The BNP-to-Albumin Ratio (BAR) as a Predictor of All-Cause Mortality in Elderly Patients with Atrial Fibrillation.

Introduction

Atrial fibrillation (AF) is the most common type of arrhythmia in clinical practice and is linked to a range of adverse consequences, including a fivefold increase in stroke incidence, left ventricular dysfunction, and heart failure. 1 , 2 Additionally, it plays a role in cognitive decline, vascular dementia, depression, and reduced quality of life, 1 , 2 making it a significant factor in cardiovascular morbidity and mortality. 3 , 4 Its incidence has increased with the increasing aging population, resulting in a substantial disease burden. 3 , 5 , 6 The onset and progression of AF involve a series of “upstream pathways”, including inflammation, oxidative stress, fibrosis, and metabolic disorders. 1 Atrial fibrosis is one of the key pathological foundations for the occurrence of AF. 7 Preventing atrial fibrosis is considered a potential therapeutic target for the development of AF and its associated complications. 8 Current therapeutic strategies for AF are broadly divided into rhythm control approaches (such as catheter ablation and antiarrhythmic drugs) and strategies targeting complications and pathophysiology, including anticoagulation, left atrial appendage occlusion, and upstream therapies against the renin-angiotensin-aldosterone system, oxidative stress, inflammation, and fibrosis. 1 Nevertheless, the risk of all-cause mortality in patients with AF may be twice that of the general population. 3 Therefore, how to improve the long-term prognosis in these patients is worth considering.

B-type natriuretic peptide (BNP) is a serum cardiac biomarker and a multifunctional hormone that, 7 , 9 in addition to its typical effects on sodium excretion, vasodilation and antagonism of the renin-angiotensin-aldosterone system, 9 has recently been found to inhibit fibroblast proliferation and extracellular matrix deposition in cardiac tissue. 10 In addition, it participates in systemic metabolism by activating lipolysis and promoting lipid oxidation. 10 Compared with clinical risk scores, natriuretic peptides themselves demonstrate superior prognostic discriminatory ability for cardiovascular events and progression to AF. 11

Albumin is the major protein in blood and is considered to be the most important antioxidant in whole blood, 12 , 13 with several important physiological functions: maintaining colloid osmotic pressure and vascular integrity, regulating metabolism and vascular activity, providing substance-binding ligands and exhibiting anticoagulant and antiplatelet aggregation activity. 13 , 14 Hypoproteinemia may indicate the presence of an underlying inflammatory state, 12 which may accelerate cardiac and organ failure, thereby increasing the risk of adverse cardiovascular outcomes and reducing survival. 15 There is evidence that lower serum albumin levels in older adults are associated with a greater risk of new-onset cardiovascular disease. 15

Previous studies have suggested that the BAR—a composite indicator comprising BNP, which reflects cardiac neurohormonal activation, 16 and albumin, which reflects inflammatory-metabolic and vascular homeostasis 12 , 14 , 17 —combines two independent predictors of AF severity, and that this ratio is associated with stroke risk in patients with AF. 18 Given the significant heterogeneity in the elderly population regarding physiological reserve, comorbidities, and geriatric syndromes, 19 the prognostic value of this ratio in elderly patients with AF has not yet been specifically evaluated. Therefore, this study focuses on the elderly population and aims to assess the association between BAR and long-term prognosis in elderly patients with AF.

Methods

Study Population

A retrospective study was performed on 1458 consecutive patients aged sixty years or older with AF who were admitted to the Heart Center and the Department of Geriatric Cardiology of the First Affiliated Hospital of Kunming Medical University between December 2019 and December 2023. The exclusion criteria were as follows: age ≥ 90 years; absence of necessary data, such as BNP or albumin levels; lack of follow-up data (loss of contact or refusal to cooperate during telephone follow-up); active bleeding; malignancy, including hematologic malignancies such as leukemia, multiple myeloma, and lymphoma; valvular heart disease-associated AF; severe infection (sepsis, septic shock, or bacteremia); severe anemia (hemoglobin < 60 g/L); uremia or dialysis dependence; decompensated cirrhosis; or in-hospital death. By excluding these extremely high-risk patients, we confined our analysis to those with a relatively low-risk profile and a modifiable disease window, thereby enabling the accurate detection of early and specific prognostic markers for AF. 20 After applying these criteria, 616 patients were excluded, and a total of 842 patients were included in the final analysis. Detailed information on the study population, including subgroup distribution and comparisons of baseline characteristics between patients with and without missing data, is provided in the supplementary tables ( Supplementary Tables 1 , 3 and 4 ). This retrospective cohort study was reported in accordance with the STROBE statement and the RECORD guideline 21 ( Supplementary Table 7 ).

Data Collection

Patient demographic and clinical information was collected at admission. The demographic characteristics included age, sex, history of smoking, alcohol consumption, and family history of cardiovascular disease. Anthropometric data included pulse and heart rate, weight, height, body mass index (BMI), systolic blood pressure (SBP) and diastolic blood pressure (DBP). Clinical history included hypertension, diabetes mellitus, coronary artery disease, chronic obstructive pulmonary disease (COPD), heart failure, peripheral vascular disease (PVD), a history of stroke or transient ischemic attack (TIA), hyperthyroidism, hypothyroidism, cardiomyopathy, and chronic kidney disease. All laboratory data, including results of routine blood tests including white blood cell (WBC), lymphocyte percentage (LY%), red blood cell (RBC),platelet (PLT), neutrophil, monocyte, creatine kinase-MB (CK-MB), cardiac troponin (cTn), myoglobin, B-type natriuretic peptide (BNP), high-sensitivity C-reactive protein (hs-CRP), total protein (TP), albumin, total bilirubin (STB), conjugated bilirubin, cholinesterase (ChE), uric acid (UA), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), serum potassium, serum sodium, estimated glomerular filtration rate (eGFR), thyroid function tests, and antibody tests, were used. All blood samples were based on the most recent plain blood before and after admission, and emergency blood was used if unavailable and then sent to the laboratory of the First Affiliated Hospital of Kunming Medical University for testing according to standard operating procedures.

Adjunctive examinations included cardiac ultrasound data such as left atrial diameter (LAD), left ventricular end-diastolic diameter (LVDd), right atrial long‑axis diameter (RAd), interventricular septal thickness, left ventricular posterior wall thickness, left ventricular end-systolic volume (LVESV), left ventricular ejection fraction (LVEF), pulmonary artery systolic pressure (PASP), and other data.

