Triglyceride-glucose index and all-cause mortality in patients with chronic severe heart failure

Abstract

BACKGROUND

Chronic severe heart failure is associated with high long-term mortality. The triglyceride-glucose (TyG) index is a simple surrogate marker of insulin resistance, but its prognostic significance in patients with chronic severe heart failure remains unclear.

AIM

To evaluate the association between the TyG index and 5-year all-cause mortality in patients with chronic severe heart failure.

METHODS

This retrospective cohort study included 261 consecutive patients with chronic severe heart failure hospitalized between January 2011 and June 2013. Patients were stratified into low- and high-TyG groups according to the median TyG index of 6.885. The primary endpoint was 5-year all-cause mortality. Survival was assessed using Kaplan-Meier curves and compared with the log-rank test. Cox regression models were used to evaluate the association between the TyG index and mortality. Subgroup analyses and receiver operating characteristic curve analyses were performed.

RESULTS

During 5 years of follow-up, 135 patients died. Patients in the high-TyG group had significantly higher 5-year all-cause mortality than those in the low-TyG group (59.5% vs 43.8%, P = 0.011). Kaplan-Meier analysis showed significantly poorer survival in the high-TyG group. In the fully adjusted Cox model, each 1-unit increase in the TyG index was associated with an increased risk of 5-year all-cause mortality [hazard ratio = 1.33, 95% confidence interval (CI): 1.01-1.76, P = 0.041]. This association was generally consistent across clinically relevant subgroups, with no significant interaction by diabetes status. The TyG index alone showed modest discriminative ability for 5-year mortality prediction [area under the curve (AUC) = 0.577, 95%CI: 0.507-0.646], whereas N-terminal pro-brain natriuretic peptide (NT-proBNP) showed better performance (AUC = 0.713, 95%CI: 0.651-0.775). Adding the TyG index to NT-proBNP resulted in only a small, non-significant increase in AUC (AUC = 0.722; ΔAUC = 0.009; DeLong P = 0.318).

CONCLUSION

In this single-center retrospective cohort of patients with chronic severe heart failure, a higher TyG index was independently associated with increased 5-year all-cause mortality. However, its standalone discriminative ability was limited, and its incremental value beyond NT-proBNP was small. The TyG index may serve as an adjunctive metabolic risk marker, but further prospective validation is required.

Key Words: Triglyceride-glucose index; Chronic severe heart failure; All-cause mortality; Insulin resistance; N-terminal pro-brain natriuretic peptide; Prognosis

Core Tip: In patients with chronic severe heart failure, the triglyceride-glucose (TyG) index independently predicts 5-year mortality but has only modest discriminative ability (area under the curve = 0.577). It adds minimal prognostic value beyond N-terminal pro-brain natriuretic peptide (NT-proBNP) and should not replace it in well-resourced settings. However, in resource-limited settings where NT-proBNP is unavailable, the TyG index may serve as a low-cost metabolic alert signal.

INTRODUCTION

Heart failure is a major global public health problem and is associated with substantial morbidity, frequent hospitalization, impaired quality of life, and high mortality. The prevalence of heart failure increases markedly with age, and its clinical burden remains considerable despite advances in pharmacological and device-based therapies[1]. Chronic severe heart failure represents an advanced stage of cardiovascular disease and is characterized by severe symptoms, impaired cardiac function, recurrent hospitalization, and poor long-term prognosis. Accurate risk stratification in this population remains clinically important.

Established prognostic markers in heart failure include age, renal function, left ventricular ejection fraction (LVEF), natriuretic peptides, inflammatory status, nutritional status, and use of guideline-directed medical therapy. N-terminal pro-brain natriuretic peptide (NT-proBNP) is one of the most widely used biomarkers for diagnosis, risk assessment, and prognosis in patients with heart failure[2]. However, heart failure is not only a hemodynamic disorder but also a systemic syndrome involving metabolic, inflammatory, neurohormonal, and renal abnormalities. Therefore, simple and inexpensive biomarkers reflecting metabolic dysfunction may provide additional information for risk assessment.

