Abstract
Aims
The triglyceride-cholesterol-body weight index (TCBI) has been proposed as a biomarker of the prognosis of stroke. However, the relationship between cumulative exposure and dynamic trajectories of the TCBI (cuTCBI/TCBI changes) and the risk of cardiovascular disease (CVD) remains unclear. The aim of this study is to examine the association between cuTCBI/TCBI changes and the risk of CVD.
Methods and results
The participants were from the China Health and Retirement Longitudinal Study (CHARLS). Partitioning Around Medoids (PAM) algorithm, Cox regression model, restricted cubic spline (RCS), Kaplan-Meier analysis and subgroup analysis were employed to evaluate the association between cuTCBI/TCBI changes and CVD risk. In model 3, compared to Q1, the risk of CVD increased by 83% for Q4 (HR:1.83, 95%CI: 1.49,2.24) (P for trend < 0.001). RCS analysis revealed a significant nonlinear relationship between cuTCBI and CVD risk. The results of PAM clustering demonstrated that the participants in moderate increase cluster (HR:1.35, 95%CI: 1.17,1.56, P < 0.001) and persistently high level cluster (HR:1.49, 95%CI: 1.22,1.83, P < 0.001) demonstrated a significantly elevated risk of CVD compared with the stable low risk cluster, with a significant trend towards increased risk across clusters (P for trend < 0.001). In the gender-stratified analysis, significant associations were observed in males, whereas the female TCBI change clusters did not reach statistical significance in the fully adjusted model.
Conclusion
Our research demonstrated a significant association between cuTCBI/TCBI changes and the increased risk of CVD. And the relationship between cuTCBI and CVD risk is nonlinear. The association between TCBI change clusters and outcomes may differ by gender, with statistical significance observed only in males. Our results indicated that regular monitoring of cuTCBI levels and maintaining relatively low levels may be crucial for CVD prevention.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-28341-w.
Keywords: Cardiovascular disease, Triglyceride-cholesterol-body weight index, PAM clustering, CHARLS database
Introduction
CVD is a common chronic disease among middle-aged and elderly individuals. Its prevalence rises markedly with age, making it a frequent cause of disability and death among older patients [1, 2]. Research from the Global Cardiovascular Disease Task Force predicts that, by 2025, over 5 million men and 2.8 million women worldwide will die prematurely from cardiovascular disease [3]. Informal care costs for CVD patients are rapidly increasing, reaching 61 billion dollars in 2015 and projected to double to 128 billion dollars by 2035, imposing a heavy economic burden on society and families [4].
The TCBI is a novel, simple nutritional index comprising serum triglycerides (TG), total cholesterol (TC), and body weight (BW) [5]. Compared with traditional lipid markers such as LDL, HDL, and TG/HDL, TCBI has the advantage of linking lipid characteristics to body composition (weight) and capturing the synergistic effects between metabolism and nutrition, simplifying risk stratification with a comprehensive score [6]. Previous studies have shown that TCBI is a prognostic indicator for coronary heart disease [5], heart failure [7], and has a significant negative correlation with cognitive function [8] and stroke [9]. In the current academic consensus, there are still some differences on the relationship between TCBI and the incidence rate of CVD. For instance, Rezaee et al. [10] suggested that there is a positive correlation between TCBI and increased CVD mortality. The reason for this difference is that the TCBI index obtained from blood markers at a single time point is unstable and cannot reflect the changes caused by long-term metabolic accumulation well. Therefore, we propose cumulative TCBI(cuTCBI) to represent cumulative average exposure and address the limitations of single-time-point measurements. We divided the TCBI trajectory into three different patterns using PAM clustering analysis to better describe its dynamic changes. The similar methods have been proven in previous literature [11–13].
Methods
Study design
The data for this study comes from the CHARLS database, which is a nationally representative longitudinal dataset collected using a multi-stage stratified sampling method, including middle-aged and elderly people aged at least 45 years old. In 2011, a total of 28 provinces, 150 counties, and 450 villages in China participated in the first round of baseline surveys targeting adults (aged ≥ 45). In this survey, 17,708 adults participated and filled out standard questionnaires on demographic characteristics, health status, and blood testing, which were later updated during the follow-up process.The research methods and data collection procedures of CHARLS have been described in previous literature [14].
The present research was carried out in accordance with the Declaration of Helsinki and received approval from the Biomedical Ethics Review Panel of Peking University (IRB 00001052_11015). Data were collected by trained field personnel, and informed consent was obtained from all participants [14]. For more information, refer to the official website: http://charls.pku.edu.cn/en.
Study population
In this study, the baseline time point was defined as the 2011 survey wave. The exposure assessment window spanned from 2011 to 2015, during which cumulative TCBI was calculated and PAM clusters were assigned. Participants who developed CVD or died during this window were excluded. Follow-up for CVD outcomes began in 2015.The follow-up continued until the occurrence of CVD or 2020, whichever came first. The time of CVD occurrence was defined as the number of years from the occurrence year minus 2015 [15]. Follow-up time was calculated from 2015 to the CVD occurrence or 2020. Participants were excluded if they were aged 45 years or younger, or if they had a history of heart disease, stroke, or cancer. Subjects with missing CVD data at baseline were also excluded, as well as those with missing baseline anthropometric, health-related, sociodemographic, or biomarker data. The detailed inclusion and exclusion process is shown in Fig. 1. A total of 4246 participants were included in the final analysis cohort.
Fig. 1.
Flowchart of the study population
Assessment of cuTCBI / the change in TCBI
The values of the TCBI and cuTCBI index were determined according to the following formula [5, 16]: TCBI = TG (mg/dL) × TC (mg/dL) × BW (kg)/1,000. CuTCBI = ((TCBI(2011) + TCBI(2015))/2) × (2015–2011). The calculation of the mean of the two measurements taken between 2011 and 2015 is required in order to reflect the mean exposure level. This mean should then be multiplied by the follow-up duration in order to reflect long-term cumulative exposure. This method aligns with the approach used to calculate the cumulative metabolic index in the CHARLS cohort. Subsequently, based on the distribution of cuTCBI, PAM clustering was employed using Gower distance to identify distinct trajectory patterns: the stable low-risk group (Cluster 1), the moderately increasing group (Cluster 2), and the high-risk group (Cluster 3) (Fig. 2).
Fig. 2.
Clustering analysis of TCBI and its association with CVD status. A PAM clustering of TCBI; (B)TCBI trends over time; C CVD incidence by cluster
Assessment of cardiovascular disease
The study mainly focused on the risk of CVD. CVD was defined as self-reported history of heart disease or stroke. The presence of heart disease was assessed through the question: “Has your doctor ever told you that you had a heart problem or have you received treatment for related issues?” Stroke occurrence was assessed through the question: “Has your doctor ever told you that you are suffering from a stroke or have you received treatment for problems caused by a stroke?” All data were self-reported.
Assessment of covariates
Gender, age, marital status, education level, smoking status, and drinking status were used as sociodemographic characteristics. The health indicators included height, body weight, BMI, waist circumference (WC), systolic blood pressure (SBP), diastolic blood pressure (DBP), hypertension, diabetes (DM), dyslipidemia, and depressive symptoms. Laboratory examinations measured C-reactive protein (CRP), fasting blood glucose (FBG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL) and glycated hemoglobin (HbA1c). The Center for Epidemiological Studies Depression (CESD-10) scale was utilized in order to assess depressive risk.