Pharmacological and interventional treatments during hospitalization included warfarin, new oral anticoagulants, aspirin or indobufen, clopidogrel or ticagrelor, statins, β-blockers, digoxin, diuretics, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker/angiotensin receptor-neprilysin inhibitor (ACEI/ARB/ARNI), calcium channel blockers (CCBs), amiodarone/dexamethasone, and SGLT2 inhibitors, as well as atrial fibrillation ablation (AF ablation), left atrial appendage occlusion(LAAO), and pacemaker implantation. The CHA 2 DS 2 -VASc and HAS-BLED scores were also recorded. BAR was calculated by dividing the log10-transformed BNP (multiplied by 100) by the albumin level.

Follow-Up and Outcomes

Survival data were collected through telephone follow-up with patients or their family members. Lost to follow-up was defined as failure to establish contact with both the patient and at least one family member after three or more unsuccessful attempts on separate days. Time-to-event was calculated from the date of discharge (index date) to the date of all-cause mortality, with censoring at the last known alive date. The study endpoint was defined as all-cause mortality (ACM).

Statistical Analysis

Patients were categorized into a low-BAR group (BAR < 6.05; n = 419) and a high-BAR group (BAR ≥ 6.05; n = 423) based on optimal cutoff values determined by restricted cubic splines (RCSs). Alternative grouping strategies (tertiles and quartiles) were explored in supplementary analyses to further characterize the threshold effect.( Supplementary Table 6 ). Normally distributed continuous variables are expressed as the mean (standard deviation), and comparisons between groups were performed using the t test. Continuous variables that were not normally distributed are expressed as medians (P25, P75), and comparisons between groups were made using the Mann‒Whitney U -test. Categorical variables are expressed as frequencies and percentages, and comparisons between groups were made using the chi-square test.

The Kaplan‒Meier method was used to estimate the cumulative incidence of ACM. We first carried out univariable analyses to identify potential prognostic candidates. Given the limited number of outcome events (137 deaths) and the availability of over 20 candidate predictors, we applied a clinically-oriented hierarchical variable-selection strategy to construct the final multivariable Cox model. This approach prioritized well-validated baseline predictors, including demographic features, routine laboratory measurements, cardiac structural-functional parameters, established thromboembolic-bleeding risk scores, and key interventional covariates. 22–25 Univariate and multivariate Cox proportional hazards regression analyses were fitted for ACM, with adjustment for sex, age, BMI, heart failure, CHA 2 DS 2 ‑VASc score, HAS‑BLED score, AFCA, LAAO, LAd, LVEF, platelet count, serum sodium, and eGFR. The proportional hazards assumption was verified using Schoenfeld residuals and met for all covariates (global test: Chisq = 9.774, df = 14, P = 0.779) ( Supplementary Table 5 ). Missing data were addressed via three sensitivity analyses: multiple imputation for missing covariates (m = 10, pooled by median), inverse probability weighting for missing BAR, and an extreme bound analysis assuming immediate death for lost‑to‑follow‑up cases (with worst‑case BAR assignment), as detailed in Supplementary Table 2 . To further evaluate the robustness of the association between the BAR and ACM, hierarchical multivariate models were constructed with incremental adjustment for confounders, using the unadjusted model as the reference. The results are expressed as hazard ratios (HRs) and 95% confidence intervals (CIs). The predictive value of the BAR for ACM was quantified using receiver operating characteristic (ROC) curves. The area under the ROC curve (AUC) and 95% confidence interval were calculated, and comparisons were made among different biomarkers. Model 1 (M1) was adjusted for age and sex. Model 2 (M2) was further adjusted for Lad, LVEF, BMI, Heart failure, Platelet, Sodium, and eGFR. Model 3 was additionally adjusted for the BAR based on Model2. Using a calibration curve generated via bootstrap resampling (B = 200), we evaluated the calibration of the full model (including BAR) and the clinical model (excluding BAR). The full model included Age, Sex, BMI, heart failure, LAd, LVEF, Platelet count, serum sodium, eGFR, and BAR; the clinical model included the same variables except BAR. Model performance was assessed using the C-statistic and the Brier score. The Integrated Discrimination Improvement (IDI) and Net Reclassification Improvement (NRI) were used to evaluate the incremental predictive value of BAR beyond BNP. The reference model included Age, Sex, BMI, heart failure, LAd, LVEF, Platelet count, serum sodium, eGFR, and BNP; the full model additionally included BAR. All the analyses were conducted using IBM SPSS Statistics 31.0 and R 4.6.1. A two-sided p-value < 0.05 was considered to indicate statistical significance.

