Artificial intelligence-enhanced three-dimensional echocardiography reveals left atrial-ventricular coupling index as...

BACKGROUND: The Left Atrial-Ventricular Coupling Index (LACI), which quantifies the volumetric interaction between the left atrium and ventricle, has emerged as a promising prognostic marker for cardiovascular events. METHODS: We applied three-dimensional echocardiography (3DE) and AI quantitative software to automatically obtain the new echocardiographic parameter LACI in 1084 coronary artery disease (CAD) patients at the Fourth Affiliated Hospital of China Medical University. LACI was defined as the ratio between the left atrial end-diastolic volume and the left ventricular end-diastolic volume. The primary endpoints were major adverse cardiovascular events (MACE), including unstable angina, myocardial reinfarction, heart failure (HF) and death. RESULTS: Over a median follow-up of 13 months, 216 MACE were recorded. LACI was significantly higher in patients with MACE than in those without (0.22 [IQR 0.17–0.28] vs. 0.20 [IQR 0.15–0.25], p < 0.001). ROC analysis indicated that LACI demonstrated higher predictive accuracy for MACE than traditional echocardiography parameters. Multivariable Cox regression analysis revealed that LACI was significantly correlated with MACE (HR: 1.68, 95% CI [1.25–2.27], p < 0.001), especially in patients with severe lesion subgroups according to the Gensini scores (HR 1.33, 95% CI [1.17–1.52], p < 0.002) and male (HR: 1.37, 95% CI [1.20–1.56], p < 0.002). Furthermore, LACI assessment enabled further risk stratification in high-risk patients with severe lesion (p < 0.001 on log-rank testing). Consistently, category-free NRI and IDI confirmed the improvement by LACI to stratify MACE risk in severe lesion patients (p < 0.05). CONCLUSION: Our findings demonstrated that AI-enhanced 3DE-derived LACI provided incremental prognostic value beyond conventional echocardiographic parameters in assessing the prognosis among CAD patients, especially in severe coronary lesion subgroups.