Outcomes after surgical resection of spinal ependymomas: extent of resection, imaging phenotype, and survival

To characterize functional recovery, imaging phenotype, and survival in surgically treated spinal ependymomas and to explore clinicopathological and radiologic correlates of postoperative course.

We retrospectively reviewed 48 consecutive patients (55 surgical procedures) with pathologically confirmed spinal ependymoma, including both intramedullary and intradural-extramedullary/filum lesions. Collected variables included demographics, tumor location and MRI phenotype (axial position, cyst formation, syrinx, cord edema, enhancement), extent of resection, pathological subtype/grade, Ki-67, complications, recurrence/progression, and follow-up. Functional outcomes were assessed pre- and postoperatively using the modified McCormick Scale (mMCS) and Karnofsky Performance Status (KPS). Kaplan-Meier methods estimated progression-free survival (PFS) and overall survival (OS) with log-rank comparisons; univariable Cox models (cluster-robust for repeated procedures) estimated hazard ratios (HR). Because numerous exploratory comparisons were performed, p-values were corrected with the Benjamini-Hochberg false-discovery rate.

Mean age was 37.8 ± 14.5 years; 62.5% were male. Tumors were most commonly cervical (43.6%) and lumbar (38.2%); 56.4% were intramedullary and 43.6% extramedullary. Pathology included WHO grade 2 ependymoma (63.6%), myxopapillary ependymoma (29.1%), and WHO grade 3 ependymoma (7.3%). Gross-total resection (GTR) was achieved in 85.5% of procedures. Complications occurred in 9.1% and recurrence/progression in 23.6%; median follow-up was 22 months (mean 31.2 ± 28.8). KPS improved significantly (p < 0.001), whereas mMCS did not (p = 0.25), and no between-group differences in ΔmMCS/ΔKPS reached significance. Three- and five-year PFS were 84% and 78%, and OS 94% and 88%. In univariable analyses, subtotal resection was associated with shorter PFS (HR 3.7, 95% CI 1.5-9.7) and a similar trend for OS (HR 4.7; p = 0.06); WHO grade 3 (HR 5.7) and Ki-67 > 3% (PFS HR 3.7; OS HR 7.2) were associated with worse outcomes. After false-discovery-rate correction, only pathological subtype remained significant; earlier associations of syrinx and anatomical level with PFS were not confirmed. Intramedullary and extramedullary tumors did not differ significantly in PFS or OS.

Postoperative gains were more consistently captured by KPS than mMCS. Maximal safe resection remains central to tumor control and survival, while MRI phenotype and Ki-67 may provide additional, hypothesis-generating prognostic context for surveillance planning.