Association between triglyceride-glucose index combined with a body shape index and atherosclerotic cardiovascular disease risk varies by glycemic status: insights from the UK Biobank Authors: Xinyi Shao, Zhaofu Tan, Lifu Sun, Yidian Fu, Aijun Chen, Jin Chen, Genlong Bai, Jingbo Zhang Published: 2025 ABSTRACT AbstractObjectiveThe triglyceride-glucose (TyG) index, a measure of insulin resistance, has been confirmed to be associated with adverse clinical outcomes. A new composite indicator, the TyG-A body type index (TyG-ABSI), was developed by integrating the TyG index and the A body type index. This study aimed to thoroughly investigate the association between the TyG-ABSI and the risk of atherosclerotic cardiovascular disease (ASCVD) among individuals with normoglycemia, dysglycemia, and diabetes.MethodsParticipants from the UK Biobank were included and categorized into 3 groups: normoglycemia, dysglycemia (prediabetes), and diabetes. The primary study outcome was the incidence of all-cause ASCVD. Cox regression and restricted cubic spline (RCS) analyses were performed to evaluate the linear and nonlinear associations between the TyG-ABSI and ASCVD. Furthermore, time-dependent receiver operating characteristic (ROC) curves were constructed to evaluate the discriminative performance of the TyG-ABSI and other indices. Additional analyses, including Kaplan–Meier survival curves, subgroup analyses, and sensitivity analyses, were conducted to assess robustness. A mediation analysis was performed to identify potential biomarkers.ResultsDuring an average follow-up of 14.8 years, a total of 29,680 ASCVD events were documented. The results indicated that elevated TyG-ABSI values were positively associated with the risk of ASCVD in the population with normoglycemia (nonlinear, P for nonlinear < 0.001) and diabetes (linear, P for nonlinear = 0.96) (HR = 1.07, 95% CI: 1.05–1.09; HR = 1.10, 95% CI: 1.06–1.14, respectively). No significant associations were detected in the dysglycemia group. Compared with the TyG index and other TyG-derived metrics, the TyG-ABSI demonstrated superior predictive performance (e.g., 10-year AUC = 0.634 in the normoglycemia group). Subgroup analyses and sensitivity analyses confirmed the robustness of our findings. Moreover, the results of the mediation analysis demonstrated that white blood cell (WBC) count (mediation proportions: 3.74–9.73%) and C-reactive protein (CRP) level (mediation proportions: 9.96–26.89%) significantly mediated the association between the TyG-ABSI and ASCVD.ConclusionsIn the normoglycemia subgroup, the association between the TyG-ABSI and ASCVD was nonlinear. Moreover, the TyG-ABSI was linearly associated with an increased risk of ASCVD in the diabetes subgroup but not in the dysglycemia subgroup. The predictive value of the TyG-ABSI across different glycemic statuses provides new evidence for medical practice. Furthermore, the TyG-ABSI may serve as a useful tool for identifying high-risk individuals within the seemingly low-risk normoglycemic population and for further risk identification among diabetic patients.Graphical abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12933-025-03021-y.Keywords: Triglyceride-glucose index, A body shape index, Atherosclerotic cardiovascular disease, Insulin resistance, UK Biobank Research insights What is currently known about this topic? Central obesity contributes to atherosclerosis, and insulin resistance increases ASCVD incidence. Central obesity contributes to atherosclerosis, and insulin resistance increases ASCVD incidence. The TyG index reflects insulin resistance, while ABSI better captures the risk of central obesity. The TyG index reflects insulin resistance, while ABSI better captures the risk of central obesity. Combined indices may improve comprehensive predictive utility for ASCVD incidence. Combined indices may improve comprehensive predictive utility for ASCVD incidence. What is the key research question? How does the TyG-ABSI associate with ASCVD incidence across different glycemic status? How does the TyG-ABSI associate with ASCVD incidence across different glycemic status? What is new? This is the first large-scale study demonstrating TyG-ABSI-ASCVD relationship variation. This is the first large-scale study demonstrating TyG-ABSI-ASCVD relationship variation. TyG-ABSI shows superior predictive value over other insulin resistance indices. TyG-ABSI shows superior predictive value over other insulin resistance indices. Inflammatory biomarkers (e.g. CRP & WBC) partially mediate TyG-ABSI-ASCVD relationship. Inflammatory biomarkers (e.g. CRP & WBC) partially mediate TyG-ABSI-ASCVD relationship. How might this study influence clinical practice? TyG-ABSI enables personalized risk stratification and intervention in varied glycemic populations. TyG-ABSI enables personalized risk stratification and intervention in varied glycemic populations. Introduction Atherosclerotic cardiovascular disease (ASCVD) is among the most common causes of death, accounting for more than 60% of global cardiovascular deaths, and its burden continues to increase at an alarming rate [1–3]. Metabolic dysfunction is considered the main factor of its pathogenesis; thus, optimizing the present tactics to identify and manage metabolic risk factors is critical for superior ASCVD prevention [4, 5]. Insulin resistance (IR) is a state of pathophysiological impairment in which the body’s sensitivity and response to insulin are reduced, resulting in the inability of insulin to efficiently transport glucose into cells, which causes metabolic abnormalities [6–8]. Previous studies have suggested that it is closely related to the occurrence and development of ASCVD [9]. The gold standards for evaluating IR, including high-insulin normal blood glucose clamp and intravenous glucose tolerance tests, are complex and