Yu-Chen Shih, Yu-Jui Wang, Bo-Ching Lee, Koping Chang, Jiann-Shing Jeng, Li-Kai Tsai, Hsin-Hsi Tsai
The simplified Edinburgh criteria may have limited applicability for Asian populations. The proposed Asian CT criteria offer improved discrimination and may facilitate CT-based cerebral amyloid angiopathy diagnosis in Asian patients with lobar intracerebral hemorrhage.
BACKGROUND: The simplified Edinburgh criteria enable diagnosis of cerebral amyloid angiopathy-associated lobar intracerebral hemorrhage when magnetic resonance imaging is unavailable. However, their applicability to Asian populations, which have higher frequencies of deep perforator arteriopathy, remains uncertain.
METHODS: We retrospectively analyzed 2 cohorts of patients with spontaneous lobar intracerebral hemorrhage at National Taiwan University Hospital. Patients in the magnetic resonance imaging cohort (2015-2022) who underwent acute computed tomography (CT) and magnetic resonance imaging were classified using the modified Boston criteria. CT imaging predictors were selected by elastic net penalized logistic regression following prespecified clinical exclusions, and a logistic regression model with risk stratification was developed. Bootstrap optimism correction with uniform shrinkage was applied for internal validation. Histopathologic validation was performed in an independent cohort from 2016 to 2023.
RESULTS: Eighty-seven patients were analyzed. The Edinburgh criteria yielded a suboptimal area under the receiver operating characteristic curve of 0.654 (95% CI, 0.547-0.761). The Asian CT criteria (absence of old lacune, cortical involvement, and presence of cortical atrophy) achieved an apparent area under the receiver operating characteristic curve of 0.843 (optimism-corrected 0.834; DeLong P=0.002 versus Edinburgh). The rule-in specificity for the high-risk group was 85.1% (95% CI, 72.3%-92.6%), and the rule-out sensitivity for the low-risk group was 85.0% (95% CI, 70.9%-92.9%). Histopathologic validation confirmed a rule-out sensitivity of 83.3% and a rule-in specificity of 92.9%. Decision curve analysis demonstrated superior net benefit across threshold probabilities from 2% to 79%.
CONCLUSIONS: The simplified Edinburgh criteria may have limited applicability for Asian populations. The proposed Asian CT criteria offer improved discrimination and may facilitate CT-based cerebral amyloid angiopathy diagnosis in Asian patients with lobar intracerebral hemorrhage.