Abstract
BACKGROUND AND PURPOSE: This study aimed to identify predictors of acute mortality after intracerebral hemorrhage (ICH), including voxel-wise analysis of hematoma location. METHODS: In 282 consecutive patients with acute ICH, clinical and radiological predictors of acute mortality were identified. Voxel-based lesion-symptom mapping examined spatial correlates of acute mortality, contrasting results in basal ganglia ICH and lobar ICH. RESULTS: Acute mortality was 47.9%. In bivariate analyses, one clinical (serum glucose) and two radiological (hematoma volume and intraventricular extension) measures significantly predicted mortality. The relationship was strongest for hematoma volume. Multivariable modeling identified four significant predictors of mortality (ICH volume, intraventricular extension, serum glucose, and serum hemoglobin), although this model only minimally improved the predictive value provided by ICH volume alone. Voxel-wise analysis found that for patients with lobar ICH, brain regions where acute hematoma was significantly associated with higher acute mortality included inferior parietal lobule and posterior insula; for patients with basal ganglia ICH, a large region extending from cortex to brainstem. CONCLUSIONS: For patients with lobar ICH, acute mortality is related to both hematoma size and location, with findings potentially useful for therapeutic decision making. The current findings also underscore differences between the syndromes of acute deep and lobar ICH.