Results

Baseline Patient Characteristics

This study ultimately included 842 patients with AF. The mean age of the study participants was 72.67 (7.73) years, with males accounting for 55.1% of the cohort. The median follow-up duration, calculated retrospectively from the date of discharge to the date of death or last censoring, was 916 days (IQR, 706–1325 days), during which 137 deaths (16.3%) were recorded. Compared with those in the low-BAR group, patients in the high-BAR group were significantly older and exhibited elevated heart rates (HRs) as well as increased levels of troponin, LgBNP, hs-CRP, total bilirubin, conjugated bilirubin, UA, and serum potassium. Echocardiographic assessments revealed significantly increased LAd, LVDd, RAd and LVESV (all P < 0.05). Within the CHA 2 DS 2 -VASc score range (1–9 points), a significantly greater proportion of patients in the high-BAR group were classified into sex-specific high-risk categories according to guideline definitions; they also exhibited higher prevalence rates of heart failure and cardiomyopathy, along with higher prescribed doses of beta-blockers (all P < 0.05). Conversely, the high-BAR group had lower SBP, BMI, LY%, RBC, hemoglobin, total protein, albumin, cholinesterase, total cholesterol, LDL-C, eGFR, and LVEF and lower rates of prior AF ablation and LAAO history (all P < 0.05). The median follow-up duration was longer in the high-BAR group (P < 0.05). ( Table 1 ). Table 1 Baselines Characteristics According to BAR Variables Total (n=842) Low BAR (n=419) High BAR (n=423) p value Basic characteristics Follow-up time, days 916 (706, 1325) 997 (763, 1353) 1136 (824, 1538) <0.001 Age, years 72.67 (7.73) 71.00 (7.20) 74.31 (7.88) <0.001 Sex, male, n (%) 464 (55.1%) 243(58.0%) 221 (52.2%) 0.094 HR (beats/minute) 89.08 (22.15) 83.75 (19.13) 94.45 (23.66) <0.001 Systolic BP, mmHg 126.2 (21.0) 128.2(20.5) 124.2 (21.4) 0.004 Diastolic BP, mmHg 78.6 (14.0) 78.01 (13.33) 79.26 (14.54) 0.195 BMI (kg/m2) 24.23 (3.79) 24.78 (3.64) 23.68 (3.85) <0.001 Medical history, n (%) Hypertension 597 (70.9%) 302 (72.1%) 295 (69.7%) 0.455 Diabetes 236 (28.0%) 127 (30.3%) 109 (25.8%) 0.142 Coronary heart disease (Y) 326 (38.7%) 160 (38.2%) 166 (39.2%) 0.753 Heart failure (Y) 358 (42.5%) 98 (23.4%) 260 (61.5%) <0.001 PVD 423 (50.2%) 203 (48.4%) 220 (52.0%) 0.301 TIA 171 (20.3%) 77 (18.4%) 94 (22.3%) 0.160 Cardiomyopathy 93 (11.0%) 25 (6.0%) 68 (16.1%) <0.001 Smoking Status 209 (24.8%) 104 (24.8%) 105 (24.8%) 1.000 Alcohol Consumption 99 (11.8%) 53 (12.6%) 46 (10.9%) 0.424 Laboratory data Troponin (ng/mL) 0.15 (0.007, 0.502) 0.01 (0.006, 0.019) 0.03 (0.013, 0.113) <0.001 LgBNP, ng/mL 2.38 (0.57) 1.96 (0.41) 2.80 (0.38) <0.001 WBC (10^9/L) 6.42 (2.03) 6.37 (1.97) 6.47 (2.08) 0.467 LY% 26.4 (19.75, 32.9) 29 (23.1, 35.5) 23.7 (16.8, 30.4) <0.001 RBC (10^9/L) 4.62 (0.61) 4.73 (0.57) 4.52 (0.64) <0.001 Hemoglobin (Hb), g/L 141.77(19.1) 144.88 (17.51) 138.82 (20.24) <0.001 Platelet (10^9/L) 185.0 (61.23) 188.578 (61.49) 181.44 (60.77) 0.091 hsCRP, mg/L 2.20 (0.8, 6.04) 1.2 (0.5, 3.2) 3.685 (1.3, 11.5) <0.001 Total Protein (TP) (g/L) 68.40 (6.55) 70.33 (6.43) 66.45 (6.08) <0.001 Albumin (g/L) 39.4 (36.5, 41.8) 41.44 (3.45) 36.85 (3.74) <0.001 Total bilirubin (umolL) 13.4 (10.0, 18.25) 12.5 (9.2, 17.1) 14.4 (10.5, 20) <0.001 CB (umol/L) 6.2 (4.5, 8.4) 5.5 (4.2, 7.3) 7.0 (5, 10.1) <0.001 Cholinesterase (mmol/L) 6.7 (5.5, 7.9) 7.6 (6.5, 8.7) 5.8 (4.8, 6.9) <0.001 UA, umol/L 424.30 (135.67) 393.38 (108.43) 455.07 (152.11) <0.001 Total Cholesterol, mmolL 3.68 (3.06 4.42) 3.85 (3.18, 4.62) 3.53 (2.96, 4.23) <0.001 HDLC, mmolL 1.10 (0.91, 1.3) 1.10 (0.93, 1.31) 1.09 (0.89, 1.27) 0.092 LDLC, mmolL 2.1 (1.51, 2.73) 2.19 (1.57, 2.89) 2.02 (1.44, 2.55) <0.001 Potassium, mmolL 3.87 (0.52) 3.83 (0.50) 3.91 (0.53) 0.029 Sodium, mmolL 144.51 (46.11) 145.77 (64.97) 143.26 (7.86) 0.432 eGFR, mL/min 54.95 (42.24, 69.12) 61.63 (49.18, 74.81) 48.96 (35.69, 61.53) <0.001 Echocardiography LAd, mm 42.11 (8.37) 39.63 (7.83) 44.61 (8.15) <0.001 LVDd, mm 48.19 (8.35) 46.47 (6.93) 49.90 (9.25) <0.001 RAd, mm 50.02 (11.61) 48.80 (10.45) 53.31 (12.27) <0.001 LVESV, mL 46 (33, 65) 40 (32, 56.75) 53 (36, 76.75) <0.001 LVEF, % 60 (51, 69.25) 65 (56, 71) 55 (46, 66) <0.001 Treatment modalities, n (%) Warfarin 65 (7.7%) 25 (6.0%) 40 (9.5%) 0.058 NOACs 632 (75.1%) 318 (75.9%) 314 (74.2%) 0.577 Aspirin or Indobufen 139 (16.4%) 78 (18.6%) 60 (14.2%) 0.082 Clopidogrel or Ticagrelor 171 (20.3%) 87 (20.8%) 84 (19.9%) 0.744 Statins 644 (76.5%) 331 (79.0%) 313 (74.0%) 0.087 β-blockers 585 (69.5%) 255 (60.9%) 330 (78.0%) <0.001 ACEI/ARB/ARNI 475 (56.4%) 229 (54.7%) 246 (58.2%) 0.306 Amiodarone or Dronedarone 326 (38.7%) 165 (39.4%) 161 (38.1%) 0.695 AF Ablation 142 (16.9%) 103 (24.6%) 39 (9.2%) <0.001 History of LAAO 96 (11.4%) 60 (14.3%) 36 (8.5%) 0.008 CHA 2 DS 2 VASc score Low-risk 74 (8.8%) 50 (11.9%) 24 (5.7%) 0.001 High-risk 768 (91.2%) 369 (88.1%) 399 (94.3%) 0.001 HASBLED score 2 (1, 3) 2 (1, 2) 2 (1, 3) <0.001 Notes : Data are presented as mean (SD) for normally distributed continuous variables, median (IQR) for non-normally distributed continuous variables, and n (%) for categorical variables. Abbreviations : HR heart rate; BP blood pressure; BMI body mass index; PVD Peripheral Vascular Disease; TIA History of Stroke or History of Transient Ischemic Attack; LgBNP log-transformed B-type natriuretic peptide; WBC White Blood Cell Count; LY% Lymphocyte Percentage; RBC red blood cell; hsCRP high-sensitivity C-reactive protein;CB Conjugated Bilirubin; UA Uric acid; HDLC High-Density Lipoprotein Cholesterol; LDLC Low-Density Lipoprotein Cholesterol; eGFR Glomerular Filtration Rate; LAd Left Atrial Diameter; LVDd left ventricular end-diastolic diameter; RAd Right Atrial Diameter; LVESV Left Ventricular End-Systolic Volume; LVEF left ventricular ejection fraction; NOACs Novel Oral Anticoagulants; AF Ablation Atrial Fibrillation Ablation; history of LAAO History of Left Atrial Appendage Occlusion.

