Agreement between the simplified Edinburgh CT criteria and the Boston MRI criteria v2.0 for cerebral amyloid angiopathy diagnosis in lobar intracerebral hemorrhage from a population-based perspective. Dijon Stroke Registry

BACKGROUND: Cerebral amyloid angiopathy (CAA) is a frequent cause of lobar intracerebral hemorrhage (ICH). While the Boston criteria version 2.0 enable in vivo CAA diagnosis using MRI features, the simplified Edinburgh CT criteria were proposed to overcome the limited access to MRI. We aimed to assess the agreement between them in a population-based setting. METHODS: Patients ≥50 years with lobar ICH were retrieved from the population-based Dijon Stroke Registry, France. Agreement between Edinburgh and Boston criteria was assessed using Cohen's kappa (κ), and a multinomial logistic regression was performed to investigate factors associated with discrepancies. Additionally, characteristics and outcomes were compared between included and excluded patients from the agreement analysis to investigate potential selection bias. RESULTS: Among 164 patients with lobar ICH, 71 had both brain CT and MRI available. Agreement between CT- and MRI-based criteria was poor (observed: 43.7%; κ = 0.16; SD: 0.075; 95% CI: 0.01-0.31). Patients with lack of agreement (n = 40) were older (median: 80.2 versus 73.7 years; p = 0.04), but the difference was no longer significant in multivariable model. Patients not included in the analysis (n = 93, mainly because of lack of MRI: n = 81) were older, had greater clinical, larger ICH volume, higher handicap at discharge, and lower 30-day survival than included patients. CONCLUSION: Agreement between CT and MRI criteria for CAA diagnosis in lobar ICH was poor, which raises questions about the reliability of CT alone for classifying CAA in ICH patients in our population-based registry.