Tumor volume and the dural tail sign enable the differentiation of intracranial solitary fibrous tumor/hemangiopericytoma from high-grade meningioma

• Published in: Clinical neurology and neurosurgery 2021-08, Vol.207, p.106769-106769, Article 106769
• Main Authors: Xiao, Dongdong, Shi, Jiawei, Zhou, Mengting, Yan, Ling, Zhao, Zhen, Hu, Tingting, Guo, Xuebing, Zhao, Hongyang, Yan, Pengfei, Jiang, Xiaobing
• Format: Article
• Language: English
• Subjects: Adult; Brain cancer; Brain Neoplasms - diagnosis; Decision making; Diagnosis, Differential; Differential diagnosis; Dural tail sign; Female; Hemangiopericytoma - diagnosis; High-grade meningioma; Humans; Intracranial solitary fibrous tumor/hemangiopericytoma; Lymphocytes; Magnetic Resonance Imaging; Male; Medical records; Meningeal Neoplasms - diagnosis; Meningioma; Meningioma - diagnosis; Mesenchyme; Middle Aged; Mimicry; Multivariate analysis; Nervous system; Neurology; Neutrophils; Nomograms; Patients; Peripheral blood; Predictive Value of Tests; Radiation therapy; Retrospective Studies; Solitary Fibrous Tumors - diagnosis; Surgery; Tumor Burden; Tumor volume; Tumors; Variables
• ISSN: 0303-8467; 1872-6968
• DOI: 10.1016/j.clineuro.2021.106769

Intracranial solitary fibrous tumor/hemangiopericytoma (SFT/HPC) is a rare mesenchymal neoplasm with imaging features mimicking high-grade meningioma (HGM) and can easily be misdiagnosed. We sought to determine the value of routine preoperative data in differentiating these tumors. Patients with confirmed SFT/HPC or HGM between January 2012 and June 2020 were identified. A total of 28 preoperative variables (including age, sex, tumor location, tumor volume, 10 traditional MRI features, and 14 peripheral blood indices) were collected for each patient. The top features were selected sequentially based on the least absolute shrinkage and selection operator (LASSO) and support vector machines-recursive feature elimination (SVM-RFE) methods. Differentiation and calibration of the classifiers were assessed by receiver operating characteristic (ROC) curves and calibration curves, respectively. Nomograms were constructed based on multivariate analysis. A total of 127 patients, including 29 with SFT/HPC and 98 with HGM, were analyzed. Three features were first selected using the LASSO and SVM-RFE methods, and corresponding models were developed. Although the area under the curve (AUC) of model 1 was the highest, a comprehensive analysis suggested the superiority of model 2, which consisted only of the features tumor volume (TV) and dural tail sign (DTS) (AUC: 0.942, sensitivity: 93.10%, p-value of H-L test: 0.734, Brier score: 0.07). A risk score formula and a nomogram were constructed. TV can be used to effectively identify SFT/HPC and HGM, whereas adding DTS can improve the overall prediction accuracy. As these two variables are routinely available and are easy for clinicians to master, they can provide a powerful reference for clinical decision-making. •Identification of SFT/HPC and HGM.•Tumor volume was the most potent differentiating index.•The prediction efficiency of the model was improved by adding the dural tail sign.•Peripheral blood inflammatory indicators had no clear reference value for differentiation.