Validation And Construction Of Predictive Score And Model For Left Atrial Thrombus Or Spontaneous Echo Contrast In Patients With Non-valvular Atrial Fibrillation

Part One Risk factors for left atrial thrombus and spontaneous echo contrast in patients with non-valvular atrial fibrillationObjective:The presence of left atrial thrombosis（LAT）or spontaneous echo contrast（SEC）development in patients with non-valvular atrial fibrillation（NVAF）raises the risk of stroke and systemic embolism,which contributes to increased morbidity and mortality.This study aimed to identify the risk factors for LAT/SEC in patients with NVAF.Methods:A total of 665 patients with NVAF who underwent transesophageal echocardiograpy from Cangzhou Central Hospital from January 2016 to August 2023 were retrospectively analyzed.Among them,399 patients with NVAF from January 2016 to August 2021 comprised the training cohort,while the remaining 266 patients with NVAF from September2021 to August 2023 constituted the test cohort.Risk factors for LAT/SEC were identified using univariate and multivariate logistic regression analysis.Results:Among the 665 patients with NVAF,70 patients（10.53%）had LAT/SEC.Patients with LAT/SEC,when compared with those without LAT/SEC,exhibited a higher prevalence of hypertension,a history of stroke/transient ischemic attack（TIA）,peripheral vascular disease,congestive heart failure,paroxysmal AF,and absence of anti-coagulation therapy,as well as enlarged left atrial diameter,increased left ventricular end-diastolic diameter,decreased left ventricular ejection fraction（LVEF）,reduced left atrial appendage emptying velocity,and a lower estimated glomerular filtration rate（e GFR）.Univariate logistic regression analysis revealed that previous stroke/TIA,congestive heart failure,hypertension,peripheral vascular diseases,non-paroxysmal AF,absence of anti-coagulation therapy,enlarged left atrial diameter,increased left ventricular end-diastolic diameter,decreased LVEF,slower left atrial appendage emptying velocity,and reduced e GFR were independent risk factors for LAT/SEC.Multivariate logistic regression analysis demonstrated that a history of stroke/TIA,congestive heart failure,non-paroxysmal AF,absence of anti-coagulation therapy,increased left atrial diameter,enlarged left ventricular end-diastolic diameter,decreased LVEF,slower left atrial appendage empties,and reduced e GFR were independent risk factors for LAT/SEC.Summary:1.Among the 655 NVAF patients included in this study,10.53%had LAT/SEC;2.Independent risk factors for LAT/SEC included a history of stroke/TIA,congestive heart failure,non-paroxysmal AF,absence of anti-coagulation therapy,increased left atrial diameter,enlarged left ventricular end-diastolic diameter,decreased LVEF,slower left atrial appendage empties,and reduced e GFR.Part Two Validation and comparison of CHA₂DS₂-VASc-RAF and CHA₂DS₂-VASc-LAF scores for predicting left atrial thrombus and spontaneous echo contrast in patients with non-valvular atrial fibrillationObjective:This study aimed to investigate the efficacy of CHA₂DS₂-VASc-RAF（R was renal function,AF was AF type）and CHA₂DS₂-VASc-LAF（L was left atrial diameter,AF was AF type）scores in predicting LAT/SEC in patients with NVAF and compare them with CHADS₂ and CHA₂DS₂-VASc scores.Methods:The material collection process remained consistent with the first part.The area under the receiver operating characteristic curve（AUC）was used to evaluate the diagnostic performance of prediction scores,such as CHA₂DS₂-VASc-RAF and CHA₂DS₂-VASc-LAF scores.Pairwise comparison of the four prediction scores and models were performed using the Delong method.Results:Patients in the LAT/SEC group,when compared with those in the non-LAT/SEC group,had higher CHADS₂（1.0 vs.2.0）,CHA₂DS₂-VASc（1.0 vs.2.5）,CHA₂DS₂-VASc-LAF（2.0 vs.5.0）,and CHA₂DS₂-VASc-RAF（2.0 vs.6.0）scores.In the training cohort（n=399）,the AUC values for CHADS₂,CHA₂DS₂-VASc,CHA₂DS₂-VASc-LAF,and CHA₂DS₂-VASc-RAF scores were 0.736,0.737,0.829,and 0.839,respectively.In the test cohort（n=266）,the AUC values for CHADS₂,CHA₂DS₂-VASc,CHA₂DS₂-VASc-LAF,and CHA₂DS₂-VASc-RAF scores