Insulin resistance has increasingly been recognized as an important contributor to cardiovascular disease and heart failure progression[3]. It may promote endothelial dysfunction, oxidative stress, chronic inflammation, neurohormonal activation, impaired myocardial substrate utilization, and adverse ventricular remodeling[4-7]. These mechanisms are highly relevant to the pathophysiology of advanced heart failure, in which metabolic remodeling and impaired energy utilization are common features[4,5].

The triglyceride-glucose (TyG) index, calculated from fasting triglyceride and fasting glucose levels, was originally proposed as a simple surrogate marker of insulin resistance[8]. Compared with direct insulin resistance measurements, the TyG index is inexpensive, readily available, and easy to calculate using routine laboratory tests. Previous studies have summarized the role, strengths, and limitations of the TyG index in cardiometabolic diseases[9-11]. The TyG index has also been associated with visceral obesity, stroke, coronary artery disease, repeat revascularization, in-stent restenosis, and adverse cardiovascular outcomes[12-17]. More recent evidence suggests that the TyG index may be associated with the risk of incident heart failure and worsening heart failure[18-20].

However, evidence focusing specifically on the prognostic significance of the TyG index in patients with chronic severe heart failure remains limited. Given the high mortality and distinct metabolic stress in this population, we aimed to investigate whether the TyG index is independently associated with 5-year all-cause mortality in patients with chronic severe heart failure. We further evaluated its predictive performance alone and in combination with NT-proBNP and explored whether the association was consistent across clinically relevant subgroups.

MATERIALS AND METHODS

Study population

This retrospective cohort study enrolled consecutive patients with chronic severe heart failure admitted to Beijing Chaoyang Hospital between January 2011 and June 2013. Chronic severe heart failure was defined according to the Chinese guidelines for the diagnosis and treatment of heart failure and required New York Heart Association functional class III or IV symptoms, evidence of cardiac dysfunction, and elevated NT-proBNP levels[2]. After applying the inclusion and exclusion criteria, 261 patients were included in the final analysis.

The exclusion criteria were severe concurrent non-cardiovascular illness, systemic hematological or immune disorders, severe mental illness, malignancy, acute infection at admission, or incomplete baseline or follow-up data.

Ethics statement

The study was reviewed and approved by the Ethics Committee of Beijing Chaoyang Hospital. Because this was a retrospective study based on routinely collected medical records and involved no direct patient contact, the requirement for informed consent was waived by the ethics committee. Patient privacy was protected throughout the study. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Data collection and definitions

Baseline demographic characteristics, medical history, vital signs, laboratory parameters, echocardiographic indices, and medication use were collected from electronic medical records. Laboratory tests were performed within 24 hours after admission, and echocardiography was performed within 48 hours after admission.

The TyG index was calculated as ln[fasting triglyceride (mg/dL) × fasting glucose (mg/dL)/2]. Patients were divided into low- and high-TyG groups according to the median TyG index of 6.885.

Estimated glomerular filtration rate was calculated using standard clinical equations. Diabetes mellitus, hypertension, hyperlipidemia, prior percutaneous coronary intervention (PCI), and prior coronary artery bypass grafting (CABG) were defined according to documented medical history, previous diagnosis, medication use, or discharge diagnosis.

Follow-up and endpoint

The follow-up period was 5 years. The primary endpoint was 5-year all-cause mortality. Cardiac death was defined as death attributed to heart failure progression, fatal arrhythmia, acute myocardial infarction, sudden cardiac death, or other cardiovascular causes documented in medical records or follow-up records. Survival status was obtained from medical records, outpatient visits, or telephone follow-up. Patients lost to follow-up were censored at the date of last contact.

