Prognostic stratification with composite insulin resistance–inflammation biomarkers in patients with chronic kidney disease and coronary artery disease across glycemic statuses

Abstract

Background

Patients with chronic kidney disease (CKD) and coronary artery diseases (CAD) have a poor long-term prognosis. Although insulin resistance (IR) and systemic inflammation are well-established drivers of cardiovascular risk, the prognostic value of their composite assessment across the glycemic spectrum in patients with CKD and CAD remains undetermined. This study aimed to evaluate the prognostic utility of composite IR-inflammation biomarkers for predicting mortality in patients with CKD and CAD stratified by glycemic status.

Methods

1353 patients with CKD and CAD were enrolled from National Health and Nutrition Examination Survey (NHANES) data (1999–2018). Composite biomarkers (TyG-hsCRP, TyG-CRP, and C-reactive Protein–Triglyceride Glucose Index [CTI]) were calculated. Patients were categorized by glycemic status (normoglycemia, prediabetes, diabetes) based on WHO/IEC criteria. The endpoint was all-cause and cardiovascular disease (CVD) death. Statistical analyses included Cox regression, Nelson-Aalen cumulative hazard plots with Log-rank test, restricted cubic splines, ROC curves, and reclassification metrics, adjusted for demographics, comorbidities, and treatments. Subgroup and sensitivity analyses ensured robustness.

Results

Over a median follow-up of 63-months, 744 all-cause and 323 CVD deaths occurred. Adjusted models showed elevated composite indices linked to higher mortality (e.g., CTI HR 1.43 [95% CI 1.24–1.65] for all-cause; HR 1.32 [1.06–1.64] for CVD). CTI provided good discrimination (AUC 0.700) and reclassification (IDI 0.010; NRI 0.196 for all-cause). The predictive utility of all three composite biomarkers was most pronounced in patients with diabetes, whereas CTI retained the strong association with all-cause mortality in normoglycemic and prediabetic patients. Risk stratification using both CTI and glycemic status identified patients with diabetes and high CTI as having the highest all-cause (HR 1.63 [1.22–2.17]) and CVD (HR 1.37 [0.88–2.14]) death risk.

Conclusion

Composite biomarkers integrating IR and inflammation, particularly CTI, significantly enhance mortality prediction in patients with CKD and CAD. The predictive utility is modulated by underlying glycemic status, enabling refined risk stratification and potentially guiding tailored management strategies for this complex patient population.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-026-03108-0.

Research insight

What is currently known about this topic?

Both insulin resistance and systemic inflammation are established risk factors in CAD, yet their combined prognostic value in high-risk patients with CKD and CAD across different glycemic states remains unquantified.

What is the key research question?

Can biomarkers integrating insulin resistance and inflammation improve risk stratification for mortality in CKD and CAD patients, and does their predictive utility vary across the glycemic spectrum?

What is new?

We demonstrate that the CTI provides good prognostic value, with its predictive strength being significantly modulated by underlying glycemic status—revealing distinct risk patterns in normoglycemic, prediabetic, and diabetic subgroups.

How might this study influence clinical practice?

This research enables refined risk stratification using readily available biomarkers, potentially guiding personalized management strategies by identifying high-risk CKD and CAD patients who may benefit from more intensive metabolic and anti-inflammatory interventions tailored to their glycemic status.

Introduction

Coronary artery disease (CAD) is a leading cause of death worldwide, placing a heavy burden on healthcare systems [1, 2]. Renal impairment strongly correlates with cardiovascular diseases. Patients with chronic kidney disease (CKD) face a higher risk of cardiovascular events [3, 4]. This link may stem from insulin resistance (IR) and disrupted glucose metabolism due to kidney dysfunction [5]. Once CKD patients develop CAD, their prognosis worsens, with increased risks of adverse outcomes and stress-induced hyperglycemia [6]. Thus, early interventions and secondary prevention are essential to reduce long-term cardiovascular risks and improve outcomes in this group.

However, due to the high heterogeneity of renal disease combined with cardiovascular conditions, traditional statistical methods often perform inadequately in developing predictive models, which limits their effectiveness [7]. Single biomarkers also have drawbacks: the triglyceride-glucose (TyG) index, a surrogate for IR, links to CAD severity and events but overlooks inflammation’s role [8, 9]. Similarly, inflammatory markers like C-reactive protein (CRP) or high-sensitivity CRP (hsCRP) predict atherosclerotic burden and major adverse cardiovascular events (MACE), yet they ignore IR [10–12].

IR and inflammation interact synergistically in CKD and CAD patients. IR promotes endothelial dysfunction and oxidative stress, amplifying inflammation. Meanwhile, chronic inflammation worsens IR by impairing insulin signaling pathways. Together, they accelerate atherosclerosis, plaque instability, and vascular calcification—key drivers of poor outcomes in this population [13]. The C-reactive protein–triglyceride glucose index (CTI) integrates these factors, offering a better tool to capture their combined effects [13]. Yet, the predictive value of IR and inflammation together, especially across glycemic statuses (normoglycemia, prediabetes, diabetes), remains underexplored in CKD and CAD patients.

This study aimed to assess the predictive value of composite indices (TyG-hsCRP, TyG-CRP, and CTI) for long-term mortality in CKD and CAD patients, stratified by glycemic status (normoglycemia, prediabetes, and diabetes). We also compared these biomarkers’ performance and their added value for risk stratification and evaluate their integrated stratification value alongside glycemic status. These insights could refine risk assessment and guide clinical decisions in this high-risk group.

