Independent Validation of the Hematoma Expansion Prediction Score: A Non-contrast Score Equivalent in Accuracy to the...

The computed tomography angiography (CTA) spot sign is widely used to assess the risk of hematoma expansion following acute intracerebral hemorrhage (ICH). However, not all patients can receive intravenous contrast nor are all hospital systems equipped with this technology. We aimed to independently validate the Hematoma Expansion Prediction (HEP) Score, an 18-point non-contrast prediction scale, in an external cohort and compare its diagnostic capability to the CTA spot sign.

We performed a retrospective analysis of the predicting hematoma growth and outcome in intracerebral hemorrhage using contrast bolus CT (PREDICT) Cohort Study. Primary outcome was significant hematoma expansion (≥ 6 mL or ≥ 33%). We generated a receiver operating characteristic (ROC) curve comparing the HEP score to significant expansion. We calculated sensitivity, specificity, positive and negative predictive values (PPV/NPV) for each score point. We determined independent predictors of significant hematoma expansion via logistic regression.

A total of 292 patients were included in primary analysis. Hematoma growth of ≥ 6 mL or ≥ 33% occurred in 94 patients (32%). The HEP score was associated with significant expansion (adjusted odds ratio [aOR] 1.14, 95% confidence interval [CI] 1.01-1.30). ROC curves comparing HEP score to significant expansion had an area under the curve of 0.64 (95% CI 0.57-0.71). Youden's method showed an optimum score of 4. HEP Scores ≥ 4 (n = 100, sensitivity 49%, specificity 73%, PPV 46%, NPV 75%, aOR 1.99, 95% CI 1.09-3.64) accurately predicted significant expansion. PPV increased with higher HEP scores, but at the cost of lower sensitivity. The diagnostic characteristics of the spot sign (n = 82, Sensitivity 49%, Specificity 81%, PPV 55%, NPV 76%, aOR 2.95, 95% CI 1.61-5.42) were similar to HEP scores ≥ 4.

The HEP score is predictive of significant expansion (≥ 6 mL or ≥ 33%) and is comparable to the spot sign in diagnostic accuracy. Non-contrast prediction tools may have a potential role in the recruitment of patients in future intracerebral hemorrhage trials.