Use of preoperative MR to predict dural, perineural, and venous sinus invasion of skull base tumors.

PURPOSE To assess the accuracy of MR imaging in predicting dural, venous sinus, and perineural invasion by skull base tumors. METHODS The preoperative MR images of 22 patients who had resection of skull base neoplasms were evaluated for the following characteristics: dural enhancement, pial enhancement, local perineural invasion by adjacent tumor, and venous sinus invasion by tumor. The greatest width of dural enhancement was measured, and the character of dural enhancement was noted. The pathologic and surgical reports were reviewed retrospectively with specific attention to dural, venous, and local perineural invasion. RESULTS Of the 22 patients studied, dural invasion by tumor was confirmed in eight patients, vascular invasion in six patients, and perineural invasion in four patients. The sensitivity of dural enhancement in predicting invasion was 88%, the specificity 50%, and the accuracy 64%. When enhancement and focal nodularity were present, the sensitivity remained at 88%; however, specificity was 100% and accuracy 95%. If the dural enhancement was more than 5 mm thick, sensitivity, specificity, and accuracy were 75%, 100%, and 91%, respectively. Predicting tumor invasion of the dura by the presence of pial enhancement was 50% sensitive and 100% specific. Venous sinus/jugular vein invasion was predicted with 100% sensitivity, 94% specificity, and 95% accuracy. Local perineural invasion was predicted with 100% sensitivity, 50% specificity, and 59% accuracy. CONCLUSIONS The presence of pial enhancement, focal dural nodules, or dural thickening of more than 5 mm is highly accurate in predicting the presence of neoplastic dural invasion. Linear enhancement of dura does not imply dural infiltration by tumor. Venous invasion by tumor can be predicted accurately with preoperative MR imaging.