MRI and biologic behavior of desmoid tumors in children

OBJECTIVE. The outcome of desmoid tumor in children cannot be reliably predicted on the basis of histologic findings. We sought to determine whether the postoperative presence of residual or recurrent tumor can be predicted on the basis of demographic variables and baseline MRI features of the tumor. We also aimed to determine how imaging features change during adjuvant treatment and how the imaging features relate to the histologic features. MATERIALS AND METHODS. Two radiologists retrospectively reviewed images from 281 MRI examinations performed at baseline and during postoperative therapy for desmoid tumor. The examinations had been performed on 17 children treated between September 1991 and March 2003. Tumor volume; distinctness of margins; involvement of bone and neurovascular bundle; and T1-weighted, T2-weighted, and STIR signal intensity and contrast enhancement pattern were recorded. Baseline imaging and demographic features were correlated with the postoperative presence of residual or recurrent tumor. Imaging changes during follow-up were compared with treatment response and outcome. The imaging features of eight tumors were compared with percentage cellularity and collagen deposition in biopsy samples obtained within 30 days of imaging. RESULTS. Baseline involvement of the neurovascular bundle approached significance as a predictor of the presence of residual or recurrent tumor (p = 0.08). Other baseline imaging and demographic features were not predictive (p ≥ 0.4). Changes in imaging features were variable during follow-up. T2-weighted and STIR signal intensity may be correlated with percentage cellularity and collagen deposition. CONCLUSION. MRI has limited value in prediction of the postoperative presence of residual or recurrent desmoid tumor in children. It is useful, however, for detecting disease and monitoring postoperative adjuvant therapy.