Associations of C‐Reactive Protein‐Triglyceride Glucose Index and Estimated Pulse Wave Velocity With All‐Cause Mortality: A Nationwide Prospective Cohort Study

ABSTRACT

Background and Aims

The combined prognostic value of metaflammation and vascular aging markers in cardiovascular‐kidney‐metabolic (CKM) syndrome is unclear. This study aimed to assess the independent and joint associations of the C‐reactive protein‐triglyceride glucose index (CTI) and estimated pulse wave velocity (ePWV) with all‐cause mortality.

Methods

This prospective study included 7183 participants from the CHARLS 2011 survey (original cohort) and 9594 from 2015 (temporal sensitivity analysis cohort). Multivariable Cox models and restricted cubic splines were used. Joint effects were analyzed by median‐based groups, and mediation was tested using the KHB method.

Results

During follow‐up, 906 deaths occurred in the original cohort. Each 1‐unit increase in CTI was associated with a 56% higher mortality risk (HR = 1.56, 95% CI: 1.37–1.77), and each 1 m/s increase in ePWV with an 11% higher risk (HR = 1.11, 95% CI: 1.05–1.18). The high CTI/high ePWV group had the highest risk (HR = 1.73, 95% CI: 1.31–2.29). ePWV mediated 6.48% of the CTI‐mortality association, while CTI mediated 2.62% of the ePWV‐mortality association. These findings were confirmed in the temporal sensitivity analysis.

Conclusion

CTI and ePWV independently and jointly associate with increased all‐cause mortality. The observed bidirectional mediation suggests potentially interconnected pathways that warrant further investigation. Integrating these accessible biomarkers could improve risk stratification in CKM syndrome.

Abbreviations

AHA

American Heart Association

AP

attributable proportion due to interaction

ASCVD

atherosclerotic cardiovascular disease

baPWV

brachial‐ankle pulse wave velocity

BMI

body mass index

BUN

blood urea nitrogen

cfPWV

carotid‐femoral pulse wave velocity

CHARLS

China Health and Retirement Longitudinal Study

CKD

chronic kidney disease

CKM

cardiovascular‐kidney‐metabolic

CMD

cardiometabolic disease

CRP

C‐reactive protein

CTI

C‐reactive protein–triglyceride glucose index

CVD

cardiovascular disease

DBP

diastolic blood pressure

eGFR

estimated glomerular filtration rate

ePWV

estimated pulse wave velocity

FPG

fasting plasma glucose

Hb

hemoglobin

HbA1c

glycated hemoglobin

HCT

hematocrit

HDL‐C

high‐density lipoprotein cholesterol

IR

Insulin resistance

IRS

insulin receptor substrate

LDL‐C

low‐density lipoprotein cholesterol

MBP

mean blood pressure

MCV

mean corpuscular volume

PLT

platelet

PP

pulse pressure

RERI

relative excess risk due to interaction

SBP

systolic blood pressure

sCr

serum creatinine

SI

synergy index

TC

total cholesterol

TG

triglycerides

TyG

triglyceride‐glucose index

UA

uric acid

VIF

variance inflation factor

WBC

white blood cell

1. Introduction

Cardiometabolic diseases (CMDs) and chronic kidney disease (CKD) frequently coexist and are intricately intertwined pathophysiologically. Their synergistic effect significantly increases mortality risk, posing a major global health challenge [1]. To address this complexity, the American Heart Association (AHA) introduced the concept of cardiovascular‐kidney‐metabolic (CKM) syndrome. This framework systematically describes a continuum from risk factor accumulation to subclinical damage and ultimately progression to clinical events, thereby emphasizing the necessity of early screening and multidisciplinary management [2, 3, 4]. Epidemiological data from China highlight the severity of this issue: the weighted prevalence of CKM syndrome among Chinese adults increased from 77.1% to 83.7% between 2010 and 2019 [5]. Another nationwide survey indicated that only 18.8% of adults were in CKM stage 0, while a substantial 23.6% had progressed to advanced stages, with all‐cause mortality risk rising significantly alongside CKM stage progression [6]. Although these studies have enhanced our understanding of the burden and prognostic impact of CKM syndrome in China [7, 8], the core challenge remains how to early identify individuals at high mortality risk within this large population and implement timely interventions.

The pathogenesis of CKM syndrome involves complex interactions among hemodynamic, inflammatory, and metabolic pathways [3], which mutually influence and intertwine, ultimately leading to a marked increase in mortality risk. On one hand, metabolic disorders centered on insulin resistance are deeply intertwined with a state of chronic low‐grade inflammation, forming the core pathophysiological axis of “metaflammation” [9]. Chronic inflammation activates intracellular kinases, interfering with insulin signal transduction and affecting insulin's metabolic actions in adipose tissue, liver, skeletal muscle, and vasculature, directly mediating the development of insulin resistance. Insulin resistance (IR), in turn, exacerbates hyperglycemia, which further intensifies oxidative stress and inflammatory responses through mechanisms such as increased advanced glycation end product generation and mitochondrial dysfunction, perpetuating a cycle that drives metabolic dysregulation and tissue damage [10, 11, 12, 13, 14, 15]. At the vascular level, this metaflammation directly drives structural remodeling and stiffening of the arterial wall by inducing endothelial dysfunction and promoting phenotypic switching of vascular smooth muscle cells and collagen deposition. On the other hand, vascular aging, with its hallmark feature of increased arterial stiffness [16, 17, 18], represents another critical pathway driving CKM syndrome progression. Arterial stiffness leads to adverse hemodynamic changes, including sustained hypertension, volume overload, and altered perfusion dynamics, which subsequently damage end organs such as the heart, brain, and kidneys [19, 20, 21, 22, 23]. The stiffened arterial wall fails to effectively cushion the pressure wave generated by cardiac ejection, resulting in significantly elevated systolic blood pressure (SBP) and pulse pressure (PP), directly increasing cardiac afterload and promoting left ventricular hypertrophy. Simultaneously, the prematurely reflected wave impairs coronary perfusion during diastole, while the widened pulse pressure continuously assaults the microvascular architecture of organs [10]. The volume overload, reduced tissue perfusion, and consequent tissue hypoxia resulting from stiffened arteries can further deteriorate insulin sensitivity and glucose/lipid metabolism, potentially amplifying inflammatory responses through mechanisms like stress pathway activation [24, 25]. This self‐reinforcing, vicious cycle of dynamic interaction strongly suggests that combined assessment of metaflammation and vascular aging may identify subgroups of individuals at extremely high mortality risk.