were 0.691,0.701,0.829,and 0.836,respectively.In the training and test cohorts,pairwise comparisons using the Delong method revealed no significant difference between CHA₂DS₂-VASc-LAF and CHA₂DS₂-VASc-RAF,but the AUC values for the two scores were higher than those of CHADS₂and CHA₂DS₂-VASc scores.Summary:1.Patients in the LAT/SEC group,when compared with those in the non-LAT/SEC group,had higher CHADS₂,CHA₂DS₂-VASc,CHA₂DS₂-VASc-LAF,and CHA₂DS₂-VASc-RAF scores（P<0.001）.2.The predictive efficacy of CHA₂DS₂-VASc-RAF and CHA₂DS_2-VASc-LAF scores was superior to that of the traditional CHADS₂ and CHA₂DS₂-VASc scores.Part Three Construction and validation of prediction models for left atrial thrombosis or spontaneous in patients with non-valvular atrial fibrillationObjective:The nomogram and random forest prediction model was constructed based on risk the factors for LAT/SEC,and the two models were compared with CHA₂DS₂-VASc-RAF,CHA₂DS_2-VASc-LAF,CHA₂DS₂-VASc,and CHADS₂ scores.Method:Patients with NVAF between January 2016 to August 2023comprised the training cohort,and 266 patients with NVAF from September2021 to August 2023 constituted the test cohort.1)The nomogram and random forest prediction models were constructed using the training cohort,and their predictive value was assessed by ROC curves.Subsequently,these models were validated using the test cohort.2)Nomogram and random forest models were constructed using the training cohort,and their predictive value was assessed by the AUC.Results:1.In the training cohort（n=399）,the nomogram model was constructed based on a history of stroke/TIA,congestive heart failure,non-paroxysmal AF,absence of anti-coagulation therapy,left atrial diameter,left ventricular end-diastolic diameter,LVEF,and e GFR.The ROC curve showed that the AUC of the nomogram model was 0.971,with 100%sensitivity and 81.9%specificity.In the test cohort（n=266）,the ROC curve showed that the AUC of the nomogram model was 0.941,with 87.1%sensitivity and 91.9%specificity.In the two cohorts,pairwise comparisons using the Delong method showed that the AUC of the nomogram model was higher than those of CHA₂DS₂-VASc-LAF,CHA₂DS₂-VASc-RAF,CHA₂DS₂-VASc,and CHADS₂ scores.2.The importance ranking of risk factors constructed by the random forest model was as follows:left ventricular end-diastolic diameter,e GFR,left atrial diameter,LVEF,non-paroxysmal AF,a history of stroke/TIA,absence of anti-coagulation therapy,and congestive heart failure.In the training cohort（n=399）,the AUC of the random forest model was 0.918,with 84.6%sensitivity and 89.2%specificity.In the test cohort（n=266）,the AUC of the random forest model was 0.871,with 90.0%sensitivity and 66.8%specificity.In the two cohorts,pairwise comparisons using the Delong method showed that the AUC of the random forest model was similar to those of the nomogram model,CHA₂DS₂-VASc-LAF,and CHA₂DS₂-VASc-RAF scores,but was higher than those of CHA₂DS₂-VASc and CHADS₂ scores.Summary:1.Among the two models and four scores,the AUC values from highest to lowest were as follows:nomogram model,random forest model,CHA₂DS₂-VASc-LAF,CHA₂DS₂-VASc-RAF,CHA₂DS₂-VASc,and CHADS₂scores.2.The predictive value of the nomogram model was the highest,which was similar to that of the random forest model but was higher than those of CHA₂DS₂-VASc-LAF,CHA₂DS₂-VASc-RAF,CHA₂DS₂-VASc,and CHADS₂ scores.Conclusions:1.In this study,10.53%of patients with NVAF had LAT/SEC.2.A history stroke/TIA,congestive heart failure,non-paroxysmal AF,absence of anti-coagulation therapy,increased left atrial diameter,enlarged left ventricular end-diastolic diameter,decreased LVEF,slower left atrial appendage empties,and reduced e GFR were independent risk factors for LAT/SEC.3.This study verified that the predictive efficacy of the LAT/SEC,CHA₂DS₂-VASc-LAF and CHA₂DS₂-VASc-RAF scores was recently found to be superior to that of the traditional CHADS₂ and CHA₂DS₂-VASc scores.4.The predictive value of the nomogram model was the highest,which was similar to that of the random forest model but was higher than those of CHA₂DS₂-VASc-LAF,CHA₂DS₂-VASc-RAF,CHA₂DS₂-VASc,and CHADS₂ scores.