Abstract
Background
The C-reactive protein–triglyceride glucose index (CTI) has emerged as a promising composite biomarker for cardiovascular disease (CVD) risk. However, whether incorporating obesity-related metrics such as waist circumference (WC), body mass index (BMI), or waist-to-height ratio (WHtR) into CTI to form modified indices such as CTI-WC, CTI-BMI, and CTI-WHtR improves predictive performance remains uncertain. The performance of these modified indices requires validation in large-scale prospective cohorts stratified by glycemic status.
Methods
This study used data from the China Health and Retirement Longitudinal Study (CHARLS) from 2011 to 2020, involving 7,579 participants aged ≥ 45 years. Multivariate Cox regression and restricted cubic splines (RCSs) analyses were used to assess the associations of the CTI and its modified indices with CVD risk. To compare the predictive performance, time-dependent Harrell’s C-indices, integrated discrimination improvement and net reclassification index were utilized. Weighted quantile sum (WQS) regression was used to evaluate component contributions.
Results
During a mean follow-up of 8.28 years, 1,871 (24.69%) participants experienced their first CVD event. The CTI and its modified indices were effective in predicting CVD incidence in the general population. RCS analysis revealed positive linear dose–response relationships between these indices and CVD risk in the general population, which persisted in both normal glucose regulation (NGR) and prediabetes mellitus (Pre-DM) patients. WQS regression analysis revealed that, in the general population, TG contributed the most to CVD risk in the CTI, while WC, BMI, and WHtR had greater weights in their modified indices. In the general population, all modified CTI indices demonstrated superior predictive ability than did the original CTI (C-index: CTI-WC 0.619, CTI-WHtR 0.616, CTI-BMI 0.614, and CTI 0.612). In the population with NGR, Pre-DM, and DM, the predictive capability of CTI-WC is superior to that of the original CTI, with its C-index being greater across all these populations.
Conclusions
The CTI and its modified indices are effective in predicting CVD risk in the general population. The modified CTI indices, especially the CTI-WC, show superior predictive ability across different glycemic statuses. These findings suggest that incorporating obesity-related metrics into the CTI may enhance its utility for CVD risk prediction.
Graphical abstract
Supplementary Information
The online version contains supplementary material available at 10.1186/s13098-026-02169-1.
Keywords: C-reactive protein-triglyceride glucose index, Modified CTI indices, Cardiovascular diseases, Different glycemic statuses, Prospective cohort study
Introduction
Cardiovascular disease (CVD) remains a leading cause of morbidity and mortality worldwide, posing significant challenges to public health. [1, 2] Traditional risk factors, such as diabetes, dyslipidemia, hypertension, and obesity, have long been recognized as critical contributors to CVD, particularly among middle-aged and older adults. [2, 3] However, emerging evidence suggests that inflammation [4, 5] and insulin resistance [6, 7] also play pivotal roles in the pathogenesis of CVD. In recent years, the integration of inflammatory markers and metabolic parameters has garnered increasing attention as a potential approach to enhance the prediction of CVD risk [8, 9].
C-reactive protein (CRP) is a well-established biomarker of systemic inflammation, which has been shown to be associated with increased CVD risk in numerous epidemiological studies [10–12]. Meanwhile, the triglyceride-glucose (TyG) index, derived from fasting triglyceride(TG) and fasting blood glucose (FPG) levels, has emerged as a reliable surrogate marker of insulin resistance and has demonstrated significant associations with CVD outcomes. [13, 14] Building on these findings, the CRP-triglyceride glucose index (CTI), which combines CRP, TG, and FPG, has been proposed as a novel composite indicator that integrates both inflammatory and metabolic components to predict CVD risk. [15–17] Furthermore, obesity, particularly central obesity, not only stands as an independent risk factor for CVD but also constitutes a core pathological driver of inflammation and insulin resistance. [18] Visceral adipose tissue actively secretes proinflammatory cytokines that directly elevate CRP levels while exacerbating insulin resistance and lipid metabolism disturbances. [18] Therefore, integrating central adiposity metrics such as waist circumference (WC), body mass index (BMI), or waist‑to‑height ratio (WHtR) into the CTI to form a composite phenotype capturing the triad of “inflammation–insulin resistance–central obesity” could, in theory, enhance its predictive performance. Nevertheless, the comparative performance of the CTI and its modified indices, which incorporate these additional obesity‑related metrics, remains to be fully elucidated in large‑scale prospective cohort studies, especially across different glycemic statuses.
The China Health and Retirement Longitudinal Study (CHARLS) is a comprehensive longitudinal survey that captures detailed information on health, socioeconomic status, and lifestyle factors among individuals aged 45 years and above. [19] CHARLS provides a unique opportunity to investigate the predictive capacity of the CTI and its modified indices in a nationally representative sample of middle-aged and older adults. By leveraging the rich data from the CHARLS, this study aims to comprehensively evaluate the associations between the CTI and its modified indices with incident CVD events and to compare their predictive performance across different glycemic statuses. This research may offer valuable insights into the utility of these composite indices for risk stratification and early intervention in CVD prevention.
Methods
Study design
The CHARLS is a large-scale, nationally representative longitudinal survey conducted in China. It spans 450 villages, 150 counties, and 28 provinces. The survey was launched by the National School of Development at Peking University, with the initial baseline survey (Wave 1) conducted from 2011 to 2012. To date, four follow-up surveys have been completed, in 2013 (Wave 2), 2015 (Wave 3), 2018 (Wave 4), and 2020 (Wave 5). The CHARLS received approval from the Peking University Institutional Review Board (IRB00001052—11,015). Written informed consent was obtained from all participants prior to their involvement. The study adhered to the STROBE guidelines for reporting observational epidemiological research. [20] Further information about the CHARLS has been reported in earlier studies. [19] The data can be obtained through the CHARLS website (https://charls.pku.edu.cn/) after completing the application procedures and necessary registration.
Study population
Our cohort study employed data from the 2011 baseline and subsequent 2013, 2015, 2018, and 2020 waves of the CHARLS. The initial sample comprised 17,708 participants from the 2011 baseline survey. The specific procedure for selecting the population is illustrated in Fig. 1. The criteria for excluding participants in our study were as follows: 1) age < 45 years at baseline;2) incomplete demographic information at baseline; 3) with CVD or without CVD information at baseline;4) without CTI and its modified indices data at baseline; 5) extreme abnormal values calculated for CTI and its modified indices at baseline; 6) lost participants during follow-up. As a result, 10,129 participants were excluded from the study, leaving a final cohort of 7,579 participants. In order to mitigate the bias that arises from missing variables, we utilized multiple imputation methods to manage the missing data.
Fig. 1.
Flowchart of participant selection
Data collection and definition
CTI and its modified indices
The levels of CRP, FPG, and TG were assessed following an overnight fast using an enzymatic colorimetric method on a Hitachi 7180 chemistry analyser (Hitachi, Tokyo, Japan). During the physical examination, height, weight, and waist circumference (WC) were each measured three times, with the mean values used as the final results. Height was determined using a vertical stadiometer, weight was measured with a calibrated scale, and WC was obtained using a non-stretchable tape measure placed horizontally around the waist at the level of the navel while the participants were standing. Body mass index (BMI) was calculated as “weight (kg)/height (m)2” [21], and waist-to-height ratio (WHtR) was defined as “Waist (cm)/Height (cm)” [22].
The CTI was calculated using the following formula, with modified CTI indices derived by simple multiplication with obesity-related metrics following the established methodology of Wang et al [23]. and Zeng et al. [24]:
CTI = 0.412 × Ln (CRP [mg/L]) + Ln (TG [mg/dl] × FPG [mg/dl])/2; [25].
CTI-WC = CTI × WC;
CTI-BMI = CTI × BMI;
CTI-WHtR = CTI × WHtR.
This approach leverages the pathophysiological synergy among inflammation, insulin resistance, and adiposity in CVD development, where central obesity elevates CRP and worsens insulin resistance [18]. By integrating CTI with obesity metrics, these modified indices capture this pathological triad, potentially enhancing risk stratification beyond the original CTI.
Different glycemic statuses
Glycemic status can be categorized into three groups: normal glucose regulation (NGR), prediabetes mellitus (Pre-DM), and diabetes mellitus (DM). NGR was diagnosed by an hemoglobin A1c (HBA1c) < 5.7% and an FPG < 100 mg/dL. Pre-DM was defined as an HbA1c of 5.7–6.4% or an FPG of 100–125 mg/dL. DM was defined by at least one of the following criteria: (1) HbA1c ≥ 6.5%, (2) FPG ≥ 126 mg/dL, (3) current use of antidiabetic medications, or (4) history of diabetes diagnosed by a doctor as reported by the patient.
