Synovial sarcoma: MR imaging findings in 34 patients.

OBJECTIVE: MR imaging is considered the procedure of choice for detecting and staging soft-tissue tumors. Its ability to show differences between benign and malignant soft-tissue tumors and its usefulness in suggesting a specific histologic diagnosis remain controversial. We studied the MR features of synovial sarcoma in 34 patients to determine if these tumors have specific MR findings that can be used to suggest the diagnosis.

MATERIALS AND METHODS: MR imaging studies of 34 patients with synovial sarcoma were collected from two institutions and studied to determine the following characteristics of the tumor: size, shape, location, signal intensity and homogeneity, margin definition, presence of hemorrhage, and relationships to adjacent structures. These findings were then correlated with pathologic findings.

RESULTS: The tumors tended to be deep, large (85% were > or = 5 cm in diameter), and located in the extremities with epicenters close to joints (63% within 7 cm of a joint). The lesions were usually inhomogeneous on T2-weighted images (82%) and clearly delineated from surrounding tissues (91%). Forty-four percent had high signal consistent with hemorrhage on both T1- and T2-weighted images. Fluid-fluid levels, best visualized on T2-weighted images, were present in 18% of patients. Thirty-five percent of the lesions had areas that were hyper-, iso-, and hypointense relative to fat on T2-weighted images, constituting a triple signal intensity. The tumors frequently involved adjacent bone, with 71% invading, eroding, or touching bone. No association of pathologic subtypes with specific imaging findings was noted.

CONCLUSION: Our results show a spectrum of MR imaging findings in synovial sarcoma. Nevertheless, the results suggest that synovial sarcoma should be considered when MR images show a relatively well-defined but inhomogeneous hemorrhagic lesion near a joint and in contact with bone. Fluid-fluid levels and areas hyper-, hypo-, and isointense relative to fat (triple signal) on T2-weighted sequences support the diagnosis.