Synergistic effect of AIP and cystatin C: a study on cardiovascular risk prediction in populations with different glycemic statuses

Kuangyi Wang; Yufeng Jiang; Liangping Zhao; Yafeng Zhou

doi:10.1186/s13098-026-02231-y

. 2026 Jul 17;18:215. doi: 10.1186/s13098-026-02231-y

Synergistic effect of AIP and cystatin C: a study on cardiovascular risk prediction in populations with different glycemic statuses

Kuangyi Wang ¹, Yufeng Jiang ¹, Liangping Zhao ¹, Yafeng Zhou ^1,^✉

PMCID: PMC13656459 PMID: 42464312

Abstract

Background

Cardiovascular disease (CVD) poses a major global health burden. The Atherogenic Index of Plasma (AIP) and Cystatin C are novel biomarkers reflecting lipid metabolism disorders and renal/micro-inflammatory status, respectively. Their combined indicator (AIP-Cys, the product of AIP × Cystatin C) may provide improved predictive value beyond either marker alone. This study aimed to investigate the predictive role of AIP-Cys for the risk of incident CVD across different glycemic statuses: normal glucose regulation (NGR), pre-diabetes (preDM), and diabetes mellitus (DM).

Materials and methods

This study was based on the prospective China Health and Retirement Longitudinal Study (CHARLS) cohort. A total of 6,035 participants aged ≥ 45 years without a history of CVD at baseline (2011) were included. The primary exposure was baseline AIP-Cys (the product of AIP × Cystatin C), and the outcome was the first self-reported incident heart disease or stroke event during follow-up (until 2020). Glycemic status was defined based on fasting plasma glucose and glycated hemoglobin levels. Multivariable Cox proportional hazards models were used to estimate hazard ratios (HRs). Restricted cubic splines and piecewise linear models were applied to analyze nonlinear relationships. Mediation and sensitivity analyses were conducted to verify the robustness of the results.

Results

During a median follow-up of 9.0 years, 651 incident CVD cases occurred. In the fully adjusted model, AIP-Cys was independently associated with CVD risk in the overall (HR per unit increase: 1.34; 95% CI: 1.06–1.69), NGR (HR: 1.60; 95% CI: 1.13–2.26), and preDM (HR: 1.32; 95% CI: 1.02–1.70) populations, but not in the DM population. RCS analysis revealed a significant nonlinear relationship in the NGR and preDM groups (P for nonlinearity < 0.05). Piecewise regression identified inflection points (0.47 for NGR; 0.37 for preDM), below which risk increased sharply. Quartile analysis showed the highest risk in Q3 rather than Q4, suggesting a saturation effect. Mediation analysis indicated that systolic blood pressure partially mediated this association (mediating proportion: 14.6% in overall population). The predictive performance of AIP-Cys was modest for short-term risk (1-year AUC: 0.64–0.67) but limited for long-term prediction.

Conclusions

The composite indicator AIP-Cys, derived from AIP and Cystatin C, is an independent risk factor for incident CVD in individuals with NGR and preDM. The association exhibits a nonlinear threshold pattern. However, its predictive value is limited in the DM population, and its long-term predictive performance is modest. Additionally, the self-reported nature of cardiovascular outcomes represents a limitation that should be considered when interpreting these findings. By integrating lipid metabolism and renal/inflammatory pathways, this indicator offers a novel biomarker-based approach for early cardiovascular risk identification in the pre-diabetes stage. Future research should further explore its potential for clinical translation and intervention.

Graphical abstract

graphic file with name 13098_2026_2231_Figa_HTML.webp

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13098-026-02231-y.

Keywords: Cardiovascular disease, Atherogenic index of plasma, Cystatin C, Glucose metabolic status, CHARLS

Background

Cardiovascular diseases (CVD) pose a major challenge in the field of global public health and have long been a leading cause of disability and mortality worldwide [1, 2]. Over the past half-century, accelerated global population aging, lifestyle changes driven by urbanization, and the Westernization of dietary patterns have contributed to a continuous rise in the prevalence and mortality of CVD, imposing a substantial socioeconomic burden on healthcare systems across nations [3]. Although recent breakthroughs in cardiovascular diagnostics and treatment, alongside improved control of risk factors and evidence-based pharmacological therapies, have driven a decline in age-adjusted CVD mortality in some developed regions, the absolute number of CVD cases continues to increase globally, with a notable trend toward earlier onset [4, 5]. This paradoxical observation suggests that traditional intervention strategies based on typical clinical symptoms are no longer sufficient to address the current CVD epidemic. Moving forward, the focus of CVD prevention and management should shift toward earlier risk identification and more precise preventive interventions. Particularly against the backdrop of increasingly prominent metabolic risk factors such as diabetes and obesity, there is an urgent need to develop novel risk assessment systems in this field.

Abnormal glucose metabolism serves as an independent risk factor for the development and progression of CVD, and its relationship with cardiovascular health exhibits a continuous spectrum. Numerous prospective cohort studies and meta-analyses have confirmed that the risk of CVD shows a graded increase along the progression from normal glucose regulation (NGR) to pre-diabetes (preDM) and further to diabetes mellitus (DM) [6–8]. The preDM stage not only indicates a significantly elevated risk of progressing to DM but is also closely associated with subclinical atherosclerosis, endothelial dysfunction, and microvascular disease, leading to a higher risk of cardiovascular events compared to individuals with NGR [6, 9–12]. When advancing to the DM stage, patients face a further elevated risk of major cardiovascular endpoints, including coronary heart disease, ischemic stroke, and heart failure. Moreover, CVD has become the leading cause of mortality among individuals with DM [7, 8, 11, 13, 14]. This exposure–response relationship underscores the necessity for precise cardiovascular risk assessment across different stages of glucose metabolism. Specifically, it necessitates a stratified evaluation to determine whether a novel biomarker’s predictive value is uniform or varies across this metabolic continuum. Such stratification is crucial for its precise clinical application, distinguishing universal utility from specific value in high-risk transitional states like preDM.

In the field of cardiovascular risk markers, the Atherogenic Index of Plasma (AIP) and Cystatin C have emerged widespread attention due to their unique pathophysiological significance. AIP was first proposed by Dobiásová et al. and is calculated as the logarithm of the triglyceride to high-density lipoprotein cholesterol ratio [log₁₀(TG/HDL-C)]. This index provides a more accurate reflection of the atherogenic lipoprotein phenotype, particularly the abundance of small dense low-density lipoprotein (sdLDL) [15–17]. Due to its small size, susceptibility to oxidation, and high affinity for arterial wall proteoglycans, sdLDL exhibits stronger atherogenic potential [18, 19]. Clinical studies have shown that for every one-unit increase in AIP, the risk of major adverse cardiovascular events approximately doubles, and its predictive value is significantly superior to that of conventional lipid parameters [20–24]. On the other hand, Cystatin C, a member of the cysteine protease inhibitor superfamily, is produced at a constant rate by nucleated cells, and its serum concentration is not significantly influenced by factors such as muscle mass, age, or gender, making it a sensitive indicator for assessing glomerular filtration function. Recent studies have found that elevated levels of Cystatin C not only reflect early renal impairment but also directly participate in the formation and destabilization of atherosclerotic plaques through mechanisms such as promoting vascular wall elastin degradation, activating matrix metalloproteinases, and inducing the release of inflammatory factors [25, 26]. Therefore, high levels of Cystatin C are not only a marker of renal dysfunction but also an important predictor of increased subclinical atherosclerotic burden, progression toward more complex plaque morphology, and elevated future risk of cardiovascular events such as myocardial infarction and stroke [27–29].

Given the complementary roles of AIP and Cystatin C in the pathophysiological process of CVD, where AIP mainly reflects the atherogenic potential of lipid metabolism disorders while Cystatin C indicates subclinical renal impairment and systemic inflammatory states, their combined application may be associated with enhanced predictive value compared to either marker alone. From a pathophysiological perspective, the coexistence of abnormal lipid metabolism and a chronic low-grade inflammatory state may accelerate vascular endothelial dysfunction and the progression of atherosclerosis through multiple "hit" mechanisms [30–33]. This "multi-hit" hypothesis suggests that when two distinct but interconnected pathological pathways converge, their combined effect on the cardiovascular system could be conceptualized as multiplicative rather than merely additive [34–36]. In this context, individuals with both high AIP and high Cystatin C levels may experience more complex and compounded impacts, potentially leading to a disproportionately increased risk of cardiovascular events. The product term (AIP × Cystatin C, AIP-Cys) serves as a mathematical representation of this potential synergistic interaction, as it amplifies the value when both components are elevated, whereas an additive combination would not reflect such amplification. However, despite their established independent roles, whether their combined effect, particularly when quantified as a product (AIP × Cystatin C, AIP-Cys) reflecting potential synergistic risk, can provide superior predictive value across different glucose metabolism statuses remains unclear. Large-scale prospective evidence is lacking. This combined assessment strategy holds promise for overcoming the limitations of traditional risk evaluation models, particularly in identifying special phenotypic subgroups within the continuous spectrum of cardiovascular risk associated with different glucose metabolism states, such as individuals who are "metabolically healthy but at high cardiovascular risk" or those with "metabolic abnormalities whose risk is underestimated".

To test this scientific hypothesis, this study utilized the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative large-scale cohort database. Leveraging the comprehensive information available in this database, the study will adopt a prospective cohort design to systematically investigate the predictive efficacy of the combined AIP and Cystatin C indicator (AIP-Cys) for CVD incidence across populations with three distinct glucose metabolism statuses: NGR, preDM, and DM. Specifically, we aim to prospectively examine whether AIP-Cys, as a composite marker, outperforms either AIP or Cystatin C alone in predicting CVD risk, and to explore whether its association exhibits nonlinear characteristics. The findings are expected to provide important epidemiological evidence for establishing a precision cardiovascular risk stratification system based on multidimensional biomarkers, thereby guiding the development of individualized prevention strategies tailored to populations with different glucose metabolism statuses.

Materials and methods

Study design and population

The China Health and Retirement Longitudinal Study (CHARLS) is a nationally representative large-scale social survey project led by the National School of Development at Peking University. It aims to systematically collect high-quality microdata from Chinese households and individuals aged 45 and above, comprehensively tracking the dynamics of population aging and health-economic conditions. Initiated in 2011 with a national baseline survey, the project covers 17,000 households across 28 provinces (autonomous regions, municipalities) and employs a multistage stratified probability sampling method along with electronic mapping technology to ensure sample representativeness. Follow-up surveys are conducted every two years, with five rounds completed by 2020. The data encompass multiple dimensions, including respondents’ health status, healthcare utilization, pension insurance, employment income, household assets, intergenerational support, and innovatively incorporate biomarker information such as physical measurements and blood tests. The survey data are publicly available free of charge (https://charls.pku.edu.cn/). This study has been approved by the Ethics Committee of Peking University (IRB00001052-11015), and written informed consent was obtained from all participants [37].

A standardized data collection protocol was employed in this study, with systematically trained professionals conducting anthropometric measurements and venous blood sample collection. Comprehensive demographic and health-related variables were recorded, including age, sex, height, weight, waist circumference, marital status, residential area, educational background, smoking and alcohol consumption behaviors, as well as self-reported or clinically diagnosed chronic disease histories such as hypertension, diabetes, heart disease, and stroke. Weight was measured using a precision electronic scale with participants in a shoeless state, recorded to an accuracy of 0.1 kg. Height was measured using a vertical stadiometer, accurate to 0.1 cm. Waist circumference was uniformly measured at the umbilical level using a soft tape, recorded to 0.1 cm. Blood pressure was measured using an Omron HEM-7200 automatic blood pressure monitor, with the average of three consecutive readings taken for each participant. All participants underwent fasting venous blood collection performed by clinical healthcare professionals. Collected samples were immediately frozen and stored at -20℃, then transported via a professional cold-chain logistics system to the Chinese Center for Disease Control and Prevention (Beijing) for laboratory testing. Blood biochemical analyses included: serum triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and fasting plasma glucose (FPG) measured using enzymatic colorimetric methods; creatinine measured using the rate blank and compensated Jaffe methods, with units in μmol/L; and high-sensitivity C-reactive protein (hs-CRP) quantified via immunoturbidimetry using a Hitachi 7180 automatic biochemical analyzer. All testing procedures adhered to strict quality control standards, with both intra-batch and inter-batch coefficients of variation (CV) for all blood biomarker assays maintained below 5%, ensuring high accuracy and reproducibility of the experimental data [37, 38].

In this study, the baseline cohort was defined as individuals who participated in the first survey round in 2011 and completed follow-ups in 2013, 2015, 2018, and 2020. A total of 17,708 participants were included in the 2011 baseline survey. Based on the following exclusion criteria, we further screened the study subjects (Fig. 1): (1) age under 45 years (N = 244); (2) missing data on diabetes history, or missing any blood test data for fasting plasma glucose, glycated hemoglobin, triglycerides, high-density lipoprotein, or cystatin C, or incomplete baseline sociodemographic characteristics, health-related indicators, anthropometric data, or other biomarker information (N = 8,977); (3) pre-existing history of stroke or heart disease at study initiation, incomplete relevant medical history data, or missing follow-up data on stroke and heart disease (N = 2,942). Following these exclusions, a total of 6,035 participants were ultimately included. Given that abnormal glucose metabolism represents a continuous risk spectrum for CVD, and the dominant risk drivers may differ between normoglycemic, prediabetic, and diabetic states, we stratified all analyses by glycemic status (NGR, preDM, DM) to evaluate the predictive value and applicability of our exposure variable across these distinct metabolic stages.