invasive; therefore, these methods are not suitable for clinical research [10]. Homeostasis model assessment of insulin resistance (HOMA-IR) is a validated alternative evaluation metric [11]. However, circulating insulin concentrations have not yet been widely used in clinical practice. Recently, the TyG index has been used as a simple, convenient, and cost-effective alternative for detecting IR [12]. The TyG index can predict not only the risk and severity of CVD but also adverse clinical outcomes associated with CVD [13, 14]. In addition, central obesity significantly contributes to the development of atherogenic endothelial dysfunction, partly because of metabolic disorders, eventually resulting in ASCVD [15, 16]. However, the predictive power of a single indicator is still limited. To improve predictive performance, researchers have developed a series of TyG-derived indices by combining the TyG index with obesity-related parameters, such as body mass index (BMI), waist-to-height ratio (WHtR), and body roundness index (BRI) [17–20]. When employed in cardiovascular disease (CVD) prognosis, BMI often yields an “obesity paradox” [21]. Krakauer et al. introduced the A Body Shape Index (ABSI), which standardizes waist circumference by height and BMI [22]. This index more accurately reflects abdominal fat distribution independent of total body size. Building on this concept, a novel composite indicator, the TyG-ABSI, was developed by combining the TyG index with the ABSI [23]. This indicator integrates both IR and central obesity into a single measure, showing promise in predicting adverse health outcomes for certain populations [23–25]. A previous study suggested that the TyG-ABSI is significantly associated with CVD, cardiovascular mortality, and all-cause mortality in populations with MetS, providing new evidence for medical practice and public health interventions [24]. In addition, previous studies have shown that inflammatory factors (e.g., C-reactive protein [CRP]) are markers of pathways in ASCVD, IR and central obesity and are associated with inflammatory biomarkers [26–31]. However, the association between Tyg-ABSI and ASCVD risk remains unclear, particularly in individuals with different glycemic statuses. Therefore, this study is based on the UK Biobank cohort to investigate the association of the TyG-ABSI index with ASCVD in patients with different glycemic statuses, to explore the potential mediating roles of inflammatory biomarkers in this relationship, and to provide more evidence for the practical application of the TyG-ABSI index in the real world. Materials and methods Data source and study participants The present study utilized data obtained from the UK Biobank, a national-based large prospective cohort study in which more than 500,000 participants aged 37–73 years were recruited from more than 20 assessment centers in England, Scotland, and Wales from 2006–2010 [32]. The UK Biobank has received ethical approval from the North West Multicenter Research Ethics Committee (Reference No. 21/NW/0157). This research was conducted using the UK Biobank Resource under Application Number 747364 and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines [33]. In this study, 501,955 participants were included in the longitudinal cohort. We excluded participants who were diagnosed with ASCVD at baseline (n = 33,350), those lacking TyG index and ABSI calculation-related data (n = 75,571), and those with missing values for covariates and glycemic status (n = 100,518). The detailed flow chart and inclusion and exclusion criteria used in the present study are shown in Fig. 1 and Supplementary Table 1. Finally, 292,516 participants were included in the subsequent analyses, including 241,620 in the normoglycemia group, 36,212 in the dysglycemia group, and 14,684 in the diabetes group. Flow chart of the participants included and excluded Definition of glycemic status Glycemic status was evaluated using HbA1c (mmol/mol), which was measured through high-performance liquid chromatography analysis blood samples from non-fasting individuals. The units in mmol/mol were converted to percentages (%) as shown in the formula below [34]: Normoglycemia was defined as no diagnosis of diabetes or an HbA1c < 5.7%; dysglycemia was identified as no diagnosis of diabetes with an HbA1c ≥ 5.7 and < 6.5%; and diabetes was ascertained by combining self-reported or doctor-diagnosed diabetes, reported usage of antihyperglycemic medications or insulin, or a measured HbA1c ≥ 6.5% [35–37]. Definition of the TyG index, TyG-ABSI, and TyG-related indices The levels of biochemical indicators, such as CRP, fasting blood glucose (FBG), triglyceride (TG), low-density lipoprotein cholesterol (LDL), and high-density lipoprotein cholesterol (HDL), were measured from peripheral blood samples collected at baseline by the UK Biobank. All measurements were performed using a Beckman Coulter AU5800 automated clinical chemistry analyzer in the UK Biobank central laboratory. On the basis of previously validated studies [38–41], the TyG index was calculated with the following formula: Anthropometric measurements, including waist circumference (WC), standing height, and weight, were recorded at baseline in the UK Biobank during the physical assessments. Afterward, the ABSI was calculated using the following formula: 2. Next, the TyG-ABSI was calculated via the formula below: 3. The formulas used to further calculate the other TyG-related indices are as follows: 4. 5. 6. 7. 8. We also calculated other IR-related indices, including the metabolic score for insulin resistance (METS-IR) and the atherogenic index of plasma (AIP), via the formula below: 9. 