Methods

Study population and design

A pooled analysis was performed using data from NHANES survey cycles spanning 1999 to 2018. The Centers for Disease Control and Prevention (CDC) continuously conducts the National Health and Nutrition Examination Survey (NHANES), a project that prospectively evaluates the health, diet, and lifestyle of individuals in the U.S., and this research made use of its data [14, 15]. DataDryad (10.5061/dryad.d5h62) provides access to the datasets. The CDC’s institutional review board for research involving human subjects granted permission for this study.

Our study incorporated CKD and CAD participants who were 20 years or older at the time of registration. We utilized exclusion standards to eliminate persons with incomplete data on vital parameters, such as triglycerides (TG), fasting plasma glucose (FPG), hemoglobin A1c (HbA1c), diabetes condition, creatinine concentrations, and life expectancy. A detailed flowchart of patient selection is provided in Fig. 1.

Fig. 1.

Flowchart of patient enrollment. TG Triglyceride, FBG Fasting blood glucose, CRP C-reactive protein, hsCRP High-sensitivity C-reactive protein, HbA1c Hemoglobin A1c, eGFR Estimated glomerular filtration rate

Biomarkers calculation

TyG index was calculated using the following formula: .

The composite biomarker TyG-hsCRP, integrating the TyG index and the inflammatory marker hsCRP, was calculated using the following formula:

The composite biomarker TyG-CRP, integrating the TyG index and the inflammatory marker CRP, was calculated as:

The C-reactive Protein–Triglyceride Glucose Index (CTI), which combines biomarkers of inflammation (CRP) and insulin resistance (TyG index), was calculated according to the established equation: [13].

Definition of glycemic status

Glycemic status was defined according to World Health Organization/International Expert Committee (WHO/IEC) criteria. Prediabetes was defined as an HbA1c level of 6.0–6.5% or an FPG level of 6.1–7.0 mmol/L [16]. Diabetes was defined based on a self-reported physician’s diagnosis, current use of antidiabetic medication, or newly detected hyperglycemia (defined as FPG ≥ 7.0 mmol/L or HbA1c ≥ 6.5%). Individuals with glycemic levels below the threshold for prediabetes were classified as having normoglycemia.

Measurement of inflammation biomarkers

Inflammation biomarkers, including CRP, hsCRP, were measured under ambulatory stable conditions. Blood samples were collected from non-institutionalized participants during routine visits to mobile examination centers in a community-based setting. hsCRP was quantified using high-sensitivity latex-enhanced nephelometry or immunoturbidimetric assays (e.g., on Beckman Coulter analyzers) for precise low-level detection. CRP measurements followed similar immunoturbidimetric protocols, with assay details ensuring reproducibility across NHANES cycles as per CDC laboratory guidelines.

Clinical presentations and outcomes

The endpoints were all-cause mortality and cardiovascular disease (CVD) mortality. The National Center for Health Statistics paired NHANES data with death records from the National Death Index to acquire mortality data [17]. Participants who could not be linked to death data during the follow-up period were considered alive. The evaluation was conducted on both general death rates and those related to cardiovascular diseases, with the latter being categorized according to the International Classification of Diseases10th Revision (ICD-10) codes I00-I09, I11, I13, I20-I51, and I60-I69.

The presence of CAD was confirmed through a self-declared doctor’s diagnosis of coronary heart disease, angina, heart attack, or myocardial infarction. The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation was utilized to compute the estimated glomerular filtration rate (eGFR) [18], where a CKD diagnosis was determined by an eGFR value was lower than 60 mL/min per 1.73 m².

Statistical analysis

Patient baseline characteristics are presented as medians with interquartile ranges (IQR) for continuous variables and as frequencies with percentages for categorical variables. After confirming that the data were missing at random, multiple imputation using chained equations (MICE) with the random forest method, implemented in the R mice package, was employed to handle missing data. The missing data patterns and details were showed in the Supplementary Fig. 1 and Supplementary Table 1.

Restricted cubic spline (RCS) analysis was performed to examine the linearity of the relationship between the various biomarkers and clinical outcomes. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between the biomarkers and mortality risk. The proportional hazards assumption was verified using Schoenfeld residuals. Participants were stratified into two groups using pre-specified cut-off values for the inflammatory biomarkers based on the median value (TyG-hsCRP: 11.48; TyG-CRP: 9.40; CTI: 8.45). Subsequent analyses were further stratified by glycemic status (Normoglycemia, Prediabetes, and Diabetes). Two analytical models were constructed: a crude model without adjustments, and a fully adjusted model that controlled for age, gender, body mass index (BMI), systolic blood pressure, smoking status, diabetes status, hypertension, dyslipidemia, estimated glomerular filtration rate (eGFR), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglyceride and use of antidiabetic agents, antihypertensive medication and lipid-lowering medication. The variables included for adjustment were selected based on collinearity testing using the variance inflation factor (VIF), in conjunction with prior literature and clinical expertise [19, 20].

Cumulative hazard plots were generated using the Nelson-Aalen estimator to visualize risk accumulation, with group differences tested via Log-rank statistic. Receiver operating characteristic (ROC) curves were constructed, and the areas under the curve (AUC) were compared using the DeLong test. Integrated Discrimination Improvement (IDI) and Net Reclassification Improvement (NRI) are used to evaluate the enhancement in disease risk prediction accuracy provided by biomarkers. Subgroup analyses were conducted based on age (< 65 vs. ≥ 65 years), sex (male vs. female), BMI (< 30 kg/m² vs. ≥ 30 kg/m²), smoking status (yes vs. no), hypertension (yes vs. no), and hypercholesterolemia status (yes vs. no). Sensitivity analyses were conducted by excluding variables associated with the TyG index and blood glucose levels (including diabetes status, triglyceride and hemoglobin A1c) and repeating the analyses.