The triglyceride‐glucose (TyG) index has been widely validated as a reliable surrogate marker for assessing insulin resistance and its association with cardiovascular disease risk [26, 27]. However, it cannot fully capture an individual's metaflammation burden. To address this limitation, the C‐reactive protein–triglyceride glucose index (CTI) was developed. This index simultaneously quantifies the intensity of both IR and inflammation and has demonstrated potential superior to the TyG index alone in predicting cardiovascular events and mortality [28, 29, 30]. Given the practical limitations of gold‐standard measurement methods like brachial‐ankle pulse wave velocity (baPWV) and carotid‐femoral pulse wave velocity (cfPWV), the estimated pulse wave velocity (ePWV), derived from age and blood pressure, has emerged as a simple, non‐invasive, and effective alternative. Its value in predicting cardiovascular outcomes and all‐cause mortality has been confirmed, performing comparably to measured PWV [31, 32, 33, 34, 35], thus providing a practical tool for assessing vascular health at the population level.

Despite solid pathophysiological rationale, the independent and joint effects of CTI and ePWV on all‐cause mortality in middle‐aged and older adults have not been examined in prospective cohorts. Therefore, utilizing data from the nationally representative China Health and Retirement Longitudinal Study (CHARLS) cohort, this study aims to: (1) systematically evaluate the independent associations of CTI and ePWV with all‐cause mortality in middle‐aged and older adults; (2) examine the additive and multiplicative interactions between CTI and ePWV and investigate their joint effects; and (3) explore whether ePWV or CTI mediates the association of the other with all‐cause mortality, and assess the potential modifying effect of CKM stage on these associations. The findings are expected to provide new evidence for prognostic risk assessment and early intervention strategies for CKM syndrome.

2. Methods

2.1. Study Design and Participants

This was a prospective cohort analysis utilizing data from the China Health and Retirement Longitudinal Study (CHARLS), a nationwide longitudinal survey of community‐dwelling Chinese adults aged ≥ 45 years with publicly available data from 2011 to 2020 [36].

Using the 2011 survey as baseline, we sequentially excluded participants who: (1) were aged < 45 years or had missing age/sex (n = 683); (2) had missing CTI or ePWV (n = 7511); (3) had missing CKM stage (n = 146); (4) had missing covariates (n = 1642); (5) died within the baseline year (n = 6); or (6) had missing mortality status (n = 537). A total of 7183 participants were included and followed until 2020, during which 906 deaths occurred (Figure 1). Using the 2015 survey as a different time origin and applying similar exclusion criteria, a temporal sensitivity analysis cohort of 9594 participants was constructed, with 652 deaths recorded by 2020. Given that the 2015 cohort included participants from the 2011 cohort who survived to 2015, the two cohorts were not independent; this analysis was designed to test the robustness of our findings across different time frames rather than to serve as an independent external validation. Multiple imputation was not performed because missingness was largely due to non‐participation in blood collection, which may be non‐random; complete‐case analysis was used instead.

Figure 1.

Flowchart of participants' screening.

This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [37].

2.2. Data Collection

All data were collected by uniformly trained investigators using standardized procedures, including:

• 1.Sociodemographic information: age, sex, education level (junior high school and below, high school graduate, college and above), marital status (married/cohabiting vs. not), and residence (urban/rural).
• 2.Lifestyle information: smoking status (current smoker vs. not) and alcohol consumption (drinking alcohol ≥ 1 time per week vs. not).
• 3.Comorbidity information: Hypertension, diabetes, dyslipidemia, and CKD were defined based on self‐reported physician diagnosis, meeting clinical diagnostic cut‐off points, or receiving relevant treatment. Heart disease, stroke, liver disease, digestive system diseases, asthma, lung disease, arthritis, and cancer were defined based on self‐reported physician diagnosis or receiving relevant treatment.

Additionally, participants' height, weight, systolic blood pressure (SBP), and diastolic blood pressure (DBP) were measured, and body mass index (BMI) was calculated. Participants were required to fast for at least 8 h before providing fasting venous blood samples collected in the morning for the following assays: mean corpuscular volume (MCV), hematocrit (HCT), hemoglobin (Hb), white blood cell (WBC) count, platelet (PLT) count, fasting plasma glucose (FPG), glycated hemoglobin (HbA1c), triglycerides (TG), total cholesterol (TC), high‐density lipoprotein cholesterol (HDL‐C), low‐density lipoprotein cholesterol (LDL‐C), blood urea nitrogen (BUN), uric acid (UA), serum creatinine (sCr), and C‐reactive protein (CRP). The estimated glomerular filtration rate (eGFR) was calculated using the CKD‐EPI 2009 equation [38].

