Deep Proteomic Analysis for Non-Invasive Left Atrial Appendage Thrombus Prediction in Patients with Non-Valvular Atrial Fibrillation
Ontology highlight
ABSTRACT: Methods: Deep proteomic analysis on plasma from 8 non-valvular AF patients categorized into thrombus and control groups based on LAAT presence was conducted. Identified biomarkers were validated using enzyme-linked immunosorbent assay in a cohort of 179 patients, with clinical, transthoracic echocardiography, and transesophageal echocardiography data collected. A Lasso regression-based predictive model was constructed. Results: The Thrombus group had higher CHA2DS2-VASc scores, larger left atrial appendage thrombus (LAA) diameter, and lower LAA flow velocity. Deep proteomic analysis identified 30 differentially expressed proteins, with MYL4, PCYOX1, and DCN selected as diagnostic biomarkers (P<0.0001). The model included MYL4, PCYOX1, DCN, LAA cauliflower morphology, and CHA2DS2-VASc score. The model's area under the curve is 0.983. Additionally, net reclassification improvement and integrated discrimination improvement indices were also calculated, yielding values of 1.797 and 0.740, respectively, indicating a substantial improvement in risk prediction and classification over the conventional CHA2DS2-VASc model alone. Conclusions: Elevated MYL4, decreased PCYOX1 and DCN levels, and cauliflower LAA morphology are independent risk factors. Incorporating these variables into the prediction model improves the diagnostic performance of the LAAT prediction model, potentially offering a non-invasive and clinically valuable approach to stroke prevention. Keywords: Atrial Fibrillation; Thrombosis; Proteomics
ORGANISM(S): Homo Sapiens
SUBMITTER: Yang Li
PROVIDER: PXD047778 | iProX | Tue Dec 12 00:00:00 GMT 2023
REPOSITORIES: iProX
ACCESS DATA