Cardiac CT Angiography In Etiologic Investigation Of Acute Ischemic Stroke:A Preliminary Study

BackgroundCardioembolic stroke is generally the most severe ischemic stroke subtype with the worst clinical outcomes.Atrial fibrillation is the most common high risk cardioembolic conditions.Left atrial appendage is the most frequent site of intracardiac thrombus formation in atrial fibrillation patients.Occult paroxysmal atrial fibrillation is thought to be one of the major etiologies for embolic stroke of undetermined source（ESUS）and can cause changes in the function and structure of the left atrial appendage.Cardiac CT Angiography（CTA）can well delineate the morphology and structure of the left appendage and detect intracardiac thrombi.AimThis study examines cardiac CTA images of various stroke subtypes to identify characteristics indicative of the causes of ischemic strokes,as well as clinical and image features associates with these characteristics,and evaluates the use of cardiac CTA to infer etiologies of ESUS patients.MethodsCardiac CTA images of consecutive acute ischemic stroke patients treated with reperfusion therapy during October 2012 and May 2016 were retrospectively collected.ESUS was defined by current criteria.Filling defects,enlargement of left atrial,extent of trabeculations（mild,moderate,and extensive）,number of lobes and morphology of left atrial appendage（cactus,chicken-wing,windsock,and cauliflower）were measured on raw or reconstructed images.These image biomarkers,along with additional clinical characteristics were compared among two major causative subtypes（large artery atherosclerosis and cardioembolism）according to the international Trial of Org 10172 in Acute Stroke Treatment（TOAST）criteria,to identify specific image biomarkers that might be helpful to distinguish causes of strokes.Then clinical characteristics and cardiac CTA measures were compared to determine variables associated with specific image biomarkers.ResultsAmong 216 acute ischemic stroke patients（117 males;median age 71[60～79]years）with cardiac CTA images,according to TOAST classification,142（65.74%）were classified as cardioembolism,26（12.04%）as large artery atherosclerosis,2（0.93%）as small artery occlusion,0（0.0%）as other causes,12（5.56%）as multiple potential causes,3（1.38%）as undermined due to cursory evaluation,and another 31（14.35%）cases met criteria for ESUS.Filling defects in the left heart were observed in 84 cases（38.7%）,76（accounting for 53.52%of the total cardioembolic stroke group）of which were cardioembolic strokes.Among them,14 had thrombus in the left heart（3 in the apex of left ventricular,1 in left atrium,10 in left atrial appendage）,1 had left atrial myxoma;69 had left atrial stasis.And there were 9 cases had both stasis and thrombus in the left atrial appendage.A total of 116 cases（53.7%）had enlargement in the left atrial.In patients with nonvalvular atrial fibrillation,univariate analysis revealed smoking（36.8%vs 15.8%,P=0.009）,anti-platelet medication use（35.3%vs 19.3%,P=0.071）,left atrial enlargement（85.30%vs 54.40%,P=0.000）,cauliflower morphology（37.5%vs 22.2%,P=0.032）and mild trabeculations（52.3%vs 22.8%.P=0.001）to be associated with a filling defect in the left atrium.In multivariate analysis,left atrial enlargement（OR=3.442,95%CI[1.267～9.351],P=0.015）and mild trabeculations（OR=5.153 compared with extensive trabeculations,95%CI[1.495～17.760],P＝0.019）remained independently associated with a filling defect in the left atrium.Receiver operating characteristic analysis demonstrated a moderate discriminatory capacity of left atrial enlargement（AUC 0.655,95%CI[0.556～0.753],P=0.003）and mild trabeculations（AUC 0.636,95%CI[0.538-0.733],P=0.009）in predicting left atrium filling defect in nonvalvular atrial fibrillation patients.The distribution of LAA morphology included Cactus 67（33.0%）,Chicken Wing 35（17.2%）,Windsock 41（20.2%）,and Cauliflower 60（29.6%）,and the distribution of number of lobes was 76（38.8%）with 1 lobe,39（19.9%）with 2 lobes,48（24.5%）with 3 lobes and 33（16.8%）with 4 or more.There were no significant differences in the distributions of left atrial appendage morphology（P=0.867）and number of lobes（P=0.988）among cardioembolism,large artery atherosclerosis,and ESUS group.The ESUS group were younger（P=0.002）,had lower CHADS₂（P=0.042）and CHA₂DS₂-VASc（P=0.003）scores,higher Barthel index（P=0.037）at 3 months after stroke and more extensive trabeculations（P=0.05）than the cardioembolism group,while their clinical and cardiac CTA features were quite similar to the large artery atherosclerosis group.ConclusionsCardioembolic stroke patients are more commonly have a filling defect in the left heart.In patients with nonvalvular atrial fibrillation,a filling defect in the left atrium independently associates with left atrial enlargement and mild trabeculations,which might be the result of left atrial appendage remodeling.The clinical features and cardiac images of ESUS patients are quite close to those of large artery atherosclerosis,and the diagnostic yields of cardioembolic sources by cardiac CTA are low in ESUS patients.