Abstract Number: Esoc2026ot102 the Improve Trial: A Randomised Multicentre Trial to Investigate the Clinical Impact of Mri-Guided Selection of Symptomatic Carotid Stenosis Patients for Revascularisation

Abstract Background and aims Selection of symptomatic carotid stenosis patients for revascularisation remains suboptimal, as current predominantly stenosis-based selection does not fully reflect heterogeneity in stroke risk. IMPROVE calculates the 3-year ipsilateral ischemic stroke risk based on intraplaque haemorrhage on carotid MRI, stenosis degree and clinical characteristics. In our model–based cost-effectiveness analysis IMPROVE-guided selection for revascularisation reduced ipsilateral ischaemic strokes and perioperative strokes/deaths by 35%, while also substantially lowering costs. Methods To evaluate the clinical impact and cost-effectiveness of IMPROVE-guided versus care-as-usual (CAU)-guided selection of symptomatic patients with carotid stenosis for revascularisation. Results This multicentre randomized non-inferiority trial includes 613 recent TIA and non-disabling stroke patients (mRS≤3) with ipsilateral 30–99% carotid stenosis. Patients are randomized to IMPROVE- or CAU-guided revascularisation selection. In the IMPROVE-guided arm, patients with a predicted IMPROVE 3-year ipsilateral ischemic stroke risk ≥10% are recommended for revascularisation and optimised medical therapy (OMT). For patients with 10% risk, OMT-only is recommended. The primary outcome is any stroke or death within 44 days, or ipsilateral ischemic stroke during minimal three years of follow-up. Secondary outcomes include cardiovascular events, quality adjusted life years (QALYs), functional outcomes, and costs. Conclusions We expect that IMPROVE will be non-inferior to CAU for the primary outcome. A non-inferiority design is justified because the IMPROVE-strategy is expected to maintain at least equal effectiveness for the primary outcome, while it is expected to substantially reduce revascularisations and costs. Conflict of interest