Prospective associations of triglyceride-glucose related indices with cardiovascular disease and mortality in individuals with metabolic syndrome: evidence from the UK biobank

Abstract

Background

Triglyceride‒glucose (TyG) related indices have been implicated in metabolic syndrome (MetS), cardiovascular disease (CVD), and mortality. However, their prospective associations with CVD and mortality among individuals with MetS remain limited.

Methods

In this large-scale prospective study, we included 113,699 UK Biobank participants with MetS who were free of CVD at baseline. The TyG index, TyG-body mass index (TyG-BMI), TyG-waist circumference (TyG-WC), and TyG-waist-to-height ratio (TyG-WHtR) were calculated and categorized into quartiles. Cox proportional hazards models, restricted cubic splines, two-piecewise regression models, and K-means clustering were applied to evaluate the associations of TyG-related indices with CVD and mortality outcomes. Incremental predictive performance was further assessed using the net reclassification index (NRI) and integrated discrimination improvement (IDI).

Results

Over a median follow-up of 13.7 years, higher quartiles of TyG-BMI, TyG-WC, and TyG-WHtR were positively associated with all CVD outcomes. Among these indices, TyG-WC demonstrated the strongest associations, particularly for total CVD (hazard ratios [HRs] for the highest vs. lowest quartile: 1.42 [95% CI: 1.36–1.48]). For mortality outcomes, TyG-WC also exhibited the most pronounced associations, showing an HR of 1.51 (1.32–1.72) for CVD mortality. In contrast, the TyG index alone showed inverse associations with both CVD incidence and CVD mortality. Dose–response analyses revealed that TyG-related indices were linearly associated with coronary heart disease (CHD), whereas U- or J-shaped nonlinear relationships were observed for other CVD outcomes, with risks rising steadily beyond the identified inflection points. In predictive analyses, TyG-WC provided the greatest incremental predictive value for total CVD, all-cause mortality, and CVD mortality, while TyG-WHtR yielded the largest improvements for CHD and stroke. Moreover, results from trajectory analyses showed that participants with persistently high TyG-related trajectories had substantially elevated risks of CVD. All results remained robust across subgroup and sensitivity analyses.

Conclusions

TyG-BMI, TyG-WC, and TyG-WHtR were robust predictors of CVD and mortality in individuals with MetS. These indices exhibited nonlinear threshold effects and improved predictive performance, supporting their utility in clinical risk stratification.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-025-03070-3.

Research Insights

What is currently known about this topic?

MetS represents a major global public health challenge. TyG-related indices have been associated with CVD and mortality in the general population or in individuals with specific metabolic disorders. However, evidence on their prognostic values for incident CVD in individuals with MetS remains limited.

What is the key research question?

Whether baseline levels and longitudinal trajectories of TyG-related indices predict CVD and mortality outcomes in individuals with MetS?

Which TyG-related indices provide the best predictive performance for CVD and mortality in individuals with MetS?

What is new?

Elevated baseline levels and persistently high trajectories of TyG-BMI, TyG-WC, and TyG-WHtR were prospectively associated with elevated risks of CVD and mortality in MetS patients. Nonlinear (U- or J-shaped) relationships were observed for all outcomes, except CHD. Moreover, TyG-WC and TyG-WHtR exhibit stronger associations with CVD and mortality and superior predictive performance compared with TyG-BMI.

How might this study influence clinical practice?

These findings highlight the potential utility of TyG-WC and TyG-WHtR in the early identification and dynamic risk stratification of individuals with MetS at a high risk of adverse CVD and mortality outcomes.

Introduction

Metabolic syndrome (MetS), defined by a cluster of interrelated abnormalities such as central adiposity, hypertension, dyslipidemia, and hyperglycemia, has emerged as a major global health concern [1]. Its prevalence has been steadily increasing worldwide, currently affecting over one-quarter of adults and increasing to worrisome levels among children and adolescents, with the likelihood of further rapid escalation [2, 3]. Notably, MetS has been demonstrated not only to increase the risk of major chronic diseases, including type 2 diabetes mellitus (T2DM), cardiovascular disease (CVD), and stroke, but also to contribute to premature death [3–5]. Given the well-established links between MetS and elevated risks of CVD and mortality, identifying reliable biomarkers that can predict these adverse outcomes is urgently needed to facilitate earlier prevention and management strategies.

Insulin resistance (IR), characterized by reduced sensitivity of insulin-responsive tissues, is a key pathogenic mechanism underlying major metabolic disorders, such as hypertension, T2DM, and CVD [6–8]. The gold standards for evaluating IR in the clinical setting include the hyperinsulinemic-euglycemic clamp and intravenous glucose tolerance test [9]. However, their high cost and invasiveness have limited feasibility in large epidemiological surveys. Recently, the triglyceride-glucose (TyG) index, derived from fasting blood glucose (FBG) and triglyceride (TG), has been widely recognized as a simple, reproducible, and cost-effective surrogate marker of IR [9–11]. Beyond the TyG index itself, composite indices integrating TyG with anthropometric indicators, such as TyG-body mass index (BMI), TyG-waist circumference (WC), and TyG-waist-to-height ratio (WHtR), have also been developed to further improve predictive performance in cardiovascular risk assessment [9, 12–15]. Although other anthropometric indices, such as body roundness index (BRI) and a body shape index (ABSI), have also shown better predictive value for cardiometabolic outcomes, they are less routinely used in clinical settings and have rarely been incorporated into TyG-related composite indices in previous research [16–18]. In addition, current evidence suggests that BRI and ABSI do not consistently demonstrate superior predictive performance compared with simpler central obesity measures [19–21]. Therefore, the present study focused on BMI, WC, and WHtR to ensure methodological consistency with existing literature and facilitate clinical applicability.

A growing number of studies have demonstrated that higher levels of TyG-related indices were associated with increased risks of CVD and mortality in both the general population and individuals with metabolic disorders such as hypertension, diabetes, and cardiometabolic-kidney-metabolic syndrome [22–24]. Importantly, recent evidence revealed significant associations between TyG-related indices and all-cause or cause-specific mortality among individuals with MetS [25–27]. However, despite the markedly elevated cardiovascular risk conferred by MetS, no prior research has specifically examined the associations of TyG-related indices with incident CVD in this high-risk population. Given that MetS itself is a powerful predictor of CVD, it remains essential to determine whether these indices offer incremental prognostic value beyond its clinical definition. Moreover, studies have shown that the cardiovascular risk associated with MetS may partly depend on the presence of insulin resistance, and individuals who exhibit both MetS and insulin resistance represent an especially high-risk subgroup [28]. These findings highlight the importance of evaluating TyG-related indices, which serve as surrogate markers of insulin resistance, as potential tools for refining CVD risk stratification among individuals with MetS.

Furthermore, most existing studies have relied on single time-point measurements, without accounting for longitudinal changes in TyG-related indices and their potential prognostic implications. Several cohort studies have assessed the association between dynamic changes in the TyG index and prognosis in patients with myocardial infarction and CVD [29–31]. However, these findings were constrained by a relatively small sample size, short follow-up duration, and single-center study population, limiting the robustness and generalizability of the conclusions. These knowledge gaps highlight the need for prospective investigations to clarify both the static and dynamic relationships between TyG-related indices and CVD risk among individuals with MetS.

Therefore, using data from the large-scale UK Biobank cohort, we aimed to comprehensively assess the associations of both baseline levels and longitudinal changes in TyG-related indices with the risk of incident total CVD, its major subtypes (including coronary heart disease [CHD] and stroke), all-cause mortality, and CVD mortality among participants with MetS. In addition, we compared the increment predictive performance of the four TyG-related indices for these outcomes. The insights gained from this study may facilitate earlier identification of high-risk individuals and support the development of tailored preventive interventions for individuals with MetS.

Methods

Study participants

The present study utilized data from the UK Biobank, a large-scale, population-based prospective cohort study that recruited over 500,000 individuals aged 37–73 years from 22 assessment centers across England, Wales, and Scotland between 2006 and 2010 [32]. At baseline, participants provided information on sociodemographic, lifestyle, environmental, and health-related factors through a touch-screen questionnaire and a face-to-face interview. They also underwent standardized physical examinations and provided biological samples [33]. Detailed information on the study design and procedures has been described previously. Ethical approval for the UK Biobank was granted by the Northwest Multi-center Research Ethics Committee (reference: 21/NW/0157), and written informed consent was obtained from all participants prior to study enrollment.

Of the 502,366 participants enrolled at baseline between 2006 and 2010, we excluded individuals with missing data on TyG-related indices (n = 75,146), those with missing data on MetS components (n = 21,462, including 232 on elevated BP and 21,230 on hyperglycemia), those free of MetS at baseline (n = 270,668, including 57,393 with no MetS components, 111,683 with one component, and 101,592 with two components), and those with a history of CVD at baseline (n = 21,391). Baseline CVD were identified through multiple sources, including primary care records, hospital admission data, and self-reported disease history collected via questionnaires and nurse-led verbal interviews. Specifically, individuals with a physician-diagnosed CVD event recorded in primary care or hospital databases prior to baseline assessment were classified as having prevalent CVD. Self-reported CVD history encompassed the following conditions: heart attack, angina, stroke, atrial fibrillation (AF), atrial flutter, heart failure (HF)/pulmonary odema, heart attack/myocardial infarction, cardiomyopathy, subarachnoid haemorrhage, brain haemorrhage, and ischaemic stroke. After exclusions, 113,699 participants remained for the final analysis. Among them, 3,104 individuals had repeated measurements of TyG-related indices between 2012 and 2013, allowing the evaluation of longitudinal changes in these indices in relation to the risks of CVD and mortality (Fig. 1).