Associations Between the BAR and Endpoints

As shown in Figure 1 , Kaplan‒Meier analysis demonstrated that ACM was significantly greater in AF patients with higher BAR levels than in those with those with lower BAR levels (ACM: log-rank χ2 80.677; P < 0.001). As shown in Figure 2 , the relationship between the BAR and ACM in AF patients was flexibly modeled and visualized using restricted cubic spline regression. This analysis revealed a significant nonlinear association between the BAR and ACM (P for overall < 0.001; P for nonlinear = 0.0024). Overall, the BAR was positively associated with the risk of ACM, and the curve exhibited a clear threshold effect: the risk of death remained relatively stable below the threshold but increased sharply once the threshold was surpassed. Figure 1 KaplanMeier survival curves all-cause mortality on the basis of BAR levels. Kaplan Meier curve of all cause mortality. The horizontal axis label is, Survival time (day), ranging from 0 to 2000. The vertical axis label is, Survival probability, ranging from 0.00 to 1.00. Legend text: BAR group (low BAR less than 6.05, high BAR greater than or equal to 6.05), with series labels, low BAR and high BAR. Log rank test text: chi squared (1) equals 80.677; P less than 0.001. Number at risk table aligned to the survival time axis shows: low BAR, 419, 407, 200, 69, 1; high BAR, 423, 355, 181, 70, 6. A Kaplan Meier survival curve showing all cause mortality by BAR group. Figure 2 Restricted cubic spline analysis for the associations between the BAR and all-cause mortality. A line graph showing restricted cubic spline of all cause mortality versus BAR. Note : Hazard ratios are indicated by solid lines.

A High BAR as an Independent Predictor of All-Cause Mortality

Univariate and multivariate Cox regression analyses were performed to identify predictors of ACM ( Table 2 ). In univariate analysis, when the low-BAR group was used as the reference, the high-BAR group was associated with a significantly higher risk of ACM (HR: 6.785; 95% CI: 4.197–11.015; p < 0.001). Multivariate Cox regression analysis confirmed that the BAR was an independent predictor of ACM (HR: 4.430; 95% CI: 2.460–7.980; p < 0.001). Table 2 Univariate and Multivariate Analysis Using Cox Proportional Hazard Analysis of Variables Associated with All-Cause Mortality in Patients with Atrial Fibrillation Variable Unadjusted HR (95% CI) P value Adjust P value Low BAR Reference Reference High BAR 6.785 (4.179–11.015) <0.001 4.430(2.460–7.980) <0.001 Sex (male) 1.223 (0.936–1.598) 0.140 Age 1.090 (1.071–1.110) <0.001 1.041 (1.012–1.071) 0.005 BMI 0.895 (0.859–0.931) <0.001 0.923 (0.873–0.976) 0.005 Heart failure 3.039 (2.317–3.986) <0.001 1.568 (1.021–2.409) 0.040 CHA2DS2-VASc score 1.252(1.165–1.346) <0.001 1.144(1.001–1.308) 0.049 HAS-BLED score 1.369 (1.253–1.496) <0.001 AFCA 0.231(0.132–0.405 <0.001 LAAO 0.272 (0.134 0.551) <0.001 0.247 (0.076–0.803) 0.020 LAd 1.037 (1.022–1.053) <0.001 LVEF 0.967 (0.958–0.977) <0.001 0.984 (0.971–0.998) 0.020 Platelet 0.996 (0.994–0.999) 0.003 Sodium 0.987 (0.979–0.994) <0.001 0.982(0.970–0.995) 0.006 eGFR 0.966 (0.959–0.973) <0.001 Note : Low BAR < 6.05, high BAR ≥ 6.05. Abbreviations : CI, confidence interval; BMI body mass index; AFCA Atrial fibrillation catheter ablation; LAAO Left atrial appendage occlusion; LAd, left atrial diameter; LVESV, Left Ventricular End-Systolic Volume; LVEF, left ventricular ejection fraction; eGFR, glomerular filtration rate; BAR, Log-transformed BNP*100/Albumin.

The four multivariate Cox proportional hazards models used to determine the association between the BAR groups and prognosis are presented in Table 3 . In Model 4, when the low-BAR group was used as the reference, adjustments were made for age, sex, BMI, Heart failure, CHA 2 DS 2 -VASc score, HAS-BLED score, AF ablation, LAAO, LAd, LVEF, and eGFR, and high BAR remained independently associated with a significantly increased risk of ACM (HR: 3.639; 95% CI: 2.115–6.260; p < 0.001). Table 3 Predictive Value of BAR for ACM in Cox Proportional Hazards Models All-Cause Mortality HR (95% CI) P Unadjusted 6.785 (4.179–11.015) <0.001 Model 1 5.582 (3.422–9.105) <0.001 Model 2 5.647 (3.461–9.213) <0.001 Model 3 4.067(2.447–6.759) <0.001 Model 4 3.639(2.115–6.260) <0.001 Notes : Reference: low. Low BAR < 6.05, high BAR ≥ 6.05 Model 1: Adjusted for Age. Model 2: Adjusted for Age and Sex. Model 3: Adjusted for variables included in model 2 and BMI, Heart failure. Model 4: Adjusted for variables included in model 3 and CHA2DS2-VASc score,HASBLED score,Atrial fibrillation catheter ablation,Left atrial appendage occlusion,LAd,LVEF,eGFR. Abbreviations : CI, confidence interval; HR, hazard ratio; BAR,Log-transformed BNP*100/Albumin.

Predictive Ability of the BAR in Patients with AF

ROC analysis revealed that the AUC of the BAR for predicting ACM in patients with AF was 0.778 (95% CI: 0.737–0.819), with a sensitivity of 82.5% and a specificity of 63.8%. The BAR had significantly better predictive performance than Lg-BNP (AUC = 0.731; 95% CI: 0.688–0.774; P < 0.001) and albumin (AUC = 0.728; 95% CI: 0.68–0.776; P = 0.009) ( Figure 3 ). Figure 3 ROC curves of the BAR with the reference line for all-cause mortality. A line graph showing receiver operating characteristic curves for all cause mortality.

