Correlation Between Cerebral Hemodynamic Changes And Prognosis In Patients With Intracranial Atherosclerotic Stenosis

Part I: Correlation between plaque and perfusion features in middle cerebral artery stenosis Objective:To explore the potential relationship between plaque characteristics and downstream blood perfusion in patients with severe intracranial middle cerebral artery stenosis.Methods:Patients with severe unilateral middle cerebral artery stenosis treated in the first medical center of Chinese PLA general hospital from 2013 to 2016 were enrolled retrospectively.The average cerebral blood flow(CBF)of the middle cerebral artery(MCA)territory was measured by using three dimensional pseudo continuous arterial spin labeling(3D p CASL)with a double post-labeling delay(PLD)of 1.5 s and 2.5 s.Finally,the early-arriving blood flow proportion and late-arriving retrograde blood flow proportion(corresponding to antegrade and collateral blood flow)were obtained by mathematical calculation.Meanwhile,high-resolution vessel wall imaging(HR VWI)was performed to obtain several plaque characteristics,such as area,burden,degree of enhancement,and remodeling index.The correlation between plaque and perfusion characteristics was analyzed.In addition,according to the plaque burden,the patients were divided into high(≥ 50%)and low(<50%)burden group,inter-group comparison and regression analysis were carried out.Results:A total of 58 patients were enrolled in our study(47.17±11.67 years,39 males).Plaque burden was negatively correlated with early-arriving blood flow proportion(r=-0.362,p=0.005),but no significant correlation with late-arriving retrograde blood flow proportion(r=-0.008,p=0.951).Compared with low burden group,the high burden group was older(50.38±11.33 vs 43.23±11.04,p=0.019)and had a lower early-arriving blood flow proportion(72.13%±14.86% vs 83.00%±15.67%,p=0.009);multivariate regression analysis showed that the age(OR 1.056,95%CI 1.000-1.115,p=0.050)and early-arriving blood flow proportion(per 1 quartile: OR 0.561,95%CI0.328-0.957,p=0.034)were independently related to the high burden of plaque.Conclusions:Plaque burden is related to downstream antegrade blood flow,but not to the collateral blood flow.The higher the plaque burden is,the less the antegrade blood flow is.Part II: MR perfusion-weighted imaging predict the prognosis of non-acute ischemic strokeObjective:We aimed to propose a new imaging biomarker,hypoperfusion cubage ratio(HCR),and to evaluate its effect on the prognosis of non-acute stroke.Methods:Patients with severe stenosis and occlusion of the unilateral internal carotid artery(ICA)or middle cerebral artery(MCA)who underwent magnetic resonance(MR)perfusion-weighted imaging(PWI)from 2015 to 2019 were enrolled from the first medical center of Chinese PLA general hospital.PWI lesion volumes were determined with the Neu Brain CARE software program.HCR was defined as the proportion of time to maximum of the residual function(Tmax)>6s lesion volume to Tmax >4s.According to the modified Rankin Scale(m RS)at follow-up,the patients were divided into two groups: good prognosis(m RS≤2)(event-)and poor prognosis(m RS>2)(event+),and their clinical and imaging characteristics were compared.In addition,through the method of machine learning(ML),after feature selection,it was divided into Model A(excluding HCR)and Model B(including HCR)according to whether the variable HCR was included or not.Then they were integrated with the most commonly used Naive Bayes algorithm and the diagnostic performance for poor prognosis was compared.The area under the receiver operating characteristic curve(AUC)was the most important evaluation criterion.Results:Of the 181 patients(median age,60 years;148 men)evaluated,33 experienced a poor prognosis.HCR was higher in the event+ than event-groups [32(13-49)vs.0(0-8),p<0.001],and multivariable regression analysis revealed that HCR was associated with prognosis(odds ratio =1.083,95% confidence interval [CI] 1.03-1.14,p<0.001)after adjusted by age,hypertension,Tmax>4s lesion volume and Tmax>6s lesion volume.Meanwhile,HCR was correlated with poor collateral circulation(r=-0.476,p<0.001).Besides,ten variables after feature selection were included in the ML model,and the information gain ranking(IGR)showed that HCR had the greatest impact on poor prognosis.After 10-fold cross-validation policy,the AUC of Model B was significantly higher than that of Model A(0.79 VS 0.56),and the results of the other four evaluation indexes(accuracy,precision,recall rate and F1 score)were also higher than those of model A.Conclusions:HCR is a comprehensive index reflecting hemodynamic changes and collateral circulation,which can predict the prognosis of non-acute ischemic stroke.It is easier to identify high-risk patients with poor prognosis of stroke by combining imaging index.