Fig. 1.

Flowchart of participant selection in the study

Definition of MetS

MetS was defined according to the joint criteria of the American Heart Association (AHA), the National Heart, Lung, and Blood Institute (NHLBI), and the International Diabetes Federation (IDF) [34]. Participants were classified as having MetS at baseline if they met at least three of the following five components: (1) central obesity, defined as a WC ≥ 102 cm in males or ≥ 88 cm in females; (2) elevated blood pressure, characterized by systolic blood pressure (SBP) ≥ 130 mmHg, diastolic blood pressure (DBP) ≥ 85 mmHg, or current use of antihypertensive medication; (3) hypertriglyceridemia, defined as TG levels ≥ 1.70 mmol/L (150 mg/dL) or use of lipid-lowering medication; (4) reduced high-density lipoprotein cholesterol (HDL-C), defined as HDL-C < 1.03 mmol/L (< 40 mg/dL) in males or < 1.29 mmol/L (50 mg/dL) in females; and (5) hyperglycemia, defined as glycated hemoglobin (HbA1c) ≥ 5.7% (39 mmol/mol), or current use of glucose-lowering medication. This substitution was adopted because glucose measurements in the UK Biobank were obtained primarily from non-fasting blood samples rather than under standardized fasting conditions. Prior evidence indicates that HbA1c serves as a valid and reliable alternative marker for identifying hyperglycemia in non-fasting settings, thereby providing a more stable assessment of long-term glycemic status in this cohort [35, 36]. The detailed information and related variables in the UK Biobank were summarized in Table S1.

Assessment of TyG-related indices

Random peripheral venous blood samples were collected from each participant, and biochemical markers were measured using the Beckman Coulter AU5800 chemistry analyzer. Detailed description of measurement procedures and assay quality control are shown on the UK Biobank website (https://biobank.ndph.ox.ac.uk/showcase/refer.cgi?id=5636). In this study, four TyG-related indices were evaluated: TyG, TyG-BMI, TyG-WC, and TyG-WHtR. The TyG index was calculated from TG and blood glucose levels according to previously established methods [15, 37], and three composite indices were constructed by integrating the TyG index with measures of general or central obesity. The formulas were as follows [25]:

Outcomes

The primary outcomes of this study were the incidence of total CVD and all-cause mortality in individuals with MetS. The secondary outcomes included the incidence of CHD and stroke and CVD mortality. Incident events were ascertained through linkage to inpatient admission records and death registry data, based on the International Classification of Diseases, 10th Revision (ICD-10) codes. Inpatient admission data were available from the Hospital Episode Statistics for England, the Scottish Morbidity Record for Scotland, and the Patient Episode Database for Wales. Death registry data were available from the NHS England for England and Wales, and the Information and Statistics Division for Scotland. The diagnostic ICD-10 codes were I20-I25, I48, I50, and I60-I64 for total CVD (including CHD, AF, HF, stroke), I20–I25 for CHD, I60–I64 for stroke, and I00-I99 for CVD death [17, 24]. Participants were followed from baseline enrollment until the first occurrence of CVD, death, or censoring at the end of region-specific registry updates, whichever came first. To account for the different update cycles of national hospital and death registries, follow-up was censored on 30 November 2022 for England, 31 August 2022 for Scotland, and 31 May 2022 for Wales, corresponding to the latest dates for which complete registry data were available at the time of data extraction.

Covariates

In the present study, covariates were selected based on established evidence regarding TyG-related indices and cardiovascular and mortality outcomes [13, 17, 24]. In addition, we constructed a directed acyclic graph (DAG) to systematically identify potential confounders and to minimize inappropriate adjustment for mediators (Supplementary Fig. S1). Demographic characteristics comprised age (continuous, years), sex (male vs. female), and ethnicity (white vs. non-white). Socioeconomic status was assessed via multiple indicators, including employment status (employed vs. unemployed), the Townsend deprivation index (a continuous measure of area-level socioeconomic deprivation), educational attainment (university/college vs. others), and family income (low, intermediate, and high). Lifestyle behavioral factors included smoking status (never, former, and current), alcohol consumption (never or special occasions only, 1–3 times/month, 1–2 times/week, 3–4 times/week, and daily or almost daily), sleep duration (< 7, 7–8, and > 8 hours/day), physical activity (inadequate vs. adequate), and a healthy diet score (continuous). Adequate physical activity was defined according to the World Health Organization’ recommendation of at least 150 min/week moderate-to-vigorous intensity physical activity. Clinical measures included SBP and DBP (continuous, mmHg), HbA1c (continuous, mmol/mol), and lipid profiles (including HDL-C, low-density lipoprotein cholesterol [LDL-C], and total cholesterol [TC]; all continuous, mmol/L). Medical history included self-reported hypertension and diabetes (both binary), along with current use of antihypertensive, antihyperglycemic, and lipid-lowering medications (all binary). For participants with missing covariates data, we created an additional “missing” category to maximize the sample size, a widely used method for missing covariates in previous studies.

Statistical analysis

Descriptive statistics of baseline characteristics were performed based on the quartiles of the baseline TyG level. Continuous variables were summarized as medians with interquartile ranges (IQRs) for non-normally distributed data and as means with standard deviations (SDs) for normally distributed data, and categorical variables were expressed as frequencies and percentages. Baseline characteristics were compared using one-way ANOVA for normally distributed continuous variables, the Kruskal–Wallis rank-sum test for skewed continuous variables, and Chi-square test for categorical variables.

Kaplan‒Meier survival curves were generated, and log-rank tests were applied to compare cumulative hazards across TyG quartiles. Cox proportional hazards models were then used to estimate the associations of TyG-related indices with CVD (total CVD, CHD, and stroke) and mortality (all-cause and CVD mortality) outcomes among participants with MetS, with results expressed as hazard ratios (HRs) and 95% confidence intervals (CIs). The proportional hazard assumption was checked using Schoenfeld residuals method, with no observed violations. Two multivariable-adjusted Cox models were constructed: Model 1 adjusted for sociodemographic and lifestyle covariates (age, sex, ethnicity, employment status, Townsend deprivation index, educational attainment, household income, smoking status, alcohol consumption, sleep duration, physical activity, and healthy diet score); Model 2 was further adjusted for clinical biomarkers (SBP, DBP, HbA1c, HDL-C, LDL-C, and TC) and medical history (self-reported hypertension, diabetes, and current use of antihypertensive, antihyperglycemic, and lipid-lowering medications). Associations were evaluated for per 1-SD increase in continuous TyG-related indices and quartiles (Q2–Q4) of TyG-related indices compared with the lowest quartile (Q1), respectively. Dose‒response relationships were explored using restricted cubic splines (RCS) with four knots at the 5th, 35th, 65th, and 95th percentiles, fully adjusted for covariates in Model 2. Nonlinearity was tested using likelihood ratio tests. If a threshold effect was identified, two-piecewise regression models were applied on either side of the inflection point.

To characterize longitudinal patterns of TyG-related indices, we applied K-means clustering analysis to participants with repeated biomarker measurements (n = 3,104). These repeated measurements were obtained during 2012–2013, with a median of 4.44 years (IQR: 3.67–4.94) after the baseline assessment. To determine the optimal number of clusters, we examined candidate solutions with k = 1–10 and evaluated the within-cluster sum of squares (WCSS) using the elbow method. The final choice of k = 4 was selected at the point where the decline in WCSS began to plateau, while also ensuring that the resulting trajectories were clinically interpretable (Supplementary Fig. S2). Cox proportional hazards models were subsequently applied to assess the associations of these trajectory groups with CVD risk in individuals with MetS.

To evaluate the incremental predictive value of TyG-related indices, two risk prediction models were developed: a conventional model including only Model 2 covariates and an extended model additionally incorporating TyG-related indices. Model performance was compared using the net reclassification index (NRI) and the integrated discrimination improvement (IDI), both of which evaluate whether incorporating a novel biomarker improves risk stratification beyond the basic model.

Subgroup analyses were performed by age (< 60 years vs. ≥60 years), sex (male vs. female), smoking status (never vs. ever/current), alcohol consumption (< 3 vs. ≥3 times/week), physical activity (inadequate vs. adequate), healthy diet score (low vs. high), the number of MetS components, and key cardiovascular risk factors (including diabetes, hypertension, and chronic kidney disease [CKD]). Several sensitivity analyses were subsequently performed to test the robustness of our findings. First, missing covariates were addressed using multiple imputation by chained equations (MICE) under a missing-at-random assumption, generating m = 5 imputed datasets with 100 iterations each via the mice package in R. Convergence was evaluated using trace plots, which showed well-mixed and overlapping trajectories without systematic trends, indicating good model convergence (Supplementary Figs. S3–S4). Second, participants who developed outcomes within the first two years of follow-up were excluded to minimize potential reverse causation. Third, Fine and Gray’s competing risk models were applied to account for non-CVD mortality. Fourth, categorical analyses were repeated using the TyG-related index category with the lowest predicted risk (from spline models) as the reference group. Fifth, additional adjustment was made for C-reactive protein (CRP) to consider the role of systemic inflammation. Sixth, a sensitivity analysis excluding all blood pressure–related covariates from the multivariable model was performed. Seventh, further adjustment for fasting time was also conducted.

All analyses were conducted using R version 4.4.0 (R Foundation for Statistical Computing). All statistical tests were two-sided, with P < 0.05 considered statistically significant.

