Prognosis and Prediction of Asymptomatic Intracranial Hemorrhage After Endovascular Thrombectomy: A Multi-Center Study.

Abstract

The impact of asymptomatic intracranial hemorrhage (aICH) on functional outcomes after endovascular thrombectomy (EVT) remains unclear, and tools for forecasting this complication are lacking. We aim to evaluate the clinical relevance of aICH and establish a prediction model. Data of patients who received EVT for acute anterior-circulation large vessel occlusion in 3 comprehensive hospitals were retrospectively analyzed. Asymptomatic intracranial hemorrhage was defined as any hemorrhage detected after EVT that did not fulfill the definition of symptomatic intracranial hemorrhage in the European Cooperative Acute Stroke Study. Logistic regression models were performed to assess the impact of aICH on 90-day functional outcomes and identify the predictors of aICH, which were then used to establish a prediction model. The discrimination, calibration, and clinical utility of the model were evaluated. This study included 460 patients, among whom 152 (33.0%) developed aICH after EVT. Asymptomatic intracranial hemorrhage was negatively associated with 90-day excellent outcomes (adjusted odds ratio [OR]: 0.414, 95% confidence interval [CI]: 0.230-0.745, p=0.003) and good outcome (adjusted OR: 0.603, 95% CI: 0.374-0.971, p=0.037), but not with mortality (adjusted OR: 1.110, 95% CI: 0.611-2.017, p=0.732) after adjusted for other predictors of functional outcome. Pre-stroke anticoagulant therapy (OR: 2.233, 95% CI: 1.073-4.647, p=0.032), Alberta stroke program early CT score (OR: 0.842, 95% CI: 0.754-0.939, p=0.002), site of occlusion (internal carotid artery occlusion as the reference; M1 segment of middle cerebral artery occlusion, OR: 2.827, 95% CI: 1.409-5.674, p=0.003; tandem occlusion, OR: 3.928, 95% CI: 1.752-8.806, p=0.001), intravenous thrombolysis (OR: 2.091, 95% CI: 1.362-3.209, p=0.001), and successful recanalization (OR: 0.383, 95% CI: 0.213-0.689, p=0.001) were identified as the predictors of aICH, which were incorporated into a nomogram model. The area under the receiver operating characteristic curve of the model was 0.707 (95% CI: 0.657-0.757), and the calibration plot demonstrated good consistency between actual observed and predicted probability of aICH. Decision curve analysis showed that patients might benefit from the model. Asymptomatic intracranial hemorrhage was negatively associated with favorable functional outcome after EVT. We established a nomogram model for predicting aICH, which requires external clinical validation. The impact of asymptomatic intracranial hemorrhage after endovascular thrombectomy on mid-term functional outcome has been controversial. We found that asymptomatic intracranial hemorrhage may also decreased the likelihood of 90-day favourable functional outcome after endovascular thrombectomy, supporting the notion that asymptomatic intracranial hemorrhage at the acute stage may not be benign. Moreover, we established a prediction model for this complication, which may improve clinical evaluation and management of patients who would receive endovascular thrombectomy for large vessel occlusion.