Association of the triglyceride glucose–body roundness index with mortality in HFpEF and effect modification by nutritional status: a multicenter prospective cohort study Authors: You Zhou, Yanhui Han, Yingli Xie, Liping Wang, Hengliang Zhang, Ruixiao Chen, Bingqiang Li, Jindong Zhao, Zhaokun Pu, Huifeng Zhang Published: 2025 ABSTRACT BackgroundHeart failure with preserved ejection fraction (HFpEF) is a major cause of morbidity and mortality. Insulin resistance and visceral adiposity play pivotal roles in its pathogenesis; however, the prognostic significance of the triglyceride glucose–body roundness index (TyG–BRI) and its interaction with nutritional status remain uncertain.MethodsA total of 1,015 patients with HFpEF were prospectively enrolled. The TyG–BRI was calculated as the product of the triglyceride–glucose index (TyG index) and the body roundness index (BRI). Cox regression models were used to evaluate its prognostic value for adverse outcomes, with subgroup analyses stratified by nutritional status and mediation analysis assessing the role of exercise tolerance.ResultsDuring follow-up, 232 (22.9%) all-cause and 158 (15.6%) cardiovascular (CV) deaths occurred. In multivariate Cox regression analysis, patients in the highest TyG–BRI tertile exhibited significantly higher risks of all-cause (hazard ratio [HR] = 2.59, 95% confidence interval [CI]: 1.82–3.71; P for trend < 0.001) and CV mortality (HR = 2.43, 95% CI: 1.59–3.72; P for trend < 0.001) compared with those in the lowest tertile. The prognostic value of the TyG–BRI for all-cause mortality was more prominent among participants with impaired nutritional status (malnourished or at risk of malnutrition) (P for interaction < 0.1). Incorporation of the TyG–BRI into the baseline risk model for all-cause death significantly improved model discrimination (C-statistic = 0.701 vs. 0.626; P < 0.001) and outperformed its individual components (P < 0.05). Mediation analysis further revealed that the six-minute walk distance mediated 14.6% of the association between the TyG–BRI and all-cause mortality.ConclusionsIn patients with HFpEF, the TyG–BRI was independently associated with adverse outcomes, with its prognostic value particularly evident among those with impaired nutritional status. Incorporating the TyG–BRI into the risk model modestly improved prognostic discrimination, and reduced exercise capacity appeared to partly mediate this association.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12944-025-02834-w. Introduction Heart failure has become a global epidemic, affecting more than 60 million individuals worldwide, with heart failure with preserved ejection fraction (HFpEF) accounting for approximately half of all cases. The prevalence of HFpEF continues to rise across populations and is associated with substantial morbidity and mortality, with five-year mortality approaching 50% and hospitalization rates reaching up to 80% [1, 2]. Therefore, comprehensive identification of risk factors and effective risk stratification are essential for guiding clinical management and improving outcomes in patients with HFpEF. Accurate diagnosis of HFpEF remains a major challenge because of its complex and heterogeneous nature. This heterogeneity indicates that HFpEF is driven by multiple interrelated pathophysiological mechanisms rather than a single pathway [1]. Patients with HFpEF exhibit diverse clinical phenotypes, a high burden of comorbidities, and multisystem involvement [3]. Among these mechanisms, insulin resistance (IR) has emerged as a key pathophysiological contributor to HFpEF. Experimental and clinical studies have demonstrated that IR promotes systemic inflammation, neurohormonal activation, and cytokine overproduction, which together contribute to myocardial remodeling and disease progression [4]. Obesity, particularly visceral adiposity, frequently coexists with IR in HFpEF, forming a self-perpetuating vicious cycle that accelerates cardiac remodeling through metabolic and inflammatory pathways. This interplay underscores the importance of simultaneously considering both metabolic and adiposity-related factors in risk evaluation [5]. The triglyceride–glucose (TyG) index is a well-validated surrogate marker of IR [6]. The body roundness index (BRI), a novel geometric indicator of visceral adiposity, provides a more comprehensive assessment of body fat distribution than conventional anthropometric indices [7]. Integrating the TyG index with the BRI may therefore offer a more accurate approach to risk stratification in HFpEF. Although several studies have investigated the prognostic value of combining metabolic and adiposity indices in specific disease contexts [8, 9], such evidence remains scarce in patients with HFpEF. Moreover, malnutrition is common in this population [10], and nutritional assessment constitutes an essential component of risk stratification and clinical management (Fig. 1). Whether different nutritional statuses modify the prognostic impact of this combined index remains unclear. Fig. 1Proposed mechanisms underlying the prognostic role of TyG–BRI in patients with HFpEF. HFpEF heart failure with preserved ejection fraction, TyG–BRI triglyceride glucose–body roundness index, MNA-SF Mini Nutritional Assessment-Short Form Proposed mechanisms underlying the prognostic role of TyG–BRI in patients with HFpEF. HFpEF heart failure with preserved ejection fraction, TyG–BRI triglyceride glucose–body roundness index, MNA-SF Mini Nutritional Assessment-Short Form To our knowledge, this is the first study to evaluate the combined prognostic value of the TyG index and the BRI in patients with HFpEF. Based on these two parameters, we developed the TyG–BRI to examine its prognostic significance across different nutritional statuses and to assess its incremental value in risk stratification. Methods Study design and population This was a prospective, multicenter cohort study that enrolled patients with HFpEF who were hospitalized at the First Affiliated Hospital of Henan University of Science