Results

Baseline characteristics of the participants

This study included 113,699 participants with a median age of 60 years (IQR: 53 to 64). Among them, 55,683 (48.97%) were males, and 106,028 (93.25%) were Whites. Compared with those in the lower quartiles, participants in higher TyG quartiles were likely to be male, non-White, unemployed, less educated, smokers, daily drinkers, and to report a longer sleep duration. They also presented more adverse metabolic profiles, including higher levels of TC, LDL-C, HbA1c, SBP, and DBP, alongside lower concentrations of HDL-C (all P < 0.001, Table 1).

Table 1.

Baseline characteristics of study population according to baseline TyG index

Characteristics	Total population	TyG index				P value
		Quartile 1 (< 8.86)	Quartile 2 (8.86–9.12)	Quartile 3 (9.12–9.44)	Quartile 4 (≥ 9.44)
n	113,699	28,425	28,424	28,425	28,425
Age, year	58.12 ± 7.62	58.48 ± 7.74	58.31 ± 7.53	58.23 ± 7.48	57.44 ± 7.67	< 0.001
Sex						< 0.001
Female	58,016 (51.03)	17,265 (60.74)	15,414 (54.23)	14,109 (49.64)	11,228 (39.50)
Male	55,683 (48.97)	11,160 (39.26)	13,010 (45.77)	14,316 (50.36)	17,197 (60.50)
Ethnicity						< 0.001
Whites	106,028 (93.25)	25,543 (89.86)	26,883 (94.58)	26,940 (94.78)	26,662 (93.80)
Non-Whites	7,671 (6.75)	2,882 (10.14)	1,541 (5.42)	1,485 (5.22)	1,763 (6.20)
Employed status						< 0.001
Employed	102,899 (90.50)	25,775 (90.68)	26,010 (91.51)	25,824 (90.85)	25,290 (88.97)
Unemployed	10,140 (8.92)	2465 (8.67)	2271 (7.99)	2451 (8.62)	2953 (10.39)
Missing	660 (0.58)	185 (0.65)	143 (0.50)	150 (0.53)	182 (0.64)
Educational levels						< 0.001
University or college	29,039 (25.54)	7076 (24.89)	7401 (26.04)	7394 (26.01)	7168 (25.22)
Others	82,968 (72.97)	20,889 (73.49)	20,635 (72.60)	20,641 (72.62)	20,803 (73.19)
Missing	1692 (1.49)	460 (1.62)	388 (1.37)	390 (1.37)	454 (1.60)
Townsend deprivation index	-1.92 (-3.52, 0.92)	-1.75 (-3.40, 1.26)	-2.05 (-3.60, 0.65)	-2.04 (-3.56, 0.68)	-1.84 (-3.48, 1.05)	< 0.001
Family income						< 0.001
Low	26,857 (23.62)	6905 (24.29)	6508 (22.90)	6726 (23.66)	6718 (23.63)
Intermediate	49,944 (43.93)	12,379 (43.55)	12,559 (44.18)	12,531 (44.08)	12,475 (43.89)
High	18,581 (16.34)	4238 (14.91)	4765 (16.76)	4715 (16.59)	4863 (17.11)
Missing	18,317 (16.11)	4903 (17.25)	4592 (16.16)	4453 (15.67)	4369 (15.37)
Physical activity						< 0.001
Inadequate	22,219 (19.54)	5703 (20.06)	5397 (18.99)	5549 (19.52)	5570 (19.60)
Adequate	69,667 (61.27)	17,294 (60.84)	17,839 (62.76)	17,492 (61.54)	17,042 (59.95)
Missing	21,813 (19.18)	5428 (19.10)	5188 (18.25)	5384 (18.94)	5813 (20.45)
Smoking status						< 0.001
Never smoking	57,186 (50.30)	15,038 (52.90)	14,502 (51.02)	14,250 (50.13)	13,396 (47.13)
Ever smoking	42,400 (37.29)	10,239 (36.02)	10,570 (37.19)	10,692 (37.61)	10,899 (38.34)
Current smoking	13,376 (11.76)	2971 (10.45)	3168 (11.15)	3316 (11.67)	3921 (13.79)
Missing	737 (0.65)	177 (0.62)	184 (0.65)	167 (0.59)	209 (0.74)
Frequency of drinking						< 0.001
Never or special occasions only	28,643 (25.19)	8159 (28.70)	6714 (23.62)	6748 (23.74)	7022 (24.70)
1–3 times/month	14,234 (12.52)	3676 (12.93)	3512 (12.36)	3540 (12.45)	3506 (12.33)
1–2 times/week	29,164 (25.65)	7132 (25.09)	7447 (26.20)	7433 (26.15)	7152 (25.16)
3–4 times/week	21,953 (19.31)	5092 (17.91)	5694 (20.03)	5571 (19.60)	5596 (19.69)
Daily or almost daily	19,384 (17.05)	4267 (15.01)	4990 (17.56)	5071 (17.84)	5056 (17.79)
Missing	321 (0.28)	99 (0.35)	67 (0.24)	62 (0.22)	93 (0.33)
Daily sleep duration						< 0.001
< 7 h/day	29,725 (26.14)	7686 (27.04)	7326 (25.77)	7274 (25.59)	7439 (26.17)
7–8 h/day	72,420 (63.69)	17,873 (62.88)	18,380 (64.66)	18,328 (64.48)	17,839 (62.76)
> 8 h/day	10,449 (9.19)	2542 (8.94)	2469 (8.69)	2555 (8.99)	2883 (10.14)
Missing	1105 (0.97)	324 (1.14)	249 (0.88)	268 (0.94)	264 (0.93)
Healthy diet score	3.17 ± 1.33	3.30 ± 1.31	3.21 ± 1.32	3.14 ± 1.33	3.03 ± 1.35	< 0.001
Total cholesterol, mmol/L	5.16 ± 1.10	5.73 ± 1.15	5.86 ± 1.20	5.93 ± 1.32	5.16 ± 1.10	< 0.001
Low density lipoprotein cholesterol, mmol/L	3.24 ± 0.87	3.68 ± 0.89	3.76 ± 0.92	3.70 ± 0.96	3.24 ± 0.87	< 0.001
High density lipoprotein cholesterol, mmol/L	1.33 ± 0.33	1.28 ± 0.29	1.22 ± 0.27	1.12 ± 0.25	1.33 ± 0.33	< 0.001
Glycated hemoglobin, mmol/mol	38.84 ± 6.64	37.60 ± 6.06	38.33 ± 7.07	42.49 ± 13.27	38.84 ± 6.64	< 0.001
Systolic blood pressure, mmHg	143.93 ± 16.46	144.50 ± 16.77	144.76 ± 16.85	145.14 ± 17.01	143.93 ± 16.46	< 0.001
Diastolic blood pressure, mmHg	85.57 ± 9.36	86.14 ± 9.42	86.15 ± 9.40	86.16 ± 9.68	85.57 ± 9.36	< 0.001
Self-reported history of hypertension	50,534 (44.45)	14,103 (49.61)	11,841 (41.66)	12,018 (42.28)	12,572 (44.23)	< 0.001
Self-reported history of diabetes	14,662 (12.90)	3934 (13.84)	2248 (7.91)	2820 (9.92)	5660 (19.91)	< 0.001
Use of antihypertensive medication	42,231 (37.14)	12,332 (43.38)	9602 (33.78)	9843 (34.63)	10,454 (36.78)	< 0.001
Use of antihyperglycemic medication	11,059 (9.73)	2973 (10.46)	1618 (5.69)	2031 (7.15)	4437 (15.61)	< 0.001
Use of lipid-lowering medication	35,846 (31.53)	11,991 (42.18)	7497 (26.38)	7616 (26.79)	8742 (30.75)	< 0.001

Data are presented as mean ± standard deviation for continue variables with normal distribution, median (interquartile range) for continue variables with non-normal distribution, and number (%) for categorical variables

Associations between TyG-related indices and the risk of CVD in individuals with MetS

During a median follow-up of 13.71 years (IQR: 12.95–14.45), 24,094 participants with MetS developed CVD, comprising 14,319 cases of CHD and 3,873 cases of stroke. Kaplan-Meier curves revealed that participants in the highest TyG quartile had significantly higher cumulative incidences of total CVD, CHD, and stroke compared to those in the lowest quartile (all P values for log-rank test < 0.05; Fig. 2A–C).

Fig. 2.

Kaplan‒Meier curves of cardiovascular disease and mortality outcomes according to quartiles of the baseline TyG index in participants with metabolic syndrome. Note: A. total CVD, B. CHD, C. stroke, D. all-cause mortality, E. CVD mortality. Abbreviations: CVD cardiovascular disease, CHD coronary heart disease, TyG triglycerides-glucose

After full adjustment for covariates, Cox regression analyses indicated that each 1-SD increase in the TyG index was inversely associated with risks of total CVD and CHD, whereas the TyG-BMI, TyG-WC, and TyG-WHtR indices were consistently associated with increased risks of all CVD outcomes (Table 2). For total CVD incidence, the HRs of each 1-SD increase in TyG-related indices were 0.97 (95% CI: 0.95–0.98) for TyG, 1.14 (1.13–1.16) for TyG-BMI, 1.16 (1.14–1.18) for TyG-WC, and 1.13 (1.11–1.15) for TyG-WHtR (Table 2). When categorized into quartiles, participants in the fourth quartile had significantly elevated risks of CVD compared with those in the first quartile, with HR for total CVD being 1.35 (95% CI: 1.30–1.40) for TyG-BMI, 1.42 (1.36–1.48) for TyG-WC, and 1.32 (1.27–1.37) for TyG-WHtR (Table 2). For CHD, the corresponding HRs in the fourth quartile were 1.16 (95% CI: 1.10–1.22) for TyG-BMI, 1.18 (1.12–1.25) for TyG-WC, and 1.19 (1.13–1.25) for TyG-WHtR, respectively. A statistically significant elevated hazard of incident stroke was observed only in the highest quartile of TyG-WHtR (HR = 1.11, 95% CI = 1.01–1.23).

