Integrating filling pressure and atrial function: prognostic performance of the left atrial stiffness index in patients with preserved ejection fraction

Background The left atrial stiffness index (LASI), defined as the ratio of E/e′ to left atrial reservoir strain (LARs), has been proposed as a novel marker of left atrial–ventricular coupling and diastolic function. However, its diagnostic and prognostic utility relative to its individual components remains uncertain. Purpose To evaluate the diagnostic and prognostic performance of LASI in patients with preserved ejection fraction, and to compare its contribution with E/e′ and LARs. Methods We retrospectively analysed 289 in- or out-patients (LVEF>50%, sinus rhythm), who underwent TTE with complete image datasets available for re-evaluation, and had an available NT-proBNP. Complete acquisitions for measurement of key parameters within the British Society of Echocardiography 2024 diastolic dysfunction (DD) algorithm1 (LAVi, E/A, TDI velocities, vTR, and LARs) were mandatory. All studies were analysed using an AI-assisted reporting platform (EchoConfidence, Mycardium AI), with human review and approval. Regression modelling was used to assess the relative contribution of LASI and its components for the clinical outcomes of heart failure hospitalisation (HFH) and death, along with NT-proBNP. Results 59.1% were women. Mean LVEF was 61.4 ± 6.7%. E/e', LARs and LASI demonstrated significant associations with NTproBNP (all p<0.001), with non-linear relationships observed for LASI and LARs, but a predominantly linear association for E/e′ (Figure 1). As per the BSE 2024 algorithm, 19.1% had impaired diastolic function with normal filling pressure, and 56.7% had impaired diastolic function with elevated pressure. All 3 parameters demonstrated gradual increases with worsening diastolic grade (all p<0.001), strongest for E/e′ (ρ=0.59). Over a median follow-up of 24 months, 84 composite events occurred. When stratified by cohort median (0.57) and adjusted for age, LASI had a significant impact on death, HFH and the combined endpoint (p=0.01, p=0.04, p=0.01; Figure 2). Of its two components, E/e′ had the greatest contribution, significantly improving model fit over age alone in the NT-proBNP (χ2: 59.8 vs 38.4, p<0.001) and the outcome model (χ2: 27.6 vs 21.3, p=0.012). Adding LARs provided additional explanatory value for NT-proBNP (χ2: 70.2 vs 59.8, p=0.001), but not for outcomes (χ2: 28.5 vs 27.6, p=0.341). Combining them into LASI rather than individually caused loss of information in the NT-proBNP model (χ2: 59.7 vs 70.2, p=0.02), though it was preserved in the outcome model (χ2: 30.5 vs 28.5, p=0.7). Conclusions In a real-world cohort with preserved EF, LASI is strongly associated with NT-proBNP, diastolic dysfunction severity, and adverse clinical outcomes. As an integrative marker of atrial–ventricular coupling, LASI consolidates the prognostic information of E/e′ and LARs into a single parameter, yet it does not consistently add information beyond the additive contribution of its constituents in this cohort.Figure 1Figure 2