Statistical analysis

Continuous variables were expressed as mean ± SD or median with interquartile range, depending on data distribution. Categorical variables were expressed as n (%). Between-group comparisons were performed using Student’s t-test or the Mann-Whitney U test for continuous variables and the χ² test or Fisher’s exact test for categorical variables, as appropriate.

Kaplan-Meier survival curves were generated to compare 5-year survival between the low- and high-TyG groups, and differences were assessed using the log-rank test.

Cox proportional hazards regression analysis was used to evaluate the association between the TyG index and 5-year all-cause mortality. Univariable Cox regression was first performed to identify variables associated with mortality. Multivariable Cox regression models were then constructed using clinically relevant covariates and variables associated with mortality in univariable analysis. The fully adjusted model included age, sex, body mass index (BMI), diabetes mellitus, hypertension, estimated glomerular filtration rate, LVEF, NT-proBNP, angiotensin-converting enzyme inhibitor angiotensin receptor blocker (ACEI/ARB) use, and beta-blocker use. NT-proBNP was entered into the Cox models as a standardized continuous variable, and hazard ratios for NT-proBNP were expressed per standard deviation increase. The TyG index was analyzed both as a continuous variable and as a categorical variable according to the median value.

Subgroup analyses were performed according to diabetes status, age, sex, and renal function. Interaction terms were tested to evaluate potential effect modification.

Receiver operating characteristic curve (ROC) analysis was used to evaluate the discriminative ability of the TyG index, NT-proBNP, and their combination for predicting 5-year all-cause mortality. Because follow-up completeness was high and the endpoint was assessed at a fixed 5-year time point, conventional ROC analysis was used as an exploratory assessment of discrimination. Area under the curve (AUC) were compared using the DeLong test. A two-sided P value < 0.05 was considered statistically significant.

RESULTS

Study population and follow-up

A total of 261 patients with chronic severe heart failure were included in the final analysis. The mean age was 67.8 ± 11.6 years, and 165 patients were male. The median TyG index was 6.885; therefore, 130 patients were assigned to the low-TyG group and 131 patients to the high-TyG group. During the 5-year follow-up period, 135 patients died, corresponding to an overall 5-year all-cause mortality rate of 51.7%. Follow-up was complete for 257 patients, while 4 patients were lost to follow-up and censored at the last available contact.

Baseline characteristics according to TyG index

Baseline clinical characteristics according to TyG index are shown in Table 1. Patients in the high-TyG group had a higher prevalence of diabetes mellitus and hyperlipidemia than those in the low-TyG group. The high-TyG group also had higher BMI, hemoglobin, total cholesterol, low-density lipoprotein cholesterol (LDL-C), and glycated hemoglobin (HbA1c) levels. There were no significant differences between the two groups in age, sex, smoking status, alcohol consumption, hypertension, cerebral infarction, prior PCI, prior CABG, blood pressure, inflammatory markers, renal function, NT-proBNP, LVEF, or baseline use of ACEI/ARB, beta-blockers, and digoxin.

Table 1 Baseline clinical characteristics according to triglyceride-glucose index, n (%).

Variable	Low TyG index (n = 130)	High TyG index (n = 131)	P value
Age, years	67.5 ± 11.8	68.1 ± 11.4	0.675
Age ≥ 65 years	74 (56.9)	77 (58.8)	0.761
Male sex	83 (63.8)	82 (62.6)	0.834
Smokin	73 (56.2)	68 (51.9)	0.580
Alcohol consumption	29 (22.3)	32 (24.4)	0.475
Hypertension	85 (65.4)	97 (74.0)	0.115
Diabetes mellitus	50 (38.5)	69 (52.7)	0.021
Hyperlipidemia	63 (48.5)	81 (61.8)	0.030
Cerebral infarction	16 (12.3)	24 (18.3)	0.178
Prior PCI	24 (18.5)	27 (20.6)	0.662
Prior CABG	15 (11.5)	14 (10.7)	0.827
BMI, kg/m2	24.2 ± 3.9	25.6 ± 5.2	0.012
Heart rate, bpm	80.0 (70.0-98.2)	81.0 (67.0-95.0)	0.326
SBP, mmHg	130.0 (118.7-150.0)	130.0 (120.0-150.0)	0.162
DBP, mmHg	80.0 (70.0-90.0)	80.0 (70.0-90.0)	0.100
White blood cell count, 109/L	6.41 (5.22-7.75)	7.02 (5.74-8.45)	0.127
Hemoglobin, g/L	119.5 ± 22.2	125.1 ± 21.7	0.041
NEUT%, %	66.0 ± 10.6	67.2 ± 10.6	0.399
hs-CRP, mg/L	6.45 (2.05-12.33)	8.13 (3.01-13.41)	0.593
AST, U/L	23 (19-32)	22 (17-31)	0.365
ALT, U/L	21 (17-31)	21 (17-27)	0.326
Uric acid, μmol/L	382.8 (296.8-484.4)	388.7 (305.7-481.7)	0.877
eGFR, mL/minute/1.73 m2	56.8 (35.2-75.0)	61.8 (38.3-83.5)	0.148
TC, mmol/L	3.5 (3.1-4.0)	4.1 (3.6-4.9)	< 0.001
LDL-C, mmol/L	1.88 (1.48-2.33)	2.36 (1.89-2.88)	< 0.001
HDL-C, mmol/L	1.1 (0.9-1.3)	1.0 (0.9-1.2)	0.313
HbA1c, %	6.3 (5.8-7.0)	6.7 (6.2-7.8)	< 0.001
NT-proBNP, pg/mL	6014.0 (2883.2-12896.7)	5319.5 (3277.0-12287.2)	0.695
LVEF, %	43.0 (35.0-60.0)	45.0 (32.0-63.0)	0.653
ACEI/ARB	72 (55.4)	81 (61.8)	0.290
Beta-blocker	79 (60.8)	76 (58.0)	0.651
Digoxin	75 (57.6)	65 (49.6)	0.191

Values are presented as mean ± SD, median (interquartile range).

TyG: Triglyceride-glucose; PCI: Percutaneous coronary intervention; CABG: Coronary artery bypass grafting; BMI: Body mass index; SBP: Systolic blood pressure; DBP: Diastolic blood pressure; NEUT%: Neutrophil percentage; hs-CRP: High-sensitivity C-reactive protein; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; eGFR: Estimated glomerular filtration rate; TC: Total cholesterol; LDL-C: Low-density lipoprotein cholesterol; HDL-C: High-density lipoprotein cholesterol; HbA1c: Glycated hemoglobin; NT-proBNP: N-terminal pro-brain natriuretic peptide; LVEF: Left ventricular ejection fraction; ACEI: Angiotensin-converting enzyme inhibitor; ARB: Angiotensin receptor blocker.

Clinical outcomes according to TyG index

Clinical outcomes according to TyG index are shown in Table 2. The high-TyG group had significantly higher mortality than the low-TyG group. The 1-year all-cause mortality rate was 24.4% in the high-TyG group and 11.5% in the low-TyG group (P = 0.007). The 5-year all-cause mortality rate was also significantly higher in the high-TyG group (59.5% vs 43.8%, P = 0.011). Similarly, 5-year cardiac mortality was higher in the high-TyG group than in the low-TyG group (52.7% vs 36.2%, P = 0.007).

Table 2 Clinical outcomes according to triglyceride-glucose index, n (%).

Outcome	Low TyG index (n = 130)	High TyG index (n = 131)	P value
1-year all-cause mortality	15 (11.5)	32 (24.4)	0.007
5-year all-cause mortality	57 (43.8)	78 (59.5)	0.011
5-year cardiac mortality	47 (36.2)	69 (52.7)	0.007

TyG: Triglyceride-glucose.

Kaplan-Meier survival analysis demonstrated significantly poorer 5-year survival in patients with a higher TyG index than in those with a lower TyG index (log-rank, P = 0.011; Figure 1).

Figure 1 Kaplan-Meier survival curves for patients with low vs high triglyceride-glucose index. The high-triglyceride-glucose group showed significantly worse survival (Log-rank P = 0.011). TyG: Triglyceride-glucose.

Cox regression analysis

In univariable Cox regression analysis, age ≥ 65 years, BMI, hypertension, diabetes mellitus, prior PCI, prior CABG, diastolic blood pressure, hemoglobin, neutrophil percentage, high-sensitivity C-reactive protein, estimated glomerular filtration rate, HbA1c, NT-proBNP, ACEI/ARB use, beta-blocker use, and the TyG index were associated with 5-year all-cause mortality (Table 3).

Table 3 Univariable Cox regression analysis for 5-year all-cause mortality.

Variable	HR	95%CI	P value
Age ≥ 65 years	1.92	1.35-2.73	< 0.001
Male sex	0.96	0.68-1.36	0.818
BMI, per 1 kg/m2	0.91	0.87-0.95	< 0.001
Hypertension	1.42	1.00-2.01	0.048
Diabetes mellitus	1.78	1.27-2.50	0.001
Hyperlipidemia	1.24	0.88-1.75	0.214
Prior PCI	1.48	1.00-2.19	0.049
Prior CABG	1.76	1.08-2.87	0.023
DBP, per 1 mmHg	0.985	0.972-0.998	0.024
Hemoglobin, per 10 g/L	0.83	0.76-0.91	< 0.001
NEUT%, per 1%	1.035	1.018-1.052	< 0.001
hs-CRP, per 1 mg/L	1.018	1.002-1.034	0.029
eGFR, per 10 mL/minute/1.73 m2	0.84	0.78-0.91	< 0.001
HbA1c, per 1%	1.15	1.02-1.30	0.026
LVEF, per 5%	0.96	0.89-1.03	0.245
NT-proBNP, per SD increase	1.71	1.33-2.21	< 0.001
ACEI/ARB use	0.67	0.47-0.95	0.026
Beta-blocker use	0.64	0.45-0.91	0.013
TyG index, per 1-unit increase	1.46	1.14-1.87	0.003
High TyG group	1.52	1.08-2.13	0.016

PCI: Percutaneous coronary intervention; CABG: Coronary artery bypass grafting; BMI: Body mass index; DBP: Diastolic blood pressure; NEUT%: Neutrophil percentage; hs-CRP: High-sensitivity C-reactive protein; eGFR: Estimated glomerular filtration rate; HbA1c: Glycated hemoglobin; NT-proBNP: N-terminal pro-brain natriuretic peptide; LVEF: Left ventricular ejection fraction; ACEI: Angiotensin-converting enzyme inhibitor; ARB: Angiotensin receptor blocker; HR: Hazard ratio; CI: Confidence interval; TyG: Triglyceride-glucose.

In multivariable Cox regression models, the TyG index remained independently associated with 5-year all-cause mortality. In the fully adjusted model including age, sex, BMI, diabetes mellitus, hypertension, estimated glomerular filtration rate, LVEF, NT-proBNP, ACEI/ARB use, and beta-blocker use, each 1-unit increase in the TyG index was associated with a 33% higher risk of all-cause mortality [hazard ratio (HR) = 1.33, 95% confidence interval (CI): 1.01-1.76, P = 0.041] (Table 4).

Table 4 Multivariable Cox regression models for the association between triglyceride-glucose index and 5-year all-cause mortality.

Model	Adjustment	n	Events	HR	95%CI	P value
Model 1	Unadjusted	261	135	1.46	1.14-1.87	0.003
Model 2	Age, sex, BMI	261	135	1.41	1.09-1.83	0.009
Model 3	Model 2 + diabetes mellitus, hypertension, eGFR, LVEF	261	135	1.36	1.04-1.78	0.025
Model 4	Model 3 + NT-proBNP, ACEI/ARB use, beta-blocker use	261	135	1.33	1.01-1.76	0.041

Hazard ratio refer to each 1-unit increase in triglyceride-glucose index. In models including N-terminal pro-brain natriuretic peptide (NT-proBNP), NT-proBNP was entered as a standardized continuous variable. HR: Hazard ratio; CI: Confidence interval; BMI: Body mass index; eGFR: Estimated glomerular filtration rate; LVEF: Left ventricular ejection fraction; NT-proBNP: N-terminal pro-brain natriuretic peptide; ACEI: Angiotensin-converting enzyme inhibitor; ARB: Angiotensin receptor blocker.

When the TyG index was analyzed as a categorical variable according to the median value, the high-TyG group showed a higher risk of 5-year all-cause mortality than the low-TyG group in univariable Cox regression (HR = 1.52, 95%CI: 1.08-2.13, P = 0.016) (Table 3).

Subgroup analyses

Subgroup analyses were exploratory. The association between the TyG index and 5-year all-cause mortality was generally directionally consistent across diabetes status, age, sex, and renal function, with no statistically significant interactions observed.

In subgroup analysis stratified by diabetes status, the adjusted HR was 1.28 (95%CI: 0.91-1.80) in patients with diabetes mellitus and 1.31 (95%CI: 0.89-1.94) in those without diabetes mellitus. No significant interaction was observed between diabetes status and the TyG index (P for interaction = 0.684) (Table 5).

Table 5 Exploratory subgroup analyses for the association between triglyceride-glucose index and 5-year all-cause mortality.

Subgroup	n	Events	Adjusted HR	95%CI	P value	P for interaction
Diabetes mellitus	119	77	1.28	0.91-1.80	0.156	0.684
No diabetes mellitus	142	58	1.31	0.89-1.94	0.172
Age < 65 years	110	39	1.24	0.78-1.97	0.361	0.522
Age ≥ 65 years	151	96	1.39	1.01-1.92	0.044
Male	165	84	1.36	0.98-1.89	0.067	0.744
Female	96	51	1.29	0.82-2.04	0.269
eGFR < 60 mL/minute/1.73 m2	137	86	1.42	1.01-2.01	0.046	0.438
eGFR ≥ 60 mL/minute/1.73 m2	124	49	1.21	0.79-1.86	0.382

Subgroup analyses were exploratory and were not adjusted for multiple comparisons. Adjusted hazard ratio refer to each 1-unit increase in triglyceride-glucose index. HR: Hazard ratio; CI: Confidence interval; eGFR: Estimated glomerular filtration rate.

Predictive performance

The TyG index alone showed modest discriminative ability for predicting 5-year all-cause mortality, with an AUC of 0.577 (95%CI: 0.507-0.646, P = 0.033). The optimal TyG cut-off value determined by the Youden index was 6.926, with a sensitivity of 58.5% and a specificity of 56.3%.

NT-proBNP showed better predictive performance than the TyG index, with an AUC of 0.713 (95%CI: 0.651-0.775, P < 0.001). The combination of the TyG index and NT-proBNP yielded an AUC of 0.722 (95%CI: 0.661-0.783, P < 0.001). Compared with NT-proBNP alone, the increase in AUC after adding the TyG index was small and not statistically significant by the DeLong test (ΔAUC = 0.009, P = 0.318) (Table 6 and Figure 2).

Figure 2 Receiver operating characteristic curves for the triglyceride-glucose index, N-terminal pro-brain natriuretic peptide, and their combination in predicting 5-year all-cause mortality. The combination of triglyceride-glucose index and N-terminal pro-brain natriuretic peptide (NT-proBNP) achieved the highest area under the curve (0.722), but the improvement compared with NT-proBNP alone was not statistically significant by the DeLong test (P = 0.318). TyG: Triglyceride-glucose; NT-proBNP: N-terminal pro-brain natriuretic peptide; AUC: Area under the curve; ROC: Receiver operating characteristic.

Table 6 Receiver operating characteristic analysis for prediction of 5-year all-cause mortality.

Model	AUC	95%CI	Sensitivity (%)	Specificity (%)	Youden index	P vs 0.5	DeLong P vs NT-proBNP
TyG index	0.577	0.507-0.646	58.5	56.3	0.148	0.033	-
NT-proBNP	0.713	0.651-0.775	68.9	70.2	0.391	< 0.001	Reference
TyG index + NT-proBNP	0.722	0.661-0.783	66.7	76.2	0.429	< 0.001	0.318

AUC: Area under the curve; CI: Confidence interval; TyG: Triglyceride-glucose; NT-proBNP: N-terminal pro-brain natriuretic peptide.

DISCUSSION

In this retrospective cohort study of 261 patients with chronic severe heart failure, we evaluated the association between the TyG index and 5-year all-cause mortality. The main findings were as follows. First, patients with a higher TyG index had significantly higher 1-year and 5-year all-cause mortality than those with a lower TyG index. Second, the TyG index remained independently associated with 5-year all-cause mortality after adjustment for clinically relevant confounders. Third, subgroup analyses showed generally consistent associations across clinically relevant subgroups, with no significant interaction by diabetes status. Fourth, although the TyG index was statistically associated with mortality, its standalone predictive ability was modest, and adding it to NT-proBNP produced only a small and statistically non-significant improvement in AUC.

Biological plausibility

The association between elevated TyG index and adverse outcomes in chronic severe heart failure is biologically plausible. The TyG index is widely regarded as a practical surrogate marker of insulin resistance[8-11]. Insulin resistance may contribute to cardiovascular disease and heart failure progression through multiple pathways, including endothelial dysfunction, oxidative stress, impaired nitric oxide signaling, chronic inflammation, neurohormonal activation, and abnormal myocardial substrate utilization[3,6,7]. In patients with advanced heart failure, metabolic remodeling is a central feature, with impaired myocardial energy production, altered fatty acid and glucose utilization, mitochondrial dysfunction, and systemic catabolic-anabolic imbalance[4,5].

In the present cohort, patients with a higher TyG index had a higher prevalence of diabetes mellitus and hyperlipidemia and higher BMI, HbA1c, total cholesterol, and LDL-C levels. These findings suggest that an elevated TyG index may reflect a broader metabolic risk phenotype rather than isolated glucose or lipid abnormalities. The independent association between the TyG index and mortality after multivariable adjustment suggests that metabolic dysfunction may provide prognostic information not fully captured by traditional heart failure indicators.

However, these mechanistic explanations should be interpreted cautiously. The present study did not directly measure insulin concentrations, HOMA-IR, myocardial metabolism, oxidative stress markers, or endothelial function. Therefore, the TyG index should be interpreted as a readily available clinical marker of metabolic burden rather than direct mechanistic evidence of insulin resistance-mediated cardiac injury.

Comparison with previous studies

Previous studies have reported associations between the TyG index and visceral obesity, stroke, coronary artery disease, repeat revascularization, in-stent restenosis, and adverse cardiovascular outcomes[12-17]. In a systematic review and meta-analysis, Liang et al[14] reported that the TyG index was associated with the risk, severity, and prognosis of coronary artery disease. Studies in patients undergoing PCI have also shown that elevated TyG index is associated with repeat revascularization, in-stent restenosis, and poor prognosis after myocardial infarction[15,17].

Emerging evidence also supports a relationship between the TyG index and heart failure. Li et al[18] reported that a higher TyG index was associated with increased risk of heart failure using data from two large cohorts and Mendelian randomization analysis. Zhang and Hou[19] found an association between the TyG index and heart failure in the NHANES population. More recently, Lai et al[20] reported that TyG index trajectories were associated with worsening heart failure in elderly patients with chronic heart failure and type 2 diabetes. These studies suggest that the TyG index may be relevant not only to atherosclerotic cardiovascular disease but also to the development and progression of heart failure.

Our study adds to the existing literature by focusing on patients with chronic severe heart failure, a high-risk population characterized by advanced symptoms and poor long-term prognosis. Unlike studies assessing incident heart failure in general populations, the present study evaluated long-term mortality among patients with established severe heart failure. The findings suggest that metabolic risk reflected by the TyG index may retain prognostic relevance even in advanced stages of heart failure.

Subgroup analyses and predictive value

The subgroup analyses showed no significant interaction by diabetes status, age, sex, renal function. In particular, the association between the TyG index and mortality was directionally consistent in patients with and without diabetes mellitus. This finding suggests that the prognostic relevance of the TyG index may not be entirely driven by diagnosed diabetes. However, these subgroup analyses were exploratory and limited by the modest sample size; therefore, they should be interpreted cautiously.

Although the TyG index was independently associated with mortality, its predictive value was limited. The AUC of the TyG index alone was 0.577, indicating modest discrimination. NT-proBNP remained a substantially stronger prognostic biomarker. Moreover, adding the TyG index to NT-proBNP increased the AUC by only 0.009, and this improvement was not statistically significant. Therefore, the TyG index should not be considered a replacement for established prognostic biomarkers. Instead, it may serve as an adjunctive marker reflecting metabolic risk burden.

Clinical implications

The TyG index is inexpensive, readily available, and easy to calculate using routine fasting triglyceride and glucose measurements[8,9]. These features make it potentially attractive as a simple metabolic risk marker, especially in settings where advanced biomarker testing may be unavailable. However, the present findings do not support using the TyG index as a standalone prognostic tool in patients with chronic severe heart failure. Its discriminative ability was modest, and its incremental value beyond NT-proBNP was small and statistically non-significant.

Therefore, NT-proBNP, clinical status, renal function, echocardiographic parameters, and guideline-directed medical therapy should remain central to risk assessment in patients with chronic severe heart failure[2]. The TyG index may provide additional information regarding metabolic risk burden, but it should not be used alone to guide major clinical decisions.

Causality

The present findings should be interpreted as evidence of association rather than causation. Although insulin resistance and metabolic dysfunction may play mechanistic roles in heart failure progression, the retrospective observational design of this study cannot determine whether an elevated TyG index directly contributes to mortality or merely reflects greater metabolic and systemic disease burden. Future prospective studies are needed to clarify whether interventions targeting metabolic dysfunction can improve outcomes in patients with chronic severe heart failure.

Limitations

Several limitations should be acknowledged. First, this was a single-center retrospective study with a modest sample size, which may limit the generalizability of the findings. Second, although we adjusted for several clinically relevant variables, residual confounding cannot be excluded. Data on lipid-lowering therapy, detailed heart failure etiology, medication adherence, dose titration, mineralocorticoid receptor antagonists, ARNI, SGLT2 inhibitors, and device therapy were not consistently available and therefore could not be fully incorporated into the analysis.

Third, the TyG index was measured only once at baseline. Dynamic changes in triglyceride, glucose, and insulin resistance during follow-up could not be assessed. Fourth, the optimal TyG cut-off value was derived from the same cohort using ROC analysis and should therefore be considered exploratory. External validation is required before this threshold can be applied clinically. Fifth, subgroup analyses were limited by the modest sample size and number of events and were not adjusted for multiple comparisons. Finally, the observational design precludes causal inference.

CONCLUSION

In this single-center retrospective cohort of patients with chronic severe heart failure, a higher TyG index was independently associated with increased 5-year all-cause mortality. However, the TyG index alone showed only modest discriminative ability, and its incremental value beyond NT-proBNP was small and statistically non-significant. The TyG index may reflect metabolic risk in this high-risk population, but it should be regarded as an adjunctive marker rather than a standalone prognostic tool. Further prospective multicenter studies are needed to validate these findings.