Abstract
Background
Left heart failure (LHF) is the most common in clinical setting and is more likely to experience adverse cardiovascular events. This study aims to establish a predictive nomogram of adverse events based on cardiopulmonary ultrasound (CPUS) parameters and clinical predictors for LHF patients. And the predictive performance of the nomogram was assessed.
Methods
Potential risk factors associated with the occurrence of adverse events in LHF were analyzed using one-way analysis of variance (ANOVA) or univariate logistic regression. A multivariate logistic regression model was subsequently constructed, and the results were visualized with a nomogram. Then the predictive ability and clinical application value of the nomogram were verified by the concordance index (C-index), calibration curve and decision curve.
Results
A total of 329 heart failure patients including 278 males and 51 females, with an average age of (53.62±15.85) years were enrolled in this study. Independent influencing factors of adverse events were identified through multivariate logistic regression. Finally, four indicators were included, namely, systolic pulmonary artery pressure (sPAP) [odds ratio (OR) =15.53, 95% confidence interval (CI): 1.53–157.19, P=0.02], left atrial transverse diameter (LAD) (OR =1.190, 95% CI: 1.031–1.373, P=0.02), E/e’ ratio of left ventricle (LV-E/e’) (OR =1.241, 95% CI: 1.101–1.399, P<0.001), and the number of B lines (OR =1.591, 95% CI: 1.123–2.254, P=0.009). The C-index of the nomogram model for predicting the risk of adverse events in LHF patients were 0.888 (95% CI: 0.826–0.950). Furthermore, the calibration curve demonstrated great consistency between the predicted probabilities and the observed outcomes, and the decision curve analysis confirmed the important clinical advantage of the model in LHF patients.
Conclusions
CPUS may provide valuable prognostic information in patients with LHF. The constructed prediction model for adverse events, incorporating CPUS parameters including sPAP, LAD, LV-E/e’, and number of B lines, demonstrated favorable discriminative ability and accuracy within the study cohort.
Keywords: Left heart failure (LHF), cardiopulmonary ultrasound (CPUS), prediction model, nomogram, adverse events
Highlight box.
Key findings
• This study established a nomogram based on cardiopulmonary ultrasound (CPUS) parameters and clinical influencing factors for estimating the risk of adverse events in left heart failure (LHF) patients. And the nomogram had good practical prediction ability, good differentiation and accuracy..
What is known and what is new?
• The parameters of CPUS were closely related to the occurrence of adverse events in LHF patients..
• This study provided more clinical reference for individual treatment of LHF patients, and may help improve the prognosis of LHF patients..
What is the implication, and what should change now?
• The nomogram was established with limited sample size and lack of external validation, only derivation cohort without validation cohort present in the study, further studies were needed to validate the accuracy and applicability of this predictive model.
Introduction
Left heart failure (LHF) is associated with high mortality rates and unsatisfactory long-term prognoses, currently ranking as the leading cause of death among individuals with cardiovascular disorders (1). As a critical and severe complication of LHF, pulmonary hypertension (PH) and pulmonary edema (PE) are key predictive factors for adverse outcomes. It significantly increases patients’ risks of emergency department attendance and hospital readmission, and is therefore regarded as a core therapeutic intervention target for HF management.In 2013, the Global Burden of Disease Study reported 61.7 million cases of heart failure (HF) worldwide, which represented almost a doubling since 1990 (2). Studies have shown that LHF patients with PH had more severe condition of disease and worse prognosis (3). And PE progresses rapidly and has a high mortality rate, characterized by rapid disease progression, cardiogenic PE poses a severe threat to patient prognosis (4,5). They were also more likely to experience adverse cardiovascular events such as respiratory failure, cardiogenic shock (CS) and death. Consequently, it is crucial to perform a comprehensive adverse event risk assessment and to adjust the treatment in time, in order to effectively manage adverse events and reduce mortality.
Cardiopulmonary ultrasound (CPUS), a newly developed examination method in recent years, offers great convenience for clinical diagnosis and disease management. CPUS can be used to evaluate key factors which affect patient’s prognosis including the structure and function of the heart, pulmonary artery pressure and PE. Öhman et al. research found that CPUS seemed safe and feasible for guiding acute heart failure (AHF) treatment. Ultrasound-guided treatment might be associated with improved total decongestion, shorter hospitalizations, and improved prognosis (6). In a prospective study published in 2023, Ruben et al. found that point-of-care CPUS was an invaluable tool to approach individuals with CS in order to obtain an accurate diagnosis and to favour proper clinical management (7). An expert consensus issued by the European Society of Cardiology also mentioned that the combined application of echocardiography and pulmonary ultrasound (PUS) can help improve diagnostic accuracy and better monitor intervention response (8).
At present, there were relatively few studies on the application of CPUS, most of which focus on the correlation between PUS and echocardiography. And most current models in LHF patients rely on clinical indicators alone, while the integration of clinical predictive utility of ultrasound signatures for CPUS remains underexplored for early risk prediction.
In the study, we sought to investigate the correlation between CPUS parameters and adverse events, and then to establish a nomogram based on CPUS parameters and clinical predictors for estimating the risk of adverse events in LHF patients. The predictive performance of the nomogram was assessed. The purpose of this study is essentially to create a paradigm for future studies of the management of LHF patients, facilitate early in-hospital risk stratification for patients, enabling timely intervention, and to search for evidence CPUS to serve as a non-invasive method in monitoring the prognosis of LHF patients. We present this article in accordance with the TREND reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-2226/rc).