Table 2.

Association between insulin resistance-related indices and risk of cardiovascular disease in individuals with metabolic syndrome

Exposure	Total CVD			CHD			Stroke
	Cases/ incidence rate	HR (95% CI)		Cases/ incidence rate	HR (95% CI)		Cases/ incidence rate	HR (95% CI)
		Model 1	Model 2		Model 1	Model 2		Model 1	Model 2
TyG index
Each 1-SD increase	–	1.02 (1.01–1.03)	0.97 (0.95–0.98)	–	1.07 (1.05–1.09)	0.97 (0.94–0.99)	–	1.04 (1.01–1.07)	1.00 (0.96–1.04)
Quartile 1	6144/17.83	Reference	Reference	3441/9.60	Reference	Reference	988/2.65	Reference	Reference
Quartile 2	5610/16.06	0.90 (0.86–0.93)	0.91 (0.87–0.94)	3235/8.95	0.93 (0.89–0.98)	0.91 (0.87–0.96)	911/2.43	0.93 (0.85–1.02)	0.93 (0.85–1.02)
Quartile 3	5858/16.86	0.92 (0.89–0.95)	0.90 (0.87–0.94)	3504/9.76	0.99 (0.94–1.04)	0.92 (0.87–0.97)	934/2.49	0.94 (0.86–1.03)	0.92 (0.84–1.02)
Quartile 4	6482/18.94	1.02 (0.98–1.05)	0.92 (0.88–0.97)	4139/11.69	1.14 (1.09–1.19)	0.94 (0.88–0.99)	1040/2.79	1.05 (0.96–1.15)	0.97 (0.87–1.09)
TyG-BMI index
Each 1-SD increase	–	1.17 (1.16–1.19)	1.14 (1.13–1.16)	–	1.12 (1.10–1.14)	1.06 (1.04–1.08)	–	1.07 (1.03–1.10)	1.02 (0.98–1.05)
Quartile 1	5608/16.13	Reference	Reference	3357/9.37	Reference	Reference	1005/2.70	Reference	Reference
Quartile 2	5779/16.63	1.06 (1.02–1.10)	1.06 (1.02–1.10)	3448/9.59	1.03 (0.98–1.08)	1.02 (0.97–1.07)	986/2.63	1.03 (0.94–1.12)	1.02 (0.93–1.11)
Quartile 3	5962/17.20	1.15 (1.11–1.19)	1.14 (1.10–1.18)	3578/9.98	1.11 (1.06–1.16)	1.07 (1.02–1.12)	904/2.41	1.00 (0.91–1.09)	0.96 (0.88–1.05)
Quartile 4	6745/19.73	1.44 (1.39–1.49)	1.35 (1.30–1.40)	3936/11.04	1.30 (1.24–1.36)	1.16 (1.10–1.22)	978/2.62	1.17 (1.07–1.28)	1.05 (0.95–1.16)
TyG-WC index
Each 1-SD increase	–	1.19 (1.17–1.20)	1.16 (1.14–1.18)	–	1.13 (1.11–1.15)	1.07 (1.05–1.09)	–	1.08 (1.04–1.12)	1.03 (0.99–1.07)
Quartile 1	4788/13.48	Reference	Reference	2799/7.67	Reference	Reference	899/2.39	Reference	Reference
Quartile 2	5558/15.88	1.11 (1.07–1.16)	1.11 (1.07–1.16)	3310/9.17	1.09 (1.03–1.14)	1.06 (1.01–1.12)	924/2.46	1.00 (0.91–1.10)	0.99 (0.90–1.09)
Quartile 3	6237/18.13	1.21 (1.16–1.26)	1.21 (1.16–1.26)	3741/10.48	1.14 (1.08–1.20)	1.10 (1.04–1.16)	941/2.52	1.00 (0.90–1.10)	0.97 (0.88–1.07)
Quartile 4	7511/22.46	1.49 (1.43–1.55)	1.42 (1.36–1.48)	4469/12.76	1.33 (1.26–1.40)	1.18 (1.12–1.25)	1109/3.00	1.19 (1.08–1.31)	1.09 (0.98–1.21)
TyG-WHtR index
Each 1-SD increase	–	1.16 (1.15–1.18)	1.13 (1.11–1.15)	–	1.14 (1.12–1.16)	1.08 (1.06–1.10)	–	1.09 (1.06–1.13)	1.04 (1.01–1.08)
Quartile 1	5139/14.58	Reference	Reference	2978/8.20	Reference	Reference	894/2.38	Reference	Reference
Quartile 2	5605/16.05	1.06 (1.02–1.10)	1.07 (1.03–1.11)	3349/9.29	1.08 (1.02–1.13)	1.07 (1.02–1.13)	910/2.42	1.00 (0.92–1.10)	1.00 (0.91–1.10)
Quartile 3	6048/17.49	1.13 (1.08–1.17)	1.12 (1.08–1.17)	3654/10.22	1.14 (1.08–1.19)	1.10 (1.05–1.16)	950/2.54	1.03 (0.94–1.13)	1.01 (0.92–1.11)
Quartile 4	7302/21.73	1.40 (1.35–1.45)	1.32 (1.27–1.37)	4338/12.35	1.35 (1.28–1.41)	1.19 (1.13–1.25)	1119/3.02	1.22 (1.12–1.34)	1.11 (1.01–1.23)

Incidence rates are presented as per 1000 person-years

Model 1 were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, and healthy diet score

Model 2 were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, healthy diet score, systolic/diastolic blood pressure, glycated hemoglobin, high density lipoprotein cholesterol, low density lipoprotein cholesterol, total cholesterol, self-reported history of hypertension and diabetes, and use of antihypertensive, antihyperglycemic, lipid-lowering medications

CVD cardiovascular disease, CHD coronary heart disease, SD standard deviation, HR hazard ratio, CI confidence interval, TyG triglycerides-glucose index, BMI body mass index, WC waist circumference, WHtR waist-to-height ratio

RCS analyses further revealed dose‒response relationship of the four TyG-related indices and CVD outcomes. For total CVD, the TyG index displayed a U-shaped association, whereas the TyG-BMI, TyG-WC, and TyG-WHtR indices presented a J-shaped and nonlinear association, characterized by sharply increasing risk at higher levels of exposure (all P values for nonlinearity < 0.05; Fig. 3A–D). For CHD, the TyG index demonstrated an inverse association, whereas the TyG-BMI, TyG-WC, and TyG-WHtR indices showed steadily increasing risks (all P values for nonlinearity > 0.05; Fig. 3E–H). For stroke, all TyG-related indices exhibited a U-shaped relationship, with risk decreasing below the threshold point and increasing thereafter (all P values for nonlinearity < 0.05; Fig. 3I–L).

Fig. 3.

Dose–response relationships between TyG-related indices and the risk of total CVD A–D, CHD E–H, and stroke I–L in participants with MetS. Note: Models were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, healthy diet score, systolic/diastolic blood pressure, glycated hemoglobin, high density lipoprotein cholesterol, low density lipoprotein cholesterol, total cholesterol, self-reported history of hypertension and diabetes, and use of antihypertensive, antihyperglycemic, lipid-lowering medications. Abbreviations: CVD cardiovascular disease, CHD coronary heart disease, TyG triglycerides-glucose, BMI body mass index, WC waist circumference, WHtR waist-to-height ratio

Associations between TyG-related indices and the risk of mortality in individuals with MetS

During follow-up, 12,317 deaths from all causes and 2,475 deaths from CVD were recorded. A similar trend was observed for both all-cause mortality and CVD mortality, with the highest TyG quartile exhibiting significantly elevated risks (all P values for log-rank test < 0.001; Fig. 2D–E). After fully adjusted for covariates, each 1-SD increase in the TyG-BMI, TyG-WC, and TyG-WHtR indices was significantly associated with a 9%-14% increased risks of all-cause mortality, with HRs of 1.09 (95% CI: 1.07–1.11), 1.14 (1.12–1.17), and 1.12 (1.10–1.14) (Table 3). For CVD mortality, the corresponding HRs were 1.18 (95% CI: 1.14–1.23), 1.22 (1.17–1.27), and 1.19 (1.14–1.24). Moreover, compared with the lowest quartile, participants in the highest quartiles of TyG-BMI, TyG-WC, and TyG-WHtR had the significantly heightened mortality risks, with HRs of 1.17 (95% CI: 1.11–1.23), 1.34 (1.27–1.43), and 1.29 (1.22–1.36) for all-cause mortality, and 1.36 (1.21–1.53), 1.51 (1.32–1.72), and 1.42 (1.25–1.60) for CVD mortality, respectively. However, the TyG index demonstrated no robust associations with all-cause or CVD mortality (Table 3). As shown in Fig. 4, the TyG index followed a U-shaped relationship with risk of both all-cause and CVD mortality, whereas the TyG-BMI, TyG-WC, and TyG-WHtR indices presented J-shaped curves, with markedly elevated risks above threshold points (all P values for nonlinearity < 0.05; Fig. 4).

Table 3.