The discriminative ability of the three models for predicting long-term mortality was evaluated by AUC analysis. Model 3 (M3)(AUC = 0.838; 95% CI: 0.803–0.873) had significantly better predictive accuracy for ACM than Model 1 (AUC = 0.705; 95% CI: 0.667–0.742; P < 0.001) and Model 2 (AUC = 0.788; 95% CI: 0.757–0.820; P < 0.05) ( Figure 4 ). Figure 4 The ROC curves according to the three models predictive value for all patients. M1: Model 1 was adjusted for age and sex. M2: Model 2 was further adjusted for BMI, Heart failure, Lad, LVEF, Platelet, Sodium, and eGFR. M3: Model 3 adjusted for variables included in model2 and the BAR. Lad, left atrial diameter; LVEF,left ventricular ejection fraction. ROC curves for all-cause Mortality Prediction. Line graph with three receiver operating characteristic curves and a diagonal reference line. The x-axis label is 1 minus Specificity, unit not shown, ranging from 0.00 to 1.00 with labeled ticks at 0.00, 0.25, 0.50, 0.75, 1.00. The y-axis label is Sensitivity, unit not shown, ranging from 0.00 to 1.00 with labeled ticks at 0.00, 0.25, 0.50, 0.75, 1.00. Three curves are labeled in the legend: M1 with AUC = 0.705(95% CI:0.667–0.742); M2 with AUC = 0.788 (95% CI:0.757–0.820); M3 with AUC = 0.838(95% CI:0.803–0.873). Across the plot, M3 stays above M2 and M2 stays above M1, while all three rise from near 0.00 sensitivity at 0.00 on the x-axis toward near 1.00 sensitivity at 1.00 on the x-axis. Text block lists DeLong test p-values: M1 vs M2 p less than 0.001; M1 vs M3 p less than 0.001; M2 vs M3 p equals 0.0011. A line graph of receiver operating characteristic curves comparing three models, with Model 3 highest curve.

The calibration curves indicate that both models demonstrate good calibration, with the full model having a slightly higher C-statistic (0.8222) compared with the clinical model (0.7806), and a lower Brier score (0.1113) than the clinical model (0.1176)( Figure 5 ). Figure 5 Calibration curves of the full model (with BAR) and the clinical model (without BAR). Calibration Curves for Full and Clinical Models. Single line graph with three series and a text box. Legend text: Full model (with BAR). Clinical model (without BAR). Ideal calibration. Additional text: N equals 842, Events equals 137. Performance (corrected): Full model: C-statistic: 0.8222, Brier Score: 0.1113. Clinical model: C-statistic: 0.7806, Brier Score: 0.1176. Delta C-statistic: minus 0.0416. Delta Brier Score: 0.0063. Horizontal axis label: Predicted Probability. Range shown: 0.0 to 0.7. Vertical axis label: Observed Probability. Range shown: 0.0 to 0.6. Plotted series: Ideal calibration dashed diagonal line from approximately (0.0, 0.0) to approximately (0.7, 0.7). Full model (with BAR) curve with circular markers, rising from near (0.0, 0.0) to near (0.7, 0.58). Clinical model (without BAR) curve with triangular markers, rising from near (0.0, 0.0) to near (0.7, 0.62). Readable coordinate pairs are not explicitly labeled for the model curves beyond the axis tick ranges. A line graph showing calibration curves for full and clinical models and an ideal calibration reference.

To further evaluate the incremental predictive value of BAR beyond BNP, the IDI and NRI were calculated. After incorporating BAR into the BNP model, risk reclassification performance improved significantly (IDI: 0.074, 95% CI: 0.047–0.103, P < 0.001; NRI: 0.531, 95% CI: 0.341–0.717, P < 0.001) ( Table 4 ). Table 4 Incremental Predictive Value of BAR Beyond BNP Model AIC AUC (95% CI) ΔAUC (P value)* IDI (95% CI) cNRI (95% CI) Base model 576.31 – – – – + BNP 577.87 0.801 (0.763–0.840) Reference Reference Reference + BNP + BAR 530.79 0.845 (0.810–0.880) 0.043(0.0007) 0.074 (0.047–0.103) 0.531 (0.341–0.717) Notes : The base model consisted of Age, Sex, Heart failure, BMI, LAd, LVEF, Platelet, Sodium and eGFR. The BNP model was defined as the base model plus BNP. The full model was defined as the BNP model plus BAR. *AUCs were compared using the DeLong test. IDI and cNRI were used to evaluate the incremental predictive value of the full model over the BNP model. All IDI and cNRI Pvalues were <0.001. Analyses were based on 842 participants (137 events and 705 non-events) after multiple imputation by chained equations (MICE). Abbreviations : AUC, area under the curve; IDI, integrated discrimination improvement; cNRI, continuous net reclassification improvement.

Discussion

This retrospective study is the first to show that the BAR has a significant effect on the prognosis of elderly patients with AF. Kaplan–Meier analysis revealed that the all-cause mortality rate in the high-BAR group was significantly higher than that in the low-BAR group. After adjustment using a multivariate Cox proportional hazards model, the BAR remained an independent risk factor for ACM. In stepwise Cox regression analyses, adjustment for age reduced the HR from 6.785 to 5.582; further adjustment for sex resulted in minimal change (5.647). Inclusion of BMI and heart failure substantially attenuated the HR to 4.067, whereas additional adjustment for the CHA 2 DS 2 -VASc score, HAS-BLED score, ablation history, left atrial appendage occlusion, LAd, LVEF, and eGFR only modestly attenuated the HR to 3.639. These results indicate that heart failure, BMI, and age are the major confounders in the BAR-mortality association.

ROC curve analysis further revealed that the area under ROC curve (AUC) for the ability of the BAR to predict all-cause mortality was 0.778 (95% CI: 0.737–0.819). Its predictive efficacy is superior to that of BNP alone or albumin alone. In supplementary analyses, we performed non‑parametric tests (Mann‑Whitney U -test) comparing both raw BAR and log 10 ‑transformed BAR between survivors and non‑survivors. Both metrics showed significant inter‑group differences (BAR‑log 10 : U = 21,402.000, Z = −10.323, P < 0.001; BAR‑raw: U = 24,632.000, Z = −9.098, P < 0.001). We further conducted ROC analyses for raw BAR and log 10 ‑transformed BAR, which yielded AUCs of 0.745 (95% CI: 0.703–0.788) and 0.778 (95% CI: 0.737–0.819), respectively (DeLong test: Z = −4.043, P < 0.001). ( Supplementary Figure 1 ) These findings suggest that although log‑transformation moderately improves discriminatory ability, BAR retains statistically significant and clinically meaningful predictive value under both forms. Overall, these research findings endorse the utilization of BAR as an adjunctive tool for risk stratification in elderly patients with AF.

