MRI Reflects Meningioma Biology and Molecular Risk.

Area of Science:

• Neuro-oncology
• Radiomics
• Genomics

Background:

• Meningioma molecular landscape understanding has advanced via epigenomic studies (cIMPACT-NOW update 8).
• Molecular data integration into risk-adapted treatment algorithms is increasing.
• The utility of non-invasive MRI in reflecting meningioma molecular variation and risk is unclear.

Purpose of the Study:

• To assess the capability of preoperative MRI radiomic features to predict meningioma molecular characteristics and risk.
• To correlate MRI-derived features with histopathological and molecular profiles of meningiomas.

Main Methods:

• Analysis of 225 newly diagnosed meningiomas (WHO grades 1-3) with preoperative MRI and molecular profiling.
• Automated segmentation of tumor core and edema regions from MRI data.
• Radiomic feature extraction and Random Forest model training for predicting WHO grade, molecular risk, and 1p loss.

Main Results:

• High accuracy (91%) in predicting integrated molecular risk and 87.5% for 1p chromosomal status.
• Robust performance with Area Under the Curve (AUC) > 0.89 for molecular predictions.
• Moderate accuracy (76.8%) for WHO grade prediction, indicating MRI's stronger correlation with molecular features.

Conclusions:

• Preoperative MRI effectively captures meningioma molecular biology, potentially enabling rapid risk assessment.
• Radiomic analysis of MRI shows significant promise for non-invasive molecular subtyping and risk stratification.
• Current MRI accuracy for WHO grade prediction is insufficient for direct clinical application, necessitating further research.