Establishment And Validation Of Stroke Prediction Model In Patients With Nonvalvular Paroxysmal Atrial Fibrillation By Left Atrial Automatic Functional Imaging

BackgroundNonvalvular paroxysmal atrial fibrillation（NVPAF）is the exclusive heart valve cause of atrial fibrillation.Common valvular heart diseases that cause AF include moderate and severe mitral stenosis,mitral valvuloplasty,and mechanical or biological heart valve replacement.Clinically,NV heart disease is the most common cause of AF^[1].Cerebral ischemic stroke（CIS）is the most serious complication of AF.Compared to AF caused by valvular heart disease,CIS caused by NVPAF tends to have more severe brain function damage and higher mortality and recurrence rates^[2,3].Research has shown that left atrial（LA）remodeling in patients with AF is closely related to CIS^[4].Therefore,the evaluation of AF function has been an important research topic in recent years.Automatic LA function imaging（AFILA）is a new ultrasound technique for evaluating AF function that employs the speckle tracking principle in two-dimensional echocardiography.The operator can easily and quickly obtain parameters,such as AF strain,LA volume,and emptying fraction.It can also be used in combination with three planes during the onset of paroxysmal and persistent AF.Compared to previous new ultrasonic techniques,this method is simple,has good repeatability,is less affected by image quality,and exhibits higher accuracy.At present,there are many clinical models used to predict the CIS risk in patients with AF,and each model is characterized by different risk factors.The CHADS2,ATRIA,CHA2DS2-VASc,and Essen score risk factors are associated with basic clinical diseases.The ABC and GARFIELD-AF score risk factors are biochemical blood indicators.The CHA2DS2-VASc score is the most commonly used method to clinically assess the risk of thromboembolism in patients with AF.It is a valuable reference index for predicting stroke and guiding anticoagulation.When the CHA2DS2-VASc score is≥2,the risk of thromboembolism is considered high and oral anticoagulant drugs are required.A score of 1 identifies the risk of thromboembolism,which can be treated with or without anticoagulants.A score of 0 is considered to indicate a low risk of thrombosis and does not require anticoagulant therapy.However,the risk factors in the scoring system are not comprehensive and only include clinical chronic diseases,such as hypertension,diabetes,heart failure,age,gender,and other risk factors.At present,more and more studies have found that some biological indicators and LA function assessment play an important role in the prediction of stroke.The predictive value of these excluded test indicators and potential stroke risk factors in LA function for patients with AF combined with CIS has been a research focus.Among the NVPAF patients,a CHA2DS2-VASc score of<2 is greater than that in persistent AF patients,which does not reach the threshold of clinical anticoagulation treatment.However,these patients also have the risk of stroke.The purpose of the present study was to establish a model suitable for patients with moderate and low risk of thrombosis in NVPAF to develop CIS,predict the probability of stroke risk,and conduct an external validation in order to provide a reference basis for clinical anticoagulation treatment.as to provide a certain reference basis for clinical anticoagulation treatment.Part Ⅰ Evaluation of left atrial function in patients with nonvalvular paroxysmal atrial fibrillation by left atrial automatic functional imaging ultrasonographyObjective:To evaluate left atrial（LA）volume and strain in patients with non-valvular paroxysmal atrial fibrillation（NVPAF）using the new technology of left atrial automatic myocardial function imaging（AFILA）.Methods:A total of 80 NVPAF patients and 60 normal control patients were included in the study.The LA volume and strain parameters of the two groups were analyzed.The differences in LA function parameters were compared between the two groups to generate the receiver operating characteristic curve（ROC）and calculate the area under the curve（AUC）,sensitivity,and specificity of each parameter.Results:1.There was no statistically significant difference in age and sex between the two groups.Compared to the normal group,the LA minimum volume（LAVmin）,LA maximum volume（LAVmax）,and volume at onset of LA contraction（LAVpre A）in the NVPAF group were significantly increased.The LA emptying fraction（LAEF）was significantly decreased,and LA reservoir strain（S＿R）,LA conduit strain（S＿CD）,and LA contractile strain（S＿CT）were significantly compromised（P<0.05）.There was no significant difference in LA evacuation volume（LAEV）reduction（P>0.05）.2.Logistic regression analysis of LA function parameters in NVPAF patients showed that LAEF and S＿R were independently correlated with NVPAF[odds ratio（OR）values:0.883（0.827～0.943）,P<0.001;0.916（0.569～1.474）,P=0.047].3.The ROC curve results showed that LAEF had a high test efficiency in the diagnosis of NVPAF,with P<0.001,AUC of 0.843,sensitivity of 0.788,and specificity of 0.867.For the LA strain parameters,the S＿R test efficiency was higher,with P<0.001,AUC of 0.762,sensitivity of 0.713,and specificity of 0.783.Conclusions:Compared to the normal control group,the LA volume in NVPAF patients increased significantly,the LA emptying fraction decreased significantly,while the LA reservoir strain,conduit strain,and contractile strain decreased.Among the ultrasound AFILA left atrial function parameters,LAEF and S＿R indexes were independently correlated with paroxysmal atrial fibrillation,and had higher test efficacy.Part Ⅱ Left atrial automated functional myocardial imaging to identify patients with paroxysmal atrial