Assessment of covariates
The initial data for the participants were carefully gathered by trained interviewers through the use of structured questionnaires.
1) Socio-demographic characteristics and health-related behaviors: age, sex, education, drinking, and smoking status. Education level was grouped into college/above, high school, and primary school/below. Drinking frequency was divided into more than once a month, less than once a month, and never. Smoking status was categorized as current smoker, ex-smoker, and non-smoker.
2) Body measurements: systolic blood pressure (SBP) and diastolic blood pressure (DBP). SBP and DBP were measured using an Omron HEM-7200 sphygmomanometer, with the values calculated as the means of two or three consecutive readings.
3) Data on disease and medication history: disease history included hypertension, and diabetes, as well as details on medications used for dyslipidemia, and diabetes. Hypertension was identified by an SBP ≥ 140 mmHg, a DBP ≥ 90 mmHg, the use of antihypertensive medications, or self-reported prior hypertension diagnosis.
4) Laboratory examination data: FPG, CRP, total cholesterol (TC), TG, low density lipoprotein cholesterol (LDL-c), high density lipoprotein cholesterol (HDL-c), HBA1c, blood urea nitrogen (BUN), serum creatinine (SCR), and uric acid (UA). The laboratory data were assessed following an overnight fast using an enzymatic colorimetric method on a Hitachi 7180 chemistry analyser (Hitachi, Tokyo, Japan).
Assessment of CVD incidence
CVD events were evaluated through the following inquiries: “Have you been told by a doctor that you had a stroke? or “Have you been told by a doctor that you had a heart attack, angina, coronary heart disease, heart failure, or other heart problems?” Participants with a reported history of stroke or heart disease were identified as having CVD events.
Statistical analysis
Normally distributed continuous variables were expressed as the means ± standard deviations (SDs), whereas categorical variables were depicted as numbers and percentages. For continuous variables, group comparisons were performed using the independent samples Student’s t-test if the data were normally distributed, or the Mann–Whitney U-test if not. Categorical data were analysed using the chi-square test when applicable.
Multivariate Cox regression analysis was used to examine the associations between CTI and its modified indices with CVD risk. Each of the CTI and its modified indices underwent Z-score standardization, allowing for a direct comparison of their hazard ratios (HRs) and 95% confidence intervals (CIs) for each standard deviation (SD). The CTI and its modified indices were incorporated into the models both as continuous variables and categorical variables defined by quartiles. A total of three Cox regression models were utilized in this study. The principles for selecting covariates included the following: (1) prior research findings and clinical constraints; (2) a p-value < 0.1 in the univariate analysis for the variables. Despite the p-value of UA exceeding 0.1(p = 0.212) in univariate analysis, it was included in the model due to its well-established association with CVD shown in prior studies. [26] (3) In generalized variance inflation factor (GVIF) analysis to detect multicollinearity, variables with GVIF^(1/(2*Df) greater than 2 were excluded. In this case, TC (2.44) and LDL-c (2.23) were excluded; Model 1 was unadjusted for any variables. Model 2 was adjusted for age, sex, education, smoking, drinking, DBP, antidyslipidemic medications, and UA status. Model 3 included further adjustments for SBP, DM, hypertension, HDL-c, HbA1c, and antidiabetic medications. To control for type I errors resulting from multiple comparisons, the Benjamini–Hochberg method for false discovery rate (FDR) correction was applied in all the models. The cumulative incidence of CVD was estimated using the Kaplan–Meier method. Cumulative hazard curves were constructed with this method, and the log-rank test was used to compare differences. Moreover, a fully adjusted Model 3 with restricted cubic spline (RCS) analysis was employed to elucidate the dose–response associations between the CTI and its modified indices with CVD risk. In order to evaluate the predictive ability of the CTI and its modified indices for CVD incidence, we calculated the time-dependent Harrell’s C-indices. Moreover, the integrated discrimination improvement (IDI) and net reclassification improvement (NRI) were also employed to assess the additional predictive value. To explore the associations between CVD incidence and these indices among different glycemic statuses, subgroup analysis, interaction analyses, RCS regression analyses, and time-dependent Harrell’s C-indices were also employed across different glycemic statuses. The Benjamini–Hochberg false discovery rate (FDR) correction method was employed in all subgroup analyses to adjust for type I errors resulting from multiple testing.
The CTI and its modified indices were derived using a calculation that integrates CRP, TG, FPG and an obesity metric (such as WC, BMI, or WHtR). To elucidate the respective roles of CRP, TG, FPG, and the chosen obesity parameter in predicting CVD risk, we applied a weighted quantile sum (WQS) regression framework, with bootstrap resampling conducted over 500 iterations. This method allocated weights to each factor and was bounded within the range of 0 to 1, [27] to signify their predictive influence on CVD risk. Variables with higher weights were deemed more influential in the composite effect.
All the statistical analyses were performed using R statistical software (version 4.3.3). A two-tailed P value < 0.05 was regarded as statistically significant.
Results
Baseline characteristics of the study participants
The baseline characteristics of all participants are presented in Table 1. In this study, a total of 7,579 participants were involved, of whom 3,199 (42.21%) were aged 60 years or older and 4,380 (57.79%) were under 60 years, with 47.30% being male and 52.70% being female. Among them, 2,937 (38.75%) patients had NGR, 3,630 (47.90%) patients had Pre-DM, and 1,012 (13.35%) patients had DM. The baseline mean values of the CTI, CTI-WC, CTI-BMI, and CTI-WHtR were 8.70 ± 0.79, 739.56 ± 124.95, 203.51 ± 41.95, and 4.69 ± 0.81, respectively. Significant differences were observed in the CTI and its modified indices between the CVD group and the non-CVD group (all P < 0.001). Moreover, the CTI and its modified indices were significantly higher in the CVD group than in the non-CVD group. The results also showed that participants with CVD were more likely to be older, female, have a primary school or below, be non-smokers, never drink, and have hypertension. They were also more likely to be using antidiabetic and antidyslipidemic medications. Furthermore, they exhibited higher values for SBP, DBP, BMI, WHtR, WC, TC, LDL, TG, CRP, FPG, and HbA1c.
Table 1.
