Abstract

Introduction: Defining the specific underlying pathophysiology of ischemic stroke is critical for minimizing the risk of recurrent events with personalized secondary prevention treatments. A notable portion of ischemic strokes are classified as embolic stroke of undetermined source (ESUS), leaving these patients without optimal treatment tailored to their pathophysiology. Hypothesis: Standard clinically collected data can reliably reclassify a substantial portion of ESUS patients into either a large artery atherosclerotic (LAA) or cardioembolic (CE) cause. Methods: A statistical model was developed to discriminate LAA from CE using a retrospective cohort of ischemic stroke patients treated at the University of Washington. A total of 189 patients were included (79 CE and 61 LAA to train the model, 49 ESUS patients to assess reclassification). Sixteen candidate predictors were collected across several sources: clinical risk factors, blood tests, echocardiography, ECG, and neurovascular imaging (Table 1). The LASSO (least absolute shrinkage and selection operator) was used to select important predictors in a penalized logistic regression model with stroke etiology (LAA vs. CE) as the outcome. ESUS patients were considered reclassified if the model-based probability of CE or LAA was at least 75%. Results: Of 189 patients, the mean (SD) age was 68 (14) years and 40% were women. The LASSO selected 12 predictors to discriminate CE vs. LAA (Table 1), with a corresponding cross-validated C-statistic = 0.87 (95% CI: 0.82-0.94). When the model was applied to the 49 ESUS patients, 23 (47%) were reclassified to LAA and 6 (12%) to CE. Conclusions: A multivariate model based on standard clinical data can separate LAA from CE with a high degree of discrimination. Applying this model led to reclassification to either LAA or CE in 59% of ESUS patients. Such approaches may enable more personalized secondary prevention strategies, but need to be tested in future trials.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call