Fig. 1 — Study participants flow diagram

Definition of exposure and outcome variables

This study considered the AIP-Cys as the primary exposure variable. AIP was calculated using the formula: AIP = Log₁₀(triglycerides / high-density lipoprotein cholesterol) [17, 39]. AIP-Cys was defined as the product of AIP × Cystatin C. The choice of a multiplicative, rather than additive, combination was driven by our a priori hypothesis of a potential synergistic biological interaction between the lipid metabolism disorder (reflected by AIP) and the renal/inflammatory pathway (reflected by Cystatin C). From a pathophysiological perspective, the “multi-hit” hypothesis suggests that when two distinct but interconnected pathological pathways converge, their combined effect on the cardiovascular system may be multiplicative rather than merely additive [34–36]. The product term mathematically captures this potential synergy, as it amplifies the value when both components are elevated, whereas a simple additive combination would not reflect such amplification. To statistically justify the construction of AIP-Cys as a multiplicative term, we formally tested for multiplicative interaction between AIP and Cystatin C. In the fully adjusted Cox proportional hazards model (Model IV) for the overall population, we included an interaction term (AIP * Cystatin C) alongside the main effects of AIP and Cystatin C. A statistically significant coefficient for this interaction term would indicate that the joint effect of AIP and Cystatin C is greater than the sum of their individual effects on the multiplicative scale, thereby providing empirical support for their combined use as a product. The outcome variable was cardiovascular disease, defined as the first self-reported history of heart disease and/or stroke. The identification of cardiovascular disease was based on responses to the following questions: "Have you ever been diagnosed by a doctor with a heart attack, coronary heart disease, angina, congestive heart failure, or other heart diseases? Or have you ever been diagnosed by a doctor with a stroke?" This definition is consistent with previous studies based on CHARLS data [40, 41]. It is important to acknowledge that relying on self-reported physician diagnoses may introduce misclassification bias, as it depends on participants’ recall and access to healthcare. All participants were followed from baseline, with a total of four follow-up visits completed. The follow-up period ended at the occurrence of a cardiovascular event or the conclusion of the follow-up period in 2020, whichever came first. Additionally, we explored the relationship between cumulative AIP-Cys (cuAIP-Cys) and the risk of CVD onset. cuAIP-Cys was calculated using a linear model with the following formula: cuAIP-Cys = [(AIP-Cys₂₀₁₂ + AIP-Cys₂₀₁₅) / 2] * time interval (2015–2012). Here, AIP-Cys₂₀₁₂ and AIP-Cys₂₀₁₅ represent the AIP-Cys values measured in 2012 (first wave) and 2015 (third wave), respectively, with a measurement interval of three years. The average of these two time points was calculated to reflect the mean exposure level, which was then multiplied by the duration to estimate long-term cumulative exposure. This method follows the approach used for calculating cumulative metabolic indices in the CHARLS cohort [42].

Definition of glycemic status

According to the consensus of the World Health Organization and the International Diabetes Federation, an individual’s glycemic status is defined based on two key biomarkers: fasting plasma glucose and glycated hemoglobin. Normal Glucose Regulation (NGR) is defined as fasting plasma glucose < 100 mg/dL and HbA1c < 5.7%; Pre-diabetes (preDM) is defined as fasting plasma glucose between 100 and 125 mg/dL and/or HbA1c between 5.7% and 6.4%; and Diabetes Mellitus (DM) is diagnosed when fasting plasma glucose ≥ 126 mg/dL and/or HbA1c ≥ 6.5% [38, 43].

Definition of covariates

Covariates included sociodemographic characteristics, health indicators, and blood test parameters. Sociodemographic characteristics encompassed age, sex, marital status, education level, smoking status, and alcohol consumption. Health indicators included height, weight, body mass index (BMI), waist circumference (WC), systolic blood pressure (SBP), diastolic blood pressure (DBP), as well as the presence of hypertension, dysglycemia, dyslipidemia, chronic pulmonary disease, and cancer. Medication history for hypertension, diabetes, and dyslipidemia was also recorded, along with renal function staging. Blood test parameters comprised white blood cell count, mean corpuscular volume, platelet count, blood urea nitrogen, creatinine, cystatin C, fasting plasma glucose, total cholesterol, triglycerides, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, C-reactive protein, glycated hemoglobin, uric acid, hematocrit, and hemoglobin.

Statistical analysis

Missing data for study participants were addressed using multiple imputation with five imputation cycles and 100 iterations (Table S1). Our primary analytical approach treated AIP-Cys as a continuous variable. We first assessed its association with CVD risk using multivariable Cox proportional hazards models. To investigate the dose–response relationship and identify potential nonlinearity, we employed restricted cubic splines (RCS) with knots at the 5th, 35th, 65th, and 95th percentiles of the AIP-Cys distribution. When RCS indicated a significant nonlinear association, a piecewise linear Cox proportional hazards model was applied to quantify the threshold effect. The threshold (inflection point) was determined using the maximum likelihood method, dividing the study population into low- and high-level intervals. Cox models were then fitted separately on either side of the threshold to calculate the hazard ratio (HR) and its 95% confidence interval (CI) per unit increase in AIP-Cys.

Subsequently, to provide a more clinically interpretable risk stratification and to visually describe the risk distribution observed in the continuous analysis, we categorized baseline AIP-Cys levels into quartiles: first quartile (Q1): -0.11 < AIP-Cys < 0.05; second quartile (Q2): 0.17 < AIP-Cys < 0.26; third quartile (Q3): 0.36 < AIP-Cys < 0.46; fourth quartile (Q4): 0.60 < AIP-Cys < 0.85. Baseline characteristics of the participants were summarized using descriptive statistics: categorical variables were presented as frequencies (N) and percentages (%), while all continuous variables were confirmed to follow a non-normal distribution by the Kolmogorov–Smirnov test and were summarized as medians with interquartile ranges.

The Kaplan–Meier method was used to plot cumulative incidence curves for CVD across different AIP-Cys groups, with the log-rank test applied to compare differences between groups. To assess the independent association between AIP-Cys and CVD risk, we constructed four multivariable Cox proportional hazards models. The selection of covariates was based on prior knowledge from cardiovascular epidemiology and established literature using the CHARLS cohort, including demographic characteristics (age, sex, marital status, education level, and hukou type), behavioral factors (alcohol consumption and smoking status), anthropometric measures (waist circumference and BMI), and clinical history (hypertension, dyslipidemia, etc.) [40, 41]. Model I was unadjusted; Model II adjusted for demographic characteristics, including age, sex, marital status, education level, and hukou type; Model III further incorporated behavioral and anthropometric indicators such as alcohol consumption, smoking status, waist circumference, and BMI based on Model II; and Model IV additionally adjusted for clinical comorbidities including hypertension, dyslipidemia, cancer, chronic pulmonary disease, as well as treatment for diabetes, stroke, heart disease, hypertension and hyperlipidemia (yes/no). The proportional hazards assumption for the main models was tested using Schoenfeld residuals, and no significant violation was detected (P > 0.05). The generalized variance inflation factor (GVIF^(1/(2*DF))) was used to assess multicollinearity among variables in each model, with all variance inflation factor (VIF) values below 2, indicating no significant multicollinearity issues (Table S2).

Stratified analyses were conducted by sex, age (< 61 years, ≥ 61 years; the median age of the study population), BMI (< 23 kg/m², ≥ 23 kg/m²; the cut-point approximates the median age of the study population), smoking status, alcohol consumption, hypertension, and dyslipidemia to evaluate the consistency of the relationship between AIP-Cys and CVD risk across different subgroups. Interaction terms between AIP-Cys and each stratification variable were introduced to assess potential effect modification [44, 45].

To explore potential pathways linking AIP-Cys to cardiovascular disease risk, we conducted an exploratory mediation analysis within the observational framework. A comprehensive set of candidate mediators was selected a priori based on their established roles in cardiovascular pathophysiology, including blood pressure parameters (SBP, DBP), lipid profiles (total cholesterol, triglycerides, LDL-C, HDL-C), renal function markers (creatinine, blood urea nitrogen, eGFR), inflammatory markers (hs-CRP), metabolic factors (uric acid), and other demographic characteristics (age, sex, marital status, education level, and hukou type). We employed a causal mediation analysis framework within the counterfactual approach. This analysis was performed separately for the overall population and for each glycemic status subgroup, with adjustment for all covariates used in Model IV. Only those variables demonstrating a statistically significant mediation effect (i.e., a bootstrapped confidence interval for the indirect effect that excluded the null) in the initial screening were retained as final mediators in the presented model. Bias-corrected 95% confidence interval for the direct, indirect, and mediated proportions were estimated using bootstrap resampling with 1,000 iterations to ensure robust inference.

To verify the robustness of the findings, four sensitivity analyses were conducted: (1) pooled analysis based on the five datasets generated from multiple imputation; (2) exclusion of participants with missing data; (3) exclusion of individuals diagnosed with hypertension or chronic pulmonary disease at baseline; (4) exclusion of participants who had received treatment for diabetes, heart disease, stroke, hypertension, or dyslipidemia at baseline; and (5) separate evaluation of the independent risk effects of AIP-Cys on heart disease and stroke [45]. Additionally, to assess the potential impact of unmeasured confounding, we performed an E-value analysis. The E-value quantifies the minimum strength of association, on the risk ratio scale, that an unmeasured confounder would need to have with both the exposure and the outcome, conditional on measured covariates, to fully explain away the observed exposure-outcome association. E-values were calculated for the point estimate and the lower limit of the 95% confidence interval of the hazard ratio for AIP-Cys (both continuous and quartiles) in the fully adjusted model (Model IV), using the formula proposed by VanderWeele and Ding [46].

To evaluate the incremental predictive value of the composite AIP-Cys index over its individual components (AIP and Cystatin C), we constructed four Cox proportional hazards models for CVD risk prediction: a basic model adjusted for the same conventional risk factors as Model IV, the basic model plus Cystatin C, the basic model plus AIP, and the basic model plus AIP-Cys. Model discrimination was assessed using receiver operating characteristic (ROC) curves, with area under the curve (AUC) values compared via the bootstrap method with 1,000 resamples. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated to further quantify the incremental predictive value [47, 48].

All statistical analyses were performed using R statistical software (version 4.5.1), with a two-sided P value < 0.05 considered statistically significant.

Results

Participants characteristics

According to the established inclusion and exclusion criteria, a total of 6,035 participants (47.52% male) from the CHARLS cohort were included in the baseline analysis. The baseline characteristics of the participants are presented in Table 1. During a median follow-up of 9.0 years, 651 participants who initially had no cardiovascular disease at baseline subsequently developed cardiovascular disease.

Table 1.

