relationship of pattern to risk (number of blasts, FAB subtype, or severity of pancytopenia).5,6 In this study of the femoral marrow, the MRI was successful in distinguishing these high-risk groups.

Skeptics might question the clinical value of the MRI, since the results did not really further define the risk groups when the more conventional parameters were available. For example, in this series, no additional information was provided beyond the demographic, marrow and cytogenetic findings that were already available. Yet, the MRI may indeed serve a useful role. Serial evaluations of patients with MDS might provide information to the clinician about disease stability or progression. If a shift in pattern to the scattered or uniform variant occurs, this may warrant therapeutic intervention, perhaps with investigational approaches.


    1. Bennett JM, et al. Br J Haematol 1982;51:189-199.

    2. Mufti GJ, et al. Br J Haematol 1985;59:425-433.

    3. Greenberg P, et al. Blood 1997;89:2079-2088.

    4. Tanaka J, et al. Jpn J Med Imaging 1995;14:49-61.

    5. Lewis S, et al. Br J Radiol 1995;68:121-127.

    6. Depaoli L, et al. Eur J Radiol 1992;49:105-107.