Results
Out of the 274 patients, LA strain analysis was not evaluable in 39 because of the presence of atrial fibrillation with irregular rhythm and in 7 because of poor quality of imaging. The clinical and echocardiographic characteristics of the remaining 228 are shown in Table 1. Among the patients enrolled, 17 presented atrial fibrillation with regular ventricular paced rhythm; 31 (13.6%) were in KDOQI Class 1, 111 (48.7%) in Class 2, 43 (18.9%) in Class 3a, 34 (14.9%) in Class 3b, 8 (3.5%) in Class 4, and 1 in Class 5.
Left atrial strain measures’ reproducibility. A high degree of intra-observer and intra-observer reproducibility was observed for LAr (average ICC = 0.93; 95% CI: 0.84–0.97 and ICC: 0.87; 95% CI: 0.71–0.95, respectively) as well as for LAcd (ICC: 0.93; 95% CI: 0.84–0.97 and ICC: 0.86; 95% CI: 0.68–0.94, respectively) and LAct (ICC: 0.89; 95% CI: 0.76–0.96 and ICC: 0.80; 95% CI: 0.56–0.91, respectively).
Clinical and echocardiographic correlates of left atrial strain measures. As shown in Supplementary Table 1, LAr was significantly and positively correlated with GFR-EPI, LVEF, TAPSE, e’s, e’l, and e’m and significantly and negatively correlated with age, NYHA class, heart rate, LVEDV, LVESV, LAVi, MR, TR, E/e’s, E/e’m, LV-GLS, RV-GLS, and RVfwLS. LAcd was significantly and positively correlated with age, NYHA class, heart rate, LVEDV, LVESV, LAVi, TR, E/e’s, E/e’m, LV-GLS, RV-GLS, and RVfwLS and significantly and negatively correlated with GFR-EPI, LVEF, TAPSE, e’s, e’l, and e’m. LAct was significantly and positively correlated with NYHA class, heart rate, LVEDV, LVESV, LAVi, MR, TR, E/e’s, E/e’m, LV-GLS, RV-GLS, and RVfwLS and significantly and negatively correlated with GFR-EPI, LVEF, and TAPSE. LAr/Eel, LAr/Ees, and LAr/Eem were significantly and positively correlated with SAP, GFR-EPI, LVEF, TAPSE, e’s, e’l, and e’m and significantly and negatively correlated with age, NYHA class, heart rate, LVESV, LAVi, MR, TR, E/e’s, E/e’m, LV-GLS, RV-GLS, and RVfwLS.
Left atrial strain measures and heart failure worsening. During a mean follow-up of 14±7 months, at least one event related to heart failure worsening (40 hospitalisations caused by acute decompensated heart failure, 5 heart transplantations, and 19 cardiovascular deaths) occurred in 47 patients. As shown in Table 2, during univariate Cox regression analysis, LAr, LAcd, and LAct were all associated with events related to heart failure worsening. In forward stepwise Cox regression analysis, LAr (HR: 0.94; 95% CI: 0.91–0.98; p: 0.006) and LAct (HR: 1.05; 95% CI: 1.00–1.10; p: 0.043) but not LAcd remained associated with events after correction for NYHA class, heart rate, age, MR, TR, LVEF, GFR-EPI, and TAPSE.
Analogously, LAr/Eel, LAr/Rrs, and LAr/Eem were associated with events during univariate analysis, but only LAr/Ees and LAr/Eem remained significantly associated with them during multivariate analysis (HR: 0.49; 95% CI: 0.30–0.78; p: 0.002 and HR: 0.65; 95% CI: 0.47–0.89; p: 0.008, respectively) after correction for NYHA class, heart rate, age, MR, TR, LVEF, GFR-EPI, and TAPSE. As shown in Figure 2, the presence of a reduced or mildly reduced/preserved LVEF did not influence the prognostic value of all the LA strain measures as well as of those derived by the combination with Doppler measures.
ROC curves. As shown in Supplementary Figure 1, among the LA measures derived by strain analysis, LASr/Ees (AUC 0.76; 95% CI: 0.67–0.85) showed the greater ROC AUC, followed by LASr/Eem (AUC 0.75; 95% CI: 0.65–0.84), LASr (AUC 0.71; 95% CI: 0.63–0.80), LASct (AUC 0.70; 95% CI: 0.61–0.78), and LAScd (AUC 0.60; 95% CI: 0.52–0.69). For LASr, the best cut-off was 17% (sensitivity of 72% and specificity of 60%), for LASr/Ees 1.28 (sensitivity of 80% and specificity of 64%), and for LASr/Eem 1.60 (sensitivity of 77% and specificity of 59%). Figure 3 shows the Kaplan–Meier curves of the patients dichotomised according to the best cut-off of LASr (Panel A) and LASr/Ees (Panel B).