Border-zone Infarction on Diffusion-weighted Imaging for Etiological Stratification of Anterior Circulation Large...

Accurate differentiation of intracranial atherosclerotic disease-related large vessel occlusion from cardioembolic large vessel occlusion is important for understanding stroke etiology and optimizing endovascular treatment strategies. Although computed tomography-based diagnostic models have been proposed, magnetic resonance imaging-based models remain less well established. This study investigated the diagnostic value of border-zone infarction on diffusion-weighted imaging and evaluated predictive models incorporating border-zone infarction for identifying intracranial atherosclerotic disease-related large vessel occlusion. We retrospectively analyzed consecutive patients who underwent endovascular treatment for anterior circulation large vessel occlusion between April 2015 and April 2025. Intracranial atherosclerotic disease-related large vessel occlusion was defined as >50% residual stenosis after thrombectomy, or when dynamic intraprocedural re-occlusion occurred after initial reperfusion that was indicative of in situ thrombosis. Diagnostic performance of the ABC²D score was evaluated and compared with models incorporating border-zone infarction. Among 149 patients, 24 (16%) had intracranial atherosclerotic disease-related large vessel occlusion. Multivariable analysis identified absence of atrial fibrillation, absence of the hyperdense artery sign, and presence of border-zone infarction as independent predictors of intracranial atherosclerotic disease-related large vessel occlusion. Border-zone infarction indicated high specificity (94%) and the highest positive predictive value (64%). The ABC²D model achieved an area under the curve of 0.837, which improved to 0.912 with the addition of border-zone infarction. Furthermore, a simplified 2-variable model comprising only the absence of atrial fibrillation and the presence of border-zone infarction showed a high practical discriminative ability (area under the curve, 0.883). Border-zone infarction on diffusion-weighted imaging is a highly specific marker for intracranial atherosclerotic disease-related large vessel occlusion that significantly improves established computed tomography-based diagnostic models. A simplified 2-variable model (atrial fibrillation absence and border-zone infarction) offers a practical tool for rapid etiological triage. Prospective multicenter validation is warranted.