Association between insulin resistance-related indices and risk of all-cause and cardiovascular mortality in individuals with metabolic syndrome

Exposure	All-cause mortality			CVD mortality
	Cases/incidence rate	HR (95% CI)		Cases/incidence rate	HR (95% CI)
		Model 1	Model 2		Model 1	Model 2
TyG index
Each 1-SD increase	–	1.02 (1.00-1.04)	0.99 (0.97–1.02)	–	1.00 (0.96–1.04)	0.92 (0.88–0.97)
Quartile 1	3189/8.45	Reference	Reference	671/1.78	Reference	Reference
Quartile 2	2870/7.57	0.91 (0.86–0.95)	0.94 (0.90-1.00)	547/1.44	0.82 (0.73–0.91)	0.85 (0.76–0.96)
Quartile 3	2888/7.62	0.89 (0.85–0.94)	0.92 (0.88–0.98)	562/1.48	0.81 (0.72–0.91)	0.82 (0.73–0.93)
Quartile 4	3370/8.93	1.04 (0.99–1.09)	1.00 (0.94–1.07)	695/1.84	0.97 (0.87–1.08)	0.87 (0.76-1.00)
TyG-BMI index
Each 1-SD increase	–	1.12 (1.10–1.14)	1.09 (1.07–1.11)	–	1.24 (1.20–1.29)	1.18 (1.14–1.23)
Quartile 1	3174/8.43	Reference	Reference	605/1.61	Reference	Reference
Quartile 2	2907/7.67	0.96 (0.91–1.01)	0.97 (0.92–1.02)	561/1.48	0.97 (0.86–1.09)	0.99 (0.88–1.11)
Quartile 3	2921/7.7	1.02 (0.97–1.07)	1.02 (0.97–1.08)	555/1.46	1.02 (0.91–1.15)	1.02 (0.90–1.15)
Quartile 4	3315/8.77	1.25 (1.19–1.31)	1.17 (1.11–1.23)	754/2	1.51 (1.35–1.68)	1.36 (1.21–1.53)
TyG-WC index
Each 1-SD increase	–	1.17 (1.15–1.19)	1.14 (1.12–1.17)	–	1.27 (1.22–1.32)	1.22 (1.17–1.27)
Quartile 1	2579/6.78	Reference	Reference	452/1.19	Reference	Reference
Quartile 2	2808/7.4	1.05 (0.99–1.11)	1.07 (1.01–1.13)	534/1.41	1.08 (0.95–1.23)	1.10 (0.97–1.25)
Quartile 3	3069/8.12	1.12 (1.06–1.18)	1.13 (1.07–1.20)	618/1.63	1.17 (1.03–1.33)	1.20 (1.05–1.36)
Quartile 4	3861/10.31	1.40 (1.32–1.48)	1.34 (1.27–1.43)	871/2.33	1.60 (1.41–1.81)	1.51 (1.32–1.72)
TyG-WHtR index
Each 1-SD increase	–	1.15 (1.13–1.17)	1.12 (1.10–1.14)	–	1.24 (1.20–1.29)	1.19 (1.14–1.24)
Quartile 1	2685/7.07	Reference	Reference	495/1.3	Reference	Reference
Quartile 2	2752/7.25	1.00 (0.95–1.06)	1.03 (0.97–1.08)	554/1.46	1.07 (0.95–1.21)	1.11 (0.98–1.25)
Quartile 3	2997/7.92	1.06 (1.00-1.11)	1.08 (1.02–1.14)	574/1.52	1.06 (0.94–1.20)	1.08 (0.96–1.23)
Quartile 4	3883/10.35	1.34 (1.28–1.41)	1.29 (1.22–1.36)	852/2.27	1.53 (1.37–1.72)	1.42 (1.25–1.60)

Incidence rates are presented as per 1000 person-years

Model 1 were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, and healthy diet score

Model 2 were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, healthy diet score, systolic/diastolic blood pressure, glycated hemoglobin, high density lipoprotein cholesterol, low density lipoprotein cholesterol, total cholesterol, self-reported history of hypertension and diabetes, and use of antihypertensive, antihyperglycemic, lipid-lowering medications

CVD cardiovascular disease, CHD coronary heart disease, SD standard deviation, HR hazard ratio, CI confidence interval, TyG triglycerides- glucose index, BMI body mass index, WC waist circumference, WHtR waist-to-height ratio

Fig. 4.

Dose–response relationships between TyG-related indices and the risk of all-cause mortality A–D and CVD mortality E–H in participants with MetS. Note: Models were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, healthy diet score, systolic/diastolic blood pressure, glycated hemoglobin, high density lipoprotein cholesterol, low density lipoprotein cholesterol, total cholesterol, self-reported history of hypertension and diabetes, and use of antihypertensive, antihyperglycemic, lipid-lowering medications. Abbreviations: CVD cardiovascular disease, TyG triglycerides- glucose index, BMI body mass index, WC waist circumference, WHtR waist-to-height ratio

Threshold effects of TyG-related indices on CVD and mortality in individuals with MetS

On the basis of observed nonlinear relationships, two-segmented Cox proportional hazards regression analyses were performed to quantify threshold-specific associations. The inflection points ranged from 9.08 to 9.30 for the TyG index, 240.67–294.14 for TyG-BMI, 739.55–793.34 for TyG-WC, and 4.78–5.69 for TyG-WHtR. When below the threshold point, the TyG index was inversely associated with total CVD, stroke, all-cause mortality, and CVD mortality. Conversely, above the threshold points, the TyG index was positively related with risk of these outcomes (all P values for log-likelihood ratio < 0.001) (Table 4). Similar patterns were observed for the relationship of TyG-BMI and TyG-WC with stroke, all-cause mortality, CVD mortality, for TyG-WHtR in relation to stroke and all-cause mortality (Table 4). Besides, we observed that the hazard of total CVD increased more steeply after the threshold points than below it as the TyG index increased (Table 4).

Table 4.

Two-piecewise regression models for associations of insulin resistance-related indices with cardiovascular disease and mortality in individuals with metabolic syndrome

		Two-piecewise regression			P value for log-likelihood ratio
Outcome/groups	Standard regression	Inflection point	Before inflection point	Equal to or after inflection point
Total CVD
TyG index	0.935 (0.902–0.970)	9.082	0.838 (0.793–0.887)	1.033 (0.979–1.089)	< 0.001
TyG-BMI index	1.003 (1.002–1.003)	248.603	1.000 (0.999–1.001)	1.003 (1.003–1.004)	< 0.001
TyG-WC index	1.001 (1.001–1.001)	787.879	1.000 (1.000-1.001)	1.001 (1.001–1.001)	< 0.001
TyG-WHtR index	1.193 (1.169–1.217)	5.513	1.115 (1.072–1.160)	1.250 (1.212–1.289)	< 0.001
Stroke
TyG index	0.998 (0.913–1.091)	9.303	0.894 (0.795–1.004)	1.197 (1.026–1.396)	0.005
TyG-BMI index	1.000 (1.000-1.001)	240.674	0.995 (0.992–0.998)	1.001 (1.000-1.002)	0.001
TyG-WC index	1.000 (1.000-1.001)	739.552	0.997 (0.996–0.999)	1.000 (1.000-1.001)	0.004
TyG-WHtR index	1.060 (1.008–1.114)	4.776	0.730 (0.593-0.900)	1.113 (1.052–1.178)	< 0.001
All-cause mortality
TyG index	0.982 (0.934–1.031)	9.204	0.874 (0.816–0.937)	1.136 (1.049–1.229)	< 0.001
TyG-BMI index	1.002 (1.001–1.002)	242.083	0.994 (0.992–0.995)	1.003 (1.002–1.003)	< 0.001
TyG-WC index	1.001 (1.001–1.001)	793.336	0.999 (0.998–0.999)	1.001 (1.001–1.002)	< 0.001
TyG-WHtR index	1.181 (1.149–1.214)	4.936	0.838 (0.760–0.923)	1.257 (1.217–1.298)	< 0.001
CVD mortality
TyG index	0.855 (0.769–0.950)	9.102	0.704 (0.598–0.828)	1.024 (0.875–1.198)	0.003
TyG-BMI index	1.004 (1.003–1.004)	294.138	0.999 (0.997–1.001)	1.007 (1.005–1.008)	< 0.001
TyG-WC index	1.002 (1.001–1.002)	759.734	0.996 (0.994–0.998)	1.002 (1.002–1.002)	< 0.001
TyG-WHtR index	1.279 (1.204–1.357)	5.692	1.009 (0.907–1.122)	1.554 (1.418–1.704)	< 0.001

Models were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, healthy diet score, systolic/diastolic blood pressure, glycated hemoglobin, high density lipoprotein cholesterol, low density lipoprotein cholesterol, total cholesterol, self-reported history of hypertension and diabetes, and use of antihypertensive, antihyperglycemic, lipid-lowering medications

CVD cardiovascular disease, CHD coronary heart disease, SD standard deviation, HR hazard ratio, CI confidence interval, TyG triglycerides- glucose index, BMI body mass index, WC waist circumference, WHtR waist-to-height ratio

Incremental predictive values of TyG-related indices in individuals with MetS

We further assessed the incremental predictive value of incorporating TyG-related indices into the conventional risk model for incident CVD and mortality in individuals with MetS. As shown in Table 5, the TyG index did not significantly improve predictive performance. In contrast, the incorporation of TyG-BMI, TyG-WC, and TyG-WHtR significantly enhanced both IDI and NRI for total CVD, all-cause mortality, and CVD mortality, with TyG-WC showing the greatest incremental value (IDI = 0.432, 95% CI = 0.361–0.519; 0.212, 0.140–0.275; 0.172, 0.092–0.312; NRI = 6.254, 5.683–6.955; 6.243, 5.030–7.152; 8.049, 5.495–10.343; respectively). For CHD and stroke, TyG-WHtR exhibited the largest improvement in predictive performance (IDI = 0.089, 0.053–0.142 and 0.012, 0.001–0.039; NRI = 3.199, 2.464–4.153 and 3.205, 0.391–4.453).