2.3. Calculation of CTI

CTI was calculated using the formula: 0.412*ln[CRP (mg/L)]+ln[TG (mg/dL)*FPG (mg/dL)/2] [39].

2.4. Calculation of ePWV

ePWV was calculated based on age and mean arterial pressure using the formula described by Greve et al. [40], where mean blood pressure (MBP) was calculated as DBp + 0.4*(SBP‐DBP) [41, 42]. The specific formula is:

$$\text{ePWV}=9.587-0.402\times \text{age}+4.560\times {10}^{-3}\times {\text{age}}^2-2.621\times {10}^{-5}\times {\text{age}}^2\times \text{MBP}+3.176\times {10}^{-3}\times \text{age}\times \text{MBP}-1.832\times {10}^{-2}\times \text{MBP}.$$

2.5. Definition of CKM Syndrome Stages

CKM syndrome was defined according to the 2023 American Heart Association definition [2, 3] into 5 stages:

Stage 0: No CKM health risk factors.

Stage 1: Excess or dysfunctional adiposity (including overweight, abdominal obesity, impaired glucose tolerance, or pre‐diabetes), without other metabolic risk factors or CKD.

Stage 2: Metabolic disorders (including hypertension, metabolic syndrome, hypertriglyceridemia, diabetes) or moderate‐to‐high risk CKD.

Stage 3: Subclinical cardiovascular disease (CVD) in CKM syndrome. Due to the lack of direct indicators, subclinical CVD was operationally defined by risk equivalents, including a predicted 10‐year CVD risk ≥ 20% (calculated using the PREVENT equations [43]) or a very high‐risk CKD stage (determined by KDIGO criteria [44]).

Stage 4: Clinical CVD, including coronary heart disease, myocardial infarction, heart failure, or stroke.

Detailed methods and rationale for staging followed previous similar research [29, 45]. For subsequent analyses, stages 3 and 4 were combined and defined as advanced CKM syndrome.

2.6. Ascertainment of All‐Cause Mortality

Mortality status and year of death were ascertained through follow‐up surveys in 2013, 2015, 2018, and 2020, verified via death certificates, medical records, or interviews with relatives.

2.7. Statistical Analysis

Participants were grouped based on mortality status during follow‐up. Continuous variables were tested for normal distribution using D'Agostino's K‐squared test and then described as mean ± standard deviation or median (interquartile range) accordingly. Group differences were compared using t‐tests or Wilcoxon rank‐sum tests. Categorical variables were described as frequencies (percentages), and group differences were compared using chi‐square tests.

Spearman's rank correlation coefficients were calculated to assess correlations between variables, excluding those strongly correlated with CTI or ePWV (|rho | > 0.6) (Table S1, Table S4). Univariable Cox regression was used to screen covariates associated with mortality (p < 0.20). Before multivariable adjustment, the variance inflation factor (VIF) was calculated to assess multicollinearity; variables with VIF > 5 were excluded (Table S2, Table S5). Although age was strongly correlated with ePWV, we additionally adjusted for age given its strong predictive role for mortality, with VIF values remaining below 5 after inclusion.

Multivariable Cox proportional hazards models, adjusted for different sets of covariates, were used to assess the independent associations of CTI and ePWV with all‐cause mortality risk. The proportional hazards assumption was tested using Schoenfeld residuals. To avoid over‐adjustment, the CKM stage was not included as a covariate in the primary models but was used only for subgroup stratification. Multiplicative interaction was tested by including the product term of CTI and ePWV. Restricted cubic spline (RCS) models with three knots (10th, 50th, 90th percentiles) were used to examine potential non‐linear associations. Given the absence of clinically meaningful thresholds in the RCS analyses, we used medians to dichotomize CTI and ePWV for constructing joint exposure groups (Low/Low, High/Low, Low/High, High/High); additive interaction was assessed using the relative excess risk due to interaction (RERI), the attributable proportion due to interaction (AP), and the synergy index (SI).

Mediation analysis was performed using the Karlson–Holm–Breen method [46] to test bidirectional mediation between CTI and ePWV in their associations with mortality. Subgroup analyses were stratified by age (</≥ 60 years), sex (male/female), and CKM stage (0–2/3–4). Sensitivity analyses included: (1) excluding CKM stage 0 or 4 participants; (2) excluding deaths within the first 2 years; and (3) repeating all analyses using the 2015 cohort.

All primary analyses (independent associations, joint effects, and mediation) were prespecified; subgroup and sensitivity analyses were exploratory. Given the prespecified nature of the primary analyses and consistent findings across sensitivity and temporal analyses, we did not adjust for multiple comparisons to avoid inflating type II error; exploratory results should be interpreted with caution.

All analyses were performed using STATA 18.0 (StataCorp LLC, Texas, USA) and R 4.5.1 (R Foundation for Statistical Computing, Vienna, Austria), with two‐sided p < 0.05 considered statistically significant.