Baseline characteristics stratified by the AIP-Cys index

Characteristics	AIP-cys					P-value
Characteristics	overall	Q1	Q2	Q3	Q4	P-value
Num. of subjects	6035	1509	1509	1508	1509
Gender (N, %)						<0.001
Male	2868 (47.5%)	781 (51.8%)	707 (46.9%)	629 (41.7%)	751 (49.8%)
Female	3167 (52.5%)	728 (48.2%)	802 (53.1%)	879 (58.3%)	758 (50.2%)
Age(years)	61.00 (54.00, 69.00)	61.00 (54.00, 70.00)	60.00 (53.00, 68.00)	60.00 (54.00, 68.00)	62.00 (56.00, 70.00)	<0.001
Height (cm)	157.30 (151.50, 164.00)	157.00 (151.20, 163.30)	157.30 (151.10, 163.50)	157.00 (151.40, 163.43)	158.30 (152.00, 165.00)	<0.001
Weight (kg)	57.00 (50.20, 65.00)	53.20 (47.30, 59.40)	56.00 (49.60, 63.30)	58.60 (51.70, 66.40)	61.30 (53.40, 69.50)	<0.001
BMI (kg/m²)	22.92 (20.61, 25.51)	21.40 (19.55, 23.42)	22.63 (20.48, 24.90)	23.60 (21.24, 26.27)	24.40 (21.94, 27.03)	<0.001
Waistline (cm)	84.00 (77.10, 91.05)	79.00 (74.10, 85.20)	82.80 (76.50, 89.20)	86.60 (79.00, 93.00)	88.90 (81.60, 95.40)	<0.001
SBP (mmHg)	127.00 (114.00, 143.00)	123.00 (110.00, 139.00)	125.00 (113.00, 140.00)	128.00 (115.00, 144.25)	133.00 (119.00, 149.00)	<0.001
DBP (mmHg)	74.00 (67.00, 83.00)	72.00 (64.00, 80.00)	74.00 (66.00, 82.00)	75.00 (68.00, 84.00)	77.00 (69.00, 85.00)	<0.001
Marital status (N, %)						0.052
Marred	5262 (87.2%)	1307 (86.6%)	1336 (88.5%)	1323 (87.7%)	1296 (85.9%)
Divorced	28 (0.5%)	13 (0.9%)	3 (0.2%)	6 (0.4%)	6 (0.4%)
Others	745 (12.3%)	189 (12.5%)	170 (11.3%)	179 (11.9%)	207 (13.7%)
Hukou (N, %)						<0.001
Urban	862 (14.3%)	168 (11.1%)	193 (12.8%)	228 (15.1%)	273 (18.1%)
Rural	5144 (85.2%)	1335 (88.5%)	1306 (86.5%)	1276 (84.6%)	1227 (81.3%)
Others	29 (0.5%)	6 (0.4%)	10 (0.7%)	4 (0.3%)	9 (0.6%)
Education level (N, %)						0.838
High school or above	545 (9.0%)	123 (8.2%)	139 (9.2%)	147 (9.7%)	136 (9.0%)
Middle school	1109 (18.4%)	265 (17.6%)	273 (18.1%)	287 (19.0%)	284 (18.8%)
Primary school	1328 (22.0%)	334 (22.1%)	336 (22.3%)	320 (21.2%)	338 (22.4%)
No formal education	3053 (50.6%)	787 (52.2%)	761 (50.4%)	754 (50.0%)	751 (49.8%)
Smoke (N, %)						<0.001
Yes	2387 (39.6%)	636 (42.1%)	597 (39.6%)	514 (34.1%)	640 (42.4%)
No	3648 (60.4%)	873 (57.9%)	912 (60.4%)	994 (65.9%)	869 (57.6%)
Drink (N, %)						<0.001
Yes	2390 (39.6%)	687 (45.5%)	600 (39.8%)	523 (34.7%)	580 (38.4%)
No	3645 (60.4%)	822 (54.5%)	909 (60.2%)	985 (65.3%)	929 (61.6%)
Hypertension (N, %)	1320 (21.9%)	218 (14.4%)	263 (17.4%)	364 (24.1%)	475 (31.5%)	<0.001
Dyslipidemia (N, %)	398 (6.6%)	54 (3.6%)	75 (5.0%)	112 (7.4%)	157 (10.4%)	<0.001
Diabetes (N, %)	907 (15.0%)	132 (8.7%)	162 (10.7%)	232 (15.4%)	381 (25.2%)	<0.001
Chronic Lung Disease (N, %)	597 (9.9%)	180 (11.9%)	145 (9.6%)	122 (8.1%)	150 (9.9%)	0.005
Cancer (N, %)	55 (0.9%)	14 (0.9%)	12 (0.8%)	16 (1.1%)	13 (0.9%)	0.886
Hypertension medication (N, %)	993 (16.5%)	143 (9.5%)	193 (12.8%)	265 (17.6%)	392 (26.0%)	<0.001
Dyslipidemia medication (N, %)	213 (3.5%)	26 (1.7%)	33 (2.2%)	57 (3.8%)	97 (6.4%)	<0.001
Diabetes medication (N, %)	185 (3.1%)	30 (2.0%)	32 (2.1%)	49 (3.2%)	74 (4.9%)	<0.001
WBC (×10^9)	6.00 (5.00, 7.20)	5.70 (4.70, 6.90)	5.88 (4.83, 7.20)	6.10 (5.00, 7.20)	6.30 (5.30, 7.60)	<0.001
MCV (fL)	91.50 (86.90, 95.80)	92.00 (87.00, 96.60)	91.60 (86.90, 95.70)	91.00 (86.50, 95.00)	91.30 (87.10, 95.80)	0.004
Platelets (×10^9/L)	207.00 (162.00, 255.00)	204.00 (163.00, 251.00)	206.00 (160.00, 255.00)	209.00 (164.00, 257.00)	207.00 (159.00, 259.00)	0.375
BUN (mg/dL)	15.29 (12.63, 18.29)	15.97 (12.91, 19.27)	15.07 (12.52, 18.21)	14.76 (12.27, 17.51)	15.38 (12.94, 18.26)	<0.001
FBG (mg/dL)	102.24 (94.14, 113.22)	99.54 (91.98, 108.36)	100.44 (92.70, 109.62)	102.96 (95.22, 113.85)	108.18 (98.10, 124.38)	<0.001
Creatinine (mg/dL)	0.76 (0.64, 0.88)	0.75 (0.63, 0.86)	0.73 (0.63, 0.85)	0.73 (0.64, 0.87)	0.81 (0.70, 0.96)	<0.001
Total cholesterol (mg/dL)	190.59 (167.01, 215.72)	186.34 (165.46, 209.15)	187.11 (164.30, 211.47)	192.91 (169.62, 221.14)	196.01 (170.49, 221.91)	<0.001
Triglycerides (mg/dL)	103.54 (74.34, 150.45)	61.06 (51.33, 71.68)	89.39 (77.88, 103.54)	125.67 (106.20, 149.57)	192.04 (149.57, 257.54)	<0.001
HDL (mg/dL)	49.87 (40.98, 60.31)	65.34 57.60, 75.00)	53.35 (47.17, 59.92)	46.01 (40.59, 52.19)	37.50 (32.09, 43.69)	<0.001
LDL (mg/dL)	114.43 (93.17, 137.24)	107.86 (90.08, 128.74)	116.37 (95.49, 138.40)	121.39 (99.74, 145.36)	112.11 (86.21, 136.08)	<0.001
CRP (mg/L)	1.02 (0.54, 2.14)	0.74 (0.44, 1.69)	0.87 (0.51, 1.81)	1.08 (0.59, 2.18)	1.39 (0.76, 2.78)	<0.001
HbA1c (%)	5.10 (4.90, 5.40)	5.10 (4.80, 5.40)	5.10 (4.90, 5.40)	5.20 (4.90, 5.40)	5.20 (4.90, 5.50)	<0.001
UricAcid (mg/dL)	4.29 (3.56, 5.14)	4.09 (3.44, 4.89)	4.09 (3.40, 4.88)	4.21 (3.53, 5.03)	4.79 (4.03, 5.73)	<0.001
Hematocrit (mmol/mol)	41.60 (38.00, 45.20)	41.00 (37.20, 44.70)	41.40 (37.80, 45.10)	41.50 (37.90, 45.00)	42.30 (39.00, 46.00)	<0.001
Hemoglobin (g/dL)	14.20 (13.00, 15.50)	14.00 (12.80, 15.20)	14.10 (12.90, 15.40)	14.20 (13.00, 15.50)	14.50 (13.20, 15.80)	<0.001
eGFR (ml/min/1.73m²)	79.85 (66.00, 94.44)	81.23 (66.87, 95.68)	82.63 (69.55, 97.05)	82.64 (69.03, 97.18)	73.30 (59.21, 87.19)	<0.001
GMS						<0.001
NGR	2450 (40.6%)	745 (49.4%)	712 (47.2%)	564 (37.4%)	429 (28.4%)
pre-DM	2660 (44.1%)	628 (41.6%)	633 (41.9%)	704 (46.7%)	695 (46.1%)
DM	925 (15.3%)	136 (9.0%)	164 (10.9%)	240 (15.9%)	385 (25.5%)
CystatinC (mg/L)	0.98 (0.86, 1.13)	0.97 (0.86, 1.11)	0.96 (0.85, 1.09)	0.95 (0.84, 1.09)	1.03 (0.89, 1.20)	<0.001
AIP	0.31 (0.12, 0.54)	-0.01(-0.11, 0.06)	0.22 (0.17, 0.27)	0.42 (0.36, 0.51)	0.69 (0.57, 0.86)	<0.001
AIP-Cys	0.31 (0.12, 0.53)	-0.01 (-0.11, 0.05)	0.21 (0.17, 0.26)	0.41 (0.36, 0.46)	0.69 (0.60, 0.85)	<0.001

Open in a new tab

Data are presented as medians with interquartile ranges (25th to 75th percentiles) or as numbers with percentages, N (%).

BMI body mass index, SBP systolic blood pressure, DBP diastolic blood pressure, WBC white blood cell, MCV mean corpuscular volume, BUN blood urea nitrogen, FBG fasting blood glucose, HDL high-density lipoprotein cholesterol ,LDL low-density lipoprotein cholesterol, CRP high-sensitivity C-reactive protein, HbA1c glycosylated hemoglobin A1c, eGFR estimated glomerular filtration rate, GMS glucose metabolism status, NGR normal glucose regulation, preDM pre-diabetes, DM diabetes, AIP atherogenic index of plasma, Cys Cystatin C, AIP-Cys the product of AIP × Cystatin C

Further analysis revealed that individuals in higher AIP-Cys quartile groups exhibited significantly higher levels of weight, BMI, waist circumference, systolic and diastolic blood pressure, white blood cell count, fasting plasma glucose, total cholesterol, triglycerides, low-density lipoprotein, C-reactive protein, uric acid, hemoglobin, and hematocrit, while their high-density lipoprotein levels were relatively lower. Additionally, groups with higher AIP-Cys levels showed a higher prevalence of chronic conditions (hypertension, diabetes, hyperlipidemia) and greater use of medications (antihypertensive, antidiabetic, antidyslipidemic) (all P values < 0.05).

Association between AIP-cys and CVD risk

We first assessed the association between AIP-Cys as a continuous variable and CVD risk. As shown in Table 2, in the overall population, a significant positive association was observed, with the HR decreasing progressively from 1.75 (95% CI: 1.42–2.16) in Model I to 1.34 (95% CI: 1.06–1.69) in the fully adjusted Model IV (P for trend < 0.001). Similar patterns were observed in the NGR and preDM populations (Model IV HR: NGR = 1.60, 95% CI: 1.13–2.26; preDM = 1.32, 95% CI: 1.02–1.70). In stark contrast, the association was weakest and did not reach statistical significance in the DM population (Model IV HR = 1.24, 95% CI: 0.78–1.95). To explore the potential role of medication use as a confounding factor in this subgroup, we further analyzed baseline medication rates among participants with diabetes. As shown in Table S3, diabetic participants exhibited substantially higher rates of medication use compared to non-diabetic individuals, including for hypertension (26.4% vs. 14.7%, P < 0.001) and dyslipidemia (8.8% vs. 2.5%, P < 0.001).

Table 2.

Association between the AIP-Cys index and the risk of CVD

Characteristics	Model I		Model II		Model III		Model IV
Characteristics	HR (95%CI)	P-value	HR (95%CI)	P-value	HR (95%CI)	*P-value*	HR (95%CI)	*P-value*
Total
Continuous AIP-Cys	1.75 (1.42, 2.16)	<0.001	1.71 (1.38, 2.11)	<0.001	1.43 (1.14, 1.79)	<0.001	1.34 (1.06, 1.69)	<0.001
AIP-Cys quartile
Q1	Reference		Reference		Reference		Reference
Q2	1.37 (1.07, 1.75)	<0.001	1.34 (1.06, 1.70)	0.002	1.32 (1.05, 1.65)	<0.001	1.31 (1.04, 1.64)	<0.001
Q3	1.89 (1.49, 2.38)	<0.001	1.85 (1.46, 2.34)	<0.001	1.66 (1.31, 2.10)	<0.001	1.62 (1.28, 2.06)	<0.001
Q4	1.86 (1.47, 2.35)	<0.001	1.83 (1.44, 2.31)	<0.001	1.55 (1.21, 1.97)	<0.001	1.45 (1.14, 1.85)	0.003
P for trend		<0.001		<0.001		<0.001		<0.001
NGR
Continuous AIP-Cys	1.84 (1.22, 2.77)	0.003	1.77 (1.17, 2.69)	0.007	1.65 (1.15, 2.38)	0.008	1.60 (1.13, 2.26)	0.009
AIP-Cys quartile
Q1	Reference		Reference		Reference		Reference
Q2	1.66 (1.32, 2.10)	<0.001	1.65 (1.30, 2.09)	<0.001	1.55 (1.22, 1.98)	<0.001	1.54 (1.20, 1.97)	<0.001
Q3	1.79 (1.21, 2.65)	0.003	1.75 (1.18, 2.60)	0.005	1.63 (1.09, 2.43)	0.017	1.60 (1.07, 2.38)	<0.001
Q4	1.79 (1.18, 2.71)	0.007	1.74 (1.14, 2.65)	0.010	1.61 (1.13, 2.32)	0.011	1.55 (1.10, 2.19)	0.016
P for trend		0.011		0.020		0.003		<0.001
preDM
Continuous AIP-Cys	1.49 (1.08, 2.08)	0.017	1.47 (1.05, 2.04)	0.020	1.36 (1.04, 1.78)	0.022	1.32 (1.02, 1.70)	0.021
AIP-Cys quartile
Q1	Reference		Reference		Reference		Reference
Q2	1.49 (1.20, 1.86)	0.005	1.48 (1.18, 1.85)	0.007	1.41 (1.09, 1.82)	0.009	1.38 (1.05, 1.81)	0.012
Q3	1.79 (1.29, 2.48)	<0.001	1.75 (1.25, 2.45)	<0.001	1.65 (1.18, 2.31)	0.004	1.62 (1.16, 2.28)	0.005
Q4	1.48 (1.05, 2.07)	0.025	1.47 (1.04, 2.06)	0.029	1.28 (0.90, 1.83)	0.167	1.22 (0.85, 1.74)	0.282
P for trend		0.004		<0.001		<0.001		<0.001
DM
Continuous AIP-Cys	1.51 (1.00, 2.26)	0.047	1.48 (0.98, 2.24)	0.062	1.29 (0.83, 1.99)	0.260	1.24 (0.78, 1.95)	0.361
AIP-Cys quartile
Q1	Reference		Reference		Reference		Reference
Q2	1.37 (0.65, 2.91)	0.407	1.32 (0.62, 2.80)	0.476	1.17 (0.55, 2.50)	0.688	1.27 (0.59, 2.72)	0.537
Q3	1.91 (0.97, 3.76)	0.060	1.78 (0.90, 3.52)	0.096	1.56 (0.78, 3.09)	0.206	1.55 (0.78, 3.10)	0.213
Q4	2.23 (1.18, 4.22)	0.014	2.15 (1.13, 4.08)	0.020	1.78 (0.93, 3.42)	0.084	1.73 (0.89, 3.34)	0.105
P for trend		0.0290		0.010		0.023		<0.001

Open in a new tab

Model I: Unadjusted model;

Model II: Adjusted for demographic characteristics, including age, gender, marital status, educational level, and hukou type;

Model III: Further included behavioral and anthropometric indicators on top of Model II, such as drinking status, smoking status, waist circumference, and BMI;

Model IV: On the basis of Model III, additional adjustments were made for clinical comorbidities of hypertension, dyslipidemia, cancer, chronic lung diseases, as well as treatment for diabetes, stroke, heart disease, hypertension and hyperlipidemia.

NGR normal glucose regulation, preDM pre-diabetes, DM diabetes, AIP atherogenic index of plasma, Cys Cystatin C, AIP-Cys the product of AIP and Cystatin, HR hazard ratio, BMI body mass index

We further categorized AIP-Cys into quartiles (Q1-Q4) to provide a clinically interpretable risk stratification (Table 2). In the overall population, compared to Q1 (lowest risk), the HRs for Q2, Q3, and Q4 were all significantly elevated, with the peak risk observed in Q3 (HR = 1.62, 95% CI: 1.28–2.06) rather than Q4 (HR = 1.45, 95% CI: 1.14–1.85). This pattern of risk plateauing or attenuating at the highest quartile, with Q3 often exhibiting the highest risk, was also evident in the NGR and preDM subgroups. Again, no consistent or statistically significant associations were found across quartiles in the DM population.

