Risk score for outcome prediction after microsurgical resection of spinal ependymoma (SOURSE score)

Abstract

ObjectiveMicrosurgical resection of spinal ependymomas is associated with a considerable risk of postoperative neurological deterioration. We aimed to develop a risk score for outcome prediction after surgery for spinal ependymoma. Materials and methodsAll patients who underwent microsurgical resection of spinal ependymoma between 1980 and 2015 were included. Different perioperative parameters were collected for the score construction. Poor outcome was defined as the modified McCormick Scale (MMCS) >2 at 6 months after surgery. ResultsOf 131 patients (mean age: 45.6 ± 16.7 years; 63 females), 38 cases (29%) showed poor outcome. Based on the univariate analysis, preoperative MMCS, subtotal tumor resection, proximal tumor level on the spinal cord, tumor extension, intramedullary location, and WHO grading were included in the multivariate analysis. The final risk score consisted of the following independent predictors: preoperative MMCS > 1 (1 point), proximal tumor level at Th 10 and higher (1 point), and tumor extension ≥ 3 vertebrae (1 point). The constructed score (0–3 points; Score for OUtcome after Resection of Spinal Ependymoma [SOURSE]) showed high diagnostic accuracy (area under the curve [AUC] = 0.883), which was superior to preoperative MMCS (AUC = 0.798) and Karnofsky Performance Status (AUC = 0.794). Patients scoring 0, 1, 2, and 3 points showed poor outcome in 0%, 12.9%, 54.6%, and 76.2% of the cases respectively. ConclusionThe presented SOURSE score based on preoperative neurologic condition, tumor location, and tumor extension could accurately predict the postoperative outcome in patients undergoing microsurgery of spinal ependymoma. Our data should be validated in a prospective trial.