Derivation of a data-driven follow-up protocol for resected skull base meningiomas: a Bayesian analysis

Complete resection of skull base meningiomas (SBMs) is often limited by their proximity to critical neurovascular structures, potentially increasing the risk of postoperative progression. This study quantified long-term progression risk after SBM resection, identified predictors of progression, and developed a data-driven MRI surveillance protocol using Bayesian methods.

Patients undergoing SBM resection between 2002 and 2020 at two neurosurgical centres were analysed. Kaplan-Meier and Cox regression analyses were used to estimate intervention-free survival (IFS). Conditional probability modelling was used to derive an MRI surveillance schedule that maintained a ≤ 3% risk of progression requiring intervention, stratified by extent of resection.

A total of 358 SBMs were included. Median age at diagnosis was 57 years (IQR 18), and 76.0% of patients were female. WHO Grade 1 tumours accounted for 80.3% of cases and Grade 2 for 19.7%. Median follow-up was 97 months (IQR 64-128). Progression requiring re-intervention occurred in 14.1% of patients. Subtotal resection (STR) increased the risk of re-intervention (HR 2.62, 95% CI 1.30-5.30, p = 0.009), whereas higher comorbidity burden (HR 0.78, 95% CI 0.62-0.98, p = 0.038) and incidental presentation (HR 0.11, 95% CI 0.01-0.88, p = 0.038) were associated with lower risk. Conditional probability modelling demonstrated higher annual progression risk following STR, supporting more intensive imaging surveillance. Recommended MRI schedules were: gross total resection (GTR), scans at 3 months, 2, 5, 8, and 10 years; STR, scans at 3 months, 1 year, 18 months, annually from years 2-8, and at 10 years.

Progression requiring re-intervention occurs in approximately 17% of patients after SBM resection and is strongly associated with extent of resection. A conditional-risk-based MRI surveillance protocol provides an evidence-based framework for long-term follow-up.