Methods
Study design
The individuals with HF in the study mainly coursed by coronary heart disease, dilated cardiomyopathy, alcoholic cardiomyopathy and hypertensive heart disease. The main etiology of was determined by the clinical doctor through comprehensive assessment based on the symptoms of the patients, laboratory tests, imaging examinations and other means. Patients with significant coronary artery disease were not excluded from the present cohort. For patients identified with significant coronary stenosis during clinical workup, coronary revascularization was performed as clinically indicated following standard clinical practice guidelines.
The exclusion criteria are as follows: (I) congenital heart disease; (II) pericarditis; (III) restrictive cardiomyopathy; (IV) hypertrophic cardiomyopathy; (V) cardiac valve stenosis narrow; (VI) pulmonary disease or thromboembolic PH (underwent comprehensive pulmonary function tests or highly suspected thromboembolic PH filling defects in pulmonary arteries on computed tomography pulmonary angiography (CTPA), mismatched perfusion defects on V/Q scan were excluded); (VII) severe liver and kidney failure; (VII) malignant tumors. According to the occurrence of adverse events they were divided into adverse event group (n=81) and no adverse event group (n=248). The participant selection flowchart is depicted in Figure 1.
Figure 1.
Case screening flow chart.
all adjudicators involved in endpoint assessment were strictly blinded to the predictor variables throughout the study process. To ensure the rigor of blinding, we established an independent adjudication committee composed of three experienced clinicians who had no involvement in patient recruitment, treatment, or data collection. Adjudicators only received de-identified patient information (without any identifiers related to predictor variables) when reviewing endpoint events, and any discrepancies among adjudicators were resolved through consensus discussion.
Echo-suspected PH associated with left heart HF as an estimated systolic pulmonary artery pressure (sPAP) ≥40 mmHg, which was calculated as: sPAP = 4 × (TRV)2 + right atrial pressure (RAP). RAP was estimated based on the diameter and collapsibility of the inferior vena cava (IVC): RAP was assumed to be 5 mmHg if IVC diameter <1.5 cm and collapsibility >50%; 10 mmHg if IVC diameter 1.5–2.5 cm and collapsibility 20–50%; and 15 mmHg if IVC diameter >2.5 cm and collapsibility <20%. Adequate tricuspid regurgitation (TR) signals may be absent in a proportion of patients, limiting the reliability of this non-invasive estimation. All TTE examinations were reviewed by a second independent echocardiographer to ensure the accuracy of sPAP measurement, and any discrepancies were resolved through consensus.
Population
This was a prospective study. A total of 329 HF patients admitted to the department of cardiology, the First Affiliated Hospital of Guangxi Medical University and The People’s Hospital of Guangxi Zhuang Autonomous Region & Guangxi Academy of Medical Sciences from October 2021 to October 2024 were selected. The individuals diagnosis with HF according to the New York Heart Association Classification for the first time, and the cardiac function classification was grade III and grade IV (9). Specifically, among the 329 enrolled patients, 207 cases met the criteria of echocardiographically-estimated sPAP ≥40 mmHg and were classified as the PH-LHF group, while the remaining 122 cases with sPAP <40 mmHg constituted the HF group with normal pulmonary pressure.
The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The People’s Hospital of Guangxi Zhuang Autonomous Region and all subjects gave written informed consent to participate in the study. The First Affiliated Hospital of Guangxi Medical University was also informed of and agreed to the study. In the final dataset, the personal identification numbers were replaced by a code.
Predictors
All ultrasound predictors were measured within 24 hours of hospital admission:
Echocardiographic examination
Transthoracic echocardiographic measurements (Figure 2) were performed were when the patients admitted to the hospital for first time according to the 2015 American Association of Echocardiography (ASE) recommendation (10), using the commercially available EPIQ 7C and IE33 system (Koninklijke Philips N.V., Netherlands) equipped with the S5-1 probe (frequency range of 1–5 MHz) by the same investigator with more than 3 years of experience in echocardiographic diagnosis. All measurements were carried out and interpreted by the experienced physicians who were blinded to the objectives of the study to maintain an unbiased assessment.
Figure 2.
Echocardiographic examination. (A) Parasternal LAX view to obtain LVWT, obtain RVWT and LVEDD. (B,C) Parasternal SAX to obtain PAD and PAAT. (D-I) A4C to obtain LAD, RAD, RVEDD, RVWT, RVEF, sPAP, TAPSE, LV-E/e’, RV-E/e’ and Tei index. A4C, Apex four-chamber view; LAD, left atrial transverse diameter; LAX, long-axis view; LV-E/e’, ratio of pulsed Doppler of the mitral valve inflow to tissue Doppler of the lateral mitral annulus; LVEDD, left ventricular end-diastolic diameter; LVWT, left ventricular wall thickness; PAAT, pulmonary artery acceleration time; PAD, pulmonary artery diameter; RAD, right atrium transverse diameter; RV-E/e’, ratio of pulsed Doppler of the tricuspid inflow to tissue Doppler of the lateral tricuspid annulus; RVEDD, right ventricular end-diastolic diameter; RVEF, right ventricular ejection fraction; RVWT, right ventricular wall thickness; SAX, short-axis view; sPAP, systolic pulmonary artery pressure; TAPSE, tricuspid annular plane systolic excursion.
