The Added Value of Vessel Wall MRI Over CTA for Differentiating Large-Artery Atherosclerotic and Cardioembolic Stroke in the Anterior and Posterior Circulations.

Abstract

Background and purpose: Differentiation between stroke etiologies is essential for secondary prevention; however, etiologic classification remains challenging in many ischemic strokes. We evaluated whether magnetic resonance vessel wall imaging (VWI) plaque markers improve computed tomography angiography (CTA)-based differentiation between cardioembolic (CE) and large-artery atherosclerotic (LAA) stroke in the anterior and posterior circulations, as a step toward improving etiologic classification beyond current criteria.

Methods: We retrospectively included patients with unilateral anterior or posterior circulation ischemic stroke between June 2022 and April 2024 across an academic health system, classified as LAA or CE according to modified TOAST criteria. Imaging markers included CTA-derived noncalcified plaque extent (>30% stenosis), VWI-derived intracranial plaque enhancement (IPE) extent, and intraplaque hemorrhage (IPH) presence, restricted to arterial segments upstream from the infarct. Given potential territory-specific plaque differences, analyses were stratified by circulation. Multivariable logistic regression models assessed associations of CTA and VWI markers with stroke etiology. Associations with LAA were summarized using adjusted odds ratios (aOR; plaque extent variables per 1-SD increase). Overall performance for discriminating between LAA and CE etiologies was assessed using the area under the receiver operating characteristic curve (AUC).

Results: Among 193 patients (median age, 70 years; 60% male), 137 had anterior circulation strokes (65 LAA, 72 CE) and 56 had posterior circulation strokes (38 LAA, 18 CE). In anterior circulation strokes, VWI markers improved discrimination over CTA alone (AUC, 0.57 for CTA alone vs 0.76 for CTA+VWI; p<0.001); CTA-derived noncalcified plaque extent (aOR, 1.7; p=0.02), IPH (aOR, 7.0; p<0.001), and IPE extent (aOR, 1.9; p=0.005) were independently associated with LAA. In posterior circulation strokes, IPE extent improved discrimination over CTA alone (AUC, 0.77 for CTA alone vs 0.84 for CTA+VWI; p=0.02); IPE extent (aOR, 2.2; p=0.02) and CTA-derived noncalcified plaque extent (aOR, 5.9; p=0.002) were independently associated with LAA. IPH was not significantly associated with posterior circulation stroke etiology (p=0.26).

Conclusions: VWI adds complementary value for differentiating LAA and CE stroke etiologies in both vascular territories, supporting a pragmatic CTA-VWI approach for etiologic characterization beyond stenosis assessment alone.