10. Outcomes The primary outcome in this study was incident ASCVD. The diagnosis of ASCVD was obtained primarily through links with hospital admission records, primary care records, self-reports and death registration data. Following established precedents, the International Classification of Diseases (ICD) coding system was used to recognize the diagnosis of ASCVD, and the data field and code are detailed in Supplementary Table 2 [42–44]. Participants were followed from the baseline assessment (2006–2010) until the first occurrence of any ASCVD event, death, or the censoring date (September 01, 2024), whichever occurred first. Covariates Potential confounders, identified via a literature review [28, 45–48], included a range of demographic, socioeconomic, and health-related variables, including age; sex (male and female); educational level (college or above, high school or equivalent, and below high school); race and ethnicity (White and others); Townsend deprivation index (TDI); alcohol consumption status (current drinker, previous drinker, and never); smoking status (current smoker, previous smoker, and never); physical activity group (high level, moderate level, and low level); history of hypertension (yes and no); LDL levels; and HDL levels. The TDI was derived from the postcode of residence using aggregated data on unemployment, ownership of cars and houses, and household overcrowding. Physical activity groups were computed as metabolic equivalent (MET) hours per week on the basis of short-form International Physical Activity Questionnaire (IPAQ) guidelines. Statistical analysis The characteristics of the participants at baseline are described as the means ± standard deviations (SD) for continuous variables and as the frequency and percentage (N [%]) for categorical variables. Furthermore, the follow-up time was separately calculated from the baseline date to the date of the first ASCVD occurrence, death, or update date of linkages, whichever occurred first. The participants were subsequently stratified into 4 groups according to the TyG-ABSI quartiles (Q1 < 6.13, Q2 ≥ 6.13 and < 6.65; Q3 ≥ 6.65 and < 7.16; and Q4 ≥ 7.16). Statistical differences between groups were assessed using the Kruskal–Wallis test for continuous data and the chi-square test for categorical data. For the following longitudinal analysis, we examined the association of the TyG-ABSI with ASCVD in the general population and in different glycemic status groups using multivariable Cox proportional hazards regression analysis models to estimate HRs with 95% CIs. The models were adjusted for potential confounders. Model 1 was a crude model (without adjustment). Model 2 was adjusted for age, sex and ethnicity. Model 3, building on Model 2, was further adjusted for educational level, TDI, smoking status, alcohol consumption status, physical activity group, and history of hypertension, LDL, and HDL. Furthermore, the Kaplan–Meier method was used to compare the cumulative risk of ASCVD events among participants categorized by glycemic status, and the logarithmic rank test was used for statistical evaluation. Afterward, restricted cubic splines (RCSs) based on Model 3 with 3 knots at the 10th, 50th, and 90th percentiles of the TyG-ABSI variable were applied to account for potential nonlinear dose–response relationships. Furthermore, time-dependent receiver operating characteristic (ROC) curves were constructed, and the area under the curve (AUC) values were used to estimate the predictive value of the TyG-ABSI and other TyG-related indices. We also used the time-dependent Harrell’s C index to evaluate the predictive performance of the TyG-ABSI and other TyG-related indices for ASCVD incidence. The incremental predictive value of the TyG-ABSI was assessed using the net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indices, with clinical utility evaluated through decision curve analysis (DCA). Next, we performed subgroup analysis to investigate the robustness of the results. In brief, we conducted stratification analyses by age group (< 60 vs. ≥ 60 years), sex (male vs. female), smoking status (current vs. previous vs. never), alcohol consumption status (current vs. previous vs. never), history of hypertension (yes vs. no), and physical activity group (low vs. moderate vs. high). In addition, to increase the robustness of the results, several sensitivity analyses based on Model 3 were performed. First, we repeated the main analyses after excluding the ASCVD cases with < 2 years of follow-up to minimize the effect of subclinical conditions on the observed associations. Considering the effect of cholesterol-lowering medication on the TyG-related index, we further adjusted for medication use to validate the robustness of our findings. Furthermore, we adjusted for inflammatory markers (CRP level and white blood cell [WBC] count) to reduce bias in the present correlation results. Moreover, we conducted inverse probability of treatment weighting (IPTW) analysis to reduce study bias. To explore the mechanism underlying the association between the TyG-ABSI and the incidence of ASCVD, we conducted a mediation analysis to investigate the potential inflammatory biomarkers (all the mediators were measured at baseline concurrently with the TyG-ABSI assessment) linked to the TyG-ABSI and ASCVD. This mediation method estimated total effects (TE), direct effects (DE), and indirect effects (IE) as well as the proportion of mediation calculated using the 1000 times bootstrap method with a fully adjusted model (Model 3). The data analysis in the present study was performed between September 06, 2025, and October 01, 2025. All analyses were performed using R version 4.4.3. A two-sided P value of < 0.05 was set as the threshold for statistical significance. Results Basic characteristics of the participants A total of 292,516 participants in the longitudinal cohort from the UK Biobank were enrolled, with an average age of 55.9 ± 8.1 years, and 137,819 (47.1%) participants were male. Table 