Magnetic Resonance Imaging for the Assessment of Long Bone Tumors

BACKGROUND Wide resection margins of osseous tumors are associated with a low incidence of local recurrence, making accurate measurement of the intraosseous extent of primary malignant long bone tumors is crucial. We compared the intraosseous tumor extent assessed by magnetic resonance imaging (MRI) with the gross specimen to evaluate the accuracy of MRI. METHODS A total of 255 patients with primary malignant tumors in the long bones were included. Using MRI, we defined the length of tumor as the distance from the articular surface to the boundary between abnormal and normal marrow signal. The extent of the abnormal intraosseous signal was measured on unenhanced T1-weighted (T1WI) magnetic resonance images after chemotherapy. All gross surgical specimens were sectioned, and tumor extent was measured. Wilcoxon signed-rank test was used to test the differences between MRI and gross specimen findings. Spearman's correlation analysis was used to test the correlation between groups. RESULTS Median tumor length by gross specimen (112 mm; range, 45-300 mm) was longer than that by MRI (108 mm; range, 45-304 mm; Z = -6.916, P < 0.001). Of 255 images, tumor length was accurately represented on 27 T1WI magnetic resonance images, overestimated on 79 images, and underestimated on 149 images. The median difference between imaging and gross specimen measurements was 2.0 mm (range: 1.0-15.0 mm) for the 79 cases where tumor length was overestimated, and 5.0 mm (range: 1.0-18.0 mm) for the 149 cases where tumor length was underestimated. The Spearman correlation demonstrated a high correlation of tumor length on gross specimen with the tumor length on MRI (R = 0.99, P < 0.01). CONCLUSIONS We conclude that preoperative MRI could be a useful method in determining intramedullary malignant bone tumor boundaries and may serve as an accepted assessment method of long bone tumors before limb-sparing surgery.