As shown in Fig. 2, this pattern is visually corroborated by the Kaplan–Meier survival curves. In the overall (Fig. 2A), NGR (Fig. 2B), and preDM (Fig. 2C) populations, the curves for higher AIP-Cys quartiles showed significant separation from Q1, with the Q3 curve often displaying a cumulative risk comparable to or exceeding that of Q4, especially in the later follow-up period. In the DM population (Fig. 2D), the survival curves for different quartiles showed substantial overlap, reflecting the attenuated association. The analysis of cumulative AIP-Cys (cuAIP-Cys) showed that higher cumulative levels were associated with a higher incidence of CVD (Table S4).

Justification for the multiplicative combination

To validate the conceptual framework of AIP-Cys, we examined the interaction between its components. In the fully adjusted model, we observed a significant multiplicative interaction between AIP and Cystatin C on CVD risk in the overall population (Table S5: HR for interaction term = 2.15, 95% CI: 1.74–2.68, P = 0.018). This finding indicates a statistical interaction between AIP and Cystatin C, suggesting that their combined association with CVD risk is greater than the sum of their individual associations on the multiplicative scale. This statistical interaction provides a rationale for using their product (AIP-Cys) as a composite marker in this exploratory analysis.

Nonlinear association between AIP-cys and CVD risk

To formally investigate the dose–response relationship, we employed RCS analysis (Fig. 3). After full adjustment for covariates, a significant nonlinear association was detected between continuous AIP-Cys and CVD risk in the overall population, the NGR population, and the preDM population (P for nonlinearity: 0.033, 0.041, and 0.026, respectively; Fig. 3A-C). No significant nonlinear trend was observed in the DM population (P for nonlinearity > 0.05; Fig. 3D), consistent with the lack of a significant linear association.

To quantify this nonlinear relationship, a two-piecewise Cox regression model was applied (Table 3). In the overall and NGR populations, an inflection point was identified at an AIP-Cys value of 0.47. Below this threshold, CVD risk increased sharply with increasing AIP-Cys (overall: HR = 2.27, 95% CI: 1.43–3.58, P < 0.001; NGR: HR = 3.38, 95% CI: 1.51–7.59, P < 0.001). Above 0.47, the association was no longer statistically significant. In the preDM population, a similar threshold was observed at 0.37, with a significant risk increase below this point (HR = 2.27, 95% CI: 1.09–4.75, P < 0.001) and a non-significant association above it. This threshold analysis aligns with the quartile findings, suggesting a plateauing of risk at higher AIP-Cys levels. For comparison, RCS analysis of the individual components showed a nonlinear association for AIP but a linear pattern for Cystatin C with CVD risk (Figure S1).

Table 3.

Threshold effect in the nonlinear relationship between AIP-Cys and CVD risk in the total, NGR and preDM population

Participants	Adjusted HR (95% CI)	P-value
AIP-Cys-Total
Total	1.34 (1.06, 1.69)	<0.001
Fitting by two-piecewise Cox proportional risk model
Inflection point	0.47
AIP-Cys < 0.47	2.27 (1.43, 3.58)	<0.001
AIP-Cys > 0.47	0.83 (0.53, 1.30)	0.415
P for Log-likelihood ratio		0.007
AIP-Cys-NGR
Total	1.60 (1.13, 2.26)	<0.001
Fitting by two-piecewise Cox proportional risk model
Inflection point	0.47
AIP-Cys < 0.47	3.38 (1.51, 7.59)	<0.001
AIP-Cys > 0.47	0.38 (0.11, 1.33)	0.129
P for Log-likelihood ratio		0.008
AIP-Cys-preDM
Total	1.32 (1.02, 1.70)	<0.001
Fitting by two-piecewise Cox proportional risk model
Inflection point	0.37
AIP-Cys < 0.37	2.27 (1.09, 4.75)	<0.001
AIP-Cys > 0.37	0.79 (0.42, 1.48)	0.460
P for Log-likelihood ratio		0.009

Open in a new tab

The two-stage Cox proportional hazards regression model was used to conduct a threshold effect analysis of the nonlinear relationship between AIP-Cys and the risk of CVD. Multivariable models were adjusted for age, sex, marital status, education level, household registration status, alcohol consumption, smoking status, waist circumference, BMI, hypertension, dyslipidemia, cancer, chronic pulmonary disease, as well as treatment for diabetes, stroke, heart disease, hypertension and hyperlipidemia.

NGR normal glucose regulation, preDM pre-diabetes, AIP atherogenic index of plasma, Cys Cystatin C, AIP-Cys the product of AIP × Cystatin C, HR hazard ratio

Subgroup analysis

Stratified analyses (Fig. 4) showed that the positive association between continuous AIP-Cys (per unit increase) and CVD risk was consistent across subgroups defined by sex, age (< 61 vs. ≥ 61 years), BMI (< 23 vs. ≥ 23 kg/m²), smoking status, alcohol consumption, hypertension, and dyslipidemia. The interaction terms between AIP-Cys and these stratification variables were not statistically significant (P for interaction > 0.05 for all), indicating a lack of significant effect modification.

Predictive performance of AIP-cys

Time-dependent ROC analysis (Fig. 5) indicated that AIP-Cys demonstrated modest discriminative ability for near-term (1-year) CVD risk prediction in the overall (AUC = 0.64), NGR (AUC = 0.65), and preDM (AUC = 0.67) populations. However, its predictive performance declined for medium- to long-term (4- and 8-year) risk (AUCs ranging from 0.57 to 0.58, Fig. 5A). These findings indicate that AIP-Cys is not suitable for long-term cardiovascular risk prediction, and its clinical utility should be restricted to short-term risk assessment. Given the lack of a significant association, predictive performance was not evaluated in the DM population. The composite AIP-Cys indicator showed a potential predictive advantage over models containing AIP or Cystatin C alone (Figure S2). These findings suggest that the AIP-Cys model may hold some clinical relevance only for near-term cardiovascular event risk assessment. For medium- to long-term risk stratification, the indicator lacks stable discriminative ability, implying its limited application value in long-term prognosis prediction. This observation suggests that AIP-Cys might be more suitable for short-term risk warning rather than for the dynamic monitoring and evaluation of long-term cardiovascular risk. Despite these limitations, as shown in Table S6, the composite AIP-Cys index provided statistically significant incremental predictive value over AIP or Cystatin C alone in terms of AUC, NRI, and IDI.

Mediation analysis

Mediation analysis (Fig. 6) suggested that systolic and diastolic blood pressure partially mediated the association between AIP-Cys and CVD risk in the overall, NGR, and preDM populations (mediation proportions: SBP 14.6% (Fig. 6A), 12.6% (Fig. 6C), 17.9% (Fig. 6F); DBP 10.6% (Fig. 6B), 6.9% (Fig. 6D), 12.9% (Fig. 6G), respectively). Additionally, LDL-C mediated 6.7% of the association in the NGR group (Fig. 6E).

Sensitivity analysis

To verify the robustness of the main findings, several sensitivity analyses were performed. First, the results remained consistent between the complete dataset without missing data (5,841 participants) and the multiply imputed datasets (Table S7). Second, after excluding participants with hypertension and/or chronic pulmonary disease at baseline, AIP-Cys remained significantly associated with the risk of developing CVD during follow-up (Table S8). Third, further exclusion of participants receiving treatment for diabetes, heart disease, stroke, hypertension, or dyslipidemia at baseline yielded similar results (Table S9). Fourth, when separately evaluating the independent risk effects of AIP-Cys on heart disease (Table S10) and stroke, the findings remained statistically significant (Table S11).

To further assess robustness to potential unmeasured confounding, we calculated E-values for significant associations in the fully adjusted models (Table S12). In the total population, the strongest association (AIP-Cys Q3 vs Q1: HR = 1.62, 95% CI: 1.28–2.06) yielded an E-value of 2.62 for the point estimate and 1.88 for the lower confidence limit, indicating that an unmeasured confounder would need to be associated with both exposure and outcome by a risk ratio of at least 2.62 to explain away the point estimate, and 1.88 to nullify its confidence interval. In the NGR subgroup, the continuous AIP-Cys showed the most robust association (E-value for lower CI limit = 1.51). In the preDM subgroup, while the continuous association was more sensitive to unmeasured confounding (E-value for lower CI limit = 1.16), the Q3 association remained highly robust (E-value for HR = 2.62, E-value for lower CI limit = 1.59). Overall, these analyses suggest that the observed associations, particularly the nonlinear pattern with peak risk in Q3, are moderately to highly robust and unlikely to be entirely explained by unmeasured factors.

Discussion

This study, based on the large-scale, nationally representative CHARLS cohort, is the first to systematically investigate the predictive value of the combined index of the AIP and Cystatin C (AIP-Cys) for the risk of incident CVD in populations with different glycemic statuses. Our key findings include: First, in the overall population as well as in individuals with NGR and preDM, higher levels of AIP-Cys were independently associated with a significantly increased risk of CVD, and this association exhibited a significant nonlinear pattern. This association was markedly attenuated or absent in the population with diabetes mellitus. Second, the combined AIP-Cys indicator demonstrated superior predictive performance compared to AIP or Cystatin C used individually. Third, the positive association between AIP-Cys and CVD risk remained consistent across subgroups stratified by age, sex, BMI, and other factors. Fourth, mediation analysis indicated that blood pressure (systolic and diastolic) served as a significant mediating pathway through which AIP-Cys influenced CVD risk in the overall population, the NGR group, and the preDM group. Additionally, in the NGR group, LDL also mediated the association between AIP-Cys and CVD. Fifth, higher cumulative AIP-Cys levels were associated with a higher incidence of CVD.

The strong association between AIP, as a marker reflecting atherogenic lipoprotein particles, and CVD events has been validated in numerous cross-sectional and prospective studies [16, 17, 22, 49, 50]. Early research uncovered the pathophysiological mechanisms directly linking AIP to atherosclerosis, including its significant correlation with small dense low-density lipoprotein particle size [51], its influence on HDL cholesterol esterification rates [52–54], and its mediating role in insulin resistance [55]. These mechanisms collectively explain the superiority of AIP in predicting cardiovascular events such as acute coronary syndromes and stroke. As research has progressed, scholars have found that compared to traditional lipid parameters, AIP offers a more comprehensive reflection of the overall burden of the atherogenic lipid profile and demonstrates stronger risk stratification capabilities under specific metabolic conditions like diabetes and pre-diabetes. This suggests a synergistic effect between abnormal glucose metabolism and elevated AIP levels, significantly amplifying cardiovascular risk [22, 56, 57]. Cystatin C, as a sensitive marker for glomerular filtration function superior to traditional creatinine, is associated with cardiovascular disease primarily due to the inherent link between chronic kidney disease and cardiovascular risk [58]. Several large-scale meta-analyses have confirmed that elevated cystatin C levels are significantly associated with risks of all-cause cardiovascular events, coronary heart disease, myocardial infarction, and stroke, independently of traditional risk factors [59–63]. Notably, cystatin C demonstrates better risk prediction ability than eGFR, particularly by identifying cardiovascular risk in individuals with compensatory renal function, whereas traditional eGFR only holds predictive value when renal function is significantly impaired [62]. Its mechanisms extend beyond renal function assessment to include direct involvement in inflammatory regulation and cysteine protease inhibition within the atherosclerotic process [59, 64–66]. Additionally, cystatin C is linked to an increased risk of developing diabetes and mortality among diabetic patients [67, 68]. Against this research background, this study combined cystatin C with AIP, a lipid metabolism marker, to construct a composite indicator termed AIP-Cys. This approach fully leveraged the respective sensitivity of AIP and cystatin C in predicting cardiovascular risk. Furthermore, by integrating lipid metabolism pathways with renal and inflammatory pathways, it explored the synergistic predictive value of this combined indicator for cardiovascular risk across different glycemic levels.