Left ventricular wall thickness (LVWT), left ventricular end-diastolic diameter (LVEDD), and right ventricular wall thickness (RVWT) were obtained in parasternal long-axis view (LAX) view. And pulmonary artery diameter (PAD) was obtained in parasternal short-axis view (SAX). Pulmonary artery acceleration time (PAAT) was assessed by pulse-wave Doppler in SAX view as well. And then, left atrial transverse diameter (LAD), right atrium transverse diameter (RAD), right ventricular end-diastolic transverse diameter (RVEDD) and right ventricular ejection fraction (RVEF) in apex four-chamber view (A4C). Left ventricular ejection fraction (LVEF) was indeed acquired using both the biplane Simpson’s disk‑summation and M-model in parasternal long-axis view. And we ultimately adopted the values obtained from the biplane Simpson’s method for the final analysis. Continuous Doppler was used to measure the differential velocity pressure of TR and to estimate sPAP. Tricuspid annular plane systolic excursion (TAPSE) was measured by M-mode. Diastolic tricuspid inflow velocity (E) was assessed by pulse-wave Doppler. The tricuspid annular movement was detected using tissue Doppler imaging (TDI), and early diastolic wave (e′), isovolumic contraction time (ICT), isovolumic relaxation time (IRT) and ejection time (ET) were obtained. RV myocardial work index (Tei index) = (ICT + IRT)/ET. LV-E/e’ was used to assess LV diastolic function. RV-E/e’ was used to assess RV diastolic function. And the Tei index reflects general right ventricular function.
PUS examination
A standard 8-zone image acquisition approach was used: the anterior area is from the anterior axillary line to the parasternal line, and the lateral region is from anterior axillary line to posterior axillary line. The anterior area was divided into anterior superior area and anterior inferior area, and the lateral area was divided into lateral superior area and lateral inferior area by the horizontal line equivalent to the second and third intercostal area and its extension. The pleural line, A-line, B-line, effusion and parenchyma of lung were observed. A semi-quantitative score was given. Definition of PUS score (Figure 3): (I) N: pulmonary sliding sign with A-line or less than 2 separate B-lines; (II) B1-Line: defined as multiple vertical B-lines with inter-line distance less than 7 mm, representing interlobular septal thickening; (III) B2-line: defined as vertical B-lines with inter-line distance less than 3 mm, indicating more severe alveolar-interstitial involvement; (IV) C: pleural effusion and the pulmonary present a tissue-like structure with dynamic air bronchogram. Each region is scored on the basis of the most severe performance: N is scored 0, B1-line is scored 1, B2-line is scored 2, and C is scored 3. The total score count was determined by summing the number of 8 possible zones. The range of score for each zone was 0 to 3. The range of total score was 0 to 24. In addition to the sonographer counting the score, an independent imaging score committee, composed of 2 blinded expert PUS sonographers, reviewed all images. The ICC for interobserver agreement was 0.89 [95% confidence interval (CI): 0.82–0.94], indicating excellent reproducibility of PUS scoring. The kappa value for PUS scoring was 0.78 (95% CI: 0.71–0.84), indicating substantial interobserver agreement.
Figure 3.
Definition of pulmonary ultrasound score. (A) N, pulmonary sliding sign with A-line or less than 2 separate B-lines. (B) B1-line, the spacing between adjacent B-lines is less than 7 mm. (C) B2-line, the spacing between adjacent B lines is less than 3 mm. (D) C, pleural effusion and the pulmonary present a tissue-like structure with dynamic air bronchogram. Yellow arrows: A-line; Red arrows: B-line; Arrowhead: tissue-like structure; Star: pleural effusion.
Definition of clinical and explanatory variables
Demographic (including sex, age and weight), laboratory {including cystatin-C (Cys-C), creatinine (CREA), creatinine clearance {CCR = [(140−“age”)דweight (kg)”]/[72דserum creatinine (mg/dL)”]×(0.85 for female patients)}, total cholesterol (T-CHO), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), pro brain natriuretic peptide (pro-BNP), creatine kinase (CK), creatine kinase isoenzyme MB (MBCK-MB), lactic dehydrogenase (LD), α-hydroxybutyrate dehydrogenase (α-HBD), high-sensitivity C-reactive protein (hsCRP)}, and clinical data mainly including etiology were retrieved from electronic medical records. And laboratory tests were taken when the patients admitted to the hospital for first time within 24 hours.
Outcomes
Adverse events including respiratory failure, CS, malignant arrhythmia, death, and unplanned readmission within 3 months. And only the first event was counted all of which are important adverse outcomes closely associated with the severity and prognosis of the index disease. These events share a common pathophysiological mechanism (myocardial dysfunction, impaired organ perfusion), and their combination can more comprehensively reflect the overall impact of the intervention on patient outcomes.
Among these adverse events
Respiratory failure: defined as the need for invasive or non-invasive mechanical ventilation (including high-flow nasal cannula with flow rate ≥60 L/min) due to impaired gas exchange, confirmed by arterial blood gas analysis (PaO2/FiO2 <300 mmHg for acute respiratory failure).
CS: defined based on the Third Universal Definition of Myocardial Infarction (2022), characterized by sustained hypotension (systolic blood pressure <90 mmHg for more than 30 minutes) despite adequate fluid resuscitation, accompanied by signs of end-organ hypoperfusion (oliguria, altered mental status, cool extremities).
Malignant arrhythmias: defined as clinically significant arrhythmias requiring urgent intervention (such as cardioversion, defibrillation, antiarrhythmic drug administration), including ventricular tachycardia, ventricular fibrillation, atrial fib.
Death: defined as all-cause mortality confirmed by medical records, death certificates, or official notifications from relevant authorities, with no ambiguity regarding the time and cause of death.
