NON-INVASLOGY USING MULTIPARAMETRIC MRI, RADIOMICS, AND INTEGRATED MOLECULAR-MORPHOLOGICAL MARKERS: A SINGLE-CENTER OBSERVATIONAL STUDY

Abstract

Meningiomas are biologically heterogeneous intracranial tumors ranging from indolent World Health Organization (WHO) grade 1 neoplasms to highly proliferative, brain-invasive, recurrent, and molecularly aggressive lesions. Although histopathological grading remains fundamental, morphology alone does not always accurately predict tumor behavior. This study aimed to develop an integrated clinical-radiological-radiomic model for the non-invasive prediction of aggressive meningioma biology and to assess its relationship with WHO grade, Ki-67/MIB-1 proliferative index, brain invasion, molecular risk markers, and early recurrence. This single-center observational model included 104 patients with newly diagnosed intracranial meningiomas who underwent preoperative multiparametric MRI followed by surgical resection and histopathological evaluation. MRI assessment included T1- and T2-weighted imaging, contrast-enhanced imaging, FLAIR, diffusion-weighted imaging, apparent diffusion coefficient mapping, susceptibility-weighted imaging, and perfusion imaging when available. Conventional imaging variables included tumor volume, margin configuration, peritumoral edema, enhancement heterogeneity, necrosis, bone or venous invasion, cerebrospinal fluid cleft integrity, and brain-tumor interface. Radiomic analysis evaluated intensity, shape, texture, wavelet-derived, and tumor heterogeneity features. Histopathological and molecular assessment included WHO grade, Ki-67/MIB-1, brain invasion, NF2 status, TERT promoter mutation, CDKN2A/B deletion, and DNA methylation profile. An original MRI-Based Integrated Meningioma Aggressiveness Score was developed. The cohort included 76 WHO grade 1, 24 grade 2, and 4 grade 3 tumors. High-risk meningiomas demonstrated larger volume, lower apparent diffusion coefficient, greater edema, irregular margins, heterogeneous enhancement, brain invasion, and increased radiomic heterogeneity. The area under the receiver operating characteristic curve was 0.82 for conventional MRI, 0.89 for radiomics, and 0.93 for the integrated model. Integrated multiparametric MRI, radiomics, and molecular-morphological assessment may improve preoperative risk stratification and support personalized surgical planning, postoperative surveillance, and selection for adjuvant treatment.