Related Experiment Video
Updated: Sep 16, 2025
Comprehensive Endovascular and Open Surgical Management of Cerebral Arteriovenous Malformations
Published on: October 20, 2017
Predictive factors of external ventricular drain placement in ruptured brain arteriovenous malformations with
Idriss Haffaf1, Jonathan Cortese2, Lorenzo Garzelli2
1Department of Neuroradiology, Pitié-Salpêtrière University Hospital, Sorbonne University, Paris, France. idrisshaffaf75@gmail.com.
Background Ad Purpose:
Intraventricular (IV) hemorrhage is a common finding after brain arterio-venous malformation (BAVM) rupture and has been associated with poor prognosis. This report investigates previously published IV hemorrhage quantification scores as predictors of external ventricular drain (EVD) placement following BAVM rupture, along with other clinical and radiological parameters.
Methods:
One hundred and twenty-nine patients with BAVM rupture and IV hemorrhage were included (133 ruptures), all admitted in our institution between January 1, 2003 and May 15, 2020. IV hemorrhage score, LeRoux, Graeb and modified Graeb score were compared as predictors of EVD placement using receiver operating (ROC) curves. Multivariate models were constructed for prediction of EVD placement and to define radiological hydrocephalus. Multivariate logistic regression was also used to test whether the different IV hemorrhage quantification scores were predictors of death or in-hospital mortality.
Results:
The four IV hemorrhage quantification scores performed similarly (area under ROC curves comprised between 0.65 and 0.67) for EVD placement. In multivariate analysis, only elevated S-100B protein serum level (p = 0.01) and radiological hydrocephalus (p < 0.001) were associated with EVD placement. None of the four IV hemorrhage quantification scores were predictive of EVD placement. Among all radiological features, only the occipital horn index > 0.12 (p < 0.001), the frontal horn index > 0.13 (p = 0.04) and V3 index > 0.05 (p = 0.006) were independently associated with hydrocephalus. None of the four IV hemorrhage quantification scores were predictive of death or in-hospital mortality.
Conclusions:
IV hemorrhage scores are reproducible with high inter-observer agreement but were not predictive of EVD placement. Only hydrocephalus and elevated S-100B protein serum level, a surrogate for clinical severity, were independently associated with EVD placement in our study.