3. Results

3.1. Baseline Characteristics of Participants

Baseline characteristics of the 7183 participants in the original cohort, stratified by survival status, were shown in Table 1. The median follow‐up was 8.67 years (95% CI: 8.64–8.69). Compared with survivors, decedents were older (median: 68.00 vs. 57.00 years), more likely to be male (63.47% vs. 42.70%), had lower educational attainment and poorer marital status, and exhibited less healthy lifestyles (all p < 0.05). The prevalence of most comorbidities—including hypertension, diabetes, heart disease, stroke, CKD, asthma, pulmonary disease, and cancer—was significantly higher in the deceased group (all p < 0.05), as was the proportion of participants with advanced CKM syndrome (Stages 3–4: 45.14% vs. 18.80%). Regarding clinical indicators, the deceased group had significantly higher levels of BMI, SBP, WBC, PLT, MCV, FPG, TG, TC, LDL‐C, BUN, UA, sCr, CRP, CTI, and ePWV, and lower eGFR. Baseline characteristics of the 2015 cohort (N = 9594) were presented in Supplementary Table S3, with between‐group differences largely similar to those in the original cohort.

Table 1.

Characteristics of participants included in the original cohort (N = 7183).

	Total (N = 7183)	Alive (N = 6277)	Died (N = 906)	p
Age (year), M (IQR)	58.00 (52.00–65.00)	57.00 (51.00–63.00)	68.00 (61.00–75.00)	< 0.001
Female, n (%)	3928 (54.68)	3597 (57.30)	331 (36.53)	< 0.001
Education level, n (%)				< 0.001
Junior high school or below	6579 (91.59)	5705 (90.89)	874 (96.47)
High school	545 (7.59)	518 (8.25)	27 (2.98)
College or above	59 (0.82)	54 (0.86)	5 (0.55)
Marital status, n (%)				< 0.001
With partnership	6334 (88.18)	5658 (90.14)	676 (74.61)
Without a partnership or never married	849 (11.82)	619 (9.86)	230 (25.39)
Household type, n (%)				0.425
Urban	2343 (32.62)	2058 (32.79)	285 (31.46)
Rural	4840 (67.38)	4219 (67.21)	621 (68.54)
Current Smoking, n (%)	2140 (29.79)	1782 (28.39)	358 (39.51)	< 0.001
Drinking at least once per week, n (%)	1213 (16.89)	1024 (16.31)	189 (20.86)	< 0.001
CKM syndrome stages, n (%)				< 0.001
0	743 (10.34)	685 (10.91)	58 (6.40)
1	1086 (15.12)	1015 (16.17)	71 (7.84)
2	3765 (52.42)	3397 (54.12)	368 (40.62)
3	521 (7.25)	315 (5.02)	206 (22.74)
4	1068 (14.87)	865 (13.78)	203 (22.41)
CKM syndrome advancing stage, n (%)				< 0.001
0–2	5594 (77.88)	5097 (81.20)	497 (54.86)
3–4	1589 (22.12)	1180 (18.80)	409 (45.14)
Hypertension, n (%)	2938 (40.90)	2442 (38.90)	496 (54.75)	< 0.001
Diabetes, n (%)	1208 (16.82)	1001 (15.95)	207 (22.85)	< 0.001
Dyslipidemia, n (%)	3470 (48.31)	3049 (48.57)	421 (46.47)	0.236
Heart disease, n (%)	928 (12.92)	761 (12.12)	167 (18.43)	< 0.001
Stroke, n (%)	186 (2.59)	140 (2.23)	46 (5.08)	< 0.001
CKD, n (%)	674 (9.38)	527 (8.40)	147 (16.23)	< 0.001
Liver disease, n (%)	279 (3.88)	234 (3.73)	45 (4.97)	0.071
Digestive disease, n (%)	1918 (26.70)	1697 (27.04)	221 (24.39)	0.093
Asthma, n (%)	321 (4.47)	238 (3.79)	83 (9.16)	< 0.001
Pulmonary disease, n (%)	858 (11.94)	656 (10.45)	202 (22.30)	< 0.001
Arthritis, n (%)	2786 (38.79)	2444 (38.94)	342 (37.75)	0.493
Cancer, n (%)	7 (0.10)	3 (0.05)	4 (0.44)	< 0.001
BMI (kg/m2), M (IQR)	23.17 (20.96–25.67)	23.31 (21.13–25.81)	22.05 (19.96–24.62)	< 0.001
SBP (mmHg), M (IQR)	126.00 (113.50–141.00)	125.00 (113.00–140.00)	134.00 (119.00–150.00)	< 0.001
DBP (mmHg), M (IQR)	74.00 (66.50–82.50)	74.00 (66.50–82.50)	75.50 (66.50–83.50)	0.076
WBC (10^9/L), M (IQR)	5.95 (4.95–7.20)	5.90 (4.92–7.20)	6.10 (5.00–7.60)	< 0.001
PLT (10^9/L), M (IQR)	206.00 (162.00–255.00)	207.00 (163.00–255.00)	199.00 (153.00–255.00)	0.015
MCV (fL), M (IQR)	91.30 (86.90–95.60)	91.20 (86.80–95.40)	92.80 (88.00–97.20)	< 0.001
HCT, M (IQR)	41.50 (37.90–45.00)	41.50 (37.90–45.00)	41.40 (37.50–45.10)	0.502
Hb (g/dL), M (IQR)	14.20 (13.00–15.50)	14.20 (13.00–15.50)	14.20 (13.00–15.50)	0.920
FPG (mg/dL), M (IQR)	102.42 (94.50–113.22)	102.24 (94.32–112.68)	103.86 (94.50–117.90)	0.002
HbA1c (%), M (IQR)	5.10 (4.90–5.40)	5.10 (4.90–5.40)	5.20 (4.90–5.50)	0.085
TG (mg/dL), M (IQR)	105.32 (75.22–153.99)	106.20 (75.22–154.88)	103.54 (72.57–146.02)	0.009
TC (mg/dL), M (IQR)	190.59 (167.40–215.34)	190.98 (168.17–215.34)	186.53 (160.83–213.02)	< 0.001
HDL‐C (mg/dL), M (IQR)	49.48 (40.59–59.54)	49.10 (40.59–59.54)	49.87 (40.98–61.08)	0.105
LDL‐C (mg/dL), M (IQR)	114.43 (93.17–137.63)	115.21 (93.94–138.02)	109.41 (87.37–133.76)	< 0.001
BUN (mg/dL), M (IQR)	15.13 (12.55–18.18)	15.04 (12.49–17.95)	15.66 (13.00–19.07)	< 0.001
UA (mg/dL), M (IQR)	4.25 (3.55–5.09)	4.21 (3.53–5.05)	4.57 (3.65–5.48)	< 0.001
sCr (mg/dL), M (IQR)	0.76 (0.66–0.87)	0.75 (0.64–0.86)	0.80 (0.69–0.94)	< 0.001
eGFR (ml/min/1.73m2), M (IQR)	94.80 (84.10–102.28)	95.75 (85.53–102.86)	87.85 (74.39–96.19)	< 0.001
CRP (mg/L), M (IQR)	1.03 (0.55–2.16)	0.97 (0.53–1.98)	1.48 (0.71–4.01)	< 0.001
CTI, M (IQR)	4.70 (4.33–5.12)	4.68 (4.32–5.09)	4.88 (4.46–5.36)	< 0.001
ePWV (m/s), M (IQR)	10.95 (9.67–12.55)	10.75 (9.56–12.20)	12.87 (11.27–14.51)	< 0.001