All statistical analyses were performed using R software (version 4.1.3) and Stata (version 17.0). A two-sided p-value < 0.05 was considered statistically significant.

Results

Study characteristics

A total of 1353 participants with CKD and CAD were included in the final analysis. The median age was 77 years (IQR 69–80), 47.1% were female. Glycemic status distribution was as follows: normoglycemia (34.6%), prediabetes (18.8%), and diabetes (46.6%). The median follow-up time was 63 months. During the follow-up, 744 all-cause deaths and 323 CVD deaths were recorded. Deceased patients were older, more likely to have stroke, had significantly higher CRP and hsCRP levels, and lower eGFR (all P < 0.05) (Table 1).

Table 1.

Baseline characteristics of patients with CAD and CKD

		Overall	Survival	Death	P
n		1353	609	744
Demographics
Age, y		77.00 [69.00, 80.00]	73.00 [66.00, 80.00]	79.00 [72.00, 80.00]	< 0.001
Female(%)		637(47.1)	315(51.7)	322(43.3)	0.002
Race (%)	Mexican American	106 (7.8)	46 (7.6)	60 (8.1)	< 0.001
	Non-Hispanic Black	257 (19.0)	136 (22.3)	121 (16.3)
	Non-Hispanic White	852 (63.0)	334 (54.8)	518 (69.6)
	Other Hispanic	71 (5.2)	49 (8.0)	22 (3.0)
	Other	67 (5.0)	44 (7.2)	23 (3.1)
Education (%)	Less than high school	475 (35.1)	184 (30.2)	291 (39.1)	< 0.001
	High school	341 (25.2)	151 (24.8)	190 (25.5)
	More than high school	537 (39.7)	274 (45.0)	263 (35.3)
Body mass index, kg/m2		28.73 [25.80, 33.07]	29.60 [26.40, 34.00]	28.20 [25.10, 32.15]	<0.001
Smoke (%)		159 (11.8)	86 (14.1)	73 (9.8)	0.018
Systolic blood pressure, mmHg		132.00 [118.00,149.00]	130.00 [117.00,146.00]	134.00 [118.75,152.00]	0.023
Diastolic blood pressure, mmHg		63.00 [55.00,72.00]	65.00 [57.00,74.00]	62.00 [54.75,71.00]	<0.001
Clinical presentation
Glycemic statuses (%)	Normoglycemia	468 (34.6)	208 (34.2)	260 (34.9)	0.729
	Prediabetes	254 (18.8)	120 (19.7)	134 (18.0)
	Diabetes	631 (46.6)	281 (46.1)	350 (47.0)
Stroke (%)		304 (22.5)	118 (19.4)	186 (25.0)	0.016
Hypertension (%)		1201 (88.8)	534 (87.7)	667 (89.7)	0.292
Hypercholesterolemia (%)		1250 (92.4)	598 (98.2)	652 (87.6)	<0.001
Cancer history (%)		354 (26.2)	154 (25.3)	0.547	0.547
Laboratory data
White blood cell count		7.20 [5.90, 8.60]	154 (25.3)	7.30 [6.00, 8.83]	0.027
Hemoglobin, g/dL		13.40 [12.30, 14.50]	154 (25.3)	13.30 [12.20, 14.43]	0.009
Platelet count, × 109/L		215.00 [176.00, 260.00]	211.00 [173.00, 256.00]	217.00 [178.00, 264.00]	0.089
Albumin, g/L		41.00 [39.00, 43.00]	41.00 [38.00, 43.00]	0.135	0.135
eGFR, mL/min per 1.73 m2		47.74 [37.03, 55.03]	49.21 [39.06, 55.45]	46.57 [35.33, 54.13]	<0.001
Creatinine, mg/dL		1.27 [1.08, 1.52]	1.22 [1.02, 1.41]	1.30 [1.10, 1.62]	<0.001
Cystatin, mg/L		1.48 [1.26, 1.90]	1.46 [1.21, 1.83]	1.53 [1.26, 1.95]	0.017
Glucose, mg/dL		104.00 [92.00, 127.00]	103.00 [92.00, 125.00]	104.00 [92.00, 129.00]	0.254
Calcium, mg/dL		9.40 [9.10, 9.70]	9.40 [9.10, 9.70]	9.40 [9.10, 9.70]	0.992
Sodium, mmol/L		140.00 [138.00, 141.00]	140.00 [138.00, 142.00]	139.65 [137.38, 141.00]	<0.001
Potassium, mmol/L		4.20 [3.90, 4.50]	4.20 [3.90, 4.40]	4.20 [4.00, 4.50]	0.002
HbA1c, %		5.90 [5.50, 6.60]	6.00 [5.60, 6.60]	5.90 [5.50, 6.70]	0.446
Triglycerides, mg/dL		140.00 [97.00, 196.00]	138.00 [99.00, 194.00]	141.00 [96.00, 198.25]	0.864
Total cholesterol, mg/dL		169.00 [145.00, 200.00]	167.00 [142.00, 199.00]	171.00 [147.75, 200.25]	0.032
HDL-C, mg/dL		47.00 [39.00, 59.00]	167.00 [142.00, 199.00]	47.00 [39.00, 58.00]	0.393
LDL-C, mg/dL		92.00 [72.00, 120.00]	90.00 [71.00, 119.00]	94.50 [72.00, 120.00]	0.221
C-reactive protein, mg/L		2.8 [1.3, 7.5]	2.5 [1.2,6.3]	3.0 [1.4, 8.3]	0.042
hsCRP, mg/L		2.34 [1.22, 5.90]	2.20 [1.20, 4.90]	2.50 [1.30, 6.40]	0.001
Medications
Antidiabetic medication (%)		899 (66.4)	384 (63.1)	515 (69.2)	0.020
Lipid-lowering medication (%)		1245 (92.0)	554 (91.0)	515 (69.2)	0.235
Antihypertensive medication (%)		1306 (96.5)	585 (96.1)	721 (96.9)	0.484