Inflammation is the most important factor in the development of AF, second only to atrial infarction and subsequent tissue fibrosis. 7 , 26 A study finds inflammatory cytokines selectively regulate BNP gene. 27 Inflammation and hemodynamic stress, which may be important pathophysiologic underpinnings of AF onset and left atrial appendage thrombosis, both trigger natriuretic peptide production. 28 BNP expression is primarily regulated by mechanical stretch, ischemia-hypoxia, and neurohumoral factors, among which mechanical stretch is considered the most important. 9 In addition to being a marker of inflammation and hemodynamic stress, 10 natriuretic peptides have a wide range of cardioprotective effects. Through the NPR-A/cGMP pathway, BNP exerts anti-myocardial hypertrophy and anti-fibrotic effects, suggesting its role in counteracting AF-related cardiac remodeling. 29

Enhanced BNP synthesis is a response to the increased ventricular load. 30 Elderly people often have increased left ventricular wall thickness and left ventricular mass, as well as elevated total peripheral resistance and systemic blood pressure. 31 Accordingly, plasma BNP and its precursors serve as useful biomarkers for excluding severe left ventricular dysfunction. 27 In addition, BNP is independently associated with AF burden, risk of AF progression, 11 , 32 and its elevation may also reflect increased left atrial pressure and impaired left ventricular filling. 32 Thus, elevated plasma BNP concentrations have important prognostic value in elderly patients. 31

Low serum albumin levels are associated with the increased risk of ischemic heart disease, heart failure, AF, stroke, and venous thromboembolism, 13 as well as an approximately twofold higher risk of cardiovascular mortality and all-cause mortality. 33 The pathophysiologic basis of these associations lies in albumin’s critical roles in inflammation and thrombosis. During systemic inflammation and metabolic stress, serum albumin—a negative acute-phase protein—decreases significantly due to reduced hepatic synthesis and increased capillary leakage into the extravascular space. 15 , 34 , 35 This is clinically relevant because inflammation itself is a key mechanism in the initiation and maintenance of AF 12 . Second, reduced albumin levels can further aggravate disease progression. Albumin inhibits platelet activation via binding free fatty acids. Lower albumin concentrations disrupt this homeostatic balance and favour a prothrombotic state. Meanwhile, accumulated free fatty acids may reciprocally amplify oxidative stress and inflammatory responses. 17 In addition, albumin possesses notable neuroprotective properties in patients with acute ischemic stroke. Such protective effects are mainly mediated by multiple pathways, including ameliorating cerebral microcirculation, suppressing thrombus formation, and inhibiting leukocyte adhesion to vascular endothelium. 36

These findings suggest that the pathophysiological processes reflected by these two markers are complementary, with serum albumin primarily reflecting systemic nutrition, inflammation, and hypercoagulability, and BNP primarily reflecting myocardial stress and metabolic function. The calibration curve of the BAR model demonstrated good agreement between predicted and observed outcomes, with a C-statistic of 0.8222. Notably, the addition of BAR significantly improved risk reclassification relative to BNP alone, as reflected by significant IDI and NRI.

Given that biomarkers capable of integrating multiple pathophysiological mechanisms of AF are critica l for optimizing risk stratification and prognostic assessment, 37 composite indices combining inflammatory and metabolic parameters have garnered increasing research interest. For instance, the Composite Inflammation Index (CTI) exhibited superior predictive performance than individual biomarkers including TyG and CRP for cardiovascular disease among middle-aged and elderly Chinese adults. 38 Likewise, the red cell distribution width-to-albumin ratio (RAR) and advanced lung cancer inflammation index (ALI) carry prognostic implications in stable heart failure; specifically, ALI predicts long-term outcomes, whereas RAR forecasts 1-year mortality. 39 Prior work has shown that BAR and heart rate can predict stroke risk among hospitalized patients with AF. 18 More recently, the NT-proBNP-to-albumin ratio (NTAR) – a biomarker functionally analogous to BAR – has emerged as a useful risk-stratification tool for patients with pulmonary arterial hypertension and chronic thromboembolic pulmonary hypertension, with strong discriminatory performance for predicting prolonged hospital admission and mortality. 40 Similarly, in patients with chronic heart failure, NTAR reflects the combined burden of hemodynamic stress and reduced systemic metabolic reserve, and independently predicts in-hospital mortality. 41 Building on these findings, our results show that the BAR --a composite index incorporating BNP (neurohormonal activation) and albumin (nutritional-inflammatory status) – confers additional prognostic value in elderly patients with AF. As a low-cost and routinely measurable biomarker, the BAR may therefore help estimate ACM risk, identify high-risk individuals, and guide personalized treatment and follow-up.

Limitations

This study is a retrospective study with inherent limitations. First, as a single-center study, the applicability of its results to a broader population requires the consideration of multicenter data. Second, the BAR was based on a single measurement at admission and thus could not capture dynamic changes following therapeutic interventions or longitudinal biomarker variations. Additionally, despite adjustment for multiple confounders, residual confounding cannot be completely excluded. Third, the median follow-up duration was longer in the high-BAR group, which may reflect survivor bias; however, the Cox model, which uses time as a scale, can partially account for this bias. Fourth, the BAR cutoff of 6.05 was selected based on the optimal threshold identified by restricted cubic spline (RCS) analysis. While this threshold showed an incremental predictive signal in our cohort, it may not represent a biologically or clinically optimal cutoff and should therefore be considered exploratory, warranting external validation. Finally, even though the model’s discriminatory power has improved, whether the BAR can guide clinical decision-making (such as risk stratification and follow-up strategies) still requires further clarification through validation in external cohorts and prospective intervention studies.

Conclusion

In summary, based on routine clinical measurements of serum albumin and cardiac biomarkers, this study revealed that the BNP/albumin ratio (BAR) is independently associated with all-cause mortality in elderly patients with AF. The potential prognostic value of this ratio may stem from the integration of two core pathophysiological dimensions: cardiac stress, as reflected by BNP, and systemic inflammation and metabolic burden, as reflected by albumin.

Abbreviations

BNP, B-type natriuretic peptide; AF, Atrial Fibrillation; BAR: BNP-to-albumin ratio; BMI, body mass index; SBP, systolic blood pressure; LY%, lymphocyte percentage; RBC, red blood cell; PLT, platelet; hs-CRP, high-sensitivity C-reactive protein; STB, total bilirubin; ChE, cholinesterase; UA, uric acid; LDL-C, low-density lipoprotein cholesterol; eGFR, estimated glomerular filtration rate; LAD, left atrial diameter; LVDd, left ventricular end-diastolic diameter; Rad, right atrial long‑axis diameter; LVESV, left ventricular end-systolic volume; LVEF, left ventricular ejection fraction; PASP, pulmonary artery systolic pressure; AF ablation, atrial fibrillation ablation; LAAO, left atrial appendage occlusion; HRs, heart rates; ACM all-cause mortality.

Data Sharing Statement

Due to privacy and ethical restrictions, these datasets can be obtained from the corresponding authors upon reasonable request.

Ethics Approval and Informed Consent

The study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Kunming Medical University and was conducted in accordance with the Declaration of Helsinki. The ethics approval number of the study was (2026) Ethics L No.59. We have obtained informed consent from each patient in this study.