Readmission: defined as unplanned readmission to any department within 3 months after discharge, directly related to all-cause (including worsening cardiac function, recurrent arrhythmia, or complications of the initial treatment).
The model time-zero and outcome horizon. time-zero is defined as the date of definitive LHF diagnosis. The outcome horizon is set at 3 months from time-zero, with the primary outcome being a composite of all-cause mortality and heart failure hospitalization.
Statistical analysis
Statistical analyses were performed at the group-level. And R software (version 4.0) was utilized for all statistical analyses. Student’s t-test or Mann-Whitney test was employed to compare continuous variables. Categorical data were presented as frequencies and percentages and analyzed using either the χ2 test or Fisher’s exact test.
Univariable logistic regression was used among all collected variables (including sex, heart failure etiology, NYHA class, laboratory findings, treatment variables, echocardiographic parameters, etc.) to screen potential risk factors for adverse events in LHF patients. Variables demonstrating statistical significance (P<0.05) in the univariable analysis were included in the stepwise multivariable logistic regression model. The identified independent risk factors were adopted to develop the nomogram.
The “glmnet” package was used to select the best predictive features. And the “rms” package was used to establish numerical model for multivariate logistic regression analysis. The selection of the optimal combination of influencing factors was performed using LASSO regression with ten-fold cross-validation.
A nomogram was created using the “regplot” package. The points for each independent variable were summed to obtain the total points, which were subsequently converted to predicted probabilities. The predictive performance of the nomogram was assessed using the concordance index (C-index), decision curve analysis and the correction curve.
About collinearity, linearity assumptions, or missing data, we calculated variance inflation factors (VIF) for all predictors before modeling. Predictors with VIF >10 were considered highly collinear and were removed or combined (where appropriate) to reduce multicollinearity. For continuous predictors, we assessed the linearity of their relationship with the log-odds of the target variable using partial plots. Where non-linearity was detected, we included quadratic terms or used spline functions to account for the non-linear relationship. We have specified the proportion of missing data for each predictor. For predictors with missing data <5%, we used mean/mode imputation; for those with missing data between 5% and 20%, we used multiple imputation (5 imputed datasets) to preserve statistical power.
Results
Characteristics of the study population
In this study, a total of 329 HF patients including 278 males and 51 females, with an average age of (53.62±15.85) years were enrolled. According to the occurrence of adverse events, they were divided into adverse event group (n=81, among them respiratory failure 21 case, CS 23 case, malignant arrhythmia 8 case, death 2 case, and unplanned readmission within 3 months 27 case) and no adverse event group (n=248).
Including general information, medical history, laboratory examination, echocardiography, lung ultrasound and other information. Univariate analysis showed that weight, sPAP, LAD, LVEDD, LVEF, LV-E/e’, TAPSE, RV-E/e’, the number of B-lines, lung ultrasound score, troponin-I and pro-BNP may be potential risk factors for adverse events in LHF patients (P<0.05). The other factors were considered to have no significant correlation with adverse events (P>0.05). The characteristics of the patients in both groups were outlined in table 1.
Table 1. Analysis of risk factors associated with adverse events in left heart failure patients.
| Variable | Adverse event group (n=81) | No adverse event group (n=248) | P value |
|---|---|---|---|
| Sex | 0.29 | ||
| Male | 66 (81%) | 214 (86%) | |
| Female | 15 (19%) | 34 (14%) | |
| Age (years) | 53.63±2.87 | 53.29±1.43 | 0.91 |
| Weight (kg) | 71.03±0.71 | 67.94±0.45 | <0.05 |
| Etiology | 0.23 | ||
| Coronary heart disease | 29 (36%) | 105 (42%) | |
| Dilated cardiomyopathy | 41 (51%) | 94 (38%) | |
| Alcoholic cardiomyopathy | 6 (7%) | 26 (11%) | |
| Hypertensive heart disease | 5 (6%) | 23 (9%) | |
| sPAP (mmHg) | 60.56±2.33 | 44.95±1.28 | <0.05 |
| PAD (mm) | 28.54±0.41 | 27.95±0.23 | 0.20 |
| LAD (mm) | 53.76±1.03 | 48.36±0.76 | <0.05 |
| LVEDD (mm) | 69.90±1.62 | 66.04±0.79 | <0.05 |
| LVWT (mm) | 9.80±0.33 | 10.00±0.19 | 0.60 |
