Craniopharyngioma adherence: a comprehensive topographical categorization and outcome-related risk stratification model based on the methodical examination of 500 tumors.

OBJECTIVE Craniopharyngioma (CP) adherence strongly influences the potential for achieving a radical and safe surgical treatment. However, this factor remains poorly addressed in the scientific literature. This study provides a rational, comprehensive description of CP adherence that can be used for the prediction of surgical risks associated with the removal of these challenging lesions. METHODS This study retrospectively analyzes the evidence provided in pathological, neuroradiological, and surgical CP reports concerning 3 components of the CP attachment: 1) the intracranial structures attached to the tumor; 2) the morphology of the adhesion; and 3) the adhesion strength. From a total of 1781 CP reports published between 1857 and 2016, a collection of 500 CPs providing the best information about the type of CP attachment were investigated. This cohort includes autopsy studies (n = 254); surgical studies with a detailed description or pictorial evidence of CP adherence (n = 298); and surgical CP videos (n = 61) showing the technical steps for releasing the attachment. A predictive model of CP adherence in hierarchical severity levels correlated with surgical outcomes was generated by multivariate analysis. RESULTS The anatomical location of the CP attachment occurred predominantly at the third ventricle floor (TVF) (54%, n = 268), third ventricle walls (23%, n = 114), and pituitary stalk (19%, n = 94). The optic chiasm was involved in 56% (n = 281). Six morphological patterns of CP attachment were identified: 1) fibrovascular pedicle (5.4%); 2) sessile or patch-like (21%); 3) cap-like (over the CP top, 14%); 4) bowl-like (around the CP bottom, 13.5%); 5) ring-like (encircling central band, 19%); and 6) circumferential (enveloping the entire CP, 27%). Adhesion strength was classified in 4 grades: 1) loose (easily dissectible, 8%); 2) tight (requires sharp dissection, 32%); 3) fusion (no clear cleavage plane, 40%); and 4) replacement (loss of brain tissue integrity, 20%). The types of CP attachment associated with the worst surgical outcomes are the ring-like, bowl-like, and circumferential ones with fusion to the TVF or replacement of this structure (p < 0.001). The CP topography is the variable that best predicts the type of CP attachment (p < 0.001). Ring-like and circumferential attachments were observed for CPs invading the TVF (secondary intraventricular CPs) and CPs developing within the TVF itself (infundibulo-tuberal CPs). Brain invasion and peritumoral gliosis occurred predominantly in the ring-like and circumferential adherence patterns (p < 0.001). A multivariate model including the variables CP topography, tumor consistency, and the presence of hydrocephalus, infundibulo-tuberal syndrome, and/or hypothalamic dysfunction accurately predicts the severity of CP attachment in 87% of cases. CONCLUSIONS A comprehensive descriptive model of CP adherence in 5 hierarchical levels of increased severity-mild, moderate, serious, severe, and critical-was generated. This model, based on the location, morphology, and strength of the attachment can be used to anticipate the surgical risk of hypothalamic injury and to plan the degree of removal accordingly.