Predictive Value Of The Comprehensive Stroke Recurrence Model On Stroke Risk In Patients With Transient Symptoms With Infarction

Background and Purpose:Transient ischemic attack(TIA)with a positive diffusion-weighted image(DWI)had a very high risk of stroke,which was defined as transient symptoms with infarction(TSI).In clinical practice,it was extremely important to identify TSI patients with high risks of stroke and take effective intervention to reduce the incidence of stroke.Many studies demonstrated that some detail neuroimaging features in TSI patients were associated with increasing risks of stroke occurrence.Recently,Nam et al reported that a risk score based on imaging features,which called the comprehensive stroke recurrence model(CSR),had a high ability in predicting the risk of stroke recurrence among patients with first-ever acute ischemic stroke.However,the predictive value of this model have been not validated in TSI patients.The purpose of this study was to evaluate the ability of CSR model to predict the short-term and long-term risk of stroke in Chinese patients with TSI and compare with the traditional models based on clinical risk factors.Methods:This study was continuously included eligible TIA patients who were enrolled by the First Affiliation Hospital of Zhengzhou University from October 2010 to August 2017.All patients without contradictions of magnetic resonance imaging(MRI)were recommended to undergo work-ups.The baseline data of eligible TSI patients were collected and Essen stroke risk score(ESRS),ABCD2 score,ABCD3I score and CSR score were calculated by uniformly trained neurologists,.Follow-up was conducted at 7d,90d and 1 year after TSI occurrence,secondary preventive medication and prognosis information were recorded as well.The endpoint event was ischemic stroke occurrence at 7d,90d and 1 year.The incidence of stroke was analyzed in TSI patients with different CSR scores.Univariate logistic regression analysis was conducted to determine the risk factors associated with stroke occurrence.If there were statistical differences by univariate analysis at baseline,confounding factors were selected to examine the relationship between the risk predictive scores and stroke occurrence using multivariate logistic regression.Associations were presented as odds ratio(OR)with corresponding 95%confidence interval(CI).Draw the receiver operating characteristic(ROC)curve of ESRS,ABCD2 score,ABCD3I score and CSR score,with the area under the curve(AUC)as a measure of predictive ability.The AUCs for the different scores were calculated by Delong method and were compared with the Z test and P value.Results:A total of 450 eligible TSI patients were included,with an average age of 57.7±12.0 years,156(34.7%)were females and 294(65.3%)were males.Ischemic stroke occurred in 48(10.7%),84(18.7%),and 104(23.1%)patients at 7-day,90-day and 1-year follow-up,respectively.A higher CSR score was associated with the increasing risk of stroke occurrence.The risks of stroke with different CSR scores at 7-day follow-up were 3.9%,6.1%,6.9%,3.6%,10.0%,15.3%,22.2%,29.6%and 22.2%;the risks of stroke with different CSR scores at 90-day follow-up were 3.9%,11.0%,9.7%,12.7%,14.0%,25.4%,44.4%,40.7%and 66.7%;and the risks of stroke with different CSR scores at 1-year follow-up were 7.8%,15.9%,15.3%,18.2%,16.0%,27.1%,46.7%,51.9%and 77.8%,respectively.After adjusting for potential covariates,multivariate logistic regression analysis showed that the CSR score≥5 was the risk factor of stroke occurrence in TSI patients(OR 3.555,95%CI 1.885-6.705,P<0.001 at 7 days;OR 4.901,95%CI 2.914-8.244,P<0.001 at 90 days and OR 4.006,95%CI 2.468-6.503,P<0.001 at 1 year,respectively).The predictive value of CSR(AUC=0.730,95%CI 0.638-0.752)for 90-day risk of stroke after TSI was higher than that of ESRS(AUC=0.583,95%CI 0.532-0.626,P=0.001),ABCD2 score(AUC=0.602,95%CI 0.531-0.625,P=0.004)and ABCD3I score(AUC=0.623,95%CI 0.557-0.688,P=0.017).The predictive value of CSR(AUC=0.693,95%CI 0.635-0.724)for 1-year risk of stroke after TSI was higher than that of ESRS(AUC=0.581,95%CI 0.532-0.626,P=0.004)and ABCD2 score(AUC=0.592,95%CI 0.531-0.625,P=0.012).CSR score had a higher predictive value than ABCD3I score(AUC=0.632,95%CI 0.572-0.692,P=0.118)at 1-year follow-up,but the difference was not statistically significant.Conclusions:Patients with TSI had a high risk of stroke.CSR model had a higher ability for predicting the risk of stroke after TSI in comparison with clinical scores.In clinical practice,CSR model could further identify high-risk TSI patients.