Definitions
Hypertension is diagnosed when SBP ≥ 140 mmHg, DBP ≥ 90 mmHg, currently using hypertension medication, or self-report of hypertension. Diabetes was defined as either fasting blood glucose ≥ 126 mg/dL, or HbA1c ≥ 6.5%, taking antidiabetes medications, or self-identified as having diabetes. Dyslipidemia: Total Cholesterol ≥ 240 mg/dL, Triglycerides ≥ 150 mg/dL, HDL-C ≤ 40 mg/dL, LDL-C ≥ 160 mg/dL, current use of lipid-lowering drugs or self-reported dyslipidemia diagnosed by a doctor. The variables “Lung,”“Liver” and “Kidney” were derived from the following question:“Have you been diagnosed with any of the conditions listed below by a doctor?”.
Statistical analysis
The data analysis was done with RStudio. For normally distributed continuous data, results are presented as means and standard deviations. ANOVA was used to see the differences between groups. For variables that were not normally distributed quantitatively, median and interquartile range were presented, and Kruskal-Wallis test was used for comparison between groups. Categorical variables were represented by counts and percentages, and were evaluated using the chi-squared test. Participants were classified into four categories based on cuTCBI levels: Quartile (Q) 1 ≤ 2451.14; 2451.14 < Q2 ≤ 3733.70; 3733.70 < Q3 ≤ 6022.59; Q4 ≥ 6022.59. Furthermore, cuTCBI was incorporated as a continuous variable in our research.
The incidence rates of CVD were estimated using Kaplan-Meier curves and log-rank tests. Schoenfeld residual tests confirmed that all Cox proportional hazards models met the proportional hazards assumption (all P > 0.05), supporting appropriate model specification and result reliability (Additional File 1). To this end, the same three hierarchical models were used for both the primary Cox regression and the sensitivity analysis using the Fine-Gray regression. Model 1 was unadjusted. Model 2 was adjusted for age, gender, marital status, education, smoking, and drinking. Model 3 was additionally adjusted for hypertension, diabetes mellitus, hypertension treatment, and diabetes treatment. The Fine-Gray model was employed to account for competing risks from non-cardiovascular death. The restricted cubic spline was applied to explore the relationship of cuTCBI with occurrence of CVD. Subgroup analysis was performed to determine whether the effect of cuTCBI on CVD occurrence varied by age, gender, marital status, smoking, drinking, hypertension, dyslipidemia and BMI. The net reclassification improvement (NRI) and the integrated discrimination improvement (IDI) were calculated to compare the predictive ability of different indicators for incident CVD. All statistical analyses were conducted using R version 4.5.0, with a significance level of P < 0.05.
Results
Baseline characteristics
This study included 4246 participants from CHARLS cohort. The average age of the participants was 58.03 ± 8.42 years, among whom 1981 (46.7%) were male and 2265 (53.3%) were female. Among these, 2557 (60.2%) had a BMI less than 24, 1247 (29.4%) had a BMI 24–28, and 442 (10.4%) had a BMI higher than 28 (Table 1). Furthermore, we generated a baseline table stratified by changes in TCBI (Additional file 2).
Table 1.
Baseline characteristics by cuTCBI Quartiles
| Variable | Overall (N = 4246) |
Q1 (N = 1062) |
Q2 (N = 1061) |
Q3 (N = 1061) |
Q4 (N = 1062) |
Test Statistic | Test Type | p-value |
|---|---|---|---|---|---|---|---|---|
| Age, Mean ± SD | 58.03 ± 8.42 | 59.33 ± 8.95 | 58.50 ± 8.46 | 57.74 ± 8.26 | 56.55 ± 7.74 | H = 56.876 | Kruskal-Wallis | < 0.001 |
| Age_group | χ² = 44.694 | Chi-square | < 0.001 | |||||
| 45–60 | 2,706 (63.7%) | 611 (57.5%) | 655 (61.7%) | 685 (64.6%) | 755 (71.1%) | |||
| > 60 | 1,540 (36.3%) | 451 (42.5%) | 406 (38.3%) | 376 (35.4%) | 307 (28.9%) | |||
| Gender | χ² = 38.355 | Chi-square | < 0.001 | |||||
| Female | 2,265 (53.3%) | 485 (45.7%) | 564 (53.2%) | 607 (57.2%) | 609 (57.3%) | |||
| Male | 1,981 (46.7%) | 577 (54.3%) | 497 (46.8%) | 454 (42.8%) | 453 (42.7%) | |||
| Marital | χ² = 9.4 | Chi-square | 0.024 | |||||
| No | 553 (13.0%) | 157 (14.8%) | 153 (14.4%) | 122 (11.5%) | 121 (11.4%) | |||
| Yes | 3,693 (87.0%) | 905 (85.2%) | 908 (85.6%) | 939 (88.5%) | 941 (88.6%) | |||
| Income | χ² = 12.865 | Chi-square | 0.045 | |||||
| High | 1,808 (42.6%) | 458 (43.1%) | 467 (44.0%) | 440 (41.5%) | 443 (41.7%) | |||
| Low | 116 (2.7%) | 21 (2.0%) | 30 (2.8%) | 22 (2.1%) | 43 (4.0%) | |||
| Median | 2,322 (54.7%) | 583 (54.9%) | 564 (53.2%) | 599 (56.5%) | 576 (54.2%) | |||
| Education | χ² = 14.715 | Chi-square | 0.023 | |||||
| High school or above | 381 (9.0%) | 73 (6.9%) | 94 (8.9%) | 108 (10.2%) | 106 (10.0%) | |||
| No formal education | 1,154 (27.2%) | 293 (27.6%) | 308 (29.0%) | 294 (27.7%) | 259 (24.4%) | |||
| Primary school | 2,711 (63.8%) | 696 (65.5%) | 659 (62.1%) | 659 (62.1%) | 697 (65.6%) | |||
| Location | χ² = 17.974 | Chi-square | 0.001 | |||||
| City | 220 (5.2%) | 29 (2.7%) | 59 (5.6%) | 67 (6.3%) | 65 (6.1%) | |||
| Rural | 4,026 (94.8%) | 1,033 (97.3%) | 1,002 (94.4%) | 994 (93.7%) | 997 (93.9%) | |||
| Smoking | χ² = 25.527 | Chi-square | < 0.001 | |||||
| Yes | 1,639 (38.6%) | 475 (44.7%) | 410 (38.6%) | 372 (35.1%) | 382 (36.0%) | |||
| No | 2,607 (61.4%) | 587 (55.3%) | 651 (61.4%) | 689 (64.9%) | 680 (64.0%) | |||
| Drinking | χ² = 6.393 | Chi-square | 0.094 | |||||
| No | 2,778 (65.4%) | 662 (62.3%) | 697 (65.7%) | 709 (66.8%) | 710 (66.9%) | |||
| Yes | 1,468 (34.6%) | 400 (37.7%) | 364 (34.3%) | 352 (33.2%) | 352 (33.1%) | |||
| Hypertension | χ² = 163.18 | Chi-square | < 0.001 | |||||
| No | 2,762 (65.0%) | 822 (77.4%) | 729 (68.7%) | 661 (62.3%) | 550 (51.8%) | |||
| Yes | 1,484 (35.0%) | 240 (22.6%) | 332 (31.3%) | 400 (37.7%) | 512 (48.2%) | |||
| DM | χ² = 29.855 | Chi-square | < 0.001 | |||||
| No | 4,169 (98.2%) | 1,053 (99.2%) | 1,052 (99.2%) | 1,040 (98.0%) | 1,024 (96.4%) | |||
| Yes | 77 (1.8%) | 9 (0.8%) | 9 (0.8%) | 21 (2.0%) | 38 (3.6%) | |||
| Dyslipidemia | χ² = 1384.003 | Chi-square | < 0.001 | |||||
