Hematoma expansion occurs in approximately one-third of patients with intracerebral hemorrhage (ICH), but markers for predicting expansion are not well defined. The PREDICT study, aimed at prospectively validating the spot sign on CT angiography, found the spot sign had a sensitivity of 51% for predicting expansion. Markers present on non-contrast CT head previously published elsewhere as associated with hematoma expansion or larger ICH include hematoma density, shape, presence of fluid-levels, and regularity, but predictive models utilizing these markers are not widely used. Here, the authors hypothesized that these non-contrast CT head markers (irregular margins, heterogeneous density, and fluid-blood levels) would predict expansion both independently and together with the spot sign on CTA, with the aim of identifying non-contrast measures which would be reliable markers of hematoma expansion and to create a more sensitive prediction model.

Analyzing a PREDICT cohort of 311 patients, the authors found a high inter-observer agreement for fluid levels, and relatively good agreement for scoring of heterogeneity and irregularity. All 3 non-contrast CT markers were significantly associated with hematoma expansion on follow up CT scan; in addition, all 3 non-contrast CT markers were significantly associated with the spot sign on CTA and with absolute hematoma growth at 24 hours. Margin irregularity was the most sensitive and fluid levels was the most specific, but the spot sign remained the most specific predictor of hematoma expansion, with the highest positive and negative predictive values of hematoma expansion. The predictive value of the spot sign was not altered when the non-contrast markers were added to the model, confirming the spot sign as the most reliable marker for hematoma expansion. However, given the ease of scoring, the relatively good inter-observer agreement, and the ease of obtaining a non-contrast CT scan, especially in situations where contrast may be contraindicated, it does appear that heterogeneity, irregularity, and fluid-blood levels might also be appropriate surrogate markers.