RESUMO
PURPOSE: A classification system for cystic sellar lesions does not exist. We propose a novel classification scheme for these lesions based on the heterogeneity of the cyst wall/contents and the presence of a solid component on imaging. METHODS: We retrospectively reviewed 205 patients' medical records (2008-2020) who underwent primary surgery for a cystic sellar lesion. Cysts were classified a priori into 1 of 4 cyst types based on the heterogeneity of the cyst wall/contents and the presence of a solid component imaging. There was high interrater reliability. Univariable and multivariable models were used to estimate the ability of cyst type to predict the two most common diagnoses: Rathke cleft cyst (RCC) and cystic pituitary adenoma. RESULTS: The frequencies of RCC and cystic pituitary adenoma in our cohort were 45.4% and 36.4%, respectively. Non-neoplastic lesions (e.g., arachnoid cysts and RCC) were more likely to be Type 1 or 2, whereas cystic neoplasms (e.g., pituitary adenomas and craniopharyngiomas) were more likely to be Type 3 or 4 (p<0.0001). Higher cyst types, compared to Type 1, had higher odds of being cystic pituitary adenomas compared to RCCs (OR: 23.7, p=0.033, and 342.6, p <0.0001, for Types 2 and 4, respectively). Lesions with a fluid-fluid level on preoperative MRI also had higher odds of being pituitary adenomas (OR: 12.7; p=0.023). Cystic pituitary adenomas were more common in patients with obesity (OR: 5.0, p=0.003) or symptomatic hyperprolactinemia (OR: 11.5; p<0.001, respectively). The multivariable model had a positive predictive value of 82.2% and negative predictive value of 86.4%. CONCLUSION: When applied to the diagnosis of RCC versus cystic pituitary adenoma, higher cystic lesion types (Type 2 & 4), presence of fluid-fluid level, symptomatic hyperprolactinemia, and obesity were predictors of cystic pituitary adenoma. Further validation is needed, but this classification scheme may prove to be a useful tool for the management of patients with common sellar pathology.