| No | 2,245 (52.9%) | 902 (84.9%) | 771 (72.7%) | 451 (42.5%) | 121 (11.4%) | |||
| Yes | 2,001 (47.1%) | 160 (15.1%) | 290 (27.3%) | 610 (57.5%) | 941 (88.6%) | |||
| Lung | χ² = 10.679 | Chi-square | 0.014 | |||||
| No | 3,887 (91.5%) | 949 (89.4%) | 976 (92.0%) | 972 (91.6%) | 990 (93.2%) | |||
| Yes | 359 (8.5%) | 113 (10.6%) | 85 (8.0%) | 89 (8.4%) | 72 (6.8%) | |||
| Liver | χ² = 10.415 | Chi-square | 0.015 | |||||
| No | 4,110 (96.8%) | 1,029 (96.9%) | 1,012 (95.4%) | 1,031 (97.2%) | 1,038 (97.7%) | |||
| Yes | 136 (3.2%) | 33 (3.1%) | 49 (4.6%) | 30 (2.8%) | 24 (2.3%) | |||
| Kidney | χ² = 4.223 | Chi-square | 0.200 | |||||
| No | 4,019 (94.7%) | 996 (93.8%) | 1,000 (94.3%) | 1,007 (94.9%) | 1,016 (95.7%) | |||
| Yes | 227 (5.3%) | 66 (6.2%) | 61 (5.7%) | 54 (5.1%) | 46 (4.3%) | |||
| Hypertension treatment | χ² = 122.551 | Chi-square | < 0.001 | |||||
| No | 3,389 (79.8%) | 933 (87.9%) | 891 (84.0%) | 823 (77.6%) | 742 (69.9%) | |||
| Yes | 857 (20.2%) | 129 (12.1%) | 170 (16.0%) | 238 (22.4%) | 320 (30.1%) | |||
| DM treatment | χ² = 25.813 | Chi-square | < 0.001 | |||||
| No | 4,055 (95.5%) | 1,029 (96.9%) | 1,026 (96.7%) | 1,014 (95.6%) | 986 (92.8%) | |||
| Yes | 191 (4.5%) | 33 (3.1%) | 35 (3.3%) | 47 (4.4%) | 76 (7.2%) | |||
| Dyslipidemia treatment | χ² = 90.853 | Chi-square | < 0.001 | |||||
| No | 3,938 (92.7%) | 1,029 (96.9%) | 1,002 (94.4%) | 987 (93.0%) | 920 (86.6%) | |||
| Yes | 308 (7.3%) | 33 (3.1%) | 59 (5.6%) | 74 (7.0%) | 142 (13.4%) | |||
| Depression group | χ² = 26.637 | Chi-square | < 0.001 | |||||
| low | 2,764 (65.1%) | 641 (60.4%) | 679 (64.0%) | 702 (66.2%) | 742 (69.9%) | |||
| median | 1,236 (29.1%) | 358 (33.7%) | 326 (30.7%) | 289 (27.2%) | 263 (24.8%) | |||
| high | 246 (5.8%) | 63 (5.9%) | 56 (5.3%) | 70 (6.6%) | 57 (5.4%) | |||
| TC, mg/dl | 190.59 (167.01, 214.95) | 170.49 (151.16, 190.98) | 185.57 (165.08, 208.38) | 196.01 (175.13, 220.75) | 211.86 (186.34, 238.15) | H = 758.648 | Kruskal-Wallis | < 0.001 |
| TG, mg/dl | 104.43 (74.34, 153.99) | 65.49 (53.99, 79.65) | 91.15 (73.46, 110.63) | 123.01 (98.24, 150.45) | 199.13 (150.45, 271.70) | H = 2509.869 | Kruskal-Wallis | < 0.001 |
| HdL, mg/dl | 49.48 (40.59, 60.31) | 56.83 (48.33, 67.65) | 52.96 (44.85, 63.02) | 47.94 (39.82, 57.22) | 40.59 (33.63, 49.10) | H = 776.27 | Kruskal-Wallis | < 0.001 |
| LdL, mg/dl | 113.27 (92.78, 136.47) | 98.20 (83.12, 116.37) | 113.66 (96.26, 134.15) | 122.94 (102.06, 145.75) | 123.33 (95.10, 150.39) | H = 3387.539 | Kruskal-Wallis | < 0.001 |
| HbA1c | 5.10 (4.90, 5.40) | 5.10 (4.80, 5.30) | 5.10 (4.90, 5.40) | 5.10 (4.90, 5.40) | 5.20 (4.90, 5.60) | H = 88.52 | Kruskal-Wallis | < 0.001 |
| Glucose, mg/dl | 102.24 (94.32, 112.32) | 99.00 (92.16, 107.46) | 100.62 (93.60, 108.90) | 102.60 (94.68, 112.32) | 107.46 (99.00, 122.40) | H = 248.71 | Kruskal-Wallis | < 0.001 |
| WC, cm | 84.20 (77.80, 91.20) | 78.00 (73.00, 83.00) | 83.00 (77.20, 89.00) | 86.60 (80.20, 93.00) | 91.00 (85.00, 97.00) | H = 957.583 | Kruskal-Wallis | < 0.001 |
| TyG, mg/dl | 4.64 (4.45, 4.86) | 4.39 (4.28, 4.51) | 4.56 (4.45, 4.68) | 4.72 (4.60, 4.85) | 4.99 (4.83, 5.20) | H = 2339.66 | Kruskal-Wallis | < 0.001 |
| CRP, mg/dl | 0.97 (0.53, 1.97) | 0.76 (0.44, 1.67) | 0.82 (0.49, 1.77) | 1.01 (0.55, 2.00) | 1.30 (0.71, 2.49) | H = 141.187 | Kruskal-Wallis | < 0.001 |
| UA mg/dl | 4.22 (3.51, 5.05) | 4.04 (3.34, 4.80) | 4.06 (3.42, 4.87) | 4.29 (3.55, 5.04) | 4.57 (3.78, 5.50) | H = 138.794 | Kruskal-Wallis | < 0.001 |
| SBP, mmHg | 124.50 (112.50, 139.00) | 119.00 (108.00, 131.50) | 123.50 (112.00, 138.00) | 125.50 (113.50, 141.00) | 129.00 (117.00, 144.00) | H = 154.159 | Kruskal-Wallis | < 0.001 |
| DBP, mmHg | 74.00 (66.50, 82.00) | 70.00 (63.00, 77.50) | 73.50 (66.50, 81.00) | 74.50 (67.50, 82.50) | 78.00 (70.50, 86.50) | H = 225.08 | Kruskal-Wallis | < 0.001 |
| BMI, kg/m2 | 23.13 (20.94, 25.63) | 20.86 (19.28, 22.63) | 22.69 (20.81, 24.58) | 23.82 (21.79, 26.21) | 25.47 (23.27, 27.77) | H = 1044.944 | Kruskal-Wallis | < 0.001 |
| BMI group | χ² = 784.053 | Chi-square | < 0.001 | |||||
| < 24 | 2,557 (60.2%) | 931 (87.7%) | 720 (67.9%) | 557 (52.5%) | 349 (32.9%) | |||
| > 28 | 442 (10.4%) | 17 (1.6%) | 54 (5.1%) | 123 (11.6%) | 248 (23.4%) | |||
| 24–28 | 1,247 (29.4%) | 114 (10.7%) | 287 (27.0%) | 381 (35.9%) | 465 (43.8%) | |||
| Obesity | χ² = 312.902 | Chi-square | < 0.001 | |||||
| No | 3,804 (89.6%) | 1,045 (98.4%) | 1,007 (94.9%) | 938 (88.4%) | 814 (76.6%) | |||
| Yes | 442 (10.4%) | 17 (1.6%) | 54 (5.1%) | 123 (11.6%) | 248 (23.4%) | |||
| CVD onset | χ² = 46.803 | Chi-square | < 0.001 | |||||
| No | 3,123 (77.5%) | 851 (84.6%) | 776 (78.0%) | 757 (74.9%) | 739 (72.7%) | |||
| Yes | 906 (22.5%) | 155 (15.4%) | 219 (22.0%) | 254 (25.1%) | 278 (27.3%) | |||
Continuous variables are expressed as Mean ± SD or Median (IQR), categorical variables are expressed as number (percent)
Abbreviation: cuTCBI Cumulative triglyceride-total cholesterol-body weight index, SBP Systolic blood pressure, DBP Diastolic blood pressure, BMI Body mass index, CRP C-reactive protein, TG Triglycerides, FBG Fast blood glucose, TC Total cholesterol, HDL High density lipoprotein cholesterol, LDL Low density lipoprotein cholesterol, HbA1c Glycated hemoglobin
Compared with participants in the lowest quartile (Q1) of cuTCBI, those in higher quartiles were more likely to be younger, have a higher proportion of females, and indicated metabolic disorder characteristics, including higher BMI, elevated TG levels, greater insulin resistance, higher FBG, increased blood pressure, and more pronounced central obesity. Regarding disease burden, participants in the high cuTCBI group exhibited significantly higher prevalence rates of hypertension, diabetes, and dyslipidemia, accompanied by markedly increased usage of antihypertensive, antidiabetic, and lipid-lowering medications. Most critically, the incidence of cardiovascular disease was significantly higher in the high cuTCBI group than in the low cuTCBI group.