and Technology, the First and Sixth Medical Centers of the Chinese People’s Liberation Army General Hospital, and the Affiliated Hospital of Chengde Medical University between March 2023 and September 2023. According to the 2021 European Society of Cardiology (ESC) Guidelines for the Diagnosis and Treatment of Acute and Chronic Heart Failure, the diagnosis of HFpEF required fulfillment of the following three criteria [11]: (1) symptoms and/or signs of heart failure; (2) left ventricular ejection fraction (LVEF) ≥ 50%; and (3) evidence of structural heart disease or diastolic dysfunction on echocardiography, accompanied by elevated NT-proBNP levels. All participants had symptomatic chronic HFpEF with structural heart disease, corresponding to ACC/AHA Stage C heart failure. Exclusion criteria were as follows: (1) age < 18 or > 90 years, or pregnancy; (2) severe hepatic or renal dysfunction; (3) advanced malignancy or connective tissue diseases; (4) inability to complete enrollment or follow-up for any reason, or incomplete clinical data; and (5) inability to complete the Mini Nutritional Assessment–Short Form (MNA-SF) questionnaire or the six-minute walk distance (6MWD) test. A total of 1,015 eligible patients were included in the final analysis (614 men and 401 women; mean age, 65.3 ± 12.8 years). Participants were categorized into three groups according to TyG–BRI tertiles: T1 (TyG–BRI < 24.81, n = 338), T2 (24.81 ≤ TyG–BRI < 35.75, n = 338), and T3 (TyG–BRI ≥ 35.75, n = 339) (Fig. 2). Fig. 2Flow diagram of participants selection. HFpEF heart failure with preserved ejection fraction, ESC European Society of Cardiology, MNA-SF Mini Nutritional Assessment-Short Form, 6MWD six-minute walk distance, CV death cardiovascular death, TyG–BRI triglyceride glucose–body roundness index Flow diagram of participants selection. HFpEF heart failure with preserved ejection fraction, ESC European Society of Cardiology, MNA-SF Mini Nutritional Assessment-Short Form, 6MWD six-minute walk distance, CV death cardiovascular death, TyG–BRI triglyceride glucose–body roundness index Ethics statement This study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Henan University of Science and Technology (approval number: 2023-03-K0026). Written informed consent was obtained from all participants after a detailed explanation of the study objectives and procedures. All patient data were anonymized prior to analysis to ensure confidentiality and protect participant privacy. Data collection and definitions Clinical data, including vital signs, laboratory parameters, echocardiographic measurements, comorbidities, and medication profiles, were obtained from electronic medical records. The TyG index was calculated as ln [fasting TG (mmol/L) × 88.6 × FBG (mmol/L) × 18 / 2] [12], and the BRI was derived according to the formula [8]:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\text{B}\text{R}\text{I}=364.2-365.5\times\sqrt{1-{\left(\frac{WC/2\pi\:}{0.5\times height}\right)}^{2}}$$\end{document} The TyG–BRI was calculated by multiplying the TyG index by the BRI. Mean arterial pressure (MAP) was computed as (systolic blood pressure + 2 × diastolic blood pressure) / 3. Chronic kidney disease (CKD) was identified either from medical history or by an estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m2, calculated using the CKD-EPI equation [13]. According to the original validation and guideline recommendations for the MNA-SF, scores of ≤ 7, 8–11, and 12–14 were classified as malnutrition, at risk of malnutrition, and normal nutritional status, respectively [14, 15]. Exercise capacity was further assessed using the six-minute walk distance (6MWD) test. Follow-up and outcomes Prognostic information was obtained through telephone follow-up or by reviewing electronic medical records, with a median follow-up duration of 2.0 years (IQR 1.9–2.2 years). The primary endpoints were all-cause mortality and cardiovascular (CV) mortality. Statistical analysis Baseline characteristics were summarized across tertiles of the TyG–BRI. Continuous variables were reported as mean ± SD or median (IQR), as appropriate; group differences were assessed using one-way ANOVA for normally distributed variables or the Kruskal–Wallis test for non-normal data. Categorical variables were presented as n (%) and compared using the chi-squared test or Fisher’s exact test, as appropriate. Kaplan–Meier curves were used to estimate the cumulative incidence of the primary outcomes, and between-group differences were assessed with the log-rank test. Associations between the TyG–BRI and outcomes were examined using Cox proportional hazards models, treating the TyG–BRI as a categorical variable (tertiles; lowest tertile as the reference) and as a continuous variable (per SD increase). Linear trend across tertiles was tested by entering the within-tertile median as a continuous term. In addition to the unadjusted model, two multivariable models (Model 1 and Model 2) were fitted. The proportional hazards assumption was evaluated with Schoenfeld residuals, and no violations were detected. Dose–response relationships were further explored using restricted cubic spline (RCS) with three knots to depict the association between the TyG–BRI and hazard ratios (HRs). Exploratory analyses were stratified by nutritional status, and robustness was assessed in prespecified comorbidity subgroups. Interactions were evaluated using likelihood ratio tests, with two-sided P < 0.10 considered significant for interaction. The incremental predictive value of the TyG–BRI for risk stratification was quantified by the C-statistic, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision-curve analysis (DCA), and was compared with the TyG index and BRI individually. Mediation analysis (R “mediation” package) assessed the mediating role of 6MWD in the association between the TyG–BRI and mortality, adjusting for Model 2 covariates. Evidence for mediation was defined a priori as a significant natural indirect effect, a significant total effect, and a positive