Table 5.

Increment predictive values of insulin resistance-related indices for the risks of cardiovascular disease and mortality in individuals with metabolic syndrome

	Total CVD	CHD	Stroke	All-cause mortality	CVD mortality
Conventional model	Reference	Reference	Reference	Reference	Reference
+TyG index
IDI value (%)	0.008 (0.001–0.020)	0.007 (−0.001-0.020)	−0.001 (−0.002-0.008)	0.001 (-0.001-0.009)	0.013 (−0.003-0.047)
P value	0.040	0.139	1.327	0.693	0.198
NRI value (%)	0.385 (−0.421-0.977)	0.229 (−1.404-1.118)	−1.675 (−2.121-3.415)	−1.230 (−2.571-2.389)	0.842 (−2.614-2.962)
P value	0.356	0.970	1.109	0.535	0.455
+TyG-BMI index
IDI value (%)	0.383 (0.321–0.464)	0.062 (0.032–0.103)	0.003 (-0.002-0.018)	0.091 (0.041–0.149)	0.128 (0.058–0.215)
P value	< 0.001	< 0.001	0.337	< 0.001	< 0.001
NRI value (%)	5.830 (5.093–6.464)	2.859 (2.014–3.579)	1.564 (-2.508-3.199)	4.714 (3.116–5.866)	7.094 (4.676–9.085)
P value	< 0.001	< 0.001	0.356	< 0.001	< 0.001
+TyG-WC index
IDI value (%)	0.432 (0.361–0.519)	0.075 (0.042–0.117)	0.007 (0.000-0.031)	0.212 (0.140–0.275)	0.172 (0.092–0.312)
P value	< 0.001	< 0.001	0.040	< 0.001	< 0.001
NRI value (%)	6.254 (5.683–6.955)	2.913 (2.157–3.851)	2.904 (−1.710-4.494)	6.243 (5.030–7.152)	8.049 (5.495–10.343)
P value	< 0.001	< 0.001	0.099	< 0.001	< 0.001
+TyG-WHtR index
IDI value (%)	0.331 (0.268–0.414)	0.089 (0.053–0.142)	0.012 (0.001–0.039)	0.187 (0.106–0.261)	0.142 (0.073–0.223)
P value	< 0.001	< 0.001	0.020	< 0.001	< 0.001
NRI value (%)	5.399 (4.758–6.231)	3.199 (2.464–4.153)	3.205 (0.391–4.453)	5.645 (4.035–6.659)	6.408 (4.431–8.349)
P value	< 0.001	< 0.001	0.040	< 0.001	< 0.001

Results were presented comparing the new models, which incorporate the insulin resistance-related indices into conventional model, with conventional model. Conventional models were developed based on covariates in Model 2, including age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, healthy diet score, systolic/diastolic blood pressure, glycated hemoglobin, high density lipoprotein cholesterol, low density lipoprotein cholesterol, total cholesterol, self-reported history of hypertension and diabetes, and use of antihypertensive, antihyperglycemic, lipid-lowering medications

CVD cardiovascular disease, CHD coronary heart disease, SD standard deviation, HR hazard ratio, CI confidence interval, TyG triglycerides- glucose index, BMI body mass index, WC waist circumference, WHtR waist-to-height ratio, IDI integrated discrimination improvement, NRI net reclassification index

Effects of longitudinal changes in TyG-related indices on CVD risk in individuals with MetS

As shown in Fig. S5, K-means clustering analyses identified four distinct trajectory patterns: for the TyG index, namely the consistently low group (class 1), low-to-middle increasing group (class 2), middle-to-low decreasing group (class 3), and consistently high group (class 4). For TyG-BMI, TyG-WC, and TyG-WHtR indices, the patterns were the consistently low group (class 1), low-middle group (class 2), middle group (class 3), and consistently high group (class 4), respectively. Compared with those in the consistently low group, participants in consistently high group (class 4) had a significantly increased risk of total CVD, with HRs being of 1.74 (95% CI: 1.25–2.43) for TyG-BMI, 1.44 (1.02–2.01) for TyG-WC, and 1.59 (1.14–2.22) for TyG-WHtR, respectively (Table 6). However, although hazards of CHD and stroke tended to increase in the consistently high group, these associations did not reach statistical significance, which may be due to the limited number of incident cases (n = 395 and 112, respectively). When we compared baseline characteristics between participants included in the single measurement analysis and those with repeated measurements of TyG-related indices, heterogeneity was observed between the two groups (Supplementary Table S2).

Table 6.

Changes of insulin resistance-related indices and risk of cardiovascular disease in individuals with metabolic syndrome

	Total CVD			CHD			Stroke
Groups	Cases/ incidence rate	HR (95% CI)		Cases/ incidence rate	HR (95% CI)		Cases/ incidence rate	HR (95% CI)
		Model 1	Model 2		Model 1	Model 2		Model 1	Model 2
TyG index
Class 1	179/17.62	Reference	Reference	106/10.05	Reference	Reference	35/3.15	Reference	Reference
Class 2	129/19.19	1.00 (0.81–1.23)	0.93 (0.75–1.14)	84/12.07	1.06 (0.80–1.40)	0.95 (0.71–1.26)	16/2.15	1.20 (0.72-2.00)	1.12 (0.66–1.90)
Class 3	202/16.36	1.08 (0.88–1.34)	0.99 (0.79–1.24)	115/8.99	1.18 (0.89–1.57)	1.03 (0.76–1.39)	36/2.71	1.42 (0.85–2.39)	1.31 (0.75–2.29)
Class 4	169/16.93	1.16 (0.91–1.46)	1.02 (0.76–1.36)	90/8.69	1.36 (1.00-1.85)	1.06 (0.73–1.53)	25/2.33	0.90 (0.47–1.70)	0.83 (0.39–1.80)
TyG-BMI index
Class 1	279/17.53	Reference	Reference	175/10.66	Reference	Reference	52/3.00	Reference	Reference
Class 2	171/16.44	1.12 (0.92–1.36)	1.07 (0.88–1.30)	86/7.93	1.17 (0.91–1.50)	1.09 (0.85–1.41)	28/2.50	1.55 (0.94–2.54)	1.51 (0.91–2.50)
Class 3	172/16.52	1.17 (0.94–1.45)	1.11 (0.89–1.40)	102/9.44	0.94 (0.70–1.27)	0.84 (0.62–1.15)	23/2.04	1.40 (0.79–2.46)	1.42 (0.79–2.53)
Class 4	57/22.99	1.86 (1.36–2.55)	1.74 (1.25–2.43)	32/12.35	1.59 (1.05–2.42)	1.35 (0.87–2.10)	9/3.28	1.70 (0.76–3.80)	1.82 (0.78–4.21)
TyG-WC index
Class 1	259/17.02	Reference	Reference	156/9.93	Reference	Reference	46/2.80	Reference	Reference
Class 2	240/18.95	1.25 (0.99–1.58)	1.19 (0.94–1.51)	132/9.98	1.14 (0.85–1.55)	1.06 (0.78–1.44)	42/3.01	1.40 (0.78–2.51)	1.37 (0.76–2.47)
Class 3	105/13.24	1.32 (1.03–1.69)	1.22 (0.94–1.58)	65/7.96	1.04 (0.75–1.43)	0.90 (0.65–1.26)	16/1.88	1.47 (0.79–2.74)	1.40 (0.74–2.64)
Class 4	75/22.10	1.59 (1.15–2.20)	1.44 (1.02–2.01)	42/11.83	1.23 (0.81–1.87)	0.97 (0.62–1.52)	8/2.15	0.95 (0.39–2.34)	0.92 (0.36–2.34)
TyG-WHtR index
Class 1	66/22.11	Reference	Reference	37/11.90	Reference	Reference	10/3.04	Reference	Reference
Class 2	221/18.35	1.24 (1.00-1.55)	1.18 (0.94–1.47)	122/9.70	1.17 (0.88–1.56)	1.11 (0.83–1.48)	40/3.06	1.89 (1.00-3.57)	1.82 (0.96–3.45)
Class 3	280/17.14	1.28 (1.01–1.61)	1.18 (0.93–1.50)	167/9.91	1.08 (0.80–1.47)	0.95 (0.69–1.30)	50/2.81	2.01 (1.05–3.87)	1.96 (1.00-3.82)
Class 4	112/14.27	1.74 (1.27–2.39)	1.59 (1.14–2.22)	69/8.50	1.45 (0.95–2.19)	1.17 (0.75–1.82)	12/1.42	1.86 (0.78–4.44)	1.88 (0.77–4.62)

Different groups of TyG-related indices were identified using the K-means clustering approach, and the trajectories were shown in Fig. S5

Incidence rates are presented as per 1000 person-years.

