MR imaging of cavernous sinus involvement by pituitary adenomas.

The ability of high-resolution MR imaging (1.5 T) to detect invasion of the cavernous sinuses by pituitary adenoma was determined through a retrospective review of 74 patients. These patients were divided into three groups: 25 normal subjects, 24 subjects with invasive pituitary adenomas, and 25 subjects with noninvasive pituitary adenomas. A fourth group of 30 patients, who subsequently underwent surgery for pituitary adenoma, was evaluated prospectively by MR for the presence or absence of cavernous sinus invasion. Several features were analyzed: (1) the detectability of the medial and lateral dural margins of the cavernous sinus (2) the size and variation in intensity of compartments within the cavernous sinus (3) the relationship of endocrine function to the surgical and MR appearance of the cavernous sinus and (4) carotid artery displacement or encasement by tumor. The normal cavernous sinuses were usually symmetric, but their sizes varied. The lateral dural margin of the cavernous sinus was always recognized on MR as a linear, discrete, low-intensity area. The medial dural margin (pituitary capsule) was seen on MR in only two of the 25 normal patients. In all 24 patients with cavernous sinus invasion involvement was unilateral and was most common with laterally positioned prolactin or adrenocorticotropic hormone secretory adenomas. Invasion of the cavernous sinus was suspected by MR in only two of the 13 invasive microadenomas and was questionable in three. In 10 of the 11 macroadenomas with surgically proved dural invasion, MR demonstrated an asymmetric increase in size and intensity of the superior and inferior cavernous sinus compartments. Noninvasive macroadenomas compressed and displaced the cavernous sinus bilaterally. The prospective MR evaluation of 30 patients undergoing surgery for pituitary tumor revealed a sensitivity for predicting cavernous sinus invasion of 55%, a specificity of 85.7%, a positive predictive value of 62.5%, and a negative predictive value of 81.8%. No feature permitted certain distinction between invasive and noninvasive microadenomas, as the medial dural wall of the cavernous sinus could not be reliably identified. The most specific sign of cavernous sinus invasion was carotid artery encasement.