1 displays the characteristic indicators of participants stratified by different glycemic statuses: 241,620 (82.60%) with normoglycemia, 36,212 (12.38%) with dysglycemia, and 14,684 (5.02%) with diabetes. The results revealed that the participants in the normoglycemia group markedly differed from the participants in the other two groups. Briefly, compared with the normoglycemic group, the diabetic group was older; more likely to be male, smokers, and drinkers; had lower educational and physical activity levels; had a higher rate of hypertension history; had lower HDL and LDL levels; and had higher weight, WC, BMI, FBG level and TDI score the dysglycemic group was older; had lower educational and physical activity levels; was more likely to be smokers and nondrinkers; had a higher rate of hypertension history; had lower HDL levels; had higher LDL levels; and had higher weight, WC, BMI, FBG level and TDI score (Table 1). The mean TyG-ABSI scores for the normoglycemia, dysglycemia, and diabetes groups were 6.59, 6.89, and 7.35, respectively (Supplementary Fig. 1). Table 1Characteristics of participants by different glycemic statusCharacteristicParticipantsTotal (N = 292,516)Normoglycemia (N = 241,620)Dysglycemia (N = 36,212)Diabetes (N = 14,684)P valueAge (y), mean ± SD55.9 ± 8.155.3 ± 8.159.0 ± 7.158.7 ± 7.4 < 0.001TDI, mean ± SD− 1.5 ± 3.0− 1.6 ± 2.9 − 1.2 ± 3.1— 0.7 ± 3.3 < 0.001HDL (mmol/L), mean ± SD1.5 ± 0.41.5 ± 0.41.4 ± 0.41.2 ± 0.3 < 0.001LDL (mmol/L), mean ± SD3.6 ± 0.83.6 ± 0.83.7 ± 0.92.9 ± 0.9 < 0.001Height (cm), mean ± SD169.1 ± 9.3169.2 ± 9.3168.0 ± 9.3169.6 ± 9.3 < 0.001Weight (kg), mean ± SD77.9 ± 15.776.7 ± 15.081.1 ± 17.089.8 ± 18.7 < 0.001WC (cm), mean ± SD89.7 ± 13.288.4 ± 12.693.8 ± 13.6102.0 ± 14.5 < 0.001BMI (kg/m2), mean ± SD27.2 ± 4.626.7 ± 4.328.6 ± 5.131.1 ± 5.8 < 0.001FBG (mmol/L), mean ± SD5.1 ± 1.24.9 ± 0.65.2 ± 0.87.6 ± 3.5 < 0.001HbA1c (mmol/mol), mean ± SD35.7 ± 6.233.9 ± 3.040.8 ± 1.953.2 ± 14.5 < 0.001TyG, mean ± SD8.7 ± 0.68.6 ± 0.58.9 ± 0.59.2 ± 0.7 < 0.001TyG-ABSI, mean ± SD6.7 ± 0.76.6 ± 0.76.9 ± 0.77.3 ± 0.8 < 0.001TyG-BMI, mean ± SD237.2 ± 48.4231.4 ± 44.2255.1 ± 51.9288.7 ± 62.2 < 0.001TyG-WHtR, mean ± SD4.6 ± 0.84.5 ± 0.75.0 ± 0.85.6 ± 1.0 < 0.001TyG-BRI, mean ± SD35.6 ± 14.533.7 ± 13.041.4 ± 15.752.1 ± 19.4 < 0.001Gender, N (%)Male137,819 (47.1)112,138 (46.4)16,697 (46.1)8984 (61.2) < 0.001Female154,697 (52.9)129,482 (53.6)19,515 (53.9)5700 (38.8)Race and ethnicitya, N (%)White278,481 (95.2)232,581 (96.3)32,990 (91.1)12,910 (87.9) < 0.001Othersa14,035 (4.8)9039 (3.7)3222 (8.9)1774 (12.1)Educational level, N (%) ≥ College or above106,798 (36.5)91,480 (37.9)11,088 (30.6)4230 (28.8) < 0.001High school / equivalent147,897 (50.6)121,919 (50.5)18,522 (51.1)7456 (50.8) < High school37,821 (12.9)28,221 (11.7)6602 (18.2)2998 (20.4)Smoking status, N (%)Current29,100 (9.9)22,136 (9.2)5357 (14.8)1607 (10.9) < 0.001Previous100,271 (34.3)81,585 (33.8)12,628 (34.9)6058 (41.3)Never163,145 (55.8)137,899 (57.1)18,227 (50.3)7019 (47.8)Alcohol drinking, N (%)Current272,589 (93.2)227,093 (94.0)32,815 (90.6)12,681 (86.4) < 0.001Previous9315 (3.2)6935 (2.9)1463 (4.0)917 (6.2)Never10,612 (3.6)7592 (3.1)1934 (5.3)1086 (7.4)Physical activity group, N (%)Low53,151 (18.2)42,203 (17.5)7107 (19.6)3841 (26.2) < 0.001Moderate119,118 (40.7)98,401 (40.7)14,821 (40.9)5896 (40.2)High120,247 (41.1)101,016 (41.8)14,284 (39.4)4947 (33.7)Hypertension, N (%)No221,237 (75.6)190,677 (78.9)24,285 (67.1)6275 (42.7) < 0.001Yes71,279 (24.4)50,943 (21.1)11,927 (32.9)8409 (57.3)aIn the UK Biobank, other included Asian or Asian British (including Bangladeshi, Indian, Pakistani, and any other Asian background), Black or Black British (including African, Caribbean, and any other Black background), Chinese, mixed race (including Black and White African, Black and White Caribbean, Asian and White, any other mixed background), or other ethnic groupBMI: body mass index; SD: standard deviation, TDI, Townsend deprivation index; HDL, high-density-lipoprotein; LDL, low-density-lipoprotein; FBG, fasting blood-glucose; WC, waist circumference; TyG, triglyceride-glucose Characteristics of participants by different glycemic status aIn the UK Biobank, other included Asian or Asian British (including Bangladeshi, Indian, Pakistani, and any other Asian background), Black or Black British (including African, Caribbean, and any other Black background), Chinese, mixed race (including Black and White African, Black and White Caribbean, Asian and White, any other mixed background), or other ethnic group BMI: body mass index; SD: standard deviation, TDI, Townsend deprivation index; HDL, high-density-lipoprotein; LDL, low-density-lipoprotein; FBG, fasting blood-glucose; WC, waist circumference; TyG, triglyceride-glucose Additional analyses provided further insights into the participants’ baseline characteristics. Compared with the participants we excluded, the participants we included in the final analyses were younger, more likely to be male and white, and more likely to have higher educational levels and lower TDI scores (Supplementary Table 1). Association between the TyG-ABSI and risk of ASCVD During an average follow-up period of 14.8 years, 29,680 (10.14%) participants experienced their first occurrence of ASCVD. Analysis of the Kaplan–Meier cumulative incidence curve revealed a gradual increase in ASCVD events from the normoglycemia group to the diabetes group, with a statistically significant difference observed (log-rank test P < 0.001; Fig. 2A–C). Afterward, we evaluated the association between the TyG-ABSI and the risk of ASCVD among all participants using multivariable Cox proportional hazards regression (Supplementary Table 3), and we found that each one-SD increase in the TyG-ABSI was associated with an increased risk of ASCVD (HR = 1.12; 95% CI: 1.10–1.14). We subsequently investigated