Author Contributions

All authors gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Conceptualization: Xiran Zeng, Yunhong Yang, Lixing Chen

Data curation: Xiran Zeng, Yunhong Yang, Lixing Chen, Dajin Liu, Tao Shi, Yujuan Peng and Xinuo Ma

Formal analysis: Xiran Zeng, Yunhong Yang, Lixing Chen, Dajin Liu, Baotong Hua, Qin Fu and Zhimin Zheng

Funding acquisition: Lixing Chen

Investigation: Xiran Zeng, Yunhong Yang, Lixing Chen, Tao Shi, Yujuan Peng and Xinuo Ma

Methodology: Xiran Zeng, Yunhong Yang, Lixing Chen

Supervision: Lixing Chen

Writing – original Draft: Xiran Zeng, Yunhong Yang, Lixing Chen and Dajin Liu

Writing – review and editing: Xiran Zeng, Yunhong Yang, Lixing Chen, Dajin Liu, Baotong Hua, Tao Shi, Qin Fu and Zhimin Zheng, Yujuan Peng and Xinuo Ma

All authors have read and approved the final version of the manuscript.

Disclosure

The authors declare no competing interests.

References

1. Joglar JA , Chung MK , Armbruster AL , et al. 2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation: a Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines . Circulation . 2024 ; 149 ( 1 ). doi: 10.1161/CIR.0000000000001193 PMC11095842 38033089

2. Whitfield R , Ascenção R , Da Silva GL , Almeida AG , Pinto FJ , Caldeira D . Screening strategies for atrial fibrillation in the elderly population: a systematic review and network meta-analysis . Clin Res Cardiol . 2023 ; 112 ( 6 ): 705 – 14 . doi: 10.1007/s00392-022-02117-9 36342504

3. Van Gelder IC , Rienstra M , Bunting KV , et al. ESC Guidelines for the management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS) . Eur Heart J . 2024 ; 45 ( 36 ): 3314 – 3414 . doi: 10.1093/eurheartj/ehae176 39210723

4. Lippi G , Sanchis-Gomar F , Cervellin G . Global epidemiology of atrial fibrillation: an increasing epidemic and public health challenge . Int J Stroke . doi: 10.1177/1747493019897870 31955707

5. Ko D , Rahman F , Schnabel RB , Yin X , Benjamin EJ , Christophersen IE . Atrial fibrillation in women: epidemiology, pathophysiology, presentation, and prognosis . Nat Rev Cardiol . 2016 ; 13 ( 6 ): 321 – 332 . doi: 10.1038/nrcardio.2016.45 PMC5579870 27053455

6. Elliott AD , Middeldorp ME , Van Gelder IC , Albert CM , Sanders P . Epidemiology and modifiable risk factors for atrial fibrillation . Nat Rev Cardiol . 2023 ; 20 ( 6 ): 404 – 417 . doi: 10.1038/s41569-022-00820-8 36600003

7. Li Z , Liu Q , Liu F , et al. Atrial cardiomyopathy markers and new-onset atrial fibrillation risk in patients with acute myocardial infarction . Eur J Internal Med . 2022 ; 102 : 72 – 79 . doi: 10.1016/j.ejim.2022.04.019 35513991

8. Nattel S . Atrial fibrosis, endocardial damage, and thrombosis in atrial fibrillation . JACC Clin Electrophysiol . 2023 ; 9 ( 7 ): 1169 – 1171 . doi: 10.1016/j.jacep.2023.02.012 37115117

9. Rossi R , Jabrah D , Douglas A , et al. Investigating the Role of Brain Natriuretic Peptide (BNP) and N-Terminal-proBNP in Thrombosis and Acute Ischemic Stroke Etiology . Ijms . 2024 ; 25 ( 5 ): 2999 . doi: 10.3390/ijms25052999 PMC10931830 38474245

10. Volpe M , Gallo G , Rubattu S . Endocrine functions of the heart: from bench to bedside . Eur Heart J . 2023 ; 44 ( 8 ): 643 – 655 . doi: 10.1093/eurheartj/ehac759 36582126

11. Hijazi Z , Wallentin L . Natriuretic peptide should be used as a routine tool for evaluation of patients with atrial fibrillation . Heart . 2019 ; 105 ( 5 ): 353 – 354 . doi: 10.1136/heartjnl-2018-314040 30314973

12. Tan J , Zhang X . Relationship between serum albumin and risk of atrial fibrillation: a dose-response meta-analysis . Front. Nutr . 2021 ; 8 : 728353 10.3389/fnut.2021.728353 PMC8418186 34490334

13. Arques S . Human serum albumin in cardiovascular diseases . Eur J Internal Med . 2018 ; 52 : 8 – 12 . doi: 10.1016/j.ejim.2018.04.014 29680174

14. Chien SC , Chen CY , Lin CF , Yeh HI . Critical appraisal of the role of serum albumin in cardiovascular disease . Biomark Res . 2017 ; 5 : 31 . doi: 10.1186/s40364-017-0111-x PMC5681838 29152305

15. Manolis AA , Manolis TA , Melita H , Mikhailidis DP , Manolis AS . Low serum albumin: a neglected predictor in patients with cardiovascular disease . Eur J Internal Med . 2022 ; 102 : 24 – 39 . doi: 10.1016/j.ejim.2022.05.004 35537999

16. Chen S , Redfors B , O’Neill BP , et al. Low and elevated B-type natriuretic peptide levels are associated with increased mortality in patients with preserved ejection fraction undergoing transcatheter aortic valve replacement: an analysis of the PARTNER II trial and registry . Eur Heart J . 2020 ; 41 ( 8 ): 958 – 969 . doi: 10.1093/eurheartj/ehz892 31883339

17. Dhindsa S , Ghanim H , Dandona P . Nonesterified fatty acids, albumin, and platelet aggregation . Diabetes . 2015 ; 64 ( 3 ): 703 – 705 . doi: 10.2337/db14-1481 25713195

18. Ndjana Lessomo FY , Luo X , Fan Q , Afeku K , quan WZ , Zheng Z . The BNP to albumin ratio and heart rate in stroke complicating atrial fibrillation . Curr Prob Cardiol . 2023 ; 48 ( 4 ): 101556 . doi: 10.1016/j.cpcardiol.2022.101556 36538997

19. Parks AL , Frankel DS , Kim DH , et al. Management of atrial fibrillation in older adults . BMJ . 2024 ; 386 : e076246 . doi: 10.1136/bmj-2023-076246 39288952

20. Soler-Espejo E , Rivera-Caravaca JM , Bru-Cánovas JD , et al. Patients with atrial fibrillation and common exclusion criteria from clinical trials are at high risk of clinical events: the Murcia AF Project II (MAFP-II) cohort study . Intern Emerg Med . 2024 ; 19 ( 7 ): 1941 – 1948 . doi: 10.1007/s11739-024-03701-9 PMC11467067 38965203