Abbreviations: BMI, body mass index; BUN, blood urea nitrogen; CKM, cardiovascular‐kidney‐metabolic; CRP, C‐reactive protein; CTI, triglyceride‐glucose‐body mass index; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; FPG, fasting plasma glucose; Hb, hemoglobin; HbA1c, glycosylated hemoglobin; HCT, hematocrit; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; MCV, mean corpuscular volume; PLT, platelet; SBP, systolic blood pressure; sCr, serum creatinine; TC, total cholesterol; TG, triglycerides; UA, uric acid; WBC, white blood cell count.

3.2. Independent Associations of CTI and ePWV with All‐Cause Mortality

In the original cohort, CTI and ePWV each demonstrated consistently significant independent associations with all‐cause mortality across models (Table 2). In the age‐unadjusted model (Model 3), each 1‐unit increase in CTI was associated with a 58% higher mortality risk (HR = 1.58, 95% CI: 1.40–1.80), and each 1 m/s increase in ePWV with a 36% higher risk (HR = 1.36, 95% CI: 1.31–1.41). After additional adjustment for age (Model 4), the corresponding HRs were attenuated to 1.56 (95% CI: 1.37–1.77) for CTI and 1.11 (95% CI: 1.05–1.18) for ePWV. The multiplicative interaction between CTI and ePWV (as continuous variables) was statistically significant in the unadjusted model (p = 0.031), but was attenuated and became non‐significant in all multivariable‐adjusted models (all P for multiplicative interaction > 0.05). RCS analysis suggested an approximately linear positive relationship across the clinically relevant range for both markers, with no meaningful thresholds identified (Figure 2). When dichotomized by their medians, both high‐CTI and high‐ePWV groups showed significantly increased mortality risk (all HR > 1, p < 0.05), consistent with the continuous analyses. In the 2015 cohort, the independent associations of CTI and ePWV with all‐cause mortality remained significant (Table 4).

Table 2.

Independent associations of CTI and ePWV with all‐cause mortality in the original cohort.

Model		CTI		ePWV		p for interactiona
		HR (95%CI)	p	HR (95%CI)	p
Crude	Continuous	1.41 (1.27,1.57)	< 0.001	1.41 (1.37,1.44)	< 0.001	0.031
	Binary	1.27 (1.12,1.46)	< 0.001	3.83 (3.26,4.49)	< 0.001	0.094
Model 1b	Continuous	1.50 (1.35,1.67)	< 0.001	1.36 (1.32,1.40)	< 0.001	0.125
	Binary	1.36 (1.19,1.55)	< 0.001	3.12 (2.65,3.68)	< 0.001	0.788
Model 2c	Continuous	1.48 (1.32,1.65)	< 0.001	1.42 (1.37,1.47)	< 0.001	0.194
	Binary	1.29 (1.12,1.48)	< 0.001	2.98 (2.49,3.57)	< 0.001	0.201
Model 3d	Continuous	1.58 (1.40,1.80)	< 0.001	1.36 (1.31,1.41)	< 0.001	0.243
	Binary	1.40 (1.21,1.63)	< 0.001	2.43 (2.02,2.93)	< 0.001	0.298
Model 4e	Continuous	1.56 (1.37,1.77)	< 0.001	1.11 (1.05,1.18)	0.001	0.146
	Binary	1.35 (1.16,1.56)	< 0.001	1.24 (1.00,1.52)	0.047	0.423

Abbreviations: BMI, body mass index; BUN, blood urea nitrogen; CI, confidence interval; CKD, chronic kidney disease; CTI, triglyceride‐glucose‐body mass index; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; HbA1c, glycosylated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; HR, hazard ratio; LDL‐C, low‐density lipoprotein cholesterol; MCV, mean corpuscular volume; UA, uric acid; WBC, white blood cell count.