| LVEF (%) | 28.78±1.37 | 35.00±0.10 | <0.05 |
| LV-E/e’ | 19.24±1.66 | 9.44±0.77 | <0.05 |
| RAD (mm) | 46.19±1.01 | 44.53±0.70 | 0.20 |
| RVEDD (mm) | 42.93±0.98 | 41.38±0.77 | 0.27 |
| RVWT (mm) | 5.32±0.16 | 5.03±0.10 | 0.11 |
| TAPSE (mm) | 13.90±0.58 | 16.74±0.42 | <0.05 |
| RV-E/e’ | 8.13±0.58 | 6.40±0.31 | <0.05 |
| Tei index | 0.79±0.14 | 1.14±0.53 | 0.68 |
| The number of B-lines | 21.27±1.51 | 16.55±0.81 | <0.05 |
| PUS score | 10.51±0.94 | 7.90±0.51 | <0.05 |
| Cys-C (mg/L) | 2.02±0.26 | 1.61±0.13 | 0.12 |
| CREA (μmol/L) | 204.54±37.62 | 153.11±16.73 | 0.15 |
| CCR (ml/min) | 52.16±3.68 | 60.59±2.39 | 0.06 |
| T-CHO (mmol/L) | 4.18±0.16 | 4.26±0.11 | 0.71 |
| TG (mmol/L) | 1.36±0.77 | 1.77±0.12 | 0.052 |
| HDL-C (mmol/L) | 1.00±0.05 | 1.11±0.03 | 0.08 |
| LDL-C (mmol/L) | 2.51±0.13 | 2.52±0.09 | 0.97 |
| Troponin-I (ng/mL) | 1.342±0.396 | 0.211±0.047 | <0.05 |
| Pro-BNP (pg/mL) | 11,719±1793 | 4,891±742 | <0.05 |
| CK (U/L) | 243.20±82.87 | 109.09±11.28 | 0.12 |
| MBCK-MB (U/L) | 21.98±3.76 | 21.14±2.11 | 0.84 |
| LD (U/L) | 254.76±21.99 | 253.13±11.86 | 0.95 |
| α-HBD (U/L) | 186.83±14.46 | 143.59±12.34 | 0.051 |
| hsCRP (mg/L) | 5.01±0.62 | 4.62±0.85 | 0.79 |
Values are expressed as n (%) or mean ± standard deviation. CCR, creatinine clearance; CK, creatine kinase; CREA, creatinine; Cys-C, cystatin-C; HDL-C, high-density lipoprotein cholesterol; hsCRP, high-sensitivity C-reactive protein; LAD, left atrial transverse diameter; LD, lactic dehydrogenase; LDL-C, low-density lipoprotein cholesterol; LV-E/e’, ratio of pulsed Doppler of the mitral valve inflow to tissue Doppler of the lateral mitral annulus; LVEDD, left ventricular end-diastolic diameter; LVEF, left ventricular ejection fraction; LVWT, left ventricular wall thickness; MBCK-MB, creatine kinase isoenzyme MB; PAD, pulmonary artery diameter; pro-BNP, pro brain natriuretic peptide; PUS, pulmonary ultrasound; RAD, right atrium transverse diameter; RV-E/e’, ratio of pulsed Doppler of the tricuspid inflow to tissue Doppler of the lateral tricuspid annulus; RVEDD, right ventricular end-diastolic diameter; RVWT, right ventricular wall thickness; sPAP, systolic pulmonary artery pressure; T-CHO, total cholesterol; TAPSE, tricuspid annular plane systolic excursion; TG, triglyceride; α-HBD, α-hydroxybutyrate dehydrogenase.
Selected predictors
One-way ANOVA was applied to continuous variables and univariate logistic regression for preliminary variable screening, determining the optimal parameterization. According to the result, 12 variables were identified as the optimal parameterization (P<0.05), including weight, sPAP, LAD, LVEDD, LVEF, LV-E/e’, TAPSE, RV-E/e’, The number of B-lines, PUS score, troponin-I, and pro-BNP, shown in Table 1. Out of the 12 variables, four were identified as potential predictive factors, including sPAP, LAD, LV-E/e’, and the number of B-lines. These four variables were then advanced to multivariate logistic regression analysis within the training cohort. Of these factors, sPAP [odds ratio (OR) =15.53, 95% CI: 1.534–157.19, P=0.02], LAD (OR =1.190, 95% CI: 1.031–1.373, P=0.02), LV-E/e’ (OR =1.241, 95% CI: 1.101–1.399, P<0.001), and the number of B-lines (OR =1.591, 95% CI: 1.123-2.254, P=0.009)—found to be independently associated with the occurrence of adverse events in LHF patients. Conversely, the others variables did not independently predict the occurrence of adverse events in LHF. The results were shown in Table 2.
Table 2. Independent risk factors for adverse events in LHF patients.
| Variable | P value | OR value | 95% CI |
|---|---|---|---|
| Weight | 0.16 | 0.974 | 0.937–1.010 |
| sPAP | 0.02 | 15.53 | 1.534–157.19 |
| 0: <40 mmHg | |||
| 1: ≥40 mmHg | |||
| LAD | 0.02 | 1.190 | 1.031–1.373 |
| LVEDD | 0.58 | 1.027 | 0.934–1.129 |
| LVEF | 0.83 | 0.989 | 0.896–1.092 |
| LV-E/e’ | <0.0001 | 1.241 | 1.101–1.399 |
| TAPSE | 0.11 | 0.844 | 0.686–1.037 |
| RV-E/e’ | 0.20 | 1.131 | 0.936–1.366 |
| The number of B-lines | 0.009 | 1.591 | 1.123–2.254 |
| PUS score | 0.11 | 0.504 | 0.290–0.876 |
| Troponin-I | 0.47 | 1.284 | 0.647–2.550 |
| pro-BNP | 0.12 | 1.035 | 1.014–1.357 |
CI, confidence interval; LAD, left atrial transverse diameter; LHF, left heart failure; LV-E/e’, ratio of pulsed Doppler of the mitral valve inflow to tissue Doppler of the lateral mitral annulus; LVEDD, left ventricular end-diastolic diameter; LVEF, left ventricular ejection fraction; OR, odds ratio; pro-BNP, pro brain natriuretic peptide; PUS, pulmonary ultrasound; RV-E/e’, ratio of pulsed Doppler of the tricuspid inflow to tissue Doppler of the lateral tricuspid annulus; sPAP, systolic pulmonary artery pressure; TAPSE, tricuspid annular plane systolic excursion.
