Perfusion Imaging-Based Triage for Acute Ischemic Stroke: Trends in Use and Impact on Clinical Outcomes.

Background: Perfusion imaging (PI) serves as a valuable tool for triaging patients with acute ischemic stroke for endovascular treatment (EVT). This study aims to investigate trends in PI use and its impacts on EVT rates and clinical outcomes, particularly focusing on variations across different time windows. Methods: Data from a prospective, nationwide, acute stroke registry in South Korea were analyzed retrospectively. PI was regarded as treatment-decision imaging when conducted either (1) prior to EVT, or (2) within 3 hours from hospital arrival in patients not receiving EVT. The study spanned 3 epochs: 2011-2014, 2015-2017, and 2018-2021. Based on the time from onset to arrival, patients were categorized into 2 time windows: early (0-6 hours) and late (6-24 hours). We evaluated EVT rates and clinical outcomes in patients with anterior large vessel occlusion. Results: From 2011 to 2021 among 49 449 patients with acute ischemic stroke presenting within 24 hours of onset, PI use rates declined from 36.9% to 30.1%. In the early window, rates dropped from 48.4% to 32.4%, whereas in the late window, they increased from 23.5% to 27.8%. Factors such as older age, atrial fibrillation, anterior large vessel occlusion, and severe stroke were associated with higher rates in the late window. Conversely, younger age and male sex were associated with higher rates in the early window. For patients with anterior large vessel occlusion, PI use increased the likelihood of receiving EVT in the late window and minimized the risk of symptomatic intracranial hemorrhage in the early window. However, 3-month functional outcomes and mortality were. unaffected. Conclusion: The study revealed distinct trends in PI use across early and late time windows, indicating varying roles of PI in these time frames. However, the definitive value and necessity of PI in guiding EVT decision-making remain unclear, underscoring the need for further research.