Limited usefulness of classic MR findings in the diagnosis of tenosynovial giant cell tumor.

OBJECTIVE

To determine the frequency with which MRI of tenosynovial giant cell tumor demonstrates hemosiderin, visible intralesional fat signal, and proximity to synovial tissue.

MATERIAL AND METHODS

This is a retrospective study of 31 cases of tenosynovial giant cell tumors which had concomitant MRI. Images were examined for lesion size, morphology, origin, bone erosions, MRI signal characteristics, contrast enhancement, and blooming artifact, comparing prospective and retrospective reports. Histology was reviewed for the presence of hemosiderin and xanthoma cells.

RESULTS

Eight lesions were diffuse and 23 were localized nodules. Three lesions were located in subcutaneous tissue and 4 adjacent to tendons beyond the extent of their tendon sheath. All lesions exhibited areas of low T1- and T2-weighted signal. Blooming artifact on gradient echo imaging was present in 86% of diffuse and only 27% of nodular disease. There was interobserver variability of 40% in assessing blooming. Iron was visible on H&E or iron stain in 97% of cases. Fat signal intensity was seen in only 3% of cases, although xanthoma cells were present on in 48%. The correct diagnosis was included in the prospective radiology differential diagnosis in 86% of diffuse cases and 62% of nodular cases.

CONCLUSION

Blooming on GRE MRI has low sensitivity for nodular tenosynovial giant cell tumors and is not universal in diffuse tumors. There was high interobserver variability in assessment of blooming. Intralesional fat signal is not a useful sign and may occur adjacent to tendons which lack a tendon sheath and may occur in a subcutaneous location.