proportion mediated. All statistical analyses were conducted in R (version 4.2.1; R Foundation for Statistical Computing, Vienna, Austria). Two-sided P < 0.05 was considered statistically significant. Results Characteristics of the participants A total of 1,015 participants were analyzed (mean age, 65.3 years; 60.5% male). Baseline characteristics by tertiles of the TyG–BRI are summarized in Table 1. Participants in the highest tertile exhibited a more adverse cardiometabolic profile, with higher BMI, MAP, triglycerides, fasting glucose, total cholesterol, and LDL-C, and lower HDL-C, along with modestly higher platelet count and serum sodium, and a lower LVEF (all P < 0.05). Diabetes and ischemic etiology were more prevalent in the highest tertile (P = 0.017 and 0.049), whereas age, sex, AF, CKD, hypertension, and prior stroke did not differ materially across tertiles (all P > 0.05). Consistent with the greater glycemic burden, insulin and other oral hypoglycemic agents were more frequently prescribed in T3, whereas other medication classes were broadly comparable across groups. Table 1Baseline characteristics of the study population based on tertiles of TyG–BRIVariablesTertile of TyG–BRIP valueT1 (n = 338)T2 (n = 338)T3 (n = 339)TyG–BRI< 24.8124.81–35.75≥ 35.75 - Demographics Age (years)65.0 (56.0–76.0)66.0 (57.0–75.0)66.0 (56.0–75.0)0.958 Male (%)193 (57.10)207 (61.24)214 (63.13)0.260Current/ex-Smoker (%)77 (22.78)86 (25.44)102 (30.09)0.091Medical measurements BMI (kg/m2)22.2 (20.2–23.8)25.8 (24.0-27.4)28.1 (26.6–31.2)< 0.001 MAP (mmHg)92.7 ± 15.495.8 ± 15.796.9 ± 14.20.001 HR (bpm)78.0 (68.0–90.0)77.0 (68.0–88.0)77.0 (68.0–87.0)0.571NYHA classification (%)0.255 I-II164 (48.52)169 (50.00)191 (56.34) III126 (37.28)126 (37.28)113 (33.33) IV48 (14.20)43 (12.72)35 (10.32)Medical history (%) AF141 (41.72)131 (38.76)127 (37.46)0.510 CKD100 (29.59)102 (30.18)112 (33.04)0.583 Hypertension244 (72.19)229 (67.75)237 (69.91)0.453 Diabetes145 (42.90)139 (41.12)174 (51.33)0.017 Stroke87 (25.74)89 (26.33)88 (25.96)0.984 Ischemic etiology65 (19.23)90 (26.63)87 (25.66)0.049Laboratory measurements WBC (109/L)6.38 (5.19–7.94)6.48 (5.21–8.16)6.49 (5.41–8.08)0.443 Lymphocyte (109/L)1.87 (1.46–2.46)1.90 (1.48–2.46)1.97 (1.51–2.56)0.378 Hemoglobin (g/L)128.0 (112.0-143.0)128.0 (115.0-143.9)129.0 (113.5-142.5)0.935 Albumin (g/L)39.8 (36.5–42.6)39.8 (37.0-42.1)39.3 (36.7–42.7)0.956 Platelets (109/L)188.0 (150.2–241.0)200.0 (166.2–237.0)198.0 (162.0-249.5)0.024 ALT (U/L)19.1 (14.1–28.3)19.3 (13.1–29.1)20.5 (13.2–30.4)0.677 AST (U/L)22.5 (17.8–30.7)21.5 (17.5–27.6)23.1 (17.8–33.8)0.070 Creatinine (umol/L)77.4 (62.3–97.3)77.8 (61.8–95.4)79.4 (63.2–99.3)0.614 eGFR (ml/min/1.73m2)82.4 (59.8–97.5)82.5 (61.7–97.5)79.8 (59.6–96.3)0.800 CRP (mg/L)4.80 (2.20-16.15)4.80 (2.60–14.20)5.10 (2.40-14.35)0.981 FBG (mmol/L)5.25 (4.58–6.36)5.64 (4.88–6.79)5.95 (5.02–7.32)< 0.001 TG (mmol/L)0.98 (0.70–1.47)1.21 (0.88–1.78)1.44 (1.05–1.92)< 0.001 TC (mmol/L)3.64 (2.94–4.34)3.73 (3.09–4.63)3.88 (3.27–4.62)< 0.001 LDL-C (mmol/L)2.05 (1.55–2.68)2.24 (1.65–2.88)2.27 (1.76–2.94)0.008 HDL-C (mmol/L)1.09 (0.83–1.34)1.04 (0.84–1.21)1.00 (0.82–1.19)0.014 Potassium (mmol/L)3.97 (3.69–4.31)4.01 (3.71–4.30)3.93 (3.65–4.25)0.373 Sodium (mmol/L)140.5 (137.9–142.0)140.8 (138.1-142.9)140.9 (138.6-143.2)0.026 NT-proBNP (pg/ml)1582.5(784.2-3903.8)1652.0 (613.3–4676.0)1397.0 (734.6-3941.5)0.821Echocardiography LVEF (%)60.0 (55.0–65.0)59.0 (55.0–64.0)58.0 (54.0–62.0)0.006Medications at discharge (%) Antiplatelet agents172 (50.89)186 (55.03)191 (56.34)0.332 ACEI/ARB/ARNI133 (39.35)156 (46.15)155 (45.72)0.136 Beta-blocker219 (64.79)228 (67.46)226 (66.67)0.754 CCB81 (23.96)93 (27.51)92 (27.14)0.514 Lipid-lowering drugs195 (57.69)202 (59.76)210 (61.95)0.529 Digoxin71 (21.01)66 (19.53)63 (18.58)0.727 Mineralocorticoid antagonists206 (60.95)202 (59.76)214 (63.13)0.660 Diuretics184 (54.44)178 (52.66)192 (56.64)0.582 SGLT2 inhibitors129 (38.17)135 (39.94)132 (38.94)0.894 Insulin26 (7.69)30 (8.88)45 (13.27)0.038 Other oral antidiabetic agents66 (19.53)79 (23.37)107 (31.56)0.001All-cause death Incidence/1000 person-years72.59119.47184.81-CV death Incidence/1000 person-years51.4279.65125.47-TyG–BRI triglyceride glucose–body roundness index, BMI body mass index, MAP mean arterial pressure, HR heart rate, NYHA New York Heart Association, AF atrial fibrillation, CKD chronic kidney disease, WBC white blood cell, AST aspartate aminotransferase, ALT alanine aminotransferase, eGFR estimated glomerular filtration rate, CRP C-reactive protein, FBG fasting blood glucose, TG triglyceride, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol, NT-proBNP N-terminal pro-brain natriuretic peptide, LVEF left ventricular ejection fraction, ACEI/ARB/ARNI angiotensin converting enzyme inhibitor/angiotensin receptor blocker/angiotensin receptor-neprilysin inhibitors, CCB calcium channel blockers, SGLT2 inhibitors sodium-glucose co-transporter-2 inhibitors, CV death cardiovascular death Table 1Baseline characteristics of the study population based on tertiles of TyG–BRIVariablesTertile of TyG–BRIP valueT1 (n = 338)T2 (n = 338)T3 (n = 339)TyG–BRI< 24.8124.81–35.75≥ 35.75 - Demographics Age (years)65.0 (56.0–76.0)66.0 (57.0–75.0)66.0 (56.0–75.0)0.958 Male (%)193 (57.10)207 (61.24)214 (63.13)0.260Current/ex-Smoker (%)77 (22.78)86 (25.44)102 (30.09)0.091Medical measurements BMI (kg/m2)22.2 (20.2–23.8)25.8 (24.0-27.4)28.1 (26.6–31.2)< 0.001 MAP (mmHg)92.7 ± 15.495.8 ± 15.796.9 ± 14.20.001 HR (bpm)78.0 (68.0–90.0)77.0 (68.0–88.0)77.0 (68.0–87.0)0.571NYHA classification (%)0.255 I-II164 (48.52)169 (50.00)191 (56.34) III126 (37.28)126 (37.28)113 (33.33) IV48 (14.20)43 (12.72)35 (10.32)Medical history (%) AF141 (41.72)131 (38.76)127 (37.46)0.510 CKD100 (29.59)102 (30.18)112 (33.04)0.583 Hypertension244 (72.19)229 (67.75)237 (69.91)0.453 Diabetes145 (42.90)139 (41.12)174 (51.33)0.017 Stroke87 (25.74)89 (26.33)88 (25.96)0.984 Ischemic etiology65 (19.23)90 (26.63)87 (25.66)0.049Laboratory measurements WBC (109/L)6.38 (5.19–7.94)6.48 (5.21–8.16)6.49 (5.41–8.08)0.443 