Continuous values are summarized as median (interquartile range), and categorical variables as n (%)

P values were calculated to compare baseline characteristics. Continuous variables were analysed using one-way analysis of variance (ANOVA) or the Kruskal–Wallis test, while categorical variables were compared using the chi-square or Fisher’s exact test when appropriate

eGFR Estimated glomerular filtration rate, HbA1c Haemoglobin A1c, HDL-C High-density lipoprotein cholesterol, LDL-C Low-density lipoprotein cholesterol, hsCRP High-sensitivity C-reactive protein, CKD Chronic kidney disease, CAD Coronary artery disease

Biomarkers and mortality

Higher levels of TyG-hsCRP, TyG-CRP, and CTI were associated with increased risk of all-cause mortality (Table 2). In fully adjusted models, the strongest association was observed with CTI (HR 1.43, 95% CI 1.24–1.65, P < 0.001), followed by TyG-CRP (HR 1.17, 95% CI 1.12–1.22, P < 0.001). When dichotomized at medians, elevated levels showed consistent risks (e.g., CTI > 8.45: HR 1.44, 95% CI 1.20–1.74, P < 0.001). Cumulative hazard curves demonstrated significantly higher cumulative incidence of all-cause death in groups with elevated composite indices compared to low levels (log-rank P < 0.05) (Fig. 2). RCS analyses revealed TyG-hsCRP had linear dose–response relationships for all-cause mortality (P for nonlinear > 0.05), while other biomarkers had non-linear dose-response relationships for all-cause mortality with accelerating risks at higher biomarker levels for composites and a U-shaped pattern for TyG (Fig. 3). ROC analyses indicated relatively better discriminatory performance for CTI (AUC 0.700) compared to TyG-hsCRP (AUC 0.698), TyG-CRP (AUC 0.695), and TyG (AUC 0.691) for all-cause mortality (Fig. 4).

Table 2.

Associations between various biomarkers and all-cause death

	Unadjusted model HR (95%CI)	P	Adjusted model HR (95%CI)	P
TyG	1.10(0.99–1.23)	0.096	1.14(1.00–1.34)	0.047
TyG ≤ 8.40	1(Ref.)		1(Ref.)
TyG > 8.40	1.05(1.01–1.23)	0.034	1.04(1.00–1.27)	0.029
TyG-hsCRP	1.01(1.01–1.02)	0.001	1.01(1.00–1.02)	0.002
TyG-hsCRP ≤ 11.48	1(Ref.)		1(Ref.)
TyG-hsCRP > 11.48	1.17(1.01–1.35)	0.035	1.18(1.02–1.39)	0.030
TyG-CRP	1.12(1.07–1.17)	0.001	1.17(1.12–1.22)	< 0.001
TyG-CRP ≤ 9.40	1(Ref.)		1(Ref.)
TyG-CRP > 9.40	1.17(1.01–1.35)	0.034	1.31(1.09–1.57)	0.004
CTI	1.16(1.07–1.26)	< 0.001	1.43(1.24–1.65)	< 0.001
CTI ≤ 8.45	1(Ref.)		1(Ref.)
CTI > 8.45	1.21(1.05–1.40)	0.008	1.44(1.20–1.74)	< 0.001

Adjusted model was fully adjusted for age, gender, BMI, systolic blood pressure, smoking status, diabetes status, hypertension, dyslipidemia, hemoglobin A1c, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, use of antidiabetic agents, antihypertensive medication and lipid-lowering medication

TyG Triglyceride-glucose index, hsCRP High-sensitivity C-reactive protein, CRP C-reactive protein, CTI C-reactive protein–triglyceride glucose index, HR Hazard ratio, CI Confidence interval

Fig. 2.

Cumulative hazard plot analysis for all-cause death stratified by a TyG-hsCRP, b TyG-CRP, c CTI and d TyG. TyG Triglyceride-glucose index, hsCRP High-sensitivity C-reactive protein, CRP C-reactive protein, CTI C-reactive protein–triglyceride glucose index

Fig. 3.

Restricted cubic spline analysis of the association between a TyG-hsCRP, b TyG-CRP, c CTI, d TyG and all-cause death in patients with CKD and CAD. TyG Triglyceride-glucose index, hsCRP High-sensitivity C-reactive protein, CRP C-reactive protein, CTI C-reactive protein–triglyceride glucose index, CKD Chronic kidney disease, CAD Coronary artery disease

Fig. 4.