The relationship between cuTCBI/TCBI changes and CVD risk
Over 5 years of follow-up, 906 participants developed their first incident of CVD.The incidence of CVD gradually increased from Q1 to Q4, with 155 cases (17.1%) in the first quartile, 219 cases (24.2%) in the second quartile, 254 cases (28.0%) in the third quartile, and 278 cases (30.7%) in the fourth quartile. The Kaplan-Meier cumulative incidence curves revealed a gradual increase in the incidence of CVD events from Q1 to Q4, which was statistically significant (log-rank test P < 0.001, Fig. 3A). A Cox proportional hazards regression model was used to explore the association between cuTCBI and CVD incidence. After adjustments were made for potential confounding variables (Model 3), compared with Q1, the CVD risk increased with the increase in cuTCBI levels, with HRs of 1.46 (95%CI: 1.18,1.79), 1.64 (95%CI: 1.34,2.01), and 1.83 (95%CI: 1.49, 2.24) for Q2, Q3, and Q4, respectively (Table 2). When cuTCBI was analyzed as a continuous variable in Model 3, each one SD increase in cuTCBI was associated with a 12% increased risk of incident CVD (HR:1.12, 95%CI: 1.05,1.18, P < 0.001). Moreover, in model 3, compared to the stable low-risk group (Cluster 1), the moderately increasing group (Cluster 2) had 35% higher risk (HR:1.35, 95%CI: 1.17,1.56), the high-risk group (Cluster 3) had a 49% higher risk(HR:1.49, 95%CI: 1.22,1.83). RCS analysis showed a significant non-linear relationship between cuTCBI and the occurrence of CVD events in the study population (P for nonlinear<0.0001, P for overall<0.0001, Fig. 3B).
Fig. 3.
Association of cuTCBI with CVD risk in survival and dose-response analyses. A Kaplan-Meier curves of CVD risk by cuTCBI group; (B) Restricted cubic spline analysis of cuTCBI group and CVD risk
Table 2.
Multivariate COX regression of the relationship between cuTCBI /TCBI changes and the risk of new-onset CVD
| Characteristic | Event, n | Model 1 | Model 2 | Model 3 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95%CI | P | HR | 95%CI | P | HR | 95%CI | P | ||
| cuTCBI (per 1 SD) | 906 | 1.14 | 1.08, 1.20 | < 0.001 | 1.17 | 1.10,1.23 | < 0.001 | 1.12 | 1.05,1.18 | < 0.001 |
| cuTCBI quartile | ||||||||||
| Q1 | 155 | Ref | Ref | Ref | ||||||
| Q2 | 219 | 1.49 | 1.21,1.83 | < 0.001 | 1.52 | 1.23,1.86 | < 0.001 | 1.46 | 1.18,1.79 | < 0.001 |
| Q3 | 254 | 1.73 | 1.42, 2.11 | < 0.001 | 1.78 | 1.45,2.17 | < 0.001 | 1.64 | 1.34,2.01 | < 0.001 |
| Q4 | 278 | 1.94 | 1.59, 2.36 | < 0.001 | 2.09 | 1.71,2.52 | < 0.001 | 1.83 | 1.49,2.24 | < 0.001 |
| P for trend | < 0.001 | < 0.001 | < 0.001 | |||||||
| TCBI changes | ||||||||||
| Cluster 1 | 409 | Ref | Ref | Ref | ||||||
| Cluster 2 | 365 | 1.43 | 1.24,1.65 | < 0.001 | 1.46 | 1.26,1.68 | < 0.001 | 1.35 | 1.17,1.56 | < 0.001 |
| Cluster 3 | 132 | 1.59 | 1.31,1.94 | < 0.001 | 1.70 | 1.39,2.07 | < 0.001 | 1.49 | 1.22,1.83 | < 0.001 |
| P for trend | < 0.001 | < 0.001 | < 0.001 | |||||||
HR Hazard Ratio, CI Confidence
Model 1: unadjusted for any covariates
Model 2: adjusted for age, gender, marital status, education level, smoking, drinking
Model 3: adjusted for age, gender, marital status, education level, smoking, drinking, Hypertension, DM, Hypertension treatment, DM treatment
Subgroup analysis
Afterwards, we conducted subgroup analysis based on age, gender, marital status, smoking, drinking, hypertension, dyslipidemia, and BMI (Fig. 4). The results showed that cuTCBI levels were associated with an increased risk of CVD in all subgroups. When using TCBI changes in the same category for subgroup analysis, TCBI changes were associated with an increased risk of CVD in most groups and gender showed a significant interaction effect (Fig. 5). To compare the predictive ability of different indicators for incident CVD, we performed analyses of the net reclassification improvement (NRI) and the integrated discrimination improvement (IDI). Although cuTyG showed the largest IDI improvement (0.006, P = 0.0004) and cuTCBI demonstrated a slightly smaller IDI (0.004, P < 0.001), cuTCBI was the only metric that achieved statistical significance for both NRI and IDI (Additional file 3).
Fig. 4.
Forest plot of subgroup analysis for CVD risk by cuTCBI group
Fig. 5.
Forest plot of subgroup analysis for CVD risk by TCBI clusters
Gender-specific associations of cuTCBI and TCBI changes with CVD risk
According to the results of interaction analysis, cox analysis was performed to examine the relationship between cuTCBI / changes in TCBI and CVD risk (Table 3). In the fully adjusted model, when analyzed as categorical variable, men in the highest quartile of cuTCBI had a HR of 2.09(HR: 2.09, 95%CI: 1.55, 2.81) compared to those in Q1. Similarly, the CVD risk of women in Q4 had a 77% increased risk (HR: 1.77, 95%CI: 1.33, 2.35). The RCS analysis revealed that there was a significant nonlinear association between cuTCBI and CVD risk in both men (P for nonlinear: 0.0002, P for overall<0.0001, Fig. 6B) and women (P for nonlinear: 0.0025, P for overall<0.0001, Fig. 6D). In addition, the Kaplan-Meier cumulative incidence curves revealed a gradual increase in the incidence of CVD events from Q1 to Q4, which was statistically significant (log-rank test P < 0.0001, Fig. 6A, C). Regarding TCBI clustering patterns, compared to cluster 1, the men in cluster 3 had an 89% increased CVD risk(HR: 1.89, 95%CI: 1.38, 2.59). However, this association was not observed in women. These findings highlight the importance of sex-specific considerations in CVD prevention strategies.