Construction and validation of the model
We constructed a nomogram incorporating the four significant variables (sPAP, LAD, LV-E/e’ and the number of B-lines) as components of the total score. The total score can be used to determine the probability of adverse event occurrence in LHF patients by referring to the corresponding probability on the outcome axis (Figure 4). The nomogram shown excellent discrimination efficiency with a C-statistic of 0.888 (Figure 5) and accuracy which of the calibration plot closely aligned with the ideal 45° diagonal line (Figure 6). Clinical decision curve demonstrated the model achieved net benefits when the threshold probability values were between 0.2 and 0.9 (Figure 7), the results suggested the net benefit across these thresholds and the predictive model significantly enhanced risk stratification for adverse event occurrence in LHF patients.
Figure 4.
A nomogram predicting the risk of adverse events in LHF patients. sPAP: 0, <40 mmHg; 1: ≥40 mmHg. LAD, left atrial transverse diameter; LHF, left heart failure; LV-E/e’, ratio of pulsed Doppler of the mitral valve inflow to tissue Doppler of the lateral mitral annulus; sPAP, systolic pulmonary artery pressure.
Figure 5.
ROC curve for evaluating curve nomogram model of the risk of adverse events in LHF patient. LHF, left heart failure; ROC, receiver operating characteristic.
Figure 6.
Calibration curves to evaluate the accuracy of the nomogram model.
Figure 7.
DCA curve analysis of nomogram prediction. DCA, decision curve analysis.
Discussion
The major findings of this study were as follows: (I) assessment of cardiopulmonary function of LHF patients by CPUS parameters, notably the sPAP, LAD, LV-E /e’ and the number of B-lines, emerged as independent influencing factors of adverse event occurrence. (II) The nomogram incorporating these four variables (sPAP, LAD, LV-E/e’ and the number of B-lines) could predict adverse event occurrence risk. (III) The selected predictors are well-established clinical and ultrasound parameters, our study found that the model’s added value lies in its integrated assessment of multiple relevant parameters, which may provide a more comprehensive and unified prediction of adverse events than using individual parameters or traditional tools alone.
LHF is one of the most common diseases in clinic and is associated with a high risk of adverse events. With the increase of left heart filling pressure in LHF patients, pulmonary venous return was blocked, resulting in the increase of pulmonary venous pressure, which lead to the increase of pulmonary artery pressure, right heart pressure load, right heart dysfunction and even death (11). On the other hand, with the increase of left heart filling pressure, the increased pulmonary capillary wedge pressure may cause pulmonary capillary damage, redistribute pulmonary fluid, and PE occur which was closely related to the prognosis of patients (4). Studies had shown that LHF patients with PH had more severe conditions and worse prognosis than those with normal pulmonary artery pressure (12). At present, there is a lack of effective prediction models that can predict the risk of adverse events in LHF patients. By analyzing the general data, CPUS results and laboratory indicators between LHF patients with PH and LHF patients with normal pulmonary pressure, this study comprehensively collected the potential risk factors for adverse events in LHF patients, combined with the case characteristics of our unit. Then independent risk factors for adverse events in LHF patients were identified, and an individualized prediction model for adverse events in LHF patients was established.
This study found that sPAP was an independent factor for adverse events in LHF patients (OR =15.53, P=0.02). Present study had confirmed that PH was an independent risk factor for poor prognosis in patients with chronic HF, and the longer with history of PH, the worse of cardiac function grade and prognosis, while acute HF patients with PH had an increased risk of death (13), which was consistent with the research results of the higher incidence of adverse events in LHF patients in this study. Some scholars guided the treatment of HF patients by monitoring their pulmonary artery pressure, which can reduce the risk of re-hospitalization (14). Moreover, study had shown that HF patients with PH had a significantly increased risk of adverse events, death, and worse prognosis (15).
Echocardiography can dynamically monitor cardiac morphology, structure and function. And our study found that the echocardiography parameters LAD and LV-E/e’ reflecting left ventricular diastolic function were independent risk factors for adverse events in LHF patients. LAD was closely related to left ventricular diastolic dysfunction. and studies had shown that the risk of sudden cardiac death increased by 15% for every 1 standard deviation increase in LAD (16). LV-E/e’ was related to left ventricular filling pressure, LV-E/e’ <8 indicated normal left ventricular filling pressure, LV-E/e’ >14 indicated increased left ventricular filling pressure, and the increase of LV-E/e’ was associated with increased risk of cardiovascular events or death, which was consistent with the results of Desai’s studies (17). The initial factor of PH in LHF patients was the increase of left ventricular filling pressure, which leaded to the progressive increase of left atrial pressure.
LHF patients were prone to cardiogenic PE. With the increase of left heart filling pressure in LHF patients, the increased pulmonary capillary wedge pressure may cause pulmonary capillary damage, redistribute pulmonary fluid, and cause PE due to excessive extravascular pulmonary water accumulation in the interstitial and alveoli (4). Early diagnosis of cardiogenic PE was particularly important to improve the survival rate of patients. PUS had been increasingly recognized by clinicians in the early diagnosis of PE. And study had shown the presence of B-lines was one of the PUS manifestations of PE in LHF patients. And the number of B-lines was significantly correlated with the extravascular lung water index (EVLW index) (18). At present, many clinical studies have confirmed that the number of B-lines in PUS images was significantly correlated with the severity of PE (19). Agricola et al. (20) found that B-line scores were linearly correlated with EVLW, wedge pressure and radioactive pulmonary water scores determined by PiCCO system, providing more reliable information for the diagnosis of PE. Therefore, the number of B-lines, as an independent risk factor for adverse events in LHF patients, which may relate to the occurrence of cardiogenic PE.
