We pooled individual patient data from observational studies conducted at our centers that assessed the prognostic value of TTE in patients with chronic HF due to ischemic etiology or idiopathic cardiomyopathy, in stable clinical conditions over the last 3 months and aged >18 years. ^,, Patients were defined as having HFrEF if left ventricular ejection fraction (LVEF) was 40% and HFpEF if LVEF was 40%. The study population included also a control group of healthy subjects and subjects with CVRF (i.e. arterial hypertension and/or diabetes mellitus and/or dyslipidemia) without overt cardiac involvement. To be eligible for inclusion, patients had to be characterized with the following TTE parameters: LVEF, sPAP and TAPSE/sPAP. Detailed information on patients’ clinical conditions, medical history, medications and laboratory data (including creatinine and natriuretic peptides) was recorded for each patient at the time of the TTE examination. Patients were followed up by periodic clinical visits and/or telephone calls. The primary endpoint of survival analysis was all-cause death.

A standardized TTE examination was carried out with commercial equipment in high-volume and high-quality echocardiographic laboratories. All measurements were assessed by an expert noninvasive cardiologist at each center according to the American Society of Echocardiography and the European Association of Cardiovascular Imaging guidelines recommendations. ^, Peak sPAP was calculated from the maximum velocity of tricuspid regurgitation (TRV) + the estimate of right atrial pressure based on inferior vena cava diameter and collapsibility. TAPSE was measured by M-mode echocardiography in apical 4-chamber view. The TAPSE/sPAP ratio was used as a noninvasive index of RV/PA coupling 4. For patients in atrial fibrillation, all measurements were repeated at least three times, and the average value was calculated.

Overall statistical analysis was conducted using an integrated dataset of participants stratified by four subgroups, including healthy controls and patients with CVRF, HRrEF and HFpEF. For the focus of the study, sPAP, TAPSE and TAPSE/sPAP ratio were adjusted for potential confounders including age and sex. Participants with complete data on echocardiographic variables, survival time, and event (all-cause death) status were included. Figure 1 shows the study flow-chart.

Study flowchart.

Overall survival was defined as time to death (in months). To assess the association between sPAP, TAPSE and TAPSE/sPAP ratio with mortality we applied Cox proportional hazards regression models, adjusted for age and sex.

For prognostic threshold identification, we applied smoothing spline regression to the predicted hazard ratio (HR) values, where HR = 1 indicates neutral risk. Without supposing a predetermined functional form, smoothing splines are adaptable, nonparametric regression approaches that enable modeling of nonlinear interactions between a continuous predictor and the outcome. Using penalized smoothing splines, which fit a smooth curve through HR estimations to minimize random fluctuations while maintaining the trend, we showed the predicted HR across the continuous range of each variable. The threshold was identified as the intersection point between the smoothed HR curve and the reference line at HR = 1, marking the parameter value at which the mortality risk transitions. For visualization, we plotted the smoothed HR curve, incorporating 95% confidence intervals (CI) to account for uncertainty. The final cut-off point was marked on the graph to enhance interpretability. Survival probabilities were visualized using Kaplan–Meier curves and risk tables were added for interpretability. To compare the performance of the sPAP or TAPSE vs TAPSE/sPAP, we used the following statistical indices: (1) Concordance Index (C-index): a measure that quantifies the model’s ability to correctly rank survival times, that is, to distinguish between patients who experience the event (e.g., death) earlier versus later. A C-index of 0.50 indicates no discriminative ability (equivalent to random chance), whereas a value of 1.00 reflects perfect discrimination; (2) Likelihood Ratio Test (LRT): used to compare the goodness-of-fit of nested models. A significant LRT (p Value <0.05) indicates that the more complex model provides a significantly better fit to the data than the simpler one; (3) Akaike Information Criterion (AIC): a measure of model quality that balances model fit and complexity. Lower AIC values indicate better models, penalizing unnecessary parameters to avoid overfitting.

Overall statistical analyses were conducted using R software (R-project.org, version 4.3.1, Austria.

The study population consisted of 2,378 HF patients, 1,660 with HFrEF and 718 with HFpEF, 216 healthy subjects and 210 subjects with CVRF. The baseline demographic, clinical and echocardiographic characteristics of the study cohorts are presented in Table 1 .

In the healthy control group, mean values of PASP were 22.3 ± 5.8 mmHg, of TAPSE were 22.9 ± 3.4 mm and of TAPSE/PASP were 1.1 ± 0.5 mm/mmHg. In subjects with CVRF, mean values of PASP were 22.3 ± 5.8 mmHg, of TAPSE were mm and of TAPSE/PASP were 1.1 ± 0.5 mm/mmHg. In the HFrEF group, mean values of PASP were 35.0 ± 13.1 mmHg, of TAPSE were 18.8 ± 4.2 mm and of TAPSE/PASP were 0.63 ± 0.32 mm/mmHg. In the HFpEF group, mean values of PASP were 29.8 ± 12.2 mmHg, of TAPSE were 20.6 ± 4.0 mm and of TAPSE/PASP were 0.80 ± 0.35 mm/mmHg. The final analyses included cases with a follow-up duration of up to 113 months. Over this period, 256 deaths were recorded in the HFrEF group, 117 in the HFpEF group, 18 among subjects with CVRF.