Lymphocyte (109/L)1.87 (1.46–2.46)1.90 (1.48–2.46)1.97 (1.51–2.56)0.378 Hemoglobin (g/L)128.0 (112.0-143.0)128.0 (115.0-143.9)129.0 (113.5-142.5)0.935 Albumin (g/L)39.8 (36.5–42.6)39.8 (37.0-42.1)39.3 (36.7–42.7)0.956 Platelets (109/L)188.0 (150.2–241.0)200.0 (166.2–237.0)198.0 (162.0-249.5)0.024 ALT (U/L)19.1 (14.1–28.3)19.3 (13.1–29.1)20.5 (13.2–30.4)0.677 AST (U/L)22.5 (17.8–30.7)21.5 (17.5–27.6)23.1 (17.8–33.8)0.070 Creatinine (umol/L)77.4 (62.3–97.3)77.8 (61.8–95.4)79.4 (63.2–99.3)0.614 eGFR (ml/min/1.73m2)82.4 (59.8–97.5)82.5 (61.7–97.5)79.8 (59.6–96.3)0.800 CRP (mg/L)4.80 (2.20-16.15)4.80 (2.60–14.20)5.10 (2.40-14.35)0.981 FBG (mmol/L)5.25 (4.58–6.36)5.64 (4.88–6.79)5.95 (5.02–7.32)< 0.001 TG (mmol/L)0.98 (0.70–1.47)1.21 (0.88–1.78)1.44 (1.05–1.92)< 0.001 TC (mmol/L)3.64 (2.94–4.34)3.73 (3.09–4.63)3.88 (3.27–4.62)< 0.001 LDL-C (mmol/L)2.05 (1.55–2.68)2.24 (1.65–2.88)2.27 (1.76–2.94)0.008 HDL-C (mmol/L)1.09 (0.83–1.34)1.04 (0.84–1.21)1.00 (0.82–1.19)0.014 Potassium (mmol/L)3.97 (3.69–4.31)4.01 (3.71–4.30)3.93 (3.65–4.25)0.373 Sodium (mmol/L)140.5 (137.9–142.0)140.8 (138.1-142.9)140.9 (138.6-143.2)0.026 NT-proBNP (pg/ml)1582.5(784.2-3903.8)1652.0 (613.3–4676.0)1397.0 (734.6-3941.5)0.821Echocardiography LVEF (%)60.0 (55.0–65.0)59.0 (55.0–64.0)58.0 (54.0–62.0)0.006Medications at discharge (%) Antiplatelet agents172 (50.89)186 (55.03)191 (56.34)0.332 ACEI/ARB/ARNI133 (39.35)156 (46.15)155 (45.72)0.136 Beta-blocker219 (64.79)228 (67.46)226 (66.67)0.754 CCB81 (23.96)93 (27.51)92 (27.14)0.514 Lipid-lowering drugs195 (57.69)202 (59.76)210 (61.95)0.529 Digoxin71 (21.01)66 (19.53)63 (18.58)0.727 Mineralocorticoid antagonists206 (60.95)202 (59.76)214 (63.13)0.660 Diuretics184 (54.44)178 (52.66)192 (56.64)0.582 SGLT2 inhibitors129 (38.17)135 (39.94)132 (38.94)0.894 Insulin26 (7.69)30 (8.88)45 (13.27)0.038 Other oral antidiabetic agents66 (19.53)79 (23.37)107 (31.56)0.001All-cause death Incidence/1000 person-years72.59119.47184.81-CV death Incidence/1000 person-years51.4279.65125.47-TyG–BRI triglyceride glucose–body roundness index, BMI body mass index, MAP mean arterial pressure, HR heart rate, NYHA New York Heart Association, AF atrial fibrillation, CKD chronic kidney disease, WBC white blood cell, AST aspartate aminotransferase, ALT alanine aminotransferase, eGFR estimated glomerular filtration rate, CRP C-reactive protein, FBG fasting blood glucose, TG triglyceride, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol, NT-proBNP N-terminal pro-brain natriuretic peptide, LVEF left ventricular ejection fraction, ACEI/ARB/ARNI angiotensin converting enzyme inhibitor/angiotensin receptor blocker/angiotensin receptor-neprilysin inhibitors, CCB calcium channel blockers, SGLT2 inhibitors sodium-glucose co-transporter-2 inhibitors, CV death cardiovascular death Baseline characteristics of the study population based on tertiles of TyG–BRI TyG–BRI triglyceride glucose–body roundness index, BMI body mass index, MAP mean arterial pressure, HR heart rate, NYHA New York Heart Association, AF atrial fibrillation, CKD chronic kidney disease, WBC white blood cell, AST aspartate aminotransferase, ALT alanine aminotransferase, eGFR estimated glomerular filtration rate, CRP C-reactive protein, FBG fasting blood glucose, TG triglyceride, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol, NT-proBNP N-terminal pro-brain natriuretic peptide, LVEF left ventricular ejection fraction, ACEI/ARB/ARNI angiotensin converting enzyme inhibitor/angiotensin receptor blocker/angiotensin receptor-neprilysin inhibitors, CCB calcium channel blockers, SGLT2 inhibitors sodium-glucose co-transporter-2 inhibitors, CV death cardiovascular death Associations between TyG–BRI and adverse outcomes During follow-up, event rates were 123.48 and 84.09 per 1,000 person-years for all-cause and CV death, respectively. Across TyG–BRI tertiles from T1 to T3, incidence increased stepwise—72.59, 119.47, and 184.81 per 1,000 person-years for all-cause death, and 51.42, 79.65, and 125.47 per 1,000 person-years for CV death. Kaplan–Meier curves showed higher cumulative incidence with increasing tertiles (Fig. 3; log-rank test, both P < 0.001). In RCS analyses, the overall association between the TyG–BRI and both all-cause and CV mortality was significant (both P for overall association < 0.001) with no evidence of nonlinearity (both P for non-linearity > 0.05), indicating an approximately linear relationship across the observed range (Fig. 4). Fig. 3Kaplan-Meier estimation of (a) all-cause death and (b) CV death by tertiles of TyG–BRI in HFpEF. CV death cardiovascular death, TyG–BRI triglyceride glucose–body roundness index, HFpEF heart failure with preserved ejection fraction Kaplan-Meier estimation of (a) all-cause death and (b) CV death by tertiles of TyG–BRI in HFpEF. CV death cardiovascular death, TyG–BRI triglyceride glucose–body roundness index, HFpEF heart failure with preserved ejection fraction Fig. 4HRs for (a) all-cause and (b) CV death in HFpEF from spline analyses adjusted for model 2. CV death cardiovascular death. Restricted cubic splines of the TyG–BRI with three knots were used HRs for (a) all-cause and (b) CV death in HFpEF from spline analyses adjusted for model 2. CV death cardiovascular death. Restricted cubic splines of the TyG–BRI with three knots were used Table 2 summarizes three Cox models evaluating the association between the TyG–BRI and outcomes. Across all models, the highest tertile (T3) was associated with greater all-cause mortality versus T1 (unadjusted HR, 2.57; 95% CI, 1.83–3.61; Model 1 h, 2.43; 95% CI, 1.72–3.41; Model 2 h, 2.59; 95% CI, 1.82–3.71; all P < 0.001). Treating the TyG–BRI as a continuous variable also indicated higher risk (Model 2 per-SD HR, 1.54; 95% CI, 1.36–1.74; P < 0.001). Similar patterns were observed for CV death (Model 2: T3 vs. T1 HR, 2.43; 95% CI, 1.59–3.72; per-SD HR, 1.54; 95% CI, 1.33–1.78; both P < 0.001). Table 2HRs (95% CI) of primary outcomes according to TyG–BRI tertilesCategoriesIncidence/1000 person-yUnadjustedModel 1Model 2HR (95% CI)P-valueP