Through long-term follow-up analysis, this study revealed a complex nonlinear relationship between AIP-Cys and cardiovascular disease risk. Specifically, individuals at intermediate levels (Q3) exhibited the highest cardiovascular risk across multiple subgroups, rather than a continuous escalation at the highest levels (Q4). This phenomenon was particularly pronounced in the NGR and preDM populations but was not evident in the DM population. This "threshold effect" was further confirmed by the piecewise linear model. Specifically, when AIP-Cys levels were below 0.47 in the NGR population and below 0.37 in the preDM population, CVD risk increased sharply with rising AIP-Cys. Beyond these inflection points, however, the increase in risk plateaued or became non-significant. This may reveal a critical biological "saturation effect": once lipid disorders and low-grade inflammation reach a certain threshold, their damage to the cardiovascular system approaches its maximum, after which the absolute increase in risk diminishes [69–71]. This phenomenon aligns with the concept of a nonlinear relationship between atherosclerotic burden and cardiovascular events, where beyond a certain plaque load, the risk of rupture and subsequent events may plateau or become less dependent on the incremental increase in the underlying pathological drivers [72]. In addition to this saturation effect, alternative explanations for the observation that risk peaked in Q3 rather than Q4 should be considered. For instance, participants in the highest quartile (Q4) may have simultaneously experienced a higher burden of subclinical or clinical comorbidities, leading to competitive mortality from non-cardiovascular causes that precluded the observation of incident CVD events during follow-up. Alternatively, these high-risk individuals have received more aggressive drug treatment interventions. As shown in Table 1, the population in Q4 had a higher rate of drug use, such as lipid-lowering or antihypertensive treatments, which could have attenuated their subsequent cardiovascular risk and contributed to the lower relative risk observed in Q4 compared to Q3. This could be another plausible explanation for the observed lower risk in the Q4 group compared to the Q3 group. Another key finding is that the predictive value of AIP-Cys was significantly attenuated in the DM population. This attenuation of the synergistic effect may be attributed to several interconnected factors. First, in individuals with established diabetes, the overwhelming pathophysiological drivers of CVD, such as severe hyperglycemia, profound insulin resistance, and advanced glycation end-products, may dominate the risk profile. The risk information captured by the lipid-renal/inflammatory pathways (AIP-Cys) might represent an earlier or parallel pathological process whose incremental contribution becomes masked or “drowned out” by these dominant diabetic risk factors. Second, and directly relevant to the present findings, baseline data from our cohort confirmed that diabetic participants received substantially more intensive pharmacological interventions than non-diabetic individuals (Table S3), including for hypertension (26.4% vs. 14.7%) and dyslipidemia (8.8% vs. 2.5%). Such aggressive clinical management, particularly the use of statins and antihypertensive drugs, could directly modify the components of AIP-Cys (e.g., lipid profile and blood pressure), thereby disrupting its physiological association with incident CVD risk. Third, it is plausible that a “ceiling effect” may exist: once the metabolic dysregulation reaches the severity seen in diabetes, the cardiovascular system may have already sustained substantial damage, and the additive risk from the specific pathways reflected by AIP-Cys no longer discriminates risk as effectively. Finally, survival bias could also play a role, as individuals with DM and extremely high AIP-Cys might have experienced events earlier, leaving a healthier surviving cohort for analysis. These potential reasons highlight that AIP-Cys is likely most valuable for risk stratification in the early, modifiable stages of metabolic disease (NGR and preDM) rather than for further differentiating risk within an already very high-risk diabetic population. This important limitation is now explicitly acknowledged in the conclusions section. Exploratory mediation analysis suggested that systolic blood pressure may partially account for the observed association between AIP-Cys and cardiovascular risk. Specifically, systolic blood pressure mediated 14.6% of the effect in the overall population, 12.6% in the NGR group, and 17.9% in the preDM group. This provides a potential mechanistic explanation for how elevated AIP-Cys may contribute to CVD risk, although causal inference is limited by the observational study design. Specifically, lipid abnormalities and low-grade inflammation may worsen vascular stiffness and endothelial dysfunction, thereby increasing blood pressure and ultimately elevating the risk of CVD events. Notably, in the NGR group alone, LDL-C also emerged as a significant mediator (6.7%), a finding not observed in the preDM population. This divergence may hint at a shift in the underlying mechanisms as glucose metabolism deteriorates. In the NGR state, the pathogenic effect of AIP-Cys may be partially mediated through traditional lipid pathways (influencing LDL-C levels) and blood pressure. However, as individuals progress to pre-diabetes, the pro-atherogenic influence of AIP-Cys might become more direct, operating through pathways less dependent on traditional lipid fractions, such as enhanced endothelial dysfunction [73, 74], oxidative stress [75], and insulin resistance [76–78]. This hypothesis of a mechanism transition warrants further investigation in experimental settings. Additionally, integrating AIP and cystatin C into a composite indicator demonstrated superior potential for predicting cardiovascular risk compared to either marker alone (Figure S2). Importantly, our formal incremental predictive analysis further substantiated this observation. In the overall, NGR, and preDM populations, the AIP-Cys model provided statistically significant improvements in AUC, NRI, and IDI compared with models containing AIP alone or Cystatin C alone (Table S6). These findings indicate that the composite index captures synergistic information from lipid and renal/inflammatory pathways that is not fully accounted for by either biomarker individually. The superior reclassification performance, particularly in the preDM group (NRI = 0.099, IDI = 0.010), suggests that AIP-Cys could help reclassify a meaningful proportion of individuals into more appropriate risk categories, thereby facilitating earlier and more targeted preventive interventions. However, predictive performance evaluation indicated that while AIP-Cys exhibited modest discriminative ability for short-term (1-year) risk assessment (AUC = 0.64), its predictive efficacy for medium- to long-term (≥ 4 years) risk was limited (AUC = 0.58), suggesting it is more suitable for short- to medium-term risk assessment. This aligns with the dynamic nature of biomarkers and underscores the necessity of regular monitoring and assessing cumulative exposure (such as cuAIP-Cys in this study) in clinical practice. The results indicated a significant positive correlation between higher cumulative AIP-Cys levels and the incidence of cardiovascular disease, a pattern similar to that observed for AIP-Cys, with the association most pronounced in the NGR and preDM populations.

These findings hold several potential clinical implications. First, as an easily calculable composite index based on routine lipid and renal function tests, AIP-Cys could serve as a pragmatic risk identification tool in community screening or health check-ups. It may help clinicians to more promptly identify individuals who are "normoglycemic or prediabetic but at significantly elevated cardiovascular risk," prompting more aggressive blood pressure and lipid management and lifestyle interventions, even in the absence of a diabetes diagnosis. Second, by integrating lipid and renal/inflammatory pathways, AIP-Cys could potentially serve as a dynamic monitoring target to evaluate the comprehensive effects of lifestyle or pharmacological interventions (e.g., statins, anti-inflammatory therapies). Third, our findings support a differentiated risk assessment strategy tailored to glycemic status. While glycemic control remains paramount for diabetic patients, a composite biomarker like AIP-Cys may provide valuable incremental risk stratification information in NGR and preDM populations, facilitating more precise primary prevention. The necessity of this research lies in addressing the pressing need to shift CVD prevention towards earlier risk identification. Our study validates a low-cost, highly accessible novel composite biomarker with specific application potential during the critical intervention window of prediabetes. It provides new epidemiological evidence to address potential underestimation of risk in these populations by traditional models, offering a fresh perspective for early and precise cardiovascular disease prevention.

Strengths and limitations

This study possesses several notable strengths. First, it is based on the CHARLS cohort, a large-scale, nationally representative prospective study with a substantial sample size and extended follow-up duration, effectively reducing selection bias and enhancing the generalizability and external validity of the findings. Second, this study innovatively integrates AIP, which reflects lipid metabolism disorders, with cystatin C, a marker of renal function and low-grade inflammatory status, to form the composite indicator AIP-Cys. By exploring its predictive value for cardiovascular risk from a multi-pathway synergistic perspective, it provides new insights for risk assessment strategies that go beyond single biomarkers. Third, we not only analyzed baseline exposure levels but also investigated the long-term effects of cumulative exposure (cuAIP-Cys). Additionally, we revealed the nonlinear relationship and threshold effects between AIP-Cys and CVD risk, offering important theoretical foundations for precisely identifying high-risk thresholds and developing differentiated intervention targets. Fourth, comprehensive mediation analysis highlighted the significant role of blood pressure in the association between AIP-Cys and CVD, deepening the understanding of potential pathophysiological mechanisms. Finally, we validated the robustness of the main findings through detailed subgroup analyses and multiple sensitivity analyses, and systematically evaluated the heterogeneity in the predictive performance of this indicator across populations with different glycemic statuses, thereby strengthening the reliability of the conclusions.

At the same time, this study has certain limitations. First, the ascertainment of cardiovascular disease outcomes relied on participants’ self-reported physician diagnoses, which may be subject to misclassification bias arising from recall inaccuracy, limited understanding of medical terminology, or differential access to healthcare services. This issue is particularly pertinent in light of the modest long-term predictive performance of AIP-Cys (e.g., 8-year AUC ≤ 0.58), because outcome misclassification may attenuate the observed associations and weaken the apparent predictive utility of the marker. Such misclassification would most likely be non-differential with respect to AIP-Cys exposure, in which case the estimated hazard ratios would be biased toward the null, potentially underestimating the true association between AIP-Cys and CVD risk. Accordingly, the significant associations identified in our study may represent conservative estimates of the underlying effect. Nevertheless, we cannot exclude the possibility of differential misclassification, whereby the accuracy of self-reported diagnoses may vary according to participants’ health status or other correlated factors. Future studies incorporating medical record linkage, endpoint adjudication by clinical committees, or national registry data will be essential to validate these findings and obtain more accurate effect estimates. Second, a related concern inherent to observational studies is the potential for residual confounding. Despite our comprehensive adjustment for a wide range of known confounders, we cannot exclude the possibility that unmeasured factors, such as detailed dietary habits, physical activity levels, or genetic predispositions, may have influenced the results. Furthermore, while we adjusted for self-reported medication use, we lacked granular data on treatment intensity, adherence, and duration. This is particularly relevant for participants in the highest AIP-Cys quartile (Q4), who may have received more aggressive pharmacological interventions. Such residual confounding by indication could have attenuated the observed risk in this group, potentially contributing to the nonlinear plateauing effect observed. To quantitatively assess the robustness of our findings to such unmeasured confounding, we conducted an E-value analysis (Table S12). This analysis revealed that for our most robust finding, the Q3 vs Q1 association in the preDM subgroup (HR = 1.62, 95% CI: 1.16–2.28), an unmeasured confounder would need to be associated with both AIP-Cys and CVD by a risk ratio of 2.62 to fully explain away the point estimate, and a risk ratio of 2.62 to nullify its confidence interval. These values exceed the typical effect sizes of common cardiovascular risk factors (e.g., smoking, physical inactivity), suggesting that this finding is unlikely to be completely explained by unmeasured confounding. In the total population, the Q3 association (HR = 1.62, 95% CI: 1.28–2.06) yielded E-values of 2.62 and 1.88 for the point estimate and lower confidence limit, respectively, indicating moderate to high robustness. However, it is important to acknowledge that some associations were more sensitive. Specifically, the continuous AIP-Cys measure in the preDM subgroup yielded an E-value of only 1.16 for the lower bound of its confidence interval, warranting a more cautious interpretation of the linear component of this association. Nevertheless, the consistency of the nonlinear pattern, particularly the peak risk consistently observed in Q3, across multiple subgroups strengthens our overall confidence in the primary conclusions. Future studies with access to prescription databases or employing advanced methods such as instrumental variable analysis could further clarify these relationships by better accounting for treatment effects and unmeasured confounding. Third, the calculation of AIP-Cys relied on biomarker measurements at specific time points and did not capture their dynamic changes during follow-up. Future research incorporating longitudinal repeated measurements for trajectory analysis would be more valuable. Fourth, CHARLS participants are primarily middle-aged and older Chinese adults aged 45 years and above. Therefore, the applicability of our findings to younger populations or other racial/ethnic groups requires further validation. Finally, the limited predictive efficacy of AIP-Cys for medium- to long-term (≥ 4 years) risk suggests that it may be more suitable for short- to medium-term risk screening and dynamic monitoring rather than for predicting long-term absolute risk.

Conclusions

Based on data from a large-scale prospective cohort, this study confirms that the composite indicator AIP-Cys, derived from the Atherogenic Index of Plasma and Cystatin C, is an independent risk factor for cardiovascular events in individuals with normal glucose regulation and pre-diabetes. This association exhibits a significant nonlinear threshold pattern. By integrating lipid metabolism and renal/inflammatory pathways, this combined indicator shows a combined predictive association that may be stronger than that of either marker alone, consistent with a statistical interaction pattern, and demonstrates good robustness across different subgroups. In contrast, its predictive value is substantially attenuated in individuals with diagnosed diabetes. Although AIP-Cys shows promise as a short-term risk marker in NGR and preDM populations, its long-term predictive performance remains modest. Therefore, the findings should be interpreted cautiously, particularly given that cardiovascular outcomes were based on self-reported physician diagnoses.

Our findings support the potential of AIP-Cys as a valuable, cost-effective novel biomarker for identifying individuals at high cardiovascular risk in the early stages before the onset of diabetes. This addresses a key need in the era of precision prevention, where strategies must be tailored to an individual’s metabolic status. By providing a tool that may improve risk stratification among those traditionally considered at lower or intermediate risk, this research contributes to the paradigm of earlier and more targeted cardiovascular disease prevention. Notably, the absence of a significant association in the DM population may be partially explained by the higher baseline use of antihypertensive and lipid-lowering medications in this subgroup, which could have modified the exposure–outcome relationship. Future research should aim to validate its utility in broader populations and explore interventions aimed at effectively modulating AIP and cystatin C levels to reduce cardiovascular risk.

Supplementary Information

Supplementary material 1^{(368KB, docx)}

Acknowledgements

We extend our sincere gratitude to the CHARLS research team for their dedication and tireless efforts in advancing the CHARLS project.

Abbreviations

CVD: Cardiovascular disease
AIP: Atherogenic index of plasma
AIP-Cys: The product of AIP × Cystatin C
cuAIP-Cys: Cumulative AIP-Cys
AIP * Cystatin C: The interaction term of AIP and Cystatin C
NGR: Normal glucose regulation
preDM: Pre-diabetes
DM: Diabetes mellitus
CHARLS: China health and retirement longitudinal study
HR: Hazard ratio
CI: Confidence interval
RCS: Restricted cubic splines
ROC: Receiver operating characteristic (curve)
AUC: Area under the ROC Curve
TG: Triglyceride
TC: Total cholesterol
HDL-C: High-density lipoprotein cholesterol
LDL-C: Low-density lipoprotein cholesterol
sdLDL: Small dense low-density lipoprotein
FPG: Fasting plasma glucose
HbA1c: Glycosylated hemoglobin A1c
SBP: Systolic blood pressure
DBP: Diastolic blood pressure
hs-CRP: High-sensitivity C-reactive protein
BMI: Body mass index
WC: Waist circumference
GVIF: Generalized variance inflation factor
VIF: Variance inflation factor
DF: Degrees of freedom
eGFR: Estimated glomerular filtration rate
NRI: Net reclassification improvement
IDI: Integrated discrimination improvement

Author contributions

KYW and YFZ conceived the study. KYW conducted the data analysis and drafted the manuscript. KYW and YFJ assisted in data collection and interpretation. LPZ and YFZ supervised the research. All authors critically reviewed the manuscript and approved the final version.