Baseline characteristics of the study participants with and without cardiovascular disease
| level | Overall | Non-CVD | CVD | P value | |
|---|---|---|---|---|---|
| 7579 | 5708 | 1871 | |||
| Age,year (%) | < 60 | 4380 (57.79) | 3452 (60.48) | 928 (49.60) | < 0.001 |
| ≥ 60 | 3199 (42.21) | 2256 (39.52) | 943 (50.40) | ||
| Sex (%) | Female | 3994 (52.70) | 2917 (51.10) | 1077 (57.56) | < 0.001 |
| Male | 3585 (47.30) | 2791 (48.90) | 794 (42.44) | ||
| Education (%) | College or above | 210 (2.77) | 147 (2.58) | 63 (3.37) | 0.008 |
| High school | 2022 (26.68) | 1568 (27.47) | 454 (24.27) | ||
| Primary school or below | 5347 (70.55) | 3993 (69.95) | 1354 (72.37) | ||
| Smoking (%) | Current smoker | 2358 (31.11) | 1841 (32.25) | 517 (27.63) | < 0.001 |
| Ex-smoker | 613 (8.09) | 433 (7.59) | 180 (9.62) | ||
| Non-smoker | 4608 (60.80) | 3434 (60.16) | 1174 (62.75) | ||
| Drinking (%) | Drink but less than once a month | 629 (8.30) | 491 (8.60) | 138 (7.38) | 0.002 |
| Drink more than once a month | 1668 (22.01) | 1299 (22.76) | 369 (19.72) | ||
| never | 5282 (69.69) | 3918 (68.64) | 1364 (72.90) | ||
| Hypertension (%) | No | 4621 (60.97) | 3686 (64.58) | 935 (49.97) | < 0.001 |
| Yes | 2958 (39.03) | 2022 (35.42) | 936 (50.03) | ||
| Antidyslipidemic medication (%) | No | 7315 (96.52) | 5562 (97.44) | 1753 (93.69) | < 0.001 |
| Yes | 264 (3.48) | 146 (2.56) | 118 (6.31) | ||
| Antidiabetic medication (%) | No | 7358 (97.08) | 5568 (97.55) | 1790 (95.67) | < 0.001 |
| Yes | 221 (2.92) | 140 (2.45) | 81 (4.33) | ||
| SBP,mmHg (mean (SD)) | 128.81 (21.07) | 127.44 (20.55) | 133.02 (22.07) | < 0.001 | |
| DBP,mmHg (mean (SD)) | 75.07 (12.10) | 74.50 (11.86) | 76.81 (12.63) | < 0.001 | |
| BMI,kg/m2 (mean (SD)) | 23.29 (3.66) | 23.08 (3.55) | 23.94 (3.93) | < 0.001 | |
| WC,cm (mean (SD)) | 84.72 (9.72) | 84.07 (9.46) | 86.69 (10.23) | < 0.001 | |
| WHtR (mean (SD)) | 0.54 (0.06) | 0.53 (0.06) | 0.55 (0.07) | < 0.001 | |
| TC,mg/dL (mean (SD)) | 193.15 (37.57) | 192.13 (37.53) | 196.27 (37.51) | < 0.001 | |
| LDL-c ,mg/dL(mean (SD)) | 116.75 (34.38) | 115.78 (34.21) | 119.71 (34.74) | < 0.001 | |
| HDL-c ,mg/dL(mean (SD)) | 51.84 (15.22) | 52.13 (15.29) | 50.96 (14.95) | 0.004 | |
| TG,mg/dL (mean (SD)) | 126.17 (85.06) | 124.34 (85.53) | 131.77 (83.38) | 0.001 | |
| FPG,mg/dL (mean (SD)) | 108.42 (31.72) | 107.58 (29.83) | 110.99 (36.79) | < 0.001 | |
| HbA1c,% (mean (SD)) | 5.24 (0.75) | 5.22 (0.73) | 5.31 (0.84) | < 0.001 | |
| BUN,mg/dL (mean (SD)) | 15.75 (4.47) | 15.79 (4.48) | 15.62 (4.44) | 0.148 | |
| SCR,mg/dL (mean (SD)) | 0.78 (0.22) | 0.78 (0.24) | 0.78 (0.19) | 0.548 | |
| UA,mg/dL (mean (SD)) | 4.44 (1.24) | 4.45 (1.22) | 4.41 (1.27) | 0.212 | |
| CRP,mg/dL (mean (SD)) | 2.53 (6.90) | 2.48 (6.88) | 2.67 (6.96) | 0.313 | |
| Follow-up time,year (mean (SD)) | 8.28 (1.68) | 9.00 (0) | 6.10 (2.26) | < 0.001 | |
| Glycemic status | |||||
| NGR No | 4642 (61.25) | 3461 (60.63) | 1181 (63.12) | 0.059 | |
| Yes | 2937 (38.75) | 2247 (39.37) | 690 (36.88) | ||
| Pre-DM No | 3949 (52.10) | 2964 (51.93) | 985 (52.65) | 0.608 | |
| Yes | 3630 (47.90) | 2744 (48.07) | 886 (47.35) | ||
| DM No | 6567 (86.65) | 4991 (87.44) | 1576 (84.23) | < 0.001 | |
| Yes | 1012 (13.35) | 717 (12.56) | 295 (15.77) | ||
| CTI and its modified indices | |||||
| CTI (mean (SD)) | 8.70 (0.79) | 8.66 (0.79) | 8.81 (0.80) | < 0.001 | |
| CTI-WC (mean (SD)) | 739.56 (124.95) | 730.86 (122.13) | 766.09 (129.67) | < 0.001 | |
| CTI-BMI (mean (SD)) | 203.51 (41.95) | 200.82 (40.87) | 211.74 (44.11) | < 0.001 | |
| CTI-WHtR (mean (SD)) | 4.69 (0.81) | 4.64 (0.79) | 4.87 (0.83) | < 0.001 | |
| CTI Quantiles (%) | Q1 | 2026 (26.73) | 1631 (28.57) | 395 (21.11) | < 0.001 |
| Q2 | 1962 (25.89) | 1479 (25.91) | 483 (25.82) | ||
| Q3 | 1912 (25.23) | 1409 (24.68) | 503 (26.88) | ||
| Q4 | 1679 (22.15) | 1189 (20.83) | 490 (26.19) | ||
| CTI-WC Quantiles (%) | Q1 | 1911 (25.21) | 1561 (27.35) | 350 (18.71) | < 0.001 |
| Q2 | 1989 (26.24) | 1531 (26.82) | 458 (24.48) | ||
| Q3 | 1950 (25.73) | 1453 (25.46) | 497 (26.56) | ||
| Q4 | 1729 (22.81) | 1163 (20.37) | 566 (30.25) | ||
| CTI-BMI Quantiles (%) | Q1 | 1996 (26.34) | 1599 (28.01) | 397 (21.22) | < 0.001 |
| Q2 | 1972 (26.02) | 1528 (26.77) | 444 (23.73) | ||
| Q3 | 1954 (25.78) | 1471 (25.77) | 483 (25.82) | ||
| Q4 | 1657 (21.86) | 1110 (19.45) | 547 (29.24) | ||
| CTI-WHtR Quantiles (%) | Q1 | 1902 (25.10) | 1563 (27.38) | 339 (18.12) | < 0.001 |
| Q2 | 2013 (26.56) | 1540 (26.98) | 473 (25.28) | ||
| Q3 | 1940 (25.60) | 1442 (25.26) | 498 (26.62) | ||
| Q4 | 1724 (22.75) | 1163 (20.37) | 561 (29.98) | ||
CVD, cardiovascular disease; SD, standard deviation; WC, waist circumference; BMI, body-mass index; WHtR, waist-to-height ratio; SBP,systolic blood pressure; DBP, diastolic blood pressure; TC, total cholesterol; LDL-c, low density lipoprotein cholesterol; HDL-c, high density lipoprotein cholesterol; TG, triglyceride; FPG, fasting blood glucose; HbA1c, hemoglobin A1c; BUN, blood urea nitrogen; SCR, serum creatinine; UA, uric acid; CRP, C-reactive protein; CTI, C-reactive protein–triglyceride glucose index; NGR, normal glucose regulation; Pre-DM, prediabetes mellitus; DM, diabetes mellitus
Associations between the CTI and its modified indices with CVD incidence in the general population
During an average follow-up period of 8.28 years, 1,871 (24.69%) participants in the general population experienced their first CVD event. The incidence of CVD increased progressively from Q1 to Q4 according to the CTI, with 395 (21.11%) in Q1, 483 (25.82%) in Q2, 503 (26.88%) in Q3, and 490 (26.19%) in Q4. The incidence of CVD increased progressively from Q1 to Q4 according to the CTI-WC, with 350 (18.71%) in Q1, 458 (24.48%) in Q2, 497 (26.56%) in Q3, and 566 (30.25%) in Q4.The incidence of CVD increased progressively from Q1 to Q4 according to the CTI-BMI, with 397 (21.22%) in Q1, 444 (23.73%) in Q2, 483 (25.82%) in Q3, and 547 (29.24%) in Q4.The incidence of CVD increased progressively from Q1 to Q4 according to the CTI-WHtR, with 339 (18.12%) in Q1, 473 (25.28%) in Q2, 498 (26.62%) in Q3, and 561 (29.98%) in Q4 (Table 1). Analysis of the Kaplan–Meier cumulative incidence curve showed that CVD events for the CTI and its modified indices increased progressively from the Q1 to Q4 groups, with a statistically significant difference (all log—rank tests P < 0.001, Fig. 2).The RCS analysis showed linear and positive dose–response relationships between CTI and its modified indices with CVD risk (CTI, P for overall = 0.017, P for nonlinear = 0.320; CTI-WC, P for overall < 0.001, P for nonlinear = 0.813; CTI-BMI, P for overall < 0.001, P for nonlinear = 0.540;CTI-WHtR, P for overall < 0.001, P for nonlinear = 0.543, Fig. 3).
Fig. 2.
K-M plot of CVD incidence according to CTI (A), CTI-WC (B), CTI-BMI (C), and CTI-WHtR (D). CTI, C-reactive protein- triglyceride glucose index, BMI, body mass index; WC, waist circumference; and WHtR, waist-to-height ratio
Fig. 3.