for trendHR (95% CI)P-valueP for trendHR (95% CI)P-valueP for trendAll-cause death Continuous variable per SD1.55 (1.38–1.74)< 0.0011.54 (1.37–1.73)< 0.0011.54 (1.36–1.74)< 0.001 Tertilea< 0.001< 0.001< 0.001 T1 (n = 338)72.59Ref.Ref.Ref. T2 (n = 338)119.471.65 (1.15–2.37)0.0071.56 (1.09–2.25)0.0161.58 (1.09–2.30)0.015 T3 (n = 339)184.812.57 (1.83–3.61)< 0.0012.43 (1.72–3.41)< 0.0012.59 (1.82–3.71)< 0.001CV death Continuous variable per SD1.56 (1.35–1.79)< 0.0011.55 (1.34–1.79)< 0.0011.54 (1.33–1.78)< 0.001 Tertilea< 0.001< 0.001< 0.001 T1 (n = 338)51.42Ref.Ref.Ref. T2 (n = 338)79.651.55 (1.00-2.40)0.0481.47 (0.95–2.28)0.0851.48 (0.94–2.32)0.084 T3 (n = 339)125.472.45 (1.63–3.68)< 0.0012.32 (1.54–3.49)< 0.0012.43 (1.59–3.72)< 0.001HR hazard ratio, CI confidence interval, TyG–BRI triglyceride glucose–body roundness index, SD standard deviation, CV death cardiovascular deathModel 1: adjusted for age, gender, smoking history, mean arterial pressure and heart rateModel 2: adjusted for Model 1 + NYHA classification, hemoglobin, AST, creatinine, sodium, TC, HDL-C, LDL, NT-proBNP, LVEF, ACEI/ARB/ARNI, beta-blocker, mineralocorticoid antagonists, diuretics, digoxin, SGLT2 inhibitors and other hypoglycemic therapyaTyG–BRI: T1 (< 24.81), T2 (24.81–35.75), T3 (≥ 35.75) Table 2HRs (95% CI) of primary outcomes according to TyG–BRI tertilesCategoriesIncidence/1000 person-yUnadjustedModel 1Model 2HR (95% CI)P-valueP for trendHR (95% CI)P-valueP for trendHR (95% CI)P-valueP for trendAll-cause death Continuous variable per SD1.55 (1.38–1.74)< 0.0011.54 (1.37–1.73)< 0.0011.54 (1.36–1.74)< 0.001 Tertilea< 0.001< 0.001< 0.001 T1 (n = 338)72.59Ref.Ref.Ref. T2 (n = 338)119.471.65 (1.15–2.37)0.0071.56 (1.09–2.25)0.0161.58 (1.09–2.30)0.015 T3 (n = 339)184.812.57 (1.83–3.61)< 0.0012.43 (1.72–3.41)< 0.0012.59 (1.82–3.71)< 0.001CV death Continuous variable per SD1.56 (1.35–1.79)< 0.0011.55 (1.34–1.79)< 0.0011.54 (1.33–1.78)< 0.001 Tertilea< 0.001< 0.001< 0.001 T1 (n = 338)51.42Ref.Ref.Ref. T2 (n = 338)79.651.55 (1.00-2.40)0.0481.47 (0.95–2.28)0.0851.48 (0.94–2.32)0.084 T3 (n = 339)125.472.45 (1.63–3.68)< 0.0012.32 (1.54–3.49)< 0.0012.43 (1.59–3.72)< 0.001HR hazard ratio, CI confidence interval, TyG–BRI triglyceride glucose–body roundness index, SD standard deviation, CV death cardiovascular deathModel 1: adjusted for age, gender, smoking history, mean arterial pressure and heart rateModel 2: adjusted for Model 1 + NYHA classification, hemoglobin, AST, creatinine, sodium, TC, HDL-C, LDL, NT-proBNP, LVEF, ACEI/ARB/ARNI, beta-blocker, mineralocorticoid antagonists, diuretics, digoxin, SGLT2 inhibitors and other hypoglycemic therapyaTyG–BRI: T1 (< 24.81), T2 (24.81–35.75), T3 (≥ 35.75) HRs (95% CI) of primary outcomes according to TyG–BRI tertiles HR hazard ratio, CI confidence interval, TyG–BRI triglyceride glucose–body roundness index, SD standard deviation, CV death cardiovascular death Model 1: adjusted for age, gender, smoking history, mean arterial pressure and heart rate Model 2: adjusted for Model 1 + NYHA classification, hemoglobin, AST, creatinine, sodium, TC, HDL-C, LDL, NT-proBNP, LVEF, ACEI/ARB/ARNI, beta-blocker, mineralocorticoid antagonists, diuretics, digoxin, SGLT2 inhibitors and other hypoglycemic therapy aTyG–BRI: T1 (< 24.81), T2 (24.81–35.75), T3 (≥ 35.75) Prognostic implications of TyG–BRI across nutritional-status strata Exploratory analyses stratified by MNA-SF indicated that the prognostic value of the TyG–BRI was confined to participants with impaired nutritional status (malnourished or at risk). In this stratum, Kaplan–Meier curves separated clearly (all log-rank P < 0.05), whereas no separation was observed in the normal nutritional status group (log-rank P > 0.05; Additional File 1: Figs. S1–S2). RCS analyses were concordant: the overall association between the TyG–BRI and mortality was significant in the impaired group (P < 0.05) but not in the normal group (P > 0.05; Additional File 1: Figs. S3–S4). Multivariable Cox models further supported these findings; after covariate adjustment, higher TyG–BRI remained significantly associated with adverse outcomes in the impaired group (P < 0.05) but not in the normal group (P > 0.05). The interaction between the TyG–BRI and nutritional status was statistically significant (P for interaction < 0.10; Fig. 5; Additional File 2: Table S1). In sensitivity analyses within the MNA-SF–defined strata, further adjustment for routinely measured baseline biomarkers closely related to nutritional and inflammatory status (serum albumin, lymphocyte count, and C-reactive protein), as well as lipid-lowering drugs, did not materially alter these associations, supporting the robustness of our finding (Additional File 2: Table S2). Fig. 5Forest plot of all-cause and cardiovascular death across tertiles of TyG–BRI stratified by nutritional status adjusted for model 2. Nutritional status was defined according to the MNA-SF: malnourished (≤ 7), at risk of malnutrition (8–11), and normal nutritional status (≥ 12). CV death cardiovascular death, TyG–BRI triglyceride glucose–body roundness index, MNA-SF Mini Nutritional Assessment-Short Form. Interaction was considered statistically significant at P < 0.1 Forest plot of all-cause and cardiovascular death across tertiles of TyG–BRI stratified by nutritional status adjusted for model 2. Nutritional status was defined according to the MNA-SF: malnourished (≤ 7), at risk of malnutrition (8–11), and normal nutritional status (≥ 12). CV death cardiovascular death, TyG–BRI triglyceride glucose–body roundness index, MNA-SF Mini Nutritional Assessment-Short Form. Interaction was considered statistically significant at P < 0.1 To assess robustness across clinical contexts, additional subgroup analyses were performed by major comorbidities. Results were consistent across methods—Kaplan–Meier, RCS, and Cox regression—demonstrating a stable, stepwise increase in risk with higher TyG–BRI (all P for trend < 0.05), with no material heterogeneity across subgroups (Additional File 1: Figs. S1–S6; Additional File 2: Table S3). Evaluating the incremental value of TyG–BRI for risk stratification The incremental predictive value of the TyG–BRI for mortality is shown in Table 3. Adding the TyG–BRI to the baseline risk model (MAGGIC score) [16] improved discrimination, increasing the C-statistic from 0.626 to 0.701 (P < 0.001). Reclassification also improved: NRI, 0.524 (95% CI, 0.377–0.672; P < 0.001); and IDI, 0.071 (95% CI, 0.052–0.091; P < 0.001). Compared with the TyG index or BRI alone, the TyG–BRI consistently provided better predictive performance (Table 3; Fig. 6; both P < 0.05 for differences). DCA demonstrated greater net benefit after adding the TyG–BRI to the baseline model, exceeding that of the TyG index or BRI alone (Additional File 1: Fig. S7). Table 3Evaluation of the incremental effect of adding TyG–BRI to the baseline risk model for predicting mortalityAUC (95% CI)P valueNRI (95% CI)P valueIDI (95% CI)P valueBaseline risk modela0.626 (0.587–0.666)Ref.-Ref.-Ref. + TyG–BRI0.701 (0.663–0.738)< 0.0010.524 (0.377–0.672)< 0.0010.071 (0.052–0.091)< 0.001 + TyG index0.661 (0.622-0.700)0.0140.264 (0.113–0.415)< 0.0010.025 (0.014–0.036)< 0.001 + BRI0.687 (0.650–0.726)< 0.0010.482 (0.334–0.631)< 0.0010.056 (0.039–0.072)< 0.001Pairwise comparison of AUC + TyG–BRI vs. + TyG index0.701 vs. 0.6610.033- - - - + TyG–BRI vs. + BRI0.701 vs. 0.6870.007- - - - + TyG index vs. + BRI0.661 vs. 0.6870.192- - - - TyG–BRI triglyceride glucose–body roundness index, TyG index triglyceride–glucose index, BRI body roundness index, AUC area under the receiver operating characteristic curve, CI confidence interval, NRI net reclassification improvement, IDI integrated discrimination improvementaThe MAGGIC score was used as the baseline risk model Table 3Evaluation of the incremental effect of adding TyG–BRI to the baseline risk model for predicting mortalityAUC (95% CI)P valueNRI (95% CI)P valueIDI (95% CI)P valueBaseline risk modela0.626 (0.587–0.666)Ref.-Ref.-Ref. + TyG–BRI0.701 (0.663–0.738)< 0.0010.524 (0.377–0.672)< 0.0010.071 (0.052–0.091)< 0.001 + TyG index0.661 (0.622-0.700)0.0140.264 (0.113–0.415)< 0.0010.025 (0.014–0.036)< 0.001 + BRI0.687 (0.650–0.726)< 0.0010.482 (0.334–0.631)< 0.0010.056 (0.039–0.072)< 0.001Pairwise comparison of AUC + TyG–BRI vs. + TyG index0.701 vs. 0.6610.033- - - - + TyG–BRI vs. + BRI0.701 vs. 0.6870.007- - - - + TyG index vs. + BRI0.661 vs. 0.6870.192- - - - TyG–BRI triglyceride glucose–body roundness index, TyG index triglyceride–glucose index, BRI body roundness index, AUC area under the receiver operating characteristic curve, CI confidence interval, NRI net reclassification improvement, IDI integrated discrimination improvementaThe MAGGIC score was used as the baseline risk model Evaluation of the incremental effect of adding TyG–BRI to the baseline risk model for predicting mortality TyG–BRI triglyceride glucose–body roundness index, TyG index triglyceride–glucose index, BRI body roundness index, AUC area under the receiver operating characteristic curve, CI confidence interval, NRI net reclassification improvement, IDI integrated discrimination improvement aThe MAGGIC score was used as the baseline risk model Fig. 6ROC curves evaluating the incremental effect of TyG–BRI beyond the baseline risk model. The MAGGIC score was used as the baseline risk model. ROC curve receiver operator characteristic curve, TyG–BRI triglyceride glucose–body roundness index, TyG index triglyceride–glucose index, BRI body roundness index, MAGGIC score Meta-analysis Global Group in Chronic Heart Failure score ROC curves evaluating the incremental effect of TyG–BRI beyond the baseline risk model. The MAGGIC score was used as the baseline risk model. ROC curve receiver operator characteristic curve, TyG–BRI triglyceride glucose–body roundness index, TyG index triglyceride–glucose index, BRI body roundness index, MAGGIC score Meta-analysis Global Group in Chronic Heart Failure score Mediating role of 6MWD in the association between TyG–BRI and mortality Using 6MWD as a marker of exercise tolerance, mediation analysis showed a significant indirect effect of the TyG–BRI on all-cause mortality (1.33; 95% CI, 0.28–2.40; P = 0.008) alongside a significant direct effect (7.61; 95% CI, 5.01–10.11; P < 0.001). The total effect was 9.16 (95% CI, 6.16–12.11; P < 0.001), with 14.6% of the association mediated through 6MWD (Fig. 7). Given the observational design of the present study, this mediation analysis should be considered exploratory and the findings interpreted with caution. Fig. 7Analysis of the mediation by 6MWD of the associations of TyG–BRI with all-cause mortality. Both TyG–BRI and 6MWD were z-standardized before analysis. 6MWD six-minute walk distance, TyG–BRI triglyceride glucose–body roundness index Analysis of the mediation by 6MWD of the associations of TyG–BRI with all-cause mortality. Both TyG–BRI and 6MWD were z-standardized before analysis. 6MWD six-minute walk distance, TyG–BRI triglyceride glucose–body roundness index Discussion This study is the first to construct and evaluate the composite indicator TyG–BRI—integrating the TyG index and BRI—to assess its prognostic significance in patients with HFpEF. The principal findings are as follows: (1) higher TyG–BRI levels were significantly associated with an increased risk of adverse outcomes; (2) the TyG–BRI provided incremental prognostic value for risk stratification and was superior to its individual components (the TyG index and BRI); (3) the prognostic effect of the TyG–BRI varied by nutritional status, being more evident among patients with impaired nutrition; and (4) mediation analysis indicated that exercise tolerance may partially mediate the association between the TyG–BRI and adverse outcomes. In summary, our findings suggest that the TyG–BRI may serve as a novel, pragmatic integrative risk marker in patients with HFpEF, rather than a comprehensive mechanistic explanation. HFpEF represents a growing public health burden, accounting for approximately half of all heart failure cases worldwide. Its prevalence continues to rise in parallel with population aging and the increasing incidence of hypertension, obesity, and type 2 diabetes [1]. The pathophysiology of HFpEF is multifactorial and involves systemic inflammation, myocardial fibrosis, endothelial dysfunction, metabolic dysregulation, and oxidative stress, which collectively contribute to impaired left ventricular relaxation. In addition, cardiovascular aging and multiorgan interactions further complicate disease progression, underscoring HFpEF as a complex clinical syndrome rather than a single cardiac disorder [3]. Given these heterogeneous mechanisms and diverse comorbidities, prognostic evaluation in HFpEF remains highly challenging yet clinically imperative for optimizing individualized management and improving outcomes. Among the multiple mechanisms involved, IR and obesity are two major, interrelated determinants of HFpEF. IR and the accompanying hyperinsulinemia promote myocardial hypertrophy, fibrosis, and endothelial dysfunction, thereby increasing the risk and progression of HFpEF. IR-related metabolic disturbances lead to concentric left ventricular remodeling and diastolic dysfunction, while activation of the MAPK pathway further enhances vasoconstriction and inflammation, forming a self-perpetuating cycle that accelerates disease progression [17]. Obesity, particularly central obesity, has also been recognized as a critical metabolic phenotype of HFpEF. Central adiposity is highly prevalent, affecting 77–96% of patients, and remains common even among those with BMI < 30 kg/m2, indicating an independent contribution beyond general obesity [18, 19]. Patients with central obesity exhibit higher rates of diabetes and dyslipidemia, elevated cardiac filling pressures, and impaired cardiopulmonary function [19]. Mechanistically, central obesity induces IR through inflammation and lipotoxicity, whereas IR aggravates fat accumulation, forming a vicious metabolic–obesity cycle that elevates cardiovascular risk [8, 20]. Together, these observations emphasize the intertwined metabolic nature of HFpEF and the importance of jointly considering obesity and IR in prognostic evaluation. In clinical research, surrogate indices derived from routinely measured clinical variables can effectively capture the degree of IR and central obesity. Among these, the TyG index has emerged as a reliable and convenient marker of IR and has been widely applied across diverse populations [21]. In a large cohort study including 138,620 participants, Xu et al. demonstrated a significant positive association between the TyG index and the risk of heart failure [22]. Specifically, individuals in the highest quartile of the TyG index had a 24% higher risk of developing heart failure compared with those in the lowest quartile (HR, 1.24; 95% CI, 1.07–1.44), and this association remained robust after multivariable adjustment. In addition, the BR. I—an anthropometric indicator integrating multiple body measurements to primarily reflect central obesity—has also been extensively investigated in clinical research [23]. In a cohort of 10,527 individuals with metabolic syndrome, the BRI was significantly associated with the prevalence of cardiovascular disease and mortality, showing a U-shaped relationship. When the BRI exceeded the threshold of 6.89, each additional unit was associated with a 9% higher risk of CV death and an 8% higher risk of all-cause death [24]. Several studies have explored the combined use of metabolic and anthropometric indices in disease prognosis—particularly in nephrology, diabetes, stroke, and geriatric medicine—showing potential advantages [8, 9, 25, 26]. However, evidence remains limited for HFpEF, in which both metabolic disturbance and central adiposity play pivotal roles [17, 19]. To address this gap, we developed a composite indicator—TyG–BRI—and comprehensively evaluated its prognostic value in patients with HFpEF. Our findings show that higher TyG–BRI levels were significantly associated with an increased risk of adverse outcomes (HR per SD increase, 1.54; 95% CI, 1.36–1.74 for all-cause death; HR per SD increase, 1.54; 95% CI, 1.33–1.78 for CV death). This combined approach may help capture key aspects of both metabolic dysregulation and fat accumulation, providing a more integrated—but still partial—assessment of cardiovascular risk rather than a comprehensive mechanistic explanation. As IR (indicated by a higher TyG index) and central obesity (indicated by a higher BRI) act synergistically [8], their integration provides novel insight into their joint contribution to adverse outcomes in HFpEF. Nutritional status not only plays a pivotal role in the prognostic evaluation of HFpEF but is also closely interrelated with IR and obesity [10, 27, 28]. Therefore, we further investigated whether the prognostic effect of the TyG–BRI differed across nutritional-status categories. Based on the MNA-SF score, patients were categorized into groups reflecting different levels of nutritional status [29]. Notably, the prognostic value of the TyG–BRI was significant only among those with impaired nutrition, but not among individuals with normal nutritional status. This may be attributable to the frequent coexistence of malnutrition with systemic inflammation, oxidative stress, and muscle wasting, which can exacerbate IR and central adiposity–related injury [30, 31]. Conversely, preserved metabolic reserves in well-nourished individuals may buffer these detrimental effects, attenuating the predictive power of the TyG–BRI. In the incremental predictive analysis, integrating the TyG–BRI into the baseline model significantly improved prognostic performance. Moreover, the magnitude of improvement exceeded that achieved by its individual components, supporting the superior predictive value of the TyG–BRI as a novel composite metabolic–anthropometric indicator in patients with HFpEF. Given that exercise capacity is not only an established prognostic determinant in HFpEF but also closely related to obesity and metabolic health [32–34], we