fibrillation at high risk of strokeObjective:The purpose of this study was to investigate the risk factors of nonvalvular paroxysmal atrial fibrillation（NVPAF）with CIS using left atrial automatic imaging（AFILA）and to determine the diagnostic value of left atrial（LA）strain in patients with CHA2DS2-VASc score of<2.Methods:A total of 205 patients with NVPAF were included in the study and divided into the no CIS group without stroke（154 patients）and the CIS group（51 patients）.Clinical baseline data,blood biochemical results,and AFILA ultrasound data were collected for all patients.Univariate analysis of the above data to determine the differences between the two groups of patients was carried out.Independent risk factors were obtained using multivariate logistic regression analysis,and the ROC curve was generated to delineate the S＿CT truncation value.Results:1.Univariate analysis showed that there were significant differences in age,CHA2DS2-VASc score,anticoagulant drug use,history of hypertension,diabetes,and coronary heart disease,as well as LAEF,S＿R,S＿CT,WBC,NEUT,HCY,urea,NDD,NT-proBNP,fibrinogen（Fib）,cardiac troponin I（cTnI）,and NLR levels between the two groups（all P<0.05）.2.Multifactor analysis results showed that age,hypertension,and levels of S＿CT,urea,NLR,Fib,and cTnI were independent risk factors associated with CIS in patients with NVPAF.After adjusting the CHA2DS2-VASc score for age,sex,and history of hypertension,S＿CT significantly led to NVPAF complicated with stroke[odds ratio（OR）value 1.259（1.095～1.447,P=0.001）].3.There were statistically significant differences in the CHA2DS2-VASc scores between the NVPAF and normal control groups.In particular,the difference in the group with a CHA2DS2-VASc score of <2 was more significant than that in the group with a score of ≥2（P=0.009）.S＿CT of-4.5 was the cut-off value for CIS in NVPAF patients（sensitivity=0.80,specificity=0.75,AUC=0.866,and P<0.0001）.Conclusions:1.Age,hypertension,as well as levels of S＿CT,urea,NLR,fibrinogen,and cTnI were independently associated risk factors for patients with combined CIS.2.LA strain measured using AFILA is a better indicator for dynamic assessment of LA function impairment in patients with paroxysmal AF combined with ischemic stroke,especially in groups with CHA2DS2-VASc score of<2,which is a more sensitive metric.S＿CT>-4.5 is considered to have a therapeutic value in clinical anticoagulation.Part Ⅲ Establishment and validation of stroke prediction model based on ultrasound AFILA parametersObjective:To generate a nomogram prediction model based on the independent risk factors for CIS in NVPAF patients and to validate it in an external validation group.Methods:A total of 205 NVPAF patients were included in the modeling group.A new nomogram prediction model was established based on the modeling group regression model.The ROC,calibration,and clinical decision curves were used to evaluate the differentiation,calibration,and clinical practicability of the nomogram model.In addition,80 NVPAF patients treated at the Department of Cardiology of Nanchang First Hospital between April 2022 and October 2022 were selected as the external validation group.The baseline data for the two groups were compared and the nomogram was externally validated.The area under the ROC curve was compared between the new model and the CHA2DS2-VASc scoring system,and the diagnostic effect of the model was tested in groups with a CHA2DS2-VASc score of<2 and≥2.Results:1.There was no statistical difference between the modeling and external verification groups in terms of 39 independent variables,including clinical routine parameters,biochemical indicators,and ultrasonic indicators,as well as one outcome variable（with or without stroke）（P>0.05）.2.A Nomogram model was established based on the risk factors screened using logistic regression.3.ROC curve analysis:Modeling group C-index=0.929,95%confidence interval（CI）:0.893–0.966,sensitivity=0.882,specificity=0.851,positive predictive value=0.662,negative predictive value=0.956,accuracy=0.859,and P<0.001.Verification group C-index=0.887,95%CI（0.798～0.976）,sensitivity=0.833,specificity=0.875,positive predictive value=0.741,negative predictive value=0.925,accuracy=0.863,and P<0.001.The model had good differentiation.H-L test results showed that X2=3.095,P=0.928 in the modeling group and X2=7.131,P=0.523 in the external verification group.The model calibration ability was good.Clinical decision curve analysis demonstrated that this model has a clinical value.4.The AUC for CHA2DS2-VASc score=0.717,95%CI:0.634～0.8,sensitivity=0.755,specificity=0.713,and P<0.001.The diagnostic efficacy of the nomogram model（AUC:0.929,95%CI:0.893～0.966,sensitivity=0.882,specificity=0.851,and P<0.001）was superior to that of the CHA2DS2-VASc scoring model.The sensitivity and specificity were high.5.The AUC values for the CHA2DS2-VASc score of≥2 and<2 were 0.856 and 0.963 with 95%CIs:0.772～0.941 and 0.896～1.00（P<0.01）,respectively.The sensitivity values were 0.833 and 0.857,respectively.The specificity values were0.738 and 0.878（P<0.01）,respectively.The model had a better diagnostic effect and higher sensitivity and specificity in patients with CHA2DS2-VASc scores of<2 than in those with≥2.Conclusions:1.The stroke nomogram model based on the LA function parameters obtained using ultrasound AFILA shows good discrimination,accuracy,and clinical practical value.2.The new nomogram model has better diagnostic efficiency and higher sensitivity and specificity than the CHA2DS2-VASc scoring model.3.Compared to the patients with a CHA2DS2-VASc score of≥2,the nomogram model has higher sensitivity,specificity,and accuracy in predicting the risk of CIS in patients with paroxysmal AF with a score of<2.