Refining the Edinburgh Computed Tomography Criteria with a 30 mL Hematoma Volume Threshold

The simplified Edinburgh criteria utilize computed tomography (CT) to identify cerebral amyloid angiopathy (CAA)-related intracerebral hemorrhage (ICH), showing good performance in the original autopsy cohort. However, diagnostic performance varies in external validation, particularly among survivors with smaller hematoma volumes. We hypothesized that applying a minimum volume threshold could improve diagnostic accuracy.

We externally validated the simplified Edinburgh criteria in a Chinese cohort. All patients underwent CT imaging as the index test. The reference standard was a diagnosis of probable CAA, with or without pathological confirmation, based on the Boston 2.0 criteria. Diagnostic indices, including the area under the curve, were assessed. Additionally, likelihood ratio differences were analyzed to explore a potential hematoma volume threshold associated with improved diagnostic performance.

A total of 171 patients were included: 52 lobar ICH cases with histological confirmation and 119 with magnetic resonance imaging (MRI) data. Patients undergoing hematoma evacuation had larger ICH volumes and were younger than those assessed by MRI. Overall, the simplified Edinburgh criteria showed limited sensitivity and specificity for both rule-in criteria (65% and 54%) and rule-out criteria (56% and 58%). The pathology-confirmed subgroup showed higher rule-in specificity than the MRI-based subgroup (87% vs. 43%). In patients with hematoma volume >30 mL, rule-in specificity increased to 71.4% and rule-out sensitivity increased to 89.3%, although overall diagnostic performance remained moderate.

The simplified Edinburgh CT criteria may yield more accurate diagnoses in patients with larger hematomas, and a 30 mL threshold may serve as a potential exploratory cutoff.