In this study, the results of serological examination of LHF patients were also collected. And the results showed that troponin-I and pro-BNP levels in LHF patients with adverse events were higher than those in patients without adverse events. Pro-BNP is mainly secreted by cardiomyocytes, and the expression concentration of pro-BNP can be significantly increased in the process of myocardial cell injury or myocardial cell compliance decline (21). Studies have shown that the increase of pro-BNP may be related to the decrease of mitochondrial activity in atrial cells of patients, leading to increased apoptosis of cardiomyocytes and finally promoting the release of pro-BNP in cells (22). Troponin-I is a specific protein present in myocardial tissue and a marker of myocardial tissue damage and necrosis. It can reflect the degree and range of myocardial injury (23). Therefore, in this study, the levels of pro-BNP and troponin-I in LHF patients with adverse events were higher than those in patients without adverse events, which was considered to be related to atrial and ventricular enlargement, increased apoptosis of myocardial cells, and more serious damage of myocardial cells in these patients. Given the collinearity existed between pro-BNP, troponin-I and several hemodynamic variables reflecting cardiac loading conditions in our dataset. High pro-BNP and troponin-I were largely captured by other included predictors, and adding pro-BNP did not further improve model discrimination (C-index/AUC). On the other hand, from the clinical perspective, both markers are well-known surrogate markers of heart failure severity. Our model aimed to identify incremental risk factors beyond routinely‑used circulating biomarkers. Therefore, we did not enter them into the final multivariable model.
Limitations of the study
While our results are promising, there are a few limitations. Firstly, detailed baseline therapeutic information, including guideline‑directed medical therapy, diuretics, inotropic agents, device implantation, mechanical circulatory support, heart transplantation eligibility and coronary revascularization, was not pre‑specified in our original study design and could not be fully retrieved within the current revision period. Differences in disease management may act as potential confounders and may affect the independent prognostic value of echocardiographic variables. Secondly, RHC was not performed in all participants due to its invasive nature and associated risks. And only very few patients had complete RHC data, thus, we could not validate echocardiographic measurements against the invasive gold standard. The non-invasive estimation of sPAP via TTE has inherent limitations compared with invasive right heart catheterization (RHC), which is the gold standard for PH diagnosis. TTE-derived sPAP is an estimation rather than a direct measurement, and its accuracy may be affected by factors such as TR quality, operator experience, and patient body habitus. Thirdly, the primary endpoint is a composite of clinically heterogeneous events including respiratory failure, CS, malignant arrhythmias, death and heart‑failure‑related readmission. Heterogeneity among component events may influence the interpretation of the composite outcome. Another important limitation is that all participants in the present study were NYHA class III or IV. Accordingly, and its findings cannot be generalized to patients with mild-to-moderate heart failure (NYHA class I-II) or stable outpatients. Finally, the nomogram was established with limited sample size and lack of external validation, only derivation cohort without validation cohort present in the study. There was substantial gender imbalance, with a relatively small number of female patients, which may impair generalizability to female populations. The inclusion of length of stay as a predictor may indeed introduce data leakage, as length of stay is a post-treatment outcome that could be influenced by the target variable. Therefore, further studies were needed to validate the accuracy and applicability of this predictive model.
Conclusions
In summary, our exploratory study suggests that CPUS may provide valuable prognostic information in patients with LHF. The constructed prediction model for adverse events, incorporating CPUS parameters including sPAP, LAD, LV-E/e’, and number of B lines, demonstrates favorable discriminative ability and accuracy within the study cohort.
These preliminary findings may offer supportive evidence for risk stratification and clinical decision-making. However, further external validation and prospective studies are warranted before the model can be widely implemented in routine clinical practice.
Supplementary
The article’s supplementary files as
Acknowledgments
The authors would like to express their gratitude to the staff at the Department of Ultrasound, Guangxi Academy of Medical Sciences &The People’s Hospital of Guangxi Zhuang Autonomous Region, and at the Department of Ultrasound, The First Affiliated Hospital of Guangxi Medical University for their help.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The People’s Hospital of Guangxi Zhuang Autonomous Region and all subjects gave written informed consent to participate in the study. The First Affiliated Hospital of Guangxi Medical University was also informed of and agreed to the study.
Footnotes
Reporting Checklist: The authors have completed the TREND reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-2226/rc
Funding: This study was financially supported by The People’s Hospital of Guangxi Zhuang Autonomous Region 2025 National Nature Hospital cultivation fund (No. 2025GPY0117) and Guangxi Key Research and Development Program (No. AB23026042).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-2226/coif). The authors have no conflicts of interest to declare.
