E/e' ratio combined with left ventricular mass index predicts HFpEF rehospitalization risk in stage 3–4 chronic kidney disease patients: a retrospective cohort study

Figures

• 

  Figure 1. Correlation between E/e’ and LVMI with kernel density distribution. Scatter plot showing the relationship between E/e’ ratio and LVMI in 254 patients with stage 3–4 CKD and HFpEF. Blue circles represent patients with negative combined indicator status (n = 166); red circles represent patients with positive combined indicator status (n = 88). The black dashed line indicates the linear regression line for the overall cohort. Marginal density plots display the kernel density distributions along each axis. Spearman's correlation coefficient and statistical significance are shown. CKD, chronic kidney disease; E/e’, ratio of early transmitral flow velocity to early diastolic mitral annular velocity; HFpEF, heart failure with preserved ejection fraction; LVMI, left ventricular mass index.

• 

  Figure 2. Cumulative HFpEF rehospitalization rate by combined indicator Status. Kaplan–Meier survival curves showing the cumulative incidence of HFpEF rehospitalization stratified by combined echocardiographic indicator status in 254 patients with stage 3–4 chronic kidney disease. The blue line represents patients with negative combined indicator status (n = 166), and the red line represents patients with positive combined indicator status (n = 88). Shaded areas indicate 95% confidence intervals. The risk table below the plot shows the number of patients at risk at each time point. During a median follow-up of 2.1 years, 68 primary endpoint events occurred (overall incidence rate 26.8%). Statistical comparison performed using log-rank test. HFpEF, heart failure with preserved ejection fraction.

• 

  Figure 3. Time-dependent ROC curves at 1-year for HFpEF rehospitalization prediction. The curves depict model discrimination for the primary endpoint, with sensitivity (true positive rate) plotted against 1—specificity (false positive rate). The diagonal dashed line represents the performance of a non-informative random classifier. Solid lines correspond to the combined model, the E/e′ model, and the LVMI model. The legend reports the area under the curve (AUC) with corresponding 95% confidence intervals, as calculated from time-dependent ROC analysis.

• 

  Figure 4. Decision curve analysis for the prediction models. Decision curve analysis showing the net benefit of different prediction strategies across threshold probabilities for HFpEF rehospitalization risk. The y-axis represents the net benefit. The x-axis represents the threshold probability (%). The gray line (“None”) assumes no patients experience the event. The red line (“All”) assumes all patients experience the event. The blue line represents the combined echocardiographic indicator model, while the orange dashed line and green dot-dash line represent the E/e’ and LVMI single-parameter models, respectively. The light blue shaded area highlights the clinically relevant threshold probability range (10%–50%). Vertical dotted lines mark the boundaries of this range. E/e’, ratio of early transmitral flow velocity to early diastolic mitral annular velocity; HFpEF, heart failure with preserved ejection fraction; LVMI, left ventricular mass index.

• 

  Figure 5. Subgroup analysis of combined indicator on rehospitalization risk. The forest plot displays hazard ratios (HR) and 95% confidence intervals (CI) for the association between the combined indicator and HFpEF rehospitalization risk across predefined subgroups. Each square represents the point estimate of the hazard ratio for the respective subgroup, with the size of the square proportional to the statistical weight of the estimate. Horizontal lines extending from each square represent the 95% confidence intervals. The vertical dashed line at HR = 1.0 indicates no effect (null hypothesis). Hazard ratios greater than 1.0 (to the right of the vertical line) indicate increased risk of rehospitalization, while ratios less than 1.0 indicate decreased risk. The table on the left presents detailed information for each subgroup including sample sizes with event counts, proportions of events in combined indicator positive and negative groups, hazard ratios with 95% confidence intervals, individual p-values, and p-values for interaction testing. Hazard ratios and 95% confidence intervals were estimated using Cox proportional hazards models adjusted for all variables in the fully adjusted model…