Musculoskeletal tumours: preliminary experience with perfusion MRI

Purpose. Our study aimed to assess the role of magnetic resonance imaging (MRI) in the characterisation of musculoskeletal tumours and to identify specific perfusion patterns for the different tumours. Materials and methods. Between January 2003 and September 2005, we evaluated the conventional and perfusion MRIs of 39 patients with musculoskeletal tumours. Dynamic MRI was performed with a 1.5-T and 1.0-T MRI unit before and after the intravenous administration of contrast material, using dedicated phased-array coils appropriate for the region to be studied and fast and ultrafast consecutive sequences. Postprocessing was done on an independent workstation (Advantage Windows, GE Medical System), with Functool (GE) software, which allowed a quantitative evaluation of enhancement as a function of time. The results were compared with the histopathological diagnoses obtained by biopsy or surgery. Results. The lesions identified in the 39 patients included 23 soft tissue tumours (12 benign, 11 malignant) and 16 bone tumours (ten benign, six malignant). Comparing the time-intensity diagrams of lesions of the same histological type, we found typical enhancement patterns for some bone tumours only, especially for bone, cartilaginous, fibrohistiocytic and pseudoinflammatory lesions. No typical enhancement pattern could be detected for any of the histological types of soft tissue tumour. Analysis of the slope of the time-intensity curves has a sensitivity and specificity of 64%-58% for soft tissue tumours and 86%-67% for bone tumours in determining the biological aggressiveness of the lesions. Conclusions. Perfusion MRI had moderate sensitivity and specificity in the differential diagnosis between lesions with high or low biological activity. Only in a few cases was it possible to find some correlation between perfusion patterns and lesion histology. The slope values should therefore be used in combination with conventional spin-echo images and other imaging and clinical data in order to narrow the field of the possible differential diagnoses and reliably predict the nature of the lesion.