A dynamic nomogram for predict individual risk of malignant brain edema after endovascular thrombectomy in acute ischemic stroke

Abstract

Abstract The aim of this study was to develop a dynamic nomogram combining clinical and imaging data to predict malignant brain edema (MBE) after endovascular thrombectomy (EVT) in patients with large vessel occlusion stroke (LVOS). We analyzed the data of LVOS patients receiving EVT at our center from October 2018 to February 2023, and divided a 7:3 ratio into the training cohort and internal validation cohort, and we also prospectively collected patients from another stroke center for external validation. MBE was defined as a midline shift or pineal gland shift > 5 mm, as determined by computed tomography (CT) scans obtained within 7 days after EVT. A nomogram was constructed using logistic regression analysis, and its receiver operating characteristic curve (ROC) and calibration were assessed in three cohorts. A total of 432 patients were enrolled in this study, with 247 in the training cohort, 100 in the internal validation cohort, and 85 in the external validation cohort. MBE occurred in 24% (59) in the training cohort, 16% (16) in the internal validation cohort and 14% (12) in the external validation cohort. After adjusting for various confounding factors, we constructed a nomogram including the clot burden score (CBS), baseline neutrophil count, core infarct volume on CTP before EVT, collateral index, and the number of retrieval attempts. The AUCs of the training cohorts were 0.891 (95% CI 0.840–0.942), the Hosmer–Lemeshow test showed good calibration of the nomogram (P = 0.879). And our nomogram performed well in both internal and external validation data. Our nomogram demonstrates promising potential in identifying patients at elevated risk of MBE following EVT for LVOS.

Keywords

• Predictive model
• Nomogram
• Malignant brain edema
• Endovascular thrombectomy