Cerebral amyloid angiopathy-related acute lobar intra-cerebral hemorrhage: diagnostic value of plain CT

Abstract

Diagnosing probable cerebral amyloid angiopathy (CAA) after lobar intra-cerebral hemorrhage (l-ICH) currently relies on the MR-based modified Boston criteria (mBC). However, MRI has limited availability and the mBC have moderate sensitivity, with isolated l-ICH being classified as “possible CAA”. A recent autopsy-based study reported potential value of finger-like projections (FLP) and subarachnoid hemorrhage (SAH) on acute CT. Here we assessed these markers’ performance in a cohort most of whom survived the index episode. We included all patients from a prospective pathology database with non-traumatic l-ICH, admission CT and available tissue sample showing no alternative cause. CT was assessed by two blinded independent neuroradiologists. Interobserver reproducibility was almost perfect for SAH and substantial for FLP. Sixteen patients were eligible [age 65.8\u2009±\u20097.2 yrs; hematoma volume: 39(26, 71)mls; hematoma evacuation sample 15 patients; autopsy one patient]. MRI was available in 11 patients. ICH-related death affected six patients. Aβ40–42 immunohistochemistry revealed CAA in seven patients (44%). SAH and FLP were present in 12/16 (75%) and 10/16 (62%) patients, respectively. SAH had 100% sensitivity for CAA but low specificity; FLP had lower performance. Using either pathology or MRI as reference standard yielded essentially similar results. All patients with possible CAA on MRI but CAA on pathology had SAH. In patients with moderate-size l-ICH who mostly survived the index event, SAH had perfect sensitivity and better performance than FLP. In addition, SAH appeared to add onto MRI in possible CAA, the clinically most relevant scenario. Studies in larger samples are however warranted.