the association between the TyG-ABSI and the risk of ASCVD in participants with different glycemic statuses (Table 2, Supplementary Table 4). When the TyG-ABSI was treated as a continuous variable, the fully adjusted model (Model 3) indicated that each one-SD increase in the TyG-ABSI was associated with a significant increase in the risk of ASCVD in both normoglycemic and diabetic participants (HR = 1.07; 95% CI: 1.05–1.09; HR = 1.10; 95% CI: 1.06–1.14, respectively). When the TyG-ABSI was categorized into quartiles, a distinct dose–response relationship emerged in the normoglycemia and diabetes groups (P for trend < 0.001). Interestingly, when extreme quartiles were compared, the participants in the highest quartile (Q4) exhibited a markedly increased risk: the HR in the normoglycemia group was 1.20 (95% CI: 1.14–1.27), the HR in the dysglycemia group was 1.14 (95% CI: 1.01–1.29), and the HR in the diabetes group was 1.27 (95% CI: 1.05–1.55). The findings also revealed no statistically significant correlation between the continuous variables in the dysglycemia group (P > 0.05) (Table 2). Furthermore, we constructed smooth curves to investigate the potential nonlinear relationship between the TyG-ABSI and ASCVD based on RCS regression analysis (Fig. 2D–F). After adjusting for all potential confounders, a significant nonlinear (reversed L-shaped) correlation was revealed between the TyG-ABSI and ASCVD in participants with normoglycemia (P for nonlinearity < 0.001), and statistically significant linear associations were observed in the diabetes group (Fig. 2D and F). However, no relationship was found between the TyG-ABSI and ASCVD in the dysglycemia group (all P > 0.05) (Fig. 2E). The Kaplan–Meier curves analysis and RCS analysis of TyG-ABSI and ASCVD in different glycemic status. The Kaplan–Meier curves demonstrate clinical progression over time for incident ASCVD (normoglycemia A, dysglycemia B, and diabetes C), visualizing individuals with quartile TyG-ABSI Q1 (red lines), Q2 (green lines), Q3 (blue lines), and Q4 (purple lines) (P < 0.001). The shaded areas represent the 95% CIs. RCS regression for evaluated TyG-ABSI of ASCVD incidence according to continuous levels of the TyG-ABSI after adjusted as Model 3 (normoglycemia D, dysglycemia E, and diabetes F) Table 2Associations between TyG-ABSI and risk of ASCVDModel 1aModel 2bModel 3cOutcomesCases/NHR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P valueNormoglycemiaper SD increase21,360/2416201.54 (1.52, 1.57) < 0.0011.24 (1.22, 1.26) < 0.0011.07 (1.05, 1.09) < 0.001QuartilesQ12949/668501 (Reference)1 (Reference)1 (Reference)Q24878/629851.79 (1.71, 1.87) < 0.0011.31 (1.25, 1.38) < 0.0011.13 (1.08, 1.18) < 0.001Q36381/597062.51 (2.40, 2.62) < 0.0011.52 (1.45, 1.60) < 0.0011.17 (1.11, 1.23) < 0.001Q47152/520793.28 (3.15, 3.43) < 0.0011.75 (1.67, 1.84) < 0.0011.20 (1.14, 1.27) < 0.001P for trend < 0.001 < 0.001 < 0.001Dysglycemiaper SD increase4991/362121.38 (1.34, 1.42) < 0.0011.19 (1.15, 1.23) < 0.0011.04 (1.00, 1.08)0.05QuartilesQ1378/50991 (Reference)1 (Reference)1 (Reference)Q2903/81621.52 (1.35, 1.71) < 0.0011.29 (1.14, 1.46) < 0.0011.10 (0.97, 1.24)0.13Q31427/101661.97 (1.76, 2.21) < 0.0011.46 (1.29, 1.64) < 0.0011.11 (0.99, 1.26)0.08Q42283/127852.56 (2.29, 2.85) < 0.0011.66 (1.48, 1.87) < 0.0011.14 (1.01, 1.29)0.04P for trend < 0.001 < 0.0010.08Diabetesper SD increase3329/146841.23 (1.19, 1.27) < 0.0011.16 (1.12, 1.20) < 0.0011.10 (1.06, 1.14) < 0.001QuartilesQ1123/9791 (Reference)1 (Reference)1 (Reference)Q2304/17831.40 (1.14, 1.73)0.0021.19 (0.97, 1.47)0.101.08 (0.87, 1.33)0.49Q3661/32331.71 (1.41, 2.07) < 0.0011.30 (1.07, 1.58)0.0081.10 (0.91, 1.35)0.33Q42241/86892.23 (1.86, 2.68) < 0.0011.61 (1.33, 1.93) < 0.0011.27 (1.05, 1.55)0.02P for trend < 0.001 < 0.001 < 0.001CI: confidence interval; HR, hazardous ratio, BMI, body mass index; TDI, Townsend deprivation index; HDL, high-density-lipoprotein; LDL, low-density-lipoprotein; TyG, triglyceride-glucose; ABSI, a body shape indexaModel 1: Without adjustmentbModel 2: Adjusted for age, gender, and race/ethnicitycModel 3: Adjusted for age, gender, and race/ethnicity, education level, TDI, smoking status, alcohol drinking, physical activity group, history of hypertension, HDL, and LDL Associations between TyG-ABSI and risk of ASCVD CI: confidence interval; HR, hazardous ratio, BMI, body mass index; TDI, Townsend deprivation index; HDL, high-density-lipoprotein; LDL, low-density-lipoprotein; TyG, triglyceride-glucose; ABSI, a body shape index aModel 1: Without adjustment bModel 2: Adjusted for age, gender, and race/ethnicity cModel 3: Adjusted for age, gender, and race/ethnicity, education level, TDI, smoking status, alcohol drinking, physical activity group, history of hypertension, HDL, and LDL Results of ROC analysis In our previous analysis, the TyG-ABSI was not significantly associated with ASCVD in the dysglycemic subpopulation. Therefore, the following analyses were conducted only in the normoglycemia and diabetes groups. On the basis of time-dependent ROC curve evaluation, the TyG-ABSI had a modest predictive capacity for ASCVD in the normoglycemia group (AUC: 0.643, 0.635, and 0.634) and in the diabetes group (AUC: 0.582, 0.582, and 0.567) at 3-year, 5-year, and 10-year intervals, respectively, outperforming the TyG index (normoglycemia group; AUC: 0.597, 0.595, and 0.594), other TyG-derived indices (e.g., the TyG-BMI in the normoglycemia group; AUC: 0.602, 0.601, and 0.595), and other insulin resistance indices (e.g., the AIP in the normoglycemia group; AUC: 0.612, 0.608, and 0.604) (Fig. 3A–F; Supplementary Fig. 2). Furthermore, the long-term discriminatory ability of the TyG-ABSI was evaluated over a 3- to 10-year predictive timeframe using the C-index (Supplementary Fig. 3). The TyG-ABSI outperformed the other TyG-derived indices in predicting