Model 1 were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, and healthy diet score

Model 2 were adjusted for age, sex, ethnicity, employed status, Townsend deprivation index, educational level, family income, smoking status, alcohol drinking, sleep duration, physical activity, healthy diet score, systolic/diastolic blood pressure, glycated hemoglobin, high density lipoprotein cholesterol, low density lipoprotein cholesterol, total cholesterol, self-reported history of hypertension and diabetes, and use of antihypertensive, antihyperglycemic, lipid-lowering medications

Abbreviations: CVD cardiovascular disease, CHD coronary heart disease, SD standard deviation, HR hazard ratio, CI confidence interval, TyG triglycerides- glucose index, BMI body mass index, WC waist circumference, WHtR waist-to-height ratio

Sensitivity and subgroup analyses

Subgroup analyses showed that the associations of TyG-related indices with total CVD and all-cause mortality were stronger among participants aged ≥ 60 years than those < 60 years (Supplementary Figs. S6–S7) and more pronounced in females for TyG-WC with CVD incidence and CVD mortality (Supplementary Figs. S8–S9). Associations with CVD incidence were generally consistent across smoking status, whereas associations with all-cause mortality were stronger in never smokers (Supplementary Figs. S10–S11). No substantial differences were observed across groups defined by drinking frequency or physical activity (Supplementary Figs. S12–S15). Notably, associations of TyG-related indices with CVD incidence and mortality were more pronounced in participants with lower diet quality (Supplementary Figs. S16–S17). In addition, the presence of CKD, diabetes, or hypertension did not modify the associations between TyG-related indices and the risks of CVD and mortality outcomes (Supplementary Figs. S18–S23). In contrast, the number of MetS components appeared to modify these associations, with stronger effects observed for the TyG index on stroke risk, TyG-BMI on all-cause mortality and CVD mortality, TyG-WC on all-cause mortality, and TyG-WHtR on both all-cause and CVD mortality among participants with a higher number of MetS components (all P values for interaction < 0.05, Supplementary Table S3).

Sensitivity analyses confirmed the robustness of findings after imputing missing covariates (Supplementary Figs. S24–S25), excluding events within the first two years of follow-up (Supplementary Figs. S26–S27), and accounting for competing risks of non-CVD mortality (Supplementary Fig. S28). When the lowest-risk group was used as the reference, individuals in the highest quartile (Q4) had HRs greater than 1 for incident CVD and mortality, although some associations did not reach statistical significance in the fully adjusted model (Supplementary Table S4). Furthermore, the results remained consistent after additional adjustment for CRP and fasting time (Supplementary Figs. S29–S32). Similarly, when blood pressure–related covariates were not included, the findings remained robust (Supplementary Figs. S33–S34).

Discussion

In this large prospective cohort study, we comprehensively evaluated the associations of TyG-related indices with CVD and mortality in individuals with MetS, as well as the impact of their longitudinal changes on these outcomes. Compared with the TyG index alone, the combined indices, (e.g., TyG-BMI, TyG-WC, and TyG-WHtR) demonstrated stronger and more consistent associations with CVD incidence, CVD mortality, and all-cause mortality, and exhibited superior predictive performance. Most associations were nonlinear, with risks rising sharply beyond specific inflection points. Persistently high levels of TyG-related indices from baseline through follow-up were also significantly associated with increased risks, underscoring the importance of long-term monitoring. These findings extend previous evidence from the general populations to a high-risk MetS cohort, highlighting the potential clinical utility of TyG-related indices for dynamic risk stratification and early identification of individuals at heightened cardiometabolic risk.

Although the individual components underlying TyG-related indices, including FBG, TG, BMI, WC, and WHtR, are established cardiovascular risk markers, recent studies have increasingly focused on TyG-related indices and demonstrated that these indices provide greater clinical utility in predicting cardiometabolic risk. Previous studies have established associations between TyG-related indices and CVD and mortality outcomes in the general population. For instance, Dang et al. reported that TyG, TyG-WC, TyG-WHtR, and TyG-BMI were positively associated with CVD incidence and CVD mortality in the U.S. population [13]. Similarly, Su et al. and Mo et al. observed positive associations between the TyG index and the risks of CVD incidence, CVD mortality, and all-cause mortality in the Chinese population [23, 38]. Alavi et al. further demonstrated that the risks of CVD and all-cause mortality increased across TyG tertiles in 5432 Iranian adults [12]. MetS affects a substantial proportion of the population and is closely linked to cardiovascular and mortality outcomes [39], making the identification of reliable risk and prognostic factors is critical for optimizing management and reducing healthcare costs in this high-risk group. Evidence has also confirmed the associations between TyG-related indices and mortality outcomes among individuals with MetS. For instance, data from the National Health and Nutrition Examination Survey (NHANES) indicated that elevated TyG-related indices were significantly associated with both CVD and all-cause mortality in this population [25, 40]. However, the relationship between TyG indices and CVD incidence in patients with MetS has not been previously investigated, leaving uncertainty regarding potential synergistic effects of insulin resistance and MetS on CVD development. In the present study, we demonstrated for the first time that TyG-BMI, TyG-WC, and TyG-WHtR were significantly associated with CVD incidence (including total CVD, CHD, and stroke), CVD mortality, and all-cause mortality, whereas associations of the TyG index alone with these outcomes were less consistent. These findings highlight the potential utility of obesity-adjusted TyG-related indices as more reliable and cost-effective prognostic indicators for adverse outcomes in patients with MetS. The associations between TyG-related indices and CVD incidence, CVD mortality, and all-cause mortality may be explained by multiple interrelated mechanisms, whereby insulin resistance induces oxidative stress, exacerbates inflammatory responses, promotes foam cell formation, impairs endothelial function, and facilitates smooth muscle cell proliferation, collectively accelerating atherosclerosis and increasing cardiovascular risk [13, 41–43].

Notably, the apparent inverse association observed between the highest TyG quartile and the risk of CVD in the multivariable-adjusted model warrants cautious interpretation. Additional analyses indicate that this pattern does not reflect a true protective effect of elevated TyG but rather arises from methodological artifacts, particularly the choice of reference group. Although many previous studies have conventionally used Q1 as the reference category [13, 14, 31], RCS analyses demonstrated that the lowest-risk segment did not align with Q1 in this MetS cohort. Consequently, using Q1 as the reference may artificially produce HR below 1 for the higher TyG categories. When the reference group was redefined as the TyG category with the lowest estimated risk, the direction of association became fully consistent with the incidence rates, confirming that higher TyG levels were generally associated with increased CVD and CHD risk, although some associations were not statistically significant. Beyond the reference-group issue, other factors may contribute to the apparent paradox, including potential collider bias introduced by restricting the analysis to individuals with MetS [44, 45], residual confounding from unmeasured metabolic or inflammatory factors [46], and possible medication effects, wherein individuals with markedly elevated TyG may receive more intensive treatment [47]. Taken together, these considerations suggest that the inverse association is unlikely to represent a biologically plausible phenomenon, and future studies evaluating TyG-related indices within MetS populations should carefully consider reference-category selection and potential sources of structural bias to avoid misleading interpretations.

Overall, the associations of TyG-related indices with CVD outcomes tended to follow U-shaped nonlinear patterns, whereas their relationships with mortality outcomes were more frequently U- or J-shaped, with risks consistently rising beyond the inflection points. Previous studies, however, have reported somewhat inconsistent dose‒response relationships between TyG-related indices and CVD or mortality outcomes. For example, several studies have shown that TyG-related indices exhibit U-shaped or L-shaped relationships with all-cause and CVD mortality in populations with metabolic disorders, such as prediabetes, diabetes, hypertension, or metabolic dysfunction-associated steatotic liver disease (MASLD) [16, 41, 48]. Additionally, evidence from multiple studies suggested that the associations between TyG-related indices and CVD or mortality outcomes may vary: in the UK Biobank, TyG showed a nonlinear relationship with CVD outcomes while TyG-BMI, TyG-WC, and TyG-WHtR were predominantly linear [22]. Similarly, Dang et al. observed mostly linear associations for all four indices with CVD incidence and mortality in a U.S. cohort [13]. In patients with NAFLD, most associations of TyG and TyG-WHtR with CVD outcomes were linear, but their relationships with chronic heart failure and all-cause mortality were nonlinear [49]. These nonlinear patterns may be explained by the underlying metabolic disturbances characteristic of MetS, including dysregulated glucose and lipid metabolism, insulin resistance, chronic low-grade inflammation, and altered fat distribution [50, 51]. Both excessively low and excessively high levels of TyG-related indices could reflect metabolic imbalance or compensatory physiological responses, thereby increasing cardiovascular and mortality risks [52]. In particular, central obesity and visceral fat accumulation, which are prevalent in MetS, may amplify the adverse effects of elevated TyG-related indices through mechanisms involving inflammatory pathways, oxidative stress, and endothelial dysfunction [8, 53]. Collectively, these findings underline the complex interplay between TyG-related indices and cardiometabolic pathways in MetS. A deeper understanding of these mechanisms will be essential for refining risk stratification and guiding more precise prevention strategies in this high-risk population.

Given that TyG-related indices may fluctuate over time, it is crucial to investigate their long-term trajectories and their associations with CVD outcomes [54]. Previous studies in the general population have demonstrated that long-term trajectories of the TyG index were linked to multiple adverse health outcomes, underscoring the importance of dynamic monitoring for more precise risk prediction. For example, data from the Coronary Artery Risk Development (CARDIA) study revealed that, compared with participants in the low TyG trajectory group, those in the high TyG trajectory group had significantly greater risks of CVD events (HR = 2.35, 95% CI: 1.34–4.12) and all-cause mortality (HR = 3.04, 95% CI: 1.83–5.07) among U.S. adults [55]. Similarly, evidence from the Korea National Health Insurance Service-National Sample Cohort indicated that relative to the stable TyG group, an increasing TyG trajectory was positively associated with both all-cause mortality (HR = 1.09, 95% CI: 1.03–1.15) and CVD mortality (HR = 1.23, 95% CI: 1.01–1.50) in 233,546 Korean adults aged ≥ 19 years [56]. Findings from the Atherosclerosis Risk in Communities (ARIC) study further revealed that participants with high or very high TyG trajectories had significantly increased risks of incident peripheral artery disease compared with those with low TyG trajectories (OR = 1.40, 95% CI: 1.13–1.74; OR = 1.74, 95% CI: 1.29–2.34) among 15,792 U.S. adults aged 45–64 years [57]. In addition, based on a large Chinese prospective cohort, Xu et al. observed that both gains and losses in TyG levels, relative to stable TyG levels, were associated with increased risks of cardiometabolic diseases in 36,359 adults aged ≥ 18 years [58]. However, limited evidence has explored whether similar longitudinal associations exist among patients with MetS. To our knowledge, the present study is the first to demonstrate that persistently high levels of TyG-BMI, TyG-WC, and TyG-WHtR from baseline through follow-up were positively associated with the risk of total CVD, CHD, and stroke in this high-risk population. These findings suggest that clinical risk evaluation should not rely solely on single time-point measurements. Instead, both research and practice should incorporate dynamic monitoring of TyG-related indices to more accurately capture their long-term adverse cardiometabolic impacts.

