PURPOSE: To understand clinical significance of irregular interface between meningioma and adjacent brain parenchyma in predicting histological grading of tumor, focusing on brain parenchymal invasion. MATERIALS AND METHODS: Pathologically confirmed 79 cases with meningiomas with pathological reports about the presence of parenchymal invasion were included. We defined the presence of interface irregularity as either spiculations or fuzzy margins between the tumor and brain parenchyma. We counted number of spiculations and measured ratio of fuzzy margin length to whole length of mass with consensus of two neuroradiologists. We classified the patients into Present group and Absent group, and the two groups were compared by using the Mann-Whitney U test. Statistical correlations between the presence of an interface irregularity and brain parenchymal invasion by the tumor as well as meningioma histological grade were tested with chi-square test. The optimal cutoff values of spiculation numbers and the ratio of fuzzy margins were determined. The sensitivity and specificity of number of spiculations, ratio of fuzzy margin and the presence of irregular interface as combined parameters for predicting the parenchymal invasion were calculated using ROC curve analysis. RESULTS: Statistically significant differences were noted between the Present and Absent groups for number of spiculations and ratio of fuzzy margin (P = 0.038 and P = 0.028, respectively). The optimal cutoff value for number of spiculations (> 4.5 with 61.1% sensitivity and 68.9% specificity) and the ratio of fuzzy margin (> 0.24 with 66.7% sensitivity and 65.6% specificity) were determined. The sensitivity and specificity of interface irregularity as the combined parameters were 72% and 59%, respectively. The interface irregularity between tumor and brain parenchyma significantly correlated with not only brain parenchymal invasion (P = 0.001) and but also histological grade (P < 0.001). CONCLUSION: The interface irregularity between tumor and brain parenchyma in MRI can be a strong predictive factor for brain parenchymal invasion and high grade meningioma.
Brain* ; Consensus ; Humans ; Magnetic Resonance Imaging ; Meningioma* ; Pathology* ; ROC Curve ; Sensitivity and Specificity