Funding

This work was supported by grants from National Natural Science Foundation of China (82570323), the Science and Technology Development Program of Jiangsu Province-Clinical Frontier Technology (BE2022754), Clinical Medicine Expert Team (Class A) of Jinji Lake Health Talents Program of Suzhou Industrial Park(SZYQTD202102), Suzhou Key Discipline for Medicine(SZXK202129), Demonstration of Scientific and Technological Innovation Project (SKY2021002), Suzhou Dedicated Project on Diagnosis and Treatment Technology of Major Diseases(LCZX202132), Research on Collaborative Innovation of medical engineering combination(SZM2021014), Research on Collaborative Innovation of medical engineering combination(SZM2022003), Suzhou Key Laboratory of Diagnosis and Treatment of Panvascular Diseases(SZS2023021), Gusu Talent Program (GSWS2022119). The funders had no roles in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The dataset used in this study is publicly available from the China Health and Retirement Longitudinal Study (CHARLS). Researchers can apply for data access by completing a registration and submitting a formal request via the official CHARLS website: http://charls.pku.edu.cn.

Declarations

Ethics approval and consent to participate

The CHARLS study was conducted in compliance with the ethical principles outlined in the Declaration of Helsinki. The research protocol received approval from the Institutional Review Board of Peking University. Prior to their participation, written informed consent was obtained from all individuals involved 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.

References

1.Zhao D, Liu J, Wang M, Zhang X, Zhou M. Epidemiology of cardiovascular disease in China: current features and implications. Nat Rev Cardiol. 2019;16(4):203–12. [DOI] [PubMed] [Google Scholar]
2.Timmis A, Aboyans V, Vardas P, Townsend N, Torbica A, Kavousi M, et al. European Society of Cardiology: the 2023 atlas of cardiovascular disease statistics. Eur Heart J. 2024;45(38):4019–62. [DOI] [PubMed] [Google Scholar]
3.Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global burden of cardiovascular diseases and risk factors, 1990–2019: update from the GBD 2019 study. J Am Coll Cardiol. 2020;76(25):2982–3021. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.McClellan M, Brown N, Califf RM, Warner JJ. Call to action: urgent challenges in cardiovascular disease: a presidential advisory from the American Heart Association. Circulation. 2019;139(9):e44–54. [DOI] [PubMed] [Google Scholar]
5.Li H, Ge J. Cardiovascular diseases in China: current status and future perspectives. Int J Cardiol Heart Vasc. 2015;6:25–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Tabák AG, Herder C, Rathmann W, Brunner EJ, Kivimäki M. Prediabetes: a high-risk state for diabetes development. Lancet. 2012;379(9833):2279–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Kannel WB, McGee DL. Diabetes and cardiovascular disease the framingham study. JAMA. 1979;241(19):2035–8. [DOI] [PubMed] [Google Scholar]
8.Wei M, Gaskill SP, Haffner SM, Stern MP. Effects of diabetes and level of glycemia on all-cause and cardiovascular mortality. The San Antonio Heart Study. Diabetes Care. 1998;21(7):1167–72. [DOI] [PubMed] [Google Scholar]
9.Nathan DM, Davidson MB, DeFronzo RA, Heine RJ, Henry RR, Pratley R, et al. Impaired fasting glucose and impaired glucose tolerance: implications for care. Diabetes Care. 2007;30(3):753–9. [DOI] [PubMed] [Google Scholar]
10.Jackson AM, Rørth R, Liu J, Kristensen SL, Anand IS, Claggett BL, et al. Diabetes and pre-diabetes in patients with heart failure and preserved ejection fraction. Eur J Heart Fail. 2022;24(3):497–509. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Yusuf S, Hawken S, Ounpuu S, Dans T, Avezum A, Lanas F, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 2004;364(9438):937–52. [DOI] [PubMed] [Google Scholar]
12.Echouffo-Tcheugui JB, Selvin E. Prediabetes and what it means: the epidemiological evidence. Annu Rev Public Health. 2021;42:59–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Rawshani A, Rawshani A, Franzén S, Eliasson B, Svensson AM, Miftaraj M, et al. Mortality and cardiovascular disease in type 1 and type 2 diabetes. N Engl J Med. 2017;376(15):1407–18. [DOI] [PubMed] [Google Scholar]
14.Rawshani A, Rawshani A, Franzén S, Sattar N, Eliasson B, Svensson AM, et al. Risk factors, mortality, and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2018;379(7):633–44. [DOI] [PubMed] [Google Scholar]
15.Fernández-Macías JC, Ochoa-Martínez AC, Varela-Silva JA, Pérez-Maldonado IN. Atherogenic index of plasma: novel predictive biomarker for cardiovascular illnesses. Arch Med Res. 2019;50(5):285–94. [DOI] [PubMed] [Google Scholar]
16.Wang L, Chen F, Xiaoqi C, Yujun C, Zijie L. Atherogenic index of plasma is an independent risk factor for coronary artery disease and a higher SYNTAX score. Angiology. 2021;72(2):181–6. [DOI] [PubMed] [Google Scholar]
17.Dobiásová M, Frohlich J. The plasma parameter log (TG/HDL-C) as an atherogenic index: correlation with lipoprotein particle size and esterification rate in apoB-lipoprotein-depleted plasma (FER(HDL)). Clin Biochem. 2001;34(7):583–8. [DOI] [PubMed] [Google Scholar]
18.Mikhailidis DP, Elisaf M, Rizzo M, Berneis K, Griffin B, Zambon A, et al. “European panel on low density lipoprotein (LDL) subclasses”: a statement on the pathophysiology, atherogenicity and clinical significance of LDL subclasses: executive summary. Curr Vasc Pharmacol. 2011;9(5):531–2. [DOI] [PubMed] [Google Scholar]
19.Amrita J, Mahajan M, Bhanwer AJ, Mohan G. Oxidative stress: an effective prognostic tool for an early detection of cardiovascular disease in menopausal women. Biochem Res Int. 2016. 10.1155/2016/6157605. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Hamzeh B, Pasdar Y, Mirzaei N, Faramani RS, Najafi F, Shakiba E, et al. Visceral adiposity index and atherogenic index of plasma as useful predictors of risk of cardiovascular diseases: evidence from a cohort study in Iran. Lipids Health Dis. 2021;20(1):82. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Hu Y, Liang Y, Li J, Li X, Yu M, Cui W. Correlation between atherogenic index of plasma and cardiovascular disease risk across Cardiovascular-kidney-metabolic syndrome stages 0-3: a nationwide prospective cohort study. Cardiovasc Diabetol. 2025;24(1):40. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Min Q, Wu Z, Yao J, Wang S, Duan L, Liu S, et al. Association between atherogenic index of plasma control level and incident cardiovascular disease in middle-aged and elderly Chinese individuals with abnormal glucose metabolism. Cardiovasc Diabetol. 2024;23(1):54. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Zhou K, Qin Z, Tian J, Cui K, Yan Y, Lyu S. The atherogenic index of plasma: a powerful and reliable predictor for coronary artery disease in patients with type 2 diabetes. Angiology. 2021;72(10):934–41. [DOI] [PubMed] [Google Scholar]
24.Huang X, Li C, Zeng P, Ling Y, Tan S, Bai Z, Shen S, Chen S, Nie B, Wang H et al. Non-traditional lipid-inflammatory parameters estimate the risk of stroke in middle-aged and older Chinese adults: a nationwide prospective cohort study. J Adv Res. 2025. [DOI] [PMC free article] [PubMed]
25.Kaneko R, Sawada S, Tokita A, Honkura R, Tamura N, Kodama S, et al. Serum cystatin C level is associated with carotid arterial wall elasticity in subjects with type 2 diabetes mellitus: a potential marker of early-stage atherosclerosis. Diabetes Res Clin Pract. 2018;139:43–51. [DOI] [PubMed] [Google Scholar]
26.Dabravolski SA, Elizova NV, Korchagina ER, Maltseva ON, Pavshintsev VV, Orekhov AN. Cystatin C in atherosclerotic cardiovascular disease. Clin Chim Acta. 2025;580:120721. [DOI] [PubMed] [Google Scholar]
27.Wang Y, Zhang Y, Ma Q, Wang C, Xu Y, Sun H, et al. Determination of clinical cut-off values for serum cystatin C levels to predict ischemic stroke risk. J Stroke Cerebrovasc Dis. 2019;28(11):104345. [DOI] [PubMed] [Google Scholar]
28.Gatto L, Prati F. Subclinical atherosclerosis: how and when to treat it? Eur Heart J Suppl. 2020;22(Suppl E):E87-e90. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Grodecki K, Geers J, Kwiecinski J, Lin A, Slipczuk L, Slomka PJ, et al. Phenotyping atherosclerotic plaque and perivascular adipose tissue: signalling pathways and clinical biomarkers in atherosclerosis. Nat Rev Cardiol. 2025;22(6):443–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Ridker PM, Everett BM, Thuren T, MacFadyen JG, Chang WH, Ballantyne C, et al. Antiinflammatory therapy with canakinumab for atherosclerotic disease. N Engl J Med. 2017;377(12):1119–31. [DOI] [PubMed] [Google Scholar]
31.Tall AR, Yvan-Charvet L. Cholesterol, inflammation and innate immunity. Nat Rev Immunol. 2015;15(2):104–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Hansson GK, Hermansson A. The immune system in atherosclerosis. Nat Immunol. 2011;12(3):204–12. [DOI] [PubMed] [Google Scholar]
33.Libby P. The molecular mechanisms of the thrombotic complications of atherosclerosis. J Intern Med. 2008;263(5):517–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Zhou X-J, Zhang H. The genetics of IgA nephropathy: implications for future therapies. Semin Nephrol. 2025;44(5):151567. [DOI] [PubMed] [Google Scholar]
35.Novak J, Reily C, Steers NJ, Schumann T, Rizk DV, Julian BA, et al. Emerging biochemical and immunologic mechanisms in the pathogenesis of IgA nephropathy. Semin Nephrol. 2025;44(5):151565. [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Gleeson PJ, Monteiro RC. The role of mucosal immunity: what can we learn from animal and human studies? Semin Nephrol. 2025;44(5):151566. [DOI] [PubMed] [Google Scholar]
37.Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China health and retirement longitudinal study (CHARLS). Int J Epidemiol. 2014;43(1):61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Qu L, Fang S, Lan Z, Xu S, Jiang J, Pan Y, et al. Association between atherogenic index of plasma and new-onset stroke in individuals with different glucose metabolism status: insights from CHARLS. Cardiovasc Diabetol. 2024;23(1):215. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Assempoor R, Daneshvar MS, Taghvaei A, Abroy AS, Azimi A, Nelson JR, et al. Atherogenic index of plasma and coronary artery disease: a systematic review and meta-analysis of observational studies. Cardiovasc Diabetol. 2025;24(1):35. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Zhang L, Li S, Liu D, Gui J, Hu J, Wang Q, et al. The relationship between C-reactive protein-triglyceride-glucose index and cardiovascular disease: insights from the China health and retirement longitudinal study (CHARLS). Cardiovasc Diabetol. 2025;24(1):410. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Zhang N, Xie G, Jian G, Jiang K, Lu Z, Wang Z, et al. Association of cumulative changes in atherogenic index of plasma and its obesity-related derivatives with cardiovascular disease: emphasis on incremental risk of diabetes status. Cardiovasc Diabetol. 2025;24(1):435. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Zou Y, Lu S, Li D, Huang X, Wang C, Xie G, et al. Exposure of cumulative atherogenic index of plasma and the development of prediabetes in middle-aged and elderly individuals: evidence from the CHARLS cohort study. Cardiovasc Diabetol. 2024;23(1):355. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.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. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.VanderWeele TJ. Causal mediation analysis with survival data. Epidemiology. 2011;22(4):582–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Lai H, Tu Y, Liao C, Zhang S, He L, Li J. Joint assessment of abdominal obesity and non-traditional lipid parameters for primary prevention of cardiometabolic multimorbidity: insights from the China health and retirement longitudinal study 2011–2018. Cardiovasc Diabetol. 2025;24(1):109. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167(4):268–74. [DOI] [PubMed] [Google Scholar]
47.Huang X, Li C, Zeng P, Ling Y, Tan S, Bai Z, et al. Non-traditional lipid-inflammatory parameters estimate the risk of stroke in middle-aged and older Chinese adults: a nationwide prospective cohort study. J Adv Res. 2025. 10.1016/j.jare.2025.11.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Li C, Luo X, Chen Y, Lin J, Gu S. Predictive value of an integrated insulin resistance and lipometabolic score for cardiometabolic multimorbidity in older adults: a UK cohort study. Cardiovasc Diabetol. 2026. 10.1186/s12933-025-03064-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Rabiee Rad M, Ghasempour Dabaghi G, Darouei B, Amani-Beni R. The association of atherogenic index of plasma with cardiovascular outcomes in patients with coronary artery disease: a systematic review and meta-analysis. Cardiovasc Diabetol. 2024;23(1):119. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Zheng H, Wu K, Wu W, Chen G, Chen Z, Cai Z, et al. Relationship between the cumulative exposure to atherogenic index of plasma and ischemic stroke: a retrospective cohort study. Cardiovasc Diabetol. 2023;22(1):313. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Vekic J, Zeljkovic A, Cicero AFG, Janez A, Stoian AP, Sonmez A, et al. Atherosclerosis development and progression: the role of atherogenic small, dense LDL. Medicina (Kaunas). 2022. 10.3390/medicina58020299. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Liu J, Yang R, Zhou M, Mao W, Li H, Zhao H, et al. Fractional esterification rate of cholesterol in high-density lipoprotein associates with risk of coronary heart disease. Lipids Health Dis. 2017;16(1):162. [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Dobiásová M, Frohlich J. The new atherogenic plasma index reflects the triglyceride and HDL-cholesterol ratio, the lipoprotein particle size and the cholesterol esterification rate: changes during lipanor therapy. Vnitr Lek. 2000;46(3):152–6. [PubMed] [Google Scholar]
54.Dong J, Yu S, Yang R, Li H, Guo H, Zhao H, et al. A simple and precise method for direct measurement of fractional esterification rate of high density lipoprotein cholesterol by high performance liquid chromatography. Clin Chem Lab Med. 2014;52(4):557–64. [DOI] [PubMed] [Google Scholar]
55.Gray RS, Robbins DC, Wang W, Yeh JL, Fabsitz RR, Cowan LD, et al. Relation of LDL size to the insulin resistance syndrome and coronary heart disease in American Indians. The Strong Heart Study. Arterioscler Thromb Vasc Biol. 1997;17(11):2713–20. [DOI] [PubMed] [Google Scholar]
56.Frontoni S, Bracaglia D, Gigli F. Relationship between autonomic dysfunction, insulin resistance and hypertension, in diabetes. Nutr Metab Cardiovasc Dis. 2005;15(6):441–9. [DOI] [PubMed] [Google Scholar]
57.Rosenberg DE, Jabbour SA, Goldstein BJ. Insulin resistance, diabetes and cardiovascular risk: approaches to treatment. Diabetes Obes Metab. 2005;7(6):642–53. [DOI] [PubMed] [Google Scholar]
58.Astor BC, Shafi T, Hoogeveen RC, Matsushita K, Ballantyne CM, Inker LA, et al. Novel markers of kidney function as predictors of ESRD, cardiovascular disease, and mortality in the general population. Am J Kidney Dis. 2012;59(5):653–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Angelidis C, Deftereos S, Giannopoulos G, Anatoliotakis N, Bouras G, Hatzis G, et al. Cystatin C: an emerging biomarker in cardiovascular disease. Curr Top Med Chem. 2013;13(2):164–79. [DOI] [PubMed] [Google Scholar]
60.Lee M, Saver JL, Huang WH, Chow J, Chang KH, Ovbiagele B. Impact of elevated cystatin C level on cardiovascular disease risk in predominantly high cardiovascular risk populations: a meta-analysis. Circ Cardiovasc Qual Outcomes. 2010;3(6):675–83. [DOI] [PubMed] [Google Scholar]
61.Bi M, Huang Z, Li P, Cheng C, Huang Y, Chen W. The association between elevated cystatin C levels with myocardial infarction: a meta-analysis. Int J Clin Exp Med. 2015;8(11):20540–7. [PMC free article] [PubMed] [Google Scholar]
62.Svensson-Färbom P, Ohlson Andersson M, Almgren P, Hedblad B, Engström G, Persson M, et al. Cystatin C identifies cardiovascular risk better than creatinine-based estimates of glomerular filtration in middle-aged individuals without a history of cardiovascular disease. J Intern Med. 2014;275(5):506–21. [DOI] [PubMed] [Google Scholar]
63.Zhang Y, Yang S, Chen J, Zhang Z, He P, Zhou C, et al. Associations of serum cystatin C and its change with new-onset cardiovascular disease in Chinese general population. Nutr Metab Cardiovasc Dis. 2022;32(8):1963–71. [DOI] [PubMed] [Google Scholar]
64.Liu J, Sukhova GK, Sun JS, Xu WH, Libby P, Shi GP. Lysosomal cysteine proteases in atherosclerosis. Arterioscler Thromb Vasc Biol. 2004;24(8):1359–66. [DOI] [PubMed] [Google Scholar]
65.Frostegård J. Immunity, atherosclerosis and cardiovascular disease. BMC Med. 2013;11:117. [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Zhang Y, Sun L. Cystatin C in cerebrovascular disorders. Curr Neurovasc Res. 2017;14(4):406–14. [DOI] [PubMed] [Google Scholar]
67.González KA, Stickel AM, Kaur SS, Ramos AR, González HM, Tarraf W. Serum cystatin-C is linked to increased prevalence of diabetes and higher risk of mortality in diverse middle-aged and older adults. PLoS ONE. 2022;17(9):e0270289. [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Sahakyan K, Lee KE, Shankar A, Klein R. Serum cystatin C and the incidence of type 2 diabetes mellitus. Diabetologia. 2011;54(6):1335–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Goldstein JL, Ho YK, Basu SK, Brown MS. Binding site on macrophages that mediates uptake and degradation of acetylated low density lipoprotein, producing massive cholesterol deposition. Proc Natl Acad Sci U S A. 1979;76(1):333–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Lusis AJ. Atherosclerosis. Nature. 2000;407(6801):233–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
71.Bristow MR, Ginsburg R, Minobe W, Cubicciotti RS, Sageman WS, Lurie K, et al. Decreased catecholamine sensitivity and beta-adrenergic-receptor density in failing human hearts. N Engl J Med. 1982;307(4):205–11. [DOI] [PubMed] [Google Scholar]
72.Zaman S, Wasfy JH, Kapil V, Ziaeian B, Parsonage WA, Sriswasdi S, et al. The Lancet Commission on rethinking coronary artery disease: moving from ischaemia to atheroma. Lancet (London, England). 2025;405(10486):1264–312. [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Gaddam RR, Pathuri M, Deb P, Dwivedi S, Vikram A, Kasina V, et al. Engineered miR-122 inhibitors preserve endothelial mitochondrial function and prevent vascular dysfunction in obesity-associated prediabetes. Mol Ther Nucleic Acids. 2026;37(1):102830. [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Sardu C, Nicoletti GF, Trotta MC, Pieretti G, Gatta G, Grella R, et al. Endothelial dysfunction and cardiovascular outcomes in pre-menopausal women with high vs low breast fat accumulation the role of TSP-1. Am J Prev Cardiol. 2025;24:101349. [DOI] [PMC free article] [PubMed] [Google Scholar]
75.Liu J, Chen K, Tang M, Mu Q, Zhang S, Li J, et al. Oxidative stress and inflammation mediate the adverse effects of cadmium exposure on all-cause and cause-specific mortality in patients with diabetes and prediabetes. Cardiovasc Diabetol. 2025;24(1):145. [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Romeo S, Vidal-Puig A, Husain M, Ahima R, Arca M, Bhatt DL, et al. Clinical staging to guide management of metabolic disorders and their sequelae: a European atherosclerosis society consensus statement. Eur Heart J. 2025;46(38):3685–713. [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Abdul-Ghani MA, Tripathy D, DeFronzo RA. Contributions of beta-cell dysfunction and insulin resistance to the pathogenesis of impaired glucose tolerance and impaired fasting glucose. Diabetes Care. 2006;29(5):1130–9. [DOI] [PubMed] [Google Scholar]
78.Ingelsson E, Sundström J, Arnlöv J, Zethelius B, Lind L. Insulin resistance and risk of congestive heart failure. JAMA. 2005;294(3):334–41. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary material 1^{(368KB, docx)}