^a The multiplicative interaction between CTI and ePWV was examined.

^b Multivariable Cox proportional hazard models were adjusted for gender, education level, marital status, smoking status, and alcohol consumption.

^c Multivariable Cox proportional hazard models were adjusted for gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, and cancer.

^d Multivariable Cox proportional hazard models were adjusted for gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, cancer, BMI, WBC, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR.

^e Multivariable Cox proportional hazard models were adjusted for age, gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, cancer, BMI, WBC, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR.

Figure 2.

Nonlinear associations of CTI and ePWV with all‐cause mortality in the original cohort and the 2015 cohort. a. Multivariable Cox proportional hazard models were adjusted for age, gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, cancer, BMI, WBC, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR. b. Multivariable Cox proportional hazard models were adjusted for age, gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, arthritis, cancer, BMI, WBC, PLT, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR. BMI, body mass index; BUN, blood urea nitrogen; CI, confidence interval; CKD, chronic kidney disease; CTI, triglyceride‐glucose‐body mass index; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; HbA1c, glycosylated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; HR, hazard ratio; LDL‐C, low‐density lipoprotein cholesterol; MCV, mean corpuscular volume; PLT, platelet; UA, uric acid; WBC, white blood cell count.

Table 4.

Independent and joint associations of CTI and ePWV with all‐cause mortality in the 2015 cohort.

		HR (95%CI)a	p	p for interactionb,c
Independent association				0.788
	CTI	1.38 (1.19,1.60)	< 0.001
	ePWV	1.13 (1.06,1.22)	0.001
Joint association				0.276
	Low CTI + Low ePWV	1	ref
	High CTI + Low ePWV	1.32 (0.95,1.85)	0.100
	Low CTI + High ePWV	1.53 (1.08,2.16)	0.017
	High CTI + High ePWV	1.55 (1.11,2.17)	0.010
	p for trend	1.12 (1.03,1.21)	0.008

Abbreviations: BMI, body mass index; BUN, blood urea nitrogen; CI, confidence interval; CKD, chronic kidney disease; CTI, triglyceride‐glucose‐body mass index; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; HbA1c, glycosylated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; HR, hazard ratio; LDL‐C, low‐density lipoprotein cholesterol; MCV, mean corpuscular volume; PLT, platelet; UA, uric acid; WBC, white blood cell count.

^a Multivariable Cox proportional hazard models were adjusted for age, gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, arthritis, cancer, BMI, WBC, PLT, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR.

^b The multiplicative interaction between CTI and ePWV was examined in an analysis for their independent associations.

^c The additive interaction between CTI and ePWV was examined in the analysis for their joint associations.

3.3. Joint Associations of CTI and ePWV With All‐Cause Mortality

In the age‐adjusted model (Table 3), using the low‐CTI + low‐ePWV group as the reference, the high‐CTI + low‐ePWV group had a 33% increased mortality risk (HR = 1.33, 95% CI: 1.01–1.74), the low‐CTI + high‐ePWV group had a 49% increased risk (HR = 1.49, 95% CI: 1.11–1.99), and the high‐CTI + high‐ePWV group had the highest risk, being 1.73 times that of the reference group (HR = 1.73, 95% CI: 1.31–2.29). A significant dose‐response trend was observed across these joint categories (p for trend < 0.001). No significant additive interaction was detected between CTI and ePWV (p for RERI = 0.705). In the 2015 cohort (Table 4), the pattern of joint effects was generally consistent, with the high‐CTI + high‐ePWV group showing the highest mortality risk (HR = 1.55, 95% CI: 1.11–2.17), and no significant additive interaction was detected (p for RERI = 0.276).

Table 3.

Joint associations of CTI and ePWV with all‐cause mortality in the original cohort.

Model	Group	HR (95%CI)	p	p for RERIa
Crude model				0.453
	Low CTI + Low ePWV	1	ref
	High CTI + Low ePWV	1.58 (1.19,2.10)	0.002
	Low CTI + High ePWV	4.43 (3.49,5.62)	< 0.001
	High CTI + High ePWV	5.31 (4.23,6.67)	< 0.001
	p for trend	1.46 (1.38,1.55)	< 0.001
Adjusted model				0.705
	Low CTI + Low ePWV	1	ref
	High CTI + Low ePWV	1.33 (1.01,1.74)	0.043
	Low CTI + High ePWV	1.49 (1.11,1.99)	0.007
	High CTI + High ePWV	1.73 (1.31,2.29)	< 0.001
	p for trend	1.17 (1.09,1.15)	< 0.001

Abbreviations: BMI, body mass index; BUN, blood urea nitrogen; CI, confidence interval; CKD, chronic kidney disease; CTI, triglyceride‐glucose‐body mass index; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; HbA1c, glycosylated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; HR, hazard ratio; LDL‐C, low‐density lipoprotein cholesterol; MCV, mean corpuscular volume; RERI, relative excess risk due to interaction; UA, uric acid; WBC, white blood cell count.

^a The additive interaction between CTI and ePWV was examined.

^b Multivariable Cox proportional hazard models were adjusted for age, gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, cancer, BMI, WBC, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR.