Table 3.
Multivariate COX regression of the relationship between cuTCBI / the changes of TCBI and the risk of new-onset CVD according to gender
| Characteristic | Event, n | Model 1 | Model 2 | Model 3 | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95%CI | P | HR | 95%CI | P | HR | 95%CI | P | |||
| Male | cuTCBI (per 1 SD) | 383 | 1.13 | 1.05, 1.22 | 0.001 | 1.20 | 1.11,1.29 | < 0.001 | 1.15 | 1.06,1.25 | <0.001 |
| cuTCBI quartile | |||||||||||
| Q1 | 84 | Ref | Ref | Ref | |||||||
| Q2 | 84 | 1.20 | 0.89,1.62 | 0.243 | 1.25 | 0.92,1.69 | 0.156 | 1.18 | 0.87,1.60 | 0.294 | |
| Q3 | 98 | 1.53 | 1.15,2.05 | 0.004 | 1.71 | 1.27, 2.29 | < 0.001 | 1.61 | 1.19,2.16 | 0.002 | |
| Q4 | 117 | 1.90 | 1.43,2.51 | < 0.001 | 2.32 | 1.73, 3.09 | < 0.001 | 2.09 | 1.55,2.81 | < 0.001 | |
| P for trend | < 0.001 | < 0.001 | < 0.001 | ||||||||
| TCBI changes | |||||||||||
| Cluster 1 | 178 | Ref | Ref | Ref | |||||||
| Cluster 2 | 148 | 1.59 | 1.28,1.98 | < 0.001 | 1.75 | 1.40, 2.18 | < 0.001 | 1.69 | 1.35, 2.11 | < 0.001 | |
| Cluster 3 | 57 | 1.72 | 1.27, 2.32 | < 0.001 | 2.13 | 1.57, 2.90 | < 0.001 | 1.89 | 1.38, 2.59 | < 0.001 | |
| P for trend | < 0.001 | < 0.001 | < 0.001 | ||||||||
| Female | cuTCBI (per 1 SD) | 523 | 1.15 | 1.06, 1.24 | < 0.001 | 1.14 | 1.06, 1.23 | < 0.001 | 1.08 | 1.00, 1.18 | 0.060 |
| cuTCBI quartile | |||||||||||
| Q1 | 71 | Ref | Ref | Ref | |||||||
| Q2 | 135 | 1.77 | 1.32, 2.36 | < 0.001 | 1.83 | 1.37, 2.44 | < 0.001 | 1.76 | 1.32, 2.35 | < 0.001 | |
| Q3 | 156 | 1.89 | 1.43, 2.51 | < 0.001 | 1.91 | 1.44, 2.52 | < 0.001 | 1.75 | 1.32, 2.32 | < 0.001 | |
| Q4 | 161 | 1.99 | 1.50, 2.63 | < 0.001 | 2.05 | 1.55, 2.71 | < 0.001 | 1.77 | 1.33, 2.35 | < 0.001 | |
| P for trend | < 0.001 | < 0.001 | < 0.001 | ||||||||
| TCBI changes | |||||||||||
| Cluster 1 | 231 | Ref | Ref | Ref | |||||||
| Cluster 2 | 217 | 1.30 | 1.08, 1.56 | 0.006 | 1.29 | 1.07, 1.56 | 0.007 | 1.18 | 0.97, 1.42 | 0.091 | |
| Cluster 3 | 75 | 1.49 | 1.15, 1.93 | 0.003 | 1.47 | 1.13, 1.91 | 0.004 | 1.26 | 0.96, 1.65 | 0.096 | |
| P for trend | < 0.001 | < 0.001 | 0.048 | ||||||||
Model 1: unadjusted for any covariates
Model 2: adjusted for age, gender, marital status, education level, smoking, drinking
Model 3: adjusted for age, gender, marital status, education level, smoking, drinking, Hypertension, DM, Hypertension treatment, DM treatment
Fig. 6.
Sex-stratified analysis of the association between cuTCBI and CVD risk. A Kaplan-Meier analysis of CVD risk by cuTCBI group in males. B Restricted cubic spline analysis of CVD according to cuTCBI group in males. C Kaplan-Meier analysis of CVD risk by cuTCBI group in females. D Restricted cubic spline analysis of CVD according to cuTCBI group in females
Sensitivity analysis
To address the concern that pre-existing but undiagnosed CVD at baseline may bias the association, we performed a sensitivity analysis excluding participants with CVD prior to 2018, thereby including only incident CVD cases between 2018 and 2020 (Additional file 4). The results were directionally consistent with the primary analysis, with no substantive changes. In addition, we performed another sensitivity analysis using the Fine-Gray regression model. In the overall population, the results from Model 3 showed that, compared with the lowest quartile, the highest quartile was associated with a 78% increased risk of incident CVD (HR:1.78, 95%CI: 1.46,2.17, P < 0.001). Compared with Cluster 1, Cluster 3 was associated with a 45% increased risk of incident CVD (HR:1.45, 95%CI: 1.20,1.76, P < 0.001). These results were highly consistent with those from the Cox regression. The consistency of these results supports the robustness of the elevated risk associated with the top quartile, independent of specific cut-point selection. In gender subgroup analyses, the findings were also generally similar to those from the Cox regression (Additional file 5 and 6).
Discussion
There were 4246 participants in this study. They were divided into different groups based on cuTCBI levels. Cox proportional risk regression model was used to explore the relationship between cuTCBI and CVD risk. In our analysis, cuTCBI was treated as continuous variable and categorical variable. In the fully adjusted model, when cuTCBI was regarded as continuous variable, we found that each standard deviation increase in cuTCBI was associated with a 12% increase in CVD risk (HR:1.12, 95%CI:1.05–1.18, P < 0.001). When analyzing cuTCBI as a categorical variable, the highest quartile group had an HR of 1.83 (95%CI: 1.49,2.24, P < 0.001) compared to the lowest quartile group. Analysis of TCBI changes revealed that compared with cluster 1, cluster 2 had an HR of 1.35 (95%CI: 1.17,1.56, P < 0.001), and cluster 3 had an HR of 1.49 (95%CI: 1.22, 1.83, P < 0.001). Additionally, we identified a significant nonlinear relationship between cuTCBI and CVD risk in the general population (P for nonlinear < 0.0001; P for overall < 0.0001). K-M survival curve analysis revealed significant survival differences between different cuTCBI levels and CVD (log-rank test P < 0.0001). Subgroup analysis showed that in most subgroups, changes in TCBI and cuTCBI levels were associated with an increased risk of CVD. Interaction analysis revealed significant interactions between gender and TCBI changes on CVD risk (interaction P = 0.045). Collectively, these findings suggest that cuTCBI is a composite indicator with potential predictive value, representing a promising CVD risk biomarker warranting further investigation.
Based on these findings, our research has demonstrated that long-term cumulative exposure and dynamic model are more stable prediction factors, and that the relationship is basically nonlinear, which expands the scope of this field of research. The previous studies on the association between TCBI and CVD mostly relied on single time point measurements and could not reflect its dynamic cumulative effect [17–20]. To overcome the limitations of this approach, we adopted a method that combines cuTCBI with TCBI trajectories for analysis. By integrating different measurement results, we reduced the errors caused by short-term fluctuations in biomarkers and more accurately reflected a person’s long-term metabolic burden and potential “cumulative damage” effects. This provides a reasonable explanation for the more obvious and consistent relationship observed in this study. On the other hand, the TCBI trajectory analysis also identified different types of TCBI changes, such as “stable low risk”, “moderate increase”, and “persistently high levels”. This indicates that the dynamic process itself has the ability of predicting CVD risk, and therefore can achieve precise risk classification beyond baseline evaluation. The combined application of these methods makes the results more stable and interpretable.
