Abstract 2970: Inter-observer Agreement in Diagnosis of “Borderzone” Infarcts On Diffusion Weighted MR Imaging

• Published in: Stroke (1970) 2012-02, Vol.43 (suppl_1)
• Main Authors: Jain, Minal, Kung, Nancy T, Jain, Anunaya, Kanthala, Abhijit R, Jahromi, Babak S
• Format: Article
• Language: English
• ISSN: 0039-2499; 1524-4628
• DOI: 10.1161/str.43.suppl_1.A2970

Abstract only Aim: To evaluate consensus in diagnosis of “borderzone” infarcts on diffusion weighted MR imaging (DWI). Methods: DWI of 144 consecutive stroke patients were independently reviewed by blinded Neuroradiologists (n=2) and Vascular Neurologists (n=2) and classified into embolic versus borderzone infarct patterns. CT angiography studies of each of these patients were later reviewed to determine degree of ipsilateral carotid artery stenosis/occlusion by NASCET criteria. Subjects with any MCA occlusion/stenosis on any side were excluded. Results: The agreement between Neuroradiologists was fair (Kappa=0.22, 95% CI 0.089 to 0.351), agreeing on 61.8% as embolic and 7.64% as borderzone infarcts. The agreement between Vascular Neurologists was good (Kappa=0.722, 95% CI 0.55 to 0.893), agreeing on 84.14% as embolic and 9.66% as borderzone infarcts. Inter-rater agreement between Neuroradiologist and Vascular Neurologist pairs varied between fair to moderate (Kappa = 0.22 to 0.72). Overall Fleiss’ Kappa for all 4 raters was 0.373 (95% CI 0.306 to 0.439). Classification of infarct patterns by degree of carotid disease on CTA is shown in the table below. As is seen in the table above, infarct patterns in patients with 100% carotid occlusion were classified as borderzone by both Neuroradiologist and Vascular Neurologist pairs in less than 30% of patients. The Neuroradiologist pair commonly classified infarct patterns in 75% of patients with 70-99% stenosis as borderzone. The Vascular Neurologists too classified 80% of these patients with 70-99% stenosis as having a borderzone infarct. Amongst patients with 90% of infarct patterns as embolic. Conclusions: Carotid occlusion and near-occlusion were unexpectedly associated with a smaller proportion of borderzone infarction than 70-99% stenosis. Discrepancies between Neuroradiologists and Vascular Neurologists towards diagnosis of watershed infarcts suggest the need to develop common definitions/guidelines. MR DWI imaging may not be a sole/sufficient criterion to classify watershed versus embolic infarction.