The adjusted association between sPAP and all-cause-death is shown through the spline plot presented in the Central Illustration: the model, adjusted for age and sex, assumes a log-linear relationship between sPAP and the logarithm of the HR in the entire cohort of HF patients (n = 2,378) over a wide range with an estimated threshold of 33.34 mmHg. Specifically, the HR value for sPAP (modeled as continuous variable) of 1.03 was significant, indicating a 3% increase in mortality risk per mmHg increase in sPAP. Furthermore, the HR for age was also significantly associated with increased mortality risk (p Value = 1.92 × 10-12), while the HR for male sex was not statistically significant.

Supplementary Figure 1 shows the frequency distribution of sPAP within the entire cohort. The association between sPAP and outcome adjusted for age and sex was similar across the different subgroups of patients, i.e. in HFrEF patients as compared to HFpEF patients ( Supplementary Figure 2A ), in patients with ischemic as compared to nonischemic etiology ( Supplementary Figure 2B ), and in CVRF subjects ( Supplementary Figure 2C ). Supplementary Figure 2D shows Kaplan–Meier survival curves for HF patients (n = 2,378) with sPAP below or above the risk threshold of 33.34 mmHg. Supplementary Figure 2E–F show Kaplan–Meier survival curves respectively for HFrEF and for HFpEF patients with sPAP below or above the threshold of 33.34 mmHg.

The adjusted association between TAPSE and all-cause death adjusted for age and sex in all HF patients is shown in Supplementary Figure 3A . When TAPSE was modelled continuously in the entire cohort of HF patients (N = 2,378), the all-cause mortality HR adjusted for age and sex showed a progressive nonlinear increase in the risk of death as TAPSE decreased below a threshold at 19.25 mm.

The association between TAPSE and all-cause mortality was significant in HFrEF patients but not in patients with HFpEF ( Supplementary Figure 3B ). The association between TAPSE and outcome was similar in patients with ischemic as compared to nonischemic etiology ( Supplementary Figure 3C ) and in CVRF subjects ( Supplementary Figure 3D ).

Supplementary Figure 4 shows the frequency distribution of TAPSE/sPAP within the entire cohort. The adjusted association between TAPSE/sPAP and all-cause death is shown in the Central Illustration. When TAPSE/sPAP was modelled continuously in the entire cohort of HF patients (N = 2,378), the all-cause mortality HR adjusted for age and sex showed a progressive nonlinear increase in the risk of death as TAPSE/sPAP decreased below a threshold at 0.68 mm/mmHg. Specifically, the HR value for TAPSE/sPAP (modeled as continuous variable) of 0.16 was significant, indicating a 84% decrease in mortality risk per mm/mmHg increase in TAPSE/sPAP. Figure 2 shows Kaplan–Meier survival curves for HF patients with TAPSE/sPAP >0.68 mm/mmHg (red line) or 0.68 mm/mmHg (green line). The association between TAPSE/sPAP and all-cause death was similar across the different subgroups of patients, i.e. in HFrEF patients as compared to HFpEF patients ( Figure 3A ), in patients with ischemic as compared to nonischemic etiology ( Figure 3 B), as well as in CVRF subjects ( Figure 3 C).

Kaplan–Meier survival curves for HF patients with TAPSE/sPAP >0.68 mm/mmHg (red line) or 0.68 mm/mmHg (green line).

(A) Adjusted association between TAPSE/sPAP and all-cause death in HFrEF patients (solid line) as compared to HFpEF patients (dotted line). (B) Adjusted association between TAPSE/sPAP and all-cause death in patients with ischemic (solid line) as compared to non-ischemic etiology (dotted line). (C) Adjusted association between TAPSE/sPAP and all-cause death in CVRF subjects.

In Table 2 different metrics are reported to compare the accuracy of TAPSE, sPAP and TAPSE/sPAP in predicting all cause death in the overall population of HF patients. The TAPSE/sPAP demonstrated better performances compared to TAPSE and sPAP in predicting mortality as shown by improved model fit (likelihood ratio test), predictive accuracy (C-index) and AIC (Akaike information criterion). The HRs further underlined the protective effect of a higher TAPSE/sPAP ratio. Table 2 and Table 3 show the accuracy of TAPSE, sPAP and TAPSE/sPAP in predicting all cause death respectively in HFpEF and HFrEF patients. In patients with HFpEF, TAPSE/sPAP performed better than TAPSE and similar to sPAP, while in patients with HFrEF, TAPSE/sPAP performed better than both TAPSE and sPAP. Table 4

Table 2

Accuracy of sPAP or TAPSE alone or TAPSE/sPAP in predicting all-cause mortality in the overall heart failure patient population based on different metrics

Metric	Adjusted TAPSE/sPAP	Adjusted sPAP	Adjusted TAPSE
n. events	373	373	373
HR	0.16 (95% CI: 0.10–0.23)	1.03 (95% CI: 1.02–1.03)	0.92 (95% CI: 0.89–0.94)
C-index	0.68	0.67	0.65
Likelihood ratio test	168.80	148.40	118.20
AIC	4,904.14	4,923.41	4,954.65

Only gold members can continue reading. Log In or Register to continue

Share this:

• X
• Facebook

Related

Related posts:

1. Native Coronary Artery Versus Saphenous Vein Graft Percutaneous Coronary Intervention After Coronary Artery Bypass Grafting: A Contemporary Synthesis
2. Extended Pharmacoinvasive PCI Compared to Primary PCI: Insights From Madras Medical College STEMI Registry
3. Prevalence of Desmoplakin Variants in Patients With Recurrent Myocarditis
4. National Patterns of Pulmonary Hypertension Mortality in the United States, 2010 to 2024

Stay updated, free articles. Join our Telegram channel

Join