316 Acute Ischemic Stroke Presentation Labs and Imaging Can Be Used to Model Thrombus Proteomic Composition

INTRODUCTION: The literature points to a strong relationship between the composition of acute ischemic stroke thrombi and their susceptibility to thrombolytic or endovascular treatment. Composition can be determined by molecular studies of the clot itself, but there is tremendous utility in being able to predict clot composition solely on data available at patient presentation. METHODS: Quantitative proteomes were generated using mass spectrometry of thromboembolic material retrieved by thrombectomy from 59 stroke patients. Of the 2,790 total identified proteins, marker proteins were selected to identify broad groups of thrombus components: RBCs (12 proteins), platelets (9 proteins), histones (23 proteins) and neutrophils (13 proteins). Dimensional reduction of marker proteins was done by principal component (PC) analysis for each group. A broad survey of demographic, lab, and radiographic data was gathered by retrospective chart review. Regression analysis was used to identify the combination of the four variables on chart review that most strongly associated with the PCs for each thrombus component group. RESULTS: After dimensional reduction, a significant amount of the variance in marker protein abundance for each group was represented by the first PC for each group (R2 > 0.94 for all). It was possible to generate linear models representing the PC associated with RBCs (R2 = 0.37, p = 0.02), platelets (R2 = 0.40, p = 0.0004), histones (R2 = 0.52 and p = 0.001), and neutrophils (R2 = 0.81, p = 1E-12) using combinations of four data points at presentation. CONCLUSIONS: This preliminary work suggests that the abundance of clot components may be predicted by modeling patient data available at presentation, which has the potential to drive future choice of therapy. Next steps include refining and validating the models on a prospectively maintained stroke database at our home institution.