further included exercise tolerance as a potential mediator in the pathway linking the TyG–BRI to adverse outcomes. Mediation analysis revealed that exercise tolerance partially mediated the association between the TyG–BRI and mortality, suggesting that impaired metabolic and anthropometric profiles may contribute to poor prognosis partly through reduced functional capacity. These findings highlight the potential clinical relevance of the TyG–BRI beyond its metabolic implications and raise the hypothesis that interventions targeting metabolic optimization and improvement of exercise capacity could potentially mitigate the risk of adverse outcomes in HFpEF. However, given the observational nature of our study, these results should be regarded as hypothesis-generating rather than causal evidence. Strengths and limitations This study has several strengths. First, its multicenter prospective design enhances the robustness and representativeness of the findings. Second, it is the first to evaluate the prognostic value of the composite index TyG–BRI in patients with HFpEF. Third, the predictive superiority of the TyG–BRI was consistently observed across subgroups and further confirmed through model-improvement analyses, underscoring its incremental prognostic value. Several limitations should be acknowledged. First, although this was a multicenter prospective cohort, the number of participating centers was limited and all patients were recruited from hospital-based settings, so selection bias cannot be excluded. Moreover, serial echocardiographic data during follow-up were not systematically collected, precluding evaluation of longitudinal changes in LVEF. Second, IR and central adiposity were not directly quantified by reference or imaging-based methods, so the TyG–BRI should be regarded as a surrogate index that only partially reflects the underlying metabolic and adipose-tissue abnormalities. Third, as an observational study, residual and unmeasured confounding cannot be excluded, no causal inference can be drawn—particularly regarding the relationships among the TyG–BRI, exercise capacity, and adverse outcomes—and the mediation analysis should therefore be interpreted as exploratory and hypothesis-generating. Finally, the pathophysiology of HFpEF is inherently complex and involves multiple mechanisms beyond metabolic burden and central adiposity; accordingly, the TyG–BRI should be viewed as a pragmatic integrative risk marker rather than a comprehensive mechanistic indicator, and further mechanistic and prospective studies are warranted. Conclusions In patients with HFpEF, the composite TyG–BRI was independently associated with adverse outcomes, with stronger prognostic relevance among those with impaired nutritional status, and provided modest incremental prognostic information beyond established risk factors. These findings suggest that the TyG–BRI may serve as a simple integrative marker to aid risk stratification in HFpEF, but its clinical utility requires further validation in external cohorts. Supplementary Information Supplementary Material 1: Fig S1. Kaplan–Meier curves for all-cause mortality by TyG–BRI tertiles across subgroups. TyG–BRI triglyceride-glucose body roundness index. Fig S2. Kaplan–Meier curves for CV mortality by TyG–BRI tertiles across subgroups. TyG–BRI triglyceride-glucose body roundness index, CV death cardiovascular death. Fig S3. HRs for all-cause mortality across subgroups by TyG–BRI from restricted cubic spline analyses adjusted for Model 2. TyG–BRI triglyceride-glucose body roundness index. Fig S4. HRs for CV death across subgroups by TyG–BRI from restricted cubic spline analyses adjusted for Model 2. CV death cardiovascular death. Fig S5. Forest plot of all-cause mortality by TyG–BRI tertiles stratified by comorbidities adjusted for Model 2. Fig S6. Forest plot of CV death by TyG–BRI tertiles stratified by comorbidities adjusted for Model 2. Fig S7. Decision curve analysis (DCA) comparing the baseline model with TyG–BRI, TyG index, and BRI for mortality prediction. The MAGGIC score was used as the baseline risk model. TyG–BRI triglyceride glucose–body roundness index, TyG index triglyceride–glucose index, BRI body roundness index. Supplementary Material 1: Fig S1. Kaplan–Meier curves for all-cause mortality by TyG–BRI tertiles across subgroups. TyG–BRI triglyceride-glucose body roundness index. Fig S2. Kaplan–Meier curves for CV mortality by TyG–BRI tertiles across subgroups. TyG–BRI triglyceride-glucose body roundness index, CV death cardiovascular death. Fig S3. HRs for all-cause mortality across subgroups by TyG–BRI from restricted cubic spline analyses adjusted for Model 2. TyG–BRI triglyceride-glucose body roundness index. Fig S4. HRs for CV death across subgroups by TyG–BRI from restricted cubic spline analyses adjusted for Model 2. CV death cardiovascular death. Fig S5. Forest plot of all-cause mortality by TyG–BRI tertiles stratified by comorbidities adjusted for Model 2. Fig S6. Forest plot of CV death by TyG–BRI tertiles stratified by comorbidities adjusted for Model 2. Fig S7. Decision curve analysis (DCA) comparing the baseline model with TyG–BRI, TyG index, and BRI for mortality prediction. The MAGGIC score was used as the baseline risk model. TyG–BRI triglyceride glucose–body roundness index, TyG index triglyceride–glucose index, BRI body roundness index. Supplementary Material 2: Table S1. HRs of primary outcomes by TyG–BRI tertiles across MNA-SF–defined subgroups. Table S2. HRs of primary outcomes by TyG–BRI tertiles across MNA-SF–defined subgroups in sensitivity analyses. Table S3. HRs of primary outcomes according to TyG–BRI tertiles across subgroups stratified by comorbidity status. Supplementary Material 2: Table S1. HRs of primary outcomes by TyG–BRI tertiles across MNA-SF–defined subgroups. Table S2. HRs of primary outcomes by TyG–BRI tertiles across MNA-SF–defined subgroups in sensitivity analyses. Table S3. HRs of primary outcomes according to TyG–BRI tertiles across subgroups stratified by comorbidity status.