Do intracerebral hemorrhage nonexpanders actually expand into the ventricular space?

Background and Purpose-The computed tomographic angiography spot sign as a predictor of hematoma expansion is limited by its modest sensitivity and positive predictive value. It is possible that hematoma expansion in spot-positive patients is missed because of decompression of intracerebral hemorrhage (ICH) into the ventricular space. We hypothesized that revising hematoma expansion definitions to include intraventricular hemorrhage (IVH) expansion will improve the predictive performance of the spot sign. Our objectives were to determine the proportion of ICH nonexpanders who actually have IVH expansion, determine the proportion of false-positive spot signs that have IVH expansion, and compare the known predictive performance of the spot sign to a revised definition incorporating IVH expansion. Methods-We analyzed patients from the multicenter PREDICT ICH spot sign study. We defined hematoma expansion as =6 mL or =33% ICH expansion or >2 mL IVH expansion and compared spot sign performance using this revised definition with the conventional 6 mL/33% definition using receiver operating curve analysis. Results-Of 311 patients, 213 did not meet the 6-mL/33% expansion definition (nonexpanders). Only 13 of 213 (6.1%) nonexpanders had =2 mL IVH expansion. Of the false-positive spot signs, 4 of 40 (10%) had >2 mL ventricular expansion. The area under the curve for spot sign to predict significant ICH expansion was 0.65 (95% confidence interval, 0.58-0.72), which was no different than when IVH expansion was added to the definition (area under the curve, 0.66; 95% confidence interval, 0.58-0.71). Conclusions-Although IVH expansion does indeed occur in a minority of ICH nonexpanders, its inclusion into a revised hematoma expansion definition does not alter the predictive performance of the spot sign.