Receiver operating characteristic (ROC) curves of different biomarkers for predicting all-cause death in patients with CKD and CAD: a TyG-hsCRP, b TyG-CRP, c CTI, d TyG. TyG Triglyceride-glucose index, hsCRP High-sensitivity C-reactive protein, CRP C-reactive Protein, CTI C-reactive protein–triglyceride glucose index, AUC Area under the Curve, CKD Chronic kidney disease, CAD Coronary artery disease

Similar patterns were observed for CVD mortality, with composite indices showing robust associations in adjusted models (e.g., TyG-hsCRP HR 1.01, 95% CI 1.00–1.02, P = 0.016; TyG-CRP HR 1.16, 95% CI 1.12–1.22, P < 0.001; CTI HR 1.32, 95% CI 1.06–1.64, P = 0.012; Supplementary Table 2). Cumulative hazard curves for CVD death also showed higher cumulative hazards in elevated biomarker groups (log-rank P < 0.05 for composites) (Supplementary Fig. 2). RCS for CVD mortality mirrored those for all-cause, with linear increases for TyG-hsCRP and non-linear increases for other composites (Supplementary Fig. 3). ROC curves for CVD mortality confirmed better performance of composites (CTI AUC 0.640, TyG-hsCRP AUC 0.641, TyG-CRP AUC 0.638) compared with TyG (AUC 0.637) (Supplementary Fig. 4).

Reclassification analyses showed that adding biomarkers to the basic model improved prediction: for all-cause death, CTI had the highest IDI (0.010, 95% CI 0.006–0.013, P < 0.001) and NRI (0.196, 95% CI 0.106–0.284, P < 0.001). For CVD death, TyG-CRP had the highest IDI (0.015, 95% CI 0.011–0.019, P < 0.001) and NRI (0.452, 95% CI 0.301–0.583, P < 0.001), followed by CTI (IDI 0.013, NRI 0.386), TyG (IDI 0.011, NRI 0.362) (Supplementary Table 3).

Associations across different glycemic status

The prognostic value of the biomarkers for all-cause mortality varied across glycemic statuses (Table 3). In normoglycemia, adjusted associations were significant for TyG-CRP (HR 1.14, 95% CI 1.06–1.23, P = 0.001) and CTI (HR 1.50, 95% CI 1.16–1.94, P = 0.002). In prediabetes, all composites were predictive (P < 0.05). In diabetes, associations held for TyG-hsCRP (HR 1.01, 95% CI 1.00–1.03, P = 0.026), TyG-CRP (HR 1.18, 95% CI 1.10–1.26, P < 0.001), and CTI (HR 1.33, 95% CI 1.08–1.66, P = 0.008). Dichotomized analyses showed amplified risks in dysglycemic groups (e.g., CTI > 8.45: normoglycemia HR 1.92, P < 0.001; prediabetes HR 1.73, P = 0.001; diabetes HR 1.94, P = 0.031). RCS stratified by glycemic status illustrated linear associations between TyG-hsCRP and all-cause death in patients with diabetes and non-linear associations between other biomarkers and all-cause death across glycemic status (Supplementary Fig. 5, 6, 7).

Table 3.

Associations between various biomarkers and all-cause death across different glycemic statuses

	Unadjusted model HR (95%CI)	P	Adjusted model HR (95%CI)	P
Normoglycemia
TyG	1.09(0.85–1.39)	0.496	1.04(0.64–2.09)	0.966
TyG ≤ 8.40	1(Ref.)		1(Ref.)
TyG > 8.40	1.09(0.85–1.41)	0.508	1.20(0.80–1.80)	0.379
TyG-hsCRP	1.01(1.00–1.02)	0.026	1.01(1.00–1.02)	0.049
TyG-hsCRP ≤ 11.48	1(Ref.)		1(Ref.)
TyG-hsCRP > 11.48	1.09(0.87–1.35)	0.469	1.06(0.84–1.34)	0.603
TyG-CRP	1.07(1.01–1.19)	0.033	1.14(1.06–1.23)	0.001
TyG-CRP ≤ 9.40	1(Ref.)		1(Ref.)
TyG-CRP > 9.40	1.06(0.84–1.31)	0.625	1.15(1.02–1.32)	0.03
CTI	1.13(1.04–1.34)	0.024	1.50(1.16–1.94)	0.002
CTI ≤ 8.45	1(Ref.)		1(Ref.)
CTI > 8.45	1.28(1.00–1.64)	0.047	1.92(1.38–2.68)	0.001
Prediabetes
TyG	1.00(0.96–1.05)	0.474	1.00(0.99–1.05)	0.833
TyG ≤ 8.40	1(Ref.)		1(Ref.)
TyG > 8.40	1.01(0.65–1.29)	0.601	1.03(0.96–1.09)	0.158
TyG-hsCRP	1.02(1.00–1.03)	0.05	1.02(1.00–1.03)	0.042
TyG-hsCRP ≤ 11.48	1(Ref.)		1(Ref.)
TyG-hsCRP > 11.48	1.24(0.88–1.75)	0.218	1.18(1.02–1.42)	0.013
TyG-CRP	1.30(1.15–1.47)	0.001	1.28(1.14–1.44)	0.001
TyG-CRP ≤ 9.40	1(Ref.)		1(Ref.)
TyG-CRP > 9.40	1.13(0.80–1.60)	0.483	1.28(0.84–1.97)	0.252
CTI	1.27(1.00–1.63)	0.045	1.70(1.19–2.42)	0.003
CTI ≤ 8.45	1(Ref.)		1(Ref.)
CTI > 8.45	1.29(1.01–1.83)	0.014	1.73(1.14–2.62)	0.001
Diabetes
TyG	1.05(0.91–1.23)	0.459	1.04(0.73–1.13)	0.318
TyG ≤ 8.40	1(Ref.)		1(Ref.)
TyG > 8.40	1.02(0.77–1.21)	0.787	1.01(0.57–1.19)	0.454
TyG-hsCRP	1.02(1.00–1.04)	0.012	1.01(1.00–1.03)	0.026
TyG-hsCRP ≤ 11.48	1(Ref.)		1(Ref.)
TyG-hsCRP > 11.48	1.10(0.86–1.41)	0.44	1.28(0.99–1.67)	0.06
TyG-CRP	1.12(1.04–1.21)	0.003	1.18(1.10–1.26)	0.001
TyG-CRP ≤ 9.40	1(Ref.)		1(Ref.)
TyG-CRP > 9.40	1.18(0.92–1.51)	0.202	1.70(1.23–2.37)	0.001
CTI	1.10(1.00–1.24)	0.018	1.33(1.08–1.66)	0.008
CTI ≤ 8.45	1(Ref.)		1(Ref.)
CTI > 8.45	1.24(1.07–1.42)	0.034	1.94(1.67–2.12)	0.031

