Stroke pattern interpretation: the variability of hypertensive versus amyloid angiopathy hemorrhage

It is commonly felt that cerebral amyloid angiopathy (CAA) related intracerebral hemorrhage (ICH) can be distinguished from hypertension (HTN)-related ICH by certain typical features on computerized tomography (CT) and magnetic resonance imaging (MRI). The purpose of this study was to investigate the performance of clinicians who were asked to differentiate between CAA and HTN based on hemorrhage pattern interpretation and to assess the feasibility of such classification.

The admission scans from 83 patients who were admitted to our service with an acute ICH were presented to 5 clinicians in a randomized and blinded fashion (1 junior, and 1 senior neurosurgical resident, 1 attending neurosurgeon, and 2 neurosurgeon-neuroradiologists). There were no patients who received oral anticoagulants other than low-dose aspirin, or who suffered from vascular malformations or tumors. Scans from 41 patients with a histologically proven diagnosis of CAA and from 42 patients with a clear history of HTN were investigated. Hematoma evacuation was done in all CAA patients and in 59% of HTN patients (n = 25).

The overall average classification accuracy was 66.8% (range: 62.7-69.9). For correct HTN classification it was 69.5% (range: 64.3-81), and 63.9% for CAA, respectively (range: 48.9-75.6). There were negligible differences in classification accuracy among all observers. Patients with a CAA-related ICH were significantly older than patients with a HTN-related ICH (74 vs. 66.5 years, p < 0.05). There was a significantly higher number of hematomas >30 ml in CAA (85.3%) when compared with HTN (59.5%). No basal ganglionic hemorrhage was seen in CAA, but in 40.5% in HTN. Intraventricular hemorrhage was seen in 24.4% in CAA, and in 26.2% in HTN. Two patients (4.9%) with CAA, and 7 patients with HTN (16.7%) presented with cerebellar hematomas.

Three of 10 scans were not correctly diagnosed regardless of the examiner's level of training. This calls into question the reliability of classifying the underlying pathological condition based on hemorrhage pattern interpretation on CT or MRI. The definite diagnosis of CAA- versus HTN-related hemorrhage requires a histopathological confirmation and should not be based solely on hemorrhage pattern interpretation.