Impact Of Internal Carotid Artery Calcification In Stroke Patients With Intravenous Thrombolysis And The Establishment Of Prognostic Model

Part I Internal Carotid Artery Calcification: Prevalence and Determinants in Acute Ischemic Stroke Patients with Intravenous ThrombolysisBackground and aims: Internal carotid artery calcification(ICAC)is a common imaging finding on unenhanced brain CT,especially in the elderly.Previous studies have reported that ICAC was associated with the occurrence,prognosis and complications of acute ischemic stroke(AIS).Therefore,it is of great significance to clarify the prevalence and determinants of ICAC for the management and prognostic prediction of AIS patients.However,most studies on this topic are from western cohorts and studies based on patients or populations in China are limited.The aim of this study was to analyze the prevalence and to investigate the determinants of ICAC based on AIS patients in our center.Methods: We retrospectively analyzed a prospectively maintained database of our hospital from July 2015 to January 2021.AIS patients received intravenous thrombolysis(IVT)were continuously enrolled.All patients underwent unenhanced CT scan of the brain.The third-party software 3Dslicer was used to assess the presence and the volume of ICAC on CT image.Regions with more than 2 consecutive voxels above 130 Hu was defined as calcification.We first described the distribution and prevalence of ICAC by sex and age groups.Then,multiple logistic regression analysis was used to explore the determinants of ICAC.A significance level of p < 0.05 was applied.Results: We included 1022 patients with a median age of 63(interquartile range 55-71),of whom 747(73.1%)patients were male.The prevalence of ICAC was 82.1% and the median volume of calcification was 86.55mm3(interquartile range 12.25 mm3-286.65 mm3).The prevalence and volume of ICAC were increased with age,but not much different between both sides and both genders.Multivariate logistic regression analysis showed that age(OR 3.019,95% CI 2.506-3.638,p< 0.001,per 10 years),diabetes mellitus(OR 2.213,95% CI 1.237-3.957,p= 0.007)and previous history of stroke(OR 2.134,95% CI 1.127-4.043,p= 0.020)were independently associated with the presence of ICAC.Conclusion: ICAC is highly prevalence in patients with IVT and tends to increase with age.Aging,diabetes mellitus and prior stroke history are important determinants of ICAC.Part II Internal Carotid Artery Calcification and the Clinical Outcome in Acute Ischemic Stroke Patients with Intravenous ThrombolysisBackground and aims: IVT with recombinant tissue plasminogen activator is the firstline treatment for AIS to achieve ultra-early cerebral ischemia reperfusion.Timely restoration of cerebral blood perfusion in AIS patients can effectively reduce the final infarction,improve the neurological defects,and ultimately achieve the purpose of reducing the mortality and disability rate of stroke.Nevertheless,IVT significantly increases the risk of hemorrhagic transformation,and the prognosis of some patients remains poor despite timely treatment.ICAC can lead to vascular wall stiffness and cerebral hemodynamic instability,which in turn will affect downstream microcirculation and neurons.Thus,ICAC may have an impact on the clinical outcome in AIS patients with IVT.However,studies focused on this topic are limited,and only qualitative methods were used.This part of the study was aimed to investigate the association between ICAC,assessed by quantitative method,and the clinical outcome of AIS patients in anterior circulation who received IVT in our center.Methods: We retrospectively analyzed the prospectively and continuously enrolled AIS patients in the anterior circulation who received IVT with recombinant tissue plasminogen activator in our hospital from July 2015 to January 2021.All patients undergone non-enhanced brain CT at baseline and 24 h after IVT.Baseline CT was used to assess ICAC,including the presence of ICAC,calcification volume and calcification score.Follow-up CT at 24 h after IVT was used to define hemorrhagic transformation according to ECASS-II classification criteria.Functional outcome was assessed at 3 months after IVT on the modified Rankin Scale(m RS),and poor prognosis was defined as an m RS score of 3 or higher.Multivariate logistic regression analysis was used to analyze the association between ICAC and clinical outcome after IVT.Due to the skewed distribution of calcification volume,and patients with a calcification volume of zero,we transformed the calcification volume with the natural logarithm after adding 1.00mm3 to it [ln(calcification volume +1.0mm3)].Meanwhile,we also grouped the calcification volume by quartiles to further confirm the findings.The same method was also applied for the analysis of calcification score.P < 0.05 was defined as statistically significant.Results: Amongst 600 patients [median age,63 years(interquartile range 54-71 years);70.7% male],214(35.7%)patients showed poor prognosis,and 53(8.8%)patients showed hemorrhagic transformation.Meanwhile,489(79.7%)patients were present with ICAC,with a median calcification volume of 70.10mm3(interquartile range 5.38 mm3-274.66 mm3)and a median calcification score of 159.77(interquartile range 6.35-671.31).The univariate analysis showed that higher percentage of ICAC(p<0.001),larger calcification volume(p=0.005)and a higher calcification score(p=0.008)were significantly associated with poor outcome at 3 months.Higher calcification volume(p=0.037)was significantly associated with hemorrhagic transformation,while calcification score only showed marginal significance(p=0.052).In sensitivity analysis,after adjusting confounding factors,the presence of ICAC was independently associated with poor prognosis at 3 months(OR 1.866,95% CI: 1.073-3.244,p = 0.027).However,neither calcification volume nor calcification score was independently associated with hemorrhagic transformation.Conclusion: In AIS patients with anterior circulation who received IVT,ICAC is independently associated with poor prognosis and calcification volume of internal carotid artery is correlated with hemorrhage transformation,however,it is not an independent predictor for it.Part III Use of Machine Learning Algorithm to Predict Clinical Outcome in Acute Ischemic Stroke Patients with Intravenous Thrombolysis on the Basis of Baseline Information Before Intravenous ThrombolysisBackground and aims: The prediction of functional outcome in AIS patients is useful for individualized management and therapeutic stratification.Meanwhile,it can also be used to guide patients' expectations for treatment and rehabilitation,set reasonable goals,and improve patient-physician communication.In recent years,machine learning has attracted more and more attention in the field of healthcare due to its powerful processing ability of complex data.Compared with traditional statistical models,machine learning models show better predictive efficiency.In this part of the study,we constructed prediction models based on the data available before thrombolytic therapy using machine learning algorithm to predict the poor prognosis of patients at 3 months.Meanwhile,we analyzed the feature importance in the prediction model,and intended to further verify the predictive effect of ICAC on the prognosis of stroke patients through machine learning approach.Methods: This study retrospectively analyzed the prospectively enrolled AIS patients in anterior circulation who received IVT with alteplase in our hospital from July 2015 to January 2021.All biomarkers available before IVT were collected.The presence of ICAC was assessed on unenhanced CT of the brain.To simplify the model,features were selected based on previous published literatures and clinical experience.The data were cleaned,and then normalized by min-max normalization method.Poor prognosis was defined as m RS score of 3 or higher at 3 months after IVT.The data were randomly divided into training set and test set in a ratio of 8:2.Logistic regression,random forest and extreme gradient boosting were used to developed the predict models.The effectiveness of the model was evaluated by receiver operating characteristic curve(ROC)and confusion matrix.Then the analysis was stratified by age groups as follow: <70 years and ? 70 years.To avoid a preference for consecutive variants,shapley additive explanations(SHAP)was used to unlock the importance of ICAC in the predicting model.Results: A total of 600 patients were enrolled in the study,of whom 214(35.7%)showed poor outcomes.A total of 39 features were extracted from the baseline data and 10 features were further selected.When all features were used for modeling,the areas under the curve(AUC)of logistic regression,random forest and extreme gradient boosting models were 0.738,0.769 and 0.756,respectively;the accuracy were 0.700,0.716 and 0.675,respectively.When the selected features were used to build the models,the AUC were 0.745,0.784 and 0.767,respectively;the accuracy was 0.733,0.750 and 0.725,respectively.When the random forest model was built with all features,the AUC of ?70 years old and < 70 years old were 0.824 and 0.717,respectively.When modeling with selected features,the AUC of ?70 years old and < 70 years old was 0.891 and 0.728,respectively.The feature importance of ICAC ranked fifth both in all the 39 features and the selected 10 features,which showed a positive effect in the prediction of poor functional outcomes.Conclusion: The random forest model is more effective than other algorithms in predicting the poor prognosis 3 months after IVT in acute anterior circulation ischemic stroke patients.And it shows a better performance in patients over 70 years old.What is more,ICAC is an important predictor for the prognosis of AIS patients with IVT.