Our findings also showed that incorporating anthropometric measures into the TyG index substantially enhanced its predictive ability, with TyG-WC and TyG-WHtR exhibiting the strongest predictive performance for CVD incidence, CVD mortality, and all-cause mortality. These results were consistent with previous large-scale studies demonstrating that TyG-WHtR, TyG-WC, and TyG-BMI improve discrimination and reclassification metrics for cardiometabolic and mortality outcomes across diverse populations. For example, Liu et al. reported that TyG-WHtR, TyG-WC, and TyG-BMI markedly improved NRI, IDI, and AUC values among 282,920 participants with cardiovascular-kidney-metabolic (CKM) syndrome stages 0–3 [17]. Similarly, Qiao et al. observed that TyG-WC and TyG-WHtR yielded higher C-index, NRI, and IDI values in predicting CVD and mortality in patients with MASLD [22]. Additional studies further support these findings: TyG-BMI demonstrated superior diagnostic efficacy for all-cause and cardiovascular in CKD patients; TyG-WHtR showed superior diagnostic accuracy for predicting CKD in 10,660 U.S. participants; and TyG-WHtR and TyG-WC provided greater discrimination for CVD incidence and mortality among 11,937 U.S. adults [13]. Enhanced predictive performance of TyG-WHtR and TyG-WC was also observed for diabetes risk in a large Japanese cohort [59].

Notably, TyG-WC and TyG-WHtR outperformed TyG-BMI, partly because WC and WHtR better capture central adiposity and visceral fat accumulation, which are more strongly associated with metabolic abnormalities and cardiovascular dysfunction than overall adiposity reflected by BMI [60–62]. Abdominal or visceral adiposity contributes to cardiometabolic risk through multiple biological pathways, including systemic inflammation, leptin resistance, dyslipidemia, and endothelial dysfunction [63–65]. These pathological processes exacerbate insulin resistance, hypertension, and vascular and myocardial injury, collectively amplifying cardiovascular risk [65]. Given that individuals with MetS often exhibit excessive visceral fat deposition, indices integrating TyG with WC or WHtR may more effectively capture the combined effects of insulin resistance and abdominal adiposity [13]. Previous studies have shown that the TyG-WHtR and TyG-WC indices were superior to the TyG-BMI in identifying the risk of cardiovascular disease risk and diabetes mellitus [13, 61, 66]. In addition, TyG-WC provided the greatest improvements in IDI and NRI for total CVD, CVD mortality, and all-cause mortality, suggesting it is the most informative index for overall cardiovascular risk, whereas TyG-WHtR showed superior value for predicting CHD and stroke, highlighting its utility for atherosclerotic or cerebrovascular risk assessment. These findings highlight the clinical value of tailored application of TyG-related indices, with TyG-WC prioritized for broad cardiovascular and mortality risk stratification, and TyG-WHtR for CHD and stroke risk assessment, enabling more precise and personalized preventive strategies in patients with MetS.

Our subgroup analysis stratified by the number of MetS components suggested that the severity of MetS may modify the association between TyG-related indices and study outcomes. Specifically, the related hazard generally increased with a greater number of MetS components. These results can be explained by the synergistic influence of MetS components on cardiovascular risk, reflecting the underlying cumulative burden of metabolic dysfunction. Moreover, the observed age- and sex-specific heterogeneities were consistent with previous evidence in the MetS population, which may be attributed to the physiological differences between genders, varied initial cardiovascular risk, distinct lifestyle and behaviors patterns, and genetic background.

Based on a large prospective cohort study, we comprehensively evaluated the predictive and prognostic performance of TyG-related indices for CVD and mortality outcomes in individuals with MetS. Importantly, this study extended the clinical utility of TyG-related indices beyond single time-point measurements and highlighted their potential value in dynamic risk monitoring and stratification. Nevertheless, several limitations should be acknowledged. First, some metabolic variables (e.g., HbA1c, LDL-C, HDL-C, SBP, DBP) may lie on the causal pathway, and adjusting for them could cause over-adjustment. To address this, we conducted a sensitivity analysis excluding SBP, DBP, self-reported hypertension, and antihypertensive medication use from the models. The results were consistent with our main findings, indicating that potential over-adjustment did not materially affect the observed associations. Second, the longitudinal sub-cohort included only 3,104 participants (2.7% of the analytic sample), which may limit statistical power and raise concerns about potential selection bias, as participants who returned for repeat assessments may differ from the overall cohort. Moreover, due to the data availability, we only used two time-point measurement of TyG-related indices to identify trajectories, which limited our ability to identify more nuanced changes. These factors may have limited the precision and reliability of trajectory estimation, reduced statistical power, and partly explained the non-significant associations observed for certain outcomes such as CHD and stroke. Therefore, findings from the trajectory analyses should be interpreted with caution. Third, the use of non-fasting blood samples may introduce measurement error in the TyG index, which was originally defined using fasting glucose and triglycerides, potentially attenuating or biasing the observed associations. To address this concern, we conducted additional sensitivity analyses by further adjusting for fasting time, and the results remained consistent with the main findings. Nevertheless, some residual measurement error is still possible, and future studies using strictly fasting samples or repeated biomarker assessments are warranted to further clarify this issue. Fourth, the study population was heterogeneous in terms of age, comorbidities, and medication use, which may have introduced variability in baseline risk profiles, differential susceptibility to cardiovascular events, and potential interactions with TyG-related indices. Finally, the generalizability of our findings to populations with different ethnic backgrounds, healthcare systems, or lifestyle and environmental contexts remains uncertain and warrants further validation in diverse cohorts.

Conclusion

In summary, this study provided comprehensive evidence that TyG-related indices, including TyG-BMI, TyG-WC, and TyG-WHtR, were significant predictors of CVD incidence, CVD mortality, and all-cause mortality in individuals with MetS. These associations were frequently nonlinear, underscoring the importance of threshold effects in risk assessment. Notably, longitudinal increases in these indices further improved CVD risk prediction, underscoring the importance of dynamic monitoring. Furthermore, TyG-WC demonstrated the greatest incremental value for total CVD, CVD mortality, and all-cause mortality, making it the preferred index for assessing overall cardiovascular and mortality risk. TyG-WHtR performed best for CHD and stroke, suggesting greater utility in identifying patients at risk of specific atherosclerotic events. These findings highlight the clinical application of TyG-related indices as simple and cost-effective tools for outcome-specific risk stratification in MetS. Using TyG-WC or TyG-WHtR according to the cardiovascular endpoint may support earlier identification of high-risk patients and more targeted prevention. Further studies should validate these results in diverse populations and establish clinically meaningful thresholds.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ABSI

A body shape index

AF

Atrial fibrillation

AHA

American heart association

ARIC

Atherosclerosis risk in communities

BMI

Body mass index

BRI

Body roundness index

CARDIA

Coronary artery risk development

CHD

Coronary heart disease

CIs

Confidence intervals

CKD

Chronic kidney disease

CKM

Cardiovascular-kidney-metabolic

CRP

C-reactive protein

CVD

Cardiovascular disease

FBG

Fasting blood glucose

HbA1c

Glycated hemoglobin

HBP

High blood pressure

HDL-C

High-density lipoprotein cholesterol

HF

Heart failure

HRs

Hazard ratios

IDF

International diabetes federation

IDI

Integrated discrimination improvement

IQR

Interquartile ranges

IR

Insulin resistance

LDL-C

Low-density lipoprotein cholesterol

MASLD

Metabolic dysfunction-associated steatotic liver disease

MetS

Metabolic syndrome

NAFLD

Non-alcoholic fatty liver disease

NHANES

National health and nutrition examination survey

NHLBI

National heart, lung, and blood institute

NRI

Net reclassification index

RCS

Restricted cubic splines

SD

Standard deviations

T2DM

Type 2 diabetes mellitus

TG

Triglyceride

TyG

Triglyceride-glucose

WC

Waist circumference

WHtR

Waist-to-height ratio

Author contributions

DLH and YQH conceived and designed the study, DLH wrote the initial manuscript, YQH performed the data analysis and visualization, XXN and ZYB revised the manuscript critically for important intellectual content. All authors contributed to the interpretations of the findings and reviewed the manuscript. All authors read and approved the final manuscript.

Funding

This research received no external funding.

Data availability

Data used in this study are available from the UK Biobank team ([http://www.ukbiobank.ac.uk/]) after a successful application. The data and methods that support the findings of this study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Ethical approval for the UK Biobank study was granted by the Northwest Multicenter Research Ethics Committee (Ref: 21/NW/0157), and all participants provided written informed consent prior to enrollment.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.