Location-specific differences in hematoma volume predict outcomes in patients with spontaneous intracerebral hemorrhage

Background and objective Functional outcome after spontaneous intracerebral hemorrhage (ICH) may vary depending on hematoma volume and location. We assessed the interaction between hematoma volume and location, and modified the original ICH score to include such an interaction. Methods Consecutive ICH patients were enrolled in the Intracerebral Hemorrhage Outcomes Project from 2009 to 2017. Inclusion criteria were age >= 18 years, baseline modified Rankin Scale (mRS) score 0-2, neuroimaging, and follow-up. Functional dependence and mortality were defined as 90-day mRS>2 and death, respectively. A location ICH score was developed using multivariable regression and area under the receiver operator characteristic curve (AUROC) analyses. Results The study cohort comprised 311 patients, and the derivation and validation cohorts comprised 209 and 102 patients, respectively. Interactions between hematoma volume and location predicted functional dependence (p = 0.008) and mortality (p = 0.025). The location ICH score comprised age >= 80 years (1 point), Glasgow Coma Scale score (3-9 = 2 points; 10-13 = 1 point), volume-location (lobar:>= 24 mL=2 points, 21-24 mL=1 point; deep:>= 8 mL=2 points, 7-8 mL=1 point; brainstem:>= 6 mL=2 points, 3-6 mL=1 point; cerebellum:>= 24 mL=2 points, 12-24 mL=1 point), and intraventricular hemorrhage (1 point). AUROC of the location ICH score was higher in functional dependence (0.883 vs. 0.770, p = 0.002) but not mortality (0.838 vs. 0.841, p = 0.918) discrimination compared to the original ICH score. Conclusions The interaction between hematoma volume and location exerted an independent effect on outcomes. Excellent discrimination of functional dependence and mortality was observed with incorporation of location-specific volume thresholds into a prediction model. Therefore, the volume-location relationship plays an important role in ICH outcome prediction.