Adjusted model was fully adjusted for age, gender, BMI, systolic blood pressure, smoking status, hypertension, dyslipidemia, hemoglobin A1c, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, use of antidiabetic agents, antihypertensive medication and lipid-lowering medication

TyG Triglyceride-glucose index, hsCRP High-sensitivity C-reactive protein, CRP C-reactive protein, CTI C-reactive protein–triglyceride glucose index, HR Hazard ratio, CI Confidence interval

For CVD mortality, associations were attenuated but followed similar trends (Supplementary Table 4). In normoglycemia, only higher CTI (> 8.45) was significant (adjusted HR 1.89, 95% CI 1.11–33.23, P = 0.020). In prediabetes, TyG-CRP (HR 1.38, 95% CI 1.17–1.64, P < 0.001) predicted risk with statistically significant. In diabetes, TyG-CRP and CTI were associated (TyG-CRP HR 1.16, 95% CI 1.05–1.28, P = 0.002; CTI HR 1.42, 95% CI 1.03–1.95, P = 0.031). Stratified RCS for CVD mortality showed non-linear associations between in biomarkers and CVD death across different glycemic status (Supplementary Figs. 8, 9, 10).

Joint biomarkers and glycemic status interactions

Using combined stratification based on biomarker levels and glycemic status, patients were divided into six groups (Fig. 5). The highest risk of all-cause death was observed in patients with high CTI and diabetes. Specifically, patients with CTI > 8.45 and diabetes had a markedly higher risk (HR 1.63, 95%CI 1.22–2.17, P = 0.001) compared to the reference group (CTI ≤ 8.45 and normoglycemia). The risk increased progressively with worsening glycemic status (P for trend < 0.001). This pattern was confirmed in joint analyses using TyG-hsCRP and TyG-CRP with glycemic status. Similar joint effects were noted for CVD mortality, with the highest risks in high-biomarker dysglycemic groups (Supplementary Fig. 11).

Fig. 5.

Forest plots of Cox regression analysis for the association between TyG-hsCRP, TyG-CRP, CTI, TyG and all-cause death in patients with CKD and CAD across different glycemic statuses in unadjusted models and adjusted models. Adjusted model was fully adjusted for age, gender, BMI, systolic blood pressure, smoking status, hypertension, dyslipidemia, hemoglobin A1c, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, use of antidiabetic agents, antihypertensive medication and lipid-lowering medication. TyG Triglyceride-glucose index, hsCRP High-sensitivity C-reactive protein, CRP C-reactive protein, CTI C-reactive protein–triglyceride glucose index, HR Hazard ratio, CI Confidence interval, CKD Chronic kidney disease, CAD Coronary artery disease

Subgroup analysis

Subgroup analyses showed generally consistent associations across age, sex, BMI, smoking, hypertension, and hypercholesterolemia strata (Table 4, Supplementary Table 5, 6). Significant interactions were observed between TyG-CRP/glycemic status and age (P for interaction = 0.021), with stronger associations in participants aged ≥ 65 years. Borderline interactions were also noted between TyG-hsCRP and BMI (P = 0.047).

Table 4.

