Combining D-dimer and LDL/HDL ratio to predict the absence of atrial fibrillation in patients with an Implantable Loop Recorder for embolic stroke of Undetermined source

Background and Objective: Embolic stroke of undetermined source (ESUS) patients undergoing long-term rhythm monitoring with Implantable Loop Recorder (ILR) have an atrial fibrillation (AF) detection rate of approximately 12 % at 1 year and 30 % with extended follow-up over 3 years. However, research specifically focusing on the majority of patients in whom AF is not detected through implantable cardiac monitors remains limited. Abnormal lipid profiles may be associated with embolic risks from non-AF sources. This study aimed to develop a model to predict the absence of AF in patients with ESUS using multiple variables, including lipid profiles. Methods: A retrospective, multicenter cohort study was conducted across four institutions, involving 99 ESUS patients. Patients were categorized based on AF detection via ILR. Patient characteristics, blood test results, and echocardiographic findings were assessed through univariate and multivariate logistic regression analyses. ROC curve analysis was performed to evaluate the biomarkers' predictive accuracy. Results: AF was detected in 30.3 % of patients over a median follow-up of 25.5 months. Multivariate analysis confirmed elevated D-dimer (OR: 2.77, p = 0.002), low LDL/HDL ratio (OR: 2.0, p = 0.01), and CHA2DS2-VASc score (OR: 1.4, p = 0.04) as independent predictors of AF detection. The CHA2DS2-VASc score was excluded due to multicollinearity, and patients with D-dimer < 0.9 mu g/ml and LDL/HDL ratio > 1.98 had significantly lower AF detection rates (6.8 %, P < 0.001; sensitivity 93.1 %, specificity 44.2 %). Conclusion: Combining D-dimer and LDL/HDL ratios provides an effective and accessible method for predicting the absence of AF in patients with an ILR for ESUS.