Abstract TMP37: Gene Expression May Predict Symptom Onset Time In Acute Ischemic Stroke

Abstract

Introduction: Imprecise timing of symptom onset (SO) in patients with acute ischemic stroke (AIS) may impact eligibility for reperfusion therapy. Unreliable “last-known-well” (LKW) estimates may reduce the use of thrombolytic or endovascular treatments. A blood biomarker that serves as a “stroke clock” could minimize uncertainty and augment treatment rates. Methods: The single-center VuEssence Stroke Trial (VEST) evaluated gene expression biomarkers in adult ED patients with suspected AIS, a LKW time < 24hrs, and timing of SO accurate within a 0-3 hour time frame. At ED presentation and before reperfusion treatment, whole blood was collected in PAXgene RNA tubes for RNA-sequencing and gene expression analysis. Researchers blinded to molecular results adjudicated final diagnosis (AIS vs stroke mimic) and estimated time of SO based on times of witnessed onset (WO), LKW, and symptom discovery (SD). The confirmed AIS subgroup with WO timing, or ≤60 minutes between LKW and SD, was analyzed. Results: Of AIS patients, 56 met criteria (52% men, mean age 71.6). The average time from SO was 148 minutes (range 48-490). After RNA-sequencing, linear regression showed 684 genes with a significant linear time-related expression change in reference to time from SO (<0.2 Benjamini-Hochberg p-adjusted). A subgroup of 210 genes with <0.004 p-values and linear coefficients >0.25 were used to generate multiple classifiers of up to 12 genes that identified AIS <2.5 hrs from SO. Each gene is used in at least one classifier with other genes selected to maximize a measure of accuracy. The top panels showed both high sensitivity and specificity ( table ). Conclusion: Using a biological “stroke clock” biomarker instead of SO timing may streamline evaluations of suspected AIS, augment reperfusion therapy rates, and improve outcomes. VEST identified time-related gene expression changes that may accurately classify AIS <2.5 hrs from SO. These gene panels will require further validation.