ASCVD in both the normoglycemia and diabetes groups. Notably, the ability of the TyG-ABSI to predict ASCVD is better in individuals with normoglycemia than in individuals with diabetes. Time-dependent Receiver Operating Characteristic analysis of TyG-ABSI in predicting ASCVD. ROC curves and AUC for the associations of TyG index, TyG-ABSI, and other TyG-related indices with ASCVD in participants with normoglycemia at 3-year A, 5-year B, and 10-year C. ROC curves and AUC for the associations of TyG index, TyG-ABSI, and other TyG-related indices with ASCVD in participants in diabetes at 3-year D, 5-year E, and 10-year (F) Subgroup analyses To delve deeper into the correlation between the TyG-ABSI and the likelihood of ASCVD in participants with different glycemic statuses, we performed subgroup and interaction analyses between the TyG-ABSI and ASCVD and identified population variances, as presented in Fig. 4. The results revealed that age, sex, and history of hypertension significantly affected the association between the TyG-ABSI and ASCVD in participants with normoglycemia (P for interaction < 0.001) (Fig. 4A). However, in the diabetes group, there was no significant interaction effect on the association between the TyG-ABSI and ASCVD (all P values for interactions > 0.05) (Fig. 4B). Subgroup analysis for incident ASCVD. Each stratification was adjusted for all co-variables except the stratification factor itself (normoglycemia A and diabetes B). Squares indicate the HRs, while the lines showing their 95% CIs Sensitivity analyses To evaluate the robustness of our findings regarding the association between the TyG-ABSI and ASCVD in the normoglycemia and diabetes groups, we performed the following sensitivity analyses (Table 3). To reduce bias from undiagnosed preexisting conditions, we excluded participants who developed ASCVD within 2 years of follow-up. Participants with elevated TyG-ABSI scores had a significantly greater risk of ASCVD than those with lower TyG-ABSI scores did in both groups (HR = 1.07; 95% CI: 1.05–1.09; HR = 1.10; 95% CI: 1.05–1.14, respectively) (Table 3). Furthermore, in the IPTW analyses, the association of the TyG-ABSI with ASCVD in the normoglycemia and diabetes groups was also confirmed (HR = 1.08, 95% CI: 1.05–1.12; HR = 1.12, 95% CI: 1.05–1.19, respectively) (Table 3). Table 3Sensitivity analysesAnalysisAdjusted HRa (95% CI)P valueExcluding participants diagnosed within 2 years follow-upNormoglycemia1.07 (1.05, 1.09) < 0.001Diabetes1.10 (1.05, 1.14) < 0.001Adjusted for medication usingNormoglycemia1.07 (1.05, 1.09) < 0.001Diabetes1.10 (1.06, 1.14) < 0.001Adjusted for inflammatory markersNormoglycemia1.06 (1.04, 1.09) < 0.001Diabetes1.09 (1.05, 1.13) < 0.001IPTW analysesNormoglycemia1.08 (1.05, 1.12) < 0.001Diabetes1.12 (1.05, 1.19) < 0.001CI: confidence interval; HR, hazardous ratio; TDI, Townsend deprivation index; IPTW, Inverse Probability Treatment Weight; HDL, high-density-lipoprotein; LDL, low-density-lipoproteinaAdjusted for age, gender, and race/ethnicity, education level, TDI, smoking status, alcohol drinking, physical activity group, history of hypertension, history of diabetes, HDL, and LDL Sensitivity analyses CI: confidence interval; HR, hazardous ratio; TDI, Townsend deprivation index; IPTW, Inverse Probability Treatment Weight; HDL, high-density-lipoprotein; LDL, low-density-lipoprotein aAdjusted for age, gender, and race/ethnicity, education level, TDI, smoking status, alcohol drinking, physical activity group, history of hypertension, history of diabetes, HDL, and LDL Incremental value and clinical utility of TyG-ABSI To further validate the incremental predictive value of the TyG-ABSI, the NRI and IDI were evaluated (Supplementary Table 5). Notably, when the TyG-ABSI was incorporated into the model comprising core variables from the Systematic Coronary Risk Estimation 2 (SCORE2), which typically incorporates age, sex, non-HDL, smoking status, and systolic blood pressure, improvements were observed. The NRI of the normoglycemia group was 0.120 (P < 0.001), with a significant IDI of 0.001 (P < 0.001). The NRI of the diabetes group was 0.114 (P < 0.001), with a significant IDI of 0.002 (P < 0.001). In addition, the consistent superiority of the TyG-ABSI in terms of net benefit, as determined by DCA across 3-, 5-, and 10-year prediction windows for ASCVD outcomes in both the normoglycemic and diabetes groups, supports its superior clinical utility in risk stratification (Supplementary Fig. 4). Mediation analyses We performed mediation analyses in the normoglycemic and diabetic population, and the results indicated that several inflammatory biomarkers, particularly CRP and WBC, partially mediated this relationship (Fig. 5 and Supplementary Table 6). Briefly, for the TyG-ABSI-CRP pathway, the IE values of ASCVD in the normoglycemia and diabetes groups were 0.0007 and 0.0009, respectively (Fig. 5A–B). With respect to the TyG-ABSI-WBC pathway, the IE values of ASCVD in the two groups were 0.0018 and 0.0022, respectively (Fig. 5C–D). Notably, there were higher mediation proportions as a result of WBCs (from 9.96 to 26.89%) than there were from CRP levels (from 3.74 to 9.73%) across different glycemic status populations (Fig. 5A–D). Mediation analyses between TyG-ABSI and ASCVD. Mediating effect of CRP on the association between TyG-ABSI and ASCVD in normoglycemia group A and diabetes group B. Mediating effect of WBC on the association between TyG-ABSI and ASCVD in normoglycemia group C and diabetes group D Discussion This is the first study to assess the association of the TyG-ABSI index with ASCVD in patients with different glycemic statuses. Our study suggests that the TyG-ABSI is associated with a higher risk of ASCVD in participants with normoglycemia, while no significant association was observed in the dysglycemia group. Among