Data Availability Statement

[CR1] 1.Zhao D, Liu J, Wang M, Zhang X, Zhou M. Epidemiology of cardiovascular disease in China: current features and implications. Nat Rev Cardiol. 2019;16(4):203–12. [DOI] [PubMed] [Google Scholar]

[CR2] 2.Timmis A, Aboyans V, Vardas P, Townsend N, Torbica A, Kavousi M, et al. European Society of Cardiology: the 2023 atlas of cardiovascular disease statistics. Eur Heart J. 2024;45(38):4019–62. [DOI] [PubMed] [Google Scholar]

[CR3] 3.Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global burden of cardiovascular diseases and risk factors, 1990–2019: update from the GBD 2019 study. J Am Coll Cardiol. 2020;76(25):2982–3021. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR4] 4.McClellan M, Brown N, Califf RM, Warner JJ. Call to action: urgent challenges in cardiovascular disease: a presidential advisory from the American Heart Association. Circulation. 2019;139(9):e44–54. [DOI] [PubMed] [Google Scholar]

[CR5] 5.Li H, Ge J. Cardiovascular diseases in China: current status and future perspectives. Int J Cardiol Heart Vasc. 2015;6:25–31. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR6] 6.Tabák AG, Herder C, Rathmann W, Brunner EJ, Kivimäki M. Prediabetes: a high-risk state for diabetes development. Lancet. 2012;379(9833):2279–90. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Kannel WB, McGee DL. Diabetes and cardiovascular disease the framingham study. JAMA. 1979;241(19):2035–8. [DOI] [PubMed] [Google Scholar]

[CR8] 8.Wei M, Gaskill SP, Haffner SM, Stern MP. Effects of diabetes and level of glycemia on all-cause and cardiovascular mortality. The San Antonio Heart Study. Diabetes Care. 1998;21(7):1167–72. [DOI] [PubMed] [Google Scholar]

[CR9] 9.Nathan DM, Davidson MB, DeFronzo RA, Heine RJ, Henry RR, Pratley R, et al. Impaired fasting glucose and impaired glucose tolerance: implications for care. Diabetes Care. 2007;30(3):753–9. [DOI] [PubMed] [Google Scholar]

[CR10] 10.Jackson AM, Rørth R, Liu J, Kristensen SL, Anand IS, Claggett BL, et al. Diabetes and pre-diabetes in patients with heart failure and preserved ejection fraction. Eur J Heart Fail. 2022;24(3):497–509. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.Yusuf S, Hawken S, Ounpuu S, Dans T, Avezum A, Lanas F, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 2004;364(9438):937–52. [DOI] [PubMed] [Google Scholar]

[CR12] 12.Echouffo-Tcheugui JB, Selvin E. Prediabetes and what it means: the epidemiological evidence. Annu Rev Public Health. 2021;42:59–77. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR13] 13.Rawshani A, Rawshani A, Franzén S, Eliasson B, Svensson AM, Miftaraj M, et al. Mortality and cardiovascular disease in type 1 and type 2 diabetes. N Engl J Med. 2017;376(15):1407–18. [DOI] [PubMed] [Google Scholar]

[CR14] 14.Rawshani A, Rawshani A, Franzén S, Sattar N, Eliasson B, Svensson AM, et al. Risk factors, mortality, and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2018;379(7):633–44. [DOI] [PubMed] [Google Scholar]

[CR15] 15.Fernández-Macías JC, Ochoa-Martínez AC, Varela-Silva JA, Pérez-Maldonado IN. Atherogenic index of plasma: novel predictive biomarker for cardiovascular illnesses. Arch Med Res. 2019;50(5):285–94. [DOI] [PubMed] [Google Scholar]

[CR16] 16.Wang L, Chen F, Xiaoqi C, Yujun C, Zijie L. Atherogenic index of plasma is an independent risk factor for coronary artery disease and a higher SYNTAX score. Angiology. 2021;72(2):181–6. [DOI] [PubMed] [Google Scholar]

[CR17] 17.Dobiásová M, Frohlich J. The plasma parameter log (TG/HDL-C) as an atherogenic index: correlation with lipoprotein particle size and esterification rate in apoB-lipoprotein-depleted plasma (FER(HDL)). Clin Biochem. 2001;34(7):583–8. [DOI] [PubMed] [Google Scholar]

[CR18] 18.Mikhailidis DP, Elisaf M, Rizzo M, Berneis K, Griffin B, Zambon A, et al. “European panel on low density lipoprotein (LDL) subclasses”: a statement on the pathophysiology, atherogenicity and clinical significance of LDL subclasses: executive summary. Curr Vasc Pharmacol. 2011;9(5):531–2. [DOI] [PubMed] [Google Scholar]

[CR19] 19.Amrita J, Mahajan M, Bhanwer AJ, Mohan G. Oxidative stress: an effective prognostic tool for an early detection of cardiovascular disease in menopausal women. Biochem Res Int. 2016. 10.1155/2016/6157605. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR20] 20.Hamzeh B, Pasdar Y, Mirzaei N, Faramani RS, Najafi F, Shakiba E, et al. Visceral adiposity index and atherogenic index of plasma as useful predictors of risk of cardiovascular diseases: evidence from a cohort study in Iran. Lipids Health Dis. 2021;20(1):82. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Hu Y, Liang Y, Li J, Li X, Yu M, Cui W. Correlation between atherogenic index of plasma and cardiovascular disease risk across Cardiovascular-kidney-metabolic syndrome stages 0-3: a nationwide prospective cohort study. Cardiovasc Diabetol. 2025;24(1):40. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Min Q, Wu Z, Yao J, Wang S, Duan L, Liu S, et al. Association between atherogenic index of plasma control level and incident cardiovascular disease in middle-aged and elderly Chinese individuals with abnormal glucose metabolism. Cardiovasc Diabetol. 2024;23(1):54. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Zhou K, Qin Z, Tian J, Cui K, Yan Y, Lyu S. The atherogenic index of plasma: a powerful and reliable predictor for coronary artery disease in patients with type 2 diabetes. Angiology. 2021;72(10):934–41. [DOI] [PubMed] [Google Scholar]

[CR24] 24.Huang X, Li C, Zeng P, Ling Y, Tan S, Bai Z, Shen S, Chen S, Nie B, Wang H et al. Non-traditional lipid-inflammatory parameters estimate the risk of stroke in middle-aged and older Chinese adults: a nationwide prospective cohort study. J Adv Res. 2025. [DOI] [PMC free article] [PubMed]

[CR25] 25.Kaneko R, Sawada S, Tokita A, Honkura R, Tamura N, Kodama S, et al. Serum cystatin C level is associated with carotid arterial wall elasticity in subjects with type 2 diabetes mellitus: a potential marker of early-stage atherosclerosis. Diabetes Res Clin Pract. 2018;139:43–51. [DOI] [PubMed] [Google Scholar]

[CR26] 26.Dabravolski SA, Elizova NV, Korchagina ER, Maltseva ON, Pavshintsev VV, Orekhov AN. Cystatin C in atherosclerotic cardiovascular disease. Clin Chim Acta. 2025;580:120721. [DOI] [PubMed] [Google Scholar]

[CR27] 27.Wang Y, Zhang Y, Ma Q, Wang C, Xu Y, Sun H, et al. Determination of clinical cut-off values for serum cystatin C levels to predict ischemic stroke risk. J Stroke Cerebrovasc Dis. 2019;28(11):104345. [DOI] [PubMed] [Google Scholar]