Dose–response relationships between CTI (A), CTI-WC (B), CTI-BMI (C), or CTI-WHtR (D) and cardiovascular disease risk after adjustments for Model 3. CTI, C-reactive protein- triglyceride glucose index, BMI, body mass index; WC, waist circumference; and WHtR, waist-to-height ratio
The associations between the CTI and its modified indices with CVD risk are depicted in Table 2. When the CTI and its modified indices were treated as continuous dependent variables, the risk of CVD events increased with each increase in the SD. After adjusting for potential confounders in Model 3, a 21.0% increase in the risk of CVD incidence was observed for every SD in CTI-WC, which was the most significant increase compared with that in CTI (HR: 1.09, 95%CI: 1.03,1.15), CTI-BMI (HR: 1.18, 95%CI: 1.12,1.24), and CTI-WHtR (HR: 1.17, 95%CI: 1.10,1.24). Similarly, when stratified by quartiles, participants in the highest quartile of CTI, CTI-WC, CTI-BMI, and CTI-WHtR exhibited a significantly greater CVD risk compared to those in the lowest quartile, with HRs (95% CIs) of 1.30(1.11,1.52), 1.67(1.42,1.96), 1.59(1.36,1.86), and 1.56(1.32,1.84), respectively. Compared with their reference (Q1), the greatest increase in CVD risk of 67% was observed in the highest quartile of the CTI-WC. Even after applying the FDR correction, the above results still held statistical significance (all p < 0.05).
Table 2.
Multivariate Cox regression analysis of the associations between the C-reactive protein- triglyceride glucose index and its modified indices with cardiovascular disease risk
| Model 1 HR(95%CI) |
P value | P valueFDR | Model 2 HR(95%CI) |
P value | P value FDR | Model 3 HR(95%CI) |
P value | P value FDR | |
|---|---|---|---|---|---|---|---|---|---|
| CTI | |||||||||
| Continuous (per 1 SD) | 1.17(1.12,1.22) | ＜0.001 | ＜0.001 | 1.11(1.06,1.17) | ＜0.001 | ＜0.001 | 1.09(1.03,1.15) | 0.004 | 0.013 |
| Quartile 1 | Ref | Ref | Ref | ||||||
| Quartile 2 | 1.30(1.13,1.48) | 0.001 | ＜0.001 | 1.22(1.07,1.40) | 0.003 | 0.005 | 1.21(1.06,1.38) | 0.006 | 0.015 |
| Quartile 3 | 1.41(1.23,1.60) | ＜0.001 | ＜0.001 | 1.27(1.11,1.45) | ＜0.001 | 0.001 | 1.22(1.06,1.41) | 0.005 | 0.015 |
| Quartile 4 | 1.59(1.39,1.82) | ＜0.001 | ＜0.001 | 1.39(1.21,1.60) | ＜0.001 | ＜0.001 | 1.30(1.11,1.52) | 0.001 | 0.005 |
| P for trend | ＜0.001 | ＜0.001 | ＜0.001 | ||||||
| CTI-WC | |||||||||
| Continuous (per 1 SD) | 1.27(1.22,1.33) | ＜0.001 | ＜0.001 | 1.21(1.16,1.27) | ＜0.001 | ＜0.001 | 1.21(1.14,1.28) | ＜0.001 | ＜0.001 |
| Quartile 1 | Ref | Ref | Ref | ||||||
| Quartile 2 | 1.29(1.12,1.48) | ＜0.001 | ＜0.001 | 1.26(1.09,1.44) | 0.001 | 0.003 | 1.26(1.10,1.45) | 0.001 | 0.004 |
| Quartile 3 | 1.45(1.26,1.66) | ＜0.001 | ＜0.001 | 1.34(1.17,1.54) | ＜0.001 | ＜0.001 | 1.34(1.16,1.55) | ＜0.001 | ＜0.001 |
| Quartile 4 | 1.95(1.71,2.23) | ＜0.001 | ＜0.001 | 1.71(1.48,1.96) | ＜0.001 | ＜0.001 | 1.67(1.42,1.96) | ＜0.001 | ＜0.001 |
| P for trend | ＜0.001 | ＜0.001 | ＜0.001 | ||||||
| CTI-BMI | |||||||||
| Continuous (per 1 SD) | 1.23(1.18,1.28) | ＜0.001 | ＜0.001 | 1.19(1.14,1.24) | ＜0.001 | ＜0.001 | 1.18(1.12,1.24) | ＜0.001 | ＜0.001 |
| Quartile 1 | Ref | Ref | Ref | ||||||
| Quartile 2 | 1.15(1.00,1.31) | 0.048 | 0.048 | 1.14(0.99,1.31) | 0.061 | 0.077 | 1.13(0.98,1.30) | 0.083 | 0.150 |
| Quartile 3 | 1.27(1.12,1.46) | ＜0.001 | ＜0.001 | 1.22(1.06,1.39) | 0.005 | 0,010 | 1.20(1.04,1.38) | 0.013 | 0.034 |
| Quartile 4 | 1.80(1.58,2.05) | ＜0.001 | ＜0.001 | 1.64(1.43,1.89) | ＜0.001 | ＜0.001 | 1.59(1.36,1.86) | 0.001 | ＜0.001 |
| P for trend | ＜0.001 | ＜0.001 | ＜0.001 | ||||||
| CTI-WHtR | |||||||||
| Continuous (per 1 SD) | 1.28(1.22,1.34) | ＜0.001 | ＜0.001 | 1.19(1.13,1.25) | ＜0.001 | ＜0.001 | 1.17(1.10,1.24) | ＜0.001 | ＜0.001 |
| Quartile 1 | Ref | Ref | Ref | ||||||
| Quartile 2 | 1.37(1.19,1.57) | ＜0.001 | ＜0.001 | 1.30(1.13,1.50) | ＜0.001 | ＜0.001 | 1.29(1.12,1.49) | ＜0.001 | 0.001 |
| Quartile 3 | 1.51(1.31,1.73) | ＜0.001 | ＜0.001 | 1.351.17,1.55) | ＜0.001 | ＜0.001 | 1.32(1.14,1.53) | ＜0.001 | 0.001 |
| Quartile 4 | 2.01(1.75,2.30) | ＜0.001 | ＜0.001 | 1.64(1.42,1.90) | ＜0.001 | ＜0.001 | 1.56(1.32,1.84) | ＜0.001 | ＜0.001 |
| P for trend | ＜0.001 | ＜0.001 | ＜0.001 |
Model 1 was unadjusted
Model 2 was adjusted for age, sex, education, smoking, drinking, DBP, antidyslipidemic medication, UA
Model 3 included further adjustments for hypertension, DM, SBP, HDL-c, HbA1c and medication use (antidiabetic and antidyslipidemic medications)
HR, Hazard Ratio; CI, Confidence Interval; WC, waist circumference; BMI, body-mass index; WHtR, waist-to-height ratio; SBP,systolic blood pressure; DBP, diastolic blood pressure
HDL-c, high density lipoprotein cholesterol; HbA1c, hemoglobin A1c; UA, uric acid; CRP, C-reactive protein; CTI, C-reactive protein–triglyceride glucose index; WC, waist circumference
BMI, body-mass index; WHtR,waist-to-height ratio; NGR, normal glucose regulation; Pre-DM, prediabetes mellitus; DM, diabetes mellitus
*P value adjusted using the False Discovery Rate (FDR) correction
The predictive performance was evaluated using C-index analysis, the NRI, and IDI analysis, based on the adjusted Cox regression Model 3. The overall C-index value was 0.619 for CTI-WC, followed by 0.616 for CTI-WHtR, 0.614 for CTI-BMI, and 0.612 for CTI. The time-dependent Harrell’s C-indices of the CTI and its modified indices are shown in Fig. 4A. The IDI and NRI indices comparing the models are depicted in Fig. 4B-C. The IDI of 0(95% CI:−0.003,0.002) (p = 0.806) and NRI of −0.002(95% CI:−0.040,0.032) (p = 0.866)were not significant when the CTI-WC was compared with the CTI-BMI, which was similar to the comparison between the CTI-BMI and CTI-WHtR, with an IDI of −0.002(95% CI:−0.004,0.001) (p = 0.100) and NRI of −0.024(95% CI:−0.073,0.005) (p = 0.119). Compared with CTI alone, CTI-WC, CTI-BMI, and TyG-WHtR all produced significant IDI indices [TyG-WC: 0.005(95% CI:0.002,0.009), p < 0.001; CTI-BMI: 0.005(95% CI:0.002,0.008), p < 0.001; CTI-WHtR: 0.003(95% CI:0,0.006), p = 0.010] and RNI indices [CTI-WC: 0.082(95% CI:0.045,0.106), p < 0.001; CTI-BMI:0.058(95% CI:0.026,0.082), p < 0.001;CTI-WHtR: 0.049(95% CI:0.014,0.082), P = 0.010]. Furthermore, the comparison of CTI-WC with CTI-WHtR revealed significant results, with an IDI value of −0.002(95% CI:−0.003,−0.001) (p < 0.001) and an RNI value of −0.078(95% CI:−0.102,−0.049) (p < 0.001).
Fig. 4.
Time-dependent predictive capacity of the CTI and its modified indices for CVD incidence (A); IDI (B) and NRI(C) indices of the CTI and its modified indices. CTI, C-reactive protein- triglyceride glucose index; IDI, integrated discrimination improvement;NRI,net reclassification improvement
WQS analyses (Fig. 5) were performed using the more robust comprehensive indicator. The results showed that among the variables assessed, the TG component of the CTI had the highest relative contribution weight to CVD risk at 0.62 (P < 0.05), highlighting its significant role in modifying CVD risk over time. In the modified CTI indices, however, WC in the CTI-WC, BMI in the CTI-BMI, and WHtR in the CTI-WHtR had higher weights of 0.46, 0.41, and 0.44 (all P < 0.05), respectively.
Fig. 5.
Weighted Quantile Sum regression analysis was used to assess the effects of the weights of each component of CTI (A), CTI-WC (B), CTI-BMI (C), and CTI-WHtR (D) on CVD risk after adjustment for Model 3. The estimated weight quantifies each component's proportional contribution to the composite risk, identifying where higher values denote stronger relative importance. A bar that crosses the red dashed line denotes a significant association (P<0.05), identifying that the component as a pivotal driver of CVD risk. CVD, cardiovascular disease; CTI, triglyceride-glucose index; BMI, body mass index; WC, waist circumference; and WHtR, waist-to-height ratio
Associations between the CTI and its modified indices with CVD incidence across different glycemic statuses
To further investigate the associations between the CTI and its modified indices with CVD risk, subgroup and interaction analyses were performed after fully adjusting for potential confounders in Model 2, and the results are shown in Tables 3-4. There were no interactions between the CTI and its modified indices with different glycemic statuses. However, interactions were observed for the CTI, CTI-WC, and CTI-WHtR across age groups, and for the CTI-BMI across sex groups. Each index was a predictive factor for CVD incidence in the population aged younger and older than 60 years. These associations held significance in both the female and male populations. The CTI-WC, CTI-BMI, and CTI-WHtR still had predictive value of CVD incidence among populations with NGR [HR: CTI-WC, 1.25(95% CI:1.14,1.36); CTI-BMI, 1.18(95% CI:1.09,1.28), CTI-WHtR, 1.20(95% CI:1.10,1.31)], Pre-DM [HR: CTI-WC, 1.17(95% CI:1.09,1.26); CTI-BMI, 1.20(95% CI:1.12,1.28), CTI-WHtR,1.16(95% CI:1.07,1.25)]and DM[HR: CTI-WC, 1.27(95% CI:1.13,1.42); CTI-BMI,1.19(95% CI:1.08,1.32), CTI-WHtR,1.24(95% CI:1.10,1.40)]. The CTI was a predictive factor among the population with NGR [HR: 1.12(95% CI:1.03,1.23)] and Pre-DM [HR: 1.10(95% CI:1.02,1.18)]. The significance of the results (all p < 0.05) was maintained even after the FDR adjustment (Table 3).
Table 3.
Subgroup analyses of the relationships between the C-reactive protein- triglyceride glucose index and its modified indices with CVD incidence in the general population
| Characteristics | Count of participants |
CTI | CTI-WC | CTI-BMI | CTI-WHtR | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HR(95%CI) | P value | Adjusted P value | HR(95%CI) | P value | Adjusted P value | HR(95%CI) | P value | Adjusted P value | HR(95%CI) | P value | Adjusted P value | ||
| Sex | |||||||||||||
| Female | 3994 | 1.15(1.08,1.22) | < 0.001 | < 0.001 | 1.22(1.15,1.30) | < 0.001 | < 0.001 | 1.19(1.12,1.26) | < 0.001 | < 0.001 | 1.19(1.12,1.27) | < 0.001 | < 0.001 |
| Male | 3585 | 1.08(1.01,1.16) | 0.034 | 0.039 | 1.20(1.12,1.29) | < 0.001 | < 0.001 | 1.21(1.12,1.30) | < 0.001 | < 0.001 | 1.19(1.10,1.29) | < 0.001 | < 0.001 |
| Age(years) | |||||||||||||
| < 60 | 4380 | 1.14(1.07,1.21) | < 0.001 | < 0.001 | 1.22(1.14,1.31) | < 0.001 | < 0.001 | 1.17(1.10,1.25) | < 0.001 | < 0.001 | 1.21(1.13,1.29) | < 0.001 | < 0.001 |
| ≥ 60 | 3199 | 1.08(1.01,1.16) | 0.022 | 0.029 | 1.19(1.12,1.28) | < 0.001 | < 0.001 | 1.19(1.12,1.27) | < 0.001 | < 0.001 | 1.16(1.09,1.25) | < 0.001 | < 0.001 |
| Glycemic statuses | |||||||||||||
| NGR | 2937 | 1.12(1.03,1.23) | 0.007 | 0.014 | 1.25(1.14,1.36) | < 0.001 | < 0.001 | 1.18(1.09,1.28) | < 0.001 | < 0.001 | 1.20(1.10,1.31) | < 0.001 | < 0.001 |
| Pre-DM | 3630 | 1.10(1.02,1.18) | 0.012 | 0.019 | 1.17(1.09,1.26) | < 0.001 | < 0.001 | 1.20(1.12,1.28) | < 0.001 | < 0.001 | 1.16(1.07,1.25) | < 0.001 | < 0.001 |
| DM | 1012 | 1.10(0.99,1.23) | 0.082 | 0.082 | 1.27(1.13,1.42) | < 0.001 | < 0.001 | 1.19(1.08,1.32) | 0.001 | 0.001 | 1.24(1.10,1.40) | < 0.001 | < 0.001 |
CTI, C-reactive protein–triglyceride glucose index; WC, waist circumference; BMI, body-mass index; WHtR, waist-to-height ratio; NGR, normal glucose regulation; Pre-DM, prediabetes mellitus; DM, diabetes mellitus
Table 4.
Interaction analyses of CTI and modified indices stratified by age, gender and glycemic statuses
| Variables* | Multiplicative interaction | |
|---|---|---|
| Coefficient (95% CI) | P-value | |
| Interaction terms: | ||
| Age (years) | ||
| CTI: ≥ 60 | −0.095(0.909,0.046) | 0.037 |
| CTI-WC: ≥ 60 | −0.100(0.905,0.045) | 0.025 |
| CTI-BMI: ≥ 60 | −0.063(0.939,0.041) | 0.126 |
| CTI-WHtR: ≥ 60 | −0.100(0.904,0.045) | 0.026 |
| Gender | ||
| CTI: Male | −0.001(0.999,0.046) | 0.987 |
| CTI-WC: Male | 0.058(1.059,0.045) | 0.201 |
| CTI-BMI: Male | 0.094(1.098,0.043) | 0.027 |
| CTI-WHtR: Male | 0.082(1.085,0.048) | 0.083 |
| Glycemic statuses | ||
| CTI: Pre-DM | 0.013(1.013,0.055) | 0.815 |
| CTI-WC: Pre-DM | −0.015(0.985,0.052) | 0.769 |
| CTI-BMI: Pre-DM | 0.046(1.047,0.049) | 0.341 |
| CTI-WHtR: Pre-DM | 0.006(1.006,0.053) | 0.907 |
| CTI: DM | 0.009(1.009,0.068) | 0.901 |
| CTI-WC: DM | 0.051(1.053,0.068) | 0.451 |
| CTI-BMI: DM | 0.053(1.054,0.058) | 0.364 |
| CTI-WHtR: DM | 0.081(1.084,0.069) | 0.240 |
* All the exposure factors (CTI, CTI-BMI, CTI-WC and CTI-WHtR) were analyzed as continuous variable (per 1 SD). And the details of stratified variables are: (1) age: divided into < 60 years old and ≥ 60 years old; (2) gender: divided into male and female; (3) Glycemic statuses: divided into NGR, Pre-DM and DM
** The details of interaction terms: < 60 years old, female and NGR were set as the reference groups. NGR, normal glucose regulation;Pre-DM, prediabetes mellitus; and DM, diabetes mellitus
The RCS analysis of the associations between the CTI and its modified indices with CVD incidence across different glycemic statuses is also shown in Fig. 6. In the population with NGR, the relationships between the CTI and its modified indices with CVD incidence were significant (CTI, P for overall = 0.047, P for nonlinear = 0.569; CTI-WC, P for overall < 0.001 P for nonlinear = 0.871; CTI-BMI, P for overall = 0.001 P for nonlinear = 0.283; CTI-WHtR, P for overall < 0.001 P for nonlinear = 0.554). In the population with Pre-DM, the associations between the CTI and its modified indices with CVD incidence were also significant (CTI, P for overall = 0.028, P for nonlinear = 0.239; CTI-WC, P for overall < 0.001, P for nonlinear = 0.792; CTI-BMI, P for overall < 0.001, P for nonlinear = 0.873; CTI-WHtR, P for overall = 0.002, P for nonlinear = 0.782). In the population with DM, the association between CTI with CVD incidence were insignificant (P for overall = 0.380, P for nonlinear = 0.986), but the associations between modified CTI indices and CVD incidence was significant (CTI-WC, P for overall = 0.001,P for nonlinear = 0.822; CTI-BMI, P for overall = 0.001, P for nonlinear = 0.108; CTI-WHtR, P for overall = 0.003, P for nonlinear = 0.369).
Fig. 6.
Dose–response associations between CTI and modified indices and CVD risk after adjustments for Model 2 across different glycemic statuses (A-D:NGR;E-H:Pre-DM;I-L:DM). CVD, cardiovascular disease; CTI, C-reactive protein- triglyceride glucose index; BMI, body mass index; WC, waist circumference; WHtR, waist-to-height ratioand; NGR, normal glucose regulation; Pre-DM, prediabetes mellitus; and DM, diabetes mellitus
The time-dependent Harrell’s C-indices of the CTI and its modified indices across different glycemic statuses are shown in Supplementary Material Fig. 1. The NRI and IDI indices of the models are depicted in Supplementary Material Figs. 2, 3, 4. In the population with NGR, the overall C-index value was 0.615 for CTI-WC, followed by 0.608 for CTI-WHtR, 0.605 for CTI-BMI, and 0.606 for CTI. Compared with CTI alone, CTI-WC produced significant IDI indices [0.008(95% CI:0.002,0.014), p < 0.001] and RNI indices[0.094(95% CI:0.025,0.153), p < 0.001], but CTI-BMI and CTI-WHtR produced insignificant IDI indices [CTI-BMI, 0.004(0,0.009),P = 0.159;CTI-WHtR,0.005(95%CI:0,0.011), P = 0.050] and RNI indices [CTI-BMI,0.049(95%CI:−0.015,0.099),P = 0.209;CTI-WHtR,0.056(95% CI:−0.011,0.110), P = 0.100]. Furthermore, the comparison of CTI-WC with CTI-WHtR yielded significant results, with an IDI value of −0.003(95% CI:−0.006,−0.001) (p < 0.001) and an RNI value of −0.094(95% CI:−0.149, −0.054) (p < 0.001). In the population with Pre-DM, the overall C-index values were 0.609 for CTI-WC, followed by 0.608 for CTI-WHtR, 0.608 for CTI-BMI, and 0.603 for CTI. Compared with CTI alone, CTI-BMI produced significant IDI indices [0.005(95% CI:0.001,0.011), p < 0.001] and RNI indices [0.059(95% CI:0.011,0.104), p = 0.030], but CTI-WC and CTI-WHtR produced insignificant IDI indices [CTI-WC,0.003(95% CI:0,0.007), P = 0.050;CTI-WHtR,0.002(95% CI:−0.001,0.006), P = 0.070] and RNI indices [CTI-WC,0.063(95% CI:−0.001,0.096),P = 0.070;CTI-WHtR,0.057(95% CI:−0.004,0.090), P = 0.090]. Furthermore, the comparison of CTI-WC with CTI-WHtR yielded significant results, with an IDI value of −0.001(95% CI:−0.003,0) (p = 0.030) and an RNI value of −0.069(95% CI:−0.106, −0.001) (p = 0.030). In the population with DM, the overall C-index values were 0.649 for CTI-WC and CTI-BMI, followed by 0.645 for CTI-WHtR, and 0.629 for CTI. Compared with CTI alone, CTI-WC and CTI-WHtR produced significant IDI indices (CTI-WC,0.014(95% CI:0.002,0.028), p < 0.001; CTI-WHtR,0.009(95% CI:0.001,0.022), P = 0.020] and RNI indices (CTI-WC,0.121(95% CI:0.038,0.187), P = 0.010; CTI-WHtR,0.077(95% CI:0.015,0.166), P = 0.020], but CTI-BMI produced insignificant IDI indices [0.011(95% CI:−0.001,0024), p = 0.070] and RNI indices [0.109(95% CI:−0.006,0.178), P = 0.080], Furthermore, the comparison of CTI-WC with CTI-WHtR yielded significant results, with an IDI value of −0.004(95% CI:−0.009,0) (p = 0.040) and an RNI value of −0.120(95% CI:−0.194,−0.013) (p = 0.040).
Discussion
Main findings
Our study provides novel insights into the associations between the CTI and its modified indices with CVD risk across different glycemic statuses in a large, nationally representative cohort of middle-aged and older adults from China. The results demonstrated that the CTI and its modified indices are effective in predicting CVD risk. While the improvements in discrimination metrics observed with the modified CTI indices were statistically significant, we acknowledge that the magnitude of these improvements was modest (C-index increases of 0.002–0.007; IDI range: 0.003–0.005; NRI range: 0.049–0.082). Nevertheless, even small improvements in risk prediction at the population level can translate into meaningful clinical benefits when applied to large cohorts. For instance, an NRI of 0.082 for CTI-WC versus CTI indicates that approximately 8 in 100 individuals would be correctly reclassified to a more appropriate risk category, potentially altering clinical management decisions. In the context of CVD prevention, where small shifts in risk stratification can guide decisions regarding statin initiation, lifestyle intervention intensity, or monitoring frequency, such refinements may have substantial public health implications when scaled across millions of at-risk individuals. The modified CTI indices, particularly CTI-WC, could be readily incorporated into existing risk assessment workflows as a cost-effective enhancement, requiring only routinely available clinical measurements, such as CRP, TG, FPG, and WC. Future studies should evaluate whether risk-based management guided by these modified indices leads to improved clinical outcomes compared to standard approaches. In the CTI, TG emerged as the dominant contributor to CVD risk in the general population, while WC, BMI, and WHtR also had significant weights in the modified CTI indices, underscoring their importance in enhancing the predictive capacity of the CTI. These results highlight the importance of integrating inflammatory markers, metabolic parameters, and obesity-related metrics to enhance the prediction of CVD risk. This approach could facilitate early identification of high-risk individuals and guide targeted preventive interventions, such as lifestyle modifications and pharmacotherapy, to reduce the burden of CVD.
Comparison with previous studies
Extensive studies [12, 28, 29] have consistently identified CRP as a systemic inflammatory marker linked to greater CVD risk. Meanwhile, a growing corpus of evidence [30, 31] has affirmed that elevated TyG index values are reliably associated with increased CVD risk. Aligning with our findings, prior investigations [15–17] have demonstrated that both the CTI and the combined CRP-TyG are significantly independently related to CVD risk. Additional obesity-related metrics, such as WC, BMI, and WHtR, have been robustly validated as powerful predictors of CVD risk. [32, 33] However, the integration of the CTI into its subsequent modifications with obesity-related metrics has not been extensively explored in large-scale prospective cohort studies. Our results first demonstrated that both the CTI and its modified indices are significantly associated with increased CVD risk and exhibit better predictive performance than the original CTI. Similarly, other studies [13, 23] have also shown that insulin resistance indices, such as the TyG index and atherogenic index of plasma, after adding obesity-related metrics, including WC, BMI and WHtR, have greater predictive value than the original indices do. Our study also, for the first time, revealed that the modified CTI indices, especially the CTI-WC, demonstrated superior predictive ability across different glycemic statuses. WQS regression analysis further revealed that WC, BMI, and WHtR had substantial weights in their respective modified CTI indices, indicating their significant contributions to CVD risk. This highlighted the importance of considering obesity metrics in risk prediction. The WC, in particular, emerged as a strong predictor across different glycemic statuses, suggesting that central adiposity may play a more critical role in CVD pathogenesis compared to overall adiposity measures.
Potential mechanisms
The mechanistic underpinnings of the superior predictive power observed with CTI-WC, CTI-BMI and CTI-WHtR rely on convergent biology whereby systemic inflammation, insulin resistance and adiposity act synergistically to initiate, propagate and perpetuate CVD. CRP is more than a passive biomarker. By activating the classical complement pathway, inducing endothelial nitric-oxide synthase uncoupling and up-regulating vascular adhesion molecules, CRP directly fosters endothelial dysfunction and plaque instability. [34, 35] This inflammatory milieu is amplified when insulin resistance—captured indirectly by the TyG component—co-exists, as hyperglycaemia and ectopic lipid accumulation generate advanced glycation end-products and reactive oxygen species that further impair vasodilatory capacity and promote smooth-muscle proliferation. [36, 37] Visceral adiposity, quantified by WC, BMI or WHtR, supplies the third pathophysiological pillar. Excess visceral fat releases a torrent of pro-inflammatory adipokines while suppressing adiponectin, creating a self-reinforcing cycle of lipotoxicity and systemic inflammation; concomitantly, portal free-fatty-acid flux aggravates hepatic insulin resistance and drives atherogenic dyslipidemia. [18] The integration of these obesity variables into the CTI therefore does not merely add statistical weight but captures the anatomical source of the inflammatory and metabolic insult.
Notably, the predictive superiority of modified CTI indices manifested differently across glycemic strata, suggesting context-dependent pathophysiological predominance. In the population with established DM, the overwhelming burden of chronic hyperglycemia and advanced vascular damage may obscure the distinct contribution of adiposity-driven inflammation, thereby attenuating the incremental discriminative capacity of obesity metrics. In contrast, among individuals with NGR or Pre-DM, where metabolic perturbations are less severe, visceral adiposity and its inflammatory sequelae may constitute a more distinct and independent determinant of CVD risk, allowing obesity-enhanced indices to exhibit greater predictive enhancement. This glycemic status-dependent divergence implies that the relative prognostic weight of central adiposity diminishes as hyperglycemia-mediated injury becomes the dominant pathogenic driver, a hypothesis that merits validation in dedicated mechanistic investigations. A metagenomic study further suggested that visceral adiposity alters the gut-microbiota composition, increasing lipopolysaccharide translocation that triggers Toll-like receptor 4-mediated inflammatory cascades, [38] thereby linking central obesity with both CRP elevation and insulin resistance. Epigenetic modifications, such as DNA hypermethylation induced by hyperglycemia and adiposity, may lock this pathological triad into a chronic state [39, 40], explaining why the CTI-WC, CTI-BMI and CTI-WHtR outperform the original CTI in long-term risk prediction. Collectively, these intertwined pathways illustrate that the modified CTI indices do not simply aggregate risk factors; they reflect the anatomical and molecular axes through which inflammation, insulin resistance and central obesity converge to accelerate atherosclerosis and cardiovascular events.
Strengths and limitations
This study has several notable strengths that enhance the validity and reliability of its findings. The use of data from the CHARLS provides a large, nationally representative sample of middle-aged and older adults, ensuring that the results are broadly applicable to this demographic group. The longitudinal design, with a mean follow-up of 8.28 years, allows for the robust assessment of the predictive power of the CTI and its modified indices over an extended period. Additionally, comprehensive data collection, including detailed measurements of both traditional and novel risk factors, enables a thorough evaluation of the indices. The application of advanced statistical methods, such as multivariate Cox regression, RCS, and WQS regression, further strengthens the analysis by accounting for potential confounders and non-linear relationships. Finally, the study’s focus on different glycemic statuses provides valuable insights into the predictive utility of these indices across various metabolic profiles.
Despite these strengths, the study also has several limitations that should be considered when the results are interpreted. First, the study population is limited to individuals aged 45 years and above, which may restrict the generalizability of the findings to younger populations. Second, the diagnosis of CVD and certain variables, such as smoking and drinking status, was based on self-reported data, which may introduce potential biases. However, previous validation studies in CHARLS showed that self-reported medical history, such as hypertension and diabetes, was reasonably accurate. [41] Notably, self-reported physician diagnosis of cardiovascular outcomes could lead to outcome misclassification, potentially underestimating or overestimating the true associations observed in this study.Additionally, the use of multiple imputation methods to handle missing data, while necessary, may introduce some degree of uncertainty into the results. Third, although we adjusted for a wide range of potential confounders, residual confounders, such as hypogonadism [42] and liver fibrosis [43], may still exist. Finally, the generalizability of our findings to other populations, particularly those from different ethnic or geographical backgrounds, needs to be confirmed in future studies.
Future directions
Future research should focus on validating the predictive performance of the CTI and its modified indices in different ethnicities. Extended follow-up periods in existing cohorts, as well as the establishment of new longitudinal studies, will provide more robust data on the long-term predictive value of the CTI and its modified indices. Additionally, interventions targeting the components of the CTI and its modified indices, such as obesity, inflammation and insulin resistance, should be explored to determine their impact on reducing CVD risk. Finally, the development of more comprehensive and personalized risk prediction models that integrate multiple risk factors, including genetic and lifestyle factors, may further enhance the accuracy of CVD risk stratification.
Conclusion
Our study demonstrated that the CTI and its modified indices incorporating obesity-related metrics are effective predictors of CVD risk. The modified CTI indices, particularly the CTI-WC, showed better predictive performance than the original CTI across different glycemic statuses. These findings highlight the potential utility of incorporating obesity-related metrics into the CTI for enhanced CVD risk prediction. Further research is needed to validate these findings and explore the underlying mechanisms and potential interventions involved.
Supplementary Information
Acknowledgements
This study leveraged data from the CHARLS database. The authors convey their appreciation to the CHARLS research team and all contributors to the study.
Author contributions
Data collection and analyses: XD and BW. Study design: TZ and JY. Manuscript writing: all authors. Result interpretations: all authors. Manuscript proofing: all authors.
Funding
This study was supported by the Beijing Hospital Medical-Engineering Special Project on AI-Driven Multimodal Wearable Intelligent Monitoring System for Cardiovascular Diseases: From Development to Clinical Translation (Grant No. BJ-2025—157).
Data availability
The data supporting the findings of this study can be accessed through the China Health and Retirement Longitudinal Study (CHARLS) repository by registering and submitting a request on the official CHARLS website at http://charls.pku.edu.cn.
Declarations
Ethics approval
The CHARLS study adhered to the principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Peking University. Written informed consent was obtained from all participants prior to their involvement in the study.
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.
Contributor Information
Jiefu Yang, Email: yangjiefu2011@126.com.
Tong Zou, Email: zoutong2001@163.com.
References
- 1.Mensah GA, Fuster V, Murray CJL, Roth GA, Global Burden of Cardiovascular Diseases and Risks Collaborators. Global burden of cardiovascular diseases and risks, 1990-2022. J Am Coll Cardiol. 2023;82(25):2350–473. 10.1016/j.jacc.2023.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Vaduganathan M, Mensah GA, Turco JV, Fuster V, Roth GA. The global burden of cardiovascular diseases and risk: A compass for future health. J Am Coll Cardiol. 2022;80(25):2361–71. 10.1016/j.jacc.2022.11.005. [DOI] [PubMed] [Google Scholar]
- 3.Dong X, Wang J, Wang C, et al. Worldwide burden of metabolic risk-related cardiovascular disease from 1990 to 2021, with projections to 2050: A systematic analysis for the Global Burden of Disease Study 2021. Diabetes Obes Metab. 2025;27(9):4859–82. 10.1111/dom.16529. [DOI] [PubMed] [Google Scholar]
- 4.Liu Y, Guan S, Xu H, Zhang N, Huang M, Liu Z. Inflammation biomarkers are associated with the incidence of cardiovascular disease: A meta-analysis. Front Cardiovasc Med. 2023;10:1175174. 10.3389/fcvm.2023.1175174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Barac A, Wang H, Shara NM, et al. Markers of inflammation, metabolic risk factors, and incident heart failure in American Indians: The Strong Heart Study. J Clin Hypertens (Greenwich). 2012;14(1):13–9. 10.1111/j.1751-7176.2011.00560.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Feng Y, Yin L, Huang H, Hu Y, Lin S. Assessing the impact of insulin resistance trajectories on cardiovascular disease risk using longitudinal targeted maximum likelihood estimation. Cardiovasc Diabetol. 2025;24(1):112. 10.1186/s12933-025-02651-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kosmas CE, Bousvarou MD, Kostara CE, Papakonstantinou EJ, Salamou E, Guzman E. Insulin resistance and cardiovascular disease. J Int Med Res. 2023;51(3):3000605231164548. 10.1177/03000605231164548. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Huang J, Zhang J, Li L, et al. Triglyceride-glucose index and hsCRP-to-albumin ratio as predictors of major adverse cardiovascular events in STEMI patients with hypertension. Sci Rep. 2024;14(1):28112. 10.1038/s41598-024-79673-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gui Z, Chen X, Wang D, et al. Inflammatory and metabolic markers mediate the association of hepatic steatosis and fibrosis with 10-year ASCVD risk. Ann Med. 2025;57(1):2486594. 10.1080/07853890.2025.2486594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Markozannes G, Koutsioumpa C, Cividini S, et al. Global assessment of C-reactive protein and health-related outcomes: an umbrella review of evidence from observational studies and Mendelian randomization studies. Eur J Epidemiol. 2021;36(1):11–36. 10.1007/s10654-020-00681-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chia JE, Ang SP. Elevated C-reactive protein and cardiovascular risk. Curr Opin Cardiol. 2025;40(4):237–43. 10.1097/HCO.0000000000001215. [DOI] [PubMed] [Google Scholar]