participants with diabetes, there was a linear relationship between the TyG-ABSI and ASCVD risk. These associations exhibited a clear dose–response relationship, remained robust across a series of sensitivity analyses, and were partly mediated by systemic inflammation markers (CRP level and WBC count). Furthermore, time-dependent ROC analyses indicated that compared with the TyG index or other related parameters alone, the TyG-ABSI had a superior ability to predict ASCVD, particularly over the medium to long term. Although several studies have investigated the association between the TyG index and ASCVD risk, most examined the TyG index independently and did not include anthropometric measurements [19, 49–52]. The absence of this could undermine the accuracy of predictions. For example, Xia X et al. reported that compared with TyG, TyG-WC and TyG-WHtR had superior abilities to predict incident ASCVD [19]. Therefore, recent research has increasingly focused on combining the TyG index with obesity-related indicators such as BMI, WC, and WHtR to develop composite indices, particularly in CVD studies [53–56]. In this context, this study utilized nationally representative UK Biobank data to comprehensively examine the association between the newly developed TyG-ABSI and ASCVD risk in patients with different glycemic statuses and compared its predictive performance with that of other established TyG-related indices. Given its recent development, the efficacy of the TyG-ABSI in ASCVD populations with different glycemic statuses is still unclear. Our research findings address this gap and emphasize its outstanding predictive ability. In our study, we analyzed data from 292,516 participants included in the UK Biobank database. The results revealed a significant correlation between the increase in the TyG-ABSI and the incidence of ASCVD, and the association was most pronounced in the normoglycemia group. These findings indicate that the TyG-ABSI is particularly valuable for identifying individuals at high risk of ASCVD in seemingly low-risk populations with a normal blood glycemic status. Despite having a normal blood glycemic status, these individuals still exhibited IR and central obesity. Early identification of this high-risk subgroup can provide a critical window for enhancing lifestyle interventions to prevent the onset of ASCVD. To elucidate the potential mechanisms linking the TyG-ABSI to ASCVD, we conducted mediation analyses. Our results indicated that systemic inflammatory markers, namely, the CRP level and WBC count, mediated a modest portion of this relationship. However, most of the association remains unexplained by the inflammatory markers we examined in the present study, which highlights an interesting area for future research. This study suggested that an elevated TyG-ABSI may be associated with the risk of ASCVD in populations with different blood glucose levels. The exact biological mechanism may require further exploration, but there are several possible explanations, such as the combined effect of insulin resistance and abdominal obesity. IR can reduce cellular glucose uptake, inducing metabolic disorders, including hyperglycemia and dyslipidemia [57, 58]. These metabolic abnormalities can generate excess reactive oxygen species (ROS) through the formation of advanced glycation end products and activation of the protein kinase C (PKC)–nicotinamide adenine dinucleotide phosphate oxidase (NADPH) pathway, reducing the bioavailability of nitric oxide. This pathway impairs endothelial function and upregulates the expression of adhesion molecules such as vascular cell adhesion molecule-1 (VCAM-1), leading to the release of proinflammatory cytokines (such as IL-6 and TNF-α) and accelerating atherosclerosis, which further leads to ASCVD [59–64]. In addition, inflammation can exacerbate IR, thereby releasing inflammatory mediators from tissues and further exacerbating systemic inflammatory responses. Inflammation and insulin resistance synergistically promote and exacerbate one another, thereby increasing the risk of ASCVD. Compared with the TyG, the TyG-ABSI includes a measurement method for abdominal obesity. Abdominal obesity, particularly visceral fat, can result in the production of more inflammatory factors and fatty acids, further exacerbating IR [65]. An imbalance in the metabolism of inflammatory cytokines and adipokines can interfere with insulin signaling, exacerbate systemic inflammation, promote macrophage-mediated vascular damage, and ultimately lead to the occurrence of ASCVD [66]. The results of this study provide additional indirect support for the findings of previous studies, indicating that systemic inflammation may act as a mediator between the TyG-ABSI and ASCVD in patients with different glycemic statuses. This study elucidated the role of the TyG-ABSI in ASCVD, including with different glycemic statuses, to provide a comprehensive approach for clinical decision-making on the basis of previous evidence. Its predictive ability stems from the synergistic effect of IR and central obesity, surpassing traditional risk factors and providing important value for early risk stratification. The number of ASCVD patients is increasing globally, particularly those with inflammation and IR, who are more prone to ASCVD events. Therefore, it is necessary to promptly identify high-risk patients with ASCVD. However, while the TyG-ABSI provides a convenient summary measure at the population level, its interpretation in clinical practice requires caution, and future studies should explore more sophisticated modeling approaches. This study presented several significant advantages. First, this study explored for the first time the correlation between the TyG-ABSI and ASCVD incidence according to blood glucose status. Second, designing this study as a prospective, nationwide