3.4. Mediating Effects of CTI and ePWV

In the original cohort (Figure 3A), ePWV partially mediated the association between CTI and all‐cause mortality, with a mediation proportion of 6.48%. Conversely, CTI also partially mediated the association between ePWV and all‐cause mortality, with a mediation proportion of 2.62%. A similar pattern was observed in the 2015 cohort (Figure 3B), with mediation proportions of 5.21% for ePWV and 4.51% for CTI.

Figure 3.

Mediating effects of CTI and ePWV on each other's associations with all‐cause mortality in the original cohort and 2015 cohort. (A) Multivariable logistic regression models in the original cohort were adjusted for age, gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, cancer, BMI, WBC, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR. (B) Multivariable logistic regression models in the 2015 cohort were adjusted for age, gender, education level, marital status, smoking status, alcohol consumption, hypertension, diabetes, heart disease, stroke, CKD, liver disease, digestive disease, asthma, pulmonary disease, arthritis, cancer, BMI, WBC, PLT, MCV, HbA1c, HDL‐C, LDL‐C, BUN, UA, and eGFR. BMI, body mass index; BUN, blood urea nitrogen; CI, confidence interval; CKD, chronic kidney disease; CTI, triglyceride‐glucose‐body mass index; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; HbA1c, glycosylated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; MCV, mean corpuscular volume; OR, odds ratio; PLT, platelet; UA, uric acid; WBC, white blood cell count.

3.5. Subgroup Analyses

Subgroup analyses were performed based on sex, age, and CKM stage. The independent positive associations of CTI with all‐cause mortality remained significant across all examined subgroups (Table S6). For ePWV, the positive associations persisted in subgroups of males, females, those aged ≥ 60 years, and those with CKM stages 3–4, but were not significant in the < 60 years or CKM stages 0–2 subgroups (Table S6). Joint effect analysis showed that the high‐CTI + high‐ePWV group was associated with the highest mortality risk in most subgroups (Table S7).

3.6. Sensitivity Analyses

Sensitivity analyses confirmed the overall robustness of the main findings. The independent associations of CTI with all‐cause mortality remained significant across all sensitivity scenarios; for ePWV, the association was attenuated after excluding participants with CKM stage 4 (HR = 1.23, 95% CI: 0.97–1.56, p = 0.088) but remained significant in other scenarios (Table S8). The high‐CTI + high‐ePWV group consistently showed the highest mortality risk in joint effect analyses, with no significant additive interaction detected in any sensitivity analysis (Table S9). Mediation analyses remained broadly consistent, except that the mediation effects were not significant after excluding CKM stage 4 (Table S10).

4. Discussion

This large‐scale prospective cohort study is the first to systematically demonstrate the independent and joint predictive value of the metabolic inflammation marker CTI and the vascular aging indicator ePWV for all‐cause mortality among middle‐aged and older Chinese adults. Our principal findings are threefold: First, both CTI and ePWV were independent risk factors for all‐cause mortality, with robust associations persisting after extensive multivariable adjustment. Second, individuals with concurrent elevations in both biomarkers faced a substantially heightened mortality risk, demonstrating a clear dose‐response relationship. Third, the observed mediation effects suggested bidirectional pathways that may link metaflammation and vascular aging to increased mortality risk, though these findings remain exploratory.

Our findings align with and extend a growing body of evidence. CTI was independently associated with all‐cause and cardiovascular mortality among US adults in early CKM syndrome stages [28], and similar findings were reported in a nationwide Chinese cohort [29]. Together with earlier meta‐analyses linking the TyG index [47] and CRP [48] to all‐cause mortality, these studies reinforce the biological plausibility of CTI as a marker of metaflammation. Furthermore, a recent systematic review confirmed the association of ePWV with cardiovascular outcomes and all‐cause mortality [49], an association that remains significant across different CKM stages [50, 51, 52, 53, 54]. These findings are consistent with previous meta‐analyses establishing the prognostic value of PWV [55], and recent studies have further highlighted the clinical relevance of vascular elasticity in diverse settings [56, 57]. This collective evidence supports the reliability of our findings.

The results suggest a pathophysiological intertwining between metabolic inflammation and vascular aging that may contribute to mortality risk. Pro‐inflammatory cytokines (e.g., TNF‐α, IL‐6, IL‐1β) may impair insulin signaling via activation of intracellular kinases such as JNK and IKKβ, leading to insulin receptor substrate phosphorylation and degradation, thereby inducing systemic insulin resistance [9, 58]. In this process, activation of the NLRP3 inflammasome may further amplify the inflammatory cascade [59], contributing to elevations in CTI components (CRP and insulin resistance). This inflammatory cascade not only disrupts metabolic homeostasis but may also directly affect the vasculature by inducing endothelial dysfunction, promoting vascular smooth muscle cell phenotypic switching and collagen deposition, and accelerating accumulation of advanced glycation end‐products, collectively contributing to increased arterial stiffness [16, 60]. Conversely, increased aortic stiffness could exacerbate metabolic dysfunction through multiple mechanisms, including impaired microvascular perfusion, increased cardiac afterload, and activation of the renin‐angiotensin‐aldosterone system and sympathetic nervous system, potentially worsening insulin sensitivity and amplifying inflammatory responses [24, 25]. While previous research has indicated that arterial stiffness mediates the risk of IR‐related atherosclerotic cardiovascular disease (ASCVD) [61], whether IR in turn mediates the arterial stiffness‐CVD risk remains unclear. The bidirectional mediation effects observed in our study provide exploratory epidemiological evidence for this hypothesized interconnection, rather than definitive proof of causal pathways. Furthermore, the emerging field of immunometabolism offers a more granular perspective on this interplay. Key metabolites such as lactate, glutamine, and cholesterol homeostasis may influence macrophage polarization (pro‐inflammatory M1 vs. anti‐inflammatory M2) and atherosclerotic plaque stability [60, 62, 63, 64]. The glucose, lipid, and CRP levels integrated into CTI may indirectly reflect this underlying immunometabolic disequilibrium, thereby positioning CTI as a potential bridge connecting macroscopic risk with microscopic mechanisms.