- 12.Zheng L, Ye J, Liao X, Li J, Wang Q, Wang F. Frailty, high-sensitivity C-reactive protein and cardiovascular disease: a nationwide prospective cohort study. Aging Clin Exp Res. 2025;37(1):58. 10.1007/s40520-025-02928-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Dang K, Wang X, Hu J, et al. The association between triglyceride-glucose index and its combination with obesity indicators and cardiovascular disease: NHANES 2003-2018. Cardiovasc Diabetol. 2024;23(1):8. 10.1186/s12933-023-02115-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhang Q, Xiao S, Jiao X, Shen Y. The triglyceride-glucose index is a predictor for cardiovascular and all-cause mortality in CVD patients with diabetes or pre-diabetes: evidence from NHANES 2001-2018. Cardiovasc Diabetol. 2023;22(1):279. 10.1186/s12933-023-02030-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cui C, Liu L, Qi Y, et al. Joint association of TyG index and high sensitivity C-reactive protein with cardiovascular disease: a national cohort study. Cardiovasc Diabetol. 2024;23(1):156. 10.1186/s12933-024-02244-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Huo G, Tang Y, Liu Z, Cao J, Yao Z, Zhou D. Association between C-reactive protein-triglyceride glucose index and stroke risk in different glycemic status: insights from the China Health and Retirement Longitudinal Study (CHARLS). Cardiovasc Diabetol. 2025;24(1):142. 10.1186/s12933-025-02686-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Xu M, Zhang L, Xu D, Shi W, Zhang W. Usefulness of C-reactive protein-triglyceride glucose index in detecting prevalent coronary heart disease: findings from the national health and nutrition examination survey 1999-2018. Front Cardiovasc Med. 2024;11:1485538. 10.3389/fcvm.2024.1485538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Neeland IJ, Ross R, Després JP, et al. Visceral and ectopic fat, atherosclerosis, and cardiometabolic disease: a position statement. Lancet Diabetes Endocrinol. 2019;7(9):715–25. 10.1016/S2213-8587(19)30084-1. [DOI] [PubMed] [Google Scholar]
- 19.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. 10.1093/ije/dys203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.von Elm E, Altman DG, Egger M, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–9. 10.1016/j.jclinepi.2007.11.008. [DOI] [PubMed] [Google Scholar]
- 21.Flegal KM. Body-mass index and all-cause mortality. Lancet. 2017;389(10086):2284–5. 10.1016/S0140-6736(17)31437-X. [DOI] [PubMed] [Google Scholar]
- 22.Cui C, Qi Y, Song J, et al. Comparison of triglyceride glucose index and modified triglyceride glucose indices in prediction of cardiovascular diseases in middle aged and older Chinese adults. Cardiovasc Diabetol. 2024;23(1):185. 10.1186/s12933-024-02278-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wang X, Wen P, Liao Y, Wu T, Zeng L, Huang Y, et al. Association of atherogenic index of plasma and its modified indices with stroke risk in individuals with cardiovascular-kidney-metabolic syndrome stages 0-3: a longitudinal analysis based on CHARLS. Cardiovasc Diabetol. 2025;24(1):254. 10.1186/s12933-025-02784-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Zeng Y, Zhou D, Chen Y, Huo G. Association of cholesterol, high-density lipoprotein, and glucose index and its modified indices with the risk of stroke: insights from CHARLS. BMC Neurol. 2025. 10.1186/s12883-025-04575-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ruan GT, Xie HL, Zhang HY, et al. A novel inflammation and insulin resistance related indicator to predict the survival of patients with cancer. Front Endocrinol (Lausanne). 2022;13:905266. 10.3389/fendo.2022.905266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Zhang S, Wang Y, Cheng J, et al. Hyperuricemia and cardiovascular disease. Curr Pharm Des. 2019;25(6):700–9. 10.2174/1381612825666190408122557. [DOI] [PubMed] [Google Scholar]
- 27.Tanner EM, Bornehag CG, Gennings C. Repeated holdout validation for weighted quantile sum regression. MethodsX. 2019;6:2855–60. 10.1016/j.mex.2019.11.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Liu Y, He W, Ji Y, Wang Q, Li X. A linear positive association between high-sensitivity C-reactive protein and the prevalence of cardiovascular disease among individuals with diabetes. BMC Cardiovasc Disord. 2024;24(1):411. 10.1186/s12872-024-04091-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Seven E, Husemoen LL, Sehested TS, et al. Adipocytokines, C-reactive protein, and cardiovascular disease: a population-based prospective study. PLoS ONE. 2015;10(6):e0128987. 10.1371/journal.pone.0128987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Liu X, Tan Z, Huang Y, et al. Relationship between the triglyceride-glucose index and risk of cardiovascular diseases and mortality in the general population: a systematic review and meta-analysis. Cardiovasc Diabetol. 2022;21(1):124. 10.1186/s12933-022-01546-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Liu C, Liang D. The association between the triglyceride-glucose index and the risk of cardiovascular disease in US population aged ≤ 65 years with prediabetes or diabetes: a population-based study. Cardiovasc Diabetol. 2024;23(1):168. 10.1186/s12933-024-02261-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Wu T, Wei B, Song YP, et al. Predictive power of a body shape index and traditional anthropometric indicators for cardiovascular disease: a cohort study in rural Xinjiang, China. Ann Hum Biol. 2022;49(1):27–34. 10.1080/03014460.2022.2049874. [DOI] [PubMed] [Google Scholar]
- 33.He Y, Shi L. Anthropometric indicators and cardiovascular diseases risk in pre-diabetic and diabetic adults: NHANES 1999-2018 cross-sectional analysis. Exp Gerontol. 2024;194:112516. 10.1016/j.exger.2024.112516. [DOI] [PubMed] [Google Scholar]
- 34.Libby P, Buring JE, Badimon L, et al. Atherosclerosis. Nat Rev Dis Primers. 2019;5(1):56. 10.1038/s41572-019-0106-z. [DOI] [PubMed] [Google Scholar]
- 35.Zhu Y, Xian X, Wang Z, et al. Research progress on the relationship between atherosclerosis and inflammation. Biomolecules. 2018;8(3):80. 10.3390/biom8030080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Giacco F, Brownlee M. Oxidative stress and diabetic complications. Circ Res. 2010;107(9):1058–70. 10.1161/CIRCRESAHA.110.223545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Mao L, Yin R, Yang L, Zhao D. Role of advanced glycation end products on vascular smooth muscle cells under diabetic atherosclerosis. Front Endocrinol (Lausanne). 2022;13:983723. 10.3389/fendo.2022.983723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Tang WH, Kitai T, Hazen SL. Gut microbiota in cardiovascular health and disease. Circ Res. 2017;120(7):1183–96. 10.1161/CIRCRESAHA.117.309715. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Geißler C, Krause C, Neumann AM, et al. Dietary induction of obesity and insulin resistance is associated with changes in Fgf21 DNA methylation in liver of mice. J Nutr Biochem. 2022;100:108907. 10.1016/j.jnutbio.2021.108907. [DOI] [PubMed] [Google Scholar]
- 40.Long Y, Mao C, Liu S, Tao Y, Xiao D. Epigenetic modifications in obesity-associated diseases. MedComm (2020). 2024;5(2):e496. 10.1002/mco2.496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ning M, Zhang Q, Yang M. Comparison of self-reported and biomedical data on hypertension and diabetes: findings from the China Health and Retirement Longitudinal Study (CHARLS). BMJ Open. 2016;6(1):e009836. 10.1136/bmjopen-2015-009836. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Noble M, Cahill D. Hypogonadism: cardiometabolism and gonadal function in men. Metab Target Organ Damage. 2024;4:14. 10.20517/mtod.2023.58. [Google Scholar]
- 43.Lonardo A, Ballestri S, Baffy G, Weiskirchen R. Liver fibrosis as a barometer of systemic health by gauging the risk of extrahepatic disease. Metab Target Organ Damage. 2024;4:41. 10.20517/mtod.2024.42. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data supporting the findings of this study can be accessed through the China Health and Retirement Longitudinal Study (CHARLS) repository by registering and submitting a request on the official CHARLS website at http://charls.pku.edu.cn.