longitudinal cohort study with a large sample size made our results more stable and universal. Third, we conducted a subgroup analysis to evaluate the consistency of the results across different population characteristics, which provides valuable insights for clinical practice. In addition, the TyG-ABSI serves as a simple and easy-to-use assessment tool in clinical practice, enhancing its practical application and assisting clinicians in implementing interventions such as improving dietary habits and promoting physical exercise. Finally, the results of the mediation study suggest that the CRP level and WBC count mediate the relationship between the TyG-ABSI and ASCVD, highlighting a simple and affordable biomarker of inflammation levels and providing mechanistic support for ASCVD progression. Several limitations in the present study warrant consideration. First, this was an observational study; thus, we cannot infer a causal role of the TyG-ABSI in ASCVD. Although our study identified this interaction, establishing causality requires further investigations, including experimental designs and randomized controlled trials. Second, our exposure variables were measured only at baseline. These parameters can fluctuate over time, and a single measurement may not fully represent long-term exposure levels, which could lead to an underestimation of their true association with ASCVD risk. Future research should consider the dynamic effects of blood glucose status and TyG-ABSI levels over time to more comprehensively evaluate this relationship. Third, despite adjusting for many covariates, residual bias may remain due to unmeasured or uncontrolled variables. Our null finding in the dysglycemia (prediabetes) group limited the clinical utility of the TyG-ABSI, which requires further investigation in other cohorts. Moreover, the modest predictive performance may limit the results of the present study. Finally, as the UK Biobank database primarily comprises UK adults aged 37 years and older who tend to be healthier, have a higher socioeconomic status, and may exhibit a “healthy volunteer” bias, the generalizability of our findings to younger populations, other ethnic populations, and other countries with different income levels and lifestyle patterns requires validation in future studies. Conclusion The relationship between the TyG-ABSI and ASCVD varied according to different glycemic statuses. In the normoglycemia subgroup, the association between the TyG-ABSI and ASCVD was nonlinear. The TyG-ABSI was linearly associated with an increased risk of ASCVD in the diabetes subgroup but not in the dysglycemia subgroup. Compared with the TyG index and other TyG-related indices, the TyG-ABSI was a better predictor of ASCVD in normoglycemic or diabetic subjects. Our results in the present study provide actionable evidence for health interventions in populations with different glycemic statuses and improve resource allocation in the management of ASCVD. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements All authors show their gratitude to the personnel who contributed to the UK Biobank (Approved Research ID: 747364, Approval date: 24 June 2025). Abbreviations Analysis of variance Body mass index Confidence intervals Hazard ratios International classification of diseases International physical activity questionnaire Inverse probability of treatment weighting Metabolic equivalent Standard deviation Strengthening the Reporting of observational studies in epidemiology Townsend deprivation index Triglyceride-glucose A body shape index Waist-to-height ratio Body roundness index Cardiovascular disease Atherosclerotic cardiovascular disease C-reactive protein Insulin resistance Fasting blood glucose Triglycerides Low-density lipoprotein cholesterol High-density lipoprotein cholesterol Waist circumference Metabolic score for insulin resistance Atherogenic index of plasma Receiver operating characteristic Area under the curve Integrated discrimination improvement Net reclassification improvement Decision curve analysis White blood cell count Reactive oxygen species Protein kinase C Nicotinamide adenine dinucleotide phosphate oxidase Vascular cell adhesion molecule-1 Author contributions Conceptualization: JB Zhang, XY Shao; Methodology and Statistical analysis: JB Zhang, XY Shao; Data visualization: JB Zhang, XY Shao, LF Sun; Original draft preparation: JB Zhang, XY Shao, ZF Tan; Draft review and editing: YD Fu, XY Shao, JB Zhang, ZF Tan, GL Bai, LF Sun, AJ Chen, and J Chen; Funding acquisition: JB Zhang, XY Shao, ZF Tan, and AJ Chen. All authors have reviewed and consented to the final version of the manuscript. Funding This research was funded by China Postdoctoral Science Foundation (grant number: 2024M763893); Outstanding Youth Foundation of Chongqing Province (grant number: CSTB2024YCJH-KYXM0101); The Innovative Research Project for Doctoral Students of the First Clinical College of Chongqing Medical University (grant number: CYYY-BSYJSKYCXXM202428). The First Clinical College Clinical medicine first-class discipline construction project to department of Dermatology and Venereology to Zhaofu Tan (CYYY-BSYJSCXXM-202333); Natural Science Foundation Project of Chongqing (grant number: 2024NSCQ-LZX0086). Data availability Data could be accessed directly through the UK Biobank. The code will be shared on request. Declarations Ethics approval and consent to participate The study was approved by the North West Multi-center Research Ethics Committee (Reference No. 11/NW/0382) in 2011 and renewed in 2016 (16/NW/0274) and 2021 (21/NW/0157). Study participants provided written informed consent before enrolment. Consent for publication Not Applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement Data could be accessed directly through the UK Biobank. The code will be shared on request.