A key finding of this study is that RCS analysis shows a significant and complex non-linear relationship between cuTCBI and CVD risk (P for nonlinear < 0.0001; P for overall < 0.0001). Above approximately 5000 units, cuTCBI showed a significantly elevated risk within an initial range. At higher values, although the HR remained above 1.0, the 95% CI included 1.0, indicating a non-significant positive trend. The wide confidence intervals at the upper tail of the cuTCBI distribution indicate statistical uncertainty in risk at extreme values. The quartile-based analysis (showing a significantly elevated risk in Q4) provides a complementary, population-relevant assessment of risk across the distribution. It is not a methodological flaw but rather reveals that the CVD risk is inherently non-uniform across populations. Therefore, in clinical practice, the focus should shift from a binary classification of “elevated” cuTCBI towards identifying individuals who maintain persistently high levels. This approach is crucial for advancing the precision prevention of cardiovascular disease. For example, in the fully adjusted model, quartile-based analysis of cuTCBI showed a significantly elevated risk in Q4 compared with Q1 (HR :1.83, 95%CI: 1.49, 2.24). When analyzed as a continuous variable, each one-SD increase in cuTCBI was associated with a 12% increase in CVD risk (HR:1.12, 95%CI: 1.02, 1.18). This is consistent with the association between TCBI and CVD risk reported in previous literature [10]. From a physiological mechanism perspective, insulin resistance may be a key factor. TyG and its derivatives have long been widely recognized as effective alternative markers for assessing insulin resistance [21, 22], and there is close correlation with atherosclerosis development and CVD [22–24]. The TCBI shares conceptual similarities with the TyG index. Although TCBI does not directly include glucose parameters, weight is one of the key markers of obesity related metabolic abnormalities, which is closely related to insulin resistance [25, 26]. Therefore, we speculate that the continuously rising TCBI could serve as an indirect biomarker for persistent insulin resistance status. The long-term accumulation of metabolic burden through cuTCBI expression may mainly exert its detrimental effect by exacerbating insulin resistance, in other words, the continuously increasing cuTCBI actually contributes to a chronic state of insulin resistance, which may subsequently contribute to chronic inflammation [27], oxidative stress [28, 29], and endothelial dysfunction [30, 31]. This also explains why “cumulative exposure” has higher predictive power than “single point measurement”. The nonlinear relationship we have studied provide us with a new perspective to the mode of action of TCBI. Here we propose a two-phase model to interpret this complex association. In the first phase, at low cuTCBI levels, the observed significant protective effect may reflect an optimal state of metabolic health, characterized by higher muscle mass [32], a healthier lipid profile [33], and maintained insulin sensitivity [34], collectively contributing to a lower risk of CVD. In the second phase, beyond a certain threshold, a high cuTCBI may overwhelm homeostatic capacity, leading to a decline in insulin sensitivity. In recent years, many studies have confirmed that the TyG index has significant non-linear associations with all-cause mortality, cardiovascular mortality and major adverse cardiovascular events in different populations, including patients with diabetes [35, 36], coronary heart disease [37, 38], and hypertension [39]. Thus, we speculate that the body will maintain homeostasis below a certain threshold through hyperinsulinemia and other means. Once the metabolic burden becomes too heavy, decompensation may occur, contributing to the risk of cardiovascular disease. In fact, the latest research suggests that hyperinsulinemia is not only a compensatory mechanism, but may also be one of the causes of insulin resistance and metabolic syndrome [40]. The mechanism of this process may be that compensatory hyperinsulinemia itself becomes a pathogenic factor, which can activate the SGK-1 signaling pathway to damage endothelial function and increase vascular stiffness, thereby increasing vascular risk [41].This also explains why this study used cuTCBI as a categorical variable to show that there is a huge risk for Q4 (HR:1.83, 95%CI: 1.49, 2.24). Using it as a continuous variable in complex models can dilute its effectiveness, and categorical analysis is easier to find a specific “high-risk interval”. The dynamic changes in TCBI also indicate that different long-term metabolic patterns can lead to different CVD risks. The population in cluster 2 and cluster 3 may have experienced longer and greater exposure to TCBI compared to those in cluster 1. The cumulative damage to vascular endothelium caused by TCBI exposure and metabolic burden may be the reason for the significant increase in risk observed in this cohort.
The research results we have obtained have significant clinical and public health implications for the prevention and treatment of cardiovascular disease. The TCBI index and the changes of TCBI could provide a practical method for identifying the people with high risk of CVD. Integrating these assessment methods into health monitoring protocols enables earlier detection of the people with underlying metabolic disorders—even when their single-timepoint test results fall within normal ranges. The threshold effect identified suggests that timely lifestyle interventions for TCBI patients on an upward trajectory or near the risk threshold may help delay the progression of CVD, and clinically, should include long-term monitoring of TCBI trends to more accurately evaluate individual prognosis and assess intervention effectiveness; The observed gender interaction effect reminds us to pay attention to future cardiovascular disease risk assessment and prevention strategies, and to fully consider the unique aspects of gender in order to promote the continuous improvement of personalized prevention and treatment systems. Of note, in the gender-stratified analysis, the female TCBI change clusters were not significant in Model 3, suggesting that the observed associations in this study may be gender-specific. Future studies are needed to further validate this finding and explore the underlying mechanisms.
The main advantage of this study is the large-scale prospective cohort design and reliable endpoint event data obtained through long-term follow-up. In terms of methodology, we attempted to combine cumulative exposure modeling with trajectory analysis to better reflect the long-term dynamic characteristics of TCBI. Moreover, the application of restricted cubic spline analysis has shown non-linear correlations between variables, which makes the research results appear more scientifically rigorous and reliable. However, this study also has certain shortcomings. Firstly, The main limitation of this study is that CVD outcomes were based on self-report, without validation by medical records or adjudicated endpoint events. Although the health-related questions in the CHARLS questionnaire have been widely used in similar studies, and we have verified the robustness of our results through sensitivity analyses, misclassification bias cannot be completely excluded. Future studies should incorporate medical insurance data or clinical records for validation. Secondly, although many known confounding factors were considered in the statistical model, the interference caused by residual confounding factors cannot be completely ruled out, such as dietary factors not being carefully measured, and genetic background also having an impact; In addition, this population belongs to a specific queue member range, and whether these conclusions are applicable to other races or geographical regions requires external verification work in the future. Another limitation of our clustering analysis must be acknowledged. Given that data were available for only two time points (2011 and 2015), genuine long-term trajectory modeling was not feasible. Finally, calculating cumulative TCBI required complete measurements at both 2011 and 2015, which may have introduced selection bias, as participants who completed follow-up were likely healthier. However, baseline characteristics were comparable between the included and excluded populations (all SMDs < 0.14), indicating that the bias was likely small and the results are robust (Additional file 7).
Conclusion
This study revealed a complex nonlinear association between cumulative exposure of cuTCBI and CVD risk. Due to limited data in the extremely high range of cuTCBI, the RCS analysis did not reach statistical significance in this interval. However, quartile-based analysis showed that individuals with the highest cuTCBI levels (Q4) had a significantly elevated risk of cardiovascular disease, and these individuals also belonged to the high-risk trajectory group identified by the TCBI dynamic trajectory analysis. Future studies with larger sample sizes are needed to precisely assess the risk at very high cuTCBI levels. In addition, sex-based intervention strategies should be developed to reduce the risk of CVD onset. Our findings highlight the importance of long-term monitoring of TCBI dynamic changes in clinical practice. Therefore, early identification of individuals on a high-risk trajectory and timely intervention may represent a novel strategy for the precise prevention of CVD.
