Development of a Clinicopathological-Radiomics Model for Predicting Progression and Recurrence in Meningioma Patients
Mengna He,Xiaolan Wang,Chencui Huang,Xiting Peng,Ning Li,Feng Li,Hao Dong,Zhengyang Wang,Lingli Zhao,Fengping Wu,Minming Zhang,Xiaojun Guan,Xiaojun Xu
DOI: https://doi.org/10.1016/j.acra.2023.10.059
IF: 5.482
2023-11-01
Academic Radiology
Abstract:RATIONALE AND OBJECTIVES: Tumor progression and recurrence(P/R)after surgical resection are common in meningioma patients and can indicate poor prognosis. This study aimed to investigate the values of clinicopathological information and preoperative magnetic resonance imaging (MRI) radiomics in predicting P/R and progression-free survival (PFS) in meningioma patients.METHODS AND MATERIALS: A total of 169 patients with pathologically confirmed meningioma were included in this study, 54 of whom experienced P/R. Clinicopathological information, including age, gender, Simpson grading, World Health Organization (WHO) grading, Ki-67 index, and radiotherapy history, as well as preoperative traditional radiographic findings and radiomics features for each MRI modality (T1-weighted, T2-weighted, and enhanced T1-weighted images) were initially extracted. After feature selection, the optimal performance was estimated among the models established using different feature sets. Finally, Cox survival analysis was further used to predict PFS.RESULTS: Ki-67 index, Simpson grading, WHO grading, and radiotherapy history were found to be independent predictors for P/R in the multivariate regression analysis. This clinicopathological model had an area under the curve (AUC) of 0.865 and 0.817 in the training and testing sets, respectively. The performance of the combined radiomics model reached 0.85 and 0.84, respectively. A clinicopathological-radiomics model was then established, which significantly improved the prediction of meningioma P/R (AUC = 0.93 and 0.88, respectively). Finally, the risk ratio was estimated for each selected feature, and the C-index of 0.749 was obtained.CONCLUSION: Radiomics signatures of preoperative MRI have the ability to predict meningioma at the risk of P/R. By integrating clinicopathological information, the best performance was achieved.
radiology, nuclear medicine & medical imaging