Validation of a simple clinical tool for screening of acute lacunar stroke—A substudy of the WAKE-UP trial

Abstract

Introduction: Lacunar stroke represents around a quarter of all ischemic strokes; however, their identification with computed tomography in the hyperacute setting is challenging. We aimed to validate a clinical score to identify lacunar stroke in the acute setting, independently, with data from the WAKE-UP trial using magnetic resonance imaging. Methods: We analyzed data from the WAKE-UP trial and extracted Oxfordshire Community Stroke Project (OCSP) classification. Lacunar score was defined by National Institutes of Health Stroke Scale (NIHSS) < 7 and OCSP lacunar syndrome. Assessment of lacunar infarct by two independent investigators was blinded to clinical data. We calculated sensitivity, specificity, negative and positive predictive value (NPV and PPV, respectively) of lacunar score. Results: We included 503 patients in the analysis, mean (±SD) age 65.2 (±11.6) years, 325 (65%) males, median (IQR) NIHSS = 6 (4–9); 108 (22%) lacunar infarcts were identified on magnetic resonance (MR), patients fulfilling lacunar score criteria were 120 (24%), of which 47 (44%) had a lacunar infarct. Lacunar score was negative in 322 (82%) of patients without lacunar infarct. Patients with lacunar score had lower NIHSS (4 vs 7, p < 0.001), higher systolic (157 vs 151 mmHg, p = 0.001) and diastolic (86 vs 83 mmHg, p = 0.013) blood pressure and smaller infarct volume (2.4 vs 9.5 mL, p < 0.001). Performance of lacunar score was as follows: sensitivity 0.44; specificity 0.82; PPV 0.39; NPV 0.84; and accuracy 0.73. Assuming a prevalence of lacunar stroke of 13%, PPV lowered to 0.30 but NPV was 0.90. Lacunar score performed better for supratentorial lacunar infarcts. Conclusion: Lacunar score had a very good specificity and NPV for screening of lacunar stroke. Implementation of this simple tool into clinical practice may help hyperacute management and guide patient selection in clinical trials. Data access statement: Data supporting the results of this paper are available upon reasonable request to the corresponding author.