[CR28] 28.Gatto L, Prati F. Subclinical atherosclerosis: how and when to treat it? Eur Heart J Suppl. 2020;22(Suppl E):E87-e90. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR29] 29.Grodecki K, Geers J, Kwiecinski J, Lin A, Slipczuk L, Slomka PJ, et al. Phenotyping atherosclerotic plaque and perivascular adipose tissue: signalling pathways and clinical biomarkers in atherosclerosis. Nat Rev Cardiol. 2025;22(6):443–55. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Ridker PM, Everett BM, Thuren T, MacFadyen JG, Chang WH, Ballantyne C, et al. Antiinflammatory therapy with canakinumab for atherosclerotic disease. N Engl J Med. 2017;377(12):1119–31. [DOI] [PubMed] [Google Scholar]

[CR31] 31.Tall AR, Yvan-Charvet L. Cholesterol, inflammation and innate immunity. Nat Rev Immunol. 2015;15(2):104–16. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Hansson GK, Hermansson A. The immune system in atherosclerosis. Nat Immunol. 2011;12(3):204–12. [DOI] [PubMed] [Google Scholar]

[CR33] 33.Libby P. The molecular mechanisms of the thrombotic complications of atherosclerosis. J Intern Med. 2008;263(5):517–27. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Zhou X-J, Zhang H. The genetics of IgA nephropathy: implications for future therapies. Semin Nephrol. 2025;44(5):151567. [DOI] [PubMed] [Google Scholar]

[CR35] 35.Novak J, Reily C, Steers NJ, Schumann T, Rizk DV, Julian BA, et al. Emerging biochemical and immunologic mechanisms in the pathogenesis of IgA nephropathy. Semin Nephrol. 2025;44(5):151565. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR36] 36.Gleeson PJ, Monteiro RC. The role of mucosal immunity: what can we learn from animal and human studies? Semin Nephrol. 2025;44(5):151566. [DOI] [PubMed] [Google Scholar]

[CR37] 37.Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China health and retirement longitudinal study (CHARLS). Int J Epidemiol. 2014;43(1):61–8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR38] 38.Qu L, Fang S, Lan Z, Xu S, Jiang J, Pan Y, et al. Association between atherogenic index of plasma and new-onset stroke in individuals with different glucose metabolism status: insights from CHARLS. Cardiovasc Diabetol. 2024;23(1):215. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR39] 39.Assempoor R, Daneshvar MS, Taghvaei A, Abroy AS, Azimi A, Nelson JR, et al. Atherogenic index of plasma and coronary artery disease: a systematic review and meta-analysis of observational studies. Cardiovasc Diabetol. 2025;24(1):35. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR40] 40.Zhang L, Li S, Liu D, Gui J, Hu J, Wang Q, et al. The relationship between C-reactive protein-triglyceride-glucose index and cardiovascular disease: insights from the China health and retirement longitudinal study (CHARLS). Cardiovasc Diabetol. 2025;24(1):410. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR41] 41.Zhang N, Xie G, Jian G, Jiang K, Lu Z, Wang Z, et al. Association of cumulative changes in atherogenic index of plasma and its obesity-related derivatives with cardiovascular disease: emphasis on incremental risk of diabetes status. Cardiovasc Diabetol. 2025;24(1):435. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR42] 42.Zou Y, Lu S, Li D, Huang X, Wang C, Xie G, et al. Exposure of cumulative atherogenic index of plasma and the development of prediabetes in middle-aged and elderly individuals: evidence from the CHARLS cohort study. Cardiovasc Diabetol. 2024;23(1):355. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.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. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.VanderWeele TJ. Causal mediation analysis with survival data. Epidemiology. 2011;22(4):582–5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR45] 45.Lai H, Tu Y, Liao C, Zhang S, He L, Li J. Joint assessment of abdominal obesity and non-traditional lipid parameters for primary prevention of cardiometabolic multimorbidity: insights from the China health and retirement longitudinal study 2011–2018. Cardiovasc Diabetol. 2025;24(1):109. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167(4):268–74. [DOI] [PubMed] [Google Scholar]

[CR47] 47.Huang X, Li C, Zeng P, Ling Y, Tan S, Bai Z, et al. Non-traditional lipid-inflammatory parameters estimate the risk of stroke in middle-aged and older Chinese adults: a nationwide prospective cohort study. J Adv Res. 2025. 10.1016/j.jare.2025.11.029. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Li C, Luo X, Chen Y, Lin J, Gu S. Predictive value of an integrated insulin resistance and lipometabolic score for cardiometabolic multimorbidity in older adults: a UK cohort study. Cardiovasc Diabetol. 2026. 10.1186/s12933-025-03064-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR49] 49.Rabiee Rad M, Ghasempour Dabaghi G, Darouei B, Amani-Beni R. The association of atherogenic index of plasma with cardiovascular outcomes in patients with coronary artery disease: a systematic review and meta-analysis. Cardiovasc Diabetol. 2024;23(1):119. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR50] 50.Zheng H, Wu K, Wu W, Chen G, Chen Z, Cai Z, et al. Relationship between the cumulative exposure to atherogenic index of plasma and ischemic stroke: a retrospective cohort study. Cardiovasc Diabetol. 2023;22(1):313. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR51] 51.Vekic J, Zeljkovic A, Cicero AFG, Janez A, Stoian AP, Sonmez A, et al. Atherosclerosis development and progression: the role of atherogenic small, dense LDL. Medicina (Kaunas). 2022. 10.3390/medicina58020299. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR52] 52.Liu J, Yang R, Zhou M, Mao W, Li H, Zhao H, et al. Fractional esterification rate of cholesterol in high-density lipoprotein associates with risk of coronary heart disease. Lipids Health Dis. 2017;16(1):162. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR53] 53.Dobiásová M, Frohlich J. The new atherogenic plasma index reflects the triglyceride and HDL-cholesterol ratio, the lipoprotein particle size and the cholesterol esterification rate: changes during lipanor therapy. Vnitr Lek. 2000;46(3):152–6. [PubMed] [Google Scholar]

[CR54] 54.Dong J, Yu S, Yang R, Li H, Guo H, Zhao H, et al. A simple and precise method for direct measurement of fractional esterification rate of high density lipoprotein cholesterol by high performance liquid chromatography. Clin Chem Lab Med. 2014;52(4):557–64. [DOI] [PubMed] [Google Scholar]

[CR55] 55.Gray RS, Robbins DC, Wang W, Yeh JL, Fabsitz RR, Cowan LD, et al. Relation of LDL size to the insulin resistance syndrome and coronary heart disease in American Indians. The Strong Heart Study. Arterioscler Thromb Vasc Biol. 1997;17(11):2713–20. [DOI] [PubMed] [Google Scholar]

[CR56] 56.Frontoni S, Bracaglia D, Gigli F. Relationship between autonomic dysfunction, insulin resistance and hypertension, in diabetes. Nutr Metab Cardiovasc Dis. 2005;15(6):441–9. [DOI] [PubMed] [Google Scholar]

[CR57] 57.Rosenberg DE, Jabbour SA, Goldstein BJ. Insulin resistance, diabetes and cardiovascular risk: approaches to treatment. Diabetes Obes Metab. 2005;7(6):642–53. [DOI] [PubMed] [Google Scholar]

[CR58] 58.Astor BC, Shafi T, Hoogeveen RC, Matsushita K, Ballantyne CM, Inker LA, et al. Novel markers of kidney function as predictors of ESRD, cardiovascular disease, and mortality in the general population. Am J Kidney Dis. 2012;59(5):653–62. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR59] 59.Angelidis C, Deftereos S, Giannopoulos G, Anatoliotakis N, Bouras G, Hatzis G, et al. Cystatin C: an emerging biomarker in cardiovascular disease. Curr Top Med Chem. 2013;13(2):164–79. [DOI] [PubMed] [Google Scholar]

[CR60] 60.Lee M, Saver JL, Huang WH, Chow J, Chang KH, Ovbiagele B. Impact of elevated cystatin C level on cardiovascular disease risk in predominantly high cardiovascular risk populations: a meta-analysis. Circ Cardiovasc Qual Outcomes. 2010;3(6):675–83. [DOI] [PubMed] [Google Scholar]

[CR61] 61.Bi M, Huang Z, Li P, Cheng C, Huang Y, Chen W. The association between elevated cystatin C levels with myocardial infarction: a meta-analysis. Int J Clin Exp Med. 2015;8(11):20540–7. [PMC free article] [PubMed] [Google Scholar]

[CR62] 62.Svensson-Färbom P, Ohlson Andersson M, Almgren P, Hedblad B, Engström G, Persson M, et al. Cystatin C identifies cardiovascular risk better than creatinine-based estimates of glomerular filtration in middle-aged individuals without a history of cardiovascular disease. J Intern Med. 2014;275(5):506–21. [DOI] [PubMed] [Google Scholar]

[CR63] 63.Zhang Y, Yang S, Chen J, Zhang Z, He P, Zhou C, et al. Associations of serum cystatin C and its change with new-onset cardiovascular disease in Chinese general population. Nutr Metab Cardiovasc Dis. 2022;32(8):1963–71. [DOI] [PubMed] [Google Scholar]

[CR64] 64.Liu J, Sukhova GK, Sun JS, Xu WH, Libby P, Shi GP. Lysosomal cysteine proteases in atherosclerosis. Arterioscler Thromb Vasc Biol. 2004;24(8):1359–66. [DOI] [PubMed] [Google Scholar]

[CR65] 65.Frostegård J. Immunity, atherosclerosis and cardiovascular disease. BMC Med. 2013;11:117. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR66] 66.Zhang Y, Sun L. Cystatin C in cerebrovascular disorders. Curr Neurovasc Res. 2017;14(4):406–14. [DOI] [PubMed] [Google Scholar]

[CR67] 67.González KA, Stickel AM, Kaur SS, Ramos AR, González HM, Tarraf W. Serum cystatin-C is linked to increased prevalence of diabetes and higher risk of mortality in diverse middle-aged and older adults. PLoS ONE. 2022;17(9):e0270289. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR68] 68.Sahakyan K, Lee KE, Shankar A, Klein R. Serum cystatin C and the incidence of type 2 diabetes mellitus. Diabetologia. 2011;54(6):1335–40. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR69] 69.Goldstein JL, Ho YK, Basu SK, Brown MS. Binding site on macrophages that mediates uptake and degradation of acetylated low density lipoprotein, producing massive cholesterol deposition. Proc Natl Acad Sci U S A. 1979;76(1):333–7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR70] 70.Lusis AJ. Atherosclerosis. Nature. 2000;407(6801):233–41. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR71] 71.Bristow MR, Ginsburg R, Minobe W, Cubicciotti RS, Sageman WS, Lurie K, et al. Decreased catecholamine sensitivity and beta-adrenergic-receptor density in failing human hearts. N Engl J Med. 1982;307(4):205–11. [DOI] [PubMed] [Google Scholar]

[CR72] 72.Zaman S, Wasfy JH, Kapil V, Ziaeian B, Parsonage WA, Sriswasdi S, et al. The Lancet Commission on rethinking coronary artery disease: moving from ischaemia to atheroma. Lancet (London, England). 2025;405(10486):1264–312. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR73] 73.Gaddam RR, Pathuri M, Deb P, Dwivedi S, Vikram A, Kasina V, et al. Engineered miR-122 inhibitors preserve endothelial mitochondrial function and prevent vascular dysfunction in obesity-associated prediabetes. Mol Ther Nucleic Acids. 2026;37(1):102830. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR74] 74.Sardu C, Nicoletti GF, Trotta MC, Pieretti G, Gatta G, Grella R, et al. Endothelial dysfunction and cardiovascular outcomes in pre-menopausal women with high vs low breast fat accumulation the role of TSP-1. Am J Prev Cardiol. 2025;24:101349. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR75] 75.Liu J, Chen K, Tang M, Mu Q, Zhang S, Li J, et al. Oxidative stress and inflammation mediate the adverse effects of cadmium exposure on all-cause and cause-specific mortality in patients with diabetes and prediabetes. Cardiovasc Diabetol. 2025;24(1):145. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR76] 76.Romeo S, Vidal-Puig A, Husain M, Ahima R, Arca M, Bhatt DL, et al. Clinical staging to guide management of metabolic disorders and their sequelae: a European atherosclerosis society consensus statement. Eur Heart J. 2025;46(38):3685–713. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR77] 77.Abdul-Ghani MA, Tripathy D, DeFronzo RA. Contributions of beta-cell dysfunction and insulin resistance to the pathogenesis of impaired glucose tolerance and impaired fasting glucose. Diabetes Care. 2006;29(5):1130–9. [DOI] [PubMed] [Google Scholar]

[CR78] 78.Ingelsson E, Sundström J, Arnlöv J, Zethelius B, Lind L. Insulin resistance and risk of congestive heart failure. JAMA. 2005;294(3):334–41. [DOI] [PubMed] [Google Scholar]

PERMALINK

Synergistic effect of AIP and cystatin C: a study on cardiovascular risk prediction in populations with different glycemic statuses

Kuangyi Wang

Yufeng Jiang

Liangping Zhao

Yafeng Zhou

Abstract

Background

Materials and methods

Results

Conclusions

Graphical abstract

Supplementary Information

Background

Materials and methods

Study design and population

Fig. 1.

Definition of exposure and outcome variables

Definition of glycemic status

Definition of covariates

Statistical analysis

Results

Participants characteristics

Table 1.

Association between AIP-cys and CVD risk

Table 2.

Fig. 2.

Justification for the multiplicative combination

Nonlinear association between AIP-cys and CVD risk

Fig. 3.

Table 3.

Subgroup analysis

Fig. 4.

Predictive performance of AIP-cys

Fig. 5.

Mediation analysis

Fig. 6.

Sensitivity analysis

Discussion

Strengths and limitations

Conclusions

Supplementary Information

Acknowledgements

Abbreviations

Author contributions

Funding

Data availability

Declarations

Ethics approval and consent to participate

Competing interests

Footnotes

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases