Meta-Analysis of Accuracy of the Spot Sign for Predicting Hematoma Growth and Clinical Outcomes

Abstract

Background and Purpose— The computed tomography angiographic spot sign refers to contrast leakage within intracerebral hemorrhage (ICH). It has been proposed as a surrogate radiological marker for ICH growth. We conducted a meta-analysis to study the accuracy of the spot sign for predicting ICH growth and mortality. Methods— PubMed, Medline, conference proceedings, and article references in English up to June 2017 were searched for studies reporting “computed tomography angiography” and “spot sign” or “intracerebral hemorrhage” and “spot sign.” Each study was ranked on 27 criteria resulting in a quality rating score. Bivariate random effect meta-analysis was used to calculate positive and negative likelihood ratios and area under summary receiver operating characteristics curve for ICH growth and mortality. Hematoma growth was defined using the change in ≥6 mL or ≥33% increase in volume. Results— There were 26 studies describing 5085 patients, including 15 studies not used in previous meta-analyses. Positive likelihood ratio and negative likelihood ratio for ICH growth were 4.85 (95% CI, 3.85–6.02; I 2 =76.1%) and 0.49 (95% CI, 0.40–0.58) and mortality were 4.65 (95% CI, 3.67–5.90) and 0.55 (95% CI, 0.40–0.69), respectively. For ICH growth, the pooled sensitivity was 0.57 (95% CI, 0.49–0.64) and pooled false positive rate was 0.12 (95% CI, 0.09–0.14). The post-test probability of ICH growth was 0.57. The area under the curve for ICH growth and mortality was 0.86 and 0.87 (CIs are not provided in bivariate method). Meta-regression showed sensitivity of the test to decline significantly with subsequent year of publication (β=−0.148; 95% CI, −0.295 to −0.001; P =0.05). Higher quality assessment is associated with lower false positive rate (β=−0.074; 95% CI, −0.126 to −0.022; P =0.006). Conclusions— The high area under the curve potentially suggests that the spot sign can predict hematoma growth and mortality. Caution is recommended in its application given the heterogeneity across studies, which is appropriate given the data.