Simpson Grading and Risk of Recurrence in 4074 Meningioma Patients With Gross Total Resection.

Abstract

Simpson grading remains the most widely used clinical classification of extent of resection in meningioma surgery, yet its relevance in modern neurosurgical practice remains debated. The aim was to evaluate whether Simpson grading is associated with recurrence risk after gross total resection (GTR) of intracranial meningiomas within the context of tumor biology and anatomic location. We analyzed 4074 patients undergoing GTR (Simpson grades 1-3) for primary intracranial meningioma, compiling data from 42 institutions worldwide. Cumulative incidence of recurrence was estimated using Aalen-Johansen methods. Multivariable inverse probability of censoring weighted logistic regression models were used to estimate adjusted recurrence risks at 5 and 10 years, adjusting for age, sex, World Health Organization (WHO) grade, Ki-67 proliferation index, and healthcare system setting. Higher Simpson grades were consistently associated with increased recurrence risk in WHO-1 and WHO-2. Among WHO-1, Simpson grade 2 had more than twice the recurrence odds compared with grade 1 at 5 years, whereas grade 3 showed even higher odds ratio of 3.47 (95% CI, 1.98-6.08). Similar patterns were observed in WHO-2 tumors. Interaction analyses suggested that differences between Simpson grades increase with higher Ki-67 proliferation indices. The prognostic significance of Simpson grade varied across anatomic locations, with clear associations observed for convexity and skull-base meningiomas, but limited relevance for parasagittal/falx meningiomas irrespective of WHO grade. Simpson grading remained strongly associated with recurrence after GTR. These findings suggest that the extent of dural and bony clearance remains clinically relevant; however, Simpson grading should be contextualized within tumor biology and anatomic constraints. Future recurrence models integrating molecular and surgical variables may refine prognostic frameworks rather than current standalone classification systems.