Data Sharing Statement
Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-2226/dss
References
- 1. Costanzo MR. The Cardiorenal Syndrome in Heart Failure. Cardiol Clin 2022;40:219-35.. 10.1016/j.ccl.2021.12.010 [DOI] [PubMed] [Google Scholar]
- 2. Global Burden of Disease Study 2013 Collaborators . Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 2015;386:743-800.. 10.1016/S0140-6736(15)60692-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Apostolo A, Vignati C, Cittar M, et al. Determinants of exercise performance in heart failure patients with extremely reduced cardiac output and left ventricular assist device. Eur J Prev Cardiol 2023;30:ii63-9.. 10.1093/eurjpc/zwad239 [DOI] [PubMed] [Google Scholar]
- 4. Keskinis C, Panagoutsos S, Kyriklidou P, et al. Scanning more corresponds to more accuracy in hemodialysis patients: 28-zone protocol's superior findings from an observational study. J Ultrasound 2025;28:89-99.. 10.1007/s40477-024-00964-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Degefu N, Jambo A, Nigusse S, et al. The Burden and Contributing Factors of Cardiogenic Pulmonary Edema Among Acute Heart Failure Patients Admitted to Tertiary Hospital, Eastern Ethiopia. Open Access Emerg Med 2023;15:405-14.. 10.2147/OAEM.S436352 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Öhman J, Harjola VP, Karjalainen P, et al. Focused echocardiography and lung ultrasound protocol for guiding treatment in acute heart failure. ESC Heart Fail 2018;5:120-8.. 10.1002/ehf2.12208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Ruben M, Molinas MS, Paladini H, et al. Emerging concepts in heart failure management and treatment: focus on point-of-care ultrasound in cardiogenic shock. Drugs Context 2023;12:2022-5-8.. 10.7573/dic.2022-5-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Čelutkienė J, Lainscak M, Anderson L, et al. Imaging in patients with suspected acute heart failure: timeline approach position statement on behalf of the Heart Failure Association of the European Society of Cardiology. Eur J Heart Fail 2020;22:181-95.. 10.1002/ejhf.1678 [DOI] [PubMed] [Google Scholar]
- 9. Caraballo C, Desai NR, Mulder H, et al. Clinical Implications of the New York Heart Association Classification. J Am Heart Assoc 2019;8:e014240.. 10.1161/JAHA.119.014240 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Recommendations for Cardiac Chamber Quantification by Echocardiography in Adults: An Update from the American Society of Echocardiography and the European Association of, Cardiovascular Imaging. Eur Heart J Cardiovasc Imaging 2016;17:412.. 10.1093/ehjci/jew041 [DOI] [PubMed] [Google Scholar]
- 11. Zhu YJ, Kou JJ, Bo YC, et al. Amonafide Targeting NTSR1-PI3K/AKT/mTOR Signaling Attenuates Vascular Remodeling in Pulmonary Arterial Hypertension. J Am Heart Assoc 2026;15:e045580.. 10.1161/JAHA.125.045580 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. de Jesus Perez VA, Lai YC. TAZ and RUNX2 awareness in pulmonary hypertension due to left heart disease. Eur Respir J 2024;64:2400905.. 10.1183/13993003.00905-2024 [DOI] [PubMed] [Google Scholar]
- 13. Omote K, Sorimachi H, Obokata M, et al. Pulmonary vascular disease in pulmonary hypertension due to left heart disease: pathophysiologic implications. Eur Heart J 2022;43:3417-31.. 10.1093/eurheartj/ehac184 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Guichard JL, Bonno EL, Nassif ME, et al. Seated Pulmonary Artery Pressure Monitoring in Patients With Heart Failure: Results of the PROACTIVE-HF Trial. JACC Heart Fail 2024;12:1879-93.. 10.1016/j.jchf.2024.05.017 [DOI] [PubMed] [Google Scholar]
- 15. Gomez Rojas O, Mitchell S, Peterson JA, et al. Real-World Performance of Sotatercept in PAH Patients With Cardiopulmonary Comorbidities: A Retrospective Single Center Experience. Pulm Circ 2025;15:e70226.. 10.1002/pul2.70226 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Konety SH, Koene RJ, Norby FL, et al. Echocardiographic Predictors of Sudden Cardiac Death: The Atherosclerosis Risk in Communities Study and Cardiovascular Health Study. Circ Cardiovasc Imaging 2016;9:10.1161/CIRCIMAGING.115.004431 e004431.. 10.1161/CIRCIMAGING.115.004431 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Desai CS, Colangelo LA, Liu K, et al. Prevalence, prospective risk markers, and prognosis associated with the presence of left ventricular diastolic dysfunction in young adults: the coronary artery risk development in young adults study. Am J Epidemiol 2013;177:20-32.. 10.1093/aje/kws224 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Pramudyo M, Kamarullah W, Pranata R, et al. Low-dose versus high-dose intravenous nitroglycerin in the treatment of sympathetic crashing acute pulmonary oedema: a systematic review and meta-analysis focusing on efficacy, safety and outcomes. BMJ Open 2025;15:e099142.. 10.1136/bmjopen-2025-099142 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Hegemann N, Sang P, Kim JH, et al. Ultrasonographic assessment of pulmonary and Central venous congestion in experimental heart failure. Am J Physiol Heart Circ Physiol 2024;326:H433-40.. 10.1152/ajpheart.00735.2023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Agricola E, Bove T, Oppizzi M, et al. "Ultrasound comet-tail images": a marker of pulmonary edema: a comparative study with wedge pressure and extravascular lung water. Chest 2005;127:1690-5.. 10.1378/chest.127.5.1690 [DOI] [PubMed] [Google Scholar]
- 21. Cordero A, Martínez Rey-Rañal E, Moreno MJ, et al. Predictive Value of Pro-BNP for Heart Failure Readmission after an Acute Coronary Syndrome. J Clin Med 2021;10:1653.. 10.3390/jcm10081653 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. . [Google Scholar]
- 23. Empana JP, Lerner I, Perier MC, et al. Ultrasensitive Troponin I and Incident Cardiovascular Disease. Arterioscler Thromb Vasc Biol 2022;42:1471-81.. 10.1161/ATVBAHA.122.317961 [DOI] [PubMed] [Google Scholar]