Results of subgroup analysis stratified by CTI and glycemic status

CTI	Normoglycemia		Prediabetes		Diabetes		P for interaction
	CTI ≤ 8.45	CTI > 8.45	CTI ≤ 8.45	CTI > 8.45	CTI ≤ 8.45	CTI > 8.45
Female	1(Ref.)	1.01(0.70–1.46)	1.07(0.61–1.55)	1.33(0.99–1.80)	1.18(0.82–1.72)	1.41(0.86–2.30)	0.298
Male	1(Ref.)	1.27(0.85–1.87)	0.98(0.70–1.38)	1.45(1.09–1.93–2.55)	1.39(1.03–1.87)	1.76(1.26–2.47)
BMI < 30	1(Ref.)	1.18(0.88–1.60)	1.08(0.80–1.46)	1.38(0.97–1.97)	1.36(1.02–1.81)	1.40(1.07–1.84)	0.176
BMI ≥ 30	1(Ref.)	1.65(1.03–3.12)	1.03(0.55–1.92)	1.62(0.86–3.05)	1.76(1.07–2.87)	1.89(1.20–2.96)
Age < 65	1(Ref.)	0.82(0.26–2.60)	0.96(0.19–4.96)	1.63(0.39–6.88)	1.62(0.51–5.10)	2.46(0.97–6.23)	0.973
Age ≥ 65	1(Ref.)	1.36(1.00–1.86)	1.00(0.76–1.31)	1.48(1.19–1.84)	1.34(1.06–1.70)	1.59(1.23–2.05)
Smoking	1(Ref.)	2.31(0.83–6.46)	1.08(0.28–4.18)	2.72(1.07–6.89)	2.80(1.18–6.61)	2.95(1.15–7.62)	0.655
No smoking	1(Ref.)	0.87(0.48–1.59)	1.05(0.70–1.18)	1.08(0.58–2.00)	1.22(0.81–1.85)	1.36(0.87–2.13)
Hypertension	1(Ref.)	1.21(0.93–1.57)	1.01(0.73–1.30)	1.35(0.98–1.86)	1.17(0.91–1.50)	1.37(1.02–1.68)	0.997
No hypertension	1(Ref.)	1.23(0.54–2.82)	1.07(0.36–3.14)	1.61(0.67–3.85)	2.18(1.18–4.06)	3.34(1.73–6.47)
Hypercholesterolemia	1(Ref.)	1.27(0.96–1.66)	1.08(0.81–1.45)	1.42(1.02–1.96)	1.36(1.04–1.74)	1.49(1.21–1.74)	0.291
No hypercholesterolemia	1(Ref.)	1.14(0.35–3.73)	0.66(0.28–1.57)	1.51(0.82–2.77)	1.27(0.70–2.25)	1.65(0.91–2.99)

Adjusted model was fully adjusted for age, gender, BMI, systolic blood pressure, smoking status, hypertension, dyslipidemia, hemoglobin A1c, estimated glomerular filtration rate, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, use of antidiabetic agents, antihypertensive medication and lipid-lowering medication. (When a variable is used as a research variable in the subgroup analysis, that variable will be excluded during adjustment.)

CTI C-reactive protein–triglyceride glucose index, BMI Body mass index

Sensitivity analysis

Sensitivity analyses excluding TyG- and glucose-related variables (diabetes status, triglyceride and HbA1c) yielded results consistent with the primary analysis. The associations between composite indices and both all-cause (e.g., CTI HR 1.29, 95% CI 1.18–1.43, P < 0.001) and CVD (e.g., TyG-CRP HR 1.15, 95% CI 1.08–1.23, P < 0.001) mortality remained statistically significant and similar in magnitude (Supplementary Table 7, 8). Stratified sensitivities by glycemic status yielded similar results, with preserved significance in dysglycemia (Supplementary Table 9).

Discussion

In this population-based cohort study of 1353 patients with CKD and CAD, composite indices integrating IR and inflammation—namely TyG-hsCRP, TyG-CRP, and CTI—demonstrated good prognostic value for all-cause and CVD mortality. Elevated levels of these indices were independently associated with increased mortality risk, with CTI exhibiting the strongest adjusted HR. Stratified analyses revealed amplified risks in prediabetes and diabetes subgroups, where joint effects with dysglycemia conferred the highest mortality hazards. Reclassification metrics confirmed enhanced predictive performance, with CTI yielding the highest integrated discrimination improvement and net reclassification improvement for all-cause mortality. Risk stratification integrating the composite biomarkers with glycemic status successfully identified patient subgroups at the highest all-cause and CVD death risk. Sensitivity analyses, including exclusion of glucose-related variables, affirmed the robustness of these associations across glycemic statuses.

IR plays a central role in the pathogenesis of CAD [21]. Its detrimental effects extend beyond the promotion of traditional risk factors to directly drive the process of coronary artery disease and kidney diseases [8, 22]. IR can promote active vascular calcification through multiple pathways, including the induction of endothelial dysfunction, activation of the renin-angiotensin system, and the promotion of an osteogenic phenotypic switch in vascular smooth muscle cells [23, 24]. Consequently, IR is not merely a causative factor in CKD and CAD but also a pivotal biological marker for assessing its activity and prognosis.

Operating in parallel with IR, chronic inflammatory states constitute another cornerstone throughout the continuum of atherosclerosis [12]. In advanced lesions, persistent inflammation contributes to plaque destabilization by weakening the fibrous cap via actions of substances like matrix metalloproteinases, while simultaneously exacerbating a pro-calcific microenvironment within the intima [25, 26]. For patients with CKD, a high inflammatory burden often heralds more rapid plaque progression, a greater risk of microvascular dysfunction, and more frequent cardiovascular events, which explains why inflammatory markers carry independent prognostic value in this high-risk population [10, 11, 27, 28].

The novel composite biomarkers examined in this study—TyG-hsCRP, TyG-CRP, and the CTI—offer the distinct advantage of concurrently capturing the dual pathological processes of metabolic dysregulation and inflammatory activation. Our finding that these composite biomarkers demonstrated good predictive performance aligns with the pathophysiological concept that IR and inflammation synergistically contribute to cardiovascular risk. While previous studies have established the prognostic value of either the TyG index or inflammatory markers alone in CAD populations, few have specifically examined their combination in CKD and CAD patients. The CTI is not a simple linear addition of the TyG index and CRP; instead, it optimizes their combined contribution to risk by assigning a specific weight to CRP in its algorithm [29–32]. This mathematical refinement likely underlies its marginally superior predictive performance in our cohort. Furthermore, the significant associations observed for TyG-hsCRP suggest that the core “metabolism-inflammation” axis is a robust predictor, regardless of the specific inflammatory marker used, though the strength of association may vary. Our findings provide new clinical evidence supporting the application of these composite biomarkers in the CKD and CAD population. Their potential mechanism for enhancing prognostic prediction lies in their ability to provide an integrated assessment of this core axis, thereby yielding pathophysiological information beyond that offered by traditional risk factors.