Supplementary Information
Supplementary Material 8. Schoenfeld residual tests for proportional hazards assumption in Cox regression models
Supplementary Material 9. Baseline characteristics by TCBI cluster
Supplementary Material 10. NRI and IDI for biomarkers
Supplementary Material 11. Sensitivity analysis excluding participants with pre-existing CVD prior to 2018
Supplementary Material 12. Multivariate Fine-Gray regression of the relationship between cuTCBI /TCBI changes and the risk of new-onset CVD
Supplementary Material 13. Multivariate Fine-Gray regression of the relationship between cuTCBI / the changes of TCBI and the risk of new-onset CVD according to gender
Supplementary Material 14. Comparison of baseline characteristics between original and study populations
Acknowledgements
AcknowledgementsThe authors would like to express their appreciation to the CHARLS database for making the data available. This work would not have been possible without the dedication of all participants.
Authors’ contributions
JL analyzed the data and wrote the main manuscript text. QF analyzed the data. DZ, YF, SL, LZ, YL , PW and QA conducted the literature search and prepared figures. HS and NL designed the study and reviewed the manuscript, while also securing funding. All authors reviewed and approved the final manuscript.
Funding
The Jilin Provincial Department of Finance Project (2023SCZ51), the Jilin Provincial Department of Finance Project(2024SCZ19), the China-Japan Union Hospital of Jilin University Spring Bud Plan (2023CL01).
Data availability
This study analyzed publicly available data from the CHARLS database, which can be accessed at the website below: http://charls.pku.edu.cn/.
Declarations
Ethics approval and consent to participate
This analysis uses data from the CHARLS database, which obtained ethical approval from the Institutional Review Board of Peking University (IRB00001052-11015 for survey; IRB00001052-11014 for blood samples) and secured written informed consent from all participants.
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.
Junyu Liu and Qingfeng Fu contributed equally to this work.
Contributor Information
Hui Sun, Email: s_h@jlu.edu.cn.
Nan Liang, Email: liangnan2006@jlu.edu.cn.
References
- 1.Corbett S, Courtiol A, Lummaa V, Moorad J, Stearns S. The transition to modernity and chronic disease: mismatch and natural selection. Nat Rev Genet. 2018;19(7):419–30. [DOI] [PubMed] [Google Scholar]
- 2.Global regional. national age-sex specific mortality for 264 causes of death, 1980–2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet. 2017;390(10100):1151–210. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sacco RL, Roth GA, Reddy KS, Arnett DK, Bonita R, Gaziano TA, Heidenreich PA, Huffman MD, Mayosi BM, Mendis S, et al. The Heart of 25 by 25: Achieving the Goal of Reducing Global and Regional Premature Deaths From Cardiovascular Diseases and Stroke: A Modeling Study From the American Heart Association and World Heart Federation. Circulation. 2016;133(23):e674–690. [DOI] [PubMed] [Google Scholar]
- 4.Dunbar SB, Khavjou OA, Bakas T, Hunt G, Kirch RA, Leib AR, Morrison RS, Poehler DC, Roger VL, Whitsel LP. Projected Costs of Informal Caregiving for Cardiovascular Disease: 2015 to 2035: A Policy Statement From the American Heart Association. Circulation. 2018;137(19):e558–77. [DOI] [PubMed] [Google Scholar]
- 5.Doi S, Iwata H, Wada H, Funamizu T, Shitara J, Endo H, Naito R, Konishi H, Tsuboi S, Ogita M, et al. A novel and simply calculated nutritional index serves as a useful prognostic indicator in patients with coronary artery disease. Int J Cardiol. 2018;262:92–8. [DOI] [PubMed] [Google Scholar]
- 6.Ding H, Li C, Zhao X. The relationship between intrinsic capacity and sarcopenia in middle-aged and older Chinese populations: the mediating influence of a novel nutritional index. Front Public Health. 2025;13:1605158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ishiwata S, Yatsu S, Kasai T, Sato A, Matsumoto H, Shitara J, Shimizu M, Murata A, Kato T, Suda S, Doi S, Hiki M, Matsue Y, Naito R, Iwata H, Takagi A, Daida H. Prognostic Effect of a Novel Simply Calculated Nutritional Index in Acute Decompensated Heart Failure. Nutrients. 2020;12(11):3311. 10.3390/nu12113311. [DOI] [PMC free article] [PubMed]
- 8.Liu G, Zhang J. Association of a novel nutritional index with cognitive impairment in middle-aged and elderly Chinese adults: a cross-sectional analysis from the China Health and Retirement Longitudinal Study. Front Nutr. 2025;12:1486917. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Shi Y, Wang X, Yu C, Zhou W, Wang T, Zhu L, Bao H, Cheng X. Association of a novel nutritional index with stroke in Chinese population with hypertension: Insight from the China H-type hypertension registry study. Front Nutr. 2023;10:997180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Rezaee M, Kamrani F, Imannezhad M, Shahri HH, Saihood WK, Rezvani A, Far PM, Mahaki H, Esmaily H, Moohebati M, et al. Beyond traditional metrics: evaluating the triglyceride-total cholesterol-body weight index (TCBI) in cardiovascular risk assessment. BMC Cardiovasc Disord. 2025;25(1):39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ma X, Ma X, Wang Y, Qiu G, Zhang C. Associations of cumulative exposure and dynamic trajectories of the C-reactive protein-triglyceride-glucose index with incident cardiovascular disease in middle-aged and older Chinese adults: a nationwide cohort study. Cardiovasc Diabetol. 2025;24(1):303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Yang Y, Liu A. Associations of cumulative exposure and dynamic trajectories of the C-reactive protein-triglyceride-glucose index with incident stroke in middle-aged and older Chinese adults: a longitudinal analysis based on CHARLS. Cardiovasc Diabetol. 2025;24(1):386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Li F, Wang Y, Shi B, Sun S, Wang S, Pang S, Wu X. Association between the cumulative average triglyceride glucose-body mass index and cardiovascular disease incidence among the middle-aged and older population: a prospective nationwide cohort study in China. Cardiovasc Diabetol. 2024;23(1):16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Chen X, Wang C, Huang Q, Luo Y, Wu Z, Chen J, Gong H. Associations of cumulative exposure and dynamic trajectories of the triglyceride-total cholesterol-body weight index with new-onset cardiometabolic multimorbidity in middle-aged and older Chinese adults: evidence from a nationwide prospective cohort study. Cardiovasc Diabetol. 2026;25(1):146. 10.1186/s12933. [DOI] [PMC free article] [PubMed]
- 16.Minami-Takano A, Iwata H, Miyosawa K, Kubota K, Kimura A, Osawa S, Shitara M, Okazaki S, Suwa S, Miyauchi K, Sumiyoshi M, Amano A, Daida H. A Novel Nutritional Index Serves as A Useful Prognostic Indicator in Cardiac Critical Patients Requiring Mechanical Circulatory Support. Nutrients. 2019;11(6):1420. 10.3390/nu11061420. [DOI] [PMC free article] [PubMed]
- 17.Li G, Li S. Exploring the prognostic value of the novel nutritional index for in-hospital mortality in acute coronary syndrome: a sex-specific analysis. Front Med (Lausanne). 2025;12:1498260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Liang L, Zhao X, Huang L, Tian P, Huang B, Feng J, Zhou P, Wang J, Zhang J, Zhang Y. Prevalence and prognostic importance of malnutrition, as assessed by four different scoring systems, in elder patients with heart failure. Nutr Metab Cardiovasc Dis. 2023;33(5):978–86. [DOI] [PubMed] [Google Scholar]
