Predicting ischemic stroke after carotid artery stenting based on proximal calcification and the jellyfish sign

OBJECTIVE Carotid artery stenting (CAS) has been considered to prevent ischemic strokes caused by stenosis of the cervical carotid artery. The most common complication of CAS is new cerebral infarction. The authors have previously reported that the jellyfish sign-the rise and fall of the mobile component of the carotid plaque surface detected by carotid ultrasonography-suggests thinning and rupture of the fibrous cap over the unstable plaque content, such as the lipid-rich necrotic core or internal plaque hemorrhage. The authors' aim in the present study was to evaluate the risk of a new ischemic lesion after CAS by using many risk factors including calcification (size and location) and the jellyfish sign. METHODS Eighty-six lesions (77 patients) were treated with CAS. The presence of ischemic stroke was determined using diffusion-weighted imaging (DWI). Risk factors included calcification of the plaque (classified into 5 groups for size and 3 groups for location) and the jellyfish sign, among others. Multiple linear regression analysis (stepwise analysis and partial least squares [PLS] analysis) was conducted, followed by a machine learning analysis using an artificial neural network (ANN) based on the log-linearized gaussian mixture network (LLGMN). The additive effects of the jellyfish sign and calcification on ischemic stroke after CAS were examined using the Kruskal-Wallis test, followed by the Steel-Dwass test. RESULTS The stepwise analysis selected the jellyfish sign, proximal calcification (proximal Ca), low-density lipoprotein (LDL) cholesterol, and patient age for the prediction model to predict new DWI lesions. The PLS analysis revealed the same top 3 variables (jellyfish sign, proximal Ca, and LDL cholesterol) according to the variable importance in projection scores. The ANN was then used, showing that these 3 variables remained. The accuracy of the ANN improved; areas under the receiver operating characteristic curves of the stepwise analysis, the PLS analysis, and the ANN were 0.719, 0.727, and 0.768, respectively. The combination of the jellyfish sign and proximal Ca indicates a significantly increased risk for ischemic stroke after CAS. CONCLUSIONS The jellyfish sign, proximal Ca, and LDL cholesterol were considered to be important predictors for new DWI lesions after CAS. These 3 factors can be easily determined during a standard clinical visit. Thus, these 3 variables-especially the jellyfish sign and proximal Ca-may be useful for reducing the ischemic stroke risk in patients with stenosis of the cervical carotid artery.