In this study, we compared and validated various composite indices of IR and inflammation (TyG-hsCRP, TyG-CRP, and CTI). It is important to note that their underlying algorithms differ; for instance, CTI incorporates logarithmic transformations, whereas TyG-CRP is a simple summation. The primary objective of this research was to validate the predictive value of these novel indices in patients with CKD and CAD, rather than to develop new biomarkers, which justified this comparative approach.

The primary clinical significance of our study lies in providing a more precise risk stratification tool for patients with CKD and CAD—a complex CAD subgroup characterized by high procedural risk and poor long-term outcomes. The heterogeneous associations across glycemic status offer refined insights for personalized management. The good prognostic value of CTI and TyG-CRP in the non-diabetic subgroup may be explained by the fact that in these patients, death risk is less dominated by severe hyperglycemia, allowing the more subtle interplay between IR and inflammation to emerge as a dominant risk driver. In contrast, in established diabetes, the metabolic milieu is so profoundly disturbed that multiple pathways, reflected by various composite indices, contribute more uniformly to adverse outcomes [33, 34]. By utilizing readily available composite indices like the CTI in conjunction with a patient’s glycemic status, clinicians can readily identify those “very high-risk” individuals within an already high-risk population. For these patients, more aggressive management strategies—including intensive lipid-lowering, stringent glycemic and inflammation control, and closer long-term follow-up—can be implemented.

In the subgroup analysis, a notable interaction was observed between TyG-CRP/glycemic status and age. Although variations in CRP levels across age groups may partially explain this interaction, the findings in our study are more likely attributable to reduced statistical power due to insufficient sample sizes in certain subgroups.

Several limitations of our study warrant careful consideration when interpreting the results. First, the observational design may introduce selection bias and limits generalizability to other ethnic populations. Future multi-ethnic studies are essential for validation and should also aim to establish clinically applicable cut-off values for these composite indices to guide decision-making. Second, using WHO/IEC criteria for prediabetes resulted in a smaller cohort (18.8%) compared to ADA standards, possibly underrepresenting this group and attenuating subgroup associations. Additionally, inflammatory markers such as hsCRP and CRP may be influenced by acute infections, intercurrent illnesses, or transient factors, potentially affecting biomarker interpretation. However, in the NHANES cohort, blood samples were collected under non-emergency conditions, which may mitigate such interference to some extent. Furthermore, for CKD patients, the etiology of CKD, staging based on eGFR and proteinuria levels, and the use of medications such as SGLT2 inhibitors or ACEI/ARBs—which can improve prognosis in this population—were not analyzed due to database limitations; we anticipate further exploration in future clinical studies. Finally, residual confounding from unmeasured factors, such as dialysis status or the severity of coronary lesions, cannot be entirely excluded despite extensive adjustment.

Conclusion

In conclusion, this study demonstrates that composite biomarkers integrating insulin resistance and systemic inflammation, particularly the CTI, provide significant prognostic value for mortality in patients with CKD and CAD. The predictive utility of these biomarkers varies across glycemic status. The integration of these biomarkers with glycemic status enables enhanced risk stratification, potentially guiding more personalized management strategies for this high-risk population.

Abbreviations

AUC

Area under the curve

BMI

Body mass index

CABG

Coronary artery bypass grafting

CAD

Coronary artery disease

CI

Confidence interval

CKD

Chronic kidney disease

CRP

C-reactive protein

CTI

C-reactive protein–triglyceride glucose index

DBP

Diastolic blood pressure

eGFR

Estimated glomerular filtration rate

FBG

Fasting blood glucose

FPG

Fasting plasma glucose

HbA1c

Hemoglobin A1c

HDL-C

High-density lipoprotein cholesterol

HR

Hazard ratio

hsCRP

High-sensitivity C-reactive protein

IR

Insulin resistance

LDL-C

Low-density lipoprotein cholesterol

PCI

Percutaneous coronary intervention

ROC

Receiver operating characteristic

SBP

Systolic blood pressure

TC

Total cholesterol

TyG

Triglyceride-glucose index

Author contributions

Z.Y and K.D contributed to the study design. Z.Y and E.X contributed to data collection, manuscript writing, data processing, and figure mapping. C.S contributed to the data proofreading. Z.Y. and R.Z contributed to formal analysis; writing—original draft preparation; C.W and H.W contributed to review and to edit. All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by research grants from the Noncommunicable Chronic Diseases-National Science and Technology Major Project (Grant number: 2025ZD0548200), Beijing Natural Science Foundation (Grant number:L246034), Noncommunicable Chronic Diseases-National Science and Technology Major Project (Grant number: 2023ZD0513900), CAMS Innovation Fund for Medical Sciences (CIFMS, Grant number: 2025-I2M-XHXX-032), Fundamental Research Funds for the Central Universities (Grant number: 3332024033), and Beijing Physician Scientist Training Project (Grant number: BJPSTP-2025-09).

Data availability

The data are publicly available in DataDryad (10.5061/dryad.d5h62).

Declarations

Ethical approval and consent to participate

The research adhered to the ethical principles outlined in the Declaration of Helsinki and received approval from the CDC’s institutional review board.

Competing interest

The authors declare no competing interests.