- 19.Zhang G, Pan Y, Zhang R, Wang M, Meng X, Li Z, Li H, Wang Y, Zhao X, Wang Y, et al. A novel nutritional index and adverse outcomes in ischemic stroke: Results from the third China National Stroke Registry. Nutr Metab Cardiovasc Dis. 2022;32(6):1477–84. [DOI] [PubMed] [Google Scholar]
- 20.Kim HR, Kang MG, Kim K, Koh JS, Park JR, Hwang SJ, Jeong YH, Ahn JH, Park Y, Bae JS, et al. Comparative analysis of three nutrition scores in predicting mortality after acute myocardial infarction. Nutrition. 2021;90:111243. [DOI] [PubMed] [Google Scholar]
- 21.Simental-Mendía LE, Ortega-Pacheco CJ, García-Guerrero E, Sicsik-Aragón MA, Guerrero-Romero F, Martínez-Aguilar G. The triglycerides and glucose index is strongly associated with hepatic steatosis in children with overweight or obesity. Eur J Pediatr. 2021;180(6):1755–60. [DOI] [PubMed] [Google Scholar]
- 22.da Silva A, Caldas APS, Hermsdorff HHM, Bersch-Ferreira ÂC, Torreglosa CR, Weber B, Bressan J. Triglyceride-glucose index is associated with symptomatic coronary artery disease in patients in secondary care. Cardiovasc Diabetol. 2019;18(1):89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ren Q, Huang Y, Liu Q, Chu T, Li G, Wu Z. Association between triglyceride glucose-waist height ratio index and cardiovascular disease in middle-aged and older Chinese individuals: a nationwide cohort study. Cardiovasc Diabetol. 2024;23(1):247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Che B, Zhong C, Zhang R, Pu L, Zhao T, Zhang Y, Han L. Triglyceride-glucose index and triglyceride to high-density lipoprotein cholesterol ratio as potential cardiovascular disease risk factors: an analysis of UK biobank data. Cardiovasc Diabetol. 2023;22(1):34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Szukiewicz D. Molecular Mechanisms for the Vicious Cycle between Insulin Resistance and the Inflammatory Response in Obesity. Int J Mol Sci. 2023;24(12):9818. 10.3390/ijms24129818. [DOI] [PMC free article] [PubMed]
- 26.Mari A, Stefanski A, van Raalte DH, Ma X, LaBell ES, Fan L, Lee CJ, Thomas MK, Bunck MC, Ferrannini E. Tirzepatide Treatment and Associated Changes in β-Cell Function and Insulin Sensitivity in People With Obesity or Overweight With Prediabetes or Normoglycemia: A Post Hoc Analysis From the SURMOUNT-1 Trial. Diabetes Care. 2025;48(9):1622–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Libby P. Inflammation in Atherosclerosis-No Longer a Theory. Clin Chem. 2021;67(1):131–42. [DOI] [PubMed] [Google Scholar]
- 28.Myszko M, Bychowski J, Skrzydlewska E, Łuczaj W. The Dual Role of Oxidative Stress in Atherosclerosis and Coronary Artery Disease: Pathological Mechanisms and Diagnostic Potential. Antioxidants (Basel). 2025;14(3):275. 10.3390/antiox14030275. [DOI] [PMC free article] [PubMed]
- 29.Förstermann U, Xia N, Li H. Roles of Vascular Oxidative Stress and Nitric Oxide in the Pathogenesis of Atherosclerosis. Circ Res. 2017;120(4):713–35. [DOI] [PubMed] [Google Scholar]
- 30.Xu S, Ilyas I, Little PJ, Li H, Kamato D, Zheng X, Luo S, Li Z, Liu P, Han J, et al. Endothelial Dysfunction in Atherosclerotic Cardiovascular Diseases and Beyond: From Mechanism to Pharmacotherapies. Pharmacol Rev. 2021;73(3):924–67. [DOI] [PubMed] [Google Scholar]
- 31.Gomez Toledo A, Golden GJ, Cummings RD, Malmström J, Esko JD. Endothelial Glycocalyx Turnover in Vascular Health and Disease: Rethinking Endothelial Dysfunction. Annu Rev Biochem. 2025;94(1):561–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gao K, Cao LF, Ma WZ, Gao YJ, Luo MS, Zhu J, Li T, Zhou D. Association between sarcopenia and cardiovascular disease among middle-aged and older adults: Findings from the China health and retirement longitudinal study. EClinicalMedicine. 2022;44:101264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Cooney MT, Dudina A, De Bacquer D, Wilhelmsen L, Sans S, Menotti A, De Backer G, Jousilahti P, Keil U, Thomsen T, et al. HDL cholesterol protects against cardiovascular disease in both genders, at all ages and at all levels of risk. Atherosclerosis. 2009;206(2):611–6. [DOI] [PubMed] [Google Scholar]
- 34.Fonseca VA. Management of diabetes mellitus and insulin resistance in patients with cardiovascular disease. Am J Cardiol. 2003;92(4a):j50–60. [DOI] [PubMed] [Google Scholar]
- 35.Liu C, Liang D, Xiao K, Xie L. Association between the triglyceride-glucose index and all-cause and CVD mortality in the young population with diabetes. Cardiovasc Diabetol. 2024;23(1):171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Fang C, Peng N, Cheng J, Zhang X, Gu W, Zhu Z, Yin X, Yan Z, Zhang J, Yu P, et al. The association between TyG index and cardiovascular mortality is modified by antidiabetic or lipid-lowering agent: a prospective cohort study. Cardiovasc Diabetol. 2025;24(1):65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Huang LX, Sun T, Sun J, Wu ZM, Zhao YB, Li MY, Huo QY, Zhang BY, Ling C, Chen C, et al. Association of the Triglyceride-Glucose Index With Stroke and All-Cause Mortality in Patients With Coronary Artery Disease: A Nationwide Cohort Study From 1999 to 2018. MedComm. 2020;2025(610):e70422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Hao B, Lyu L, Xu J, Zhu X, Xu C, Gao W, Qin J, Huang T, Ding Y, Zhang Z, et al. The relationship between triglyceride-glucose index and prospective key clinical outcomes in patients hospitalised for coronary artery disease. Cardiovasc Diabetol. 2024;23(1):40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Zhou D, Liu XC, Kenneth L, Huang YQ, Feng YQ. A Non-Linear Association of Triglyceride Glycemic Index With Cardiovascular and All-Cause Mortality Among Patients With Hypertension. Front Cardiovasc Med. 2021;8:778038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Janssen J. Hyperinsulinemia and Its Pivotal Role in Aging, Obesity, Type 2 Diabetes, Cardiovascular Disease and Cancer. Int J Mol Sci. 2021;22(15):7797. 10.3390/ijms22157797. [DOI] [PMC free article] [PubMed]
- 41.Hill MA, Yang Y, Zhang L, Sun Z, Jia G, Parrish AR, Sowers JR. Insulin resistance, cardiovascular stiffening and cardiovascular disease. Metabolism. 2021;119:154766. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 8. Schoenfeld residual tests for proportional hazards assumption in Cox regression models
Supplementary Material 9. Baseline characteristics by TCBI cluster
Supplementary Material 10. NRI and IDI for biomarkers
Supplementary Material 11. Sensitivity analysis excluding participants with pre-existing CVD prior to 2018
Supplementary Material 12. Multivariate Fine-Gray regression of the relationship between cuTCBI /TCBI changes and the risk of new-onset CVD
Supplementary Material 13. Multivariate Fine-Gray regression of the relationship between cuTCBI / the changes of TCBI and the risk of new-onset CVD according to gender
Supplementary Material 14. Comparison of baseline characteristics between original and study populations
Data Availability Statement
This study analyzed publicly available data from the CHARLS database, which can be accessed at the website below: http://charls.pku.edu.cn/.