MRI and CT imaging biomarkers of cerebral amyloid angiopathy in lobar intracerebral hemorrhage

Abstract

Background: Cerebral amyloid angiopathy (CAA), a common cause of intracerebral hemorrhage (ICH), is diagnosed using the Boston criteria including magnetic resonance imaging (MRI) biomarkers (cerebral microbleeds (CMBs) and cortical superficial siderosis (cSS). The simplified Edinburgh criteria include computed tomography (CT) biomarkers (subarachnoid extension (SAE) and finger-like projections (FLPs)). The underlying mechanisms and diagnostic accuracy of CT compared to MRI biomarkers of CAA are unknown. Methods: We included 140 survivors of spontaneous lobar supratentorial ICH with both acute CT and MRI. We assessed associations between MRI and CT biomarkers and the diagnostic accuracy of CT- compared to MRI-based criteria. Results: FLPs were more common in patients with strictly lobar CMB (44.7% vs 23.5%; p = 0.014) and SAE was more common in patients with cSS (61.3% vs 31.2%; p = 0.002). The high probability of the CAA category of the simplified Edinburgh criteria showed 87.2% (95% confidence interval (CI): 78.3–93.4) specificity, 29.6% (95% CI: 18.0–43.6) sensitivity, 59.3% (95% CI: 38.8–77.6) positive predictive value, and 66.4% (95%: CI 56.9–75.0) negative predictive value, 2.3 (95% CI: 1.2–4.6) positive likelihood ratio and 0.8 (95% CI 0.7–1.0) negative likelihood ratio for probable CAA (vs non-probable CAA), defined by the modified Boston criteria; the area under the receiver operating characteristic curve (AUROC) was 0.62 (95% CI: 0.54–0.71). Conclusion: In lobar ICH survivors, we found associations between putative biomarkers of parenchymal CAA (FLP and strictly lobar CMBs) and putative biomarkers of leptomeningeal CAA (SAE and cSS). In a hospital population, CT biomarkers might help rule-in probable CAA (diagnosed using the Boston criteria), but their absence is probably not as useful to rule it out, suggesting an important continued role for MRI in ICH survivors with suspected CAA.