Toward 3-Dimensional Tumor Analysis for Planning Optimal Resection in Intradural Extramedullary Tumors: A Single-Institution Feasibility Study.

BACKGROUND AND OBJECTIVES: Intradural extramedullary spinal tumors (IDEMs) are ideally managed with gross total resection (GTR) for optimal local disease control in the absence of established adjuvant regimens. In this retrospective cohort study, we investigated the utility of morphology-based analysis of optimal IDEM resection quality compared with current linear measurement benchmarks. METHODS: Tumors were manually segmented from preoperative contrast-enhanced MRI. Morphological features of sphericity, elongation, and volume were extracted from masks, along with manual anteroposterior, craniocaudal, and transverse linear measurements. Optimal resection was defined as Simpson Grade 1 or 2 for meningiomas or en bloc GTR for myxopapillary ependymomas (MPEs) and was predicted for dorsal and ventral spinal meningiomas plus conus and nonconus MPEs. RESULTS: We identified 103 tumors, including 71 meningiomas and 32 MPEs. Linear measurement did not reliably predict optimal resection across anatomic subtypes of meningiomas or MPEs (all P > .05). Instead, optimal resection was determined by higher sphericity across meningiomas and, specifically, dorsal meningiomas (optimal resection median 0.80 (IQR: 0.77-0.82) vs suboptimal 0.76 (0.72-0.77), P = .027). Shape features were not predictive in ventral meningiomas. En bloc GTR was limited to nonconus MPEs, in which sphericity again emerged as a predictor (0.74 (0.69-0.77) for en bloc vs 0.68 (0.64-0.72) for piecemeal, P = .047), likely reflecting resectability for encapsulated tumors. After excluding patients with <1 year of surveillance to determine nonrecurrence, 62 patients without genetic syndromes remained for recurrence analysis with a median follow-up of 36.1 months. Optimal resection was associated with significantly lower recurrence (optimal 8.6% vs suboptimal 29.6%, P = .045). This protective effect remained, although statistically insignificant after adjusting for adjuvant radiation and tumor grade (recurrence odds ratio = 0.22, 95% CI: 0.04-1.07, P = .077). CONCLUSION: IDEM tumor morphology is predictive of optimal resection for dorsal meningiomas and nonconus MPEs, whereas linear measurements offer less consistent predictive value. Morphological assessment is feasible with standard MRI and may be further automated.