頭蓋内髄膜腫における高悪性度を特定するためのルーチンMRIパラメータの診断価値

To evaluate whether preoperative MRI features can accurately predict WHO grade 2/3 meningiomas using routinely available imaging parameters.

This retrospective single-centre study included 282 patients (mean age 58.4 ± 13.4 years; 176 women) who underwent surgery for histologically confirmed meningiomas between 2010 and 2024. Preoperative MRI was assessed for tumour volume, peritumoral oedema, heterogeneous enhancement, cystic components, midline shift, and normalized apparent diffusion coefficient (NADC). Univariable and multivariable logistic regression analyses identified independent predictors of high-grade meningiomas (WHO grade 2/3). Model performance was evaluated with ROC curve analysis.

Among 282 meningiomas, 78% were WHO grade 1 and 22% were grade 2 or 3. High-grade tumours were significantly associated with tumour volume ≥ 45 cc (p= 0.016), presence of cystic components (p = 0.005), midline shift > 3 mm (p = 0.024), and NADC < 1.07 (p = 0.028). These features remained independently associated with higher grade in the multivariable model. The resulting model achieved an AUC of 0.768 (95% CI: 0.689-0.847), with 75.5% sensitivity and 62.8% specificity at the optimal threshold.

We obtained a multivariable MRI-based model combining morphologic and diffusion features that potentially enables preoperative risk stratification of meningiomas and supports surgical decision-making. Further validation in prospective, multicentre cohorts is warranted.