21. von Elm E , Altman DG , Egger M , Gotzsche PC , Vandenbroucke JP . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies . Rev Saude Publica . 2007 ; 41 ( 6 ): 931‑937

22. Calkins H , Kuck KH , Cappato R , et al. HRS/EHRA/ECAS expert consensus statement on catheter and surgical ablation of atrial fibrillation: recommendations for patient selection, procedural techniques, patient management and follow-up, definitions, endpoints, and research trial design . Europace . 2012 ; 14 ( 4 ): 528 – 606 . doi: 10.1093/europace/eus027 22389422

23. Park J , Cha MJ , Jung CY , et al. Prognostic efficacy of platelet count in patients with nonvalvular atrial fibrillation . Heart Rhythm . 2019 ; 16 ( 2 ): 197 – 203 . doi: 10.1016/j.hrthm.2018.08.023 30170225

24. Zhou Y , Lin D , Wu S , et al. Dysnatremia is associated with increased risk of all-cause mortality within 365 days post-discharge in patients with atrial fibrillation without heart failure: a prospective cohort study . Front Cardiovasc Med . 2022 ; 9 : 963103 . doi: 10.3389/fcvm.2022.963103 PMC9597697 36312256

25. Ha JT , Freedman SB , Kelly DM , et al. Kidney function, albuminuria, and risk of incident atrial fibrillation: a systematic review and meta-analysis . Am J Kidney Dis . 2024 ; 83 ( 3 ): 350 – 359.e1 . doi: 10.1053/j.ajkd.2023.07.023 37777059

26. Sagris M , Vardas EP , Theofilis P , Antonopoulos AS , Oikonomou E , Tousoulis D . Atrial Fibrillation: pathogenesis, Predisposing Factors, and Genetics . Ijms . 2021 ; 23 ( 1 ): 6 . doi: 10.3390/ijms23010006 PMC8744894 35008432

27. Goetze JP . Atrial secretion of B-type natriuretic peptide . Eur Heart J . 2006 ; 27 ( 14 ): 1648 – 1650 . doi: 10.1093/eurheartj/ehl109 16785247

28. Goetze JP , Bruneau BG , Ramos HR , Ogawa T , de Bold MK , de Bold AJ . Cardiac natriuretic peptides . Nat Rev Cardiol . 2020 ; 17 ( 11 ): 698 – 717 . doi: 10.1038/s41569-020-0381-0 32444692

29. Nishikimi T , Maeda N , Matsuoka H . The role of natriuretic peptides in cardioprotection . Cardiovasc Res . 2006 ; 69 ( 2 ): 318 – 328 . doi: 10.1016/j.cardiores.2005.10.001 16289003

30. Hama N , Itoh H , Shirakami G , et al. Rapid Ventricular Induction of Brain Natriuretic Peptide Gene Expression in Experimental Acute Myocardial Infarction . Circulation . 1995 ; 92 ( 6 ): 1558 – 1564 . doi: 10.1161/01.CIR.92.6.1558 7664440

31. Wallén T , Landahl S , Hedner T , Saito Y , Masuda I , Nakao K . Brain natriuretic peptide in an elderly population . J Internal Med . 1997 ; 242 ( 4 ): 307 – 311 . doi: 10.1046/j.1365-2796.1997.00184.x 9366809

32. Inohara T , Kim S , Pieper K , et al. B-type natriuretic peptide, disease progression and clinical outcomes in atrial fibrillation . Heart . 2019 ; 105 ( 5 ): 370 – 377 . doi: 10.1136/heartjnl-2018-313642 30228248

33. Djoussé L , Rothman KJ , Cupples LA , Levy D , Ellison RC . Serum albumin and risk of myocardial infarction and all-cause mortality in the framingham offspring study . Circulation . 2002 ; 106 ( 23 ): 2919 – 2924 . doi: 10.1161/01.CIR.0000042673.07632.76 12460872

34. Gruys E , Toussaint MJM , Niewold TA , Koopmans SJ . Acute phase reaction and acute phase proteins . J Zheijang Univ Sci B . 2005 ; 6 ( 11 ): 1045 – 1056 . doi: 10.1631/jzus.2005.B1045 PMC1390650 16252337

35. Bao J , Hu J , Lv X . Association between CRP-TyG index and hepatic synthetic function in post-myocardial infarction ICU patients . Front Med . 2025 ; 12 : 1710430 . doi: 10.3389/fmed.2025.1710430 PMC12738846 41458520

36. Dziedzic T , Slowik A , Szczudlik A . Serum Albumin Level as a Predictor of Ischemic Stroke Outcome . Stroke . 2004 ; 35 ( 6 ). doi: 10.1161/01.STR.0000126609.18735.be 15073386

37. Stoll L , Ha EE , Lo JC . Metabolic and Inflammatory Mechanisms Driving Atrial Fibrillation . 2025 . 10.1146/annurev-pharmtox-062124-025403 PMC12581197 40956862

38. Zhang L , Li S , Liu D , et al. The relationship between C-reactive protein-triglyceride-glucose index and cardiovascular disease: insights from the China health and retirement longitudinal study (CHARLS) . Cardiovasc Diabetol . 2025 ; 24 ( 1 ): 410 . doi: 10.1186/s12933-025-02960-w PMC12560464 41152893

39. Liu S , Fu T , Deng T , Cai X , Zhan Y , Zhu H . Association between inflammation- and nutrition-related indicators and mortality in patients with heart failure: a cohort study . Front Nutr . 2025 ; 12 : 1617069 . doi: 10.3389/fnut.2025.1617069 PMC12602239 41229545

40. Iancu DG , Cristescu L , Mares RG , Varga A , Tilea I . Evaluating NT-proBNP-to-Albumin (NTAR) and RDW-to-eGFR (RGR) Ratios as Biomarkers for Predicting Hospitalization Duration and Mortality in Pulmonary Arterial Hypertension (PAH) and Chronic Thromboembolic Pulmonary Hypertension (CTEPH) . Diagnostics . 2025 ; 15 ( 17 ): 2126 . doi: 10.3390/diagnostics15172126 PMC12428362 40941614

41. Cristescu L , Mares RG , Iancu DG , Marusteri MS , Varga A , Tilea I . Prognostic Value of the NT-proBNP-to-Albumin Ratio (NTAR) for In-Hospital Mortality in Chronic Heart Failure Patients . Biomedicines . 2025 ; 13 ( 9 ): 2091 . doi: 10.3390/biomedicines13092091 PMC12467257 41007652