Our findings have translational potential. The calculation of both CTI and ePWV relies on routine clinical parameters, making them suitable for integration into cardiovascular risk stratification frameworks, particularly within the context of CKM syndrome. Identifying the “high CTI + high ePWV” phenotype may enable early identification of individuals who would benefit most from intensive intervention. Lifestyle modification forms the cornerstone: healthy diets, regular physical activity, smoking cessation, and weight management have been shown to concurrently improve insulin sensitivity and arterial compliance [65, 66, 67]. Pharmacologically, novel agents such as SGLT2 inhibitors and GLP‐1 receptor agonists exhibit pleiotropic effects beyond glucose‐lowering, including anti‐inflammatory actions and reduced arterial stiffness [68, 69, 70, 71, 72], theoretically targeting pathways reflected by both CTI and ePWV. Statins may also offer additional benefits through their lipid‐independent anti‐inflammatory and vascular protective effects [73, 74]. While these considerations underscore the translational potential, we acknowledge that we did not formally quantify the incremental predictive value of CTI and ePWV over traditional risk factors; future dedicated prediction modeling studies are warranted to address this question.

Strengths of this study include its large‐scale, nationally representative sample, prospective design, long‐term follow‐up, comprehensive adjustment for confounders, and temporal sensitivity analysis using the 2015 cohort. The investigation of joint effects and bidirectional mediation adds novel insights. However, several limitations should be acknowledged. First, the observational nature precludes definitive causal inference, and residual confounding cannot be entirely ruled out. Given the cross‐sectional measurement of mediators and exposures at baseline, the mediation findings should be interpreted as hypothesis‐generating. Second, ePWV is an estimated index derived from age and blood pressure rather than a direct measure of arterial stiffness; although we adjusted for age and blood pressure‐related variables, residual confounding from these core components cannot be fully excluded. Third, we assessed only all‐cause mortality due to the lack of cause‐specific death data in CHARLS, which limits our ability to delineate whether the observed associations are driven primarily by vascular, metabolic, or combined mechanisms. Additionally, a substantial number of participants were excluded due to missing CTI or ePWV data at baseline, which may have introduced selection bias and limited the generalizability of our findings. Fourth, the study population was restricted to middle‐aged and older Chinese adults, cautioning against generalization to other ethnicities or younger populations. Finally, a standardized method for defining CKM syndrome in the Chinese population has not been established, and its evolution might affect the long‐term applicability of the results.

5. Conclusions

This study found that both CTI and ePWV were independent risk factors for all‐cause mortality after multivariable adjustment. Individuals with concurrent elevations in both biomarkers faced the highest mortality risk. Mediation analyses also indicated that CTI and ePWV partially mediated each other's association with all‐cause mortality, though these exploratory findings require cautious interpretation. These associations were consistent across various subgroups, particularly among patients with both early and advanced CKM syndrome stages. In summary, CTI and ePWV, as simple and accessible clinical parameters, hold significant value for identifying high‐risk individuals. Integrating these biomarkers, which reflect metabolic inflammation and vascular aging, respectively, into risk stratification could facilitate early and precise prevention strategies for CKM syndrome.

Author Contributions

Huixiao Yuan: conceptualization, formal analysis, methodology, writing – original draft. Qingqing Li: conceptualization, methodology, formal analysis, writing – original draft. Lei Sun: methodology, validation, writing – original draft. Jianli Ge: data curation, writing – original draft. Shasha Geng: data curation, writing – original draft. Xin Chen: data curation, writing – original draft. Yingqian Zhu: data curation, writing – original draft. Hua Jiang: funding acquisition, writing – review and editing. Yang Li: validation, supervision, writing – review and editing.

Ethics Statement

The CHARLS study protocol for survey and blood sample collection was approved by the Peking University Biomedical Institutional Review Board (IRB00001052‐11014, IRB00001052‐11015). The study was conducted in accordance with the ethical principles outlined in the 2008 Declaration of Helsinki. All participants were informed about the purpose of the study, assured of confidentiality, and provided written consent prior to participation. Participation was voluntary, and respondents could withdraw at any time without consequence. There was no requirement for additional ethical approval from the approved data users.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

Yang Li affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.

Supporting information

Supporting File

Yuan H., Li Q., Sun L., et al., “Associations of C‐Reactive Protein‐Triglyceride Glucose Index and Estimated Pulse Wave Velocity With All‐Cause Mortality: A Nationwide Prospective Cohort Study,” Health Science Reports 9 (2026): e72870, 10.1002/hsr2.72870.

Data Availability Statement

The data that support the findings of this study are available on the CHARLS project website at https://charls.pku.edu.cn/en/. Derived data supporting the findings of this study are available from the corresponding author L.Y., on request.